1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
2 <html><head><title>LLVM Assembly Language Reference Manual</title></head>
5 <table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
6 <tr><td> <font size=+5 color="#EEEEFF" face="Georgia,Palatino,Times,Roman"><b>LLVM Language Reference Manual</b></font></td>
10 <li><a href="#abstract">Abstract</a>
11 <li><a href="#introduction">Introduction</a>
12 <li><a href="#identifiers">Identifiers</a>
13 <li><a href="#typesystem">Type System</a>
15 <li><a href="#t_primitive">Primitive Types</a>
17 <li><a href="#t_classifications">Type Classifications</a>
19 <li><a href="#t_derived">Derived Types</a>
21 <li><a href="#t_array" >Array Type</a>
22 <li><a href="#t_function">Function Type</a>
23 <li><a href="#t_pointer">Pointer Type</a>
24 <li><a href="#t_struct" >Structure Type</a>
25 <!-- <li><a href="#t_packed" >Packed Type</a> -->
28 <li><a href="#highlevel">High Level Structure</a>
30 <li><a href="#modulestructure">Module Structure</a>
31 <li><a href="#globalvars">Global Variables</a>
32 <li><a href="#functionstructure">Function Structure</a>
34 <li><a href="#instref">Instruction Reference</a>
36 <li><a href="#terminators">Terminator Instructions</a>
38 <li><a href="#i_ret" >'<tt>ret</tt>' Instruction</a>
39 <li><a href="#i_br" >'<tt>br</tt>' Instruction</a>
40 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a>
41 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a>
43 <li><a href="#binaryops">Binary Operations</a>
45 <li><a href="#i_add" >'<tt>add</tt>' Instruction</a>
46 <li><a href="#i_sub" >'<tt>sub</tt>' Instruction</a>
47 <li><a href="#i_mul" >'<tt>mul</tt>' Instruction</a>
48 <li><a href="#i_div" >'<tt>div</tt>' Instruction</a>
49 <li><a href="#i_rem" >'<tt>rem</tt>' Instruction</a>
50 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a>
52 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
54 <li><a href="#i_and">'<tt>and</tt>' Instruction</a>
55 <li><a href="#i_or" >'<tt>or</tt>' Instruction</a>
56 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a>
57 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a>
58 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a>
60 <li><a href="#memoryops">Memory Access Operations</a>
62 <li><a href="#i_malloc" >'<tt>malloc</tt>' Instruction</a>
63 <li><a href="#i_free" >'<tt>free</tt>' Instruction</a>
64 <li><a href="#i_alloca" >'<tt>alloca</tt>' Instruction</a>
65 <li><a href="#i_load" >'<tt>load</tt>' Instruction</a>
66 <li><a href="#i_store" >'<tt>store</tt>' Instruction</a>
67 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
69 <li><a href="#otherops">Other Operations</a>
71 <li><a href="#i_phi" >'<tt>phi</tt>' Instruction</a>
72 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a>
73 <li><a href="#i_call" >'<tt>call</tt>' Instruction</a>
74 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a>
77 <li><a href="#intrinsics">Intrinsic Functions</a>
79 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
81 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
82 <li><a href="#i_va_end" >'<tt>llvm.va_end</tt>' Intrinsic</a>
83 <li><a href="#i_va_copy" >'<tt>llvm.va_copy</tt>' Intrinsic</a>
84 <li><a href="#i_unwind" >'<tt>llvm.unwind</tt>' Intrinsic</a>
88 <p><b>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a> and <A href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></b><p>
94 <!-- *********************************************************************** -->
95 <p><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
96 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
97 <a name="abstract">Abstract
98 </b></font></td></tr></table><ul>
99 <!-- *********************************************************************** -->
102 This document is a reference manual for the LLVM assembly language. LLVM is
103 an SSA based representation that provides type safety, low level operations,
104 flexibility, and the capability of representing 'all' high level languages
105 cleanly. It is the common code representation used throughout all phases of
106 the LLVM compilation strategy.
112 <!-- *********************************************************************** -->
113 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
114 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
115 <a name="introduction">Introduction
116 </b></font></td></tr></table><ul>
117 <!-- *********************************************************************** -->
119 The LLVM code representation is designed to be used in three different forms: as
120 an in-memory compiler IR, as an on-disk bytecode representation, suitable for
121 fast loading by a dynamic compiler, and as a human readable assembly language
122 representation. This allows LLVM to provide a powerful intermediate
123 representation for efficient compiler transformations and analysis, while
124 providing a natural means to debug and visualize the transformations. The three
125 different forms of LLVM are all equivalent. This document describes the human
126 readable representation and notation.<p>
128 The LLVM representation aims to be a light weight and low level while being
129 expressive, typed, and extensible at the same time. It aims to be a "universal
130 IR" of sorts, by being at a low enough level that high level ideas may be
131 cleanly mapped to it (similar to how microprocessors are "universal IR's",
132 allowing many source languages to be mapped to them). By providing type
133 information, LLVM can be used as the target of optimizations: for example,
134 through pointer analysis, it can be proven that a C automatic variable is never
135 accessed outside of the current function... allowing it to be promoted to a
136 simple SSA value instead of a memory location.<p>
138 <!-- _______________________________________________________________________ -->
139 </ul><a name="wellformed"><h4><hr size=0>Well Formedness</h4><ul>
141 It is important to note that this document describes 'well formed' LLVM assembly
142 language. There is a difference between what the parser accepts and what is
143 considered 'well formed'. For example, the following instruction is
144 syntactically okay, but not well formed:<p>
147 %x = <a href="#i_add">add</a> int 1, %x
150 ...because the definition of <tt>%x</tt> does not dominate all of its uses. The
151 LLVM infrastructure provides a verification pass that may be used to verify that
152 an LLVM module is well formed. This pass is automatically run by the parser
153 after parsing input assembly, and by the optimizer before it outputs bytecode.
154 The violations pointed out by the verifier pass indicate bugs in transformation
155 passes or input to the parser.<p>
157 <!-- Describe the typesetting conventions here. -->
160 <!-- *********************************************************************** -->
161 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
162 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
163 <a name="identifiers">Identifiers
164 </b></font></td></tr></table><ul>
165 <!-- *********************************************************************** -->
167 LLVM uses three different forms of identifiers, for different purposes:<p>
170 <li>Numeric constants are represented as you would expect: 12, -3 123.421, etc. Floating point constants have an optional hexidecimal notation.
171 <li>Named values are represented as a string of characters with a '%' prefix. For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
172 <li>Unnamed values are represented as an unsigned numeric value with a '%' prefix. For example, %12, %2, %44.
