1 ==============================
2 LLVM Language Reference Manual
3 ==============================
12 This document is a reference manual for the LLVM assembly language. LLVM
13 is a Static Single Assignment (SSA) based representation that provides
14 type safety, low-level operations, flexibility, and the capability of
15 representing 'all' high-level languages cleanly. It is the common code
16 representation used throughout all phases of the LLVM compilation
22 The LLVM code representation is designed to be used in three different
23 forms: as an in-memory compiler IR, as an on-disk bitcode representation
24 (suitable for fast loading by a Just-In-Time compiler), and as a human
25 readable assembly language representation. This allows LLVM to provide a
26 powerful intermediate representation for efficient compiler
27 transformations and analysis, while providing a natural means to debug
28 and visualize the transformations. The three different forms of LLVM are
29 all equivalent. This document describes the human readable
30 representation and notation.
32 The LLVM representation aims to be light-weight and low-level while
33 being expressive, typed, and extensible at the same time. It aims to be
34 a "universal IR" of sorts, by being at a low enough level that
35 high-level ideas may be cleanly mapped to it (similar to how
36 microprocessors are "universal IR's", allowing many source languages to
37 be mapped to them). By providing type information, LLVM can be used as
38 the target of optimizations: for example, through pointer analysis, it
39 can be proven that a C automatic variable is never accessed outside of
40 the current function, allowing it to be promoted to a simple SSA value
41 instead of a memory location.
48 It is important to note that this document describes 'well formed' LLVM
49 assembly language. There is a difference between what the parser accepts
50 and what is considered 'well formed'. For example, the following
51 instruction is syntactically okay, but not well formed:
57 because the definition of ``%x`` does not dominate all of its uses. The
58 LLVM infrastructure provides a verification pass that may be used to
59 verify that an LLVM module is well formed. This pass is automatically
60 run by the parser after parsing input assembly and by the optimizer
61 before it outputs bitcode. The violations pointed out by the verifier
62 pass indicate bugs in transformation passes or input to the parser.
69 LLVM identifiers come in two basic types: global and local. Global
70 identifiers (functions, global variables) begin with the ``'@'``
71 character. Local identifiers (register names, types) begin with the
72 ``'%'`` character. Additionally, there are three different formats for
73 identifiers, for different purposes:
75 #. Named values are represented as a string of characters with their
76 prefix. For example, ``%foo``, ``@DivisionByZero``,
77 ``%a.really.long.identifier``. The actual regular expression used is
78 '``[%@][-a-zA-Z$._][-a-zA-Z$._0-9]*``'. Identifiers that require other
79 characters in their names can be surrounded with quotes. Special
80 characters may be escaped using ``"\xx"`` where ``xx`` is the ASCII
81 code for the character in hexadecimal. In this way, any character can
82 be used in a name value, even quotes themselves. The ``"\01"`` prefix
83 can be used on global variables to suppress mangling.
84 #. Unnamed values are represented as an unsigned numeric value with
85 their prefix. For example, ``%12``, ``@2``, ``%44``.
86 #. Constants, which are described in the section Constants_ below.
88 LLVM requires that values start with a prefix for two reasons: Compilers
89 don't need to worry about name clashes with reserved words, and the set
90 of reserved words may be expanded in the future without penalty.
91 Additionally, unnamed identifiers allow a compiler to quickly come up
92 with a temporary variable without having to avoid symbol table
95 Reserved words in LLVM are very similar to reserved words in other
96 languages. There are keywords for different opcodes ('``add``',
97 '``bitcast``', '``ret``', etc...), for primitive type names ('``void``',
98 '``i32``', etc...), and others. These reserved words cannot conflict
99 with variable names, because none of them start with a prefix character
100 (``'%'`` or ``'@'``).
102 Here is an example of LLVM code to multiply the integer variable
109 %result = mul i32 %X, 8
111 After strength reduction:
115 %result = shl i32 %X, 3
121 %0 = add i32 %X, %X ; yields i32:%0
122 %1 = add i32 %0, %0 ; yields i32:%1
123 %result = add i32 %1, %1
125 This last way of multiplying ``%X`` by 8 illustrates several important
126 lexical features of LLVM:
128 #. Comments are delimited with a '``;``' and go until the end of line.
129 #. Unnamed temporaries are created when the result of a computation is
130 not assigned to a named value.
131 #. Unnamed temporaries are numbered sequentially (using a per-function
132 incrementing counter, starting with 0). Note that basic blocks and unnamed
133 function parameters are included in this numbering. For example, if the
134 entry basic block is not given a label name and all function parameters are
135 named, then it will get number 0.
137 It also shows a convention that we follow in this document. When
138 demonstrating instructions, we will follow an instruction with a comment
139 that defines the type and name of value produced.
147 LLVM programs are composed of ``Module``'s, each of which is a
148 translation unit of the input programs. Each module consists of
149 functions, global variables, and symbol table entries. Modules may be
150 combined together with the LLVM linker, which merges function (and
151 global variable) definitions, resolves forward declarations, and merges
152 symbol table entries. Here is an example of the "hello world" module:
156 ; Declare the string constant as a global constant.
157 @.str = private unnamed_addr constant [13 x i8] c"hello world\0A\00"
159 ; External declaration of the puts function
160 declare i32 @puts(i8* nocapture) nounwind
162 ; Definition of main function
163 define i32 @main() { ; i32()*
164 ; Convert [13 x i8]* to i8 *...
165 %cast210 = getelementptr [13 x i8]* @.str, i64 0, i64 0
167 ; Call puts function to write out the string to stdout.
168 call i32 @puts(i8* %cast210)
173 !0 = !{i32 42, null, !"string"}
176 This example is made up of a :ref:`global variable <globalvars>` named
177 "``.str``", an external declaration of the "``puts``" function, a
178 :ref:`function definition <functionstructure>` for "``main``" and
179 :ref:`named metadata <namedmetadatastructure>` "``foo``".
181 In general, a module is made up of a list of global values (where both
182 functions and global variables are global values). Global values are
183 represented by a pointer to a memory location (in this case, a pointer
184 to an array of char, and a pointer to a function), and have one of the
185 following :ref:`linkage types <linkage>`.
192 All Global Variables and Functions have one of the following types of
196 Global values with "``private``" linkage are only directly
197 accessible by objects in the current module. In particular, linking
198 code into a module with an private global value may cause the
199 private to be renamed as necessary to avoid collisions. Because the
200 symbol is private to the module, all references can be updated. This
201 doesn't show up in any symbol table in the object file.
203 Similar to private, but the value shows as a local symbol
204 (``STB_LOCAL`` in the case of ELF) in the object file. This
205 corresponds to the notion of the '``static``' keyword in C.
206 ``available_externally``
207 Globals with "``available_externally``" linkage are never emitted
208 into the object file corresponding to the LLVM module. They exist to
209 allow inlining and other optimizations to take place given knowledge
210 of the definition of the global, which is known to be somewhere
211 outside the module. Globals with ``available_externally`` linkage
212 are allowed to be discarded at will, and are otherwise the same as
213 ``linkonce_odr``. This linkage type is only allowed on definitions,
216 Globals with "``linkonce``" linkage are merged with other globals of
217 the same name when linkage occurs. This can be used to implement
218 some forms of inline functions, templates, or other code which must
219 be generated in each translation unit that uses it, but where the
220 body may be overridden with a more definitive definition later.
221 Unreferenced ``linkonce`` globals are allowed to be discarded. Note
222 that ``linkonce`` linkage does not actually allow the optimizer to
223 inline the body of this function into callers because it doesn't
224 know if this definition of the function is the definitive definition
225 within the program or whether it will be overridden by a stronger
226 definition. To enable inlining and other optimizations, use
227 "``linkonce_odr``" linkage.
229 "``weak``" linkage has the same merging semantics as ``linkonce``
230 linkage, except that unreferenced globals with ``weak`` linkage may
231 not be discarded. This is used for globals that are declared "weak"
234 "``common``" linkage is most similar to "``weak``" linkage, but they
235 are used for tentative definitions in C, such as "``int X;``" at
236 global scope. Symbols with "``common``" linkage are merged in the
237 same way as ``weak symbols``, and they may not be deleted if
238 unreferenced. ``common`` symbols may not have an explicit section,
239 must have a zero initializer, and may not be marked
240 ':ref:`constant <globalvars>`'. Functions and aliases may not have
243 .. _linkage_appending:
246 "``appending``" linkage may only be applied to global variables of
247 pointer to array type. When two global variables with appending
248 linkage are linked together, the two global arrays are appended
249 together. This is the LLVM, typesafe, equivalent of having the
250 system linker append together "sections" with identical names when
253 The semantics of this linkage follow the ELF object file model: the
254 symbol is weak until linked, if not linked, the symbol becomes null
255 instead of being an undefined reference.
256 ``linkonce_odr``, ``weak_odr``
257 Some languages allow differing globals to be merged, such as two
258 functions with different semantics. Other languages, such as
259 ``C++``, ensure that only equivalent globals are ever merged (the
260 "one definition rule" --- "ODR"). Such languages can use the
261 ``linkonce_odr`` and ``weak_odr`` linkage types to indicate that the
262 global will only be merged with equivalent globals. These linkage
263 types are otherwise the same as their non-``odr`` versions.
265 If none of the above identifiers are used, the global is externally
266 visible, meaning that it participates in linkage and can be used to
267 resolve external symbol references.
269 It is illegal for a function *declaration* to have any linkage type
270 other than ``external`` or ``extern_weak``.
277 LLVM :ref:`functions <functionstructure>`, :ref:`calls <i_call>` and
278 :ref:`invokes <i_invoke>` can all have an optional calling convention
279 specified for the call. The calling convention of any pair of dynamic
280 caller/callee must match, or the behavior of the program is undefined.
281 The following calling conventions are supported by LLVM, and more may be
284 "``ccc``" - The C calling convention
285 This calling convention (the default if no other calling convention
286 is specified) matches the target C calling conventions. This calling
287 convention supports varargs function calls and tolerates some
288 mismatch in the declared prototype and implemented declaration of
289 the function (as does normal C).
290 "``fastcc``" - The fast calling convention
291 This calling convention attempts to make calls as fast as possible
292 (e.g. by passing things in registers). This calling convention
293 allows the target to use whatever tricks it wants to produce fast
294 code for the target, without having to conform to an externally
295 specified ABI (Application Binary Interface). `Tail calls can only
296 be optimized when this, the GHC or the HiPE convention is
297 used. <CodeGenerator.html#id80>`_ This calling convention does not
298 support varargs and requires the prototype of all callees to exactly
299 match the prototype of the function definition.
300 "``coldcc``" - The cold calling convention
301 This calling convention attempts to make code in the caller as
302 efficient as possible under the assumption that the call is not
303 commonly executed. As such, these calls often preserve all registers
304 so that the call does not break any live ranges in the caller side.
305 This calling convention does not support varargs and requires the
306 prototype of all callees to exactly match the prototype of the
307 function definition. Furthermore the inliner doesn't consider such function
309 "``cc 10``" - GHC convention
310 This calling convention has been implemented specifically for use by
311 the `Glasgow Haskell Compiler (GHC) <http://www.haskell.org/ghc>`_.
312 It passes everything in registers, going to extremes to achieve this
313 by disabling callee save registers. This calling convention should
314 not be used lightly but only for specific situations such as an
315 alternative to the *register pinning* performance technique often
316 used when implementing functional programming languages. At the
317 moment only X86 supports this convention and it has the following
320 - On *X86-32* only supports up to 4 bit type parameters. No
321 floating point types are supported.
322 - On *X86-64* only supports up to 10 bit type parameters and 6
323 floating point parameters.
325 This calling convention supports `tail call
326 optimization <CodeGenerator.html#id80>`_ but requires both the
327 caller and callee are using it.
328 "``cc 11``" - The HiPE calling convention
329 This calling convention has been implemented specifically for use by
330 the `High-Performance Erlang
331 (HiPE) <http://www.it.uu.se/research/group/hipe/>`_ compiler, *the*
332 native code compiler of the `Ericsson's Open Source Erlang/OTP
333 system <http://www.erlang.org/download.shtml>`_. It uses more
334 registers for argument passing than the ordinary C calling
335 convention and defines no callee-saved registers. The calling
336 convention properly supports `tail call
337 optimization <CodeGenerator.html#id80>`_ but requires that both the
338 caller and the callee use it. It uses a *register pinning*
339 mechanism, similar to GHC's convention, for keeping frequently
340 accessed runtime components pinned to specific hardware registers.
341 At the moment only X86 supports this convention (both 32 and 64
343 "``webkit_jscc``" - WebKit's JavaScript calling convention
344 This calling convention has been implemented for `WebKit FTL JIT
345 <https://trac.webkit.org/wiki/FTLJIT>`_. It passes arguments on the
346 stack right to left (as cdecl does), and returns a value in the
347 platform's customary return register.
348 "``anyregcc``" - Dynamic calling convention for code patching
349 This is a special convention that supports patching an arbitrary code
350 sequence in place of a call site. This convention forces the call
351 arguments into registers but allows them to be dynamcially
352 allocated. This can currently only be used with calls to
353 llvm.experimental.patchpoint because only this intrinsic records
354 the location of its arguments in a side table. See :doc:`StackMaps`.
355 "``preserve_mostcc``" - The `PreserveMost` calling convention
356 This calling convention attempts to make the code in the caller as little
357 intrusive as possible. This calling convention behaves identical to the `C`
358 calling convention on how arguments and return values are passed, but it
359 uses a different set of caller/callee-saved registers. This alleviates the
360 burden of saving and recovering a large register set before and after the
361 call in the caller. If the arguments are passed in callee-saved registers,
362 then they will be preserved by the callee across the call. This doesn't
363 apply for values returned in callee-saved registers.
365 - On X86-64 the callee preserves all general purpose registers, except for
366 R11. R11 can be used as a scratch register. Floating-point registers
367 (XMMs/YMMs) are not preserved and need to be saved by the caller.
369 The idea behind this convention is to support calls to runtime functions
370 that have a hot path and a cold path. The hot path is usually a small piece
371 of code that doesn't 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>`, an opening
639 curly brace, a list of basic blocks, and a closing curly brace.
641 LLVM function declarations consist of the "``declare``" keyword, an
642 optional :ref:`linkage type <linkage>`, an optional :ref:`visibility
643 style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`,
644 an optional :ref:`calling convention <callingconv>`,
645 an optional ``unnamed_addr`` attribute, a return type, an optional
646 :ref:`parameter attribute <paramattrs>` for the return type, a function
647 name, a possibly empty list of arguments, an optional alignment, an optional
648 :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`,
649 and an optional :ref:`prologue <prologuedata>`.
651 A function definition contains a list of basic blocks, forming the CFG (Control
652 Flow Graph) for the function. Each basic block may optionally start with a label
653 (giving the basic block a symbol table entry), contains a list of instructions,
654 and ends with a :ref:`terminator <terminators>` instruction (such as a branch or
655 function return). If an explicit label is not provided, a block is assigned an
656 implicit numbered label, using the next value from the same counter as used for
657 unnamed temporaries (:ref:`see above<identifiers>`). For example, if a function
658 entry block does not have an explicit label, it will be assigned label "%0",
659 then the first unnamed temporary in that block will be "%1", etc.
661 The first basic block in a function is special in two ways: it is
662 immediately executed on entrance to the function, and it is not allowed
663 to have predecessor basic blocks (i.e. there can not be any branches to
664 the entry block of a function). Because the block can have no
665 predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`.
667 LLVM allows an explicit section to be specified for functions. If the
668 target supports it, it will emit functions to the section specified.
669 Additionally, the function can be placed in a COMDAT.
671 An explicit alignment may be specified for a function. If not present,
672 or if the alignment is set to zero, the alignment of the function is set
673 by the target to whatever it feels convenient. If an explicit alignment
674 is specified, the function is forced to have at least that much
675 alignment. All alignments must be a power of 2.
677 If the ``unnamed_addr`` attribute is given, the address is known to not
678 be significant and two identical functions can be merged.
