1 ==============================
2 LLVM Language Reference Manual
3 ==============================
12 This document is a reference manual for the LLVM assembly language. LLVM
13 is a Static Single Assignment (SSA) based representation that provides
14 type safety, low-level operations, flexibility, and the capability of
15 representing 'all' high-level languages cleanly. It is the common code
16 representation used throughout all phases of the LLVM compilation
22 The LLVM code representation is designed to be used in three different
23 forms: as an in-memory compiler IR, as an on-disk bitcode representation
24 (suitable for fast loading by a Just-In-Time compiler), and as a human
25 readable assembly language representation. This allows LLVM to provide a
26 powerful intermediate representation for efficient compiler
27 transformations and analysis, while providing a natural means to debug
28 and visualize the transformations. The three different forms of LLVM are
29 all equivalent. This document describes the human readable
30 representation and notation.
32 The LLVM representation aims to be light-weight and low-level while
33 being expressive, typed, and extensible at the same time. It aims to be
34 a "universal IR" of sorts, by being at a low enough level that
35 high-level ideas may be cleanly mapped to it (similar to how
36 microprocessors are "universal IR's", allowing many source languages to
37 be mapped to them). By providing type information, LLVM can be used as
38 the target of optimizations: for example, through pointer analysis, it
39 can be proven that a C automatic variable is never accessed outside of
40 the current function, allowing it to be promoted to a simple SSA value
41 instead of a memory location.
48 It is important to note that this document describes 'well formed' LLVM
49 assembly language. There is a difference between what the parser accepts
50 and what is considered 'well formed'. For example, the following
51 instruction is syntactically okay, but not well formed:
57 because the definition of ``%x`` does not dominate all of its uses. The
58 LLVM infrastructure provides a verification pass that may be used to
59 verify that an LLVM module is well formed. This pass is automatically
60 run by the parser after parsing input assembly and by the optimizer
61 before it outputs bitcode. The violations pointed out by the verifier
62 pass indicate bugs in transformation passes or input to the parser.
69 LLVM identifiers come in two basic types: global and local. Global
70 identifiers (functions, global variables) begin with the ``'@'``
71 character. Local identifiers (register names, types) begin with the
72 ``'%'`` character. Additionally, there are three different formats for
73 identifiers, for different purposes:
75 #. Named values are represented as a string of characters with their
76 prefix. For example, ``%foo``, ``@DivisionByZero``,
77 ``%a.really.long.identifier``. The actual regular expression used is
78 '``[%@][a-zA-Z$._][a-zA-Z$._0-9]*``'. Identifiers which require other
79 characters in their names can be surrounded with quotes. Special
80 characters may be escaped using ``"\xx"`` where ``xx`` is the ASCII
81 code for the character in hexadecimal. In this way, any character can
82 be used in a name value, even quotes themselves.
83 #. Unnamed values are represented as an unsigned numeric value with
84 their prefix. For example, ``%12``, ``@2``, ``%44``.
85 #. Constants, which are described in the section Constants_ below.
87 LLVM requires that values start with a prefix for two reasons: Compilers
88 don't need to worry about name clashes with reserved words, and the set
89 of reserved words may be expanded in the future without penalty.
90 Additionally, unnamed identifiers allow a compiler to quickly come up
91 with a temporary variable without having to avoid symbol table
94 Reserved words in LLVM are very similar to reserved words in other
95 languages. There are keywords for different opcodes ('``add``',
96 '``bitcast``', '``ret``', etc...), for primitive type names ('``void``',
97 '``i32``', etc...), and others. These reserved words cannot conflict
98 with variable names, because none of them start with a prefix character
101 Here is an example of LLVM code to multiply the integer variable
108 %result = mul i32 %X, 8
110 After strength reduction:
114 %result = shl i32 %X, 3
120 %0 = add i32 %X, %X ; yields {i32}:%0
121 %1 = add i32 %0, %0 ; yields {i32}:%1
122 %result = add i32 %1, %1
124 This last way of multiplying ``%X`` by 8 illustrates several important
125 lexical features of LLVM:
127 #. Comments are delimited with a '``;``' and go until the end of line.
128 #. Unnamed temporaries are created when the result of a computation is
129 not assigned to a named value.
130 #. Unnamed temporaries are numbered sequentially (using a per-function
131 incrementing counter, starting with 0). Note that basic blocks are
132 included in this numbering. For example, if the entry basic block is not
133 given a label name, then it will get number 0.
135 It also shows a convention that we follow in this document. When
136 demonstrating instructions, we will follow an instruction with a comment
137 that defines the type and name of value produced.
145 LLVM programs are composed of ``Module``'s, each of which is a
146 translation unit of the input programs. Each module consists of
147 functions, global variables, and symbol table entries. Modules may be
148 combined together with the LLVM linker, which merges function (and
149 global variable) definitions, resolves forward declarations, and merges
150 symbol table entries. Here is an example of the "hello world" module:
154 ; Declare the string constant as a global constant.
155 @.str = private unnamed_addr constant [13 x i8] c"hello world\0A\00"
157 ; External declaration of the puts function
158 declare i32 @puts(i8* nocapture) nounwind
160 ; Definition of main function
161 define i32 @main() { ; i32()*
162 ; Convert [13 x i8]* to i8 *...
163 %cast210 = getelementptr [13 x i8]* @.str, i64 0, i64 0
165 ; Call puts function to write out the string to stdout.
166 call i32 @puts(i8* %cast210)
171 !1 = metadata !{i32 42}
174 This example is made up of a :ref:`global variable <globalvars>` named
175 "``.str``", an external declaration of the "``puts``" function, a
176 :ref:`function definition <functionstructure>` for "``main``" and
177 :ref:`named metadata <namedmetadatastructure>` "``foo``".
179 In general, a module is made up of a list of global values (where both
180 functions and global variables are global values). Global values are
181 represented by a pointer to a memory location (in this case, a pointer
182 to an array of char, and a pointer to a function), and have one of the
183 following :ref:`linkage types <linkage>`.
190 All Global Variables and Functions have one of the following types of
194 Global values with "``private``" linkage are only directly
195 accessible by objects in the current module. In particular, linking
196 code into a module with an private global value may cause the
197 private to be renamed as necessary to avoid collisions. Because the
198 symbol is private to the module, all references can be updated. This
199 doesn't show up in any symbol table in the object file.
201 Similar to ``private``, but the symbol is passed through the
202 assembler and evaluated by the linker. Unlike normal strong symbols,
203 they are removed by the linker from the final linked image
204 (executable or dynamic library).
205 ``linker_private_weak``
206 Similar to "``linker_private``", but the symbol is weak. Note that
207 ``linker_private_weak`` symbols are subject to coalescing by the
208 linker. The symbols are removed by the linker from the final linked
209 image (executable or dynamic library).
211 Similar to private, but the value shows as a local symbol
212 (``STB_LOCAL`` in the case of ELF) in the object file. This
213 corresponds to the notion of the '``static``' keyword in C.
214 ``available_externally``
215 Globals with "``available_externally``" linkage are never emitted
216 into the object file corresponding to the LLVM module. They exist to
217 allow inlining and other optimizations to take place given knowledge
218 of the definition of the global, which is known to be somewhere
219 outside the module. Globals with ``available_externally`` linkage
220 are allowed to be discarded at will, and are otherwise the same as
221 ``linkonce_odr``. This linkage type is only allowed on definitions,
224 Globals with "``linkonce``" linkage are merged with other globals of
225 the same name when linkage occurs. This can be used to implement
226 some forms of inline functions, templates, or other code which must
227 be generated in each translation unit that uses it, but where the
228 body may be overridden with a more definitive definition later.
229 Unreferenced ``linkonce`` globals are allowed to be discarded. Note
230 that ``linkonce`` linkage does not actually allow the optimizer to
231 inline the body of this function into callers because it doesn't
232 know if this definition of the function is the definitive definition
233 within the program or whether it will be overridden by a stronger
234 definition. To enable inlining and other optimizations, use
235 "``linkonce_odr``" linkage.
237 "``weak``" linkage has the same merging semantics as ``linkonce``
238 linkage, except that unreferenced globals with ``weak`` linkage may
239 not be discarded. This is used for globals that are declared "weak"
242 "``common``" linkage is most similar to "``weak``" linkage, but they
243 are used for tentative definitions in C, such as "``int X;``" at
244 global scope. Symbols with "``common``" linkage are merged in the
245 same way as ``weak symbols``, and they may not be deleted if
246 unreferenced. ``common`` symbols may not have an explicit section,
247 must have a zero initializer, and may not be marked
248 ':ref:`constant <globalvars>`'. Functions and aliases may not have
251 .. _linkage_appending:
254 "``appending``" linkage may only be applied to global variables of
255 pointer to array type. When two global variables with appending
256 linkage are linked together, the two global arrays are appended
257 together. This is the LLVM, typesafe, equivalent of having the
258 system linker append together "sections" with identical names when
261 The semantics of this linkage follow the ELF object file model: the
262 symbol is weak until linked, if not linked, the symbol becomes null
263 instead of being an undefined reference.
264 ``linkonce_odr``, ``weak_odr``
265 Some languages allow differing globals to be merged, such as two
266 functions with different semantics. Other languages, such as
267 ``C++``, ensure that only equivalent globals are ever merged (the
268 "one definition rule" --- "ODR"). Such languages can use the
269 ``linkonce_odr`` and ``weak_odr`` linkage types to indicate that the
270 global will only be merged with equivalent globals. These linkage
271 types are otherwise the same as their non-``odr`` versions.
273 If none of the above identifiers are used, the global is externally
274 visible, meaning that it participates in linkage and can be used to
275 resolve external symbol references.
277 It is illegal for a function *declaration* to have any linkage type
278 other than ``external`` or ``extern_weak``.
285 LLVM :ref:`functions <functionstructure>`, :ref:`calls <i_call>` and
286 :ref:`invokes <i_invoke>` can all have an optional calling convention
287 specified for the call. The calling convention of any pair of dynamic
288 caller/callee must match, or the behavior of the program is undefined.
289 The following calling conventions are supported by LLVM, and more may be
292 "``ccc``" - The C calling convention
293 This calling convention (the default if no other calling convention
294 is specified) matches the target C calling conventions. This calling
295 convention supports varargs function calls and tolerates some
296 mismatch in the declared prototype and implemented declaration of
297 the function (as does normal C).
298 "``fastcc``" - The fast calling convention
299 This calling convention attempts to make calls as fast as possible
300 (e.g. by passing things in registers). This calling convention
301 allows the target to use whatever tricks it wants to produce fast
302 code for the target, without having to conform to an externally
303 specified ABI (Application Binary Interface). `Tail calls can only
304 be optimized when this, the GHC or the HiPE convention is
305 used. <CodeGenerator.html#id80>`_ This calling convention does not
306 support varargs and requires the prototype of all callees to exactly
307 match the prototype of the function definition.
308 "``coldcc``" - The cold calling convention
309 This calling convention attempts to make code in the caller as
310 efficient as possible under the assumption that the call is not
311 commonly executed. As such, these calls often preserve all registers
312 so that the call does not break any live ranges in the caller side.
313 This calling convention does not support varargs and requires the
314 prototype of all callees to exactly match the prototype of the
315 function definition. Furthermore the inliner doesn't consider such function
317 "``cc 10``" - GHC convention
318 This calling convention has been implemented specifically for use by
319 the `Glasgow Haskell Compiler (GHC) <http://www.haskell.org/ghc>`_.
320 It passes everything in registers, going to extremes to achieve this
321 by disabling callee save registers. This calling convention should
322 not be used lightly but only for specific situations such as an
323 alternative to the *register pinning* performance technique often
324 used when implementing functional programming languages. At the
325 moment only X86 supports this convention and it has the following
328 - On *X86-32* only supports up to 4 bit type parameters. No
329 floating point types are supported.
330 - On *X86-64* only supports up to 10 bit type parameters and 6
331 floating point parameters.
333 This calling convention supports `tail call
334 optimization <CodeGenerator.html#id80>`_ but requires both the
335 caller and callee are using it.
336 "``cc 11``" - The HiPE calling convention
337 This calling convention has been implemented specifically for use by
338 the `High-Performance Erlang
339 (HiPE) <http://www.it.uu.se/research/group/hipe/>`_ compiler, *the*
340 native code compiler of the `Ericsson's Open Source Erlang/OTP
341 system <http://www.erlang.org/download.shtml>`_. It uses more
342 registers for argument passing than the ordinary C calling
343 convention and defines no callee-saved registers. The calling
344 convention properly supports `tail call
345 optimization <CodeGenerator.html#id80>`_ but requires that both the
346 caller and the callee use it. It uses a *register pinning*
347 mechanism, similar to GHC's convention, for keeping frequently
348 accessed runtime components pinned to specific hardware registers.
349 At the moment only X86 supports this convention (both 32 and 64
351 "``webkit_jscc``" - WebKit's JavaScript calling convention
352 This calling convention has been implemented for `WebKit FTL JIT
353 <https://trac.webkit.org/wiki/FTLJIT>`_. It passes arguments on the
354 stack right to left (as cdecl does), and returns a value in the
355 platform's customary return register.
356 "``anyregcc``" - Dynamic calling convention for code patching
357 This is a special convention that supports patching an arbitrary code
358 sequence in place of a call site. This convention forces the call
359 arguments into registers but allows them to be dynamcially
360 allocated. This can currently only be used with calls to
361 llvm.experimental.patchpoint because only this intrinsic records
362 the location of its arguments in a side table. See :doc:`StackMaps`.
363 "``preserve_mostcc``" - The `PreserveMost` calling convention
364 This calling convention attempts to make the code in the caller as little
365 intrusive as possible. This calling convention behaves identical to the `C`
366 calling convention on how arguments and return values are passed, but it
367 uses a different set of caller/callee-saved registers. This alleviates the
368 burden of saving and recovering a large register set before and after the
369 call in the caller. If the arguments are passed in callee-saved registers,
370 then they will be preserved by the callee across the call. This doesn't
371 apply for values returned in callee-saved registers.
373 - On X86-64 the callee preserves all general purpose registers, except for
374 R11. R11 can be used as a scratch register. Floating-point registers
375 (XMMs/YMMs) are not preserved and need to be saved by the caller.
377 The idea behind this convention is to support calls to runtime functions
378 that have a hot path and a cold path. The hot path is usually a small piece
379 of code that doesn't many registers. The cold path might need to call out to
380 another function and therefore only needs to preserve the caller-saved
381 registers, which haven't already been saved by the caller. The
382 `PreserveMost` calling convention is very similar to the `cold` calling
383 convention in terms of caller/callee-saved registers, but they are used for
384 different types of function calls. `coldcc` is for function calls that are
385 rarely executed, whereas `preserve_mostcc` function calls are intended to be
386 on the hot path and definitely executed a lot. Furthermore `preserve_mostcc`
387 doesn't prevent the inliner from inlining the function call.
389 This calling convention will be used by a future version of the ObjectiveC
390 runtime and should therefore still be considered experimental at this time.
391 Although this convention was created to optimize certain runtime calls to
392 the ObjectiveC runtime, it is not limited to this runtime and might be used
393 by other runtimes in the future too. The current implementation only
394 supports X86-64, but the intention is to support more architectures in the
396 "``preserve_allcc``" - The `PreserveAll` calling convention
397 This calling convention attempts to make the code in the caller even less
398 intrusive than the `PreserveMost` calling convention. This calling
399 convention also behaves identical to the `C` calling convention on how
400 arguments and return values are passed, but it uses a different set of
401 caller/callee-saved registers. This removes the burden of saving and
402 recovering a large register set before and after the call in the caller. If
403 the arguments are passed in callee-saved registers, then they will be
404 preserved by the callee across the call. This doesn't apply for values
405 returned in callee-saved registers.
407 - On X86-64 the callee preserves all general purpose registers, except for
408 R11. R11 can be used as a scratch register. Furthermore it also preserves
409 all floating-point registers (XMMs/YMMs).
411 The idea behind this convention is to support calls to runtime functions
412 that don't need to call out to any other functions.
414 This calling convention, like the `PreserveMost` calling convention, will be
415 used by a future version of the ObjectiveC runtime and should be considered
416 experimental at this time.
417 "``cc <n>``" - Numbered convention
418 Any calling convention may be specified by number, allowing
419 target-specific calling conventions to be used. Target specific
420 calling conventions start at 64.
422 More calling conventions can be added/defined on an as-needed basis, to
423 support Pascal conventions or any other well-known target-independent
426 .. _visibilitystyles:
431 All Global Variables and Functions have one of the following visibility
434 "``default``" - Default style
435 On targets that use the ELF object file format, default visibility
436 means that the declaration is visible to other modules and, in
437 shared libraries, means that the declared entity may be overridden.
438 On Darwin, default visibility means that the declaration is visible
439 to other modules. Default visibility corresponds to "external
440 linkage" in the language.
441 "``hidden``" - Hidden style
442 Two declarations of an object with hidden visibility refer to the
443 same object if they are in the same shared object. Usually, hidden
444 visibility indicates that the symbol will not be placed into the
445 dynamic symbol table, so no other module (executable or shared
446 library) can reference it directly.
447 "``protected``" - Protected style
448 On ELF, protected visibility indicates that the symbol will be
449 placed in the dynamic symbol table, but that references within the
450 defining module will bind to the local symbol. That is, the symbol
451 cannot be overridden by another module.
458 All Global Variables, Functions and Aliases can have one of the following
462 "``dllimport``" causes the compiler to reference a function or variable via
463 a global pointer to a pointer that is set up by the DLL exporting the
464 symbol. On Microsoft Windows targets, the pointer name is formed by
465 combining ``__imp_`` and the function or variable name.
467 "``dllexport``" causes the compiler to provide a global pointer to a pointer
468 in a DLL, so that it can be referenced with the ``dllimport`` attribute. On
469 Microsoft Windows targets, the pointer name is formed by combining
470 ``__imp_`` and the function or variable name. Since this storage class
471 exists for defining a dll interface, the compiler, assembler and linker know
472 it is externally referenced and must refrain from deleting the symbol.
477 LLVM IR allows you to specify both "identified" and "literal" :ref:`structure
478 types <t_struct>`. Literal types are uniqued structurally, but identified types
479 are never uniqued. An :ref:`opaque structural type <t_opaque>` can also be used
480 to forward declare a type which is not yet available.
482 An example of a identified structure specification is:
486 %mytype = type { %mytype*, i32 }
488 Prior to the LLVM 3.0 release, identified types were structurally uniqued. Only
489 literal types are uniqued in recent versions of LLVM.
496 Global variables define regions of memory allocated at compilation time
499 Global variables definitions must be initialized, may have an explicit section
500 to be placed in, and may have an optional explicit alignment specified.
502 Global variables in other translation units can also be declared, in which
503 case they don't have an initializer.
505 A variable may be defined as ``thread_local``, which means that it will
506 not be shared by threads (each thread will have a separated copy of the
507 variable). Not all targets support thread-local variables. Optionally, a
508 TLS model may be specified:
511 For variables that are only used within the current shared library.
513 For variables in modules that will not be loaded dynamically.
515 For variables defined in the executable and only used within it.
517 The models correspond to the ELF TLS models; see `ELF Handling For
518 Thread-Local Storage <http://people.redhat.com/drepper/tls.pdf>`_ for
519 more information on under which circumstances the different models may
520 be used. The target may choose a different TLS model if the specified
521 model is not supported, or if a better choice of model can be made.
523 A variable may be defined as a global ``constant``, which indicates that
524 the contents of the variable will **never** be modified (enabling better
525 optimization, allowing the global data to be placed in the read-only
526 section of an executable, etc). Note that variables that need runtime
527 initialization cannot be marked ``constant`` as there is a store to the
530 LLVM explicitly allows *declarations* of global variables to be marked
531 constant, even if the final definition of the global is not. This
532 capability can be used to enable slightly better optimization of the
533 program, but requires the language definition to guarantee that
534 optimizations based on the 'constantness' are valid for the translation
535 units that do not include the definition.
537 As SSA values, global variables define pointer values that are in scope
538 (i.e. they dominate) all basic blocks in the program. Global variables
539 always define a pointer to their "content" type because they describe a
540 region of memory, and all memory objects in LLVM are accessed through
543 Global variables can be marked with ``unnamed_addr`` which indicates
544 that the address is not significant, only the content. Constants marked
545 like this can be merged with other constants if they have the same
546 initializer. Note that a constant with significant address *can* be
547 merged with a ``unnamed_addr`` constant, the result being a constant
548 whose address is significant.
550 A global variable may be declared to reside in a target-specific
551 numbered address space. For targets that support them, address spaces
552 may affect how optimizations are performed and/or what target
553 instructions are used to access the variable. The default address space
554 is zero. The address space qualifier must precede any other attributes.
556 LLVM allows an explicit section to be specified for globals. If the
557 target supports it, it will emit globals to the section specified.
559 By default, global initializers are optimized by assuming that global
560 variables defined within the module are not modified from their
561 initial values before the start of the global initializer. This is
562 true even for variables potentially accessible from outside the
563 module, including those with external linkage or appearing in
564 ``@llvm.used`` or dllexported variables. This assumption may be suppressed
565 by marking the variable with ``externally_initialized``.
567 An explicit alignment may be specified for a global, which must be a
568 power of 2. If not present, or if the alignment is set to zero, the
569 alignment of the global is set by the target to whatever it feels
570 convenient. If an explicit alignment is specified, the global is forced
571 to have exactly that alignment. Targets and optimizers are not allowed
572 to over-align the global if the global has an assigned section. In this
573 case, the extra alignment could be observable: for example, code could
574 assume that the globals are densely packed in their section and try to
575 iterate over them as an array, alignment padding would break this
578 Globals can also have a :ref:`DLL storage class <dllstorageclass>`.
582 [@<GlobalVarName> =] [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal]
583 [AddrSpace] [unnamed_addr] [ExternallyInitialized]
584 <global | constant> <Type>
585 [, section "name"] [, align <Alignment>]
587 For example, the following defines a global in a numbered address space
588 with an initializer, section, and alignment:
592 @G = addrspace(5) constant float 1.0, section "foo", align 4
594 The following example just declares a global variable
598 @G = external global i32
600 The following example defines a thread-local global with the
601 ``initialexec`` TLS model:
605 @G = thread_local(initialexec) global i32 0, align 4
607 .. _functionstructure:
612 LLVM function definitions consist of the "``define``" keyword, an
613 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
614 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
615 an optional :ref:`calling convention <callingconv>`,
616 an optional ``unnamed_addr`` attribute, a return type, an optional
617 :ref:`parameter attribute <paramattrs>` for the return type, a function
618 name, a (possibly empty) argument list (each with optional :ref:`parameter
619 attributes <paramattrs>`), optional :ref:`function attributes <fnattrs>`,
620 an optional section, an optional alignment, an optional :ref:`garbage
621 collector name <gc>`, an optional :ref:`prefix <prefixdata>`, an opening
622 curly brace, a list of basic blocks, and a closing curly brace.
624 LLVM function declarations consist of the "``declare``" keyword, an
625 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
626 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
627 an optional :ref:`calling convention <callingconv>`,
628 an optional ``unnamed_addr`` attribute, a return type, an optional
629 :ref:`parameter attribute <paramattrs>` for the return type, a function
630 name, a possibly empty list of arguments, an optional alignment, an optional
631 :ref:`garbage collector name <gc>` and an optional :ref:`prefix <prefixdata>`.
633 A function definition contains a list of basic blocks, forming the CFG (Control
634 Flow Graph) for the function. Each basic block may optionally start with a label
635 (giving the basic block a symbol table entry), contains a list of instructions,
636 and ends with a :ref:`terminator <terminators>` instruction (such as a branch or
637 function return). If an explicit label is not provided, a block is assigned an
638 implicit numbered label, using the next value from the same counter as used for
639 unnamed temporaries (:ref:`see above<identifiers>`). For example, if a function
640 entry block does not have an explicit label, it will be assigned label "%0",
641 then the first unnamed temporary in that block will be "%1", etc.
643 The first basic block in a function is special in two ways: it is
644 immediately executed on entrance to the function, and it is not allowed
645 to have predecessor basic blocks (i.e. there can not be any branches to
646 the entry block of a function). Because the block can have no
647 predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`.
649 LLVM allows an explicit section to be specified for functions. If the
650 target supports it, it will emit functions to the section specified.
652 An explicit alignment may be specified for a function. If not present,
653 or if the alignment is set to zero, the alignment of the function is set
654 by the target to whatever it feels convenient. If an explicit alignment
655 is specified, the function is forced to have at least that much
656 alignment. All alignments must be a power of 2.
658 If the ``unnamed_addr`` attribute is given, the address is know to not
659 be significant and two identical functions can be merged.
663 define [linkage] [visibility] [DLLStorageClass]
665 <ResultType> @<FunctionName> ([argument list])
666 [fn Attrs] [section "name"] [align N]
667 [gc] [prefix Constant] { ... }
674 Aliases act as "second name" for the aliasee value (which can be either
675 function, global variable, another alias or bitcast of global value).
676 Aliases may have an optional :ref:`linkage type <linkage>`, an optional
677 :ref:`visibility style <visibility>`, and an optional :ref:`DLL storage class
682 @<Name> = [Visibility] [DLLStorageClass] alias [Linkage] <AliaseeTy> @<Aliasee>
684 The linkage must be one of ``private``, ``linker_private``,
685 ``linker_private_weak``, ``internal``, ``linkonce``, ``weak``,
686 ``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers
687 might not correctly handle dropping a weak symbol that is aliased by a non-weak
690 .. _namedmetadatastructure:
695 Named metadata is a collection of metadata. :ref:`Metadata
696 nodes <metadata>` (but not metadata strings) are the only valid
697 operands for a named metadata.
701 ; Some unnamed metadata nodes, which are referenced by the named metadata.
702 !0 = metadata !{metadata !"zero"}
703 !1 = metadata !{metadata !"one"}
704 !2 = metadata !{metadata !"two"}
706 !name = !{!0, !1, !2}
713 The return type and each parameter of a function type may have a set of
714 *parameter attributes* associated with them. Parameter attributes are
715 used to communicate additional information about the result or
716 parameters of a function. Parameter attributes are considered to be part
717 of the function, not of the function type, so functions with different
718 parameter attributes can have the same function type.
720 Parameter attributes are simple keywords that follow the type specified.
721 If multiple parameter attributes are needed, they are space separated.
726 declare i32 @printf(i8* noalias nocapture, ...)
727 declare i32 @atoi(i8 zeroext)
728 declare signext i8 @returns_signed_char()
730 Note that any attributes for the function result (``nounwind``,
731 ``readonly``) come immediately after the argument list.
733 Currently, only the following parameter attributes are defined:
736 This indicates to the code generator that the parameter or return
737 value should be zero-extended to the extent required by the target's
738 ABI (which is usually 32-bits, but is 8-bits for a i1 on x86-64) by
739 the caller (for a parameter) or the callee (for a return value).
741 This indicates to the code generator that the parameter or return
742 value should be sign-extended to the extent required by the target's
743 ABI (which is usually 32-bits) by the caller (for a parameter) or
744 the callee (for a return value).
746 This indicates that this parameter or return value should be treated
747 in a special target-dependent fashion during while emitting code for
748 a function call or return (usually, by putting it in a register as
749 opposed to memory, though some targets use it to distinguish between
750 two different kinds of registers). Use of this attribute is
753 This indicates that the pointer parameter should really be passed by
754 value to the function. The attribute implies that a hidden copy of
755 the pointee is made between the caller and the callee, so the callee
756 is unable to modify the value in the caller. This attribute is only
757 valid on LLVM pointer arguments. It is generally used to pass
758 structs and arrays by value, but is also valid on pointers to
759 scalars. The copy is considered to belong to the caller not the
760 callee (for example, ``readonly`` functions should not write to
761 ``byval`` parameters). This is not a valid attribute for return
764 The byval attribute also supports specifying an alignment with the
765 align attribute. It indicates the alignment of the stack slot to
766 form and the known alignment of the pointer specified to the call
767 site. If the alignment is not specified, then the code generator
768 makes a target-specific assumption.
774 .. Warning:: This feature is unstable and not fully implemented.
776 The ``inalloca`` argument attribute allows the caller to take the
777 address of outgoing stack arguments. An ``inalloca`` argument must
778 be a pointer to stack memory produced by an ``alloca`` instruction.
779 The alloca, or argument allocation, must also be tagged with the
780 inalloca keyword. Only the past argument may have the ``inalloca``
781 attribute, and that argument is guaranteed to be passed in memory.
783 An argument allocation may be used by a call at most once because
784 the call may deallocate it. The ``inalloca`` attribute cannot be
785 used in conjunction with other attributes that affect argument
786 storage, like ``inreg``, ``nest``, ``sret``, or ``byval``. The
787 ``inalloca`` attribute also disables LLVM's implicit lowering of
788 large aggregate return values, which means that frontend authors
789 must lower them with ``sret`` pointers.
791 When the call site is reached, the argument allocation must have
792 been the most recent stack allocation that is still live, or the
793 results are undefined. It is possible to allocate additional stack
794 space after an argument allocation and before its call site, but it
795 must be cleared off with :ref:`llvm.stackrestore
798 See :doc:`InAlloca` for more information on how to use this
802 This indicates that the pointer parameter specifies the address of a
803 structure that is the return value of the function in the source
804 program. This pointer must be guaranteed by the caller to be valid:
805 loads and stores to the structure may be assumed by the callee
806 not to trap and to be properly aligned. This may only be applied to
807 the first parameter. This is not a valid attribute for return
810 This indicates that pointer values :ref:`based <pointeraliasing>` on
811 the argument or return value do not alias pointer values which are
812 not *based* on it, ignoring certain "irrelevant" dependencies. For a
813 call to the parent function, dependencies between memory references
814 from before or after the call and from those during the call are
815 "irrelevant" to the ``noalias`` keyword for the arguments and return
816 value used in that call. The caller shares the responsibility with
817 the callee for ensuring that these requirements are met. For further
818 details, please see the discussion of the NoAlias response in `alias
819 analysis <AliasAnalysis.html#MustMayNo>`_.
821 Note that this definition of ``noalias`` is intentionally similar
822 to the definition of ``restrict`` in C99 for function arguments,
823 though it is slightly weaker.
