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>
131 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
134 <li><a href="#int_os">Operating System Intrinsics</a>
136 <li><a href="#i_readport">'<tt>llvm.readport</tt>' Intrinsic</a></li>
137 <li><a href="#i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a></li>
138 <li><a href="#i_readio">'<tt>llvm.readio</tt>' Intrinsic</a></li>
139 <li><a href="#i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a></li>
141 <li><a href="#int_libc">Standard C Library Intrinsics</a>
143 <li><a href="#i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a></li>
144 <li><a href="#i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a></li>
145 <li><a href="#i_memset">'<tt>llvm.memset</tt>' Intrinsic</a></li>
146 <li><a href="#i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a></li>
147 <li><a href="#i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a></li>
151 <li><a href="#int_count">Bit counting Intrinsics</a>
153 <li><a href="#int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic </a></li>
154 <li><a href="#int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic </a></li>
155 <li><a href="#int_cttz">'<tt>llvm.cttz</tt>' Intrinsic </a></li>
158 <li><a href="#int_debugger">Debugger intrinsics</a></li>
163 <div class="doc_author">
164 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
165 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
168 <!-- *********************************************************************** -->
169 <div class="doc_section"> <a name="abstract">Abstract </a></div>
170 <!-- *********************************************************************** -->
172 <div class="doc_text">
173 <p>This document is a reference manual for the LLVM assembly language.
174 LLVM is an SSA based representation that provides type safety,
175 low-level operations, flexibility, and the capability of representing
176 'all' high-level languages cleanly. It is the common code
177 representation used throughout all phases of the LLVM compilation
181 <!-- *********************************************************************** -->
182 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
183 <!-- *********************************************************************** -->
185 <div class="doc_text">
187 <p>The LLVM code representation is designed to be used in three
188 different forms: as an in-memory compiler IR, as an on-disk bytecode
189 representation (suitable for fast loading by a Just-In-Time compiler),
190 and as a human readable assembly language representation. This allows
191 LLVM to provide a powerful intermediate representation for efficient
192 compiler transformations and analysis, while providing a natural means
193 to debug and visualize the transformations. The three different forms
194 of LLVM are all equivalent. This document describes the human readable
195 representation and notation.</p>
197 <p>The LLVM representation aims to be light-weight and low-level
198 while being expressive, typed, and extensible at the same time. It
199 aims to be a "universal IR" of sorts, by being at a low enough level
200 that high-level ideas may be cleanly mapped to it (similar to how
201 microprocessors are "universal IR's", allowing many source languages to
202 be mapped to them). By providing type information, LLVM can be used as
203 the target of optimizations: for example, through pointer analysis, it
204 can be proven that a C automatic variable is never accessed outside of
205 the current function... allowing it to be promoted to a simple SSA
206 value instead of a memory location.</p>
210 <!-- _______________________________________________________________________ -->
211 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
213 <div class="doc_text">
215 <p>It is important to note that this document describes 'well formed'
216 LLVM assembly language. There is a difference between what the parser
217 accepts and what is considered 'well formed'. For example, the
218 following instruction is syntactically okay, but not well formed:</p>
221 %x = <a href="#i_add">add</a> int 1, %x
224 <p>...because the definition of <tt>%x</tt> does not dominate all of
225 its uses. The LLVM infrastructure provides a verification pass that may
226 be used to verify that an LLVM module is well formed. This pass is
227 automatically run by the parser after parsing input assembly and by
228 the optimizer before it outputs bytecode. The violations pointed out
229 by the verifier pass indicate bugs in transformation passes or input to
232 <!-- Describe the typesetting conventions here. --> </div>
234 <!-- *********************************************************************** -->
235 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
236 <!-- *********************************************************************** -->
238 <div class="doc_text">
240 <p>LLVM uses three different forms of identifiers, for different
244 <li>Named values are represented as a string of characters with a '%' prefix.
245 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
246 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
247 Identifiers which require other characters in their names can be surrounded
248 with quotes. In this way, anything except a <tt>"</tt> character can be used
251 <li>Unnamed values are represented as an unsigned numeric value with a '%'
252 prefix. For example, %12, %2, %44.</li>
254 <li>Constants, which are described in a <a href="#constants">section about
255 constants</a>, below.</li>
258 <p>LLVM requires that values start with a '%' sign for two reasons: Compilers
259 don't need to worry about name clashes with reserved words, and the set of
260 reserved words may be expanded in the future without penalty. Additionally,
261 unnamed identifiers allow a compiler to quickly come up with a temporary
262 variable without having to avoid symbol table conflicts.</p>
264 <p>Reserved words in LLVM are very similar to reserved words in other
265 languages. There are keywords for different opcodes ('<tt><a
266 href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
267 href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
268 href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
269 and others. These reserved words cannot conflict with variable names, because
270 none of them start with a '%' character.</p>
272 <p>Here is an example of LLVM code to multiply the integer variable
273 '<tt>%X</tt>' by 8:</p>
278 %result = <a href="#i_mul">mul</a> uint %X, 8
281 <p>After strength reduction:</p>
284 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
287 <p>And the hard way:</p>
290 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
291 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
292 %result = <a href="#i_add">add</a> uint %1, %1
295 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
296 important lexical features of LLVM:</p>
300 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
303 <li>Unnamed temporaries are created when the result of a computation is not
304 assigned to a named value.</li>
306 <li>Unnamed temporaries are numbered sequentially</li>
310 <p>...and it also shows a convention that we follow in this document. When
311 demonstrating instructions, we will follow an instruction with a comment that
312 defines the type and name of value produced. Comments are shown in italic
317 <!-- *********************************************************************** -->
318 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
319 <!-- *********************************************************************** -->
321 <!-- ======================================================================= -->
322 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
325 <div class="doc_text">
327 <p>LLVM programs are composed of "Module"s, each of which is a
328 translation unit of the input programs. Each module consists of
329 functions, global variables, and symbol table entries. Modules may be
330 combined together with the LLVM linker, which merges function (and
331 global variable) definitions, resolves forward declarations, and merges
332 symbol table entries. Here is an example of the "hello world" module:</p>
334 <pre><i>; Declare the string constant as a global constant...</i>
335 <a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
336 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
338 <i>; External declaration of the puts function</i>
339 <a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
341 <i>; Definition of main function</i>
342 int %main() { <i>; int()* </i>
343 <i>; Convert [13x sbyte]* to sbyte *...</i>
345 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
347 <i>; Call puts function to write out the string to stdout...</i>
349 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
351 href="#i_ret">ret</a> int 0<br>}<br></pre>
353 <p>This example is made up of a <a href="#globalvars">global variable</a>
354 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
355 function, and a <a href="#functionstructure">function definition</a>
356 for "<tt>main</tt>".</p>
358 <p>In general, a module is made up of a list of global values,
359 where both functions and global variables are global values. Global values are
360 represented by a pointer to a memory location (in this case, a pointer to an
361 array of char, and a pointer to a function), and have one of the following <a
362 href="#linkage">linkage types</a>.</p>
366 <!-- ======================================================================= -->
367 <div class="doc_subsection">
368 <a name="linkage">Linkage Types</a>
371 <div class="doc_text">
374 All Global Variables and Functions have one of the following types of linkage:
379 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
381 <dd>Global values with internal linkage are only directly accessible by
382 objects in the current module. In particular, linking code into a module with
383 an internal global value may cause the internal to be renamed as necessary to
384 avoid collisions. Because the symbol is internal to the module, all
385 references can be updated. This corresponds to the notion of the
386 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
389 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
391 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
392 the twist that linking together two modules defining the same
393 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
394 is typically used to implement inline functions. Unreferenced
395 <tt>linkonce</tt> globals are allowed to be discarded.
398 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
400 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
401 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
402 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
405 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
407 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
408 pointer to array type. When two global variables with appending linkage are
409 linked together, the two global arrays are appended together. This is the
410 LLVM, typesafe, equivalent of having the system linker append together
411 "sections" with identical names when .o files are linked.
414 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
416 <dd>If none of the above identifiers are used, the global is externally
417 visible, meaning that it participates in linkage and can be used to resolve
418 external symbol references.
422 <p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
423 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
424 variable and was linked with this one, one of the two would be renamed,
425 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
426 external (i.e., lacking any linkage declarations), they are accessible
427 outside of the current module. It is illegal for a function <i>declaration</i>
428 to have any linkage type other than "externally visible".</a></p>
432 <!-- ======================================================================= -->
433 <div class="doc_subsection">
434 <a name="callingconv">Calling Conventions</a>
437 <div class="doc_text">
439 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
440 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
441 specified for the call. The calling convention of any pair of dynamic
442 caller/callee must match, or the behavior of the program is undefined. The
443 following calling conventions are supported by LLVM, and more may be added in
447 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
449 <dd>This calling convention (the default if no other calling convention is
450 specified) matches the target C calling conventions. This calling convention
451 supports varargs function calls and tolerates some mismatch in the declared
452 prototype and implemented declaration of the function (as does normal C).