682 define [linkage] [visibility] [DLLStorageClass]
684 <ResultType> @<FunctionName> ([argument list])
685 [unnamed_addr] [fn Attrs] [section "name"] [comdat [($name)]]
686 [align N] [gc] [prefix Constant] [prologue Constant] { ... }
688 The argument list is a comma seperated sequence of arguments where each
689 argument is of the following form
693 <type> [parameter Attrs] [name]
701 Aliases, unlike function or variables, don't create any new data. They
702 are just a new symbol and metadata for an existing position.
704 Aliases have a name and an aliasee that is either a global value or a
707 Aliases may have an optional :ref:`linkage type <linkage>`, an optional
708 :ref:`visibility style <visibility>`, an optional :ref:`DLL storage class
709 <dllstorageclass>` and an optional :ref:`tls model <tls_model>`.
713 @<Name> = [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal] [unnamed_addr] alias <AliaseeTy> @<Aliasee>
715 The linkage must be one of ``private``, ``internal``, ``linkonce``, ``weak``,
716 ``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers
717 might not correctly handle dropping a weak symbol that is aliased.
719 Aliases that are not ``unnamed_addr`` are guaranteed to have the same address as
720 the aliasee expression. ``unnamed_addr`` ones are only guaranteed to point
723 Since aliases are only a second name, some restrictions apply, of which
724 some can only be checked when producing an object file:
726 * The expression defining the aliasee must be computable at assembly
727 time. Since it is just a name, no relocations can be used.
729 * No alias in the expression can be weak as the possibility of the
730 intermediate alias being overridden cannot be represented in an
733 * No global value in the expression can be a declaration, since that
734 would require a relocation, which is not possible.
741 Comdat IR provides access to COFF and ELF object file COMDAT functionality.
743 Comdats have a name which represents the COMDAT key. All global objects that
744 specify this key will only end up in the final object file if the linker chooses
745 that key over some other key. Aliases are placed in the same COMDAT that their
746 aliasee computes to, if any.
748 Comdats have a selection kind to provide input on how the linker should
749 choose between keys in two different object files.
753 $<Name> = comdat SelectionKind
755 The selection kind must be one of the following:
758 The linker may choose any COMDAT key, the choice is arbitrary.
760 The linker may choose any COMDAT key but the sections must contain the
763 The linker will choose the section containing the largest COMDAT key.
765 The linker requires that only section with this COMDAT key exist.
767 The linker may choose any COMDAT key but the sections must contain the
770 Note that the Mach-O platform doesn't support COMDATs and ELF only supports
771 ``any`` as a selection kind.
773 Here is an example of a COMDAT group where a function will only be selected if
774 the COMDAT key's section is the largest:
778 $foo = comdat largest
779 @foo = global i32 2, comdat($foo)
781 define void @bar() comdat($foo) {
785 As a syntactic sugar the ``$name`` can be omitted if the name is the same as
791 @foo = global i32 2, comdat
794 In a COFF object file, this will create a COMDAT section with selection kind
795 ``IMAGE_COMDAT_SELECT_LARGEST`` containing the contents of the ``@foo`` symbol
796 and another COMDAT section with selection kind
797 ``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT
798 section and contains the contents of the ``@bar`` symbol.
800 There are some restrictions on the properties of the global object.
801 It, or an alias to it, must have the same name as the COMDAT group when
803 The contents and size of this object may be used during link-time to determine
804 which COMDAT groups get selected depending on the selection kind.
805 Because the name of the object must match the name of the COMDAT group, the
806 linkage of the global object must not be local; local symbols can get renamed
807 if a collision occurs in the symbol table.
809 The combined use of COMDATS and section attributes may yield surprising results.
816 @g1 = global i32 42, section "sec", comdat($foo)
817 @g2 = global i32 42, section "sec", comdat($bar)
819 From the object file perspective, this requires the creation of two sections
820 with the same name. This is necessary because both globals belong to different
821 COMDAT groups and COMDATs, at the object file level, are represented by
824 Note that certain IR constructs like global variables and functions may create
825 COMDATs in the object file in addition to any which are specified using COMDAT
826 IR. This arises, for example, when a global variable has linkonce_odr linkage.
828 .. _namedmetadatastructure:
833 Named metadata is a collection of metadata. :ref:`Metadata
834 nodes <metadata>` (but not metadata strings) are the only valid
835 operands for a named metadata.
839 ; Some unnamed metadata nodes, which are referenced by the named metadata.
844 !name = !{!0, !1, !2}
851 The return type and each parameter of a function type may have a set of
852 *parameter attributes* associated with them. Parameter attributes are
853 used to communicate additional information about the result or
854 parameters of a function. Parameter attributes are considered to be part
855 of the function, not of the function type, so functions with different
856 parameter attributes can have the same function type.
858 Parameter attributes are simple keywords that follow the type specified.
859 If multiple parameter attributes are needed, they are space separated.
864 declare i32 @printf(i8* noalias nocapture, ...)
865 declare i32 @atoi(i8 zeroext)
866 declare signext i8 @returns_signed_char()
868 Note that any attributes for the function result (``nounwind``,
869 ``readonly``) come immediately after the argument list.
871 Currently, only the following parameter attributes are defined:
874 This indicates to the code generator that the parameter or return
875 value should be zero-extended to the extent required by the target's
876 ABI (which is usually 32-bits, but is 8-bits for a i1 on x86-64) by
877 the caller (for a parameter) or the callee (for a return value).
879 This indicates to the code generator that the parameter or return
880 value should be sign-extended to the extent required by the target's
881 ABI (which is usually 32-bits) by the caller (for a parameter) or
882 the callee (for a return value).
884 This indicates that this parameter or return value should be treated
885 in a special target-dependent fashion during while emitting code for
886 a function call or return (usually, by putting it in a register as
887 opposed to memory, though some targets use it to distinguish between
888 two different kinds of registers). Use of this attribute is
891 This indicates that the pointer parameter should really be passed by
892 value to the function. The attribute implies that a hidden copy of
893 the pointee is made between the caller and the callee, so the callee
894 is unable to modify the value in the caller. This attribute is only
895 valid on LLVM pointer arguments. It is generally used to pass
896 structs and arrays by value, but is also valid on pointers to
897 scalars. The copy is considered to belong to the caller not the
898 callee (for example, ``readonly`` functions should not write to
899 ``byval`` parameters). This is not a valid attribute for return
902 The byval attribute also supports specifying an alignment with the
903 align attribute. It indicates the alignment of the stack slot to
904 form and the known alignment of the pointer specified to the call
905 site. If the alignment is not specified, then the code generator
906 makes a target-specific assumption.
912 The ``inalloca`` argument attribute allows the caller to take the
913 address of outgoing stack arguments. An ``inalloca`` argument must
914 be a pointer to stack memory produced by an ``alloca`` instruction.
915 The alloca, or argument allocation, must also be tagged with the
916 inalloca keyword. Only the last argument may have the ``inalloca``
917 attribute, and that argument is guaranteed to be passed in memory.
919 An argument allocation may be used by a call at most once because
920 the call may deallocate it. The ``inalloca`` attribute cannot be
921 used in conjunction with other attributes that affect argument
922 storage, like ``inreg``, ``nest``, ``sret``, or ``byval``. The
923 ``inalloca`` attribute also disables LLVM's implicit lowering of
924 large aggregate return values, which means that frontend authors
925 must lower them with ``sret`` pointers.
927 When the call site is reached, the argument allocation must have
928 been the most recent stack allocation that is still live, or the
929 results are undefined. It is possible to allocate additional stack
930 space after an argument allocation and before its call site, but it
931 must be cleared off with :ref:`llvm.stackrestore
934 See :doc:`InAlloca` for more information on how to use this
938 This indicates that the pointer parameter specifies the address of a
939 structure that is the return value of the function in the source
940 program. This pointer must be guaranteed by the caller to be valid:
941 loads and stores to the structure may be assumed by the callee
942 not to trap and to be properly aligned. This may only be applied to
943 the first parameter. This is not a valid attribute for return
947 This indicates that the pointer value may be assumed by the optimizer to
948 have the specified alignment.
950 Note that this attribute has additional semantics when combined with the
956 This indicates that objects accessed via pointer values
957 :ref:`based <pointeraliasing>` on the argument or return value are not also
958 accessed, during the execution of the function, via pointer values not
959 *based* on the argument or return value. The attribute on a return value
960 also has additional semantics described below. The caller shares the
961 responsibility with the callee for ensuring that these requirements are met.
962 For further details, please see the discussion of the NoAlias response in
963 :ref:`alias analysis <Must, May, or No>`.
965 Note that this definition of ``noalias`` is intentionally similar
966 to the definition of ``restrict`` in C99 for function arguments.
968 For function return values, C99's ``restrict`` is not meaningful,
969 while LLVM's ``noalias`` is. Furthermore, the semantics of the ``noalias``
970 attribute on return values are stronger than the semantics of the attribute
971 when used on function arguments. On function return values, the ``noalias``
972 attribute indicates that the function acts like a system memory allocation
973 function, returning a pointer to allocated storage disjoint from the
974 storage for any other object accessible to the caller.
977 This indicates that the callee does not make any copies of the
978 pointer that outlive the callee itself. This is not a valid
979 attribute for return values.
984 This indicates that the pointer parameter can be excised using the
985 :ref:`trampoline intrinsics <int_trampoline>`. This is not a valid
986 attribute for return values and can only be applied to one parameter.
989 This indicates that the function always returns the argument as its return
990 value. This is an optimization hint to the code generator when generating
991 the caller, allowing tail call optimization and omission of register saves
992 and restores in some cases; it is not checked or enforced when generating
993 the callee. The parameter and the function return type must be valid
994 operands for the :ref:`bitcast instruction <i_bitcast>`. This is not a
995 valid attribute for return values and can only be applied to one parameter.
998 This indicates that the parameter or return pointer is not null. This
999 attribute may only be applied to pointer typed parameters. This is not
1000 checked or enforced by LLVM, the caller must ensure that the pointer
1001 passed in is non-null, or the callee must ensure that the returned pointer
1004 ``dereferenceable(<n>)``
1005 This indicates that the parameter or return pointer is dereferenceable. This
1006 attribute may only be applied to pointer typed parameters. A pointer that
1007 is dereferenceable can be loaded from speculatively without a risk of
1008 trapping. The number of bytes known to be dereferenceable must be provided
1009 in parentheses. It is legal for the number of bytes to be less than the
1010 size of the pointee type. The ``nonnull`` attribute does not imply
1011 dereferenceability (consider a pointer to one element past the end of an
1012 array), however ``dereferenceable(<n>)`` does imply ``nonnull`` in
1013 ``addrspace(0)`` (which is the default address space).
1017 Garbage Collector Strategy Names
1018 --------------------------------
1020 Each function may specify a garbage collector strategy name, which is simply a
1023 .. code-block:: llvm
1025 define void @f() gc "name" { ... }
1027 The supported values of *name* includes those :ref:`built in to LLVM
1028 <builtin-gc-strategies>` and any provided by loaded plugins. Specifying a GC
1029 strategy will cause the compiler to alter its output in order to support the
1030 named garbage collection algorithm. Note that LLVM itself does not contain a
1031 garbage collector, this functionality is restricted to generating machine code
1032 which can interoperate with a collector provided externally.
1039 Prefix data is data associated with a function which the code
1040 generator will emit immediately before the function's entrypoint.
1041 The purpose of this feature is to allow frontends to associate
1042 language-specific runtime metadata with specific functions and make it
1043 available through the function pointer while still allowing the
1044 function pointer to be called.
1046 To access the data for a given function, a program may bitcast the
1047 function pointer to a pointer to the constant's type and dereference
1048 index -1. This implies that the IR symbol points just past the end of
1049 the prefix data. For instance, take the example of a function annotated
1050 with a single ``i32``,
1052 .. code-block:: llvm
1054 define void @f() prefix i32 123 { ... }
1056 The prefix data can be referenced as,
1058 .. code-block:: llvm
1060 %0 = bitcast *void () @f to *i32
1061 %a = getelementptr inbounds *i32 %0, i32 -1
1064 Prefix data is laid out as if it were an initializer for a global variable
1065 of the prefix data's type. The function will be placed such that the
1066 beginning of the prefix data is aligned. This means that if the size
1067 of the prefix data is not a multiple of the alignment size, the
1068 function's entrypoint will not be aligned. If alignment of the
1069 function's entrypoint is desired, padding must be added to the prefix
1072 A function may have prefix data but no body. This has similar semantics
1073 to the ``available_externally`` linkage in that the data may be used by the
1074 optimizers but will not be emitted in the object file.
1081 The ``prologue`` attribute allows arbitrary code (encoded as bytes) to
1082 be inserted prior to the function body. This can be used for enabling
1083 function hot-patching and instrumentation.
1085 To maintain the semantics of ordinary function calls, the prologue data must
1086 have a particular format. Specifically, it must begin with a sequence of
1087 bytes which decode to a sequence of machine instructions, valid for the
1088 module's target, which transfer control to the point immediately succeeding
1089 the prologue data, without performing any other visible action. This allows
1090 the inliner and other passes to reason about the semantics of the function
1091 definition without needing to reason about the prologue data. Obviously this
1092 makes the format of the prologue data highly target dependent.
1094 A trivial example of valid prologue data for the x86 architecture is ``i8 144``,
1095 which encodes the ``nop`` instruction:
1097 .. code-block:: llvm
1099 define void @f() prologue i8 144 { ... }
1101 Generally prologue data can be formed by encoding a relative branch instruction
1102 which skips the metadata, as in this example of valid prologue data for the
1103 x86_64 architecture, where the first two bytes encode ``jmp .+10``:
1105 .. code-block:: llvm
1107 %0 = type <{ i8, i8, i8* }>
1109 define void @f() prologue %0 <{ i8 235, i8 8, i8* @md}> { ... }
1111 A function may have prologue data but no body. This has similar semantics
1112 to the ``available_externally`` linkage in that the data may be used by the
1113 optimizers but will not be emitted in the object file.
1120 Attribute groups are groups of attributes that are referenced by objects within
1121 the IR. They are important for keeping ``.ll`` files readable, because a lot of
1122 functions will use the same set of attributes. In the degenerative case of a
1123 ``.ll`` file that corresponds to a single ``.c`` file, the single attribute
1124 group will capture the important command line flags used to build that file.
1126 An attribute group is a module-level object. To use an attribute group, an
1127 object references the attribute group's ID (e.g. ``#37``). An object may refer
1128 to more than one attribute group. In that situation, the attributes from the
1129 different groups are merged.
1131 Here is an example of attribute groups for a function that should always be
1132 inlined, has a stack alignment of 4, and which shouldn't use SSE instructions:
1134 .. code-block:: llvm
1136 ; Target-independent attributes:
1137 attributes #0 = { alwaysinline alignstack=4 }
1139 ; Target-dependent attributes:
1140 attributes #1 = { "no-sse" }
1142 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse".
1143 define void @f() #0 #1 { ... }
1150 Function attributes are set to communicate additional information about
1151 a function. Function attributes are considered to be part of the
1152 function, not of the function type, so functions with different function
1153 attributes can have the same function type.
1155 Function attributes are simple keywords that follow the type specified.
1156 If multiple attributes are needed, they are space separated. For
1159 .. code-block:: llvm
1161 define void @f() noinline { ... }
1162 define void @f() alwaysinline { ... }
1163 define void @f() alwaysinline optsize { ... }
1164 define void @f() optsize { ... }
1167 This attribute indicates that, when emitting the prologue and
1168 epilogue, the backend should forcibly align the stack pointer.
1169 Specify the desired alignment, which must be a power of two, in
1172 This attribute indicates that the inliner should attempt to inline
1173 this function into callers whenever possible, ignoring any active
1174 inlining size threshold for this caller.
1176 This indicates that the callee function at a call site should be
1177 recognized as a built-in function, even though the function's declaration
1178 uses the ``nobuiltin`` attribute. This is only valid at call sites for
1179 direct calls to functions that are declared with the ``nobuiltin``
1182 This attribute indicates that this function is rarely called. When
1183 computing edge weights, basic blocks post-dominated by a cold
1184 function call are also considered to be cold; and, thus, given low
1187 This attribute indicates that the source code contained a hint that
1188 inlining this function is desirable (such as the "inline" keyword in
1189 C/C++). It is just a hint; it imposes no requirements on the
1192 This attribute indicates that the function should be added to a
1193 jump-instruction table at code-generation time, and that all address-taken
1194 references to this function should be replaced with a reference to the
1195 appropriate jump-instruction-table function pointer. Note that this creates
1196 a new pointer for the original function, which means that code that depends
1197 on function-pointer identity can break. So, any function annotated with
1198 ``jumptable`` must also be ``unnamed_addr``.
