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], [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 dynamically
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
357 unintrusive as possible. This convention behaves identically 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 use 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 variable 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 aliases 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>`,
639 an optional :ref:`personality <personalityfn>`,
640 an opening curly brace, a list of basic blocks, and a closing curly brace.
642 LLVM function declarations consist of the "``declare``" keyword, an
643 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
644 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
645 an optional :ref:`calling convention <callingconv>`,
646 an optional ``unnamed_addr`` attribute, a return type, an optional
647 :ref:`parameter attribute <paramattrs>` for the return type, a function
648 name, a possibly empty list of arguments, an optional alignment, an optional
649 :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`,
650 and an optional :ref:`prologue <prologuedata>`.
652 A function definition contains a list of basic blocks, forming the CFG (Control
653 Flow Graph) for the function. Each basic block may optionally start with a label
654 (giving the basic block a symbol table entry), contains a list of instructions,
655 and ends with a :ref:`terminator <terminators>` instruction (such as a branch or
656 function return). If an explicit label is not provided, a block is assigned an
657 implicit numbered label, using the next value from the same counter as used for
658 unnamed temporaries (:ref:`see above<identifiers>`). For example, if a function
659 entry block does not have an explicit label, it will be assigned label "%0",
660 then the first unnamed temporary in that block will be "%1", etc.
662 The first basic block in a function is special in two ways: it is
663 immediately executed on entrance to the function, and it is not allowed
664 to have predecessor basic blocks (i.e. there can not be any branches to
665 the entry block of a function). Because the block can have no
666 predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`.
668 LLVM allows an explicit section to be specified for functions. If the
669 target supports it, it will emit functions to the section specified.
670 Additionally, the function can be placed in a COMDAT.
672 An explicit alignment may be specified for a function. If not present,
673 or if the alignment is set to zero, the alignment of the function is set
674 by the target to whatever it feels convenient. If an explicit alignment
675 is specified, the function is forced to have at least that much
676 alignment. All alignments must be a power of 2.
678 If the ``unnamed_addr`` attribute is given, the address is known to not
679 be significant and two identical functions can be merged.
683 define [linkage] [visibility] [DLLStorageClass]
685 <ResultType> @<FunctionName> ([argument list])
686 [unnamed_addr] [fn Attrs] [section "name"] [comdat [($name)]]
687 [align N] [gc] [prefix Constant] [prologue Constant]
688 [personality Constant] { ... }
690 The argument list is a comma seperated sequence of arguments where each
691 argument is of the following form
695 <type> [parameter Attrs] [name]
703 Aliases, unlike function or variables, don't create any new data. They
704 are just a new symbol and metadata for an existing position.
706 Aliases have a name and an aliasee that is either a global value or a
709 Aliases may have an optional :ref:`linkage type <linkage>`, an optional
710 :ref:`visibility style <visibility>`, an optional :ref:`DLL storage class
711 <dllstorageclass>` and an optional :ref:`tls model <tls_model>`.
715 @<Name> = [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal] [unnamed_addr] alias <AliaseeTy> @<Aliasee>
717 The linkage must be one of ``private``, ``internal``, ``linkonce``, ``weak``,
718 ``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers
719 might not correctly handle dropping a weak symbol that is aliased.
721 Aliases that are not ``unnamed_addr`` are guaranteed to have the same address as
722 the aliasee expression. ``unnamed_addr`` ones are only guaranteed to point
725 Since aliases are only a second name, some restrictions apply, of which
726 some can only be checked when producing an object file:
728 * The expression defining the aliasee must be computable at assembly
729 time. Since it is just a name, no relocations can be used.
731 * No alias in the expression can be weak as the possibility of the
732 intermediate alias being overridden cannot be represented in an
735 * No global value in the expression can be a declaration, since that
736 would require a relocation, which is not possible.
743 Comdat IR provides access to COFF and ELF object file COMDAT functionality.
745 Comdats have a name which represents the COMDAT key. All global objects that
746 specify this key will only end up in the final object file if the linker chooses
747 that key over some other key. Aliases are placed in the same COMDAT that their
748 aliasee computes to, if any.
750 Comdats have a selection kind to provide input on how the linker should
751 choose between keys in two different object files.
755 $<Name> = comdat SelectionKind
757 The selection kind must be one of the following:
760 The linker may choose any COMDAT key, the choice is arbitrary.
762 The linker may choose any COMDAT key but the sections must contain the
765 The linker will choose the section containing the largest COMDAT key.
767 The linker requires that only section with this COMDAT key exist.
769 The linker may choose any COMDAT key but the sections must contain the
772 Note that the Mach-O platform doesn't support COMDATs and ELF only supports
773 ``any`` as a selection kind.
775 Here is an example of a COMDAT group where a function will only be selected if
776 the COMDAT key's section is the largest:
780 $foo = comdat largest
781 @foo = global i32 2, comdat($foo)
783 define void @bar() comdat($foo) {
787 As a syntactic sugar the ``$name`` can be omitted if the name is the same as
793 @foo = global i32 2, comdat
796 In a COFF object file, this will create a COMDAT section with selection kind
797 ``IMAGE_COMDAT_SELECT_LARGEST`` containing the contents of the ``@foo`` symbol
798 and another COMDAT section with selection kind
799 ``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT
800 section and contains the contents of the ``@bar`` symbol.
802 There are some restrictions on the properties of the global object.
803 It, or an alias to it, must have the same name as the COMDAT group when
805 The contents and size of this object may be used during link-time to determine
806 which COMDAT groups get selected depending on the selection kind.
807 Because the name of the object must match the name of the COMDAT group, the
808 linkage of the global object must not be local; local symbols can get renamed
809 if a collision occurs in the symbol table.
811 The combined use of COMDATS and section attributes may yield surprising results.
818 @g1 = global i32 42, section "sec", comdat($foo)
819 @g2 = global i32 42, section "sec", comdat($bar)
821 From the object file perspective, this requires the creation of two sections
822 with the same name. This is necessary because both globals belong to different
823 COMDAT groups and COMDATs, at the object file level, are represented by
826 Note that certain IR constructs like global variables and functions may
827 create COMDATs in the object file in addition to any which are specified using
828 COMDAT IR. This arises when the code generator is configured to emit globals
829 in individual sections (e.g. when `-data-sections` or `-function-sections`
830 is supplied to `llc`).
832 .. _namedmetadatastructure:
837 Named metadata is a collection of metadata. :ref:`Metadata
838 nodes <metadata>` (but not metadata strings) are the only valid
839 operands for a named metadata.
841 #. Named metadata are represented as a string of characters with the
842 metadata prefix. The rules for metadata names are the same as for
843 identifiers, but quoted names are not allowed. ``"\xx"`` type escapes
844 are still valid, which allows any character to be part of a name.
848 ; Some unnamed metadata nodes, which are referenced by the named metadata.
853 !name = !{!0, !1, !2}
860 The return type and each parameter of a function type may have a set of
861 *parameter attributes* associated with them. Parameter attributes are
862 used to communicate additional information about the result or
863 parameters of a function. Parameter attributes are considered to be part
864 of the function, not of the function type, so functions with different
865 parameter attributes can have the same function type.
867 Parameter attributes are simple keywords that follow the type specified.
868 If multiple parameter attributes are needed, they are space separated.
873 declare i32 @printf(i8* noalias nocapture, ...)
874 declare i32 @atoi(i8 zeroext)
875 declare signext i8 @returns_signed_char()
877 Note that any attributes for the function result (``nounwind``,
878 ``readonly``) come immediately after the argument list.
880 Currently, only the following parameter attributes are defined:
883 This indicates to the code generator that the parameter or return
884 value should be zero-extended to the extent required by the target's
885 ABI (which is usually 32-bits, but is 8-bits for a i1 on x86-64) by
886 the caller (for a parameter) or the callee (for a return value).
888 This indicates to the code generator that the parameter or return
889 value should be sign-extended to the extent required by the target's
890 ABI (which is usually 32-bits) by the caller (for a parameter) or
891 the callee (for a return value).
893 This indicates that this parameter or return value should be treated
894 in a special target-dependent fashion during while emitting code for
895 a function call or return (usually, by putting it in a register as
896 opposed to memory, though some targets use it to distinguish between
897 two different kinds of registers). Use of this attribute is
900 This indicates that the pointer parameter should really be passed by
901 value to the function. The attribute implies that a hidden copy of
902 the pointee is made between the caller and the callee, so the callee
903 is unable to modify the value in the caller. This attribute is only
904 valid on LLVM pointer arguments. It is generally used to pass
905 structs and arrays by value, but is also valid on pointers to
906 scalars. The copy is considered to belong to the caller not the
907 callee (for example, ``readonly`` functions should not write to
908 ``byval`` parameters). This is not a valid attribute for return
911 The byval attribute also supports specifying an alignment with the
912 align attribute. It indicates the alignment of the stack slot to
913 form and the known alignment of the pointer specified to the call
914 site. If the alignment is not specified, then the code generator
915 makes a target-specific assumption.
921 The ``inalloca`` argument attribute allows the caller to take the
922 address of outgoing stack arguments. An ``inalloca`` argument must
923 be a pointer to stack memory produced by an ``alloca`` instruction.
924 The alloca, or argument allocation, must also be tagged with the
925 inalloca keyword. Only the last argument may have the ``inalloca``
926 attribute, and that argument is guaranteed to be passed in memory.
928 An argument allocation may be used by a call at most once because
929 the call may deallocate it. The ``inalloca`` attribute cannot be
930 used in conjunction with other attributes that affect argument
931 storage, like ``inreg``, ``nest``, ``sret``, or ``byval``. The
932 ``inalloca`` attribute also disables LLVM's implicit lowering of
933 large aggregate return values, which means that frontend authors
934 must lower them with ``sret`` pointers.
936 When the call site is reached, the argument allocation must have
937 been the most recent stack allocation that is still live, or the
938 results are undefined. It is possible to allocate additional stack
939 space after an argument allocation and before its call site, but it
940 must be cleared off with :ref:`llvm.stackrestore
943 See :doc:`InAlloca` for more information on how to use this
947 This indicates that the pointer parameter specifies the address of a
948 structure that is the return value of the function in the source
949 program. This pointer must be guaranteed by the caller to be valid:
950 loads and stores to the structure may be assumed by the callee
951 not to trap and to be properly aligned. This may only be applied to
952 the first parameter. This is not a valid attribute for return
956 This indicates that the pointer value may be assumed by the optimizer to
957 have the specified alignment.
959 Note that this attribute has additional semantics when combined with the
965 This indicates that objects accessed via pointer values
966 :ref:`based <pointeraliasing>` on the argument or return value are not also
967 accessed, during the execution of the function, via pointer values not
968 *based* on the argument or return value. The attribute on a return value
969 also has additional semantics described below. The caller shares the
970 responsibility with the callee for ensuring that these requirements are met.
971 For further details, please see the discussion of the NoAlias response in
972 :ref:`alias analysis <Must, May, or No>`.
974 Note that this definition of ``noalias`` is intentionally similar
975 to the definition of ``restrict`` in C99 for function arguments.
977 For function return values, C99's ``restrict`` is not meaningful,
978 while LLVM's ``noalias`` is. Furthermore, the semantics of the ``noalias``
979 attribute on return values are stronger than the semantics of the attribute
980 when used on function arguments. On function return values, the ``noalias``
981 attribute indicates that the function acts like a system memory allocation
982 function, returning a pointer to allocated storage disjoint from the
983 storage for any other object accessible to the caller.
986 This indicates that the callee does not make any copies of the
987 pointer that outlive the callee itself. This is not a valid
988 attribute for return values.
993 This indicates that the pointer parameter can be excised using the
994 :ref:`trampoline intrinsics <int_trampoline>`. This is not a valid
995 attribute for return values and can only be applied to one parameter.
998 This indicates that the function always returns the argument as its return
999 value. This is an optimization hint to the code generator when generating
1000 the caller, allowing tail call optimization and omission of register saves
1001 and restores in some cases; it is not checked or enforced when generating
1002 the callee. The parameter and the function return type must be valid
1003 operands for the :ref:`bitcast instruction <i_bitcast>`. This is not a
1004 valid attribute for return values and can only be applied to one parameter.
1007 This indicates that the parameter or return pointer is not null. This
1008 attribute may only be applied to pointer typed parameters. This is not
1009 checked or enforced by LLVM, the caller must ensure that the pointer
1010 passed in is non-null, or the callee must ensure that the returned pointer
1013 ``dereferenceable(<n>)``
1014 This indicates that the parameter or return pointer is dereferenceable. This
1015 attribute may only be applied to pointer typed parameters. A pointer that
1016 is dereferenceable can be loaded from speculatively without a risk of
1017 trapping. The number of bytes known to be dereferenceable must be provided
1018 in parentheses. It is legal for the number of bytes to be less than the
1019 size of the pointee type. The ``nonnull`` attribute does not imply
1020 dereferenceability (consider a pointer to one element past the end of an
1021 array), however ``dereferenceable(<n>)`` does imply ``nonnull`` in
1022 ``addrspace(0)`` (which is the default address space).
1024 ``dereferenceable_or_null(<n>)``
1025 This indicates that the parameter or return value isn't both
1026 non-null and non-dereferenceable (up to ``<n>`` bytes) at the same
1027 time. All non-null pointers tagged with
1028 ``dereferenceable_or_null(<n>)`` are ``dereferenceable(<n>)``.
1029 For address space 0 ``dereferenceable_or_null(<n>)`` implies that
1030 a pointer is exactly one of ``dereferenceable(<n>)`` or ``null``,
1031 and in other address spaces ``dereferenceable_or_null(<n>)``
1032 implies that a pointer is at least one of ``dereferenceable(<n>)``
1033 or ``null`` (i.e. it may be both ``null`` and
1034 ``dereferenceable(<n>)``). This attribute may only be applied to
1035 pointer typed parameters.
1039 Garbage Collector Strategy Names
1040 --------------------------------
1042 Each function may specify a garbage collector strategy name, which is simply a
1045 .. code-block:: llvm
1047 define void @f() gc "name" { ... }
1049 The supported values of *name* includes those :ref:`built in to LLVM
1050 <builtin-gc-strategies>` and any provided by loaded plugins. Specifying a GC
1051 strategy will cause the compiler to alter its output in order to support the
1052 named garbage collection algorithm. Note that LLVM itself does not contain a
1053 garbage collector, this functionality is restricted to generating machine code
1054 which can interoperate with a collector provided externally.
1061 Prefix data is data associated with a function which the code
1062 generator will emit immediately before the function's entrypoint.
1063 The purpose of this feature is to allow frontends to associate
1064 language-specific runtime metadata with specific functions and make it
1065 available through the function pointer while still allowing the
1066 function pointer to be called.
1068 To access the data for a given function, a program may bitcast the
1069 function pointer to a pointer to the constant's type and dereference
1070 index -1. This implies that the IR symbol points just past the end of
1071 the prefix data. For instance, take the example of a function annotated
1072 with a single ``i32``,
1074 .. code-block:: llvm
1076 define void @f() prefix i32 123 { ... }
1078 The prefix data can be referenced as,
1080 .. code-block:: llvm
1082 %0 = bitcast void* () @f to i32*
1083 %a = getelementptr inbounds i32, i32* %0, i32 -1
1084 %b = load i32, i32* %a
1086 Prefix data is laid out as if it were an initializer for a global variable
1087 of the prefix data's type. The function will be placed such that the
1088 beginning of the prefix data is aligned. This means that if the size
1089 of the prefix data is not a multiple of the alignment size, the
1090 function's entrypoint will not be aligned. If alignment of the
1091 function's entrypoint is desired, padding must be added to the prefix
1094 A function may have prefix data but no body. This has similar semantics
1095 to the ``available_externally`` linkage in that the data may be used by the
1096 optimizers but will not be emitted in the object file.
1103 The ``prologue`` attribute allows arbitrary code (encoded as bytes) to
1104 be inserted prior to the function body. This can be used for enabling
1105 function hot-patching and instrumentation.
1107 To maintain the semantics of ordinary function calls, the prologue data must
1108 have a particular format. Specifically, it must begin with a sequence of
1109 bytes which decode to a sequence of machine instructions, valid for the
1110 module's target, which transfer control to the point immediately succeeding
1111 the prologue data, without performing any other visible action. This allows
1112 the inliner and other passes to reason about the semantics of the function
1113 definition without needing to reason about the prologue data. Obviously this
1114 makes the format of the prologue data highly target dependent.
1116 A trivial example of valid prologue data for the x86 architecture is ``i8 144``,
1117 which encodes the ``nop`` instruction:
1119 .. code-block:: llvm
1121 define void @f() prologue i8 144 { ... }
1123 Generally prologue data can be formed by encoding a relative branch instruction
1124 which skips the metadata, as in this example of valid prologue data for the
1125 x86_64 architecture, where the first two bytes encode ``jmp .+10``:
1127 .. code-block:: llvm
1129 %0 = type <{ i8, i8, i8* }>
1131 define void @f() prologue %0 <{ i8 235, i8 8, i8* @md}> { ... }
1133 A function may have prologue data but no body. This has similar semantics
1134 to the ``available_externally`` linkage in that the data may be used by the
1135 optimizers but will not be emitted in the object file.
1139 Personality Function
1140 --------------------
1142 The ``personality`` attribute permits functions to specify what function
1143 to use for exception handling.
1150 Attribute groups are groups of attributes that are referenced by objects within
1151 the IR. They are important for keeping ``.ll`` files readable, because a lot of
1152 functions will use the same set of attributes. In the degenerative case of a
1153 ``.ll`` file that corresponds to a single ``.c`` file, the single attribute
1154 group will capture the important command line flags used to build that file.
1156 An attribute group is a module-level object. To use an attribute group, an
1157 object references the attribute group's ID (e.g. ``#37``). An object may refer
1158 to more than one attribute group. In that situation, the attributes from the
1159 different groups are merged.
1161 Here is an example of attribute groups for a function that should always be
1162 inlined, has a stack alignment of 4, and which shouldn't use SSE instructions:
1164 .. code-block:: llvm
1166 ; Target-independent attributes:
1167 attributes #0 = { alwaysinline alignstack=4 }
1169 ; Target-dependent attributes:
1170 attributes #1 = { "no-sse" }
1172 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse".
1173 define void @f() #0 #1 { ... }
1180 Function attributes are set to communicate additional information about
1181 a function. Function attributes are considered to be part of the
1182 function, not of the function type, so functions with different function
1183 attributes can have the same function type.
1185 Function attributes are simple keywords that follow the type specified.
1186 If multiple attributes are needed, they are space separated. For
1189 .. code-block:: llvm
1191 define void @f() noinline { ... }
1192 define void @f() alwaysinline { ... }
1193 define void @f() alwaysinline optsize { ... }
1194 define void @f() optsize { ... }
1197 This attribute indicates that, when emitting the prologue and
1198 epilogue, the backend should forcibly align the stack pointer.
1199 Specify the desired alignment, which must be a power of two, in
1202 This attribute indicates that the inliner should attempt to inline
1203 this function into callers whenever possible, ignoring any active
1204 inlining size threshold for this caller.
1206 This indicates that the callee function at a call site should be
1207 recognized as a built-in function, even though the function's declaration
1208 uses the ``nobuiltin`` attribute. This is only valid at call sites for
1209 direct calls to functions that are declared with the ``nobuiltin``
1212 This attribute indicates that this function is rarely called. When
1213 computing edge weights, basic blocks post-dominated by a cold
1214 function call are also considered to be cold; and, thus, given low
1217 This attribute indicates that the callee is dependent on a convergent
1218 thread execution pattern under certain parallel execution models.
1219 Transformations that are execution model agnostic may only move or
1220 tranform this call if the final location is control equivalent to its
1221 original position in the program, where control equivalence is defined as
1222 A dominates B and B post-dominates A, or vice versa.
1224 This attribute indicates that the source code contained a hint that
1225 inlining this function is desirable (such as the "inline" keyword in
1226 C/C++). It is just a hint; it imposes no requirements on the
1229 This attribute indicates that the function should be added to a
1230 jump-instruction table at code-generation time, and that all address-taken
1231 references to this function should be replaced with a reference to the
1232 appropriate jump-instruction-table function pointer. Note that this creates
1233 a new pointer for the original function, which means that code that depends
1234 on function-pointer identity can break. So, any function annotated with
1235 ``jumptable`` must also be ``unnamed_addr``.
1237 This attribute suggests that optimization passes and code generator
1238 passes make choices that keep the code size of this function as small
1239 as possible and perform optimizations that may sacrifice runtime
1240 performance in order to minimize the size of the generated code.
1242 This attribute disables prologue / epilogue emission for the
1243 function. This can have very system-specific consequences.
1245 This indicates that the callee function at a call site is not recognized as
1246 a built-in function. LLVM will retain the original call and not replace it
1247 with equivalent code based on the semantics of the built-in function, unless
1248 the call site uses the ``builtin`` attribute. This is valid at call sites
1249 and on function declarations and definitions.
1251 This attribute indicates that calls to the function cannot be
1252 duplicated. A call to a ``noduplicate`` function may be moved
1253 within its parent function, but may not be duplicated within
1254 its parent function.
1256 A function containing a ``noduplicate`` call may still
1257 be an inlining candidate, provided that the call is not
1258 duplicated by inlining. That implies that the function has
1259 internal linkage and only has one call site, so the original
1260 call is dead after inlining.
1262 This attributes disables implicit floating point instructions.
1264 This attribute indicates that the inliner should never inline this
1265 function in any situation. This attribute may not be used together
1266 with the ``alwaysinline`` attribute.
1268 This attribute suppresses lazy symbol binding for the function. This
1269 may make calls to the function faster, at the cost of extra program
1270 startup time if the function is not called during program startup.
1272 This attribute indicates that the code generator should not use a
1273 red zone, even if the target-specific ABI normally permits it.
1275 This function attribute indicates that the function never returns
1276 normally. This produces undefined behavior at runtime if the
1277 function ever does dynamically return.
1279 This function attribute indicates that the function never raises an
1280 exception. If the function does raise an exception, its runtime
1281 behavior is undefined. However, functions marked nounwind may still
1282 trap or generate asynchronous exceptions. Exception handling schemes
1283 that are recognized by LLVM to handle asynchronous exceptions, such
1284 as SEH, will still provide their implementation defined semantics.
1286 This function attribute indicates that the function is not optimized
1287 by any optimization or code generator passes with the
1288 exception of interprocedural optimization passes.
1289 This attribute cannot be used together with the ``alwaysinline``
1290 attribute; this attribute is also incompatible
1291 with the ``minsize`` attribute and the ``optsize`` attribute.
1293 This attribute requires the ``noinline`` attribute to be specified on
1294 the function as well, so the function is never inlined into any caller.
1295 Only functions with the ``alwaysinline`` attribute are valid
1296 candidates for inlining into the body of this function.
1298 This attribute suggests that optimization passes and code generator
1299 passes make choices that keep the code size of this function low,
1300 and otherwise do optimizations specifically to reduce code size as
1301 long as they do not significantly impact runtime performance.
1303 On a function, this attribute indicates that the function computes its
1304 result (or decides to unwind an exception) based strictly on its arguments,
1305 without dereferencing any pointer arguments or otherwise accessing
1306 any mutable state (e.g. memory, control registers, etc) visible to
1307 caller functions. It does not write through any pointer arguments
1308 (including ``byval`` arguments) and never changes any state visible
1309 to callers. This means that it cannot unwind exceptions by calling
1310 the ``C++`` exception throwing methods.
1312 On an argument, this attribute indicates that the function does not
1313 dereference that pointer argument, even though it may read or write the
1314 memory that the pointer points to if accessed through other pointers.
1316 On a function, this attribute indicates that the function does not write
1317 through any pointer arguments (including ``byval`` arguments) or otherwise
1318 modify any state (e.g. memory, control registers, etc) visible to
1319 caller functions. It may dereference pointer arguments and read
1320 state that may be set in the caller. A readonly function always
1321 returns the same value (or unwinds an exception identically) when
1322 called with the same set of arguments and global state. It cannot
1323 unwind an exception by calling the ``C++`` exception throwing
1326 On an argument, this attribute indicates that the function does not write
1327 through this pointer argument, even though it may write to the memory that
1328 the pointer points to.
1330 This attribute indicates that the only memory accesses inside function are
1331 loads and stores from objects pointed to by its pointer-typed arguments,
1332 with arbitrary offsets. Or in other words, all memory operations in the
1333 function can refer to memory only using pointers based on its function
1335 Note that ``argmemonly`` can be used together with ``readonly`` attribute
1336 in order to specify that function reads only from its arguments.
1338 This attribute indicates that this function can return twice. The C
1339 ``setjmp`` is an example of such a function. The compiler disables
1340 some optimizations (like tail calls) in the caller of these
1343 This attribute indicates that
1344 `SafeStack <http://clang.llvm.org/docs/SafeStack.html>`_
1345 protection is enabled for this function.
1347 If a function that has a ``safestack`` attribute is inlined into a
1348 function that doesn't have a ``safestack`` attribute or which has an
1349 ``ssp``, ``sspstrong`` or ``sspreq`` attribute, then the resulting
1350 function will have a ``safestack`` attribute.
1351 ``sanitize_address``
1352 This attribute indicates that AddressSanitizer checks
1353 (dynamic address safety analysis) are enabled for this function.
1355 This attribute indicates that MemorySanitizer checks (dynamic detection
1356 of accesses to uninitialized memory) are enabled for this function.
1358 This attribute indicates that ThreadSanitizer checks
1359 (dynamic thread safety analysis) are enabled for this function.
1361 This attribute indicates that the function should emit a stack
1362 smashing protector. It is in the form of a "canary" --- a random value
1363 placed on the stack before the local variables that's checked upon
1364 return from the function to see if it has been overwritten. A
1365 heuristic is used to determine if a function needs stack protectors
1366 or not. The heuristic used will enable protectors for functions with:
1368 - Character arrays larger than ``ssp-buffer-size`` (default 8).
1369 - Aggregates containing character arrays larger than ``ssp-buffer-size``.
1370 - Calls to alloca() with variable sizes or constant sizes greater than
1371 ``ssp-buffer-size``.
1373 Variables that are identified as requiring a protector will be arranged
1374 on the stack such that they are adjacent to the stack protector guard.
1376 If a function that has an ``ssp`` attribute is inlined into a
1377 function that doesn't have an ``ssp`` attribute, then the resulting
1378 function will have an ``ssp`` attribute.
1380 This attribute indicates that the function should *always* emit a
1381 stack smashing protector. This overrides the ``ssp`` function
1384 Variables that are identified as requiring a protector will be arranged
1385 on the stack such that they are adjacent to the stack protector guard.
1386 The specific layout rules are:
1388 #. Large arrays and structures containing large arrays
1389 (``>= ssp-buffer-size``) are closest to the stack protector.
1390 #. Small arrays and structures containing small arrays
1391 (``< ssp-buffer-size``) are 2nd closest to the protector.
1392 #. Variables that have had their address taken are 3rd closest to the
1395 If a function that has an ``sspreq`` attribute is inlined into a
1396 function that doesn't have an ``sspreq`` attribute or which has an
1397 ``ssp`` or ``sspstrong`` attribute, then the resulting function will have
1398 an ``sspreq`` attribute.
1400 This attribute indicates that the function should emit a stack smashing
1401 protector. This attribute causes a strong heuristic to be used when
1402 determining if a function needs stack protectors. The strong heuristic
1403 will enable protectors for functions with:
1405 - Arrays of any size and type
1406 - Aggregates containing an array of any size and type.
1407 - Calls to alloca().
1408 - Local variables that have had their address taken.
1410 Variables that are identified as requiring a protector will be arranged
1411 on the stack such that they are adjacent to the stack protector guard.
1412 The specific layout rules are:
1414 #. Large arrays and structures containing large arrays
1415 (``>= ssp-buffer-size``) are closest to the stack protector.
1416 #. Small arrays and structures containing small arrays
1417 (``< ssp-buffer-size``) are 2nd closest to the protector.
1418 #. Variables that have had their address taken are 3rd closest to the
1421 This overrides the ``ssp`` function attribute.
1423 If a function that has an ``sspstrong`` attribute is inlined into a
1424 function that doesn't have an ``sspstrong`` attribute, then the
1425 resulting function will have an ``sspstrong`` attribute.
1427 This attribute indicates that the function will delegate to some other
1428 function with a tail call. The prototype of a thunk should not be used for
1429 optimization purposes. The caller is expected to cast the thunk prototype to
1430 match the thunk target prototype.
1432 This attribute indicates that the ABI being targeted requires that
1433 an unwind table entry be produce for this function even if we can
1434 show that no exceptions passes by it. This is normally the case for
1435 the ELF x86-64 abi, but it can be disabled for some compilation
1440 Module-Level Inline Assembly
1441 ----------------------------
1443 Modules may contain "module-level inline asm" blocks, which corresponds
1444 to the GCC "file scope inline asm" blocks. These blocks are internally
1445 concatenated by LLVM and treated as a single unit, but may be separated
1446 in the ``.ll`` file if desired. The syntax is very simple:
1448 .. code-block:: llvm
1450 module asm "inline asm code goes here"
1451 module asm "more can go here"
1453 The strings can contain any character by escaping non-printable
1454 characters. The escape sequence used is simply "\\xx" where "xx" is the
1455 two digit hex code for the number.
1457 Note that the assembly string *must* be parseable by LLVM's integrated assembler
1458 (unless it is disabled), even when emitting a ``.s`` file.
1460 .. _langref_datalayout:
1465 A module may specify a target specific data layout string that specifies
1466 how data is to be laid out in memory. The syntax for the data layout is
1469 .. code-block:: llvm
1471 target datalayout = "layout specification"
1473 The *layout specification* consists of a list of specifications
1474 separated by the minus sign character ('-'). Each specification starts
1475 with a letter and may include other information after the letter to
1476 define some aspect of the data layout. The specifications accepted are
1480 Specifies that the target lays out data in big-endian form. That is,
1481 the bits with the most significance have the lowest address
1484 Specifies that the target lays out data in little-endian form. That
1485 is, the bits with the least significance have the lowest address
1488 Specifies the natural alignment of the stack in bits. Alignment
1489 promotion of stack variables is limited to the natural stack
1490 alignment to avoid dynamic stack realignment. The stack alignment
1491 must be a multiple of 8-bits. If omitted, the natural stack
1492 alignment defaults to "unspecified", which does not prevent any
1493 alignment promotions.
1494 ``p[n]:<size>:<abi>:<pref>``
1495 This specifies the *size* of a pointer and its ``<abi>`` and
1496 ``<pref>``\erred alignments for address space ``n``. All sizes are in
1497 bits. The address space, ``n`` is optional, and if not specified,
1498 denotes the default address space 0. The value of ``n`` must be
1499 in the range [1,2^23).
1500 ``i<size>:<abi>:<pref>``
1501 This specifies the alignment for an integer type of a given bit
1502 ``<size>``. The value of ``<size>`` must be in the range [1,2^23).
1503 ``v<size>:<abi>:<pref>``
1504 This specifies the alignment for a vector type of a given bit
1506 ``f<size>:<abi>:<pref>``
1507 This specifies the alignment for a floating point type of a given bit
1508 ``<size>``. Only values of ``<size>`` that are supported by the target
1509 will work. 32 (float) and 64 (double) are supported on all targets; 80
1510 or 128 (different flavors of long double) are also supported on some
1513 This specifies the alignment for an object of aggregate type.
1515 If present, specifies that llvm names are mangled in the output. The
1518 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix.
1519 * ``m``: Mips mangling: Private symbols get a ``$`` prefix.
1520 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other
1521 symbols get a ``_`` prefix.
1522 * ``w``: Windows COFF prefix: Similar to Mach-O, but stdcall and fastcall
1523 functions also get a suffix based on the frame size.
1524 ``n<size1>:<size2>:<size3>...``
1525 This specifies a set of native integer widths for the target CPU in
1526 bits. For example, it might contain ``n32`` for 32-bit PowerPC,
1527 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of
1528 this set are considered to support most general arithmetic operations
1531 On every specification that takes a ``<abi>:<pref>``, specifying the
1532 ``<pref>`` alignment is optional. If omitted, the preceding ``:``
1533 should be omitted too and ``<pref>`` will be equal to ``<abi>``.
1535 When constructing the data layout for a given target, LLVM starts with a
1536 default set of specifications which are then (possibly) overridden by
1537 the specifications in the ``datalayout`` keyword. The default
1538 specifications are given in this list:
1540 - ``E`` - big endian
1541 - ``p:64:64:64`` - 64-bit pointers with 64-bit alignment.
1542 - ``p[n]:64:64:64`` - Other address spaces are assumed to be the
1543 same as the default address space.
1544 - ``S0`` - natural stack alignment is unspecified
1545 - ``i1:8:8`` - i1 is 8-bit (byte) aligned
1546 - ``i8:8:8`` - i8 is 8-bit (byte) aligned
1547 - ``i16:16:16`` - i16 is 16-bit aligned
1548 - ``i32:32:32`` - i32 is 32-bit aligned
1549 - ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred
1550 alignment of 64-bits
1551 - ``f16:16:16`` - half is 16-bit aligned
1552 - ``f32:32:32`` - float is 32-bit aligned
1553 - ``f64:64:64`` - double is 64-bit aligned
1554 - ``f128:128:128`` - quad is 128-bit aligned
1555 - ``v64:64:64`` - 64-bit vector is 64-bit aligned
1556 - ``v128:128:128`` - 128-bit vector is 128-bit aligned
1557 - ``a:0:64`` - aggregates are 64-bit aligned
1559 When LLVM is determining the alignment for a given type, it uses the
1562 #. If the type sought is an exact match for one of the specifications,
1563 that specification is used.
1564 #. If no match is found, and the type sought is an integer type, then
1565 the smallest integer type that is larger than the bitwidth of the
1566 sought type is used. If none of the specifications are larger than
1567 the bitwidth then the largest integer type is used. For example,
1568 given the default specifications above, the i7 type will use the
1569 alignment of i8 (next largest) while both i65 and i256 will use the
1570 alignment of i64 (largest specified).
1571 #. If no match is found, and the type sought is a vector type, then the
1572 largest vector type that is smaller than the sought vector type will
1573 be used as a fall back. This happens because <128 x double> can be
1574 implemented in terms of 64 <2 x double>, for example.
1576 The function of the data layout string may not be what you expect.
1577 Notably, this is not a specification from the frontend of what alignment
1578 the code generator should use.
1580 Instead, if specified, the target data layout is required to match what
1581 the ultimate *code generator* expects. This string is used by the
1582 mid-level optimizers to improve code, and this only works if it matches
1583 what the ultimate code generator uses. There is no way to generate IR
1584 that does not embed this target-specific detail into the IR. If you
1585 don't specify the string, the default specifications will be used to
1586 generate a Data Layout and the optimization phases will operate
1587 accordingly and introduce target specificity into the IR with respect to
1588 these default specifications.
1595 A module may specify a target triple string that describes the target
1596 host. The syntax for the target triple is simply:
1598 .. code-block:: llvm
1600 target triple = "x86_64-apple-macosx10.7.0"
1602 The *target triple* string consists of a series of identifiers delimited
1603 by the minus sign character ('-'). The canonical forms are:
1607 ARCHITECTURE-VENDOR-OPERATING_SYSTEM
1608 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT
1610 This information is passed along to the backend so that it generates
1611 code for the proper architecture. It's possible to override this on the
1612 command line with the ``-mtriple`` command line option.
1614 .. _pointeraliasing:
1616 Pointer Aliasing Rules
1617 ----------------------
1619 Any memory access must be done through a pointer value associated with
1620 an address range of the memory access, otherwise the behavior is
1621 undefined. Pointer values are associated with address ranges according
1622 to the following rules:
1624 - A pointer value is associated with the addresses associated with any
1625 value it is *based* on.
1626 - An address of a global variable is associated with the address range
1627 of the variable's storage.
1628 - The result value of an allocation instruction is associated with the
1629 address range of the allocated storage.
1630 - A null pointer in the default address-space is associated with no
1632 - An integer constant other than zero or a pointer value returned from
1633 a function not defined within LLVM may be associated with address
1634 ranges allocated through mechanisms other than those provided by
1635 LLVM. Such ranges shall not overlap with any ranges of addresses
1636 allocated by mechanisms provided by LLVM.
1638 A pointer value is *based* on another pointer value according to the
1641 - A pointer value formed from a ``getelementptr`` operation is *based*
1642 on the first value operand of the ``getelementptr``.
1643 - The result value of a ``bitcast`` is *based* on the operand of the
1645 - A pointer value formed by an ``inttoptr`` is *based* on all pointer
1646 values that contribute (directly or indirectly) to the computation of
1647 the pointer's value.
1648 - The "*based* on" relationship is transitive.
1650 Note that this definition of *"based"* is intentionally similar to the
1651 definition of *"based"* in C99, though it is slightly weaker.
1653 LLVM IR does not associate types with memory. The result type of a
1654 ``load`` merely indicates the size and alignment of the memory from
1655 which to load, as well as the interpretation of the value. The first
1656 operand type of a ``store`` similarly only indicates the size and
1657 alignment of the store.
1659 Consequently, type-based alias analysis, aka TBAA, aka
1660 ``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR.
1661 :ref:`Metadata <metadata>` may be used to encode additional information
1662 which specialized optimization passes may use to implement type-based
1667 Volatile Memory Accesses
1668 ------------------------
1670 Certain memory accesses, such as :ref:`load <i_load>`'s,
1671 :ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be
1672 marked ``volatile``. The optimizers must not change the number of
1673 volatile operations or change their order of execution relative to other
1674 volatile operations. The optimizers *may* change the order of volatile
1675 operations relative to non-volatile operations. This is not Java's
1676 "volatile" and has no cross-thread synchronization behavior.
1678 IR-level volatile loads and stores cannot safely be optimized into
1679 llvm.memcpy or llvm.memmove intrinsics even when those intrinsics are
1680 flagged volatile. Likewise, the backend should never split or merge
1681 target-legal volatile load/store instructions.