455 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
457 <dd>This calling convention attempts to make calls as fast as possible
458 (e.g. by passing things in registers). This calling convention allows the
459 target to use whatever tricks it wants to produce fast code for the target,
460 without having to conform to an externally specified ABI. Implementations of
461 this convention should allow arbitrary tail call optimization to be supported.
462 This calling convention does not support varargs and requires the prototype of
463 all callees to exactly match the prototype of the function definition.
466 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
468 <dd>This calling convention attempts to make code in the caller as efficient
469 as possible under the assumption that the call is not commonly executed. As
470 such, these calls often preserve all registers so that the call does not break
471 any live ranges in the caller side. This calling convention does not support
472 varargs and requires the prototype of all callees to exactly match the
473 prototype of the function definition.
476 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
478 <dd>Any calling convention may be specified by number, allowing
479 target-specific calling conventions to be used. Target specific calling
480 conventions start at 64.
484 <p>More calling conventions can be added/defined on an as-needed basis, to
485 support pascal conventions or any other well-known target-independent
490 <!-- ======================================================================= -->
491 <div class="doc_subsection">
492 <a name="globalvars">Global Variables</a>
495 <div class="doc_text">
497 <p>Global variables define regions of memory allocated at compilation time
498 instead of run-time. Global variables may optionally be initialized, and may
499 have an optional explicit alignment specified. A
500 variable may be defined as a global "constant," which indicates that the
501 contents of the variable will <b>never</b> be modified (enabling better
502 optimization, allowing the global data to be placed in the read-only section of
503 an executable, etc). Note that variables that need runtime initialization
504 cannot be marked "constant" as there is a store to the variable.</p>
507 LLVM explicitly allows <em>declarations</em> of global variables to be marked
508 constant, even if the final definition of the global is not. This capability
509 can be used to enable slightly better optimization of the program, but requires
510 the language definition to guarantee that optimizations based on the
511 'constantness' are valid for the translation units that do not include the
515 <p>As SSA values, global variables define pointer values that are in
516 scope (i.e. they dominate) all basic blocks in the program. Global
517 variables always define a pointer to their "content" type because they
518 describe a region of memory, and all memory objects in LLVM are
519 accessed through pointers.</p>
521 <p>An explicit alignment may be specified for a global. If not present, or if
522 the alignment is set to zero, the alignment of the global is set by the target
523 to whatever it feels convenient. If an explicit alignment is specified, the
524 global is forced to have at least that much alignment. All alignments must be
530 <!-- ======================================================================= -->
531 <div class="doc_subsection">
532 <a name="functionstructure">Functions</a>
535 <div class="doc_text">
537 <p>LLVM function definitions consist of an optional <a href="#linkage">linkage
538 type</a>, an optional <a href="#callingconv">calling convention</a>, a return
539 type, a function name, a (possibly empty) argument list, an optional alignment,
540 an opening curly brace,
541 a list of basic blocks, and a closing curly brace. LLVM function declarations
542 are defined with the "<tt>declare</tt>" keyword, an optional <a
543 href="#callingconv">calling convention</a>, a return type, a function name,
544 a possibly empty list of arguments, and an optional alignment.</p>
546 <p>A function definition contains a list of basic blocks, forming the CFG for
547 the function. Each basic block may optionally start with a label (giving the
548 basic block a symbol table entry), contains a list of instructions, and ends
549 with a <a href="#terminators">terminator</a> instruction (such as a branch or
550 function return).</p>
552 <p>The first basic block in a program is special in two ways: it is immediately
553 executed on entrance to the function, and it is not allowed to have predecessor
554 basic blocks (i.e. there can not be any branches to the entry block of a
555 function). Because the block can have no predecessors, it also cannot have any
556 <a href="#i_phi">PHI nodes</a>.</p>
558 <p>LLVM functions are identified by their name and type signature. Hence, two
559 functions with the same name but different parameter lists or return values are
560 considered different functions, and LLVM will resolve references to each
563 <p>An explicit alignment may be specified for a function. If not present, or if
564 the alignment is set to zero, the alignment of the function is set by the target
565 to whatever it feels convenient. If an explicit alignment is specified, the
566 function is forced to have at least that much alignment. All alignments must be
573 <!-- *********************************************************************** -->
574 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
575 <!-- *********************************************************************** -->
577 <div class="doc_text">
579 <p>The LLVM type system is one of the most important features of the
580 intermediate representation. Being typed enables a number of
581 optimizations to be performed on the IR directly, without having to do
582 extra analyses on the side before the transformation. A strong type
583 system makes it easier to read the generated code and enables novel
584 analyses and transformations that are not feasible to perform on normal
585 three address code representations.</p>
589 <!-- ======================================================================= -->
590 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
591 <div class="doc_text">
592 <p>The primitive types are the fundamental building blocks of the LLVM
593 system. The current set of primitive types is as follows:</p>
595 <table class="layout">
600 <tr><th>Type</th><th>Description</th></tr>
601 <tr><td><tt>void</tt></td><td>No value</td></tr>
602 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
603 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
604 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
605 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
606 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
607 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
614 <tr><th>Type</th><th>Description</th></tr>
615 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
616 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
617 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
618 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
619 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
620 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
628 <!-- _______________________________________________________________________ -->
629 <div class="doc_subsubsection"> <a name="t_classifications">Type
630 Classifications</a> </div>
631 <div class="doc_text">
632 <p>These different primitive types fall into a few useful
635 <table border="1" cellspacing="0" cellpadding="4">
637 <tr><th>Classification</th><th>Types</th></tr>
639 <td><a name="t_signed">signed</a></td>
640 <td><tt>sbyte, short, int, long, float, double</tt></td>
643 <td><a name="t_unsigned">unsigned</a></td>
644 <td><tt>ubyte, ushort, uint, ulong</tt></td>
647 <td><a name="t_integer">integer</a></td>
648 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
651 <td><a name="t_integral">integral</a></td>
652 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
656 <td><a name="t_floating">floating point</a></td>
657 <td><tt>float, double</tt></td>
660 <td><a name="t_firstclass">first class</a></td>
661 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
662 float, double, <a href="#t_pointer">pointer</a>,
663 <a href="#t_packed">packed</a></tt></td>
668 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
669 most important. Values of these types are the only ones which can be
670 produced by instructions, passed as arguments, or used as operands to
671 instructions. This means that all structures and arrays must be
672 manipulated either by pointer or by component.</p>
675 <!-- ======================================================================= -->
676 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
678 <div class="doc_text">
680 <p>The real power in LLVM comes from the derived types in the system.
681 This is what allows a programmer to represent arrays, functions,
682 pointers, and other useful types. Note that these derived types may be
683 recursive: For example, it is possible to have a two dimensional array.</p>
687 <!-- _______________________________________________________________________ -->
688 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
690 <div class="doc_text">
694 <p>The array type is a very simple derived type that arranges elements
695 sequentially in memory. The array type requires a size (number of
696 elements) and an underlying data type.</p>
701 [<# elements> x <elementtype>]
704 <p>The number of elements is a constant integer value; elementtype may
705 be any type with a size.</p>
708 <table class="layout">
711 <tt>[40 x int ]</tt><br/>
712 <tt>[41 x int ]</tt><br/>
713 <tt>[40 x uint]</tt><br/>
716 Array of 40 integer values.<br/>
717 Array of 41 integer values.<br/>
718 Array of 40 unsigned integer values.<br/>
722 <p>Here are some examples of multidimensional arrays:</p>
723 <table class="layout">
726 <tt>[3 x [4 x int]]</tt><br/>
727 <tt>[12 x [10 x float]]</tt><br/>
728 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
731 3x4 array of integer values.<br/>
732 12x10 array of single precision floating point values.<br/>
733 2x3x4 array of unsigned integer values.<br/>
738 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
739 length array. Normally, accesses past the end of an array are undefined in
740 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
741 As a special case, however, zero length arrays are recognized to be variable
742 length. This allows implementation of 'pascal style arrays' with the LLVM
743 type "{ int, [0 x float]}", for example.</p>
747 <!-- _______________________________________________________________________ -->
748 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
749 <div class="doc_text">
751 <p>The function type can be thought of as a function signature. It
752 consists of a return type and a list of formal parameter types.
753 Function types are usually used to build virtual function tables
754 (which are structures of pointers to functions), for indirect function
755 calls, and when defining a function.</p>
757 The return type of a function type cannot be an aggregate type.
760 <pre> <returntype> (<parameter list>)<br></pre>
761 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
762 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
763 which indicates that the function takes a variable number of arguments.