175 LLVM requires the values start with a '%' sign for two reasons: Compilers don't
176 need to worry about name clashes with reserved words, and the set of reserved
177 words may be expanded in the future without penalty. Additionally, unnamed
178 identifiers allow a compiler to quickly come up with a temporary variable
179 without having to avoid symbol table conflicts.<p>
181 Reserved words in LLVM are very similar to reserved words in other languages.
182 There are keywords for different opcodes ('<tt><a href="#i_add">add</a></tt>',
183 '<tt><a href="#i_cast">cast</a></tt>', '<tt><a href="#i_ret">ret</a></tt>',
184 etc...), for primitive type names ('<tt><a href="#t_void">void</a></tt>',
185 '<tt><a href="#t_uint">uint</a></tt>', etc...), and others. These reserved
186 words cannot conflict with variable names, because none of them start with a '%'
189 Here is an example of LLVM code to multiply the integer variable '<tt>%X</tt>'
194 %result = <a href="#i_mul">mul</a> uint %X, 8
197 After strength reduction:
199 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
204 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
205 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
206 %result = <a href="#i_add">add</a> uint %1, %1
209 This last way of multiplying <tt>%X</tt> by 8 illustrates several important lexical features of LLVM:<p>
212 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of line.
213 <li>Unnamed temporaries are created when the result of a computation is not
214 assigned to a named value.
215 <li>Unnamed temporaries are numbered sequentially
218 ...and it also show a convention that we follow in this document. When
219 demonstrating instructions, we will follow an instruction with a comment that
220 defines the type and name of value produced. Comments are shown in italic
223 The one non-intuitive notation for constants is the optional hexidecimal form of
224 floating point constants. For example, the form '<tt>double
225 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
226 4.5e+15</tt>' which is also supported by the parser. The only time hexadecimal
227 floating point constants are useful (and the only time that they are generated
228 by the disassembler) is when an FP constant has to be emitted that is not
229 representable as a decimal floating point number exactly. For example, NaN's,
230 infinities, and other special cases are represented in their IEEE hexadecimal
231 format so that assembly and disassembly do not cause any bits to change in the
235 <!-- *********************************************************************** -->
236 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
237 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
238 <a name="typesystem">Type System
239 </b></font></td></tr></table><ul>
240 <!-- *********************************************************************** -->
242 The LLVM type system is one of the most important features of the intermediate
243 representation. Being typed enables a number of optimizations to be performed
244 on the IR directly, without having to do extra analyses on the side before the
245 transformation. A strong type system makes it easier to read the generated code
246 and enables novel analyses and transformations that are not feasible to perform
247 on normal three address code representations.<p>
249 <!-- The written form for the type system was heavily influenced by the
250 syntactic problems with types in the C language<sup><a
251 href="#rw_stroustrup">1</a></sup>.<p> -->
255 <!-- ======================================================================= -->
256 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
257 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
258 <a name="t_primitive">Primitive Types
259 </b></font></td></tr></table><ul>
261 The primitive types are the fundemental building blocks of the LLVM system. The
262 current set of primitive types are as follows:<p>
264 <table border=0 align=center><tr><td>
266 <table border=1 cellspacing=0 cellpadding=4 align=center>
267 <tr><td><tt>void</tt></td> <td>No value</td></tr>
268 <tr><td><tt>ubyte</tt></td> <td>Unsigned 8 bit value</td></tr>
269 <tr><td><tt>ushort</tt></td><td>Unsigned 16 bit value</td></tr>
270 <tr><td><tt>uint</tt></td> <td>Unsigned 32 bit value</td></tr>
271 <tr><td><tt>ulong</tt></td> <td>Unsigned 64 bit value</td></tr>
272 <tr><td><tt>float</tt></td> <td>32 bit floating point value</td></tr>
273 <tr><td><tt>label</tt></td> <td>Branch destination</td></tr>
278 <table border=1 cellspacing=0 cellpadding=4 align=center>
279 <tr><td><tt>bool</tt></td> <td>True or False value</td></tr>
280 <tr><td><tt>sbyte</tt></td> <td>Signed 8 bit value</td></tr>
281 <tr><td><tt>short</tt></td> <td>Signed 16 bit value</td></tr>
282 <tr><td><tt>int</tt></td> <td>Signed 32 bit value</td></tr>
283 <tr><td><tt>long</tt></td> <td>Signed 64 bit value</td></tr>
284 <tr><td><tt>double</tt></td><td>64 bit floating point value</td></tr>
287 </td></tr></table><p>
291 <!-- _______________________________________________________________________ -->
292 </ul><a name="t_classifications"><h4><hr size=0>Type Classifications</h4><ul>
294 These different primitive types fall into a few useful classifications:<p>
296 <table border=1 cellspacing=0 cellpadding=4 align=center>
297 <tr><td><a name="t_signed">signed</td> <td><tt>sbyte, short, int, long, float, double</tt></td></tr>
298 <tr><td><a name="t_unsigned">unsigned</td><td><tt>ubyte, ushort, uint, ulong</tt></td></tr>
299 <tr><td><a name="t_integer">integer</td><td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td></tr>
300 <tr><td><a name="t_integral">integral</td><td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td></tr>
301 <tr><td><a name="t_floating">floating point</td><td><tt>float, double</tt></td></tr>
302 <tr><td><a name="t_firstclass">first class</td><td><tt>bool, ubyte, sbyte, ushort, short,<br> uint, int, ulong, long, float, double, <a href="#t_pointer">pointer</a></tt></td></tr>
309 <!-- ======================================================================= -->
310 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0><tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
311 <a name="t_derived">Derived Types
312 </b></font></td></tr></table><ul>
314 The real power in LLVM comes from the derived types in the system. This is what
315 allows a programmer to represent arrays, functions, pointers, and other useful
316 types. Note that these derived types may be recursive: For example, it is
317 possible to have a two dimensional array.<p>
321 <!-- _______________________________________________________________________ -->
322 </ul><a name="t_array"><h4><hr size=0>Array Type</h4><ul>
326 The array type is a very simple derived type that arranges elements sequentially
327 in memory. The array type requires a size (number of elements) and an
328 underlying data type.<p>
332 [<# elements> x <elementtype>]
335 The number of elements is a constant integer value, elementtype may be any type
340 <tt>[40 x int ]</tt>: Array of 40 integer values.<br>
341 <tt>[41 x int ]</tt>: Array of 41 integer values.<br>
342 <tt>[40 x uint]</tt>: Array of 40 unsigned integer values.<p>
345 Here are some examples of multidimensional arrays:<p>
347 <table border=0 cellpadding=0 cellspacing=0>
348 <tr><td><tt>[3 x [4 x int]]</tt></td><td>: 3x4 array integer values.</td></tr>
349 <tr><td><tt>[12 x [10 x float]]</tt></td><td>: 2x10 array of single precision floating point values.</td></tr>
350 <tr><td><tt>[2 x [3 x [4 x uint]]]</tt></td><td>: 2x3x4 array of unsigned integer values.</td></tr>
355 <!-- _______________________________________________________________________ -->
356 </ul><a name="t_function"><h4><hr size=0>Function Type</h4><ul>
360 The function type can be thought of as a function signature. It consists of a
361 return type and a list of formal parameter types. Function types are usually
362 used when to build virtual function tables (which are structures of pointers to
363 functions), for indirect function calls, and when defining a function.<p>
367 <returntype> (<parameter list>)
370 Where '<tt><parameter list></tt>' is a comma-separated list of type
371 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
