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 that 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. The ``"\01"`` prefix
83 can be used on global variables to suppress mangling.
84 #. Unnamed values are represented as an unsigned numeric value with
85 their prefix. For example, ``%12``, ``@2``, ``%44``.
86 #. Constants, which are described in the section Constants_ below.
88 LLVM requires that values start with a prefix for two reasons: Compilers
89 don't need to worry about name clashes with reserved words, and the set
90 of reserved words may be expanded in the future without penalty.
91 Additionally, unnamed identifiers allow a compiler to quickly come up
92 with a temporary variable without having to avoid symbol table
95 Reserved words in LLVM are very similar to reserved words in other
96 languages. There are keywords for different opcodes ('``add``',
97 '``bitcast``', '``ret``', etc...), for primitive type names ('``void``',
98 '``i32``', etc...), and others. These reserved words cannot conflict
99 with variable names, because none of them start with a prefix character
100 (``'%'`` or ``'@'``).
102 Here is an example of LLVM code to multiply the integer variable
109 %result = mul i32 %X, 8
111 After strength reduction:
115 %result = shl i32 %X, 3
121 %0 = add i32 %X, %X ; yields i32:%0
122 %1 = add i32 %0, %0 ; yields i32:%1
123 %result = add i32 %1, %1
125 This last way of multiplying ``%X`` by 8 illustrates several important
126 lexical features of LLVM:
128 #. Comments are delimited with a '``;``' and go until the end of line.
129 #. Unnamed temporaries are created when the result of a computation is
130 not assigned to a named value.
131 #. Unnamed temporaries are numbered sequentially (using a per-function
132 incrementing counter, starting with 0). Note that basic blocks and unnamed
133 function parameters are included in this numbering. For example, if the
134 entry basic block is not given a label name and all function parameters are
135 named, then it will get number 0.
137 It also shows a convention that we follow in this document. When
138 demonstrating instructions, we will follow an instruction with a comment
139 that defines the type and name of value produced.
147 LLVM programs are composed of ``Module``'s, each of which is a
148 translation unit of the input programs. Each module consists of
149 functions, global variables, and symbol table entries. Modules may be
150 combined together with the LLVM linker, which merges function (and
151 global variable) definitions, resolves forward declarations, and merges
152 symbol table entries. Here is an example of the "hello world" module:
156 ; Declare the string constant as a global constant.
157 @.str = private unnamed_addr constant [13 x i8] c"hello world\0A\00"
159 ; External declaration of the puts function
160 declare i32 @puts(i8* nocapture) nounwind
162 ; Definition of main function
163 define i32 @main() { ; i32()*
164 ; Convert [13 x i8]* to i8 *...
165 %cast210 = getelementptr [13 x i8]* @.str, i64 0, i64 0
167 ; Call puts function to write out the string to stdout.
168 call i32 @puts(i8* %cast210)
173 !0 = !{i32 42, null, !"string"}
176 This example is made up of a :ref:`global variable <globalvars>` named
177 "``.str``", an external declaration of the "``puts``" function, a
178 :ref:`function definition <functionstructure>` for "``main``" and
179 :ref:`named metadata <namedmetadatastructure>` "``foo``".
181 In general, a module is made up of a list of global values (where both
182 functions and global variables are global values). Global values are
183 represented by a pointer to a memory location (in this case, a pointer
184 to an array of char, and a pointer to a function), and have one of the
185 following :ref:`linkage types <linkage>`.
192 All Global Variables and Functions have one of the following types of
196 Global values with "``private``" linkage are only directly
197 accessible by objects in the current module. In particular, linking
198 code into a module with an private global value may cause the
199 private to be renamed as necessary to avoid collisions. Because the
200 symbol is private to the module, all references can be updated. This
201 doesn't show up in any symbol table in the object file.
203 Similar to private, but the value shows as a local symbol
204 (``STB_LOCAL`` in the case of ELF) in the object file. This
205 corresponds to the notion of the '``static``' keyword in C.
206 ``available_externally``
207 Globals with "``available_externally``" linkage are never emitted
208 into the object file corresponding to the LLVM module. They exist to
209 allow inlining and other optimizations to take place given knowledge
210 of the definition of the global, which is known to be somewhere
211 outside the module. Globals with ``available_externally`` linkage
212 are allowed to be discarded at will, and are otherwise the same as
213 ``linkonce_odr``. This linkage type is only allowed on definitions,
216 Globals with "``linkonce``" linkage are merged with other globals of
217 the same name when linkage occurs. This can be used to implement
218 some forms of inline functions, templates, or other code which must
219 be generated in each translation unit that uses it, but where the
220 body may be overridden with a more definitive definition later.
221 Unreferenced ``linkonce`` globals are allowed to be discarded. Note
222 that ``linkonce`` linkage does not actually allow the optimizer to
223 inline the body of this function into callers because it doesn't
224 know if this definition of the function is the definitive definition
225 within the program or whether it will be overridden by a stronger
226 definition. To enable inlining and other optimizations, use
227 "``linkonce_odr``" linkage.
229 "``weak``" linkage has the same merging semantics as ``linkonce``
230 linkage, except that unreferenced globals with ``weak`` linkage may
231 not be discarded. This is used for globals that are declared "weak"
234 "``common``" linkage is most similar to "``weak``" linkage, but they
235 are used for tentative definitions in C, such as "``int X;``" at
236 global scope. Symbols with "``common``" linkage are merged in the
237 same way as ``weak symbols``, and they may not be deleted if
238 unreferenced. ``common`` symbols may not have an explicit section,
239 must have a zero initializer, and may not be marked
240 ':ref:`constant <globalvars>`'. Functions and aliases may not have
243 .. _linkage_appending:
246 "``appending``" linkage may only be applied to global variables of
247 pointer to array type. When two global variables with appending
248 linkage are linked together, the two global arrays are appended
249 together. This is the LLVM, typesafe, equivalent of having the
250 system linker append together "sections" with identical names when
253 The semantics of this linkage follow the ELF object file model: the
254 symbol is weak until linked, if not linked, the symbol becomes null
255 instead of being an undefined reference.
256 ``linkonce_odr``, ``weak_odr``
257 Some languages allow differing globals to be merged, such as two
258 functions with different semantics. Other languages, such as
259 ``C++``, ensure that only equivalent globals are ever merged (the
260 "one definition rule" --- "ODR"). Such languages can use the
261 ``linkonce_odr`` and ``weak_odr`` linkage types to indicate that the
262 global will only be merged with equivalent globals. These linkage
263 types are otherwise the same as their non-``odr`` versions.
265 If none of the above identifiers are used, the global is externally
266 visible, meaning that it participates in linkage and can be used to
267 resolve external symbol references.
269 It is illegal for a function *declaration* to have any linkage type
270 other than ``external`` or ``extern_weak``.
277 LLVM :ref:`functions <functionstructure>`, :ref:`calls <i_call>` and
278 :ref:`invokes <i_invoke>` can all have an optional calling convention
279 specified for the call. The calling convention of any pair of dynamic
280 caller/callee must match, or the behavior of the program is undefined.
281 The following calling conventions are supported by LLVM, and more may be
284 "``ccc``" - The C calling convention
285 This calling convention (the default if no other calling convention
286 is specified) matches the target C calling conventions. This calling
287 convention supports varargs function calls and tolerates some
288 mismatch in the declared prototype and implemented declaration of
289 the function (as does normal C).
290 "``fastcc``" - The fast calling convention
291 This calling convention attempts to make calls as fast as possible
292 (e.g. by passing things in registers). This calling convention
293 allows the target to use whatever tricks it wants to produce fast
294 code for the target, without having to conform to an externally
295 specified ABI (Application Binary Interface). `Tail calls can only
296 be optimized when this, the GHC or the HiPE convention is
297 used. <CodeGenerator.html#id80>`_ This calling convention does not
298 support varargs and requires the prototype of all callees to exactly
299 match the prototype of the function definition.
300 "``coldcc``" - The cold calling convention
301 This calling convention attempts to make code in the caller as
302 efficient as possible under the assumption that the call is not
303 commonly executed. As such, these calls often preserve all registers
304 so that the call does not break any live ranges in the caller side.
305 This calling convention does not support varargs and requires the
306 prototype of all callees to exactly match the prototype of the
307 function definition. Furthermore the inliner doesn't consider such function
309 "``cc 10``" - GHC convention
310 This calling convention has been implemented specifically for use by
311 the `Glasgow Haskell Compiler (GHC) <http://www.haskell.org/ghc>`_.
312 It passes everything in registers, going to extremes to achieve this
313 by disabling callee save registers. This calling convention should
314 not be used lightly but only for specific situations such as an
315 alternative to the *register pinning* performance technique often
316 used when implementing functional programming languages. At the
317 moment only X86 supports this convention and it has the following
320 - On *X86-32* only supports up to 4 bit type parameters. No
321 floating point types are supported.
322 - On *X86-64* only supports up to 10 bit type parameters and 6
323 floating point parameters.
325 This calling convention supports `tail call
326 optimization <CodeGenerator.html#id80>`_ but requires both the
327 caller and callee are using it.
328 "``cc 11``" - The HiPE calling convention
329 This calling convention has been implemented specifically for use by
330 the `High-Performance Erlang
331 (HiPE) <http://www.it.uu.se/research/group/hipe/>`_ compiler, *the*
332 native code compiler of the `Ericsson's Open Source Erlang/OTP
333 system <http://www.erlang.org/download.shtml>`_. It uses more
334 registers for argument passing than the ordinary C calling
335 convention and defines no callee-saved registers. The calling
336 convention properly supports `tail call
337 optimization <CodeGenerator.html#id80>`_ but requires that both the
338 caller and the callee use it. It uses a *register pinning*
339 mechanism, similar to GHC's convention, for keeping frequently
340 accessed runtime components pinned to specific hardware registers.
341 At the moment only X86 supports this convention (both 32 and 64
343 "``webkit_jscc``" - WebKit's JavaScript calling convention
344 This calling convention has been implemented for `WebKit FTL JIT
345 <https://trac.webkit.org/wiki/FTLJIT>`_. It passes arguments on the
346 stack right to left (as cdecl does), and returns a value in the
347 platform's customary return register.
348 "``anyregcc``" - Dynamic calling convention for code patching
349 This is a special convention that supports patching an arbitrary code
350 sequence in place of a call site. This convention forces the call
351 arguments into registers but allows them to be dynamcially
352 allocated. This can currently only be used with calls to
353 llvm.experimental.patchpoint because only this intrinsic records
354 the location of its arguments in a side table. See :doc:`StackMaps`.
355 "``preserve_mostcc``" - The `PreserveMost` calling convention
356 This calling convention attempts to make the code in the caller as little
357 intrusive as possible. This calling convention behaves identical to the `C`
358 calling convention on how arguments and return values are passed, but it
359 uses a different set of caller/callee-saved registers. This alleviates the
360 burden of saving and recovering a large register set before and after the
361 call in the caller. If the arguments are passed in callee-saved registers,
362 then they will be preserved by the callee across the call. This doesn't
363 apply for values returned in callee-saved registers.
365 - On X86-64 the callee preserves all general purpose registers, except for
366 R11. R11 can be used as a scratch register. Floating-point registers
367 (XMMs/YMMs) are not preserved and need to be saved by the caller.
369 The idea behind this convention is to support calls to runtime functions
370 that have a hot path and a cold path. The hot path is usually a small piece
371 of code that doesn't many registers. The cold path might need to call out to
372 another function and therefore only needs to preserve the caller-saved
373 registers, which haven't already been saved by the caller. The
374 `PreserveMost` calling convention is very similar to the `cold` calling
375 convention in terms of caller/callee-saved registers, but they are used for
376 different types of function calls. `coldcc` is for function calls that are
377 rarely executed, whereas `preserve_mostcc` function calls are intended to be
378 on the hot path and definitely executed a lot. Furthermore `preserve_mostcc`
379 doesn't prevent the inliner from inlining the function call.
381 This calling convention will be used by a future version of the ObjectiveC
382 runtime and should therefore still be considered experimental at this time.
383 Although this convention was created to optimize certain runtime calls to
384 the ObjectiveC runtime, it is not limited to this runtime and might be used
385 by other runtimes in the future too. The current implementation only
386 supports X86-64, but the intention is to support more architectures in the
388 "``preserve_allcc``" - The `PreserveAll` calling convention
389 This calling convention attempts to make the code in the caller even less
390 intrusive than the `PreserveMost` calling convention. This calling
391 convention also behaves identical to the `C` calling convention on how
392 arguments and return values are passed, but it uses a different set of
393 caller/callee-saved registers. This removes the burden of saving and
394 recovering a large register set before and after the call in the caller. If
395 the arguments are passed in callee-saved registers, then they will be
396 preserved by the callee across the call. This doesn't apply for values
397 returned in callee-saved registers.
399 - On X86-64 the callee preserves all general purpose registers, except for
400 R11. R11 can be used as a scratch register. Furthermore it also preserves
401 all floating-point registers (XMMs/YMMs).
403 The idea behind this convention is to support calls to runtime functions
404 that don't need to call out to any other functions.
406 This calling convention, like the `PreserveMost` calling convention, will be
407 used by a future version of the ObjectiveC runtime and should be considered
408 experimental at this time.
409 "``cc <n>``" - Numbered convention
410 Any calling convention may be specified by number, allowing
411 target-specific calling conventions to be used. Target specific
412 calling conventions start at 64.
414 More calling conventions can be added/defined on an as-needed basis, to
415 support Pascal conventions or any other well-known target-independent
418 .. _visibilitystyles:
423 All Global Variables and Functions have one of the following visibility
426 "``default``" - Default style
427 On targets that use the ELF object file format, default visibility
428 means that the declaration is visible to other modules and, in
429 shared libraries, means that the declared entity may be overridden.
430 On Darwin, default visibility means that the declaration is visible
431 to other modules. Default visibility corresponds to "external
432 linkage" in the language.
433 "``hidden``" - Hidden style
434 Two declarations of an object with hidden visibility refer to the
435 same object if they are in the same shared object. Usually, hidden
436 visibility indicates that the symbol will not be placed into the
437 dynamic symbol table, so no other module (executable or shared
438 library) can reference it directly.
439 "``protected``" - Protected style
440 On ELF, protected visibility indicates that the symbol will be
441 placed in the dynamic symbol table, but that references within the
442 defining module will bind to the local symbol. That is, the symbol
443 cannot be overridden by another module.
445 A symbol with ``internal`` or ``private`` linkage must have ``default``
453 All Global Variables, Functions and Aliases can have one of the following
457 "``dllimport``" causes the compiler to reference a function or variable via
458 a global pointer to a pointer that is set up by the DLL exporting the
459 symbol. On Microsoft Windows targets, the pointer name is formed by
460 combining ``__imp_`` and the function or variable name.
462 "``dllexport``" causes the compiler to provide a global pointer to a pointer
463 in a DLL, so that it can be referenced with the ``dllimport`` attribute. On
464 Microsoft Windows targets, the pointer name is formed by combining
465 ``__imp_`` and the function or variable name. Since this storage class
466 exists for defining a dll interface, the compiler, assembler and linker know
467 it is externally referenced and must refrain from deleting the symbol.
471 Thread Local Storage Models
472 ---------------------------
474 A variable may be defined as ``thread_local``, which means that it will
475 not be shared by threads (each thread will have a separated copy of the
476 variable). Not all targets support thread-local variables. Optionally, a
477 TLS model may be specified:
480 For variables that are only used within the current shared library.
482 For variables in modules that will not be loaded dynamically.
484 For variables defined in the executable and only used within it.
486 If no explicit model is given, the "general dynamic" model is used.
488 The models correspond to the ELF TLS models; see `ELF Handling For
489 Thread-Local Storage <http://people.redhat.com/drepper/tls.pdf>`_ for
490 more information on under which circumstances the different models may
491 be used. The target may choose a different TLS model if the specified
492 model is not supported, or if a better choice of model can be made.
494 A model can also be specified in a alias, but then it only governs how
495 the alias is accessed. It will not have any effect in the aliasee.
502 LLVM IR allows you to specify both "identified" and "literal" :ref:`structure
503 types <t_struct>`. Literal types are uniqued structurally, but identified types
504 are never uniqued. An :ref:`opaque structural type <t_opaque>` can also be used
505 to forward declare a type that is not yet available.
507 An example of a identified structure specification is:
511 %mytype = type { %mytype*, i32 }
513 Prior to the LLVM 3.0 release, identified types were structurally uniqued. Only
514 literal types are uniqued in recent versions of LLVM.
521 Global variables define regions of memory allocated at compilation time
524 Global variables definitions must be initialized.
526 Global variables in other translation units can also be declared, in which
527 case they don't have an initializer.
529 Either global variable definitions or declarations may have an explicit section
530 to be placed in and may have an optional explicit alignment specified.
532 A variable may be defined as a global ``constant``, which indicates that
533 the contents of the variable will **never** be modified (enabling better
534 optimization, allowing the global data to be placed in the read-only
535 section of an executable, etc). Note that variables that need runtime
536 initialization cannot be marked ``constant`` as there is a store to the
539 LLVM explicitly allows *declarations* of global variables to be marked
540 constant, even if the final definition of the global is not. This
541 capability can be used to enable slightly better optimization of the
542 program, but requires the language definition to guarantee that
543 optimizations based on the 'constantness' are valid for the translation
544 units that do not include the definition.
546 As SSA values, global variables define pointer values that are in scope
547 (i.e. they dominate) all basic blocks in the program. Global variables
548 always define a pointer to their "content" type because they describe a
549 region of memory, and all memory objects in LLVM are accessed through
552 Global variables can be marked with ``unnamed_addr`` which indicates
553 that the address is not significant, only the content. Constants marked
554 like this can be merged with other constants if they have the same
555 initializer. Note that a constant with significant address *can* be
556 merged with a ``unnamed_addr`` constant, the result being a constant
557 whose address is significant.
559 A global variable may be declared to reside in a target-specific
560 numbered address space. For targets that support them, address spaces
561 may affect how optimizations are performed and/or what target
562 instructions are used to access the variable. The default address space
563 is zero. The address space qualifier must precede any other attributes.
565 LLVM allows an explicit section to be specified for globals. If the
566 target supports it, it will emit globals to the section specified.
567 Additionally, the global can placed in a comdat if the target has the necessary
570 By default, global initializers are optimized by assuming that global
571 variables defined within the module are not modified from their
572 initial values before the start of the global initializer. This is
573 true even for variables potentially accessible from outside the
574 module, including those with external linkage or appearing in
575 ``@llvm.used`` or dllexported variables. This assumption may be suppressed
576 by marking the variable with ``externally_initialized``.
578 An explicit alignment may be specified for a global, which must be a
579 power of 2. If not present, or if the alignment is set to zero, the
580 alignment of the global is set by the target to whatever it feels
581 convenient. If an explicit alignment is specified, the global is forced
582 to have exactly that alignment. Targets and optimizers are not allowed
583 to over-align the global if the global has an assigned section. In this
584 case, the extra alignment could be observable: for example, code could
585 assume that the globals are densely packed in their section and try to
586 iterate over them as an array, alignment padding would break this
587 iteration. The maximum alignment is ``1 << 29``.
589 Globals can also have a :ref:`DLL storage class <dllstorageclass>`.
591 Variables and aliasaes can have a
592 :ref:`Thread Local Storage Model <tls_model>`.
596 [@<GlobalVarName> =] [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal]
597 [unnamed_addr] [AddrSpace] [ExternallyInitialized]
598 <global | constant> <Type> [<InitializerConstant>]
599 [, section "name"] [, comdat [($name)]]
600 [, align <Alignment>]
602 For example, the following defines a global in a numbered address space
603 with an initializer, section, and alignment:
607 @G = addrspace(5) constant float 1.0, section "foo", align 4
609 The following example just declares a global variable
613 @G = external global i32
615 The following example defines a thread-local global with the
616 ``initialexec`` TLS model:
620 @G = thread_local(initialexec) global i32 0, align 4
622 .. _functionstructure:
627 LLVM function definitions consist of the "``define``" keyword, an
628 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
629 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
630 an optional :ref:`calling convention <callingconv>`,
631 an optional ``unnamed_addr`` attribute, a return type, an optional
632 :ref:`parameter attribute <paramattrs>` for the return type, a function
633 name, a (possibly empty) argument list (each with optional :ref:`parameter
634 attributes <paramattrs>`), optional :ref:`function attributes <fnattrs>`,
635 an optional section, an optional alignment,
636 an optional :ref:`comdat <langref_comdats>`,
637 an optional :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`,
638 an optional :ref:`prologue <prologuedata>`, an opening
639 curly brace, a list of basic blocks, and a closing curly brace.
641 LLVM function declarations consist of the "``declare``" keyword, an
642 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
643 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
644 an optional :ref:`calling convention <callingconv>`,
645 an optional ``unnamed_addr`` attribute, a return type, an optional
646 :ref:`parameter attribute <paramattrs>` for the return type, a function
647 name, a possibly empty list of arguments, an optional alignment, an optional
648 :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`,
649 and an optional :ref:`prologue <prologuedata>`.
651 A function definition contains a list of basic blocks, forming the CFG (Control
652 Flow Graph) for the function. Each basic block may optionally start with a label
653 (giving the basic block a symbol table entry), contains a list of instructions,
654 and ends with a :ref:`terminator <terminators>` instruction (such as a branch or
655 function return). If an explicit label is not provided, a block is assigned an
656 implicit numbered label, using the next value from the same counter as used for
657 unnamed temporaries (:ref:`see above<identifiers>`). For example, if a function
658 entry block does not have an explicit label, it will be assigned label "%0",
659 then the first unnamed temporary in that block will be "%1", etc.
661 The first basic block in a function is special in two ways: it is
662 immediately executed on entrance to the function, and it is not allowed
663 to have predecessor basic blocks (i.e. there can not be any branches to
664 the entry block of a function). Because the block can have no
665 predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`.
667 LLVM allows an explicit section to be specified for functions. If the
668 target supports it, it will emit functions to the section specified.
669 Additionally, the function can placed in a COMDAT.
671 An explicit alignment may be specified for a function. If not present,
672 or if the alignment is set to zero, the alignment of the function is set
673 by the target to whatever it feels convenient. If an explicit alignment
674 is specified, the function is forced to have at least that much
675 alignment. All alignments must be a power of 2.
677 If the ``unnamed_addr`` attribute is given, the address is know to not
678 be significant and two identical functions can be merged.
682 define [linkage] [visibility] [DLLStorageClass]
684 <ResultType> @<FunctionName> ([argument list])
685 [unnamed_addr] [fn Attrs] [section "name"] [comdat [($name)]]
686 [align N] [gc] [prefix Constant] [prologue Constant] { ... }
688 The argument list is a comma seperated sequence of arguments where each
689 argument is of the following form
693 <type> [parameter Attrs] [name]
701 Aliases, unlike function or variables, don't create any new data. They
702 are just a new symbol and metadata for an existing position.
704 Aliases have a name and an aliasee that is either a global value or a
707 Aliases may have an optional :ref:`linkage type <linkage>`, an optional
708 :ref:`visibility style <visibility>`, an optional :ref:`DLL storage class
709 <dllstorageclass>` and an optional :ref:`tls model <tls_model>`.
713 @<Name> = [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal] [unnamed_addr] alias <AliaseeTy> @<Aliasee>
715 The linkage must be one of ``private``, ``internal``, ``linkonce``, ``weak``,
716 ``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers
717 might not correctly handle dropping a weak symbol that is aliased.
719 Alias that are not ``unnamed_addr`` are guaranteed to have the same address as
720 the aliasee expression. ``unnamed_addr`` ones are only guaranteed to point
723 Since aliases are only a second name, some restrictions apply, of which
724 some can only be checked when producing an object file:
726 * The expression defining the aliasee must be computable at assembly
727 time. Since it is just a name, no relocations can be used.
729 * No alias in the expression can be weak as the possibility of the
730 intermediate alias being overridden cannot be represented in an
733 * No global value in the expression can be a declaration, since that
734 would require a relocation, which is not possible.
741 Comdat IR provides access to COFF and ELF object file COMDAT functionality.
743 Comdats have a name which represents the COMDAT key. All global objects that
744 specify this key will only end up in the final object file if the linker chooses
745 that key over some other key. Aliases are placed in the same COMDAT that their
746 aliasee computes to, if any.
748 Comdats have a selection kind to provide input on how the linker should
749 choose between keys in two different object files.
753 $<Name> = comdat SelectionKind
755 The selection kind must be one of the following:
758 The linker may choose any COMDAT key, the choice is arbitrary.
760 The linker may choose any COMDAT key but the sections must contain the
763 The linker will choose the section containing the largest COMDAT key.
765 The linker requires that only section with this COMDAT key exist.
767 The linker may choose any COMDAT key but the sections must contain the
770 Note that the Mach-O platform doesn't support COMDATs and ELF only supports
771 ``any`` as a selection kind.
773 Here is an example of a COMDAT group where a function will only be selected if
774 the COMDAT key's section is the largest:
778 $foo = comdat largest
779 @foo = global i32 2, comdat($foo)
781 define void @bar() comdat($foo) {
785 As a syntactic sugar the ``$name`` can be omitted if the name is the same as
791 @foo = global i32 2, comdat
794 In a COFF object file, this will create a COMDAT section with selection kind
795 ``IMAGE_COMDAT_SELECT_LARGEST`` containing the contents of the ``@foo`` symbol
796 and another COMDAT section with selection kind
797 ``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT
798 section and contains the contents of the ``@bar`` symbol.
800 There are some restrictions on the properties of the global object.
801 It, or an alias to it, must have the same name as the COMDAT group when
803 The contents and size of this object may be used during link-time to determine
804 which COMDAT groups get selected depending on the selection kind.
805 Because the name of the object must match the name of the COMDAT group, the
806 linkage of the global object must not be local; local symbols can get renamed
807 if a collision occurs in the symbol table.
809 The combined use of COMDATS and section attributes may yield surprising results.
816 @g1 = global i32 42, section "sec", comdat($foo)
817 @g2 = global i32 42, section "sec", comdat($bar)
819 From the object file perspective, this requires the creation of two sections
820 with the same name. This is necessary because both globals belong to different
821 COMDAT groups and COMDATs, at the object file level, are represented by
824 Note that certain IR constructs like global variables and functions may create
825 COMDATs in the object file in addition to any which are specified using COMDAT
826 IR. This arises, for example, when a global variable has linkonce_odr linkage.
828 .. _namedmetadatastructure:
833 Named metadata is a collection of metadata. :ref:`Metadata
834 nodes <metadata>` (but not metadata strings) are the only valid
835 operands for a named metadata.
839 ; Some unnamed metadata nodes, which are referenced by the named metadata.
844 !name = !{!0, !1, !2}
851 The return type and each parameter of a function type may have a set of
852 *parameter attributes* associated with them. Parameter attributes are
853 used to communicate additional information about the result or
854 parameters of a function. Parameter attributes are considered to be part
855 of the function, not of the function type, so functions with different
856 parameter attributes can have the same function type.
858 Parameter attributes are simple keywords that follow the type specified.
859 If multiple parameter attributes are needed, they are space separated.
864 declare i32 @printf(i8* noalias nocapture, ...)
865 declare i32 @atoi(i8 zeroext)
866 declare signext i8 @returns_signed_char()
868 Note that any attributes for the function result (``nounwind``,
869 ``readonly``) come immediately after the argument list.
871 Currently, only the following parameter attributes are defined:
874 This indicates to the code generator that the parameter or return
875 value should be zero-extended to the extent required by the target's
876 ABI (which is usually 32-bits, but is 8-bits for a i1 on x86-64) by
877 the caller (for a parameter) or the callee (for a return value).
879 This indicates to the code generator that the parameter or return
880 value should be sign-extended to the extent required by the target's
881 ABI (which is usually 32-bits) by the caller (for a parameter) or
882 the callee (for a return value).
884 This indicates that this parameter or return value should be treated
885 in a special target-dependent fashion during while emitting code for
886 a function call or return (usually, by putting it in a register as
887 opposed to memory, though some targets use it to distinguish between
888 two different kinds of registers). Use of this attribute is
891 This indicates that the pointer parameter should really be passed by
892 value to the function. The attribute implies that a hidden copy of
893 the pointee is made between the caller and the callee, so the callee
894 is unable to modify the value in the caller. This attribute is only
895 valid on LLVM pointer arguments. It is generally used to pass
896 structs and arrays by value, but is also valid on pointers to
897 scalars. The copy is considered to belong to the caller not the
898 callee (for example, ``readonly`` functions should not write to
899 ``byval`` parameters). This is not a valid attribute for return
902 The byval attribute also supports specifying an alignment with the
903 align attribute. It indicates the alignment of the stack slot to
904 form and the known alignment of the pointer specified to the call
905 site. If the alignment is not specified, then the code generator
906 makes a target-specific assumption.
912 The ``inalloca`` argument attribute allows the caller to take the
913 address of outgoing stack arguments. An ``inalloca`` argument must
914 be a pointer to stack memory produced by an ``alloca`` instruction.
915 The alloca, or argument allocation, must also be tagged with the
916 inalloca keyword. Only the last argument may have the ``inalloca``
917 attribute, and that argument is guaranteed to be passed in memory.
919 An argument allocation may be used by a call at most once because
920 the call may deallocate it. The ``inalloca`` attribute cannot be
921 used in conjunction with other attributes that affect argument
922 storage, like ``inreg``, ``nest``, ``sret``, or ``byval``. The
923 ``inalloca`` attribute also disables LLVM's implicit lowering of
924 large aggregate return values, which means that frontend authors
925 must lower them with ``sret`` pointers.
927 When the call site is reached, the argument allocation must have
928 been the most recent stack allocation that is still live, or the
929 results are undefined. It is possible to allocate additional stack
930 space after an argument allocation and before its call site, but it
931 must be cleared off with :ref:`llvm.stackrestore
934 See :doc:`InAlloca` for more information on how to use this
938 This indicates that the pointer parameter specifies the address of a
939 structure that is the return value of the function in the source
940 program. This pointer must be guaranteed by the caller to be valid:
941 loads and stores to the structure may be assumed by the callee
942 not to trap and to be properly aligned. This may only be applied to
943 the first parameter. This is not a valid attribute for return
947 This indicates that the pointer value may be assumed by the optimizer to
948 have the specified alignment.
950 Note that this attribute has additional semantics when combined with the
956 This indicates that objects accessed via pointer values
957 :ref:`based <pointeraliasing>` on the argument or return value are not also
958 accessed, during the execution of the function, via pointer values not
959 *based* on the argument or return value. The attribute on a return value
960 also has additional semantics described below. The caller shares the
961 responsibility with the callee for ensuring that these requirements are met.
962 For further details, please see the discussion of the NoAlias response in
963 :ref:`alias analysis <Must, May, or No>`.
965 Note that this definition of ``noalias`` is intentionally similar
966 to the definition of ``restrict`` in C99 for function arguments.
968 For function return values, C99's ``restrict`` is not meaningful,
969 while LLVM's ``noalias`` is. Furthermore, the semantics of the ``noalias``
970 attribute on return values are stronger than the semantics of the attribute
971 when used on function arguments. On function return values, the ``noalias``
972 attribute indicates that the function acts like a system memory allocation
973 function, returning a pointer to allocated storage disjoint from the
974 storage for any other object accessible to the caller.
977 This indicates that the callee does not make any copies of the
978 pointer that outlive the callee itself. This is not a valid
979 attribute for return values.
984 This indicates that the pointer parameter can be excised using the
985 :ref:`trampoline intrinsics <int_trampoline>`. This is not a valid
986 attribute for return values and can only be applied to one parameter.
989 This indicates that the function always returns the argument as its return
990 value. This is an optimization hint to the code generator when generating
991 the caller, allowing tail call optimization and omission of register saves
992 and restores in some cases; it is not checked or enforced when generating
993 the callee. The parameter and the function return type must be valid
994 operands for the :ref:`bitcast instruction <i_bitcast>`. This is not a
995 valid attribute for return values and can only be applied to one parameter.
998 This indicates that the parameter or return pointer is not null. This
999 attribute may only be applied to pointer typed parameters. This is not
1000 checked or enforced by LLVM, the caller must ensure that the pointer
1001 passed in is non-null, or the callee must ensure that the returned pointer
1004 ``dereferenceable(<n>)``
1005 This indicates that the parameter or return pointer is dereferenceable. This
1006 attribute may only be applied to pointer typed parameters. A pointer that
1007 is dereferenceable can be loaded from speculatively without a risk of
1008 trapping. The number of bytes known to be dereferenceable must be provided
1009 in parentheses. It is legal for the number of bytes to be less than the
1010 size of the pointee type. The ``nonnull`` attribute does not imply
1011 dereferenceability (consider a pointer to one element past the end of an
1012 array), however ``dereferenceable(<n>)`` does imply ``nonnull`` in
1013 ``addrspace(0)`` (which is the default address space).
1017 Garbage Collector Names
1018 -----------------------
1020 Each function may specify a garbage collector name, which is simply a
1023 .. code-block:: llvm
1025 define void @f() gc "name" { ... }
1027 The compiler declares the supported values of *name*. Specifying a
1028 collector will cause the compiler to alter its output in order to
1029 support the named garbage collection algorithm.
1036 Prefix data is data associated with a function which the code
1037 generator will emit immediately before the function's entrypoint.