764 Variable argument functions can access their arguments with the <a
765 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
767 <table class="layout">
770 <tt>int (int)</tt> <br/>
771 <tt>float (int, int *) *</tt><br/>
772 <tt>int (sbyte *, ...)</tt><br/>
775 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
776 <a href="#t_pointer">Pointer</a> to a function that takes an
777 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
778 returning <tt>float</tt>.<br/>
779 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
780 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
781 the signature for <tt>printf</tt> in LLVM.<br/>
787 <!-- _______________________________________________________________________ -->
788 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
789 <div class="doc_text">
791 <p>The structure type is used to represent a collection of data members
792 together in memory. The packing of the field types is defined to match
793 the ABI of the underlying processor. The elements of a structure may
794 be any type that has a size.</p>
795 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
796 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
797 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
800 <pre> { <type list> }<br></pre>
802 <table class="layout">
805 <tt>{ int, int, int }</tt><br/>
806 <tt>{ float, int (int) * }</tt><br/>
809 a triple of three <tt>int</tt> values<br/>
810 A pair, where the first element is a <tt>float</tt> and the second element
811 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
812 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
818 <!-- _______________________________________________________________________ -->
819 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
820 <div class="doc_text">
822 <p>As in many languages, the pointer type represents a pointer or
823 reference to another object, which must live in memory.</p>
825 <pre> <type> *<br></pre>
827 <table class="layout">
830 <tt>[4x int]*</tt><br/>
831 <tt>int (int *) *</tt><br/>
834 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
835 four <tt>int</tt> values<br/>
836 A <a href="#t_pointer">pointer</a> to a <a
837 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
844 <!-- _______________________________________________________________________ -->
845 <div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
846 <div class="doc_text">
850 <p>A packed type is a simple derived type that represents a vector
851 of elements. Packed types are used when multiple primitive data
852 are operated in parallel using a single instruction (SIMD).
853 A packed type requires a size (number of
854 elements) and an underlying primitive data type. Vectors must have a power
855 of two length (1, 2, 4, 8, 16 ...). Packed types are
856 considered <a href="#t_firstclass">first class</a>.</p>
861 < <# elements> x <elementtype> >
864 <p>The number of elements is a constant integer value; elementtype may
865 be any integral or floating point type.</p>
869 <table class="layout">
872 <tt><4 x int></tt><br/>
873 <tt><8 x float></tt><br/>
874 <tt><2 x uint></tt><br/>
877 Packed vector of 4 integer values.<br/>
878 Packed vector of 8 floating-point values.<br/>
879 Packed vector of 2 unsigned integer values.<br/>
885 <!-- _______________________________________________________________________ -->
886 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
887 <div class="doc_text">
891 <p>Opaque types are used to represent unknown types in the system. This
892 corresponds (for example) to the C notion of a foward declared structure type.
893 In LLVM, opaque types can eventually be resolved to any type (not just a
904 <table class="layout">
917 <!-- *********************************************************************** -->
918 <div class="doc_section"> <a name="constants">Constants</a> </div>
919 <!-- *********************************************************************** -->
921 <div class="doc_text">
923 <p>LLVM has several different basic types of constants. This section describes
924 them all and their syntax.</p>
928 <!-- ======================================================================= -->
929 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
931 <div class="doc_text">
934 <dt><b>Boolean constants</b></dt>
936 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
937 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
940 <dt><b>Integer constants</b></dt>
942 <dd>Standard integers (such as '4') are constants of the <a
943 href="#t_integer">integer</a> type. Negative numbers may be used with signed
947 <dt><b>Floating point constants</b></dt>
949 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
950 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
951 notation (see below). Floating point constants must have a <a
952 href="#t_floating">floating point</a> type. </dd>
954 <dt><b>Null pointer constants</b></dt>
956 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
957 and must be of <a href="#t_pointer">pointer type</a>.</dd>
961 <p>The one non-intuitive notation for constants is the optional hexadecimal form
962 of floating point constants. For example, the form '<tt>double
963 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
964 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
965 (and the only time that they are generated by the disassembler) is when a
966 floating point constant must be emitted but it cannot be represented as a
967 decimal floating point number. For example, NaN's, infinities, and other
968 special values are represented in their IEEE hexadecimal format so that
969 assembly and disassembly do not cause any bits to change in the constants.</p>
973 <!-- ======================================================================= -->
974 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
977 <div class="doc_text">
978 <p>Aggregate constants arise from aggregation of simple constants
979 and smaller aggregate constants.</p>
982 <dt><b>Structure constants</b></dt>
984 <dd>Structure constants are represented with notation similar to structure
985 type definitions (a comma separated list of elements, surrounded by braces
986 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
987 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
988 must have <a href="#t_struct">structure type</a>, and the number and
989 types of elements must match those specified by the type.
992 <dt><b>Array constants</b></dt>
994 <dd>Array constants are represented with notation similar to array type
995 definitions (a comma separated list of elements, surrounded by square brackets
996 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
997 constants must have <a href="#t_array">array type</a>, and the number and
998 types of elements must match those specified by the type.
1001 <dt><b>Packed constants</b></dt>
1003 <dd>Packed constants are represented with notation similar to packed type
1004 definitions (a comma separated list of elements, surrounded by
1005 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< int 42,
1006 int 11, int 74, int 100 ></tt>". Packed constants must have <a
1007 href="#t_packed">packed type</a>, and the number and types of elements must
1008 match those specified by the type.
1011 <dt><b>Zero initialization</b></dt>
1013 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1014 value to zero of <em>any</em> type, including scalar and aggregate types.
1015 This is often used to avoid having to print large zero initializers (e.g. for
1016 large arrays) and is always exactly equivalent to using explicit zero
1023 <!-- ======================================================================= -->
1024 <div class="doc_subsection">
1025 <a name="globalconstants">Global Variable and Function Addresses</a>
1028 <div class="doc_text">
1030 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1031 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1032 constants. These constants are explicitly referenced when the <a
1033 href="#identifiers">identifier for the global</a> is used and always have <a
1034 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1040 %Z = global [2 x int*] [ int* %X, int* %Y ]
1045 <!-- ======================================================================= -->
1046 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1047 <div class="doc_text">
1048 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1049 no specific value. Undefined values may be of any type and be used anywhere
1050 a constant is permitted.</p>
1052 <p>Undefined values indicate to the compiler that the program is well defined
1053 no matter what value is used, giving the compiler more freedom to optimize.
1057 <!-- ======================================================================= -->
1058 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1061 <div class="doc_text">
1063 <p>Constant expressions are used to allow expressions involving other constants
1064 to be used as constants. Constant expressions may be of any <a
1065 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1066 that does not have side effects (e.g. load and call are not supported). The
1067 following is the syntax for constant expressions:</p>
1070 <dt><b><tt>cast ( CST to TYPE )</tt></b></dt>
1072 <dd>Cast a constant to another type.</dd>
1074 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1076 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1077 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1078 instruction, the index list may have zero or more indexes, which are required
1079 to make sense for the type of "CSTPTR".</dd>
1081 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1083 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1084 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1085 binary</a> operations. The constraints on operands are the same as those for
1086 the corresponding instruction (e.g. no bitwise operations on floating point
1087 values are allowed).</dd>
1091 <!-- *********************************************************************** -->
1092 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1093 <!-- *********************************************************************** -->
1095 <div class="doc_text">
1097 <p>The LLVM instruction set consists of several different
1098 classifications of instructions: <a href="#terminators">terminator
1099 instructions</a>, <a href="#binaryops">binary instructions</a>,
1100 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1101 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1102 instructions</a>.</p>
1106 <!-- ======================================================================= -->
1107 <div class="doc_subsection"> <a name="terminators">Terminator
1108 Instructions</a> </div>
1110 <div class="doc_text">
1112 <p>As mentioned <a href="#functionstructure">previously</a>, every
1113 basic block in a program ends with a "Terminator" instruction, which
1114 indicates which block should be executed after the current block is
1115 finished. These terminator instructions typically yield a '<tt>void</tt>'
1116 value: they produce control flow, not values (the one exception being
1117 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1118 <p>There are six different terminator instructions: the '<a
1119 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1120 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1121 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1122 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1123 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1127 <!-- _______________________________________________________________________ -->
1128 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1129 Instruction</a> </div>
1130 <div class="doc_text">
1132 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1133 ret void <i>; Return from void function</i>
1136 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1137 value) from a function back to the caller.</p>
1138 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1139 returns a value and then causes control flow, and one that just causes
1140 control flow to occur.</p>
1142 <p>The '<tt>ret</tt>' instruction may return any '<a
1143 href="#t_firstclass">first class</a>' type. Notice that a function is
1144 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1145 instruction inside of the function that returns a value that does not
1146 match the return type of the function.</p>
1148 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1149 returns back to the calling function's context. If the caller is a "<a
1150 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1151 the instruction after the call. If the caller was an "<a
1152 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1153 at the beginning of the "normal" destination block. If the instruction
1154 returns a value, that value shall set the call or invoke instruction's
1157 <pre> ret int 5 <i>; Return an integer value of 5</i>
1158 ret void <i>; Return from a void function</i>
1161 <!-- _______________________________________________________________________ -->
1162 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1163 <div class="doc_text">
1165 <pre> br bool <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1168 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1169 transfer to a different basic block in the current function. There are
1170 two forms of this instruction, corresponding to a conditional branch
1171 and an unconditional branch.</p>
1173 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1174 single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1175 unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1176 value as a target.</p>
1178 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1179 argument is evaluated. If the value is <tt>true</tt>, control flows
1180 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1181 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1183 <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
1184 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
1186 <!-- _______________________________________________________________________ -->
1187 <div class="doc_subsubsection">
1188 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1191 <div class="doc_text">
1195 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1200 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1201 several different places. It is a generalization of the '<tt>br</tt>'
1202 instruction, allowing a branch to occur to one of many possible
1208 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1209 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1210 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1211 table is not allowed to contain duplicate constant entries.</p>
1215 <p>The <tt>switch</tt> instruction specifies a table of values and
1216 destinations. When the '<tt>switch</tt>' instruction is executed, this
1217 table is searched for the given value. If the value is found, control flow is
1218 transfered to the corresponding destination; otherwise, control flow is
1219 transfered to the default destination.</p>
1221 <h5>Implementation:</h5>
1223 <p>Depending on properties of the target machine and the particular
1224 <tt>switch</tt> instruction, this instruction may be code generated in different
1225 ways. For example, it could be generated as a series of chained conditional
1226 branches or with a lookup table.</p>
1231 <i>; Emulate a conditional br instruction</i>
1232 %Val = <a href="#i_cast">cast</a> bool %value to int
1233 switch int %Val, label %truedest [int 0, label %falsedest ]
1235 <i>; Emulate an unconditional br instruction</i>
1236 switch uint 0, label %dest [ ]
1238 <i>; Implement a jump table:</i>
1239 switch uint %val, label %otherwise [ uint 0, label %onzero
1240 uint 1, label %onone
1241 uint 2, label %ontwo ]
1245 <!-- _______________________________________________________________________ -->
1246 <div class="doc_subsubsection">
1247 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1250 <div class="doc_text">
1255 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1256 to label <normal label> except label <exception label>
1261 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1262 function, with the possibility of control flow transfer to either the
1263 '<tt>normal</tt>' label or the
1264 '<tt>exception</tt>' label. If the callee function returns with the
1265 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1266 "normal" label. If the callee (or any indirect callees) returns with the "<a
1267 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1268 continued at the dynamically nearest "exception" label.</p>
1272 <p>This instruction requires several arguments:</p>
1276 The optional "cconv" marker indicates which <a href="callingconv">calling
1277 convention</a> the call should use. If none is specified, the call defaults
1278 to using C calling conventions.