825 For function return values, C99's ``restrict`` is not meaningful,
826 while LLVM's ``noalias`` is.
828 This indicates that the callee does not make any copies of the
829 pointer that outlive the callee itself. This is not a valid
830 attribute for return values.
835 This indicates that the pointer parameter can be excised using the
836 :ref:`trampoline intrinsics <int_trampoline>`. This is not a valid
837 attribute for return values and can only be applied to one parameter.
840 This indicates that the function always returns the argument as its return
841 value. This is an optimization hint to the code generator when generating
842 the caller, allowing tail call optimization and omission of register saves
843 and restores in some cases; it is not checked or enforced when generating
844 the callee. The parameter and the function return type must be valid
845 operands for the :ref:`bitcast instruction <i_bitcast>`. This is not a
846 valid attribute for return values and can only be applied to one parameter.
850 Garbage Collector Names
851 -----------------------
853 Each function may specify a garbage collector name, which is simply a
858 define void @f() gc "name" { ... }
860 The compiler declares the supported values of *name*. Specifying a
861 collector which will cause the compiler to alter its output in order to
862 support the named garbage collection algorithm.
869 Prefix data is data associated with a function which the code generator
870 will emit immediately before the function body. The purpose of this feature
871 is to allow frontends to associate language-specific runtime metadata with
872 specific functions and make it available through the function pointer while
873 still allowing the function pointer to be called. To access the data for a
874 given function, a program may bitcast the function pointer to a pointer to
875 the constant's type. This implies that the IR symbol points to the start
878 To maintain the semantics of ordinary function calls, the prefix data must
879 have a particular format. Specifically, it must begin with a sequence of
880 bytes which decode to a sequence of machine instructions, valid for the
881 module's target, which transfer control to the point immediately succeeding
882 the prefix data, without performing any other visible action. This allows
883 the inliner and other passes to reason about the semantics of the function
884 definition without needing to reason about the prefix data. Obviously this
885 makes the format of the prefix data highly target dependent.
887 Prefix data is laid out as if it were an initializer for a global variable
888 of the prefix data's type. No padding is automatically placed between the
889 prefix data and the function body. If padding is required, it must be part
892 A trivial example of valid prefix data for the x86 architecture is ``i8 144``,
893 which encodes the ``nop`` instruction:
897 define void @f() prefix i8 144 { ... }
899 Generally prefix data can be formed by encoding a relative branch instruction
900 which skips the metadata, as in this example of valid prefix data for the
901 x86_64 architecture, where the first two bytes encode ``jmp .+10``:
905 %0 = type <{ i8, i8, i8* }>
907 define void @f() prefix %0 <{ i8 235, i8 8, i8* @md}> { ... }
909 A function may have prefix data but no body. This has similar semantics
910 to the ``available_externally`` linkage in that the data may be used by the
911 optimizers but will not be emitted in the object file.
918 Attribute groups are groups of attributes that are referenced by objects within
919 the IR. They are important for keeping ``.ll`` files readable, because a lot of
920 functions will use the same set of attributes. In the degenerative case of a
921 ``.ll`` file that corresponds to a single ``.c`` file, the single attribute
922 group will capture the important command line flags used to build that file.
924 An attribute group is a module-level object. To use an attribute group, an
925 object references the attribute group's ID (e.g. ``#37``). An object may refer
926 to more than one attribute group. In that situation, the attributes from the
927 different groups are merged.
929 Here is an example of attribute groups for a function that should always be
930 inlined, has a stack alignment of 4, and which shouldn't use SSE instructions:
934 ; Target-independent attributes:
935 attributes #0 = { alwaysinline alignstack=4 }
937 ; Target-dependent attributes:
938 attributes #1 = { "no-sse" }
940 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse".
941 define void @f() #0 #1 { ... }
948 Function attributes are set to communicate additional information about
949 a function. Function attributes are considered to be part of the
950 function, not of the function type, so functions with different function
951 attributes can have the same function type.
953 Function attributes are simple keywords that follow the type specified.
954 If multiple attributes are needed, they are space separated. For
959 define void @f() noinline { ... }
960 define void @f() alwaysinline { ... }
961 define void @f() alwaysinline optsize { ... }
962 define void @f() optsize { ... }
965 This attribute indicates that, when emitting the prologue and
966 epilogue, the backend should forcibly align the stack pointer.
967 Specify the desired alignment, which must be a power of two, in
970 This attribute indicates that the inliner should attempt to inline
971 this function into callers whenever possible, ignoring any active
972 inlining size threshold for this caller.
974 This indicates that the callee function at a call site should be
975 recognized as a built-in function, even though the function's declaration
976 uses the ``nobuiltin`` attribute. This is only valid at call sites for
977 direct calls to functions which are declared with the ``nobuiltin``
980 This attribute indicates that this function is rarely called. When
981 computing edge weights, basic blocks post-dominated by a cold
982 function call are also considered to be cold; and, thus, given low
985 This attribute indicates that the source code contained a hint that
986 inlining this function is desirable (such as the "inline" keyword in
987 C/C++). It is just a hint; it imposes no requirements on the
990 This attribute suggests that optimization passes and code generator
991 passes make choices that keep the code size of this function as small
992 as possible and perform optimizations that may sacrifice runtime
993 performance in order to minimize the size of the generated code.
995 This attribute disables prologue / epilogue emission for the
996 function. This can have very system-specific consequences.
998 This indicates that the callee function at a call site is not recognized as
999 a built-in function. LLVM will retain the original call and not replace it
1000 with equivalent code based on the semantics of the built-in function, unless
1001 the call site uses the ``builtin`` attribute. This is valid at call sites
1002 and on function declarations and definitions.
1004 This attribute indicates that calls to the function cannot be
1005 duplicated. A call to a ``noduplicate`` function may be moved
1006 within its parent function, but may not be duplicated within
1007 its parent function.
1009 A function containing a ``noduplicate`` call may still
1010 be an inlining candidate, provided that the call is not
1011 duplicated by inlining. That implies that the function has
1012 internal linkage and only has one call site, so the original
1013 call is dead after inlining.
1015 This attributes disables implicit floating point instructions.
1017 This attribute indicates that the inliner should never inline this
1018 function in any situation. This attribute may not be used together
1019 with the ``alwaysinline`` attribute.
1021 This attribute suppresses lazy symbol binding for the function. This
1022 may make calls to the function faster, at the cost of extra program
1023 startup time if the function is not called during program startup.
1025 This attribute indicates that the code generator should not use a
1026 red zone, even if the target-specific ABI normally permits it.
1028 This function attribute indicates that the function never returns
1029 normally. This produces undefined behavior at runtime if the
1030 function ever does dynamically return.
1032 This function attribute indicates that the function never returns
1033 with an unwind or exceptional control flow. If the function does
1034 unwind, its runtime behavior is undefined.
1036 This function attribute indicates that the function is not optimized
1037 by any optimization or code generator passes with the
1038 exception of interprocedural optimization passes.
1039 This attribute cannot be used together with the ``alwaysinline``
1040 attribute; this attribute is also incompatible
1041 with the ``minsize`` attribute and the ``optsize`` attribute.
1043 This attribute requires the ``noinline`` attribute to be specified on
1044 the function as well, so the function is never inlined into any caller.
1045 Only functions with the ``alwaysinline`` attribute are valid
1046 candidates for inlining into the body of this function.
1048 This attribute suggests that optimization passes and code generator
1049 passes make choices that keep the code size of this function low,
1050 and otherwise do optimizations specifically to reduce code size as
1051 long as they do not significantly impact runtime performance.
1053 On a function, this attribute indicates that the function computes its
1054 result (or decides to unwind an exception) based strictly on its arguments,
1055 without dereferencing any pointer arguments or otherwise accessing
1056 any mutable state (e.g. memory, control registers, etc) visible to
1057 caller functions. It does not write through any pointer arguments
1058 (including ``byval`` arguments) and never changes any state visible
1059 to callers. This means that it cannot unwind exceptions by calling
1060 the ``C++`` exception throwing methods.
1062 On an argument, this attribute indicates that the function does not
1063 dereference that pointer argument, even though it may read or write the
1064 memory that the pointer points to if accessed through other pointers.
1066 On a function, this attribute indicates that the function does not write
1067 through any pointer arguments (including ``byval`` arguments) or otherwise
1068 modify any state (e.g. memory, control registers, etc) visible to
1069 caller functions. It may dereference pointer arguments and read
1070 state that may be set in the caller. A readonly function always
1071 returns the same value (or unwinds an exception identically) when
1072 called with the same set of arguments and global state. It cannot
1073 unwind an exception by calling the ``C++`` exception throwing
1076 On an argument, this attribute indicates that the function does not write
1077 through this pointer argument, even though it may write to the memory that
1078 the pointer points to.
1080 This attribute indicates that this function can return twice. The C
1081 ``setjmp`` is an example of such a function. The compiler disables
1082 some optimizations (like tail calls) in the caller of these
1084 ``sanitize_address``
1085 This attribute indicates that AddressSanitizer checks
1086 (dynamic address safety analysis) are enabled for this function.
1088 This attribute indicates that MemorySanitizer checks (dynamic detection
1089 of accesses to uninitialized memory) are enabled for this function.
1091 This attribute indicates that ThreadSanitizer checks
1092 (dynamic thread safety analysis) are enabled for this function.
1094 This attribute indicates that the function should emit a stack
1095 smashing protector. It is in the form of a "canary" --- a random value
1096 placed on the stack before the local variables that's checked upon
1097 return from the function to see if it has been overwritten. A
1098 heuristic is used to determine if a function needs stack protectors
1099 or not. The heuristic used will enable protectors for functions with:
1101 - Character arrays larger than ``ssp-buffer-size`` (default 8).
1102 - Aggregates containing character arrays larger than ``ssp-buffer-size``.
1103 - Calls to alloca() with variable sizes or constant sizes greater than
1104 ``ssp-buffer-size``.
1106 Variables that are identified as requiring a protector will be arranged
1107 on the stack such that they are adjacent to the stack protector guard.
1109 If a function that has an ``ssp`` attribute is inlined into a
1110 function that doesn't have an ``ssp`` attribute, then the resulting
1111 function will have an ``ssp`` attribute.
1113 This attribute indicates that the function should *always* emit a
1114 stack smashing protector. This overrides the ``ssp`` function
1117 Variables that are identified as requiring a protector will be arranged
1118 on the stack such that they are adjacent to the stack protector guard.
1119 The specific layout rules are:
1121 #. Large arrays and structures containing large arrays
1122 (``>= ssp-buffer-size``) are closest to the stack protector.
1123 #. Small arrays and structures containing small arrays
1124 (``< ssp-buffer-size``) are 2nd closest to the protector.
1125 #. Variables that have had their address taken are 3rd closest to the
1128 If a function that has an ``sspreq`` attribute is inlined into a
1129 function that doesn't have an ``sspreq`` attribute or which has an
1130 ``ssp`` or ``sspstrong`` attribute, then the resulting function will have
1131 an ``sspreq`` attribute.
1133 This attribute indicates that the function should emit a stack smashing
1134 protector. This attribute causes a strong heuristic to be used when
1135 determining if a function needs stack protectors. The strong heuristic
1136 will enable protectors for functions with:
1138 - Arrays of any size and type
1139 - Aggregates containing an array of any size and type.
1140 - Calls to alloca().
1141 - Local variables that have had their address taken.
1143 Variables that are identified as requiring a protector will be arranged
1144 on the stack such that they are adjacent to the stack protector guard.
1145 The specific layout rules are:
1147 #. Large arrays and structures containing large arrays
1148 (``>= ssp-buffer-size``) are closest to the stack protector.
1149 #. Small arrays and structures containing small arrays
1150 (``< ssp-buffer-size``) are 2nd closest to the protector.
1151 #. Variables that have had their address taken are 3rd closest to the
1154 This overrides the ``ssp`` function attribute.
1156 If a function that has an ``sspstrong`` attribute is inlined into a
1157 function that doesn't have an ``sspstrong`` attribute, then the
1158 resulting function will have an ``sspstrong`` attribute.
1160 This attribute indicates that the ABI being targeted requires that
1161 an unwind table entry be produce for this function even if we can
1162 show that no exceptions passes by it. This is normally the case for
1163 the ELF x86-64 abi, but it can be disabled for some compilation
1168 Module-Level Inline Assembly
1169 ----------------------------
1171 Modules may contain "module-level inline asm" blocks, which corresponds
1172 to the GCC "file scope inline asm" blocks. These blocks are internally
1173 concatenated by LLVM and treated as a single unit, but may be separated
1174 in the ``.ll`` file if desired. The syntax is very simple:
1176 .. code-block:: llvm
1178 module asm "inline asm code goes here"
1179 module asm "more can go here"
1181 The strings can contain any character by escaping non-printable
1182 characters. The escape sequence used is simply "\\xx" where "xx" is the
1183 two digit hex code for the number.
1185 The inline asm code is simply printed to the machine code .s file when
1186 assembly code is generated.
1188 .. _langref_datalayout:
1193 A module may specify a target specific data layout string that specifies
1194 how data is to be laid out in memory. The syntax for the data layout is
1197 .. code-block:: llvm
1199 target datalayout = "layout specification"
1201 The *layout specification* consists of a list of specifications
1202 separated by the minus sign character ('-'). Each specification starts
1203 with a letter and may include other information after the letter to
1204 define some aspect of the data layout. The specifications accepted are
1208 Specifies that the target lays out data in big-endian form. That is,
1209 the bits with the most significance have the lowest address
1212 Specifies that the target lays out data in little-endian form. That
1213 is, the bits with the least significance have the lowest address
1216 Specifies the natural alignment of the stack in bits. Alignment
1217 promotion of stack variables is limited to the natural stack
1218 alignment to avoid dynamic stack realignment. The stack alignment
1219 must be a multiple of 8-bits. If omitted, the natural stack
1220 alignment defaults to "unspecified", which does not prevent any
1221 alignment promotions.
1222 ``p[n]:<size>:<abi>:<pref>``
1223 This specifies the *size* of a pointer and its ``<abi>`` and
1224 ``<pref>``\erred alignments for address space ``n``. All sizes are in
1225 bits. The address space, ``n`` is optional, and if not specified,
1226 denotes the default address space 0. The value of ``n`` must be
1227 in the range [1,2^23).
1228 ``i<size>:<abi>:<pref>``
1229 This specifies the alignment for an integer type of a given bit
1230 ``<size>``. The value of ``<size>`` must be in the range [1,2^23).
1231 ``v<size>:<abi>:<pref>``
1232 This specifies the alignment for a vector type of a given bit
1234 ``f<size>:<abi>:<pref>``
1235 This specifies the alignment for a floating point type of a given bit
1236 ``<size>``. Only values of ``<size>`` that are supported by the target
1237 will work. 32 (float) and 64 (double) are supported on all targets; 80
1238 or 128 (different flavors of long double) are also supported on some
1241 This specifies the alignment for an object of aggregate type.
1243 If present, specifies that llvm names are mangled in the output. The
1246 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix.
1247 * ``m``: Mips mangling: Private symbols get a ``$`` prefix.
1248 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other
1249 symbols get a ``_`` prefix.
1250 * ``w``: Windows COFF prefix: Similar to Mach-O, but stdcall and fastcall
1251 functions also get a suffix based on the frame size.
1252 ``n<size1>:<size2>:<size3>...``
1253 This specifies a set of native integer widths for the target CPU in
1254 bits. For example, it might contain ``n32`` for 32-bit PowerPC,
1255 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of
1256 this set are considered to support most general arithmetic operations
1259 On every specification that takes a ``<abi>:<pref>``, specifying the
1260 ``<pref>`` alignment is optional. If omitted, the preceding ``:``
1261 should be omitted too and ``<pref>`` will be equal to ``<abi>``.
1263 When constructing the data layout for a given target, LLVM starts with a
1264 default set of specifications which are then (possibly) overridden by
1265 the specifications in the ``datalayout`` keyword. The default
1266 specifications are given in this list:
1268 - ``E`` - big endian
1269 - ``p:64:64:64`` - 64-bit pointers with 64-bit alignment.
1270 - ``p[n]:64:64:64`` - Other address spaces are assumed to be the
1271 same as the default address space.
1272 - ``S0`` - natural stack alignment is unspecified
1273 - ``i1:8:8`` - i1 is 8-bit (byte) aligned
1274 - ``i8:8:8`` - i8 is 8-bit (byte) aligned
1275 - ``i16:16:16`` - i16 is 16-bit aligned
1276 - ``i32:32:32`` - i32 is 32-bit aligned
1277 - ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred
1278 alignment of 64-bits
1279 - ``f16:16:16`` - half is 16-bit aligned
1280 - ``f32:32:32`` - float is 32-bit aligned
1281 - ``f64:64:64`` - double is 64-bit aligned
1282 - ``f128:128:128`` - quad is 128-bit aligned
1283 - ``v64:64:64`` - 64-bit vector is 64-bit aligned
1284 - ``v128:128:128`` - 128-bit vector is 128-bit aligned
1285 - ``a:0:64`` - aggregates are 64-bit aligned
1287 When LLVM is determining the alignment for a given type, it uses the
1290 #. If the type sought is an exact match for one of the specifications,
1291 that specification is used.
1292 #. If no match is found, and the type sought is an integer type, then
1293 the smallest integer type that is larger than the bitwidth of the
1294 sought type is used. If none of the specifications are larger than
1295 the bitwidth then the largest integer type is used. For example,
1296 given the default specifications above, the i7 type will use the
1297 alignment of i8 (next largest) while both i65 and i256 will use the
1298 alignment of i64 (largest specified).
1299 #. If no match is found, and the type sought is a vector type, then the
1300 largest vector type that is smaller than the sought vector type will
1301 be used as a fall back. This happens because <128 x double> can be
1302 implemented in terms of 64 <2 x double>, for example.
1304 The function of the data layout string may not be what you expect.
1305 Notably, this is not a specification from the frontend of what alignment
1306 the code generator should use.
1308 Instead, if specified, the target data layout is required to match what
1309 the ultimate *code generator* expects. This string is used by the
1310 mid-level optimizers to improve code, and this only works if it matches
1311 what the ultimate code generator uses. If you would like to generate IR
1312 that does not embed this target-specific detail into the IR, then you
1313 don't have to specify the string. This will disable some optimizations
1314 that require precise layout information, but this also prevents those
1315 optimizations from introducing target specificity into the IR.
1322 A module may specify a target triple string that describes the target
1323 host. The syntax for the target triple is simply:
1325 .. code-block:: llvm
1327 target triple = "x86_64-apple-macosx10.7.0"
1329 The *target triple* string consists of a series of identifiers delimited
1330 by the minus sign character ('-'). The canonical forms are:
1334 ARCHITECTURE-VENDOR-OPERATING_SYSTEM
1335 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT
1337 This information is passed along to the backend so that it generates
1338 code for the proper architecture. It's possible to override this on the
1339 command line with the ``-mtriple`` command line option.
1341 .. _pointeraliasing:
1343 Pointer Aliasing Rules
1344 ----------------------
1346 Any memory access must be done through a pointer value associated with
1347 an address range of the memory access, otherwise the behavior is
1348 undefined. Pointer values are associated with address ranges according
1349 to the following rules:
1351 - A pointer value is associated with the addresses associated with any
1352 value it is *based* on.
1353 - An address of a global variable is associated with the address range
1354 of the variable's storage.
1355 - The result value of an allocation instruction is associated with the
1356 address range of the allocated storage.
1357 - A null pointer in the default address-space is associated with no
1359 - An integer constant other than zero or a pointer value returned from
1360 a function not defined within LLVM may be associated with address
1361 ranges allocated through mechanisms other than those provided by
1362 LLVM. Such ranges shall not overlap with any ranges of addresses
1363 allocated by mechanisms provided by LLVM.
1365 A pointer value is *based* on another pointer value according to the
1368 - A pointer value formed from a ``getelementptr`` operation is *based*
1369 on the first operand of the ``getelementptr``.
1370 - The result value of a ``bitcast`` is *based* on the operand of the
1372 - A pointer value formed by an ``inttoptr`` is *based* on all pointer
1373 values that contribute (directly or indirectly) to the computation of
1374 the pointer's value.
1375 - The "*based* on" relationship is transitive.
1377 Note that this definition of *"based"* is intentionally similar to the
1378 definition of *"based"* in C99, though it is slightly weaker.
1380 LLVM IR does not associate types with memory. The result type of a
1381 ``load`` merely indicates the size and alignment of the memory from
1382 which to load, as well as the interpretation of the value. The first
1383 operand type of a ``store`` similarly only indicates the size and
1384 alignment of the store.
1386 Consequently, type-based alias analysis, aka TBAA, aka
1387 ``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR.
1388 :ref:`Metadata <metadata>` may be used to encode additional information
1389 which specialized optimization passes may use to implement type-based
1394 Volatile Memory Accesses
1395 ------------------------
1397 Certain memory accesses, such as :ref:`load <i_load>`'s,
1398 :ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be
1399 marked ``volatile``. The optimizers must not change the number of
1400 volatile operations or change their order of execution relative to other
1401 volatile operations. The optimizers *may* change the order of volatile
1402 operations relative to non-volatile operations. This is not Java's
1403 "volatile" and has no cross-thread synchronization behavior.
1405 IR-level volatile loads and stores cannot safely be optimized into
1406 llvm.memcpy or llvm.memmove intrinsics even when those intrinsics are
1407 flagged volatile. Likewise, the backend should never split or merge
1408 target-legal volatile load/store instructions.
1410 .. admonition:: Rationale
1412 Platforms may rely on volatile loads and stores of natively supported
1413 data width to be executed as single instruction. For example, in C
1414 this holds for an l-value of volatile primitive type with native
1415 hardware support, but not necessarily for aggregate types. The
1416 frontend upholds these expectations, which are intentionally
1417 unspecified in the IR. The rules above ensure that IR transformation
1418 do not violate the frontend's contract with the language.
1422 Memory Model for Concurrent Operations
1423 --------------------------------------
1425 The LLVM IR does not define any way to start parallel threads of
1426 execution or to register signal handlers. Nonetheless, there are
1427 platform-specific ways to create them, and we define LLVM IR's behavior
1428 in their presence. This model is inspired by the C++0x memory model.
1430 For a more informal introduction to this model, see the :doc:`Atomics`.
1432 We define a *happens-before* partial order as the least partial order
1435 - Is a superset of single-thread program order, and
1436 - When a *synchronizes-with* ``b``, includes an edge from ``a`` to
1437 ``b``. *Synchronizes-with* pairs are introduced by platform-specific
1438 techniques, like pthread locks, thread creation, thread joining,
1439 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering
1440 Constraints <ordering>`).
1442 Note that program order does not introduce *happens-before* edges
1443 between a thread and signals executing inside that thread.
1445 Every (defined) read operation (load instructions, memcpy, atomic
1446 loads/read-modify-writes, etc.) R reads a series of bytes written by
1447 (defined) write operations (store instructions, atomic
1448 stores/read-modify-writes, memcpy, etc.). For the purposes of this
1449 section, initialized globals are considered to have a write of the
1450 initializer which is atomic and happens before any other read or write
1451 of the memory in question. For each byte of a read R, R\ :sub:`byte`
1452 may see any write to the same byte, except:
1454 - If write\ :sub:`1` happens before write\ :sub:`2`, and
1455 write\ :sub:`2` happens before R\ :sub:`byte`, then
1456 R\ :sub:`byte` does not see write\ :sub:`1`.
1457 - If R\ :sub:`byte` happens before write\ :sub:`3`, then
1458 R\ :sub:`byte` does not see write\ :sub:`3`.
1460 Given that definition, R\ :sub:`byte` is defined as follows:
1462 - If R is volatile, the result is target-dependent. (Volatile is
1463 supposed to give guarantees which can support ``sig_atomic_t`` in
1464 C/C++, and may be used for accesses to addresses which do not behave
1465 like normal memory. It does not generally provide cross-thread
1467 - Otherwise, if there is no write to the same byte that happens before
1468 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte.
1469 - Otherwise, if R\ :sub:`byte` may see exactly one write,
1470 R\ :sub:`byte` returns the value written by that write.
1471 - Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may
1472 see are atomic, it chooses one of the values written. See the :ref:`Atomic
1473 Memory Ordering Constraints <ordering>` section for additional
1474 constraints on how the choice is made.
1475 - Otherwise R\ :sub:`byte` returns ``undef``.
1477 R returns the value composed of the series of bytes it read. This
1478 implies that some bytes within the value may be ``undef`` **without**
1479 the entire value being ``undef``. Note that this only defines the
1480 semantics of the operation; it doesn't mean that targets will emit more
1481 than one instruction to read the series of bytes.
1483 Note that in cases where none of the atomic intrinsics are used, this
1484 model places only one restriction on IR transformations on top of what
1485 is required for single-threaded execution: introducing a store to a byte
1486 which might not otherwise be stored is not allowed in general.
1487 (Specifically, in the case where another thread might write to and read
1488 from an address, introducing a store can change a load that may see
1489 exactly one write into a load that may see multiple writes.)
1493 Atomic Memory Ordering Constraints
1494 ----------------------------------
1496 Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`,
1497 :ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`,
1498 :ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take
1499 an ordering parameter that determines which other atomic instructions on
1500 the same address they *synchronize with*. These semantics are borrowed
1501 from Java and C++0x, but are somewhat more colloquial. If these
1502 descriptions aren't precise enough, check those specs (see spec
1503 references in the :doc:`atomics guide <Atomics>`).
1504 :ref:`fence <i_fence>` instructions treat these orderings somewhat
1505 differently since they don't take an address. See that instruction's
1506 documentation for details.
1508 For a simpler introduction to the ordering constraints, see the
1512 The set of values that can be read is governed by the happens-before
1513 partial order. A value cannot be read unless some operation wrote
1514 it. This is intended to provide a guarantee strong enough to model
1515 Java's non-volatile shared variables. This ordering cannot be
1516 specified for read-modify-write operations; it is not strong enough
1517 to make them atomic in any interesting way.
1519 In addition to the guarantees of ``unordered``, there is a single
1520 total order for modifications by ``monotonic`` operations on each
1521 address. All modification orders must be compatible with the
1522 happens-before order. There is no guarantee that the modification
1523 orders can be combined to a global total order for the whole program
1524 (and this often will not be possible). The read in an atomic
1525 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and
1526 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification
1527 order immediately before the value it writes. If one atomic read
1528 happens before another atomic read of the same address, the later
1529 read must see the same value or a later value in the address's
1530 modification order. This disallows reordering of ``monotonic`` (or
1531 stronger) operations on the same address. If an address is written
1532 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally
1533 read that address repeatedly, the other threads must eventually see
1534 the write. This corresponds to the C++0x/C1x
1535 ``memory_order_relaxed``.
1537 In addition to the guarantees of ``monotonic``, a
1538 *synchronizes-with* edge may be formed with a ``release`` operation.
1539 This is intended to model C++'s ``memory_order_acquire``.
1541 In addition to the guarantees of ``monotonic``, if this operation
1542 writes a value which is subsequently read by an ``acquire``
1543 operation, it *synchronizes-with* that operation. (This isn't a
1544 complete description; see the C++0x definition of a release
1545 sequence.) This corresponds to the C++0x/C1x
1546 ``memory_order_release``.
1547 ``acq_rel`` (acquire+release)
1548 Acts as both an ``acquire`` and ``release`` operation on its
1549 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``.
1550 ``seq_cst`` (sequentially consistent)
1551 In addition to the guarantees of ``acq_rel`` (``acquire`` for an
1552 operation which only reads, ``release`` for an operation which only
1553 writes), there is a global total order on all
1554 sequentially-consistent operations on all addresses, which is
1555 consistent with the *happens-before* partial order and with the
1556 modification orders of all the affected addresses. Each
1557 sequentially-consistent read sees the last preceding write to the
1558 same address in this global order. This corresponds to the C++0x/C1x
1559 ``memory_order_seq_cst`` and Java volatile.
1563 If an atomic operation is marked ``singlethread``, it only *synchronizes
1564 with* or participates in modification and seq\_cst total orderings with
1565 other operations running in the same thread (for example, in signal
1573 LLVM IR floating-point binary ops (:ref:`fadd <i_fadd>`,
1574 :ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`,
1575 :ref:`frem <i_frem>`) have the following flags that can set to enable
1576 otherwise unsafe floating point operations
1579 No NaNs - Allow optimizations to assume the arguments and result are not
1580 NaN. Such optimizations are required to retain defined behavior over
1581 NaNs, but the value of the result is undefined.
1584 No Infs - Allow optimizations to assume the arguments and result are not
1585 +/-Inf. Such optimizations are required to retain defined behavior over
1586 +/-Inf, but the value of the result is undefined.
1589 No Signed Zeros - Allow optimizations to treat the sign of a zero
1590 argument or result as insignificant.
1593 Allow Reciprocal - Allow optimizations to use the reciprocal of an
1594 argument rather than perform division.
1597 Fast - Allow algebraically equivalent transformations that may
1598 dramatically change results in floating point (e.g. reassociate). This
1599 flag implies all the others.
1606 The LLVM type system is one of the most important features of the
1607 intermediate representation. Being typed enables a number of
1608 optimizations to be performed on the intermediate representation
1609 directly, without having to do extra analyses on the side before the
1610 transformation. A strong type system makes it easier to read the
1611 generated code and enables novel analyses and transformations that are
1612 not feasible to perform on normal three address code representations.
1622 The void type does not represent any value and has no size.
1640 The function type can be thought of as a function signature. It consists of a
1641 return type and a list of formal parameter types. The return type of a function
1642 type is a void type or first class type --- except for :ref:`label <t_label>`
1643 and :ref:`metadata <t_metadata>` types.
1649 <returntype> (<parameter list>)
1651 ...where '``<parameter list>``' is a comma-separated list of type
1652 specifiers. Optionally, the parameter list may include a type ``...``, which
1653 indicates that the function takes a variable number of arguments. Variable
1654 argument functions can access their arguments with the :ref:`variable argument
1655 handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type
1656 except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`.