1683 .. admonition:: Rationale
1685 Platforms may rely on volatile loads and stores of natively supported
1686 data width to be executed as single instruction. For example, in C
1687 this holds for an l-value of volatile primitive type with native
1688 hardware support, but not necessarily for aggregate types. The
1689 frontend upholds these expectations, which are intentionally
1690 unspecified in the IR. The rules above ensure that IR transformation
1691 do not violate the frontend's contract with the language.
1695 Memory Model for Concurrent Operations
1696 --------------------------------------
1698 The LLVM IR does not define any way to start parallel threads of
1699 execution or to register signal handlers. Nonetheless, there are
1700 platform-specific ways to create them, and we define LLVM IR's behavior
1701 in their presence. This model is inspired by the C++0x memory model.
1703 For a more informal introduction to this model, see the :doc:`Atomics`.
1705 We define a *happens-before* partial order as the least partial order
1708 - Is a superset of single-thread program order, and
1709 - When a *synchronizes-with* ``b``, includes an edge from ``a`` to
1710 ``b``. *Synchronizes-with* pairs are introduced by platform-specific
1711 techniques, like pthread locks, thread creation, thread joining,
1712 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering
1713 Constraints <ordering>`).
1715 Note that program order does not introduce *happens-before* edges
1716 between a thread and signals executing inside that thread.
1718 Every (defined) read operation (load instructions, memcpy, atomic
1719 loads/read-modify-writes, etc.) R reads a series of bytes written by
1720 (defined) write operations (store instructions, atomic
1721 stores/read-modify-writes, memcpy, etc.). For the purposes of this
1722 section, initialized globals are considered to have a write of the
1723 initializer which is atomic and happens before any other read or write
1724 of the memory in question. For each byte of a read R, R\ :sub:`byte`
1725 may see any write to the same byte, except:
1727 - If write\ :sub:`1` happens before write\ :sub:`2`, and
1728 write\ :sub:`2` happens before R\ :sub:`byte`, then
1729 R\ :sub:`byte` does not see write\ :sub:`1`.
1730 - If R\ :sub:`byte` happens before write\ :sub:`3`, then
1731 R\ :sub:`byte` does not see write\ :sub:`3`.
1733 Given that definition, R\ :sub:`byte` is defined as follows:
1735 - If R is volatile, the result is target-dependent. (Volatile is
1736 supposed to give guarantees which can support ``sig_atomic_t`` in
1737 C/C++, and may be used for accesses to addresses that do not behave
1738 like normal memory. It does not generally provide cross-thread
1740 - Otherwise, if there is no write to the same byte that happens before
1741 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte.
1742 - Otherwise, if R\ :sub:`byte` may see exactly one write,
1743 R\ :sub:`byte` returns the value written by that write.
1744 - Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may
1745 see are atomic, it chooses one of the values written. See the :ref:`Atomic
1746 Memory Ordering Constraints <ordering>` section for additional
1747 constraints on how the choice is made.
1748 - Otherwise R\ :sub:`byte` returns ``undef``.
1750 R returns the value composed of the series of bytes it read. This
1751 implies that some bytes within the value may be ``undef`` **without**
1752 the entire value being ``undef``. Note that this only defines the
1753 semantics of the operation; it doesn't mean that targets will emit more
1754 than one instruction to read the series of bytes.
1756 Note that in cases where none of the atomic intrinsics are used, this
1757 model places only one restriction on IR transformations on top of what
1758 is required for single-threaded execution: introducing a store to a byte
1759 which might not otherwise be stored is not allowed in general.
1760 (Specifically, in the case where another thread might write to and read
1761 from an address, introducing a store can change a load that may see
1762 exactly one write into a load that may see multiple writes.)
1766 Atomic Memory Ordering Constraints
1767 ----------------------------------
1769 Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`,
1770 :ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`,
1771 :ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take
1772 ordering parameters that determine which other atomic instructions on
1773 the same address they *synchronize with*. These semantics are borrowed
1774 from Java and C++0x, but are somewhat more colloquial. If these
1775 descriptions aren't precise enough, check those specs (see spec
1776 references in the :doc:`atomics guide <Atomics>`).
1777 :ref:`fence <i_fence>` instructions treat these orderings somewhat
1778 differently since they don't take an address. See that instruction's
1779 documentation for details.
1781 For a simpler introduction to the ordering constraints, see the
1785 The set of values that can be read is governed by the happens-before
1786 partial order. A value cannot be read unless some operation wrote
1787 it. This is intended to provide a guarantee strong enough to model
1788 Java's non-volatile shared variables. This ordering cannot be
1789 specified for read-modify-write operations; it is not strong enough
1790 to make them atomic in any interesting way.
1792 In addition to the guarantees of ``unordered``, there is a single
1793 total order for modifications by ``monotonic`` operations on each
1794 address. All modification orders must be compatible with the
1795 happens-before order. There is no guarantee that the modification
1796 orders can be combined to a global total order for the whole program
1797 (and this often will not be possible). The read in an atomic
1798 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and
1799 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification
1800 order immediately before the value it writes. If one atomic read
1801 happens before another atomic read of the same address, the later
1802 read must see the same value or a later value in the address's
1803 modification order. This disallows reordering of ``monotonic`` (or
1804 stronger) operations on the same address. If an address is written
1805 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally
1806 read that address repeatedly, the other threads must eventually see
1807 the write. This corresponds to the C++0x/C1x
1808 ``memory_order_relaxed``.
1810 In addition to the guarantees of ``monotonic``, a
1811 *synchronizes-with* edge may be formed with a ``release`` operation.
1812 This is intended to model C++'s ``memory_order_acquire``.
1814 In addition to the guarantees of ``monotonic``, if this operation
1815 writes a value which is subsequently read by an ``acquire``
1816 operation, it *synchronizes-with* that operation. (This isn't a
1817 complete description; see the C++0x definition of a release
1818 sequence.) This corresponds to the C++0x/C1x
1819 ``memory_order_release``.
1820 ``acq_rel`` (acquire+release)
1821 Acts as both an ``acquire`` and ``release`` operation on its
1822 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``.
1823 ``seq_cst`` (sequentially consistent)
1824 In addition to the guarantees of ``acq_rel`` (``acquire`` for an
1825 operation that only reads, ``release`` for an operation that only
1826 writes), there is a global total order on all
1827 sequentially-consistent operations on all addresses, which is
1828 consistent with the *happens-before* partial order and with the
1829 modification orders of all the affected addresses. Each
1830 sequentially-consistent read sees the last preceding write to the
1831 same address in this global order. This corresponds to the C++0x/C1x
1832 ``memory_order_seq_cst`` and Java volatile.
1836 If an atomic operation is marked ``singlethread``, it only *synchronizes
1837 with* or participates in modification and seq\_cst total orderings with
1838 other operations running in the same thread (for example, in signal
1846 LLVM IR floating-point binary ops (:ref:`fadd <i_fadd>`,
1847 :ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`,
1848 :ref:`frem <i_frem>`, :ref:`fcmp <i_fcmp>`) have the following flags that can
1849 be set to enable otherwise unsafe floating point operations
1852 No NaNs - Allow optimizations to assume the arguments and result are not
1853 NaN. Such optimizations are required to retain defined behavior over
1854 NaNs, but the value of the result is undefined.
1857 No Infs - Allow optimizations to assume the arguments and result are not
1858 +/-Inf. Such optimizations are required to retain defined behavior over
1859 +/-Inf, but the value of the result is undefined.
1862 No Signed Zeros - Allow optimizations to treat the sign of a zero
1863 argument or result as insignificant.
1866 Allow Reciprocal - Allow optimizations to use the reciprocal of an
1867 argument rather than perform division.
1870 Fast - Allow algebraically equivalent transformations that may
1871 dramatically change results in floating point (e.g. reassociate). This
1872 flag implies all the others.
1876 Use-list Order Directives
1877 -------------------------
1879 Use-list directives encode the in-memory order of each use-list, allowing the
1880 order to be recreated. ``<order-indexes>`` is a comma-separated list of
1881 indexes that are assigned to the referenced value's uses. The referenced
1882 value's use-list is immediately sorted by these indexes.
1884 Use-list directives may appear at function scope or global scope. They are not
1885 instructions, and have no effect on the semantics of the IR. When they're at
1886 function scope, they must appear after the terminator of the final basic block.
1888 If basic blocks have their address taken via ``blockaddress()`` expressions,
1889 ``uselistorder_bb`` can be used to reorder their use-lists from outside their
1896 uselistorder <ty> <value>, { <order-indexes> }
1897 uselistorder_bb @function, %block { <order-indexes> }
1903 define void @foo(i32 %arg1, i32 %arg2) {
1905 ; ... instructions ...
1907 ; ... instructions ...
1909 ; At function scope.
1910 uselistorder i32 %arg1, { 1, 0, 2 }
1911 uselistorder label %bb, { 1, 0 }
1915 uselistorder i32* @global, { 1, 2, 0 }
1916 uselistorder i32 7, { 1, 0 }
1917 uselistorder i32 (i32) @bar, { 1, 0 }
1918 uselistorder_bb @foo, %bb, { 5, 1, 3, 2, 0, 4 }
1925 The LLVM type system is one of the most important features of the
1926 intermediate representation. Being typed enables a number of
1927 optimizations to be performed on the intermediate representation
1928 directly, without having to do extra analyses on the side before the
1929 transformation. A strong type system makes it easier to read the
1930 generated code and enables novel analyses and transformations that are
1931 not feasible to perform on normal three address code representations.
1941 The void type does not represent any value and has no size.
1959 The function type can be thought of as a function signature. It consists of a
1960 return type and a list of formal parameter types. The return type of a function
1961 type is a void type or first class type --- except for :ref:`label <t_label>`
1962 and :ref:`metadata <t_metadata>` types.
1968 <returntype> (<parameter list>)
1970 ...where '``<parameter list>``' is a comma-separated list of type
1971 specifiers. Optionally, the parameter list may include a type ``...``, which
1972 indicates that the function takes a variable number of arguments. Variable
1973 argument functions can access their arguments with the :ref:`variable argument
1974 handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type
1975 except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`.
1979 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1980 | ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` |
1981 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1982 | ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. |
1983 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1984 | ``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. |
1985 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1986 | ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values |
1987 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1994 The :ref:`first class <t_firstclass>` types are perhaps the most important.
1995 Values of these types are the only ones which can be produced by
2003 These are the types that are valid in registers from CodeGen's perspective.
2012 The integer type is a very simple type that simply specifies an
2013 arbitrary bit width for the integer type desired. Any bit width from 1
2014 bit to 2\ :sup:`23`\ -1 (about 8 million) can be specified.
2022 The number of bits the integer will occupy is specified by the ``N``
2028 +----------------+------------------------------------------------+
2029 | ``i1`` | a single-bit integer. |
2030 +----------------+------------------------------------------------+
2031 | ``i32`` | a 32-bit integer. |
2032 +----------------+------------------------------------------------+
2033 | ``i1942652`` | a really big integer of over 1 million bits. |
2034 +----------------+------------------------------------------------+
2038 Floating Point Types
2039 """"""""""""""""""""
2048 - 16-bit floating point value
2051 - 32-bit floating point value
2054 - 64-bit floating point value
2057 - 128-bit floating point value (112-bit mantissa)
2060 - 80-bit floating point value (X87)
2063 - 128-bit floating point value (two 64-bits)
2070 The x86_mmx type represents a value held in an MMX register on an x86
2071 machine. The operations allowed on it are quite limited: parameters and
2072 return values, load and store, and bitcast. User-specified MMX
2073 instructions are represented as intrinsic or asm calls with arguments
2074 and/or results of this type. There are no arrays, vectors or constants
2091 The pointer type is used to specify memory locations. Pointers are
2092 commonly used to reference objects in memory.
2094 Pointer types may have an optional address space attribute defining the
2095 numbered address space where the pointed-to object resides. The default
2096 address space is number zero. The semantics of non-zero address spaces
2097 are target-specific.
2099 Note that LLVM does not permit pointers to void (``void*``) nor does it
2100 permit pointers to labels (``label*``). Use ``i8*`` instead.
2110 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2111 | ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. |
2112 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2113 | ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. |
2114 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2115 | ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space #5. |
2116 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2125 A vector type is a simple derived type that represents a vector of
2126 elements. Vector types are used when multiple primitive data are
2127 operated in parallel using a single instruction (SIMD). A vector type
2128 requires a size (number of elements) and an underlying primitive data
2129 type. Vector types are considered :ref:`first class <t_firstclass>`.
2135 < <# elements> x <elementtype> >
2137 The number of elements is a constant integer value larger than 0;
2138 elementtype may be any integer, floating point or pointer type. Vectors
2139 of size zero are not allowed.
2143 +-------------------+--------------------------------------------------+
2144 | ``<4 x i32>`` | Vector of 4 32-bit integer values. |
2145 +-------------------+--------------------------------------------------+
2146 | ``<8 x float>`` | Vector of 8 32-bit floating-point values. |
2147 +-------------------+--------------------------------------------------+
2148 | ``<2 x i64>`` | Vector of 2 64-bit integer values. |
2149 +-------------------+--------------------------------------------------+
2150 | ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. |
2151 +-------------------+--------------------------------------------------+
2160 The label type represents code labels.
2175 The metadata type represents embedded metadata. No derived types may be
2176 created from metadata except for :ref:`function <t_function>` arguments.
2189 Aggregate Types are a subset of derived types that can contain multiple
2190 member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are
2191 aggregate types. :ref:`Vectors <t_vector>` are not considered to be
2201 The array type is a very simple derived type that arranges elements
2202 sequentially in memory. The array type requires a size (number of
2203 elements) and an underlying data type.
2209 [<# elements> x <elementtype>]
2211 The number of elements is a constant integer value; ``elementtype`` may
2212 be any type with a size.
2216 +------------------+--------------------------------------+
2217 | ``[40 x i32]`` | Array of 40 32-bit integer values. |
2218 +------------------+--------------------------------------+
2219 | ``[41 x i32]`` | Array of 41 32-bit integer values. |
2220 +------------------+--------------------------------------+
2221 | ``[4 x i8]`` | Array of 4 8-bit integer values. |
2222 +------------------+--------------------------------------+
2224 Here are some examples of multidimensional arrays:
2226 +-----------------------------+----------------------------------------------------------+
2227 | ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. |
2228 +-----------------------------+----------------------------------------------------------+
2229 | ``[12 x [10 x float]]`` | 12x10 array of single precision floating point values. |
2230 +-----------------------------+----------------------------------------------------------+
2231 | ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. |
2232 +-----------------------------+----------------------------------------------------------+
2234 There is no restriction on indexing beyond the end of the array implied
2235 by a static type (though there are restrictions on indexing beyond the
2236 bounds of an allocated object in some cases). This means that
2237 single-dimension 'variable sized array' addressing can be implemented in
2238 LLVM with a zero length array type. An implementation of 'pascal style
2239 arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for
2249 The structure type is used to represent a collection of data members
2250 together in memory. The elements of a structure may be any type that has
2253 Structures in memory are accessed using '``load``' and '``store``' by
2254 getting a pointer to a field with the '``getelementptr``' instruction.
2255 Structures in registers are accessed using the '``extractvalue``' and
2256 '``insertvalue``' instructions.
2258 Structures may optionally be "packed" structures, which indicate that
2259 the alignment of the struct is one byte, and that there is no padding
2260 between the elements. In non-packed structs, padding between field types
2261 is inserted as defined by the DataLayout string in the module, which is
2262 required to match what the underlying code generator expects.
2264 Structures can either be "literal" or "identified". A literal structure
2265 is defined inline with other types (e.g. ``{i32, i32}*``) whereas
2266 identified types are always defined at the top level with a name.
2267 Literal types are uniqued by their contents and can never be recursive
2268 or opaque since there is no way to write one. Identified types can be
2269 recursive, can be opaqued, and are never uniqued.
2275 %T1 = type { <type list> } ; Identified normal struct type
2276 %T2 = type <{ <type list> }> ; Identified packed struct type
2280 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2281 | ``{ i32, i32, i32 }`` | A triple of three ``i32`` values |
2282 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2283 | ``{ 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``. |
2284 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2285 | ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. |
2286 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2290 Opaque Structure Types
2291 """"""""""""""""""""""
2295 Opaque structure types are used to represent named structure types that
2296 do not have a body specified. This corresponds (for example) to the C
2297 notion of a forward declared structure.
2308 +--------------+-------------------+
2309 | ``opaque`` | An opaque type. |
2310 +--------------+-------------------+
2317 LLVM has several different basic types of constants. This section
2318 describes them all and their syntax.
2323 **Boolean constants**
2324 The two strings '``true``' and '``false``' are both valid constants
2326 **Integer constants**
2327 Standard integers (such as '4') are constants of the
2328 :ref:`integer <t_integer>` type. Negative numbers may be used with
2330 **Floating point constants**
2331 Floating point constants use standard decimal notation (e.g.
2332 123.421), exponential notation (e.g. 1.23421e+2), or a more precise
2333 hexadecimal notation (see below). The assembler requires the exact
2334 decimal value of a floating-point constant. For example, the
2335 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating
2336 decimal in binary. Floating point constants must have a :ref:`floating
2337 point <t_floating>` type.
2338 **Null pointer constants**
2339 The identifier '``null``' is recognized as a null pointer constant
2340 and must be of :ref:`pointer type <t_pointer>`.
2342 The one non-intuitive notation for constants is the hexadecimal form of
2343 floating point constants. For example, the form
2344 '``double 0x432ff973cafa8000``' is equivalent to (but harder to read
2345 than) '``double 4.5e+15``'. The only time hexadecimal floating point
2346 constants are required (and the only time that they are generated by the
2347 disassembler) is when a floating point constant must be emitted but it
2348 cannot be represented as a decimal floating point number in a reasonable
2349 number of digits. For example, NaN's, infinities, and other special
2350 values are represented in their IEEE hexadecimal format so that assembly
2351 and disassembly do not cause any bits to change in the constants.
2353 When using the hexadecimal form, constants of types half, float, and
2354 double are represented using the 16-digit form shown above (which
2355 matches the IEEE754 representation for double); half and float values
2356 must, however, be exactly representable as IEEE 754 half and single
2357 precision, respectively. Hexadecimal format is always used for long
2358 double, and there are three forms of long double. The 80-bit format used
2359 by x86 is represented as ``0xK`` followed by 20 hexadecimal digits. The
2360 128-bit format used by PowerPC (two adjacent doubles) is represented by
2361 ``0xM`` followed by 32 hexadecimal digits. The IEEE 128-bit format is
2362 represented by ``0xL`` followed by 32 hexadecimal digits. Long doubles
2363 will only work if they match the long double format on your target.
2364 The IEEE 16-bit format (half precision) is represented by ``0xH``
2365 followed by 4 hexadecimal digits. All hexadecimal formats are big-endian
2366 (sign bit at the left).
2368 There are no constants of type x86_mmx.
2370 .. _complexconstants:
2375 Complex constants are a (potentially recursive) combination of simple
2376 constants and smaller complex constants.
2378 **Structure constants**
2379 Structure constants are represented with notation similar to
2380 structure type definitions (a comma separated list of elements,
2381 surrounded by braces (``{}``)). For example:
2382 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as
2383 "``@G = external global i32``". Structure constants must have
2384 :ref:`structure type <t_struct>`, and the number and types of elements
2385 must match those specified by the type.
2387 Array constants are represented with notation similar to array type
2388 definitions (a comma separated list of elements, surrounded by
2389 square brackets (``[]``)). For example:
2390 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have
2391 :ref:`array type <t_array>`, and the number and types of elements must
2392 match those specified by the type. As a special case, character array
2393 constants may also be represented as a double-quoted string using the ``c``
2394 prefix. For example: "``c"Hello World\0A\00"``".
2395 **Vector constants**
2396 Vector constants are represented with notation similar to vector
2397 type definitions (a comma separated list of elements, surrounded by
2398 less-than/greater-than's (``<>``)). For example:
2399 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants
2400 must have :ref:`vector type <t_vector>`, and the number and types of
2401 elements must match those specified by the type.
2402 **Zero initialization**
2403 The string '``zeroinitializer``' can be used to zero initialize a
2404 value to zero of *any* type, including scalar and
2405 :ref:`aggregate <t_aggregate>` types. This is often used to avoid
2406 having to print large zero initializers (e.g. for large arrays) and
2407 is always exactly equivalent to using explicit zero initializers.
2409 A metadata node is a constant tuple without types. For example:
2410 "``!{!0, !{!2, !0}, !"test"}``". Metadata can reference constant values,
2411 for example: "``!{!0, i32 0, i8* @global, i64 (i64)* @function, !"str"}``".
2412 Unlike other typed constants that are meant to be interpreted as part of
2413 the instruction stream, metadata is a place to attach additional
2414 information such as debug info.
2416 Global Variable and Function Addresses
2417 --------------------------------------
2419 The addresses of :ref:`global variables <globalvars>` and
2420 :ref:`functions <functionstructure>` are always implicitly valid
2421 (link-time) constants. These constants are explicitly referenced when
2422 the :ref:`identifier for the global <identifiers>` is used and always have
2423 :ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM
2426 .. code-block:: llvm
2430 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2437 The string '``undef``' can be used anywhere a constant is expected, and
2438 indicates that the user of the value may receive an unspecified
2439 bit-pattern. Undefined values may be of any type (other than '``label``'
2440 or '``void``') and be used anywhere a constant is permitted.
2442 Undefined values are useful because they indicate to the compiler that
2443 the program is well defined no matter what value is used. This gives the
2444 compiler more freedom to optimize. Here are some examples of
2445 (potentially surprising) transformations that are valid (in pseudo IR):
2447 .. code-block:: llvm
2457 This is safe because all of the output bits are affected by the undef
2458 bits. Any output bit can have a zero or one depending on the input bits.
2460 .. code-block:: llvm
2471 These logical operations have bits that are not always affected by the
2472 input. For example, if ``%X`` has a zero bit, then the output of the
2473 '``and``' operation will always be a zero for that bit, no matter what
2474 the corresponding bit from the '``undef``' is. As such, it is unsafe to
2475 optimize or assume that the result of the '``and``' is '``undef``'.
2476 However, it is safe to assume that all bits of the '``undef``' could be
2477 0, and optimize the '``and``' to 0. Likewise, it is safe to assume that
2478 all the bits of the '``undef``' operand to the '``or``' could be set,
2479 allowing the '``or``' to be folded to -1.
2481 .. code-block:: llvm
2483 %A = select undef, %X, %Y
2484 %B = select undef, 42, %Y
2485 %C = select %X, %Y, undef
2495 This set of examples shows that undefined '``select``' (and conditional
2496 branch) conditions can go *either way*, but they have to come from one
2497 of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were
2498 both known to have a clear low bit, then ``%A`` would have to have a
2499 cleared low bit. However, in the ``%C`` example, the optimizer is
2500 allowed to assume that the '``undef``' operand could be the same as
2501 ``%Y``, allowing the whole '``select``' to be eliminated.
2503 .. code-block:: llvm
2505 %A = xor undef, undef
2522 This example points out that two '``undef``' operands are not
2523 necessarily the same. This can be surprising to people (and also matches
2524 C semantics) where they assume that "``X^X``" is always zero, even if
2525 ``X`` is undefined. This isn't true for a number of reasons, but the
2526 short answer is that an '``undef``' "variable" can arbitrarily change
2527 its value over its "live range". This is true because the variable
2528 doesn't actually *have a live range*. Instead, the value is logically
2529 read from arbitrary registers that happen to be around when needed, so
2530 the value is not necessarily consistent over time. In fact, ``%A`` and
2531 ``%C`` need to have the same semantics or the core LLVM "replace all
2532 uses with" concept would not hold.
2534 .. code-block:: llvm
2542 These examples show the crucial difference between an *undefined value*
2543 and *undefined behavior*. An undefined value (like '``undef``') is
2544 allowed to have an arbitrary bit-pattern. This means that the ``%A``
2545 operation can be constant folded to '``undef``', because the '``undef``'
2546 could be an SNaN, and ``fdiv`` is not (currently) defined on SNaN's.
2547 However, in the second example, we can make a more aggressive
2548 assumption: because the ``undef`` is allowed to be an arbitrary value,
2549 we are allowed to assume that it could be zero. Since a divide by zero
2550 has *undefined behavior*, we are allowed to assume that the operation
2551 does not execute at all. This allows us to delete the divide and all
2552 code after it. Because the undefined operation "can't happen", the
2553 optimizer can assume that it occurs in dead code.
2555 .. code-block:: llvm
2557 a: store undef -> %X
2558 b: store %X -> undef
2563 These examples reiterate the ``fdiv`` example: a store *of* an undefined
2564 value can be assumed to not have any effect; we can assume that the
2565 value is overwritten with bits that happen to match what was already
2566 there. However, a store *to* an undefined location could clobber
2567 arbitrary memory, therefore, it has undefined behavior.
2574 Poison values are similar to :ref:`undef values <undefvalues>`, however
2575 they also represent the fact that an instruction or constant expression
2576 that cannot evoke side effects has nevertheless detected a condition
2577 that results in undefined behavior.
2579 There is currently no way of representing a poison value in the IR; they
2580 only exist when produced by operations such as :ref:`add <i_add>` with
2583 Poison value behavior is defined in terms of value *dependence*:
2585 - Values other than :ref:`phi <i_phi>` nodes depend on their operands.
2586 - :ref:`Phi <i_phi>` nodes depend on the operand corresponding to
2587 their dynamic predecessor basic block.
2588 - Function arguments depend on the corresponding actual argument values
2589 in the dynamic callers of their functions.
2590 - :ref:`Call <i_call>` instructions depend on the :ref:`ret <i_ret>`
2591 instructions that dynamically transfer control back to them.
2592 - :ref:`Invoke <i_invoke>` instructions depend on the
2593 :ref:`ret <i_ret>`, :ref:`resume <i_resume>`, or exception-throwing
2594 call instructions that dynamically transfer control back to them.
2595 - Non-volatile loads and stores depend on the most recent stores to all
2596 of the referenced memory addresses, following the order in the IR
2597 (including loads and stores implied by intrinsics such as
2598 :ref:`@llvm.memcpy <int_memcpy>`.)
2599 - An instruction with externally visible side effects depends on the
2600 most recent preceding instruction with externally visible side
2601 effects, following the order in the IR. (This includes :ref:`volatile
2602 operations <volatile>`.)
2603 - An instruction *control-depends* on a :ref:`terminator
2604 instruction <terminators>` if the terminator instruction has
2605 multiple successors and the instruction is always executed when
2606 control transfers to one of the successors, and may not be executed
2607 when control is transferred to another.
2608 - Additionally, an instruction also *control-depends* on a terminator
2609 instruction if the set of instructions it otherwise depends on would
2610 be different if the terminator had transferred control to a different
2612 - Dependence is transitive.
2614 Poison values have the same behavior as :ref:`undef values <undefvalues>`,
2615 with the additional effect that any instruction that has a *dependence*
2616 on a poison value has undefined behavior.
2618 Here are some examples:
2620 .. code-block:: llvm
2623 %poison = sub nuw i32 0, 1 ; Results in a poison value.
2624 %still_poison = and i32 %poison, 0 ; 0, but also poison.
2625 %poison_yet_again = getelementptr i32, i32* @h, i32 %still_poison
2626 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
2628 store i32 %poison, i32* @g ; Poison value stored to memory.
2629 %poison2 = load i32, i32* @g ; Poison value loaded back from memory.
2631 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
2633 %narrowaddr = bitcast i32* @g to i16*
2634 %wideaddr = bitcast i32* @g to i64*
2635 %poison3 = load i16, i16* %narrowaddr ; Returns a poison value.
2636 %poison4 = load i64, i64* %wideaddr ; Returns a poison value.
2638 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2639 br i1 %cmp, label %true, label %end ; Branch to either destination.
2642 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2643 ; it has undefined behavior.
2647 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2648 ; Both edges into this PHI are
2649 ; control-dependent on %cmp, so this
2650 ; always results in a poison value.
2652 store volatile i32 0, i32* @g ; This would depend on the store in %true
2653 ; if %cmp is true, or the store in %entry
2654 ; otherwise, so this is undefined behavior.
2656 br i1 %cmp, label %second_true, label %second_end
2657 ; The same branch again, but this time the
2658 ; true block doesn't have side effects.
2665 store volatile i32 0, i32* @g ; This time, the instruction always depends
2666 ; on the store in %end. Also, it is
2667 ; control-equivalent to %end, so this is
2668 ; well-defined (ignoring earlier undefined
2669 ; behavior in this example).
2673 Addresses of Basic Blocks
2674 -------------------------
2676 ``blockaddress(@function, %block)``
2678 The '``blockaddress``' constant computes the address of the specified
2679 basic block in the specified function, and always has an ``i8*`` type.
2680 Taking the address of the entry block is illegal.
2682 This value only has defined behavior when used as an operand to the
2683 ':ref:`indirectbr <i_indirectbr>`' instruction, or for comparisons
2684 against null. Pointer equality tests between labels addresses results in
2685 undefined behavior --- though, again, comparison against null is ok, and
2686 no label is equal to the null pointer. This may be passed around as an
2687 opaque pointer sized value as long as the bits are not inspected. This
2688 allows ``ptrtoint`` and arithmetic to be performed on these values so
2689 long as the original value is reconstituted before the ``indirectbr``
2692 Finally, some targets may provide defined semantics when using the value
2693 as the operand to an inline assembly, but that is target specific.
2697 Constant Expressions
2698 --------------------
2700 Constant expressions are used to allow expressions involving other
2701 constants to be used as constants. Constant expressions may be of any
2702 :ref:`first class <t_firstclass>` type and may involve any LLVM operation
2703 that does not have side effects (e.g. load and call are not supported).
2704 The following is the syntax for constant expressions:
2706 ``trunc (CST to TYPE)``
2707 Truncate a constant to another type. The bit size of CST must be
2708 larger than the bit size of TYPE. Both types must be integers.
2709 ``zext (CST to TYPE)``
2710 Zero extend a constant to another type. The bit size of CST must be
2711 smaller than the bit size of TYPE. Both types must be integers.
2712 ``sext (CST to TYPE)``
2713 Sign extend a constant to another type. The bit size of CST must be
2714 smaller than the bit size of TYPE. Both types must be integers.
2715 ``fptrunc (CST to TYPE)``
2716 Truncate a floating point constant to another floating point type.
2717 The size of CST must be larger than the size of TYPE. Both types
2718 must be floating point.
2719 ``fpext (CST to TYPE)``
2720 Floating point extend a constant to another type. The size of CST
2721 must be smaller or equal to the size of TYPE. Both types must be
2723 ``fptoui (CST to TYPE)``
2724 Convert a floating point constant to the corresponding unsigned
2725 integer constant. TYPE must be a scalar or vector integer type. CST
2726 must be of scalar or vector floating point type. Both CST and TYPE
2727 must be scalars, or vectors of the same number of elements. If the
2728 value won't fit in the integer type, the results are undefined.
2729 ``fptosi (CST to TYPE)``
2730 Convert a floating point constant to the corresponding signed
2731 integer constant. TYPE must be a scalar or vector integer type. CST
2732 must be of scalar or vector floating point type. Both CST and TYPE
2733 must be scalars, or vectors of the same number of elements. If the
2734 value won't fit in the integer type, the results are undefined.
2735 ``uitofp (CST to TYPE)``
2736 Convert an unsigned integer constant to the corresponding floating
2737 point constant. TYPE must be a scalar or vector floating point type.
2738 CST must be of scalar or vector integer type. Both CST and TYPE must
2739 be scalars, or vectors of the same number of elements. If the value
2740 won't fit in the floating point type, the results are undefined.
2741 ``sitofp (CST to TYPE)``
2742 Convert a signed integer constant to the corresponding floating
2743 point constant. TYPE must be a scalar or vector floating point type.
2744 CST must be of scalar or vector integer type. Both CST and TYPE must
2745 be scalars, or vectors of the same number of elements. If the value
2746 won't fit in the floating point type, the results are undefined.
2747 ``ptrtoint (CST to TYPE)``
2748 Convert a pointer typed constant to the corresponding integer
2749 constant. ``TYPE`` must be an integer type. ``CST`` must be of
2750 pointer type. The ``CST`` value is zero extended, truncated, or
2751 unchanged to make it fit in ``TYPE``.
2752 ``inttoptr (CST to TYPE)``
2753 Convert an integer constant to a pointer constant. TYPE must be a
2754 pointer type. CST must be of integer type. The CST value is zero
2755 extended, truncated, or unchanged to make it fit in a pointer size.
2756 This one is *really* dangerous!
2757 ``bitcast (CST to TYPE)``
2758 Convert a constant, CST, to another TYPE. The constraints of the
2759 operands are the same as those for the :ref:`bitcast
2760 instruction <i_bitcast>`.
2761 ``addrspacecast (CST to TYPE)``
2762 Convert a constant pointer or constant vector of pointer, CST, to another
2763 TYPE in a different address space. The constraints of the operands are the
2764 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`.
2765 ``getelementptr (TY, CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (TY, CSTPTR, IDX0, IDX1, ...)``
2766 Perform the :ref:`getelementptr operation <i_getelementptr>` on
2767 constants. As with the :ref:`getelementptr <i_getelementptr>`
2768 instruction, the index list may have zero or more indexes, which are
2769 required to make sense for the type of "pointer to TY".
2770 ``select (COND, VAL1, VAL2)``
2771 Perform the :ref:`select operation <i_select>` on constants.
2772 ``icmp COND (VAL1, VAL2)``
2773 Performs the :ref:`icmp operation <i_icmp>` on constants.
2774 ``fcmp COND (VAL1, VAL2)``
2775 Performs the :ref:`fcmp operation <i_fcmp>` on constants.
2776 ``extractelement (VAL, IDX)``
2777 Perform the :ref:`extractelement operation <i_extractelement>` on
2779 ``insertelement (VAL, ELT, IDX)``
2780 Perform the :ref:`insertelement operation <i_insertelement>` on
2782 ``shufflevector (VEC1, VEC2, IDXMASK)``
2783 Perform the :ref:`shufflevector operation <i_shufflevector>` on
2785 ``extractvalue (VAL, IDX0, IDX1, ...)``
2786 Perform the :ref:`extractvalue operation <i_extractvalue>` on
2787 constants. The index list is interpreted in a similar manner as
2788 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At
2789 least one index value must be specified.
2790 ``insertvalue (VAL, ELT, IDX0, IDX1, ...)``
2791 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants.
2792 The index list is interpreted in a similar manner as indices in a
2793 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index
2794 value must be specified.
2795 ``OPCODE (LHS, RHS)``
2796 Perform the specified operation of the LHS and RHS constants. OPCODE
2797 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise
2798 binary <bitwiseops>` operations. The constraints on operands are
2799 the same as those for the corresponding instruction (e.g. no bitwise
2800 operations on floating point values are allowed).
2807 Inline Assembler Expressions
2808 ----------------------------
2810 LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level
2811 Inline Assembly <moduleasm>`) through the use of a special value. This value
2812 represents the inline assembler as a template string (containing the
2813 instructions to emit), a list of operand constraints (stored as a string), a
2814 flag that indicates whether or not the inline asm expression has side effects,
2815 and a flag indicating whether the function containing the asm needs to align its
2816 stack conservatively.
2818 The template string supports argument substitution of the operands using "``$``"
2819 followed by a number, to indicate substitution of the given register/memory
2820 location, as specified by the constraint string. "``${NUM:MODIFIER}``" may also
2821 be used, where ``MODIFIER`` is a target-specific annotation for how to print the
2822 operand (See :ref:`inline-asm-modifiers`).
2824 A literal "``$``" may be included by using "``$$``" in the template. To include
2825 other special characters into the output, the usual "``\XX``" escapes may be
2826 used, just as in other strings. Note that after template substitution, the
2827 resulting assembly string is parsed by LLVM's integrated assembler unless it is
2828 disabled -- even when emitting a ``.s`` file -- and thus must contain assembly
2829 syntax known to LLVM.
2831 LLVM's support for inline asm is modeled closely on the requirements of Clang's
2832 GCC-compatible inline-asm support. Thus, the feature-set and the constraint and
2833 modifier codes listed here are similar or identical to those in GCC's inline asm
2834 support. However, to be clear, the syntax of the template and constraint strings
2835 described here is *not* the same as the syntax accepted by GCC and Clang, and,
2836 while most constraint letters are passed through as-is by Clang, some get
2837 translated to other codes when converting from the C source to the LLVM
2840 An example inline assembler expression is:
2842 .. code-block:: llvm
2844 i32 (i32) asm "bswap $0", "=r,r"
2846 Inline assembler expressions may **only** be used as the callee operand
2847 of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction.
2848 Thus, typically we have:
2850 .. code-block:: llvm
2852 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y)
2854 Inline asms with side effects not visible in the constraint list must be
2855 marked as having side effects. This is done through the use of the
2856 '``sideeffect``' keyword, like so:
2858 .. code-block:: llvm
2860 call void asm sideeffect "eieio", ""()
2862 In some cases inline asms will contain code that will not work unless
2863 the stack is aligned in some way, such as calls or SSE instructions on
2864 x86, yet will not contain code that does that alignment within the asm.
2865 The compiler should make conservative assumptions about what the asm
2866 might contain and should generate its usual stack alignment code in the
2867 prologue if the '``alignstack``' keyword is present:
2869 .. code-block:: llvm
2871 call void asm alignstack "eieio", ""()
2873 Inline asms also support using non-standard assembly dialects. The
2874 assumed dialect is ATT. When the '``inteldialect``' keyword is present,
2875 the inline asm is using the Intel dialect. Currently, ATT and Intel are
2876 the only supported dialects. An example is:
2878 .. code-block:: llvm
2880 call void asm inteldialect "eieio", ""()
2882 If multiple keywords appear the '``sideeffect``' keyword must come
2883 first, the '``alignstack``' keyword second and the '``inteldialect``'
2886 Inline Asm Constraint String
2887 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2889 The constraint list is a comma-separated string, each element containing one or
2890 more constraint codes.