1280 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1281 function value being invoked. In most cases, this is a direct function
1282 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1283 an arbitrary pointer to function value.
1286 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1287 function to be invoked. </li>
1289 <li>'<tt>function args</tt>': argument list whose types match the function
1290 signature argument types. If the function signature indicates the function
1291 accepts a variable number of arguments, the extra arguments can be
1294 <li>'<tt>normal label</tt>': the label reached when the called function
1295 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1297 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1298 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1304 <p>This instruction is designed to operate as a standard '<tt><a
1305 href="#i_call">call</a></tt>' instruction in most regards. The primary
1306 difference is that it establishes an association with a label, which is used by
1307 the runtime library to unwind the stack.</p>
1309 <p>This instruction is used in languages with destructors to ensure that proper
1310 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1311 exception. Additionally, this is important for implementation of
1312 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1316 %retval = invoke int %Test(int 15) to label %Continue
1317 except label %TestCleanup <i>; {int}:retval set</i>
1318 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
1319 except label %TestCleanup <i>; {int}:retval set</i>
1324 <!-- _______________________________________________________________________ -->
1326 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1327 Instruction</a> </div>
1329 <div class="doc_text">
1338 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1339 at the first callee in the dynamic call stack which used an <a
1340 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1341 primarily used to implement exception handling.</p>
1345 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1346 immediately halt. The dynamic call stack is then searched for the first <a
1347 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1348 execution continues at the "exceptional" destination block specified by the
1349 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1350 dynamic call chain, undefined behavior results.</p>
1353 <!-- _______________________________________________________________________ -->
1355 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1356 Instruction</a> </div>
1358 <div class="doc_text">
1367 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1368 instruction is used to inform the optimizer that a particular portion of the
1369 code is not reachable. This can be used to indicate that the code after a
1370 no-return function cannot be reached, and other facts.</p>
1374 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1379 <!-- ======================================================================= -->
1380 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1381 <div class="doc_text">
1382 <p>Binary operators are used to do most of the computation in a
1383 program. They require two operands, execute an operation on them, and
1384 produce a single value. The operands might represent
1385 multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1386 The result value of a binary operator is not
1387 necessarily the same type as its operands.</p>
1388 <p>There are several different binary operators:</p>
1390 <!-- _______________________________________________________________________ -->
1391 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1392 Instruction</a> </div>
1393 <div class="doc_text">
1395 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1398 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1400 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1401 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1402 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1403 Both arguments must have identical types.</p>
1405 <p>The value produced is the integer or floating point sum of the two
1408 <pre> <result> = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
1411 <!-- _______________________________________________________________________ -->
1412 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1413 Instruction</a> </div>
1414 <div class="doc_text">
1416 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1419 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1421 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1422 instruction present in most other intermediate representations.</p>
1424 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1425 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1427 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1428 Both arguments must have identical types.</p>
1430 <p>The value produced is the integer or floating point difference of
1431 the two operands.</p>
1433 <pre> <result> = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
1434 <result> = sub int 0, %val <i>; yields {int}:result = -%var</i>
1437 <!-- _______________________________________________________________________ -->
1438 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1439 Instruction</a> </div>
1440 <div class="doc_text">
1442 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1445 <p>The '<tt>mul</tt>' instruction returns the product of its two
1448 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1449 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1451 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1452 Both arguments must have identical types.</p>
1454 <p>The value produced is the integer or floating point product of the
1456 <p>There is no signed vs unsigned multiplication. The appropriate
1457 action is taken based on the type of the operand.</p>
1459 <pre> <result> = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
1462 <!-- _______________________________________________________________________ -->
1463 <div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1464 Instruction</a> </div>
1465 <div class="doc_text">
1467 <pre> <result> = div <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1470 <p>The '<tt>div</tt>' instruction returns the quotient of its two
1473 <p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1474 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1476 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1477 Both arguments must have identical types.</p>
1479 <p>The value produced is the integer or floating point quotient of the
1482 <pre> <result> = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1485 <!-- _______________________________________________________________________ -->
1486 <div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1487 Instruction</a> </div>
1488 <div class="doc_text">
1490 <pre> <result> = rem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1493 <p>The '<tt>rem</tt>' instruction returns the remainder from the
1494 division of its two operands.</p>
1496 <p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1497 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1499 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1500 Both arguments must have identical types.</p>
1502 <p>This returns the <i>remainder</i> of a division (where the result
1503 has the same sign as the divisor), not the <i>modulus</i> (where the
1504 result has the same sign as the dividend) of a value. For more
1505 information about the difference, see <a
1506 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1509 <pre> <result> = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1512 <!-- _______________________________________________________________________ -->
1513 <div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1514 Instructions</a> </div>
1515 <div class="doc_text">
1517 <pre> <result> = seteq <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1518 <result> = setne <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1519 <result> = setlt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1520 <result> = setgt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1521 <result> = setle <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1522 <result> = setge <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1525 <p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1526 value based on a comparison of their two operands.</p>
1528 <p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1529 be of <a href="#t_firstclass">first class</a> type (it is not possible
1530 to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1531 or '<tt>void</tt>' values, etc...). Both arguments must have identical
1534 <p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1535 value if both operands are equal.<br>
1536 The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1537 value if both operands are unequal.<br>
1538 The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1539 value if the first operand is less than the second operand.<br>
1540 The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1541 value if the first operand is greater than the second operand.<br>
1542 The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1543 value if the first operand is less than or equal to the second operand.<br>
1544 The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1545 value if the first operand is greater than or equal to the second
1548 <pre> <result> = seteq int 4, 5 <i>; yields {bool}:result = false</i>
1549 <result> = setne float 4, 5 <i>; yields {bool}:result = true</i>
1550 <result> = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1551 <result> = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1552 <result> = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1553 <result> = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1556 <!-- ======================================================================= -->
1557 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1558 Operations</a> </div>
1559 <div class="doc_text">
1560 <p>Bitwise binary operators are used to do various forms of
1561 bit-twiddling in a program. They are generally very efficient
1562 instructions and can commonly be strength reduced from other
1563 instructions. They require two operands, execute an operation on them,
1564 and produce a single value. The resulting value of the bitwise binary
1565 operators is always the same type as its first operand.</p>
1567 <!-- _______________________________________________________________________ -->
1568 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1569 Instruction</a> </div>
1570 <div class="doc_text">
1572 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1575 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1576 its two operands.</p>
1578 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
1579 href="#t_integral">integral</a> values. Both arguments must have
1580 identical types.</p>
1582 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
1584 <div style="align: center">
1585 <table border="1" cellspacing="0" cellpadding="4">
1616 <pre> <result> = and int 4, %var <i>; yields {int}:result = 4 & %var</i>
1617 <result> = and int 15, 40 <i>; yields {int}:result = 8</i>
1618 <result> = and int 4, 8 <i>; yields {int}:result = 0</i>
1621 <!-- _______________________________________________________________________ -->
1622 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
1623 <div class="doc_text">
1625 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1628 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1629 or of its two operands.</p>
1631 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
1632 href="#t_integral">integral</a> values. Both arguments must have
1633 identical types.</p>
1635 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
1637 <div style="align: center">
1638 <table border="1" cellspacing="0" cellpadding="4">
1669 <pre> <result> = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
1670 <result> = or int 15, 40 <i>; yields {int}:result = 47</i>
1671 <result> = or int 4, 8 <i>; yields {int}:result = 12</i>
1674 <!-- _______________________________________________________________________ -->
1675 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1676 Instruction</a> </div>
1677 <div class="doc_text">
1679 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1682 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1683 or of its two operands. The <tt>xor</tt> is used to implement the
1684 "one's complement" operation, which is the "~" operator in C.</p>
1686 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
1687 href="#t_integral">integral</a> values. Both arguments must have
1688 identical types.</p>
1690 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
1692 <div style="align: center">
1693 <table border="1" cellspacing="0" cellpadding="4">
1725 <pre> <result> = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
1726 <result> = xor int 15, 40 <i>; yields {int}:result = 39</i>
1727 <result> = xor int 4, 8 <i>; yields {int}:result = 12</i>
1728 <result> = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
1731 <!-- _______________________________________________________________________ -->
1732 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1733 Instruction</a> </div>
1734 <div class="doc_text">
1736 <pre> <result> = shl <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1739 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1740 the left a specified number of bits.</p>
1742 <p>The first argument to the '<tt>shl</tt>' instruction must be an <a
1743 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1746 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
1748 <pre> <result> = shl int 4, ubyte %var <i>; yields {int}:result = 4 << %var</i>
1749 <result> = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1750 <result> = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1753 <!-- _______________________________________________________________________ -->
1754 <div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1755 Instruction</a> </div>
1756 <div class="doc_text">
1758 <pre> <result> = shr <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1761 <p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1762 the right a specified number of bits.</p>
1764 <p>The first argument to the '<tt>shr</tt>' instruction must be an <a
1765 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1768 <p>If the first argument is a <a href="#t_signed">signed</a> type, the
1769 most significant bit is duplicated in the newly free'd bit positions.