1038 The purpose of this feature is to allow frontends to associate
1039 language-specific runtime metadata with specific functions and make it
1040 available through the function pointer while still allowing the
1041 function pointer to be called.
1043 To access the data for a given function, a program may bitcast the
1044 function pointer to a pointer to the constant's type and dereference
1045 index -1. This implies that the IR symbol points just past the end of
1046 the prefix data. For instance, take the example of a function annotated
1047 with a single ``i32``,
1049 .. code-block:: llvm
1051 define void @f() prefix i32 123 { ... }
1053 The prefix data can be referenced as,
1055 .. code-block:: llvm
1057 %0 = bitcast *void () @f to *i32
1058 %a = getelementptr inbounds *i32 %0, i32 -1
1061 Prefix data is laid out as if it were an initializer for a global variable
1062 of the prefix data's type. The function will be placed such that the
1063 beginning of the prefix data is aligned. This means that if the size
1064 of the prefix data is not a multiple of the alignment size, the
1065 function's entrypoint will not be aligned. If alignment of the
1066 function's entrypoint is desired, padding must be added to the prefix
1069 A function may have prefix data but no body. This has similar semantics
1070 to the ``available_externally`` linkage in that the data may be used by the
1071 optimizers but will not be emitted in the object file.
1078 The ``prologue`` attribute allows arbitrary code (encoded as bytes) to
1079 be inserted prior to the function body. This can be used for enabling
1080 function hot-patching and instrumentation.
1082 To maintain the semantics of ordinary function calls, the prologue data must
1083 have a particular format. Specifically, it must begin with a sequence of
1084 bytes which decode to a sequence of machine instructions, valid for the
1085 module's target, which transfer control to the point immediately succeeding
1086 the prologue data, without performing any other visible action. This allows
1087 the inliner and other passes to reason about the semantics of the function
1088 definition without needing to reason about the prologue data. Obviously this
1089 makes the format of the prologue data highly target dependent.
1091 A trivial example of valid prologue data for the x86 architecture is ``i8 144``,
1092 which encodes the ``nop`` instruction:
1094 .. code-block:: llvm
1096 define void @f() prologue i8 144 { ... }
1098 Generally prologue data can be formed by encoding a relative branch instruction
1099 which skips the metadata, as in this example of valid prologue data for the
1100 x86_64 architecture, where the first two bytes encode ``jmp .+10``:
1102 .. code-block:: llvm
1104 %0 = type <{ i8, i8, i8* }>
1106 define void @f() prologue %0 <{ i8 235, i8 8, i8* @md}> { ... }
1108 A function may have prologue data but no body. This has similar semantics
1109 to the ``available_externally`` linkage in that the data may be used by the
1110 optimizers but will not be emitted in the object file.
1117 Attribute groups are groups of attributes that are referenced by objects within
1118 the IR. They are important for keeping ``.ll`` files readable, because a lot of
1119 functions will use the same set of attributes. In the degenerative case of a
1120 ``.ll`` file that corresponds to a single ``.c`` file, the single attribute
1121 group will capture the important command line flags used to build that file.
1123 An attribute group is a module-level object. To use an attribute group, an
1124 object references the attribute group's ID (e.g. ``#37``). An object may refer
1125 to more than one attribute group. In that situation, the attributes from the
1126 different groups are merged.
1128 Here is an example of attribute groups for a function that should always be
1129 inlined, has a stack alignment of 4, and which shouldn't use SSE instructions:
1131 .. code-block:: llvm
1133 ; Target-independent attributes:
1134 attributes #0 = { alwaysinline alignstack=4 }
1136 ; Target-dependent attributes:
1137 attributes #1 = { "no-sse" }
1139 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse".
1140 define void @f() #0 #1 { ... }
1147 Function attributes are set to communicate additional information about
1148 a function. Function attributes are considered to be part of the
1149 function, not of the function type, so functions with different function
1150 attributes can have the same function type.
1152 Function attributes are simple keywords that follow the type specified.
1153 If multiple attributes are needed, they are space separated. For
1156 .. code-block:: llvm
1158 define void @f() noinline { ... }
1159 define void @f() alwaysinline { ... }
1160 define void @f() alwaysinline optsize { ... }
1161 define void @f() optsize { ... }
1164 This attribute indicates that, when emitting the prologue and
1165 epilogue, the backend should forcibly align the stack pointer.
1166 Specify the desired alignment, which must be a power of two, in
1169 This attribute indicates that the inliner should attempt to inline
1170 this function into callers whenever possible, ignoring any active
1171 inlining size threshold for this caller.
1173 This indicates that the callee function at a call site should be
1174 recognized as a built-in function, even though the function's declaration
1175 uses the ``nobuiltin`` attribute. This is only valid at call sites for
1176 direct calls to functions that are declared with the ``nobuiltin``
1179 This attribute indicates that this function is rarely called. When
1180 computing edge weights, basic blocks post-dominated by a cold
1181 function call are also considered to be cold; and, thus, given low
1184 This attribute indicates that the source code contained a hint that
1185 inlining this function is desirable (such as the "inline" keyword in
1186 C/C++). It is just a hint; it imposes no requirements on the
1189 This attribute indicates that the function should be added to a
1190 jump-instruction table at code-generation time, and that all address-taken
1191 references to this function should be replaced with a reference to the
1192 appropriate jump-instruction-table function pointer. Note that this creates
1193 a new pointer for the original function, which means that code that depends
1194 on function-pointer identity can break. So, any function annotated with
1195 ``jumptable`` must also be ``unnamed_addr``.
1197 This attribute suggests that optimization passes and code generator
1198 passes make choices that keep the code size of this function as small
1199 as possible and perform optimizations that may sacrifice runtime
1200 performance in order to minimize the size of the generated code.
1202 This attribute disables prologue / epilogue emission for the
1203 function. This can have very system-specific consequences.
1205 This indicates that the callee function at a call site is not recognized as
1206 a built-in function. LLVM will retain the original call and not replace it
1207 with equivalent code based on the semantics of the built-in function, unless
1208 the call site uses the ``builtin`` attribute. This is valid at call sites
1209 and on function declarations and definitions.
1211 This attribute indicates that calls to the function cannot be
1212 duplicated. A call to a ``noduplicate`` function may be moved
1213 within its parent function, but may not be duplicated within
1214 its parent function.
1216 A function containing a ``noduplicate`` call may still
1217 be an inlining candidate, provided that the call is not
1218 duplicated by inlining. That implies that the function has
1219 internal linkage and only has one call site, so the original
1220 call is dead after inlining.
1222 This attributes disables implicit floating point instructions.
1224 This attribute indicates that the inliner should never inline this
1225 function in any situation. This attribute may not be used together
1226 with the ``alwaysinline`` attribute.
1228 This attribute suppresses lazy symbol binding for the function. This
1229 may make calls to the function faster, at the cost of extra program
1230 startup time if the function is not called during program startup.
1232 This attribute indicates that the code generator should not use a
1233 red zone, even if the target-specific ABI normally permits it.
1235 This function attribute indicates that the function never returns
1236 normally. This produces undefined behavior at runtime if the
1237 function ever does dynamically return.
1239 This function attribute indicates that the function never returns
1240 with an unwind or exceptional control flow. If the function does
1241 unwind, its runtime behavior is undefined.
1243 This function attribute indicates that the function is not optimized
1244 by any optimization or code generator passes with the
1245 exception of interprocedural optimization passes.
1246 This attribute cannot be used together with the ``alwaysinline``
1247 attribute; this attribute is also incompatible
1248 with the ``minsize`` attribute and the ``optsize`` attribute.
1250 This attribute requires the ``noinline`` attribute to be specified on
1251 the function as well, so the function is never inlined into any caller.
1252 Only functions with the ``alwaysinline`` attribute are valid
1253 candidates for inlining into the body of this function.
1255 This attribute suggests that optimization passes and code generator
1256 passes make choices that keep the code size of this function low,
1257 and otherwise do optimizations specifically to reduce code size as
1258 long as they do not significantly impact runtime performance.
1260 On a function, this attribute indicates that the function computes its
1261 result (or decides to unwind an exception) based strictly on its arguments,
1262 without dereferencing any pointer arguments or otherwise accessing
1263 any mutable state (e.g. memory, control registers, etc) visible to
1264 caller functions. It does not write through any pointer arguments
1265 (including ``byval`` arguments) and never changes any state visible
1266 to callers. This means that it cannot unwind exceptions by calling
1267 the ``C++`` exception throwing methods.
1269 On an argument, this attribute indicates that the function does not
1270 dereference that pointer argument, even though it may read or write the
1271 memory that the pointer points to if accessed through other pointers.
1273 On a function, this attribute indicates that the function does not write
1274 through any pointer arguments (including ``byval`` arguments) or otherwise
1275 modify any state (e.g. memory, control registers, etc) visible to
1276 caller functions. It may dereference pointer arguments and read
1277 state that may be set in the caller. A readonly function always
1278 returns the same value (or unwinds an exception identically) when
1279 called with the same set of arguments and global state. It cannot
1280 unwind an exception by calling the ``C++`` exception throwing
1283 On an argument, this attribute indicates that the function does not write
1284 through this pointer argument, even though it may write to the memory that
1285 the pointer points to.
1287 This attribute indicates that this function can return twice. The C
1288 ``setjmp`` is an example of such a function. The compiler disables
1289 some optimizations (like tail calls) in the caller of these
1291 ``sanitize_address``
1292 This attribute indicates that AddressSanitizer checks
1293 (dynamic address safety analysis) are enabled for this function.
1295 This attribute indicates that MemorySanitizer checks (dynamic detection
1296 of accesses to uninitialized memory) are enabled for this function.
1298 This attribute indicates that ThreadSanitizer checks
1299 (dynamic thread safety analysis) are enabled for this function.
1301 This attribute indicates that the function should emit a stack
1302 smashing protector. It is in the form of a "canary" --- a random value
1303 placed on the stack before the local variables that's checked upon
1304 return from the function to see if it has been overwritten. A
1305 heuristic is used to determine if a function needs stack protectors
1306 or not. The heuristic used will enable protectors for functions with:
1308 - Character arrays larger than ``ssp-buffer-size`` (default 8).
1309 - Aggregates containing character arrays larger than ``ssp-buffer-size``.
1310 - Calls to alloca() with variable sizes or constant sizes greater than
1311 ``ssp-buffer-size``.
1313 Variables that are identified as requiring a protector will be arranged
1314 on the stack such that they are adjacent to the stack protector guard.
1316 If a function that has an ``ssp`` attribute is inlined into a
1317 function that doesn't have an ``ssp`` attribute, then the resulting
1318 function will have an ``ssp`` attribute.
1320 This attribute indicates that the function should *always* emit a
1321 stack smashing protector. This overrides the ``ssp`` function
1324 Variables that are identified as requiring a protector will be arranged
1325 on the stack such that they are adjacent to the stack protector guard.
1326 The specific layout rules are:
1328 #. Large arrays and structures containing large arrays
1329 (``>= ssp-buffer-size``) are closest to the stack protector.
1330 #. Small arrays and structures containing small arrays
1331 (``< ssp-buffer-size``) are 2nd closest to the protector.
1332 #. Variables that have had their address taken are 3rd closest to the
1335 If a function that has an ``sspreq`` attribute is inlined into a
1336 function that doesn't have an ``sspreq`` attribute or which has an
1337 ``ssp`` or ``sspstrong`` attribute, then the resulting function will have
1338 an ``sspreq`` attribute.
1340 This attribute indicates that the function should emit a stack smashing
1341 protector. This attribute causes a strong heuristic to be used when
1342 determining if a function needs stack protectors. The strong heuristic
1343 will enable protectors for functions with:
1345 - Arrays of any size and type
1346 - Aggregates containing an array of any size and type.
1347 - Calls to alloca().
1348 - Local variables that have had their address taken.
1350 Variables that are identified as requiring a protector will be arranged
1351 on the stack such that they are adjacent to the stack protector guard.
1352 The specific layout rules are:
1354 #. Large arrays and structures containing large arrays
1355 (``>= ssp-buffer-size``) are closest to the stack protector.
1356 #. Small arrays and structures containing small arrays
1357 (``< ssp-buffer-size``) are 2nd closest to the protector.
1358 #. Variables that have had their address taken are 3rd closest to the
1361 This overrides the ``ssp`` function attribute.
1363 If a function that has an ``sspstrong`` attribute is inlined into a
1364 function that doesn't have an ``sspstrong`` attribute, then the
1365 resulting function will have an ``sspstrong`` attribute.
1367 This attribute indicates that the ABI being targeted requires that
1368 an unwind table entry be produce for this function even if we can
1369 show that no exceptions passes by it. This is normally the case for
1370 the ELF x86-64 abi, but it can be disabled for some compilation
1375 Module-Level Inline Assembly
1376 ----------------------------
1378 Modules may contain "module-level inline asm" blocks, which corresponds
1379 to the GCC "file scope inline asm" blocks. These blocks are internally
1380 concatenated by LLVM and treated as a single unit, but may be separated
1381 in the ``.ll`` file if desired. The syntax is very simple:
1383 .. code-block:: llvm
1385 module asm "inline asm code goes here"
1386 module asm "more can go here"
1388 The strings can contain any character by escaping non-printable
1389 characters. The escape sequence used is simply "\\xx" where "xx" is the
1390 two digit hex code for the number.
1392 The inline asm code is simply printed to the machine code .s file when
1393 assembly code is generated.
1395 .. _langref_datalayout:
1400 A module may specify a target specific data layout string that specifies
1401 how data is to be laid out in memory. The syntax for the data layout is
1404 .. code-block:: llvm
1406 target datalayout = "layout specification"
1408 The *layout specification* consists of a list of specifications
1409 separated by the minus sign character ('-'). Each specification starts
1410 with a letter and may include other information after the letter to
1411 define some aspect of the data layout. The specifications accepted are
1415 Specifies that the target lays out data in big-endian form. That is,
1416 the bits with the most significance have the lowest address
1419 Specifies that the target lays out data in little-endian form. That
1420 is, the bits with the least significance have the lowest address
1423 Specifies the natural alignment of the stack in bits. Alignment
1424 promotion of stack variables is limited to the natural stack
1425 alignment to avoid dynamic stack realignment. The stack alignment
1426 must be a multiple of 8-bits. If omitted, the natural stack
1427 alignment defaults to "unspecified", which does not prevent any
1428 alignment promotions.
1429 ``p[n]:<size>:<abi>:<pref>``
1430 This specifies the *size* of a pointer and its ``<abi>`` and
1431 ``<pref>``\erred alignments for address space ``n``. All sizes are in
1432 bits. The address space, ``n`` is optional, and if not specified,
1433 denotes the default address space 0. The value of ``n`` must be
1434 in the range [1,2^23).
1435 ``i<size>:<abi>:<pref>``
1436 This specifies the alignment for an integer type of a given bit
1437 ``<size>``. The value of ``<size>`` must be in the range [1,2^23).
1438 ``v<size>:<abi>:<pref>``
1439 This specifies the alignment for a vector type of a given bit
1441 ``f<size>:<abi>:<pref>``
1442 This specifies the alignment for a floating point type of a given bit
1443 ``<size>``. Only values of ``<size>`` that are supported by the target
1444 will work. 32 (float) and 64 (double) are supported on all targets; 80
1445 or 128 (different flavors of long double) are also supported on some
1448 This specifies the alignment for an object of aggregate type.
1450 If present, specifies that llvm names are mangled in the output. The
1453 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix.
1454 * ``m``: Mips mangling: Private symbols get a ``$`` prefix.
1455 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other
1456 symbols get a ``_`` prefix.
1457 * ``w``: Windows COFF prefix: Similar to Mach-O, but stdcall and fastcall
1458 functions also get a suffix based on the frame size.
1459 ``n<size1>:<size2>:<size3>...``
1460 This specifies a set of native integer widths for the target CPU in
1461 bits. For example, it might contain ``n32`` for 32-bit PowerPC,
1462 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of
1463 this set are considered to support most general arithmetic operations
1466 On every specification that takes a ``<abi>:<pref>``, specifying the
1467 ``<pref>`` alignment is optional. If omitted, the preceding ``:``
1468 should be omitted too and ``<pref>`` will be equal to ``<abi>``.
1470 When constructing the data layout for a given target, LLVM starts with a
1471 default set of specifications which are then (possibly) overridden by
1472 the specifications in the ``datalayout`` keyword. The default
1473 specifications are given in this list:
1475 - ``E`` - big endian
1476 - ``p:64:64:64`` - 64-bit pointers with 64-bit alignment.
1477 - ``p[n]:64:64:64`` - Other address spaces are assumed to be the
1478 same as the default address space.
1479 - ``S0`` - natural stack alignment is unspecified
1480 - ``i1:8:8`` - i1 is 8-bit (byte) aligned
1481 - ``i8:8:8`` - i8 is 8-bit (byte) aligned
1482 - ``i16:16:16`` - i16 is 16-bit aligned
1483 - ``i32:32:32`` - i32 is 32-bit aligned
1484 - ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred
1485 alignment of 64-bits
1486 - ``f16:16:16`` - half is 16-bit aligned
1487 - ``f32:32:32`` - float is 32-bit aligned
1488 - ``f64:64:64`` - double is 64-bit aligned
1489 - ``f128:128:128`` - quad is 128-bit aligned
1490 - ``v64:64:64`` - 64-bit vector is 64-bit aligned
1491 - ``v128:128:128`` - 128-bit vector is 128-bit aligned
1492 - ``a:0:64`` - aggregates are 64-bit aligned
1494 When LLVM is determining the alignment for a given type, it uses the
1497 #. If the type sought is an exact match for one of the specifications,
1498 that specification is used.
1499 #. If no match is found, and the type sought is an integer type, then
1500 the smallest integer type that is larger than the bitwidth of the
1501 sought type is used. If none of the specifications are larger than
1502 the bitwidth then the largest integer type is used. For example,
1503 given the default specifications above, the i7 type will use the
1504 alignment of i8 (next largest) while both i65 and i256 will use the
1505 alignment of i64 (largest specified).
1506 #. If no match is found, and the type sought is a vector type, then the
1507 largest vector type that is smaller than the sought vector type will
1508 be used as a fall back. This happens because <128 x double> can be
1509 implemented in terms of 64 <2 x double>, for example.
1511 The function of the data layout string may not be what you expect.
1512 Notably, this is not a specification from the frontend of what alignment
1513 the code generator should use.
1515 Instead, if specified, the target data layout is required to match what
1516 the ultimate *code generator* expects. This string is used by the
1517 mid-level optimizers to improve code, and this only works if it matches
1518 what the ultimate code generator uses. If you would like to generate IR
1519 that does not embed this target-specific detail into the IR, then you
1520 don't have to specify the string. This will disable some optimizations
1521 that require precise layout information, but this also prevents those
1522 optimizations from introducing target specificity into the IR.
1529 A module may specify a target triple string that describes the target
1530 host. The syntax for the target triple is simply:
1532 .. code-block:: llvm
1534 target triple = "x86_64-apple-macosx10.7.0"
1536 The *target triple* string consists of a series of identifiers delimited
1537 by the minus sign character ('-'). The canonical forms are:
1541 ARCHITECTURE-VENDOR-OPERATING_SYSTEM
1542 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT
1544 This information is passed along to the backend so that it generates
1545 code for the proper architecture. It's possible to override this on the
1546 command line with the ``-mtriple`` command line option.
1548 .. _pointeraliasing:
1550 Pointer Aliasing Rules
1551 ----------------------
1553 Any memory access must be done through a pointer value associated with
1554 an address range of the memory access, otherwise the behavior is
1555 undefined. Pointer values are associated with address ranges according
1556 to the following rules:
1558 - A pointer value is associated with the addresses associated with any
1559 value it is *based* on.
1560 - An address of a global variable is associated with the address range
1561 of the variable's storage.
1562 - The result value of an allocation instruction is associated with the
1563 address range of the allocated storage.
1564 - A null pointer in the default address-space is associated with no
1566 - An integer constant other than zero or a pointer value returned from
1567 a function not defined within LLVM may be associated with address
1568 ranges allocated through mechanisms other than those provided by
1569 LLVM. Such ranges shall not overlap with any ranges of addresses
1570 allocated by mechanisms provided by LLVM.
1572 A pointer value is *based* on another pointer value according to the
1575 - A pointer value formed from a ``getelementptr`` operation is *based*
1576 on the first operand of the ``getelementptr``.
1577 - The result value of a ``bitcast`` is *based* on the operand of the
1579 - A pointer value formed by an ``inttoptr`` is *based* on all pointer
1580 values that contribute (directly or indirectly) to the computation of
1581 the pointer's value.
1582 - The "*based* on" relationship is transitive.
1584 Note that this definition of *"based"* is intentionally similar to the
1585 definition of *"based"* in C99, though it is slightly weaker.
1587 LLVM IR does not associate types with memory. The result type of a
1588 ``load`` merely indicates the size and alignment of the memory from
1589 which to load, as well as the interpretation of the value. The first
1590 operand type of a ``store`` similarly only indicates the size and
1591 alignment of the store.
1593 Consequently, type-based alias analysis, aka TBAA, aka
1594 ``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR.
1595 :ref:`Metadata <metadata>` may be used to encode additional information
1596 which specialized optimization passes may use to implement type-based
1601 Volatile Memory Accesses
1602 ------------------------
1604 Certain memory accesses, such as :ref:`load <i_load>`'s,
1605 :ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be
1606 marked ``volatile``. The optimizers must not change the number of
1607 volatile operations or change their order of execution relative to other
1608 volatile operations. The optimizers *may* change the order of volatile
1609 operations relative to non-volatile operations. This is not Java's
1610 "volatile" and has no cross-thread synchronization behavior.
1612 IR-level volatile loads and stores cannot safely be optimized into
1613 llvm.memcpy or llvm.memmove intrinsics even when those intrinsics are
1614 flagged volatile. Likewise, the backend should never split or merge
1615 target-legal volatile load/store instructions.
1617 .. admonition:: Rationale
1619 Platforms may rely on volatile loads and stores of natively supported
1620 data width to be executed as single instruction. For example, in C
1621 this holds for an l-value of volatile primitive type with native
1622 hardware support, but not necessarily for aggregate types. The
1623 frontend upholds these expectations, which are intentionally
1624 unspecified in the IR. The rules above ensure that IR transformation
1625 do not violate the frontend's contract with the language.
1629 Memory Model for Concurrent Operations
1630 --------------------------------------
1632 The LLVM IR does not define any way to start parallel threads of
1633 execution or to register signal handlers. Nonetheless, there are
1634 platform-specific ways to create them, and we define LLVM IR's behavior
1635 in their presence. This model is inspired by the C++0x memory model.
1637 For a more informal introduction to this model, see the :doc:`Atomics`.
1639 We define a *happens-before* partial order as the least partial order
1642 - Is a superset of single-thread program order, and
1643 - When a *synchronizes-with* ``b``, includes an edge from ``a`` to
1644 ``b``. *Synchronizes-with* pairs are introduced by platform-specific
1645 techniques, like pthread locks, thread creation, thread joining,
1646 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering
1647 Constraints <ordering>`).
1649 Note that program order does not introduce *happens-before* edges
1650 between a thread and signals executing inside that thread.
1652 Every (defined) read operation (load instructions, memcpy, atomic
1653 loads/read-modify-writes, etc.) R reads a series of bytes written by
1654 (defined) write operations (store instructions, atomic
1655 stores/read-modify-writes, memcpy, etc.). For the purposes of this
1656 section, initialized globals are considered to have a write of the
1657 initializer which is atomic and happens before any other read or write
1658 of the memory in question. For each byte of a read R, R\ :sub:`byte`
1659 may see any write to the same byte, except:
1661 - If write\ :sub:`1` happens before write\ :sub:`2`, and
1662 write\ :sub:`2` happens before R\ :sub:`byte`, then
1663 R\ :sub:`byte` does not see write\ :sub:`1`.
1664 - If R\ :sub:`byte` happens before write\ :sub:`3`, then
1665 R\ :sub:`byte` does not see write\ :sub:`3`.
1667 Given that definition, R\ :sub:`byte` is defined as follows:
1669 - If R is volatile, the result is target-dependent. (Volatile is
1670 supposed to give guarantees which can support ``sig_atomic_t`` in
1671 C/C++, and may be used for accesses to addresses that do not behave
1672 like normal memory. It does not generally provide cross-thread
1674 - Otherwise, if there is no write to the same byte that happens before
1675 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte.
1676 - Otherwise, if R\ :sub:`byte` may see exactly one write,
1677 R\ :sub:`byte` returns the value written by that write.
1678 - Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may
1679 see are atomic, it chooses one of the values written. See the :ref:`Atomic
1680 Memory Ordering Constraints <ordering>` section for additional
1681 constraints on how the choice is made.
1682 - Otherwise R\ :sub:`byte` returns ``undef``.
1684 R returns the value composed of the series of bytes it read. This
1685 implies that some bytes within the value may be ``undef`` **without**
1686 the entire value being ``undef``. Note that this only defines the
1687 semantics of the operation; it doesn't mean that targets will emit more
1688 than one instruction to read the series of bytes.
1690 Note that in cases where none of the atomic intrinsics are used, this
1691 model places only one restriction on IR transformations on top of what
1692 is required for single-threaded execution: introducing a store to a byte
1693 which might not otherwise be stored is not allowed in general.
1694 (Specifically, in the case where another thread might write to and read
1695 from an address, introducing a store can change a load that may see
1696 exactly one write into a load that may see multiple writes.)
1700 Atomic Memory Ordering Constraints
1701 ----------------------------------
1703 Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`,
1704 :ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`,
1705 :ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take
1706 ordering parameters that determine which other atomic instructions on
1707 the same address they *synchronize with*. These semantics are borrowed
1708 from Java and C++0x, but are somewhat more colloquial. If these
1709 descriptions aren't precise enough, check those specs (see spec
1710 references in the :doc:`atomics guide <Atomics>`).
1711 :ref:`fence <i_fence>` instructions treat these orderings somewhat
1712 differently since they don't take an address. See that instruction's
1713 documentation for details.
1715 For a simpler introduction to the ordering constraints, see the
1719 The set of values that can be read is governed by the happens-before
1720 partial order. A value cannot be read unless some operation wrote
1721 it. This is intended to provide a guarantee strong enough to model
1722 Java's non-volatile shared variables. This ordering cannot be
1723 specified for read-modify-write operations; it is not strong enough
1724 to make them atomic in any interesting way.
1726 In addition to the guarantees of ``unordered``, there is a single
1727 total order for modifications by ``monotonic`` operations on each
1728 address. All modification orders must be compatible with the
1729 happens-before order. There is no guarantee that the modification
1730 orders can be combined to a global total order for the whole program
1731 (and this often will not be possible). The read in an atomic
1732 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and
1733 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification
1734 order immediately before the value it writes. If one atomic read
1735 happens before another atomic read of the same address, the later
1736 read must see the same value or a later value in the address's
1737 modification order. This disallows reordering of ``monotonic`` (or
1738 stronger) operations on the same address. If an address is written
1739 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally
1740 read that address repeatedly, the other threads must eventually see
1741 the write. This corresponds to the C++0x/C1x
1742 ``memory_order_relaxed``.
1744 In addition to the guarantees of ``monotonic``, a
1745 *synchronizes-with* edge may be formed with a ``release`` operation.
1746 This is intended to model C++'s ``memory_order_acquire``.
1748 In addition to the guarantees of ``monotonic``, if this operation
1749 writes a value which is subsequently read by an ``acquire``
1750 operation, it *synchronizes-with* that operation. (This isn't a
1751 complete description; see the C++0x definition of a release
1752 sequence.) This corresponds to the C++0x/C1x
1753 ``memory_order_release``.
1754 ``acq_rel`` (acquire+release)
1755 Acts as both an ``acquire`` and ``release`` operation on its
1756 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``.
1757 ``seq_cst`` (sequentially consistent)
1758 In addition to the guarantees of ``acq_rel`` (``acquire`` for an
1759 operation that only reads, ``release`` for an operation that only
1760 writes), there is a global total order on all
1761 sequentially-consistent operations on all addresses, which is
1762 consistent with the *happens-before* partial order and with the
1763 modification orders of all the affected addresses. Each
1764 sequentially-consistent read sees the last preceding write to the
1765 same address in this global order. This corresponds to the C++0x/C1x
1766 ``memory_order_seq_cst`` and Java volatile.
1770 If an atomic operation is marked ``singlethread``, it only *synchronizes
1771 with* or participates in modification and seq\_cst total orderings with
1772 other operations running in the same thread (for example, in signal
1780 LLVM IR floating-point binary ops (:ref:`fadd <i_fadd>`,
1781 :ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`,
1782 :ref:`frem <i_frem>`) have the following flags that can set to enable
1783 otherwise unsafe floating point operations
1786 No NaNs - Allow optimizations to assume the arguments and result are not
1787 NaN. Such optimizations are required to retain defined behavior over
1788 NaNs, but the value of the result is undefined.
1791 No Infs - Allow optimizations to assume the arguments and result are not
1792 +/-Inf. Such optimizations are required to retain defined behavior over
1793 +/-Inf, but the value of the result is undefined.
1796 No Signed Zeros - Allow optimizations to treat the sign of a zero
1797 argument or result as insignificant.
1800 Allow Reciprocal - Allow optimizations to use the reciprocal of an
1801 argument rather than perform division.
1804 Fast - Allow algebraically equivalent transformations that may
1805 dramatically change results in floating point (e.g. reassociate). This
1806 flag implies all the others.
1810 Use-list Order Directives
1811 -------------------------
1813 Use-list directives encode the in-memory order of each use-list, allowing the
1814 order to be recreated. ``<order-indexes>`` is a comma-separated list of
1815 indexes that are assigned to the referenced value's uses. The referenced
1816 value's use-list is immediately sorted by these indexes.
1818 Use-list directives may appear at function scope or global scope. They are not
1819 instructions, and have no effect on the semantics of the IR. When they're at
1820 function scope, they must appear after the terminator of the final basic block.
1822 If basic blocks have their address taken via ``blockaddress()`` expressions,
1823 ``uselistorder_bb`` can be used to reorder their use-lists from outside their
1830 uselistorder <ty> <value>, { <order-indexes> }
1831 uselistorder_bb @function, %block { <order-indexes> }
1837 define void @foo(i32 %arg1, i32 %arg2) {
1839 ; ... instructions ...
1841 ; ... instructions ...
1843 ; At function scope.
1844 uselistorder i32 %arg1, { 1, 0, 2 }
1845 uselistorder label %bb, { 1, 0 }
1849 uselistorder i32* @global, { 1, 2, 0 }
1850 uselistorder i32 7, { 1, 0 }
1851 uselistorder i32 (i32) @bar, { 1, 0 }
1852 uselistorder_bb @foo, %bb, { 5, 1, 3, 2, 0, 4 }
1859 The LLVM type system is one of the most important features of the
1860 intermediate representation. Being typed enables a number of
1861 optimizations to be performed on the intermediate representation
1862 directly, without having to do extra analyses on the side before the
1863 transformation. A strong type system makes it easier to read the
1864 generated code and enables novel analyses and transformations that are
1865 not feasible to perform on normal three address code representations.
1875 The void type does not represent any value and has no size.
1893 The function type can be thought of as a function signature. It consists of a
1894 return type and a list of formal parameter types. The return type of a function
1895 type is a void type or first class type --- except for :ref:`label <t_label>`
1896 and :ref:`metadata <t_metadata>` types.
1902 <returntype> (<parameter list>)
1904 ...where '``<parameter list>``' is a comma-separated list of type
1905 specifiers. Optionally, the parameter list may include a type ``...``, which
1906 indicates that the function takes a variable number of arguments. Variable
1907 argument functions can access their arguments with the :ref:`variable argument
1908 handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type
1909 except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`.
1913 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1914 | ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` |
1915 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1916 | ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. |
1917 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1918 | ``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. |
1919 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1920 | ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values |
1921 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1928 The :ref:`first class <t_firstclass>` types are perhaps the most important.
1929 Values of these types are the only ones which can be produced by
1937 These are the types that are valid in registers from CodeGen's perspective.
1946 The integer type is a very simple type that simply specifies an
1947 arbitrary bit width for the integer type desired. Any bit width from 1
1948 bit to 2\ :sup:`23`\ -1 (about 8 million) can be specified.
1956 The number of bits the integer will occupy is specified by the ``N``
1962 +----------------+------------------------------------------------+
1963 | ``i1`` | a single-bit integer. |
1964 +----------------+------------------------------------------------+
1965 | ``i32`` | a 32-bit integer. |
1966 +----------------+------------------------------------------------+
1967 | ``i1942652`` | a really big integer of over 1 million bits. |
1968 +----------------+------------------------------------------------+
1972 Floating Point Types
1973 """"""""""""""""""""
1982 - 16-bit floating point value
1985 - 32-bit floating point value
1988 - 64-bit floating point value
1991 - 128-bit floating point value (112-bit mantissa)
1994 - 80-bit floating point value (X87)
1997 - 128-bit floating point value (two 64-bits)
2004 The x86_mmx type represents a value held in an MMX register on an x86
2005 machine. The operations allowed on it are quite limited: parameters and
2006 return values, load and store, and bitcast. User-specified MMX
2007 instructions are represented as intrinsic or asm calls with arguments
2008 and/or results of this type. There are no arrays, vectors or constants
2025 The pointer type is used to specify memory locations. Pointers are
2026 commonly used to reference objects in memory.