2892 For each element in the constraint list an appropriate register or memory
2893 operand will be chosen, and it will be made available to assembly template
2894 string expansion as ``$0`` for the first constraint in the list, ``$1`` for the
2897 There are three different types of constraints, which are distinguished by a
2898 prefix symbol in front of the constraint code: Output, Input, and Clobber. The
2899 constraints must always be given in that order: outputs first, then inputs, then
2900 clobbers. They cannot be intermingled.
2902 There are also three different categories of constraint codes:
2904 - Register constraint. This is either a register class, or a fixed physical
2905 register. This kind of constraint will allocate a register, and if necessary,
2906 bitcast the argument or result to the appropriate type.
2907 - Memory constraint. This kind of constraint is for use with an instruction
2908 taking a memory operand. Different constraints allow for different addressing
2909 modes used by the target.
2910 - Immediate value constraint. This kind of constraint is for an integer or other
2911 immediate value which can be rendered directly into an instruction. The
2912 various target-specific constraints allow the selection of a value in the
2913 proper range for the instruction you wish to use it with.
2918 Output constraints are specified by an "``=``" prefix (e.g. "``=r``"). This
2919 indicates that the assembly will write to this operand, and the operand will
2920 then be made available as a return value of the ``asm`` expression. Output
2921 constraints do not consume an argument from the call instruction. (Except, see
2922 below about indirect outputs).
2924 Normally, it is expected that no output locations are written to by the assembly
2925 expression until *all* of the inputs have been read. As such, LLVM may assign
2926 the same register to an output and an input. If this is not safe (e.g. if the
2927 assembly contains two instructions, where the first writes to one output, and
2928 the second reads an input and writes to a second output), then the "``&``"
2929 modifier must be used (e.g. "``=&r``") to specify that the output is an
2930 "early-clobber" output. Marking an ouput as "early-clobber" ensures that LLVM
2931 will not use the same register for any inputs (other than an input tied to this
2937 Input constraints do not have a prefix -- just the constraint codes. Each input
2938 constraint will consume one argument from the call instruction. It is not
2939 permitted for the asm to write to any input register or memory location (unless
2940 that input is tied to an output). Note also that multiple inputs may all be
2941 assigned to the same register, if LLVM can determine that they necessarily all
2942 contain the same value.
2944 Instead of providing a Constraint Code, input constraints may also "tie"
2945 themselves to an output constraint, by providing an integer as the constraint
2946 string. Tied inputs still consume an argument from the call instruction, and
2947 take up a position in the asm template numbering as is usual -- they will simply
2948 be constrained to always use the same register as the output they've been tied
2949 to. For example, a constraint string of "``=r,0``" says to assign a register for
2950 output, and use that register as an input as well (it being the 0'th
2953 It is permitted to tie an input to an "early-clobber" output. In that case, no
2954 *other* input may share the same register as the input tied to the early-clobber
2955 (even when the other input has the same value).
2957 You may only tie an input to an output which has a register constraint, not a
2958 memory constraint. Only a single input may be tied to an output.
2960 There is also an "interesting" feature which deserves a bit of explanation: if a
2961 register class constraint allocates a register which is too small for the value
2962 type operand provided as input, the input value will be split into multiple
2963 registers, and all of them passed to the inline asm.
2965 However, this feature is often not as useful as you might think.
2967 Firstly, the registers are *not* guaranteed to be consecutive. So, on those
2968 architectures that have instructions which operate on multiple consecutive
2969 instructions, this is not an appropriate way to support them. (e.g. the 32-bit
2970 SparcV8 has a 64-bit load, which instruction takes a single 32-bit register. The
2971 hardware then loads into both the named register, and the next register. This
2972 feature of inline asm would not be useful to support that.)
2974 A few of the targets provide a template string modifier allowing explicit access
2975 to the second register of a two-register operand (e.g. MIPS ``L``, ``M``, and
2976 ``D``). On such an architecture, you can actually access the second allocated
2977 register (yet, still, not any subsequent ones). But, in that case, you're still
2978 probably better off simply splitting the value into two separate operands, for
2979 clarity. (e.g. see the description of the ``A`` constraint on X86, which,
2980 despite existing only for use with this feature, is not really a good idea to
2983 Indirect inputs and outputs
2984 """""""""""""""""""""""""""
2986 Indirect output or input constraints can be specified by the "``*``" modifier
2987 (which goes after the "``=``" in case of an output). This indicates that the asm
2988 will write to or read from the contents of an *address* provided as an input
2989 argument. (Note that in this way, indirect outputs act more like an *input* than
2990 an output: just like an input, they consume an argument of the call expression,
2991 rather than producing a return value. An indirect output constraint is an
2992 "output" only in that the asm is expected to write to the contents of the input
2993 memory location, instead of just read from it).
2995 This is most typically used for memory constraint, e.g. "``=*m``", to pass the
2996 address of a variable as a value.
2998 It is also possible to use an indirect *register* constraint, but only on output
2999 (e.g. "``=*r``"). This will cause LLVM to allocate a register for an output
3000 value normally, and then, separately emit a store to the address provided as
3001 input, after the provided inline asm. (It's not clear what value this
3002 functionality provides, compared to writing the store explicitly after the asm
3003 statement, and it can only produce worse code, since it bypasses many
3004 optimization passes. I would recommend not using it.)
3010 A clobber constraint is indicated by a "``~``" prefix. A clobber does not
3011 consume an input operand, nor generate an output. Clobbers cannot use any of the
3012 general constraint code letters -- they may use only explicit register
3013 constraints, e.g. "``~{eax}``". The one exception is that a clobber string of
3014 "``~{memory}``" indicates that the assembly writes to arbitrary undeclared
3015 memory locations -- not only the memory pointed to by a declared indirect
3021 After a potential prefix comes constraint code, or codes.
3023 A Constraint Code is either a single letter (e.g. "``r``"), a "``^``" character
3024 followed by two letters (e.g. "``^wc``"), or "``{``" register-name "``}``"
3027 The one and two letter constraint codes are typically chosen to be the same as
3028 GCC's constraint codes.
3030 A single constraint may include one or more than constraint code in it, leaving
3031 it up to LLVM to choose which one to use. This is included mainly for
3032 compatibility with the translation of GCC inline asm coming from clang.
3034 There are two ways to specify alternatives, and either or both may be used in an
3035 inline asm constraint list:
3037 1) Append the codes to each other, making a constraint code set. E.g. "``im``"
3038 or "``{eax}m``". This means "choose any of the options in the set". The
3039 choice of constraint is made independently for each constraint in the
3042 2) Use "``|``" between constraint code sets, creating alternatives. Every
3043 constraint in the constraint list must have the same number of alternative
3044 sets. With this syntax, the same alternative in *all* of the items in the
3045 constraint list will be chosen together.
3047 Putting those together, you might have a two operand constraint string like
3048 ``"rm|r,ri|rm"``. This indicates that if operand 0 is ``r`` or ``m``, then
3049 operand 1 may be one of ``r`` or ``i``. If operand 0 is ``r``, then operand 1
3050 may be one of ``r`` or ``m``. But, operand 0 and 1 cannot both be of type m.
3052 However, the use of either of the alternatives features is *NOT* recommended, as
3053 LLVM is not able to make an intelligent choice about which one to use. (At the
3054 point it currently needs to choose, not enough information is available to do so
3055 in a smart way.) Thus, it simply tries to make a choice that's most likely to
3056 compile, not one that will be optimal performance. (e.g., given "``rm``", it'll
3057 always choose to use memory, not registers). And, if given multiple registers,
3058 or multiple register classes, it will simply choose the first one. (In fact, it
3059 doesn't currently even ensure explicitly specified physical registers are
3060 unique, so specifying multiple physical registers as alternatives, like
3061 ``{r11}{r12},{r11}{r12}``, will assign r11 to both operands, not at all what was
3064 Supported Constraint Code List
3065 """"""""""""""""""""""""""""""
3067 The constraint codes are, in general, expected to behave the same way they do in
3068 GCC. LLVM's support is often implemented on an 'as-needed' basis, to support C
3069 inline asm code which was supported by GCC. A mismatch in behavior between LLVM
3070 and GCC likely indicates a bug in LLVM.
3072 Some constraint codes are typically supported by all targets:
3074 - ``r``: A register in the target's general purpose register class.
3075 - ``m``: A memory address operand. It is target-specific what addressing modes
3076 are supported, typical examples are register, or register + register offset,
3077 or register + immediate offset (of some target-specific size).
3078 - ``i``: An integer constant (of target-specific width). Allows either a simple
3079 immediate, or a relocatable value.
3080 - ``n``: An integer constant -- *not* including relocatable values.
3081 - ``s``: An integer constant, but allowing *only* relocatable values.
3082 - ``X``: Allows an operand of any kind, no constraint whatsoever. Typically
3083 useful to pass a label for an asm branch or call.
3085 .. FIXME: but that surely isn't actually okay to jump out of an asm
3086 block without telling llvm about the control transfer???)
3088 - ``{register-name}``: Requires exactly the named physical register.
3090 Other constraints are target-specific:
3094 - ``z``: An immediate integer 0. Outputs ``WZR`` or ``XZR``, as appropriate.
3095 - ``I``: An immediate integer valid for an ``ADD`` or ``SUB`` instruction,
3096 i.e. 0 to 4095 with optional shift by 12.
3097 - ``J``: An immediate integer that, when negated, is valid for an ``ADD`` or
3098 ``SUB`` instruction, i.e. -1 to -4095 with optional left shift by 12.
3099 - ``K``: An immediate integer that is valid for the 'bitmask immediate 32' of a
3100 logical instruction like ``AND``, ``EOR``, or ``ORR`` with a 32-bit register.
3101 - ``L``: An immediate integer that is valid for the 'bitmask immediate 64' of a
3102 logical instruction like ``AND``, ``EOR``, or ``ORR`` with a 64-bit register.
3103 - ``M``: An immediate integer for use with the ``MOV`` assembly alias on a
3104 32-bit register. This is a superset of ``K``: in addition to the bitmask
3105 immediate, also allows immediate integers which can be loaded with a single
3106 ``MOVZ`` or ``MOVL`` instruction.
3107 - ``N``: An immediate integer for use with the ``MOV`` assembly alias on a
3108 64-bit register. This is a superset of ``L``.
3109 - ``Q``: Memory address operand must be in a single register (no
3110 offsets). (However, LLVM currently does this for the ``m`` constraint as
3112 - ``r``: A 32 or 64-bit integer register (W* or X*).
3113 - ``w``: A 32, 64, or 128-bit floating-point/SIMD register.
3114 - ``x``: A lower 128-bit floating-point/SIMD register (``V0`` to ``V15``).
3118 - ``r``: A 32 or 64-bit integer register.
3119 - ``[0-9]v``: The 32-bit VGPR register, number 0-9.
3120 - ``[0-9]s``: The 32-bit SGPR register, number 0-9.
3125 - ``Q``, ``Um``, ``Un``, ``Uq``, ``Us``, ``Ut``, ``Uv``, ``Uy``: Memory address
3126 operand. Treated the same as operand ``m``, at the moment.
3128 ARM and ARM's Thumb2 mode:
3130 - ``j``: An immediate integer between 0 and 65535 (valid for ``MOVW``)
3131 - ``I``: An immediate integer valid for a data-processing instruction.
3132 - ``J``: An immediate integer between -4095 and 4095.
3133 - ``K``: An immediate integer whose bitwise inverse is valid for a
3134 data-processing instruction. (Can be used with template modifier "``B``" to
3135 print the inverted value).
3136 - ``L``: An immediate integer whose negation is valid for a data-processing
3137 instruction. (Can be used with template modifier "``n``" to print the negated
3139 - ``M``: A power of two or a integer between 0 and 32.
3140 - ``N``: Invalid immediate constraint.
3141 - ``O``: Invalid immediate constraint.
3142 - ``r``: A general-purpose 32-bit integer register (``r0-r15``).
3143 - ``l``: In Thumb2 mode, low 32-bit GPR registers (``r0-r7``). In ARM mode, same
3145 - ``h``: In Thumb2 mode, a high 32-bit GPR register (``r8-r15``). In ARM mode,
3147 - ``w``: A 32, 64, or 128-bit floating-point/SIMD register: ``s0-s31``,
3148 ``d0-d31``, or ``q0-q15``.
3149 - ``x``: A 32, 64, or 128-bit floating-point/SIMD register: ``s0-s15``,
3150 ``d0-d7``, or ``q0-q3``.
3151 - ``t``: A floating-point/SIMD register, only supports 32-bit values:
3156 - ``I``: An immediate integer between 0 and 255.
3157 - ``J``: An immediate integer between -255 and -1.
3158 - ``K``: An immediate integer between 0 and 255, with optional left-shift by
3160 - ``L``: An immediate integer between -7 and 7.
3161 - ``M``: An immediate integer which is a multiple of 4 between 0 and 1020.
3162 - ``N``: An immediate integer between 0 and 31.
3163 - ``O``: An immediate integer which is a multiple of 4 between -508 and 508.
3164 - ``r``: A low 32-bit GPR register (``r0-r7``).
3165 - ``l``: A low 32-bit GPR register (``r0-r7``).
3166 - ``h``: A high GPR register (``r0-r7``).
3167 - ``w``: A 32, 64, or 128-bit floating-point/SIMD register: ``s0-s31``,
3168 ``d0-d31``, or ``q0-q15``.
3169 - ``x``: A 32, 64, or 128-bit floating-point/SIMD register: ``s0-s15``,
3170 ``d0-d7``, or ``q0-q3``.
3171 - ``t``: A floating-point/SIMD register, only supports 32-bit values:
3177 - ``o``, ``v``: A memory address operand, treated the same as constraint ``m``,
3179 - ``r``: A 32 or 64-bit register.
3183 - ``r``: An 8 or 16-bit register.
3187 - ``I``: An immediate signed 16-bit integer.
3188 - ``J``: An immediate integer zero.
3189 - ``K``: An immediate unsigned 16-bit integer.
3190 - ``L``: An immediate 32-bit integer, where the lower 16 bits are 0.
3191 - ``N``: An immediate integer between -65535 and -1.
3192 - ``O``: An immediate signed 15-bit integer.
3193 - ``P``: An immediate integer between 1 and 65535.
3194 - ``m``: A memory address operand. In MIPS-SE mode, allows a base address
3195 register plus 16-bit immediate offset. In MIPS mode, just a base register.
3196 - ``R``: A memory address operand. In MIPS-SE mode, allows a base address
3197 register plus a 9-bit signed offset. In MIPS mode, the same as constraint
3199 - ``ZC``: A memory address operand, suitable for use in a ``pref``, ``ll``, or
3200 ``sc`` instruction on the given subtarget (details vary).
3201 - ``r``, ``d``, ``y``: A 32 or 64-bit GPR register.
3202 - ``f``: A 32 or 64-bit FPU register (``F0-F31``), or a 128-bit MSA register
3203 (``W0-W31``). In the case of MSA registers, it is recommended to use the ``w``
3204 argument modifier for compatibility with GCC.
3205 - ``c``: A 32-bit or 64-bit GPR register suitable for indirect jump (always
3207 - ``l``: The ``lo`` register, 32 or 64-bit.
3212 - ``b``: A 1-bit integer register.
3213 - ``c`` or ``h``: A 16-bit integer register.
3214 - ``r``: A 32-bit integer register.
3215 - ``l`` or ``N``: A 64-bit integer register.
3216 - ``f``: A 32-bit float register.
3217 - ``d``: A 64-bit float register.
3222 - ``I``: An immediate signed 16-bit integer.
3223 - ``J``: An immediate unsigned 16-bit integer, shifted left 16 bits.
3224 - ``K``: An immediate unsigned 16-bit integer.
3225 - ``L``: An immediate signed 16-bit integer, shifted left 16 bits.
3226 - ``M``: An immediate integer greater than 31.
3227 - ``N``: An immediate integer that is an exact power of 2.
3228 - ``O``: The immediate integer constant 0.
3229 - ``P``: An immediate integer constant whose negation is a signed 16-bit
3231 - ``es``, ``o``, ``Q``, ``Z``, ``Zy``: A memory address operand, currently
3232 treated the same as ``m``.
3233 - ``r``: A 32 or 64-bit integer register.
3234 - ``b``: A 32 or 64-bit integer register, excluding ``R0`` (that is:
3236 - ``f``: A 32 or 64-bit float register (``F0-F31``), or when QPX is enabled, a
3237 128 or 256-bit QPX register (``Q0-Q31``; aliases the ``F`` registers).
3238 - ``v``: For ``4 x f32`` or ``4 x f64`` types, when QPX is enabled, a
3239 128 or 256-bit QPX register (``Q0-Q31``), otherwise a 128-bit
3240 altivec vector register (``V0-V31``).
3242 .. FIXME: is this a bug that v accepts QPX registers? I think this
3243 is supposed to only use the altivec vector registers?
3245 - ``y``: Condition register (``CR0-CR7``).
3246 - ``wc``: An individual CR bit in a CR register.
3247 - ``wa``, ``wd``, ``wf``: Any 128-bit VSX vector register, from the full VSX
3248 register set (overlapping both the floating-point and vector register files).
3249 - ``ws``: A 32 or 64-bit floating point register, from the full VSX register
3254 - ``I``: An immediate 13-bit signed integer.
3255 - ``r``: A 32-bit integer register.
3259 - ``I``: An immediate unsigned 8-bit integer.
3260 - ``J``: An immediate unsigned 12-bit integer.
3261 - ``K``: An immediate signed 16-bit integer.
3262 - ``L``: An immediate signed 20-bit integer.
3263 - ``M``: An immediate integer 0x7fffffff.
3264 - ``Q``, ``R``, ``S``, ``T``: A memory address operand, treated the same as
3265 ``m``, at the moment.
3266 - ``r`` or ``d``: A 32, 64, or 128-bit integer register.
3267 - ``a``: A 32, 64, or 128-bit integer address register (excludes R0, which in an
3268 address context evaluates as zero).
3269 - ``h``: A 32-bit value in the high part of a 64bit data register
3271 - ``f``: A 32, 64, or 128-bit floating point register.
3275 - ``I``: An immediate integer between 0 and 31.
3276 - ``J``: An immediate integer between 0 and 64.
3277 - ``K``: An immediate signed 8-bit integer.
3278 - ``L``: An immediate integer, 0xff or 0xffff or (in 64-bit mode only)
3280 - ``M``: An immediate integer between 0 and 3.
3281 - ``N``: An immediate unsigned 8-bit integer.
3282 - ``O``: An immediate integer between 0 and 127.
3283 - ``e``: An immediate 32-bit signed integer.
3284 - ``Z``: An immediate 32-bit unsigned integer.
3285 - ``o``, ``v``: Treated the same as ``m``, at the moment.
3286 - ``q``: An 8, 16, 32, or 64-bit register which can be accessed as an 8-bit
3287 ``l`` integer register. On X86-32, this is the ``a``, ``b``, ``c``, and ``d``
3288 registers, and on X86-64, it is all of the integer registers.
3289 - ``Q``: An 8, 16, 32, or 64-bit register which can be accessed as an 8-bit
3290 ``h`` integer register. This is the ``a``, ``b``, ``c``, and ``d`` registers.
3291 - ``r`` or ``l``: An 8, 16, 32, or 64-bit integer register.
3292 - ``R``: An 8, 16, 32, or 64-bit "legacy" integer register -- one which has
3293 existed since i386, and can be accessed without the REX prefix.
3294 - ``f``: A 32, 64, or 80-bit '387 FPU stack pseudo-register.
3295 - ``y``: A 64-bit MMX register, if MMX is enabled.
3296 - ``x``: If SSE is enabled: a 32 or 64-bit scalar operand, or 128-bit vector
3297 operand in a SSE register. If AVX is also enabled, can also be a 256-bit
3298 vector operand in an AVX register. If AVX-512 is also enabled, can also be a
3299 512-bit vector operand in an AVX512 register, Otherwise, an error.
3300 - ``Y``: The same as ``x``, if *SSE2* is enabled, otherwise an error.
3301 - ``A``: Special case: allocates EAX first, then EDX, for a single operand (in
3302 32-bit mode, a 64-bit integer operand will get split into two registers). It
3303 is not recommended to use this constraint, as in 64-bit mode, the 64-bit
3304 operand will get allocated only to RAX -- if two 32-bit operands are needed,
3305 you're better off splitting it yourself, before passing it to the asm
3310 - ``r``: A 32-bit integer register.
3313 .. _inline-asm-modifiers:
3315 Asm template argument modifiers
3316 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3318 In the asm template string, modifiers can be used on the operand reference, like
3321 The modifiers are, in general, expected to behave the same way they do in
3322 GCC. LLVM's support is often implemented on an 'as-needed' basis, to support C
3323 inline asm code which was supported by GCC. A mismatch in behavior between LLVM
3324 and GCC likely indicates a bug in LLVM.
3328 - ``c``: Print an immediate integer constant unadorned, without
3329 the target-specific immediate punctuation (e.g. no ``$`` prefix).
3330 - ``n``: Negate and print immediate integer constant unadorned, without the
3331 target-specific immediate punctuation (e.g. no ``$`` prefix).
3332 - ``l``: Print as an unadorned label, without the target-specific label
3333 punctuation (e.g. no ``$`` prefix).
3337 - ``w``: Print a GPR register with a ``w*`` name instead of ``x*`` name. E.g.,
3338 instead of ``x30``, print ``w30``.
3339 - ``x``: Print a GPR register with a ``x*`` name. (this is the default, anyhow).
3340 - ``b``, ``h``, ``s``, ``d``, ``q``: Print a floating-point/SIMD register with a
3341 ``b*``, ``h*``, ``s*``, ``d*``, or ``q*`` name, rather than the default of
3350 - ``a``: Print an operand as an address (with ``[`` and ``]`` surrounding a
3354 - ``y``: Print a VFP single-precision register as an indexed double (e.g. print
3355 as ``d4[1]`` instead of ``s9``)
3356 - ``B``: Bitwise invert and print an immediate integer constant without ``#``
3358 - ``L``: Print the low 16-bits of an immediate integer constant.
3359 - ``M``: Print as a register set suitable for ldm/stm. Also prints *all*
3360 register operands subsequent to the specified one (!), so use carefully.
3361 - ``Q``: Print the low-order register of a register-pair, or the low-order
3362 register of a two-register operand.
3363 - ``R``: Print the high-order register of a register-pair, or the high-order
3364 register of a two-register operand.
3365 - ``H``: Print the second register of a register-pair. (On a big-endian system,
3366 ``H`` is equivalent to ``Q``, and on little-endian system, ``H`` is equivalent
3369 .. FIXME: H doesn't currently support printing the second register
3370 of a two-register operand.
3372 - ``e``: Print the low doubleword register of a NEON quad register.
3373 - ``f``: Print the high doubleword register of a NEON quad register.
3374 - ``m``: Print the base register of a memory operand without the ``[`` and ``]``
3379 - ``L``: Print the second register of a two-register operand. Requires that it
3380 has been allocated consecutively to the first.
3382 .. FIXME: why is it restricted to consecutive ones? And there's
3383 nothing that ensures that happens, is there?
3385 - ``I``: Print the letter 'i' if the operand is an integer constant, otherwise
3386 nothing. Used to print 'addi' vs 'add' instructions.
3390 No additional modifiers.
3394 - ``X``: Print an immediate integer as hexadecimal
3395 - ``x``: Print the low 16 bits of an immediate integer as hexadecimal.
3396 - ``d``: Print an immediate integer as decimal.
3397 - ``m``: Subtract one and print an immediate integer as decimal.
3398 - ``z``: Print $0 if an immediate zero, otherwise print normally.
3399 - ``L``: Print the low-order register of a two-register operand, or prints the
3400 address of the low-order word of a double-word memory operand.
3402 .. FIXME: L seems to be missing memory operand support.
3404 - ``M``: Print the high-order register of a two-register operand, or prints the
3405 address of the high-order word of a double-word memory operand.
3407 .. FIXME: M seems to be missing memory operand support.
3409 - ``D``: Print the second register of a two-register operand, or prints the
3410 second word of a double-word memory operand. (On a big-endian system, ``D`` is
3411 equivalent to ``L``, and on little-endian system, ``D`` is equivalent to
3413 - ``w``: No effect. Provided for compatibility with GCC which requires this
3414 modifier in order to print MSA registers (``W0-W31``) with the ``f``
3423 - ``L``: Print the second register of a two-register operand. Requires that it
3424 has been allocated consecutively to the first.
3426 .. FIXME: why is it restricted to consecutive ones? And there's
3427 nothing that ensures that happens, is there?
3429 - ``I``: Print the letter 'i' if the operand is an integer constant, otherwise
3430 nothing. Used to print 'addi' vs 'add' instructions.
3431 - ``y``: For a memory operand, prints formatter for a two-register X-form
3432 instruction. (Currently always prints ``r0,OPERAND``).
3433 - ``U``: Prints 'u' if the memory operand is an update form, and nothing
3434 otherwise. (NOTE: LLVM does not support update form, so this will currently
3435 always print nothing)
3436 - ``X``: Prints 'x' if the memory operand is an indexed form. (NOTE: LLVM does
3437 not support indexed form, so this will currently always print nothing)
3445 SystemZ implements only ``n``, and does *not* support any of the other
3446 target-independent modifiers.
3450 - ``c``: Print an unadorned integer or symbol name. (The latter is
3451 target-specific behavior for this typically target-independent modifier).
3452 - ``A``: Print a register name with a '``*``' before it.
3453 - ``b``: Print an 8-bit register name (e.g. ``al``); do nothing on a memory
3455 - ``h``: Print the upper 8-bit register name (e.g. ``ah``); do nothing on a
3457 - ``w``: Print the 16-bit register name (e.g. ``ax``); do nothing on a memory
3459 - ``k``: Print the 32-bit register name (e.g. ``eax``); do nothing on a memory
3461 - ``q``: Print the 64-bit register name (e.g. ``rax``), if 64-bit registers are
3462 available, otherwise the 32-bit register name; do nothing on a memory operand.
3463 - ``n``: Negate and print an unadorned integer, or, for operands other than an
3464 immediate integer (e.g. a relocatable symbol expression), print a '-' before
3465 the operand. (The behavior for relocatable symbol expressions is a
3466 target-specific behavior for this typically target-independent modifier)
3467 - ``H``: Print a memory reference with additional offset +8.
3468 - ``P``: Print a memory reference or operand for use as the argument of a call
3469 instruction. (E.g. omit ``(rip)``, even though it's PC-relative.)
3473 No additional modifiers.
3479 The call instructions that wrap inline asm nodes may have a
3480 "``!srcloc``" MDNode attached to it that contains a list of constant
3481 integers. If present, the code generator will use the integer as the
3482 location cookie value when report errors through the ``LLVMContext``
3483 error reporting mechanisms. This allows a front-end to correlate backend
3484 errors that occur with inline asm back to the source code that produced
3487 .. code-block:: llvm
3489 call void asm sideeffect "something bad", ""(), !srcloc !42
3491 !42 = !{ i32 1234567 }
3493 It is up to the front-end to make sense of the magic numbers it places
3494 in the IR. If the MDNode contains multiple constants, the code generator
3495 will use the one that corresponds to the line of the asm that the error
3503 LLVM IR allows metadata to be attached to instructions in the program
3504 that can convey extra information about the code to the optimizers and
3505 code generator. One example application of metadata is source-level
3506 debug information. There are two metadata primitives: strings and nodes.
3508 Metadata does not have a type, and is not a value. If referenced from a
3509 ``call`` instruction, it uses the ``metadata`` type.
3511 All metadata are identified in syntax by a exclamation point ('``!``').
3513 .. _metadata-string:
3515 Metadata Nodes and Metadata Strings
3516 -----------------------------------
3518 A metadata string is a string surrounded by double quotes. It can
3519 contain any character by escaping non-printable characters with
3520 "``\xx``" where "``xx``" is the two digit hex code. For example:
3523 Metadata nodes are represented with notation similar to structure
3524 constants (a comma separated list of elements, surrounded by braces and
3525 preceded by an exclamation point). Metadata nodes can have any values as
3526 their operand. For example:
3528 .. code-block:: llvm
3530 !{ !"test\00", i32 10}
3532 Metadata nodes that aren't uniqued use the ``distinct`` keyword. For example:
3534 .. code-block:: llvm
3536 !0 = distinct !{!"test\00", i32 10}
3538 ``distinct`` nodes are useful when nodes shouldn't be merged based on their
3539 content. They can also occur when transformations cause uniquing collisions
3540 when metadata operands change.
3542 A :ref:`named metadata <namedmetadatastructure>` is a collection of
3543 metadata nodes, which can be looked up in the module symbol table. For
3546 .. code-block:: llvm
3550 Metadata can be used as function arguments. Here ``llvm.dbg.value``
3551 function is using two metadata arguments:
3553 .. code-block:: llvm
3555 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
3557 Metadata can be attached with an instruction. Here metadata ``!21`` is
3558 attached to the ``add`` instruction using the ``!dbg`` identifier:
3560 .. code-block:: llvm
3562 %indvar.next = add i64 %indvar, 1, !dbg !21
3564 More information about specific metadata nodes recognized by the
3565 optimizers and code generator is found below.
3567 .. _specialized-metadata:
3569 Specialized Metadata Nodes
3570 ^^^^^^^^^^^^^^^^^^^^^^^^^^
3572 Specialized metadata nodes are custom data structures in metadata (as opposed
3573 to generic tuples). Their fields are labelled, and can be specified in any
3576 These aren't inherently debug info centric, but currently all the specialized
3577 metadata nodes are related to debug info.
3584 ``DICompileUnit`` nodes represent a compile unit. The ``enums:``,
3585 ``retainedTypes:``, ``subprograms:``, ``globals:`` and ``imports:`` fields are
3586 tuples containing the debug info to be emitted along with the compile unit,
3587 regardless of code optimizations (some nodes are only emitted if there are
3588 references to them from instructions).
3590 .. code-block:: llvm
3592 !0 = !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang",
3593 isOptimized: true, flags: "-O2", runtimeVersion: 2,
3594 splitDebugFilename: "abc.debug", emissionKind: 1,
3595 enums: !2, retainedTypes: !3, subprograms: !4,
3596 globals: !5, imports: !6)
3598 Compile unit descriptors provide the root scope for objects declared in a
3599 specific compilation unit. File descriptors are defined using this scope.
3600 These descriptors are collected by a named metadata ``!llvm.dbg.cu``. They
3601 keep track of subprograms, global variables, type information, and imported
3602 entities (declarations and namespaces).
3609 ``DIFile`` nodes represent files. The ``filename:`` can include slashes.
3611 .. code-block:: llvm
3613 !0 = !DIFile(filename: "path/to/file", directory: "/path/to/dir")
3615 Files are sometimes used in ``scope:`` fields, and are the only valid target
3616 for ``file:`` fields.
3623 ``DIBasicType`` nodes represent primitive types, such as ``int``, ``bool`` and
3624 ``float``. ``tag:`` defaults to ``DW_TAG_base_type``.
3626 .. code-block:: llvm
3628 !0 = !DIBasicType(name: "unsigned char", size: 8, align: 8,
3629 encoding: DW_ATE_unsigned_char)
3630 !1 = !DIBasicType(tag: DW_TAG_unspecified_type, name: "decltype(nullptr)")
3632 The ``encoding:`` describes the details of the type. Usually it's one of the
3635 .. code-block:: llvm
3641 DW_ATE_signed_char = 6
3643 DW_ATE_unsigned_char = 8
3645 .. _DISubroutineType:
3650 ``DISubroutineType`` nodes represent subroutine types. Their ``types:`` field
3651 refers to a tuple; the first operand is the return type, while the rest are the
3652 types of the formal arguments in order. If the first operand is ``null``, that
3653 represents a function with no return value (such as ``void foo() {}`` in C++).
3655 .. code-block:: llvm
3657 !0 = !BasicType(name: "int", size: 32, align: 32, DW_ATE_signed)
3658 !1 = !BasicType(name: "char", size: 8, align: 8, DW_ATE_signed_char)
3659 !2 = !DISubroutineType(types: !{null, !0, !1}) ; void (int, char)
3666 ``DIDerivedType`` nodes represent types derived from other types, such as
3669 .. code-block:: llvm
3671 !0 = !DIBasicType(name: "unsigned char", size: 8, align: 8,
3672 encoding: DW_ATE_unsigned_char)
3673 !1 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !0, size: 32,
3676 The following ``tag:`` values are valid:
3678 .. code-block:: llvm
3680 DW_TAG_formal_parameter = 5
3682 DW_TAG_pointer_type = 15
3683 DW_TAG_reference_type = 16
3685 DW_TAG_ptr_to_member_type = 31
3686 DW_TAG_const_type = 38
3687 DW_TAG_volatile_type = 53
3688 DW_TAG_restrict_type = 55
3690 ``DW_TAG_member`` is used to define a member of a :ref:`composite type
3691 <DICompositeType>` or :ref:`subprogram <DISubprogram>`. The type of the member
3692 is the ``baseType:``. The ``offset:`` is the member's bit offset.
3693 ``DW_TAG_formal_parameter`` is used to define a member which is a formal
3694 argument of a subprogram.
3696 ``DW_TAG_typedef`` is used to provide a name for the ``baseType:``.
3698 ``DW_TAG_pointer_type``, ``DW_TAG_reference_type``, ``DW_TAG_const_type``,
3699 ``DW_TAG_volatile_type`` and ``DW_TAG_restrict_type`` are used to qualify the
3702 Note that the ``void *`` type is expressed as a type derived from NULL.
3704 .. _DICompositeType:
3709 ``DICompositeType`` nodes represent types composed of other types, like
3710 structures and unions. ``elements:`` points to a tuple of the composed types.
3712 If the source language supports ODR, the ``identifier:`` field gives the unique
3713 identifier used for type merging between modules. When specified, other types
3714 can refer to composite types indirectly via a :ref:`metadata string
3715 <metadata-string>` that matches their identifier.
3717 .. code-block:: llvm
3719 !0 = !DIEnumerator(name: "SixKind", value: 7)
3720 !1 = !DIEnumerator(name: "SevenKind", value: 7)
3721 !2 = !DIEnumerator(name: "NegEightKind", value: -8)
3722 !3 = !DICompositeType(tag: DW_TAG_enumeration_type, name: "Enum", file: !12,
3723 line: 2, size: 32, align: 32, identifier: "_M4Enum",
3724 elements: !{!0, !1, !2})
3726 The following ``tag:`` values are valid:
3728 .. code-block:: llvm
3730 DW_TAG_array_type = 1
3731 DW_TAG_class_type = 2
3732 DW_TAG_enumeration_type = 4
3733 DW_TAG_structure_type = 19
3734 DW_TAG_union_type = 23
3735 DW_TAG_subroutine_type = 21
3736 DW_TAG_inheritance = 28
3739 For ``DW_TAG_array_type``, the ``elements:`` should be :ref:`subrange
3740 descriptors <DISubrange>`, each representing the range of subscripts at that
3741 level of indexing. The ``DIFlagVector`` flag to ``flags:`` indicates that an
3742 array type is a native packed vector.
3744 For ``DW_TAG_enumeration_type``, the ``elements:`` should be :ref:`enumerator
3745 descriptors <DIEnumerator>`, each representing the definition of an enumeration
3746 value for the set. All enumeration type descriptors are collected in the
3747 ``enums:`` field of the :ref:`compile unit <DICompileUnit>`.
3749 For ``DW_TAG_structure_type``, ``DW_TAG_class_type``, and
3750 ``DW_TAG_union_type``, the ``elements:`` should be :ref:`derived types
3751 <DIDerivedType>` with ``tag: DW_TAG_member`` or ``tag: DW_TAG_inheritance``.
3758 ``DISubrange`` nodes are the elements for ``DW_TAG_array_type`` variants of
3759 :ref:`DICompositeType`. ``count: -1`` indicates an empty array.
3761 .. code-block:: llvm
3763 !0 = !DISubrange(count: 5, lowerBound: 0) ; array counting from 0
3764 !1 = !DISubrange(count: 5, lowerBound: 1) ; array counting from 1
3765 !2 = !DISubrange(count: -1) ; empty array.
3772 ``DIEnumerator`` nodes are the elements for ``DW_TAG_enumeration_type``
3773 variants of :ref:`DICompositeType`.
3775 .. code-block:: llvm
3777 !0 = !DIEnumerator(name: "SixKind", value: 7)
3778 !1 = !DIEnumerator(name: "SevenKind", value: 7)
3779 !2 = !DIEnumerator(name: "NegEightKind", value: -8)
3781 DITemplateTypeParameter
3782 """""""""""""""""""""""
3784 ``DITemplateTypeParameter`` nodes represent type parameters to generic source
3785 language constructs. They are used (optionally) in :ref:`DICompositeType` and
3786 :ref:`DISubprogram` ``templateParams:`` fields.
3788 .. code-block:: llvm
3790 !0 = !DITemplateTypeParameter(name: "Ty", type: !1)
3792 DITemplateValueParameter
3793 """"""""""""""""""""""""
3795 ``DITemplateValueParameter`` nodes represent value parameters to generic source
3796 language constructs. ``tag:`` defaults to ``DW_TAG_template_value_parameter``,
3797 but if specified can also be set to ``DW_TAG_GNU_template_template_param`` or
3798 ``DW_TAG_GNU_template_param_pack``. They are used (optionally) in
3799 :ref:`DICompositeType` and :ref:`DISubprogram` ``templateParams:`` fields.
3801 .. code-block:: llvm
3803 !0 = !DITemplateValueParameter(name: "Ty", type: !1, value: i32 7)
3808 ``DINamespace`` nodes represent namespaces in the source language.
3810 .. code-block:: llvm
3812 !0 = !DINamespace(name: "myawesomeproject", scope: !1, file: !2, line: 7)
3817 ``DIGlobalVariable`` nodes represent global variables in the source language.