1660 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1661 | ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` |
1662 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1663 | ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. |
1664 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1665 | ``i32 (i8*, ...)`` | A vararg function that takes at least one :ref:`pointer <t_pointer>` to ``i8`` (char in C), which returns an integer. This is the signature for ``printf`` in LLVM. |
1666 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1667 | ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values |
1668 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1675 The :ref:`first class <t_firstclass>` types are perhaps the most important.
1676 Values of these types are the only ones which can be produced by
1684 These are the types that are valid in registers from CodeGen's perspective.
1693 The integer type is a very simple type that simply specifies an
1694 arbitrary bit width for the integer type desired. Any bit width from 1
1695 bit to 2\ :sup:`23`\ -1 (about 8 million) can be specified.
1703 The number of bits the integer will occupy is specified by the ``N``
1709 +----------------+------------------------------------------------+
1710 | ``i1`` | a single-bit integer. |
1711 +----------------+------------------------------------------------+
1712 | ``i32`` | a 32-bit integer. |
1713 +----------------+------------------------------------------------+
1714 | ``i1942652`` | a really big integer of over 1 million bits. |
1715 +----------------+------------------------------------------------+
1719 Floating Point Types
1720 """"""""""""""""""""
1729 - 16-bit floating point value
1732 - 32-bit floating point value
1735 - 64-bit floating point value
1738 - 128-bit floating point value (112-bit mantissa)
1741 - 80-bit floating point value (X87)
1744 - 128-bit floating point value (two 64-bits)
1753 The x86mmx type represents a value held in an MMX register on an x86
1754 machine. The operations allowed on it are quite limited: parameters and
1755 return values, load and store, and bitcast. User-specified MMX
1756 instructions are represented as intrinsic or asm calls with arguments
1757 and/or results of this type. There are no arrays, vectors or constants
1774 The pointer type is used to specify memory locations. Pointers are
1775 commonly used to reference objects in memory.
1777 Pointer types may have an optional address space attribute defining the
1778 numbered address space where the pointed-to object resides. The default
1779 address space is number zero. The semantics of non-zero address spaces
1780 are target-specific.
1782 Note that LLVM does not permit pointers to void (``void*``) nor does it
1783 permit pointers to labels (``label*``). Use ``i8*`` instead.
1793 +-------------------------+--------------------------------------------------------------------------------------------------------------+
1794 | ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. |
1795 +-------------------------+--------------------------------------------------------------------------------------------------------------+
1796 | ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. |
1797 +-------------------------+--------------------------------------------------------------------------------------------------------------+
1798 | ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space #5. |
1799 +-------------------------+--------------------------------------------------------------------------------------------------------------+
1808 A vector type is a simple derived type that represents a vector of
1809 elements. Vector types are used when multiple primitive data are
1810 operated in parallel using a single instruction (SIMD). A vector type
1811 requires a size (number of elements) and an underlying primitive data
1812 type. Vector types are considered :ref:`first class <t_firstclass>`.
1818 < <# elements> x <elementtype> >
1820 The number of elements is a constant integer value larger than 0;
1821 elementtype may be any integer or floating point type, or a pointer to
1822 these types. Vectors of size zero are not allowed.
1826 +-------------------+--------------------------------------------------+
1827 | ``<4 x i32>`` | Vector of 4 32-bit integer values. |
1828 +-------------------+--------------------------------------------------+
1829 | ``<8 x float>`` | Vector of 8 32-bit floating-point values. |
1830 +-------------------+--------------------------------------------------+
1831 | ``<2 x i64>`` | Vector of 2 64-bit integer values. |
1832 +-------------------+--------------------------------------------------+
1833 | ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. |
1834 +-------------------+--------------------------------------------------+
1843 The label type represents code labels.
1858 The metadata type represents embedded metadata. No derived types may be
1859 created from metadata except for :ref:`function <t_function>` arguments.
1872 Aggregate Types are a subset of derived types that can contain multiple
1873 member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are
1874 aggregate types. :ref:`Vectors <t_vector>` are not considered to be
1884 The array type is a very simple derived type that arranges elements
1885 sequentially in memory. The array type requires a size (number of
1886 elements) and an underlying data type.
1892 [<# elements> x <elementtype>]
1894 The number of elements is a constant integer value; ``elementtype`` may
1895 be any type with a size.
1899 +------------------+--------------------------------------+
1900 | ``[40 x i32]`` | Array of 40 32-bit integer values. |
1901 +------------------+--------------------------------------+
1902 | ``[41 x i32]`` | Array of 41 32-bit integer values. |
1903 +------------------+--------------------------------------+
1904 | ``[4 x i8]`` | Array of 4 8-bit integer values. |
1905 +------------------+--------------------------------------+
1907 Here are some examples of multidimensional arrays:
1909 +-----------------------------+----------------------------------------------------------+
1910 | ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. |
1911 +-----------------------------+----------------------------------------------------------+
1912 | ``[12 x [10 x float]]`` | 12x10 array of single precision floating point values. |
1913 +-----------------------------+----------------------------------------------------------+
1914 | ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. |
1915 +-----------------------------+----------------------------------------------------------+
1917 There is no restriction on indexing beyond the end of the array implied
1918 by a static type (though there are restrictions on indexing beyond the
1919 bounds of an allocated object in some cases). This means that
1920 single-dimension 'variable sized array' addressing can be implemented in
1921 LLVM with a zero length array type. An implementation of 'pascal style
1922 arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for
1932 The structure type is used to represent a collection of data members
1933 together in memory. The elements of a structure may be any type that has
1936 Structures in memory are accessed using '``load``' and '``store``' by
1937 getting a pointer to a field with the '``getelementptr``' instruction.
1938 Structures in registers are accessed using the '``extractvalue``' and
1939 '``insertvalue``' instructions.
1941 Structures may optionally be "packed" structures, which indicate that
1942 the alignment of the struct is one byte, and that there is no padding
1943 between the elements. In non-packed structs, padding between field types
1944 is inserted as defined by the DataLayout string in the module, which is
1945 required to match what the underlying code generator expects.
1947 Structures can either be "literal" or "identified". A literal structure
1948 is defined inline with other types (e.g. ``{i32, i32}*``) whereas
1949 identified types are always defined at the top level with a name.
1950 Literal types are uniqued by their contents and can never be recursive
1951 or opaque since there is no way to write one. Identified types can be
1952 recursive, can be opaqued, and are never uniqued.
1958 %T1 = type { <type list> } ; Identified normal struct type
1959 %T2 = type <{ <type list> }> ; Identified packed struct type
1963 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1964 | ``{ i32, i32, i32 }`` | A triple of three ``i32`` values |
1965 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1966 | ``{ float, i32 (i32) * }`` | A pair, where the first element is a ``float`` and the second element is a :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32``, returning an ``i32``. |
1967 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1968 | ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. |
1969 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1973 Opaque Structure Types
1974 """"""""""""""""""""""
1978 Opaque structure types are used to represent named structure types that
1979 do not have a body specified. This corresponds (for example) to the C
1980 notion of a forward declared structure.
1991 +--------------+-------------------+
1992 | ``opaque`` | An opaque type. |
1993 +--------------+-------------------+
1998 LLVM has several different basic types of constants. This section
1999 describes them all and their syntax.
2004 **Boolean constants**
2005 The two strings '``true``' and '``false``' are both valid constants
2007 **Integer constants**
2008 Standard integers (such as '4') are constants of the
2009 :ref:`integer <t_integer>` type. Negative numbers may be used with
2011 **Floating point constants**
2012 Floating point constants use standard decimal notation (e.g.
2013 123.421), exponential notation (e.g. 1.23421e+2), or a more precise
2014 hexadecimal notation (see below). The assembler requires the exact
2015 decimal value of a floating-point constant. For example, the
2016 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating
2017 decimal in binary. Floating point constants must have a :ref:`floating
2018 point <t_floating>` type.
2019 **Null pointer constants**
2020 The identifier '``null``' is recognized as a null pointer constant
2021 and must be of :ref:`pointer type <t_pointer>`.
2023 The one non-intuitive notation for constants is the hexadecimal form of
2024 floating point constants. For example, the form
2025 '``double 0x432ff973cafa8000``' is equivalent to (but harder to read
2026 than) '``double 4.5e+15``'. The only time hexadecimal floating point
2027 constants are required (and the only time that they are generated by the
2028 disassembler) is when a floating point constant must be emitted but it
2029 cannot be represented as a decimal floating point number in a reasonable
2030 number of digits. For example, NaN's, infinities, and other special
2031 values are represented in their IEEE hexadecimal format so that assembly
2032 and disassembly do not cause any bits to change in the constants.
2034 When using the hexadecimal form, constants of types half, float, and
2035 double are represented using the 16-digit form shown above (which
2036 matches the IEEE754 representation for double); half and float values
2037 must, however, be exactly representable as IEEE 754 half and single
2038 precision, respectively. Hexadecimal format is always used for long
2039 double, and there are three forms of long double. The 80-bit format used
2040 by x86 is represented as ``0xK`` followed by 20 hexadecimal digits. The
2041 128-bit format used by PowerPC (two adjacent doubles) is represented by
2042 ``0xM`` followed by 32 hexadecimal digits. The IEEE 128-bit format is
2043 represented by ``0xL`` followed by 32 hexadecimal digits. Long doubles
2044 will only work if they match the long double format on your target.
2045 The IEEE 16-bit format (half precision) is represented by ``0xH``
2046 followed by 4 hexadecimal digits. All hexadecimal formats are big-endian
2047 (sign bit at the left).
2049 There are no constants of type x86mmx.
2051 .. _complexconstants:
2056 Complex constants are a (potentially recursive) combination of simple
2057 constants and smaller complex constants.
2059 **Structure constants**
2060 Structure constants are represented with notation similar to
2061 structure type definitions (a comma separated list of elements,
2062 surrounded by braces (``{}``)). For example:
2063 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as
2064 "``@G = external global i32``". Structure constants must have
2065 :ref:`structure type <t_struct>`, and the number and types of elements
2066 must match those specified by the type.
2068 Array constants are represented with notation similar to array type
2069 definitions (a comma separated list of elements, surrounded by
2070 square brackets (``[]``)). For example:
2071 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have
2072 :ref:`array type <t_array>`, and the number and types of elements must
2073 match those specified by the type.
2074 **Vector constants**
2075 Vector constants are represented with notation similar to vector
2076 type definitions (a comma separated list of elements, surrounded by
2077 less-than/greater-than's (``<>``)). For example:
2078 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants
2079 must have :ref:`vector type <t_vector>`, and the number and types of
2080 elements must match those specified by the type.
2081 **Zero initialization**
2082 The string '``zeroinitializer``' can be used to zero initialize a
2083 value to zero of *any* type, including scalar and
2084 :ref:`aggregate <t_aggregate>` types. This is often used to avoid
2085 having to print large zero initializers (e.g. for large arrays) and
2086 is always exactly equivalent to using explicit zero initializers.
2088 A metadata node is a structure-like constant with :ref:`metadata
2089 type <t_metadata>`. For example:
2090 "``metadata !{ i32 0, metadata !"test" }``". Unlike other
2091 constants that are meant to be interpreted as part of the
2092 instruction stream, metadata is a place to attach additional
2093 information such as debug info.
2095 Global Variable and Function Addresses
2096 --------------------------------------
2098 The addresses of :ref:`global variables <globalvars>` and
2099 :ref:`functions <functionstructure>` are always implicitly valid
2100 (link-time) constants. These constants are explicitly referenced when
2101 the :ref:`identifier for the global <identifiers>` is used and always have
2102 :ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM
2105 .. code-block:: llvm
2109 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2116 The string '``undef``' can be used anywhere a constant is expected, and
2117 indicates that the user of the value may receive an unspecified
2118 bit-pattern. Undefined values may be of any type (other than '``label``'
2119 or '``void``') and be used anywhere a constant is permitted.
2121 Undefined values are useful because they indicate to the compiler that
2122 the program is well defined no matter what value is used. This gives the
2123 compiler more freedom to optimize. Here are some examples of
2124 (potentially surprising) transformations that are valid (in pseudo IR):
2126 .. code-block:: llvm
2136 This is safe because all of the output bits are affected by the undef
2137 bits. Any output bit can have a zero or one depending on the input bits.
2139 .. code-block:: llvm
2150 These logical operations have bits that are not always affected by the
2151 input. For example, if ``%X`` has a zero bit, then the output of the
2152 '``and``' operation will always be a zero for that bit, no matter what
2153 the corresponding bit from the '``undef``' is. As such, it is unsafe to
2154 optimize or assume that the result of the '``and``' is '``undef``'.
2155 However, it is safe to assume that all bits of the '``undef``' could be
2156 0, and optimize the '``and``' to 0. Likewise, it is safe to assume that
2157 all the bits of the '``undef``' operand to the '``or``' could be set,
2158 allowing the '``or``' to be folded to -1.
2160 .. code-block:: llvm
2162 %A = select undef, %X, %Y
2163 %B = select undef, 42, %Y
2164 %C = select %X, %Y, undef
2174 This set of examples shows that undefined '``select``' (and conditional
2175 branch) conditions can go *either way*, but they have to come from one
2176 of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were
2177 both known to have a clear low bit, then ``%A`` would have to have a
2178 cleared low bit. However, in the ``%C`` example, the optimizer is
2179 allowed to assume that the '``undef``' operand could be the same as
2180 ``%Y``, allowing the whole '``select``' to be eliminated.
2182 .. code-block:: llvm
2184 %A = xor undef, undef
2201 This example points out that two '``undef``' operands are not
2202 necessarily the same. This can be surprising to people (and also matches
2203 C semantics) where they assume that "``X^X``" is always zero, even if
2204 ``X`` is undefined. This isn't true for a number of reasons, but the
2205 short answer is that an '``undef``' "variable" can arbitrarily change
2206 its value over its "live range". This is true because the variable
2207 doesn't actually *have a live range*. Instead, the value is logically
2208 read from arbitrary registers that happen to be around when needed, so
2209 the value is not necessarily consistent over time. In fact, ``%A`` and
2210 ``%C`` need to have the same semantics or the core LLVM "replace all
2211 uses with" concept would not hold.
2213 .. code-block:: llvm
2221 These examples show the crucial difference between an *undefined value*
2222 and *undefined behavior*. An undefined value (like '``undef``') is
2223 allowed to have an arbitrary bit-pattern. This means that the ``%A``
2224 operation can be constant folded to '``undef``', because the '``undef``'
2225 could be an SNaN, and ``fdiv`` is not (currently) defined on SNaN's.
2226 However, in the second example, we can make a more aggressive
2227 assumption: because the ``undef`` is allowed to be an arbitrary value,
2228 we are allowed to assume that it could be zero. Since a divide by zero
2229 has *undefined behavior*, we are allowed to assume that the operation
2230 does not execute at all. This allows us to delete the divide and all
2231 code after it. Because the undefined operation "can't happen", the
2232 optimizer can assume that it occurs in dead code.
2234 .. code-block:: llvm
2236 a: store undef -> %X
2237 b: store %X -> undef
2242 These examples reiterate the ``fdiv`` example: a store *of* an undefined
2243 value can be assumed to not have any effect; we can assume that the
2244 value is overwritten with bits that happen to match what was already
2245 there. However, a store *to* an undefined location could clobber
2246 arbitrary memory, therefore, it has undefined behavior.
2253 Poison values are similar to :ref:`undef values <undefvalues>`, however
2254 they also represent the fact that an instruction or constant expression
2255 which cannot evoke side effects has nevertheless detected a condition
2256 which results in undefined behavior.
2258 There is currently no way of representing a poison value in the IR; they
2259 only exist when produced by operations such as :ref:`add <i_add>` with
2262 Poison value behavior is defined in terms of value *dependence*:
2264 - Values other than :ref:`phi <i_phi>` nodes depend on their operands.
2265 - :ref:`Phi <i_phi>` nodes depend on the operand corresponding to
2266 their dynamic predecessor basic block.
2267 - Function arguments depend on the corresponding actual argument values
2268 in the dynamic callers of their functions.
2269 - :ref:`Call <i_call>` instructions depend on the :ref:`ret <i_ret>`
2270 instructions that dynamically transfer control back to them.
2271 - :ref:`Invoke <i_invoke>` instructions depend on the
2272 :ref:`ret <i_ret>`, :ref:`resume <i_resume>`, or exception-throwing
2273 call instructions that dynamically transfer control back to them.
2274 - Non-volatile loads and stores depend on the most recent stores to all
2275 of the referenced memory addresses, following the order in the IR
2276 (including loads and stores implied by intrinsics such as
2277 :ref:`@llvm.memcpy <int_memcpy>`.)
2278 - An instruction with externally visible side effects depends on the
2279 most recent preceding instruction with externally visible side
2280 effects, following the order in the IR. (This includes :ref:`volatile
2281 operations <volatile>`.)
2282 - An instruction *control-depends* on a :ref:`terminator
2283 instruction <terminators>` if the terminator instruction has
2284 multiple successors and the instruction is always executed when
2285 control transfers to one of the successors, and may not be executed
2286 when control is transferred to another.
2287 - Additionally, an instruction also *control-depends* on a terminator
2288 instruction if the set of instructions it otherwise depends on would
2289 be different if the terminator had transferred control to a different
2291 - Dependence is transitive.
2293 Poison Values have the same behavior as :ref:`undef values <undefvalues>`,
2294 with the additional affect that any instruction which has a *dependence*
2295 on a poison value has undefined behavior.
2297 Here are some examples:
2299 .. code-block:: llvm
2302 %poison = sub nuw i32 0, 1 ; Results in a poison value.
2303 %still_poison = and i32 %poison, 0 ; 0, but also poison.
2304 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
2305 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
2307 store i32 %poison, i32* @g ; Poison value stored to memory.
2308 %poison2 = load i32* @g ; Poison value loaded back from memory.
2310 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
2312 %narrowaddr = bitcast i32* @g to i16*
2313 %wideaddr = bitcast i32* @g to i64*
2314 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2315 %poison4 = load i64* %wideaddr ; Returns a poison value.
2317 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2318 br i1 %cmp, label %true, label %end ; Branch to either destination.
2321 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2322 ; it has undefined behavior.
2326 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2327 ; Both edges into this PHI are
2328 ; control-dependent on %cmp, so this
2329 ; always results in a poison value.
2331 store volatile i32 0, i32* @g ; This would depend on the store in %true
2332 ; if %cmp is true, or the store in %entry
2333 ; otherwise, so this is undefined behavior.
2335 br i1 %cmp, label %second_true, label %second_end
2336 ; The same branch again, but this time the
2337 ; true block doesn't have side effects.
2344 store volatile i32 0, i32* @g ; This time, the instruction always depends
2345 ; on the store in %end. Also, it is
2346 ; control-equivalent to %end, so this is
2347 ; well-defined (ignoring earlier undefined
2348 ; behavior in this example).
2352 Addresses of Basic Blocks
2353 -------------------------
2355 ``blockaddress(@function, %block)``
2357 The '``blockaddress``' constant computes the address of the specified
2358 basic block in the specified function, and always has an ``i8*`` type.
2359 Taking the address of the entry block is illegal.
2361 This value only has defined behavior when used as an operand to the
2362 ':ref:`indirectbr <i_indirectbr>`' instruction, or for comparisons
2363 against null. Pointer equality tests between labels addresses results in
2364 undefined behavior --- though, again, comparison against null is ok, and
2365 no label is equal to the null pointer. This may be passed around as an
2366 opaque pointer sized value as long as the bits are not inspected. This
2367 allows ``ptrtoint`` and arithmetic to be performed on these values so
2368 long as the original value is reconstituted before the ``indirectbr``
2371 Finally, some targets may provide defined semantics when using the value
2372 as the operand to an inline assembly, but that is target specific.
2376 Constant Expressions
2377 --------------------
2379 Constant expressions are used to allow expressions involving other
2380 constants to be used as constants. Constant expressions may be of any
2381 :ref:`first class <t_firstclass>` type and may involve any LLVM operation
2382 that does not have side effects (e.g. load and call are not supported).
2383 The following is the syntax for constant expressions:
2385 ``trunc (CST to TYPE)``
2386 Truncate a constant to another type. The bit size of CST must be
2387 larger than the bit size of TYPE. Both types must be integers.
2388 ``zext (CST to TYPE)``
2389 Zero extend a constant to another type. The bit size of CST must be
2390 smaller than the bit size of TYPE. Both types must be integers.
2391 ``sext (CST to TYPE)``
2392 Sign extend a constant to another type. The bit size of CST must be
2393 smaller than the bit size of TYPE. Both types must be integers.
2394 ``fptrunc (CST to TYPE)``
2395 Truncate a floating point constant to another floating point type.
2396 The size of CST must be larger than the size of TYPE. Both types
2397 must be floating point.
2398 ``fpext (CST to TYPE)``
2399 Floating point extend a constant to another type. The size of CST
2400 must be smaller or equal to the size of TYPE. Both types must be
2402 ``fptoui (CST to TYPE)``
2403 Convert a floating point constant to the corresponding unsigned
2404 integer constant. TYPE must be a scalar or vector integer type. CST
2405 must be of scalar or vector floating point type. Both CST and TYPE
2406 must be scalars, or vectors of the same number of elements. If the
2407 value won't fit in the integer type, the results are undefined.
2408 ``fptosi (CST to TYPE)``
2409 Convert a floating point constant to the corresponding signed
2410 integer constant. TYPE must be a scalar or vector integer type. CST
2411 must be of scalar or vector floating point type. Both CST and TYPE
2412 must be scalars, or vectors of the same number of elements. If the
2413 value won't fit in the integer type, the results are undefined.
2414 ``uitofp (CST to TYPE)``
2415 Convert an unsigned integer constant to the corresponding floating
2416 point constant. TYPE must be a scalar or vector floating point type.
2417 CST must be of scalar or vector integer type. Both CST and TYPE must
2418 be scalars, or vectors of the same number of elements. If the value
2419 won't fit in the floating point type, the results are undefined.
2420 ``sitofp (CST to TYPE)``
2421 Convert a signed integer constant to the corresponding floating
2422 point constant. TYPE must be a scalar or vector floating point type.
2423 CST must be of scalar or vector integer type. Both CST and TYPE must
2424 be scalars, or vectors of the same number of elements. If the value
2425 won't fit in the floating point type, the results are undefined.
2426 ``ptrtoint (CST to TYPE)``
2427 Convert a pointer typed constant to the corresponding integer
2428 constant. ``TYPE`` must be an integer type. ``CST`` must be of
2429 pointer type. The ``CST`` value is zero extended, truncated, or
2430 unchanged to make it fit in ``TYPE``.
2431 ``inttoptr (CST to TYPE)``
2432 Convert an integer constant to a pointer constant. TYPE must be a
2433 pointer type. CST must be of integer type. The CST value is zero
2434 extended, truncated, or unchanged to make it fit in a pointer size.
2435 This one is *really* dangerous!
2436 ``bitcast (CST to TYPE)``
2437 Convert a constant, CST, to another TYPE. The constraints of the
2438 operands are the same as those for the :ref:`bitcast
2439 instruction <i_bitcast>`.
2440 ``addrspacecast (CST to TYPE)``
2441 Convert a constant pointer or constant vector of pointer, CST, to another
2442 TYPE in a different address space. The constraints of the operands are the
2443 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`.
2444 ``getelementptr (CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)``
2445 Perform the :ref:`getelementptr operation <i_getelementptr>` on
2446 constants. As with the :ref:`getelementptr <i_getelementptr>`
2447 instruction, the index list may have zero or more indexes, which are
2448 required to make sense for the type of "CSTPTR".
2449 ``select (COND, VAL1, VAL2)``
2450 Perform the :ref:`select operation <i_select>` on constants.
2451 ``icmp COND (VAL1, VAL2)``
2452 Performs the :ref:`icmp operation <i_icmp>` on constants.
2453 ``fcmp COND (VAL1, VAL2)``
2454 Performs the :ref:`fcmp operation <i_fcmp>` on constants.
2455 ``extractelement (VAL, IDX)``
2456 Perform the :ref:`extractelement operation <i_extractelement>` on
2458 ``insertelement (VAL, ELT, IDX)``
2459 Perform the :ref:`insertelement operation <i_insertelement>` on
2461 ``shufflevector (VEC1, VEC2, IDXMASK)``
2462 Perform the :ref:`shufflevector operation <i_shufflevector>` on
2464 ``extractvalue (VAL, IDX0, IDX1, ...)``
2465 Perform the :ref:`extractvalue operation <i_extractvalue>` on
2466 constants. The index list is interpreted in a similar manner as
2467 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At
2468 least one index value must be specified.
2469 ``insertvalue (VAL, ELT, IDX0, IDX1, ...)``
2470 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants.
2471 The index list is interpreted in a similar manner as indices in a
2472 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index
2473 value must be specified.
2474 ``OPCODE (LHS, RHS)``
2475 Perform the specified operation of the LHS and RHS constants. OPCODE
2476 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise
2477 binary <bitwiseops>` operations. The constraints on operands are
2478 the same as those for the corresponding instruction (e.g. no bitwise
2479 operations on floating point values are allowed).
2486 Inline Assembler Expressions
2487 ----------------------------
2489 LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level
2490 Inline Assembly <moduleasm>`) through the use of a special value. This
2491 value represents the inline assembler as a string (containing the
2492 instructions to emit), a list of operand constraints (stored as a
2493 string), a flag that indicates whether or not the inline asm expression
2494 has side effects, and a flag indicating whether the function containing
2495 the asm needs to align its stack conservatively. An example inline
2496 assembler expression is:
2498 .. code-block:: llvm
2500 i32 (i32) asm "bswap $0", "=r,r"
2502 Inline assembler expressions may **only** be used as the callee operand
2503 of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction.
2504 Thus, typically we have:
2506 .. code-block:: llvm
2508 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y)
2510 Inline asms with side effects not visible in the constraint list must be
2511 marked as having side effects. This is done through the use of the
2512 '``sideeffect``' keyword, like so:
2514 .. code-block:: llvm
2516 call void asm sideeffect "eieio", ""()
2518 In some cases inline asms will contain code that will not work unless
2519 the stack is aligned in some way, such as calls or SSE instructions on
2520 x86, yet will not contain code that does that alignment within the asm.
2521 The compiler should make conservative assumptions about what the asm
2522 might contain and should generate its usual stack alignment code in the
2523 prologue if the '``alignstack``' keyword is present:
2525 .. code-block:: llvm
2527 call void asm alignstack "eieio", ""()
2529 Inline asms also support using non-standard assembly dialects. The
2530 assumed dialect is ATT. When the '``inteldialect``' keyword is present,
2531 the inline asm is using the Intel dialect. Currently, ATT and Intel are
2532 the only supported dialects. An example is:
2534 .. code-block:: llvm
2536 call void asm inteldialect "eieio", ""()
2538 If multiple keywords appear the '``sideeffect``' keyword must come
2539 first, the '``alignstack``' keyword second and the '``inteldialect``'
2545 The call instructions that wrap inline asm nodes may have a
2546 "``!srcloc``" MDNode attached to it that contains a list of constant
2547 integers. If present, the code generator will use the integer as the
2548 location cookie value when report errors through the ``LLVMContext``
2549 error reporting mechanisms. This allows a front-end to correlate backend
2550 errors that occur with inline asm back to the source code that produced
2553 .. code-block:: llvm
2555 call void asm sideeffect "something bad", ""(), !srcloc !42
2557 !42 = !{ i32 1234567 }
2559 It is up to the front-end to make sense of the magic numbers it places
2560 in the IR. If the MDNode contains multiple constants, the code generator
2561 will use the one that corresponds to the line of the asm that the error
2566 Metadata Nodes and Metadata Strings
2567 -----------------------------------
2569 LLVM IR allows metadata to be attached to instructions in the program
2570 that can convey extra information about the code to the optimizers and
2571 code generator. One example application of metadata is source-level
2572 debug information. There are two metadata primitives: strings and nodes.
2573 All metadata has the ``metadata`` type and is identified in syntax by a
2574 preceding exclamation point ('``!``').
2576 A metadata string is a string surrounded by double quotes. It can
2577 contain any character by escaping non-printable characters with
2578 "``\xx``" where "``xx``" is the two digit hex code. For example:
2581 Metadata nodes are represented with notation similar to structure
2582 constants (a comma separated list of elements, surrounded by braces and
2583 preceded by an exclamation point). Metadata nodes can have any values as
2584 their operand. For example:
2586 .. code-block:: llvm
2588 !{ metadata !"test\00", i32 10}
2590 A :ref:`named metadata <namedmetadatastructure>` is a collection of
2591 metadata nodes, which can be looked up in the module symbol table. For
2594 .. code-block:: llvm
2596 !foo = metadata !{!4, !3}
2598 Metadata can be used as function arguments. Here ``llvm.dbg.value``
2599 function is using two metadata arguments:
2601 .. code-block:: llvm
2603 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2605 Metadata can be attached with an instruction. Here metadata ``!21`` is
2606 attached to the ``add`` instruction using the ``!dbg`` identifier:
2608 .. code-block:: llvm
2610 %indvar.next = add i64 %indvar, 1, !dbg !21
2612 More information about specific metadata nodes recognized by the
2613 optimizers and code generator is found below.
2618 In LLVM IR, memory does not have types, so LLVM's own type system is not
2619 suitable for doing TBAA. Instead, metadata is added to the IR to
2620 describe a type system of a higher level language. This can be used to
2621 implement typical C/C++ TBAA, but it can also be used to implement
2622 custom alias analysis behavior for other languages.
2624 The current metadata format is very simple. TBAA metadata nodes have up
2625 to three fields, e.g.:
2627 .. code-block:: llvm
2629 !0 = metadata !{ metadata !"an example type tree" }
2630 !1 = metadata !{ metadata !"int", metadata !0 }
2631 !2 = metadata !{ metadata !"float", metadata !0 }
2632 !3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
2634 The first field is an identity field. It can be any value, usually a
2635 metadata string, which uniquely identifies the type. The most important
2636 name in the tree is the name of the root node. Two trees with different
2637 root node names are entirely disjoint, even if they have leaves with
2640 The second field identifies the type's parent node in the tree, or is
2641 null or omitted for a root node. A type is considered to alias all of
2642 its descendants and all of its ancestors in the tree. Also, a type is
2643 considered to alias all types in other trees, so that bitcode produced
2644 from multiple front-ends is handled conservatively.