2028 Pointer types may have an optional address space attribute defining the
2029 numbered address space where the pointed-to object resides. The default
2030 address space is number zero. The semantics of non-zero address spaces
2031 are target-specific.
2033 Note that LLVM does not permit pointers to void (``void*``) nor does it
2034 permit pointers to labels (``label*``). Use ``i8*`` instead.
2044 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2045 | ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. |
2046 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2047 | ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. |
2048 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2049 | ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space #5. |
2050 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2059 A vector type is a simple derived type that represents a vector of
2060 elements. Vector types are used when multiple primitive data are
2061 operated in parallel using a single instruction (SIMD). A vector type
2062 requires a size (number of elements) and an underlying primitive data
2063 type. Vector types are considered :ref:`first class <t_firstclass>`.
2069 < <# elements> x <elementtype> >
2071 The number of elements is a constant integer value larger than 0;
2072 elementtype may be any integer, floating point or pointer type. Vectors
2073 of size zero are not allowed.
2077 +-------------------+--------------------------------------------------+
2078 | ``<4 x i32>`` | Vector of 4 32-bit integer values. |
2079 +-------------------+--------------------------------------------------+
2080 | ``<8 x float>`` | Vector of 8 32-bit floating-point values. |
2081 +-------------------+--------------------------------------------------+
2082 | ``<2 x i64>`` | Vector of 2 64-bit integer values. |
2083 +-------------------+--------------------------------------------------+
2084 | ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. |
2085 +-------------------+--------------------------------------------------+
2094 The label type represents code labels.
2109 The metadata type represents embedded metadata. No derived types may be
2110 created from metadata except for :ref:`function <t_function>` arguments.
2123 Aggregate Types are a subset of derived types that can contain multiple
2124 member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are
2125 aggregate types. :ref:`Vectors <t_vector>` are not considered to be
2135 The array type is a very simple derived type that arranges elements
2136 sequentially in memory. The array type requires a size (number of
2137 elements) and an underlying data type.
2143 [<# elements> x <elementtype>]
2145 The number of elements is a constant integer value; ``elementtype`` may
2146 be any type with a size.
2150 +------------------+--------------------------------------+
2151 | ``[40 x i32]`` | Array of 40 32-bit integer values. |
2152 +------------------+--------------------------------------+
2153 | ``[41 x i32]`` | Array of 41 32-bit integer values. |
2154 +------------------+--------------------------------------+
2155 | ``[4 x i8]`` | Array of 4 8-bit integer values. |
2156 +------------------+--------------------------------------+
2158 Here are some examples of multidimensional arrays:
2160 +-----------------------------+----------------------------------------------------------+
2161 | ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. |
2162 +-----------------------------+----------------------------------------------------------+
2163 | ``[12 x [10 x float]]`` | 12x10 array of single precision floating point values. |
2164 +-----------------------------+----------------------------------------------------------+
2165 | ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. |
2166 +-----------------------------+----------------------------------------------------------+
2168 There is no restriction on indexing beyond the end of the array implied
2169 by a static type (though there are restrictions on indexing beyond the
2170 bounds of an allocated object in some cases). This means that
2171 single-dimension 'variable sized array' addressing can be implemented in
2172 LLVM with a zero length array type. An implementation of 'pascal style
2173 arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for
2183 The structure type is used to represent a collection of data members
2184 together in memory. The elements of a structure may be any type that has
2187 Structures in memory are accessed using '``load``' and '``store``' by
2188 getting a pointer to a field with the '``getelementptr``' instruction.
2189 Structures in registers are accessed using the '``extractvalue``' and
2190 '``insertvalue``' instructions.
2192 Structures may optionally be "packed" structures, which indicate that
2193 the alignment of the struct is one byte, and that there is no padding
2194 between the elements. In non-packed structs, padding between field types
2195 is inserted as defined by the DataLayout string in the module, which is
2196 required to match what the underlying code generator expects.
2198 Structures can either be "literal" or "identified". A literal structure
2199 is defined inline with other types (e.g. ``{i32, i32}*``) whereas
2200 identified types are always defined at the top level with a name.
2201 Literal types are uniqued by their contents and can never be recursive
2202 or opaque since there is no way to write one. Identified types can be
2203 recursive, can be opaqued, and are never uniqued.
2209 %T1 = type { <type list> } ; Identified normal struct type
2210 %T2 = type <{ <type list> }> ; Identified packed struct type
2214 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2215 | ``{ i32, i32, i32 }`` | A triple of three ``i32`` values |
2216 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2217 | ``{ 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``. |
2218 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2219 | ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. |
2220 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2224 Opaque Structure Types
2225 """"""""""""""""""""""
2229 Opaque structure types are used to represent named structure types that
2230 do not have a body specified. This corresponds (for example) to the C
2231 notion of a forward declared structure.
2242 +--------------+-------------------+
2243 | ``opaque`` | An opaque type. |
2244 +--------------+-------------------+
2251 LLVM has several different basic types of constants. This section
2252 describes them all and their syntax.
2257 **Boolean constants**
2258 The two strings '``true``' and '``false``' are both valid constants
2260 **Integer constants**
2261 Standard integers (such as '4') are constants of the
2262 :ref:`integer <t_integer>` type. Negative numbers may be used with
2264 **Floating point constants**
2265 Floating point constants use standard decimal notation (e.g.
2266 123.421), exponential notation (e.g. 1.23421e+2), or a more precise
2267 hexadecimal notation (see below). The assembler requires the exact
2268 decimal value of a floating-point constant. For example, the
2269 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating
2270 decimal in binary. Floating point constants must have a :ref:`floating
2271 point <t_floating>` type.
2272 **Null pointer constants**
2273 The identifier '``null``' is recognized as a null pointer constant
2274 and must be of :ref:`pointer type <t_pointer>`.
2276 The one non-intuitive notation for constants is the hexadecimal form of
2277 floating point constants. For example, the form
2278 '``double 0x432ff973cafa8000``' is equivalent to (but harder to read
2279 than) '``double 4.5e+15``'. The only time hexadecimal floating point
2280 constants are required (and the only time that they are generated by the
2281 disassembler) is when a floating point constant must be emitted but it
2282 cannot be represented as a decimal floating point number in a reasonable
2283 number of digits. For example, NaN's, infinities, and other special
2284 values are represented in their IEEE hexadecimal format so that assembly
2285 and disassembly do not cause any bits to change in the constants.
2287 When using the hexadecimal form, constants of types half, float, and
2288 double are represented using the 16-digit form shown above (which
2289 matches the IEEE754 representation for double); half and float values
2290 must, however, be exactly representable as IEEE 754 half and single
2291 precision, respectively. Hexadecimal format is always used for long
2292 double, and there are three forms of long double. The 80-bit format used
2293 by x86 is represented as ``0xK`` followed by 20 hexadecimal digits. The
2294 128-bit format used by PowerPC (two adjacent doubles) is represented by
2295 ``0xM`` followed by 32 hexadecimal digits. The IEEE 128-bit format is
2296 represented by ``0xL`` followed by 32 hexadecimal digits. Long doubles
2297 will only work if they match the long double format on your target.
2298 The IEEE 16-bit format (half precision) is represented by ``0xH``
2299 followed by 4 hexadecimal digits. All hexadecimal formats are big-endian
2300 (sign bit at the left).
2302 There are no constants of type x86_mmx.
2304 .. _complexconstants:
2309 Complex constants are a (potentially recursive) combination of simple
2310 constants and smaller complex constants.
2312 **Structure constants**
2313 Structure constants are represented with notation similar to
2314 structure type definitions (a comma separated list of elements,
2315 surrounded by braces (``{}``)). For example:
2316 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as
2317 "``@G = external global i32``". Structure constants must have
2318 :ref:`structure type <t_struct>`, and the number and types of elements
2319 must match those specified by the type.
2321 Array constants are represented with notation similar to array type
2322 definitions (a comma separated list of elements, surrounded by
2323 square brackets (``[]``)). For example:
2324 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have
2325 :ref:`array type <t_array>`, and the number and types of elements must
2326 match those specified by the type. As a special case, character array
2327 constants may also be represented as a double-quoted string using the ``c``
2328 prefix. For example: "``c"Hello World\0A\00"``".
2329 **Vector constants**
2330 Vector constants are represented with notation similar to vector
2331 type definitions (a comma separated list of elements, surrounded by
2332 less-than/greater-than's (``<>``)). For example:
2333 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants
2334 must have :ref:`vector type <t_vector>`, and the number and types of
2335 elements must match those specified by the type.
2336 **Zero initialization**
2337 The string '``zeroinitializer``' can be used to zero initialize a
2338 value to zero of *any* type, including scalar and
2339 :ref:`aggregate <t_aggregate>` types. This is often used to avoid
2340 having to print large zero initializers (e.g. for large arrays) and
2341 is always exactly equivalent to using explicit zero initializers.
2343 A metadata node is a constant tuple without types. For example:
2344 "``!{!0, !{!2, !0}, !"test"}``". Metadata can reference constant values,
2345 for example: "``!{!0, i32 0, i8* @global, i64 (i64)* @function, !"str"}``".
2346 Unlike other typed constants that are meant to be interpreted as part of
2347 the instruction stream, metadata is a place to attach additional
2348 information such as debug info.
2350 Global Variable and Function Addresses
2351 --------------------------------------
2353 The addresses of :ref:`global variables <globalvars>` and
2354 :ref:`functions <functionstructure>` are always implicitly valid
2355 (link-time) constants. These constants are explicitly referenced when
2356 the :ref:`identifier for the global <identifiers>` is used and always have
2357 :ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM
2360 .. code-block:: llvm
2364 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2371 The string '``undef``' can be used anywhere a constant is expected, and
2372 indicates that the user of the value may receive an unspecified
2373 bit-pattern. Undefined values may be of any type (other than '``label``'
2374 or '``void``') and be used anywhere a constant is permitted.
2376 Undefined values are useful because they indicate to the compiler that
2377 the program is well defined no matter what value is used. This gives the
2378 compiler more freedom to optimize. Here are some examples of
2379 (potentially surprising) transformations that are valid (in pseudo IR):
2381 .. code-block:: llvm
2391 This is safe because all of the output bits are affected by the undef
2392 bits. Any output bit can have a zero or one depending on the input bits.
2394 .. code-block:: llvm
2405 These logical operations have bits that are not always affected by the
2406 input. For example, if ``%X`` has a zero bit, then the output of the
2407 '``and``' operation will always be a zero for that bit, no matter what
2408 the corresponding bit from the '``undef``' is. As such, it is unsafe to
2409 optimize or assume that the result of the '``and``' is '``undef``'.
2410 However, it is safe to assume that all bits of the '``undef``' could be
2411 0, and optimize the '``and``' to 0. Likewise, it is safe to assume that
2412 all the bits of the '``undef``' operand to the '``or``' could be set,
2413 allowing the '``or``' to be folded to -1.
2415 .. code-block:: llvm
2417 %A = select undef, %X, %Y
2418 %B = select undef, 42, %Y
2419 %C = select %X, %Y, undef
2429 This set of examples shows that undefined '``select``' (and conditional
2430 branch) conditions can go *either way*, but they have to come from one
2431 of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were
2432 both known to have a clear low bit, then ``%A`` would have to have a
2433 cleared low bit. However, in the ``%C`` example, the optimizer is
2434 allowed to assume that the '``undef``' operand could be the same as
2435 ``%Y``, allowing the whole '``select``' to be eliminated.
2437 .. code-block:: llvm
2439 %A = xor undef, undef
2456 This example points out that two '``undef``' operands are not
2457 necessarily the same. This can be surprising to people (and also matches
2458 C semantics) where they assume that "``X^X``" is always zero, even if
2459 ``X`` is undefined. This isn't true for a number of reasons, but the
2460 short answer is that an '``undef``' "variable" can arbitrarily change
2461 its value over its "live range". This is true because the variable
2462 doesn't actually *have a live range*. Instead, the value is logically
2463 read from arbitrary registers that happen to be around when needed, so
2464 the value is not necessarily consistent over time. In fact, ``%A`` and
2465 ``%C`` need to have the same semantics or the core LLVM "replace all
2466 uses with" concept would not hold.
2468 .. code-block:: llvm
2476 These examples show the crucial difference between an *undefined value*
2477 and *undefined behavior*. An undefined value (like '``undef``') is
2478 allowed to have an arbitrary bit-pattern. This means that the ``%A``
2479 operation can be constant folded to '``undef``', because the '``undef``'
2480 could be an SNaN, and ``fdiv`` is not (currently) defined on SNaN's.
2481 However, in the second example, we can make a more aggressive
2482 assumption: because the ``undef`` is allowed to be an arbitrary value,
2483 we are allowed to assume that it could be zero. Since a divide by zero
2484 has *undefined behavior*, we are allowed to assume that the operation
2485 does not execute at all. This allows us to delete the divide and all
2486 code after it. Because the undefined operation "can't happen", the
2487 optimizer can assume that it occurs in dead code.
2489 .. code-block:: llvm
2491 a: store undef -> %X
2492 b: store %X -> undef
2497 These examples reiterate the ``fdiv`` example: a store *of* an undefined
2498 value can be assumed to not have any effect; we can assume that the
2499 value is overwritten with bits that happen to match what was already
2500 there. However, a store *to* an undefined location could clobber
2501 arbitrary memory, therefore, it has undefined behavior.
2508 Poison values are similar to :ref:`undef values <undefvalues>`, however
2509 they also represent the fact that an instruction or constant expression
2510 that cannot evoke side effects has nevertheless detected a condition
2511 that results in undefined behavior.
2513 There is currently no way of representing a poison value in the IR; they
2514 only exist when produced by operations such as :ref:`add <i_add>` with
2517 Poison value behavior is defined in terms of value *dependence*:
2519 - Values other than :ref:`phi <i_phi>` nodes depend on their operands.
2520 - :ref:`Phi <i_phi>` nodes depend on the operand corresponding to
2521 their dynamic predecessor basic block.
2522 - Function arguments depend on the corresponding actual argument values
2523 in the dynamic callers of their functions.
2524 - :ref:`Call <i_call>` instructions depend on the :ref:`ret <i_ret>`
2525 instructions that dynamically transfer control back to them.
2526 - :ref:`Invoke <i_invoke>` instructions depend on the
2527 :ref:`ret <i_ret>`, :ref:`resume <i_resume>`, or exception-throwing
2528 call instructions that dynamically transfer control back to them.
2529 - Non-volatile loads and stores depend on the most recent stores to all
2530 of the referenced memory addresses, following the order in the IR
2531 (including loads and stores implied by intrinsics such as
2532 :ref:`@llvm.memcpy <int_memcpy>`.)
2533 - An instruction with externally visible side effects depends on the
2534 most recent preceding instruction with externally visible side
2535 effects, following the order in the IR. (This includes :ref:`volatile
2536 operations <volatile>`.)
2537 - An instruction *control-depends* on a :ref:`terminator
2538 instruction <terminators>` if the terminator instruction has
2539 multiple successors and the instruction is always executed when
2540 control transfers to one of the successors, and may not be executed
2541 when control is transferred to another.
2542 - Additionally, an instruction also *control-depends* on a terminator
2543 instruction if the set of instructions it otherwise depends on would
2544 be different if the terminator had transferred control to a different
2546 - Dependence is transitive.
2548 Poison values have the same behavior as :ref:`undef values <undefvalues>`,
2549 with the additional effect that any instruction that has a *dependence*
2550 on a poison value has undefined behavior.
2552 Here are some examples:
2554 .. code-block:: llvm
2557 %poison = sub nuw i32 0, 1 ; Results in a poison value.
2558 %still_poison = and i32 %poison, 0 ; 0, but also poison.
2559 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
2560 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
2562 store i32 %poison, i32* @g ; Poison value stored to memory.
2563 %poison2 = load i32* @g ; Poison value loaded back from memory.
2565 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
2567 %narrowaddr = bitcast i32* @g to i16*
2568 %wideaddr = bitcast i32* @g to i64*
2569 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2570 %poison4 = load i64* %wideaddr ; Returns a poison value.
2572 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2573 br i1 %cmp, label %true, label %end ; Branch to either destination.
2576 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2577 ; it has undefined behavior.
2581 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2582 ; Both edges into this PHI are
2583 ; control-dependent on %cmp, so this
2584 ; always results in a poison value.
2586 store volatile i32 0, i32* @g ; This would depend on the store in %true
2587 ; if %cmp is true, or the store in %entry
2588 ; otherwise, so this is undefined behavior.
2590 br i1 %cmp, label %second_true, label %second_end
2591 ; The same branch again, but this time the
2592 ; true block doesn't have side effects.
2599 store volatile i32 0, i32* @g ; This time, the instruction always depends
2600 ; on the store in %end. Also, it is
2601 ; control-equivalent to %end, so this is
2602 ; well-defined (ignoring earlier undefined
2603 ; behavior in this example).
2607 Addresses of Basic Blocks
2608 -------------------------
2610 ``blockaddress(@function, %block)``
2612 The '``blockaddress``' constant computes the address of the specified
2613 basic block in the specified function, and always has an ``i8*`` type.
2614 Taking the address of the entry block is illegal.
2616 This value only has defined behavior when used as an operand to the
2617 ':ref:`indirectbr <i_indirectbr>`' instruction, or for comparisons
2618 against null. Pointer equality tests between labels addresses results in
2619 undefined behavior --- though, again, comparison against null is ok, and
2620 no label is equal to the null pointer. This may be passed around as an
2621 opaque pointer sized value as long as the bits are not inspected. This
2622 allows ``ptrtoint`` and arithmetic to be performed on these values so
2623 long as the original value is reconstituted before the ``indirectbr``
2626 Finally, some targets may provide defined semantics when using the value
2627 as the operand to an inline assembly, but that is target specific.
2631 Constant Expressions
2632 --------------------
2634 Constant expressions are used to allow expressions involving other
2635 constants to be used as constants. Constant expressions may be of any
2636 :ref:`first class <t_firstclass>` type and may involve any LLVM operation
2637 that does not have side effects (e.g. load and call are not supported).
2638 The following is the syntax for constant expressions:
2640 ``trunc (CST to TYPE)``
2641 Truncate a constant to another type. The bit size of CST must be
2642 larger than the bit size of TYPE. Both types must be integers.
2643 ``zext (CST to TYPE)``
2644 Zero extend a constant to another type. The bit size of CST must be
2645 smaller than the bit size of TYPE. Both types must be integers.
2646 ``sext (CST to TYPE)``
2647 Sign extend a constant to another type. The bit size of CST must be
2648 smaller than the bit size of TYPE. Both types must be integers.
2649 ``fptrunc (CST to TYPE)``
2650 Truncate a floating point constant to another floating point type.
2651 The size of CST must be larger than the size of TYPE. Both types
2652 must be floating point.
2653 ``fpext (CST to TYPE)``
2654 Floating point extend a constant to another type. The size of CST
2655 must be smaller or equal to the size of TYPE. Both types must be
2657 ``fptoui (CST to TYPE)``
2658 Convert a floating point constant to the corresponding unsigned
2659 integer constant. TYPE must be a scalar or vector integer type. CST
2660 must be of scalar or vector floating point type. Both CST and TYPE
2661 must be scalars, or vectors of the same number of elements. If the
2662 value won't fit in the integer type, the results are undefined.
2663 ``fptosi (CST to TYPE)``
2664 Convert a floating point constant to the corresponding signed
2665 integer constant. TYPE must be a scalar or vector integer type. CST
2666 must be of scalar or vector floating point type. Both CST and TYPE
2667 must be scalars, or vectors of the same number of elements. If the
2668 value won't fit in the integer type, the results are undefined.
2669 ``uitofp (CST to TYPE)``
2670 Convert an unsigned integer constant to the corresponding floating
2671 point constant. TYPE must be a scalar or vector floating point type.
2672 CST must be of scalar or vector integer type. Both CST and TYPE must
2673 be scalars, or vectors of the same number of elements. If the value
2674 won't fit in the floating point type, the results are undefined.
2675 ``sitofp (CST to TYPE)``
2676 Convert a signed integer constant to the corresponding floating
2677 point constant. TYPE must be a scalar or vector floating point type.
2678 CST must be of scalar or vector integer type. Both CST and TYPE must
2679 be scalars, or vectors of the same number of elements. If the value
2680 won't fit in the floating point type, the results are undefined.
2681 ``ptrtoint (CST to TYPE)``
2682 Convert a pointer typed constant to the corresponding integer
2683 constant. ``TYPE`` must be an integer type. ``CST`` must be of
2684 pointer type. The ``CST`` value is zero extended, truncated, or
2685 unchanged to make it fit in ``TYPE``.
2686 ``inttoptr (CST to TYPE)``
2687 Convert an integer constant to a pointer constant. TYPE must be a
2688 pointer type. CST must be of integer type. The CST value is zero
2689 extended, truncated, or unchanged to make it fit in a pointer size.
2690 This one is *really* dangerous!
2691 ``bitcast (CST to TYPE)``
2692 Convert a constant, CST, to another TYPE. The constraints of the
2693 operands are the same as those for the :ref:`bitcast
2694 instruction <i_bitcast>`.
2695 ``addrspacecast (CST to TYPE)``
2696 Convert a constant pointer or constant vector of pointer, CST, to another
2697 TYPE in a different address space. The constraints of the operands are the
2698 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`.
2699 ``getelementptr (CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)``
2700 Perform the :ref:`getelementptr operation <i_getelementptr>` on
2701 constants. As with the :ref:`getelementptr <i_getelementptr>`
2702 instruction, the index list may have zero or more indexes, which are
2703 required to make sense for the type of "CSTPTR".
2704 ``select (COND, VAL1, VAL2)``
2705 Perform the :ref:`select operation <i_select>` on constants.
2706 ``icmp COND (VAL1, VAL2)``
2707 Performs the :ref:`icmp operation <i_icmp>` on constants.
2708 ``fcmp COND (VAL1, VAL2)``
2709 Performs the :ref:`fcmp operation <i_fcmp>` on constants.
2710 ``extractelement (VAL, IDX)``
2711 Perform the :ref:`extractelement operation <i_extractelement>` on
2713 ``insertelement (VAL, ELT, IDX)``
2714 Perform the :ref:`insertelement operation <i_insertelement>` on
2716 ``shufflevector (VEC1, VEC2, IDXMASK)``
2717 Perform the :ref:`shufflevector operation <i_shufflevector>` on
2719 ``extractvalue (VAL, IDX0, IDX1, ...)``
2720 Perform the :ref:`extractvalue operation <i_extractvalue>` on
2721 constants. The index list is interpreted in a similar manner as
2722 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At
2723 least one index value must be specified.
2724 ``insertvalue (VAL, ELT, IDX0, IDX1, ...)``
2725 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants.
2726 The index list is interpreted in a similar manner as indices in a
2727 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index
2728 value must be specified.
2729 ``OPCODE (LHS, RHS)``
2730 Perform the specified operation of the LHS and RHS constants. OPCODE
2731 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise
2732 binary <bitwiseops>` operations. The constraints on operands are
2733 the same as those for the corresponding instruction (e.g. no bitwise
2734 operations on floating point values are allowed).
2741 Inline Assembler Expressions
2742 ----------------------------
2744 LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level
2745 Inline Assembly <moduleasm>`) through the use of a special value. This
2746 value represents the inline assembler as a string (containing the
2747 instructions to emit), a list of operand constraints (stored as a
2748 string), a flag that indicates whether or not the inline asm expression
2749 has side effects, and a flag indicating whether the function containing
2750 the asm needs to align its stack conservatively. An example inline
2751 assembler expression is:
2753 .. code-block:: llvm
2755 i32 (i32) asm "bswap $0", "=r,r"
2757 Inline assembler expressions may **only** be used as the callee operand
2758 of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction.
2759 Thus, typically we have:
2761 .. code-block:: llvm
2763 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y)
2765 Inline asms with side effects not visible in the constraint list must be
2766 marked as having side effects. This is done through the use of the
2767 '``sideeffect``' keyword, like so:
2769 .. code-block:: llvm
2771 call void asm sideeffect "eieio", ""()
2773 In some cases inline asms will contain code that will not work unless
2774 the stack is aligned in some way, such as calls or SSE instructions on
2775 x86, yet will not contain code that does that alignment within the asm.
2776 The compiler should make conservative assumptions about what the asm
2777 might contain and should generate its usual stack alignment code in the
2778 prologue if the '``alignstack``' keyword is present:
2780 .. code-block:: llvm
2782 call void asm alignstack "eieio", ""()
2784 Inline asms also support using non-standard assembly dialects. The
2785 assumed dialect is ATT. When the '``inteldialect``' keyword is present,
2786 the inline asm is using the Intel dialect. Currently, ATT and Intel are
2787 the only supported dialects. An example is:
2789 .. code-block:: llvm
2791 call void asm inteldialect "eieio", ""()
2793 If multiple keywords appear the '``sideeffect``' keyword must come
2794 first, the '``alignstack``' keyword second and the '``inteldialect``'
2800 The call instructions that wrap inline asm nodes may have a
2801 "``!srcloc``" MDNode attached to it that contains a list of constant
2802 integers. If present, the code generator will use the integer as the
2803 location cookie value when report errors through the ``LLVMContext``
2804 error reporting mechanisms. This allows a front-end to correlate backend
2805 errors that occur with inline asm back to the source code that produced
2808 .. code-block:: llvm
2810 call void asm sideeffect "something bad", ""(), !srcloc !42
2812 !42 = !{ i32 1234567 }
2814 It is up to the front-end to make sense of the magic numbers it places
2815 in the IR. If the MDNode contains multiple constants, the code generator
2816 will use the one that corresponds to the line of the asm that the error
2824 LLVM IR allows metadata to be attached to instructions in the program
2825 that can convey extra information about the code to the optimizers and
2826 code generator. One example application of metadata is source-level
2827 debug information. There are two metadata primitives: strings and nodes.
2829 Metadata does not have a type, and is not a value. If referenced from a
2830 ``call`` instruction, it uses the ``metadata`` type.
2832 All metadata are identified in syntax by a exclamation point ('``!``').
2834 Metadata Nodes and Metadata Strings
2835 -----------------------------------
2837 A metadata string is a string surrounded by double quotes. It can
2838 contain any character by escaping non-printable characters with
2839 "``\xx``" where "``xx``" is the two digit hex code. For example:
2842 Metadata nodes are represented with notation similar to structure
2843 constants (a comma separated list of elements, surrounded by braces and
2844 preceded by an exclamation point). Metadata nodes can have any values as
2845 their operand. For example:
2847 .. code-block:: llvm
2849 !{ !"test\00", i32 10}
2851 Metadata nodes that aren't uniqued use the ``distinct`` keyword. For example:
2853 .. code-block:: llvm
2855 !0 = distinct !{!"test\00", i32 10}
2857 ``distinct`` nodes are useful when nodes shouldn't be merged based on their
2858 content. They can also occur when transformations cause uniquing collisions
2859 when metadata operands change.
2861 A :ref:`named metadata <namedmetadatastructure>` is a collection of
2862 metadata nodes, which can be looked up in the module symbol table. For
2865 .. code-block:: llvm
2869 Metadata can be used as function arguments. Here ``llvm.dbg.value``
2870 function is using two metadata arguments:
2872 .. code-block:: llvm
2874 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2876 Metadata can be attached with an instruction. Here metadata ``!21`` is
2877 attached to the ``add`` instruction using the ``!dbg`` identifier:
2879 .. code-block:: llvm
2881 %indvar.next = add i64 %indvar, 1, !dbg !21
2883 More information about specific metadata nodes recognized by the
2884 optimizers and code generator is found below.
2889 In LLVM IR, memory does not have types, so LLVM's own type system is not
2890 suitable for doing TBAA. Instead, metadata is added to the IR to
2891 describe a type system of a higher level language. This can be used to
2892 implement typical C/C++ TBAA, but it can also be used to implement
2893 custom alias analysis behavior for other languages.
2895 The current metadata format is very simple. TBAA metadata nodes have up
2896 to three fields, e.g.:
2898 .. code-block:: llvm
2900 !0 = !{ !"an example type tree" }
2901 !1 = !{ !"int", !0 }
2902 !2 = !{ !"float", !0 }
2903 !3 = !{ !"const float", !2, i64 1 }
2905 The first field is an identity field. It can be any value, usually a
2906 metadata string, which uniquely identifies the type. The most important
2907 name in the tree is the name of the root node. Two trees with different
2908 root node names are entirely disjoint, even if they have leaves with
2911 The second field identifies the type's parent node in the tree, or is
2912 null or omitted for a root node. A type is considered to alias all of
2913 its descendants and all of its ancestors in the tree. Also, a type is
2914 considered to alias all types in other trees, so that bitcode produced
2915 from multiple front-ends is handled conservatively.
2917 If the third field is present, it's an integer which if equal to 1
2918 indicates that the type is "constant" (meaning
2919 ``pointsToConstantMemory`` should return true; see `other useful
2920 AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_).
2922 '``tbaa.struct``' Metadata
2923 ^^^^^^^^^^^^^^^^^^^^^^^^^^
2925 The :ref:`llvm.memcpy <int_memcpy>` is often used to implement
2926 aggregate assignment operations in C and similar languages, however it
2927 is defined to copy a contiguous region of memory, which is more than
2928 strictly necessary for aggregate types which contain holes due to
2929 padding. Also, it doesn't contain any TBAA information about the fields
2932 ``!tbaa.struct`` metadata can describe which memory subregions in a
2933 memcpy are padding and what the TBAA tags of the struct are.
2935 The current metadata format is very simple. ``!tbaa.struct`` metadata
2936 nodes are a list of operands which are in conceptual groups of three.
2937 For each group of three, the first operand gives the byte offset of a
2938 field in bytes, the second gives its size in bytes, and the third gives
2941 .. code-block:: llvm
2943 !4 = !{ i64 0, i64 4, !1, i64 8, i64 4, !2 }
2945 This describes a struct with two fields. The first is at offset 0 bytes
2946 with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes
2947 and has size 4 bytes and has tbaa tag !2.
2949 Note that the fields need not be contiguous. In this example, there is a
2950 4 byte gap between the two fields. This gap represents padding which
2951 does not carry useful data and need not be preserved.
2953 '``noalias``' and '``alias.scope``' Metadata
2954 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2956 ``noalias`` and ``alias.scope`` metadata provide the ability to specify generic
2957 noalias memory-access sets. This means that some collection of memory access
2958 instructions (loads, stores, memory-accessing calls, etc.) that carry
2959 ``noalias`` metadata can specifically be specified not to alias with some other
2960 collection of memory access instructions that carry ``alias.scope`` metadata.
2961 Each type of metadata specifies a list of scopes where each scope has an id and
2962 a domain. When evaluating an aliasing query, if for some some domain, the set
2963 of scopes with that domain in one instruction's ``alias.scope`` list is a
2964 subset of (or qual to) the set of scopes for that domain in another
2965 instruction's ``noalias`` list, then the two memory accesses are assumed not to
2968 The metadata identifying each domain is itself a list containing one or two
2969 entries. The first entry is the name of the domain. Note that if the name is a
2970 string then it can be combined accross functions and translation units. A
2971 self-reference can be used to create globally unique domain names. A
2972 descriptive string may optionally be provided as a second list entry.
2974 The metadata identifying each scope is also itself a list containing two or
2975 three entries. The first entry is the name of the scope. Note that if the name
2976 is a string then it can be combined accross functions and translation units. A
2977 self-reference can be used to create globally unique scope names. A metadata
2978 reference to the scope's domain is the second entry. A descriptive string may
2979 optionally be provided as a third list entry.
2983 .. code-block:: llvm
2985 ; Two scope domains:
2989 ; Some scopes in these domains:
2995 !5 = !{!4} ; A list containing only scope !4
2999 ; These two instructions don't alias:
3000 %0 = load float* %c, align 4, !alias.scope !5
3001 store float %0, float* %arrayidx.i, align 4, !noalias !5
3003 ; These two instructions also don't alias (for domain !1, the set of scopes
3004 ; in the !alias.scope equals that in the !noalias list):
3005 %2 = load float* %c, align 4, !alias.scope !5
3006 store float %2, float* %arrayidx.i2, align 4, !noalias !6
3008 ; These two instructions don't alias (for domain !0, the set of scopes in
3009 ; the !noalias list is not a superset of, or equal to, the scopes in the
3010 ; !alias.scope list):
3011 %2 = load float* %c, align 4, !alias.scope !6
3012 store float %0, float* %arrayidx.i, align 4, !noalias !7
3014 '``fpmath``' Metadata
3015 ^^^^^^^^^^^^^^^^^^^^^
3017 ``fpmath`` metadata may be attached to any instruction of floating point
3018 type. It can be used to express the maximum acceptable error in the
3019 result of that instruction, in ULPs, thus potentially allowing the
3020 compiler to use a more efficient but less accurate method of computing
3021 it. ULP is defined as follows:
3023 If ``x`` is a real number that lies between two finite consecutive
3024 floating-point numbers ``a`` and ``b``, without being equal to one
3025 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the
3026 distance between the two non-equal finite floating-point numbers
3027 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``.