1200 This attribute suggests that optimization passes and code generator
1201 passes make choices that keep the code size of this function as small
1202 as possible and perform optimizations that may sacrifice runtime
1203 performance in order to minimize the size of the generated code.
1205 This attribute disables prologue / epilogue emission for the
1206 function. This can have very system-specific consequences.
1208 This indicates that the callee function at a call site is not recognized as
1209 a built-in function. LLVM will retain the original call and not replace it
1210 with equivalent code based on the semantics of the built-in function, unless
1211 the call site uses the ``builtin`` attribute. This is valid at call sites
1212 and on function declarations and definitions.
1214 This attribute indicates that calls to the function cannot be
1215 duplicated. A call to a ``noduplicate`` function may be moved
1216 within its parent function, but may not be duplicated within
1217 its parent function.
1219 A function containing a ``noduplicate`` call may still
1220 be an inlining candidate, provided that the call is not
1221 duplicated by inlining. That implies that the function has
1222 internal linkage and only has one call site, so the original
1223 call is dead after inlining.
1225 This attributes disables implicit floating point instructions.
1227 This attribute indicates that the inliner should never inline this
1228 function in any situation. This attribute may not be used together
1229 with the ``alwaysinline`` attribute.
1231 This attribute suppresses lazy symbol binding for the function. This
1232 may make calls to the function faster, at the cost of extra program
1233 startup time if the function is not called during program startup.
1235 This attribute indicates that the code generator should not use a
1236 red zone, even if the target-specific ABI normally permits it.
1238 This function attribute indicates that the function never returns
1239 normally. This produces undefined behavior at runtime if the
1240 function ever does dynamically return.
1242 This function attribute indicates that the function never raises an
1243 exception. If the function does raise an exception, its runtime
1244 behavior is undefined. However, functions marked nounwind may still
1245 trap or generate asynchronous exceptions. Exception handling schemes
1246 that are recognized by LLVM to handle asynchronous exceptions, such
1247 as SEH, will still provide their implementation defined semantics.
1249 This function attribute indicates that the function is not optimized
1250 by any optimization or code generator passes with the
1251 exception of interprocedural optimization passes.
1252 This attribute cannot be used together with the ``alwaysinline``
1253 attribute; this attribute is also incompatible
1254 with the ``minsize`` attribute and the ``optsize`` attribute.
1256 This attribute requires the ``noinline`` attribute to be specified on
1257 the function as well, so the function is never inlined into any caller.
1258 Only functions with the ``alwaysinline`` attribute are valid
1259 candidates for inlining into the body of this function.
1261 This attribute suggests that optimization passes and code generator
1262 passes make choices that keep the code size of this function low,
1263 and otherwise do optimizations specifically to reduce code size as
1264 long as they do not significantly impact runtime performance.
1266 On a function, this attribute indicates that the function computes its
1267 result (or decides to unwind an exception) based strictly on its arguments,
1268 without dereferencing any pointer arguments or otherwise accessing
1269 any mutable state (e.g. memory, control registers, etc) visible to
1270 caller functions. It does not write through any pointer arguments
1271 (including ``byval`` arguments) and never changes any state visible
1272 to callers. This means that it cannot unwind exceptions by calling
1273 the ``C++`` exception throwing methods.
1275 On an argument, this attribute indicates that the function does not
1276 dereference that pointer argument, even though it may read or write the
1277 memory that the pointer points to if accessed through other pointers.
1279 On a function, this attribute indicates that the function does not write
1280 through any pointer arguments (including ``byval`` arguments) or otherwise
1281 modify any state (e.g. memory, control registers, etc) visible to
1282 caller functions. It may dereference pointer arguments and read
1283 state that may be set in the caller. A readonly function always
1284 returns the same value (or unwinds an exception identically) when
1285 called with the same set of arguments and global state. It cannot
1286 unwind an exception by calling the ``C++`` exception throwing
1289 On an argument, this attribute indicates that the function does not write
1290 through this pointer argument, even though it may write to the memory that
1291 the pointer points to.
1293 This attribute indicates that this function can return twice. The C
1294 ``setjmp`` is an example of such a function. The compiler disables
1295 some optimizations (like tail calls) in the caller of these
1297 ``sanitize_address``
1298 This attribute indicates that AddressSanitizer checks
1299 (dynamic address safety analysis) are enabled for this function.
1301 This attribute indicates that MemorySanitizer checks (dynamic detection
1302 of accesses to uninitialized memory) are enabled for this function.
1304 This attribute indicates that ThreadSanitizer checks
1305 (dynamic thread safety analysis) are enabled for this function.
1307 This attribute indicates that the function should emit a stack
1308 smashing protector. It is in the form of a "canary" --- a random value
1309 placed on the stack before the local variables that's checked upon
1310 return from the function to see if it has been overwritten. A
1311 heuristic is used to determine if a function needs stack protectors
1312 or not. The heuristic used will enable protectors for functions with:
1314 - Character arrays larger than ``ssp-buffer-size`` (default 8).
1315 - Aggregates containing character arrays larger than ``ssp-buffer-size``.
1316 - Calls to alloca() with variable sizes or constant sizes greater than
1317 ``ssp-buffer-size``.
1319 Variables that are identified as requiring a protector will be arranged
1320 on the stack such that they are adjacent to the stack protector guard.
1322 If a function that has an ``ssp`` attribute is inlined into a
1323 function that doesn't have an ``ssp`` attribute, then the resulting
1324 function will have an ``ssp`` attribute.
1326 This attribute indicates that the function should *always* emit a
1327 stack smashing protector. This overrides the ``ssp`` function
1330 Variables that are identified as requiring a protector will be arranged
1331 on the stack such that they are adjacent to the stack protector guard.
1332 The specific layout rules are:
1334 #. Large arrays and structures containing large arrays
1335 (``>= ssp-buffer-size``) are closest to the stack protector.
1336 #. Small arrays and structures containing small arrays
1337 (``< ssp-buffer-size``) are 2nd closest to the protector.
1338 #. Variables that have had their address taken are 3rd closest to the
1341 If a function that has an ``sspreq`` attribute is inlined into a
1342 function that doesn't have an ``sspreq`` attribute or which has an
1343 ``ssp`` or ``sspstrong`` attribute, then the resulting function will have
1344 an ``sspreq`` attribute.
1346 This attribute indicates that the function should emit a stack smashing
1347 protector. This attribute causes a strong heuristic to be used when
1348 determining if a function needs stack protectors. The strong heuristic
1349 will enable protectors for functions with:
1351 - Arrays of any size and type
1352 - Aggregates containing an array of any size and type.
1353 - Calls to alloca().
1354 - Local variables that have had their address taken.
1356 Variables that are identified as requiring a protector will be arranged
1357 on the stack such that they are adjacent to the stack protector guard.
1358 The specific layout rules are:
1360 #. Large arrays and structures containing large arrays
1361 (``>= ssp-buffer-size``) are closest to the stack protector.
1362 #. Small arrays and structures containing small arrays
1363 (``< ssp-buffer-size``) are 2nd closest to the protector.
1364 #. Variables that have had their address taken are 3rd closest to the
1367 This overrides the ``ssp`` function attribute.
1369 If a function that has an ``sspstrong`` attribute is inlined into a
1370 function that doesn't have an ``sspstrong`` attribute, then the
1371 resulting function will have an ``sspstrong`` attribute.
1373 This attribute indicates that the ABI being targeted requires that
1374 an unwind table entry be produce for this function even if we can
1375 show that no exceptions passes by it. This is normally the case for
1376 the ELF x86-64 abi, but it can be disabled for some compilation
1381 Module-Level Inline Assembly
1382 ----------------------------
1384 Modules may contain "module-level inline asm" blocks, which corresponds
1385 to the GCC "file scope inline asm" blocks. These blocks are internally
1386 concatenated by LLVM and treated as a single unit, but may be separated
1387 in the ``.ll`` file if desired. The syntax is very simple:
1389 .. code-block:: llvm
1391 module asm "inline asm code goes here"
1392 module asm "more can go here"
1394 The strings can contain any character by escaping non-printable
1395 characters. The escape sequence used is simply "\\xx" where "xx" is the
1396 two digit hex code for the number.
1398 The inline asm code is simply printed to the machine code .s file when
1399 assembly code is generated.
1401 .. _langref_datalayout:
1406 A module may specify a target specific data layout string that specifies
1407 how data is to be laid out in memory. The syntax for the data layout is
1410 .. code-block:: llvm
1412 target datalayout = "layout specification"
1414 The *layout specification* consists of a list of specifications
1415 separated by the minus sign character ('-'). Each specification starts
1416 with a letter and may include other information after the letter to
1417 define some aspect of the data layout. The specifications accepted are
1421 Specifies that the target lays out data in big-endian form. That is,
1422 the bits with the most significance have the lowest address
1425 Specifies that the target lays out data in little-endian form. That
1426 is, the bits with the least significance have the lowest address
1429 Specifies the natural alignment of the stack in bits. Alignment
1430 promotion of stack variables is limited to the natural stack
1431 alignment to avoid dynamic stack realignment. The stack alignment
1432 must be a multiple of 8-bits. If omitted, the natural stack
1433 alignment defaults to "unspecified", which does not prevent any
1434 alignment promotions.
1435 ``p[n]:<size>:<abi>:<pref>``
1436 This specifies the *size* of a pointer and its ``<abi>`` and
1437 ``<pref>``\erred alignments for address space ``n``. All sizes are in
1438 bits. The address space, ``n`` is optional, and if not specified,
1439 denotes the default address space 0. The value of ``n`` must be
1440 in the range [1,2^23).
1441 ``i<size>:<abi>:<pref>``
1442 This specifies the alignment for an integer type of a given bit
1443 ``<size>``. The value of ``<size>`` must be in the range [1,2^23).
1444 ``v<size>:<abi>:<pref>``
1445 This specifies the alignment for a vector type of a given bit
1447 ``f<size>:<abi>:<pref>``
1448 This specifies the alignment for a floating point type of a given bit
1449 ``<size>``. Only values of ``<size>`` that are supported by the target
1450 will work. 32 (float) and 64 (double) are supported on all targets; 80
1451 or 128 (different flavors of long double) are also supported on some
1454 This specifies the alignment for an object of aggregate type.
1456 If present, specifies that llvm names are mangled in the output. The
1459 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix.
1460 * ``m``: Mips mangling: Private symbols get a ``$`` prefix.
1461 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other
1462 symbols get a ``_`` prefix.
1463 * ``w``: Windows COFF prefix: Similar to Mach-O, but stdcall and fastcall
1464 functions also get a suffix based on the frame size.
1465 ``n<size1>:<size2>:<size3>...``
1466 This specifies a set of native integer widths for the target CPU in
1467 bits. For example, it might contain ``n32`` for 32-bit PowerPC,
1468 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of
1469 this set are considered to support most general arithmetic operations
1472 On every specification that takes a ``<abi>:<pref>``, specifying the
1473 ``<pref>`` alignment is optional. If omitted, the preceding ``:``
1474 should be omitted too and ``<pref>`` will be equal to ``<abi>``.
1476 When constructing the data layout for a given target, LLVM starts with a
1477 default set of specifications which are then (possibly) overridden by
1478 the specifications in the ``datalayout`` keyword. The default
1479 specifications are given in this list:
1481 - ``E`` - big endian
1482 - ``p:64:64:64`` - 64-bit pointers with 64-bit alignment.
1483 - ``p[n]:64:64:64`` - Other address spaces are assumed to be the
1484 same as the default address space.
1485 - ``S0`` - natural stack alignment is unspecified
1486 - ``i1:8:8`` - i1 is 8-bit (byte) aligned
1487 - ``i8:8:8`` - i8 is 8-bit (byte) aligned
1488 - ``i16:16:16`` - i16 is 16-bit aligned
1489 - ``i32:32:32`` - i32 is 32-bit aligned
1490 - ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred
1491 alignment of 64-bits
1492 - ``f16:16:16`` - half is 16-bit aligned
1493 - ``f32:32:32`` - float is 32-bit aligned
1494 - ``f64:64:64`` - double is 64-bit aligned
1495 - ``f128:128:128`` - quad is 128-bit aligned
1496 - ``v64:64:64`` - 64-bit vector is 64-bit aligned
1497 - ``v128:128:128`` - 128-bit vector is 128-bit aligned
1498 - ``a:0:64`` - aggregates are 64-bit aligned
1500 When LLVM is determining the alignment for a given type, it uses the
1503 #. If the type sought is an exact match for one of the specifications,
1504 that specification is used.
1505 #. If no match is found, and the type sought is an integer type, then
1506 the smallest integer type that is larger than the bitwidth of the
1507 sought type is used. If none of the specifications are larger than
1508 the bitwidth then the largest integer type is used. For example,
1509 given the default specifications above, the i7 type will use the
1510 alignment of i8 (next largest) while both i65 and i256 will use the
1511 alignment of i64 (largest specified).
1512 #. If no match is found, and the type sought is a vector type, then the
1513 largest vector type that is smaller than the sought vector type will
1514 be used as a fall back. This happens because <128 x double> can be
1515 implemented in terms of 64 <2 x double>, for example.
1517 The function of the data layout string may not be what you expect.
1518 Notably, this is not a specification from the frontend of what alignment
1519 the code generator should use.
1521 Instead, if specified, the target data layout is required to match what
1522 the ultimate *code generator* expects. This string is used by the
1523 mid-level optimizers to improve code, and this only works if it matches
1524 what the ultimate code generator uses. If you would like to generate IR
1525 that does not embed this target-specific detail into the IR, then you
1526 don't have to specify the string. This will disable some optimizations
1527 that require precise layout information, but this also prevents those
1528 optimizations from introducing target specificity into the IR.
1535 A module may specify a target triple string that describes the target
1536 host. The syntax for the target triple is simply:
1538 .. code-block:: llvm
1540 target triple = "x86_64-apple-macosx10.7.0"
1542 The *target triple* string consists of a series of identifiers delimited
1543 by the minus sign character ('-'). The canonical forms are:
1547 ARCHITECTURE-VENDOR-OPERATING_SYSTEM
1548 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT
1550 This information is passed along to the backend so that it generates
1551 code for the proper architecture. It's possible to override this on the
1552 command line with the ``-mtriple`` command line option.
1554 .. _pointeraliasing:
1556 Pointer Aliasing Rules
1557 ----------------------
1559 Any memory access must be done through a pointer value associated with
1560 an address range of the memory access, otherwise the behavior is
1561 undefined. Pointer values are associated with address ranges according
1562 to the following rules:
1564 - A pointer value is associated with the addresses associated with any
1565 value it is *based* on.
1566 - An address of a global variable is associated with the address range
1567 of the variable's storage.
1568 - The result value of an allocation instruction is associated with the
1569 address range of the allocated storage.
1570 - A null pointer in the default address-space is associated with no
1572 - An integer constant other than zero or a pointer value returned from
1573 a function not defined within LLVM may be associated with address
1574 ranges allocated through mechanisms other than those provided by
1575 LLVM. Such ranges shall not overlap with any ranges of addresses
1576 allocated by mechanisms provided by LLVM.
1578 A pointer value is *based* on another pointer value according to the
1581 - A pointer value formed from a ``getelementptr`` operation is *based*
1582 on the first operand of the ``getelementptr``.
1583 - The result value of a ``bitcast`` is *based* on the operand of the
1585 - A pointer value formed by an ``inttoptr`` is *based* on all pointer
1586 values that contribute (directly or indirectly) to the computation of
1587 the pointer's value.
1588 - The "*based* on" relationship is transitive.
1590 Note that this definition of *"based"* is intentionally similar to the
1591 definition of *"based"* in C99, though it is slightly weaker.