372 which indicates that the function takes a variable number of arguments. Note
373 that there currently is no way to define a function in LLVM that takes a
374 variable number of arguments, but it is possible to <b>call</b> a function that
379 <table border=0 cellpadding=0 cellspacing=0>
381 <tr><td><tt>int (int)</tt></td><td>: function taking an <tt>int</tt>, returning
382 an <tt>int</tt></td></tr>
384 <tr><td><tt>float (int, int *) *</tt></td><td>: <a href="#t_pointer">Pointer</a>
385 to a function that takes an <tt>int</tt> and a <a href="#t_pointer">pointer</a>
386 to <tt>int</tt>, returning <tt>float</tt>.</td></tr>
388 <tr><td><tt>int (sbyte *, ...)</tt></td><td>: A vararg function that takes at
389 least one <a href="#t_pointer">pointer</a> to <tt>sbyte</tt> (signed char in C),
390 which returns an integer. This is the signature for <tt>printf</tt> in
398 <!-- _______________________________________________________________________ -->
399 </ul><a name="t_struct"><h4><hr size=0>Structure Type</h4><ul>
403 The structure type is used to represent a collection of data members together in
404 memory. The packing of the field types is defined to match the ABI of the
405 underlying processor. The elements of a structure may be any type that has a
408 Structures are accessed using '<tt><a href="#i_load">load</a></tt> and '<tt><a
409 href="#i_store">store</a></tt>' by getting a pointer to a field with the '<tt><a
410 href="#i_getelementptr">getelementptr</a></tt>' instruction.<p>
414 { <type list> }
419 <table border=0 cellpadding=0 cellspacing=0>
421 <tr><td><tt>{ int, int, int }</tt></td><td>: a triple of three <tt>int</tt>
424 <tr><td><tt>{ float, int (int) * }</tt></td><td>: A pair, where the first
425 element is a <tt>float</tt> and the second element is a <a
426 href="#t_pointer">pointer</a> to a <a href="t_function">function</a> that takes
427 an <tt>int</tt>, returning an <tt>int</tt>.</td></tr>
432 <!-- _______________________________________________________________________ -->
433 </ul><a name="t_pointer"><h4><hr size=0>Pointer Type</h4><ul>
437 As in many languages, the pointer type represents a pointer or reference to
438 another object, which must live in memory.<p>
447 <table border=0 cellpadding=0 cellspacing=0>
449 <tr><td><tt>[4x int]*</tt></td><td>: <a href="#t_pointer">pointer</a> to <a
450 href="#t_array">array</a> of four <tt>int</tt> values</td></tr>
452 <tr><td><tt>int (int *) *</tt></td><td>: A <a href="#t_pointer">pointer</a> to a
453 <a href="t_function">function</a> that takes an <tt>int</tt>, returning an
454 <tt>int</tt>.</td></tr>
460 <!-- _______________________________________________________________________ -->
462 </ul><a name="t_packed"><h4><hr size=0>Packed Type</h4><ul>
464 Mention/decide that packed types work with saturation or not. Maybe have a packed+saturated type in addition to just a packed type.<p>
466 Packed types should be 'nonsaturated' because standard data types are not saturated. Maybe have a saturated packed type?<p>
471 <!-- *********************************************************************** -->
472 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
473 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
474 <a name="highlevel">High Level Structure
475 </b></font></td></tr></table><ul>
476 <!-- *********************************************************************** -->
479 <!-- ======================================================================= -->
480 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
481 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
482 <a name="modulestructure">Module Structure
483 </b></font></td></tr></table><ul>
485 LLVM programs are composed of "Module"s, each of which is a translation unit of
486 the input programs. Each module consists of functions, global variables, and
487 symbol table entries. Modules may be combined together with the LLVM linker,
488 which merges function (and global variable) definitions, resolves forward
489 declarations, and merges symbol table entries. Here is an example of the "hello world" module:<p>
492 <i>; Declare the string constant as a global constant...</i>
493 <a href="#identifiers">%.LC0</a> = <a href="#linkage_decl">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
495 <i>; Forward declaration of puts</i>
496 <a href="#functionstructure">declare</a> int "puts"(sbyte*) <i>; int(sbyte*)* </i>
498 <i>; Definition of main function</i>
499 int "main"() { <i>; int()* </i>
500 <i>; Convert [13x sbyte]* to sbyte *...</i>
501 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
503 <i>; Call puts function to write out the string to stdout...</i>
504 <a href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
505 <a href="#i_ret">ret</a> int 0
509 This example is made up of a <a href="#globalvars">global variable</a> named
510 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and a
511 <a href="#functionstructure">function definition</a> for "<tt>main</tt>".<p>
513 <a name="linkage_decl">
514 In general, a module is made up of a list of global values, where both functions
515 and global variables are global values. Global values are represented by a
516 pointer to a memory location (in this case, a pointer to an array of char, and a
517 pointer to a function), and can be either "internal" or externally accessible
518 (which corresponds to the static keyword in C, when used at global scope).<p>
520 For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
521 another module defined a "<tt>.LC0</tt>" variable and was linked with this one,
522 one of the two would be renamed, preventing a collision. Since "<tt>main</tt>"
523 and "<tt>puts</tt>" are external (i.e., lacking "<tt>internal</tt>"
524 declarations), they are accessible outside of the current module. It is illegal
525 for a function declaration to be "<tt>internal</tt>".<p>
528 <!-- ======================================================================= -->
529 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
530 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
531 <a name="globalvars">Global Variables
532 </b></font></td></tr></table><ul>
534 Global variables define regions of memory allocated at compilation time instead
535 of run-time. Global variables may optionally be initialized. A variable may
536 be defined as a global "constant", which indicates that the contents of the
537 variable will never be modified (opening options for optimization). Constants
538 must always have an initial value.<p>
540 As SSA values, global variables define pointer values that are in scope
541 (i.e. they dominate) for all basic blocks in the program. Global variables
542 always define a pointer to their "content" type because they describe a region
543 of memory, and all memory objects in LLVM are accessed through pointers.<p>
547 <!-- ======================================================================= -->
548 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
549 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
550 <a name="functionstructure">Function Structure
551 </b></font></td></tr></table><ul>
553 LLVM functions definitions are composed of a (possibly empty) argument list, an
554 opening curly brace, a list of basic blocks, and a closing curly brace. LLVM
555 function declarations are defined with the "<tt>declare</tt>" keyword, a
556 function name and a function signature.<p>
558 A function definition contains a list of basic blocks, forming the CFG for the
559 function. Each basic block may optionally start with a label (giving the basic
560 block a symbol table entry), contains a list of instructions, and ends with a <a
561 href="#terminators">terminator</a> instruction (such as a branch or function
564 The first basic block in program is special in two ways: it is immediately
565 executed on entrance to the function, and it is not allowed to have predecessor
566 basic blocks (i.e. there can not be any branches to the entry block of a
570 <!-- *********************************************************************** -->
571 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
572 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
573 <a name="instref">Instruction Reference