1770 If the first argument is unsigned, zero bits shall fill the empty
1773 <pre> <result> = shr int 4, ubyte %var <i>; yields {int}:result = 4 >> %var</i>
1774 <result> = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
1775 <result> = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
1776 <result> = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1777 <result> = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
1781 <!-- ======================================================================= -->
1782 <div class="doc_subsection">
1783 <a name="memoryops">Memory Access Operations</a>
1786 <div class="doc_text">
1788 <p>A key design point of an SSA-based representation is how it
1789 represents memory. In LLVM, no memory locations are in SSA form, which
1790 makes things very simple. This section describes how to read, write,
1791 allocate, and free memory in LLVM.</p>
1795 <!-- _______________________________________________________________________ -->
1796 <div class="doc_subsubsection">
1797 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
1800 <div class="doc_text">
1805 <result> = malloc <type>[, uint <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
1810 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
1811 heap and returns a pointer to it.</p>
1815 <p>The '<tt>malloc</tt>' instruction allocates
1816 <tt>sizeof(<type>)*NumElements</tt>
1817 bytes of memory from the operating system and returns a pointer of the
1818 appropriate type to the program. If "NumElements" is specified, it is the
1819 number of elements allocated. If an alignment is specified, the value result
1820 of the allocation is guaranteed to be aligned to at least that boundary. If
1821 not specified, or if zero, the target can choose to align the allocation on any
1822 convenient boundary.</p>
1824 <p>'<tt>type</tt>' must be a sized type.</p>
1828 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
1829 a pointer is returned.</p>
1834 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
1836 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
1837 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1838 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
1839 %array3 = malloc int, uint 4, align 1024 <i>; yields {int*}:array3</i>
1840 %array4 = malloc int, align 1024 <i>; yields {int*}:array4</i>
1844 <!-- _______________________________________________________________________ -->
1845 <div class="doc_subsubsection">
1846 <a name="i_free">'<tt>free</tt>' Instruction</a>
1849 <div class="doc_text">
1854 free <type> <value> <i>; yields {void}</i>
1859 <p>The '<tt>free</tt>' instruction returns memory back to the unused
1860 memory heap to be reallocated in the future.</p>
1864 <p>'<tt>value</tt>' shall be a pointer value that points to a value
1865 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
1870 <p>Access to the memory pointed to by the pointer is no longer defined
1871 after this instruction executes.</p>
1876 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
1877 free [4 x ubyte]* %array
1881 <!-- _______________________________________________________________________ -->
1882 <div class="doc_subsubsection">
1883 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
1886 <div class="doc_text">
1891 <result> = alloca <type>[, uint <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
1896 <p>The '<tt>alloca</tt>' instruction allocates memory on the current
1897 stack frame of the procedure that is live until the current function
1898 returns to its caller.</p>
1902 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
1903 bytes of memory on the runtime stack, returning a pointer of the
1904 appropriate type to the program. If "NumElements" is specified, it is the
1905 number of elements allocated. If an alignment is specified, the value result
1906 of the allocation is guaranteed to be aligned to at least that boundary. If
1907 not specified, or if zero, the target can choose to align the allocation on any
1908 convenient boundary.</p>
1910 <p>'<tt>type</tt>' may be any sized type.</p>
1914 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
1915 memory is automatically released when the function returns. The '<tt>alloca</tt>'
1916 instruction is commonly used to represent automatic variables that must
1917 have an address available. When the function returns (either with the <tt><a
1918 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
1919 instructions), the memory is reclaimed.</p>
1924 %ptr = alloca int <i>; yields {int*}:ptr</i>
1925 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
1926 %ptr = alloca int, uint 4, align 1024 <i>; yields {int*}:ptr</i>
1927 %ptr = alloca int, align 1024 <i>; yields {int*}:ptr</i>
1931 <!-- _______________________________________________________________________ -->
1932 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
1933 Instruction</a> </div>
1934 <div class="doc_text">
1936 <pre> <result> = load <ty>* <pointer><br> <result> = volatile load <ty>* <pointer><br></pre>
1938 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
1940 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
1941 address from which to load. The pointer must point to a <a
1942 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
1943 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
1944 the number or order of execution of this <tt>load</tt> with other
1945 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
1948 <p>The location of memory pointed to is loaded.</p>
1950 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1952 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1953 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1956 <!-- _______________________________________________________________________ -->
1957 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
1958 Instruction</a> </div>
1960 <pre> store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
1961 volatile store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
1964 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
1966 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
1967 to store and an address in which to store it. The type of the '<tt><pointer></tt>'
1968 operand must be a pointer to the type of the '<tt><value></tt>'
1969 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
1970 optimizer is not allowed to modify the number or order of execution of
1971 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
1972 href="#i_store">store</a></tt> instructions.</p>
1974 <p>The contents of memory are updated to contain '<tt><value></tt>'
1975 at the location specified by the '<tt><pointer></tt>' operand.</p>
1977 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1979 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1980 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1982 <!-- _______________________________________________________________________ -->
1983 <div class="doc_subsubsection">
1984 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
1987 <div class="doc_text">
1990 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
1996 The '<tt>getelementptr</tt>' instruction is used to get the address of a
1997 subelement of an aggregate data structure.</p>
2001 <p>This instruction takes a list of integer constants that indicate what
2002 elements of the aggregate object to index to. The actual types of the arguments
2003 provided depend on the type of the first pointer argument. The
2004 '<tt>getelementptr</tt>' instruction is used to index down through the type
2005 levels of a structure or to a specific index in an array. When indexing into a
2006 structure, only <tt>uint</tt>
2007 integer constants are allowed. When indexing into an array or pointer,
2008 <tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
2010 <p>For example, let's consider a C code fragment and how it gets
2011 compiled to LLVM:</p>
2025 int *foo(struct ST *s) {
2026 return &s[1].Z.B[5][13];
2030 <p>The LLVM code generated by the GCC frontend is:</p>
2033 %RT = type { sbyte, [10 x [20 x int]], sbyte }
2034 %ST = type { int, double, %RT }
2038 int* %foo(%ST* %s) {
2040 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
2047 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2048 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2049 and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
2050 <tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
2051 types require <tt>uint</tt> <b>constants</b>.</p>
2053 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2054 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
2055 }</tt>' type, a structure. The second index indexes into the third element of
2056 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
2057 sbyte }</tt>' type, another structure. The third index indexes into the second
2058 element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
2059 array. The two dimensions of the array are subscripted into, yielding an
2060 '<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2061 to this element, thus computing a value of '<tt>int*</tt>' type.</p>
2063 <p>Note that it is perfectly legal to index partially through a
2064 structure, returning a pointer to an inner element. Because of this,
2065 the LLVM code for the given testcase is equivalent to:</p>
2068 int* %foo(%ST* %s) {
2069 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
2070 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
2071 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
2072 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
2073 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2078 <p>Note that it is undefined to access an array out of bounds: array and
2079 pointer indexes must always be within the defined bounds of the array type.