3819 .. code-block:: llvm
3821 !0 = !DIGlobalVariable(name: "foo", linkageName: "foo", scope: !1,
3822 file: !2, line: 7, type: !3, isLocal: true,
3823 isDefinition: false, variable: i32* @foo,
3826 All global variables should be referenced by the `globals:` field of a
3827 :ref:`compile unit <DICompileUnit>`.
3834 ``DISubprogram`` nodes represent functions from the source language. The
3835 ``variables:`` field points at :ref:`variables <DILocalVariable>` that must be
3836 retained, even if their IR counterparts are optimized out of the IR. The
3837 ``type:`` field must point at an :ref:`DISubroutineType`.
3839 .. code-block:: llvm
3841 !0 = !DISubprogram(name: "foo", linkageName: "_Zfoov", scope: !1,
3842 file: !2, line: 7, type: !3, isLocal: true,
3843 isDefinition: false, scopeLine: 8, containingType: !4,
3844 virtuality: DW_VIRTUALITY_pure_virtual, virtualIndex: 10,
3845 flags: DIFlagPrototyped, isOptimized: true,
3846 function: void ()* @_Z3foov,
3847 templateParams: !5, declaration: !6, variables: !7)
3854 ``DILexicalBlock`` nodes describe nested blocks within a :ref:`subprogram
3855 <DISubprogram>`. The line number and column numbers are used to dinstinguish
3856 two lexical blocks at same depth. They are valid targets for ``scope:``
3859 .. code-block:: llvm
3861 !0 = distinct !DILexicalBlock(scope: !1, file: !2, line: 7, column: 35)
3863 Usually lexical blocks are ``distinct`` to prevent node merging based on
3866 .. _DILexicalBlockFile:
3871 ``DILexicalBlockFile`` nodes are used to discriminate between sections of a
3872 :ref:`lexical block <DILexicalBlock>`. The ``file:`` field can be changed to
3873 indicate textual inclusion, or the ``discriminator:`` field can be used to
3874 discriminate between control flow within a single block in the source language.
3876 .. code-block:: llvm
3878 !0 = !DILexicalBlock(scope: !3, file: !4, line: 7, column: 35)
3879 !1 = !DILexicalBlockFile(scope: !0, file: !4, discriminator: 0)
3880 !2 = !DILexicalBlockFile(scope: !0, file: !4, discriminator: 1)
3887 ``DILocation`` nodes represent source debug locations. The ``scope:`` field is
3888 mandatory, and points at an :ref:`DILexicalBlockFile`, an
3889 :ref:`DILexicalBlock`, or an :ref:`DISubprogram`.
3891 .. code-block:: llvm
3893 !0 = !DILocation(line: 2900, column: 42, scope: !1, inlinedAt: !2)
3895 .. _DILocalVariable:
3900 ``DILocalVariable`` nodes represent local variables in the source language.
3901 Instead of ``DW_TAG_variable``, they use LLVM-specific fake tags to
3902 discriminate between local variables (``DW_TAG_auto_variable``) and subprogram
3903 arguments (``DW_TAG_arg_variable``). In the latter case, the ``arg:`` field
3904 specifies the argument position, and this variable will be included in the
3905 ``variables:`` field of its :ref:`DISubprogram`.
3907 .. code-block:: llvm
3909 !0 = !DILocalVariable(tag: DW_TAG_arg_variable, name: "this", arg: 0,
3910 scope: !3, file: !2, line: 7, type: !3,
3911 flags: DIFlagArtificial)
3912 !1 = !DILocalVariable(tag: DW_TAG_arg_variable, name: "x", arg: 1,
3913 scope: !4, file: !2, line: 7, type: !3)
3914 !1 = !DILocalVariable(tag: DW_TAG_auto_variable, name: "y",
3915 scope: !5, file: !2, line: 7, type: !3)
3920 ``DIExpression`` nodes represent DWARF expression sequences. They are used in
3921 :ref:`debug intrinsics<dbg_intrinsics>` (such as ``llvm.dbg.declare``) to
3922 describe how the referenced LLVM variable relates to the source language
3925 The current supported vocabulary is limited:
3927 - ``DW_OP_deref`` dereferences the working expression.
3928 - ``DW_OP_plus, 93`` adds ``93`` to the working expression.
3929 - ``DW_OP_bit_piece, 16, 8`` specifies the offset and size (``16`` and ``8``
3930 here, respectively) of the variable piece from the working expression.
3932 .. code-block:: llvm
3934 !0 = !DIExpression(DW_OP_deref)
3935 !1 = !DIExpression(DW_OP_plus, 3)
3936 !2 = !DIExpression(DW_OP_bit_piece, 3, 7)
3937 !3 = !DIExpression(DW_OP_deref, DW_OP_plus, 3, DW_OP_bit_piece, 3, 7)
3942 ``DIObjCProperty`` nodes represent Objective-C property nodes.
3944 .. code-block:: llvm
3946 !3 = !DIObjCProperty(name: "foo", file: !1, line: 7, setter: "setFoo",
3947 getter: "getFoo", attributes: 7, type: !2)
3952 ``DIImportedEntity`` nodes represent entities (such as modules) imported into a
3955 .. code-block:: llvm
3957 !2 = !DIImportedEntity(tag: DW_TAG_imported_module, name: "foo", scope: !0,
3958 entity: !1, line: 7)
3963 In LLVM IR, memory does not have types, so LLVM's own type system is not
3964 suitable for doing TBAA. Instead, metadata is added to the IR to
3965 describe a type system of a higher level language. This can be used to
3966 implement typical C/C++ TBAA, but it can also be used to implement
3967 custom alias analysis behavior for other languages.
3969 The current metadata format is very simple. TBAA metadata nodes have up
3970 to three fields, e.g.:
3972 .. code-block:: llvm
3974 !0 = !{ !"an example type tree" }
3975 !1 = !{ !"int", !0 }
3976 !2 = !{ !"float", !0 }
3977 !3 = !{ !"const float", !2, i64 1 }
3979 The first field is an identity field. It can be any value, usually a
3980 metadata string, which uniquely identifies the type. The most important
3981 name in the tree is the name of the root node. Two trees with different
3982 root node names are entirely disjoint, even if they have leaves with
3985 The second field identifies the type's parent node in the tree, or is
3986 null or omitted for a root node. A type is considered to alias all of
3987 its descendants and all of its ancestors in the tree. Also, a type is
3988 considered to alias all types in other trees, so that bitcode produced
3989 from multiple front-ends is handled conservatively.
3991 If the third field is present, it's an integer which if equal to 1
3992 indicates that the type is "constant" (meaning
3993 ``pointsToConstantMemory`` should return true; see `other useful
3994 AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_).
3996 '``tbaa.struct``' Metadata
3997 ^^^^^^^^^^^^^^^^^^^^^^^^^^
3999 The :ref:`llvm.memcpy <int_memcpy>` is often used to implement
4000 aggregate assignment operations in C and similar languages, however it
4001 is defined to copy a contiguous region of memory, which is more than
4002 strictly necessary for aggregate types which contain holes due to
4003 padding. Also, it doesn't contain any TBAA information about the fields
4006 ``!tbaa.struct`` metadata can describe which memory subregions in a
4007 memcpy are padding and what the TBAA tags of the struct are.
4009 The current metadata format is very simple. ``!tbaa.struct`` metadata
4010 nodes are a list of operands which are in conceptual groups of three.
4011 For each group of three, the first operand gives the byte offset of a
4012 field in bytes, the second gives its size in bytes, and the third gives
4015 .. code-block:: llvm
4017 !4 = !{ i64 0, i64 4, !1, i64 8, i64 4, !2 }
4019 This describes a struct with two fields. The first is at offset 0 bytes
4020 with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes
4021 and has size 4 bytes and has tbaa tag !2.
4023 Note that the fields need not be contiguous. In this example, there is a
4024 4 byte gap between the two fields. This gap represents padding which
4025 does not carry useful data and need not be preserved.
4027 '``noalias``' and '``alias.scope``' Metadata
4028 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4030 ``noalias`` and ``alias.scope`` metadata provide the ability to specify generic
4031 noalias memory-access sets. This means that some collection of memory access
4032 instructions (loads, stores, memory-accessing calls, etc.) that carry
4033 ``noalias`` metadata can specifically be specified not to alias with some other
4034 collection of memory access instructions that carry ``alias.scope`` metadata.
4035 Each type of metadata specifies a list of scopes where each scope has an id and
4036 a domain. When evaluating an aliasing query, if for some domain, the set
4037 of scopes with that domain in one instruction's ``alias.scope`` list is a
4038 subset of (or equal to) the set of scopes for that domain in another
4039 instruction's ``noalias`` list, then the two memory accesses are assumed not to
4042 The metadata identifying each domain is itself a list containing one or two
4043 entries. The first entry is the name of the domain. Note that if the name is a
4044 string then it can be combined accross functions and translation units. A
4045 self-reference can be used to create globally unique domain names. A
4046 descriptive string may optionally be provided as a second list entry.
4048 The metadata identifying each scope is also itself a list containing two or
4049 three entries. The first entry is the name of the scope. Note that if the name
4050 is a string then it can be combined accross functions and translation units. A
4051 self-reference can be used to create globally unique scope names. A metadata
4052 reference to the scope's domain is the second entry. A descriptive string may
4053 optionally be provided as a third list entry.
4057 .. code-block:: llvm
4059 ; Two scope domains:
4063 ; Some scopes in these domains:
4069 !5 = !{!4} ; A list containing only scope !4
4073 ; These two instructions don't alias:
4074 %0 = load float, float* %c, align 4, !alias.scope !5
4075 store float %0, float* %arrayidx.i, align 4, !noalias !5
4077 ; These two instructions also don't alias (for domain !1, the set of scopes
4078 ; in the !alias.scope equals that in the !noalias list):
4079 %2 = load float, float* %c, align 4, !alias.scope !5
4080 store float %2, float* %arrayidx.i2, align 4, !noalias !6
4082 ; These two instructions may alias (for domain !0, the set of scopes in
4083 ; the !noalias list is not a superset of, or equal to, the scopes in the
4084 ; !alias.scope list):
4085 %2 = load float, float* %c, align 4, !alias.scope !6
4086 store float %0, float* %arrayidx.i, align 4, !noalias !7
4088 '``fpmath``' Metadata
4089 ^^^^^^^^^^^^^^^^^^^^^
4091 ``fpmath`` metadata may be attached to any instruction of floating point
4092 type. It can be used to express the maximum acceptable error in the
4093 result of that instruction, in ULPs, thus potentially allowing the
4094 compiler to use a more efficient but less accurate method of computing
4095 it. ULP is defined as follows:
4097 If ``x`` is a real number that lies between two finite consecutive
4098 floating-point numbers ``a`` and ``b``, without being equal to one
4099 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the
4100 distance between the two non-equal finite floating-point numbers
4101 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``.
4103 The metadata node shall consist of a single positive floating point
4104 number representing the maximum relative error, for example:
4106 .. code-block:: llvm
4108 !0 = !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
4112 '``range``' Metadata
4113 ^^^^^^^^^^^^^^^^^^^^
4115 ``range`` metadata may be attached only to ``load``, ``call`` and ``invoke`` of
4116 integer types. It expresses the possible ranges the loaded value or the value
4117 returned by the called function at this call site is in. The ranges are
4118 represented with a flattened list of integers. The loaded value or the value
4119 returned is known to be in the union of the ranges defined by each consecutive
4120 pair. Each pair has the following properties:
4122 - The type must match the type loaded by the instruction.
4123 - The pair ``a,b`` represents the range ``[a,b)``.
4124 - Both ``a`` and ``b`` are constants.
4125 - The range is allowed to wrap.
4126 - The range should not represent the full or empty set. That is,
4129 In addition, the pairs must be in signed order of the lower bound and
4130 they must be non-contiguous.
4134 .. code-block:: llvm
4136 %a = load i8, i8* %x, align 1, !range !0 ; Can only be 0 or 1
4137 %b = load i8, i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
4138 %c = call i8 @foo(), !range !2 ; Can only be 0, 1, 3, 4 or 5
4139 %d = invoke i8 @bar() to label %cont
4140 unwind label %lpad, !range !3 ; Can only be -2, -1, 3, 4 or 5
4142 !0 = !{ i8 0, i8 2 }
4143 !1 = !{ i8 255, i8 2 }
4144 !2 = !{ i8 0, i8 2, i8 3, i8 6 }
4145 !3 = !{ i8 -2, i8 0, i8 3, i8 6 }
4150 It is sometimes useful to attach information to loop constructs. Currently,
4151 loop metadata is implemented as metadata attached to the branch instruction
4152 in the loop latch block. This type of metadata refer to a metadata node that is
4153 guaranteed to be separate for each loop. The loop identifier metadata is
4154 specified with the name ``llvm.loop``.
4156 The loop identifier metadata is implemented using a metadata that refers to
4157 itself to avoid merging it with any other identifier metadata, e.g.,
4158 during module linkage or function inlining. That is, each loop should refer
4159 to their own identification metadata even if they reside in separate functions.
4160 The following example contains loop identifier metadata for two separate loop
4163 .. code-block:: llvm
4168 The loop identifier metadata can be used to specify additional
4169 per-loop metadata. Any operands after the first operand can be treated
4170 as user-defined metadata. For example the ``llvm.loop.unroll.count``
4171 suggests an unroll factor to the loop unroller:
4173 .. code-block:: llvm
4175 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0
4178 !1 = !{!"llvm.loop.unroll.count", i32 4}
4180 '``llvm.loop.vectorize``' and '``llvm.loop.interleave``'
4181 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4183 Metadata prefixed with ``llvm.loop.vectorize`` or ``llvm.loop.interleave`` are
4184 used to control per-loop vectorization and interleaving parameters such as
4185 vectorization width and interleave count. These metadata should be used in
4186 conjunction with ``llvm.loop`` loop identification metadata. The
4187 ``llvm.loop.vectorize`` and ``llvm.loop.interleave`` metadata are only
4188 optimization hints and the optimizer will only interleave and vectorize loops if
4189 it believes it is safe to do so. The ``llvm.mem.parallel_loop_access`` metadata
4190 which contains information about loop-carried memory dependencies can be helpful
4191 in determining the safety of these transformations.
4193 '``llvm.loop.interleave.count``' Metadata
4194 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4196 This metadata suggests an interleave count to the loop interleaver.
4197 The first operand is the string ``llvm.loop.interleave.count`` and the
4198 second operand is an integer specifying the interleave count. For
4201 .. code-block:: llvm
4203 !0 = !{!"llvm.loop.interleave.count", i32 4}
4205 Note that setting ``llvm.loop.interleave.count`` to 1 disables interleaving
4206 multiple iterations of the loop. If ``llvm.loop.interleave.count`` is set to 0
4207 then the interleave count will be determined automatically.
4209 '``llvm.loop.vectorize.enable``' Metadata
4210 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4212 This metadata selectively enables or disables vectorization for the loop. The
4213 first operand is the string ``llvm.loop.vectorize.enable`` and the second operand
4214 is a bit. If the bit operand value is 1 vectorization is enabled. A value of
4215 0 disables vectorization:
4217 .. code-block:: llvm
4219 !0 = !{!"llvm.loop.vectorize.enable", i1 0}
4220 !1 = !{!"llvm.loop.vectorize.enable", i1 1}
4222 '``llvm.loop.vectorize.width``' Metadata
4223 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4225 This metadata sets the target width of the vectorizer. The first
4226 operand is the string ``llvm.loop.vectorize.width`` and the second
4227 operand is an integer specifying the width. For example:
4229 .. code-block:: llvm
4231 !0 = !{!"llvm.loop.vectorize.width", i32 4}
4233 Note that setting ``llvm.loop.vectorize.width`` to 1 disables
4234 vectorization of the loop. If ``llvm.loop.vectorize.width`` is set to
4235 0 or if the loop does not have this metadata the width will be
4236 determined automatically.
4238 '``llvm.loop.unroll``'
4239 ^^^^^^^^^^^^^^^^^^^^^^
4241 Metadata prefixed with ``llvm.loop.unroll`` are loop unrolling
4242 optimization hints such as the unroll factor. ``llvm.loop.unroll``
4243 metadata should be used in conjunction with ``llvm.loop`` loop
4244 identification metadata. The ``llvm.loop.unroll`` metadata are only
4245 optimization hints and the unrolling will only be performed if the
4246 optimizer believes it is safe to do so.
4248 '``llvm.loop.unroll.count``' Metadata
4249 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4251 This metadata suggests an unroll factor to the loop unroller. The
4252 first operand is the string ``llvm.loop.unroll.count`` and the second
4253 operand is a positive integer specifying the unroll factor. For
4256 .. code-block:: llvm
4258 !0 = !{!"llvm.loop.unroll.count", i32 4}
4260 If the trip count of the loop is less than the unroll count the loop
4261 will be partially unrolled.
4263 '``llvm.loop.unroll.disable``' Metadata
4264 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4266 This metadata disables loop unrolling. The metadata has a single operand
4267 which is the string ``llvm.loop.unroll.disable``. For example:
4269 .. code-block:: llvm
4271 !0 = !{!"llvm.loop.unroll.disable"}
4273 '``llvm.loop.unroll.runtime.disable``' Metadata
4274 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4276 This metadata disables runtime loop unrolling. The metadata has a single
4277 operand which is the string ``llvm.loop.unroll.runtime.disable``. For example:
4279 .. code-block:: llvm
4281 !0 = !{!"llvm.loop.unroll.runtime.disable"}
4283 '``llvm.loop.unroll.full``' Metadata
4284 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4286 This metadata suggests that the loop should be unrolled fully. The
4287 metadata has a single operand which is the string ``llvm.loop.unroll.full``.
4290 .. code-block:: llvm
4292 !0 = !{!"llvm.loop.unroll.full"}
4297 Metadata types used to annotate memory accesses with information helpful
4298 for optimizations are prefixed with ``llvm.mem``.
4300 '``llvm.mem.parallel_loop_access``' Metadata
4301 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4303 The ``llvm.mem.parallel_loop_access`` metadata refers to a loop identifier,
4304 or metadata containing a list of loop identifiers for nested loops.
4305 The metadata is attached to memory accessing instructions and denotes that
4306 no loop carried memory dependence exist between it and other instructions denoted
4307 with the same loop identifier.
4309 Precisely, given two instructions ``m1`` and ``m2`` that both have the
4310 ``llvm.mem.parallel_loop_access`` metadata, with ``L1`` and ``L2`` being the
4311 set of loops associated with that metadata, respectively, then there is no loop
4312 carried dependence between ``m1`` and ``m2`` for loops in both ``L1`` and
4315 As a special case, if all memory accessing instructions in a loop have
4316 ``llvm.mem.parallel_loop_access`` metadata that refers to that loop, then the
4317 loop has no loop carried memory dependences and is considered to be a parallel
4320 Note that if not all memory access instructions have such metadata referring to
4321 the loop, then the loop is considered not being trivially parallel. Additional
4322 memory dependence analysis is required to make that determination. As a fail
4323 safe mechanism, this causes loops that were originally parallel to be considered
4324 sequential (if optimization passes that are unaware of the parallel semantics
4325 insert new memory instructions into the loop body).
4327 Example of a loop that is considered parallel due to its correct use of
4328 both ``llvm.loop`` and ``llvm.mem.parallel_loop_access``
4329 metadata types that refer to the same loop identifier metadata.
4331 .. code-block:: llvm
4335 %val0 = load i32, i32* %arrayidx, !llvm.mem.parallel_loop_access !0
4337 store i32 %val0, i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
4339 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0
4345 It is also possible to have nested parallel loops. In that case the
4346 memory accesses refer to a list of loop identifier metadata nodes instead of
4347 the loop identifier metadata node directly:
4349 .. code-block:: llvm
4353 %val1 = load i32, i32* %arrayidx3, !llvm.mem.parallel_loop_access !2
4355 br label %inner.for.body
4359 %val0 = load i32, i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
4361 store i32 %val0, i32* %arrayidx2, !llvm.mem.parallel_loop_access !0
4363 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1
4367 store i32 %val1, i32* %arrayidx4, !llvm.mem.parallel_loop_access !2
4369 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2
4371 outer.for.end: ; preds = %for.body
4373 !0 = !{!1, !2} ; a list of loop identifiers
4374 !1 = !{!1} ; an identifier for the inner loop
4375 !2 = !{!2} ; an identifier for the outer loop
4380 The ``llvm.bitsets`` global metadata is used to implement
4381 :doc:`bitsets <BitSets>`.
4383 Module Flags Metadata
4384 =====================
4386 Information about the module as a whole is difficult to convey to LLVM's
4387 subsystems. The LLVM IR isn't sufficient to transmit this information.
4388 The ``llvm.module.flags`` named metadata exists in order to facilitate
4389 this. These flags are in the form of key / value pairs --- much like a
4390 dictionary --- making it easy for any subsystem who cares about a flag to
4393 The ``llvm.module.flags`` metadata contains a list of metadata triplets.
4394 Each triplet has the following form:
4396 - The first element is a *behavior* flag, which specifies the behavior
4397 when two (or more) modules are merged together, and it encounters two
4398 (or more) metadata with the same ID. The supported behaviors are
4400 - The second element is a metadata string that is a unique ID for the
4401 metadata. Each module may only have one flag entry for each unique ID (not
4402 including entries with the **Require** behavior).
4403 - The third element is the value of the flag.
4405 When two (or more) modules are merged together, the resulting
4406 ``llvm.module.flags`` metadata is the union of the modules' flags. That is, for
4407 each unique metadata ID string, there will be exactly one entry in the merged
4408 modules ``llvm.module.flags`` metadata table, and the value for that entry will
4409 be determined by the merge behavior flag, as described below. The only exception
4410 is that entries with the *Require* behavior are always preserved.
4412 The following behaviors are supported:
4423 Emits an error if two values disagree, otherwise the resulting value
4424 is that of the operands.
4428 Emits a warning if two values disagree. The result value will be the
4429 operand for the flag from the first module being linked.
4433 Adds a requirement that another module flag be present and have a
4434 specified value after linking is performed. The value must be a
4435 metadata pair, where the first element of the pair is the ID of the
4436 module flag to be restricted, and the second element of the pair is
4437 the value the module flag should be restricted to. This behavior can
4438 be used to restrict the allowable results (via triggering of an
4439 error) of linking IDs with the **Override** behavior.
4443 Uses the specified value, regardless of the behavior or value of the
4444 other module. If both modules specify **Override**, but the values
4445 differ, an error will be emitted.
4449 Appends the two values, which are required to be metadata nodes.
4453 Appends the two values, which are required to be metadata
4454 nodes. However, duplicate entries in the second list are dropped
4455 during the append operation.
4457 It is an error for a particular unique flag ID to have multiple behaviors,
4458 except in the case of **Require** (which adds restrictions on another metadata
4459 value) or **Override**.
4461 An example of module flags:
4463 .. code-block:: llvm
4465 !0 = !{ i32 1, !"foo", i32 1 }
4466 !1 = !{ i32 4, !"bar", i32 37 }
4467 !2 = !{ i32 2, !"qux", i32 42 }
4468 !3 = !{ i32 3, !"qux",
4473 !llvm.module.flags = !{ !0, !1, !2, !3 }
4475 - Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior
4476 if two or more ``!"foo"`` flags are seen is to emit an error if their
4477 values are not equal.
4479 - Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The
4480 behavior if two or more ``!"bar"`` flags are seen is to use the value
4483 - Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The
4484 behavior if two or more ``!"qux"`` flags are seen is to emit a
4485 warning if their values are not equal.
4487 - Metadata ``!3`` has the ID ``!"qux"`` and the value:
4493 The behavior is to emit an error if the ``llvm.module.flags`` does not
4494 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is
4497 Objective-C Garbage Collection Module Flags Metadata
4498 ----------------------------------------------------
4500 On the Mach-O platform, Objective-C stores metadata about garbage
4501 collection in a special section called "image info". The metadata
4502 consists of a version number and a bitmask specifying what types of
4503 garbage collection are supported (if any) by the file. If two or more
4504 modules are linked together their garbage collection metadata needs to
4505 be merged rather than appended together.
4507 The Objective-C garbage collection module flags metadata consists of the
4508 following key-value pairs:
4517 * - ``Objective-C Version``
4518 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2.
4520 * - ``Objective-C Image Info Version``
4521 - **[Required]** --- The version of the image info section. Currently
4524 * - ``Objective-C Image Info Section``
4525 - **[Required]** --- The section to place the metadata. Valid values are
4526 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and
4527 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for
4528 Objective-C ABI version 2.
4530 * - ``Objective-C Garbage Collection``
4531 - **[Required]** --- Specifies whether garbage collection is supported or
4532 not. Valid values are 0, for no garbage collection, and 2, for garbage
4533 collection supported.
4535 * - ``Objective-C GC Only``
4536 - **[Optional]** --- Specifies that only garbage collection is supported.
4537 If present, its value must be 6. This flag requires that the
4538 ``Objective-C Garbage Collection`` flag have the value 2.
4540 Some important flag interactions:
4542 - If a module with ``Objective-C Garbage Collection`` set to 0 is
4543 merged with a module with ``Objective-C Garbage Collection`` set to
4544 2, then the resulting module has the
4545 ``Objective-C Garbage Collection`` flag set to 0.
4546 - A module with ``Objective-C Garbage Collection`` set to 0 cannot be
4547 merged with a module with ``Objective-C GC Only`` set to 6.
4549 Automatic Linker Flags Module Flags Metadata
4550 --------------------------------------------
4552 Some targets support embedding flags to the linker inside individual object
4553 files. Typically this is used in conjunction with language extensions which
4554 allow source files to explicitly declare the libraries they depend on, and have
4555 these automatically be transmitted to the linker via object files.
4557 These flags are encoded in the IR using metadata in the module flags section,
4558 using the ``Linker Options`` key. The merge behavior for this flag is required
4559 to be ``AppendUnique``, and the value for the key is expected to be a metadata
4560 node which should be a list of other metadata nodes, each of which should be a
4561 list of metadata strings defining linker options.
4563 For example, the following metadata section specifies two separate sets of
4564 linker options, presumably to link against ``libz`` and the ``Cocoa``
4567 !0 = !{ i32 6, !"Linker Options",
4570 !{ !"-framework", !"Cocoa" } } }
4571 !llvm.module.flags = !{ !0 }
4573 The metadata encoding as lists of lists of options, as opposed to a collapsed
4574 list of options, is chosen so that the IR encoding can use multiple option
4575 strings to specify e.g., a single library, while still having that specifier be
4576 preserved as an atomic element that can be recognized by a target specific
4577 assembly writer or object file emitter.
4579 Each individual option is required to be either a valid option for the target's
4580 linker, or an option that is reserved by the target specific assembly writer or
4581 object file emitter. No other aspect of these options is defined by the IR.
4583 C type width Module Flags Metadata
4584 ----------------------------------
4586 The ARM backend emits a section into each generated object file describing the
4587 options that it was compiled with (in a compiler-independent way) to prevent
4588 linking incompatible objects, and to allow automatic library selection. Some
4589 of these options are not visible at the IR level, namely wchar_t width and enum
4592 To pass this information to the backend, these options are encoded in module
4593 flags metadata, using the following key-value pairs:
4603 - * 0 --- sizeof(wchar_t) == 4
4604 * 1 --- sizeof(wchar_t) == 2
4607 - * 0 --- Enums are at least as large as an ``int``.
4608 * 1 --- Enums are stored in the smallest integer type which can
4609 represent all of its values.
4611 For example, the following metadata section specifies that the module was
4612 compiled with a ``wchar_t`` width of 4 bytes, and the underlying type of an
4613 enum is the smallest type which can represent all of its values::
4615 !llvm.module.flags = !{!0, !1}
4616 !0 = !{i32 1, !"short_wchar", i32 1}
4617 !1 = !{i32 1, !"short_enum", i32 0}
4619 .. _intrinsicglobalvariables:
4621 Intrinsic Global Variables
4622 ==========================
4624 LLVM has a number of "magic" global variables that contain data that
4625 affect code generation or other IR semantics. These are documented here.
4626 All globals of this sort should have a section specified as
4627 "``llvm.metadata``". This section and all globals that start with
4628 "``llvm.``" are reserved for use by LLVM.
4632 The '``llvm.used``' Global Variable
4633 -----------------------------------
4635 The ``@llvm.used`` global is an array which has
4636 :ref:`appending linkage <linkage_appending>`. This array contains a list of
4637 pointers to named global variables, functions and aliases which may optionally
4638 have a pointer cast formed of bitcast or getelementptr. For example, a legal
4641 .. code-block:: llvm
4646 @llvm.used = appending global [2 x i8*] [
4648 i8* bitcast (i32* @Y to i8*)
4649 ], section "llvm.metadata"
4651 If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler,
4652 and linker are required to treat the symbol as if there is a reference to the
4653 symbol that it cannot see (which is why they have to be named). For example, if
4654 a variable has internal linkage and no references other than that from the
4655 ``@llvm.used`` list, it cannot be deleted. This is commonly used to represent
4656 references from inline asms and other things the compiler cannot "see", and
4657 corresponds to "``attribute((used))``" in GNU C.
4659 On some targets, the code generator must emit a directive to the
4660 assembler or object file to prevent the assembler and linker from
4661 molesting the symbol.
4663 .. _gv_llvmcompilerused:
4665 The '``llvm.compiler.used``' Global Variable
4666 --------------------------------------------
4668 The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used``
4669 directive, except that it only prevents the compiler from touching the
4670 symbol. On targets that support it, this allows an intelligent linker to
4671 optimize references to the symbol without being impeded as it would be
4674 This is a rare construct that should only be used in rare circumstances,
4675 and should not be exposed to source languages.
4677 .. _gv_llvmglobalctors:
4679 The '``llvm.global_ctors``' Global Variable
4680 -------------------------------------------
4682 .. code-block:: llvm
4684 %0 = type { i32, void ()*, i8* }
4685 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor, i8* @data }]
4687 The ``@llvm.global_ctors`` array contains a list of constructor
4688 functions, priorities, and an optional associated global or function.
4689 The functions referenced by this array will be called in ascending order
4690 of priority (i.e. lowest first) when the module is loaded. The order of
4691 functions with the same priority is not defined.
4693 If the third field is present, non-null, and points to a global variable
4694 or function, the initializer function will only run if the associated
4695 data from the current module is not discarded.
4697 .. _llvmglobaldtors:
4699 The '``llvm.global_dtors``' Global Variable
4700 -------------------------------------------
4702 .. code-block:: llvm
4704 %0 = type { i32, void ()*, i8* }
4705 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor, i8* @data }]
4707 The ``@llvm.global_dtors`` array contains a list of destructor
4708 functions, priorities, and an optional associated global or function.
4709 The functions referenced by this array will be called in descending
4710 order of priority (i.e. highest first) when the module is unloaded. The
4711 order of functions with the same priority is not defined.
4713 If the third field is present, non-null, and points to a global variable
4714 or function, the destructor function will only run if the associated
4715 data from the current module is not discarded.
4717 Instruction Reference
4718 =====================
4720 The LLVM instruction set consists of several different classifications
4721 of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary
4722 instructions <binaryops>`, :ref:`bitwise binary
4723 instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and
4724 :ref:`other instructions <otherops>`.
4728 Terminator Instructions
4729 -----------------------
4731 As mentioned :ref:`previously <functionstructure>`, every basic block in a
4732 program ends with a "Terminator" instruction, which indicates which
4733 block should be executed after the current block is finished. These
4734 terminator instructions typically yield a '``void``' value: they produce
4735 control flow, not values (the one exception being the
4736 ':ref:`invoke <i_invoke>`' instruction).
4738 The terminator instructions are: ':ref:`ret <i_ret>`',
4739 ':ref:`br <i_br>`', ':ref:`switch <i_switch>`',
4740 ':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`',
4741 ':ref:`resume <i_resume>`', and ':ref:`unreachable <i_unreachable>`'.
4745 '``ret``' Instruction
4746 ^^^^^^^^^^^^^^^^^^^^^
4753 ret <type> <value> ; Return a value from a non-void function
4754 ret void ; Return from void function
4759 The '``ret``' instruction is used to return control flow (and optionally
4760 a value) from a function back to the caller.
4762 There are two forms of the '``ret``' instruction: one that returns a
4763 value and then causes control flow, and one that just causes control
4769 The '``ret``' instruction optionally accepts a single argument, the
4770 return value. The type of the return value must be a ':ref:`first
4771 class <t_firstclass>`' type.
4773 A function is not :ref:`well formed <wellformed>` if it it has a non-void
4774 return type and contains a '``ret``' instruction with no return value or
4775 a return value with a type that does not match its type, or if it has a
4776 void return type and contains a '``ret``' instruction with a return
4782 When the '``ret``' instruction is executed, control flow returns back to
4783 the calling function's context. If the caller is a
4784 ":ref:`call <i_call>`" instruction, execution continues at the
4785 instruction after the call. If the caller was an
4786 ":ref:`invoke <i_invoke>`" instruction, execution continues at the
4787 beginning of the "normal" destination block. If the instruction returns
4788 a value, that value shall set the call or invoke instruction's return
4794 .. code-block:: llvm
4796 ret i32 5 ; Return an integer value of 5
4797 ret void ; Return from a void function
4798 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2
4802 '``br``' Instruction
4803 ^^^^^^^^^^^^^^^^^^^^
4810 br i1 <cond>, label <iftrue>, label <iffalse>
4811 br label <dest> ; Unconditional branch
4816 The '``br``' instruction is used to cause control flow to transfer to a
4817 different basic block in the current function. There are two forms of
4818 this instruction, corresponding to a conditional branch and an
4819 unconditional branch.
4824 The conditional branch form of the '``br``' instruction takes a single
4825 '``i1``' value and two '``label``' values. The unconditional form of the
4826 '``br``' instruction takes a single '``label``' value as a target.
4831 Upon execution of a conditional '``br``' instruction, the '``i1``'
4832 argument is evaluated. If the value is ``true``, control flows to the
4833 '``iftrue``' ``label`` argument. If "cond" is ``false``, control flows
4834 to the '``iffalse``' ``label`` argument.
4839 .. code-block:: llvm
4842 %cond = icmp eq i32 %a, %b
4843 br i1 %cond, label %IfEqual, label %IfUnequal
4851 '``switch``' Instruction
4852 ^^^^^^^^^^^^^^^^^^^^^^^^
4859 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
4864 The '``switch``' instruction is used to transfer control flow to one of
4865 several different places. It is a generalization of the '``br``'
4866 instruction, allowing a branch to occur to one of many possible
4872 The '``switch``' instruction uses three parameters: an integer
4873 comparison value '``value``', a default '``label``' destination, and an
4874 array of pairs of comparison value constants and '``label``'s. The table
4875 is not allowed to contain duplicate constant entries.
4880 The ``switch`` instruction specifies a table of values and destinations.
4881 When the '``switch``' instruction is executed, this table is searched
4882 for the given value. If the value is found, control flow is transferred
4883 to the corresponding destination; otherwise, control flow is transferred
4884 to the default destination.
4889 Depending on properties of the target machine and the particular
4890 ``switch`` instruction, this instruction may be code generated in
4891 different ways. For example, it could be generated as a series of
4892 chained conditional branches or with a lookup table.
4897 .. code-block:: llvm
4899 ; Emulate a conditional br instruction
4900 %Val = zext i1 %value to i32
4901 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
4903 ; Emulate an unconditional br instruction
4904 switch i32 0, label %dest [ ]
4906 ; Implement a jump table:
4907 switch i32 %val, label %otherwise [ i32 0, label %onzero
4909 i32 2, label %ontwo ]
4913 '``indirectbr``' Instruction
4914 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4921 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
4926 The '``indirectbr``' instruction implements an indirect branch to a
4927 label within the current function, whose address is specified by
4928 "``address``". Address must be derived from a
4929 :ref:`blockaddress <blockaddress>` constant.
4934 The '``address``' argument is the address of the label to jump to. The
4935 rest of the arguments indicate the full set of possible destinations
4936 that the address may point to. Blocks are allowed to occur multiple
4937 times in the destination list, though this isn't particularly useful.
4939 This destination list is required so that dataflow analysis has an
4940 accurate understanding of the CFG.
4945 Control transfers to the block specified in the address argument. All
4946 possible destination blocks must be listed in the label list, otherwise
4947 this instruction has undefined behavior. This implies that jumps to
4948 labels defined in other functions have undefined behavior as well.
4953 This is typically implemented with a jump through a register.
4958 .. code-block:: llvm
4960 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
4964 '``invoke``' Instruction
4965 ^^^^^^^^^^^^^^^^^^^^^^^^
4972 <result> = invoke [cconv] [ret attrs] <ptr to function ty> <function ptr val>(<function args>) [fn attrs]
4973 to label <normal label> unwind label <exception label>
4978 The '``invoke``' instruction causes control to transfer to a specified
4979 function, with the possibility of control flow transfer to either the
4980 '``normal``' label or the '``exception``' label. If the callee function
4981 returns with the "``ret``" instruction, control flow will return to the
4982 "normal" label. If the callee (or any indirect callees) returns via the
4983 ":ref:`resume <i_resume>`" instruction or other exception handling
4984 mechanism, control is interrupted and continued at the dynamically
4985 nearest "exception" label.
4987 The '``exception``' label is a `landing
4988 pad <ExceptionHandling.html#overview>`_ for the exception. As such,
4989 '``exception``' label is required to have the
4990 ":ref:`landingpad <i_landingpad>`" instruction, which contains the
4991 information about the behavior of the program after unwinding happens,
4992 as its first non-PHI instruction. The restrictions on the
4993 "``landingpad``" instruction's tightly couples it to the "``invoke``"
4994 instruction, so that the important information contained within the
4995 "``landingpad``" instruction can't be lost through normal code motion.
5000 This instruction requires several arguments:
5002 #. The optional "cconv" marker indicates which :ref:`calling
5003 convention <callingconv>` the call should use. If none is
5004 specified, the call defaults to using C calling conventions.