1593 LLVM IR does not associate types with memory. The result type of a
1594 ``load`` merely indicates the size and alignment of the memory from
1595 which to load, as well as the interpretation of the value. The first
1596 operand type of a ``store`` similarly only indicates the size and
1597 alignment of the store.
1599 Consequently, type-based alias analysis, aka TBAA, aka
1600 ``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR.
1601 :ref:`Metadata <metadata>` may be used to encode additional information
1602 which specialized optimization passes may use to implement type-based
1607 Volatile Memory Accesses
1608 ------------------------
1610 Certain memory accesses, such as :ref:`load <i_load>`'s,
1611 :ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be
1612 marked ``volatile``. The optimizers must not change the number of
1613 volatile operations or change their order of execution relative to other
1614 volatile operations. The optimizers *may* change the order of volatile
1615 operations relative to non-volatile operations. This is not Java's
1616 "volatile" and has no cross-thread synchronization behavior.
1618 IR-level volatile loads and stores cannot safely be optimized into
1619 llvm.memcpy or llvm.memmove intrinsics even when those intrinsics are
1620 flagged volatile. Likewise, the backend should never split or merge
1621 target-legal volatile load/store instructions.
1623 .. admonition:: Rationale
1625 Platforms may rely on volatile loads and stores of natively supported
1626 data width to be executed as single instruction. For example, in C
1627 this holds for an l-value of volatile primitive type with native
1628 hardware support, but not necessarily for aggregate types. The
1629 frontend upholds these expectations, which are intentionally
1630 unspecified in the IR. The rules above ensure that IR transformation
1631 do not violate the frontend's contract with the language.
1635 Memory Model for Concurrent Operations
1636 --------------------------------------
1638 The LLVM IR does not define any way to start parallel threads of
1639 execution or to register signal handlers. Nonetheless, there are
1640 platform-specific ways to create them, and we define LLVM IR's behavior
1641 in their presence. This model is inspired by the C++0x memory model.
1643 For a more informal introduction to this model, see the :doc:`Atomics`.
1645 We define a *happens-before* partial order as the least partial order
1648 - Is a superset of single-thread program order, and
1649 - When a *synchronizes-with* ``b``, includes an edge from ``a`` to
1650 ``b``. *Synchronizes-with* pairs are introduced by platform-specific
1651 techniques, like pthread locks, thread creation, thread joining,
1652 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering
1653 Constraints <ordering>`).
1655 Note that program order does not introduce *happens-before* edges
1656 between a thread and signals executing inside that thread.
1658 Every (defined) read operation (load instructions, memcpy, atomic
1659 loads/read-modify-writes, etc.) R reads a series of bytes written by
1660 (defined) write operations (store instructions, atomic
1661 stores/read-modify-writes, memcpy, etc.). For the purposes of this
1662 section, initialized globals are considered to have a write of the
1663 initializer which is atomic and happens before any other read or write
1664 of the memory in question. For each byte of a read R, R\ :sub:`byte`
1665 may see any write to the same byte, except:
1667 - If write\ :sub:`1` happens before write\ :sub:`2`, and
1668 write\ :sub:`2` happens before R\ :sub:`byte`, then
1669 R\ :sub:`byte` does not see write\ :sub:`1`.
1670 - If R\ :sub:`byte` happens before write\ :sub:`3`, then
1671 R\ :sub:`byte` does not see write\ :sub:`3`.
1673 Given that definition, R\ :sub:`byte` is defined as follows:
1675 - If R is volatile, the result is target-dependent. (Volatile is
1676 supposed to give guarantees which can support ``sig_atomic_t`` in
1677 C/C++, and may be used for accesses to addresses that do not behave
1678 like normal memory. It does not generally provide cross-thread
1680 - Otherwise, if there is no write to the same byte that happens before
1681 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte.
1682 - Otherwise, if R\ :sub:`byte` may see exactly one write,
1683 R\ :sub:`byte` returns the value written by that write.
1684 - Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may
1685 see are atomic, it chooses one of the values written. See the :ref:`Atomic
1686 Memory Ordering Constraints <ordering>` section for additional
1687 constraints on how the choice is made.
1688 - Otherwise R\ :sub:`byte` returns ``undef``.
1690 R returns the value composed of the series of bytes it read. This
1691 implies that some bytes within the value may be ``undef`` **without**
1692 the entire value being ``undef``. Note that this only defines the
1693 semantics of the operation; it doesn't mean that targets will emit more
1694 than one instruction to read the series of bytes.
1696 Note that in cases where none of the atomic intrinsics are used, this
1697 model places only one restriction on IR transformations on top of what
1698 is required for single-threaded execution: introducing a store to a byte
1699 which might not otherwise be stored is not allowed in general.
1700 (Specifically, in the case where another thread might write to and read
1701 from an address, introducing a store can change a load that may see
1702 exactly one write into a load that may see multiple writes.)
1706 Atomic Memory Ordering Constraints
1707 ----------------------------------
1709 Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`,
1710 :ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`,
1711 :ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take
1712 ordering parameters that determine which other atomic instructions on
1713 the same address they *synchronize with*. These semantics are borrowed
1714 from Java and C++0x, but are somewhat more colloquial. If these
1715 descriptions aren't precise enough, check those specs (see spec
1716 references in the :doc:`atomics guide <Atomics>`).
1717 :ref:`fence <i_fence>` instructions treat these orderings somewhat
1718 differently since they don't take an address. See that instruction's
1719 documentation for details.
1721 For a simpler introduction to the ordering constraints, see the
1725 The set of values that can be read is governed by the happens-before
1726 partial order. A value cannot be read unless some operation wrote
1727 it. This is intended to provide a guarantee strong enough to model
1728 Java's non-volatile shared variables. This ordering cannot be
1729 specified for read-modify-write operations; it is not strong enough
1730 to make them atomic in any interesting way.
1732 In addition to the guarantees of ``unordered``, there is a single
1733 total order for modifications by ``monotonic`` operations on each
1734 address. All modification orders must be compatible with the
1735 happens-before order. There is no guarantee that the modification
1736 orders can be combined to a global total order for the whole program
1737 (and this often will not be possible). The read in an atomic
1738 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and
1739 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification
1740 order immediately before the value it writes. If one atomic read
1741 happens before another atomic read of the same address, the later
1742 read must see the same value or a later value in the address's
1743 modification order. This disallows reordering of ``monotonic`` (or
1744 stronger) operations on the same address. If an address is written
1745 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally
1746 read that address repeatedly, the other threads must eventually see
1747 the write. This corresponds to the C++0x/C1x
1748 ``memory_order_relaxed``.
1750 In addition to the guarantees of ``monotonic``, a
1751 *synchronizes-with* edge may be formed with a ``release`` operation.
1752 This is intended to model C++'s ``memory_order_acquire``.
1754 In addition to the guarantees of ``monotonic``, if this operation
1755 writes a value which is subsequently read by an ``acquire``
1756 operation, it *synchronizes-with* that operation. (This isn't a
1757 complete description; see the C++0x definition of a release
1758 sequence.) This corresponds to the C++0x/C1x
1759 ``memory_order_release``.
1760 ``acq_rel`` (acquire+release)
1761 Acts as both an ``acquire`` and ``release`` operation on its
1762 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``.
1763 ``seq_cst`` (sequentially consistent)
1764 In addition to the guarantees of ``acq_rel`` (``acquire`` for an
1765 operation that only reads, ``release`` for an operation that only
1766 writes), there is a global total order on all
1767 sequentially-consistent operations on all addresses, which is
1768 consistent with the *happens-before* partial order and with the
1769 modification orders of all the affected addresses. Each
1770 sequentially-consistent read sees the last preceding write to the
1771 same address in this global order. This corresponds to the C++0x/C1x
1772 ``memory_order_seq_cst`` and Java volatile.
1776 If an atomic operation is marked ``singlethread``, it only *synchronizes
1777 with* or participates in modification and seq\_cst total orderings with
1778 other operations running in the same thread (for example, in signal
1786 LLVM IR floating-point binary ops (:ref:`fadd <i_fadd>`,
1787 :ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`,
1788 :ref:`frem <i_frem>`) have the following flags that can be set to enable
1789 otherwise unsafe floating point operations
1792 No NaNs - Allow optimizations to assume the arguments and result are not
1793 NaN. Such optimizations are required to retain defined behavior over
1794 NaNs, but the value of the result is undefined.
1797 No Infs - Allow optimizations to assume the arguments and result are not
1798 +/-Inf. Such optimizations are required to retain defined behavior over
1799 +/-Inf, but the value of the result is undefined.
1802 No Signed Zeros - Allow optimizations to treat the sign of a zero
1803 argument or result as insignificant.
1806 Allow Reciprocal - Allow optimizations to use the reciprocal of an
1807 argument rather than perform division.
1810 Fast - Allow algebraically equivalent transformations that may
1811 dramatically change results in floating point (e.g. reassociate). This
1812 flag implies all the others.
1816 Use-list Order Directives
1817 -------------------------
1819 Use-list directives encode the in-memory order of each use-list, allowing the
1820 order to be recreated. ``<order-indexes>`` is a comma-separated list of
1821 indexes that are assigned to the referenced value's uses. The referenced
1822 value's use-list is immediately sorted by these indexes.
1824 Use-list directives may appear at function scope or global scope. They are not
1825 instructions, and have no effect on the semantics of the IR. When they're at
1826 function scope, they must appear after the terminator of the final basic block.
1828 If basic blocks have their address taken via ``blockaddress()`` expressions,
1829 ``uselistorder_bb`` can be used to reorder their use-lists from outside their
1836 uselistorder <ty> <value>, { <order-indexes> }
1837 uselistorder_bb @function, %block { <order-indexes> }
1843 define void @foo(i32 %arg1, i32 %arg2) {
1845 ; ... instructions ...
1847 ; ... instructions ...
1849 ; At function scope.
1850 uselistorder i32 %arg1, { 1, 0, 2 }
1851 uselistorder label %bb, { 1, 0 }
1855 uselistorder i32* @global, { 1, 2, 0 }
1856 uselistorder i32 7, { 1, 0 }
1857 uselistorder i32 (i32) @bar, { 1, 0 }
1858 uselistorder_bb @foo, %bb, { 5, 1, 3, 2, 0, 4 }
1865 The LLVM type system is one of the most important features of the
1866 intermediate representation. Being typed enables a number of
1867 optimizations to be performed on the intermediate representation
1868 directly, without having to do extra analyses on the side before the
1869 transformation. A strong type system makes it easier to read the
1870 generated code and enables novel analyses and transformations that are
1871 not feasible to perform on normal three address code representations.
1881 The void type does not represent any value and has no size.
1899 The function type can be thought of as a function signature. It consists of a
1900 return type and a list of formal parameter types. The return type of a function
1901 type is a void type or first class type --- except for :ref:`label <t_label>`
1902 and :ref:`metadata <t_metadata>` types.
1908 <returntype> (<parameter list>)
1910 ...where '``<parameter list>``' is a comma-separated list of type
1911 specifiers. Optionally, the parameter list may include a type ``...``, which
1912 indicates that the function takes a variable number of arguments. Variable
1913 argument functions can access their arguments with the :ref:`variable argument
1914 handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type
1915 except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`.
1919 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1920 | ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` |
1921 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1922 | ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. |
1923 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1924 | ``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. |
1925 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1926 | ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values |
1927 +---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1934 The :ref:`first class <t_firstclass>` types are perhaps the most important.
1935 Values of these types are the only ones which can be produced by
1943 These are the types that are valid in registers from CodeGen's perspective.
1952 The integer type is a very simple type that simply specifies an
1953 arbitrary bit width for the integer type desired. Any bit width from 1
1954 bit to 2\ :sup:`23`\ -1 (about 8 million) can be specified.
1962 The number of bits the integer will occupy is specified by the ``N``
1968 +----------------+------------------------------------------------+
1969 | ``i1`` | a single-bit integer. |
1970 +----------------+------------------------------------------------+
1971 | ``i32`` | a 32-bit integer. |
1972 +----------------+------------------------------------------------+
1973 | ``i1942652`` | a really big integer of over 1 million bits. |
1974 +----------------+------------------------------------------------+
1978 Floating Point Types
1979 """"""""""""""""""""
1988 - 16-bit floating point value
1991 - 32-bit floating point value
1994 - 64-bit floating point value
1997 - 128-bit floating point value (112-bit mantissa)
2000 - 80-bit floating point value (X87)
2003 - 128-bit floating point value (two 64-bits)
2010 The x86_mmx type represents a value held in an MMX register on an x86
2011 machine. The operations allowed on it are quite limited: parameters and
2012 return values, load and store, and bitcast. User-specified MMX
2013 instructions are represented as intrinsic or asm calls with arguments
2014 and/or results of this type. There are no arrays, vectors or constants
2031 The pointer type is used to specify memory locations. Pointers are
2032 commonly used to reference objects in memory.
2034 Pointer types may have an optional address space attribute defining the
2035 numbered address space where the pointed-to object resides. The default
2036 address space is number zero. The semantics of non-zero address spaces
2037 are target-specific.
2039 Note that LLVM does not permit pointers to void (``void*``) nor does it
2040 permit pointers to labels (``label*``). Use ``i8*`` instead.
2050 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2051 | ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. |
2052 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2053 | ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. |
2054 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2055 | ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space #5. |
2056 +-------------------------+--------------------------------------------------------------------------------------------------------------+
2065 A vector type is a simple derived type that represents a vector of
2066 elements. Vector types are used when multiple primitive data are
2067 operated in parallel using a single instruction (SIMD). A vector type
2068 requires a size (number of elements) and an underlying primitive data
2069 type. Vector types are considered :ref:`first class <t_firstclass>`.
2075 < <# elements> x <elementtype> >
2077 The number of elements is a constant integer value larger than 0;
2078 elementtype may be any integer, floating point or pointer type. Vectors
2079 of size zero are not allowed.
2083 +-------------------+--------------------------------------------------+
2084 | ``<4 x i32>`` | Vector of 4 32-bit integer values. |
2085 +-------------------+--------------------------------------------------+
2086 | ``<8 x float>`` | Vector of 8 32-bit floating-point values. |
2087 +-------------------+--------------------------------------------------+
2088 | ``<2 x i64>`` | Vector of 2 64-bit integer values. |
2089 +-------------------+--------------------------------------------------+
2090 | ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. |
2091 +-------------------+--------------------------------------------------+
2100 The label type represents code labels.
2115 The metadata type represents embedded metadata. No derived types may be
2116 created from metadata except for :ref:`function <t_function>` arguments.
2129 Aggregate Types are a subset of derived types that can contain multiple
2130 member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are
2131 aggregate types. :ref:`Vectors <t_vector>` are not considered to be
2141 The array type is a very simple derived type that arranges elements
2142 sequentially in memory. The array type requires a size (number of
2143 elements) and an underlying data type.
2149 [<# elements> x <elementtype>]
2151 The number of elements is a constant integer value; ``elementtype`` may
2152 be any type with a size.
2156 +------------------+--------------------------------------+
2157 | ``[40 x i32]`` | Array of 40 32-bit integer values. |
2158 +------------------+--------------------------------------+
2159 | ``[41 x i32]`` | Array of 41 32-bit integer values. |
2160 +------------------+--------------------------------------+
2161 | ``[4 x i8]`` | Array of 4 8-bit integer values. |
2162 +------------------+--------------------------------------+
2164 Here are some examples of multidimensional arrays:
2166 +-----------------------------+----------------------------------------------------------+
2167 | ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. |
2168 +-----------------------------+----------------------------------------------------------+
2169 | ``[12 x [10 x float]]`` | 12x10 array of single precision floating point values. |
2170 +-----------------------------+----------------------------------------------------------+
2171 | ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. |
2172 +-----------------------------+----------------------------------------------------------+
2174 There is no restriction on indexing beyond the end of the array implied
2175 by a static type (though there are restrictions on indexing beyond the
2176 bounds of an allocated object in some cases). This means that
2177 single-dimension 'variable sized array' addressing can be implemented in
2178 LLVM with a zero length array type. An implementation of 'pascal style
2179 arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for
2189 The structure type is used to represent a collection of data members
2190 together in memory. The elements of a structure may be any type that has
2193 Structures in memory are accessed using '``load``' and '``store``' by
2194 getting a pointer to a field with the '``getelementptr``' instruction.