574 </b></font></td></tr></table><ul>
575 <!-- *********************************************************************** -->
577 The LLVM instruction set consists of several different classifications of
578 instructions: <a href="#terminators">terminator instructions</a>, <a
579 href="#binaryops">binary instructions</a>, <a href="#memoryops">memory
580 instructions</a>, and <a href="#otherops">other instructions</a>.<p>
583 <!-- ======================================================================= -->
584 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
585 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
586 <a name="terminators">Terminator Instructions
587 </b></font></td></tr></table><ul>
589 As mentioned <a href="#functionstructure">previously</a>, every basic block in a
590 program ends with a "Terminator" instruction, which indicates which block should
591 be executed after the current block is finished. These terminator instructions
592 typically yield a '<tt>void</tt>' value: they produce control flow, not values
593 (the one exception being the '<a href="#i_invoke"><tt>invoke</tt></a>'
596 There are four different terminator instructions: the '<a
597 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a
598 href="#i_br"><tt>br</tt></a>' instruction, the '<a
599 href="#i_switch"><tt>switch</tt></a>' instruction, and the '<a
600 href="#i_invoke"><tt>invoke</tt></a>' instruction.<p>
603 <!-- _______________________________________________________________________ -->
604 </ul><a name="i_ret"><h4><hr size=0>'<tt>ret</tt>' Instruction</h4><ul>
608 ret <type> <value> <i>; Return a value from a non-void function</i>
609 ret void <i>; Return from void function</i>
614 The '<tt>ret</tt>' instruction is used to return control flow (and a value) from
615 a function, back to the caller.<p>
617 There are two forms of the '<tt>ret</tt>' instructruction: one that returns a
618 value and then causes control flow, and one that just causes control flow to
623 The '<tt>ret</tt>' instruction may return any '<a href="#t_firstclass">first
624 class</a>' type. Notice that a function is not <a href="#wellformed">well
625 formed</a> if there exists a '<tt>ret</tt>' instruction inside of the function
626 that returns a value that does not match the return type of the function.<p>
630 When the '<tt>ret</tt>' instruction is executed, control flow returns back to
631 the calling function's context. If the instruction returns a value, that value
632 shall be propagated into the calling function's data space.<p>
636 ret int 5 <i>; Return an integer value of 5</i>
637 ret void <i>; Return from a void function</i>
641 <!-- _______________________________________________________________________ -->
642 </ul><a name="i_br"><h4><hr size=0>'<tt>br</tt>' Instruction</h4><ul>
646 br bool <cond>, label <iftrue>, label <iffalse>
647 br label <dest> <i>; Unconditional branch</i>
652 The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
653 different basic block in the current function. There are two forms of this
654 instruction, corresponding to a conditional branch and an unconditional
659 The conditional branch form of the '<tt>br</tt>' instruction takes a single
660 '<tt>bool</tt>' value and two '<tt>label</tt>' values. The unconditional form
661 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
666 Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
667 argument is evaluated. If the value is <tt>true</tt>, control flows to the
668 '<tt>iftrue</tt>' '<tt>label</tt>' argument. If "cond" is <tt>false</tt>,
669 control flows to the '<tt>iffalse</tt>' '<tt>label</tt>' argument.<p>
674 %cond = <a href="#i_setcc">seteq</a> int %a, %b
675 br bool %cond, label %IfEqual, label %IfUnequal
677 <a href="#i_ret">ret</a> int 1
679 <a href="#i_ret">ret</a> int 0
683 <!-- _______________________________________________________________________ -->
684 </ul><a name="i_switch"><h4><hr size=0>'<tt>switch</tt>' Instruction</h4><ul>
688 switch int <value>, label <defaultdest> [ int <val>, label &dest>, ... ]
694 The '<tt>switch</tt>' instruction is used to transfer control flow to one of
695 several different places. It is a generalization of the '<tt>br</tt>'
696 instruction, allowing a branch to occur to one of many possible destinations.<p>
700 The '<tt>switch</tt>' instruction uses three parameters: a '<tt>uint</tt>'
701 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
702 an array of pairs of comparison value constants and '<tt>label</tt>'s.<p>
706 The <tt>switch</tt> instruction specifies a table of values and destinations.
707 When the '<tt>switch</tt>' instruction is executed, this table is searched for
708 the given value. If the value is found, the corresponding destination is
709 branched to, otherwise the default value it transfered to.<p>
711 <h5>Implementation:</h5>
713 Depending on properties of the target machine and the particular <tt>switch</tt>
714 instruction, this instruction may be code generated as a series of chained
715 conditional branches, or with a lookup table.<p>
719 <i>; Emulate a conditional br instruction</i>
720 %Val = <a href="#i_cast">cast</a> bool %value to uint
721 switch int %Val, label %truedest [int 0, label %falsedest ]
723 <i>; Emulate an unconditional br instruction</i>
724 switch int 0, label %dest [ ]
726 <i>; Implement a jump table:</i>
727 switch int %val, label %otherwise [ int 0, label %onzero,
729 int 2, label %ontwo ]
734 <!-- _______________________________________________________________________ -->
735 </ul><a name="i_invoke"><h4><hr size=0>'<tt>invoke</tt>' Instruction</h4><ul>
739 <result> = invoke <ptr to function ty> %<function ptr val>(<function args>)
740 to label <normal label> except label <exception label>
745 The '<tt>invoke</tt>' instruction causes control to transfer to a specified
746 function, with the possibility of control flow transfer to either the
747 '<tt>normal label</tt>' label or the '<tt>exception label</tt>'. If the callee
748 function invokes the "<tt><a href="#i_ret">ret</a></tt>" instruction, control
749 flow will return to the "normal" label. If the callee (or any indirect callees)
750 calls the "<a href="#i_unwind"><tt>llvm.unwind</tt></a>" intrinsic, control is
751 interrupted, and continued at the "except" label.<p>
756 This instruction requires several arguments:<p>
759 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
760 function value being invoked. In most cases, this is a direct function
761 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
762 an arbitrary pointer to function value.
764 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
765 function to be invoked.
767 <li>'<tt>function args</tt>': argument list whose types match the function
768 signature argument types. If the function signature indicates the function
769 accepts a variable number of arguments, the extra arguments can be specified.
771 <li>'<tt>normal label</tt>': the label reached when the called function executes
772 a '<tt><a href="#i_ret">ret</a></tt>' instruction.
774 <li>'<tt>exception label</tt>': the label reached when a callee calls the <a
775 href="#i_unwind"><tt>llvm.unwind</tt></a> intrinsic.
780 This instruction is designed to operate as a standard '<tt><a
781 href="#i_call">call</a></tt>' instruction in most regards. The primary
782 difference is that it establishes an association with a label, which is used by the runtime library to unwind the stack.<p>
784 This instruction is used in languages with destructors to ensure that proper
785 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
786 exception. Additionally, this is important for implementation of
787 '<tt>catch</tt>' clauses in high-level languages that support them.<p>
791 %retval = invoke int %Test(int 15)
793 except label %TestCleanup <i>; {int}:retval set</i>
798 <!-- ======================================================================= -->
799 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0><tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
800 <a name="binaryops">Binary Operations
801 </b></font></td></tr></table><ul>
803 Binary operators are used to do most of the computation in a program. They
804 require two operands, execute an operation on them, and produce a single value.