5005 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
5006 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
5008 #. '``ptr to function ty``': shall be the signature of the pointer to
5009 function value being invoked. In most cases, this is a direct
5010 function invocation, but indirect ``invoke``'s are just as possible,
5011 branching off an arbitrary pointer to function value.
5012 #. '``function ptr val``': An LLVM value containing a pointer to a
5013 function to be invoked.
5014 #. '``function args``': argument list whose types match the function
5015 signature argument types and parameter attributes. All arguments must
5016 be of :ref:`first class <t_firstclass>` type. If the function signature
5017 indicates the function accepts a variable number of arguments, the
5018 extra arguments can be specified.
5019 #. '``normal label``': the label reached when the called function
5020 executes a '``ret``' instruction.
5021 #. '``exception label``': the label reached when a callee returns via
5022 the :ref:`resume <i_resume>` instruction or other exception handling
5024 #. The optional :ref:`function attributes <fnattrs>` list. Only
5025 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
5026 attributes are valid here.
5031 This instruction is designed to operate as a standard '``call``'
5032 instruction in most regards. The primary difference is that it
5033 establishes an association with a label, which is used by the runtime
5034 library to unwind the stack.
5036 This instruction is used in languages with destructors to ensure that
5037 proper cleanup is performed in the case of either a ``longjmp`` or a
5038 thrown exception. Additionally, this is important for implementation of
5039 '``catch``' clauses in high-level languages that support them.
5041 For the purposes of the SSA form, the definition of the value returned
5042 by the '``invoke``' instruction is deemed to occur on the edge from the
5043 current block to the "normal" label. If the callee unwinds then no
5044 return value is available.
5049 .. code-block:: llvm
5051 %retval = invoke i32 @Test(i32 15) to label %Continue
5052 unwind label %TestCleanup ; i32:retval set
5053 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue
5054 unwind label %TestCleanup ; i32:retval set
5058 '``resume``' Instruction
5059 ^^^^^^^^^^^^^^^^^^^^^^^^
5066 resume <type> <value>
5071 The '``resume``' instruction is a terminator instruction that has no
5077 The '``resume``' instruction requires one argument, which must have the
5078 same type as the result of any '``landingpad``' instruction in the same
5084 The '``resume``' instruction resumes propagation of an existing
5085 (in-flight) exception whose unwinding was interrupted with a
5086 :ref:`landingpad <i_landingpad>` instruction.
5091 .. code-block:: llvm
5093 resume { i8*, i32 } %exn
5097 '``unreachable``' Instruction
5098 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5110 The '``unreachable``' instruction has no defined semantics. This
5111 instruction is used to inform the optimizer that a particular portion of
5112 the code is not reachable. This can be used to indicate that the code
5113 after a no-return function cannot be reached, and other facts.
5118 The '``unreachable``' instruction has no defined semantics.
5125 Binary operators are used to do most of the computation in a program.
5126 They require two operands of the same type, execute an operation on
5127 them, and produce a single value. The operands might represent multiple
5128 data, as is the case with the :ref:`vector <t_vector>` data type. The
5129 result value has the same type as its operands.
5131 There are several different binary operators:
5135 '``add``' Instruction
5136 ^^^^^^^^^^^^^^^^^^^^^
5143 <result> = add <ty> <op1>, <op2> ; yields ty:result
5144 <result> = add nuw <ty> <op1>, <op2> ; yields ty:result
5145 <result> = add nsw <ty> <op1>, <op2> ; yields ty:result
5146 <result> = add nuw nsw <ty> <op1>, <op2> ; yields ty:result
5151 The '``add``' instruction returns the sum of its two operands.
5156 The two arguments to the '``add``' instruction must be
5157 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5158 arguments must have identical types.
5163 The value produced is the integer sum of the two operands.
5165 If the sum has unsigned overflow, the result returned is the
5166 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
5169 Because LLVM integers use a two's complement representation, this
5170 instruction is appropriate for both signed and unsigned integers.
5172 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
5173 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
5174 result value of the ``add`` is a :ref:`poison value <poisonvalues>` if
5175 unsigned and/or signed overflow, respectively, occurs.
5180 .. code-block:: llvm
5182 <result> = add i32 4, %var ; yields i32:result = 4 + %var
5186 '``fadd``' Instruction
5187 ^^^^^^^^^^^^^^^^^^^^^^
5194 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
5199 The '``fadd``' instruction returns the sum of its two operands.
5204 The two arguments to the '``fadd``' instruction must be :ref:`floating
5205 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
5206 Both arguments must have identical types.
5211 The value produced is the floating point sum of the two operands. This
5212 instruction can also take any number of :ref:`fast-math flags <fastmath>`,
5213 which are optimization hints to enable otherwise unsafe floating point
5219 .. code-block:: llvm
5221 <result> = fadd float 4.0, %var ; yields float:result = 4.0 + %var
5223 '``sub``' Instruction
5224 ^^^^^^^^^^^^^^^^^^^^^
5231 <result> = sub <ty> <op1>, <op2> ; yields ty:result
5232 <result> = sub nuw <ty> <op1>, <op2> ; yields ty:result
5233 <result> = sub nsw <ty> <op1>, <op2> ; yields ty:result
5234 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields ty:result
5239 The '``sub``' instruction returns the difference of its two operands.
5241 Note that the '``sub``' instruction is used to represent the '``neg``'
5242 instruction present in most other intermediate representations.
5247 The two arguments to the '``sub``' instruction must be
5248 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5249 arguments must have identical types.
5254 The value produced is the integer difference of the two operands.
5256 If the difference has unsigned overflow, the result returned is the
5257 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
5260 Because LLVM integers use a two's complement representation, this
5261 instruction is appropriate for both signed and unsigned integers.
5263 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
5264 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
5265 result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if
5266 unsigned and/or signed overflow, respectively, occurs.
5271 .. code-block:: llvm
5273 <result> = sub i32 4, %var ; yields i32:result = 4 - %var
5274 <result> = sub i32 0, %val ; yields i32:result = -%var
5278 '``fsub``' Instruction
5279 ^^^^^^^^^^^^^^^^^^^^^^
5286 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
5291 The '``fsub``' instruction returns the difference of its two operands.
5293 Note that the '``fsub``' instruction is used to represent the '``fneg``'
5294 instruction present in most other intermediate representations.
5299 The two arguments to the '``fsub``' instruction must be :ref:`floating
5300 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
5301 Both arguments must have identical types.
5306 The value produced is the floating point difference of the two operands.
5307 This instruction can also take any number of :ref:`fast-math
5308 flags <fastmath>`, which are optimization hints to enable otherwise
5309 unsafe floating point optimizations:
5314 .. code-block:: llvm
5316 <result> = fsub float 4.0, %var ; yields float:result = 4.0 - %var
5317 <result> = fsub float -0.0, %val ; yields float:result = -%var
5319 '``mul``' Instruction
5320 ^^^^^^^^^^^^^^^^^^^^^
5327 <result> = mul <ty> <op1>, <op2> ; yields ty:result
5328 <result> = mul nuw <ty> <op1>, <op2> ; yields ty:result
5329 <result> = mul nsw <ty> <op1>, <op2> ; yields ty:result
5330 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields ty:result
5335 The '``mul``' instruction returns the product of its two operands.
5340 The two arguments to the '``mul``' instruction must be
5341 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5342 arguments must have identical types.
5347 The value produced is the integer product of the two operands.
5349 If the result of the multiplication has unsigned overflow, the result
5350 returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the
5351 bit width of the result.
5353 Because LLVM integers use a two's complement representation, and the
5354 result is the same width as the operands, this instruction returns the
5355 correct result for both signed and unsigned integers. If a full product
5356 (e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be
5357 sign-extended or zero-extended as appropriate to the width of the full
5360 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
5361 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
5362 result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if
5363 unsigned and/or signed overflow, respectively, occurs.
5368 .. code-block:: llvm
5370 <result> = mul i32 4, %var ; yields i32:result = 4 * %var
5374 '``fmul``' Instruction
5375 ^^^^^^^^^^^^^^^^^^^^^^
5382 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
5387 The '``fmul``' instruction returns the product of its two operands.
5392 The two arguments to the '``fmul``' instruction must be :ref:`floating
5393 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
5394 Both arguments must have identical types.
5399 The value produced is the floating point product of the two operands.
5400 This instruction can also take any number of :ref:`fast-math
5401 flags <fastmath>`, which are optimization hints to enable otherwise
5402 unsafe floating point optimizations:
5407 .. code-block:: llvm
5409 <result> = fmul float 4.0, %var ; yields float:result = 4.0 * %var
5411 '``udiv``' Instruction
5412 ^^^^^^^^^^^^^^^^^^^^^^
5419 <result> = udiv <ty> <op1>, <op2> ; yields ty:result
5420 <result> = udiv exact <ty> <op1>, <op2> ; yields ty:result
5425 The '``udiv``' instruction returns the quotient of its two operands.
5430 The two arguments to the '``udiv``' instruction must be
5431 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5432 arguments must have identical types.
5437 The value produced is the unsigned integer quotient of the two operands.
5439 Note that unsigned integer division and signed integer division are
5440 distinct operations; for signed integer division, use '``sdiv``'.
5442 Division by zero leads to undefined behavior.
5444 If the ``exact`` keyword is present, the result value of the ``udiv`` is
5445 a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as
5446 such, "((a udiv exact b) mul b) == a").
5451 .. code-block:: llvm
5453 <result> = udiv i32 4, %var ; yields i32:result = 4 / %var
5455 '``sdiv``' Instruction
5456 ^^^^^^^^^^^^^^^^^^^^^^
5463 <result> = sdiv <ty> <op1>, <op2> ; yields ty:result
5464 <result> = sdiv exact <ty> <op1>, <op2> ; yields ty:result
5469 The '``sdiv``' instruction returns the quotient of its two operands.
5474 The two arguments to the '``sdiv``' instruction must be
5475 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5476 arguments must have identical types.
5481 The value produced is the signed integer quotient of the two operands
5482 rounded towards zero.
5484 Note that signed integer division and unsigned integer division are
5485 distinct operations; for unsigned integer division, use '``udiv``'.
5487 Division by zero leads to undefined behavior. Overflow also leads to
5488 undefined behavior; this is a rare case, but can occur, for example, by
5489 doing a 32-bit division of -2147483648 by -1.
5491 If the ``exact`` keyword is present, the result value of the ``sdiv`` is
5492 a :ref:`poison value <poisonvalues>` if the result would be rounded.
5497 .. code-block:: llvm
5499 <result> = sdiv i32 4, %var ; yields i32:result = 4 / %var
5503 '``fdiv``' Instruction
5504 ^^^^^^^^^^^^^^^^^^^^^^
5511 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
5516 The '``fdiv``' instruction returns the quotient of its two operands.
5521 The two arguments to the '``fdiv``' instruction must be :ref:`floating
5522 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
5523 Both arguments must have identical types.
5528 The value produced is the floating point quotient of the two operands.
5529 This instruction can also take any number of :ref:`fast-math
5530 flags <fastmath>`, which are optimization hints to enable otherwise
5531 unsafe floating point optimizations:
5536 .. code-block:: llvm
5538 <result> = fdiv float 4.0, %var ; yields float:result = 4.0 / %var
5540 '``urem``' Instruction
5541 ^^^^^^^^^^^^^^^^^^^^^^
5548 <result> = urem <ty> <op1>, <op2> ; yields ty:result
5553 The '``urem``' instruction returns the remainder from the unsigned
5554 division of its two arguments.
5559 The two arguments to the '``urem``' instruction must be
5560 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5561 arguments must have identical types.
5566 This instruction returns the unsigned integer *remainder* of a division.
5567 This instruction always performs an unsigned division to get the
5570 Note that unsigned integer remainder and signed integer remainder are
5571 distinct operations; for signed integer remainder, use '``srem``'.
5573 Taking the remainder of a division by zero leads to undefined behavior.
5578 .. code-block:: llvm
5580 <result> = urem i32 4, %var ; yields i32:result = 4 % %var
5582 '``srem``' Instruction
5583 ^^^^^^^^^^^^^^^^^^^^^^
5590 <result> = srem <ty> <op1>, <op2> ; yields ty:result
5595 The '``srem``' instruction returns the remainder from the signed
5596 division of its two operands. This instruction can also take
5597 :ref:`vector <t_vector>` versions of the values in which case the elements
5603 The two arguments to the '``srem``' instruction must be
5604 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5605 arguments must have identical types.
5610 This instruction returns the *remainder* of a division (where the result
5611 is either zero or has the same sign as the dividend, ``op1``), not the
5612 *modulo* operator (where the result is either zero or has the same sign
5613 as the divisor, ``op2``) of a value. For more information about the
5614 difference, see `The Math
5615 Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a
5616 table of how this is implemented in various languages, please see
5618 operation <http://en.wikipedia.org/wiki/Modulo_operation>`_.
5620 Note that signed integer remainder and unsigned integer remainder are
5621 distinct operations; for unsigned integer remainder, use '``urem``'.
5623 Taking the remainder of a division by zero leads to undefined behavior.
5624 Overflow also leads to undefined behavior; this is a rare case, but can
5625 occur, for example, by taking the remainder of a 32-bit division of
5626 -2147483648 by -1. (The remainder doesn't actually overflow, but this
5627 rule lets srem be implemented using instructions that return both the
5628 result of the division and the remainder.)
5633 .. code-block:: llvm
5635 <result> = srem i32 4, %var ; yields i32:result = 4 % %var
5639 '``frem``' Instruction
5640 ^^^^^^^^^^^^^^^^^^^^^^
5647 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
5652 The '``frem``' instruction returns the remainder from the division of
5658 The two arguments to the '``frem``' instruction must be :ref:`floating
5659 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
5660 Both arguments must have identical types.
5665 This instruction returns the *remainder* of a division. The remainder
5666 has the same sign as the dividend. This instruction can also take any
5667 number of :ref:`fast-math flags <fastmath>`, which are optimization hints
5668 to enable otherwise unsafe floating point optimizations:
5673 .. code-block:: llvm
5675 <result> = frem float 4.0, %var ; yields float:result = 4.0 % %var
5679 Bitwise Binary Operations
5680 -------------------------
5682 Bitwise binary operators are used to do various forms of bit-twiddling
5683 in a program. They are generally very efficient instructions and can
5684 commonly be strength reduced from other instructions. They require two
5685 operands of the same type, execute an operation on them, and produce a
5686 single value. The resulting value is the same type as its operands.
5688 '``shl``' Instruction
5689 ^^^^^^^^^^^^^^^^^^^^^
5696 <result> = shl <ty> <op1>, <op2> ; yields ty:result
5697 <result> = shl nuw <ty> <op1>, <op2> ; yields ty:result
5698 <result> = shl nsw <ty> <op1>, <op2> ; yields ty:result
5699 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields ty:result
5704 The '``shl``' instruction returns the first operand shifted to the left
5705 a specified number of bits.
5710 Both arguments to the '``shl``' instruction must be the same
5711 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
5712 '``op2``' is treated as an unsigned value.
5717 The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`,
5718 where ``n`` is the width of the result. If ``op2`` is (statically or
5719 dynamically) equal to or larger than the number of bits in
5720 ``op1``, the result is undefined. If the arguments are vectors, each
5721 vector element of ``op1`` is shifted by the corresponding shift amount
5724 If the ``nuw`` keyword is present, then the shift produces a :ref:`poison
5725 value <poisonvalues>` if it shifts out any non-zero bits. If the
5726 ``nsw`` keyword is present, then the shift produces a :ref:`poison
5727 value <poisonvalues>` if it shifts out any bits that disagree with the
5728 resultant sign bit. As such, NUW/NSW have the same semantics as they
5729 would if the shift were expressed as a mul instruction with the same
5730 nsw/nuw bits in (mul %op1, (shl 1, %op2)).
5735 .. code-block:: llvm
5737 <result> = shl i32 4, %var ; yields i32: 4 << %var
5738 <result> = shl i32 4, 2 ; yields i32: 16
5739 <result> = shl i32 1, 10 ; yields i32: 1024
5740 <result> = shl i32 1, 32 ; undefined
5741 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4>
5743 '``lshr``' Instruction
5744 ^^^^^^^^^^^^^^^^^^^^^^
5751 <result> = lshr <ty> <op1>, <op2> ; yields ty:result
5752 <result> = lshr exact <ty> <op1>, <op2> ; yields ty:result
5757 The '``lshr``' instruction (logical shift right) returns the first
5758 operand shifted to the right a specified number of bits with zero fill.
5763 Both arguments to the '``lshr``' instruction must be the same
5764 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
5765 '``op2``' is treated as an unsigned value.
5770 This instruction always performs a logical shift right operation. The
5771 most significant bits of the result will be filled with zero bits after
5772 the shift. If ``op2`` is (statically or dynamically) equal to or larger
5773 than the number of bits in ``op1``, the result is undefined. If the
5774 arguments are vectors, each vector element of ``op1`` is shifted by the
5775 corresponding shift amount in ``op2``.
5777 If the ``exact`` keyword is present, the result value of the ``lshr`` is
5778 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
5784 .. code-block:: llvm
5786 <result> = lshr i32 4, 1 ; yields i32:result = 2
5787 <result> = lshr i32 4, 2 ; yields i32:result = 1
5788 <result> = lshr i8 4, 3 ; yields i8:result = 0
5789 <result> = lshr i8 -2, 1 ; yields i8:result = 0x7F
5790 <result> = lshr i32 1, 32 ; undefined
5791 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1>
5793 '``ashr``' Instruction
5794 ^^^^^^^^^^^^^^^^^^^^^^
5801 <result> = ashr <ty> <op1>, <op2> ; yields ty:result
5802 <result> = ashr exact <ty> <op1>, <op2> ; yields ty:result
5807 The '``ashr``' instruction (arithmetic shift right) returns the first
5808 operand shifted to the right a specified number of bits with sign
5814 Both arguments to the '``ashr``' instruction must be the same
5815 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
5816 '``op2``' is treated as an unsigned value.
5821 This instruction always performs an arithmetic shift right operation,
5822 The most significant bits of the result will be filled with the sign bit
5823 of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger
5824 than the number of bits in ``op1``, the result is undefined. If the
5825 arguments are vectors, each vector element of ``op1`` is shifted by the
5826 corresponding shift amount in ``op2``.
5828 If the ``exact`` keyword is present, the result value of the ``ashr`` is
5829 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
5835 .. code-block:: llvm
5837 <result> = ashr i32 4, 1 ; yields i32:result = 2
5838 <result> = ashr i32 4, 2 ; yields i32:result = 1
5839 <result> = ashr i8 4, 3 ; yields i8:result = 0
5840 <result> = ashr i8 -2, 1 ; yields i8:result = -1
5841 <result> = ashr i32 1, 32 ; undefined
5842 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0>
5844 '``and``' Instruction
5845 ^^^^^^^^^^^^^^^^^^^^^
5852 <result> = and <ty> <op1>, <op2> ; yields ty:result
5857 The '``and``' instruction returns the bitwise logical and of its two
5863 The two arguments to the '``and``' instruction must be
5864 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5865 arguments must have identical types.
5870 The truth table used for the '``and``' instruction is:
5887 .. code-block:: llvm
5889 <result> = and i32 4, %var ; yields i32:result = 4 & %var
5890 <result> = and i32 15, 40 ; yields i32:result = 8
5891 <result> = and i32 4, 8 ; yields i32:result = 0
5893 '``or``' Instruction
5894 ^^^^^^^^^^^^^^^^^^^^
5901 <result> = or <ty> <op1>, <op2> ; yields ty:result
5906 The '``or``' instruction returns the bitwise logical inclusive or of its
5912 The two arguments to the '``or``' instruction must be
5913 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5914 arguments must have identical types.
5919 The truth table used for the '``or``' instruction is:
5938 <result> = or i32 4, %var ; yields i32:result = 4 | %var
5939 <result> = or i32 15, 40 ; yields i32:result = 47
5940 <result> = or i32 4, 8 ; yields i32:result = 12
5942 '``xor``' Instruction
5943 ^^^^^^^^^^^^^^^^^^^^^
5950 <result> = xor <ty> <op1>, <op2> ; yields ty:result
5955 The '``xor``' instruction returns the bitwise logical exclusive or of
5956 its two operands. The ``xor`` is used to implement the "one's
5957 complement" operation, which is the "~" operator in C.
5962 The two arguments to the '``xor``' instruction must be
5963 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5964 arguments must have identical types.
5969 The truth table used for the '``xor``' instruction is:
5986 .. code-block:: llvm
5988 <result> = xor i32 4, %var ; yields i32:result = 4 ^ %var
5989 <result> = xor i32 15, 40 ; yields i32:result = 39
5990 <result> = xor i32 4, 8 ; yields i32:result = 12
5991 <result> = xor i32 %V, -1 ; yields i32:result = ~%V
5996 LLVM supports several instructions to represent vector operations in a
5997 target-independent manner. These instructions cover the element-access
5998 and vector-specific operations needed to process vectors effectively.
5999 While LLVM does directly support these vector operations, many
6000 sophisticated algorithms will want to use target-specific intrinsics to
6001 take full advantage of a specific target.
6003 .. _i_extractelement:
6005 '``extractelement``' Instruction
6006 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6013 <result> = extractelement <n x <ty>> <val>, <ty2> <idx> ; yields <ty>
6018 The '``extractelement``' instruction extracts a single scalar element
6019 from a vector at a specified index.
6024 The first operand of an '``extractelement``' instruction is a value of
6025 :ref:`vector <t_vector>` type. The second operand is an index indicating
6026 the position from which to extract the element. The index may be a
6027 variable of any integer type.
6032 The result is a scalar of the same type as the element type of ``val``.
6033 Its value is the value at position ``idx`` of ``val``. If ``idx``
6034 exceeds the length of ``val``, the results are undefined.
6039 .. code-block:: llvm
6041 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32
6043 .. _i_insertelement:
6045 '``insertelement``' Instruction
6046 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6053 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <n x <ty>>
6058 The '``insertelement``' instruction inserts a scalar element into a
6059 vector at a specified index.
6064 The first operand of an '``insertelement``' instruction is a value of
6065 :ref:`vector <t_vector>` type. The second operand is a scalar value whose
6066 type must equal the element type of the first operand. The third operand
6067 is an index indicating the position at which to insert the value. The
6068 index may be a variable of any integer type.
6073 The result is a vector of the same type as ``val``. Its element values
6074 are those of ``val`` except at position ``idx``, where it gets the value
6075 ``elt``. If ``idx`` exceeds the length of ``val``, the results are
6081 .. code-block:: llvm
6083 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32>
6085 .. _i_shufflevector:
6087 '``shufflevector``' Instruction
6088 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6095 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>>
6100 The '``shufflevector``' instruction constructs a permutation of elements
6101 from two input vectors, returning a vector with the same element type as
6102 the input and length that is the same as the shuffle mask.
6107 The first two operands of a '``shufflevector``' instruction are vectors
6108 with the same type. The third argument is a shuffle mask whose element
6109 type is always 'i32'. The result of the instruction is a vector whose
6110 length is the same as the shuffle mask and whose element type is the
6111 same as the element type of the first two operands.
6113 The shuffle mask operand is required to be a constant vector with either
6114 constant integer or undef values.
6119 The elements of the two input vectors are numbered from left to right
6120 across both of the vectors. The shuffle mask operand specifies, for each
6121 element of the result vector, which element of the two input vectors the
6122 result element gets. The element selector may be undef (meaning "don't
6123 care") and the second operand may be undef if performing a shuffle from
6129 .. code-block:: llvm
6131 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
6132 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32>
6133 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
6134 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle.
6135 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
6136 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32>
6137 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
6138 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32>
6140 Aggregate Operations
6141 --------------------
6143 LLVM supports several instructions for working with
6144 :ref:`aggregate <t_aggregate>` values.
6148 '``extractvalue``' Instruction
6149 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6156 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
6161 The '``extractvalue``' instruction extracts the value of a member field
6162 from an :ref:`aggregate <t_aggregate>` value.
6167 The first operand of an '``extractvalue``' instruction is a value of
6168 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The operands are
6169 constant indices to specify which value to extract in a similar manner
6170 as indices in a '``getelementptr``' instruction.
6172 The major differences to ``getelementptr`` indexing are:
6174 - Since the value being indexed is not a pointer, the first index is
6175 omitted and assumed to be zero.
6176 - At least one index must be specified.
6177 - Not only struct indices but also array indices must be in bounds.
6182 The result is the value at the position in the aggregate specified by
6188 .. code-block:: llvm
6190 <result> = extractvalue {i32, float} %agg, 0 ; yields i32
6194 '``insertvalue``' Instruction
6195 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6202 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type>
6207 The '``insertvalue``' instruction inserts a value into a member field in
6208 an :ref:`aggregate <t_aggregate>` value.
6213 The first operand of an '``insertvalue``' instruction is a value of
6214 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is
6215 a first-class value to insert. The following operands are constant
6216 indices indicating the position at which to insert the value in a
6217 similar manner as indices in a '``extractvalue``' instruction. The value
6218 to insert must have the same type as the value identified by the
6224 The result is an aggregate of the same type as ``val``. Its value is
6225 that of ``val`` except that the value at the position specified by the
6226 indices is that of ``elt``.
6231 .. code-block:: llvm
6233 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef}
6234 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val}
6235 %agg3 = insertvalue {i32, {float}} undef, float %val, 1, 0 ; yields {i32 undef, {float %val}}
6239 Memory Access and Addressing Operations
6240 ---------------------------------------
6242 A key design point of an SSA-based representation is how it represents
6243 memory. In LLVM, no memory locations are in SSA form, which makes things
6244 very simple. This section describes how to read, write, and allocate
6249 '``alloca``' Instruction
6250 ^^^^^^^^^^^^^^^^^^^^^^^^
6257 <result> = alloca [inalloca] <type> [, <ty> <NumElements>] [, align <alignment>] ; yields type*:result
6262 The '``alloca``' instruction allocates memory on the stack frame of the
6263 currently executing function, to be automatically released when this
6264 function returns to its caller. The object is always allocated in the
6265 generic address space (address space zero).
6270 The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements``
6271 bytes of memory on the runtime stack, returning a pointer of the
6272 appropriate type to the program. If "NumElements" is specified, it is
6273 the number of elements allocated, otherwise "NumElements" is defaulted
6274 to be one. If a constant alignment is specified, the value result of the
6275 allocation is guaranteed to be aligned to at least that boundary. The
6276 alignment may not be greater than ``1 << 29``. If not specified, or if
6277 zero, the target can choose to align the allocation on any convenient
6278 boundary compatible with the type.
6280 '``type``' may be any sized type.
6285 Memory is allocated; a pointer is returned. The operation is undefined
6286 if there is insufficient stack space for the allocation. '``alloca``'d
6287 memory is automatically released when the function returns. The
6288 '``alloca``' instruction is commonly used to represent automatic
6289 variables that must have an address available. When the function returns
6290 (either with the ``ret`` or ``resume`` instructions), the memory is
6291 reclaimed. Allocating zero bytes is legal, but the result is undefined.
6292 The order in which memory is allocated (ie., which way the stack grows)
6298 .. code-block:: llvm
6300 %ptr = alloca i32 ; yields i32*:ptr
6301 %ptr = alloca i32, i32 4 ; yields i32*:ptr
6302 %ptr = alloca i32, i32 4, align 1024 ; yields i32*:ptr
6303 %ptr = alloca i32, align 1024 ; yields i32*:ptr
6307 '``load``' Instruction
6308 ^^^^^^^^^^^^^^^^^^^^^^
6315 <result> = load [volatile] <ty>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>][, !nonnull !<index>][, !dereferenceable !<index>][, !dereferenceable_or_null !<index>]
6316 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
6317 !<index> = !{ i32 1 }
6322 The '``load``' instruction is used to read from memory.
6327 The argument to the ``load`` instruction specifies the memory address
6328 from which to load. The type specified must be a :ref:`first
6329 class <t_firstclass>` type. If the ``load`` is marked as ``volatile``,
6330 then the optimizer is not allowed to modify the number or order of
6331 execution of this ``load`` with other :ref:`volatile
6332 operations <volatile>`.
6334 If the ``load`` is marked as ``atomic``, it takes an extra
6335 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
6336 ``release`` and ``acq_rel`` orderings are not valid on ``load``
6337 instructions. Atomic loads produce :ref:`defined <memmodel>` results
6338 when they may see multiple atomic stores. The type of the pointee must
6339 be an integer type whose bit width is a power of two greater than or
6340 equal to eight and less than or equal to a target-specific size limit.
6341 ``align`` must be explicitly specified on atomic loads, and the load has
6342 undefined behavior if the alignment is not set to a value which is at
6343 least the size in bytes of the pointee. ``!nontemporal`` does not have
6344 any defined semantics for atomic loads.
6346 The optional constant ``align`` argument specifies the alignment of the
6347 operation (that is, the alignment of the memory address). A value of 0
6348 or an omitted ``align`` argument means that the operation has the ABI
6349 alignment for the target. It is the responsibility of the code emitter
6350 to ensure that the alignment information is correct. Overestimating the
6351 alignment results in undefined behavior. Underestimating the alignment
6352 may produce less efficient code. An alignment of 1 is always safe. The
6353 maximum possible alignment is ``1 << 29``.
6355 The optional ``!nontemporal`` metadata must reference a single
6356 metadata name ``<index>`` corresponding to a metadata node with one
6357 ``i32`` entry of value 1. The existence of the ``!nontemporal``
6358 metadata on the instruction tells the optimizer and code generator
6359 that this load is not expected to be reused in the cache. The code
6360 generator may select special instructions to save cache bandwidth, such
6361 as the ``MOVNT`` instruction on x86.
6363 The optional ``!invariant.load`` metadata must reference a single
6364 metadata name ``<index>`` corresponding to a metadata node with no
6365 entries. The existence of the ``!invariant.load`` metadata on the
6366 instruction tells the optimizer and code generator that the address
6367 operand to this load points to memory which can be assumed unchanged.
6368 Being invariant does not imply that a location is dereferenceable,
6369 but it does imply that once the location is known dereferenceable
6370 its value is henceforth unchanging.
6372 The optional ``!nonnull`` metadata must reference a single
6373 metadata name ``<index>`` corresponding to a metadata node with no
6374 entries. The existence of the ``!nonnull`` metadata on the
6375 instruction tells the optimizer that the value loaded is known to
6376 never be null. This is analogous to the ''nonnull'' attribute
6377 on parameters and return values. This metadata can only be applied
6378 to loads of a pointer type.
6380 The optional ``!dereferenceable`` metadata must reference a single
6381 metadata name ``<index>`` corresponding to a metadata node with one ``i64``
6382 entry. The existence of the ``!dereferenceable`` metadata on the instruction
6383 tells the optimizer that the value loaded is known to be dereferenceable.
6384 The number of bytes known to be dereferenceable is specified by the integer
6385 value in the metadata node. This is analogous to the ''dereferenceable''
6386 attribute on parameters and return values. This metadata can only be applied
6387 to loads of a pointer type.
6389 The optional ``!dereferenceable_or_null`` metadata must reference a single
6390 metadata name ``<index>`` corresponding to a metadata node with one ``i64``
6391 entry. The existence of the ``!dereferenceable_or_null`` metadata on the
6392 instruction tells the optimizer that the value loaded is known to be either
6393 dereferenceable or null.
6394 The number of bytes known to be dereferenceable is specified by the integer
6395 value in the metadata node. This is analogous to the ''dereferenceable_or_null''
6396 attribute on parameters and return values. This metadata can only be applied
6397 to loads of a pointer type.
6402 The location of memory pointed to is loaded. If the value being loaded
6403 is of scalar type then the number of bytes read does not exceed the
6404 minimum number of bytes needed to hold all bits of the type. For
6405 example, loading an ``i24`` reads at most three bytes. When loading a
6406 value of a type like ``i20`` with a size that is not an integral number
6407 of bytes, the result is undefined if the value was not originally
6408 written using a store of the same type.
6413 .. code-block:: llvm
6415 %ptr = alloca i32 ; yields i32*:ptr
6416 store i32 3, i32* %ptr ; yields void
6417 %val = load i32, i32* %ptr ; yields i32:val = i32 3
6421 '``store``' Instruction
6422 ^^^^^^^^^^^^^^^^^^^^^^^
6429 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] ; yields void
6430 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> ; yields void
6435 The '``store``' instruction is used to write to memory.
6440 There are two arguments to the ``store`` instruction: a value to store
6441 and an address at which to store it. The type of the ``<pointer>``
6442 operand must be a pointer to the :ref:`first class <t_firstclass>` type of
6443 the ``<value>`` operand. If the ``store`` is marked as ``volatile``,
6444 then the optimizer is not allowed to modify the number or order of
6445 execution of this ``store`` with other :ref:`volatile
6446 operations <volatile>`.
6448 If the ``store`` is marked as ``atomic``, it takes an extra
6449 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
6450 ``acquire`` and ``acq_rel`` orderings aren't valid on ``store``
6451 instructions. Atomic loads produce :ref:`defined <memmodel>` results
6452 when they may see multiple atomic stores. The type of the pointee must
6453 be an integer type whose bit width is a power of two greater than or
6454 equal to eight and less than or equal to a target-specific size limit.
6455 ``align`` must be explicitly specified on atomic stores, and the store
6456 has undefined behavior if the alignment is not set to a value which is
6457 at least the size in bytes of the pointee. ``!nontemporal`` does not
6458 have any defined semantics for atomic stores.
6460 The optional constant ``align`` argument specifies the alignment of the
6461 operation (that is, the alignment of the memory address). A value of 0
6462 or an omitted ``align`` argument means that the operation has the ABI
6463 alignment for the target. It is the responsibility of the code emitter
6464 to ensure that the alignment information is correct. Overestimating the
6465 alignment results in undefined behavior. Underestimating the
6466 alignment may produce less efficient code. An alignment of 1 is always
6467 safe. The maximum possible alignment is ``1 << 29``.
6469 The optional ``!nontemporal`` metadata must reference a single metadata
6470 name ``<index>`` corresponding to a metadata node with one ``i32`` entry of
6471 value 1. The existence of the ``!nontemporal`` metadata on the instruction
6472 tells the optimizer and code generator that this load is not expected to
6473 be reused in the cache. The code generator may select special
6474 instructions to save cache bandwidth, such as the MOVNT instruction on
6480 The contents of memory are updated to contain ``<value>`` at the
6481 location specified by the ``<pointer>`` operand. If ``<value>`` is
6482 of scalar type then the number of bytes written does not exceed the
6483 minimum number of bytes needed to hold all bits of the type. For
6484 example, storing an ``i24`` writes at most three bytes. When writing a
6485 value of a type like ``i20`` with a size that is not an integral number
6486 of bytes, it is unspecified what happens to the extra bits that do not
6487 belong to the type, but they will typically be overwritten.
6492 .. code-block:: llvm
6494 %ptr = alloca i32 ; yields i32*:ptr
6495 store i32 3, i32* %ptr ; yields void
6496 %val = load i32* %ptr ; yields i32:val = i32 3
6500 '``fence``' Instruction
6501 ^^^^^^^^^^^^^^^^^^^^^^^
6508 fence [singlethread] <ordering> ; yields void
6513 The '``fence``' instruction is used to introduce happens-before edges
6519 '``fence``' instructions take an :ref:`ordering <ordering>` argument which
6520 defines what *synchronizes-with* edges they add. They can only be given
6521 ``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings.
6526 A fence A which has (at least) ``release`` ordering semantics
6527 *synchronizes with* a fence B with (at least) ``acquire`` ordering
6528 semantics if and only if there exist atomic operations X and Y, both
6529 operating on some atomic object M, such that A is sequenced before X, X
6530 modifies M (either directly or through some side effect of a sequence
6531 headed by X), Y is sequenced before B, and Y observes M. This provides a
6532 *happens-before* dependency between A and B. Rather than an explicit
6533 ``fence``, one (but not both) of the atomic operations X or Y might
6534 provide a ``release`` or ``acquire`` (resp.) ordering constraint and
6535 still *synchronize-with* the explicit ``fence`` and establish the
6536 *happens-before* edge.
6538 A ``fence`` which has ``seq_cst`` ordering, in addition to having both
6539 ``acquire`` and ``release`` semantics specified above, participates in
6540 the global program order of other ``seq_cst`` operations and/or fences.
6542 The optional ":ref:`singlethread <singlethread>`" argument specifies
6543 that the fence only synchronizes with other fences in the same thread.
6544 (This is useful for interacting with signal handlers.)
6549 .. code-block:: llvm
6551 fence acquire ; yields void
6552 fence singlethread seq_cst ; yields void
6556 '``cmpxchg``' Instruction
6557 ^^^^^^^^^^^^^^^^^^^^^^^^^
6564 cmpxchg [weak] [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <success ordering> <failure ordering> ; yields { ty, i1 }
6569 The '``cmpxchg``' instruction is used to atomically modify memory. It
6570 loads a value in memory and compares it to a given value. If they are
6571 equal, it tries to store a new value into the memory.
6576 There are three arguments to the '``cmpxchg``' instruction: an address
6577 to operate on, a value to compare to the value currently be at that
6578 address, and a new value to place at that address if the compared values
6579 are equal. The type of '<cmp>' must be an integer type whose bit width
6580 is a power of two greater than or equal to eight and less than or equal
6581 to a target-specific size limit. '<cmp>' and '<new>' must have the same
6582 type, and the type of '<pointer>' must be a pointer to that type. If the
6583 ``cmpxchg`` is marked as ``volatile``, then the optimizer is not allowed
6584 to modify the number or order of execution of this ``cmpxchg`` with
6585 other :ref:`volatile operations <volatile>`.
6587 The success and failure :ref:`ordering <ordering>` arguments specify how this
6588 ``cmpxchg`` synchronizes with other atomic operations. Both ordering parameters
6589 must be at least ``monotonic``, the ordering constraint on failure must be no
6590 stronger than that on success, and the failure ordering cannot be either
6591 ``release`` or ``acq_rel``.