2646 If the third field is present, it's an integer which if equal to 1
2647 indicates that the type is "constant" (meaning
2648 ``pointsToConstantMemory`` should return true; see `other useful
2649 AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_).
2651 '``tbaa.struct``' Metadata
2652 ^^^^^^^^^^^^^^^^^^^^^^^^^^
2654 The :ref:`llvm.memcpy <int_memcpy>` is often used to implement
2655 aggregate assignment operations in C and similar languages, however it
2656 is defined to copy a contiguous region of memory, which is more than
2657 strictly necessary for aggregate types which contain holes due to
2658 padding. Also, it doesn't contain any TBAA information about the fields
2661 ``!tbaa.struct`` metadata can describe which memory subregions in a
2662 memcpy are padding and what the TBAA tags of the struct are.
2664 The current metadata format is very simple. ``!tbaa.struct`` metadata
2665 nodes are a list of operands which are in conceptual groups of three.
2666 For each group of three, the first operand gives the byte offset of a
2667 field in bytes, the second gives its size in bytes, and the third gives
2670 .. code-block:: llvm
2672 !4 = metadata !{ i64 0, i64 4, metadata !1, i64 8, i64 4, metadata !2 }
2674 This describes a struct with two fields. The first is at offset 0 bytes
2675 with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes
2676 and has size 4 bytes and has tbaa tag !2.
2678 Note that the fields need not be contiguous. In this example, there is a
2679 4 byte gap between the two fields. This gap represents padding which
2680 does not carry useful data and need not be preserved.
2682 '``fpmath``' Metadata
2683 ^^^^^^^^^^^^^^^^^^^^^
2685 ``fpmath`` metadata may be attached to any instruction of floating point
2686 type. It can be used to express the maximum acceptable error in the
2687 result of that instruction, in ULPs, thus potentially allowing the
2688 compiler to use a more efficient but less accurate method of computing
2689 it. ULP is defined as follows:
2691 If ``x`` is a real number that lies between two finite consecutive
2692 floating-point numbers ``a`` and ``b``, without being equal to one
2693 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the
2694 distance between the two non-equal finite floating-point numbers
2695 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``.
2697 The metadata node shall consist of a single positive floating point
2698 number representing the maximum relative error, for example:
2700 .. code-block:: llvm
2702 !0 = metadata !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
2704 '``range``' Metadata
2705 ^^^^^^^^^^^^^^^^^^^^
2707 ``range`` metadata may be attached only to loads of integer types. It
2708 expresses the possible ranges the loaded value is in. The ranges are
2709 represented with a flattened list of integers. The loaded value is known
2710 to be in the union of the ranges defined by each consecutive pair. Each
2711 pair has the following properties:
2713 - The type must match the type loaded by the instruction.
2714 - The pair ``a,b`` represents the range ``[a,b)``.
2715 - Both ``a`` and ``b`` are constants.
2716 - The range is allowed to wrap.
2717 - The range should not represent the full or empty set. That is,
2720 In addition, the pairs must be in signed order of the lower bound and
2721 they must be non-contiguous.
2725 .. code-block:: llvm
2727 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1
2728 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
2729 %c = load i8* %z, align 1, !range !2 ; Can only be 0, 1, 3, 4 or 5
2730 %d = load i8* %z, align 1, !range !3 ; Can only be -2, -1, 3, 4 or 5
2732 !0 = metadata !{ i8 0, i8 2 }
2733 !1 = metadata !{ i8 255, i8 2 }
2734 !2 = metadata !{ i8 0, i8 2, i8 3, i8 6 }
2735 !3 = metadata !{ i8 -2, i8 0, i8 3, i8 6 }
2740 It is sometimes useful to attach information to loop constructs. Currently,
2741 loop metadata is implemented as metadata attached to the branch instruction
2742 in the loop latch block. This type of metadata refer to a metadata node that is
2743 guaranteed to be separate for each loop. The loop identifier metadata is
2744 specified with the name ``llvm.loop``.
2746 The loop identifier metadata is implemented using a metadata that refers to
2747 itself to avoid merging it with any other identifier metadata, e.g.,
2748 during module linkage or function inlining. That is, each loop should refer
2749 to their own identification metadata even if they reside in separate functions.
2750 The following example contains loop identifier metadata for two separate loop
2753 .. code-block:: llvm
2755 !0 = metadata !{ metadata !0 }
2756 !1 = metadata !{ metadata !1 }
2758 The loop identifier metadata can be used to specify additional per-loop
2759 metadata. Any operands after the first operand can be treated as user-defined
2760 metadata. For example the ``llvm.vectorizer.unroll`` metadata is understood
2761 by the loop vectorizer to indicate how many times to unroll the loop:
2763 .. code-block:: llvm
2765 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0
2767 !0 = metadata !{ metadata !0, metadata !1 }
2768 !1 = metadata !{ metadata !"llvm.vectorizer.unroll", i32 2 }
2773 Metadata types used to annotate memory accesses with information helpful
2774 for optimizations are prefixed with ``llvm.mem``.
2776 '``llvm.mem.parallel_loop_access``' Metadata
2777 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2779 For a loop to be parallel, in addition to using
2780 the ``llvm.loop`` metadata to mark the loop latch branch instruction,
2781 also all of the memory accessing instructions in the loop body need to be
2782 marked with the ``llvm.mem.parallel_loop_access`` metadata. If there
2783 is at least one memory accessing instruction not marked with the metadata,
2784 the loop must be considered a sequential loop. This causes parallel loops to be
2785 converted to sequential loops due to optimization passes that are unaware of
2786 the parallel semantics and that insert new memory instructions to the loop
2789 Example of a loop that is considered parallel due to its correct use of
2790 both ``llvm.loop`` and ``llvm.mem.parallel_loop_access``
2791 metadata types that refer to the same loop identifier metadata.
2793 .. code-block:: llvm
2797 %0 = load i32* %arrayidx, align 4, !llvm.mem.parallel_loop_access !0
2799 store i32 %0, i32* %arrayidx4, align 4, !llvm.mem.parallel_loop_access !0
2801 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0
2805 !0 = metadata !{ metadata !0 }
2807 It is also possible to have nested parallel loops. In that case the
2808 memory accesses refer to a list of loop identifier metadata nodes instead of
2809 the loop identifier metadata node directly:
2811 .. code-block:: llvm
2818 %0 = load i32* %arrayidx, align 4, !llvm.mem.parallel_loop_access !0
2820 store i32 %0, i32* %arrayidx4, align 4, !llvm.mem.parallel_loop_access !0
2822 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1
2826 %0 = load i32* %arrayidx, align 4, !llvm.mem.parallel_loop_access !0
2828 store i32 %0, i32* %arrayidx4, align 4, !llvm.mem.parallel_loop_access !0
2830 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2
2832 outer.for.end: ; preds = %for.body
2834 !0 = metadata !{ metadata !1, metadata !2 } ; a list of loop identifiers
2835 !1 = metadata !{ metadata !1 } ; an identifier for the inner loop
2836 !2 = metadata !{ metadata !2 } ; an identifier for the outer loop
2838 '``llvm.vectorizer``'
2839 ^^^^^^^^^^^^^^^^^^^^^
2841 Metadata prefixed with ``llvm.vectorizer`` is used to control per-loop
2842 vectorization parameters such as vectorization factor and unroll factor.
2844 ``llvm.vectorizer`` metadata should be used in conjunction with ``llvm.loop``
2845 loop identification metadata.
2847 '``llvm.vectorizer.unroll``' Metadata
2848 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2850 This metadata instructs the loop vectorizer to unroll the specified
2851 loop exactly ``N`` times.
2853 The first operand is the string ``llvm.vectorizer.unroll`` and the second
2854 operand is an integer specifying the unroll factor. For example:
2856 .. code-block:: llvm
2858 !0 = metadata !{ metadata !"llvm.vectorizer.unroll", i32 4 }
2860 Note that setting ``llvm.vectorizer.unroll`` to 1 disables unrolling of the
2863 If ``llvm.vectorizer.unroll`` is set to 0 then the amount of unrolling will be
2864 determined automatically.
2866 '``llvm.vectorizer.width``' Metadata
2867 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2869 This metadata sets the target width of the vectorizer to ``N``. Without
2870 this metadata, the vectorizer will choose a width automatically.
2871 Regardless of this metadata, the vectorizer will only vectorize loops if
2872 it believes it is valid to do so.
2874 The first operand is the string ``llvm.vectorizer.width`` and the second
2875 operand is an integer specifying the width. For example:
2877 .. code-block:: llvm
2879 !0 = metadata !{ metadata !"llvm.vectorizer.width", i32 4 }
2881 Note that setting ``llvm.vectorizer.width`` to 1 disables vectorization of the
2884 If ``llvm.vectorizer.width`` is set to 0 then the width will be determined
2887 Module Flags Metadata
2888 =====================
2890 Information about the module as a whole is difficult to convey to LLVM's
2891 subsystems. The LLVM IR isn't sufficient to transmit this information.
2892 The ``llvm.module.flags`` named metadata exists in order to facilitate
2893 this. These flags are in the form of key / value pairs --- much like a
2894 dictionary --- making it easy for any subsystem who cares about a flag to
2897 The ``llvm.module.flags`` metadata contains a list of metadata triplets.
2898 Each triplet has the following form:
2900 - The first element is a *behavior* flag, which specifies the behavior
2901 when two (or more) modules are merged together, and it encounters two
2902 (or more) metadata with the same ID. The supported behaviors are
2904 - The second element is a metadata string that is a unique ID for the
2905 metadata. Each module may only have one flag entry for each unique ID (not
2906 including entries with the **Require** behavior).
2907 - The third element is the value of the flag.
2909 When two (or more) modules are merged together, the resulting
2910 ``llvm.module.flags`` metadata is the union of the modules' flags. That is, for
2911 each unique metadata ID string, there will be exactly one entry in the merged
2912 modules ``llvm.module.flags`` metadata table, and the value for that entry will
2913 be determined by the merge behavior flag, as described below. The only exception
2914 is that entries with the *Require* behavior are always preserved.
2916 The following behaviors are supported:
2927 Emits an error if two values disagree, otherwise the resulting value
2928 is that of the operands.
2932 Emits a warning if two values disagree. The result value will be the
2933 operand for the flag from the first module being linked.
2937 Adds a requirement that another module flag be present and have a
2938 specified value after linking is performed. The value must be a
2939 metadata pair, where the first element of the pair is the ID of the
2940 module flag to be restricted, and the second element of the pair is
2941 the value the module flag should be restricted to. This behavior can
2942 be used to restrict the allowable results (via triggering of an
2943 error) of linking IDs with the **Override** behavior.
2947 Uses the specified value, regardless of the behavior or value of the
2948 other module. If both modules specify **Override**, but the values
2949 differ, an error will be emitted.
2953 Appends the two values, which are required to be metadata nodes.
2957 Appends the two values, which are required to be metadata
2958 nodes. However, duplicate entries in the second list are dropped
2959 during the append operation.
2961 It is an error for a particular unique flag ID to have multiple behaviors,
2962 except in the case of **Require** (which adds restrictions on another metadata
2963 value) or **Override**.
2965 An example of module flags:
2967 .. code-block:: llvm
2969 !0 = metadata !{ i32 1, metadata !"foo", i32 1 }
2970 !1 = metadata !{ i32 4, metadata !"bar", i32 37 }
2971 !2 = metadata !{ i32 2, metadata !"qux", i32 42 }
2972 !3 = metadata !{ i32 3, metadata !"qux",
2974 metadata !"foo", i32 1
2977 !llvm.module.flags = !{ !0, !1, !2, !3 }
2979 - Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior
2980 if two or more ``!"foo"`` flags are seen is to emit an error if their
2981 values are not equal.
2983 - Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The
2984 behavior if two or more ``!"bar"`` flags are seen is to use the value
2987 - Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The
2988 behavior if two or more ``!"qux"`` flags are seen is to emit a
2989 warning if their values are not equal.
2991 - Metadata ``!3`` has the ID ``!"qux"`` and the value:
2995 metadata !{ metadata !"foo", i32 1 }
2997 The behavior is to emit an error if the ``llvm.module.flags`` does not
2998 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is
3001 Objective-C Garbage Collection Module Flags Metadata
3002 ----------------------------------------------------
3004 On the Mach-O platform, Objective-C stores metadata about garbage
3005 collection in a special section called "image info". The metadata
3006 consists of a version number and a bitmask specifying what types of
3007 garbage collection are supported (if any) by the file. If two or more
3008 modules are linked together their garbage collection metadata needs to
3009 be merged rather than appended together.
3011 The Objective-C garbage collection module flags metadata consists of the
3012 following key-value pairs:
3021 * - ``Objective-C Version``
3022 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2.
3024 * - ``Objective-C Image Info Version``
3025 - **[Required]** --- The version of the image info section. Currently
3028 * - ``Objective-C Image Info Section``
3029 - **[Required]** --- The section to place the metadata. Valid values are
3030 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and
3031 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for
3032 Objective-C ABI version 2.
3034 * - ``Objective-C Garbage Collection``
3035 - **[Required]** --- Specifies whether garbage collection is supported or
3036 not. Valid values are 0, for no garbage collection, and 2, for garbage
3037 collection supported.
3039 * - ``Objective-C GC Only``
3040 - **[Optional]** --- Specifies that only garbage collection is supported.
3041 If present, its value must be 6. This flag requires that the
3042 ``Objective-C Garbage Collection`` flag have the value 2.
3044 Some important flag interactions:
3046 - If a module with ``Objective-C Garbage Collection`` set to 0 is
3047 merged with a module with ``Objective-C Garbage Collection`` set to
3048 2, then the resulting module has the
3049 ``Objective-C Garbage Collection`` flag set to 0.
3050 - A module with ``Objective-C Garbage Collection`` set to 0 cannot be
3051 merged with a module with ``Objective-C GC Only`` set to 6.
3053 Automatic Linker Flags Module Flags Metadata
3054 --------------------------------------------
3056 Some targets support embedding flags to the linker inside individual object
3057 files. Typically this is used in conjunction with language extensions which
3058 allow source files to explicitly declare the libraries they depend on, and have
3059 these automatically be transmitted to the linker via object files.
3061 These flags are encoded in the IR using metadata in the module flags section,
3062 using the ``Linker Options`` key. The merge behavior for this flag is required
3063 to be ``AppendUnique``, and the value for the key is expected to be a metadata
3064 node which should be a list of other metadata nodes, each of which should be a
3065 list of metadata strings defining linker options.
3067 For example, the following metadata section specifies two separate sets of
3068 linker options, presumably to link against ``libz`` and the ``Cocoa``
3071 !0 = metadata !{ i32 6, metadata !"Linker Options",
3073 metadata !{ metadata !"-lz" },
3074 metadata !{ metadata !"-framework", metadata !"Cocoa" } } }
3075 !llvm.module.flags = !{ !0 }
3077 The metadata encoding as lists of lists of options, as opposed to a collapsed
3078 list of options, is chosen so that the IR encoding can use multiple option
3079 strings to specify e.g., a single library, while still having that specifier be
3080 preserved as an atomic element that can be recognized by a target specific
3081 assembly writer or object file emitter.
3083 Each individual option is required to be either a valid option for the target's
3084 linker, or an option that is reserved by the target specific assembly writer or
3085 object file emitter. No other aspect of these options is defined by the IR.
3087 .. _intrinsicglobalvariables:
3089 Intrinsic Global Variables
3090 ==========================
3092 LLVM has a number of "magic" global variables that contain data that
3093 affect code generation or other IR semantics. These are documented here.
3094 All globals of this sort should have a section specified as
3095 "``llvm.metadata``". This section and all globals that start with
3096 "``llvm.``" are reserved for use by LLVM.
3100 The '``llvm.used``' Global Variable
3101 -----------------------------------
3103 The ``@llvm.used`` global is an array which has
3104 :ref:`appending linkage <linkage_appending>`. This array contains a list of
3105 pointers to named global variables, functions and aliases which may optionally
3106 have a pointer cast formed of bitcast or getelementptr. For example, a legal
3109 .. code-block:: llvm
3114 @llvm.used = appending global [2 x i8*] [
3116 i8* bitcast (i32* @Y to i8*)
3117 ], section "llvm.metadata"
3119 If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler,
3120 and linker are required to treat the symbol as if there is a reference to the
3121 symbol that it cannot see (which is why they have to be named). For example, if
3122 a variable has internal linkage and no references other than that from the
3123 ``@llvm.used`` list, it cannot be deleted. This is commonly used to represent
3124 references from inline asms and other things the compiler cannot "see", and
3125 corresponds to "``attribute((used))``" in GNU C.
3127 On some targets, the code generator must emit a directive to the
3128 assembler or object file to prevent the assembler and linker from
3129 molesting the symbol.
3131 .. _gv_llvmcompilerused:
3133 The '``llvm.compiler.used``' Global Variable
3134 --------------------------------------------
3136 The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used``
3137 directive, except that it only prevents the compiler from touching the
3138 symbol. On targets that support it, this allows an intelligent linker to
3139 optimize references to the symbol without being impeded as it would be
3142 This is a rare construct that should only be used in rare circumstances,
3143 and should not be exposed to source languages.
3145 .. _gv_llvmglobalctors:
3147 The '``llvm.global_ctors``' Global Variable
3148 -------------------------------------------
3150 .. code-block:: llvm
3152 %0 = type { i32, void ()* }
3153 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
3155 The ``@llvm.global_ctors`` array contains a list of constructor
3156 functions and associated priorities. The functions referenced by this
3157 array will be called in ascending order of priority (i.e. lowest first)
3158 when the module is loaded. The order of functions with the same priority
3161 .. _llvmglobaldtors:
3163 The '``llvm.global_dtors``' Global Variable
3164 -------------------------------------------
3166 .. code-block:: llvm
3168 %0 = type { i32, void ()* }
3169 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
3171 The ``@llvm.global_dtors`` array contains a list of destructor functions
3172 and associated priorities. The functions referenced by this array will
3173 be called in descending order of priority (i.e. highest first) when the
3174 module is loaded. The order of functions with the same priority is not
3177 Instruction Reference
3178 =====================
3180 The LLVM instruction set consists of several different classifications
3181 of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary
3182 instructions <binaryops>`, :ref:`bitwise binary
3183 instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and
3184 :ref:`other instructions <otherops>`.
3188 Terminator Instructions
3189 -----------------------
3191 As mentioned :ref:`previously <functionstructure>`, every basic block in a
3192 program ends with a "Terminator" instruction, which indicates which
3193 block should be executed after the current block is finished. These
3194 terminator instructions typically yield a '``void``' value: they produce
3195 control flow, not values (the one exception being the
3196 ':ref:`invoke <i_invoke>`' instruction).
3198 The terminator instructions are: ':ref:`ret <i_ret>`',
3199 ':ref:`br <i_br>`', ':ref:`switch <i_switch>`',
3200 ':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`',
3201 ':ref:`resume <i_resume>`', and ':ref:`unreachable <i_unreachable>`'.
3205 '``ret``' Instruction
3206 ^^^^^^^^^^^^^^^^^^^^^
3213 ret <type> <value> ; Return a value from a non-void function
3214 ret void ; Return from void function
3219 The '``ret``' instruction is used to return control flow (and optionally
3220 a value) from a function back to the caller.
3222 There are two forms of the '``ret``' instruction: one that returns a
3223 value and then causes control flow, and one that just causes control
3229 The '``ret``' instruction optionally accepts a single argument, the
3230 return value. The type of the return value must be a ':ref:`first
3231 class <t_firstclass>`' type.
3233 A function is not :ref:`well formed <wellformed>` if it it has a non-void
3234 return type and contains a '``ret``' instruction with no return value or
3235 a return value with a type that does not match its type, or if it has a
3236 void return type and contains a '``ret``' instruction with a return
3242 When the '``ret``' instruction is executed, control flow returns back to
3243 the calling function's context. If the caller is a
3244 ":ref:`call <i_call>`" instruction, execution continues at the
3245 instruction after the call. If the caller was an
3246 ":ref:`invoke <i_invoke>`" instruction, execution continues at the
3247 beginning of the "normal" destination block. If the instruction returns
3248 a value, that value shall set the call or invoke instruction's return
3254 .. code-block:: llvm
3256 ret i32 5 ; Return an integer value of 5
3257 ret void ; Return from a void function
3258 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2
3262 '``br``' Instruction
3263 ^^^^^^^^^^^^^^^^^^^^
3270 br i1 <cond>, label <iftrue>, label <iffalse>
3271 br label <dest> ; Unconditional branch
3276 The '``br``' instruction is used to cause control flow to transfer to a
3277 different basic block in the current function. There are two forms of
3278 this instruction, corresponding to a conditional branch and an
3279 unconditional branch.
3284 The conditional branch form of the '``br``' instruction takes a single
3285 '``i1``' value and two '``label``' values. The unconditional form of the
3286 '``br``' instruction takes a single '``label``' value as a target.
3291 Upon execution of a conditional '``br``' instruction, the '``i1``'
3292 argument is evaluated. If the value is ``true``, control flows to the
3293 '``iftrue``' ``label`` argument. If "cond" is ``false``, control flows
3294 to the '``iffalse``' ``label`` argument.
3299 .. code-block:: llvm
3302 %cond = icmp eq i32 %a, %b
3303 br i1 %cond, label %IfEqual, label %IfUnequal
3311 '``switch``' Instruction
3312 ^^^^^^^^^^^^^^^^^^^^^^^^
3319 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3324 The '``switch``' instruction is used to transfer control flow to one of
3325 several different places. It is a generalization of the '``br``'
3326 instruction, allowing a branch to occur to one of many possible
3332 The '``switch``' instruction uses three parameters: an integer
3333 comparison value '``value``', a default '``label``' destination, and an
3334 array of pairs of comparison value constants and '``label``'s. The table
3335 is not allowed to contain duplicate constant entries.
3340 The ``switch`` instruction specifies a table of values and destinations.
3341 When the '``switch``' instruction is executed, this table is searched
3342 for the given value. If the value is found, control flow is transferred
3343 to the corresponding destination; otherwise, control flow is transferred
3344 to the default destination.
3349 Depending on properties of the target machine and the particular
3350 ``switch`` instruction, this instruction may be code generated in
3351 different ways. For example, it could be generated as a series of
3352 chained conditional branches or with a lookup table.
3357 .. code-block:: llvm
3359 ; Emulate a conditional br instruction
3360 %Val = zext i1 %value to i32
3361 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3363 ; Emulate an unconditional br instruction
3364 switch i32 0, label %dest [ ]
3366 ; Implement a jump table:
3367 switch i32 %val, label %otherwise [ i32 0, label %onzero
3369 i32 2, label %ontwo ]
3373 '``indirectbr``' Instruction
3374 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3381 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3386 The '``indirectbr``' instruction implements an indirect branch to a
3387 label within the current function, whose address is specified by
3388 "``address``". Address must be derived from a
3389 :ref:`blockaddress <blockaddress>` constant.
3394 The '``address``' argument is the address of the label to jump to. The
3395 rest of the arguments indicate the full set of possible destinations
3396 that the address may point to. Blocks are allowed to occur multiple
3397 times in the destination list, though this isn't particularly useful.
3399 This destination list is required so that dataflow analysis has an
3400 accurate understanding of the CFG.
3405 Control transfers to the block specified in the address argument. All
3406 possible destination blocks must be listed in the label list, otherwise
3407 this instruction has undefined behavior. This implies that jumps to
3408 labels defined in other functions have undefined behavior as well.
3413 This is typically implemented with a jump through a register.
3418 .. code-block:: llvm
3420 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3424 '``invoke``' Instruction
3425 ^^^^^^^^^^^^^^^^^^^^^^^^
3432 <result> = invoke [cconv] [ret attrs] <ptr to function ty> <function ptr val>(<function args>) [fn attrs]
3433 to label <normal label> unwind label <exception label>
3438 The '``invoke``' instruction causes control to transfer to a specified
3439 function, with the possibility of control flow transfer to either the
3440 '``normal``' label or the '``exception``' label. If the callee function
3441 returns with the "``ret``" instruction, control flow will return to the
3442 "normal" label. If the callee (or any indirect callees) returns via the
3443 ":ref:`resume <i_resume>`" instruction or other exception handling
3444 mechanism, control is interrupted and continued at the dynamically
3445 nearest "exception" label.
3447 The '``exception``' label is a `landing
3448 pad <ExceptionHandling.html#overview>`_ for the exception. As such,
3449 '``exception``' label is required to have the
3450 ":ref:`landingpad <i_landingpad>`" instruction, which contains the
3451 information about the behavior of the program after unwinding happens,
3452 as its first non-PHI instruction. The restrictions on the
3453 "``landingpad``" instruction's tightly couples it to the "``invoke``"
3454 instruction, so that the important information contained within the
3455 "``landingpad``" instruction can't be lost through normal code motion.
3460 This instruction requires several arguments:
3462 #. The optional "cconv" marker indicates which :ref:`calling
3463 convention <callingconv>` the call should use. If none is
3464 specified, the call defaults to using C calling conventions.
3465 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
3466 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
3468 #. '``ptr to function ty``': shall be the signature of the pointer to
3469 function value being invoked. In most cases, this is a direct
3470 function invocation, but indirect ``invoke``'s are just as possible,
3471 branching off an arbitrary pointer to function value.
3472 #. '``function ptr val``': An LLVM value containing a pointer to a
3473 function to be invoked.
3474 #. '``function args``': argument list whose types match the function
3475 signature argument types and parameter attributes. All arguments must
3476 be of :ref:`first class <t_firstclass>` type. If the function signature
3477 indicates the function accepts a variable number of arguments, the
3478 extra arguments can be specified.
3479 #. '``normal label``': the label reached when the called function
3480 executes a '``ret``' instruction.
3481 #. '``exception label``': the label reached when a callee returns via
3482 the :ref:`resume <i_resume>` instruction or other exception handling
3484 #. The optional :ref:`function attributes <fnattrs>` list. Only
3485 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
3486 attributes are valid here.
3491 This instruction is designed to operate as a standard '``call``'
3492 instruction in most regards. The primary difference is that it
3493 establishes an association with a label, which is used by the runtime
3494 library to unwind the stack.
3496 This instruction is used in languages with destructors to ensure that
3497 proper cleanup is performed in the case of either a ``longjmp`` or a
3498 thrown exception. Additionally, this is important for implementation of
3499 '``catch``' clauses in high-level languages that support them.
3501 For the purposes of the SSA form, the definition of the value returned
3502 by the '``invoke``' instruction is deemed to occur on the edge from the
3503 current block to the "normal" label. If the callee unwinds then no
3504 return value is available.
3509 .. code-block:: llvm
3511 %retval = invoke i32 @Test(i32 15) to label %Continue
3512 unwind label %TestCleanup ; {i32}:retval set
3513 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue
3514 unwind label %TestCleanup ; {i32}:retval set
3518 '``resume``' Instruction
3519 ^^^^^^^^^^^^^^^^^^^^^^^^
3526 resume <type> <value>
3531 The '``resume``' instruction is a terminator instruction that has no
3537 The '``resume``' instruction requires one argument, which must have the
3538 same type as the result of any '``landingpad``' instruction in the same
3544 The '``resume``' instruction resumes propagation of an existing
3545 (in-flight) exception whose unwinding was interrupted with a
3546 :ref:`landingpad <i_landingpad>` instruction.
3551 .. code-block:: llvm
3553 resume { i8*, i32 } %exn
3557 '``unreachable``' Instruction
3558 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3570 The '``unreachable``' instruction has no defined semantics. This
3571 instruction is used to inform the optimizer that a particular portion of
3572 the code is not reachable. This can be used to indicate that the code
3573 after a no-return function cannot be reached, and other facts.
3578 The '``unreachable``' instruction has no defined semantics.
3585 Binary operators are used to do most of the computation in a program.
3586 They require two operands of the same type, execute an operation on
3587 them, and produce a single value. The operands might represent multiple
3588 data, as is the case with the :ref:`vector <t_vector>` data type. The
3589 result value has the same type as its operands.
3591 There are several different binary operators:
3595 '``add``' Instruction
3596 ^^^^^^^^^^^^^^^^^^^^^
3603 <result> = add <ty> <op1>, <op2> ; yields {ty}:result
3604 <result> = add nuw <ty> <op1>, <op2> ; yields {ty}:result
3605 <result> = add nsw <ty> <op1>, <op2> ; yields {ty}:result
3606 <result> = add nuw nsw <ty> <op1>, <op2> ; yields {ty}:result
3611 The '``add``' instruction returns the sum of its two operands.
3616 The two arguments to the '``add``' instruction must be
3617 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
3618 arguments must have identical types.
3623 The value produced is the integer sum of the two operands.
3625 If the sum has unsigned overflow, the result returned is the
3626 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
3629 Because LLVM integers use a two's complement representation, this
3630 instruction is appropriate for both signed and unsigned integers.
3632 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
3633 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
3634 result value of the ``add`` is a :ref:`poison value <poisonvalues>` if
3635 unsigned and/or signed overflow, respectively, occurs.
3640 .. code-block:: llvm
3642 <result> = add i32 4, %var ; yields {i32}:result = 4 + %var
3646 '``fadd``' Instruction
3647 ^^^^^^^^^^^^^^^^^^^^^^
3654 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields {ty}:result
3659 The '``fadd``' instruction returns the sum of its two operands.
3664 The two arguments to the '``fadd``' instruction must be :ref:`floating
3665 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
3666 Both arguments must have identical types.
3671 The value produced is the floating point sum of the two operands. This
3672 instruction can also take any number of :ref:`fast-math flags <fastmath>`,
3673 which are optimization hints to enable otherwise unsafe floating point
3679 .. code-block:: llvm
3681 <result> = fadd float 4.0, %var ; yields {float}:result = 4.0 + %var
3683 '``sub``' Instruction
3684 ^^^^^^^^^^^^^^^^^^^^^
3691 <result> = sub <ty> <op1>, <op2> ; yields {ty}:result
3692 <result> = sub nuw <ty> <op1>, <op2> ; yields {ty}:result
3693 <result> = sub nsw <ty> <op1>, <op2> ; yields {ty}:result
3694 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields {ty}:result
3699 The '``sub``' instruction returns the difference of its two operands.
3701 Note that the '``sub``' instruction is used to represent the '``neg``'
3702 instruction present in most other intermediate representations.