3029 The metadata node shall consist of a single positive floating point
3030 number representing the maximum relative error, for example:
3032 .. code-block:: llvm
3034 !0 = !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
3036 '``range``' Metadata
3037 ^^^^^^^^^^^^^^^^^^^^
3039 ``range`` metadata may be attached only to ``load``, ``call`` and ``invoke`` of
3040 integer types. It expresses the possible ranges the loaded value or the value
3041 returned by the called function at this call site is in. The ranges are
3042 represented with a flattened list of integers. The loaded value or the value
3043 returned is known to be in the union of the ranges defined by each consecutive
3044 pair. Each pair has the following properties:
3046 - The type must match the type loaded by the instruction.
3047 - The pair ``a,b`` represents the range ``[a,b)``.
3048 - Both ``a`` and ``b`` are constants.
3049 - The range is allowed to wrap.
3050 - The range should not represent the full or empty set. That is,
3053 In addition, the pairs must be in signed order of the lower bound and
3054 they must be non-contiguous.
3058 .. code-block:: llvm
3060 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1
3061 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
3062 %c = call i8 @foo(), !range !2 ; Can only be 0, 1, 3, 4 or 5
3063 %d = invoke i8 @bar() to label %cont
3064 unwind label %lpad, !range !3 ; Can only be -2, -1, 3, 4 or 5
3066 !0 = !{ i8 0, i8 2 }
3067 !1 = !{ i8 255, i8 2 }
3068 !2 = !{ i8 0, i8 2, i8 3, i8 6 }
3069 !3 = !{ i8 -2, i8 0, i8 3, i8 6 }
3074 It is sometimes useful to attach information to loop constructs. Currently,
3075 loop metadata is implemented as metadata attached to the branch instruction
3076 in the loop latch block. This type of metadata refer to a metadata node that is
3077 guaranteed to be separate for each loop. The loop identifier metadata is
3078 specified with the name ``llvm.loop``.
3080 The loop identifier metadata is implemented using a metadata that refers to
3081 itself to avoid merging it with any other identifier metadata, e.g.,
3082 during module linkage or function inlining. That is, each loop should refer
3083 to their own identification metadata even if they reside in separate functions.
3084 The following example contains loop identifier metadata for two separate loop
3087 .. code-block:: llvm
3092 The loop identifier metadata can be used to specify additional
3093 per-loop metadata. Any operands after the first operand can be treated
3094 as user-defined metadata. For example the ``llvm.loop.unroll.count``
3095 suggests an unroll factor to the loop unroller:
3097 .. code-block:: llvm
3099 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0
3102 !1 = !{!"llvm.loop.unroll.count", i32 4}
3104 '``llvm.loop.vectorize``' and '``llvm.loop.interleave``'
3105 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3107 Metadata prefixed with ``llvm.loop.vectorize`` or ``llvm.loop.interleave`` are
3108 used to control per-loop vectorization and interleaving parameters such as
3109 vectorization width and interleave count. These metadata should be used in
3110 conjunction with ``llvm.loop`` loop identification metadata. The
3111 ``llvm.loop.vectorize`` and ``llvm.loop.interleave`` metadata are only
3112 optimization hints and the optimizer will only interleave and vectorize loops if
3113 it believes it is safe to do so. The ``llvm.mem.parallel_loop_access`` metadata
3114 which contains information about loop-carried memory dependencies can be helpful
3115 in determining the safety of these transformations.
3117 '``llvm.loop.interleave.count``' Metadata
3118 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3120 This metadata suggests an interleave count to the loop interleaver.
3121 The first operand is the string ``llvm.loop.interleave.count`` and the
3122 second operand is an integer specifying the interleave count. For
3125 .. code-block:: llvm
3127 !0 = !{!"llvm.loop.interleave.count", i32 4}
3129 Note that setting ``llvm.loop.interleave.count`` to 1 disables interleaving
3130 multiple iterations of the loop. If ``llvm.loop.interleave.count`` is set to 0
3131 then the interleave count will be determined automatically.
3133 '``llvm.loop.vectorize.enable``' Metadata
3134 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3136 This metadata selectively enables or disables vectorization for the loop. The
3137 first operand is the string ``llvm.loop.vectorize.enable`` and the second operand
3138 is a bit. If the bit operand value is 1 vectorization is enabled. A value of
3139 0 disables vectorization:
3141 .. code-block:: llvm
3143 !0 = !{!"llvm.loop.vectorize.enable", i1 0}
3144 !1 = !{!"llvm.loop.vectorize.enable", i1 1}
3146 '``llvm.loop.vectorize.width``' Metadata
3147 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3149 This metadata sets the target width of the vectorizer. The first
3150 operand is the string ``llvm.loop.vectorize.width`` and the second
3151 operand is an integer specifying the width. For example:
3153 .. code-block:: llvm
3155 !0 = !{!"llvm.loop.vectorize.width", i32 4}
3157 Note that setting ``llvm.loop.vectorize.width`` to 1 disables
3158 vectorization of the loop. If ``llvm.loop.vectorize.width`` is set to
3159 0 or if the loop does not have this metadata the width will be
3160 determined automatically.
3162 '``llvm.loop.unroll``'
3163 ^^^^^^^^^^^^^^^^^^^^^^
3165 Metadata prefixed with ``llvm.loop.unroll`` are loop unrolling
3166 optimization hints such as the unroll factor. ``llvm.loop.unroll``
3167 metadata should be used in conjunction with ``llvm.loop`` loop
3168 identification metadata. The ``llvm.loop.unroll`` metadata are only
3169 optimization hints and the unrolling will only be performed if the
3170 optimizer believes it is safe to do so.
3172 '``llvm.loop.unroll.count``' Metadata
3173 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3175 This metadata suggests an unroll factor to the loop unroller. The
3176 first operand is the string ``llvm.loop.unroll.count`` and the second
3177 operand is a positive integer specifying the unroll factor. For
3180 .. code-block:: llvm
3182 !0 = !{!"llvm.loop.unroll.count", i32 4}
3184 If the trip count of the loop is less than the unroll count the loop
3185 will be partially unrolled.
3187 '``llvm.loop.unroll.disable``' Metadata
3188 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3190 This metadata either disables loop unrolling. The metadata has a single operand
3191 which is the string ``llvm.loop.unroll.disable``. For example:
3193 .. code-block:: llvm
3195 !0 = !{!"llvm.loop.unroll.disable"}
3197 '``llvm.loop.unroll.full``' Metadata
3198 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3200 This metadata either suggests that the loop should be unrolled fully. The
3201 metadata has a single operand which is the string ``llvm.loop.unroll.disable``.
3204 .. code-block:: llvm
3206 !0 = !{!"llvm.loop.unroll.full"}
3211 Metadata types used to annotate memory accesses with information helpful
3212 for optimizations are prefixed with ``llvm.mem``.
3214 '``llvm.mem.parallel_loop_access``' Metadata
3215 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3217 The ``llvm.mem.parallel_loop_access`` metadata refers to a loop identifier,
3218 or metadata containing a list of loop identifiers for nested loops.
3219 The metadata is attached to memory accessing instructions and denotes that
3220 no loop carried memory dependence exist between it and other instructions denoted
3221 with the same loop identifier.
3223 Precisely, given two instructions ``m1`` and ``m2`` that both have the
3224 ``llvm.mem.parallel_loop_access`` metadata, with ``L1`` and ``L2`` being the
3225 set of loops associated with that metadata, respectively, then there is no loop
3226 carried dependence between ``m1`` and ``m2`` for loops in both ``L1`` and
3229 As a special case, if all memory accessing instructions in a loop have
3230 ``llvm.mem.parallel_loop_access`` metadata that refers to that loop, then the
3231 loop has no loop carried memory dependences and is considered to be a parallel
3234 Note that if not all memory access instructions have such metadata referring to
3235 the loop, then the loop is considered not being trivially parallel. Additional
3236 memory dependence analysis is required to make that determination. As a fail
3237 safe mechanism, this causes loops that were originally parallel to be considered
3238 sequential (if optimization passes that are unaware of the parallel semantics
3239 insert new memory instructions into the loop body).
3241 Example of a loop that is considered parallel due to its correct use of
3242 both ``llvm.loop`` and ``llvm.mem.parallel_loop_access``
3243 metadata types that refer to the same loop identifier metadata.
3245 .. code-block:: llvm
3249 %val0 = load i32* %arrayidx, !llvm.mem.parallel_loop_access !0
3251 store i32 %val0, i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
3253 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0
3259 It is also possible to have nested parallel loops. In that case the
3260 memory accesses refer to a list of loop identifier metadata nodes instead of
3261 the loop identifier metadata node directly:
3263 .. code-block:: llvm
3267 %val1 = load i32* %arrayidx3, !llvm.mem.parallel_loop_access !2
3269 br label %inner.for.body
3273 %val0 = load i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
3275 store i32 %val0, i32* %arrayidx2, !llvm.mem.parallel_loop_access !0
3277 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1
3281 store i32 %val1, i32* %arrayidx4, !llvm.mem.parallel_loop_access !2
3283 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2
3285 outer.for.end: ; preds = %for.body
3287 !0 = !{!1, !2} ; a list of loop identifiers
3288 !1 = !{!1} ; an identifier for the inner loop
3289 !2 = !{!2} ; an identifier for the outer loop
3291 Module Flags Metadata
3292 =====================
3294 Information about the module as a whole is difficult to convey to LLVM's
3295 subsystems. The LLVM IR isn't sufficient to transmit this information.
3296 The ``llvm.module.flags`` named metadata exists in order to facilitate
3297 this. These flags are in the form of key / value pairs --- much like a
3298 dictionary --- making it easy for any subsystem who cares about a flag to
3301 The ``llvm.module.flags`` metadata contains a list of metadata triplets.
3302 Each triplet has the following form:
3304 - The first element is a *behavior* flag, which specifies the behavior
3305 when two (or more) modules are merged together, and it encounters two
3306 (or more) metadata with the same ID. The supported behaviors are
3308 - The second element is a metadata string that is a unique ID for the
3309 metadata. Each module may only have one flag entry for each unique ID (not
3310 including entries with the **Require** behavior).
3311 - The third element is the value of the flag.
3313 When two (or more) modules are merged together, the resulting
3314 ``llvm.module.flags`` metadata is the union of the modules' flags. That is, for
3315 each unique metadata ID string, there will be exactly one entry in the merged
3316 modules ``llvm.module.flags`` metadata table, and the value for that entry will
3317 be determined by the merge behavior flag, as described below. The only exception
3318 is that entries with the *Require* behavior are always preserved.
3320 The following behaviors are supported:
3331 Emits an error if two values disagree, otherwise the resulting value
3332 is that of the operands.
3336 Emits a warning if two values disagree. The result value will be the
3337 operand for the flag from the first module being linked.
3341 Adds a requirement that another module flag be present and have a
3342 specified value after linking is performed. The value must be a
3343 metadata pair, where the first element of the pair is the ID of the
3344 module flag to be restricted, and the second element of the pair is
3345 the value the module flag should be restricted to. This behavior can
3346 be used to restrict the allowable results (via triggering of an
3347 error) of linking IDs with the **Override** behavior.
3351 Uses the specified value, regardless of the behavior or value of the
3352 other module. If both modules specify **Override**, but the values
3353 differ, an error will be emitted.
3357 Appends the two values, which are required to be metadata nodes.
3361 Appends the two values, which are required to be metadata
3362 nodes. However, duplicate entries in the second list are dropped
3363 during the append operation.
3365 It is an error for a particular unique flag ID to have multiple behaviors,
3366 except in the case of **Require** (which adds restrictions on another metadata
3367 value) or **Override**.
3369 An example of module flags:
3371 .. code-block:: llvm
3373 !0 = !{ i32 1, !"foo", i32 1 }
3374 !1 = !{ i32 4, !"bar", i32 37 }
3375 !2 = !{ i32 2, !"qux", i32 42 }
3376 !3 = !{ i32 3, !"qux",
3381 !llvm.module.flags = !{ !0, !1, !2, !3 }
3383 - Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior
3384 if two or more ``!"foo"`` flags are seen is to emit an error if their
3385 values are not equal.
3387 - Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The
3388 behavior if two or more ``!"bar"`` flags are seen is to use the value
3391 - Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The
3392 behavior if two or more ``!"qux"`` flags are seen is to emit a
3393 warning if their values are not equal.
3395 - Metadata ``!3`` has the ID ``!"qux"`` and the value:
3401 The behavior is to emit an error if the ``llvm.module.flags`` does not
3402 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is
3405 Objective-C Garbage Collection Module Flags Metadata
3406 ----------------------------------------------------
3408 On the Mach-O platform, Objective-C stores metadata about garbage
3409 collection in a special section called "image info". The metadata
3410 consists of a version number and a bitmask specifying what types of
3411 garbage collection are supported (if any) by the file. If two or more
3412 modules are linked together their garbage collection metadata needs to
3413 be merged rather than appended together.
3415 The Objective-C garbage collection module flags metadata consists of the
3416 following key-value pairs:
3425 * - ``Objective-C Version``
3426 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2.
3428 * - ``Objective-C Image Info Version``
3429 - **[Required]** --- The version of the image info section. Currently
3432 * - ``Objective-C Image Info Section``
3433 - **[Required]** --- The section to place the metadata. Valid values are
3434 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and
3435 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for
3436 Objective-C ABI version 2.
3438 * - ``Objective-C Garbage Collection``
3439 - **[Required]** --- Specifies whether garbage collection is supported or
3440 not. Valid values are 0, for no garbage collection, and 2, for garbage
3441 collection supported.
3443 * - ``Objective-C GC Only``
3444 - **[Optional]** --- Specifies that only garbage collection is supported.
3445 If present, its value must be 6. This flag requires that the
3446 ``Objective-C Garbage Collection`` flag have the value 2.
3448 Some important flag interactions:
3450 - If a module with ``Objective-C Garbage Collection`` set to 0 is
3451 merged with a module with ``Objective-C Garbage Collection`` set to
3452 2, then the resulting module has the
3453 ``Objective-C Garbage Collection`` flag set to 0.
3454 - A module with ``Objective-C Garbage Collection`` set to 0 cannot be
3455 merged with a module with ``Objective-C GC Only`` set to 6.
3457 Automatic Linker Flags Module Flags Metadata
3458 --------------------------------------------
3460 Some targets support embedding flags to the linker inside individual object
3461 files. Typically this is used in conjunction with language extensions which
3462 allow source files to explicitly declare the libraries they depend on, and have
3463 these automatically be transmitted to the linker via object files.
3465 These flags are encoded in the IR using metadata in the module flags section,
3466 using the ``Linker Options`` key. The merge behavior for this flag is required
3467 to be ``AppendUnique``, and the value for the key is expected to be a metadata
3468 node which should be a list of other metadata nodes, each of which should be a
3469 list of metadata strings defining linker options.
3471 For example, the following metadata section specifies two separate sets of
3472 linker options, presumably to link against ``libz`` and the ``Cocoa``
3475 !0 = !{ i32 6, !"Linker Options",
3478 !{ !"-framework", !"Cocoa" } } }
3479 !llvm.module.flags = !{ !0 }
3481 The metadata encoding as lists of lists of options, as opposed to a collapsed
3482 list of options, is chosen so that the IR encoding can use multiple option
3483 strings to specify e.g., a single library, while still having that specifier be
3484 preserved as an atomic element that can be recognized by a target specific
3485 assembly writer or object file emitter.
3487 Each individual option is required to be either a valid option for the target's
3488 linker, or an option that is reserved by the target specific assembly writer or
3489 object file emitter. No other aspect of these options is defined by the IR.
3491 C type width Module Flags Metadata
3492 ----------------------------------
3494 The ARM backend emits a section into each generated object file describing the
3495 options that it was compiled with (in a compiler-independent way) to prevent
3496 linking incompatible objects, and to allow automatic library selection. Some
3497 of these options are not visible at the IR level, namely wchar_t width and enum
3500 To pass this information to the backend, these options are encoded in module
3501 flags metadata, using the following key-value pairs:
3511 - * 0 --- sizeof(wchar_t) == 4
3512 * 1 --- sizeof(wchar_t) == 2
3515 - * 0 --- Enums are at least as large as an ``int``.
3516 * 1 --- Enums are stored in the smallest integer type which can
3517 represent all of its values.
3519 For example, the following metadata section specifies that the module was
3520 compiled with a ``wchar_t`` width of 4 bytes, and the underlying type of an
3521 enum is the smallest type which can represent all of its values::
3523 !llvm.module.flags = !{!0, !1}
3524 !0 = !{i32 1, !"short_wchar", i32 1}
3525 !1 = !{i32 1, !"short_enum", i32 0}
3527 .. _intrinsicglobalvariables:
3529 Intrinsic Global Variables
3530 ==========================
3532 LLVM has a number of "magic" global variables that contain data that
3533 affect code generation or other IR semantics. These are documented here.
3534 All globals of this sort should have a section specified as
3535 "``llvm.metadata``". This section and all globals that start with
3536 "``llvm.``" are reserved for use by LLVM.
3540 The '``llvm.used``' Global Variable
3541 -----------------------------------
3543 The ``@llvm.used`` global is an array which has
3544 :ref:`appending linkage <linkage_appending>`. This array contains a list of
3545 pointers to named global variables, functions and aliases which may optionally
3546 have a pointer cast formed of bitcast or getelementptr. For example, a legal
3549 .. code-block:: llvm
3554 @llvm.used = appending global [2 x i8*] [
3556 i8* bitcast (i32* @Y to i8*)
3557 ], section "llvm.metadata"
3559 If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler,
3560 and linker are required to treat the symbol as if there is a reference to the
3561 symbol that it cannot see (which is why they have to be named). For example, if
3562 a variable has internal linkage and no references other than that from the
3563 ``@llvm.used`` list, it cannot be deleted. This is commonly used to represent
3564 references from inline asms and other things the compiler cannot "see", and
3565 corresponds to "``attribute((used))``" in GNU C.
3567 On some targets, the code generator must emit a directive to the
3568 assembler or object file to prevent the assembler and linker from
3569 molesting the symbol.
3571 .. _gv_llvmcompilerused:
3573 The '``llvm.compiler.used``' Global Variable
3574 --------------------------------------------
3576 The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used``
3577 directive, except that it only prevents the compiler from touching the
3578 symbol. On targets that support it, this allows an intelligent linker to
3579 optimize references to the symbol without being impeded as it would be
3582 This is a rare construct that should only be used in rare circumstances,
3583 and should not be exposed to source languages.
3585 .. _gv_llvmglobalctors:
3587 The '``llvm.global_ctors``' Global Variable
3588 -------------------------------------------
3590 .. code-block:: llvm
3592 %0 = type { i32, void ()*, i8* }
3593 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor, i8* @data }]
3595 The ``@llvm.global_ctors`` array contains a list of constructor
3596 functions, priorities, and an optional associated global or function.
3597 The functions referenced by this array will be called in ascending order
3598 of priority (i.e. lowest first) when the module is loaded. The order of
3599 functions with the same priority is not defined.
3601 If the third field is present, non-null, and points to a global variable
3602 or function, the initializer function will only run if the associated
3603 data from the current module is not discarded.
3605 .. _llvmglobaldtors:
3607 The '``llvm.global_dtors``' Global Variable
3608 -------------------------------------------
3610 .. code-block:: llvm
3612 %0 = type { i32, void ()*, i8* }
3613 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor, i8* @data }]
3615 The ``@llvm.global_dtors`` array contains a list of destructor
3616 functions, priorities, and an optional associated global or function.
3617 The functions referenced by this array will be called in descending
3618 order of priority (i.e. highest first) when the module is unloaded. The
3619 order of functions with the same priority is not defined.
3621 If the third field is present, non-null, and points to a global variable
3622 or function, the destructor function will only run if the associated
3623 data from the current module is not discarded.
3625 Instruction Reference
3626 =====================
3628 The LLVM instruction set consists of several different classifications
3629 of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary
3630 instructions <binaryops>`, :ref:`bitwise binary
3631 instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and
3632 :ref:`other instructions <otherops>`.
3636 Terminator Instructions
3637 -----------------------
3639 As mentioned :ref:`previously <functionstructure>`, every basic block in a
3640 program ends with a "Terminator" instruction, which indicates which
3641 block should be executed after the current block is finished. These
3642 terminator instructions typically yield a '``void``' value: they produce
3643 control flow, not values (the one exception being the
3644 ':ref:`invoke <i_invoke>`' instruction).
3646 The terminator instructions are: ':ref:`ret <i_ret>`',
3647 ':ref:`br <i_br>`', ':ref:`switch <i_switch>`',
3648 ':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`',
3649 ':ref:`resume <i_resume>`', and ':ref:`unreachable <i_unreachable>`'.
3653 '``ret``' Instruction
3654 ^^^^^^^^^^^^^^^^^^^^^
3661 ret <type> <value> ; Return a value from a non-void function
3662 ret void ; Return from void function
3667 The '``ret``' instruction is used to return control flow (and optionally
3668 a value) from a function back to the caller.
3670 There are two forms of the '``ret``' instruction: one that returns a
3671 value and then causes control flow, and one that just causes control
3677 The '``ret``' instruction optionally accepts a single argument, the
3678 return value. The type of the return value must be a ':ref:`first
3679 class <t_firstclass>`' type.
3681 A function is not :ref:`well formed <wellformed>` if it it has a non-void
3682 return type and contains a '``ret``' instruction with no return value or
3683 a return value with a type that does not match its type, or if it has a
3684 void return type and contains a '``ret``' instruction with a return
3690 When the '``ret``' instruction is executed, control flow returns back to
3691 the calling function's context. If the caller is a
3692 ":ref:`call <i_call>`" instruction, execution continues at the
3693 instruction after the call. If the caller was an
3694 ":ref:`invoke <i_invoke>`" instruction, execution continues at the
3695 beginning of the "normal" destination block. If the instruction returns
3696 a value, that value shall set the call or invoke instruction's return
3702 .. code-block:: llvm
3704 ret i32 5 ; Return an integer value of 5
3705 ret void ; Return from a void function
3706 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2
3710 '``br``' Instruction
3711 ^^^^^^^^^^^^^^^^^^^^
3718 br i1 <cond>, label <iftrue>, label <iffalse>
3719 br label <dest> ; Unconditional branch
3724 The '``br``' instruction is used to cause control flow to transfer to a
3725 different basic block in the current function. There are two forms of
3726 this instruction, corresponding to a conditional branch and an
3727 unconditional branch.
3732 The conditional branch form of the '``br``' instruction takes a single
3733 '``i1``' value and two '``label``' values. The unconditional form of the
3734 '``br``' instruction takes a single '``label``' value as a target.
3739 Upon execution of a conditional '``br``' instruction, the '``i1``'
3740 argument is evaluated. If the value is ``true``, control flows to the
3741 '``iftrue``' ``label`` argument. If "cond" is ``false``, control flows
3742 to the '``iffalse``' ``label`` argument.
3747 .. code-block:: llvm
3750 %cond = icmp eq i32 %a, %b
3751 br i1 %cond, label %IfEqual, label %IfUnequal
3759 '``switch``' Instruction
3760 ^^^^^^^^^^^^^^^^^^^^^^^^
3767 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3772 The '``switch``' instruction is used to transfer control flow to one of
3773 several different places. It is a generalization of the '``br``'
3774 instruction, allowing a branch to occur to one of many possible
3780 The '``switch``' instruction uses three parameters: an integer
3781 comparison value '``value``', a default '``label``' destination, and an
3782 array of pairs of comparison value constants and '``label``'s. The table
3783 is not allowed to contain duplicate constant entries.
3788 The ``switch`` instruction specifies a table of values and destinations.
3789 When the '``switch``' instruction is executed, this table is searched
3790 for the given value. If the value is found, control flow is transferred
3791 to the corresponding destination; otherwise, control flow is transferred
3792 to the default destination.
3797 Depending on properties of the target machine and the particular
3798 ``switch`` instruction, this instruction may be code generated in
3799 different ways. For example, it could be generated as a series of
3800 chained conditional branches or with a lookup table.
3805 .. code-block:: llvm
3807 ; Emulate a conditional br instruction
3808 %Val = zext i1 %value to i32
3809 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3811 ; Emulate an unconditional br instruction
3812 switch i32 0, label %dest [ ]
3814 ; Implement a jump table:
3815 switch i32 %val, label %otherwise [ i32 0, label %onzero
3817 i32 2, label %ontwo ]
3821 '``indirectbr``' Instruction
3822 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3829 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3834 The '``indirectbr``' instruction implements an indirect branch to a
3835 label within the current function, whose address is specified by
3836 "``address``". Address must be derived from a
3837 :ref:`blockaddress <blockaddress>` constant.
3842 The '``address``' argument is the address of the label to jump to. The
3843 rest of the arguments indicate the full set of possible destinations
3844 that the address may point to. Blocks are allowed to occur multiple
3845 times in the destination list, though this isn't particularly useful.
3847 This destination list is required so that dataflow analysis has an
3848 accurate understanding of the CFG.
3853 Control transfers to the block specified in the address argument. All
3854 possible destination blocks must be listed in the label list, otherwise
3855 this instruction has undefined behavior. This implies that jumps to
3856 labels defined in other functions have undefined behavior as well.
3861 This is typically implemented with a jump through a register.
3866 .. code-block:: llvm
3868 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3872 '``invoke``' Instruction
3873 ^^^^^^^^^^^^^^^^^^^^^^^^
3880 <result> = invoke [cconv] [ret attrs] <ptr to function ty> <function ptr val>(<function args>) [fn attrs]
3881 to label <normal label> unwind label <exception label>
3886 The '``invoke``' instruction causes control to transfer to a specified
3887 function, with the possibility of control flow transfer to either the
3888 '``normal``' label or the '``exception``' label. If the callee function
3889 returns with the "``ret``" instruction, control flow will return to the
3890 "normal" label. If the callee (or any indirect callees) returns via the
3891 ":ref:`resume <i_resume>`" instruction or other exception handling
3892 mechanism, control is interrupted and continued at the dynamically
3893 nearest "exception" label.
3895 The '``exception``' label is a `landing
3896 pad <ExceptionHandling.html#overview>`_ for the exception. As such,
3897 '``exception``' label is required to have the
3898 ":ref:`landingpad <i_landingpad>`" instruction, which contains the
3899 information about the behavior of the program after unwinding happens,
3900 as its first non-PHI instruction. The restrictions on the
3901 "``landingpad``" instruction's tightly couples it to the "``invoke``"
3902 instruction, so that the important information contained within the
3903 "``landingpad``" instruction can't be lost through normal code motion.
3908 This instruction requires several arguments:
3910 #. The optional "cconv" marker indicates which :ref:`calling
3911 convention <callingconv>` the call should use. If none is
3912 specified, the call defaults to using C calling conventions.
3913 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
3914 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
3916 #. '``ptr to function ty``': shall be the signature of the pointer to
3917 function value being invoked. In most cases, this is a direct
3918 function invocation, but indirect ``invoke``'s are just as possible,
3919 branching off an arbitrary pointer to function value.
3920 #. '``function ptr val``': An LLVM value containing a pointer to a
3921 function to be invoked.
3922 #. '``function args``': argument list whose types match the function
3923 signature argument types and parameter attributes. All arguments must
3924 be of :ref:`first class <t_firstclass>` type. If the function signature
3925 indicates the function accepts a variable number of arguments, the
3926 extra arguments can be specified.
3927 #. '``normal label``': the label reached when the called function
3928 executes a '``ret``' instruction.
3929 #. '``exception label``': the label reached when a callee returns via
3930 the :ref:`resume <i_resume>` instruction or other exception handling
3932 #. The optional :ref:`function attributes <fnattrs>` list. Only
3933 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
3934 attributes are valid here.
3939 This instruction is designed to operate as a standard '``call``'
3940 instruction in most regards. The primary difference is that it
3941 establishes an association with a label, which is used by the runtime
3942 library to unwind the stack.
3944 This instruction is used in languages with destructors to ensure that
3945 proper cleanup is performed in the case of either a ``longjmp`` or a
3946 thrown exception. Additionally, this is important for implementation of
3947 '``catch``' clauses in high-level languages that support them.
3949 For the purposes of the SSA form, the definition of the value returned
3950 by the '``invoke``' instruction is deemed to occur on the edge from the
3951 current block to the "normal" label. If the callee unwinds then no
3952 return value is available.
3957 .. code-block:: llvm
3959 %retval = invoke i32 @Test(i32 15) to label %Continue
3960 unwind label %TestCleanup ; i32:retval set
3961 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue
3962 unwind label %TestCleanup ; i32:retval set
3966 '``resume``' Instruction
3967 ^^^^^^^^^^^^^^^^^^^^^^^^
3974 resume <type> <value>
3979 The '``resume``' instruction is a terminator instruction that has no
3985 The '``resume``' instruction requires one argument, which must have the
3986 same type as the result of any '``landingpad``' instruction in the same
3992 The '``resume``' instruction resumes propagation of an existing
3993 (in-flight) exception whose unwinding was interrupted with a
3994 :ref:`landingpad <i_landingpad>` instruction.
3999 .. code-block:: llvm
4001 resume { i8*, i32 } %exn
4005 '``unreachable``' Instruction
4006 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4018 The '``unreachable``' instruction has no defined semantics. This
4019 instruction is used to inform the optimizer that a particular portion of
4020 the code is not reachable. This can be used to indicate that the code
4021 after a no-return function cannot be reached, and other facts.
4026 The '``unreachable``' instruction has no defined semantics.
4033 Binary operators are used to do most of the computation in a program.
4034 They require two operands of the same type, execute an operation on
4035 them, and produce a single value. The operands might represent multiple
4036 data, as is the case with the :ref:`vector <t_vector>` data type. The
4037 result value has the same type as its operands.
4039 There are several different binary operators:
4043 '``add``' Instruction
4044 ^^^^^^^^^^^^^^^^^^^^^
4051 <result> = add <ty> <op1>, <op2> ; yields ty:result
4052 <result> = add nuw <ty> <op1>, <op2> ; yields ty:result
4053 <result> = add nsw <ty> <op1>, <op2> ; yields ty:result
4054 <result> = add nuw nsw <ty> <op1>, <op2> ; yields ty:result
4059 The '``add``' instruction returns the sum of its two operands.
4064 The two arguments to the '``add``' instruction must be
4065 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4066 arguments must have identical types.
4071 The value produced is the integer sum of the two operands.
4073 If the sum has unsigned overflow, the result returned is the
4074 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
4077 Because LLVM integers use a two's complement representation, this
4078 instruction is appropriate for both signed and unsigned integers.
4080 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4081 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4082 result value of the ``add`` is a :ref:`poison value <poisonvalues>` if
4083 unsigned and/or signed overflow, respectively, occurs.
4088 .. code-block:: llvm
4090 <result> = add i32 4, %var ; yields i32:result = 4 + %var
4094 '``fadd``' Instruction
4095 ^^^^^^^^^^^^^^^^^^^^^^
4102 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4107 The '``fadd``' instruction returns the sum of its two operands.
4112 The two arguments to the '``fadd``' instruction must be :ref:`floating
4113 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4114 Both arguments must have identical types.
4119 The value produced is the floating point sum of the two operands. This
4120 instruction can also take any number of :ref:`fast-math flags <fastmath>`,
4121 which are optimization hints to enable otherwise unsafe floating point
4127 .. code-block:: llvm
4129 <result> = fadd float 4.0, %var ; yields float:result = 4.0 + %var
4131 '``sub``' Instruction
4132 ^^^^^^^^^^^^^^^^^^^^^
4139 <result> = sub <ty> <op1>, <op2> ; yields ty:result
4140 <result> = sub nuw <ty> <op1>, <op2> ; yields ty:result
4141 <result> = sub nsw <ty> <op1>, <op2> ; yields ty:result
4142 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields ty:result
4147 The '``sub``' instruction returns the difference of its two operands.
4149 Note that the '``sub``' instruction is used to represent the '``neg``'
4150 instruction present in most other intermediate representations.
4155 The two arguments to the '``sub``' instruction must be
4156 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4157 arguments must have identical types.
4162 The value produced is the integer difference of the two operands.
4164 If the difference has unsigned overflow, the result returned is the
4165 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
4168 Because LLVM integers use a two's complement representation, this
4169 instruction is appropriate for both signed and unsigned integers.
4171 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4172 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4173 result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if
4174 unsigned and/or signed overflow, respectively, occurs.
4179 .. code-block:: llvm
4181 <result> = sub i32 4, %var ; yields i32:result = 4 - %var
4182 <result> = sub i32 0, %val ; yields i32:result = -%var
4186 '``fsub``' Instruction
4187 ^^^^^^^^^^^^^^^^^^^^^^
4194 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4199 The '``fsub``' instruction returns the difference of its two operands.
4201 Note that the '``fsub``' instruction is used to represent the '``fneg``'
4202 instruction present in most other intermediate representations.
4207 The two arguments to the '``fsub``' instruction must be :ref:`floating
4208 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4209 Both arguments must have identical types.
4214 The value produced is the floating point difference of the two operands.
4215 This instruction can also take any number of :ref:`fast-math
4216 flags <fastmath>`, which are optimization hints to enable otherwise
4217 unsafe floating point optimizations:
4222 .. code-block:: llvm
4224 <result> = fsub float 4.0, %var ; yields float:result = 4.0 - %var
4225 <result> = fsub float -0.0, %val ; yields float:result = -%var
4227 '``mul``' Instruction
4228 ^^^^^^^^^^^^^^^^^^^^^
4235 <result> = mul <ty> <op1>, <op2> ; yields ty:result
4236 <result> = mul nuw <ty> <op1>, <op2> ; yields ty:result
4237 <result> = mul nsw <ty> <op1>, <op2> ; yields ty:result
4238 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields ty:result
4243 The '``mul``' instruction returns the product of its two operands.
4248 The two arguments to the '``mul``' instruction must be
4249 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4250 arguments must have identical types.