2080 The one exception for this rules is zero length arrays. These arrays are
2081 defined to be accessible as variable length arrays, which requires access
2082 beyond the zero'th element.</p>
2087 <i>; yields [12 x ubyte]*:aptr</i>
2088 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2092 <!-- ======================================================================= -->
2093 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
2094 <div class="doc_text">
2095 <p>The instructions in this category are the "miscellaneous"
2096 instructions, which defy better classification.</p>
2098 <!-- _______________________________________________________________________ -->
2099 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
2100 Instruction</a> </div>
2101 <div class="doc_text">
2103 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
2105 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
2106 the SSA graph representing the function.</p>
2108 <p>The type of the incoming values are specified with the first type
2109 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
2110 as arguments, with one pair for each predecessor basic block of the
2111 current block. Only values of <a href="#t_firstclass">first class</a>
2112 type may be used as the value arguments to the PHI node. Only labels
2113 may be used as the label arguments.</p>
2114 <p>There must be no non-phi instructions between the start of a basic
2115 block and the PHI instructions: i.e. PHI instructions must be first in
2118 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
2119 value specified by the parameter, depending on which basic block we
2120 came from in the last <a href="#terminators">terminator</a> instruction.</p>
2122 <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>
2125 <!-- _______________________________________________________________________ -->
2126 <div class="doc_subsubsection">
2127 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
2130 <div class="doc_text">
2135 <result> = cast <ty> <value> to <ty2> <i>; yields ty2</i>
2141 The '<tt>cast</tt>' instruction is used as the primitive means to convert
2142 integers to floating point, change data type sizes, and break type safety (by
2150 The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
2151 class value, and a type to cast it to, which must also be a <a
2152 href="#t_firstclass">first class</a> type.
2158 This instruction follows the C rules for explicit casts when determining how the
2159 data being cast must change to fit in its new container.
2163 When casting to bool, any value that would be considered true in the context of
2164 a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
2165 all else are '<tt>false</tt>'.
2169 When extending an integral value from a type of one signness to another (for
2170 example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
2171 <b>source</b> value is signed, and zero-extended if the source value is
2172 unsigned. <tt>bool</tt> values are always zero extended into either zero or
2179 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
2180 %Y = cast int 123 to bool <i>; yields bool:true</i>
2184 <!-- _______________________________________________________________________ -->
2185 <div class="doc_subsubsection">
2186 <a name="i_select">'<tt>select</tt>' Instruction</a>
2189 <div class="doc_text">
2194 <result> = select bool <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
2200 The '<tt>select</tt>' instruction is used to choose one value based on a
2201 condition, without branching.
2208 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.
2214 If the boolean condition evaluates to true, the instruction returns the first
2215 value argument; otherwise, it returns the second value argument.
2221 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
2229 <!-- _______________________________________________________________________ -->
2230 <div class="doc_subsubsection">
2231 <a name="i_call">'<tt>call</tt>' Instruction</a>
2234 <div class="doc_text">
2238 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
2243 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
2247 <p>This instruction requires several arguments:</p>
2251 <p>The optional "tail" marker indicates whether the callee function accesses
2252 any allocas or varargs in the caller. If the "tail" marker is present, the
2253 function call is eligible for tail call optimization. Note that calls may
2254 be marked "tail" even if they do not occur before a <a
2255 href="#i_ret"><tt>ret</tt></a> instruction.
2258 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
2259 convention</a> the call should use. If none is specified, the call defaults
2260 to using C calling conventions.
2263 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
2264 being invoked. The argument types must match the types implied by this
2265 signature. This type can be omitted if the function is not varargs and
2266 if the function type does not return a pointer to a function.</p>
2269 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
2270 be invoked. In most cases, this is a direct function invocation, but
2271 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
2272 to function value.</p>
2275 <p>'<tt>function args</tt>': argument list whose types match the
2276 function signature argument types. All arguments must be of
2277 <a href="#t_firstclass">first class</a> type. If the function signature
2278 indicates the function accepts a variable number of arguments, the extra
2279 arguments can be specified.</p>
2285 <p>The '<tt>call</tt>' instruction is used to cause control flow to
2286 transfer to a specified function, with its incoming arguments bound to
2287 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
2288 instruction in the called function, control flow continues with the
2289 instruction after the function call, and the return value of the
2290 function is bound to the result argument. This is a simpler case of
2291 the <a href="#i_invoke">invoke</a> instruction.</p>
2296 %retval = call int %test(int %argc)
2297 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
2298 %X = tail call int %foo()
2299 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
2304 <!-- _______________________________________________________________________ -->
2305 <div class="doc_subsubsection">
2306 <a name="i_vaarg">'<tt>vaarg</tt>' Instruction</a>
2309 <div class="doc_text">
2314 <resultval> = va_arg <va_list*> <arglist>, <argty>
2319 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
2320 the "variable argument" area of a function call. It is used to implement the
2321 <tt>va_arg</tt> macro in C.</p>
2325 <p>This instruction takes a <tt>va_list*</tt> value and the type of
2326 the argument. It returns a value of the specified argument type and
2327 increments the <tt>va_list</tt> to point to the next argument. Again, the
2328 actual type of <tt>va_list</tt> is target specific.</p>
2332 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
2333 type from the specified <tt>va_list</tt> and causes the
2334 <tt>va_list</tt> to point to the next argument. For more information,
2335 see the variable argument handling <a href="#int_varargs">Intrinsic
2338 <p>It is legal for this instruction to be called in a function which does not
2339 take a variable number of arguments, for example, the <tt>vfprintf</tt>
2342 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
2343 href="#intrinsics">intrinsic function</a> because it takes a type as an
2348 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
2352 <!-- *********************************************************************** -->
2353 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
2354 <!-- *********************************************************************** -->
2356 <div class="doc_text">
2358 <p>LLVM supports the notion of an "intrinsic function". These functions have
2359 well known names and semantics and are required to follow certain
2360 restrictions. Overall, these instructions represent an extension mechanism for
2361 the LLVM language that does not require changing all of the transformations in
2362 LLVM to add to the language (or the bytecode reader/writer, the parser,
2365 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
2366 prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
2367 this. Intrinsic functions must always be external functions: you cannot define
2368 the body of intrinsic functions. Intrinsic functions may only be used in call
2369 or invoke instructions: it is illegal to take the address of an intrinsic
2370 function. Additionally, because intrinsic functions are part of the LLVM
2371 language, it is required that they all be documented here if any are added.</p>
2374 <p>To learn how to add an intrinsic function, please see the <a
2375 href="ExtendingLLVM.html">Extending LLVM Guide</a>.
2380 <!-- ======================================================================= -->
2381 <div class="doc_subsection">
2382 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
2385 <div class="doc_text">
2387 <p>Variable argument support is defined in LLVM with the <a
2388 href="#i_vanext"><tt>vanext</tt></a> instruction and these three
2389 intrinsic functions. These functions are related to the similarly
2390 named macros defined in the <tt><stdarg.h></tt> header file.</p>
2392 <p>All of these functions operate on arguments that use a
2393 target-specific value type "<tt>va_list</tt>". The LLVM assembly
2394 language reference manual does not define what this type is, so all
2395 transformations should be prepared to handle intrinsics with any type
2398 <p>This example shows how the <a href="#i_vanext"><tt>vanext</tt></a>
2399 instruction and the variable argument handling intrinsic functions are
2403 int %test(int %X, ...) {
2404 ; Initialize variable argument processing
2406 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
2408 ; Read a single integer argument
2409 %tmp = va_arg sbyte** %ap, int
2411 ; Demonstrate usage of llvm.va_copy and llvm.va_end
2413 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
2414 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
2416 ; Stop processing of arguments.
2417 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
2423 <!-- _______________________________________________________________________ -->
2424 <div class="doc_subsubsection">
2425 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
2429 <div class="doc_text">
2431 <pre> declare void %llvm.va_start(<va_list>* <arglist>)<br></pre>
2433 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
2434 <tt>*<arglist></tt> for subsequent use by <tt><a
2435 href="#i_va_arg">va_arg</a></tt>.</p>
2439 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
2443 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
2444 macro available in C. In a target-dependent way, it initializes the
2445 <tt>va_list</tt> element the argument points to, so that the next call to
2446 <tt>va_arg</tt> will produce the first variable argument passed to the function.