2195 Structures in registers are accessed using the '``extractvalue``' and
2196 '``insertvalue``' instructions.
2198 Structures may optionally be "packed" structures, which indicate that
2199 the alignment of the struct is one byte, and that there is no padding
2200 between the elements. In non-packed structs, padding between field types
2201 is inserted as defined by the DataLayout string in the module, which is
2202 required to match what the underlying code generator expects.
2204 Structures can either be "literal" or "identified". A literal structure
2205 is defined inline with other types (e.g. ``{i32, i32}*``) whereas
2206 identified types are always defined at the top level with a name.
2207 Literal types are uniqued by their contents and can never be recursive
2208 or opaque since there is no way to write one. Identified types can be
2209 recursive, can be opaqued, and are never uniqued.
2215 %T1 = type { <type list> } ; Identified normal struct type
2216 %T2 = type <{ <type list> }> ; Identified packed struct type
2220 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2221 | ``{ i32, i32, i32 }`` | A triple of three ``i32`` values |
2222 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2223 | ``{ 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``. |
2224 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2225 | ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. |
2226 +------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
2230 Opaque Structure Types
2231 """"""""""""""""""""""
2235 Opaque structure types are used to represent named structure types that
2236 do not have a body specified. This corresponds (for example) to the C
2237 notion of a forward declared structure.
2248 +--------------+-------------------+
2249 | ``opaque`` | An opaque type. |
2250 +--------------+-------------------+
2257 LLVM has several different basic types of constants. This section
2258 describes them all and their syntax.
2263 **Boolean constants**
2264 The two strings '``true``' and '``false``' are both valid constants
2266 **Integer constants**
2267 Standard integers (such as '4') are constants of the
2268 :ref:`integer <t_integer>` type. Negative numbers may be used with
2270 **Floating point constants**
2271 Floating point constants use standard decimal notation (e.g.
2272 123.421), exponential notation (e.g. 1.23421e+2), or a more precise
2273 hexadecimal notation (see below). The assembler requires the exact
2274 decimal value of a floating-point constant. For example, the
2275 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating
2276 decimal in binary. Floating point constants must have a :ref:`floating
2277 point <t_floating>` type.
2278 **Null pointer constants**
2279 The identifier '``null``' is recognized as a null pointer constant
2280 and must be of :ref:`pointer type <t_pointer>`.
2282 The one non-intuitive notation for constants is the hexadecimal form of
2283 floating point constants. For example, the form
2284 '``double 0x432ff973cafa8000``' is equivalent to (but harder to read
2285 than) '``double 4.5e+15``'. The only time hexadecimal floating point
2286 constants are required (and the only time that they are generated by the
2287 disassembler) is when a floating point constant must be emitted but it
2288 cannot be represented as a decimal floating point number in a reasonable
2289 number of digits. For example, NaN's, infinities, and other special
2290 values are represented in their IEEE hexadecimal format so that assembly
2291 and disassembly do not cause any bits to change in the constants.
2293 When using the hexadecimal form, constants of types half, float, and
2294 double are represented using the 16-digit form shown above (which
2295 matches the IEEE754 representation for double); half and float values
2296 must, however, be exactly representable as IEEE 754 half and single
2297 precision, respectively. Hexadecimal format is always used for long
2298 double, and there are three forms of long double. The 80-bit format used
2299 by x86 is represented as ``0xK`` followed by 20 hexadecimal digits. The
2300 128-bit format used by PowerPC (two adjacent doubles) is represented by
2301 ``0xM`` followed by 32 hexadecimal digits. The IEEE 128-bit format is
2302 represented by ``0xL`` followed by 32 hexadecimal digits. Long doubles
2303 will only work if they match the long double format on your target.
2304 The IEEE 16-bit format (half precision) is represented by ``0xH``
2305 followed by 4 hexadecimal digits. All hexadecimal formats are big-endian
2306 (sign bit at the left).
2308 There are no constants of type x86_mmx.
2310 .. _complexconstants:
2315 Complex constants are a (potentially recursive) combination of simple
2316 constants and smaller complex constants.
2318 **Structure constants**
2319 Structure constants are represented with notation similar to
2320 structure type definitions (a comma separated list of elements,
2321 surrounded by braces (``{}``)). For example:
2322 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as
2323 "``@G = external global i32``". Structure constants must have
2324 :ref:`structure type <t_struct>`, and the number and types of elements
2325 must match those specified by the type.
2327 Array constants are represented with notation similar to array type
2328 definitions (a comma separated list of elements, surrounded by
2329 square brackets (``[]``)). For example:
2330 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have
2331 :ref:`array type <t_array>`, and the number and types of elements must
2332 match those specified by the type. As a special case, character array
2333 constants may also be represented as a double-quoted string using the ``c``
2334 prefix. For example: "``c"Hello World\0A\00"``".
2335 **Vector constants**
2336 Vector constants are represented with notation similar to vector
2337 type definitions (a comma separated list of elements, surrounded by
2338 less-than/greater-than's (``<>``)). For example:
2339 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants
2340 must have :ref:`vector type <t_vector>`, and the number and types of
2341 elements must match those specified by the type.
2342 **Zero initialization**
2343 The string '``zeroinitializer``' can be used to zero initialize a
2344 value to zero of *any* type, including scalar and
2345 :ref:`aggregate <t_aggregate>` types. This is often used to avoid
2346 having to print large zero initializers (e.g. for large arrays) and
2347 is always exactly equivalent to using explicit zero initializers.
2349 A metadata node is a constant tuple without types. For example:
2350 "``!{!0, !{!2, !0}, !"test"}``". Metadata can reference constant values,
2351 for example: "``!{!0, i32 0, i8* @global, i64 (i64)* @function, !"str"}``".
2352 Unlike other typed constants that are meant to be interpreted as part of
2353 the instruction stream, metadata is a place to attach additional
2354 information such as debug info.
2356 Global Variable and Function Addresses
2357 --------------------------------------
2359 The addresses of :ref:`global variables <globalvars>` and
2360 :ref:`functions <functionstructure>` are always implicitly valid
2361 (link-time) constants. These constants are explicitly referenced when
2362 the :ref:`identifier for the global <identifiers>` is used and always have
2363 :ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM
2366 .. code-block:: llvm
2370 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2377 The string '``undef``' can be used anywhere a constant is expected, and
2378 indicates that the user of the value may receive an unspecified
2379 bit-pattern. Undefined values may be of any type (other than '``label``'
2380 or '``void``') and be used anywhere a constant is permitted.
2382 Undefined values are useful because they indicate to the compiler that
2383 the program is well defined no matter what value is used. This gives the
2384 compiler more freedom to optimize. Here are some examples of
2385 (potentially surprising) transformations that are valid (in pseudo IR):
2387 .. code-block:: llvm
2397 This is safe because all of the output bits are affected by the undef
2398 bits. Any output bit can have a zero or one depending on the input bits.
2400 .. code-block:: llvm
2411 These logical operations have bits that are not always affected by the
2412 input. For example, if ``%X`` has a zero bit, then the output of the
2413 '``and``' operation will always be a zero for that bit, no matter what
2414 the corresponding bit from the '``undef``' is. As such, it is unsafe to
2415 optimize or assume that the result of the '``and``' is '``undef``'.
2416 However, it is safe to assume that all bits of the '``undef``' could be
2417 0, and optimize the '``and``' to 0. Likewise, it is safe to assume that
2418 all the bits of the '``undef``' operand to the '``or``' could be set,
2419 allowing the '``or``' to be folded to -1.
2421 .. code-block:: llvm
2423 %A = select undef, %X, %Y
2424 %B = select undef, 42, %Y
2425 %C = select %X, %Y, undef
2435 This set of examples shows that undefined '``select``' (and conditional
2436 branch) conditions can go *either way*, but they have to come from one
2437 of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were
2438 both known to have a clear low bit, then ``%A`` would have to have a
2439 cleared low bit. However, in the ``%C`` example, the optimizer is
2440 allowed to assume that the '``undef``' operand could be the same as
2441 ``%Y``, allowing the whole '``select``' to be eliminated.
2443 .. code-block:: llvm
2445 %A = xor undef, undef
2462 This example points out that two '``undef``' operands are not
2463 necessarily the same. This can be surprising to people (and also matches
2464 C semantics) where they assume that "``X^X``" is always zero, even if
2465 ``X`` is undefined. This isn't true for a number of reasons, but the
2466 short answer is that an '``undef``' "variable" can arbitrarily change
2467 its value over its "live range". This is true because the variable
2468 doesn't actually *have a live range*. Instead, the value is logically
2469 read from arbitrary registers that happen to be around when needed, so
2470 the value is not necessarily consistent over time. In fact, ``%A`` and
2471 ``%C`` need to have the same semantics or the core LLVM "replace all
2472 uses with" concept would not hold.
2474 .. code-block:: llvm
2482 These examples show the crucial difference between an *undefined value*
2483 and *undefined behavior*. An undefined value (like '``undef``') is
2484 allowed to have an arbitrary bit-pattern. This means that the ``%A``
2485 operation can be constant folded to '``undef``', because the '``undef``'
2486 could be an SNaN, and ``fdiv`` is not (currently) defined on SNaN's.
2487 However, in the second example, we can make a more aggressive
2488 assumption: because the ``undef`` is allowed to be an arbitrary value,
2489 we are allowed to assume that it could be zero. Since a divide by zero
2490 has *undefined behavior*, we are allowed to assume that the operation
2491 does not execute at all. This allows us to delete the divide and all
2492 code after it. Because the undefined operation "can't happen", the
2493 optimizer can assume that it occurs in dead code.
2495 .. code-block:: llvm
2497 a: store undef -> %X
2498 b: store %X -> undef
2503 These examples reiterate the ``fdiv`` example: a store *of* an undefined
2504 value can be assumed to not have any effect; we can assume that the
2505 value is overwritten with bits that happen to match what was already
2506 there. However, a store *to* an undefined location could clobber
2507 arbitrary memory, therefore, it has undefined behavior.
2514 Poison values are similar to :ref:`undef values <undefvalues>`, however
2515 they also represent the fact that an instruction or constant expression
2516 that cannot evoke side effects has nevertheless detected a condition
2517 that results in undefined behavior.
2519 There is currently no way of representing a poison value in the IR; they
2520 only exist when produced by operations such as :ref:`add <i_add>` with
2523 Poison value behavior is defined in terms of value *dependence*:
2525 - Values other than :ref:`phi <i_phi>` nodes depend on their operands.
2526 - :ref:`Phi <i_phi>` nodes depend on the operand corresponding to
2527 their dynamic predecessor basic block.
2528 - Function arguments depend on the corresponding actual argument values
2529 in the dynamic callers of their functions.
2530 - :ref:`Call <i_call>` instructions depend on the :ref:`ret <i_ret>`
2531 instructions that dynamically transfer control back to them.
2532 - :ref:`Invoke <i_invoke>` instructions depend on the
2533 :ref:`ret <i_ret>`, :ref:`resume <i_resume>`, or exception-throwing
2534 call instructions that dynamically transfer control back to them.
2535 - Non-volatile loads and stores depend on the most recent stores to all
2536 of the referenced memory addresses, following the order in the IR
2537 (including loads and stores implied by intrinsics such as
2538 :ref:`@llvm.memcpy <int_memcpy>`.)
2539 - An instruction with externally visible side effects depends on the
2540 most recent preceding instruction with externally visible side
2541 effects, following the order in the IR. (This includes :ref:`volatile
2542 operations <volatile>`.)
2543 - An instruction *control-depends* on a :ref:`terminator
2544 instruction <terminators>` if the terminator instruction has
2545 multiple successors and the instruction is always executed when
2546 control transfers to one of the successors, and may not be executed
2547 when control is transferred to another.
2548 - Additionally, an instruction also *control-depends* on a terminator
2549 instruction if the set of instructions it otherwise depends on would
2550 be different if the terminator had transferred control to a different
2552 - Dependence is transitive.
2554 Poison values have the same behavior as :ref:`undef values <undefvalues>`,
2555 with the additional effect that any instruction that has a *dependence*
2556 on a poison value has undefined behavior.
2558 Here are some examples:
2560 .. code-block:: llvm
2563 %poison = sub nuw i32 0, 1 ; Results in a poison value.
2564 %still_poison = and i32 %poison, 0 ; 0, but also poison.
2565 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
2566 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
2568 store i32 %poison, i32* @g ; Poison value stored to memory.
2569 %poison2 = load i32* @g ; Poison value loaded back from memory.
2571 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
2573 %narrowaddr = bitcast i32* @g to i16*
2574 %wideaddr = bitcast i32* @g to i64*
2575 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2576 %poison4 = load i64* %wideaddr ; Returns a poison value.
2578 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2579 br i1 %cmp, label %true, label %end ; Branch to either destination.
2582 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2583 ; it has undefined behavior.
2587 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2588 ; Both edges into this PHI are
2589 ; control-dependent on %cmp, so this
2590 ; always results in a poison value.
2592 store volatile i32 0, i32* @g ; This would depend on the store in %true
2593 ; if %cmp is true, or the store in %entry
2594 ; otherwise, so this is undefined behavior.
2596 br i1 %cmp, label %second_true, label %second_end
2597 ; The same branch again, but this time the
2598 ; true block doesn't have side effects.
2605 store volatile i32 0, i32* @g ; This time, the instruction always depends
2606 ; on the store in %end. Also, it is
2607 ; control-equivalent to %end, so this is
2608 ; well-defined (ignoring earlier undefined
2609 ; behavior in this example).
2613 Addresses of Basic Blocks
2614 -------------------------
2616 ``blockaddress(@function, %block)``
2618 The '``blockaddress``' constant computes the address of the specified
2619 basic block in the specified function, and always has an ``i8*`` type.
2620 Taking the address of the entry block is illegal.
2622 This value only has defined behavior when used as an operand to the
2623 ':ref:`indirectbr <i_indirectbr>`' instruction, or for comparisons
2624 against null. Pointer equality tests between labels addresses results in
2625 undefined behavior --- though, again, comparison against null is ok, and
2626 no label is equal to the null pointer. This may be passed around as an
2627 opaque pointer sized value as long as the bits are not inspected. This
2628 allows ``ptrtoint`` and arithmetic to be performed on these values so
2629 long as the original value is reconstituted before the ``indirectbr``
2632 Finally, some targets may provide defined semantics when using the value
2633 as the operand to an inline assembly, but that is target specific.
2637 Constant Expressions
2638 --------------------
2640 Constant expressions are used to allow expressions involving other
2641 constants to be used as constants. Constant expressions may be of any
2642 :ref:`first class <t_firstclass>` type and may involve any LLVM operation
2643 that does not have side effects (e.g. load and call are not supported).
2644 The following is the syntax for constant expressions:
2646 ``trunc (CST to TYPE)``
2647 Truncate a constant to another type. The bit size of CST must be
2648 larger than the bit size of TYPE. Both types must be integers.
2649 ``zext (CST to TYPE)``
2650 Zero extend a constant to another type. The bit size of CST must be
2651 smaller than the bit size of TYPE. Both types must be integers.
2652 ``sext (CST to TYPE)``
2653 Sign extend a constant to another type. The bit size of CST must be
2654 smaller than the bit size of TYPE. Both types must be integers.
2655 ``fptrunc (CST to TYPE)``
2656 Truncate a floating point constant to another floating point type.
2657 The size of CST must be larger than the size of TYPE. Both types
2658 must be floating point.
2659 ``fpext (CST to TYPE)``
2660 Floating point extend a constant to another type. The size of CST
2661 must be smaller or equal to the size of TYPE. Both types must be
2663 ``fptoui (CST to TYPE)``
2664 Convert a floating point constant to the corresponding unsigned
2665 integer constant. TYPE must be a scalar or vector integer type. CST
2666 must be of scalar or vector floating point type. Both CST and TYPE
2667 must be scalars, or vectors of the same number of elements. If the
2668 value won't fit in the integer type, the results are undefined.