4255 The value produced is the integer product of the two operands.
4257 If the result of the multiplication has unsigned overflow, the result
4258 returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the
4259 bit width of the result.
4261 Because LLVM integers use a two's complement representation, and the
4262 result is the same width as the operands, this instruction returns the
4263 correct result for both signed and unsigned integers. If a full product
4264 (e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be
4265 sign-extended or zero-extended as appropriate to the width of the full
4268 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4269 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4270 result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if
4271 unsigned and/or signed overflow, respectively, occurs.
4276 .. code-block:: llvm
4278 <result> = mul i32 4, %var ; yields i32:result = 4 * %var
4282 '``fmul``' Instruction
4283 ^^^^^^^^^^^^^^^^^^^^^^
4290 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4295 The '``fmul``' instruction returns the product of its two operands.
4300 The two arguments to the '``fmul``' instruction must be :ref:`floating
4301 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4302 Both arguments must have identical types.
4307 The value produced is the floating point product of the two operands.
4308 This instruction can also take any number of :ref:`fast-math
4309 flags <fastmath>`, which are optimization hints to enable otherwise
4310 unsafe floating point optimizations:
4315 .. code-block:: llvm
4317 <result> = fmul float 4.0, %var ; yields float:result = 4.0 * %var
4319 '``udiv``' Instruction
4320 ^^^^^^^^^^^^^^^^^^^^^^
4327 <result> = udiv <ty> <op1>, <op2> ; yields ty:result
4328 <result> = udiv exact <ty> <op1>, <op2> ; yields ty:result
4333 The '``udiv``' instruction returns the quotient of its two operands.
4338 The two arguments to the '``udiv``' instruction must be
4339 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4340 arguments must have identical types.
4345 The value produced is the unsigned integer quotient of the two operands.
4347 Note that unsigned integer division and signed integer division are
4348 distinct operations; for signed integer division, use '``sdiv``'.
4350 Division by zero leads to undefined behavior.
4352 If the ``exact`` keyword is present, the result value of the ``udiv`` is
4353 a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as
4354 such, "((a udiv exact b) mul b) == a").
4359 .. code-block:: llvm
4361 <result> = udiv i32 4, %var ; yields i32:result = 4 / %var
4363 '``sdiv``' Instruction
4364 ^^^^^^^^^^^^^^^^^^^^^^
4371 <result> = sdiv <ty> <op1>, <op2> ; yields ty:result
4372 <result> = sdiv exact <ty> <op1>, <op2> ; yields ty:result
4377 The '``sdiv``' instruction returns the quotient of its two operands.
4382 The two arguments to the '``sdiv``' instruction must be
4383 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4384 arguments must have identical types.
4389 The value produced is the signed integer quotient of the two operands
4390 rounded towards zero.
4392 Note that signed integer division and unsigned integer division are
4393 distinct operations; for unsigned integer division, use '``udiv``'.
4395 Division by zero leads to undefined behavior. Overflow also leads to
4396 undefined behavior; this is a rare case, but can occur, for example, by
4397 doing a 32-bit division of -2147483648 by -1.
4399 If the ``exact`` keyword is present, the result value of the ``sdiv`` is
4400 a :ref:`poison value <poisonvalues>` if the result would be rounded.
4405 .. code-block:: llvm
4407 <result> = sdiv i32 4, %var ; yields i32:result = 4 / %var
4411 '``fdiv``' Instruction
4412 ^^^^^^^^^^^^^^^^^^^^^^
4419 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4424 The '``fdiv``' instruction returns the quotient of its two operands.
4429 The two arguments to the '``fdiv``' instruction must be :ref:`floating
4430 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4431 Both arguments must have identical types.
4436 The value produced is the floating point quotient of the two operands.
4437 This instruction can also take any number of :ref:`fast-math
4438 flags <fastmath>`, which are optimization hints to enable otherwise
4439 unsafe floating point optimizations:
4444 .. code-block:: llvm
4446 <result> = fdiv float 4.0, %var ; yields float:result = 4.0 / %var
4448 '``urem``' Instruction
4449 ^^^^^^^^^^^^^^^^^^^^^^
4456 <result> = urem <ty> <op1>, <op2> ; yields ty:result
4461 The '``urem``' instruction returns the remainder from the unsigned
4462 division of its two arguments.
4467 The two arguments to the '``urem``' instruction must be
4468 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4469 arguments must have identical types.
4474 This instruction returns the unsigned integer *remainder* of a division.
4475 This instruction always performs an unsigned division to get the
4478 Note that unsigned integer remainder and signed integer remainder are
4479 distinct operations; for signed integer remainder, use '``srem``'.
4481 Taking the remainder of a division by zero leads to undefined behavior.
4486 .. code-block:: llvm
4488 <result> = urem i32 4, %var ; yields i32:result = 4 % %var
4490 '``srem``' Instruction
4491 ^^^^^^^^^^^^^^^^^^^^^^
4498 <result> = srem <ty> <op1>, <op2> ; yields ty:result
4503 The '``srem``' instruction returns the remainder from the signed
4504 division of its two operands. This instruction can also take
4505 :ref:`vector <t_vector>` versions of the values in which case the elements
4511 The two arguments to the '``srem``' instruction must be
4512 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4513 arguments must have identical types.
4518 This instruction returns the *remainder* of a division (where the result
4519 is either zero or has the same sign as the dividend, ``op1``), not the
4520 *modulo* operator (where the result is either zero or has the same sign
4521 as the divisor, ``op2``) of a value. For more information about the
4522 difference, see `The Math
4523 Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a
4524 table of how this is implemented in various languages, please see
4526 operation <http://en.wikipedia.org/wiki/Modulo_operation>`_.
4528 Note that signed integer remainder and unsigned integer remainder are
4529 distinct operations; for unsigned integer remainder, use '``urem``'.
4531 Taking the remainder of a division by zero leads to undefined behavior.
4532 Overflow also leads to undefined behavior; this is a rare case, but can
4533 occur, for example, by taking the remainder of a 32-bit division of
4534 -2147483648 by -1. (The remainder doesn't actually overflow, but this
4535 rule lets srem be implemented using instructions that return both the
4536 result of the division and the remainder.)
4541 .. code-block:: llvm
4543 <result> = srem i32 4, %var ; yields i32:result = 4 % %var
4547 '``frem``' Instruction
4548 ^^^^^^^^^^^^^^^^^^^^^^
4555 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4560 The '``frem``' instruction returns the remainder from the division of
4566 The two arguments to the '``frem``' instruction must be :ref:`floating
4567 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4568 Both arguments must have identical types.
4573 This instruction returns the *remainder* of a division. The remainder
4574 has the same sign as the dividend. This instruction can also take any
4575 number of :ref:`fast-math flags <fastmath>`, which are optimization hints
4576 to enable otherwise unsafe floating point optimizations:
4581 .. code-block:: llvm
4583 <result> = frem float 4.0, %var ; yields float:result = 4.0 % %var
4587 Bitwise Binary Operations
4588 -------------------------
4590 Bitwise binary operators are used to do various forms of bit-twiddling
4591 in a program. They are generally very efficient instructions and can
4592 commonly be strength reduced from other instructions. They require two
4593 operands of the same type, execute an operation on them, and produce a
4594 single value. The resulting value is the same type as its operands.
4596 '``shl``' Instruction
4597 ^^^^^^^^^^^^^^^^^^^^^
4604 <result> = shl <ty> <op1>, <op2> ; yields ty:result
4605 <result> = shl nuw <ty> <op1>, <op2> ; yields ty:result
4606 <result> = shl nsw <ty> <op1>, <op2> ; yields ty:result
4607 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields ty:result
4612 The '``shl``' instruction returns the first operand shifted to the left
4613 a specified number of bits.
4618 Both arguments to the '``shl``' instruction must be the same
4619 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4620 '``op2``' is treated as an unsigned value.
4625 The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`,
4626 where ``n`` is the width of the result. If ``op2`` is (statically or
4627 dynamically) negative or equal to or larger than the number of bits in
4628 ``op1``, the result is undefined. If the arguments are vectors, each
4629 vector element of ``op1`` is shifted by the corresponding shift amount
4632 If the ``nuw`` keyword is present, then the shift produces a :ref:`poison
4633 value <poisonvalues>` if it shifts out any non-zero bits. If the
4634 ``nsw`` keyword is present, then the shift produces a :ref:`poison
4635 value <poisonvalues>` if it shifts out any bits that disagree with the
4636 resultant sign bit. As such, NUW/NSW have the same semantics as they
4637 would if the shift were expressed as a mul instruction with the same
4638 nsw/nuw bits in (mul %op1, (shl 1, %op2)).
4643 .. code-block:: llvm
4645 <result> = shl i32 4, %var ; yields i32: 4 << %var
4646 <result> = shl i32 4, 2 ; yields i32: 16
4647 <result> = shl i32 1, 10 ; yields i32: 1024
4648 <result> = shl i32 1, 32 ; undefined
4649 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4>
4651 '``lshr``' Instruction
4652 ^^^^^^^^^^^^^^^^^^^^^^
4659 <result> = lshr <ty> <op1>, <op2> ; yields ty:result
4660 <result> = lshr exact <ty> <op1>, <op2> ; yields ty:result
4665 The '``lshr``' instruction (logical shift right) returns the first
4666 operand shifted to the right a specified number of bits with zero fill.
4671 Both arguments to the '``lshr``' instruction must be the same
4672 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4673 '``op2``' is treated as an unsigned value.
4678 This instruction always performs a logical shift right operation. The
4679 most significant bits of the result will be filled with zero bits after
4680 the shift. If ``op2`` is (statically or dynamically) equal to or larger
4681 than the number of bits in ``op1``, the result is undefined. If the
4682 arguments are vectors, each vector element of ``op1`` is shifted by the
4683 corresponding shift amount in ``op2``.
4685 If the ``exact`` keyword is present, the result value of the ``lshr`` is
4686 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
4692 .. code-block:: llvm
4694 <result> = lshr i32 4, 1 ; yields i32:result = 2
4695 <result> = lshr i32 4, 2 ; yields i32:result = 1
4696 <result> = lshr i8 4, 3 ; yields i8:result = 0
4697 <result> = lshr i8 -2, 1 ; yields i8:result = 0x7F
4698 <result> = lshr i32 1, 32 ; undefined
4699 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1>
4701 '``ashr``' Instruction
4702 ^^^^^^^^^^^^^^^^^^^^^^
4709 <result> = ashr <ty> <op1>, <op2> ; yields ty:result
4710 <result> = ashr exact <ty> <op1>, <op2> ; yields ty:result
4715 The '``ashr``' instruction (arithmetic shift right) returns the first
4716 operand shifted to the right a specified number of bits with sign
4722 Both arguments to the '``ashr``' instruction must be the same
4723 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4724 '``op2``' is treated as an unsigned value.
4729 This instruction always performs an arithmetic shift right operation,
4730 The most significant bits of the result will be filled with the sign bit
4731 of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger
4732 than the number of bits in ``op1``, the result is undefined. If the
4733 arguments are vectors, each vector element of ``op1`` is shifted by the
4734 corresponding shift amount in ``op2``.
4736 If the ``exact`` keyword is present, the result value of the ``ashr`` is
4737 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
4743 .. code-block:: llvm
4745 <result> = ashr i32 4, 1 ; yields i32:result = 2
4746 <result> = ashr i32 4, 2 ; yields i32:result = 1
4747 <result> = ashr i8 4, 3 ; yields i8:result = 0
4748 <result> = ashr i8 -2, 1 ; yields i8:result = -1
4749 <result> = ashr i32 1, 32 ; undefined
4750 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0>
4752 '``and``' Instruction
4753 ^^^^^^^^^^^^^^^^^^^^^
4760 <result> = and <ty> <op1>, <op2> ; yields ty:result
4765 The '``and``' instruction returns the bitwise logical and of its two
4771 The two arguments to the '``and``' instruction must be
4772 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4773 arguments must have identical types.
4778 The truth table used for the '``and``' instruction is:
4795 .. code-block:: llvm
4797 <result> = and i32 4, %var ; yields i32:result = 4 & %var
4798 <result> = and i32 15, 40 ; yields i32:result = 8
4799 <result> = and i32 4, 8 ; yields i32:result = 0
4801 '``or``' Instruction
4802 ^^^^^^^^^^^^^^^^^^^^
4809 <result> = or <ty> <op1>, <op2> ; yields ty:result
4814 The '``or``' instruction returns the bitwise logical inclusive or of its
4820 The two arguments to the '``or``' instruction must be
4821 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4822 arguments must have identical types.
4827 The truth table used for the '``or``' instruction is:
4846 <result> = or i32 4, %var ; yields i32:result = 4 | %var
4847 <result> = or i32 15, 40 ; yields i32:result = 47
4848 <result> = or i32 4, 8 ; yields i32:result = 12
4850 '``xor``' Instruction
4851 ^^^^^^^^^^^^^^^^^^^^^
4858 <result> = xor <ty> <op1>, <op2> ; yields ty:result
4863 The '``xor``' instruction returns the bitwise logical exclusive or of
4864 its two operands. The ``xor`` is used to implement the "one's
4865 complement" operation, which is the "~" operator in C.
4870 The two arguments to the '``xor``' instruction must be
4871 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4872 arguments must have identical types.
4877 The truth table used for the '``xor``' instruction is:
4894 .. code-block:: llvm
4896 <result> = xor i32 4, %var ; yields i32:result = 4 ^ %var
4897 <result> = xor i32 15, 40 ; yields i32:result = 39
4898 <result> = xor i32 4, 8 ; yields i32:result = 12
4899 <result> = xor i32 %V, -1 ; yields i32:result = ~%V
4904 LLVM supports several instructions to represent vector operations in a
4905 target-independent manner. These instructions cover the element-access
4906 and vector-specific operations needed to process vectors effectively.
4907 While LLVM does directly support these vector operations, many
4908 sophisticated algorithms will want to use target-specific intrinsics to
4909 take full advantage of a specific target.
4911 .. _i_extractelement:
4913 '``extractelement``' Instruction
4914 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4921 <result> = extractelement <n x <ty>> <val>, <ty2> <idx> ; yields <ty>
4926 The '``extractelement``' instruction extracts a single scalar element
4927 from a vector at a specified index.
4932 The first operand of an '``extractelement``' instruction is a value of
4933 :ref:`vector <t_vector>` type. The second operand is an index indicating
4934 the position from which to extract the element. The index may be a
4935 variable of any integer type.
4940 The result is a scalar of the same type as the element type of ``val``.
4941 Its value is the value at position ``idx`` of ``val``. If ``idx``
4942 exceeds the length of ``val``, the results are undefined.
4947 .. code-block:: llvm
4949 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32
4951 .. _i_insertelement:
4953 '``insertelement``' Instruction
4954 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4961 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <n x <ty>>
4966 The '``insertelement``' instruction inserts a scalar element into a
4967 vector at a specified index.
4972 The first operand of an '``insertelement``' instruction is a value of
4973 :ref:`vector <t_vector>` type. The second operand is a scalar value whose
4974 type must equal the element type of the first operand. The third operand
4975 is an index indicating the position at which to insert the value. The
4976 index may be a variable of any integer type.
4981 The result is a vector of the same type as ``val``. Its element values
4982 are those of ``val`` except at position ``idx``, where it gets the value
4983 ``elt``. If ``idx`` exceeds the length of ``val``, the results are
4989 .. code-block:: llvm
4991 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32>
4993 .. _i_shufflevector:
4995 '``shufflevector``' Instruction
4996 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5003 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>>
5008 The '``shufflevector``' instruction constructs a permutation of elements
5009 from two input vectors, returning a vector with the same element type as
5010 the input and length that is the same as the shuffle mask.
5015 The first two operands of a '``shufflevector``' instruction are vectors
5016 with the same type. The third argument is a shuffle mask whose element
5017 type is always 'i32'. The result of the instruction is a vector whose
5018 length is the same as the shuffle mask and whose element type is the
5019 same as the element type of the first two operands.
5021 The shuffle mask operand is required to be a constant vector with either
5022 constant integer or undef values.
5027 The elements of the two input vectors are numbered from left to right
5028 across both of the vectors. The shuffle mask operand specifies, for each
5029 element of the result vector, which element of the two input vectors the
5030 result element gets. The element selector may be undef (meaning "don't
5031 care") and the second operand may be undef if performing a shuffle from
5037 .. code-block:: llvm
5039 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
5040 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32>
5041 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
5042 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle.
5043 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
5044 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32>
5045 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
5046 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32>
5048 Aggregate Operations
5049 --------------------
5051 LLVM supports several instructions for working with
5052 :ref:`aggregate <t_aggregate>` values.
5056 '``extractvalue``' Instruction
5057 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5064 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
5069 The '``extractvalue``' instruction extracts the value of a member field
5070 from an :ref:`aggregate <t_aggregate>` value.
5075 The first operand of an '``extractvalue``' instruction is a value of
5076 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The operands are
5077 constant indices to specify which value to extract in a similar manner
5078 as indices in a '``getelementptr``' instruction.
5080 The major differences to ``getelementptr`` indexing are:
5082 - Since the value being indexed is not a pointer, the first index is
5083 omitted and assumed to be zero.
5084 - At least one index must be specified.
5085 - Not only struct indices but also array indices must be in bounds.
5090 The result is the value at the position in the aggregate specified by
5096 .. code-block:: llvm
5098 <result> = extractvalue {i32, float} %agg, 0 ; yields i32
5102 '``insertvalue``' Instruction
5103 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5110 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type>
5115 The '``insertvalue``' instruction inserts a value into a member field in
5116 an :ref:`aggregate <t_aggregate>` value.
5121 The first operand of an '``insertvalue``' instruction is a value of
5122 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is
5123 a first-class value to insert. The following operands are constant
5124 indices indicating the position at which to insert the value in a
5125 similar manner as indices in a '``extractvalue``' instruction. The value
5126 to insert must have the same type as the value identified by the
5132 The result is an aggregate of the same type as ``val``. Its value is
5133 that of ``val`` except that the value at the position specified by the
5134 indices is that of ``elt``.
5139 .. code-block:: llvm
5141 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef}
5142 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val}
5143 %agg3 = insertvalue {i32, {float}} undef, float %val, 1, 0 ; yields {i32 undef, {float %val}}
5147 Memory Access and Addressing Operations
5148 ---------------------------------------
5150 A key design point of an SSA-based representation is how it represents
5151 memory. In LLVM, no memory locations are in SSA form, which makes things
5152 very simple. This section describes how to read, write, and allocate
5157 '``alloca``' Instruction
5158 ^^^^^^^^^^^^^^^^^^^^^^^^
5165 <result> = alloca [inalloca] <type> [, <ty> <NumElements>] [, align <alignment>] ; yields type*:result
5170 The '``alloca``' instruction allocates memory on the stack frame of the
5171 currently executing function, to be automatically released when this
5172 function returns to its caller. The object is always allocated in the
5173 generic address space (address space zero).
5178 The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements``
5179 bytes of memory on the runtime stack, returning a pointer of the
5180 appropriate type to the program. If "NumElements" is specified, it is
5181 the number of elements allocated, otherwise "NumElements" is defaulted
5182 to be one. If a constant alignment is specified, the value result of the
5183 allocation is guaranteed to be aligned to at least that boundary. The
5184 alignment may not be greater than ``1 << 29``. If not specified, or if
5185 zero, the target can choose to align the allocation on any convenient
5186 boundary compatible with the type.
5188 '``type``' may be any sized type.
5193 Memory is allocated; a pointer is returned. The operation is undefined
5194 if there is insufficient stack space for the allocation. '``alloca``'d
5195 memory is automatically released when the function returns. The
5196 '``alloca``' instruction is commonly used to represent automatic
5197 variables that must have an address available. When the function returns
5198 (either with the ``ret`` or ``resume`` instructions), the memory is
5199 reclaimed. Allocating zero bytes is legal, but the result is undefined.
5200 The order in which memory is allocated (ie., which way the stack grows)
5206 .. code-block:: llvm
5208 %ptr = alloca i32 ; yields i32*:ptr
5209 %ptr = alloca i32, i32 4 ; yields i32*:ptr
5210 %ptr = alloca i32, i32 4, align 1024 ; yields i32*:ptr
5211 %ptr = alloca i32, align 1024 ; yields i32*:ptr
5215 '``load``' Instruction
5216 ^^^^^^^^^^^^^^^^^^^^^^
5223 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>][, !nonnull !<index>]
5224 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
5225 !<index> = !{ i32 1 }
5230 The '``load``' instruction is used to read from memory.
5235 The argument to the ``load`` instruction specifies the memory address
5236 from which to load. The pointer must point to a :ref:`first
5237 class <t_firstclass>` type. If the ``load`` is marked as ``volatile``,
5238 then the optimizer is not allowed to modify the number or order of
5239 execution of this ``load`` with other :ref:`volatile
5240 operations <volatile>`.
5242 If the ``load`` is marked as ``atomic``, it takes an extra
5243 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
5244 ``release`` and ``acq_rel`` orderings are not valid on ``load``
5245 instructions. Atomic loads produce :ref:`defined <memmodel>` results
5246 when they may see multiple atomic stores. The type of the pointee must
5247 be an integer type whose bit width is a power of two greater than or
5248 equal to eight and less than or equal to a target-specific size limit.
5249 ``align`` must be explicitly specified on atomic loads, and the load has
5250 undefined behavior if the alignment is not set to a value which is at
5251 least the size in bytes of the pointee. ``!nontemporal`` does not have
5252 any defined semantics for atomic loads.
5254 The optional constant ``align`` argument specifies the alignment of the
5255 operation (that is, the alignment of the memory address). A value of 0
5256 or an omitted ``align`` argument means that the operation has the ABI
5257 alignment for the target. It is the responsibility of the code emitter
5258 to ensure that the alignment information is correct. Overestimating the
5259 alignment results in undefined behavior. Underestimating the alignment
5260 may produce less efficient code. An alignment of 1 is always safe. The
5261 maximum possible alignment is ``1 << 29``.
5263 The optional ``!nontemporal`` metadata must reference a single
5264 metadata name ``<index>`` corresponding to a metadata node with one
5265 ``i32`` entry of value 1. The existence of the ``!nontemporal``
5266 metadata on the instruction tells the optimizer and code generator
5267 that this load is not expected to be reused in the cache. The code
5268 generator may select special instructions to save cache bandwidth, such
5269 as the ``MOVNT`` instruction on x86.
5271 The optional ``!invariant.load`` metadata must reference a single
5272 metadata name ``<index>`` corresponding to a metadata node with no
5273 entries. The existence of the ``!invariant.load`` metadata on the
5274 instruction tells the optimizer and code generator that the address
5275 operand to this load points to memory which can be assumed unchanged.
5276 Being invariant does not imply that a location is dereferenceable,
5277 but it does imply that once the location is known dereferenceable
5278 its value is henceforth unchanging.
5280 The optional ``!nonnull`` metadata must reference a single
5281 metadata name ``<index>`` corresponding to a metadata node with no
5282 entries. The existence of the ``!nonnull`` metadata on the
5283 instruction tells the optimizer that the value loaded is known to
5284 never be null. This is analogous to the ''nonnull'' attribute
5285 on parameters and return values. This metadata can only be applied
5286 to loads of a pointer type.
5291 The location of memory pointed to is loaded. If the value being loaded
5292 is of scalar type then the number of bytes read does not exceed the
5293 minimum number of bytes needed to hold all bits of the type. For
5294 example, loading an ``i24`` reads at most three bytes. When loading a
5295 value of a type like ``i20`` with a size that is not an integral number
5296 of bytes, the result is undefined if the value was not originally
5297 written using a store of the same type.
5302 .. code-block:: llvm
5304 %ptr = alloca i32 ; yields i32*:ptr
5305 store i32 3, i32* %ptr ; yields void
5306 %val = load i32* %ptr ; yields i32:val = i32 3
5310 '``store``' Instruction
5311 ^^^^^^^^^^^^^^^^^^^^^^^
5318 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] ; yields void
5319 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> ; yields void
5324 The '``store``' instruction is used to write to memory.
5329 There are two arguments to the ``store`` instruction: a value to store
5330 and an address at which to store it. The type of the ``<pointer>``
5331 operand must be a pointer to the :ref:`first class <t_firstclass>` type of
5332 the ``<value>`` operand. If the ``store`` is marked as ``volatile``,
5333 then the optimizer is not allowed to modify the number or order of
5334 execution of this ``store`` with other :ref:`volatile
5335 operations <volatile>`.
5337 If the ``store`` is marked as ``atomic``, it takes an extra
5338 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
5339 ``acquire`` and ``acq_rel`` orderings aren't valid on ``store``
5340 instructions. Atomic loads produce :ref:`defined <memmodel>` results
5341 when they may see multiple atomic stores. The type of the pointee must
5342 be an integer type whose bit width is a power of two greater than or
5343 equal to eight and less than or equal to a target-specific size limit.
5344 ``align`` must be explicitly specified on atomic stores, and the store
5345 has undefined behavior if the alignment is not set to a value which is
5346 at least the size in bytes of the pointee. ``!nontemporal`` does not
5347 have any defined semantics for atomic stores.
5349 The optional constant ``align`` argument specifies the alignment of the
5350 operation (that is, the alignment of the memory address). A value of 0
5351 or an omitted ``align`` argument means that the operation has the ABI
5352 alignment for the target. It is the responsibility of the code emitter
5353 to ensure that the alignment information is correct. Overestimating the
5354 alignment results in undefined behavior. Underestimating the
5355 alignment may produce less efficient code. An alignment of 1 is always
5356 safe. The maximum possible alignment is ``1 << 29``.
5358 The optional ``!nontemporal`` metadata must reference a single metadata
5359 name ``<index>`` corresponding to a metadata node with one ``i32`` entry of
5360 value 1. The existence of the ``!nontemporal`` metadata on the instruction
5361 tells the optimizer and code generator that this load is not expected to
5362 be reused in the cache. The code generator may select special
5363 instructions to save cache bandwidth, such as the MOVNT instruction on
5369 The contents of memory are updated to contain ``<value>`` at the
5370 location specified by the ``<pointer>`` operand. If ``<value>`` is
5371 of scalar type then the number of bytes written does not exceed the
5372 minimum number of bytes needed to hold all bits of the type. For
5373 example, storing an ``i24`` writes at most three bytes. When writing a
5374 value of a type like ``i20`` with a size that is not an integral number
5375 of bytes, it is unspecified what happens to the extra bits that do not
5376 belong to the type, but they will typically be overwritten.
5381 .. code-block:: llvm
5383 %ptr = alloca i32 ; yields i32*:ptr
5384 store i32 3, i32* %ptr ; yields void
5385 %val = load i32* %ptr ; yields i32:val = i32 3
5389 '``fence``' Instruction
5390 ^^^^^^^^^^^^^^^^^^^^^^^
5397 fence [singlethread] <ordering> ; yields void
5402 The '``fence``' instruction is used to introduce happens-before edges
5408 '``fence``' instructions take an :ref:`ordering <ordering>` argument which
5409 defines what *synchronizes-with* edges they add. They can only be given
5410 ``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings.
5415 A fence A which has (at least) ``release`` ordering semantics
5416 *synchronizes with* a fence B with (at least) ``acquire`` ordering
5417 semantics if and only if there exist atomic operations X and Y, both
5418 operating on some atomic object M, such that A is sequenced before X, X
5419 modifies M (either directly or through some side effect of a sequence
5420 headed by X), Y is sequenced before B, and Y observes M. This provides a
5421 *happens-before* dependency between A and B. Rather than an explicit
5422 ``fence``, one (but not both) of the atomic operations X or Y might
5423 provide a ``release`` or ``acquire`` (resp.) ordering constraint and
5424 still *synchronize-with* the explicit ``fence`` and establish the
5425 *happens-before* edge.
5427 A ``fence`` which has ``seq_cst`` ordering, in addition to having both
5428 ``acquire`` and ``release`` semantics specified above, participates in
5429 the global program order of other ``seq_cst`` operations and/or fences.
5431 The optional ":ref:`singlethread <singlethread>`" argument specifies
5432 that the fence only synchronizes with other fences in the same thread.
5433 (This is useful for interacting with signal handlers.)
5438 .. code-block:: llvm
5440 fence acquire ; yields void
5441 fence singlethread seq_cst ; yields void
5445 '``cmpxchg``' Instruction
5446 ^^^^^^^^^^^^^^^^^^^^^^^^^
5453 cmpxchg [weak] [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <success ordering> <failure ordering> ; yields { ty, i1 }
5458 The '``cmpxchg``' instruction is used to atomically modify memory. It
5459 loads a value in memory and compares it to a given value. If they are
5460 equal, it tries to store a new value into the memory.
5465 There are three arguments to the '``cmpxchg``' instruction: an address
5466 to operate on, a value to compare to the value currently be at that
5467 address, and a new value to place at that address if the compared values
5468 are equal. The type of '<cmp>' must be an integer type whose bit width
5469 is a power of two greater than or equal to eight and less than or equal
5470 to a target-specific size limit. '<cmp>' and '<new>' must have the same
5471 type, and the type of '<pointer>' must be a pointer to that type. If the
5472 ``cmpxchg`` is marked as ``volatile``, then the optimizer is not allowed
5473 to modify the number or order of execution of this ``cmpxchg`` with
5474 other :ref:`volatile operations <volatile>`.
5476 The success and failure :ref:`ordering <ordering>` arguments specify how this
5477 ``cmpxchg`` synchronizes with other atomic operations. Both ordering parameters
5478 must be at least ``monotonic``, the ordering constraint on failure must be no
5479 stronger than that on success, and the failure ordering cannot be either
5480 ``release`` or ``acq_rel``.
5482 The optional "``singlethread``" argument declares that the ``cmpxchg``
5483 is only atomic with respect to code (usually signal handlers) running in
5484 the same thread as the ``cmpxchg``. Otherwise the cmpxchg is atomic with
5485 respect to all other code in the system.
5487 The pointer passed into cmpxchg must have alignment greater than or
5488 equal to the size in memory of the operand.
5493 The contents of memory at the location specified by the '``<pointer>``' operand
5494 is read and compared to '``<cmp>``'; if the read value is the equal, the
5495 '``<new>``' is written. The original value at the location is returned, together
5496 with a flag indicating success (true) or failure (false).
5498 If the cmpxchg operation is marked as ``weak`` then a spurious failure is
5499 permitted: the operation may not write ``<new>`` even if the comparison
5502 If the cmpxchg operation is strong (the default), the i1 value is 1 if and only
5503 if the value loaded equals ``cmp``.
5505 A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of
5506 identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic
5507 load with an ordering parameter determined the second ordering parameter.
5512 .. code-block:: llvm
5515 %orig = atomic load i32* %ptr unordered ; yields i32
5519 %cmp = phi i32 [ %orig, %entry ], [%old, %loop]
5520 %squared = mul i32 %cmp, %cmp
5521 %val_success = cmpxchg i32* %ptr, i32 %cmp, i32 %squared acq_rel monotonic ; yields { i32, i1 }
5522 %value_loaded = extractvalue { i32, i1 } %val_success, 0
5523 %success = extractvalue { i32, i1 } %val_success, 1
5524 br i1 %success, label %done, label %loop
5531 '``atomicrmw``' Instruction
5532 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
5539 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> ; yields ty
5544 The '``atomicrmw``' instruction is used to atomically modify memory.
5549 There are three arguments to the '``atomicrmw``' instruction: an
5550 operation to apply, an address whose value to modify, an argument to the
5551 operation. The operation must be one of the following keywords:
5565 The type of '<value>' must be an integer type whose bit width is a power
5566 of two greater than or equal to eight and less than or equal to a
5567 target-specific size limit. The type of the '``<pointer>``' operand must
5568 be a pointer to that type. If the ``atomicrmw`` is marked as
5569 ``volatile``, then the optimizer is not allowed to modify the number or
5570 order of execution of this ``atomicrmw`` with other :ref:`volatile
5571 operations <volatile>`.
5576 The contents of memory at the location specified by the '``<pointer>``'
5577 operand are atomically read, modified, and written back. The original
5578 value at the location is returned. The modification is specified by the
5581 - xchg: ``*ptr = val``
5582 - add: ``*ptr = *ptr + val``
5583 - sub: ``*ptr = *ptr - val``
5584 - and: ``*ptr = *ptr & val``
5585 - nand: ``*ptr = ~(*ptr & val)``
5586 - or: ``*ptr = *ptr | val``
5587 - xor: ``*ptr = *ptr ^ val``
5588 - max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison)
5589 - min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison)
5590 - umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned
5592 - umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned
5598 .. code-block:: llvm
5600 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields i32
5602 .. _i_getelementptr:
5604 '``getelementptr``' Instruction
5605 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5612 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
5613 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
5614 <result> = getelementptr <ptr vector> ptrval, <vector index type> idx
5619 The '``getelementptr``' instruction is used to get the address of a
5620 subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs
5621 address calculation only and does not access memory.