805 The result value of a binary operator is not neccesarily the same type as its
808 There are several different binary operators:<p>
811 <!-- _______________________________________________________________________ -->
812 </ul><a name="i_add"><h4><hr size=0>'<tt>add</tt>' Instruction</h4><ul>
816 <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
820 The '<tt>add</tt>' instruction returns the sum of its two operands.<p>
823 The two arguments to the '<tt>add</tt>' instruction must be either <a href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values. Both arguments must have identical types.<p>
827 The value produced is the integer or floating point sum of the two operands.<p>
831 <result> = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
835 <!-- _______________________________________________________________________ -->
836 </ul><a name="i_sub"><h4><hr size=0>'<tt>sub</tt>' Instruction</h4><ul>
840 <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
845 The '<tt>sub</tt>' instruction returns the difference of its two operands.<p>
847 Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
848 instruction present in most other intermediate representations.<p>
852 The two arguments to the '<tt>sub</tt>' instruction must be either <a
853 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
854 values. Both arguments must have identical types.<p>
858 The value produced is the integer or floating point difference of the two
863 <result> = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
864 <result> = sub int 0, %val <i>; yields {int}:result = -%var</i>
867 <!-- _______________________________________________________________________ -->
868 </ul><a name="i_mul"><h4><hr size=0>'<tt>mul</tt>' Instruction</h4><ul>
872 <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
876 The '<tt>mul</tt>' instruction returns the product of its two operands.<p>
879 The two arguments to the '<tt>mul</tt>' instruction must be either <a href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values. Both arguments must have identical types.<p>
883 The value produced is the integer or floating point product of the two
886 There is no signed vs unsigned multiplication. The appropriate action is taken
887 based on the type of the operand. <p>
892 <result> = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
896 <!-- _______________________________________________________________________ -->
897 </ul><a name="i_div"><h4><hr size=0>'<tt>div</tt>' Instruction</h4><ul>
901 <result> = div <ty> <var1>, <var2> <i>; yields {ty}:result</i>
906 The '<tt>div</tt>' instruction returns the quotient of its two operands.<p>
910 The two arguments to the '<tt>div</tt>' instruction must be either <a
911 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
912 values. Both arguments must have identical types.<p>
916 The value produced is the integer or floating point quotient of the two
921 <result> = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
925 <!-- _______________________________________________________________________ -->
926 </ul><a name="i_rem"><h4><hr size=0>'<tt>rem</tt>' Instruction</h4><ul>
930 <result> = rem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
934 The '<tt>rem</tt>' instruction returns the remainder from the division of its two operands.<p>
937 The two arguments to the '<tt>rem</tt>' instruction must be either <a href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values. Both arguments must have identical types.<p>
941 This returns the <i>remainder</i> of a division (where the result has the same
942 sign as the divisor), not the <i>modulus</i> (where the result has the same sign
943 as the dividend) of a value. For more information about the difference, see: <a
944 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The Math
949 <result> = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
953 <!-- _______________________________________________________________________ -->
954 </ul><a name="i_setcc"><h4><hr size=0>'<tt>set<i>cc</i></tt>' Instructions</h4><ul>
958 <result> = seteq <ty> <var1>, <var2> <i>; yields {bool}:result</i>
959 <result> = setne <ty> <var1>, <var2> <i>; yields {bool}:result</i>
960 <result> = setlt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
961 <result> = setgt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
962 <result> = setle <ty> <var1>, <var2> <i>; yields {bool}:result</i>
963 <result> = setge <ty> <var1>, <var2> <i>; yields {bool}:result</i>
966 <h5>Overview:</h5> The '<tt>set<i>cc</i></tt>' family of instructions returns a
967 boolean value based on a comparison of their two operands.<p>
969 <h5>Arguments:</h5> The two arguments to the '<tt>set<i>cc</i></tt>'
970 instructions must be of <a href="#t_firstclass">first class</a> or <a
971 href="#t_pointer">pointer</a> type (it is not possible to compare
972 '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>' or '<tt>void</tt>'
973 values, etc...). Both arguments must have identical types.<p>
975 The '<tt>setlt</tt>', '<tt>setgt</tt>', '<tt>setle</tt>', and '<tt>setge</tt>'
976 instructions do not operate on '<tt>bool</tt>' typed arguments.<p>
980 The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
981 both operands are equal.<br>
983 The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
984 both operands are unequal.<br>
986 The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
987 the first operand is less than the second operand.<br>
989 The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
990 the first operand is greater than the second operand.<br>
992 The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
993 the first operand is less than or equal to the second operand.<br>
995 The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>' value if
996 the first operand is greater than or equal to the second operand.<p>
1000 <result> = seteq int 4, 5 <i>; yields {bool}:result = false</i>
1001 <result> = setne float 4, 5 <i>; yields {bool}:result = true</i>
1002 <result> = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1003 <result> = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1004 <result> = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1005 <result> = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1010 <!-- ======================================================================= -->
1011 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
1012 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
1013 <a name="bitwiseops">Bitwise Binary Operations
1014 </b></font></td></tr></table><ul>
1016 Bitwise binary operators are used to do various forms of bit-twiddling in a
1017 program. They are generally very efficient instructions, and can commonly be
1018 strength reduced from other instructions. They require two operands, execute an
1019 operation on them, and produce a single value. The resulting value of the
1020 bitwise binary operators is always the same type as its first operand.<p>
1022 <!-- _______________________________________________________________________ -->
1023 </ul><a name="i_and"><h4><hr size=0>'<tt>and</tt>' Instruction</h4><ul>
1027 <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1031 The '<tt>and</tt>' instruction returns the bitwise logical and of its two operands.<p>
1035 The two arguments to the '<tt>and</tt>' instruction must be <a
1036 href="#t_integral">integral</a> values. Both arguments must have identical
1042 The truth table used for the '<tt>and</tt>' instruction is:<p>
1044 <center><table border=1 cellspacing=0 cellpadding=4>
1045 <tr><td>In0</td> <td>In1</td> <td>Out</td></tr>
1046 <tr><td>0</td> <td>0</td> <td>0</td></tr>
1047 <tr><td>0</td> <td>1</td> <td>0</td></tr>
1048 <tr><td>1</td> <td>0</td> <td>0</td></tr>
1049 <tr><td>1</td> <td>1</td> <td>1</td></tr>
1050 </table></center><p>
1055 <result> = and int 4, %var <i>; yields {int}:result = 4 & %var</i>
1056 <result> = and int 15, 40 <i>; yields {int}:result = 8</i>
1057 <result> = and int 4, 8 <i>; yields {int}:result = 0</i>
1062 <!-- _______________________________________________________________________ -->
1063 </ul><a name="i_or"><h4><hr size=0>'<tt>or</tt>' Instruction</h4><ul>
1067 <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1070 <h5>Overview:</h5> The '<tt>or</tt>' instruction returns the bitwise logical
1071 inclusive or of its two operands.<p>
1075 The two arguments to the '<tt>or</tt>' instruction must be <a
1076 href="#t_integral">integral</a> values. Both arguments must have identical
1082 The truth table used for the '<tt>or</tt>' instruction is:<p>
1084 <center><table border=1 cellspacing=0 cellpadding=4>
1085 <tr><td>In0</td> <td>In1</td> <td>Out</td></tr>
1086 <tr><td>0</td> <td>0</td> <td>0</td></tr>
1087 <tr><td>0</td> <td>1</td> <td>1</td></tr>
1088 <tr><td>1</td> <td>0</td> <td>1</td></tr>
1089 <tr><td>1</td> <td>1</td> <td>1</td></tr>
1090 </table></center><p>
1095 <result> = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
1096 <result> = or int 15, 40 <i>; yields {int}:result = 47</i>
1097 <result> = or int 4, 8 <i>; yields {int}:result = 12</i>
1101 <!-- _______________________________________________________________________ -->
1102 </ul><a name="i_xor"><h4><hr size=0>'<tt>xor</tt>' Instruction</h4><ul>
1106 <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1111 The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of its
1116 The two arguments to the '<tt>xor</tt>' instruction must be <a
1117 href="#t_integral">integral</a> values. Both arguments must have identical
1123 The truth table used for the '<tt>xor</tt>' instruction is:<p>
1125 <center><table border=1 cellspacing=0 cellpadding=4>
1126 <tr><td>In0</td> <td>In1</td> <td>Out</td></tr>
1127 <tr><td>0</td> <td>0</td> <td>0</td></tr>
1128 <tr><td>0</td> <td>1</td> <td>1</td></tr>
1129 <tr><td>1</td> <td>0</td> <td>1</td></tr>
1130 <tr><td>1</td> <td>1</td> <td>0</td></tr>
1131 </table></center><p>
1136 <result> = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
1137 <result> = xor int 15, 40 <i>; yields {int}:result = 39</i>
1138 <result> = xor int 4, 8 <i>; yields {int}:result = 12</i>
1142 <!-- _______________________________________________________________________ -->
1143 </ul><a name="i_shl"><h4><hr size=0>'<tt>shl</tt>' Instruction</h4><ul>
1147 <result> = shl <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1152 The '<tt>shl</tt>' instruction returns the first operand shifted to the left a
1153 specified number of bits.