6593 The optional "``singlethread``" argument declares that the ``cmpxchg``
6594 is only atomic with respect to code (usually signal handlers) running in
6595 the same thread as the ``cmpxchg``. Otherwise the cmpxchg is atomic with
6596 respect to all other code in the system.
6598 The pointer passed into cmpxchg must have alignment greater than or
6599 equal to the size in memory of the operand.
6604 The contents of memory at the location specified by the '``<pointer>``' operand
6605 is read and compared to '``<cmp>``'; if the read value is the equal, the
6606 '``<new>``' is written. The original value at the location is returned, together
6607 with a flag indicating success (true) or failure (false).
6609 If the cmpxchg operation is marked as ``weak`` then a spurious failure is
6610 permitted: the operation may not write ``<new>`` even if the comparison
6613 If the cmpxchg operation is strong (the default), the i1 value is 1 if and only
6614 if the value loaded equals ``cmp``.
6616 A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of
6617 identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic
6618 load with an ordering parameter determined the second ordering parameter.
6623 .. code-block:: llvm
6626 %orig = atomic load i32, i32* %ptr unordered ; yields i32
6630 %cmp = phi i32 [ %orig, %entry ], [%old, %loop]
6631 %squared = mul i32 %cmp, %cmp
6632 %val_success = cmpxchg i32* %ptr, i32 %cmp, i32 %squared acq_rel monotonic ; yields { i32, i1 }
6633 %value_loaded = extractvalue { i32, i1 } %val_success, 0
6634 %success = extractvalue { i32, i1 } %val_success, 1
6635 br i1 %success, label %done, label %loop
6642 '``atomicrmw``' Instruction
6643 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
6650 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> ; yields ty
6655 The '``atomicrmw``' instruction is used to atomically modify memory.
6660 There are three arguments to the '``atomicrmw``' instruction: an
6661 operation to apply, an address whose value to modify, an argument to the
6662 operation. The operation must be one of the following keywords:
6676 The type of '<value>' must be an integer type whose bit width is a power
6677 of two greater than or equal to eight and less than or equal to a
6678 target-specific size limit. The type of the '``<pointer>``' operand must
6679 be a pointer to that type. If the ``atomicrmw`` is marked as
6680 ``volatile``, then the optimizer is not allowed to modify the number or
6681 order of execution of this ``atomicrmw`` with other :ref:`volatile
6682 operations <volatile>`.
6687 The contents of memory at the location specified by the '``<pointer>``'
6688 operand are atomically read, modified, and written back. The original
6689 value at the location is returned. The modification is specified by the
6692 - xchg: ``*ptr = val``
6693 - add: ``*ptr = *ptr + val``
6694 - sub: ``*ptr = *ptr - val``
6695 - and: ``*ptr = *ptr & val``
6696 - nand: ``*ptr = ~(*ptr & val)``
6697 - or: ``*ptr = *ptr | val``
6698 - xor: ``*ptr = *ptr ^ val``
6699 - max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison)
6700 - min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison)
6701 - umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned
6703 - umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned
6709 .. code-block:: llvm
6711 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields i32
6713 .. _i_getelementptr:
6715 '``getelementptr``' Instruction
6716 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6723 <result> = getelementptr <ty>, <ty>* <ptrval>{, <ty> <idx>}*
6724 <result> = getelementptr inbounds <ty>, <ty>* <ptrval>{, <ty> <idx>}*
6725 <result> = getelementptr <ty>, <ptr vector> <ptrval>, <vector index type> <idx>
6730 The '``getelementptr``' instruction is used to get the address of a
6731 subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs
6732 address calculation only and does not access memory. The instruction can also
6733 be used to calculate a vector of such addresses.
6738 The first argument is always a type used as the basis for the calculations.
6739 The second argument is always a pointer or a vector of pointers, and is the
6740 base address to start from. The remaining arguments are indices
6741 that indicate which of the elements of the aggregate object are indexed.
6742 The interpretation of each index is dependent on the type being indexed
6743 into. The first index always indexes the pointer value given as the
6744 first argument, the second index indexes a value of the type pointed to
6745 (not necessarily the value directly pointed to, since the first index
6746 can be non-zero), etc. The first type indexed into must be a pointer
6747 value, subsequent types can be arrays, vectors, and structs. Note that
6748 subsequent types being indexed into can never be pointers, since that
6749 would require loading the pointer before continuing calculation.
6751 The type of each index argument depends on the type it is indexing into.
6752 When indexing into a (optionally packed) structure, only ``i32`` integer
6753 **constants** are allowed (when using a vector of indices they must all
6754 be the **same** ``i32`` integer constant). When indexing into an array,
6755 pointer or vector, integers of any width are allowed, and they are not
6756 required to be constant. These integers are treated as signed values
6759 For example, let's consider a C code fragment and how it gets compiled
6775 int *foo(struct ST *s) {
6776 return &s[1].Z.B[5][13];
6779 The LLVM code generated by Clang is:
6781 .. code-block:: llvm
6783 %struct.RT = type { i8, [10 x [20 x i32]], i8 }
6784 %struct.ST = type { i32, double, %struct.RT }
6786 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
6788 %arrayidx = getelementptr inbounds %struct.ST, %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
6795 In the example above, the first index is indexing into the
6796 '``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``'
6797 = '``{ i32, double, %struct.RT }``' type, a structure. The second index
6798 indexes into the third element of the structure, yielding a
6799 '``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another
6800 structure. The third index indexes into the second element of the
6801 structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two
6802 dimensions of the array are subscripted into, yielding an '``i32``'
6803 type. The '``getelementptr``' instruction returns a pointer to this
6804 element, thus computing a value of '``i32*``' type.
6806 Note that it is perfectly legal to index partially through a structure,
6807 returning a pointer to an inner element. Because of this, the LLVM code
6808 for the given testcase is equivalent to:
6810 .. code-block:: llvm
6812 define i32* @foo(%struct.ST* %s) {
6813 %t1 = getelementptr %struct.ST, %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1
6814 %t2 = getelementptr %struct.ST, %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2
6815 %t3 = getelementptr %struct.RT, %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3
6816 %t4 = getelementptr [10 x [20 x i32]], [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4
6817 %t5 = getelementptr [20 x i32], [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5
6821 If the ``inbounds`` keyword is present, the result value of the
6822 ``getelementptr`` is a :ref:`poison value <poisonvalues>` if the base
6823 pointer is not an *in bounds* address of an allocated object, or if any
6824 of the addresses that would be formed by successive addition of the
6825 offsets implied by the indices to the base address with infinitely
6826 precise signed arithmetic are not an *in bounds* address of that
6827 allocated object. The *in bounds* addresses for an allocated object are
6828 all the addresses that point into the object, plus the address one byte
6829 past the end. In cases where the base is a vector of pointers the
6830 ``inbounds`` keyword applies to each of the computations element-wise.
6832 If the ``inbounds`` keyword is not present, the offsets are added to the
6833 base address with silently-wrapping two's complement arithmetic. If the
6834 offsets have a different width from the pointer, they are sign-extended
6835 or truncated to the width of the pointer. The result value of the
6836 ``getelementptr`` may be outside the object pointed to by the base
6837 pointer. The result value may not necessarily be used to access memory
6838 though, even if it happens to point into allocated storage. See the
6839 :ref:`Pointer Aliasing Rules <pointeraliasing>` section for more
6842 The getelementptr instruction is often confusing. For some more insight
6843 into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`.
6848 .. code-block:: llvm
6850 ; yields [12 x i8]*:aptr
6851 %aptr = getelementptr {i32, [12 x i8]}, {i32, [12 x i8]}* %saptr, i64 0, i32 1
6853 %vptr = getelementptr {i32, <2 x i8>}, {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
6855 %eptr = getelementptr [12 x i8], [12 x i8]* %aptr, i64 0, i32 1
6857 %iptr = getelementptr [10 x i32], [10 x i32]* @arr, i16 0, i16 0
6862 The ``getelementptr`` returns a vector of pointers, instead of a single address,
6863 when one or more of its arguments is a vector. In such cases, all vector
6864 arguments should have the same number of elements, and every scalar argument
6865 will be effectively broadcast into a vector during address calculation.
6867 .. code-block:: llvm
6869 ; All arguments are vectors:
6870 ; A[i] = ptrs[i] + offsets[i]*sizeof(i8)
6871 %A = getelementptr i8, <4 x i8*> %ptrs, <4 x i64> %offsets
6873 ; Add the same scalar offset to each pointer of a vector:
6874 ; A[i] = ptrs[i] + offset*sizeof(i8)
6875 %A = getelementptr i8, <4 x i8*> %ptrs, i64 %offset
6877 ; Add distinct offsets to the same pointer:
6878 ; A[i] = ptr + offsets[i]*sizeof(i8)
6879 %A = getelementptr i8, i8* %ptr, <4 x i64> %offsets
6881 ; In all cases described above the type of the result is <4 x i8*>
6883 The two following instructions are equivalent:
6885 .. code-block:: llvm
6887 getelementptr %struct.ST, <4 x %struct.ST*> %s, <4 x i64> %ind1,
6888 <4 x i32> <i32 2, i32 2, i32 2, i32 2>,
6889 <4 x i32> <i32 1, i32 1, i32 1, i32 1>,
6891 <4 x i64> <i64 13, i64 13, i64 13, i64 13>
6893 getelementptr %struct.ST, <4 x %struct.ST*> %s, <4 x i64> %ind1,
6894 i32 2, i32 1, <4 x i32> %ind4, i64 13
6896 Let's look at the C code, where the vector version of ``getelementptr``
6901 // Let's assume that we vectorize the following loop:
6902 double *A, B; int *C;
6903 for (int i = 0; i < size; ++i) {
6907 .. code-block:: llvm
6909 ; get pointers for 8 elements from array B
6910 %ptrs = getelementptr double, double* %B, <8 x i32> %C
6911 ; load 8 elements from array B into A
6912 %A = call <8 x double> @llvm.masked.gather.v8f64(<8 x double*> %ptrs,
6913 i32 8, <8 x i1> %mask, <8 x double> %passthru)
6915 Conversion Operations
6916 ---------------------
6918 The instructions in this category are the conversion instructions
6919 (casting) which all take a single operand and a type. They perform
6920 various bit conversions on the operand.
6922 '``trunc .. to``' Instruction
6923 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6930 <result> = trunc <ty> <value> to <ty2> ; yields ty2
6935 The '``trunc``' instruction truncates its operand to the type ``ty2``.
6940 The '``trunc``' instruction takes a value to trunc, and a type to trunc
6941 it to. Both types must be of :ref:`integer <t_integer>` types, or vectors
6942 of the same number of integers. The bit size of the ``value`` must be
6943 larger than the bit size of the destination type, ``ty2``. Equal sized
6944 types are not allowed.
6949 The '``trunc``' instruction truncates the high order bits in ``value``
6950 and converts the remaining bits to ``ty2``. Since the source size must
6951 be larger than the destination size, ``trunc`` cannot be a *no-op cast*.
6952 It will always truncate bits.
6957 .. code-block:: llvm
6959 %X = trunc i32 257 to i8 ; yields i8:1
6960 %Y = trunc i32 123 to i1 ; yields i1:true
6961 %Z = trunc i32 122 to i1 ; yields i1:false
6962 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7>
6964 '``zext .. to``' Instruction
6965 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6972 <result> = zext <ty> <value> to <ty2> ; yields ty2
6977 The '``zext``' instruction zero extends its operand to type ``ty2``.
6982 The '``zext``' instruction takes a value to cast, and a type to cast it
6983 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
6984 the same number of integers. The bit size of the ``value`` must be
6985 smaller than the bit size of the destination type, ``ty2``.
6990 The ``zext`` fills the high order bits of the ``value`` with zero bits
6991 until it reaches the size of the destination type, ``ty2``.
6993 When zero extending from i1, the result will always be either 0 or 1.
6998 .. code-block:: llvm
7000 %X = zext i32 257 to i64 ; yields i64:257
7001 %Y = zext i1 true to i32 ; yields i32:1
7002 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
7004 '``sext .. to``' Instruction
7005 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7012 <result> = sext <ty> <value> to <ty2> ; yields ty2
7017 The '``sext``' sign extends ``value`` to the type ``ty2``.
7022 The '``sext``' instruction takes a value to cast, and a type to cast it
7023 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
7024 the same number of integers. The bit size of the ``value`` must be
7025 smaller than the bit size of the destination type, ``ty2``.
7030 The '``sext``' instruction performs a sign extension by copying the sign
7031 bit (highest order bit) of the ``value`` until it reaches the bit size
7032 of the type ``ty2``.
7034 When sign extending from i1, the extension always results in -1 or 0.
7039 .. code-block:: llvm
7041 %X = sext i8 -1 to i16 ; yields i16 :65535
7042 %Y = sext i1 true to i32 ; yields i32:-1
7043 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
7045 '``fptrunc .. to``' Instruction
7046 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7053 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2
7058 The '``fptrunc``' instruction truncates ``value`` to type ``ty2``.
7063 The '``fptrunc``' instruction takes a :ref:`floating point <t_floating>`
7064 value to cast and a :ref:`floating point <t_floating>` type to cast it to.
7065 The size of ``value`` must be larger than the size of ``ty2``. This
7066 implies that ``fptrunc`` cannot be used to make a *no-op cast*.
7071 The '``fptrunc``' instruction truncates a ``value`` from a larger
7072 :ref:`floating point <t_floating>` type to a smaller :ref:`floating
7073 point <t_floating>` type. If the value cannot fit within the
7074 destination type, ``ty2``, then the results are undefined.
7079 .. code-block:: llvm
7081 %X = fptrunc double 123.0 to float ; yields float:123.0
7082 %Y = fptrunc double 1.0E+300 to float ; yields undefined
7084 '``fpext .. to``' Instruction
7085 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7092 <result> = fpext <ty> <value> to <ty2> ; yields ty2
7097 The '``fpext``' extends a floating point ``value`` to a larger floating
7103 The '``fpext``' instruction takes a :ref:`floating point <t_floating>`
7104 ``value`` to cast, and a :ref:`floating point <t_floating>` type to cast it
7105 to. The source type must be smaller than the destination type.
7110 The '``fpext``' instruction extends the ``value`` from a smaller
7111 :ref:`floating point <t_floating>` type to a larger :ref:`floating
7112 point <t_floating>` type. The ``fpext`` cannot be used to make a
7113 *no-op cast* because it always changes bits. Use ``bitcast`` to make a
7114 *no-op cast* for a floating point cast.
7119 .. code-block:: llvm
7121 %X = fpext float 3.125 to double ; yields double:3.125000e+00
7122 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000
7124 '``fptoui .. to``' Instruction
7125 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7132 <result> = fptoui <ty> <value> to <ty2> ; yields ty2
7137 The '``fptoui``' converts a floating point ``value`` to its unsigned
7138 integer equivalent of type ``ty2``.
7143 The '``fptoui``' instruction takes a value to cast, which must be a
7144 scalar or vector :ref:`floating point <t_floating>` value, and a type to
7145 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
7146 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
7147 type with the same number of elements as ``ty``
7152 The '``fptoui``' instruction converts its :ref:`floating
7153 point <t_floating>` operand into the nearest (rounding towards zero)
7154 unsigned integer value. If the value cannot fit in ``ty2``, the results
7160 .. code-block:: llvm
7162 %X = fptoui double 123.0 to i32 ; yields i32:123
7163 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1
7164 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1
7166 '``fptosi .. to``' Instruction
7167 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7174 <result> = fptosi <ty> <value> to <ty2> ; yields ty2
7179 The '``fptosi``' instruction converts :ref:`floating point <t_floating>`
7180 ``value`` to type ``ty2``.
7185 The '``fptosi``' instruction takes a value to cast, which must be a
7186 scalar or vector :ref:`floating point <t_floating>` value, and a type to
7187 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
7188 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
7189 type with the same number of elements as ``ty``
7194 The '``fptosi``' instruction converts its :ref:`floating
7195 point <t_floating>` operand into the nearest (rounding towards zero)
7196 signed integer value. If the value cannot fit in ``ty2``, the results
7202 .. code-block:: llvm
7204 %X = fptosi double -123.0 to i32 ; yields i32:-123
7205 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1
7206 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1
7208 '``uitofp .. to``' Instruction
7209 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7216 <result> = uitofp <ty> <value> to <ty2> ; yields ty2
7221 The '``uitofp``' instruction regards ``value`` as an unsigned integer
7222 and converts that value to the ``ty2`` type.
7227 The '``uitofp``' instruction takes a value to cast, which must be a
7228 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
7229 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
7230 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
7231 type with the same number of elements as ``ty``
7236 The '``uitofp``' instruction interprets its operand as an unsigned
7237 integer quantity and converts it to the corresponding floating point
7238 value. If the value cannot fit in the floating point value, the results
7244 .. code-block:: llvm
7246 %X = uitofp i32 257 to float ; yields float:257.0
7247 %Y = uitofp i8 -1 to double ; yields double:255.0
7249 '``sitofp .. to``' Instruction
7250 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7257 <result> = sitofp <ty> <value> to <ty2> ; yields ty2
7262 The '``sitofp``' instruction regards ``value`` as a signed integer and
7263 converts that value to the ``ty2`` type.
7268 The '``sitofp``' instruction takes a value to cast, which must be a
7269 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
7270 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
7271 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
7272 type with the same number of elements as ``ty``
7277 The '``sitofp``' instruction interprets its operand as a signed integer
7278 quantity and converts it to the corresponding floating point value. If
7279 the value cannot fit in the floating point value, the results are
7285 .. code-block:: llvm
7287 %X = sitofp i32 257 to float ; yields float:257.0
7288 %Y = sitofp i8 -1 to double ; yields double:-1.0
7292 '``ptrtoint .. to``' Instruction
7293 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7300 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2
7305 The '``ptrtoint``' instruction converts the pointer or a vector of
7306 pointers ``value`` to the integer (or vector of integers) type ``ty2``.
7311 The '``ptrtoint``' instruction takes a ``value`` to cast, which must be
7312 a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a
7313 type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or
7314 a vector of integers type.
7319 The '``ptrtoint``' instruction converts ``value`` to integer type
7320 ``ty2`` by interpreting the pointer value as an integer and either
7321 truncating or zero extending that value to the size of the integer type.
7322 If ``value`` is smaller than ``ty2`` then a zero extension is done. If
7323 ``value`` is larger than ``ty2`` then a truncation is done. If they are
7324 the same size, then nothing is done (*no-op cast*) other than a type
7330 .. code-block:: llvm
7332 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture
7333 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture
7334 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture
7338 '``inttoptr .. to``' Instruction
7339 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7346 <result> = inttoptr <ty> <value> to <ty2> ; yields ty2
7351 The '``inttoptr``' instruction converts an integer ``value`` to a
7352 pointer type, ``ty2``.
7357 The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to
7358 cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>`
7364 The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by
7365 applying either a zero extension or a truncation depending on the size
7366 of the integer ``value``. If ``value`` is larger than the size of a
7367 pointer then a truncation is done. If ``value`` is smaller than the size
7368 of a pointer then a zero extension is done. If they are the same size,
7369 nothing is done (*no-op cast*).
7374 .. code-block:: llvm
7376 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture
7377 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture
7378 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture
7379 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers
7383 '``bitcast .. to``' Instruction
7384 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7391 <result> = bitcast <ty> <value> to <ty2> ; yields ty2
7396 The '``bitcast``' instruction converts ``value`` to type ``ty2`` without
7402 The '``bitcast``' instruction takes a value to cast, which must be a
7403 non-aggregate first class value, and a type to cast it to, which must
7404 also be a non-aggregate :ref:`first class <t_firstclass>` type. The
7405 bit sizes of ``value`` and the destination type, ``ty2``, must be
7406 identical. If the source type is a pointer, the destination type must
7407 also be a pointer of the same size. This instruction supports bitwise
7408 conversion of vectors to integers and to vectors of other types (as
7409 long as they have the same size).
7414 The '``bitcast``' instruction converts ``value`` to type ``ty2``. It
7415 is always a *no-op cast* because no bits change with this
7416 conversion. The conversion is done as if the ``value`` had been stored
7417 to memory and read back as type ``ty2``. Pointer (or vector of
7418 pointers) types may only be converted to other pointer (or vector of
7419 pointers) types with the same address space through this instruction.
7420 To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>`
7421 or :ref:`ptrtoint <i_ptrtoint>` instructions first.
7426 .. code-block:: llvm
7428 %X = bitcast i8 255 to i8 ; yields i8 :-1
7429 %Y = bitcast i32* %x to sint* ; yields sint*:%x
7430 %Z = bitcast <2 x int> %V to i64; ; yields i64: %V
7431 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*>
7433 .. _i_addrspacecast:
7435 '``addrspacecast .. to``' Instruction
7436 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7443 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2
7448 The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in
7449 address space ``n`` to type ``pty2`` in address space ``m``.
7454 The '``addrspacecast``' instruction takes a pointer or vector of pointer value
7455 to cast and a pointer type to cast it to, which must have a different
7461 The '``addrspacecast``' instruction converts the pointer value
7462 ``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex
7463 value modification, depending on the target and the address space
7464 pair. Pointer conversions within the same address space must be
7465 performed with the ``bitcast`` instruction. Note that if the address space
7466 conversion is legal then both result and operand refer to the same memory
7472 .. code-block:: llvm
7474 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x
7475 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y
7476 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z
7483 The instructions in this category are the "miscellaneous" instructions,
7484 which defy better classification.
7488 '``icmp``' Instruction
7489 ^^^^^^^^^^^^^^^^^^^^^^
7496 <result> = icmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
7501 The '``icmp``' instruction returns a boolean value or a vector of
7502 boolean values based on comparison of its two integer, integer vector,
7503 pointer, or pointer vector operands.
7508 The '``icmp``' instruction takes three operands. The first operand is
7509 the condition code indicating the kind of comparison to perform. It is
7510 not a value, just a keyword. The possible condition code are:
7513 #. ``ne``: not equal
7514 #. ``ugt``: unsigned greater than
7515 #. ``uge``: unsigned greater or equal
7516 #. ``ult``: unsigned less than
7517 #. ``ule``: unsigned less or equal
7518 #. ``sgt``: signed greater than
7519 #. ``sge``: signed greater or equal
7520 #. ``slt``: signed less than
7521 #. ``sle``: signed less or equal
7523 The remaining two arguments must be :ref:`integer <t_integer>` or
7524 :ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They
7525 must also be identical types.
7530 The '``icmp``' compares ``op1`` and ``op2`` according to the condition
7531 code given as ``cond``. The comparison performed always yields either an
7532 :ref:`i1 <t_integer>` or vector of ``i1`` result, as follows:
7534 #. ``eq``: yields ``true`` if the operands are equal, ``false``
7535 otherwise. No sign interpretation is necessary or performed.
7536 #. ``ne``: yields ``true`` if the operands are unequal, ``false``
7537 otherwise. No sign interpretation is necessary or performed.
7538 #. ``ugt``: interprets the operands as unsigned values and yields
7539 ``true`` if ``op1`` is greater than ``op2``.
7540 #. ``uge``: interprets the operands as unsigned values and yields
7541 ``true`` if ``op1`` is greater than or equal to ``op2``.
7542 #. ``ult``: interprets the operands as unsigned values and yields
7543 ``true`` if ``op1`` is less than ``op2``.
7544 #. ``ule``: interprets the operands as unsigned values and yields
7545 ``true`` if ``op1`` is less than or equal to ``op2``.
7546 #. ``sgt``: interprets the operands as signed values and yields ``true``
7547 if ``op1`` is greater than ``op2``.
7548 #. ``sge``: interprets the operands as signed values and yields ``true``
7549 if ``op1`` is greater than or equal to ``op2``.
7550 #. ``slt``: interprets the operands as signed values and yields ``true``
7551 if ``op1`` is less than ``op2``.
7552 #. ``sle``: interprets the operands as signed values and yields ``true``
7553 if ``op1`` is less than or equal to ``op2``.
7555 If the operands are :ref:`pointer <t_pointer>` typed, the pointer values
7556 are compared as if they were integers.
7558 If the operands are integer vectors, then they are compared element by
7559 element. The result is an ``i1`` vector with the same number of elements
7560 as the values being compared. Otherwise, the result is an ``i1``.
7565 .. code-block:: llvm
7567 <result> = icmp eq i32 4, 5 ; yields: result=false
7568 <result> = icmp ne float* %X, %X ; yields: result=false
7569 <result> = icmp ult i16 4, 5 ; yields: result=true
7570 <result> = icmp sgt i16 4, 5 ; yields: result=false
7571 <result> = icmp ule i16 -4, 5 ; yields: result=false
7572 <result> = icmp sge i16 4, 5 ; yields: result=false
7574 Note that the code generator does not yet support vector types with the
7575 ``icmp`` instruction.
7579 '``fcmp``' Instruction
7580 ^^^^^^^^^^^^^^^^^^^^^^
7587 <result> = fcmp [fast-math flags]* <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
7592 The '``fcmp``' instruction returns a boolean value or vector of boolean
7593 values based on comparison of its operands.
7595 If the operands are floating point scalars, then the result type is a
7596 boolean (:ref:`i1 <t_integer>`).
7598 If the operands are floating point vectors, then the result type is a
7599 vector of boolean with the same number of elements as the operands being
7605 The '``fcmp``' instruction takes three operands. The first operand is
7606 the condition code indicating the kind of comparison to perform. It is
7607 not a value, just a keyword. The possible condition code are:
7609 #. ``false``: no comparison, always returns false
7610 #. ``oeq``: ordered and equal
7611 #. ``ogt``: ordered and greater than
7612 #. ``oge``: ordered and greater than or equal
7613 #. ``olt``: ordered and less than
7614 #. ``ole``: ordered and less than or equal
7615 #. ``one``: ordered and not equal
7616 #. ``ord``: ordered (no nans)
7617 #. ``ueq``: unordered or equal
7618 #. ``ugt``: unordered or greater than
7619 #. ``uge``: unordered or greater than or equal
7620 #. ``ult``: unordered or less than
7621 #. ``ule``: unordered or less than or equal
7622 #. ``une``: unordered or not equal
7623 #. ``uno``: unordered (either nans)
7624 #. ``true``: no comparison, always returns true
7626 *Ordered* means that neither operand is a QNAN while *unordered* means
7627 that either operand may be a QNAN.
7629 Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating
7630 point <t_floating>` type or a :ref:`vector <t_vector>` of floating point
7631 type. They must have identical types.
7636 The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the
7637 condition code given as ``cond``. If the operands are vectors, then the
7638 vectors are compared element by element. Each comparison performed
7639 always yields an :ref:`i1 <t_integer>` result, as follows:
7641 #. ``false``: always yields ``false``, regardless of operands.
7642 #. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1``
7643 is equal to ``op2``.
7644 #. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1``
7645 is greater than ``op2``.
7646 #. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1``
7647 is greater than or equal to ``op2``.
7648 #. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1``
7649 is less than ``op2``.
7650 #. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1``
7651 is less than or equal to ``op2``.
7652 #. ``one``: yields ``true`` if both operands are not a QNAN and ``op1``
7653 is not equal to ``op2``.
7654 #. ``ord``: yields ``true`` if both operands are not a QNAN.
7655 #. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is
7657 #. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is
7658 greater than ``op2``.
7659 #. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is
7660 greater than or equal to ``op2``.
7661 #. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is
7663 #. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is
7664 less than or equal to ``op2``.
7665 #. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is
7666 not equal to ``op2``.
7667 #. ``uno``: yields ``true`` if either operand is a QNAN.
7668 #. ``true``: always yields ``true``, regardless of operands.
7670 The ``fcmp`` instruction can also optionally take any number of
7671 :ref:`fast-math flags <fastmath>`, which are optimization hints to enable
7672 otherwise unsafe floating point optimizations.
7674 Any set of fast-math flags are legal on an ``fcmp`` instruction, but the
7675 only flags that have any effect on its semantics are those that allow
7676 assumptions to be made about the values of input arguments; namely
7677 ``nnan``, ``ninf``, and ``nsz``. See :ref:`fastmath` for more information.
7682 .. code-block:: llvm
7684 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false
7685 <result> = fcmp one float 4.0, 5.0 ; yields: result=true
7686 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true
7687 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false
7689 Note that the code generator does not yet support vector types with the
7690 ``fcmp`` instruction.
7694 '``phi``' Instruction
7695 ^^^^^^^^^^^^^^^^^^^^^
7702 <result> = phi <ty> [ <val0>, <label0>], ...
7707 The '``phi``' instruction is used to implement the φ node in the SSA
7708 graph representing the function.
7713 The type of the incoming values is specified with the first type field.
7714 After this, the '``phi``' instruction takes a list of pairs as
7715 arguments, with one pair for each predecessor basic block of the current
7716 block. Only values of :ref:`first class <t_firstclass>` type may be used as
7717 the value arguments to the PHI node. Only labels may be used as the
7720 There must be no non-phi instructions between the start of a basic block
7721 and the PHI instructions: i.e. PHI instructions must be first in a basic
7724 For the purposes of the SSA form, the use of each incoming value is
7725 deemed to occur on the edge from the corresponding predecessor block to
7726 the current block (but after any definition of an '``invoke``'
7727 instruction's return value on the same edge).
7732 At runtime, the '``phi``' instruction logically takes on the value
7733 specified by the pair corresponding to the predecessor basic block that
7734 executed just prior to the current block.
7739 .. code-block:: llvm
7741 Loop: ; Infinite loop that counts from 0 on up...
7742 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
7743 %nextindvar = add i32 %indvar, 1
7748 '``select``' Instruction
7749 ^^^^^^^^^^^^^^^^^^^^^^^^
7756 <result> = select selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty
7758 selty is either i1 or {<N x i1>}
7763 The '``select``' instruction is used to choose one value based on a
7764 condition, without IR-level branching.
7769 The '``select``' instruction requires an 'i1' value or a vector of 'i1'
7770 values indicating the condition, and two values of the same :ref:`first
7771 class <t_firstclass>` type.
7776 If the condition is an i1 and it evaluates to 1, the instruction returns
7777 the first value argument; otherwise, it returns the second value
7780 If the condition is a vector of i1, then the value arguments must be
7781 vectors of the same size, and the selection is done element by element.
7783 If the condition is an i1 and the value arguments are vectors of the
7784 same size, then an entire vector is selected.
7789 .. code-block:: llvm
7791 %X = select i1 true, i8 17, i8 42 ; yields i8:17
7795 '``call``' Instruction
7796 ^^^^^^^^^^^^^^^^^^^^^^
7803 <result> = [tail | musttail] call [cconv] [ret attrs] <ty> [<fnty>*] <fnptrval>(<function args>) [fn attrs]
7808 The '``call``' instruction represents a simple function call.
7813 This instruction requires several arguments:
7815 #. The optional ``tail`` and ``musttail`` markers indicate that the optimizers
7816 should perform tail call optimization. The ``tail`` marker is a hint that
7817 `can be ignored <CodeGenerator.html#sibcallopt>`_. The ``musttail`` marker
7818 means that the call must be tail call optimized in order for the program to
7819 be correct. The ``musttail`` marker provides these guarantees:
7821 #. The call will not cause unbounded stack growth if it is part of a
7822 recursive cycle in the call graph.
7823 #. Arguments with the :ref:`inalloca <attr_inalloca>` attribute are
7826 Both markers imply that the callee does not access allocas or varargs from
7827 the caller. Calls marked ``musttail`` must obey the following additional
7830 - The call must immediately precede a :ref:`ret <i_ret>` instruction,
7831 or a pointer bitcast followed by a ret instruction.
7832 - The ret instruction must return the (possibly bitcasted) value
7833 produced by the call or void.
7834 - The caller and callee prototypes must match. Pointer types of
7835 parameters or return types may differ in pointee type, but not
7837 - The calling conventions of the caller and callee must match.
7838 - All ABI-impacting function attributes, such as sret, byval, inreg,
7839 returned, and inalloca, must match.
7840 - The callee must be varargs iff the caller is varargs. Bitcasting a
7841 non-varargs function to the appropriate varargs type is legal so
7842 long as the non-varargs prefixes obey the other rules.
7844 Tail call optimization for calls marked ``tail`` is guaranteed to occur if
7845 the following conditions are met:
7847 - Caller and callee both have the calling convention ``fastcc``.
7848 - The call is in tail position (ret immediately follows call and ret
7849 uses value of call or is void).
7850 - Option ``-tailcallopt`` is enabled, or
7851 ``llvm::GuaranteedTailCallOpt`` is ``true``.
7852 - `Platform-specific constraints are
7853 met. <CodeGenerator.html#tailcallopt>`_
7855 #. The optional "cconv" marker indicates which :ref:`calling
7856 convention <callingconv>` the call should use. If none is
7857 specified, the call defaults to using C calling conventions. The
7858 calling convention of the call must match the calling convention of
7859 the target function, or else the behavior is undefined.
7860 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
7861 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
7863 #. '``ty``': the type of the call instruction itself which is also the
7864 type of the return value. Functions that return no value are marked
7866 #. '``fnty``': shall be the signature of the pointer to function value
7867 being invoked. The argument types must match the types implied by
7868 this signature. This type can be omitted if the function is not
7869 varargs and if the function type does not return a pointer to a
7871 #. '``fnptrval``': An LLVM value containing a pointer to a function to
7872 be invoked. In most cases, this is a direct function invocation, but
7873 indirect ``call``'s are just as possible, calling an arbitrary pointer
7875 #. '``function args``': argument list whose types match the function
7876 signature argument types and parameter attributes. All arguments must
7877 be of :ref:`first class <t_firstclass>` type. If the function signature
7878 indicates the function accepts a variable number of arguments, the
7879 extra arguments can be specified.
7880 #. The optional :ref:`function attributes <fnattrs>` list. Only
7881 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
7882 attributes are valid here.
7887 The '``call``' instruction is used to cause control flow to transfer to
7888 a specified function, with its incoming arguments bound to the specified
7889 values. Upon a '``ret``' instruction in the called function, control
7890 flow continues with the instruction after the function call, and the
7891 return value of the function is bound to the result argument.
7896 .. code-block:: llvm
7898 %retval = call i32 @test(i32 %argc)
7899 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32
7900 %X = tail call i32 @foo() ; yields i32
7901 %Y = tail call fastcc i32 @foo() ; yields i32
7902 call void %foo(i8 97 signext)
7904 %struct.A = type { i32, i8 }
7905 %r = call %struct.A @foo() ; yields { i32, i8 }
7906 %gr = extractvalue %struct.A %r, 0 ; yields i32
7907 %gr1 = extractvalue %struct.A %r, 1 ; yields i8
7908 %Z = call void @foo() noreturn ; indicates that %foo never returns normally
7909 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended
7911 llvm treats calls to some functions with names and arguments that match
7912 the standard C99 library as being the C99 library functions, and may
7913 perform optimizations or generate code for them under that assumption.
7914 This is something we'd like to change in the future to provide better
7915 support for freestanding environments and non-C-based languages.
7919 '``va_arg``' Instruction
7920 ^^^^^^^^^^^^^^^^^^^^^^^^
7927 <resultval> = va_arg <va_list*> <arglist>, <argty>
7932 The '``va_arg``' instruction is used to access arguments passed through
7933 the "variable argument" area of a function call. It is used to implement
7934 the ``va_arg`` macro in C.
7939 This instruction takes a ``va_list*`` value and the type of the
7940 argument. It returns a value of the specified argument type and
7941 increments the ``va_list`` to point to the next argument. The actual
7942 type of ``va_list`` is target specific.
7947 The '``va_arg``' instruction loads an argument of the specified type
7948 from the specified ``va_list`` and causes the ``va_list`` to point to
7949 the next argument. For more information, see the variable argument
7950 handling :ref:`Intrinsic Functions <int_varargs>`.
7952 It is legal for this instruction to be called in a function which does
7953 not take a variable number of arguments, for example, the ``vfprintf``
7956 ``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic
7957 function <intrinsics>` because it takes a type as an argument.
7962 See the :ref:`variable argument processing <int_varargs>` section.
7964 Note that the code generator does not yet fully support va\_arg on many
7965 targets. Also, it does not currently support va\_arg with aggregate
7966 types on any target.
7970 '``landingpad``' Instruction
7971 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7978 <resultval> = landingpad <resultty> <clause>+
7979 <resultval> = landingpad <resultty> cleanup <clause>*
7981 <clause> := catch <type> <value>
7982 <clause> := filter <array constant type> <array constant>
7987 The '``landingpad``' instruction is used by `LLVM's exception handling
7988 system <ExceptionHandling.html#overview>`_ to specify that a basic block
7989 is a landing pad --- one where the exception lands, and corresponds to the
7990 code found in the ``catch`` portion of a ``try``/``catch`` sequence. It
7991 defines values supplied by the :ref:`personality function <personalityfn>` upon
7992 re-entry to the function. The ``resultval`` has the type ``resultty``.
7998 ``cleanup`` flag indicates that the landing pad block is a cleanup.
8000 A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and
8001 contains the global variable representing the "type" that may be caught
8002 or filtered respectively. Unlike the ``catch`` clause, the ``filter``
8003 clause takes an array constant as its argument. Use
8004 "``[0 x i8**] undef``" for a filter which cannot throw. The
8005 '``landingpad``' instruction must contain *at least* one ``clause`` or
8006 the ``cleanup`` flag.
8011 The '``landingpad``' instruction defines the values which are set by the
8012 :ref:`personality function <personalityfn>` upon re-entry to the function, and
8013 therefore the "result type" of the ``landingpad`` instruction. As with
8014 calling conventions, how the personality function results are
8015 represented in LLVM IR is target specific.
8017 The clauses are applied in order from top to bottom. If two
8018 ``landingpad`` instructions are merged together through inlining, the
8019 clauses from the calling function are appended to the list of clauses.