5626 The first argument is always a pointer or a vector of pointers, and
5627 forms the basis of the calculation. The remaining arguments are indices
5628 that indicate which of the elements of the aggregate object are indexed.
5629 The interpretation of each index is dependent on the type being indexed
5630 into. The first index always indexes the pointer value given as the
5631 first argument, the second index indexes a value of the type pointed to
5632 (not necessarily the value directly pointed to, since the first index
5633 can be non-zero), etc. The first type indexed into must be a pointer
5634 value, subsequent types can be arrays, vectors, and structs. Note that
5635 subsequent types being indexed into can never be pointers, since that
5636 would require loading the pointer before continuing calculation.
5638 The type of each index argument depends on the type it is indexing into.
5639 When indexing into a (optionally packed) structure, only ``i32`` integer
5640 **constants** are allowed (when using a vector of indices they must all
5641 be the **same** ``i32`` integer constant). When indexing into an array,
5642 pointer or vector, integers of any width are allowed, and they are not
5643 required to be constant. These integers are treated as signed values
5646 For example, let's consider a C code fragment and how it gets compiled
5662 int *foo(struct ST *s) {
5663 return &s[1].Z.B[5][13];
5666 The LLVM code generated by Clang is:
5668 .. code-block:: llvm
5670 %struct.RT = type { i8, [10 x [20 x i32]], i8 }
5671 %struct.ST = type { i32, double, %struct.RT }
5673 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
5675 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5682 In the example above, the first index is indexing into the
5683 '``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``'
5684 = '``{ i32, double, %struct.RT }``' type, a structure. The second index
5685 indexes into the third element of the structure, yielding a
5686 '``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another
5687 structure. The third index indexes into the second element of the
5688 structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two
5689 dimensions of the array are subscripted into, yielding an '``i32``'
5690 type. The '``getelementptr``' instruction returns a pointer to this
5691 element, thus computing a value of '``i32*``' type.
5693 Note that it is perfectly legal to index partially through a structure,
5694 returning a pointer to an inner element. Because of this, the LLVM code
5695 for the given testcase is equivalent to:
5697 .. code-block:: llvm
5699 define i32* @foo(%struct.ST* %s) {
5700 %t1 = getelementptr %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1
5701 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2
5702 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3
5703 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4
5704 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5
5708 If the ``inbounds`` keyword is present, the result value of the
5709 ``getelementptr`` is a :ref:`poison value <poisonvalues>` if the base
5710 pointer is not an *in bounds* address of an allocated object, or if any
5711 of the addresses that would be formed by successive addition of the
5712 offsets implied by the indices to the base address with infinitely
5713 precise signed arithmetic are not an *in bounds* address of that
5714 allocated object. The *in bounds* addresses for an allocated object are
5715 all the addresses that point into the object, plus the address one byte
5716 past the end. In cases where the base is a vector of pointers the
5717 ``inbounds`` keyword applies to each of the computations element-wise.
5719 If the ``inbounds`` keyword is not present, the offsets are added to the
5720 base address with silently-wrapping two's complement arithmetic. If the
5721 offsets have a different width from the pointer, they are sign-extended
5722 or truncated to the width of the pointer. The result value of the
5723 ``getelementptr`` may be outside the object pointed to by the base
5724 pointer. The result value may not necessarily be used to access memory
5725 though, even if it happens to point into allocated storage. See the
5726 :ref:`Pointer Aliasing Rules <pointeraliasing>` section for more
5729 The getelementptr instruction is often confusing. For some more insight
5730 into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`.
5735 .. code-block:: llvm
5737 ; yields [12 x i8]*:aptr
5738 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5740 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
5742 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
5744 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
5746 In cases where the pointer argument is a vector of pointers, each index
5747 must be a vector with the same number of elements. For example:
5749 .. code-block:: llvm
5751 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5753 Conversion Operations
5754 ---------------------
5756 The instructions in this category are the conversion instructions
5757 (casting) which all take a single operand and a type. They perform
5758 various bit conversions on the operand.
5760 '``trunc .. to``' Instruction
5761 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5768 <result> = trunc <ty> <value> to <ty2> ; yields ty2
5773 The '``trunc``' instruction truncates its operand to the type ``ty2``.
5778 The '``trunc``' instruction takes a value to trunc, and a type to trunc
5779 it to. Both types must be of :ref:`integer <t_integer>` types, or vectors
5780 of the same number of integers. The bit size of the ``value`` must be
5781 larger than the bit size of the destination type, ``ty2``. Equal sized
5782 types are not allowed.
5787 The '``trunc``' instruction truncates the high order bits in ``value``
5788 and converts the remaining bits to ``ty2``. Since the source size must
5789 be larger than the destination size, ``trunc`` cannot be a *no-op cast*.
5790 It will always truncate bits.
5795 .. code-block:: llvm
5797 %X = trunc i32 257 to i8 ; yields i8:1
5798 %Y = trunc i32 123 to i1 ; yields i1:true
5799 %Z = trunc i32 122 to i1 ; yields i1:false
5800 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7>
5802 '``zext .. to``' Instruction
5803 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5810 <result> = zext <ty> <value> to <ty2> ; yields ty2
5815 The '``zext``' instruction zero extends its operand to type ``ty2``.
5820 The '``zext``' instruction takes a value to cast, and a type to cast it
5821 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
5822 the same number of integers. The bit size of the ``value`` must be
5823 smaller than the bit size of the destination type, ``ty2``.
5828 The ``zext`` fills the high order bits of the ``value`` with zero bits
5829 until it reaches the size of the destination type, ``ty2``.
5831 When zero extending from i1, the result will always be either 0 or 1.
5836 .. code-block:: llvm
5838 %X = zext i32 257 to i64 ; yields i64:257
5839 %Y = zext i1 true to i32 ; yields i32:1
5840 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
5842 '``sext .. to``' Instruction
5843 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5850 <result> = sext <ty> <value> to <ty2> ; yields ty2
5855 The '``sext``' sign extends ``value`` to the type ``ty2``.
5860 The '``sext``' instruction takes a value to cast, and a type to cast it
5861 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
5862 the same number of integers. The bit size of the ``value`` must be
5863 smaller than the bit size of the destination type, ``ty2``.
5868 The '``sext``' instruction performs a sign extension by copying the sign
5869 bit (highest order bit) of the ``value`` until it reaches the bit size
5870 of the type ``ty2``.
5872 When sign extending from i1, the extension always results in -1 or 0.
5877 .. code-block:: llvm
5879 %X = sext i8 -1 to i16 ; yields i16 :65535
5880 %Y = sext i1 true to i32 ; yields i32:-1
5881 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
5883 '``fptrunc .. to``' Instruction
5884 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5891 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2
5896 The '``fptrunc``' instruction truncates ``value`` to type ``ty2``.
5901 The '``fptrunc``' instruction takes a :ref:`floating point <t_floating>`
5902 value to cast and a :ref:`floating point <t_floating>` type to cast it to.
5903 The size of ``value`` must be larger than the size of ``ty2``. This
5904 implies that ``fptrunc`` cannot be used to make a *no-op cast*.
5909 The '``fptrunc``' instruction truncates a ``value`` from a larger
5910 :ref:`floating point <t_floating>` type to a smaller :ref:`floating
5911 point <t_floating>` type. If the value cannot fit within the
5912 destination type, ``ty2``, then the results are undefined.
5917 .. code-block:: llvm
5919 %X = fptrunc double 123.0 to float ; yields float:123.0
5920 %Y = fptrunc double 1.0E+300 to float ; yields undefined
5922 '``fpext .. to``' Instruction
5923 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5930 <result> = fpext <ty> <value> to <ty2> ; yields ty2
5935 The '``fpext``' extends a floating point ``value`` to a larger floating
5941 The '``fpext``' instruction takes a :ref:`floating point <t_floating>`
5942 ``value`` to cast, and a :ref:`floating point <t_floating>` type to cast it
5943 to. The source type must be smaller than the destination type.
5948 The '``fpext``' instruction extends the ``value`` from a smaller
5949 :ref:`floating point <t_floating>` type to a larger :ref:`floating
5950 point <t_floating>` type. The ``fpext`` cannot be used to make a
5951 *no-op cast* because it always changes bits. Use ``bitcast`` to make a
5952 *no-op cast* for a floating point cast.
5957 .. code-block:: llvm
5959 %X = fpext float 3.125 to double ; yields double:3.125000e+00
5960 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000
5962 '``fptoui .. to``' Instruction
5963 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5970 <result> = fptoui <ty> <value> to <ty2> ; yields ty2
5975 The '``fptoui``' converts a floating point ``value`` to its unsigned
5976 integer equivalent of type ``ty2``.
5981 The '``fptoui``' instruction takes a value to cast, which must be a
5982 scalar or vector :ref:`floating point <t_floating>` value, and a type to
5983 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
5984 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
5985 type with the same number of elements as ``ty``
5990 The '``fptoui``' instruction converts its :ref:`floating
5991 point <t_floating>` operand into the nearest (rounding towards zero)
5992 unsigned integer value. If the value cannot fit in ``ty2``, the results
5998 .. code-block:: llvm
6000 %X = fptoui double 123.0 to i32 ; yields i32:123
6001 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1
6002 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1
6004 '``fptosi .. to``' Instruction
6005 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6012 <result> = fptosi <ty> <value> to <ty2> ; yields ty2
6017 The '``fptosi``' instruction converts :ref:`floating point <t_floating>`
6018 ``value`` to type ``ty2``.
6023 The '``fptosi``' instruction takes a value to cast, which must be a
6024 scalar or vector :ref:`floating point <t_floating>` value, and a type to
6025 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
6026 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
6027 type with the same number of elements as ``ty``
6032 The '``fptosi``' instruction converts its :ref:`floating
6033 point <t_floating>` operand into the nearest (rounding towards zero)
6034 signed integer value. If the value cannot fit in ``ty2``, the results
6040 .. code-block:: llvm
6042 %X = fptosi double -123.0 to i32 ; yields i32:-123
6043 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1
6044 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1
6046 '``uitofp .. to``' Instruction
6047 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6054 <result> = uitofp <ty> <value> to <ty2> ; yields ty2
6059 The '``uitofp``' instruction regards ``value`` as an unsigned integer
6060 and converts that value to the ``ty2`` type.
6065 The '``uitofp``' instruction takes a value to cast, which must be a
6066 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
6067 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
6068 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
6069 type with the same number of elements as ``ty``
6074 The '``uitofp``' instruction interprets its operand as an unsigned
6075 integer quantity and converts it to the corresponding floating point
6076 value. If the value cannot fit in the floating point value, the results
6082 .. code-block:: llvm
6084 %X = uitofp i32 257 to float ; yields float:257.0
6085 %Y = uitofp i8 -1 to double ; yields double:255.0
6087 '``sitofp .. to``' Instruction
6088 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6095 <result> = sitofp <ty> <value> to <ty2> ; yields ty2
6100 The '``sitofp``' instruction regards ``value`` as a signed integer and
6101 converts that value to the ``ty2`` type.
6106 The '``sitofp``' instruction takes a value to cast, which must be a
6107 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
6108 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
6109 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
6110 type with the same number of elements as ``ty``
6115 The '``sitofp``' instruction interprets its operand as a signed integer
6116 quantity and converts it to the corresponding floating point value. If
6117 the value cannot fit in the floating point value, the results are
6123 .. code-block:: llvm
6125 %X = sitofp i32 257 to float ; yields float:257.0
6126 %Y = sitofp i8 -1 to double ; yields double:-1.0
6130 '``ptrtoint .. to``' Instruction
6131 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6138 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2
6143 The '``ptrtoint``' instruction converts the pointer or a vector of
6144 pointers ``value`` to the integer (or vector of integers) type ``ty2``.
6149 The '``ptrtoint``' instruction takes a ``value`` to cast, which must be
6150 a a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a
6151 type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or
6152 a vector of integers type.
6157 The '``ptrtoint``' instruction converts ``value`` to integer type
6158 ``ty2`` by interpreting the pointer value as an integer and either
6159 truncating or zero extending that value to the size of the integer type.
6160 If ``value`` is smaller than ``ty2`` then a zero extension is done. If
6161 ``value`` is larger than ``ty2`` then a truncation is done. If they are
6162 the same size, then nothing is done (*no-op cast*) other than a type
6168 .. code-block:: llvm
6170 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture
6171 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture
6172 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture
6176 '``inttoptr .. to``' Instruction
6177 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6184 <result> = inttoptr <ty> <value> to <ty2> ; yields ty2
6189 The '``inttoptr``' instruction converts an integer ``value`` to a
6190 pointer type, ``ty2``.
6195 The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to
6196 cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>`
6202 The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by
6203 applying either a zero extension or a truncation depending on the size
6204 of the integer ``value``. If ``value`` is larger than the size of a
6205 pointer then a truncation is done. If ``value`` is smaller than the size
6206 of a pointer then a zero extension is done. If they are the same size,
6207 nothing is done (*no-op cast*).
6212 .. code-block:: llvm
6214 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture
6215 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture
6216 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture
6217 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers
6221 '``bitcast .. to``' Instruction
6222 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6229 <result> = bitcast <ty> <value> to <ty2> ; yields ty2
6234 The '``bitcast``' instruction converts ``value`` to type ``ty2`` without
6240 The '``bitcast``' instruction takes a value to cast, which must be a
6241 non-aggregate first class value, and a type to cast it to, which must
6242 also be a non-aggregate :ref:`first class <t_firstclass>` type. The
6243 bit sizes of ``value`` and the destination type, ``ty2``, must be
6244 identical. If the source type is a pointer, the destination type must
6245 also be a pointer of the same size. This instruction supports bitwise
6246 conversion of vectors to integers and to vectors of other types (as
6247 long as they have the same size).
6252 The '``bitcast``' instruction converts ``value`` to type ``ty2``. It
6253 is always a *no-op cast* because no bits change with this
6254 conversion. The conversion is done as if the ``value`` had been stored
6255 to memory and read back as type ``ty2``. Pointer (or vector of
6256 pointers) types may only be converted to other pointer (or vector of
6257 pointers) types with the same address space through this instruction.
6258 To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>`
6259 or :ref:`ptrtoint <i_ptrtoint>` instructions first.
6264 .. code-block:: llvm
6266 %X = bitcast i8 255 to i8 ; yields i8 :-1
6267 %Y = bitcast i32* %x to sint* ; yields sint*:%x
6268 %Z = bitcast <2 x int> %V to i64; ; yields i64: %V
6269 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*>
6271 .. _i_addrspacecast:
6273 '``addrspacecast .. to``' Instruction
6274 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6281 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2
6286 The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in
6287 address space ``n`` to type ``pty2`` in address space ``m``.
6292 The '``addrspacecast``' instruction takes a pointer or vector of pointer value
6293 to cast and a pointer type to cast it to, which must have a different
6299 The '``addrspacecast``' instruction converts the pointer value
6300 ``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex
6301 value modification, depending on the target and the address space
6302 pair. Pointer conversions within the same address space must be
6303 performed with the ``bitcast`` instruction. Note that if the address space
6304 conversion is legal then both result and operand refer to the same memory
6310 .. code-block:: llvm
6312 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x
6313 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y
6314 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z
6321 The instructions in this category are the "miscellaneous" instructions,
6322 which defy better classification.
6326 '``icmp``' Instruction
6327 ^^^^^^^^^^^^^^^^^^^^^^
6334 <result> = icmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
6339 The '``icmp``' instruction returns a boolean value or a vector of
6340 boolean values based on comparison of its two integer, integer vector,
6341 pointer, or pointer vector operands.
6346 The '``icmp``' instruction takes three operands. The first operand is
6347 the condition code indicating the kind of comparison to perform. It is
6348 not a value, just a keyword. The possible condition code are:
6351 #. ``ne``: not equal
6352 #. ``ugt``: unsigned greater than
6353 #. ``uge``: unsigned greater or equal
6354 #. ``ult``: unsigned less than
6355 #. ``ule``: unsigned less or equal
6356 #. ``sgt``: signed greater than
6357 #. ``sge``: signed greater or equal
6358 #. ``slt``: signed less than
6359 #. ``sle``: signed less or equal
6361 The remaining two arguments must be :ref:`integer <t_integer>` or
6362 :ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They
6363 must also be identical types.
6368 The '``icmp``' compares ``op1`` and ``op2`` according to the condition
6369 code given as ``cond``. The comparison performed always yields either an
6370 :ref:`i1 <t_integer>` or vector of ``i1`` result, as follows:
6372 #. ``eq``: yields ``true`` if the operands are equal, ``false``
6373 otherwise. No sign interpretation is necessary or performed.
6374 #. ``ne``: yields ``true`` if the operands are unequal, ``false``
6375 otherwise. No sign interpretation is necessary or performed.
6376 #. ``ugt``: interprets the operands as unsigned values and yields
6377 ``true`` if ``op1`` is greater than ``op2``.
6378 #. ``uge``: interprets the operands as unsigned values and yields
6379 ``true`` if ``op1`` is greater than or equal to ``op2``.
6380 #. ``ult``: interprets the operands as unsigned values and yields
6381 ``true`` if ``op1`` is less than ``op2``.
6382 #. ``ule``: interprets the operands as unsigned values and yields
6383 ``true`` if ``op1`` is less than or equal to ``op2``.
6384 #. ``sgt``: interprets the operands as signed values and yields ``true``
6385 if ``op1`` is greater than ``op2``.
6386 #. ``sge``: interprets the operands as signed values and yields ``true``
6387 if ``op1`` is greater than or equal to ``op2``.
6388 #. ``slt``: interprets the operands as signed values and yields ``true``
6389 if ``op1`` is less than ``op2``.
6390 #. ``sle``: interprets the operands as signed values and yields ``true``
6391 if ``op1`` is less than or equal to ``op2``.
6393 If the operands are :ref:`pointer <t_pointer>` typed, the pointer values
6394 are compared as if they were integers.
6396 If the operands are integer vectors, then they are compared element by
6397 element. The result is an ``i1`` vector with the same number of elements
6398 as the values being compared. Otherwise, the result is an ``i1``.
6403 .. code-block:: llvm
6405 <result> = icmp eq i32 4, 5 ; yields: result=false
6406 <result> = icmp ne float* %X, %X ; yields: result=false
6407 <result> = icmp ult i16 4, 5 ; yields: result=true
6408 <result> = icmp sgt i16 4, 5 ; yields: result=false
6409 <result> = icmp ule i16 -4, 5 ; yields: result=false
6410 <result> = icmp sge i16 4, 5 ; yields: result=false
6412 Note that the code generator does not yet support vector types with the
6413 ``icmp`` instruction.
6417 '``fcmp``' Instruction
6418 ^^^^^^^^^^^^^^^^^^^^^^
6425 <result> = fcmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
6430 The '``fcmp``' instruction returns a boolean value or vector of boolean
6431 values based on comparison of its operands.
6433 If the operands are floating point scalars, then the result type is a
6434 boolean (:ref:`i1 <t_integer>`).
6436 If the operands are floating point vectors, then the result type is a
6437 vector of boolean with the same number of elements as the operands being
6443 The '``fcmp``' instruction takes three operands. The first operand is
6444 the condition code indicating the kind of comparison to perform. It is
6445 not a value, just a keyword. The possible condition code are:
6447 #. ``false``: no comparison, always returns false
6448 #. ``oeq``: ordered and equal
6449 #. ``ogt``: ordered and greater than
6450 #. ``oge``: ordered and greater than or equal
6451 #. ``olt``: ordered and less than
6452 #. ``ole``: ordered and less than or equal
6453 #. ``one``: ordered and not equal
6454 #. ``ord``: ordered (no nans)
6455 #. ``ueq``: unordered or equal
6456 #. ``ugt``: unordered or greater than
6457 #. ``uge``: unordered or greater than or equal
6458 #. ``ult``: unordered or less than
6459 #. ``ule``: unordered or less than or equal
6460 #. ``une``: unordered or not equal
6461 #. ``uno``: unordered (either nans)
6462 #. ``true``: no comparison, always returns true
6464 *Ordered* means that neither operand is a QNAN while *unordered* means
6465 that either operand may be a QNAN.
6467 Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating
6468 point <t_floating>` type or a :ref:`vector <t_vector>` of floating point
6469 type. They must have identical types.
6474 The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the
6475 condition code given as ``cond``. If the operands are vectors, then the
6476 vectors are compared element by element. Each comparison performed
6477 always yields an :ref:`i1 <t_integer>` result, as follows:
6479 #. ``false``: always yields ``false``, regardless of operands.
6480 #. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1``
6481 is equal to ``op2``.
6482 #. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1``
6483 is greater than ``op2``.
6484 #. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1``
6485 is greater than or equal to ``op2``.
6486 #. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1``
6487 is less than ``op2``.
6488 #. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1``
6489 is less than or equal to ``op2``.
6490 #. ``one``: yields ``true`` if both operands are not a QNAN and ``op1``
6491 is not equal to ``op2``.
6492 #. ``ord``: yields ``true`` if both operands are not a QNAN.
6493 #. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is
6495 #. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is
6496 greater than ``op2``.
6497 #. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is
6498 greater than or equal to ``op2``.
6499 #. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is
6501 #. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is
6502 less than or equal to ``op2``.
6503 #. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is
6504 not equal to ``op2``.
6505 #. ``uno``: yields ``true`` if either operand is a QNAN.
6506 #. ``true``: always yields ``true``, regardless of operands.
6511 .. code-block:: llvm
6513 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false
6514 <result> = fcmp one float 4.0, 5.0 ; yields: result=true
6515 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true
6516 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false
6518 Note that the code generator does not yet support vector types with the
6519 ``fcmp`` instruction.
6523 '``phi``' Instruction
6524 ^^^^^^^^^^^^^^^^^^^^^
6531 <result> = phi <ty> [ <val0>, <label0>], ...
6536 The '``phi``' instruction is used to implement the φ node in the SSA
6537 graph representing the function.
6542 The type of the incoming values is specified with the first type field.
6543 After this, the '``phi``' instruction takes a list of pairs as
6544 arguments, with one pair for each predecessor basic block of the current
6545 block. Only values of :ref:`first class <t_firstclass>` type may be used as
6546 the value arguments to the PHI node. Only labels may be used as the
6549 There must be no non-phi instructions between the start of a basic block
6550 and the PHI instructions: i.e. PHI instructions must be first in a basic
6553 For the purposes of the SSA form, the use of each incoming value is
6554 deemed to occur on the edge from the corresponding predecessor block to
6555 the current block (but after any definition of an '``invoke``'
6556 instruction's return value on the same edge).
6561 At runtime, the '``phi``' instruction logically takes on the value
6562 specified by the pair corresponding to the predecessor basic block that
6563 executed just prior to the current block.
6568 .. code-block:: llvm
6570 Loop: ; Infinite loop that counts from 0 on up...
6571 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6572 %nextindvar = add i32 %indvar, 1
6577 '``select``' Instruction
6578 ^^^^^^^^^^^^^^^^^^^^^^^^
6585 <result> = select selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty
6587 selty is either i1 or {<N x i1>}
6592 The '``select``' instruction is used to choose one value based on a
6593 condition, without IR-level branching.
6598 The '``select``' instruction requires an 'i1' value or a vector of 'i1'
6599 values indicating the condition, and two values of the same :ref:`first
6600 class <t_firstclass>` type. If the val1/val2 are vectors and the
6601 condition is a scalar, then entire vectors are selected, not individual
6607 If the condition is an i1 and it evaluates to 1, the instruction returns
6608 the first value argument; otherwise, it returns the second value
6611 If the condition is a vector of i1, then the value arguments must be
6612 vectors of the same size, and the selection is done element by element.
6617 .. code-block:: llvm
6619 %X = select i1 true, i8 17, i8 42 ; yields i8:17
6623 '``call``' Instruction
6624 ^^^^^^^^^^^^^^^^^^^^^^
6631 <result> = [tail | musttail] call [cconv] [ret attrs] <ty> [<fnty>*] <fnptrval>(<function args>) [fn attrs]
6636 The '``call``' instruction represents a simple function call.
6641 This instruction requires several arguments:
6643 #. The optional ``tail`` and ``musttail`` markers indicate that the optimizers
6644 should perform tail call optimization. The ``tail`` marker is a hint that
6645 `can be ignored <CodeGenerator.html#sibcallopt>`_. The ``musttail`` marker
6646 means that the call must be tail call optimized in order for the program to
6647 be correct. The ``musttail`` marker provides these guarantees:
6649 #. The call will not cause unbounded stack growth if it is part of a
6650 recursive cycle in the call graph.
6651 #. Arguments with the :ref:`inalloca <attr_inalloca>` attribute are
6654 Both markers imply that the callee does not access allocas or varargs from
6655 the caller. Calls marked ``musttail`` must obey the following additional
6658 - The call must immediately precede a :ref:`ret <i_ret>` instruction,
6659 or a pointer bitcast followed by a ret instruction.
6660 - The ret instruction must return the (possibly bitcasted) value
6661 produced by the call or void.
6662 - The caller and callee prototypes must match. Pointer types of
6663 parameters or return types may differ in pointee type, but not
6665 - The calling conventions of the caller and callee must match.
6666 - All ABI-impacting function attributes, such as sret, byval, inreg,
6667 returned, and inalloca, must match.
6668 - The callee must be varargs iff the caller is varargs. Bitcasting a
6669 non-varargs function to the appropriate varargs type is legal so
6670 long as the non-varargs prefixes obey the other rules.
6672 Tail call optimization for calls marked ``tail`` is guaranteed to occur if
6673 the following conditions are met:
6675 - Caller and callee both have the calling convention ``fastcc``.
6676 - The call is in tail position (ret immediately follows call and ret
6677 uses value of call or is void).
6678 - Option ``-tailcallopt`` is enabled, or
6679 ``llvm::GuaranteedTailCallOpt`` is ``true``.
6680 - `Platform-specific constraints are
6681 met. <CodeGenerator.html#tailcallopt>`_
6683 #. The optional "cconv" marker indicates which :ref:`calling
6684 convention <callingconv>` the call should use. If none is
6685 specified, the call defaults to using C calling conventions. The
6686 calling convention of the call must match the calling convention of
6687 the target function, or else the behavior is undefined.
6688 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
6689 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
6691 #. '``ty``': the type of the call instruction itself which is also the
6692 type of the return value. Functions that return no value are marked
6694 #. '``fnty``': shall be the signature of the pointer to function value
6695 being invoked. The argument types must match the types implied by
6696 this signature. This type can be omitted if the function is not
6697 varargs and if the function type does not return a pointer to a
6699 #. '``fnptrval``': An LLVM value containing a pointer to a function to
6700 be invoked. In most cases, this is a direct function invocation, but
6701 indirect ``call``'s are just as possible, calling an arbitrary pointer
6703 #. '``function args``': argument list whose types match the function
6704 signature argument types and parameter attributes. All arguments must
6705 be of :ref:`first class <t_firstclass>` type. If the function signature
6706 indicates the function accepts a variable number of arguments, the
6707 extra arguments can be specified.
6708 #. The optional :ref:`function attributes <fnattrs>` list. Only
6709 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
6710 attributes are valid here.
6715 The '``call``' instruction is used to cause control flow to transfer to
6716 a specified function, with its incoming arguments bound to the specified
6717 values. Upon a '``ret``' instruction in the called function, control
6718 flow continues with the instruction after the function call, and the
6719 return value of the function is bound to the result argument.
6724 .. code-block:: llvm
6726 %retval = call i32 @test(i32 %argc)
6727 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32
6728 %X = tail call i32 @foo() ; yields i32
6729 %Y = tail call fastcc i32 @foo() ; yields i32
6730 call void %foo(i8 97 signext)
6732 %struct.A = type { i32, i8 }
6733 %r = call %struct.A @foo() ; yields { i32, i8 }
6734 %gr = extractvalue %struct.A %r, 0 ; yields i32
6735 %gr1 = extractvalue %struct.A %r, 1 ; yields i8
6736 %Z = call void @foo() noreturn ; indicates that %foo never returns normally
6737 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended
6739 llvm treats calls to some functions with names and arguments that match
6740 the standard C99 library as being the C99 library functions, and may
6741 perform optimizations or generate code for them under that assumption.
6742 This is something we'd like to change in the future to provide better
6743 support for freestanding environments and non-C-based languages.
6747 '``va_arg``' Instruction
6748 ^^^^^^^^^^^^^^^^^^^^^^^^
6755 <resultval> = va_arg <va_list*> <arglist>, <argty>
6760 The '``va_arg``' instruction is used to access arguments passed through
6761 the "variable argument" area of a function call. It is used to implement
6762 the ``va_arg`` macro in C.
6767 This instruction takes a ``va_list*`` value and the type of the
6768 argument. It returns a value of the specified argument type and
6769 increments the ``va_list`` to point to the next argument. The actual
6770 type of ``va_list`` is target specific.
6775 The '``va_arg``' instruction loads an argument of the specified type
6776 from the specified ``va_list`` and causes the ``va_list`` to point to
6777 the next argument. For more information, see the variable argument
6778 handling :ref:`Intrinsic Functions <int_varargs>`.
6780 It is legal for this instruction to be called in a function which does
6781 not take a variable number of arguments, for example, the ``vfprintf``
6784 ``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic
6785 function <intrinsics>` because it takes a type as an argument.
6790 See the :ref:`variable argument processing <int_varargs>` section.
6792 Note that the code generator does not yet fully support va\_arg on many
6793 targets. Also, it does not currently support va\_arg with aggregate
6794 types on any target.
6798 '``landingpad``' Instruction
6799 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6806 <resultval> = landingpad <resultty> personality <type> <pers_fn> <clause>+
6807 <resultval> = landingpad <resultty> personality <type> <pers_fn> cleanup <clause>*
6809 <clause> := catch <type> <value>
6810 <clause> := filter <array constant type> <array constant>
6815 The '``landingpad``' instruction is used by `LLVM's exception handling
6816 system <ExceptionHandling.html#overview>`_ to specify that a basic block
6817 is a landing pad --- one where the exception lands, and corresponds to the
6818 code found in the ``catch`` portion of a ``try``/``catch`` sequence. It
6819 defines values supplied by the personality function (``pers_fn``) upon
6820 re-entry to the function. The ``resultval`` has the type ``resultty``.
6825 This instruction takes a ``pers_fn`` value. This is the personality
6826 function associated with the unwinding mechanism. The optional
6827 ``cleanup`` flag indicates that the landing pad block is a cleanup.
6829 A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and
6830 contains the global variable representing the "type" that may be caught
6831 or filtered respectively. Unlike the ``catch`` clause, the ``filter``
6832 clause takes an array constant as its argument. Use
6833 "``[0 x i8**] undef``" for a filter which cannot throw. The
6834 '``landingpad``' instruction must contain *at least* one ``clause`` or
6835 the ``cleanup`` flag.
6840 The '``landingpad``' instruction defines the values which are set by the
6841 personality function (``pers_fn``) upon re-entry to the function, and
6842 therefore the "result type" of the ``landingpad`` instruction. As with
6843 calling conventions, how the personality function results are
6844 represented in LLVM IR is target specific.
6846 The clauses are applied in order from top to bottom. If two
6847 ``landingpad`` instructions are merged together through inlining, the
6848 clauses from the calling function are appended to the list of clauses.
6849 When the call stack is being unwound due to an exception being thrown,
6850 the exception is compared against each ``clause`` in turn. If it doesn't
6851 match any of the clauses, and the ``cleanup`` flag is not set, then
6852 unwinding continues further up the call stack.
6854 The ``landingpad`` instruction has several restrictions:
6856 - A landing pad block is a basic block which is the unwind destination
6857 of an '``invoke``' instruction.
6858 - A landing pad block must have a '``landingpad``' instruction as its
6859 first non-PHI instruction.
6860 - There can be only one '``landingpad``' instruction within the landing
6862 - A basic block that is not a landing pad block may not include a
6863 '``landingpad``' instruction.
6864 - All '``landingpad``' instructions in a function must have the same
6865 personality function.
6870 .. code-block:: llvm
6872 ;; A landing pad which can catch an integer.
6873 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6875 ;; A landing pad that is a cleanup.
6876 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6878 ;; A landing pad which can catch an integer and can only throw a double.
6879 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6881 filter [1 x i8**] [@_ZTId]
6888 LLVM supports the notion of an "intrinsic function". These functions
6889 have well known names and semantics and are required to follow certain
6890 restrictions. Overall, these intrinsics represent an extension mechanism
6891 for the LLVM language that does not require changing all of the
6892 transformations in LLVM when adding to the language (or the bitcode
6893 reader/writer, the parser, etc...).
6895 Intrinsic function names must all start with an "``llvm.``" prefix. This
6896 prefix is reserved in LLVM for intrinsic names; thus, function names may
6897 not begin with this prefix. Intrinsic functions must always be external
6898 functions: you cannot define the body of intrinsic functions. Intrinsic
6899 functions may only be used in call or invoke instructions: it is illegal
6900 to take the address of an intrinsic function. Additionally, because
6901 intrinsic functions are part of the LLVM language, it is required if any
6902 are added that they be documented here.