1157 The first argument to the '<tt>shl</tt>' instruction must be an <a
1158 href="#t_integer">integer</a> type. The second argument must be an
1159 '<tt>ubyte</tt>' type.<p>
1163 The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.<p>
1168 <result> = shl int 4, ubyte %var <i>; yields {int}:result = 4 << %var</i>
1169 <result> = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1170 <result> = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1174 <!-- _______________________________________________________________________ -->
1175 </ul><a name="i_shr"><h4><hr size=0>'<tt>shr</tt>' Instruction</h4><ul>
1180 <result> = shr <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1184 The '<tt>shr</tt>' instruction returns the first operand shifted to the right a specified number of bits.
1187 The first argument to the '<tt>shr</tt>' instruction must be an <a href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>' type.<p>
1191 If the first argument is a <a href="#t_signed">signed</a> type, the most
1192 significant bit is duplicated in the newly free'd bit positions. If the first
1193 argument is unsigned, zero bits shall fill the empty positions.<p>
1197 <result> = shr int 4, ubyte %var <i>; yields {int}:result = 4 >> %var</i>
1198 <result> = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
1199 <result> = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
1200 <result> = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1201 <result> = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
1208 <!-- ======================================================================= -->
1209 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
1210 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
1211 <a name="memoryops">Memory Access Operations
1212 </b></font></td></tr></table><ul>
1214 Accessing memory in SSA form is, well, sticky at best. This section describes how to read, write, allocate and free memory in LLVM.<p>
1217 <!-- _______________________________________________________________________ -->
1218 </ul><a name="i_malloc"><h4><hr size=0>'<tt>malloc</tt>' Instruction</h4><ul>
1222 <result> = malloc <type>, uint <NumElements> <i>; yields {type*}:result</i>
1223 <result> = malloc <type> <i>; yields {type*}:result</i>
1227 The '<tt>malloc</tt>' instruction allocates memory from the system heap and returns a pointer to it.<p>
1231 The the '<tt>malloc</tt>' instruction allocates
1232 <tt>sizeof(<type>)*NumElements</tt> bytes of memory from the operating
1233 system, and returns a pointer of the appropriate type to the program. The
1234 second form of the instruction is a shorter version of the first instruction
1235 that defaults to allocating one element.<p>
1237 '<tt>type</tt>' must be a sized type<p>
1240 Memory is allocated, a pointer is returned.<p>
1244 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
1246 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
1247 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1248 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
1252 <!-- _______________________________________________________________________ -->
1253 </ul><a name="i_free"><h4><hr size=0>'<tt>free</tt>' Instruction</h4><ul>
1257 free <type> <value> <i>; yields {void}</i>
1262 The '<tt>free</tt>' instruction returns memory back to the unused memory heap, to be reallocated in the future.<p>
1267 '<tt>value</tt>' shall be a pointer value that points to a value that was
1268 allocated with the '<tt><a href="#i_malloc">malloc</a></tt>' instruction.<p>
1273 Access to the memory pointed to by the pointer is not longer defined after this instruction executes.<p>
1277 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
1278 free [4 x ubyte]* %array
1282 <!-- _______________________________________________________________________ -->
1283 </ul><a name="i_alloca"><h4><hr size=0>'<tt>alloca</tt>' Instruction</h4><ul>
1287 <result> = alloca <type>, uint <NumElements> <i>; yields {type*}:result</i>
1288 <result> = alloca <type> <i>; yields {type*}:result</i>
1293 The '<tt>alloca</tt>' instruction allocates memory on the current stack frame of
1294 the procedure that is live until the current function returns to its caller.<p>
1298 The the '<tt>alloca</tt>' instruction allocates
1299 <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the runtime stack,
1300 returning a pointer of the appropriate type to the program. The second form of
1301 the instruction is a shorter version of the first that defaults to allocating
1304 '<tt>type</tt>' may be any sized type.<p>
1308 Memory is allocated, a pointer is returned. '<tt>alloca</tt>'d memory is
1309 automatically released when the function returns. The '<tt>alloca</tt>'
1310 instruction is commonly used to represent automatic variables that must have an
1311 address available, as well as spilled variables.<p>
1315 %ptr = alloca int <i>; yields {int*}:ptr</i>
1316 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
1320 <!-- _______________________________________________________________________ -->
1321 </ul><a name="i_load"><h4><hr size=0>'<tt>load</tt>' Instruction</h4><ul>
1325 <result> = load <ty>* <pointer>
1329 The '<tt>load</tt>' instruction is used to read from memory.<p>
1333 The argument to the '<tt>load</tt>' instruction specifies the memory address to load from. The pointer must point to a <a href="t_firstclass">first class</a> type.<p>