2669 ``fptosi (CST to TYPE)``
2670 Convert a floating point constant to the corresponding signed
2671 integer constant. TYPE must be a scalar or vector integer type. CST
2672 must be of scalar or vector floating point type. Both CST and TYPE
2673 must be scalars, or vectors of the same number of elements. If the
2674 value won't fit in the integer type, the results are undefined.
2675 ``uitofp (CST to TYPE)``
2676 Convert an unsigned integer constant to the corresponding floating
2677 point constant. TYPE must be a scalar or vector floating point type.
2678 CST must be of scalar or vector integer type. Both CST and TYPE must
2679 be scalars, or vectors of the same number of elements. If the value
2680 won't fit in the floating point type, the results are undefined.
2681 ``sitofp (CST to TYPE)``
2682 Convert a signed integer constant to the corresponding floating
2683 point constant. TYPE must be a scalar or vector floating point type.
2684 CST must be of scalar or vector integer type. Both CST and TYPE must
2685 be scalars, or vectors of the same number of elements. If the value
2686 won't fit in the floating point type, the results are undefined.
2687 ``ptrtoint (CST to TYPE)``
2688 Convert a pointer typed constant to the corresponding integer
2689 constant. ``TYPE`` must be an integer type. ``CST`` must be of
2690 pointer type. The ``CST`` value is zero extended, truncated, or
2691 unchanged to make it fit in ``TYPE``.
2692 ``inttoptr (CST to TYPE)``
2693 Convert an integer constant to a pointer constant. TYPE must be a
2694 pointer type. CST must be of integer type. The CST value is zero
2695 extended, truncated, or unchanged to make it fit in a pointer size.
2696 This one is *really* dangerous!
2697 ``bitcast (CST to TYPE)``
2698 Convert a constant, CST, to another TYPE. The constraints of the
2699 operands are the same as those for the :ref:`bitcast
2700 instruction <i_bitcast>`.
2701 ``addrspacecast (CST to TYPE)``
2702 Convert a constant pointer or constant vector of pointer, CST, to another
2703 TYPE in a different address space. The constraints of the operands are the
2704 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`.
2705 ``getelementptr (CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)``
2706 Perform the :ref:`getelementptr operation <i_getelementptr>` on
2707 constants. As with the :ref:`getelementptr <i_getelementptr>`
2708 instruction, the index list may have zero or more indexes, which are
2709 required to make sense for the type of "CSTPTR".
2710 ``select (COND, VAL1, VAL2)``
2711 Perform the :ref:`select operation <i_select>` on constants.
2712 ``icmp COND (VAL1, VAL2)``
2713 Performs the :ref:`icmp operation <i_icmp>` on constants.
2714 ``fcmp COND (VAL1, VAL2)``
2715 Performs the :ref:`fcmp operation <i_fcmp>` on constants.
2716 ``extractelement (VAL, IDX)``
2717 Perform the :ref:`extractelement operation <i_extractelement>` on
2719 ``insertelement (VAL, ELT, IDX)``
2720 Perform the :ref:`insertelement operation <i_insertelement>` on
2722 ``shufflevector (VEC1, VEC2, IDXMASK)``
2723 Perform the :ref:`shufflevector operation <i_shufflevector>` on
2725 ``extractvalue (VAL, IDX0, IDX1, ...)``
2726 Perform the :ref:`extractvalue operation <i_extractvalue>` on
2727 constants. The index list is interpreted in a similar manner as
2728 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At
2729 least one index value must be specified.
2730 ``insertvalue (VAL, ELT, IDX0, IDX1, ...)``
2731 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants.
2732 The index list is interpreted in a similar manner as indices in a
2733 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index
2734 value must be specified.
2735 ``OPCODE (LHS, RHS)``
2736 Perform the specified operation of the LHS and RHS constants. OPCODE
2737 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise
2738 binary <bitwiseops>` operations. The constraints on operands are
2739 the same as those for the corresponding instruction (e.g. no bitwise
2740 operations on floating point values are allowed).
2747 Inline Assembler Expressions
2748 ----------------------------
2750 LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level
2751 Inline Assembly <moduleasm>`) through the use of a special value. This
2752 value represents the inline assembler as a string (containing the
2753 instructions to emit), a list of operand constraints (stored as a
2754 string), a flag that indicates whether or not the inline asm expression
2755 has side effects, and a flag indicating whether the function containing
2756 the asm needs to align its stack conservatively. An example inline
2757 assembler expression is:
2759 .. code-block:: llvm
2761 i32 (i32) asm "bswap $0", "=r,r"
2763 Inline assembler expressions may **only** be used as the callee operand
2764 of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction.
2765 Thus, typically we have:
2767 .. code-block:: llvm
2769 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y)
2771 Inline asms with side effects not visible in the constraint list must be
2772 marked as having side effects. This is done through the use of the
2773 '``sideeffect``' keyword, like so:
2775 .. code-block:: llvm
2777 call void asm sideeffect "eieio", ""()
2779 In some cases inline asms will contain code that will not work unless
2780 the stack is aligned in some way, such as calls or SSE instructions on
2781 x86, yet will not contain code that does that alignment within the asm.
2782 The compiler should make conservative assumptions about what the asm
2783 might contain and should generate its usual stack alignment code in the
2784 prologue if the '``alignstack``' keyword is present:
2786 .. code-block:: llvm
2788 call void asm alignstack "eieio", ""()
2790 Inline asms also support using non-standard assembly dialects. The
2791 assumed dialect is ATT. When the '``inteldialect``' keyword is present,
2792 the inline asm is using the Intel dialect. Currently, ATT and Intel are
2793 the only supported dialects. An example is:
2795 .. code-block:: llvm
2797 call void asm inteldialect "eieio", ""()
2799 If multiple keywords appear the '``sideeffect``' keyword must come
2800 first, the '``alignstack``' keyword second and the '``inteldialect``'
2806 The call instructions that wrap inline asm nodes may have a
2807 "``!srcloc``" MDNode attached to it that contains a list of constant
2808 integers. If present, the code generator will use the integer as the
2809 location cookie value when report errors through the ``LLVMContext``
2810 error reporting mechanisms. This allows a front-end to correlate backend
2811 errors that occur with inline asm back to the source code that produced
2814 .. code-block:: llvm
2816 call void asm sideeffect "something bad", ""(), !srcloc !42
2818 !42 = !{ i32 1234567 }
2820 It is up to the front-end to make sense of the magic numbers it places
2821 in the IR. If the MDNode contains multiple constants, the code generator
2822 will use the one that corresponds to the line of the asm that the error
2830 LLVM IR allows metadata to be attached to instructions in the program
2831 that can convey extra information about the code to the optimizers and
2832 code generator. One example application of metadata is source-level
2833 debug information. There are two metadata primitives: strings and nodes.
2835 Metadata does not have a type, and is not a value. If referenced from a
2836 ``call`` instruction, it uses the ``metadata`` type.
2838 All metadata are identified in syntax by a exclamation point ('``!``').
2840 .. _metadata-string:
2842 Metadata Nodes and Metadata Strings
2843 -----------------------------------
2845 A metadata string is a string surrounded by double quotes. It can
2846 contain any character by escaping non-printable characters with
2847 "``\xx``" where "``xx``" is the two digit hex code. For example:
2850 Metadata nodes are represented with notation similar to structure
2851 constants (a comma separated list of elements, surrounded by braces and
2852 preceded by an exclamation point). Metadata nodes can have any values as
2853 their operand. For example:
2855 .. code-block:: llvm
2857 !{ !"test\00", i32 10}
2859 Metadata nodes that aren't uniqued use the ``distinct`` keyword. For example:
2861 .. code-block:: llvm
2863 !0 = distinct !{!"test\00", i32 10}
2865 ``distinct`` nodes are useful when nodes shouldn't be merged based on their
2866 content. They can also occur when transformations cause uniquing collisions
2867 when metadata operands change.
2869 A :ref:`named metadata <namedmetadatastructure>` is a collection of
2870 metadata nodes, which can be looked up in the module symbol table. For
2873 .. code-block:: llvm
2877 Metadata can be used as function arguments. Here ``llvm.dbg.value``
2878 function is using two metadata arguments:
2880 .. code-block:: llvm
2882 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2884 Metadata can be attached with an instruction. Here metadata ``!21`` is
2885 attached to the ``add`` instruction using the ``!dbg`` identifier:
2887 .. code-block:: llvm
2889 %indvar.next = add i64 %indvar, 1, !dbg !21
2891 More information about specific metadata nodes recognized by the
2892 optimizers and code generator is found below.
2894 Specialized Metadata Nodes
2895 ^^^^^^^^^^^^^^^^^^^^^^^^^^
2897 Specialized metadata nodes are custom data structures in metadata (as opposed
2898 to generic tuples). Their fields are labelled, and can be specified in any
2901 These aren't inherently debug info centric, but currently all the specialized
2902 metadata nodes are related to debug info.
2907 ``MDCompileUnit`` nodes represent a compile unit. The ``enums:``,
2908 ``retainedTypes:``, ``subprograms:``, ``globals:`` and ``imports:`` fields are
2909 tuples containing the debug info to be emitted along with the compile unit,
2910 regardless of code optimizations (some nodes are only emitted if there are
2911 references to them from instructions).
2913 .. code-block:: llvm
2915 !0 = !MDCompileUnit(language: DW_LANG_C99, file: !1, producer: "clang",
2916 isOptimized: true, flags: "-O2", runtimeVersion: 2,
2917 splitDebugFilename: "abc.debug", emissionKind: 1,
2918 enums: !2, retainedTypes: !3, subprograms: !4,
2919 globals: !5, imports: !6)
2924 ``MDFile`` nodes represent files. The ``filename:`` can include slashes.
2926 .. code-block:: llvm
2928 !0 = !MDFile(filename: "path/to/file", directory: "/path/to/dir")
2935 ``MDBasicType`` nodes represent primitive types. ``tag:`` defaults to
2936 ``DW_TAG_base_type``.
2938 .. code-block:: llvm
2940 !0 = !MDBasicType(name: "unsigned char", size: 8, align: 8,
2941 encoding: DW_ATE_unsigned_char)
2942 !1 = !MDBasicType(tag: DW_TAG_unspecified_type, name: "decltype(nullptr)")
2944 .. _MDSubroutineType:
2949 ``MDSubroutineType`` nodes represent subroutine types. Their ``types:`` field
2950 refers to a tuple; the first operand is the return type, while the rest are the
2951 types of the formal arguments in order. If the first operand is ``null``, that
2952 represents a function with no return value (such as ``void foo() {}`` in C++).
2954 .. code-block:: llvm
2956 !0 = !BasicType(name: "int", size: 32, align: 32, DW_ATE_signed)
2957 !1 = !BasicType(name: "char", size: 8, align: 8, DW_ATE_signed_char)
2958 !2 = !MDSubroutineType(types: !{null, !0, !1}) ; void (int, char)
2963 ``MDDerivedType`` nodes represent types derived from other types, such as
2966 .. code-block:: llvm
2968 !0 = !MDBasicType(name: "unsigned char", size: 8, align: 8,
2969 encoding: DW_ATE_unsigned_char)
2970 !1 = !MDDerivedType(tag: DW_TAG_pointer_type, baseType: !0, size: 32,
2973 .. _MDCompositeType:
2978 ``MDCompositeType`` nodes represent types composed of other types, like
2979 structures and unions. ``elements:`` points to a tuple of the composed types.
2981 If the source language supports ODR, the ``identifier:`` field gives the unique
2982 identifier used for type merging between modules. When specified, other types
2983 can refer to composite types indirectly via a :ref:`metadata string
2984 <metadata-string>` that matches their identifier.
2986 .. code-block:: llvm
2988 !0 = !MDEnumerator(name: "SixKind", value: 7)
2989 !1 = !MDEnumerator(name: "SevenKind", value: 7)
2990 !2 = !MDEnumerator(name: "NegEightKind", value: -8)
2991 !3 = !MDCompositeType(tag: DW_TAG_enumeration_type, name: "Enum", file: !12,
2992 line: 2, size: 32, align: 32, identifier: "_M4Enum",
2993 elements: !{!0, !1, !2})
2998 ``MDSubrange`` nodes are the elements for ``DW_TAG_array_type`` variants of
2999 :ref:`MDCompositeType`. ``count: -1`` indicates an empty array.
3001 .. code-block:: llvm
3003 !0 = !MDSubrange(count: 5, lowerBound: 0) ; array counting from 0
3004 !1 = !MDSubrange(count: 5, lowerBound: 1) ; array counting from 1
3005 !2 = !MDSubrange(count: -1) ; empty array.
3010 ``MDEnumerator`` nodes are the elements for ``DW_TAG_enumeration_type``
3011 variants of :ref:`MDCompositeType`.
3013 .. code-block:: llvm
3015 !0 = !MDEnumerator(name: "SixKind", value: 7)
3016 !1 = !MDEnumerator(name: "SevenKind", value: 7)
3017 !2 = !MDEnumerator(name: "NegEightKind", value: -8)
3019 MDTemplateTypeParameter
3020 """""""""""""""""""""""
3022 ``MDTemplateTypeParameter`` nodes represent type parameters to generic source
3023 language constructs. They are used (optionally) in :ref:`MDCompositeType` and
3024 :ref:`MDSubprogram` ``templateParams:`` fields.
3026 .. code-block:: llvm
3028 !0 = !MDTemplateTypeParameter(name: "Ty", type: !1)
3030 MDTemplateValueParameter
3031 """"""""""""""""""""""""
3033 ``MDTemplateValueParameter`` nodes represent value parameters to generic source
3034 language constructs. ``tag:`` defaults to ``DW_TAG_template_value_parameter``,
3035 but if specified can also be set to ``DW_TAG_GNU_template_template_param`` or
3036 ``DW_TAG_GNU_template_param_pack``. They are used (optionally) in
3037 :ref:`MDCompositeType` and :ref:`MDSubprogram` ``templateParams:`` fields.
3039 .. code-block:: llvm
3041 !0 = !MDTemplateValueParameter(name: "Ty", type: !1, value: i32 7)
3046 ``MDNamespace`` nodes represent namespaces in the source language.
3048 .. code-block:: llvm
3050 !0 = !MDNamespace(name: "myawesomeproject", scope: !1, file: !2, line: 7)
3055 ``MDGlobalVariable`` nodes represent global variables in the source language.
3057 .. code-block:: llvm
3059 !0 = !MDGlobalVariable(name: "foo", linkageName: "foo", scope: !1,
3060 file: !2, line: 7, type: !3, isLocal: true,
3061 isDefinition: false, variable: i32* @foo,
3069 ``MDSubprogram`` nodes represent functions from the source language. The
3070 ``variables:`` field points at :ref:`variables <MDLocalVariable>` that must be
3071 retained, even if their IR counterparts are optimized out of the IR. The
3072 ``type:`` field must point at an :ref:`MDSubroutineType`.
3074 .. code-block:: llvm
3076 !0 = !MDSubprogram(name: "foo", linkageName: "_Zfoov", scope: !1,
3077 file: !2, line: 7, type: !3, isLocal: true,
3078 isDefinition: false, scopeLine: 8, containingType: !4,
3079 virtuality: DW_VIRTUALITY_pure_virtual, virtualIndex: 10,
3080 flags: DIFlagPrototyped, isOptimized: true,
3081 function: void ()* @_Z3foov,
3082 templateParams: !5, declaration: !6, variables: !7)
3089 ``MDLexicalBlock`` nodes represent lexical blocks in the source language (a
3092 .. code-block:: llvm
3094 !0 = !MDLexicalBlock(scope: !1, file: !2, line: 7, column: 35)
3096 .. _MDLexicalBlockFile:
3101 ``MDLexicalBlockFile`` nodes are used to discriminate between sections of a
3102 :ref:`lexical block <MDLexicalBlock>`. The ``file:`` field can be changed to
3103 indicate textual inclusion, or the ``discriminator:`` field can be used to
3104 discriminate between control flow within a single block in the source language.