3707 The two arguments to the '``sub``' instruction must be
3708 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
3709 arguments must have identical types.
3714 The value produced is the integer difference of the two operands.
3716 If the difference has unsigned overflow, the result returned is the
3717 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
3720 Because LLVM integers use a two's complement representation, this
3721 instruction is appropriate for both signed and unsigned integers.
3723 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
3724 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
3725 result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if
3726 unsigned and/or signed overflow, respectively, occurs.
3731 .. code-block:: llvm
3733 <result> = sub i32 4, %var ; yields {i32}:result = 4 - %var
3734 <result> = sub i32 0, %val ; yields {i32}:result = -%var
3738 '``fsub``' Instruction
3739 ^^^^^^^^^^^^^^^^^^^^^^
3746 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields {ty}:result
3751 The '``fsub``' instruction returns the difference of its two operands.
3753 Note that the '``fsub``' instruction is used to represent the '``fneg``'
3754 instruction present in most other intermediate representations.
3759 The two arguments to the '``fsub``' instruction must be :ref:`floating
3760 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
3761 Both arguments must have identical types.
3766 The value produced is the floating point difference of the two operands.
3767 This instruction can also take any number of :ref:`fast-math
3768 flags <fastmath>`, which are optimization hints to enable otherwise
3769 unsafe floating point optimizations:
3774 .. code-block:: llvm
3776 <result> = fsub float 4.0, %var ; yields {float}:result = 4.0 - %var
3777 <result> = fsub float -0.0, %val ; yields {float}:result = -%var
3779 '``mul``' Instruction
3780 ^^^^^^^^^^^^^^^^^^^^^
3787 <result> = mul <ty> <op1>, <op2> ; yields {ty}:result
3788 <result> = mul nuw <ty> <op1>, <op2> ; yields {ty}:result
3789 <result> = mul nsw <ty> <op1>, <op2> ; yields {ty}:result
3790 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields {ty}:result
3795 The '``mul``' instruction returns the product of its two operands.
3800 The two arguments to the '``mul``' instruction must be
3801 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
3802 arguments must have identical types.
3807 The value produced is the integer product of the two operands.
3809 If the result of the multiplication has unsigned overflow, the result
3810 returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the
3811 bit width of the result.
3813 Because LLVM integers use a two's complement representation, and the
3814 result is the same width as the operands, this instruction returns the
3815 correct result for both signed and unsigned integers. If a full product
3816 (e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be
3817 sign-extended or zero-extended as appropriate to the width of the full
3820 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
3821 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
3822 result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if
3823 unsigned and/or signed overflow, respectively, occurs.
3828 .. code-block:: llvm
3830 <result> = mul i32 4, %var ; yields {i32}:result = 4 * %var
3834 '``fmul``' Instruction
3835 ^^^^^^^^^^^^^^^^^^^^^^
3842 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields {ty}:result
3847 The '``fmul``' instruction returns the product of its two operands.
3852 The two arguments to the '``fmul``' instruction must be :ref:`floating
3853 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
3854 Both arguments must have identical types.
3859 The value produced is the floating point product of the two operands.
3860 This instruction can also take any number of :ref:`fast-math
3861 flags <fastmath>`, which are optimization hints to enable otherwise
3862 unsafe floating point optimizations:
3867 .. code-block:: llvm
3869 <result> = fmul float 4.0, %var ; yields {float}:result = 4.0 * %var
3871 '``udiv``' Instruction
3872 ^^^^^^^^^^^^^^^^^^^^^^
3879 <result> = udiv <ty> <op1>, <op2> ; yields {ty}:result
3880 <result> = udiv exact <ty> <op1>, <op2> ; yields {ty}:result
3885 The '``udiv``' instruction returns the quotient of its two operands.
3890 The two arguments to the '``udiv``' instruction must be
3891 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
3892 arguments must have identical types.
3897 The value produced is the unsigned integer quotient of the two operands.
3899 Note that unsigned integer division and signed integer division are
3900 distinct operations; for signed integer division, use '``sdiv``'.
3902 Division by zero leads to undefined behavior.
3904 If the ``exact`` keyword is present, the result value of the ``udiv`` is
3905 a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as
3906 such, "((a udiv exact b) mul b) == a").
3911 .. code-block:: llvm
3913 <result> = udiv i32 4, %var ; yields {i32}:result = 4 / %var
3915 '``sdiv``' Instruction
3916 ^^^^^^^^^^^^^^^^^^^^^^
3923 <result> = sdiv <ty> <op1>, <op2> ; yields {ty}:result
3924 <result> = sdiv exact <ty> <op1>, <op2> ; yields {ty}:result
3929 The '``sdiv``' instruction returns the quotient of its two operands.
3934 The two arguments to the '``sdiv``' instruction must be
3935 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
3936 arguments must have identical types.
3941 The value produced is the signed integer quotient of the two operands
3942 rounded towards zero.
3944 Note that signed integer division and unsigned integer division are
3945 distinct operations; for unsigned integer division, use '``udiv``'.
3947 Division by zero leads to undefined behavior. Overflow also leads to
3948 undefined behavior; this is a rare case, but can occur, for example, by
3949 doing a 32-bit division of -2147483648 by -1.
3951 If the ``exact`` keyword is present, the result value of the ``sdiv`` is
3952 a :ref:`poison value <poisonvalues>` if the result would be rounded.
3957 .. code-block:: llvm
3959 <result> = sdiv i32 4, %var ; yields {i32}:result = 4 / %var
3963 '``fdiv``' Instruction
3964 ^^^^^^^^^^^^^^^^^^^^^^
3971 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields {ty}:result
3976 The '``fdiv``' instruction returns the quotient of its two operands.
3981 The two arguments to the '``fdiv``' instruction must be :ref:`floating
3982 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
3983 Both arguments must have identical types.
3988 The value produced is the floating point quotient of the two operands.
3989 This instruction can also take any number of :ref:`fast-math
3990 flags <fastmath>`, which are optimization hints to enable otherwise
3991 unsafe floating point optimizations:
3996 .. code-block:: llvm
3998 <result> = fdiv float 4.0, %var ; yields {float}:result = 4.0 / %var
4000 '``urem``' Instruction
4001 ^^^^^^^^^^^^^^^^^^^^^^
4008 <result> = urem <ty> <op1>, <op2> ; yields {ty}:result
4013 The '``urem``' instruction returns the remainder from the unsigned
4014 division of its two arguments.
4019 The two arguments to the '``urem``' instruction must be
4020 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4021 arguments must have identical types.
4026 This instruction returns the unsigned integer *remainder* of a division.
4027 This instruction always performs an unsigned division to get the
4030 Note that unsigned integer remainder and signed integer remainder are
4031 distinct operations; for signed integer remainder, use '``srem``'.
4033 Taking the remainder of a division by zero leads to undefined behavior.
4038 .. code-block:: llvm
4040 <result> = urem i32 4, %var ; yields {i32}:result = 4 % %var
4042 '``srem``' Instruction
4043 ^^^^^^^^^^^^^^^^^^^^^^
4050 <result> = srem <ty> <op1>, <op2> ; yields {ty}:result
4055 The '``srem``' instruction returns the remainder from the signed
4056 division of its two operands. This instruction can also take
4057 :ref:`vector <t_vector>` versions of the values in which case the elements
4063 The two arguments to the '``srem``' instruction must be
4064 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4065 arguments must have identical types.
4070 This instruction returns the *remainder* of a division (where the result
4071 is either zero or has the same sign as the dividend, ``op1``), not the
4072 *modulo* operator (where the result is either zero or has the same sign
4073 as the divisor, ``op2``) of a value. For more information about the
4074 difference, see `The Math
4075 Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a
4076 table of how this is implemented in various languages, please see
4078 operation <http://en.wikipedia.org/wiki/Modulo_operation>`_.
4080 Note that signed integer remainder and unsigned integer remainder are
4081 distinct operations; for unsigned integer remainder, use '``urem``'.
4083 Taking the remainder of a division by zero leads to undefined behavior.
4084 Overflow also leads to undefined behavior; this is a rare case, but can
4085 occur, for example, by taking the remainder of a 32-bit division of
4086 -2147483648 by -1. (The remainder doesn't actually overflow, but this
4087 rule lets srem be implemented using instructions that return both the
4088 result of the division and the remainder.)
4093 .. code-block:: llvm
4095 <result> = srem i32 4, %var ; yields {i32}:result = 4 % %var
4099 '``frem``' Instruction
4100 ^^^^^^^^^^^^^^^^^^^^^^
4107 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields {ty}:result
4112 The '``frem``' instruction returns the remainder from the division of
4118 The two arguments to the '``frem``' instruction must be :ref:`floating
4119 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4120 Both arguments must have identical types.
4125 This instruction returns the *remainder* of a division. The remainder
4126 has the same sign as the dividend. This instruction can also take any
4127 number of :ref:`fast-math flags <fastmath>`, which are optimization hints
4128 to enable otherwise unsafe floating point optimizations:
4133 .. code-block:: llvm
4135 <result> = frem float 4.0, %var ; yields {float}:result = 4.0 % %var
4139 Bitwise Binary Operations
4140 -------------------------
4142 Bitwise binary operators are used to do various forms of bit-twiddling
4143 in a program. They are generally very efficient instructions and can
4144 commonly be strength reduced from other instructions. They require two
4145 operands of the same type, execute an operation on them, and produce a
4146 single value. The resulting value is the same type as its operands.
4148 '``shl``' Instruction
4149 ^^^^^^^^^^^^^^^^^^^^^
4156 <result> = shl <ty> <op1>, <op2> ; yields {ty}:result
4157 <result> = shl nuw <ty> <op1>, <op2> ; yields {ty}:result
4158 <result> = shl nsw <ty> <op1>, <op2> ; yields {ty}:result
4159 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields {ty}:result
4164 The '``shl``' instruction returns the first operand shifted to the left
4165 a specified number of bits.
4170 Both arguments to the '``shl``' instruction must be the same
4171 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4172 '``op2``' is treated as an unsigned value.
4177 The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`,
4178 where ``n`` is the width of the result. If ``op2`` is (statically or
4179 dynamically) negative or equal to or larger than the number of bits in
4180 ``op1``, the result is undefined. If the arguments are vectors, each
4181 vector element of ``op1`` is shifted by the corresponding shift amount
4184 If the ``nuw`` keyword is present, then the shift produces a :ref:`poison
4185 value <poisonvalues>` if it shifts out any non-zero bits. If the
4186 ``nsw`` keyword is present, then the shift produces a :ref:`poison
4187 value <poisonvalues>` if it shifts out any bits that disagree with the
4188 resultant sign bit. As such, NUW/NSW have the same semantics as they
4189 would if the shift were expressed as a mul instruction with the same
4190 nsw/nuw bits in (mul %op1, (shl 1, %op2)).
4195 .. code-block:: llvm
4197 <result> = shl i32 4, %var ; yields {i32}: 4 << %var
4198 <result> = shl i32 4, 2 ; yields {i32}: 16
4199 <result> = shl i32 1, 10 ; yields {i32}: 1024
4200 <result> = shl i32 1, 32 ; undefined
4201 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4>
4203 '``lshr``' Instruction
4204 ^^^^^^^^^^^^^^^^^^^^^^
4211 <result> = lshr <ty> <op1>, <op2> ; yields {ty}:result
4212 <result> = lshr exact <ty> <op1>, <op2> ; yields {ty}:result
4217 The '``lshr``' instruction (logical shift right) returns the first
4218 operand shifted to the right a specified number of bits with zero fill.
4223 Both arguments to the '``lshr``' instruction must be the same
4224 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4225 '``op2``' is treated as an unsigned value.
4230 This instruction always performs a logical shift right operation. The
4231 most significant bits of the result will be filled with zero bits after
4232 the shift. If ``op2`` is (statically or dynamically) equal to or larger
4233 than the number of bits in ``op1``, the result is undefined. If the
4234 arguments are vectors, each vector element of ``op1`` is shifted by the
4235 corresponding shift amount in ``op2``.
4237 If the ``exact`` keyword is present, the result value of the ``lshr`` is
4238 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
4244 .. code-block:: llvm
4246 <result> = lshr i32 4, 1 ; yields {i32}:result = 2
4247 <result> = lshr i32 4, 2 ; yields {i32}:result = 1
4248 <result> = lshr i8 4, 3 ; yields {i8}:result = 0
4249 <result> = lshr i8 -2, 1 ; yields {i8}:result = 0x7F
4250 <result> = lshr i32 1, 32 ; undefined
4251 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1>
4253 '``ashr``' Instruction
4254 ^^^^^^^^^^^^^^^^^^^^^^
4261 <result> = ashr <ty> <op1>, <op2> ; yields {ty}:result
4262 <result> = ashr exact <ty> <op1>, <op2> ; yields {ty}:result
4267 The '``ashr``' instruction (arithmetic shift right) returns the first
4268 operand shifted to the right a specified number of bits with sign
4274 Both arguments to the '``ashr``' instruction must be the same
4275 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4276 '``op2``' is treated as an unsigned value.
4281 This instruction always performs an arithmetic shift right operation,
4282 The most significant bits of the result will be filled with the sign bit
4283 of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger
4284 than the number of bits in ``op1``, the result is undefined. If the
4285 arguments are vectors, each vector element of ``op1`` is shifted by the
4286 corresponding shift amount in ``op2``.
4288 If the ``exact`` keyword is present, the result value of the ``ashr`` is
4289 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
4295 .. code-block:: llvm
4297 <result> = ashr i32 4, 1 ; yields {i32}:result = 2
4298 <result> = ashr i32 4, 2 ; yields {i32}:result = 1
4299 <result> = ashr i8 4, 3 ; yields {i8}:result = 0
4300 <result> = ashr i8 -2, 1 ; yields {i8}:result = -1
4301 <result> = ashr i32 1, 32 ; undefined
4302 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0>
4304 '``and``' Instruction
4305 ^^^^^^^^^^^^^^^^^^^^^
4312 <result> = and <ty> <op1>, <op2> ; yields {ty}:result
4317 The '``and``' instruction returns the bitwise logical and of its two
4323 The two arguments to the '``and``' instruction must be
4324 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4325 arguments must have identical types.
4330 The truth table used for the '``and``' instruction is:
4347 .. code-block:: llvm
4349 <result> = and i32 4, %var ; yields {i32}:result = 4 & %var
4350 <result> = and i32 15, 40 ; yields {i32}:result = 8
4351 <result> = and i32 4, 8 ; yields {i32}:result = 0
4353 '``or``' Instruction
4354 ^^^^^^^^^^^^^^^^^^^^
4361 <result> = or <ty> <op1>, <op2> ; yields {ty}:result
4366 The '``or``' instruction returns the bitwise logical inclusive or of its
4372 The two arguments to the '``or``' instruction must be
4373 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4374 arguments must have identical types.
4379 The truth table used for the '``or``' instruction is:
4398 <result> = or i32 4, %var ; yields {i32}:result = 4 | %var
4399 <result> = or i32 15, 40 ; yields {i32}:result = 47
4400 <result> = or i32 4, 8 ; yields {i32}:result = 12
4402 '``xor``' Instruction
4403 ^^^^^^^^^^^^^^^^^^^^^
4410 <result> = xor <ty> <op1>, <op2> ; yields {ty}:result
4415 The '``xor``' instruction returns the bitwise logical exclusive or of
4416 its two operands. The ``xor`` is used to implement the "one's
4417 complement" operation, which is the "~" operator in C.
4422 The two arguments to the '``xor``' instruction must be
4423 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4424 arguments must have identical types.
4429 The truth table used for the '``xor``' instruction is:
4446 .. code-block:: llvm
4448 <result> = xor i32 4, %var ; yields {i32}:result = 4 ^ %var
4449 <result> = xor i32 15, 40 ; yields {i32}:result = 39
4450 <result> = xor i32 4, 8 ; yields {i32}:result = 12
4451 <result> = xor i32 %V, -1 ; yields {i32}:result = ~%V
4456 LLVM supports several instructions to represent vector operations in a
4457 target-independent manner. These instructions cover the element-access
4458 and vector-specific operations needed to process vectors effectively.
4459 While LLVM does directly support these vector operations, many
4460 sophisticated algorithms will want to use target-specific intrinsics to
4461 take full advantage of a specific target.
4463 .. _i_extractelement:
4465 '``extractelement``' Instruction
4466 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4473 <result> = extractelement <n x <ty>> <val>, i32 <idx> ; yields <ty>
4478 The '``extractelement``' instruction extracts a single scalar element
4479 from a vector at a specified index.
4484 The first operand of an '``extractelement``' instruction is a value of
4485 :ref:`vector <t_vector>` type. The second operand is an index indicating
4486 the position from which to extract the element. The index may be a
4492 The result is a scalar of the same type as the element type of ``val``.
4493 Its value is the value at position ``idx`` of ``val``. If ``idx``
4494 exceeds the length of ``val``, the results are undefined.
4499 .. code-block:: llvm
4501 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32
4503 .. _i_insertelement:
4505 '``insertelement``' Instruction
4506 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4513 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> ; yields <n x <ty>>
4518 The '``insertelement``' instruction inserts a scalar element into a
4519 vector at a specified index.
4524 The first operand of an '``insertelement``' instruction is a value of
4525 :ref:`vector <t_vector>` type. The second operand is a scalar value whose
4526 type must equal the element type of the first operand. The third operand
4527 is an index indicating the position at which to insert the value. The
4528 index may be a variable.
4533 The result is a vector of the same type as ``val``. Its element values
4534 are those of ``val`` except at position ``idx``, where it gets the value
4535 ``elt``. If ``idx`` exceeds the length of ``val``, the results are
4541 .. code-block:: llvm
4543 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32>
4545 .. _i_shufflevector:
4547 '``shufflevector``' Instruction
4548 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4555 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>>
4560 The '``shufflevector``' instruction constructs a permutation of elements
4561 from two input vectors, returning a vector with the same element type as
4562 the input and length that is the same as the shuffle mask.
4567 The first two operands of a '``shufflevector``' instruction are vectors
4568 with the same type. The third argument is a shuffle mask whose element
4569 type is always 'i32'. The result of the instruction is a vector whose
4570 length is the same as the shuffle mask and whose element type is the
4571 same as the element type of the first two operands.
4573 The shuffle mask operand is required to be a constant vector with either
4574 constant integer or undef values.
4579 The elements of the two input vectors are numbered from left to right
4580 across both of the vectors. The shuffle mask operand specifies, for each
4581 element of the result vector, which element of the two input vectors the
4582 result element gets. The element selector may be undef (meaning "don't
4583 care") and the second operand may be undef if performing a shuffle from
4589 .. code-block:: llvm
4591 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4592 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32>
4593 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4594 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle.
4595 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4596 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32>
4597 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4598 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32>
4600 Aggregate Operations
4601 --------------------
4603 LLVM supports several instructions for working with
4604 :ref:`aggregate <t_aggregate>` values.
4608 '``extractvalue``' Instruction
4609 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4616 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4621 The '``extractvalue``' instruction extracts the value of a member field
4622 from an :ref:`aggregate <t_aggregate>` value.
4627 The first operand of an '``extractvalue``' instruction is a value of
4628 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The operands are
4629 constant indices to specify which value to extract in a similar manner
4630 as indices in a '``getelementptr``' instruction.
4632 The major differences to ``getelementptr`` indexing are:
4634 - Since the value being indexed is not a pointer, the first index is
4635 omitted and assumed to be zero.
4636 - At least one index must be specified.
4637 - Not only struct indices but also array indices must be in bounds.
4642 The result is the value at the position in the aggregate specified by
4648 .. code-block:: llvm
4650 <result> = extractvalue {i32, float} %agg, 0 ; yields i32
4654 '``insertvalue``' Instruction
4655 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4662 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type>
4667 The '``insertvalue``' instruction inserts a value into a member field in
4668 an :ref:`aggregate <t_aggregate>` value.
4673 The first operand of an '``insertvalue``' instruction is a value of
4674 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is
4675 a first-class value to insert. The following operands are constant
4676 indices indicating the position at which to insert the value in a
4677 similar manner as indices in a '``extractvalue``' instruction. The value
4678 to insert must have the same type as the value identified by the
4684 The result is an aggregate of the same type as ``val``. Its value is
4685 that of ``val`` except that the value at the position specified by the
4686 indices is that of ``elt``.
4691 .. code-block:: llvm
4693 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef}
4694 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val}
4695 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 ; yields {i32 1, float %val}
4699 Memory Access and Addressing Operations
4700 ---------------------------------------
4702 A key design point of an SSA-based representation is how it represents
4703 memory. In LLVM, no memory locations are in SSA form, which makes things
4704 very simple. This section describes how to read, write, and allocate
4709 '``alloca``' Instruction
4710 ^^^^^^^^^^^^^^^^^^^^^^^^
4717 <result> = alloca <type>[, inalloca][, <ty> <NumElements>][, align <alignment>] ; yields {type*}:result
4722 The '``alloca``' instruction allocates memory on the stack frame of the
4723 currently executing function, to be automatically released when this
4724 function returns to its caller. The object is always allocated in the
4725 generic address space (address space zero).
4730 The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements``
4731 bytes of memory on the runtime stack, returning a pointer of the
4732 appropriate type to the program. If "NumElements" is specified, it is
4733 the number of elements allocated, otherwise "NumElements" is defaulted
4734 to be one. If a constant alignment is specified, the value result of the
4735 allocation is guaranteed to be aligned to at least that boundary. If not
4736 specified, or if zero, the target can choose to align the allocation on
4737 any convenient boundary compatible with the type.
4739 '``type``' may be any sized type.
4744 Memory is allocated; a pointer is returned. The operation is undefined
4745 if there is insufficient stack space for the allocation. '``alloca``'d
4746 memory is automatically released when the function returns. The
4747 '``alloca``' instruction is commonly used to represent automatic
4748 variables that must have an address available. When the function returns
4749 (either with the ``ret`` or ``resume`` instructions), the memory is
4750 reclaimed. Allocating zero bytes is legal, but the result is undefined.
4751 The order in which memory is allocated (ie., which way the stack grows)
4757 .. code-block:: llvm
4759 %ptr = alloca i32 ; yields {i32*}:ptr
4760 %ptr = alloca i32, i32 4 ; yields {i32*}:ptr
4761 %ptr = alloca i32, i32 4, align 1024 ; yields {i32*}:ptr
4762 %ptr = alloca i32, align 1024 ; yields {i32*}:ptr
4766 '``load``' Instruction
4767 ^^^^^^^^^^^^^^^^^^^^^^
4774 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>]
4775 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
4776 !<index> = !{ i32 1 }
4781 The '``load``' instruction is used to read from memory.
4786 The argument to the ``load`` instruction specifies the memory address
4787 from which to load. The pointer must point to a :ref:`first
4788 class <t_firstclass>` type. If the ``load`` is marked as ``volatile``,
4789 then the optimizer is not allowed to modify the number or order of
4790 execution of this ``load`` with other :ref:`volatile
4791 operations <volatile>`.
4793 If the ``load`` is marked as ``atomic``, it takes an extra
4794 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
4795 ``release`` and ``acq_rel`` orderings are not valid on ``load``
4796 instructions. Atomic loads produce :ref:`defined <memmodel>` results
4797 when they may see multiple atomic stores. The type of the pointee must
4798 be an integer type whose bit width is a power of two greater than or
4799 equal to eight and less than or equal to a target-specific size limit.
4800 ``align`` must be explicitly specified on atomic loads, and the load has
4801 undefined behavior if the alignment is not set to a value which is at
4802 least the size in bytes of the pointee. ``!nontemporal`` does not have
4803 any defined semantics for atomic loads.
4805 The optional constant ``align`` argument specifies the alignment of the
4806 operation (that is, the alignment of the memory address). A value of 0
4807 or an omitted ``align`` argument means that the operation has the ABI
4808 alignment for the target. It is the responsibility of the code emitter
4809 to ensure that the alignment information is correct. Overestimating the
4810 alignment results in undefined behavior. Underestimating the alignment
4811 may produce less efficient code. An alignment of 1 is always safe.
4813 The optional ``!nontemporal`` metadata must reference a single
4814 metadata name ``<index>`` corresponding to a metadata node with one
4815 ``i32`` entry of value 1. The existence of the ``!nontemporal``
4816 metadata on the instruction tells the optimizer and code generator
4817 that this load is not expected to be reused in the cache. The code
4818 generator may select special instructions to save cache bandwidth, such
4819 as the ``MOVNT`` instruction on x86.
4821 The optional ``!invariant.load`` metadata must reference a single
4822 metadata name ``<index>`` corresponding to a metadata node with no
4823 entries. The existence of the ``!invariant.load`` metadata on the
4824 instruction tells the optimizer and code generator that this load
4825 address points to memory which does not change value during program
4826 execution. The optimizer may then move this load around, for example, by
4827 hoisting it out of loops using loop invariant code motion.
4832 The location of memory pointed to is loaded. If the value being loaded
4833 is of scalar type then the number of bytes read does not exceed the
4834 minimum number of bytes needed to hold all bits of the type. For
4835 example, loading an ``i24`` reads at most three bytes. When loading a
4836 value of a type like ``i20`` with a size that is not an integral number
4837 of bytes, the result is undefined if the value was not originally
4838 written using a store of the same type.
4843 .. code-block:: llvm
4845 %ptr = alloca i32 ; yields {i32*}:ptr
4846 store i32 3, i32* %ptr ; yields {void}
4847 %val = load i32* %ptr ; yields {i32}:val = i32 3
4851 '``store``' Instruction
4852 ^^^^^^^^^^^^^^^^^^^^^^^
4859 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] ; yields {void}
4860 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> ; yields {void}
4865 The '``store``' instruction is used to write to memory.
4870 There are two arguments to the ``store`` instruction: a value to store
4871 and an address at which to store it. The type of the ``<pointer>``
4872 operand must be a pointer to the :ref:`first class <t_firstclass>` type of
4873 the ``<value>`` operand. If the ``store`` is marked as ``volatile``,
4874 then the optimizer is not allowed to modify the number or order of
4875 execution of this ``store`` with other :ref:`volatile
4876 operations <volatile>`.
4878 If the ``store`` is marked as ``atomic``, it takes an extra
4879 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
4880 ``acquire`` and ``acq_rel`` orderings aren't valid on ``store``
4881 instructions. Atomic loads produce :ref:`defined <memmodel>` results
4882 when they may see multiple atomic stores. The type of the pointee must
4883 be an integer type whose bit width is a power of two greater than or
4884 equal to eight and less than or equal to a target-specific size limit.
4885 ``align`` must be explicitly specified on atomic stores, and the store
4886 has undefined behavior if the alignment is not set to a value which is
4887 at least the size in bytes of the pointee. ``!nontemporal`` does not
4888 have any defined semantics for atomic stores.
4890 The optional constant ``align`` argument specifies the alignment of the
4891 operation (that is, the alignment of the memory address). A value of 0
4892 or an omitted ``align`` argument means that the operation has the ABI
4893 alignment for the target. It is the responsibility of the code emitter
4894 to ensure that the alignment information is correct. Overestimating the
4895 alignment results in undefined behavior. Underestimating the
4896 alignment may produce less efficient code. An alignment of 1 is always
4899 The optional ``!nontemporal`` metadata must reference a single metadata
4900 name ``<index>`` corresponding to a metadata node with one ``i32`` entry of
4901 value 1. The existence of the ``!nontemporal`` metadata on the instruction
4902 tells the optimizer and code generator that this load is not expected to
4903 be reused in the cache. The code generator may select special
4904 instructions to save cache bandwidth, such as the MOVNT instruction on
4910 The contents of memory are updated to contain ``<value>`` at the
4911 location specified by the ``<pointer>`` operand. If ``<value>`` is
4912 of scalar type then the number of bytes written does not exceed the
4913 minimum number of bytes needed to hold all bits of the type. For
4914 example, storing an ``i24`` writes at most three bytes. When writing a
4915 value of a type like ``i20`` with a size that is not an integral number
4916 of bytes, it is unspecified what happens to the extra bits that do not
4917 belong to the type, but they will typically be overwritten.
4922 .. code-block:: llvm
4924 %ptr = alloca i32 ; yields {i32*}:ptr
4925 store i32 3, i32* %ptr ; yields {void}
4926 %val = load i32* %ptr ; yields {i32}:val = i32 3
4930 '``fence``' Instruction
4931 ^^^^^^^^^^^^^^^^^^^^^^^
4938 fence [singlethread] <ordering> ; yields {void}
4943 The '``fence``' instruction is used to introduce happens-before edges
4949 '``fence``' instructions take an :ref:`ordering <ordering>` argument which
4950 defines what *synchronizes-with* edges they add. They can only be given
4951 ``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings.
4956 A fence A which has (at least) ``release`` ordering semantics
4957 *synchronizes with* a fence B with (at least) ``acquire`` ordering
4958 semantics if and only if there exist atomic operations X and Y, both
4959 operating on some atomic object M, such that A is sequenced before X, X
4960 modifies M (either directly or through some side effect of a sequence
4961 headed by X), Y is sequenced before B, and Y observes M. This provides a
4962 *happens-before* dependency between A and B. Rather than an explicit
4963 ``fence``, one (but not both) of the atomic operations X or Y might
4964 provide a ``release`` or ``acquire`` (resp.) ordering constraint and
4965 still *synchronize-with* the explicit ``fence`` and establish the
4966 *happens-before* edge.
4968 A ``fence`` which has ``seq_cst`` ordering, in addition to having both
4969 ``acquire`` and ``release`` semantics specified above, participates in
4970 the global program order of other ``seq_cst`` operations and/or fences.
4972 The optional ":ref:`singlethread <singlethread>`" argument specifies
4973 that the fence only synchronizes with other fences in the same thread.
4974 (This is useful for interacting with signal handlers.)
4979 .. code-block:: llvm
4981 fence acquire ; yields {void}
4982 fence singlethread seq_cst ; yields {void}
4986 '``cmpxchg``' Instruction
4987 ^^^^^^^^^^^^^^^^^^^^^^^^^
4994 cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering> ; yields {ty}
4999 The '``cmpxchg``' instruction is used to atomically modify memory. It
5000 loads a value in memory and compares it to a given value. If they are
5001 equal, it stores a new value into the memory.