6904 Some intrinsic functions can be overloaded, i.e., the intrinsic
6905 represents a family of functions that perform the same operation but on
6906 different data types. Because LLVM can represent over 8 million
6907 different integer types, overloading is used commonly to allow an
6908 intrinsic function to operate on any integer type. One or more of the
6909 argument types or the result type can be overloaded to accept any
6910 integer type. Argument types may also be defined as exactly matching a
6911 previous argument's type or the result type. This allows an intrinsic
6912 function which accepts multiple arguments, but needs all of them to be
6913 of the same type, to only be overloaded with respect to a single
6914 argument or the result.
6916 Overloaded intrinsics will have the names of its overloaded argument
6917 types encoded into its function name, each preceded by a period. Only
6918 those types which are overloaded result in a name suffix. Arguments
6919 whose type is matched against another type do not. For example, the
6920 ``llvm.ctpop`` function can take an integer of any width and returns an
6921 integer of exactly the same integer width. This leads to a family of
6922 functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and
6923 ``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is
6924 overloaded, and only one type suffix is required. Because the argument's
6925 type is matched against the return type, it does not require its own
6928 To learn how to add an intrinsic function, please see the `Extending
6929 LLVM Guide <ExtendingLLVM.html>`_.
6933 Variable Argument Handling Intrinsics
6934 -------------------------------------
6936 Variable argument support is defined in LLVM with the
6937 :ref:`va_arg <i_va_arg>` instruction and these three intrinsic
6938 functions. These functions are related to the similarly named macros
6939 defined in the ``<stdarg.h>`` header file.
6941 All of these functions operate on arguments that use a target-specific
6942 value type "``va_list``". The LLVM assembly language reference manual
6943 does not define what this type is, so all transformations should be
6944 prepared to handle these functions regardless of the type used.
6946 This example shows how the :ref:`va_arg <i_va_arg>` instruction and the
6947 variable argument handling intrinsic functions are used.
6949 .. code-block:: llvm
6951 ; This struct is different for every platform. For most platforms,
6952 ; it is merely an i8*.
6953 %struct.va_list = type { i8* }
6955 ; For Unix x86_64 platforms, va_list is the following struct:
6956 ; %struct.va_list = type { i32, i32, i8*, i8* }
6958 define i32 @test(i32 %X, ...) {
6959 ; Initialize variable argument processing
6960 %ap = alloca %struct.va_list
6961 %ap2 = bitcast %struct.va_list* %ap to i8*
6962 call void @llvm.va_start(i8* %ap2)
6964 ; Read a single integer argument
6965 %tmp = va_arg i8* %ap2, i32
6967 ; Demonstrate usage of llvm.va_copy and llvm.va_end
6969 %aq2 = bitcast i8** %aq to i8*
6970 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
6971 call void @llvm.va_end(i8* %aq2)
6973 ; Stop processing of arguments.
6974 call void @llvm.va_end(i8* %ap2)
6978 declare void @llvm.va_start(i8*)
6979 declare void @llvm.va_copy(i8*, i8*)
6980 declare void @llvm.va_end(i8*)
6984 '``llvm.va_start``' Intrinsic
6985 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6992 declare void @llvm.va_start(i8* <arglist>)
6997 The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for
6998 subsequent use by ``va_arg``.
7003 The argument is a pointer to a ``va_list`` element to initialize.
7008 The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro
7009 available in C. In a target-dependent way, it initializes the
7010 ``va_list`` element to which the argument points, so that the next call
7011 to ``va_arg`` will produce the first variable argument passed to the
7012 function. Unlike the C ``va_start`` macro, this intrinsic does not need
7013 to know the last argument of the function as the compiler can figure
7016 '``llvm.va_end``' Intrinsic
7017 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7024 declare void @llvm.va_end(i8* <arglist>)
7029 The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been
7030 initialized previously with ``llvm.va_start`` or ``llvm.va_copy``.
7035 The argument is a pointer to a ``va_list`` to destroy.
7040 The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro
7041 available in C. In a target-dependent way, it destroys the ``va_list``
7042 element to which the argument points. Calls to
7043 :ref:`llvm.va_start <int_va_start>` and
7044 :ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to
7049 '``llvm.va_copy``' Intrinsic
7050 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7057 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
7062 The '``llvm.va_copy``' intrinsic copies the current argument position
7063 from the source argument list to the destination argument list.
7068 The first argument is a pointer to a ``va_list`` element to initialize.
7069 The second argument is a pointer to a ``va_list`` element to copy from.
7074 The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro
7075 available in C. In a target-dependent way, it copies the source
7076 ``va_list`` element into the destination ``va_list`` element. This
7077 intrinsic is necessary because the `` llvm.va_start`` intrinsic may be
7078 arbitrarily complex and require, for example, memory allocation.
7080 Accurate Garbage Collection Intrinsics
7081 --------------------------------------
7083 LLVM support for `Accurate Garbage Collection <GarbageCollection.html>`_
7084 (GC) requires the implementation and generation of these intrinsics.
7085 These intrinsics allow identification of :ref:`GC roots on the
7086 stack <int_gcroot>`, as well as garbage collector implementations that
7087 require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers.
7088 Front-ends for type-safe garbage collected languages should generate
7089 these intrinsics to make use of the LLVM garbage collectors. For more
7090 details, see `Accurate Garbage Collection with
7091 LLVM <GarbageCollection.html>`_.
7093 The garbage collection intrinsics only operate on objects in the generic
7094 address space (address space zero).
7098 '``llvm.gcroot``' Intrinsic
7099 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7106 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
7111 The '``llvm.gcroot``' intrinsic declares the existence of a GC root to
7112 the code generator, and allows some metadata to be associated with it.
7117 The first argument specifies the address of a stack object that contains
7118 the root pointer. The second pointer (which must be either a constant or
7119 a global value address) contains the meta-data to be associated with the
7125 At runtime, a call to this intrinsic stores a null pointer into the
7126 "ptrloc" location. At compile-time, the code generator generates
7127 information to allow the runtime to find the pointer at GC safe points.
7128 The '``llvm.gcroot``' intrinsic may only be used in a function which
7129 :ref:`specifies a GC algorithm <gc>`.
7133 '``llvm.gcread``' Intrinsic
7134 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7141 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
7146 The '``llvm.gcread``' intrinsic identifies reads of references from heap
7147 locations, allowing garbage collector implementations that require read
7153 The second argument is the address to read from, which should be an
7154 address allocated from the garbage collector. The first object is a
7155 pointer to the start of the referenced object, if needed by the language
7156 runtime (otherwise null).
7161 The '``llvm.gcread``' intrinsic has the same semantics as a load
7162 instruction, but may be replaced with substantially more complex code by
7163 the garbage collector runtime, as needed. The '``llvm.gcread``'
7164 intrinsic may only be used in a function which :ref:`specifies a GC
7169 '``llvm.gcwrite``' Intrinsic
7170 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7177 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
7182 The '``llvm.gcwrite``' intrinsic identifies writes of references to heap
7183 locations, allowing garbage collector implementations that require write
7184 barriers (such as generational or reference counting collectors).
7189 The first argument is the reference to store, the second is the start of
7190 the object to store it to, and the third is the address of the field of
7191 Obj to store to. If the runtime does not require a pointer to the
7192 object, Obj may be null.
7197 The '``llvm.gcwrite``' intrinsic has the same semantics as a store
7198 instruction, but may be replaced with substantially more complex code by
7199 the garbage collector runtime, as needed. The '``llvm.gcwrite``'
7200 intrinsic may only be used in a function which :ref:`specifies a GC
7203 Code Generator Intrinsics
7204 -------------------------
7206 These intrinsics are provided by LLVM to expose special features that
7207 may only be implemented with code generator support.
7209 '``llvm.returnaddress``' Intrinsic
7210 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7217 declare i8 *@llvm.returnaddress(i32 <level>)
7222 The '``llvm.returnaddress``' intrinsic attempts to compute a
7223 target-specific value indicating the return address of the current
7224 function or one of its callers.
7229 The argument to this intrinsic indicates which function to return the
7230 address for. Zero indicates the calling function, one indicates its
7231 caller, etc. The argument is **required** to be a constant integer
7237 The '``llvm.returnaddress``' intrinsic either returns a pointer
7238 indicating the return address of the specified call frame, or zero if it
7239 cannot be identified. The value returned by this intrinsic is likely to
7240 be incorrect or 0 for arguments other than zero, so it should only be
7241 used for debugging purposes.
7243 Note that calling this intrinsic does not prevent function inlining or
7244 other aggressive transformations, so the value returned may not be that
7245 of the obvious source-language caller.
7247 '``llvm.frameaddress``' Intrinsic
7248 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7255 declare i8* @llvm.frameaddress(i32 <level>)
7260 The '``llvm.frameaddress``' intrinsic attempts to return the
7261 target-specific frame pointer value for the specified stack frame.
7266 The argument to this intrinsic indicates which function to return the
7267 frame pointer for. Zero indicates the calling function, one indicates
7268 its caller, etc. The argument is **required** to be a constant integer
7274 The '``llvm.frameaddress``' intrinsic either returns a pointer
7275 indicating the frame address of the specified call frame, or zero if it
7276 cannot be identified. The value returned by this intrinsic is likely to
7277 be incorrect or 0 for arguments other than zero, so it should only be
7278 used for debugging purposes.
7280 Note that calling this intrinsic does not prevent function inlining or
7281 other aggressive transformations, so the value returned may not be that
7282 of the obvious source-language caller.
7284 '``llvm.frameallocate``' and '``llvm.framerecover``' Intrinsics
7285 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7292 declare i8* @llvm.frameallocate(i32 %size)
7293 declare i8* @llvm.framerecover(i8* %func, i8* %fp)
7298 The '``llvm.frameallocate``' intrinsic allocates stack memory at some fixed
7299 offset from the frame pointer, and the '``llvm.framerecover``'
7300 intrinsic applies that offset to a live frame pointer to recover the address of
7301 the allocation. The offset is computed during frame layout of the caller of
7302 ``llvm.frameallocate``.
7307 The ``size`` argument to '``llvm.frameallocate``' must be a constant integer
7308 indicating the amount of stack memory to allocate. As with allocas, allocating
7309 zero bytes is legal, but the result is undefined.
7311 The ``func`` argument to '``llvm.framerecover``' must be a constant
7312 bitcasted pointer to a function defined in the current module. The code
7313 generator cannot determine the frame allocation offset of functions defined in
7316 The ``fp`` argument to '``llvm.framerecover``' must be a frame
7317 pointer of a call frame that is currently live. The return value of
7318 '``llvm.frameaddress``' is one way to produce such a value, but most platforms
7319 also expose the frame pointer through stack unwinding mechanisms.
7324 These intrinsics allow a group of functions to access one stack memory
7325 allocation in an ancestor stack frame. The memory returned from
7326 '``llvm.frameallocate``' may be allocated prior to stack realignment, so the
7327 memory is only aligned to the ABI-required stack alignment. Each function may
7328 only call '``llvm.frameallocate``' one or zero times from the function entry
7329 block. The frame allocation intrinsic inhibits inlining, as any frame
7330 allocations in the inlined function frame are likely to be at a different
7331 offset from the one used by '``llvm.framerecover``' called with the
7334 .. _int_read_register:
7335 .. _int_write_register:
7337 '``llvm.read_register``' and '``llvm.write_register``' Intrinsics
7338 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7345 declare i32 @llvm.read_register.i32(metadata)
7346 declare i64 @llvm.read_register.i64(metadata)
7347 declare void @llvm.write_register.i32(metadata, i32 @value)
7348 declare void @llvm.write_register.i64(metadata, i64 @value)
7354 The '``llvm.read_register``' and '``llvm.write_register``' intrinsics
7355 provides access to the named register. The register must be valid on
7356 the architecture being compiled to. The type needs to be compatible
7357 with the register being read.
7362 The '``llvm.read_register``' intrinsic returns the current value of the
7363 register, where possible. The '``llvm.write_register``' intrinsic sets
7364 the current value of the register, where possible.
7366 This is useful to implement named register global variables that need
7367 to always be mapped to a specific register, as is common practice on
7368 bare-metal programs including OS kernels.
7370 The compiler doesn't check for register availability or use of the used
7371 register in surrounding code, including inline assembly. Because of that,
7372 allocatable registers are not supported.
7374 Warning: So far it only works with the stack pointer on selected
7375 architectures (ARM, AArch64, PowerPC and x86_64). Significant amount of
7376 work is needed to support other registers and even more so, allocatable
7381 '``llvm.stacksave``' Intrinsic
7382 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7389 declare i8* @llvm.stacksave()
7394 The '``llvm.stacksave``' intrinsic is used to remember the current state
7395 of the function stack, for use with
7396 :ref:`llvm.stackrestore <int_stackrestore>`. This is useful for
7397 implementing language features like scoped automatic variable sized
7403 This intrinsic returns a opaque pointer value that can be passed to
7404 :ref:`llvm.stackrestore <int_stackrestore>`. When an
7405 ``llvm.stackrestore`` intrinsic is executed with a value saved from
7406 ``llvm.stacksave``, it effectively restores the state of the stack to
7407 the state it was in when the ``llvm.stacksave`` intrinsic executed. In
7408 practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that
7409 were allocated after the ``llvm.stacksave`` was executed.
7411 .. _int_stackrestore:
7413 '``llvm.stackrestore``' Intrinsic
7414 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7421 declare void @llvm.stackrestore(i8* %ptr)
7426 The '``llvm.stackrestore``' intrinsic is used to restore the state of
7427 the function stack to the state it was in when the corresponding
7428 :ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is
7429 useful for implementing language features like scoped automatic variable
7430 sized arrays in C99.
7435 See the description for :ref:`llvm.stacksave <int_stacksave>`.
7437 '``llvm.prefetch``' Intrinsic
7438 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7445 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
7450 The '``llvm.prefetch``' intrinsic is a hint to the code generator to
7451 insert a prefetch instruction if supported; otherwise, it is a noop.
7452 Prefetches have no effect on the behavior of the program but can change
7453 its performance characteristics.
7458 ``address`` is the address to be prefetched, ``rw`` is the specifier
7459 determining if the fetch should be for a read (0) or write (1), and
7460 ``locality`` is a temporal locality specifier ranging from (0) - no
7461 locality, to (3) - extremely local keep in cache. The ``cache type``
7462 specifies whether the prefetch is performed on the data (1) or
7463 instruction (0) cache. The ``rw``, ``locality`` and ``cache type``
7464 arguments must be constant integers.
7469 This intrinsic does not modify the behavior of the program. In
7470 particular, prefetches cannot trap and do not produce a value. On
7471 targets that support this intrinsic, the prefetch can provide hints to
7472 the processor cache for better performance.
7474 '``llvm.pcmarker``' Intrinsic
7475 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7482 declare void @llvm.pcmarker(i32 <id>)
7487 The '``llvm.pcmarker``' intrinsic is a method to export a Program
7488 Counter (PC) in a region of code to simulators and other tools. The
7489 method is target specific, but it is expected that the marker will use
7490 exported symbols to transmit the PC of the marker. The marker makes no
7491 guarantees that it will remain with any specific instruction after
7492 optimizations. It is possible that the presence of a marker will inhibit
7493 optimizations. The intended use is to be inserted after optimizations to
7494 allow correlations of simulation runs.
7499 ``id`` is a numerical id identifying the marker.
7504 This intrinsic does not modify the behavior of the program. Backends
7505 that do not support this intrinsic may ignore it.
7507 '``llvm.readcyclecounter``' Intrinsic
7508 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7515 declare i64 @llvm.readcyclecounter()
7520 The '``llvm.readcyclecounter``' intrinsic provides access to the cycle
7521 counter register (or similar low latency, high accuracy clocks) on those
7522 targets that support it. On X86, it should map to RDTSC. On Alpha, it
7523 should map to RPCC. As the backing counters overflow quickly (on the
7524 order of 9 seconds on alpha), this should only be used for small
7530 When directly supported, reading the cycle counter should not modify any
7531 memory. Implementations are allowed to either return a application
7532 specific value or a system wide value. On backends without support, this
7533 is lowered to a constant 0.
7535 Note that runtime support may be conditional on the privilege-level code is
7536 running at and the host platform.
7538 '``llvm.clear_cache``' Intrinsic
7539 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7546 declare void @llvm.clear_cache(i8*, i8*)
7551 The '``llvm.clear_cache``' intrinsic ensures visibility of modifications
7552 in the specified range to the execution unit of the processor. On
7553 targets with non-unified instruction and data cache, the implementation
7554 flushes the instruction cache.
7559 On platforms with coherent instruction and data caches (e.g. x86), this
7560 intrinsic is a nop. On platforms with non-coherent instruction and data
7561 cache (e.g. ARM, MIPS), the intrinsic is lowered either to appropriate
7562 instructions or a system call, if cache flushing requires special
7565 The default behavior is to emit a call to ``__clear_cache`` from the run
7568 This instrinsic does *not* empty the instruction pipeline. Modifications
7569 of the current function are outside the scope of the intrinsic.
7571 '``llvm.instrprof_increment``' Intrinsic
7572 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7579 declare void @llvm.instrprof_increment(i8* <name>, i64 <hash>,
7580 i32 <num-counters>, i32 <index>)
7585 The '``llvm.instrprof_increment``' intrinsic can be emitted by a
7586 frontend for use with instrumentation based profiling. These will be
7587 lowered by the ``-instrprof`` pass to generate execution counts of a
7593 The first argument is a pointer to a global variable containing the
7594 name of the entity being instrumented. This should generally be the
7595 (mangled) function name for a set of counters.
7597 The second argument is a hash value that can be used by the consumer
7598 of the profile data to detect changes to the instrumented source, and
7599 the third is the number of counters associated with ``name``. It is an
7600 error if ``hash`` or ``num-counters`` differ between two instances of
7601 ``instrprof_increment`` that refer to the same name.
7603 The last argument refers to which of the counters for ``name`` should
7604 be incremented. It should be a value between 0 and ``num-counters``.
7609 This intrinsic represents an increment of a profiling counter. It will
7610 cause the ``-instrprof`` pass to generate the appropriate data
7611 structures and the code to increment the appropriate value, in a
7612 format that can be written out by a compiler runtime and consumed via
7613 the ``llvm-profdata`` tool.
7615 Standard C Library Intrinsics
7616 -----------------------------
7618 LLVM provides intrinsics for a few important standard C library
7619 functions. These intrinsics allow source-language front-ends to pass
7620 information about the alignment of the pointer arguments to the code
7621 generator, providing opportunity for more efficient code generation.
7625 '``llvm.memcpy``' Intrinsic
7626 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7631 This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any
7632 integer bit width and for different address spaces. Not all targets
7633 support all bit widths however.
7637 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
7638 i32 <len>, i32 <align>, i1 <isvolatile>)
7639 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
7640 i64 <len>, i32 <align>, i1 <isvolatile>)
7645 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
7646 source location to the destination location.
7648 Note that, unlike the standard libc function, the ``llvm.memcpy.*``
7649 intrinsics do not return a value, takes extra alignment/isvolatile
7650 arguments and the pointers can be in specified address spaces.
7655 The first argument is a pointer to the destination, the second is a
7656 pointer to the source. The third argument is an integer argument
7657 specifying the number of bytes to copy, the fourth argument is the
7658 alignment of the source and destination locations, and the fifth is a
7659 boolean indicating a volatile access.
7661 If the call to this intrinsic has an alignment value that is not 0 or 1,
7662 then the caller guarantees that both the source and destination pointers
7663 are aligned to that boundary.
7665 If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is
7666 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
7667 very cleanly specified and it is unwise to depend on it.
7672 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
7673 source location to the destination location, which are not allowed to
7674 overlap. It copies "len" bytes of memory over. If the argument is known
7675 to be aligned to some boundary, this can be specified as the fourth
7676 argument, otherwise it should be set to 0 or 1 (both meaning no alignment).
7678 '``llvm.memmove``' Intrinsic
7679 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7684 This is an overloaded intrinsic. You can use llvm.memmove on any integer
7685 bit width and for different address space. Not all targets support all
7690 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
7691 i32 <len>, i32 <align>, i1 <isvolatile>)
7692 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
7693 i64 <len>, i32 <align>, i1 <isvolatile>)
7698 The '``llvm.memmove.*``' intrinsics move a block of memory from the
7699 source location to the destination location. It is similar to the
7700 '``llvm.memcpy``' intrinsic but allows the two memory locations to
7703 Note that, unlike the standard libc function, the ``llvm.memmove.*``
7704 intrinsics do not return a value, takes extra alignment/isvolatile
7705 arguments and the pointers can be in specified address spaces.
7710 The first argument is a pointer to the destination, the second is a
7711 pointer to the source. The third argument is an integer argument
7712 specifying the number of bytes to copy, the fourth argument is the
7713 alignment of the source and destination locations, and the fifth is a
7714 boolean indicating a volatile access.
7716 If the call to this intrinsic has an alignment value that is not 0 or 1,
7717 then the caller guarantees that the source and destination pointers are
7718 aligned to that boundary.
7720 If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call
7721 is a :ref:`volatile operation <volatile>`. The detailed access behavior is
7722 not very cleanly specified and it is unwise to depend on it.
7727 The '``llvm.memmove.*``' intrinsics copy a block of memory from the
7728 source location to the destination location, which may overlap. It
7729 copies "len" bytes of memory over. If the argument is known to be
7730 aligned to some boundary, this can be specified as the fourth argument,
7731 otherwise it should be set to 0 or 1 (both meaning no alignment).
7733 '``llvm.memset.*``' Intrinsics
7734 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7739 This is an overloaded intrinsic. You can use llvm.memset on any integer
7740 bit width and for different address spaces. However, not all targets
7741 support all bit widths.
7745 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
7746 i32 <len>, i32 <align>, i1 <isvolatile>)
7747 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
7748 i64 <len>, i32 <align>, i1 <isvolatile>)
7753 The '``llvm.memset.*``' intrinsics fill a block of memory with a
7754 particular byte value.
7756 Note that, unlike the standard libc function, the ``llvm.memset``
7757 intrinsic does not return a value and takes extra alignment/volatile
7758 arguments. Also, the destination can be in an arbitrary address space.
7763 The first argument is a pointer to the destination to fill, the second
7764 is the byte value with which to fill it, the third argument is an
7765 integer argument specifying the number of bytes to fill, and the fourth
7766 argument is the known alignment of the destination location.
7768 If the call to this intrinsic has an alignment value that is not 0 or 1,
7769 then the caller guarantees that the destination pointer is aligned to
7772 If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is
7773 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
7774 very cleanly specified and it is unwise to depend on it.
7779 The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting
7780 at the destination location. If the argument is known to be aligned to
7781 some boundary, this can be specified as the fourth argument, otherwise
7782 it should be set to 0 or 1 (both meaning no alignment).
7784 '``llvm.sqrt.*``' Intrinsic
7785 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7790 This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any
7791 floating point or vector of floating point type. Not all targets support
7796 declare float @llvm.sqrt.f32(float %Val)
7797 declare double @llvm.sqrt.f64(double %Val)
7798 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
7799 declare fp128 @llvm.sqrt.f128(fp128 %Val)
7800 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
7805 The '``llvm.sqrt``' intrinsics return the sqrt of the specified operand,
7806 returning the same value as the libm '``sqrt``' functions would. Unlike
7807 ``sqrt`` in libm, however, ``llvm.sqrt`` has undefined behavior for
7808 negative numbers other than -0.0 (which allows for better optimization,
7809 because there is no need to worry about errno being set).
7810 ``llvm.sqrt(-0.0)`` is defined to return -0.0 like IEEE sqrt.
7815 The argument and return value are floating point numbers of the same
7821 This function returns the sqrt of the specified operand if it is a
7822 nonnegative floating point number.
7824 '``llvm.powi.*``' Intrinsic
7825 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7830 This is an overloaded intrinsic. You can use ``llvm.powi`` on any
7831 floating point or vector of floating point type. Not all targets support
7836 declare float @llvm.powi.f32(float %Val, i32 %power)
7837 declare double @llvm.powi.f64(double %Val, i32 %power)
7838 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7839 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7840 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
7845 The '``llvm.powi.*``' intrinsics return the first operand raised to the
7846 specified (positive or negative) power. The order of evaluation of
7847 multiplications is not defined. When a vector of floating point type is
7848 used, the second argument remains a scalar integer value.
7853 The second argument is an integer power, and the first is a value to
7854 raise to that power.
7859 This function returns the first value raised to the second power with an
7860 unspecified sequence of rounding operations.
7862 '``llvm.sin.*``' Intrinsic
7863 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7868 This is an overloaded intrinsic. You can use ``llvm.sin`` on any
7869 floating point or vector of floating point type. Not all targets support
7874 declare float @llvm.sin.f32(float %Val)
7875 declare double @llvm.sin.f64(double %Val)
7876 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7877 declare fp128 @llvm.sin.f128(fp128 %Val)
7878 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7883 The '``llvm.sin.*``' intrinsics return the sine of the operand.
7888 The argument and return value are floating point numbers of the same
7894 This function returns the sine of the specified operand, returning the
7895 same values as the libm ``sin`` functions would, and handles error
7896 conditions in the same way.
7898 '``llvm.cos.*``' Intrinsic
7899 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7904 This is an overloaded intrinsic. You can use ``llvm.cos`` on any
7905 floating point or vector of floating point type. Not all targets support
7910 declare float @llvm.cos.f32(float %Val)
7911 declare double @llvm.cos.f64(double %Val)
7912 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7913 declare fp128 @llvm.cos.f128(fp128 %Val)
7914 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7919 The '``llvm.cos.*``' intrinsics return the cosine of the operand.
7924 The argument and return value are floating point numbers of the same
7930 This function returns the cosine of the specified operand, returning the
7931 same values as the libm ``cos`` functions would, and handles error
7932 conditions in the same way.
7934 '``llvm.pow.*``' Intrinsic
7935 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7940 This is an overloaded intrinsic. You can use ``llvm.pow`` on any
7941 floating point or vector of floating point type. Not all targets support
7946 declare float @llvm.pow.f32(float %Val, float %Power)
7947 declare double @llvm.pow.f64(double %Val, double %Power)
7948 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7949 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7950 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7955 The '``llvm.pow.*``' intrinsics return the first operand raised to the
7956 specified (positive or negative) power.
7961 The second argument is a floating point power, and the first is a value
7962 to raise to that power.
7967 This function returns the first value raised to the second power,
7968 returning the same values as the libm ``pow`` functions would, and
7969 handles error conditions in the same way.
7971 '``llvm.exp.*``' Intrinsic
7972 ^^^^^^^^^^^^^^^^^^^^^^^^^^
7977 This is an overloaded intrinsic. You can use ``llvm.exp`` on any
7978 floating point or vector of floating point type. Not all targets support
7983 declare float @llvm.exp.f32(float %Val)
7984 declare double @llvm.exp.f64(double %Val)
7985 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7986 declare fp128 @llvm.exp.f128(fp128 %Val)
7987 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7992 The '``llvm.exp.*``' intrinsics perform the exp function.
7997 The argument and return value are floating point numbers of the same
8003 This function returns the same values as the libm ``exp`` functions
8004 would, and handles error conditions in the same way.
8006 '``llvm.exp2.*``' Intrinsic
8007 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8012 This is an overloaded intrinsic. You can use ``llvm.exp2`` on any
8013 floating point or vector of floating point type. Not all targets support
8018 declare float @llvm.exp2.f32(float %Val)
8019 declare double @llvm.exp2.f64(double %Val)
8020 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val)
8021 declare fp128 @llvm.exp2.f128(fp128 %Val)
8022 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val)
8027 The '``llvm.exp2.*``' intrinsics perform the exp2 function.
8032 The argument and return value are floating point numbers of the same
8038 This function returns the same values as the libm ``exp2`` functions
8039 would, and handles error conditions in the same way.
8041 '``llvm.log.*``' Intrinsic
8042 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8047 This is an overloaded intrinsic. You can use ``llvm.log`` on any
8048 floating point or vector of floating point type. Not all targets support
8053 declare float @llvm.log.f32(float %Val)
8054 declare double @llvm.log.f64(double %Val)
8055 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
8056 declare fp128 @llvm.log.f128(fp128 %Val)
8057 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
8062 The '``llvm.log.*``' intrinsics perform the log function.
8067 The argument and return value are floating point numbers of the same
8073 This function returns the same values as the libm ``log`` functions
8074 would, and handles error conditions in the same way.
8076 '``llvm.log10.*``' Intrinsic
8077 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8082 This is an overloaded intrinsic. You can use ``llvm.log10`` on any
8083 floating point or vector of floating point type. Not all targets support
8088 declare float @llvm.log10.f32(float %Val)
8089 declare double @llvm.log10.f64(double %Val)
8090 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val)
8091 declare fp128 @llvm.log10.f128(fp128 %Val)
8092 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val)
8097 The '``llvm.log10.*``' intrinsics perform the log10 function.
8102 The argument and return value are floating point numbers of the same
8108 This function returns the same values as the libm ``log10`` functions
8109 would, and handles error conditions in the same way.
8111 '``llvm.log2.*``' Intrinsic
8112 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8117 This is an overloaded intrinsic. You can use ``llvm.log2`` on any
8118 floating point or vector of floating point type. Not all targets support
8123 declare float @llvm.log2.f32(float %Val)
8124 declare double @llvm.log2.f64(double %Val)
8125 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val)
8126 declare fp128 @llvm.log2.f128(fp128 %Val)
8127 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val)
8132 The '``llvm.log2.*``' intrinsics perform the log2 function.
8137 The argument and return value are floating point numbers of the same
8143 This function returns the same values as the libm ``log2`` functions
8144 would, and handles error conditions in the same way.
8146 '``llvm.fma.*``' Intrinsic
8147 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8152 This is an overloaded intrinsic. You can use ``llvm.fma`` on any
8153 floating point or vector of floating point type. Not all targets support
8158 declare float @llvm.fma.f32(float %a, float %b, float %c)
8159 declare double @llvm.fma.f64(double %a, double %b, double %c)
8160 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
8161 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
8162 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
8167 The '``llvm.fma.*``' intrinsics perform the fused multiply-add
8173 The argument and return value are floating point numbers of the same
8179 This function returns the same values as the libm ``fma`` functions
8180 would, and does not set errno.
8182 '``llvm.fabs.*``' Intrinsic
8183 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8188 This is an overloaded intrinsic. You can use ``llvm.fabs`` on any
8189 floating point or vector of floating point type. Not all targets support
8194 declare float @llvm.fabs.f32(float %Val)
8195 declare double @llvm.fabs.f64(double %Val)
8196 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val)
8197 declare fp128 @llvm.fabs.f128(fp128 %Val)
8198 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val)
8203 The '``llvm.fabs.*``' intrinsics return the absolute value of the
8209 The argument and return value are floating point numbers of the same
8215 This function returns the same values as the libm ``fabs`` functions
8216 would, and handles error conditions in the same way.
8218 '``llvm.minnum.*``' Intrinsic
8219 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8224 This is an overloaded intrinsic. You can use ``llvm.minnum`` on any
8225 floating point or vector of floating point type. Not all targets support
8230 declare float @llvm.minnum.f32(float %Val0, float %Val1)
8231 declare double @llvm.minnum.f64(double %Val0, double %Val1)
8232 declare x86_fp80 @llvm.minnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
8233 declare fp128 @llvm.minnum.f128(fp128 %Val0, fp128 %Val1)
8234 declare ppc_fp128 @llvm.minnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
8239 The '``llvm.minnum.*``' intrinsics return the minimum of the two
8246 The arguments and return value are floating point numbers of the same
8252 Follows the IEEE-754 semantics for minNum, which also match for libm's
8255 If either operand is a NaN, returns the other non-NaN operand. Returns
8256 NaN only if both operands are NaN. If the operands compare equal,
8257 returns a value that compares equal to both operands. This means that
8258 fmin(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
8260 '``llvm.maxnum.*``' Intrinsic
8261 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8266 This is an overloaded intrinsic. You can use ``llvm.maxnum`` on any
8267 floating point or vector of floating point type. Not all targets support
8272 declare float @llvm.maxnum.f32(float %Val0, float %Val1l)
8273 declare double @llvm.maxnum.f64(double %Val0, double %Val1)
8274 declare x86_fp80 @llvm.maxnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
8275 declare fp128 @llvm.maxnum.f128(fp128 %Val0, fp128 %Val1)
8276 declare ppc_fp128 @llvm.maxnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
8281 The '``llvm.maxnum.*``' intrinsics return the maximum of the two
8288 The arguments and return value are floating point numbers of the same
8293 Follows the IEEE-754 semantics for maxNum, which also match for libm's
8296 If either operand is a NaN, returns the other non-NaN operand. Returns
8297 NaN only if both operands are NaN. If the operands compare equal,
8298 returns a value that compares equal to both operands. This means that
8299 fmax(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
8301 '``llvm.copysign.*``' Intrinsic
8302 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8307 This is an overloaded intrinsic. You can use ``llvm.copysign`` on any
8308 floating point or vector of floating point type. Not all targets support
8313 declare float @llvm.copysign.f32(float %Mag, float %Sgn)
8314 declare double @llvm.copysign.f64(double %Mag, double %Sgn)
8315 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn)
8316 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn)
8317 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn)
8322 The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the
8323 first operand and the sign of the second operand.
8328 The arguments and return value are floating point numbers of the same
8334 This function returns the same values as the libm ``copysign``
8335 functions would, and handles error conditions in the same way.