3106 .. code-block:: llvm
3108 !0 = !MDLexicalBlock(scope: !3, file: !4, line: 7, column: 35)
3109 !1 = !MDLexicalBlockFile(scope: !0, file: !4, discriminator: 0)
3110 !2 = !MDLexicalBlockFile(scope: !0, file: !4, discriminator: 1)
3115 ``MDLocation`` nodes represent source debug locations. The ``scope:`` field is
3116 mandatory, and points at an :ref:`MDLexicalBlockFile`, an
3117 :ref:`MDLexicalBlock`, or an :ref:`MDSubprogram`.
3119 .. code-block:: llvm
3121 !0 = !MDLocation(line: 2900, column: 42, scope: !1, inlinedAt: !2)
3123 .. _MDLocalVariable:
3128 ``MDLocalVariable`` nodes represent local variables in the source language.
3129 Instead of ``DW_TAG_variable``, they use LLVM-specific fake tags to
3130 discriminate between local variables (``DW_TAG_auto_variable``) and subprogram
3131 arguments (``DW_TAG_arg_variable``). In the latter case, the ``arg:`` field
3132 specifies the argument position, and this variable will be included in the
3133 ``variables:`` field of its :ref:`MDSubprogram`.
3135 If set, the ``inlinedAt:`` field points at an :ref:`MDLocation`, and the
3136 variable represents an inlined version of a variable (with all other fields
3137 duplicated from the non-inlined version).
3139 .. code-block:: llvm
3141 !0 = !MDLocalVariable(tag: DW_TAG_arg_variable, name: "this", arg: 0,
3142 scope: !3, file: !2, line: 7, type: !3,
3143 flags: DIFlagArtificial, inlinedAt: !4)
3144 !1 = !MDLocalVariable(tag: DW_TAG_arg_variable, name: "x", arg: 1,
3145 scope: !4, file: !2, line: 7, type: !3,
3147 !1 = !MDLocalVariable(tag: DW_TAG_auto_variable, name: "y",
3148 scope: !5, file: !2, line: 7, type: !3,
3154 ``MDExpression`` nodes represent DWARF expression sequences. They are used in
3155 :ref:`debug intrinsics<dbg_intrinsics>` (such as ``llvm.dbg.declare``) to
3156 describe how the referenced LLVM variable relates to the source language
3159 The current supported vocabulary is limited:
3161 - ``DW_OP_deref`` dereferences the working expression.
3162 - ``DW_OP_plus, 93`` adds ``93`` to the working expression.
3163 - ``DW_OP_bit_piece, 16, 8`` specifies the offset and size (``16`` and ``8``
3164 here, respectively) of the variable piece from the working expression.
3166 .. code-block:: llvm
3168 !0 = !MDExpression(DW_OP_deref)
3169 !1 = !MDExpression(DW_OP_plus, 3)
3170 !2 = !MDExpression(DW_OP_bit_piece, 3, 7)
3171 !3 = !MDExpression(DW_OP_deref, DW_OP_plus, 3, DW_OP_bit_piece, 3, 7)
3176 ``MDObjCProperty`` nodes represent Objective-C property nodes.
3178 .. code-block:: llvm
3180 !3 = !MDObjCProperty(name: "foo", file: !1, line: 7, setter: "setFoo",
3181 getter: "getFoo", attributes: 7, type: !2)
3186 ``MDImportedEntity`` nodes represent entities (such as modules) imported into a
3189 .. code-block:: llvm
3191 !2 = !MDImportedEntity(tag: DW_TAG_imported_module, name: "foo", scope: !0,
3192 entity: !1, line: 7)
3197 In LLVM IR, memory does not have types, so LLVM's own type system is not
3198 suitable for doing TBAA. Instead, metadata is added to the IR to
3199 describe a type system of a higher level language. This can be used to
3200 implement typical C/C++ TBAA, but it can also be used to implement
3201 custom alias analysis behavior for other languages.
3203 The current metadata format is very simple. TBAA metadata nodes have up
3204 to three fields, e.g.:
3206 .. code-block:: llvm
3208 !0 = !{ !"an example type tree" }
3209 !1 = !{ !"int", !0 }
3210 !2 = !{ !"float", !0 }
3211 !3 = !{ !"const float", !2, i64 1 }
3213 The first field is an identity field. It can be any value, usually a
3214 metadata string, which uniquely identifies the type. The most important
3215 name in the tree is the name of the root node. Two trees with different
3216 root node names are entirely disjoint, even if they have leaves with
3219 The second field identifies the type's parent node in the tree, or is
3220 null or omitted for a root node. A type is considered to alias all of
3221 its descendants and all of its ancestors in the tree. Also, a type is
3222 considered to alias all types in other trees, so that bitcode produced
3223 from multiple front-ends is handled conservatively.
3225 If the third field is present, it's an integer which if equal to 1
3226 indicates that the type is "constant" (meaning
3227 ``pointsToConstantMemory`` should return true; see `other useful
3228 AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_).
3230 '``tbaa.struct``' Metadata
3231 ^^^^^^^^^^^^^^^^^^^^^^^^^^
3233 The :ref:`llvm.memcpy <int_memcpy>` is often used to implement
3234 aggregate assignment operations in C and similar languages, however it
3235 is defined to copy a contiguous region of memory, which is more than
3236 strictly necessary for aggregate types which contain holes due to
3237 padding. Also, it doesn't contain any TBAA information about the fields
3240 ``!tbaa.struct`` metadata can describe which memory subregions in a
3241 memcpy are padding and what the TBAA tags of the struct are.
3243 The current metadata format is very simple. ``!tbaa.struct`` metadata
3244 nodes are a list of operands which are in conceptual groups of three.
3245 For each group of three, the first operand gives the byte offset of a
3246 field in bytes, the second gives its size in bytes, and the third gives
3249 .. code-block:: llvm
3251 !4 = !{ i64 0, i64 4, !1, i64 8, i64 4, !2 }
3253 This describes a struct with two fields. The first is at offset 0 bytes
3254 with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes
3255 and has size 4 bytes and has tbaa tag !2.
3257 Note that the fields need not be contiguous. In this example, there is a
3258 4 byte gap between the two fields. This gap represents padding which
3259 does not carry useful data and need not be preserved.
3261 '``noalias``' and '``alias.scope``' Metadata
3262 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3264 ``noalias`` and ``alias.scope`` metadata provide the ability to specify generic
3265 noalias memory-access sets. This means that some collection of memory access
3266 instructions (loads, stores, memory-accessing calls, etc.) that carry
3267 ``noalias`` metadata can specifically be specified not to alias with some other
3268 collection of memory access instructions that carry ``alias.scope`` metadata.
3269 Each type of metadata specifies a list of scopes where each scope has an id and
3270 a domain. When evaluating an aliasing query, if for some some domain, the set
3271 of scopes with that domain in one instruction's ``alias.scope`` list is a
3272 subset of (or equal to) the set of scopes for that domain in another
3273 instruction's ``noalias`` list, then the two memory accesses are assumed not to
3276 The metadata identifying each domain is itself a list containing one or two
3277 entries. The first entry is the name of the domain. Note that if the name is a
3278 string then it can be combined accross functions and translation units. A
3279 self-reference can be used to create globally unique domain names. A
3280 descriptive string may optionally be provided as a second list entry.
3282 The metadata identifying each scope is also itself a list containing two or
3283 three entries. The first entry is the name of the scope. Note that if the name
3284 is a string then it can be combined accross functions and translation units. A
3285 self-reference can be used to create globally unique scope names. A metadata
3286 reference to the scope's domain is the second entry. A descriptive string may
3287 optionally be provided as a third list entry.
3291 .. code-block:: llvm
3293 ; Two scope domains:
3297 ; Some scopes in these domains:
3303 !5 = !{!4} ; A list containing only scope !4
3307 ; These two instructions don't alias:
3308 %0 = load float* %c, align 4, !alias.scope !5
3309 store float %0, float* %arrayidx.i, align 4, !noalias !5
3311 ; These two instructions also don't alias (for domain !1, the set of scopes
3312 ; in the !alias.scope equals that in the !noalias list):
3313 %2 = load float* %c, align 4, !alias.scope !5
3314 store float %2, float* %arrayidx.i2, align 4, !noalias !6
3316 ; These two instructions don't alias (for domain !0, the set of scopes in
3317 ; the !noalias list is not a superset of, or equal to, the scopes in the
3318 ; !alias.scope list):
3319 %2 = load float* %c, align 4, !alias.scope !6
3320 store float %0, float* %arrayidx.i, align 4, !noalias !7
3322 '``fpmath``' Metadata
3323 ^^^^^^^^^^^^^^^^^^^^^
3325 ``fpmath`` metadata may be attached to any instruction of floating point
3326 type. It can be used to express the maximum acceptable error in the
3327 result of that instruction, in ULPs, thus potentially allowing the
3328 compiler to use a more efficient but less accurate method of computing
3329 it. ULP is defined as follows:
3331 If ``x`` is a real number that lies between two finite consecutive
3332 floating-point numbers ``a`` and ``b``, without being equal to one
3333 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the
3334 distance between the two non-equal finite floating-point numbers
3335 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``.
3337 The metadata node shall consist of a single positive floating point
3338 number representing the maximum relative error, for example:
3340 .. code-block:: llvm
3342 !0 = !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
3346 '``range``' Metadata
3347 ^^^^^^^^^^^^^^^^^^^^
3349 ``range`` metadata may be attached only to ``load``, ``call`` and ``invoke`` of
3350 integer types. It expresses the possible ranges the loaded value or the value
3351 returned by the called function at this call site is in. The ranges are
3352 represented with a flattened list of integers. The loaded value or the value
3353 returned is known to be in the union of the ranges defined by each consecutive
3354 pair. Each pair has the following properties:
3356 - The type must match the type loaded by the instruction.
3357 - The pair ``a,b`` represents the range ``[a,b)``.
3358 - Both ``a`` and ``b`` are constants.
3359 - The range is allowed to wrap.
3360 - The range should not represent the full or empty set. That is,
3363 In addition, the pairs must be in signed order of the lower bound and
3364 they must be non-contiguous.
3368 .. code-block:: llvm
3370 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1
3371 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
3372 %c = call i8 @foo(), !range !2 ; Can only be 0, 1, 3, 4 or 5
3373 %d = invoke i8 @bar() to label %cont
3374 unwind label %lpad, !range !3 ; Can only be -2, -1, 3, 4 or 5
3376 !0 = !{ i8 0, i8 2 }
3377 !1 = !{ i8 255, i8 2 }
3378 !2 = !{ i8 0, i8 2, i8 3, i8 6 }
3379 !3 = !{ i8 -2, i8 0, i8 3, i8 6 }
3384 It is sometimes useful to attach information to loop constructs. Currently,
3385 loop metadata is implemented as metadata attached to the branch instruction
3386 in the loop latch block. This type of metadata refer to a metadata node that is
3387 guaranteed to be separate for each loop. The loop identifier metadata is
3388 specified with the name ``llvm.loop``.
3390 The loop identifier metadata is implemented using a metadata that refers to
3391 itself to avoid merging it with any other identifier metadata, e.g.,
3392 during module linkage or function inlining. That is, each loop should refer
3393 to their own identification metadata even if they reside in separate functions.
3394 The following example contains loop identifier metadata for two separate loop
3397 .. code-block:: llvm
3402 The loop identifier metadata can be used to specify additional
3403 per-loop metadata. Any operands after the first operand can be treated
3404 as user-defined metadata. For example the ``llvm.loop.unroll.count``
3405 suggests an unroll factor to the loop unroller:
3407 .. code-block:: llvm
3409 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0
3412 !1 = !{!"llvm.loop.unroll.count", i32 4}
3414 '``llvm.loop.vectorize``' and '``llvm.loop.interleave``'
3415 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3417 Metadata prefixed with ``llvm.loop.vectorize`` or ``llvm.loop.interleave`` are
3418 used to control per-loop vectorization and interleaving parameters such as
3419 vectorization width and interleave count. These metadata should be used in
3420 conjunction with ``llvm.loop`` loop identification metadata. The
3421 ``llvm.loop.vectorize`` and ``llvm.loop.interleave`` metadata are only
3422 optimization hints and the optimizer will only interleave and vectorize loops if
3423 it believes it is safe to do so. The ``llvm.mem.parallel_loop_access`` metadata
3424 which contains information about loop-carried memory dependencies can be helpful
3425 in determining the safety of these transformations.
3427 '``llvm.loop.interleave.count``' Metadata
3428 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3430 This metadata suggests an interleave count to the loop interleaver.
3431 The first operand is the string ``llvm.loop.interleave.count`` and the
3432 second operand is an integer specifying the interleave count. For
3435 .. code-block:: llvm
3437 !0 = !{!"llvm.loop.interleave.count", i32 4}
3439 Note that setting ``llvm.loop.interleave.count`` to 1 disables interleaving
3440 multiple iterations of the loop. If ``llvm.loop.interleave.count`` is set to 0
3441 then the interleave count will be determined automatically.
3443 '``llvm.loop.vectorize.enable``' Metadata
3444 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3446 This metadata selectively enables or disables vectorization for the loop. The
3447 first operand is the string ``llvm.loop.vectorize.enable`` and the second operand
3448 is a bit. If the bit operand value is 1 vectorization is enabled. A value of
3449 0 disables vectorization:
3451 .. code-block:: llvm
3453 !0 = !{!"llvm.loop.vectorize.enable", i1 0}
3454 !1 = !{!"llvm.loop.vectorize.enable", i1 1}
3456 '``llvm.loop.vectorize.width``' Metadata
3457 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3459 This metadata sets the target width of the vectorizer. The first
3460 operand is the string ``llvm.loop.vectorize.width`` and the second
3461 operand is an integer specifying the width. For example:
3463 .. code-block:: llvm
3465 !0 = !{!"llvm.loop.vectorize.width", i32 4}
3467 Note that setting ``llvm.loop.vectorize.width`` to 1 disables
3468 vectorization of the loop. If ``llvm.loop.vectorize.width`` is set to
3469 0 or if the loop does not have this metadata the width will be
3470 determined automatically.
3472 '``llvm.loop.unroll``'
3473 ^^^^^^^^^^^^^^^^^^^^^^
3475 Metadata prefixed with ``llvm.loop.unroll`` are loop unrolling
3476 optimization hints such as the unroll factor. ``llvm.loop.unroll``
3477 metadata should be used in conjunction with ``llvm.loop`` loop
3478 identification metadata. The ``llvm.loop.unroll`` metadata are only
3479 optimization hints and the unrolling will only be performed if the
3480 optimizer believes it is safe to do so.
3482 '``llvm.loop.unroll.count``' Metadata
3483 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3485 This metadata suggests an unroll factor to the loop unroller. The
3486 first operand is the string ``llvm.loop.unroll.count`` and the second
3487 operand is a positive integer specifying the unroll factor. For
3490 .. code-block:: llvm
3492 !0 = !{!"llvm.loop.unroll.count", i32 4}
3494 If the trip count of the loop is less than the unroll count the loop
3495 will be partially unrolled.
3497 '``llvm.loop.unroll.disable``' Metadata
3498 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3500 This metadata either disables loop unrolling. The metadata has a single operand
3501 which is the string ``llvm.loop.unroll.disable``. For example:
3503 .. code-block:: llvm
3505 !0 = !{!"llvm.loop.unroll.disable"}
3507 '``llvm.loop.unroll.full``' Metadata
3508 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3510 This metadata either suggests that the loop should be unrolled fully. The
3511 metadata has a single operand which is the string ``llvm.loop.unroll.disable``.
3514 .. code-block:: llvm
3516 !0 = !{!"llvm.loop.unroll.full"}
3521 Metadata types used to annotate memory accesses with information helpful
3522 for optimizations are prefixed with ``llvm.mem``.
3524 '``llvm.mem.parallel_loop_access``' Metadata
3525 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3527 The ``llvm.mem.parallel_loop_access`` metadata refers to a loop identifier,
3528 or metadata containing a list of loop identifiers for nested loops.
3529 The metadata is attached to memory accessing instructions and denotes that
3530 no loop carried memory dependence exist between it and other instructions denoted
3531 with the same loop identifier.