8020 When the call stack is being unwound due to an exception being thrown,
8021 the exception is compared against each ``clause`` in turn. If it doesn't
8022 match any of the clauses, and the ``cleanup`` flag is not set, then
8023 unwinding continues further up the call stack.
8025 The ``landingpad`` instruction has several restrictions:
8027 - A landing pad block is a basic block which is the unwind destination
8028 of an '``invoke``' instruction.
8029 - A landing pad block must have a '``landingpad``' instruction as its
8030 first non-PHI instruction.
8031 - There can be only one '``landingpad``' instruction within the landing
8033 - A basic block that is not a landing pad block may not include a
8034 '``landingpad``' instruction.
8039 .. code-block:: llvm
8041 ;; A landing pad which can catch an integer.
8042 %res = landingpad { i8*, i32 }
8044 ;; A landing pad that is a cleanup.
8045 %res = landingpad { i8*, i32 }
8047 ;; A landing pad which can catch an integer and can only throw a double.
8048 %res = landingpad { i8*, i32 }
8050 filter [1 x i8**] [@_ZTId]
8057 LLVM supports the notion of an "intrinsic function". These functions
8058 have well known names and semantics and are required to follow certain
8059 restrictions. Overall, these intrinsics represent an extension mechanism
8060 for the LLVM language that does not require changing all of the
8061 transformations in LLVM when adding to the language (or the bitcode
8062 reader/writer, the parser, etc...).
8064 Intrinsic function names must all start with an "``llvm.``" prefix. This
8065 prefix is reserved in LLVM for intrinsic names; thus, function names may
8066 not begin with this prefix. Intrinsic functions must always be external
8067 functions: you cannot define the body of intrinsic functions. Intrinsic
8068 functions may only be used in call or invoke instructions: it is illegal
8069 to take the address of an intrinsic function. Additionally, because
8070 intrinsic functions are part of the LLVM language, it is required if any
8071 are added that they be documented here.
8073 Some intrinsic functions can be overloaded, i.e., the intrinsic
8074 represents a family of functions that perform the same operation but on
8075 different data types. Because LLVM can represent over 8 million
8076 different integer types, overloading is used commonly to allow an
8077 intrinsic function to operate on any integer type. One or more of the
8078 argument types or the result type can be overloaded to accept any
8079 integer type. Argument types may also be defined as exactly matching a
8080 previous argument's type or the result type. This allows an intrinsic
8081 function which accepts multiple arguments, but needs all of them to be
8082 of the same type, to only be overloaded with respect to a single
8083 argument or the result.
8085 Overloaded intrinsics will have the names of its overloaded argument
8086 types encoded into its function name, each preceded by a period. Only
8087 those types which are overloaded result in a name suffix. Arguments
8088 whose type is matched against another type do not. For example, the
8089 ``llvm.ctpop`` function can take an integer of any width and returns an
8090 integer of exactly the same integer width. This leads to a family of
8091 functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and
8092 ``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is
8093 overloaded, and only one type suffix is required. Because the argument's
8094 type is matched against the return type, it does not require its own
8097 To learn how to add an intrinsic function, please see the `Extending
8098 LLVM Guide <ExtendingLLVM.html>`_.
8102 Variable Argument Handling Intrinsics
8103 -------------------------------------
8105 Variable argument support is defined in LLVM with the
8106 :ref:`va_arg <i_va_arg>` instruction and these three intrinsic
8107 functions. These functions are related to the similarly named macros
8108 defined in the ``<stdarg.h>`` header file.
8110 All of these functions operate on arguments that use a target-specific
8111 value type "``va_list``". The LLVM assembly language reference manual
8112 does not define what this type is, so all transformations should be
8113 prepared to handle these functions regardless of the type used.
8115 This example shows how the :ref:`va_arg <i_va_arg>` instruction and the
8116 variable argument handling intrinsic functions are used.
8118 .. code-block:: llvm
8120 ; This struct is different for every platform. For most platforms,
8121 ; it is merely an i8*.
8122 %struct.va_list = type { i8* }
8124 ; For Unix x86_64 platforms, va_list is the following struct:
8125 ; %struct.va_list = type { i32, i32, i8*, i8* }
8127 define i32 @test(i32 %X, ...) {
8128 ; Initialize variable argument processing
8129 %ap = alloca %struct.va_list
8130 %ap2 = bitcast %struct.va_list* %ap to i8*
8131 call void @llvm.va_start(i8* %ap2)
8133 ; Read a single integer argument
8134 %tmp = va_arg i8* %ap2, i32
8136 ; Demonstrate usage of llvm.va_copy and llvm.va_end
8138 %aq2 = bitcast i8** %aq to i8*
8139 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
8140 call void @llvm.va_end(i8* %aq2)
8142 ; Stop processing of arguments.
8143 call void @llvm.va_end(i8* %ap2)
8147 declare void @llvm.va_start(i8*)
8148 declare void @llvm.va_copy(i8*, i8*)
8149 declare void @llvm.va_end(i8*)
8153 '``llvm.va_start``' Intrinsic
8154 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8161 declare void @llvm.va_start(i8* <arglist>)
8166 The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for
8167 subsequent use by ``va_arg``.
8172 The argument is a pointer to a ``va_list`` element to initialize.
8177 The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro
8178 available in C. In a target-dependent way, it initializes the
8179 ``va_list`` element to which the argument points, so that the next call
8180 to ``va_arg`` will produce the first variable argument passed to the
8181 function. Unlike the C ``va_start`` macro, this intrinsic does not need
8182 to know the last argument of the function as the compiler can figure
8185 '``llvm.va_end``' Intrinsic
8186 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8193 declare void @llvm.va_end(i8* <arglist>)
8198 The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been
8199 initialized previously with ``llvm.va_start`` or ``llvm.va_copy``.
8204 The argument is a pointer to a ``va_list`` to destroy.
8209 The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro
8210 available in C. In a target-dependent way, it destroys the ``va_list``
8211 element to which the argument points. Calls to
8212 :ref:`llvm.va_start <int_va_start>` and
8213 :ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to
8218 '``llvm.va_copy``' Intrinsic
8219 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8226 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
8231 The '``llvm.va_copy``' intrinsic copies the current argument position
8232 from the source argument list to the destination argument list.
8237 The first argument is a pointer to a ``va_list`` element to initialize.
8238 The second argument is a pointer to a ``va_list`` element to copy from.
8243 The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro
8244 available in C. In a target-dependent way, it copies the source
8245 ``va_list`` element into the destination ``va_list`` element. This
8246 intrinsic is necessary because the `` llvm.va_start`` intrinsic may be
8247 arbitrarily complex and require, for example, memory allocation.
8249 Accurate Garbage Collection Intrinsics
8250 --------------------------------------
8252 LLVM's support for `Accurate Garbage Collection <GarbageCollection.html>`_
8253 (GC) requires the frontend to generate code containing appropriate intrinsic
8254 calls and select an appropriate GC strategy which knows how to lower these
8255 intrinsics in a manner which is appropriate for the target collector.
8257 These intrinsics allow identification of :ref:`GC roots on the
8258 stack <int_gcroot>`, as well as garbage collector implementations that
8259 require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers.
8260 Frontends for type-safe garbage collected languages should generate
8261 these intrinsics to make use of the LLVM garbage collectors. For more
8262 details, see `Garbage Collection with LLVM <GarbageCollection.html>`_.
8264 Experimental Statepoint Intrinsics
8265 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8267 LLVM provides an second experimental set of intrinsics for describing garbage
8268 collection safepoints in compiled code. These intrinsics are an alternative
8269 to the ``llvm.gcroot`` intrinsics, but are compatible with the ones for
8270 :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers. The
8271 differences in approach are covered in the `Garbage Collection with LLVM
8272 <GarbageCollection.html>`_ documentation. The intrinsics themselves are
8273 described in :doc:`Statepoints`.
8277 '``llvm.gcroot``' Intrinsic
8278 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8285 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
8290 The '``llvm.gcroot``' intrinsic declares the existence of a GC root to
8291 the code generator, and allows some metadata to be associated with it.
8296 The first argument specifies the address of a stack object that contains
8297 the root pointer. The second pointer (which must be either a constant or
8298 a global value address) contains the meta-data to be associated with the
8304 At runtime, a call to this intrinsic stores a null pointer into the
8305 "ptrloc" location. At compile-time, the code generator generates
8306 information to allow the runtime to find the pointer at GC safe points.
8307 The '``llvm.gcroot``' intrinsic may only be used in a function which
8308 :ref:`specifies a GC algorithm <gc>`.
8312 '``llvm.gcread``' Intrinsic
8313 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8320 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
8325 The '``llvm.gcread``' intrinsic identifies reads of references from heap
8326 locations, allowing garbage collector implementations that require read
8332 The second argument is the address to read from, which should be an
8333 address allocated from the garbage collector. The first object is a
8334 pointer to the start of the referenced object, if needed by the language
8335 runtime (otherwise null).
8340 The '``llvm.gcread``' intrinsic has the same semantics as a load
8341 instruction, but may be replaced with substantially more complex code by
8342 the garbage collector runtime, as needed. The '``llvm.gcread``'
8343 intrinsic may only be used in a function which :ref:`specifies a GC
8348 '``llvm.gcwrite``' Intrinsic
8349 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8356 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
8361 The '``llvm.gcwrite``' intrinsic identifies writes of references to heap
8362 locations, allowing garbage collector implementations that require write
8363 barriers (such as generational or reference counting collectors).
8368 The first argument is the reference to store, the second is the start of
8369 the object to store it to, and the third is the address of the field of
8370 Obj to store to. If the runtime does not require a pointer to the
8371 object, Obj may be null.
8376 The '``llvm.gcwrite``' intrinsic has the same semantics as a store
8377 instruction, but may be replaced with substantially more complex code by
8378 the garbage collector runtime, as needed. The '``llvm.gcwrite``'
8379 intrinsic may only be used in a function which :ref:`specifies a GC
8382 Code Generator Intrinsics
8383 -------------------------
8385 These intrinsics are provided by LLVM to expose special features that
8386 may only be implemented with code generator support.
8388 '``llvm.returnaddress``' Intrinsic
8389 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8396 declare i8 *@llvm.returnaddress(i32 <level>)
8401 The '``llvm.returnaddress``' intrinsic attempts to compute a
8402 target-specific value indicating the return address of the current
8403 function or one of its callers.
8408 The argument to this intrinsic indicates which function to return the
8409 address for. Zero indicates the calling function, one indicates its
8410 caller, etc. The argument is **required** to be a constant integer
8416 The '``llvm.returnaddress``' intrinsic either returns a pointer
8417 indicating the return address of the specified call frame, or zero if it
8418 cannot be identified. The value returned by this intrinsic is likely to
8419 be incorrect or 0 for arguments other than zero, so it should only be
8420 used for debugging purposes.
8422 Note that calling this intrinsic does not prevent function inlining or
8423 other aggressive transformations, so the value returned may not be that
8424 of the obvious source-language caller.
8426 '``llvm.frameaddress``' Intrinsic
8427 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8434 declare i8* @llvm.frameaddress(i32 <level>)
8439 The '``llvm.frameaddress``' intrinsic attempts to return the
8440 target-specific frame pointer value for the specified stack frame.
8445 The argument to this intrinsic indicates which function to return the
8446 frame pointer for. Zero indicates the calling function, one indicates
8447 its caller, etc. The argument is **required** to be a constant integer
8453 The '``llvm.frameaddress``' intrinsic either returns a pointer
8454 indicating the frame address of the specified call frame, or zero if it
8455 cannot be identified. The value returned by this intrinsic is likely to
8456 be incorrect or 0 for arguments other than zero, so it should only be
8457 used for debugging purposes.
8459 Note that calling this intrinsic does not prevent function inlining or
8460 other aggressive transformations, so the value returned may not be that
8461 of the obvious source-language caller.
8463 '``llvm.localescape``' and '``llvm.localrecover``' Intrinsics
8464 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8471 declare void @llvm.localescape(...)
8472 declare i8* @llvm.localrecover(i8* %func, i8* %fp, i32 %idx)
8477 The '``llvm.localescape``' intrinsic escapes offsets of a collection of static
8478 allocas, and the '``llvm.localrecover``' intrinsic applies those offsets to a
8479 live frame pointer to recover the address of the allocation. The offset is
8480 computed during frame layout of the caller of ``llvm.localescape``.
8485 All arguments to '``llvm.localescape``' must be pointers to static allocas or
8486 casts of static allocas. Each function can only call '``llvm.localescape``'
8487 once, and it can only do so from the entry block.
8489 The ``func`` argument to '``llvm.localrecover``' must be a constant
8490 bitcasted pointer to a function defined in the current module. The code
8491 generator cannot determine the frame allocation offset of functions defined in
8494 The ``fp`` argument to '``llvm.localrecover``' must be a frame pointer of a
8495 call frame that is currently live. The return value of '``llvm.localaddress``'
8496 is one way to produce such a value, but various runtimes also expose a suitable
8497 pointer in platform-specific ways.
8499 The ``idx`` argument to '``llvm.localrecover``' indicates which alloca passed to
8500 '``llvm.localescape``' to recover. It is zero-indexed.
8505 These intrinsics allow a group of functions to share access to a set of local
8506 stack allocations of a one parent function. The parent function may call the
8507 '``llvm.localescape``' intrinsic once from the function entry block, and the
8508 child functions can use '``llvm.localrecover``' to access the escaped allocas.
8509 The '``llvm.localescape``' intrinsic blocks inlining, as inlining changes where
8510 the escaped allocas are allocated, which would break attempts to use
8511 '``llvm.localrecover``'.
8513 .. _int_read_register:
8514 .. _int_write_register:
8516 '``llvm.read_register``' and '``llvm.write_register``' Intrinsics
8517 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8524 declare i32 @llvm.read_register.i32(metadata)
8525 declare i64 @llvm.read_register.i64(metadata)
8526 declare void @llvm.write_register.i32(metadata, i32 @value)
8527 declare void @llvm.write_register.i64(metadata, i64 @value)
8533 The '``llvm.read_register``' and '``llvm.write_register``' intrinsics
8534 provides access to the named register. The register must be valid on
8535 the architecture being compiled to. The type needs to be compatible
8536 with the register being read.
8541 The '``llvm.read_register``' intrinsic returns the current value of the
8542 register, where possible. The '``llvm.write_register``' intrinsic sets
8543 the current value of the register, where possible.
8545 This is useful to implement named register global variables that need
8546 to always be mapped to a specific register, as is common practice on
8547 bare-metal programs including OS kernels.
8549 The compiler doesn't check for register availability or use of the used
8550 register in surrounding code, including inline assembly. Because of that,
8551 allocatable registers are not supported.
8553 Warning: So far it only works with the stack pointer on selected
8554 architectures (ARM, AArch64, PowerPC and x86_64). Significant amount of
8555 work is needed to support other registers and even more so, allocatable
8560 '``llvm.stacksave``' Intrinsic
8561 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8568 declare i8* @llvm.stacksave()
8573 The '``llvm.stacksave``' intrinsic is used to remember the current state
8574 of the function stack, for use with
8575 :ref:`llvm.stackrestore <int_stackrestore>`. This is useful for
8576 implementing language features like scoped automatic variable sized
8582 This intrinsic returns a opaque pointer value that can be passed to
8583 :ref:`llvm.stackrestore <int_stackrestore>`. When an
8584 ``llvm.stackrestore`` intrinsic is executed with a value saved from
8585 ``llvm.stacksave``, it effectively restores the state of the stack to
8586 the state it was in when the ``llvm.stacksave`` intrinsic executed. In
8587 practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that
8588 were allocated after the ``llvm.stacksave`` was executed.
8590 .. _int_stackrestore:
8592 '``llvm.stackrestore``' Intrinsic
8593 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8600 declare void @llvm.stackrestore(i8* %ptr)
8605 The '``llvm.stackrestore``' intrinsic is used to restore the state of
8606 the function stack to the state it was in when the corresponding
8607 :ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is
8608 useful for implementing language features like scoped automatic variable
8609 sized arrays in C99.
8614 See the description for :ref:`llvm.stacksave <int_stacksave>`.
8616 '``llvm.prefetch``' Intrinsic
8617 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8624 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
8629 The '``llvm.prefetch``' intrinsic is a hint to the code generator to
8630 insert a prefetch instruction if supported; otherwise, it is a noop.
8631 Prefetches have no effect on the behavior of the program but can change
8632 its performance characteristics.
8637 ``address`` is the address to be prefetched, ``rw`` is the specifier
8638 determining if the fetch should be for a read (0) or write (1), and
8639 ``locality`` is a temporal locality specifier ranging from (0) - no
8640 locality, to (3) - extremely local keep in cache. The ``cache type``
8641 specifies whether the prefetch is performed on the data (1) or
8642 instruction (0) cache. The ``rw``, ``locality`` and ``cache type``
8643 arguments must be constant integers.
8648 This intrinsic does not modify the behavior of the program. In
8649 particular, prefetches cannot trap and do not produce a value. On
8650 targets that support this intrinsic, the prefetch can provide hints to
8651 the processor cache for better performance.
8653 '``llvm.pcmarker``' Intrinsic
8654 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8661 declare void @llvm.pcmarker(i32 <id>)
8666 The '``llvm.pcmarker``' intrinsic is a method to export a Program
8667 Counter (PC) in a region of code to simulators and other tools. The
8668 method is target specific, but it is expected that the marker will use
8669 exported symbols to transmit the PC of the marker. The marker makes no
8670 guarantees that it will remain with any specific instruction after
8671 optimizations. It is possible that the presence of a marker will inhibit
8672 optimizations. The intended use is to be inserted after optimizations to
8673 allow correlations of simulation runs.
8678 ``id`` is a numerical id identifying the marker.
8683 This intrinsic does not modify the behavior of the program. Backends
8684 that do not support this intrinsic may ignore it.
8686 '``llvm.readcyclecounter``' Intrinsic
8687 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8694 declare i64 @llvm.readcyclecounter()
8699 The '``llvm.readcyclecounter``' intrinsic provides access to the cycle
8700 counter register (or similar low latency, high accuracy clocks) on those
8701 targets that support it. On X86, it should map to RDTSC. On Alpha, it
8702 should map to RPCC. As the backing counters overflow quickly (on the
8703 order of 9 seconds on alpha), this should only be used for small
8709 When directly supported, reading the cycle counter should not modify any
8710 memory. Implementations are allowed to either return a application
8711 specific value or a system wide value. On backends without support, this
8712 is lowered to a constant 0.
8714 Note that runtime support may be conditional on the privilege-level code is
8715 running at and the host platform.
8717 '``llvm.clear_cache``' Intrinsic
8718 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8725 declare void @llvm.clear_cache(i8*, i8*)
8730 The '``llvm.clear_cache``' intrinsic ensures visibility of modifications
8731 in the specified range to the execution unit of the processor. On
8732 targets with non-unified instruction and data cache, the implementation
8733 flushes the instruction cache.
8738 On platforms with coherent instruction and data caches (e.g. x86), this
8739 intrinsic is a nop. On platforms with non-coherent instruction and data
8740 cache (e.g. ARM, MIPS), the intrinsic is lowered either to appropriate
8741 instructions or a system call, if cache flushing requires special
8744 The default behavior is to emit a call to ``__clear_cache`` from the run
8747 This instrinsic does *not* empty the instruction pipeline. Modifications
8748 of the current function are outside the scope of the intrinsic.
8750 '``llvm.instrprof_increment``' Intrinsic
8751 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8758 declare void @llvm.instrprof_increment(i8* <name>, i64 <hash>,
8759 i32 <num-counters>, i32 <index>)
8764 The '``llvm.instrprof_increment``' intrinsic can be emitted by a
8765 frontend for use with instrumentation based profiling. These will be
8766 lowered by the ``-instrprof`` pass to generate execution counts of a
8772 The first argument is a pointer to a global variable containing the
8773 name of the entity being instrumented. This should generally be the
8774 (mangled) function name for a set of counters.
8776 The second argument is a hash value that can be used by the consumer
8777 of the profile data to detect changes to the instrumented source, and
8778 the third is the number of counters associated with ``name``. It is an
8779 error if ``hash`` or ``num-counters`` differ between two instances of
8780 ``instrprof_increment`` that refer to the same name.
8782 The last argument refers to which of the counters for ``name`` should
8783 be incremented. It should be a value between 0 and ``num-counters``.
8788 This intrinsic represents an increment of a profiling counter. It will
8789 cause the ``-instrprof`` pass to generate the appropriate data
8790 structures and the code to increment the appropriate value, in a
8791 format that can be written out by a compiler runtime and consumed via
8792 the ``llvm-profdata`` tool.
8794 Standard C Library Intrinsics
8795 -----------------------------
8797 LLVM provides intrinsics for a few important standard C library
8798 functions. These intrinsics allow source-language front-ends to pass
8799 information about the alignment of the pointer arguments to the code
8800 generator, providing opportunity for more efficient code generation.
8804 '``llvm.memcpy``' Intrinsic
8805 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8810 This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any
8811 integer bit width and for different address spaces. Not all targets
8812 support all bit widths however.
8816 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
8817 i32 <len>, i32 <align>, i1 <isvolatile>)
8818 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
8819 i64 <len>, i32 <align>, i1 <isvolatile>)
8824 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
8825 source location to the destination location.
8827 Note that, unlike the standard libc function, the ``llvm.memcpy.*``
8828 intrinsics do not return a value, takes extra alignment/isvolatile
8829 arguments and the pointers can be in specified address spaces.
8834 The first argument is a pointer to the destination, the second is a
8835 pointer to the source. The third argument is an integer argument
8836 specifying the number of bytes to copy, the fourth argument is the
8837 alignment of the source and destination locations, and the fifth is a
8838 boolean indicating a volatile access.
8840 If the call to this intrinsic has an alignment value that is not 0 or 1,
8841 then the caller guarantees that both the source and destination pointers
8842 are aligned to that boundary.
8844 If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is
8845 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
8846 very cleanly specified and it is unwise to depend on it.
8851 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
8852 source location to the destination location, which are not allowed to
8853 overlap. It copies "len" bytes of memory over. If the argument is known
8854 to be aligned to some boundary, this can be specified as the fourth
8855 argument, otherwise it should be set to 0 or 1 (both meaning no alignment).
8857 '``llvm.memmove``' Intrinsic
8858 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8863 This is an overloaded intrinsic. You can use llvm.memmove on any integer
8864 bit width and for different address space. Not all targets support all
8869 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
8870 i32 <len>, i32 <align>, i1 <isvolatile>)
8871 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
8872 i64 <len>, i32 <align>, i1 <isvolatile>)
8877 The '``llvm.memmove.*``' intrinsics move a block of memory from the
8878 source location to the destination location. It is similar to the
8879 '``llvm.memcpy``' intrinsic but allows the two memory locations to
8882 Note that, unlike the standard libc function, the ``llvm.memmove.*``
8883 intrinsics do not return a value, takes extra alignment/isvolatile
8884 arguments and the pointers can be in specified address spaces.
8889 The first argument is a pointer to the destination, the second is a
8890 pointer to the source. The third argument is an integer argument
8891 specifying the number of bytes to copy, the fourth argument is the
8892 alignment of the source and destination locations, and the fifth is a
8893 boolean indicating a volatile access.
8895 If the call to this intrinsic has an alignment value that is not 0 or 1,
8896 then the caller guarantees that the source and destination pointers are
8897 aligned to that boundary.
8899 If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call
8900 is a :ref:`volatile operation <volatile>`. The detailed access behavior is
8901 not very cleanly specified and it is unwise to depend on it.
8906 The '``llvm.memmove.*``' intrinsics copy a block of memory from the
8907 source location to the destination location, which may overlap. It
8908 copies "len" bytes of memory over. If the argument is known to be
8909 aligned to some boundary, this can be specified as the fourth argument,
8910 otherwise it should be set to 0 or 1 (both meaning no alignment).
8912 '``llvm.memset.*``' Intrinsics
8913 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8918 This is an overloaded intrinsic. You can use llvm.memset on any integer
8919 bit width and for different address spaces. However, not all targets
8920 support all bit widths.
8924 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
8925 i32 <len>, i32 <align>, i1 <isvolatile>)
8926 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
8927 i64 <len>, i32 <align>, i1 <isvolatile>)
8932 The '``llvm.memset.*``' intrinsics fill a block of memory with a
8933 particular byte value.
8935 Note that, unlike the standard libc function, the ``llvm.memset``
8936 intrinsic does not return a value and takes extra alignment/volatile
8937 arguments. Also, the destination can be in an arbitrary address space.
8942 The first argument is a pointer to the destination to fill, the second
8943 is the byte value with which to fill it, the third argument is an
8944 integer argument specifying the number of bytes to fill, and the fourth
8945 argument is the known alignment of the destination location.
8947 If the call to this intrinsic has an alignment value that is not 0 or 1,
8948 then the caller guarantees that the destination pointer is aligned to
8951 If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is
8952 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
8953 very cleanly specified and it is unwise to depend on it.
8958 The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting
8959 at the destination location. If the argument is known to be aligned to
8960 some boundary, this can be specified as the fourth argument, otherwise
8961 it should be set to 0 or 1 (both meaning no alignment).
8963 '``llvm.sqrt.*``' Intrinsic
8964 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8969 This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any
8970 floating point or vector of floating point type. Not all targets support
8975 declare float @llvm.sqrt.f32(float %Val)
8976 declare double @llvm.sqrt.f64(double %Val)
8977 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
8978 declare fp128 @llvm.sqrt.f128(fp128 %Val)
8979 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
8984 The '``llvm.sqrt``' intrinsics return the sqrt of the specified operand,
8985 returning the same value as the libm '``sqrt``' functions would. Unlike
8986 ``sqrt`` in libm, however, ``llvm.sqrt`` has undefined behavior for
8987 negative numbers other than -0.0 (which allows for better optimization,
8988 because there is no need to worry about errno being set).
8989 ``llvm.sqrt(-0.0)`` is defined to return -0.0 like IEEE sqrt.
8994 The argument and return value are floating point numbers of the same
9000 This function returns the sqrt of the specified operand if it is a
9001 nonnegative floating point number.
9003 '``llvm.powi.*``' Intrinsic
9004 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9009 This is an overloaded intrinsic. You can use ``llvm.powi`` on any
9010 floating point or vector of floating point type. Not all targets support
9015 declare float @llvm.powi.f32(float %Val, i32 %power)
9016 declare double @llvm.powi.f64(double %Val, i32 %power)
9017 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
9018 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
9019 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
9024 The '``llvm.powi.*``' intrinsics return the first operand raised to the
9025 specified (positive or negative) power. The order of evaluation of
9026 multiplications is not defined. When a vector of floating point type is
9027 used, the second argument remains a scalar integer value.
9032 The second argument is an integer power, and the first is a value to
9033 raise to that power.
9038 This function returns the first value raised to the second power with an
9039 unspecified sequence of rounding operations.
9041 '``llvm.sin.*``' Intrinsic
9042 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9047 This is an overloaded intrinsic. You can use ``llvm.sin`` on any
9048 floating point or vector of floating point type. Not all targets support
9053 declare float @llvm.sin.f32(float %Val)
9054 declare double @llvm.sin.f64(double %Val)
9055 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
9056 declare fp128 @llvm.sin.f128(fp128 %Val)
9057 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
9062 The '``llvm.sin.*``' intrinsics return the sine of the operand.
9067 The argument and return value are floating point numbers of the same
9073 This function returns the sine of the specified operand, returning the
9074 same values as the libm ``sin`` functions would, and handles error
9075 conditions in the same way.
9077 '``llvm.cos.*``' Intrinsic
9078 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9083 This is an overloaded intrinsic. You can use ``llvm.cos`` on any
9084 floating point or vector of floating point type. Not all targets support
9089 declare float @llvm.cos.f32(float %Val)
9090 declare double @llvm.cos.f64(double %Val)
9091 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
9092 declare fp128 @llvm.cos.f128(fp128 %Val)
9093 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
9098 The '``llvm.cos.*``' intrinsics return the cosine of the operand.
9103 The argument and return value are floating point numbers of the same
9109 This function returns the cosine of the specified operand, returning the
9110 same values as the libm ``cos`` functions would, and handles error
9111 conditions in the same way.
9113 '``llvm.pow.*``' Intrinsic
9114 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9119 This is an overloaded intrinsic. You can use ``llvm.pow`` on any
9120 floating point or vector of floating point type. Not all targets support
9125 declare float @llvm.pow.f32(float %Val, float %Power)
9126 declare double @llvm.pow.f64(double %Val, double %Power)
9127 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
9128 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
9129 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
9134 The '``llvm.pow.*``' intrinsics return the first operand raised to the
9135 specified (positive or negative) power.
9140 The second argument is a floating point power, and the first is a value
9141 to raise to that power.
9146 This function returns the first value raised to the second power,
9147 returning the same values as the libm ``pow`` functions would, and
9148 handles error conditions in the same way.
9150 '``llvm.exp.*``' Intrinsic
9151 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9156 This is an overloaded intrinsic. You can use ``llvm.exp`` on any
9157 floating point or vector of floating point type. Not all targets support
9162 declare float @llvm.exp.f32(float %Val)
9163 declare double @llvm.exp.f64(double %Val)
9164 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
9165 declare fp128 @llvm.exp.f128(fp128 %Val)
9166 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
9171 The '``llvm.exp.*``' intrinsics perform the exp function.
9176 The argument and return value are floating point numbers of the same
9182 This function returns the same values as the libm ``exp`` functions
9183 would, and handles error conditions in the same way.
9185 '``llvm.exp2.*``' Intrinsic
9186 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9191 This is an overloaded intrinsic. You can use ``llvm.exp2`` on any
9192 floating point or vector of floating point type. Not all targets support
9197 declare float @llvm.exp2.f32(float %Val)
9198 declare double @llvm.exp2.f64(double %Val)
9199 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val)
9200 declare fp128 @llvm.exp2.f128(fp128 %Val)
9201 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val)
9206 The '``llvm.exp2.*``' intrinsics perform the exp2 function.
9211 The argument and return value are floating point numbers of the same
9217 This function returns the same values as the libm ``exp2`` functions
9218 would, and handles error conditions in the same way.
9220 '``llvm.log.*``' Intrinsic
9221 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9226 This is an overloaded intrinsic. You can use ``llvm.log`` on any
9227 floating point or vector of floating point type. Not all targets support
9232 declare float @llvm.log.f32(float %Val)
9233 declare double @llvm.log.f64(double %Val)
9234 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
9235 declare fp128 @llvm.log.f128(fp128 %Val)
9236 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
9241 The '``llvm.log.*``' intrinsics perform the log function.
9246 The argument and return value are floating point numbers of the same
9252 This function returns the same values as the libm ``log`` functions
9253 would, and handles error conditions in the same way.
9255 '``llvm.log10.*``' Intrinsic
9256 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9261 This is an overloaded intrinsic. You can use ``llvm.log10`` on any
9262 floating point or vector of floating point type. Not all targets support
9267 declare float @llvm.log10.f32(float %Val)
9268 declare double @llvm.log10.f64(double %Val)
9269 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val)
9270 declare fp128 @llvm.log10.f128(fp128 %Val)
9271 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val)
9276 The '``llvm.log10.*``' intrinsics perform the log10 function.
9281 The argument and return value are floating point numbers of the same
9287 This function returns the same values as the libm ``log10`` functions
9288 would, and handles error conditions in the same way.
9290 '``llvm.log2.*``' Intrinsic
9291 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9296 This is an overloaded intrinsic. You can use ``llvm.log2`` on any
9297 floating point or vector of floating point type. Not all targets support
9302 declare float @llvm.log2.f32(float %Val)
9303 declare double @llvm.log2.f64(double %Val)
9304 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val)
9305 declare fp128 @llvm.log2.f128(fp128 %Val)
9306 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val)
9311 The '``llvm.log2.*``' intrinsics perform the log2 function.
9316 The argument and return value are floating point numbers of the same
9322 This function returns the same values as the libm ``log2`` functions
9323 would, and handles error conditions in the same way.
9325 '``llvm.fma.*``' Intrinsic
9326 ^^^^^^^^^^^^^^^^^^^^^^^^^^
9331 This is an overloaded intrinsic. You can use ``llvm.fma`` on any
9332 floating point or vector of floating point type. Not all targets support
9337 declare float @llvm.fma.f32(float %a, float %b, float %c)
9338 declare double @llvm.fma.f64(double %a, double %b, double %c)
9339 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
9340 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
9341 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
9346 The '``llvm.fma.*``' intrinsics perform the fused multiply-add
9352 The argument and return value are floating point numbers of the same
9358 This function returns the same values as the libm ``fma`` functions
9359 would, and does not set errno.
9361 '``llvm.fabs.*``' Intrinsic
9362 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9367 This is an overloaded intrinsic. You can use ``llvm.fabs`` on any
9368 floating point or vector of floating point type. Not all targets support
9373 declare float @llvm.fabs.f32(float %Val)
9374 declare double @llvm.fabs.f64(double %Val)
9375 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val)
9376 declare fp128 @llvm.fabs.f128(fp128 %Val)
9377 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val)
9382 The '``llvm.fabs.*``' intrinsics return the absolute value of the
9388 The argument and return value are floating point numbers of the same
9394 This function returns the same values as the libm ``fabs`` functions
9395 would, and handles error conditions in the same way.
9397 '``llvm.minnum.*``' Intrinsic
9398 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9403 This is an overloaded intrinsic. You can use ``llvm.minnum`` on any
9404 floating point or vector of floating point type. Not all targets support
9409 declare float @llvm.minnum.f32(float %Val0, float %Val1)
9410 declare double @llvm.minnum.f64(double %Val0, double %Val1)
9411 declare x86_fp80 @llvm.minnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
9412 declare fp128 @llvm.minnum.f128(fp128 %Val0, fp128 %Val1)
9413 declare ppc_fp128 @llvm.minnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
9418 The '``llvm.minnum.*``' intrinsics return the minimum of the two
9425 The arguments and return value are floating point numbers of the same
9431 Follows the IEEE-754 semantics for minNum, which also match for libm's
9434 If either operand is a NaN, returns the other non-NaN operand. Returns
9435 NaN only if both operands are NaN. If the operands compare equal,
9436 returns a value that compares equal to both operands. This means that
9437 fmin(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
9439 '``llvm.maxnum.*``' Intrinsic
9440 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9445 This is an overloaded intrinsic. You can use ``llvm.maxnum`` on any
9446 floating point or vector of floating point type. Not all targets support
9451 declare float @llvm.maxnum.f32(float %Val0, float %Val1l)
9452 declare double @llvm.maxnum.f64(double %Val0, double %Val1)
9453 declare x86_fp80 @llvm.maxnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
9454 declare fp128 @llvm.maxnum.f128(fp128 %Val0, fp128 %Val1)
9455 declare ppc_fp128 @llvm.maxnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
9460 The '``llvm.maxnum.*``' intrinsics return the maximum of the two
9467 The arguments and return value are floating point numbers of the same
9472 Follows the IEEE-754 semantics for maxNum, which also match for libm's
9475 If either operand is a NaN, returns the other non-NaN operand. Returns
9476 NaN only if both operands are NaN. If the operands compare equal,
9477 returns a value that compares equal to both operands. This means that
9478 fmax(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
9480 '``llvm.copysign.*``' Intrinsic
9481 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9486 This is an overloaded intrinsic. You can use ``llvm.copysign`` on any
9487 floating point or vector of floating point type. Not all targets support
9492 declare float @llvm.copysign.f32(float %Mag, float %Sgn)
9493 declare double @llvm.copysign.f64(double %Mag, double %Sgn)
9494 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn)
9495 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn)
9496 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn)
9501 The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the
9502 first operand and the sign of the second operand.
9507 The arguments and return value are floating point numbers of the same
9513 This function returns the same values as the libm ``copysign``
9514 functions would, and handles error conditions in the same way.
9516 '``llvm.floor.*``' Intrinsic
9517 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9522 This is an overloaded intrinsic. You can use ``llvm.floor`` on any
9523 floating point or vector of floating point type. Not all targets support
9528 declare float @llvm.floor.f32(float %Val)
9529 declare double @llvm.floor.f64(double %Val)
9530 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val)
9531 declare fp128 @llvm.floor.f128(fp128 %Val)
9532 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val)
9537 The '``llvm.floor.*``' intrinsics return the floor of the operand.
9542 The argument and return value are floating point numbers of the same
9548 This function returns the same values as the libm ``floor`` functions
9549 would, and handles error conditions in the same way.
9551 '``llvm.ceil.*``' Intrinsic
9552 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9557 This is an overloaded intrinsic. You can use ``llvm.ceil`` on any
9558 floating point or vector of floating point type. Not all targets support
9563 declare float @llvm.ceil.f32(float %Val)
9564 declare double @llvm.ceil.f64(double %Val)
9565 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val)
9566 declare fp128 @llvm.ceil.f128(fp128 %Val)
9567 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val)
9572 The '``llvm.ceil.*``' intrinsics return the ceiling of the operand.
9577 The argument and return value are floating point numbers of the same
9583 This function returns the same values as the libm ``ceil`` functions
9584 would, and handles error conditions in the same way.
9586 '``llvm.trunc.*``' Intrinsic
9587 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9592 This is an overloaded intrinsic. You can use ``llvm.trunc`` on any
9593 floating point or vector of floating point type. Not all targets support
9598 declare float @llvm.trunc.f32(float %Val)
9599 declare double @llvm.trunc.f64(double %Val)
9600 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val)
9601 declare fp128 @llvm.trunc.f128(fp128 %Val)
9602 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val)
9607 The '``llvm.trunc.*``' intrinsics returns the operand rounded to the
9608 nearest integer not larger in magnitude than the operand.
9613 The argument and return value are floating point numbers of the same
9619 This function returns the same values as the libm ``trunc`` functions
9620 would, and handles error conditions in the same way.