2447 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
2448 last argument of the function, the compiler can figure that out.</p>
2452 <!-- _______________________________________________________________________ -->
2453 <div class="doc_subsubsection">
2454 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
2457 <div class="doc_text">
2459 <pre> declare void %llvm.va_end(<va_list*> <arglist>)<br></pre>
2461 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt><arglist></tt>
2462 which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
2463 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
2465 <p>The argument is a <tt>va_list</tt> to destroy.</p>
2467 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
2468 macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
2469 Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
2470 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
2471 with calls to <tt>llvm.va_end</tt>.</p>
2474 <!-- _______________________________________________________________________ -->
2475 <div class="doc_subsubsection">
2476 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
2479 <div class="doc_text">
2484 declare void %llvm.va_copy(<va_list>* <destarglist>,
2485 <va_list>* <srcarglist>)
2490 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
2491 the source argument list to the destination argument list.</p>
2495 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
2496 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
2501 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
2502 available in C. In a target-dependent way, it copies the source
2503 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
2504 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
2505 arbitrarily complex and require memory allocation, for example.</p>
2509 <!-- ======================================================================= -->
2510 <div class="doc_subsection">
2511 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
2514 <div class="doc_text">
2517 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
2518 Collection</a> requires the implementation and generation of these intrinsics.
2519 These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
2520 stack</a>, as well as garbage collector implementations that require <a
2521 href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
2522 Front-ends for type-safe garbage collected languages should generate these
2523 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
2524 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
2528 <!-- _______________________________________________________________________ -->
2529 <div class="doc_subsubsection">
2530 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
2533 <div class="doc_text">
2538 declare void %llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
2543 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
2544 the code generator, and allows some metadata to be associated with it.</p>
2548 <p>The first argument specifies the address of a stack object that contains the
2549 root pointer. The second pointer (which must be either a constant or a global
2550 value address) contains the meta-data to be associated with the root.</p>
2554 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
2555 location. At compile-time, the code generator generates information to allow
2556 the runtime to find the pointer at GC safe points.
2562 <!-- _______________________________________________________________________ -->
2563 <div class="doc_subsubsection">
2564 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
2567 <div class="doc_text">
2572 declare sbyte* %llvm.gcread(sbyte** %Ptr)
2577 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
2578 locations, allowing garbage collector implementations that require read
2583 <p>The argument is the address to read from, which should be an address
2584 allocated from the garbage collector.</p>
2588 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
2589 instruction, but may be replaced with substantially more complex code by the
2590 garbage collector runtime, as needed.</p>
2595 <!-- _______________________________________________________________________ -->
2596 <div class="doc_subsubsection">
2597 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
2600 <div class="doc_text">
2605 declare void %llvm.gcwrite(sbyte* %P1, sbyte** %P2)
2610 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
2611 locations, allowing garbage collector implementations that require write
2612 barriers (such as generational or reference counting collectors).</p>
2616 <p>The first argument is the reference to store, and the second is the heap
2617 location to store to.</p>
2621 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
2622 instruction, but may be replaced with substantially more complex code by the
2623 garbage collector runtime, as needed.</p>
2629 <!-- ======================================================================= -->
2630 <div class="doc_subsection">
2631 <a name="int_codegen">Code Generator Intrinsics</a>
2634 <div class="doc_text">
2636 These intrinsics are provided by LLVM to expose special features that may only
2637 be implemented with code generator support.
2642 <!-- _______________________________________________________________________ -->
2643 <div class="doc_subsubsection">
2644 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
2647 <div class="doc_text">
2651 declare void* %llvm.returnaddress(uint <level>)
2657 The '<tt>llvm.returnaddress</tt>' intrinsic returns a target-specific value
2658 indicating the return address of the current function or one of its callers.
2664 The argument to this intrinsic indicates which function to return the address
2665 for. Zero indicates the calling function, one indicates its caller, etc. The
2666 argument is <b>required</b> to be a constant integer value.
2672 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
2673 the return address of the specified call frame, or zero if it cannot be
2674 identified. The value returned by this intrinsic is likely to be incorrect or 0
2675 for arguments other than zero, so it should only be used for debugging purposes.
2679 Note that calling this intrinsic does not prevent function inlining or other
2680 aggressive transformations, so the value returned may not be that of the obvious
2681 source-language caller.
2686 <!-- _______________________________________________________________________ -->
2687 <div class="doc_subsubsection">
2688 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
2691 <div class="doc_text">
2695 declare void* %llvm.frameaddress(uint <level>)
2701 The '<tt>llvm.frameaddress</tt>' intrinsic returns the target-specific frame
2702 pointer value for the specified stack frame.
2708 The argument to this intrinsic indicates which function to return the frame
2709 pointer for. Zero indicates the calling function, one indicates its caller,
2710 etc. The argument is <b>required</b> to be a constant integer value.
2716 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
2717 the frame address of the specified call frame, or zero if it cannot be
2718 identified. The value returned by this intrinsic is likely to be incorrect or 0
2719 for arguments other than zero, so it should only be used for debugging purposes.
2723 Note that calling this intrinsic does not prevent function inlining or other
2724 aggressive transformations, so the value returned may not be that of the obvious
2725 source-language caller.
2729 <!-- _______________________________________________________________________ -->
2730 <div class="doc_subsubsection">
2731 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
2734 <div class="doc_text">
2738 declare void %llvm.prefetch(sbyte * <address>,
2739 uint <rw>, uint <locality>)
2746 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
2747 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
2749 effect on the behavior of the program but can change its performance
2756 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
2757 determining if the fetch should be for a read (0) or write (1), and
2758 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
2759 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
2760 <tt>locality</tt> arguments must be constant integers.
2766 This intrinsic does not modify the behavior of the program. In particular,
2767 prefetches cannot trap and do not produce a value. On targets that support this
2768 intrinsic, the prefetch can provide hints to the processor cache for better
2774 <!-- _______________________________________________________________________ -->
2775 <div class="doc_subsubsection">
2776 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
2779 <div class="doc_text">
2783 declare void %llvm.pcmarker( uint <id> )
2790 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
2792 code to simulators and other tools. The method is target specific, but it is
2793 expected that the marker will use exported symbols to transmit the PC of the marker.
2794 The marker makes no guarantees that it will remain with any specific instruction
2795 after optimizations. It is possible that the presense of a marker will inhibit
2796 optimizations. The intended use is to be inserted after optmizations to allow
2797 correlations of simulation runs.
2803 <tt>id</tt> is a numerical id identifying the marker.
2809 This intrinsic does not modify the behavior of the program. Backends that do not
2810 support this intrinisic may ignore it.
2815 <!-- _______________________________________________________________________ -->
2816 <div class="doc_subsubsection">
2817 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
2820 <div class="doc_text">
2824 declare ulong %llvm.readcyclecounter( )
2831 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
2832 counter register (or similar low latency, high accuracy clocks) on those targets
2833 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
2834 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
2835 should only be used for small timings.
2841 When directly supported, reading the cycle counter should not modify any memory.
2842 Implementations are allowed to either return a application specific value or a
2843 system wide value. On backends without support, this is lowered to a constant 0.
2849 <!-- ======================================================================= -->
2850 <div class="doc_subsection">
2851 <a name="int_os">Operating System Intrinsics</a>
2854 <div class="doc_text">
2856 These intrinsics are provided by LLVM to support the implementation of
2857 operating system level code.
2862 <!-- _______________________________________________________________________ -->
2863 <div class="doc_subsubsection">
2864 <a name="i_readport">'<tt>llvm.readport</tt>' Intrinsic</a>
2867 <div class="doc_text">
2871 declare <integer type> %llvm.readport (<integer type> <address>)
2877 The '<tt>llvm.readport</tt>' intrinsic reads data from the specified hardware
2884 The argument to this intrinsic indicates the hardware I/O address from which
2885 to read the data. The address is in the hardware I/O address namespace (as
2886 opposed to being a memory location for memory mapped I/O).
2892 The '<tt>llvm.readport</tt>' intrinsic reads data from the hardware I/O port
2893 specified by <i>address</i> and returns the value. The address and return
2894 value must be integers, but the size is dependent upon the platform upon which
2895 the program is code generated. For example, on x86, the address must be an
2896 unsigned 16-bit value, and the return value must be 8, 16, or 32 bits.
2901 <!-- _______________________________________________________________________ -->
2902 <div class="doc_subsubsection">
2903 <a name="i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a>
2906 <div class="doc_text">
2910 call void (<integer type>, <integer type>)*
2911 %llvm.writeport (<integer type> <value>,
2912 <integer type> <address>)
2918 The '<tt>llvm.writeport</tt>' intrinsic writes data to the specified hardware
2925 The first argument is the value to write to the I/O port.
2929 The second argument indicates the hardware I/O address to which data should be
2930 written. The address is in the hardware I/O address namespace (as opposed to
2931 being a memory location for memory mapped I/O).
2937 The '<tt>llvm.writeport</tt>' intrinsic writes <i>value</i> to the I/O port
2938 specified by <i>address</i>. The address and value must be integers, but the
2939 size is dependent upon the platform upon which the program is code generated.
2940 For example, on x86, the address must be an unsigned 16-bit value, and the
2941 value written must be 8, 16, or 32 bits in length.
2946 <!-- _______________________________________________________________________ -->
2947 <div class="doc_subsubsection">
2948 <a name="i_readio">'<tt>llvm.readio</tt>' Intrinsic</a>
2951 <div class="doc_text">
2955 declare <result> %llvm.readio (<ty> * <pointer>)
2961 The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2968 The argument to this intrinsic is a pointer indicating the memory address from
2969 which to read the data. The data must be a
2970 <a href="#t_firstclass">first class</a> type.