1337 The location of memory pointed to is loaded.
1341 %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1342 <a href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1343 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1349 <!-- _______________________________________________________________________ -->
1350 </ul><a name="i_store"><h4><hr size=0>'<tt>store</tt>' Instruction</h4><ul>
1354 store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
1358 The '<tt>store</tt>' instruction is used to write to memory.<p>
1362 There are two arguments to the '<tt>store</tt>' instruction: a value to store
1363 and an address to store it into. The type of the '<tt><pointer></tt>'
1364 operand must be a pointer to the type of the '<tt><value></tt>'
1367 <h5>Semantics:</h5> The contents of memory are updated to contain
1368 '<tt><value></tt>' at the location specified by the
1369 '<tt><pointer></tt>' operand.<p>
1373 %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1374 <a href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1375 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1381 <!-- _______________________________________________________________________ -->
1382 </ul><a name="i_getelementptr"><h4><hr size=0>'<tt>getelementptr</tt>' Instruction</h4><ul>
1386 <result> = getelementptr <ty>* <ptrval>{, long <aidx>|, ubyte <sidx>}*
1391 The '<tt>getelementptr</tt>' instruction is used to get the address of a
1392 subelement of an aggregate data structure.<p>
1396 This instruction takes a list of <tt>long</tt> values and <tt>ubyte</tt>
1397 constants that indicate what form of addressing to perform. The actual types of
1398 the arguments provided depend on the type of the first pointer argument. The
1399 '<tt>getelementptr</tt>' instruction is used to index down through the type
1400 levels of a structure.<p>
1402 For example, lets consider a C code fragment and how it gets compiled to
1417 int *foo(struct ST *s) {
1418 return &s[1].Z.B[5][13];
1422 The LLVM code generated by the GCC frontend is:
1425 %RT = type { sbyte, [10 x [20 x int]], sbyte }
1426 %ST = type { int, double, %RT }
1428 int* "foo"(%ST* %s) {
1429 %reg = getelementptr %ST* %s, long 1, ubyte 2, ubyte 1, long 5, long 13
1436 The index types specified for the '<tt>getelementptr</tt>' instruction depend on
1437 the pointer type that is being index into. <a href="t_pointer">Pointer</a> and
1438 <a href="t_array">array</a> types require '<tt>long</tt>' values, and <a
1439 href="t_struct">structure</a> types require '<tt>ubyte</tt>'
1440 <b>constants</b>.<p>
1442 In the example above, the first index is indexing into the '<tt>%ST*</tt>' type,
1443 which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT }</tt>'
1444 type, a structure. The second index indexes into the third element of the
1445 structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]], sbyte
1446 }</tt>' type, another structure. The third index indexes into the second
1447 element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
1448 array. The two dimensions of the array are subscripted into, yielding an
1449 '<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction return a pointer
1450 to this element, thus yielding a '<tt>int*</tt>' type.<p>
1452 Note that it is perfectly legal to index partially through a structure,
1453 returning a pointer to an inner element. Because of this, the LLVM code for the
1454 given testcase is equivalent to:<p>
1457 int* "foo"(%ST* %s) {
1458 %t1 = getelementptr %ST* %s , long 1 <i>; yields %ST*:%t1</i>
1459 %t2 = getelementptr %ST* %t1, long 0, ubyte 2 <i>; yields %RT*:%t2</i>
1460 %t3 = getelementptr %RT* %t2, long 0, ubyte 1 <i>; yields [10 x [20 x int]]*:%t3</i>
1461 %t4 = getelementptr [10 x [20 x int]]* %t3, long 0, long 5 <i>; yields [20 x int]*:%t4</i>
1462 %t5 = getelementptr [20 x int]* %t4, long 0, long 13 <i>; yields int*:%t5</i>
1471 <i>; yields [12 x ubyte]*:aptr</i>
1472 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, ubyte 1
1477 <!-- ======================================================================= -->
1478 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
1479 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
1480 <a name="otherops">Other Operations
1481 </b></font></td></tr></table><ul>
1483 The instructions in this catagory are the "miscellaneous" functions, that defy better classification.<p>
1486 <!-- _______________________________________________________________________ -->
1487 </ul><a name="i_phi"><h4><hr size=0>'<tt>phi</tt>' Instruction</h4><ul>
1491 <result> = phi <ty> [ <val0>, <label0>], ...
1496 The '<tt>phi</tt>' instruction is used to implement the φ node in the SSA
1497 graph representing the function.<p>
1501 The type of the incoming values are specified with the first type field. After
1502 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
1503 one pair for each predecessor basic block of the current block.<p>
1505 There must be no non-phi instructions between the start of a basic block and the
1506 PHI instructions: i.e. PHI instructions must be first in a basic block.<p>
1510 At runtime, the '<tt>phi</tt>' instruction logically takes on the value
1511 specified by the parameter, depending on which basic block we came from in the
1512 last <a href="#terminators">terminator</a> instruction.<p>
1517 Loop: ; Infinite loop that counts from 0 on up...
1518 %indvar = phi uint [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
1519 %nextindvar = add uint %indvar, 1
1524 <!-- _______________________________________________________________________ -->
1525 </ul><a name="i_cast"><h4><hr size=0>'<tt>cast .. to</tt>' Instruction</h4><ul>
1529 <result> = cast <ty> <value> to <ty2> <i>; yields ty2</i>
1534 The '<tt>cast</tt>' instruction is used as the primitive means to convert
1535 integers to floating point, change data type sizes, and break type safety (by
1536 casting pointers).<p>
1540 The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
1541 class value, and a type to cast it to, which must also be a first class type.<p>
1545 This instruction follows the C rules for explicit casts when determining how the
1546 data being cast must change to fit in its new container.<p>
1548 When casting to bool, any value that would be considered true in the context of
1549 a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
1550 all else are '<tt>false</tt>'.<p>
1552 When extending an integral value from a type of one signness to another (for
1553 example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
1554 <b>source</b> value is signed, and zero-extended if the source value is
1555 unsigned. <tt>bool</tt> values are always zero extended into either zero or
1560 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
1561 %Y = cast int 123 to bool <i>; yields bool:true</i>
1566 <!-- _______________________________________________________________________ -->
1567 </ul><a name="i_call"><h4><hr size=0>'<tt>call</tt>' Instruction</h4><ul>
1571 <result> = call <ty>* <fnptrval>(<param list>)
1576 The '<tt>call</tt>' instruction represents a simple function call.<p>
1580 This instruction requires several arguments:<p>
1583 <li>'<tt>ty</tt>': shall be the signature of the pointer to function value being
1584 invoked. The argument types must match the types implied by this signature.<p>
1586 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to be
1587 invoked. In most cases, this is a direct function invocation, but indirect
1588 <tt>call</tt>s are just as possible, calling an arbitrary pointer to function
1591 <li>'<tt>function args</tt>': argument list whose types match the function
1592 signature argument types. If the function signature indicates the function
1593 accepts a variable number of arguments, the extra arguments can be specified.
1598 The '<tt>call</tt>' instruction is used to cause control flow to transfer to a
1599 specified function, with its incoming arguments bound to the specified values.