5006 There are three arguments to the '``cmpxchg``' instruction: an address
5007 to operate on, a value to compare to the value currently be at that
5008 address, and a new value to place at that address if the compared values
5009 are equal. The type of '<cmp>' must be an integer type whose bit width
5010 is a power of two greater than or equal to eight and less than or equal
5011 to a target-specific size limit. '<cmp>' and '<new>' must have the same
5012 type, and the type of '<pointer>' must be a pointer to that type. If the
5013 ``cmpxchg`` is marked as ``volatile``, then the optimizer is not allowed
5014 to modify the number or order of execution of this ``cmpxchg`` with
5015 other :ref:`volatile operations <volatile>`.
5017 The :ref:`ordering <ordering>` argument specifies how this ``cmpxchg``
5018 synchronizes with other atomic operations.
5020 The optional "``singlethread``" argument declares that the ``cmpxchg``
5021 is only atomic with respect to code (usually signal handlers) running in
5022 the same thread as the ``cmpxchg``. Otherwise the cmpxchg is atomic with
5023 respect to all other code in the system.
5025 The pointer passed into cmpxchg must have alignment greater than or
5026 equal to the size in memory of the operand.
5031 The contents of memory at the location specified by the '``<pointer>``'
5032 operand is read and compared to '``<cmp>``'; if the read value is the
5033 equal, '``<new>``' is written. The original value at the location is
5036 A successful ``cmpxchg`` is a read-modify-write instruction for the purpose
5037 of identifying release sequences. A failed ``cmpxchg`` is equivalent to an
5038 atomic load with an ordering parameter determined by dropping any
5039 ``release`` part of the ``cmpxchg``'s ordering.
5044 .. code-block:: llvm
5047 %orig = atomic load i32* %ptr unordered ; yields {i32}
5051 %cmp = phi i32 [ %orig, %entry ], [%old, %loop]
5052 %squared = mul i32 %cmp, %cmp
5053 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared ; yields {i32}
5054 %success = icmp eq i32 %cmp, %old
5055 br i1 %success, label %done, label %loop
5062 '``atomicrmw``' Instruction
5063 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
5070 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> ; yields {ty}
5075 The '``atomicrmw``' instruction is used to atomically modify memory.
5080 There are three arguments to the '``atomicrmw``' instruction: an
5081 operation to apply, an address whose value to modify, an argument to the
5082 operation. The operation must be one of the following keywords:
5096 The type of '<value>' must be an integer type whose bit width is a power
5097 of two greater than or equal to eight and less than or equal to a
5098 target-specific size limit. The type of the '``<pointer>``' operand must
5099 be a pointer to that type. If the ``atomicrmw`` is marked as
5100 ``volatile``, then the optimizer is not allowed to modify the number or
5101 order of execution of this ``atomicrmw`` with other :ref:`volatile
5102 operations <volatile>`.
5107 The contents of memory at the location specified by the '``<pointer>``'
5108 operand are atomically read, modified, and written back. The original
5109 value at the location is returned. The modification is specified by the
5112 - xchg: ``*ptr = val``
5113 - add: ``*ptr = *ptr + val``
5114 - sub: ``*ptr = *ptr - val``
5115 - and: ``*ptr = *ptr & val``
5116 - nand: ``*ptr = ~(*ptr & val)``
5117 - or: ``*ptr = *ptr | val``
5118 - xor: ``*ptr = *ptr ^ val``
5119 - max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison)
5120 - min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison)
5121 - umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned
5123 - umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned
5129 .. code-block:: llvm
5131 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields {i32}
5133 .. _i_getelementptr:
5135 '``getelementptr``' Instruction
5136 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5143 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
5144 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
5145 <result> = getelementptr <ptr vector> ptrval, <vector index type> idx
5150 The '``getelementptr``' instruction is used to get the address of a
5151 subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs
5152 address calculation only and does not access memory.
5157 The first argument is always a pointer or a vector of pointers, and
5158 forms the basis of the calculation. The remaining arguments are indices
5159 that indicate which of the elements of the aggregate object are indexed.
5160 The interpretation of each index is dependent on the type being indexed
5161 into. The first index always indexes the pointer value given as the
5162 first argument, the second index indexes a value of the type pointed to
5163 (not necessarily the value directly pointed to, since the first index
5164 can be non-zero), etc. The first type indexed into must be a pointer
5165 value, subsequent types can be arrays, vectors, and structs. Note that
5166 subsequent types being indexed into can never be pointers, since that
5167 would require loading the pointer before continuing calculation.
5169 The type of each index argument depends on the type it is indexing into.
5170 When indexing into a (optionally packed) structure, only ``i32`` integer
5171 **constants** are allowed (when using a vector of indices they must all
5172 be the **same** ``i32`` integer constant). When indexing into an array,
5173 pointer or vector, integers of any width are allowed, and they are not
5174 required to be constant. These integers are treated as signed values
5177 For example, let's consider a C code fragment and how it gets compiled
5193 int *foo(struct ST *s) {
5194 return &s[1].Z.B[5][13];
5197 The LLVM code generated by Clang is:
5199 .. code-block:: llvm
5201 %struct.RT = type { i8, [10 x [20 x i32]], i8 }
5202 %struct.ST = type { i32, double, %struct.RT }
5204 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
5206 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5213 In the example above, the first index is indexing into the
5214 '``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``'
5215 = '``{ i32, double, %struct.RT }``' type, a structure. The second index
5216 indexes into the third element of the structure, yielding a
5217 '``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another
5218 structure. The third index indexes into the second element of the
5219 structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two
5220 dimensions of the array are subscripted into, yielding an '``i32``'
5221 type. The '``getelementptr``' instruction returns a pointer to this
5222 element, thus computing a value of '``i32*``' type.
5224 Note that it is perfectly legal to index partially through a structure,
5225 returning a pointer to an inner element. Because of this, the LLVM code
5226 for the given testcase is equivalent to:
5228 .. code-block:: llvm
5230 define i32* @foo(%struct.ST* %s) {
5231 %t1 = getelementptr %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1
5232 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2
5233 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3
5234 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4
5235 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5
5239 If the ``inbounds`` keyword is present, the result value of the
5240 ``getelementptr`` is a :ref:`poison value <poisonvalues>` if the base
5241 pointer is not an *in bounds* address of an allocated object, or if any
5242 of the addresses that would be formed by successive addition of the
5243 offsets implied by the indices to the base address with infinitely
5244 precise signed arithmetic are not an *in bounds* address of that
5245 allocated object. The *in bounds* addresses for an allocated object are
5246 all the addresses that point into the object, plus the address one byte
5247 past the end. In cases where the base is a vector of pointers the
5248 ``inbounds`` keyword applies to each of the computations element-wise.
5250 If the ``inbounds`` keyword is not present, the offsets are added to the
5251 base address with silently-wrapping two's complement arithmetic. If the
5252 offsets have a different width from the pointer, they are sign-extended
5253 or truncated to the width of the pointer. The result value of the
5254 ``getelementptr`` may be outside the object pointed to by the base
5255 pointer. The result value may not necessarily be used to access memory
5256 though, even if it happens to point into allocated storage. See the
5257 :ref:`Pointer Aliasing Rules <pointeraliasing>` section for more
5260 The getelementptr instruction is often confusing. For some more insight
5261 into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`.
5266 .. code-block:: llvm
5268 ; yields [12 x i8]*:aptr
5269 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5271 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
5273 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
5275 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
5277 In cases where the pointer argument is a vector of pointers, each index
5278 must be a vector with the same number of elements. For example:
5280 .. code-block:: llvm
5282 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5284 Conversion Operations
5285 ---------------------
5287 The instructions in this category are the conversion instructions
5288 (casting) which all take a single operand and a type. They perform
5289 various bit conversions on the operand.
5291 '``trunc .. to``' Instruction
5292 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5299 <result> = trunc <ty> <value> to <ty2> ; yields ty2
5304 The '``trunc``' instruction truncates its operand to the type ``ty2``.
5309 The '``trunc``' instruction takes a value to trunc, and a type to trunc
5310 it to. Both types must be of :ref:`integer <t_integer>` types, or vectors
5311 of the same number of integers. The bit size of the ``value`` must be
5312 larger than the bit size of the destination type, ``ty2``. Equal sized
5313 types are not allowed.
5318 The '``trunc``' instruction truncates the high order bits in ``value``
5319 and converts the remaining bits to ``ty2``. Since the source size must
5320 be larger than the destination size, ``trunc`` cannot be a *no-op cast*.
5321 It will always truncate bits.
5326 .. code-block:: llvm
5328 %X = trunc i32 257 to i8 ; yields i8:1
5329 %Y = trunc i32 123 to i1 ; yields i1:true
5330 %Z = trunc i32 122 to i1 ; yields i1:false
5331 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7>
5333 '``zext .. to``' Instruction
5334 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5341 <result> = zext <ty> <value> to <ty2> ; yields ty2
5346 The '``zext``' instruction zero extends its operand to type ``ty2``.
5351 The '``zext``' instruction takes a value to cast, and a type to cast it
5352 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
5353 the same number of integers. The bit size of the ``value`` must be
5354 smaller than the bit size of the destination type, ``ty2``.
5359 The ``zext`` fills the high order bits of the ``value`` with zero bits
5360 until it reaches the size of the destination type, ``ty2``.
5362 When zero extending from i1, the result will always be either 0 or 1.
5367 .. code-block:: llvm
5369 %X = zext i32 257 to i64 ; yields i64:257
5370 %Y = zext i1 true to i32 ; yields i32:1
5371 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
5373 '``sext .. to``' Instruction
5374 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5381 <result> = sext <ty> <value> to <ty2> ; yields ty2
5386 The '``sext``' sign extends ``value`` to the type ``ty2``.
5391 The '``sext``' instruction takes a value to cast, and a type to cast it
5392 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
5393 the same number of integers. The bit size of the ``value`` must be
5394 smaller than the bit size of the destination type, ``ty2``.
5399 The '``sext``' instruction performs a sign extension by copying the sign
5400 bit (highest order bit) of the ``value`` until it reaches the bit size
5401 of the type ``ty2``.
5403 When sign extending from i1, the extension always results in -1 or 0.
5408 .. code-block:: llvm
5410 %X = sext i8 -1 to i16 ; yields i16 :65535
5411 %Y = sext i1 true to i32 ; yields i32:-1
5412 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
5414 '``fptrunc .. to``' Instruction
5415 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5422 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2
5427 The '``fptrunc``' instruction truncates ``value`` to type ``ty2``.
5432 The '``fptrunc``' instruction takes a :ref:`floating point <t_floating>`
5433 value to cast and a :ref:`floating point <t_floating>` type to cast it to.
5434 The size of ``value`` must be larger than the size of ``ty2``. This
5435 implies that ``fptrunc`` cannot be used to make a *no-op cast*.
5440 The '``fptrunc``' instruction truncates a ``value`` from a larger
5441 :ref:`floating point <t_floating>` type to a smaller :ref:`floating
5442 point <t_floating>` type. If the value cannot fit within the
5443 destination type, ``ty2``, then the results are undefined.
5448 .. code-block:: llvm
5450 %X = fptrunc double 123.0 to float ; yields float:123.0
5451 %Y = fptrunc double 1.0E+300 to float ; yields undefined
5453 '``fpext .. to``' Instruction
5454 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5461 <result> = fpext <ty> <value> to <ty2> ; yields ty2
5466 The '``fpext``' extends a floating point ``value`` to a larger floating
5472 The '``fpext``' instruction takes a :ref:`floating point <t_floating>`
5473 ``value`` to cast, and a :ref:`floating point <t_floating>` type to cast it
5474 to. The source type must be smaller than the destination type.
5479 The '``fpext``' instruction extends the ``value`` from a smaller
5480 :ref:`floating point <t_floating>` type to a larger :ref:`floating
5481 point <t_floating>` type. The ``fpext`` cannot be used to make a
5482 *no-op cast* because it always changes bits. Use ``bitcast`` to make a
5483 *no-op cast* for a floating point cast.
5488 .. code-block:: llvm
5490 %X = fpext float 3.125 to double ; yields double:3.125000e+00
5491 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000
5493 '``fptoui .. to``' Instruction
5494 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5501 <result> = fptoui <ty> <value> to <ty2> ; yields ty2
5506 The '``fptoui``' converts a floating point ``value`` to its unsigned
5507 integer equivalent of type ``ty2``.
5512 The '``fptoui``' instruction takes a value to cast, which must be a
5513 scalar or vector :ref:`floating point <t_floating>` value, and a type to
5514 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
5515 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
5516 type with the same number of elements as ``ty``
5521 The '``fptoui``' instruction converts its :ref:`floating
5522 point <t_floating>` operand into the nearest (rounding towards zero)
5523 unsigned integer value. If the value cannot fit in ``ty2``, the results
5529 .. code-block:: llvm
5531 %X = fptoui double 123.0 to i32 ; yields i32:123
5532 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1
5533 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1
5535 '``fptosi .. to``' Instruction
5536 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5543 <result> = fptosi <ty> <value> to <ty2> ; yields ty2
5548 The '``fptosi``' instruction converts :ref:`floating point <t_floating>`
5549 ``value`` to type ``ty2``.
5554 The '``fptosi``' instruction takes a value to cast, which must be a
5555 scalar or vector :ref:`floating point <t_floating>` value, and a type to
5556 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
5557 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
5558 type with the same number of elements as ``ty``
5563 The '``fptosi``' instruction converts its :ref:`floating
5564 point <t_floating>` operand into the nearest (rounding towards zero)
5565 signed integer value. If the value cannot fit in ``ty2``, the results
5571 .. code-block:: llvm
5573 %X = fptosi double -123.0 to i32 ; yields i32:-123
5574 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1
5575 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1
5577 '``uitofp .. to``' Instruction
5578 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5585 <result> = uitofp <ty> <value> to <ty2> ; yields ty2
5590 The '``uitofp``' instruction regards ``value`` as an unsigned integer
5591 and converts that value to the ``ty2`` type.
5596 The '``uitofp``' instruction takes a value to cast, which must be a
5597 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
5598 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
5599 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
5600 type with the same number of elements as ``ty``
5605 The '``uitofp``' instruction interprets its operand as an unsigned
5606 integer quantity and converts it to the corresponding floating point
5607 value. If the value cannot fit in the floating point value, the results
5613 .. code-block:: llvm
5615 %X = uitofp i32 257 to float ; yields float:257.0
5616 %Y = uitofp i8 -1 to double ; yields double:255.0
5618 '``sitofp .. to``' Instruction
5619 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5626 <result> = sitofp <ty> <value> to <ty2> ; yields ty2
5631 The '``sitofp``' instruction regards ``value`` as a signed integer and
5632 converts that value to the ``ty2`` type.
5637 The '``sitofp``' instruction takes a value to cast, which must be a
5638 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
5639 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
5640 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
5641 type with the same number of elements as ``ty``
5646 The '``sitofp``' instruction interprets its operand as a signed integer
5647 quantity and converts it to the corresponding floating point value. If
5648 the value cannot fit in the floating point value, the results are
5654 .. code-block:: llvm
5656 %X = sitofp i32 257 to float ; yields float:257.0
5657 %Y = sitofp i8 -1 to double ; yields double:-1.0
5661 '``ptrtoint .. to``' Instruction
5662 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5669 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2
5674 The '``ptrtoint``' instruction converts the pointer or a vector of
5675 pointers ``value`` to the integer (or vector of integers) type ``ty2``.
5680 The '``ptrtoint``' instruction takes a ``value`` to cast, which must be
5681 a a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a
5682 type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or
5683 a vector of integers type.
5688 The '``ptrtoint``' instruction converts ``value`` to integer type
5689 ``ty2`` by interpreting the pointer value as an integer and either
5690 truncating or zero extending that value to the size of the integer type.
5691 If ``value`` is smaller than ``ty2`` then a zero extension is done. If
5692 ``value`` is larger than ``ty2`` then a truncation is done. If they are
5693 the same size, then nothing is done (*no-op cast*) other than a type
5699 .. code-block:: llvm
5701 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture
5702 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture
5703 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture
5707 '``inttoptr .. to``' Instruction
5708 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5715 <result> = inttoptr <ty> <value> to <ty2> ; yields ty2
5720 The '``inttoptr``' instruction converts an integer ``value`` to a
5721 pointer type, ``ty2``.
5726 The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to
5727 cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>`
5733 The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by
5734 applying either a zero extension or a truncation depending on the size
5735 of the integer ``value``. If ``value`` is larger than the size of a
5736 pointer then a truncation is done. If ``value`` is smaller than the size
5737 of a pointer then a zero extension is done. If they are the same size,
5738 nothing is done (*no-op cast*).
5743 .. code-block:: llvm
5745 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture
5746 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture
5747 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture
5748 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers
5752 '``bitcast .. to``' Instruction
5753 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5760 <result> = bitcast <ty> <value> to <ty2> ; yields ty2
5765 The '``bitcast``' instruction converts ``value`` to type ``ty2`` without
5771 The '``bitcast``' instruction takes a value to cast, which must be a
5772 non-aggregate first class value, and a type to cast it to, which must
5773 also be a non-aggregate :ref:`first class <t_firstclass>` type. The
5774 bit sizes of ``value`` and the destination type, ``ty2``, must be
5775 identical. If the source type is a pointer, the destination type must
5776 also be a pointer of the same size. This instruction supports bitwise
5777 conversion of vectors to integers and to vectors of other types (as
5778 long as they have the same size).
5783 The '``bitcast``' instruction converts ``value`` to type ``ty2``. It
5784 is always a *no-op cast* because no bits change with this
5785 conversion. The conversion is done as if the ``value`` had been stored
5786 to memory and read back as type ``ty2``. Pointer (or vector of
5787 pointers) types may only be converted to other pointer (or vector of
5788 pointers) types with the same address space through this instruction.
5789 To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>`
5790 or :ref:`ptrtoint <i_ptrtoint>` instructions first.
5795 .. code-block:: llvm
5797 %X = bitcast i8 255 to i8 ; yields i8 :-1
5798 %Y = bitcast i32* %x to sint* ; yields sint*:%x
5799 %Z = bitcast <2 x int> %V to i64; ; yields i64: %V
5800 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*>
5802 .. _i_addrspacecast:
5804 '``addrspacecast .. to``' Instruction
5805 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5812 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2
5817 The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in
5818 address space ``n`` to type ``pty2`` in address space ``m``.
5823 The '``addrspacecast``' instruction takes a pointer or vector of pointer value
5824 to cast and a pointer type to cast it to, which must have a different
5830 The '``addrspacecast``' instruction converts the pointer value
5831 ``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex
5832 value modification, depending on the target and the address space
5833 pair. Pointer conversions within the same address space must be
5834 performed with the ``bitcast`` instruction. Note that if the address space
5835 conversion is legal then both result and operand refer to the same memory
5841 .. code-block:: llvm
5843 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x
5844 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y
5845 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z
5852 The instructions in this category are the "miscellaneous" instructions,
5853 which defy better classification.
5857 '``icmp``' Instruction
5858 ^^^^^^^^^^^^^^^^^^^^^^
5865 <result> = icmp <cond> <ty> <op1>, <op2> ; yields {i1} or {<N x i1>}:result
5870 The '``icmp``' instruction returns a boolean value or a vector of
5871 boolean values based on comparison of its two integer, integer vector,
5872 pointer, or pointer vector operands.
5877 The '``icmp``' instruction takes three operands. The first operand is
5878 the condition code indicating the kind of comparison to perform. It is
5879 not a value, just a keyword. The possible condition code are:
5882 #. ``ne``: not equal
5883 #. ``ugt``: unsigned greater than
5884 #. ``uge``: unsigned greater or equal
5885 #. ``ult``: unsigned less than
5886 #. ``ule``: unsigned less or equal
5887 #. ``sgt``: signed greater than
5888 #. ``sge``: signed greater or equal
5889 #. ``slt``: signed less than
5890 #. ``sle``: signed less or equal
5892 The remaining two arguments must be :ref:`integer <t_integer>` or
5893 :ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They
5894 must also be identical types.
5899 The '``icmp``' compares ``op1`` and ``op2`` according to the condition
5900 code given as ``cond``. The comparison performed always yields either an
5901 :ref:`i1 <t_integer>` or vector of ``i1`` result, as follows:
5903 #. ``eq``: yields ``true`` if the operands are equal, ``false``
5904 otherwise. No sign interpretation is necessary or performed.
5905 #. ``ne``: yields ``true`` if the operands are unequal, ``false``
5906 otherwise. No sign interpretation is necessary or performed.
5907 #. ``ugt``: interprets the operands as unsigned values and yields
5908 ``true`` if ``op1`` is greater than ``op2``.
5909 #. ``uge``: interprets the operands as unsigned values and yields
5910 ``true`` if ``op1`` is greater than or equal to ``op2``.
5911 #. ``ult``: interprets the operands as unsigned values and yields
5912 ``true`` if ``op1`` is less than ``op2``.
5913 #. ``ule``: interprets the operands as unsigned values and yields
5914 ``true`` if ``op1`` is less than or equal to ``op2``.
5915 #. ``sgt``: interprets the operands as signed values and yields ``true``
5916 if ``op1`` is greater than ``op2``.
5917 #. ``sge``: interprets the operands as signed values and yields ``true``
5918 if ``op1`` is greater than or equal to ``op2``.
5919 #. ``slt``: interprets the operands as signed values and yields ``true``
5920 if ``op1`` is less than ``op2``.
5921 #. ``sle``: interprets the operands as signed values and yields ``true``
5922 if ``op1`` is less than or equal to ``op2``.
5924 If the operands are :ref:`pointer <t_pointer>` typed, the pointer values
5925 are compared as if they were integers.
5927 If the operands are integer vectors, then they are compared element by
5928 element. The result is an ``i1`` vector with the same number of elements
5929 as the values being compared. Otherwise, the result is an ``i1``.
5934 .. code-block:: llvm
5936 <result> = icmp eq i32 4, 5 ; yields: result=false
5937 <result> = icmp ne float* %X, %X ; yields: result=false
5938 <result> = icmp ult i16 4, 5 ; yields: result=true
5939 <result> = icmp sgt i16 4, 5 ; yields: result=false
5940 <result> = icmp ule i16 -4, 5 ; yields: result=false
5941 <result> = icmp sge i16 4, 5 ; yields: result=false
5943 Note that the code generator does not yet support vector types with the
5944 ``icmp`` instruction.
5948 '``fcmp``' Instruction
5949 ^^^^^^^^^^^^^^^^^^^^^^
5956 <result> = fcmp <cond> <ty> <op1>, <op2> ; yields {i1} or {<N x i1>}:result
5961 The '``fcmp``' instruction returns a boolean value or vector of boolean
5962 values based on comparison of its operands.
5964 If the operands are floating point scalars, then the result type is a
5965 boolean (:ref:`i1 <t_integer>`).
5967 If the operands are floating point vectors, then the result type is a
5968 vector of boolean with the same number of elements as the operands being
5974 The '``fcmp``' instruction takes three operands. The first operand is
5975 the condition code indicating the kind of comparison to perform. It is
5976 not a value, just a keyword. The possible condition code are:
5978 #. ``false``: no comparison, always returns false
5979 #. ``oeq``: ordered and equal
5980 #. ``ogt``: ordered and greater than
5981 #. ``oge``: ordered and greater than or equal
5982 #. ``olt``: ordered and less than
5983 #. ``ole``: ordered and less than or equal
5984 #. ``one``: ordered and not equal
5985 #. ``ord``: ordered (no nans)
5986 #. ``ueq``: unordered or equal
5987 #. ``ugt``: unordered or greater than
5988 #. ``uge``: unordered or greater than or equal
5989 #. ``ult``: unordered or less than
5990 #. ``ule``: unordered or less than or equal
5991 #. ``une``: unordered or not equal
5992 #. ``uno``: unordered (either nans)
5993 #. ``true``: no comparison, always returns true
5995 *Ordered* means that neither operand is a QNAN while *unordered* means
5996 that either operand may be a QNAN.
5998 Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating
5999 point <t_floating>` type or a :ref:`vector <t_vector>` of floating point
6000 type. They must have identical types.
6005 The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the
6006 condition code given as ``cond``. If the operands are vectors, then the
6007 vectors are compared element by element. Each comparison performed
6008 always yields an :ref:`i1 <t_integer>` result, as follows:
6010 #. ``false``: always yields ``false``, regardless of operands.
6011 #. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1``
6012 is equal to ``op2``.
6013 #. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1``
6014 is greater than ``op2``.
6015 #. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1``
6016 is greater than or equal to ``op2``.
6017 #. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1``
6018 is less than ``op2``.
6019 #. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1``
6020 is less than or equal to ``op2``.
6021 #. ``one``: yields ``true`` if both operands are not a QNAN and ``op1``
6022 is not equal to ``op2``.
6023 #. ``ord``: yields ``true`` if both operands are not a QNAN.
6024 #. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is
6026 #. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is
6027 greater than ``op2``.
6028 #. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is
6029 greater than or equal to ``op2``.
6030 #. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is
6032 #. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is
6033 less than or equal to ``op2``.
6034 #. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is
6035 not equal to ``op2``.
6036 #. ``uno``: yields ``true`` if either operand is a QNAN.
6037 #. ``true``: always yields ``true``, regardless of operands.
6042 .. code-block:: llvm
6044 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false
6045 <result> = fcmp one float 4.0, 5.0 ; yields: result=true
6046 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true
6047 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false
6049 Note that the code generator does not yet support vector types with the
6050 ``fcmp`` instruction.
6054 '``phi``' Instruction
6055 ^^^^^^^^^^^^^^^^^^^^^
6062 <result> = phi <ty> [ <val0>, <label0>], ...
6067 The '``phi``' instruction is used to implement the φ node in the SSA
6068 graph representing the function.
6073 The type of the incoming values is specified with the first type field.
6074 After this, the '``phi``' instruction takes a list of pairs as
6075 arguments, with one pair for each predecessor basic block of the current
6076 block. Only values of :ref:`first class <t_firstclass>` type may be used as
6077 the value arguments to the PHI node. Only labels may be used as the
6080 There must be no non-phi instructions between the start of a basic block
6081 and the PHI instructions: i.e. PHI instructions must be first in a basic
6084 For the purposes of the SSA form, the use of each incoming value is
6085 deemed to occur on the edge from the corresponding predecessor block to
6086 the current block (but after any definition of an '``invoke``'
6087 instruction's return value on the same edge).
6092 At runtime, the '``phi``' instruction logically takes on the value
6093 specified by the pair corresponding to the predecessor basic block that
6094 executed just prior to the current block.
6099 .. code-block:: llvm
6101 Loop: ; Infinite loop that counts from 0 on up...
6102 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6103 %nextindvar = add i32 %indvar, 1
6108 '``select``' Instruction
6109 ^^^^^^^^^^^^^^^^^^^^^^^^
6116 <result> = select selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty
6118 selty is either i1 or {<N x i1>}
6123 The '``select``' instruction is used to choose one value based on a
6124 condition, without branching.
6129 The '``select``' instruction requires an 'i1' value or a vector of 'i1'
6130 values indicating the condition, and two values of the same :ref:`first
6131 class <t_firstclass>` type. If the val1/val2 are vectors and the
6132 condition is a scalar, then entire vectors are selected, not individual
6138 If the condition is an i1 and it evaluates to 1, the instruction returns
6139 the first value argument; otherwise, it returns the second value
6142 If the condition is a vector of i1, then the value arguments must be
6143 vectors of the same size, and the selection is done element by element.
6148 .. code-block:: llvm
6150 %X = select i1 true, i8 17, i8 42 ; yields i8:17
6154 '``call``' Instruction
6155 ^^^^^^^^^^^^^^^^^^^^^^
6162 <result> = [tail] call [cconv] [ret attrs] <ty> [<fnty>*] <fnptrval>(<function args>) [fn attrs]
6167 The '``call``' instruction represents a simple function call.
6172 This instruction requires several arguments:
6174 #. The optional "tail" marker indicates that the callee function does
6175 not access any allocas or varargs in the caller. Note that calls may
6176 be marked "tail" even if they do not occur before a
6177 :ref:`ret <i_ret>` instruction. If the "tail" marker is present, the
6178 function call is eligible for tail call optimization, but `might not
6179 in fact be optimized into a jump <CodeGenerator.html#tailcallopt>`_.
6180 The code generator may optimize calls marked "tail" with either 1)
6181 automatic `sibling call
6182 optimization <CodeGenerator.html#sibcallopt>`_ when the caller and
6183 callee have matching signatures, or 2) forced tail call optimization
6184 when the following extra requirements are met:
6186 - Caller and callee both have the calling convention ``fastcc``.
6187 - The call is in tail position (ret immediately follows call and ret
6188 uses value of call or is void).
6189 - Option ``-tailcallopt`` is enabled, or
6190 ``llvm::GuaranteedTailCallOpt`` is ``true``.
6191 - `Platform specific constraints are
6192 met. <CodeGenerator.html#tailcallopt>`_
6194 #. The optional "cconv" marker indicates which :ref:`calling
6195 convention <callingconv>` the call should use. If none is
6196 specified, the call defaults to using C calling conventions. The
6197 calling convention of the call must match the calling convention of
6198 the target function, or else the behavior is undefined.
6199 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
6200 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
6202 #. '``ty``': the type of the call instruction itself which is also the
6203 type of the return value. Functions that return no value are marked
6205 #. '``fnty``': shall be the signature of the pointer to function value
6206 being invoked. The argument types must match the types implied by
6207 this signature. This type can be omitted if the function is not
6208 varargs and if the function type does not return a pointer to a
6210 #. '``fnptrval``': An LLVM value containing a pointer to a function to
6211 be invoked. In most cases, this is a direct function invocation, but
6212 indirect ``call``'s are just as possible, calling an arbitrary pointer
6214 #. '``function args``': argument list whose types match the function
6215 signature argument types and parameter attributes. All arguments must
6216 be of :ref:`first class <t_firstclass>` type. If the function signature
6217 indicates the function accepts a variable number of arguments, the
6218 extra arguments can be specified.
6219 #. The optional :ref:`function attributes <fnattrs>` list. Only
6220 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
6221 attributes are valid here.