2976 The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2977 location specified by <i>pointer</i> and returns the value. The argument must
2978 be a pointer, and the return value must be a
2979 <a href="#t_firstclass">first class</a> type. However, certain architectures
2980 may not support I/O on all first class types. For example, 32-bit processors
2981 may only support I/O on data types that are 32 bits or less.
2985 This intrinsic enforces an in-order memory model for llvm.readio and
2986 llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2987 scheduled processors may execute loads and stores out of order, re-ordering at
2988 run time accesses to memory mapped I/O registers. Using these intrinsics
2989 ensures that accesses to memory mapped I/O registers occur in program order.
2994 <!-- _______________________________________________________________________ -->
2995 <div class="doc_subsubsection">
2996 <a name="i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a>
2999 <div class="doc_text">
3003 declare void %llvm.writeio (<ty1> <value>, <ty2> * <pointer>)
3009 The '<tt>llvm.writeio</tt>' intrinsic writes data to the specified memory
3016 The first argument is the value to write to the memory mapped I/O location.
3017 The second argument is a pointer indicating the memory address to which the
3018 data should be written.
3024 The '<tt>llvm.writeio</tt>' intrinsic writes <i>value</i> to the memory mapped
3025 I/O address specified by <i>pointer</i>. The value must be a
3026 <a href="#t_firstclass">first class</a> type. However, certain architectures
3027 may not support I/O on all first class types. For example, 32-bit processors
3028 may only support I/O on data types that are 32 bits or less.
3032 This intrinsic enforces an in-order memory model for llvm.readio and
3033 llvm.writeio calls on machines that use dynamic scheduling. Dynamically
3034 scheduled processors may execute loads and stores out of order, re-ordering at
3035 run time accesses to memory mapped I/O registers. Using these intrinsics
3036 ensures that accesses to memory mapped I/O registers occur in program order.
3041 <!-- ======================================================================= -->
3042 <div class="doc_subsection">
3043 <a name="int_libc">Standard C Library Intrinsics</a>
3046 <div class="doc_text">
3048 LLVM provides intrinsics for a few important standard C library functions.
3049 These intrinsics allow source-language front-ends to pass information about the
3050 alignment of the pointer arguments to the code generator, providing opportunity
3051 for more efficient code generation.
3056 <!-- _______________________________________________________________________ -->
3057 <div class="doc_subsubsection">
3058 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
3061 <div class="doc_text">
3065 declare void %llvm.memcpy(sbyte* <dest>, sbyte* <src>,
3066 uint <len>, uint <align>)
3072 The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
3073 location to the destination location.
3077 Note that, unlike the standard libc function, the <tt>llvm.memcpy</tt> intrinsic
3078 does not return a value, and takes an extra alignment argument.
3084 The first argument is a pointer to the destination, the second is a pointer to
3085 the source. The third argument is an (arbitrarily sized) integer argument
3086 specifying the number of bytes to copy, and the fourth argument is the alignment
3087 of the source and destination locations.
3091 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3092 the caller guarantees that the size of the copy is a multiple of the alignment
3093 and that both the source and destination pointers are aligned to that boundary.
3099 The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
3100 location to the destination location, which are not allowed to overlap. It
3101 copies "len" bytes of memory over. If the argument is known to be aligned to
3102 some boundary, this can be specified as the fourth argument, otherwise it should
3108 <!-- _______________________________________________________________________ -->
3109 <div class="doc_subsubsection">
3110 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3113 <div class="doc_text">
3117 declare void %llvm.memmove(sbyte* <dest>, sbyte* <src>,
3118 uint <len>, uint <align>)
3124 The '<tt>llvm.memmove</tt>' intrinsic moves a block of memory from the source
3125 location to the destination location. It is similar to the '<tt>llvm.memcpy</tt>'
3126 intrinsic but allows the two memory locations to overlap.
3130 Note that, unlike the standard libc function, the <tt>llvm.memmove</tt> intrinsic
3131 does not return a value, and takes an extra alignment argument.
3137 The first argument is a pointer to the destination, the second is a pointer to
3138 the source. The third argument is an (arbitrarily sized) integer argument
3139 specifying the number of bytes to copy, and the fourth argument is the alignment
3140 of the source and destination locations.
3144 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3145 the caller guarantees that the size of the copy is a multiple of the alignment
3146 and that both the source and destination pointers are aligned to that boundary.
3152 The '<tt>llvm.memmove</tt>' intrinsic copies a block of memory from the source
3153 location to the destination location, which may overlap. It
3154 copies "len" bytes of memory over. If the argument is known to be aligned to
3155 some boundary, this can be specified as the fourth argument, otherwise it should
3161 <!-- _______________________________________________________________________ -->
3162 <div class="doc_subsubsection">
3163 <a name="i_memset">'<tt>llvm.memset</tt>' Intrinsic</a>
3166 <div class="doc_text">
3170 declare void %llvm.memset(sbyte* <dest>, ubyte <val>,
3171 uint <len>, uint <align>)
3177 The '<tt>llvm.memset</tt>' intrinsic fills a block of memory with a particular
3182 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
3183 does not return a value, and takes an extra alignment argument.
3189 The first argument is a pointer to the destination to fill, the second is the
3190 byte value to fill it with, the third argument is an (arbitrarily sized) integer
3191 argument specifying the number of bytes to fill, and the fourth argument is the
3192 known alignment of destination location.
3196 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3197 the caller guarantees that the size of the copy is a multiple of the alignment
3198 and that the destination pointer is aligned to that boundary.
3204 The '<tt>llvm.memset</tt>' intrinsic fills "len" bytes of memory starting at the
3205 destination location. If the argument is known to be aligned to some boundary,
3206 this can be specified as the fourth argument, otherwise it should be set to 0 or
3212 <!-- _______________________________________________________________________ -->
3213 <div class="doc_subsubsection">
3214 <a name="i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a>
3217 <div class="doc_text">
3221 declare bool %llvm.isunordered(<float or double> Val1, <float or double> Val2)
3227 The '<tt>llvm.isunordered</tt>' intrinsic returns true if either or both of the
3228 specified floating point values is a NAN.
3234 The arguments are floating point numbers of the same type.
3240 If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
3246 <!-- _______________________________________________________________________ -->
3247 <div class="doc_subsubsection">
3248 <a name="i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a>
3251 <div class="doc_text">
3255 declare <float or double> %llvm.sqrt(<float or double> Val)
3261 The '<tt>llvm.sqrt</tt>' intrinsic returns the sqrt of the specified operand,
3262 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
3263 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
3264 negative numbers (which allows for better optimization).
3270 The argument and return value are floating point numbers of the same type.
3276 This function returns the sqrt of the specified operand if it is a positive
3277 floating point number.
3281 <!-- ======================================================================= -->
3282 <div class="doc_subsection">
3283 <a name="int_count">Bit Counting Intrinsics</a>
3286 <div class="doc_text">
3288 LLVM provides intrinsics for a few important bit counting operations.
3289 These allow efficient code generation for some algorithms.
3294 <!-- _______________________________________________________________________ -->
3295 <div class="doc_subsubsection">
3296 <a name="int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic</a>
3299 <div class="doc_text">
3303 declare int %llvm.ctpop(int <src>)
3310 The '<tt>llvm.ctpop</tt>' intrinsic counts the number of ones in a variable.
3316 The only argument is the value to be counted. The argument may be of any
3317 integer type. The return type must match the argument type.
3323 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
3327 <!-- _______________________________________________________________________ -->
3328 <div class="doc_subsubsection">
3329 <a name="int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic</a>
3332 <div class="doc_text">
3336 declare int %llvm.ctlz(int <src>)
3343 The '<tt>llvm.ctlz</tt>' intrinsic counts the number of leading zeros in a
3350 The only argument is the value to be counted. The argument may be of any
3351 integer type. The return type must match the argument type.
3357 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
3358 in a variable. If the src == 0 then the result is the size in bits of the type
3359 of src. For example, <tt>llvm.cttz(int 2) = 30</tt>.
3365 <!-- _______________________________________________________________________ -->
3366 <div class="doc_subsubsection">
3367 <a name="int_cttz">'<tt>llvm.cttz</tt>' Intrinsic</a>
3370 <div class="doc_text">
3374 declare int %llvm.cttz(int <src>)
3381 The '<tt>llvm.cttz</tt>' intrinsic counts the number of trailing zeros.
3387 The only argument is the value to be counted. The argument may be of any
3388 integer type. The return type must match the argument type.
3394 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
3395 in a variable. If the src == 0 then the result is the size in bits of the type
3396 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
3400 <!-- ======================================================================= -->
3401 <div class="doc_subsection">
3402 <a name="int_debugger">Debugger Intrinsics</a>
3405 <div class="doc_text">
3407 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
3408 are described in the <a
3409 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
3410 Debugging</a> document.
3415 <!-- *********************************************************************** -->
3418 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3419 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3420 <a href="http://validator.w3.org/check/referer"><img
3421 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3423 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
3424 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
3425 Last modified: $Date$