1600 Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called function,
1601 control flow continues with the instruction after the function call, and the
1602 return value of the function is bound to the result argument. This is a simpler
1603 case of the <a href="#i_invoke">invoke</a> instruction.<p>
1607 %retval = call int %test(int %argc)
1608 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
1612 <!-- _______________________________________________________________________ -->
1613 </ul><a name="i_va_arg"><h4><hr size=0>'<tt>va_arg</tt>' Instruction</h4><ul>
1617 <result> = va_arg <va_list>* <arglist>, <retty>
1622 The '<tt>va_arg</tt>' instruction is used to access arguments passed through the
1623 "variable argument" area of a function call. It corresponds directly to the
1624 <tt>va_arg</tt> macro in C.<p>
1628 This instruction takes a pointer to a <tt>valist</tt> value to read a new
1629 argument from. The return type of the instruction is defined by the second
1630 argument, a type.<p>
1634 The '<tt>va_arg</tt>' instruction works just like the <tt>va_arg</tt> macro
1635 available in C. In a target-dependent way, it reads the argument indicated by
1636 the value the arglist points to, updates the arglist, then returns a value of
1637 the specified type. This instruction should be used in conjunction with the
1638 variable argument handling <a href="#int_varargs">Intrinsic Functions</a>.<p>
1640 It is legal for this instruction to be called in a function which does not take
1641 a variable number of arguments, for example, the <tt>vfprintf</tt> function.<p>
1643 <tt>va_arg</tt> is an LLVM instruction instead of an <a
1644 href="#intrinsics">intrinsic function</a> because the return type depends on an
1649 See the <a href="#int_varargs">variable argument processing</a> section.<p>
1651 <!-- *********************************************************************** -->
1652 </ul><table width="100%" bgcolor="#330077" border=0 cellpadding=4 cellspacing=0>
1653 <tr><td align=center><font color="#EEEEFF" size=+2 face="Georgia,Palatino"><b>
1654 <a name="intrinsics">Intrinsic Functions
1655 </b></font></td></tr></table><ul>
1656 <!-- *********************************************************************** -->
1658 LLVM supports the notion of an "intrinsic function". These functions have well
1659 known names and semantics, and are required to follow certain restrictions.
1660 Overall, these instructions represent an extension mechanism for the LLVM
1661 language that does not require changing all of the transformations in LLVM to
1662 add to the language (or the bytecode reader/writer, the parser, etc...).<p>
1664 Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix, this
1665 prefix is reserved in LLVM for intrinsic names, thus functions may not be named
1666 this. Intrinsic functions must always be external functions: you cannot define
1667 the body of intrinsic functions. Intrinsic functions may only be used in call
1668 or invoke instructions: it is illegal to take the address of an intrinsic
1669 function. Additionally, because intrinsic functions are part of the LLVM
1670 language, it is required that they all be documented here if any are added.<p>
1672 Unless an intrinsic function is target-specific, there must be a lowering pass
1673 to eliminate the intrinsic or all backends must support the intrinsic
1677 <!-- ======================================================================= -->
1678 </ul><table width="100%" bgcolor="#441188" border=0 cellpadding=4 cellspacing=0>
1679 <tr><td> </td><td width="100%"> <font color="#EEEEFF" face="Georgia,Palatino"><b>
1680 <a name="int_varargs">Variable Argument Handling Intrinsics
1681 </b></font></td></tr></table><ul>
1683 Variable argument support is defined in LLVM with the <a
1684 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three intrinsic
1685 functions. These function correspond almost directly to the similarly named
1686 macros defined in the <tt><stdarg.h></tt> header file.<p>
1688 All of these functions operate on arguments that use a target-specific type
1689 "<tt>va_list</tt>". The LLVM assembly language reference manual does not define
1690 what this type is, so all transformations should be prepared to handle
1691 intrinsics with any type used.<p>
1693 This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction
1694 and the variable argument handling intrinsic functions are used.<p>
1697 int %test(int %X, ...) {
1698 ; Allocate two va_list items. On this target, va_list is of type sbyte*
1702 ; Initialize variable argument processing
1703 call void (sbyte**)* %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
1705 ; Read a single integer argument
1706 %tmp = <a href="#i_va_arg">va_arg</a> sbyte** %ap, int
1708 ; Demonstrate usage of llvm.va_copy and llvm_va_end
1709 %apv = load sbyte** %ap
1710 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte* %apv)
1711 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
1713 ; Stop processing of arguments.
1714 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
1719 <!-- _______________________________________________________________________ -->
1720 </ul><a name="i_va_start"><h4><hr size=0>'<tt>llvm.va_start</tt>' Intrinsic</h4><ul>
1724 call void (va_list*)* %llvm.va_start(<va_list>* <arglist>)
1729 The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt> for
1730 subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt> and <tt><a
1731 href="#i_va_end">llvm.va_end</a></tt>, and must be called before either are
1736 The argument is a pointer to a <tt>va_list</tt> element to initialize.<p>
1740 The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
1741 macro available in C. In a target-dependent way, it initializes the
1742 <tt>va_list</tt> element the argument points to, so that the next call to
1743 <tt>va_arg</tt> will produce the first variable argument passed to the function.
1744 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
1745 last argument of the function, the compiler can figure that out.<p>
1748 <!-- _______________________________________________________________________ -->
1749 </ul><a name="i_va_end"><h4><hr size=0>'<tt>llvm.va_end</tt>' Intrinsic</h4><ul>
1753 call void (va_list*)* %llvm.va_end(<va_list>* <arglist>)
1758 The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt> which
1759 has been initialized previously with <tt><a
1760 href="#i_va_begin">llvm.va_begin</a></tt>.<p>
1764 The argument is a pointer to a <tt>va_list</tt> element to destroy.<p>
1768 The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt> macro
1769 available in C. In a target-dependent way, it destroys the <tt>va_list</tt>
1770 that the argument points to. Calls to <a
1771 href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
1772 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly with calls
1773 to <tt>llvm.va_end</tt>.<p>
1777 <!-- _______________________________________________________________________ -->
1778 </ul><a name="i_va_copy"><h4><hr size=0>'<tt>llvm.va_copy</tt>' Intrinsic</h4><ul>
1782 call void (va_list*, va_list)* %va_copy(<va_list>* <destarglist>,
1783 <va_list> <srcarglist>)
1788 The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
1789 the source argument list to the destination argument list.<p>
1793 The first argument is a pointer to a <tt>va_list</tt> element to initialize.
1794 The second argument is a <tt>va_list</tt> element to copy from.<p>
1799 The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
1800 available in C. In a target-dependent way, it copies the source
1801 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
1802 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
1803 arbitrarily complex and require memory allocation, for example.<p>
1806 <!-- _______________________________________________________________________ -->
1807 </ul><a name="i_unwind"><h4><hr size=0>'<tt>llvm.unwind</tt>' Intrinsic</h4><ul>
1811 call void (void)* %llvm.unwind()
1816 The '<tt>llvm.unwind</tt>' intrinsic unwinds the stack, continuing control flow
1817 at the first callee in the dynamic call stack which used an <a
1818 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1819 primarily used to implement exception handling.
1823 The '<tt>llvm.unwind</tt>' intrinsic causes execution of the current function to
1824 immediately halt. The dynamic call stack is then searched for the first <a
1825 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1826 execution continues at the "exceptional" destination block specified by the
1827 invoke instruction. If there is no <tt>invoke</tt> instruction in the dynamic
1828 call chain, undefined behavior results.
1832 <!-- *********************************************************************** -->
1834 <!-- *********************************************************************** -->
1839 <address><a href="mailto:sabre@nondot.org">Chris Lattner</a></address>
1840 <!-- Created: Tue Jan 23 15:19:28 CST 2001 -->
1841 <!-- hhmts start -->
1842 Last modified: Thu Aug 28 17:11:50 CDT 2003