8337 '``llvm.floor.*``' Intrinsic
8338 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8343 This is an overloaded intrinsic. You can use ``llvm.floor`` on any
8344 floating point or vector of floating point type. Not all targets support
8349 declare float @llvm.floor.f32(float %Val)
8350 declare double @llvm.floor.f64(double %Val)
8351 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val)
8352 declare fp128 @llvm.floor.f128(fp128 %Val)
8353 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val)
8358 The '``llvm.floor.*``' intrinsics return the floor of the operand.
8363 The argument and return value are floating point numbers of the same
8369 This function returns the same values as the libm ``floor`` functions
8370 would, and handles error conditions in the same way.
8372 '``llvm.ceil.*``' Intrinsic
8373 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8378 This is an overloaded intrinsic. You can use ``llvm.ceil`` on any
8379 floating point or vector of floating point type. Not all targets support
8384 declare float @llvm.ceil.f32(float %Val)
8385 declare double @llvm.ceil.f64(double %Val)
8386 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val)
8387 declare fp128 @llvm.ceil.f128(fp128 %Val)
8388 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val)
8393 The '``llvm.ceil.*``' intrinsics return the ceiling of the operand.
8398 The argument and return value are floating point numbers of the same
8404 This function returns the same values as the libm ``ceil`` functions
8405 would, and handles error conditions in the same way.
8407 '``llvm.trunc.*``' Intrinsic
8408 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8413 This is an overloaded intrinsic. You can use ``llvm.trunc`` on any
8414 floating point or vector of floating point type. Not all targets support
8419 declare float @llvm.trunc.f32(float %Val)
8420 declare double @llvm.trunc.f64(double %Val)
8421 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val)
8422 declare fp128 @llvm.trunc.f128(fp128 %Val)
8423 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val)
8428 The '``llvm.trunc.*``' intrinsics returns the operand rounded to the
8429 nearest integer not larger in magnitude than the operand.
8434 The argument and return value are floating point numbers of the same
8440 This function returns the same values as the libm ``trunc`` functions
8441 would, and handles error conditions in the same way.
8443 '``llvm.rint.*``' Intrinsic
8444 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8449 This is an overloaded intrinsic. You can use ``llvm.rint`` on any
8450 floating point or vector of floating point type. Not all targets support
8455 declare float @llvm.rint.f32(float %Val)
8456 declare double @llvm.rint.f64(double %Val)
8457 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val)
8458 declare fp128 @llvm.rint.f128(fp128 %Val)
8459 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val)
8464 The '``llvm.rint.*``' intrinsics returns the operand rounded to the
8465 nearest integer. It may raise an inexact floating-point exception if the
8466 operand isn't an integer.
8471 The argument and return value are floating point numbers of the same
8477 This function returns the same values as the libm ``rint`` functions
8478 would, and handles error conditions in the same way.
8480 '``llvm.nearbyint.*``' Intrinsic
8481 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8486 This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any
8487 floating point or vector of floating point type. Not all targets support
8492 declare float @llvm.nearbyint.f32(float %Val)
8493 declare double @llvm.nearbyint.f64(double %Val)
8494 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val)
8495 declare fp128 @llvm.nearbyint.f128(fp128 %Val)
8496 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val)
8501 The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the
8507 The argument and return value are floating point numbers of the same
8513 This function returns the same values as the libm ``nearbyint``
8514 functions would, and handles error conditions in the same way.
8516 '``llvm.round.*``' Intrinsic
8517 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8522 This is an overloaded intrinsic. You can use ``llvm.round`` on any
8523 floating point or vector of floating point type. Not all targets support
8528 declare float @llvm.round.f32(float %Val)
8529 declare double @llvm.round.f64(double %Val)
8530 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val)
8531 declare fp128 @llvm.round.f128(fp128 %Val)
8532 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val)
8537 The '``llvm.round.*``' intrinsics returns the operand rounded to the
8543 The argument and return value are floating point numbers of the same
8549 This function returns the same values as the libm ``round``
8550 functions would, and handles error conditions in the same way.
8552 Bit Manipulation Intrinsics
8553 ---------------------------
8555 LLVM provides intrinsics for a few important bit manipulation
8556 operations. These allow efficient code generation for some algorithms.
8558 '``llvm.bswap.*``' Intrinsics
8559 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8564 This is an overloaded intrinsic function. You can use bswap on any
8565 integer type that is an even number of bytes (i.e. BitWidth % 16 == 0).
8569 declare i16 @llvm.bswap.i16(i16 <id>)
8570 declare i32 @llvm.bswap.i32(i32 <id>)
8571 declare i64 @llvm.bswap.i64(i64 <id>)
8576 The '``llvm.bswap``' family of intrinsics is used to byte swap integer
8577 values with an even number of bytes (positive multiple of 16 bits).
8578 These are useful for performing operations on data that is not in the
8579 target's native byte order.
8584 The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high
8585 and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32``
8586 intrinsic returns an i32 value that has the four bytes of the input i32
8587 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the
8588 returned i32 will have its bytes in 3, 2, 1, 0 order. The
8589 ``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this
8590 concept to additional even-byte lengths (6 bytes, 8 bytes and more,
8593 '``llvm.ctpop.*``' Intrinsic
8594 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8599 This is an overloaded intrinsic. You can use llvm.ctpop on any integer
8600 bit width, or on any vector with integer elements. Not all targets
8601 support all bit widths or vector types, however.
8605 declare i8 @llvm.ctpop.i8(i8 <src>)
8606 declare i16 @llvm.ctpop.i16(i16 <src>)
8607 declare i32 @llvm.ctpop.i32(i32 <src>)
8608 declare i64 @llvm.ctpop.i64(i64 <src>)
8609 declare i256 @llvm.ctpop.i256(i256 <src>)
8610 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
8615 The '``llvm.ctpop``' family of intrinsics counts the number of bits set
8621 The only argument is the value to be counted. The argument may be of any
8622 integer type, or a vector with integer elements. The return type must
8623 match the argument type.
8628 The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within
8629 each element of a vector.
8631 '``llvm.ctlz.*``' Intrinsic
8632 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8637 This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any
8638 integer bit width, or any vector whose elements are integers. Not all
8639 targets support all bit widths or vector types, however.
8643 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>)
8644 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>)
8645 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>)
8646 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>)
8647 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>)
8648 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
8653 The '``llvm.ctlz``' family of intrinsic functions counts the number of
8654 leading zeros in a variable.
8659 The first argument is the value to be counted. This argument may be of
8660 any integer type, or a vector with integer element type. The return
8661 type must match the first argument type.
8663 The second argument must be a constant and is a flag to indicate whether
8664 the intrinsic should ensure that a zero as the first argument produces a
8665 defined result. Historically some architectures did not provide a
8666 defined result for zero values as efficiently, and many algorithms are
8667 now predicated on avoiding zero-value inputs.
8672 The '``llvm.ctlz``' intrinsic counts the leading (most significant)
8673 zeros in a variable, or within each element of the vector. If
8674 ``src == 0`` then the result is the size in bits of the type of ``src``
8675 if ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
8676 ``llvm.ctlz(i32 2) = 30``.
8678 '``llvm.cttz.*``' Intrinsic
8679 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8684 This is an overloaded intrinsic. You can use ``llvm.cttz`` on any
8685 integer bit width, or any vector of integer elements. Not all targets
8686 support all bit widths or vector types, however.
8690 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>)
8691 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>)
8692 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>)
8693 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>)
8694 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>)
8695 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
8700 The '``llvm.cttz``' family of intrinsic functions counts the number of
8706 The first argument is the value to be counted. This argument may be of
8707 any integer type, or a vector with integer element type. The return
8708 type must match the first argument type.
8710 The second argument must be a constant and is a flag to indicate whether
8711 the intrinsic should ensure that a zero as the first argument produces a
8712 defined result. Historically some architectures did not provide a
8713 defined result for zero values as efficiently, and many algorithms are
8714 now predicated on avoiding zero-value inputs.
8719 The '``llvm.cttz``' intrinsic counts the trailing (least significant)
8720 zeros in a variable, or within each element of a vector. If ``src == 0``
8721 then the result is the size in bits of the type of ``src`` if
8722 ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
8723 ``llvm.cttz(2) = 1``.
8725 Arithmetic with Overflow Intrinsics
8726 -----------------------------------
8728 LLVM provides intrinsics for some arithmetic with overflow operations.
8730 '``llvm.sadd.with.overflow.*``' Intrinsics
8731 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8736 This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow``
8737 on any integer bit width.
8741 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
8742 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
8743 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
8748 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
8749 a signed addition of the two arguments, and indicate whether an overflow
8750 occurred during the signed summation.
8755 The arguments (%a and %b) and the first element of the result structure
8756 may be of integer types of any bit width, but they must have the same
8757 bit width. The second element of the result structure must be of type
8758 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8764 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
8765 a signed addition of the two variables. They return a structure --- the
8766 first element of which is the signed summation, and the second element
8767 of which is a bit specifying if the signed summation resulted in an
8773 .. code-block:: llvm
8775 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
8776 %sum = extractvalue {i32, i1} %res, 0
8777 %obit = extractvalue {i32, i1} %res, 1
8778 br i1 %obit, label %overflow, label %normal
8780 '``llvm.uadd.with.overflow.*``' Intrinsics
8781 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8786 This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow``
8787 on any integer bit width.
8791 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
8792 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
8793 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
8798 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
8799 an unsigned addition of the two arguments, and indicate whether a carry
8800 occurred during the unsigned summation.
8805 The arguments (%a and %b) and the first element of the result structure
8806 may be of integer types of any bit width, but they must have the same
8807 bit width. The second element of the result structure must be of type
8808 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
8814 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
8815 an unsigned addition of the two arguments. They return a structure --- the
8816 first element of which is the sum, and the second element of which is a
8817 bit specifying if the unsigned summation resulted in a carry.
8822 .. code-block:: llvm
8824 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
8825 %sum = extractvalue {i32, i1} %res, 0
8826 %obit = extractvalue {i32, i1} %res, 1
8827 br i1 %obit, label %carry, label %normal
8829 '``llvm.ssub.with.overflow.*``' Intrinsics
8830 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8835 This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow``
8836 on any integer bit width.
8840 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
8841 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
8842 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
8847 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
8848 a signed subtraction of the two arguments, and indicate whether an
8849 overflow occurred during the signed subtraction.
8854 The arguments (%a and %b) and the first element of the result structure
8855 may be of integer types of any bit width, but they must have the same
8856 bit width. The second element of the result structure must be of type
8857 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8863 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
8864 a signed subtraction of the two arguments. They return a structure --- the
8865 first element of which is the subtraction, and the second element of
8866 which is a bit specifying if the signed subtraction resulted in an
8872 .. code-block:: llvm
8874 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
8875 %sum = extractvalue {i32, i1} %res, 0
8876 %obit = extractvalue {i32, i1} %res, 1
8877 br i1 %obit, label %overflow, label %normal
8879 '``llvm.usub.with.overflow.*``' Intrinsics
8880 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8885 This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow``
8886 on any integer bit width.
8890 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
8891 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
8892 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
8897 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
8898 an unsigned subtraction of the two arguments, and indicate whether an
8899 overflow occurred during the unsigned subtraction.
8904 The arguments (%a and %b) and the first element of the result structure
8905 may be of integer types of any bit width, but they must have the same
8906 bit width. The second element of the result structure must be of type
8907 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
8913 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
8914 an unsigned subtraction of the two arguments. They return a structure ---
8915 the first element of which is the subtraction, and the second element of
8916 which is a bit specifying if the unsigned subtraction resulted in an
8922 .. code-block:: llvm
8924 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
8925 %sum = extractvalue {i32, i1} %res, 0
8926 %obit = extractvalue {i32, i1} %res, 1
8927 br i1 %obit, label %overflow, label %normal
8929 '``llvm.smul.with.overflow.*``' Intrinsics
8930 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8935 This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow``
8936 on any integer bit width.
8940 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
8941 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
8942 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
8947 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
8948 a signed multiplication of the two arguments, and indicate whether an
8949 overflow occurred during the signed multiplication.
8954 The arguments (%a and %b) and the first element of the result structure
8955 may be of integer types of any bit width, but they must have the same
8956 bit width. The second element of the result structure must be of type
8957 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
8963 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
8964 a signed multiplication of the two arguments. They return a structure ---
8965 the first element of which is the multiplication, and the second element
8966 of which is a bit specifying if the signed multiplication resulted in an
8972 .. code-block:: llvm
8974 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
8975 %sum = extractvalue {i32, i1} %res, 0
8976 %obit = extractvalue {i32, i1} %res, 1
8977 br i1 %obit, label %overflow, label %normal
8979 '``llvm.umul.with.overflow.*``' Intrinsics
8980 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8985 This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow``
8986 on any integer bit width.
8990 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
8991 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
8992 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
8997 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
8998 a unsigned multiplication of the two arguments, and indicate whether an
8999 overflow occurred during the unsigned multiplication.
9004 The arguments (%a and %b) and the first element of the result structure
9005 may be of integer types of any bit width, but they must have the same
9006 bit width. The second element of the result structure must be of type
9007 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
9013 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
9014 an unsigned multiplication of the two arguments. They return a structure ---
9015 the first element of which is the multiplication, and the second
9016 element of which is a bit specifying if the unsigned multiplication
9017 resulted in an overflow.
9022 .. code-block:: llvm
9024 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
9025 %sum = extractvalue {i32, i1} %res, 0
9026 %obit = extractvalue {i32, i1} %res, 1
9027 br i1 %obit, label %overflow, label %normal
9029 Specialised Arithmetic Intrinsics
9030 ---------------------------------
9032 '``llvm.fmuladd.*``' Intrinsic
9033 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9040 declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
9041 declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
9046 The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add
9047 expressions that can be fused if the code generator determines that (a) the
9048 target instruction set has support for a fused operation, and (b) that the
9049 fused operation is more efficient than the equivalent, separate pair of mul
9050 and add instructions.
9055 The '``llvm.fmuladd.*``' intrinsics each take three arguments: two
9056 multiplicands, a and b, and an addend c.
9065 %0 = call float @llvm.fmuladd.f32(%a, %b, %c)
9067 is equivalent to the expression a \* b + c, except that rounding will
9068 not be performed between the multiplication and addition steps if the
9069 code generator fuses the operations. Fusion is not guaranteed, even if
9070 the target platform supports it. If a fused multiply-add is required the
9071 corresponding llvm.fma.\* intrinsic function should be used
9072 instead. This never sets errno, just as '``llvm.fma.*``'.
9077 .. code-block:: llvm
9079 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields float:r2 = (a * b) + c
9081 Half Precision Floating Point Intrinsics
9082 ----------------------------------------
9084 For most target platforms, half precision floating point is a
9085 storage-only format. This means that it is a dense encoding (in memory)
9086 but does not support computation in the format.
9088 This means that code must first load the half-precision floating point
9089 value as an i16, then convert it to float with
9090 :ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can
9091 then be performed on the float value (including extending to double
9092 etc). To store the value back to memory, it is first converted to float
9093 if needed, then converted to i16 with
9094 :ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an
9097 .. _int_convert_to_fp16:
9099 '``llvm.convert.to.fp16``' Intrinsic
9100 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9107 declare i16 @llvm.convert.to.fp16.f32(float %a)
9108 declare i16 @llvm.convert.to.fp16.f64(double %a)
9113 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
9114 conventional floating point type to half precision floating point format.
9119 The intrinsic function contains single argument - the value to be
9125 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
9126 conventional floating point format to half precision floating point format. The
9127 return value is an ``i16`` which contains the converted number.
9132 .. code-block:: llvm
9134 %res = call i16 @llvm.convert.to.fp16.f32(float %a)
9135 store i16 %res, i16* @x, align 2
9137 .. _int_convert_from_fp16:
9139 '``llvm.convert.from.fp16``' Intrinsic
9140 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9147 declare float @llvm.convert.from.fp16.f32(i16 %a)
9148 declare double @llvm.convert.from.fp16.f64(i16 %a)
9153 The '``llvm.convert.from.fp16``' intrinsic function performs a
9154 conversion from half precision floating point format to single precision
9155 floating point format.
9160 The intrinsic function contains single argument - the value to be
9166 The '``llvm.convert.from.fp16``' intrinsic function performs a
9167 conversion from half single precision floating point format to single
9168 precision floating point format. The input half-float value is
9169 represented by an ``i16`` value.
9174 .. code-block:: llvm
9176 %a = load i16* @x, align 2
9177 %res = call float @llvm.convert.from.fp16(i16 %a)
9182 The LLVM debugger intrinsics (which all start with ``llvm.dbg.``
9183 prefix), are described in the `LLVM Source Level
9184 Debugging <SourceLevelDebugging.html#format_common_intrinsics>`_
9187 Exception Handling Intrinsics
9188 -----------------------------
9190 The LLVM exception handling intrinsics (which all start with
9191 ``llvm.eh.`` prefix), are described in the `LLVM Exception
9192 Handling <ExceptionHandling.html#format_common_intrinsics>`_ document.
9196 Trampoline Intrinsics
9197 ---------------------
9199 These intrinsics make it possible to excise one parameter, marked with
9200 the :ref:`nest <nest>` attribute, from a function. The result is a
9201 callable function pointer lacking the nest parameter - the caller does
9202 not need to provide a value for it. Instead, the value to use is stored
9203 in advance in a "trampoline", a block of memory usually allocated on the
9204 stack, which also contains code to splice the nest value into the
9205 argument list. This is used to implement the GCC nested function address
9208 For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)``
9209 then the resulting function pointer has signature ``i32 (i32, i32)*``.
9210 It can be created as follows:
9212 .. code-block:: llvm
9214 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
9215 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
9216 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
9217 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
9218 %fp = bitcast i8* %p to i32 (i32, i32)*
9220 The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to
9221 ``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``.
9225 '``llvm.init.trampoline``' Intrinsic
9226 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9233 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
9238 This fills the memory pointed to by ``tramp`` with executable code,
9239 turning it into a trampoline.
9244 The ``llvm.init.trampoline`` intrinsic takes three arguments, all
9245 pointers. The ``tramp`` argument must point to a sufficiently large and
9246 sufficiently aligned block of memory; this memory is written to by the
9247 intrinsic. Note that the size and the alignment are target-specific -
9248 LLVM currently provides no portable way of determining them, so a
9249 front-end that generates this intrinsic needs to have some
9250 target-specific knowledge. The ``func`` argument must hold a function
9251 bitcast to an ``i8*``.
9256 The block of memory pointed to by ``tramp`` is filled with target
9257 dependent code, turning it into a function. Then ``tramp`` needs to be
9258 passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can
9259 be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new
9260 function's signature is the same as that of ``func`` with any arguments
9261 marked with the ``nest`` attribute removed. At most one such ``nest``
9262 argument is allowed, and it must be of pointer type. Calling the new
9263 function is equivalent to calling ``func`` with the same argument list,
9264 but with ``nval`` used for the missing ``nest`` argument. If, after
9265 calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is
9266 modified, then the effect of any later call to the returned function
9267 pointer is undefined.
9271 '``llvm.adjust.trampoline``' Intrinsic
9272 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9279 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
9284 This performs any required machine-specific adjustment to the address of
9285 a trampoline (passed as ``tramp``).
9290 ``tramp`` must point to a block of memory which already has trampoline
9291 code filled in by a previous call to
9292 :ref:`llvm.init.trampoline <int_it>`.
9297 On some architectures the address of the code to be executed needs to be
9298 different than the address where the trampoline is actually stored. This
9299 intrinsic returns the executable address corresponding to ``tramp``
9300 after performing the required machine specific adjustments. The pointer
9301 returned can then be :ref:`bitcast and executed <int_trampoline>`.
9303 Masked Vector Load and Store Intrinsics
9304 ---------------------------------------
9306 LLVM provides intrinsics for predicated vector load and store operations. The predicate is specified by a mask operand, which holds one bit per vector element, switching the associated vector lane on or off. The memory addresses corresponding to the "off" lanes are not accessed. When all bits of the mask are on, the intrinsic is identical to a regular vector load or store. When all bits are off, no memory is accessed.
9310 '``llvm.masked.load.*``' Intrinsics
9311 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9315 This is an overloaded intrinsic. The loaded data is a vector of any integer or floating point data type.
9319 declare <16 x float> @llvm.masked.load.v16f32 (<16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>)
9320 declare <2 x double> @llvm.masked.load.v2f64 (<2 x double>* <ptr>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>)
9325 Reads a vector from memory according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes. The masked-off lanes in the result vector are taken from the corresponding lanes in the passthru operand.
9331 The first operand is the base pointer for the load. The second operand is the alignment of the source location. It must be a constant integer value. The third operand, mask, is a vector of boolean 'i1' values with the same number of elements as the return type. The fourth is a pass-through value that is used to fill the masked-off lanes of the result. The return type, underlying type of the base pointer and the type of passthru operand are the same vector types.
9337 The '``llvm.masked.load``' intrinsic is designed for conditional reading of selected vector elements in a single IR operation. It is useful for targets that support vector masked loads and allows vectorizing predicated basic blocks on these targets. Other targets may support this intrinsic differently, for example by lowering it into a sequence of branches that guard scalar load operations.
9338 The result of this operation is equivalent to a regular vector load instruction followed by a 'select' between the loaded and the passthru values, predicated on the same mask. However, using this intrinsic prevents exceptions on memory access to masked-off lanes.
9343 %res = call <16 x float> @llvm.masked.load.v16f32 (<16 x float>* %ptr, i32 4, <16 x i1>%mask, <16 x float> %passthru)
9345 ;; The result of the two following instructions is identical aside from potential memory access exception
9346 %loadlal = load <16 x float>* %ptr, align 4
9347 %res = select <16 x i1> %mask, <16 x float> %loadlal, <16 x float> %passthru
9351 '``llvm.masked.store.*``' Intrinsics
9352 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9356 This is an overloaded intrinsic. The data stored in memory is a vector of any integer or floating point data type.
9360 declare void @llvm.masked.store.v8i32 (<8 x i32> <value>, <8 x i32> * <ptr>, i32 <alignment>, <8 x i1> <mask>)
9361 declare void @llvm.masked.store.v16f32(<16 x i32> <value>, <16 x i32>* <ptr>, i32 <alignment>, <16 x i1> <mask>)
9366 Writes a vector to memory according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes.
9371 The first operand is the vector value to be written to memory. The second operand is the base pointer for the store, it has the same underlying type as the value operand. The third operand is the alignment of the destination location. The fourth operand, mask, is a vector of boolean values. The types of the mask and the value operand must have the same number of vector elements.
9377 The '``llvm.masked.store``' intrinsics is designed for conditional writing of selected vector elements in a single IR operation. It is useful for targets that support vector masked store and allows vectorizing predicated basic blocks on these targets. Other targets may support this intrinsic differently, for example by lowering it into a sequence of branches that guard scalar store operations.
9378 The result of this operation is equivalent to a load-modify-store sequence. However, using this intrinsic prevents exceptions and data races on memory access to masked-off lanes.
9382 call void @llvm.masked.store.v16f32(<16 x float> %value, <16 x float>* %ptr, i32 4, <16 x i1> %mask)
9384 ;; The result of the following instructions is identical aside from potential data races and memory access exceptions
9385 %oldval = load <16 x float>* %ptr, align 4
9386 %res = select <16 x i1> %mask, <16 x float> %value, <16 x float> %oldval
9387 store <16 x float> %res, <16 x float>* %ptr, align 4
9393 This class of intrinsics provides information about the lifetime of
9394 memory objects and ranges where variables are immutable.
9398 '``llvm.lifetime.start``' Intrinsic
9399 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9406 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
9411 The '``llvm.lifetime.start``' intrinsic specifies the start of a memory
9417 The first argument is a constant integer representing the size of the
9418 object, or -1 if it is variable sized. The second argument is a pointer
9424 This intrinsic indicates that before this point in the code, the value
9425 of the memory pointed to by ``ptr`` is dead. This means that it is known
9426 to never be used and has an undefined value. A load from the pointer
9427 that precedes this intrinsic can be replaced with ``'undef'``.
9431 '``llvm.lifetime.end``' Intrinsic
9432 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9439 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
9444 The '``llvm.lifetime.end``' intrinsic specifies the end of a memory
9450 The first argument is a constant integer representing the size of the
9451 object, or -1 if it is variable sized. The second argument is a pointer
9457 This intrinsic indicates that after this point in the code, the value of
9458 the memory pointed to by ``ptr`` is dead. This means that it is known to
9459 never be used and has an undefined value. Any stores into the memory
9460 object following this intrinsic may be removed as dead.
9462 '``llvm.invariant.start``' Intrinsic
9463 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9470 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
9475 The '``llvm.invariant.start``' intrinsic specifies that the contents of
9476 a memory object will not change.
9481 The first argument is a constant integer representing the size of the
9482 object, or -1 if it is variable sized. The second argument is a pointer
9488 This intrinsic indicates that until an ``llvm.invariant.end`` that uses
9489 the return value, the referenced memory location is constant and
9492 '``llvm.invariant.end``' Intrinsic
9493 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9500 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
9505 The '``llvm.invariant.end``' intrinsic specifies that the contents of a
9506 memory object are mutable.
9511 The first argument is the matching ``llvm.invariant.start`` intrinsic.
9512 The second argument is a constant integer representing the size of the
9513 object, or -1 if it is variable sized and the third argument is a
9514 pointer to the object.
9519 This intrinsic indicates that the memory is mutable again.
9524 This class of intrinsics is designed to be generic and has no specific
9527 '``llvm.var.annotation``' Intrinsic
9528 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9535 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
9540 The '``llvm.var.annotation``' intrinsic.
9545 The first argument is a pointer to a value, the second is a pointer to a
9546 global string, the third is a pointer to a global string which is the
9547 source file name, and the last argument is the line number.
9552 This intrinsic allows annotation of local variables with arbitrary
9553 strings. This can be useful for special purpose optimizations that want
9554 to look for these annotations. These have no other defined use; they are
9555 ignored by code generation and optimization.
9557 '``llvm.ptr.annotation.*``' Intrinsic
9558 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9563 This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a
9564 pointer to an integer of any width. *NOTE* you must specify an address space for
9565 the pointer. The identifier for the default address space is the integer
9570 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
9571 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>)
9572 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>)
9573 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>)
9574 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>)
9579 The '``llvm.ptr.annotation``' intrinsic.
9584 The first argument is a pointer to an integer value of arbitrary bitwidth
9585 (result of some expression), the second is a pointer to a global string, the
9586 third is a pointer to a global string which is the source file name, and the
9587 last argument is the line number. It returns the value of the first argument.
9592 This intrinsic allows annotation of a pointer to an integer with arbitrary
9593 strings. This can be useful for special purpose optimizations that want to look
9594 for these annotations. These have no other defined use; they are ignored by code
9595 generation and optimization.
9597 '``llvm.annotation.*``' Intrinsic
9598 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9603 This is an overloaded intrinsic. You can use '``llvm.annotation``' on
9604 any integer bit width.
9608 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
9609 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
9610 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
9611 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
9612 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
9617 The '``llvm.annotation``' intrinsic.
9622 The first argument is an integer value (result of some expression), the
9623 second is a pointer to a global string, the third is a pointer to a
9624 global string which is the source file name, and the last argument is
9625 the line number. It returns the value of the first argument.
9630 This intrinsic allows annotations to be put on arbitrary expressions
9631 with arbitrary strings. This can be useful for special purpose
9632 optimizations that want to look for these annotations. These have no
9633 other defined use; they are ignored by code generation and optimization.
9635 '``llvm.trap``' Intrinsic
9636 ^^^^^^^^^^^^^^^^^^^^^^^^^
9643 declare void @llvm.trap() noreturn nounwind
9648 The '``llvm.trap``' intrinsic.
9658 This intrinsic is lowered to the target dependent trap instruction. If
9659 the target does not have a trap instruction, this intrinsic will be
9660 lowered to a call of the ``abort()`` function.
9662 '``llvm.debugtrap``' Intrinsic
9663 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9670 declare void @llvm.debugtrap() nounwind
9675 The '``llvm.debugtrap``' intrinsic.
9685 This intrinsic is lowered to code which is intended to cause an
9686 execution trap with the intention of requesting the attention of a
9689 '``llvm.stackprotector``' Intrinsic
9690 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9697 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
9702 The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it
9703 onto the stack at ``slot``. The stack slot is adjusted to ensure that it
9704 is placed on the stack before local variables.
9709 The ``llvm.stackprotector`` intrinsic requires two pointer arguments.
9710 The first argument is the value loaded from the stack guard
9711 ``@__stack_chk_guard``. The second variable is an ``alloca`` that has
9712 enough space to hold the value of the guard.
9717 This intrinsic causes the prologue/epilogue inserter to force the position of
9718 the ``AllocaInst`` stack slot to be before local variables on the stack. This is
9719 to ensure that if a local variable on the stack is overwritten, it will destroy
9720 the value of the guard. When the function exits, the guard on the stack is
9721 checked against the original guard by ``llvm.stackprotectorcheck``. If they are
9722 different, then ``llvm.stackprotectorcheck`` causes the program to abort by
9723 calling the ``__stack_chk_fail()`` function.
9725 '``llvm.stackprotectorcheck``' Intrinsic
9726 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9733 declare void @llvm.stackprotectorcheck(i8** <guard>)
9738 The ``llvm.stackprotectorcheck`` intrinsic compares ``guard`` against an already
9739 created stack protector and if they are not equal calls the
9740 ``__stack_chk_fail()`` function.
9745 The ``llvm.stackprotectorcheck`` intrinsic requires one pointer argument, the
9746 the variable ``@__stack_chk_guard``.
9751 This intrinsic is provided to perform the stack protector check by comparing
9752 ``guard`` with the stack slot created by ``llvm.stackprotector`` and if the
9753 values do not match call the ``__stack_chk_fail()`` function.
9755 The reason to provide this as an IR level intrinsic instead of implementing it
9756 via other IR operations is that in order to perform this operation at the IR
9757 level without an intrinsic, one would need to create additional basic blocks to
9758 handle the success/failure cases. This makes it difficult to stop the stack
9759 protector check from disrupting sibling tail calls in Codegen. With this
9760 intrinsic, we are able to generate the stack protector basic blocks late in
9761 codegen after the tail call decision has occurred.
9763 '``llvm.objectsize``' Intrinsic
9764 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9771 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>)
9772 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>)
9777 The ``llvm.objectsize`` intrinsic is designed to provide information to
9778 the optimizers to determine at compile time whether a) an operation
9779 (like memcpy) will overflow a buffer that corresponds to an object, or
9780 b) that a runtime check for overflow isn't necessary. An object in this
9781 context means an allocation of a specific class, structure, array, or
9787 The ``llvm.objectsize`` intrinsic takes two arguments. The first
9788 argument is a pointer to or into the ``object``. The second argument is
9789 a boolean and determines whether ``llvm.objectsize`` returns 0 (if true)
9790 or -1 (if false) when the object size is unknown. The second argument
9791 only accepts constants.
9796 The ``llvm.objectsize`` intrinsic is lowered to a constant representing
9797 the size of the object concerned. If the size cannot be determined at
9798 compile time, ``llvm.objectsize`` returns ``i32/i64 -1 or 0`` (depending
9799 on the ``min`` argument).
9801 '``llvm.expect``' Intrinsic
9802 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9807 This is an overloaded intrinsic. You can use ``llvm.expect`` on any
9812 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>)
9813 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>)
9814 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>)
9819 The ``llvm.expect`` intrinsic provides information about expected (the
9820 most probable) value of ``val``, which can be used by optimizers.
9825 The ``llvm.expect`` intrinsic takes two arguments. The first argument is
9826 a value. The second argument is an expected value, this needs to be a
9827 constant value, variables are not allowed.
9832 This intrinsic is lowered to the ``val``.
9834 '``llvm.assume``' Intrinsic
9835 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9842 declare void @llvm.assume(i1 %cond)
9847 The ``llvm.assume`` allows the optimizer to assume that the provided
9848 condition is true. This information can then be used in simplifying other parts
9854 The condition which the optimizer may assume is always true.
9859 The intrinsic allows the optimizer to assume that the provided condition is
9860 always true whenever the control flow reaches the intrinsic call. No code is
9861 generated for this intrinsic, and instructions that contribute only to the
9862 provided condition are not used for code generation. If the condition is
9863 violated during execution, the behavior is undefined.
9865 Please note that optimizer might limit the transformations performed on values
9866 used by the ``llvm.assume`` intrinsic in order to preserve the instructions
9867 only used to form the intrinsic's input argument. This might prove undesirable
9868 if the extra information provided by the ``llvm.assume`` intrinsic does cause
9869 sufficient overall improvement in code quality. For this reason,
9870 ``llvm.assume`` should not be used to document basic mathematical invariants
9871 that the optimizer can otherwise deduce or facts that are of little use to the
9874 '``llvm.donothing``' Intrinsic
9875 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9882 declare void @llvm.donothing() nounwind readnone
9887 The ``llvm.donothing`` intrinsic doesn't perform any operation. It's one of only
9888 two intrinsics (besides ``llvm.experimental.patchpoint``) that can be called
9889 with an invoke instruction.
9899 This intrinsic does nothing, and it's removed by optimizers and ignored
9902 Stack Map Intrinsics
9903 --------------------
9905 LLVM provides experimental intrinsics to support runtime patching
9906 mechanisms commonly desired in dynamic language JITs. These intrinsics
9907 are described in :doc:`StackMaps`.