9622 '``llvm.rint.*``' Intrinsic
9623 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9628 This is an overloaded intrinsic. You can use ``llvm.rint`` on any
9629 floating point or vector of floating point type. Not all targets support
9634 declare float @llvm.rint.f32(float %Val)
9635 declare double @llvm.rint.f64(double %Val)
9636 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val)
9637 declare fp128 @llvm.rint.f128(fp128 %Val)
9638 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val)
9643 The '``llvm.rint.*``' intrinsics returns the operand rounded to the
9644 nearest integer. It may raise an inexact floating-point exception if the
9645 operand isn't an integer.
9650 The argument and return value are floating point numbers of the same
9656 This function returns the same values as the libm ``rint`` functions
9657 would, and handles error conditions in the same way.
9659 '``llvm.nearbyint.*``' Intrinsic
9660 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9665 This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any
9666 floating point or vector of floating point type. Not all targets support
9671 declare float @llvm.nearbyint.f32(float %Val)
9672 declare double @llvm.nearbyint.f64(double %Val)
9673 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val)
9674 declare fp128 @llvm.nearbyint.f128(fp128 %Val)
9675 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val)
9680 The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the
9686 The argument and return value are floating point numbers of the same
9692 This function returns the same values as the libm ``nearbyint``
9693 functions would, and handles error conditions in the same way.
9695 '``llvm.round.*``' Intrinsic
9696 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9701 This is an overloaded intrinsic. You can use ``llvm.round`` on any
9702 floating point or vector of floating point type. Not all targets support
9707 declare float @llvm.round.f32(float %Val)
9708 declare double @llvm.round.f64(double %Val)
9709 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val)
9710 declare fp128 @llvm.round.f128(fp128 %Val)
9711 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val)
9716 The '``llvm.round.*``' intrinsics returns the operand rounded to the
9722 The argument and return value are floating point numbers of the same
9728 This function returns the same values as the libm ``round``
9729 functions would, and handles error conditions in the same way.
9731 Bit Manipulation Intrinsics
9732 ---------------------------
9734 LLVM provides intrinsics for a few important bit manipulation
9735 operations. These allow efficient code generation for some algorithms.
9737 '``llvm.bswap.*``' Intrinsics
9738 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9743 This is an overloaded intrinsic function. You can use bswap on any
9744 integer type that is an even number of bytes (i.e. BitWidth % 16 == 0).
9748 declare i16 @llvm.bswap.i16(i16 <id>)
9749 declare i32 @llvm.bswap.i32(i32 <id>)
9750 declare i64 @llvm.bswap.i64(i64 <id>)
9755 The '``llvm.bswap``' family of intrinsics is used to byte swap integer
9756 values with an even number of bytes (positive multiple of 16 bits).
9757 These are useful for performing operations on data that is not in the
9758 target's native byte order.
9763 The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high
9764 and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32``
9765 intrinsic returns an i32 value that has the four bytes of the input i32
9766 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the
9767 returned i32 will have its bytes in 3, 2, 1, 0 order. The
9768 ``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this
9769 concept to additional even-byte lengths (6 bytes, 8 bytes and more,
9772 '``llvm.ctpop.*``' Intrinsic
9773 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9778 This is an overloaded intrinsic. You can use llvm.ctpop on any integer
9779 bit width, or on any vector with integer elements. Not all targets
9780 support all bit widths or vector types, however.
9784 declare i8 @llvm.ctpop.i8(i8 <src>)
9785 declare i16 @llvm.ctpop.i16(i16 <src>)
9786 declare i32 @llvm.ctpop.i32(i32 <src>)
9787 declare i64 @llvm.ctpop.i64(i64 <src>)
9788 declare i256 @llvm.ctpop.i256(i256 <src>)
9789 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
9794 The '``llvm.ctpop``' family of intrinsics counts the number of bits set
9800 The only argument is the value to be counted. The argument may be of any
9801 integer type, or a vector with integer elements. The return type must
9802 match the argument type.
9807 The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within
9808 each element of a vector.
9810 '``llvm.ctlz.*``' Intrinsic
9811 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9816 This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any
9817 integer bit width, or any vector whose elements are integers. Not all
9818 targets support all bit widths or vector types, however.
9822 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>)
9823 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>)
9824 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>)
9825 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>)
9826 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>)
9827 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
9832 The '``llvm.ctlz``' family of intrinsic functions counts the number of
9833 leading zeros in a variable.
9838 The first argument is the value to be counted. This argument may be of
9839 any integer type, or a vector with integer element type. The return
9840 type must match the first argument type.
9842 The second argument must be a constant and is a flag to indicate whether
9843 the intrinsic should ensure that a zero as the first argument produces a
9844 defined result. Historically some architectures did not provide a
9845 defined result for zero values as efficiently, and many algorithms are
9846 now predicated on avoiding zero-value inputs.
9851 The '``llvm.ctlz``' intrinsic counts the leading (most significant)
9852 zeros in a variable, or within each element of the vector. If
9853 ``src == 0`` then the result is the size in bits of the type of ``src``
9854 if ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
9855 ``llvm.ctlz(i32 2) = 30``.
9857 '``llvm.cttz.*``' Intrinsic
9858 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9863 This is an overloaded intrinsic. You can use ``llvm.cttz`` on any
9864 integer bit width, or any vector of integer elements. Not all targets
9865 support all bit widths or vector types, however.
9869 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>)
9870 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>)
9871 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>)
9872 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>)
9873 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>)
9874 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
9879 The '``llvm.cttz``' family of intrinsic functions counts the number of
9885 The first argument is the value to be counted. This argument may be of
9886 any integer type, or a vector with integer element type. The return
9887 type must match the first argument type.
9889 The second argument must be a constant and is a flag to indicate whether
9890 the intrinsic should ensure that a zero as the first argument produces a
9891 defined result. Historically some architectures did not provide a
9892 defined result for zero values as efficiently, and many algorithms are
9893 now predicated on avoiding zero-value inputs.
9898 The '``llvm.cttz``' intrinsic counts the trailing (least significant)
9899 zeros in a variable, or within each element of a vector. If ``src == 0``
9900 then the result is the size in bits of the type of ``src`` if
9901 ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
9902 ``llvm.cttz(2) = 1``.
9906 Arithmetic with Overflow Intrinsics
9907 -----------------------------------
9909 LLVM provides intrinsics for some arithmetic with overflow operations.
9911 '``llvm.sadd.with.overflow.*``' Intrinsics
9912 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9917 This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow``
9918 on any integer bit width.
9922 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
9923 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
9924 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
9929 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
9930 a signed addition of the two arguments, and indicate whether an overflow
9931 occurred during the signed summation.
9936 The arguments (%a and %b) and the first element of the result structure
9937 may be of integer types of any bit width, but they must have the same
9938 bit width. The second element of the result structure must be of type
9939 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
9945 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
9946 a signed addition of the two variables. They return a structure --- the
9947 first element of which is the signed summation, and the second element
9948 of which is a bit specifying if the signed summation resulted in an
9954 .. code-block:: llvm
9956 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
9957 %sum = extractvalue {i32, i1} %res, 0
9958 %obit = extractvalue {i32, i1} %res, 1
9959 br i1 %obit, label %overflow, label %normal
9961 '``llvm.uadd.with.overflow.*``' Intrinsics
9962 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9967 This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow``
9968 on any integer bit width.
9972 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
9973 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
9974 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
9979 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
9980 an unsigned addition of the two arguments, and indicate whether a carry
9981 occurred during the unsigned summation.
9986 The arguments (%a and %b) and the first element of the result structure
9987 may be of integer types of any bit width, but they must have the same
9988 bit width. The second element of the result structure must be of type
9989 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
9995 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
9996 an unsigned addition of the two arguments. They return a structure --- the
9997 first element of which is the sum, and the second element of which is a
9998 bit specifying if the unsigned summation resulted in a carry.
10003 .. code-block:: llvm
10005 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
10006 %sum = extractvalue {i32, i1} %res, 0
10007 %obit = extractvalue {i32, i1} %res, 1
10008 br i1 %obit, label %carry, label %normal
10010 '``llvm.ssub.with.overflow.*``' Intrinsics
10011 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10016 This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow``
10017 on any integer bit width.
10021 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
10022 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
10023 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
10028 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
10029 a signed subtraction of the two arguments, and indicate whether an
10030 overflow occurred during the signed subtraction.
10035 The arguments (%a and %b) and the first element of the result structure
10036 may be of integer types of any bit width, but they must have the same
10037 bit width. The second element of the result structure must be of type
10038 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
10044 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
10045 a signed subtraction of the two arguments. They return a structure --- the
10046 first element of which is the subtraction, and the second element of
10047 which is a bit specifying if the signed subtraction resulted in an
10053 .. code-block:: llvm
10055 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
10056 %sum = extractvalue {i32, i1} %res, 0
10057 %obit = extractvalue {i32, i1} %res, 1
10058 br i1 %obit, label %overflow, label %normal
10060 '``llvm.usub.with.overflow.*``' Intrinsics
10061 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10066 This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow``
10067 on any integer bit width.
10071 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
10072 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
10073 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
10078 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
10079 an unsigned subtraction of the two arguments, and indicate whether an
10080 overflow occurred during the unsigned subtraction.
10085 The arguments (%a and %b) and the first element of the result structure
10086 may be of integer types of any bit width, but they must have the same
10087 bit width. The second element of the result structure must be of type
10088 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
10094 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
10095 an unsigned subtraction of the two arguments. They return a structure ---
10096 the first element of which is the subtraction, and the second element of
10097 which is a bit specifying if the unsigned subtraction resulted in an
10103 .. code-block:: llvm
10105 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
10106 %sum = extractvalue {i32, i1} %res, 0
10107 %obit = extractvalue {i32, i1} %res, 1
10108 br i1 %obit, label %overflow, label %normal
10110 '``llvm.smul.with.overflow.*``' Intrinsics
10111 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10116 This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow``
10117 on any integer bit width.
10121 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
10122 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
10123 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
10128 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
10129 a signed multiplication of the two arguments, and indicate whether an
10130 overflow occurred during the signed multiplication.
10135 The arguments (%a and %b) and the first element of the result structure
10136 may be of integer types of any bit width, but they must have the same
10137 bit width. The second element of the result structure must be of type
10138 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
10144 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
10145 a signed multiplication of the two arguments. They return a structure ---
10146 the first element of which is the multiplication, and the second element
10147 of which is a bit specifying if the signed multiplication resulted in an
10153 .. code-block:: llvm
10155 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
10156 %sum = extractvalue {i32, i1} %res, 0
10157 %obit = extractvalue {i32, i1} %res, 1
10158 br i1 %obit, label %overflow, label %normal
10160 '``llvm.umul.with.overflow.*``' Intrinsics
10161 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10166 This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow``
10167 on any integer bit width.
10171 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
10172 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
10173 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
10178 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
10179 a unsigned multiplication of the two arguments, and indicate whether an
10180 overflow occurred during the unsigned multiplication.
10185 The arguments (%a and %b) and the first element of the result structure
10186 may be of integer types of any bit width, but they must have the same
10187 bit width. The second element of the result structure must be of type
10188 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
10194 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
10195 an unsigned multiplication of the two arguments. They return a structure ---
10196 the first element of which is the multiplication, and the second
10197 element of which is a bit specifying if the unsigned multiplication
10198 resulted in an overflow.
10203 .. code-block:: llvm
10205 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
10206 %sum = extractvalue {i32, i1} %res, 0
10207 %obit = extractvalue {i32, i1} %res, 1
10208 br i1 %obit, label %overflow, label %normal
10210 Specialised Arithmetic Intrinsics
10211 ---------------------------------
10213 '``llvm.canonicalize.*``' Intrinsic
10214 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10221 declare float @llvm.canonicalize.f32(float %a)
10222 declare double @llvm.canonicalize.f64(double %b)
10227 The '``llvm.canonicalize.*``' intrinsic returns the platform specific canonical
10228 encoding of a floating point number. This canonicalization is useful for
10229 implementing certain numeric primitives such as frexp. The canonical encoding is
10230 defined by IEEE-754-2008 to be:
10234 2.1.8 canonical encoding: The preferred encoding of a floating-point
10235 representation in a format. Applied to declets, significands of finite
10236 numbers, infinities, and NaNs, especially in decimal formats.
10238 This operation can also be considered equivalent to the IEEE-754-2008
10239 conversion of a floating-point value to the same format. NaNs are handled
10240 according to section 6.2.
10242 Examples of non-canonical encodings:
10244 - x87 pseudo denormals, pseudo NaNs, pseudo Infinity, Unnormals. These are
10245 converted to a canonical representation per hardware-specific protocol.
10246 - Many normal decimal floating point numbers have non-canonical alternative
10248 - Some machines, like GPUs or ARMv7 NEON, do not support subnormal values.
10249 These are treated as non-canonical encodings of zero and with be flushed to
10250 a zero of the same sign by this operation.
10252 Note that per IEEE-754-2008 6.2, systems that support signaling NaNs with
10253 default exception handling must signal an invalid exception, and produce a
10256 This function should always be implementable as multiplication by 1.0, provided
10257 that the compiler does not constant fold the operation. Likewise, division by
10258 1.0 and ``llvm.minnum(x, x)`` are possible implementations. Addition with
10259 -0.0 is also sufficient provided that the rounding mode is not -Infinity.
10261 ``@llvm.canonicalize`` must preserve the equality relation. That is:
10263 - ``(@llvm.canonicalize(x) == x)`` is equivalent to ``(x == x)``
10264 - ``(@llvm.canonicalize(x) == @llvm.canonicalize(y))`` is equivalent to
10267 Additionally, the sign of zero must be conserved:
10268 ``@llvm.canonicalize(-0.0) = -0.0`` and ``@llvm.canonicalize(+0.0) = +0.0``
10270 The payload bits of a NaN must be conserved, with two exceptions.
10271 First, environments which use only a single canonical representation of NaN
10272 must perform said canonicalization. Second, SNaNs must be quieted per the
10275 The canonicalization operation may be optimized away if:
10277 - The input is known to be canonical. For example, it was produced by a
10278 floating-point operation that is required by the standard to be canonical.
10279 - The result is consumed only by (or fused with) other floating-point
10280 operations. That is, the bits of the floating point value are not examined.
10282 '``llvm.fmuladd.*``' Intrinsic
10283 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10290 declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
10291 declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
10296 The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add
10297 expressions that can be fused if the code generator determines that (a) the
10298 target instruction set has support for a fused operation, and (b) that the
10299 fused operation is more efficient than the equivalent, separate pair of mul
10300 and add instructions.
10305 The '``llvm.fmuladd.*``' intrinsics each take three arguments: two
10306 multiplicands, a and b, and an addend c.
10315 %0 = call float @llvm.fmuladd.f32(%a, %b, %c)
10317 is equivalent to the expression a \* b + c, except that rounding will
10318 not be performed between the multiplication and addition steps if the
10319 code generator fuses the operations. Fusion is not guaranteed, even if
10320 the target platform supports it. If a fused multiply-add is required the
10321 corresponding llvm.fma.\* intrinsic function should be used
10322 instead. This never sets errno, just as '``llvm.fma.*``'.
10327 .. code-block:: llvm
10329 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields float:r2 = (a * b) + c
10331 Half Precision Floating Point Intrinsics
10332 ----------------------------------------
10334 For most target platforms, half precision floating point is a
10335 storage-only format. This means that it is a dense encoding (in memory)
10336 but does not support computation in the format.
10338 This means that code must first load the half-precision floating point
10339 value as an i16, then convert it to float with
10340 :ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can
10341 then be performed on the float value (including extending to double
10342 etc). To store the value back to memory, it is first converted to float
10343 if needed, then converted to i16 with
10344 :ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an
10347 .. _int_convert_to_fp16:
10349 '``llvm.convert.to.fp16``' Intrinsic
10350 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10357 declare i16 @llvm.convert.to.fp16.f32(float %a)
10358 declare i16 @llvm.convert.to.fp16.f64(double %a)
10363 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
10364 conventional floating point type to half precision floating point format.
10369 The intrinsic function contains single argument - the value to be
10375 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
10376 conventional floating point format to half precision floating point format. The
10377 return value is an ``i16`` which contains the converted number.
10382 .. code-block:: llvm
10384 %res = call i16 @llvm.convert.to.fp16.f32(float %a)
10385 store i16 %res, i16* @x, align 2
10387 .. _int_convert_from_fp16:
10389 '``llvm.convert.from.fp16``' Intrinsic
10390 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10397 declare float @llvm.convert.from.fp16.f32(i16 %a)
10398 declare double @llvm.convert.from.fp16.f64(i16 %a)
10403 The '``llvm.convert.from.fp16``' intrinsic function performs a
10404 conversion from half precision floating point format to single precision
10405 floating point format.
10410 The intrinsic function contains single argument - the value to be
10416 The '``llvm.convert.from.fp16``' intrinsic function performs a
10417 conversion from half single precision floating point format to single
10418 precision floating point format. The input half-float value is
10419 represented by an ``i16`` value.
10424 .. code-block:: llvm
10426 %a = load i16, i16* @x, align 2
10427 %res = call float @llvm.convert.from.fp16(i16 %a)
10429 .. _dbg_intrinsics:
10431 Debugger Intrinsics
10432 -------------------
10434 The LLVM debugger intrinsics (which all start with ``llvm.dbg.``
10435 prefix), are described in the `LLVM Source Level
10436 Debugging <SourceLevelDebugging.html#format_common_intrinsics>`_
10439 Exception Handling Intrinsics
10440 -----------------------------
10442 The LLVM exception handling intrinsics (which all start with
10443 ``llvm.eh.`` prefix), are described in the `LLVM Exception
10444 Handling <ExceptionHandling.html#format_common_intrinsics>`_ document.
10446 .. _int_trampoline:
10448 Trampoline Intrinsics
10449 ---------------------
10451 These intrinsics make it possible to excise one parameter, marked with
10452 the :ref:`nest <nest>` attribute, from a function. The result is a
10453 callable function pointer lacking the nest parameter - the caller does
10454 not need to provide a value for it. Instead, the value to use is stored
10455 in advance in a "trampoline", a block of memory usually allocated on the
10456 stack, which also contains code to splice the nest value into the
10457 argument list. This is used to implement the GCC nested function address
10460 For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)``
10461 then the resulting function pointer has signature ``i32 (i32, i32)*``.
10462 It can be created as follows:
10464 .. code-block:: llvm
10466 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
10467 %tramp1 = getelementptr [10 x i8], [10 x i8]* %tramp, i32 0, i32 0
10468 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
10469 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
10470 %fp = bitcast i8* %p to i32 (i32, i32)*
10472 The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to
10473 ``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``.
10477 '``llvm.init.trampoline``' Intrinsic
10478 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10485 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
10490 This fills the memory pointed to by ``tramp`` with executable code,
10491 turning it into a trampoline.
10496 The ``llvm.init.trampoline`` intrinsic takes three arguments, all
10497 pointers. The ``tramp`` argument must point to a sufficiently large and
10498 sufficiently aligned block of memory; this memory is written to by the
10499 intrinsic. Note that the size and the alignment are target-specific -
10500 LLVM currently provides no portable way of determining them, so a
10501 front-end that generates this intrinsic needs to have some
10502 target-specific knowledge. The ``func`` argument must hold a function
10503 bitcast to an ``i8*``.
10508 The block of memory pointed to by ``tramp`` is filled with target
10509 dependent code, turning it into a function. Then ``tramp`` needs to be
10510 passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can
10511 be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new
10512 function's signature is the same as that of ``func`` with any arguments
10513 marked with the ``nest`` attribute removed. At most one such ``nest``
10514 argument is allowed, and it must be of pointer type. Calling the new
10515 function is equivalent to calling ``func`` with the same argument list,
10516 but with ``nval`` used for the missing ``nest`` argument. If, after
10517 calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is
10518 modified, then the effect of any later call to the returned function
10519 pointer is undefined.
10523 '``llvm.adjust.trampoline``' Intrinsic
10524 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10531 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
10536 This performs any required machine-specific adjustment to the address of
10537 a trampoline (passed as ``tramp``).
10542 ``tramp`` must point to a block of memory which already has trampoline
10543 code filled in by a previous call to
10544 :ref:`llvm.init.trampoline <int_it>`.
10549 On some architectures the address of the code to be executed needs to be
10550 different than the address where the trampoline is actually stored. This
10551 intrinsic returns the executable address corresponding to ``tramp``
10552 after performing the required machine specific adjustments. The pointer
10553 returned can then be :ref:`bitcast and executed <int_trampoline>`.
10555 .. _int_mload_mstore:
10557 Masked Vector Load and Store Intrinsics
10558 ---------------------------------------
10560 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.
10564 '``llvm.masked.load.*``' Intrinsics
10565 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10569 This is an overloaded intrinsic. The loaded data is a vector of any integer or floating point data type.
10573 declare <16 x float> @llvm.masked.load.v16f32 (<16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>)
10574 declare <2 x double> @llvm.masked.load.v2f64 (<2 x double>* <ptr>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>)
10579 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 of the '``passthru``' operand.
10585 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 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 the '``passthru``' operand are the same vector types.
10591 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.
10592 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.
10597 %res = call <16 x float> @llvm.masked.load.v16f32 (<16 x float>* %ptr, i32 4, <16 x i1>%mask, <16 x float> %passthru)
10599 ;; The result of the two following instructions is identical aside from potential memory access exception
10600 %loadlal = load <16 x float>, <16 x float>* %ptr, align 4
10601 %res = select <16 x i1> %mask, <16 x float> %loadlal, <16 x float> %passthru
10605 '``llvm.masked.store.*``' Intrinsics
10606 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10610 This is an overloaded intrinsic. The data stored in memory is a vector of any integer or floating point data type.
10614 declare void @llvm.masked.store.v8i32 (<8 x i32> <value>, <8 x i32> * <ptr>, i32 <alignment>, <8 x i1> <mask>)
10615 declare void @llvm.masked.store.v16f32(<16 x i32> <value>, <16 x i32>* <ptr>, i32 <alignment>, <16 x i1> <mask>)
10620 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.
10625 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.
10631 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.
10632 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.
10636 call void @llvm.masked.store.v16f32(<16 x float> %value, <16 x float>* %ptr, i32 4, <16 x i1> %mask)
10638 ;; The result of the following instructions is identical aside from potential data races and memory access exceptions
10639 %oldval = load <16 x float>, <16 x float>* %ptr, align 4
10640 %res = select <16 x i1> %mask, <16 x float> %value, <16 x float> %oldval
10641 store <16 x float> %res, <16 x float>* %ptr, align 4
10644 Masked Vector Gather and Scatter Intrinsics
10645 -------------------------------------------
10647 LLVM provides intrinsics for vector gather and scatter operations. They are similar to :ref:`Masked Vector Load and Store <int_mload_mstore>`, except they are designed for arbitrary memory accesses, rather than sequential memory accesses. Gather and scatter also employ 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 are off, no memory is accessed.
10651 '``llvm.masked.gather.*``' Intrinsics
10652 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10656 This is an overloaded intrinsic. The loaded data are multiple scalar values of any integer or floating point data type gathered together into one vector.
10660 declare <16 x float> @llvm.masked.gather.v16f32 (<16 x float*> <ptrs>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>)
10661 declare <2 x double> @llvm.masked.gather.v2f64 (<2 x double*> <ptrs>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>)
10666 Reads scalar values from arbitrary memory locations and gathers them into one vector. The memory locations are provided in the vector of pointers '``ptrs``'. The memory is accessed 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 of the '``passthru``' operand.
10672 The first operand is a vector of pointers which holds all memory addresses to read. The second operand is an alignment of the source addresses. It must be a constant integer value. The third operand, mask, is a vector of boolean 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 vector of pointers and the type of the '``passthru``' operand are the same vector types.
10678 The '``llvm.masked.gather``' intrinsic is designed for conditional reading of multiple scalar values from arbitrary memory locations in a single IR operation. It is useful for targets that support vector masked gathers and allows vectorizing basic blocks with data and control divergence. Other targets may support this intrinsic differently, for example by lowering it into a sequence of scalar load operations.
10679 The semantics of this operation are equivalent to a sequence of conditional scalar loads with subsequent gathering all loaded values into a single vector. The mask restricts memory access to certain lanes and facilitates vectorization of predicated basic blocks.
10684 %res = call <4 x double> @llvm.masked.gather.v4f64 (<4 x double*> %ptrs, i32 8, <4 x i1>%mask, <4 x double> <true, true, true, true>)
10686 ;; The gather with all-true mask is equivalent to the following instruction sequence
10687 %ptr0 = extractelement <4 x double*> %ptrs, i32 0
10688 %ptr1 = extractelement <4 x double*> %ptrs, i32 1
10689 %ptr2 = extractelement <4 x double*> %ptrs, i32 2
10690 %ptr3 = extractelement <4 x double*> %ptrs, i32 3
10692 %val0 = load double, double* %ptr0, align 8
10693 %val1 = load double, double* %ptr1, align 8
10694 %val2 = load double, double* %ptr2, align 8
10695 %val3 = load double, double* %ptr3, align 8
10697 %vec0 = insertelement <4 x double>undef, %val0, 0
10698 %vec01 = insertelement <4 x double>%vec0, %val1, 1
10699 %vec012 = insertelement <4 x double>%vec01, %val2, 2
10700 %vec0123 = insertelement <4 x double>%vec012, %val3, 3
10704 '``llvm.masked.scatter.*``' Intrinsics
10705 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10709 This is an overloaded intrinsic. The data stored in memory is a vector of any integer or floating point data type. Each vector element is stored in an arbitrary memory addresses. Scatter with overlapping addresses is guaranteed to be ordered from least-significant to most-significant element.
10713 declare void @llvm.masked.scatter.v8i32 (<8 x i32> <value>, <8 x i32*> <ptrs>, i32 <alignment>, <8 x i1> <mask>)
10714 declare void @llvm.masked.scatter.v16f32(<16 x i32> <value>, <16 x i32*> <ptrs>, i32 <alignment>, <16 x i1> <mask>)
10719 Writes each element from the value vector to the corresponding memory address. The memory addresses are represented as a vector of pointers. Writing is done 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.
10724 The first operand is a vector value to be written to memory. The second operand is a vector of pointers, pointing to where the value elements should be stored. It has the same underlying type as the value operand. The third operand is an alignment of the destination addresses. 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.
10730 The '``llvm.masked.scatter``' intrinsics is designed for writing selected vector elements to arbitrary memory addresses in a single IR operation. The operation may be conditional, when not all bits in the mask are switched on. It is useful for targets that support vector masked scatter and allows vectorizing basic blocks with data and control divergency. Other targets may support this intrinsic differently, for example by lowering it into a sequence of branches that guard scalar store operations.
10734 ;; This instruction unconditionaly stores data vector in multiple addresses
10735 call @llvm.masked.scatter.v8i32 (<8 x i32> %value, <8 x i32*> %ptrs, i32 4, <8 x i1> <true, true, .. true>)
10737 ;; It is equivalent to a list of scalar stores
10738 %val0 = extractelement <8 x i32> %value, i32 0
10739 %val1 = extractelement <8 x i32> %value, i32 1
10741 %val7 = extractelement <8 x i32> %value, i32 7
10742 %ptr0 = extractelement <8 x i32*> %ptrs, i32 0
10743 %ptr1 = extractelement <8 x i32*> %ptrs, i32 1
10745 %ptr7 = extractelement <8 x i32*> %ptrs, i32 7
10746 ;; Note: the order of the following stores is important when they overlap:
10747 store i32 %val0, i32* %ptr0, align 4
10748 store i32 %val1, i32* %ptr1, align 4
10750 store i32 %val7, i32* %ptr7, align 4
10756 This class of intrinsics provides information about the lifetime of
10757 memory objects and ranges where variables are immutable.
10761 '``llvm.lifetime.start``' Intrinsic
10762 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10769 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
10774 The '``llvm.lifetime.start``' intrinsic specifies the start of a memory
10780 The first argument is a constant integer representing the size of the
10781 object, or -1 if it is variable sized. The second argument is a pointer
10787 This intrinsic indicates that before this point in the code, the value
10788 of the memory pointed to by ``ptr`` is dead. This means that it is known
10789 to never be used and has an undefined value. A load from the pointer
10790 that precedes this intrinsic can be replaced with ``'undef'``.
10794 '``llvm.lifetime.end``' Intrinsic
10795 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10802 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
10807 The '``llvm.lifetime.end``' intrinsic specifies the end of a memory
10813 The first argument is a constant integer representing the size of the
10814 object, or -1 if it is variable sized. The second argument is a pointer
10820 This intrinsic indicates that after this point in the code, the value of
10821 the memory pointed to by ``ptr`` is dead. This means that it is known to
10822 never be used and has an undefined value. Any stores into the memory
10823 object following this intrinsic may be removed as dead.
10825 '``llvm.invariant.start``' Intrinsic
10826 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10833 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
10838 The '``llvm.invariant.start``' intrinsic specifies that the contents of
10839 a memory object will not change.
10844 The first argument is a constant integer representing the size of the
10845 object, or -1 if it is variable sized. The second argument is a pointer
10851 This intrinsic indicates that until an ``llvm.invariant.end`` that uses
10852 the return value, the referenced memory location is constant and
10855 '``llvm.invariant.end``' Intrinsic
10856 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10863 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
10868 The '``llvm.invariant.end``' intrinsic specifies that the contents of a
10869 memory object are mutable.
10874 The first argument is the matching ``llvm.invariant.start`` intrinsic.
10875 The second argument is a constant integer representing the size of the
10876 object, or -1 if it is variable sized and the third argument is a
10877 pointer to the object.
10882 This intrinsic indicates that the memory is mutable again.
10887 This class of intrinsics is designed to be generic and has no specific
10890 '``llvm.var.annotation``' Intrinsic
10891 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10898 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
10903 The '``llvm.var.annotation``' intrinsic.
10908 The first argument is a pointer to a value, the second is a pointer to a
10909 global string, the third is a pointer to a global string which is the
10910 source file name, and the last argument is the line number.
10915 This intrinsic allows annotation of local variables with arbitrary
10916 strings. This can be useful for special purpose optimizations that want
10917 to look for these annotations. These have no other defined use; they are
10918 ignored by code generation and optimization.
10920 '``llvm.ptr.annotation.*``' Intrinsic
10921 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10926 This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a
10927 pointer to an integer of any width. *NOTE* you must specify an address space for
10928 the pointer. The identifier for the default address space is the integer
10933 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
10934 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>)
10935 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>)
10936 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>)
10937 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>)
10942 The '``llvm.ptr.annotation``' intrinsic.
10947 The first argument is a pointer to an integer value of arbitrary bitwidth
10948 (result of some expression), the second is a pointer to a global string, the
10949 third is a pointer to a global string which is the source file name, and the
10950 last argument is the line number. It returns the value of the first argument.
10955 This intrinsic allows annotation of a pointer to an integer with arbitrary
10956 strings. This can be useful for special purpose optimizations that want to look
10957 for these annotations. These have no other defined use; they are ignored by code
10958 generation and optimization.
10960 '``llvm.annotation.*``' Intrinsic
10961 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10966 This is an overloaded intrinsic. You can use '``llvm.annotation``' on
10967 any integer bit width.
10971 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
10972 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
10973 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
10974 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
10975 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
10980 The '``llvm.annotation``' intrinsic.
10985 The first argument is an integer value (result of some expression), the
10986 second is a pointer to a global string, the third is a pointer to a
10987 global string which is the source file name, and the last argument is
10988 the line number. It returns the value of the first argument.
10993 This intrinsic allows annotations to be put on arbitrary expressions
10994 with arbitrary strings. This can be useful for special purpose
10995 optimizations that want to look for these annotations. These have no
10996 other defined use; they are ignored by code generation and optimization.
10998 '``llvm.trap``' Intrinsic
10999 ^^^^^^^^^^^^^^^^^^^^^^^^^
11006 declare void @llvm.trap() noreturn nounwind
11011 The '``llvm.trap``' intrinsic.
11021 This intrinsic is lowered to the target dependent trap instruction. If
11022 the target does not have a trap instruction, this intrinsic will be
11023 lowered to a call of the ``abort()`` function.
11025 '``llvm.debugtrap``' Intrinsic
11026 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11033 declare void @llvm.debugtrap() nounwind
11038 The '``llvm.debugtrap``' intrinsic.
11048 This intrinsic is lowered to code which is intended to cause an
11049 execution trap with the intention of requesting the attention of a
11052 '``llvm.stackprotector``' Intrinsic
11053 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11060 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
11065 The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it
11066 onto the stack at ``slot``. The stack slot is adjusted to ensure that it
11067 is placed on the stack before local variables.
11072 The ``llvm.stackprotector`` intrinsic requires two pointer arguments.
11073 The first argument is the value loaded from the stack guard
11074 ``@__stack_chk_guard``. The second variable is an ``alloca`` that has
11075 enough space to hold the value of the guard.
11080 This intrinsic causes the prologue/epilogue inserter to force the position of
11081 the ``AllocaInst`` stack slot to be before local variables on the stack. This is
11082 to ensure that if a local variable on the stack is overwritten, it will destroy
11083 the value of the guard. When the function exits, the guard on the stack is
11084 checked against the original guard by ``llvm.stackprotectorcheck``. If they are
11085 different, then ``llvm.stackprotectorcheck`` causes the program to abort by
11086 calling the ``__stack_chk_fail()`` function.
11088 '``llvm.stackprotectorcheck``' Intrinsic
11089 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11096 declare void @llvm.stackprotectorcheck(i8** <guard>)
11101 The ``llvm.stackprotectorcheck`` intrinsic compares ``guard`` against an already
11102 created stack protector and if they are not equal calls the
11103 ``__stack_chk_fail()`` function.
11108 The ``llvm.stackprotectorcheck`` intrinsic requires one pointer argument, the
11109 the variable ``@__stack_chk_guard``.
11114 This intrinsic is provided to perform the stack protector check by comparing
11115 ``guard`` with the stack slot created by ``llvm.stackprotector`` and if the
11116 values do not match call the ``__stack_chk_fail()`` function.
11118 The reason to provide this as an IR level intrinsic instead of implementing it
11119 via other IR operations is that in order to perform this operation at the IR
11120 level without an intrinsic, one would need to create additional basic blocks to
11121 handle the success/failure cases. This makes it difficult to stop the stack
11122 protector check from disrupting sibling tail calls in Codegen. With this
11123 intrinsic, we are able to generate the stack protector basic blocks late in
11124 codegen after the tail call decision has occurred.
11126 '``llvm.objectsize``' Intrinsic
11127 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11134 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>)
11135 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>)
11140 The ``llvm.objectsize`` intrinsic is designed to provide information to
11141 the optimizers to determine at compile time whether a) an operation
11142 (like memcpy) will overflow a buffer that corresponds to an object, or
11143 b) that a runtime check for overflow isn't necessary. An object in this
11144 context means an allocation of a specific class, structure, array, or
11150 The ``llvm.objectsize`` intrinsic takes two arguments. The first
11151 argument is a pointer to or into the ``object``. The second argument is
11152 a boolean and determines whether ``llvm.objectsize`` returns 0 (if true)
11153 or -1 (if false) when the object size is unknown. The second argument
11154 only accepts constants.
11159 The ``llvm.objectsize`` intrinsic is lowered to a constant representing
11160 the size of the object concerned. If the size cannot be determined at
11161 compile time, ``llvm.objectsize`` returns ``i32/i64 -1 or 0`` (depending
11162 on the ``min`` argument).
11164 '``llvm.expect``' Intrinsic
11165 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
11170 This is an overloaded intrinsic. You can use ``llvm.expect`` on any
11175 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>)
11176 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>)
11177 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>)
11182 The ``llvm.expect`` intrinsic provides information about expected (the
11183 most probable) value of ``val``, which can be used by optimizers.
11188 The ``llvm.expect`` intrinsic takes two arguments. The first argument is
11189 a value. The second argument is an expected value, this needs to be a
11190 constant value, variables are not allowed.
11195 This intrinsic is lowered to the ``val``.
11199 '``llvm.assume``' Intrinsic
11200 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11207 declare void @llvm.assume(i1 %cond)
11212 The ``llvm.assume`` allows the optimizer to assume that the provided
11213 condition is true. This information can then be used in simplifying other parts
11219 The condition which the optimizer may assume is always true.
11224 The intrinsic allows the optimizer to assume that the provided condition is
11225 always true whenever the control flow reaches the intrinsic call. No code is
11226 generated for this intrinsic, and instructions that contribute only to the
11227 provided condition are not used for code generation. If the condition is
11228 violated during execution, the behavior is undefined.
11230 Note that the optimizer might limit the transformations performed on values
11231 used by the ``llvm.assume`` intrinsic in order to preserve the instructions
11232 only used to form the intrinsic's input argument. This might prove undesirable
11233 if the extra information provided by the ``llvm.assume`` intrinsic does not cause
11234 sufficient overall improvement in code quality. For this reason,
11235 ``llvm.assume`` should not be used to document basic mathematical invariants
11236 that the optimizer can otherwise deduce or facts that are of little use to the
11241 '``llvm.bitset.test``' Intrinsic
11242 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11249 declare i1 @llvm.bitset.test(i8* %ptr, metadata %bitset) nounwind readnone
11255 The first argument is a pointer to be tested. The second argument is a
11256 metadata string containing the name of a :doc:`bitset <BitSets>`.
11261 The ``llvm.bitset.test`` intrinsic tests whether the given pointer is a
11262 member of the given bitset.
11264 '``llvm.donothing``' Intrinsic
11265 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11272 declare void @llvm.donothing() nounwind readnone
11277 The ``llvm.donothing`` intrinsic doesn't perform any operation. It's one of only
11278 two intrinsics (besides ``llvm.experimental.patchpoint``) that can be called
11279 with an invoke instruction.
11289 This intrinsic does nothing, and it's removed by optimizers and ignored
11292 Stack Map Intrinsics
11293 --------------------
11295 LLVM provides experimental intrinsics to support runtime patching
11296 mechanisms commonly desired in dynamic language JITs. These intrinsics
11297 are described in :doc:`StackMaps`.