6226 The '``call``' instruction is used to cause control flow to transfer to
6227 a specified function, with its incoming arguments bound to the specified
6228 values. Upon a '``ret``' instruction in the called function, control
6229 flow continues with the instruction after the function call, and the
6230 return value of the function is bound to the result argument.
6235 .. code-block:: llvm
6237 %retval = call i32 @test(i32 %argc)
6238 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32
6239 %X = tail call i32 @foo() ; yields i32
6240 %Y = tail call fastcc i32 @foo() ; yields i32
6241 call void %foo(i8 97 signext)
6243 %struct.A = type { i32, i8 }
6244 %r = call %struct.A @foo() ; yields { 32, i8 }
6245 %gr = extractvalue %struct.A %r, 0 ; yields i32
6246 %gr1 = extractvalue %struct.A %r, 1 ; yields i8
6247 %Z = call void @foo() noreturn ; indicates that %foo never returns normally
6248 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended
6250 llvm treats calls to some functions with names and arguments that match
6251 the standard C99 library as being the C99 library functions, and may
6252 perform optimizations or generate code for them under that assumption.
6253 This is something we'd like to change in the future to provide better
6254 support for freestanding environments and non-C-based languages.
6258 '``va_arg``' Instruction
6259 ^^^^^^^^^^^^^^^^^^^^^^^^
6266 <resultval> = va_arg <va_list*> <arglist>, <argty>
6271 The '``va_arg``' instruction is used to access arguments passed through
6272 the "variable argument" area of a function call. It is used to implement
6273 the ``va_arg`` macro in C.
6278 This instruction takes a ``va_list*`` value and the type of the
6279 argument. It returns a value of the specified argument type and
6280 increments the ``va_list`` to point to the next argument. The actual
6281 type of ``va_list`` is target specific.
6286 The '``va_arg``' instruction loads an argument of the specified type
6287 from the specified ``va_list`` and causes the ``va_list`` to point to
6288 the next argument. For more information, see the variable argument
6289 handling :ref:`Intrinsic Functions <int_varargs>`.
6291 It is legal for this instruction to be called in a function which does
6292 not take a variable number of arguments, for example, the ``vfprintf``
6295 ``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic
6296 function <intrinsics>` because it takes a type as an argument.
6301 See the :ref:`variable argument processing <int_varargs>` section.
6303 Note that the code generator does not yet fully support va\_arg on many
6304 targets. Also, it does not currently support va\_arg with aggregate
6305 types on any target.
6309 '``landingpad``' Instruction
6310 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6317 <resultval> = landingpad <resultty> personality <type> <pers_fn> <clause>+
6318 <resultval> = landingpad <resultty> personality <type> <pers_fn> cleanup <clause>*
6320 <clause> := catch <type> <value>
6321 <clause> := filter <array constant type> <array constant>
6326 The '``landingpad``' instruction is used by `LLVM's exception handling
6327 system <ExceptionHandling.html#overview>`_ to specify that a basic block
6328 is a landing pad --- one where the exception lands, and corresponds to the
6329 code found in the ``catch`` portion of a ``try``/``catch`` sequence. It
6330 defines values supplied by the personality function (``pers_fn``) upon
6331 re-entry to the function. The ``resultval`` has the type ``resultty``.
6336 This instruction takes a ``pers_fn`` value. This is the personality
6337 function associated with the unwinding mechanism. The optional
6338 ``cleanup`` flag indicates that the landing pad block is a cleanup.
6340 A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and
6341 contains the global variable representing the "type" that may be caught
6342 or filtered respectively. Unlike the ``catch`` clause, the ``filter``
6343 clause takes an array constant as its argument. Use
6344 "``[0 x i8**] undef``" for a filter which cannot throw. The
6345 '``landingpad``' instruction must contain *at least* one ``clause`` or
6346 the ``cleanup`` flag.
6351 The '``landingpad``' instruction defines the values which are set by the
6352 personality function (``pers_fn``) upon re-entry to the function, and
6353 therefore the "result type" of the ``landingpad`` instruction. As with
6354 calling conventions, how the personality function results are
6355 represented in LLVM IR is target specific.
6357 The clauses are applied in order from top to bottom. If two
6358 ``landingpad`` instructions are merged together through inlining, the
6359 clauses from the calling function are appended to the list of clauses.
6360 When the call stack is being unwound due to an exception being thrown,
6361 the exception is compared against each ``clause`` in turn. If it doesn't
6362 match any of the clauses, and the ``cleanup`` flag is not set, then
6363 unwinding continues further up the call stack.
6365 The ``landingpad`` instruction has several restrictions:
6367 - A landing pad block is a basic block which is the unwind destination
6368 of an '``invoke``' instruction.
6369 - A landing pad block must have a '``landingpad``' instruction as its
6370 first non-PHI instruction.
6371 - There can be only one '``landingpad``' instruction within the landing
6373 - A basic block that is not a landing pad block may not include a
6374 '``landingpad``' instruction.
6375 - All '``landingpad``' instructions in a function must have the same
6376 personality function.
6381 .. code-block:: llvm
6383 ;; A landing pad which can catch an integer.
6384 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6386 ;; A landing pad that is a cleanup.
6387 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6389 ;; A landing pad which can catch an integer and can only throw a double.
6390 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6392 filter [1 x i8**] [@_ZTId]
6399 LLVM supports the notion of an "intrinsic function". These functions
6400 have well known names and semantics and are required to follow certain
6401 restrictions. Overall, these intrinsics represent an extension mechanism
6402 for the LLVM language that does not require changing all of the
6403 transformations in LLVM when adding to the language (or the bitcode
6404 reader/writer, the parser, etc...).
6406 Intrinsic function names must all start with an "``llvm.``" prefix. This
6407 prefix is reserved in LLVM for intrinsic names; thus, function names may
6408 not begin with this prefix. Intrinsic functions must always be external
6409 functions: you cannot define the body of intrinsic functions. Intrinsic
6410 functions may only be used in call or invoke instructions: it is illegal
6411 to take the address of an intrinsic function. Additionally, because
6412 intrinsic functions are part of the LLVM language, it is required if any
6413 are added that they be documented here.
6415 Some intrinsic functions can be overloaded, i.e., the intrinsic
6416 represents a family of functions that perform the same operation but on
6417 different data types. Because LLVM can represent over 8 million
6418 different integer types, overloading is used commonly to allow an
6419 intrinsic function to operate on any integer type. One or more of the
6420 argument types or the result type can be overloaded to accept any
6421 integer type. Argument types may also be defined as exactly matching a
6422 previous argument's type or the result type. This allows an intrinsic
6423 function which accepts multiple arguments, but needs all of them to be
6424 of the same type, to only be overloaded with respect to a single
6425 argument or the result.
6427 Overloaded intrinsics will have the names of its overloaded argument
6428 types encoded into its function name, each preceded by a period. Only
6429 those types which are overloaded result in a name suffix. Arguments
6430 whose type is matched against another type do not. For example, the
6431 ``llvm.ctpop`` function can take an integer of any width and returns an
6432 integer of exactly the same integer width. This leads to a family of
6433 functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and
6434 ``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is
6435 overloaded, and only one type suffix is required. Because the argument's
6436 type is matched against the return type, it does not require its own
6439 To learn how to add an intrinsic function, please see the `Extending
6440 LLVM Guide <ExtendingLLVM.html>`_.
6444 Variable Argument Handling Intrinsics
6445 -------------------------------------
6447 Variable argument support is defined in LLVM with the
6448 :ref:`va_arg <i_va_arg>` instruction and these three intrinsic
6449 functions. These functions are related to the similarly named macros
6450 defined in the ``<stdarg.h>`` header file.
6452 All of these functions operate on arguments that use a target-specific
6453 value type "``va_list``". The LLVM assembly language reference manual
6454 does not define what this type is, so all transformations should be
6455 prepared to handle these functions regardless of the type used.
6457 This example shows how the :ref:`va_arg <i_va_arg>` instruction and the
6458 variable argument handling intrinsic functions are used.
6460 .. code-block:: llvm
6462 define i32 @test(i32 %X, ...) {
6463 ; Initialize variable argument processing
6465 %ap2 = bitcast i8** %ap to i8*
6466 call void @llvm.va_start(i8* %ap2)
6468 ; Read a single integer argument
6469 %tmp = va_arg i8** %ap, i32
6471 ; Demonstrate usage of llvm.va_copy and llvm.va_end
6473 %aq2 = bitcast i8** %aq to i8*
6474 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
6475 call void @llvm.va_end(i8* %aq2)
6477 ; Stop processing of arguments.
6478 call void @llvm.va_end(i8* %ap2)
6482 declare void @llvm.va_start(i8*)
6483 declare void @llvm.va_copy(i8*, i8*)
6484 declare void @llvm.va_end(i8*)
6488 '``llvm.va_start``' Intrinsic
6489 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6496 declare void @llvm.va_start(i8* <arglist>)
6501 The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for
6502 subsequent use by ``va_arg``.
6507 The argument is a pointer to a ``va_list`` element to initialize.
6512 The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro
6513 available in C. In a target-dependent way, it initializes the
6514 ``va_list`` element to which the argument points, so that the next call
6515 to ``va_arg`` will produce the first variable argument passed to the
6516 function. Unlike the C ``va_start`` macro, this intrinsic does not need
6517 to know the last argument of the function as the compiler can figure
6520 '``llvm.va_end``' Intrinsic
6521 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
6528 declare void @llvm.va_end(i8* <arglist>)
6533 The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been
6534 initialized previously with ``llvm.va_start`` or ``llvm.va_copy``.
6539 The argument is a pointer to a ``va_list`` to destroy.
6544 The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro
6545 available in C. In a target-dependent way, it destroys the ``va_list``
6546 element to which the argument points. Calls to
6547 :ref:`llvm.va_start <int_va_start>` and
6548 :ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to
6553 '``llvm.va_copy``' Intrinsic
6554 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6561 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
6566 The '``llvm.va_copy``' intrinsic copies the current argument position
6567 from the source argument list to the destination argument list.
6572 The first argument is a pointer to a ``va_list`` element to initialize.
6573 The second argument is a pointer to a ``va_list`` element to copy from.
6578 The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro
6579 available in C. In a target-dependent way, it copies the source
6580 ``va_list`` element into the destination ``va_list`` element. This
6581 intrinsic is necessary because the `` llvm.va_start`` intrinsic may be
6582 arbitrarily complex and require, for example, memory allocation.
6584 Accurate Garbage Collection Intrinsics
6585 --------------------------------------
6587 LLVM support for `Accurate Garbage Collection <GarbageCollection.html>`_
6588 (GC) requires the implementation and generation of these intrinsics.
6589 These intrinsics allow identification of :ref:`GC roots on the
6590 stack <int_gcroot>`, as well as garbage collector implementations that
6591 require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers.
6592 Front-ends for type-safe garbage collected languages should generate
6593 these intrinsics to make use of the LLVM garbage collectors. For more
6594 details, see `Accurate Garbage Collection with
6595 LLVM <GarbageCollection.html>`_.
6597 The garbage collection intrinsics only operate on objects in the generic
6598 address space (address space zero).
6602 '``llvm.gcroot``' Intrinsic
6603 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
6610 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
6615 The '``llvm.gcroot``' intrinsic declares the existence of a GC root to
6616 the code generator, and allows some metadata to be associated with it.
6621 The first argument specifies the address of a stack object that contains
6622 the root pointer. The second pointer (which must be either a constant or
6623 a global value address) contains the meta-data to be associated with the
6629 At runtime, a call to this intrinsic stores a null pointer into the
6630 "ptrloc" location. At compile-time, the code generator generates
6631 information to allow the runtime to find the pointer at GC safe points.
6632 The '``llvm.gcroot``' intrinsic may only be used in a function which
6633 :ref:`specifies a GC algorithm <gc>`.
6637 '``llvm.gcread``' Intrinsic
6638 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
6645 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
6650 The '``llvm.gcread``' intrinsic identifies reads of references from heap
6651 locations, allowing garbage collector implementations that require read
6657 The second argument is the address to read from, which should be an
6658 address allocated from the garbage collector. The first object is a
6659 pointer to the start of the referenced object, if needed by the language
6660 runtime (otherwise null).
6665 The '``llvm.gcread``' intrinsic has the same semantics as a load
6666 instruction, but may be replaced with substantially more complex code by
6667 the garbage collector runtime, as needed. The '``llvm.gcread``'
6668 intrinsic may only be used in a function which :ref:`specifies a GC
6673 '``llvm.gcwrite``' Intrinsic
6674 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6681 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
6686 The '``llvm.gcwrite``' intrinsic identifies writes of references to heap
6687 locations, allowing garbage collector implementations that require write
6688 barriers (such as generational or reference counting collectors).
6693 The first argument is the reference to store, the second is the start of
6694 the object to store it to, and the third is the address of the field of
6695 Obj to store to. If the runtime does not require a pointer to the
6696 object, Obj may be null.
6701 The '``llvm.gcwrite``' intrinsic has the same semantics as a store
6702 instruction, but may be replaced with substantially more complex code by
6703 the garbage collector runtime, as needed. The '``llvm.gcwrite``'
6704 intrinsic may only be used in a function which :ref:`specifies a GC
6707 Code Generator Intrinsics
6708 -------------------------
6710 These intrinsics are provided by LLVM to expose special features that
6711 may only be implemented with code generator support.
6713 '``llvm.returnaddress``' Intrinsic
6714 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6721 declare i8 *@llvm.returnaddress(i32 <level>)
6726 The '``llvm.returnaddress``' intrinsic attempts to compute a
6727 target-specific value indicating the return address of the current
6728 function or one of its callers.
6733 The argument to this intrinsic indicates which function to return the
6734 address for. Zero indicates the calling function, one indicates its
6735 caller, etc. The argument is **required** to be a constant integer
6741 The '``llvm.returnaddress``' intrinsic either returns a pointer
6742 indicating the return address of the specified call frame, or zero if it
6743 cannot be identified. The value returned by this intrinsic is likely to
6744 be incorrect or 0 for arguments other than zero, so it should only be
6745 used for debugging purposes.
6747 Note that calling this intrinsic does not prevent function inlining or
6748 other aggressive transformations, so the value returned may not be that
6749 of the obvious source-language caller.
6751 '``llvm.frameaddress``' Intrinsic
6752 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6759 declare i8* @llvm.frameaddress(i32 <level>)
6764 The '``llvm.frameaddress``' intrinsic attempts to return the
6765 target-specific frame pointer value for the specified stack frame.
6770 The argument to this intrinsic indicates which function to return the
6771 frame pointer for. Zero indicates the calling function, one indicates
6772 its caller, etc. The argument is **required** to be a constant integer
6778 The '``llvm.frameaddress``' intrinsic either returns a pointer
6779 indicating the frame address of the specified call frame, or zero if it
6780 cannot be identified. The value returned by this intrinsic is likely to
6781 be incorrect or 0 for arguments other than zero, so it should only be
6782 used for debugging purposes.
6784 Note that calling this intrinsic does not prevent function inlining or
6785 other aggressive transformations, so the value returned may not be that
6786 of the obvious source-language caller.
6790 '``llvm.stacksave``' Intrinsic
6791 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6798 declare i8* @llvm.stacksave()
6803 The '``llvm.stacksave``' intrinsic is used to remember the current state
6804 of the function stack, for use with
6805 :ref:`llvm.stackrestore <int_stackrestore>`. This is useful for
6806 implementing language features like scoped automatic variable sized
6812 This intrinsic returns a opaque pointer value that can be passed to
6813 :ref:`llvm.stackrestore <int_stackrestore>`. When an
6814 ``llvm.stackrestore`` intrinsic is executed with a value saved from
6815 ``llvm.stacksave``, it effectively restores the state of the stack to
6816 the state it was in when the ``llvm.stacksave`` intrinsic executed. In
6817 practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that
6818 were allocated after the ``llvm.stacksave`` was executed.
6820 .. _int_stackrestore:
6822 '``llvm.stackrestore``' Intrinsic
6823 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6830 declare void @llvm.stackrestore(i8* %ptr)
6835 The '``llvm.stackrestore``' intrinsic is used to restore the state of
6836 the function stack to the state it was in when the corresponding
6837 :ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is
6838 useful for implementing language features like scoped automatic variable
6839 sized arrays in C99.
6844 See the description for :ref:`llvm.stacksave <int_stacksave>`.
6846 '``llvm.prefetch``' Intrinsic
6847 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6854 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
6859 The '``llvm.prefetch``' intrinsic is a hint to the code generator to
6860 insert a prefetch instruction if supported; otherwise, it is a noop.
6861 Prefetches have no effect on the behavior of the program but can change
6862 its performance characteristics.
6867 ``address`` is the address to be prefetched, ``rw`` is the specifier
6868 determining if the fetch should be for a read (0) or write (1), and
6869 ``locality`` is a temporal locality specifier ranging from (0) - no
6870 locality, to (3) - extremely local keep in cache. The ``cache type``
6871 specifies whether the prefetch is performed on the data (1) or
6872 instruction (0) cache. The ``rw``, ``locality`` and ``cache type``
6873 arguments must be constant integers.
6878 This intrinsic does not modify the behavior of the program. In
6879 particular, prefetches cannot trap and do not produce a value. On
6880 targets that support this intrinsic, the prefetch can provide hints to
6881 the processor cache for better performance.
6883 '``llvm.pcmarker``' Intrinsic
6884 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6891 declare void @llvm.pcmarker(i32 <id>)
6896 The '``llvm.pcmarker``' intrinsic is a method to export a Program
6897 Counter (PC) in a region of code to simulators and other tools. The
6898 method is target specific, but it is expected that the marker will use
6899 exported symbols to transmit the PC of the marker. The marker makes no
6900 guarantees that it will remain with any specific instruction after
6901 optimizations. It is possible that the presence of a marker will inhibit
6902 optimizations. The intended use is to be inserted after optimizations to
6903 allow correlations of simulation runs.
6908 ``id`` is a numerical id identifying the marker.
6913 This intrinsic does not modify the behavior of the program. Backends
6914 that do not support this intrinsic may ignore it.
6916 '``llvm.readcyclecounter``' Intrinsic
6917 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6924 declare i64 @llvm.readcyclecounter()
6929 The '``llvm.readcyclecounter``' intrinsic provides access to the cycle
6930 counter register (or similar low latency, high accuracy clocks) on those
6931 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6932 should map to RPCC. As the backing counters overflow quickly (on the
6933 order of 9 seconds on alpha), this should only be used for small
6939 When directly supported, reading the cycle counter should not modify any
6940 memory. Implementations are allowed to either return a application
6941 specific value or a system wide value. On backends without support, this
6942 is lowered to a constant 0.
6944 Note that runtime support may be conditional on the privilege-level code is
6945 running at and the host platform.
6947 Standard C Library Intrinsics
6948 -----------------------------
6950 LLVM provides intrinsics for a few important standard C library
6951 functions. These intrinsics allow source-language front-ends to pass
6952 information about the alignment of the pointer arguments to the code
6953 generator, providing opportunity for more efficient code generation.
6957 '``llvm.memcpy``' Intrinsic
6958 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
6963 This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any
6964 integer bit width and for different address spaces. Not all targets
6965 support all bit widths however.
6969 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6970 i32 <len>, i32 <align>, i1 <isvolatile>)
6971 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6972 i64 <len>, i32 <align>, i1 <isvolatile>)
6977 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
6978 source location to the destination location.
6980 Note that, unlike the standard libc function, the ``llvm.memcpy.*``
6981 intrinsics do not return a value, takes extra alignment/isvolatile
6982 arguments and the pointers can be in specified address spaces.
6987 The first argument is a pointer to the destination, the second is a
6988 pointer to the source. The third argument is an integer argument
6989 specifying the number of bytes to copy, the fourth argument is the
6990 alignment of the source and destination locations, and the fifth is a
6991 boolean indicating a volatile access.
6993 If the call to this intrinsic has an alignment value that is not 0 or 1,
6994 then the caller guarantees that both the source and destination pointers
6995 are aligned to that boundary.
6997 If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is
6998 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
6999 very cleanly specified and it is unwise to depend on it.
7004 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
7005 source location to the destination location, which are not allowed to
7006 overlap. It copies "len" bytes of memory over. If the argument is known
7007 to be aligned to some boundary, this can be specified as the fourth
7008 argument, otherwise it should be set to 0 or 1 (both meaning no alignment).
7010 '``llvm.memmove``' Intrinsic
7011 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7016 This is an overloaded intrinsic. You can use llvm.memmove on any integer
7017 bit width and for different address space. Not all targets support all
7022 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
7023 i32 <len>, i32 <align>, i1 <isvolatile>)
7024 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
7025 i64 <len>, i32 <align>, i1 <isvolatile>)
7030 The '``llvm.memmove.*``' intrinsics move a block of memory from the
7031 source location to the destination location. It is similar to the
7032 '``llvm.memcpy``' intrinsic but allows the two memory locations to
7035 Note that, unlike the standard libc function, the ``llvm.memmove.*``
7036 intrinsics do not return a value, takes extra alignment/isvolatile
7037 arguments and the pointers can be in specified address spaces.
7042 The first argument is a pointer to the destination, the second is a
7043 pointer to the source. The third argument is an integer argument
7044 specifying the number of bytes to copy, the fourth argument is the
7045 alignment of the source and destination locations, and the fifth is a
7046 boolean indicating a volatile access.
7048 If the call to this intrinsic has an alignment value that is not 0 or 1,
7049 then the caller guarantees that the source and destination pointers are
7050 aligned to that boundary.
7052 If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call
7053 is a :ref:`volatile operation <volatile>`. The detailed access behavior is
7054 not very cleanly specified and it is unwise to depend on it.
7059 The '``llvm.memmove.*``' intrinsics copy a block of memory from the
7060 source location to the destination location, which may overlap. It
7061 copies "len" bytes of memory over. If the argument is known to be
7062 aligned to some boundary, this can be specified as the fourth argument,
7063 otherwise it should be set to 0 or 1 (both meaning no alignment).
7065 '``llvm.memset.*``' Intrinsics
7066 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7071 This is an overloaded intrinsic. You can use llvm.memset on any integer
7072 bit width and for different address spaces. However, not all targets
7073 support all bit widths.
7077 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
7078 i32 <len>, i32 <align>, i1 <isvolatile>)
7079 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
7080 i64 <len>, i32 <align>, i1 <isvolatile>)
7085 The '``llvm.memset.*``' intrinsics fill a block of memory with a
7086 particular byte value.
7088 Note that, unlike the standard libc function, the ``llvm.memset``
7089 intrinsic does not return a value and takes extra alignment/volatile
7090 arguments. Also, the destination can be in an arbitrary address space.
7095 The first argument is a pointer to the destination to fill, the second
7096 is the byte value with which to fill it, the third argument is an
7097 integer argument specifying the number of bytes to fill, and the fourth
7098 argument is the known alignment of the destination location.
7100 If the call to this intrinsic has an alignment value that is not 0 or 1,
7101 then the caller guarantees that the destination pointer is aligned to
7104 If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is
7105 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
7106 very cleanly specified and it is unwise to depend on it.
7111 The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting
7112 at the destination location. If the argument is known to be aligned to
7113 some boundary, this can be specified as the fourth argument, otherwise
7114 it should be set to 0 or 1 (both meaning no alignment).
7116 '``llvm.sqrt.*``' Intrinsic
7117 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7122 This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any
7123 floating point or vector of floating point type. Not all targets support
7128 declare float @llvm.sqrt.f32(float %Val)
7129 declare double @llvm.sqrt.f64(double %Val)
7130 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
7131 declare fp128 @llvm.sqrt.f128(fp128 %Val)
7132 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
7137 The '``llvm.sqrt``' intrinsics return the sqrt of the specified operand,
7138 returning the same value as the libm '``sqrt``' functions would. Unlike
7139 ``sqrt`` in libm, however, ``llvm.sqrt`` has undefined behavior for
7140 negative numbers other than -0.0 (which allows for better optimization,
7141 because there is no need to worry about errno being set).
7142 ``llvm.sqrt(-0.0)`` is defined to return -0.0 like IEEE sqrt.
7147 The argument and return value are floating point numbers of the same
7153 This function returns the sqrt of the specified operand if it is a
7154 nonnegative floating point number.
7156 '``llvm.powi.*``' Intrinsic
7157 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7162 This is an overloaded intrinsic. You can use ``llvm.powi`` on any
7163 floating point or vector of floating point type. Not all targets support
7168 declare float @llvm.powi.f32(float %Val, i32 %power)
7169 declare double @llvm.powi.f64(double %Val, i32 %power)
7170 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7171 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7172 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
7177 The '``llvm.powi.*``' intrinsics return the first operand raised to the
7178 specified (positive or negative) power. The order of evaluation of
7179 multiplications is not defined. When a vector of floating point type is
7180 used, the second argument remains a scalar integer value.
7185 The second argument is an integer power, and the first is a value to
7186 raise to that power.
7191 This function returns the first value raised to the second power with an
7192 unspecified sequence of rounding operations.
7194 '``llvm.sin.*``' Intrinsic
7195 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7200 This is an overloaded intrinsic. You can use ``llvm.sin`` on any
7201 floating point or vector of floating point type. Not all targets support
7206 declare float @llvm.sin.f32(float %Val)
7207 declare double @llvm.sin.f64(double %Val)
7208 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7209 declare fp128 @llvm.sin.f128(fp128 %Val)
7210 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7215 The '``llvm.sin.*``' intrinsics return the sine of the operand.
7220 The argument and return value are floating point numbers of the same
7226 This function returns the sine of the specified operand, returning the
7227 same values as the libm ``sin`` functions would, and handles error
7228 conditions in the same way.
7230 '``llvm.cos.*``' Intrinsic
7231 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7236 This is an overloaded intrinsic. You can use ``llvm.cos`` on any
7237 floating point or vector of floating point type. Not all targets support
7242 declare float @llvm.cos.f32(float %Val)
7243 declare double @llvm.cos.f64(double %Val)
7244 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7245 declare fp128 @llvm.cos.f128(fp128 %Val)
7246 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7251 The '``llvm.cos.*``' intrinsics return the cosine of the operand.
7256 The argument and return value are floating point numbers of the same
7262 This function returns the cosine of the specified operand, returning the
7263 same values as the libm ``cos`` functions would, and handles error
7264 conditions in the same way.
7266 '``llvm.pow.*``' Intrinsic
7267 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7272 This is an overloaded intrinsic. You can use ``llvm.pow`` on any
7273 floating point or vector of floating point type. Not all targets support
7278 declare float @llvm.pow.f32(float %Val, float %Power)
7279 declare double @llvm.pow.f64(double %Val, double %Power)
7280 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7281 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7282 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7287 The '``llvm.pow.*``' intrinsics return the first operand raised to the
7288 specified (positive or negative) power.
7293 The second argument is a floating point power, and the first is a value
7294 to raise to that power.
7299 This function returns the first value raised to the second power,
7300 returning the same values as the libm ``pow`` functions would, and
7301 handles error conditions in the same way.
7303 '``llvm.exp.*``' Intrinsic
7304 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7309 This is an overloaded intrinsic. You can use ``llvm.exp`` on any
7310 floating point or vector of floating point type. Not all targets support
7315 declare float @llvm.exp.f32(float %Val)
7316 declare double @llvm.exp.f64(double %Val)
7317 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7318 declare fp128 @llvm.exp.f128(fp128 %Val)
7319 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7324 The '``llvm.exp.*``' intrinsics perform the exp function.
7329 The argument and return value are floating point numbers of the same
7335 This function returns the same values as the libm ``exp`` functions
7336 would, and handles error conditions in the same way.
7338 '``llvm.exp2.*``' Intrinsic
7339 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7344 This is an overloaded intrinsic. You can use ``llvm.exp2`` on any
7345 floating point or vector of floating point type. Not all targets support
7350 declare float @llvm.exp2.f32(float %Val)
7351 declare double @llvm.exp2.f64(double %Val)
7352 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val)
7353 declare fp128 @llvm.exp2.f128(fp128 %Val)
7354 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val)
7359 The '``llvm.exp2.*``' intrinsics perform the exp2 function.
7364 The argument and return value are floating point numbers of the same
7370 This function returns the same values as the libm ``exp2`` functions
7371 would, and handles error conditions in the same way.
7373 '``llvm.log.*``' Intrinsic
7374 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7379 This is an overloaded intrinsic. You can use ``llvm.log`` on any
7380 floating point or vector of floating point type. Not all targets support
7385 declare float @llvm.log.f32(float %Val)
7386 declare double @llvm.log.f64(double %Val)
7387 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7388 declare fp128 @llvm.log.f128(fp128 %Val)
7389 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7394 The '``llvm.log.*``' intrinsics perform the log function.
7399 The argument and return value are floating point numbers of the same
7405 This function returns the same values as the libm ``log`` functions
7406 would, and handles error conditions in the same way.
7408 '``llvm.log10.*``' Intrinsic
7409 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7414 This is an overloaded intrinsic. You can use ``llvm.log10`` on any
7415 floating point or vector of floating point type. Not all targets support
7420 declare float @llvm.log10.f32(float %Val)
7421 declare double @llvm.log10.f64(double %Val)
7422 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val)
7423 declare fp128 @llvm.log10.f128(fp128 %Val)
7424 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val)
7429 The '``llvm.log10.*``' intrinsics perform the log10 function.
7434 The argument and return value are floating point numbers of the same
7440 This function returns the same values as the libm ``log10`` functions
7441 would, and handles error conditions in the same way.
7443 '``llvm.log2.*``' Intrinsic
7444 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7449 This is an overloaded intrinsic. You can use ``llvm.log2`` on any
7450 floating point or vector of floating point type. Not all targets support
7455 declare float @llvm.log2.f32(float %Val)
7456 declare double @llvm.log2.f64(double %Val)
7457 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val)
7458 declare fp128 @llvm.log2.f128(fp128 %Val)
7459 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val)
7464 The '``llvm.log2.*``' intrinsics perform the log2 function.
7469 The argument and return value are floating point numbers of the same
7475 This function returns the same values as the libm ``log2`` functions
7476 would, and handles error conditions in the same way.
7478 '``llvm.fma.*``' Intrinsic
7479 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7484 This is an overloaded intrinsic. You can use ``llvm.fma`` on any
7485 floating point or vector of floating point type. Not all targets support
7490 declare float @llvm.fma.f32(float %a, float %b, float %c)
7491 declare double @llvm.fma.f64(double %a, double %b, double %c)
7492 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7493 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7494 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7499 The '``llvm.fma.*``' intrinsics perform the fused multiply-add
7505 The argument and return value are floating point numbers of the same
7511 This function returns the same values as the libm ``fma`` functions
7512 would, and does not set errno.
7514 '``llvm.fabs.*``' Intrinsic
7515 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7520 This is an overloaded intrinsic. You can use ``llvm.fabs`` on any
7521 floating point or vector of floating point type. Not all targets support
7526 declare float @llvm.fabs.f32(float %Val)
7527 declare double @llvm.fabs.f64(double %Val)
7528 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val)
7529 declare fp128 @llvm.fabs.f128(fp128 %Val)
7530 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val)
7535 The '``llvm.fabs.*``' intrinsics return the absolute value of the
7541 The argument and return value are floating point numbers of the same
7547 This function returns the same values as the libm ``fabs`` functions
7548 would, and handles error conditions in the same way.
7550 '``llvm.copysign.*``' Intrinsic
7551 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7556 This is an overloaded intrinsic. You can use ``llvm.copysign`` on any
7557 floating point or vector of floating point type. Not all targets support
7562 declare float @llvm.copysign.f32(float %Mag, float %Sgn)
7563 declare double @llvm.copysign.f64(double %Mag, double %Sgn)
7564 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn)
7565 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn)
7566 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn)
7571 The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the
7572 first operand and the sign of the second operand.
7577 The arguments and return value are floating point numbers of the same
7583 This function returns the same values as the libm ``copysign``
7584 functions would, and handles error conditions in the same way.
7586 '``llvm.floor.*``' Intrinsic
7587 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7592 This is an overloaded intrinsic. You can use ``llvm.floor`` on any
7593 floating point or vector of floating point type. Not all targets support
7598 declare float @llvm.floor.f32(float %Val)
7599 declare double @llvm.floor.f64(double %Val)
7600 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val)
7601 declare fp128 @llvm.floor.f128(fp128 %Val)
7602 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val)
7607 The '``llvm.floor.*``' intrinsics return the floor of the operand.
7612 The argument and return value are floating point numbers of the same
7618 This function returns the same values as the libm ``floor`` functions
7619 would, and handles error conditions in the same way.
7621 '``llvm.ceil.*``' Intrinsic
7622 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7627 This is an overloaded intrinsic. You can use ``llvm.ceil`` on any
7628 floating point or vector of floating point type. Not all targets support
7633 declare float @llvm.ceil.f32(float %Val)
7634 declare double @llvm.ceil.f64(double %Val)
7635 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val)
7636 declare fp128 @llvm.ceil.f128(fp128 %Val)
7637 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val)
7642 The '``llvm.ceil.*``' intrinsics return the ceiling of the operand.
7647 The argument and return value are floating point numbers of the same
7653 This function returns the same values as the libm ``ceil`` functions
7654 would, and handles error conditions in the same way.
7656 '``llvm.trunc.*``' Intrinsic
7657 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7662 This is an overloaded intrinsic. You can use ``llvm.trunc`` on any
7663 floating point or vector of floating point type. Not all targets support
7668 declare float @llvm.trunc.f32(float %Val)
7669 declare double @llvm.trunc.f64(double %Val)
7670 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val)
7671 declare fp128 @llvm.trunc.f128(fp128 %Val)
7672 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val)
7677 The '``llvm.trunc.*``' intrinsics returns the operand rounded to the
7678 nearest integer not larger in magnitude than the operand.
7683 The argument and return value are floating point numbers of the same
7689 This function returns the same values as the libm ``trunc`` functions
7690 would, and handles error conditions in the same way.
7692 '``llvm.rint.*``' Intrinsic
7693 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7698 This is an overloaded intrinsic. You can use ``llvm.rint`` on any
7699 floating point or vector of floating point type. Not all targets support
7704 declare float @llvm.rint.f32(float %Val)
7705 declare double @llvm.rint.f64(double %Val)
7706 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val)
7707 declare fp128 @llvm.rint.f128(fp128 %Val)
7708 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val)
7713 The '``llvm.rint.*``' intrinsics returns the operand rounded to the
7714 nearest integer. It may raise an inexact floating-point exception if the
7715 operand isn't an integer.
7720 The argument and return value are floating point numbers of the same
7726 This function returns the same values as the libm ``rint`` functions
7727 would, and handles error conditions in the same way.
7729 '``llvm.nearbyint.*``' Intrinsic
7730 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7735 This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any
7736 floating point or vector of floating point type. Not all targets support
7741 declare float @llvm.nearbyint.f32(float %Val)
7742 declare double @llvm.nearbyint.f64(double %Val)
7743 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val)
7744 declare fp128 @llvm.nearbyint.f128(fp128 %Val)
7745 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val)
7750 The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the
7756 The argument and return value are floating point numbers of the same
7762 This function returns the same values as the libm ``nearbyint``
7763 functions would, and handles error conditions in the same way.
7765 '``llvm.round.*``' Intrinsic
7766 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7771 This is an overloaded intrinsic. You can use ``llvm.round`` on any
7772 floating point or vector of floating point type. Not all targets support
7777 declare float @llvm.round.f32(float %Val)
7778 declare double @llvm.round.f64(double %Val)
7779 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val)
7780 declare fp128 @llvm.round.f128(fp128 %Val)
7781 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val)
7786 The '``llvm.round.*``' intrinsics returns the operand rounded to the
7792 The argument and return value are floating point numbers of the same
7798 This function returns the same values as the libm ``round``
7799 functions would, and handles error conditions in the same way.
7801 Bit Manipulation Intrinsics
7802 ---------------------------
7804 LLVM provides intrinsics for a few important bit manipulation
7805 operations. These allow efficient code generation for some algorithms.
7807 '``llvm.bswap.*``' Intrinsics
7808 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7813 This is an overloaded intrinsic function. You can use bswap on any
7814 integer type that is an even number of bytes (i.e. BitWidth % 16 == 0).
7818 declare i16 @llvm.bswap.i16(i16 <id>)
7819 declare i32 @llvm.bswap.i32(i32 <id>)
7820 declare i64 @llvm.bswap.i64(i64 <id>)
7825 The '``llvm.bswap``' family of intrinsics is used to byte swap integer
7826 values with an even number of bytes (positive multiple of 16 bits).
7827 These are useful for performing operations on data that is not in the
7828 target's native byte order.
7833 The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high
7834 and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32``
7835 intrinsic returns an i32 value that has the four bytes of the input i32
7836 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the
7837 returned i32 will have its bytes in 3, 2, 1, 0 order. The
7838 ``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this
7839 concept to additional even-byte lengths (6 bytes, 8 bytes and more,
7842 '``llvm.ctpop.*``' Intrinsic
7843 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7848 This is an overloaded intrinsic. You can use llvm.ctpop on any integer
7849 bit width, or on any vector with integer elements. Not all targets
7850 support all bit widths or vector types, however.
7854 declare i8 @llvm.ctpop.i8(i8 <src>)
7855 declare i16 @llvm.ctpop.i16(i16 <src>)
7856 declare i32 @llvm.ctpop.i32(i32 <src>)
7857 declare i64 @llvm.ctpop.i64(i64 <src>)
7858 declare i256 @llvm.ctpop.i256(i256 <src>)
7859 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
7864 The '``llvm.ctpop``' family of intrinsics counts the number of bits set
7870 The only argument is the value to be counted. The argument may be of any
7871 integer type, or a vector with integer elements. The return type must
7872 match the argument type.
7877 The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within
7878 each element of a vector.
7880 '``llvm.ctlz.*``' Intrinsic
7881 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7886 This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any
7887 integer bit width, or any vector whose elements are integers. Not all
7888 targets support all bit widths or vector types, however.
7892 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>)
7893 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>)
7894 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>)
7895 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>)
7896 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>)
7897 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
7902 The '``llvm.ctlz``' family of intrinsic functions counts the number of
7903 leading zeros in a variable.
7908 The first argument is the value to be counted. This argument may be of
7909 any integer type, or a vectory with integer element type. The return
7910 type must match the first argument type.
7912 The second argument must be a constant and is a flag to indicate whether
7913 the intrinsic should ensure that a zero as the first argument produces a
7914 defined result. Historically some architectures did not provide a
7915 defined result for zero values as efficiently, and many algorithms are
7916 now predicated on avoiding zero-value inputs.
7921 The '``llvm.ctlz``' intrinsic counts the leading (most significant)
7922 zeros in a variable, or within each element of the vector. If
7923 ``src == 0`` then the result is the size in bits of the type of ``src``
7924 if ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
7925 ``llvm.ctlz(i32 2) = 30``.
7927 '``llvm.cttz.*``' Intrinsic
7928 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7933 This is an overloaded intrinsic. You can use ``llvm.cttz`` on any
7934 integer bit width, or any vector of integer elements. Not all targets
7935 support all bit widths or vector types, however.
7939 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>)
7940 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>)
7941 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>)
7942 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>)
7943 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>)
7944 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
7949 The '``llvm.cttz``' family of intrinsic functions counts the number of
7955 The first argument is the value to be counted. This argument may be of
7956 any integer type, or a vectory with integer element type. The return
7957 type must match the first argument type.
7959 The second argument must be a constant and is a flag to indicate whether
7960 the intrinsic should ensure that a zero as the first argument produces a
7961 defined result. Historically some architectures did not provide a
7962 defined result for zero values as efficiently, and many algorithms are
7963 now predicated on avoiding zero-value inputs.
7968 The '``llvm.cttz``' intrinsic counts the trailing (least significant)
7969 zeros in a variable, or within each element of a vector. If ``src == 0``
7970 then the result is the size in bits of the type of ``src`` if
7971 ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
7972 ``llvm.cttz(2) = 1``.
7974 Arithmetic with Overflow Intrinsics
7975 -----------------------------------
7977 LLVM provides intrinsics for some arithmetic with overflow operations.
7979 '``llvm.sadd.with.overflow.*``' Intrinsics
7980 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7985 This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow``
7986 on any integer bit width.
7990 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7991 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7992 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7997 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
7998 a signed addition of the two arguments, and indicate whether an overflow
7999 occurred during the signed summation.
8004 The arguments (%a and %b) and the first element of the result structure
8005 may be of integer types of any bit width, but they must have the same
8006 bit width. The second element of the result structure must be of type
8007 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8013 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
8014 a signed addition of the two variables. They return a structure --- the
8015 first element of which is the signed summation, and the second element
8016 of which is a bit specifying if the signed summation resulted in an
8022 .. code-block:: llvm
8024 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
8025 %sum = extractvalue {i32, i1} %res, 0
8026 %obit = extractvalue {i32, i1} %res, 1
8027 br i1 %obit, label %overflow, label %normal
8029 '``llvm.uadd.with.overflow.*``' Intrinsics
8030 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8035 This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow``
8036 on any integer bit width.
8040 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
8041 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
8042 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
8047 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
8048 an unsigned addition of the two arguments, and indicate whether a carry
8049 occurred during the unsigned summation.
8054 The arguments (%a and %b) and the first element of the result structure
8055 may be of integer types of any bit width, but they must have the same
8056 bit width. The second element of the result structure must be of type
8057 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
8063 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
8064 an unsigned addition of the two arguments. They return a structure --- the
8065 first element of which is the sum, and the second element of which is a
8066 bit specifying if the unsigned summation resulted in a carry.
8071 .. code-block:: llvm
8073 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
8074 %sum = extractvalue {i32, i1} %res, 0
8075 %obit = extractvalue {i32, i1} %res, 1
8076 br i1 %obit, label %carry, label %normal
8078 '``llvm.ssub.with.overflow.*``' Intrinsics
8079 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8084 This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow``
8085 on any integer bit width.
8089 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
8090 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
8091 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
8096 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
8097 a signed subtraction of the two arguments, and indicate whether an
8098 overflow occurred during the signed subtraction.
8103 The arguments (%a and %b) and the first element of the result structure
8104 may be of integer types of any bit width, but they must have the same
8105 bit width. The second element of the result structure must be of type
8106 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8112 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
8113 a signed subtraction of the two arguments. They return a structure --- the
8114 first element of which is the subtraction, and the second element of
8115 which is a bit specifying if the signed subtraction resulted in an
8121 .. code-block:: llvm
8123 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
8124 %sum = extractvalue {i32, i1} %res, 0
8125 %obit = extractvalue {i32, i1} %res, 1
8126 br i1 %obit, label %overflow, label %normal
8128 '``llvm.usub.with.overflow.*``' Intrinsics
8129 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8134 This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow``
8135 on any integer bit width.
8139 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
8140 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
8141 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
8146 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
8147 an unsigned subtraction of the two arguments, and indicate whether an
8148 overflow occurred during the unsigned subtraction.
8153 The arguments (%a and %b) and the first element of the result structure
8154 may be of integer types of any bit width, but they must have the same
8155 bit width. The second element of the result structure must be of type
8156 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
8162 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
8163 an unsigned subtraction of the two arguments. They return a structure ---
8164 the first element of which is the subtraction, and the second element of
8165 which is a bit specifying if the unsigned subtraction resulted in an
8171 .. code-block:: llvm
8173 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
8174 %sum = extractvalue {i32, i1} %res, 0
8175 %obit = extractvalue {i32, i1} %res, 1
8176 br i1 %obit, label %overflow, label %normal
8178 '``llvm.smul.with.overflow.*``' Intrinsics
8179 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8184 This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow``
8185 on any integer bit width.
8189 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
8190 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
8191 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
8196 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
8197 a signed multiplication of the two arguments, and indicate whether an
8198 overflow occurred during the signed multiplication.
8203 The arguments (%a and %b) and the first element of the result structure
8204 may be of integer types of any bit width, but they must have the same
8205 bit width. The second element of the result structure must be of type
8206 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8212 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
8213 a signed multiplication of the two arguments. They return a structure ---
8214 the first element of which is the multiplication, and the second element
8215 of which is a bit specifying if the signed multiplication resulted in an
8221 .. code-block:: llvm
8223 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
8224 %sum = extractvalue {i32, i1} %res, 0
8225 %obit = extractvalue {i32, i1} %res, 1
8226 br i1 %obit, label %overflow, label %normal
8228 '``llvm.umul.with.overflow.*``' Intrinsics
8229 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8234 This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow``
8235 on any integer bit width.
8239 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
8240 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
8241 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
8246 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
8247 a unsigned multiplication of the two arguments, and indicate whether an
8248 overflow occurred during the unsigned multiplication.
8253 The arguments (%a and %b) and the first element of the result structure
8254 may be of integer types of any bit width, but they must have the same
8255 bit width. The second element of the result structure must be of type
8256 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
8262 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
8263 an unsigned multiplication of the two arguments. They return a structure ---
8264 the first element of which is the multiplication, and the second
8265 element of which is a bit specifying if the unsigned multiplication
8266 resulted in an overflow.
8271 .. code-block:: llvm
8273 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
8274 %sum = extractvalue {i32, i1} %res, 0
8275 %obit = extractvalue {i32, i1} %res, 1
8276 br i1 %obit, label %overflow, label %normal
8278 Specialised Arithmetic Intrinsics
8279 ---------------------------------
8281 '``llvm.fmuladd.*``' Intrinsic
8282 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8289 declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
8290 declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
8295 The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add
8296 expressions that can be fused if the code generator determines that (a) the
8297 target instruction set has support for a fused operation, and (b) that the
8298 fused operation is more efficient than the equivalent, separate pair of mul
8299 and add instructions.
8304 The '``llvm.fmuladd.*``' intrinsics each take three arguments: two
8305 multiplicands, a and b, and an addend c.
8314 %0 = call float @llvm.fmuladd.f32(%a, %b, %c)
8316 is equivalent to the expression a \* b + c, except that rounding will
8317 not be performed between the multiplication and addition steps if the
8318 code generator fuses the operations. Fusion is not guaranteed, even if
8319 the target platform supports it. If a fused multiply-add is required the
8320 corresponding llvm.fma.\* intrinsic function should be used
8321 instead. This never sets errno, just as '``llvm.fma.*``'.
8326 .. code-block:: llvm
8328 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields {float}:r2 = (a * b) + c
8330 Half Precision Floating Point Intrinsics
8331 ----------------------------------------
8333 For most target platforms, half precision floating point is a
8334 storage-only format. This means that it is a dense encoding (in memory)
8335 but does not support computation in the format.
8337 This means that code must first load the half-precision floating point
8338 value as an i16, then convert it to float with
8339 :ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can
8340 then be performed on the float value (including extending to double
8341 etc). To store the value back to memory, it is first converted to float
8342 if needed, then converted to i16 with
8343 :ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an
8346 .. _int_convert_to_fp16:
8348 '``llvm.convert.to.fp16``' Intrinsic
8349 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8356 declare i16 @llvm.convert.to.fp16(f32 %a)
8361 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion
8362 from single precision floating point format to half precision floating
8368 The intrinsic function contains single argument - the value to be
8374 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion
8375 from single precision floating point format to half precision floating
8376 point format. The return value is an ``i16`` which contains the
8382 .. code-block:: llvm
8384 %res = call i16 @llvm.convert.to.fp16(f32 %a)
8385 store i16 %res, i16* @x, align 2
8387 .. _int_convert_from_fp16:
8389 '``llvm.convert.from.fp16``' Intrinsic
8390 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8397 declare f32 @llvm.convert.from.fp16(i16 %a)
8402 The '``llvm.convert.from.fp16``' intrinsic function performs a
8403 conversion from half precision floating point format to single precision
8404 floating point format.
8409 The intrinsic function contains single argument - the value to be
8415 The '``llvm.convert.from.fp16``' intrinsic function performs a
8416 conversion from half single precision floating point format to single
8417 precision floating point format. The input half-float value is
8418 represented by an ``i16`` value.
8423 .. code-block:: llvm
8425 %a = load i16* @x, align 2
8426 %res = call f32 @llvm.convert.from.fp16(i16 %a)
8431 The LLVM debugger intrinsics (which all start with ``llvm.dbg.``
8432 prefix), are described in the `LLVM Source Level
8433 Debugging <SourceLevelDebugging.html#format_common_intrinsics>`_
8436 Exception Handling Intrinsics
8437 -----------------------------
8439 The LLVM exception handling intrinsics (which all start with
8440 ``llvm.eh.`` prefix), are described in the `LLVM Exception
8441 Handling <ExceptionHandling.html#format_common_intrinsics>`_ document.
8445 Trampoline Intrinsics
8446 ---------------------
8448 These intrinsics make it possible to excise one parameter, marked with
8449 the :ref:`nest <nest>` attribute, from a function. The result is a
8450 callable function pointer lacking the nest parameter - the caller does
8451 not need to provide a value for it. Instead, the value to use is stored
8452 in advance in a "trampoline", a block of memory usually allocated on the
8453 stack, which also contains code to splice the nest value into the
8454 argument list. This is used to implement the GCC nested function address
8457 For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)``
8458 then the resulting function pointer has signature ``i32 (i32, i32)*``.
8459 It can be created as follows:
8461 .. code-block:: llvm
8463 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
8464 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
8465 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
8466 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
8467 %fp = bitcast i8* %p to i32 (i32, i32)*
8469 The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to
8470 ``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``.
8474 '``llvm.init.trampoline``' Intrinsic
8475 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8482 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
8487 This fills the memory pointed to by ``tramp`` with executable code,
8488 turning it into a trampoline.
8493 The ``llvm.init.trampoline`` intrinsic takes three arguments, all
8494 pointers. The ``tramp`` argument must point to a sufficiently large and
8495 sufficiently aligned block of memory; this memory is written to by the
8496 intrinsic. Note that the size and the alignment are target-specific -
8497 LLVM currently provides no portable way of determining them, so a
8498 front-end that generates this intrinsic needs to have some
8499 target-specific knowledge. The ``func`` argument must hold a function
8500 bitcast to an ``i8*``.
8505 The block of memory pointed to by ``tramp`` is filled with target
8506 dependent code, turning it into a function. Then ``tramp`` needs to be
8507 passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can
8508 be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new
8509 function's signature is the same as that of ``func`` with any arguments
8510 marked with the ``nest`` attribute removed. At most one such ``nest``
8511 argument is allowed, and it must be of pointer type. Calling the new
8512 function is equivalent to calling ``func`` with the same argument list,
8513 but with ``nval`` used for the missing ``nest`` argument. If, after
8514 calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is
8515 modified, then the effect of any later call to the returned function
8516 pointer is undefined.
8520 '``llvm.adjust.trampoline``' Intrinsic
8521 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8528 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
8533 This performs any required machine-specific adjustment to the address of
8534 a trampoline (passed as ``tramp``).
8539 ``tramp`` must point to a block of memory which already has trampoline
8540 code filled in by a previous call to
8541 :ref:`llvm.init.trampoline <int_it>`.
8546 On some architectures the address of the code to be executed needs to be
8547 different to the address where the trampoline is actually stored. This
8548 intrinsic returns the executable address corresponding to ``tramp``
8549 after performing the required machine specific adjustments. The pointer
8550 returned can then be :ref:`bitcast and executed <int_trampoline>`.
8555 This class of intrinsics exists to information about the lifetime of
8556 memory objects and ranges where variables are immutable.
8560 '``llvm.lifetime.start``' Intrinsic
8561 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8568 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
8573 The '``llvm.lifetime.start``' intrinsic specifies the start of a memory
8579 The first argument is a constant integer representing the size of the
8580 object, or -1 if it is variable sized. The second argument is a pointer
8586 This intrinsic indicates that before this point in the code, the value
8587 of the memory pointed to by ``ptr`` is dead. This means that it is known
8588 to never be used and has an undefined value. A load from the pointer
8589 that precedes this intrinsic can be replaced with ``'undef'``.
8593 '``llvm.lifetime.end``' Intrinsic
8594 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8601 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
8606 The '``llvm.lifetime.end``' intrinsic specifies the end of a memory
8612 The first argument is a constant integer representing the size of the
8613 object, or -1 if it is variable sized. The second argument is a pointer
8619 This intrinsic indicates that after this point in the code, the value of
8620 the memory pointed to by ``ptr`` is dead. This means that it is known to
8621 never be used and has an undefined value. Any stores into the memory
8622 object following this intrinsic may be removed as dead.
8624 '``llvm.invariant.start``' Intrinsic
8625 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8632 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
8637 The '``llvm.invariant.start``' intrinsic specifies that the contents of
8638 a memory object will not change.
8643 The first argument is a constant integer representing the size of the
8644 object, or -1 if it is variable sized. The second argument is a pointer
8650 This intrinsic indicates that until an ``llvm.invariant.end`` that uses
8651 the return value, the referenced memory location is constant and
8654 '``llvm.invariant.end``' Intrinsic
8655 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8662 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
8667 The '``llvm.invariant.end``' intrinsic specifies that the contents of a
8668 memory object are mutable.
8673 The first argument is the matching ``llvm.invariant.start`` intrinsic.
8674 The second argument is a constant integer representing the size of the
8675 object, or -1 if it is variable sized and the third argument is a
8676 pointer to the object.
8681 This intrinsic indicates that the memory is mutable again.
8686 This class of intrinsics is designed to be generic and has no specific
8689 '``llvm.var.annotation``' Intrinsic
8690 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8697 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
8702 The '``llvm.var.annotation``' intrinsic.
8707 The first argument is a pointer to a value, the second is a pointer to a
8708 global string, the third is a pointer to a global string which is the
8709 source file name, and the last argument is the line number.
8714 This intrinsic allows annotation of local variables with arbitrary
8715 strings. This can be useful for special purpose optimizations that want
8716 to look for these annotations. These have no other defined use; they are
8717 ignored by code generation and optimization.
8719 '``llvm.ptr.annotation.*``' Intrinsic
8720 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8725 This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a
8726 pointer to an integer of any width. *NOTE* you must specify an address space for
8727 the pointer. The identifier for the default address space is the integer
8732 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
8733 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>)
8734 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>)
8735 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>)
8736 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>)
8741 The '``llvm.ptr.annotation``' intrinsic.
8746 The first argument is a pointer to an integer value of arbitrary bitwidth
8747 (result of some expression), the second is a pointer to a global string, the
8748 third is a pointer to a global string which is the source file name, and the
8749 last argument is the line number. It returns the value of the first argument.
8754 This intrinsic allows annotation of a pointer to an integer with arbitrary
8755 strings. This can be useful for special purpose optimizations that want to look
8756 for these annotations. These have no other defined use; they are ignored by code
8757 generation and optimization.
8759 '``llvm.annotation.*``' Intrinsic
8760 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8765 This is an overloaded intrinsic. You can use '``llvm.annotation``' on
8766 any integer bit width.
8770 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
8771 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
8772 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
8773 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
8774 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
8779 The '``llvm.annotation``' intrinsic.
8784 The first argument is an integer value (result of some expression), the
8785 second is a pointer to a global string, the third is a pointer to a
8786 global string which is the source file name, and the last argument is
8787 the line number. It returns the value of the first argument.
8792 This intrinsic allows annotations to be put on arbitrary expressions
8793 with arbitrary strings. This can be useful for special purpose
8794 optimizations that want to look for these annotations. These have no
8795 other defined use; they are ignored by code generation and optimization.
8797 '``llvm.trap``' Intrinsic
8798 ^^^^^^^^^^^^^^^^^^^^^^^^^
8805 declare void @llvm.trap() noreturn nounwind
8810 The '``llvm.trap``' intrinsic.
8820 This intrinsic is lowered to the target dependent trap instruction. If
8821 the target does not have a trap instruction, this intrinsic will be
8822 lowered to a call of the ``abort()`` function.
8824 '``llvm.debugtrap``' Intrinsic
8825 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8832 declare void @llvm.debugtrap() nounwind
8837 The '``llvm.debugtrap``' intrinsic.
8847 This intrinsic is lowered to code which is intended to cause an
8848 execution trap with the intention of requesting the attention of a
8851 '``llvm.stackprotector``' Intrinsic
8852 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8859 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
8864 The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it
8865 onto the stack at ``slot``. The stack slot is adjusted to ensure that it
8866 is placed on the stack before local variables.
8871 The ``llvm.stackprotector`` intrinsic requires two pointer arguments.
8872 The first argument is the value loaded from the stack guard
8873 ``@__stack_chk_guard``. The second variable is an ``alloca`` that has
8874 enough space to hold the value of the guard.
8879 This intrinsic causes the prologue/epilogue inserter to force the position of
8880 the ``AllocaInst`` stack slot to be before local variables on the stack. This is
8881 to ensure that if a local variable on the stack is overwritten, it will destroy
8882 the value of the guard. When the function exits, the guard on the stack is
8883 checked against the original guard by ``llvm.stackprotectorcheck``. If they are
8884 different, then ``llvm.stackprotectorcheck`` causes the program to abort by
8885 calling the ``__stack_chk_fail()`` function.
8887 '``llvm.stackprotectorcheck``' Intrinsic
8888 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8895 declare void @llvm.stackprotectorcheck(i8** <guard>)
8900 The ``llvm.stackprotectorcheck`` intrinsic compares ``guard`` against an already
8901 created stack protector and if they are not equal calls the
8902 ``__stack_chk_fail()`` function.
8907 The ``llvm.stackprotectorcheck`` intrinsic requires one pointer argument, the
8908 the variable ``@__stack_chk_guard``.
8913 This intrinsic is provided to perform the stack protector check by comparing
8914 ``guard`` with the stack slot created by ``llvm.stackprotector`` and if the
8915 values do not match call the ``__stack_chk_fail()`` function.
8917 The reason to provide this as an IR level intrinsic instead of implementing it
8918 via other IR operations is that in order to perform this operation at the IR
8919 level without an intrinsic, one would need to create additional basic blocks to
8920 handle the success/failure cases. This makes it difficult to stop the stack
8921 protector check from disrupting sibling tail calls in Codegen. With this
8922 intrinsic, we are able to generate the stack protector basic blocks late in
8923 codegen after the tail call decision has occurred.
8925 '``llvm.objectsize``' Intrinsic
8926 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8933 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>)
8934 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>)
8939 The ``llvm.objectsize`` intrinsic is designed to provide information to
8940 the optimizers to determine at compile time whether a) an operation
8941 (like memcpy) will overflow a buffer that corresponds to an object, or
8942 b) that a runtime check for overflow isn't necessary. An object in this
8943 context means an allocation of a specific class, structure, array, or
8949 The ``llvm.objectsize`` intrinsic takes two arguments. The first
8950 argument is a pointer to or into the ``object``. The second argument is
8951 a boolean and determines whether ``llvm.objectsize`` returns 0 (if true)
8952 or -1 (if false) when the object size is unknown. The second argument
8953 only accepts constants.
8958 The ``llvm.objectsize`` intrinsic is lowered to a constant representing
8959 the size of the object concerned. If the size cannot be determined at
8960 compile time, ``llvm.objectsize`` returns ``i32/i64 -1 or 0`` (depending
8961 on the ``min`` argument).
8963 '``llvm.expect``' Intrinsic
8964 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8969 This is an overloaded intrinsic. You can use ``llvm.expect`` on any
8974 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>)
8975 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>)
8976 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>)
8981 The ``llvm.expect`` intrinsic provides information about expected (the
8982 most probable) value of ``val``, which can be used by optimizers.
8987 The ``llvm.expect`` intrinsic takes two arguments. The first argument is
8988 a value. The second argument is an expected value, this needs to be a
8989 constant value, variables are not allowed.
8994 This intrinsic is lowered to the ``val``.
8996 '``llvm.donothing``' Intrinsic
8997 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9004 declare void @llvm.donothing() nounwind readnone
9009 The ``llvm.donothing`` intrinsic doesn't perform any operation. It's the
9010 only intrinsic that can be called with an invoke instruction.
9020 This intrinsic does nothing, and it's removed by optimizers and ignored
9023 Stack Map Intrinsics
9024 --------------------
9026 LLVM provides experimental intrinsics to support runtime patching
9027 mechanisms commonly desired in dynamic language JITs. These intrinsics
9028 are described in :doc:`StackMaps`.