3533 Precisely, given two instructions ``m1`` and ``m2`` that both have the
3534 ``llvm.mem.parallel_loop_access`` metadata, with ``L1`` and ``L2`` being the
3535 set of loops associated with that metadata, respectively, then there is no loop
3536 carried dependence between ``m1`` and ``m2`` for loops in both ``L1`` and
3539 As a special case, if all memory accessing instructions in a loop have
3540 ``llvm.mem.parallel_loop_access`` metadata that refers to that loop, then the
3541 loop has no loop carried memory dependences and is considered to be a parallel
3544 Note that if not all memory access instructions have such metadata referring to
3545 the loop, then the loop is considered not being trivially parallel. Additional
3546 memory dependence analysis is required to make that determination. As a fail
3547 safe mechanism, this causes loops that were originally parallel to be considered
3548 sequential (if optimization passes that are unaware of the parallel semantics
3549 insert new memory instructions into the loop body).
3551 Example of a loop that is considered parallel due to its correct use of
3552 both ``llvm.loop`` and ``llvm.mem.parallel_loop_access``
3553 metadata types that refer to the same loop identifier metadata.
3555 .. code-block:: llvm
3559 %val0 = load i32* %arrayidx, !llvm.mem.parallel_loop_access !0
3561 store i32 %val0, i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
3563 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0
3569 It is also possible to have nested parallel loops. In that case the
3570 memory accesses refer to a list of loop identifier metadata nodes instead of
3571 the loop identifier metadata node directly:
3573 .. code-block:: llvm
3577 %val1 = load i32* %arrayidx3, !llvm.mem.parallel_loop_access !2
3579 br label %inner.for.body
3583 %val0 = load i32* %arrayidx1, !llvm.mem.parallel_loop_access !0
3585 store i32 %val0, i32* %arrayidx2, !llvm.mem.parallel_loop_access !0
3587 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1
3591 store i32 %val1, i32* %arrayidx4, !llvm.mem.parallel_loop_access !2
3593 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2
3595 outer.for.end: ; preds = %for.body
3597 !0 = !{!1, !2} ; a list of loop identifiers
3598 !1 = !{!1} ; an identifier for the inner loop
3599 !2 = !{!2} ; an identifier for the outer loop
3604 The ``llvm.bitsets`` global metadata is used to implement
3605 :doc:`bitsets <BitSets>`.
3607 Module Flags Metadata
3608 =====================
3610 Information about the module as a whole is difficult to convey to LLVM's
3611 subsystems. The LLVM IR isn't sufficient to transmit this information.
3612 The ``llvm.module.flags`` named metadata exists in order to facilitate
3613 this. These flags are in the form of key / value pairs --- much like a
3614 dictionary --- making it easy for any subsystem who cares about a flag to
3617 The ``llvm.module.flags`` metadata contains a list of metadata triplets.
3618 Each triplet has the following form:
3620 - The first element is a *behavior* flag, which specifies the behavior
3621 when two (or more) modules are merged together, and it encounters two
3622 (or more) metadata with the same ID. The supported behaviors are
3624 - The second element is a metadata string that is a unique ID for the
3625 metadata. Each module may only have one flag entry for each unique ID (not
3626 including entries with the **Require** behavior).
3627 - The third element is the value of the flag.
3629 When two (or more) modules are merged together, the resulting
3630 ``llvm.module.flags`` metadata is the union of the modules' flags. That is, for
3631 each unique metadata ID string, there will be exactly one entry in the merged
3632 modules ``llvm.module.flags`` metadata table, and the value for that entry will
3633 be determined by the merge behavior flag, as described below. The only exception
3634 is that entries with the *Require* behavior are always preserved.
3636 The following behaviors are supported:
3647 Emits an error if two values disagree, otherwise the resulting value
3648 is that of the operands.
3652 Emits a warning if two values disagree. The result value will be the
3653 operand for the flag from the first module being linked.
3657 Adds a requirement that another module flag be present and have a
3658 specified value after linking is performed. The value must be a
3659 metadata pair, where the first element of the pair is the ID of the
3660 module flag to be restricted, and the second element of the pair is
3661 the value the module flag should be restricted to. This behavior can
3662 be used to restrict the allowable results (via triggering of an
3663 error) of linking IDs with the **Override** behavior.
3667 Uses the specified value, regardless of the behavior or value of the
3668 other module. If both modules specify **Override**, but the values
3669 differ, an error will be emitted.
3673 Appends the two values, which are required to be metadata nodes.
3677 Appends the two values, which are required to be metadata
3678 nodes. However, duplicate entries in the second list are dropped
3679 during the append operation.
3681 It is an error for a particular unique flag ID to have multiple behaviors,
3682 except in the case of **Require** (which adds restrictions on another metadata
3683 value) or **Override**.
3685 An example of module flags:
3687 .. code-block:: llvm
3689 !0 = !{ i32 1, !"foo", i32 1 }
3690 !1 = !{ i32 4, !"bar", i32 37 }
3691 !2 = !{ i32 2, !"qux", i32 42 }
3692 !3 = !{ i32 3, !"qux",
3697 !llvm.module.flags = !{ !0, !1, !2, !3 }
3699 - Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior
3700 if two or more ``!"foo"`` flags are seen is to emit an error if their
3701 values are not equal.
3703 - Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The
3704 behavior if two or more ``!"bar"`` flags are seen is to use the value
3707 - Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The
3708 behavior if two or more ``!"qux"`` flags are seen is to emit a
3709 warning if their values are not equal.
3711 - Metadata ``!3`` has the ID ``!"qux"`` and the value:
3717 The behavior is to emit an error if the ``llvm.module.flags`` does not
3718 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is
3721 Objective-C Garbage Collection Module Flags Metadata
3722 ----------------------------------------------------
3724 On the Mach-O platform, Objective-C stores metadata about garbage
3725 collection in a special section called "image info". The metadata
3726 consists of a version number and a bitmask specifying what types of
3727 garbage collection are supported (if any) by the file. If two or more
3728 modules are linked together their garbage collection metadata needs to
3729 be merged rather than appended together.
3731 The Objective-C garbage collection module flags metadata consists of the
3732 following key-value pairs:
3741 * - ``Objective-C Version``
3742 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2.
3744 * - ``Objective-C Image Info Version``
3745 - **[Required]** --- The version of the image info section. Currently
3748 * - ``Objective-C Image Info Section``
3749 - **[Required]** --- The section to place the metadata. Valid values are
3750 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and
3751 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for
3752 Objective-C ABI version 2.
3754 * - ``Objective-C Garbage Collection``
3755 - **[Required]** --- Specifies whether garbage collection is supported or
3756 not. Valid values are 0, for no garbage collection, and 2, for garbage
3757 collection supported.
3759 * - ``Objective-C GC Only``
3760 - **[Optional]** --- Specifies that only garbage collection is supported.
3761 If present, its value must be 6. This flag requires that the
3762 ``Objective-C Garbage Collection`` flag have the value 2.
3764 Some important flag interactions:
3766 - If a module with ``Objective-C Garbage Collection`` set to 0 is
3767 merged with a module with ``Objective-C Garbage Collection`` set to
3768 2, then the resulting module has the
3769 ``Objective-C Garbage Collection`` flag set to 0.
3770 - A module with ``Objective-C Garbage Collection`` set to 0 cannot be
3771 merged with a module with ``Objective-C GC Only`` set to 6.
3773 Automatic Linker Flags Module Flags Metadata
3774 --------------------------------------------
3776 Some targets support embedding flags to the linker inside individual object
3777 files. Typically this is used in conjunction with language extensions which
3778 allow source files to explicitly declare the libraries they depend on, and have
3779 these automatically be transmitted to the linker via object files.
3781 These flags are encoded in the IR using metadata in the module flags section,
3782 using the ``Linker Options`` key. The merge behavior for this flag is required
3783 to be ``AppendUnique``, and the value for the key is expected to be a metadata
3784 node which should be a list of other metadata nodes, each of which should be a
3785 list of metadata strings defining linker options.
3787 For example, the following metadata section specifies two separate sets of
3788 linker options, presumably to link against ``libz`` and the ``Cocoa``
3791 !0 = !{ i32 6, !"Linker Options",
3794 !{ !"-framework", !"Cocoa" } } }
3795 !llvm.module.flags = !{ !0 }
3797 The metadata encoding as lists of lists of options, as opposed to a collapsed
3798 list of options, is chosen so that the IR encoding can use multiple option
3799 strings to specify e.g., a single library, while still having that specifier be
3800 preserved as an atomic element that can be recognized by a target specific
3801 assembly writer or object file emitter.
3803 Each individual option is required to be either a valid option for the target's
3804 linker, or an option that is reserved by the target specific assembly writer or
3805 object file emitter. No other aspect of these options is defined by the IR.
3807 C type width Module Flags Metadata
3808 ----------------------------------
3810 The ARM backend emits a section into each generated object file describing the
3811 options that it was compiled with (in a compiler-independent way) to prevent
3812 linking incompatible objects, and to allow automatic library selection. Some
3813 of these options are not visible at the IR level, namely wchar_t width and enum
3816 To pass this information to the backend, these options are encoded in module
3817 flags metadata, using the following key-value pairs:
3827 - * 0 --- sizeof(wchar_t) == 4
3828 * 1 --- sizeof(wchar_t) == 2
3831 - * 0 --- Enums are at least as large as an ``int``.
3832 * 1 --- Enums are stored in the smallest integer type which can
3833 represent all of its values.
3835 For example, the following metadata section specifies that the module was
3836 compiled with a ``wchar_t`` width of 4 bytes, and the underlying type of an
3837 enum is the smallest type which can represent all of its values::
3839 !llvm.module.flags = !{!0, !1}
3840 !0 = !{i32 1, !"short_wchar", i32 1}
3841 !1 = !{i32 1, !"short_enum", i32 0}
3843 .. _intrinsicglobalvariables:
3845 Intrinsic Global Variables
3846 ==========================
3848 LLVM has a number of "magic" global variables that contain data that
3849 affect code generation or other IR semantics. These are documented here.
3850 All globals of this sort should have a section specified as
3851 "``llvm.metadata``". This section and all globals that start with
3852 "``llvm.``" are reserved for use by LLVM.
3856 The '``llvm.used``' Global Variable
3857 -----------------------------------
3859 The ``@llvm.used`` global is an array which has
3860 :ref:`appending linkage <linkage_appending>`. This array contains a list of
3861 pointers to named global variables, functions and aliases which may optionally
3862 have a pointer cast formed of bitcast or getelementptr. For example, a legal
3865 .. code-block:: llvm
3870 @llvm.used = appending global [2 x i8*] [
3872 i8* bitcast (i32* @Y to i8*)
3873 ], section "llvm.metadata"
3875 If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler,
3876 and linker are required to treat the symbol as if there is a reference to the
3877 symbol that it cannot see (which is why they have to be named). For example, if
3878 a variable has internal linkage and no references other than that from the
3879 ``@llvm.used`` list, it cannot be deleted. This is commonly used to represent
3880 references from inline asms and other things the compiler cannot "see", and
3881 corresponds to "``attribute((used))``" in GNU C.
3883 On some targets, the code generator must emit a directive to the
3884 assembler or object file to prevent the assembler and linker from
3885 molesting the symbol.
3887 .. _gv_llvmcompilerused:
3889 The '``llvm.compiler.used``' Global Variable
3890 --------------------------------------------
3892 The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used``
3893 directive, except that it only prevents the compiler from touching the
3894 symbol. On targets that support it, this allows an intelligent linker to
3895 optimize references to the symbol without being impeded as it would be
3898 This is a rare construct that should only be used in rare circumstances,
3899 and should not be exposed to source languages.
3901 .. _gv_llvmglobalctors:
3903 The '``llvm.global_ctors``' Global Variable
3904 -------------------------------------------
3906 .. code-block:: llvm
3908 %0 = type { i32, void ()*, i8* }
3909 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor, i8* @data }]
3911 The ``@llvm.global_ctors`` array contains a list of constructor
3912 functions, priorities, and an optional associated global or function.
3913 The functions referenced by this array will be called in ascending order
3914 of priority (i.e. lowest first) when the module is loaded. The order of
3915 functions with the same priority is not defined.
3917 If the third field is present, non-null, and points to a global variable
3918 or function, the initializer function will only run if the associated
3919 data from the current module is not discarded.
3921 .. _llvmglobaldtors:
3923 The '``llvm.global_dtors``' Global Variable
3924 -------------------------------------------
3926 .. code-block:: llvm
3928 %0 = type { i32, void ()*, i8* }
3929 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor, i8* @data }]
3931 The ``@llvm.global_dtors`` array contains a list of destructor
3932 functions, priorities, and an optional associated global or function.
3933 The functions referenced by this array will be called in descending
3934 order of priority (i.e. highest first) when the module is unloaded. The
3935 order of functions with the same priority is not defined.
3937 If the third field is present, non-null, and points to a global variable
3938 or function, the destructor function will only run if the associated
3939 data from the current module is not discarded.
3941 Instruction Reference
3942 =====================
3944 The LLVM instruction set consists of several different classifications
3945 of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary
3946 instructions <binaryops>`, :ref:`bitwise binary
3947 instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and
3948 :ref:`other instructions <otherops>`.
3952 Terminator Instructions
3953 -----------------------
3955 As mentioned :ref:`previously <functionstructure>`, every basic block in a
3956 program ends with a "Terminator" instruction, which indicates which
3957 block should be executed after the current block is finished. These
3958 terminator instructions typically yield a '``void``' value: they produce
3959 control flow, not values (the one exception being the
3960 ':ref:`invoke <i_invoke>`' instruction).
3962 The terminator instructions are: ':ref:`ret <i_ret>`',
3963 ':ref:`br <i_br>`', ':ref:`switch <i_switch>`',
3964 ':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`',
3965 ':ref:`resume <i_resume>`', and ':ref:`unreachable <i_unreachable>`'.
3969 '``ret``' Instruction
3970 ^^^^^^^^^^^^^^^^^^^^^
3977 ret <type> <value> ; Return a value from a non-void function
3978 ret void ; Return from void function
3983 The '``ret``' instruction is used to return control flow (and optionally
3984 a value) from a function back to the caller.
3986 There are two forms of the '``ret``' instruction: one that returns a
3987 value and then causes control flow, and one that just causes control
3993 The '``ret``' instruction optionally accepts a single argument, the
3994 return value. The type of the return value must be a ':ref:`first
3995 class <t_firstclass>`' type.
3997 A function is not :ref:`well formed <wellformed>` if it it has a non-void
3998 return type and contains a '``ret``' instruction with no return value or
3999 a return value with a type that does not match its type, or if it has a
4000 void return type and contains a '``ret``' instruction with a return
4006 When the '``ret``' instruction is executed, control flow returns back to
4007 the calling function's context. If the caller is a
4008 ":ref:`call <i_call>`" instruction, execution continues at the
4009 instruction after the call. If the caller was an
4010 ":ref:`invoke <i_invoke>`" instruction, execution continues at the
4011 beginning of the "normal" destination block. If the instruction returns
4012 a value, that value shall set the call or invoke instruction's return
4018 .. code-block:: llvm
4020 ret i32 5 ; Return an integer value of 5
4021 ret void ; Return from a void function
4022 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2
4026 '``br``' Instruction
4027 ^^^^^^^^^^^^^^^^^^^^
4034 br i1 <cond>, label <iftrue>, label <iffalse>
4035 br label <dest> ; Unconditional branch
4040 The '``br``' instruction is used to cause control flow to transfer to a
4041 different basic block in the current function. There are two forms of
4042 this instruction, corresponding to a conditional branch and an
4043 unconditional branch.
4048 The conditional branch form of the '``br``' instruction takes a single
4049 '``i1``' value and two '``label``' values. The unconditional form of the
4050 '``br``' instruction takes a single '``label``' value as a target.
4055 Upon execution of a conditional '``br``' instruction, the '``i1``'
4056 argument is evaluated. If the value is ``true``, control flows to the
4057 '``iftrue``' ``label`` argument. If "cond" is ``false``, control flows
4058 to the '``iffalse``' ``label`` argument.
4063 .. code-block:: llvm
4066 %cond = icmp eq i32 %a, %b
4067 br i1 %cond, label %IfEqual, label %IfUnequal
4075 '``switch``' Instruction
4076 ^^^^^^^^^^^^^^^^^^^^^^^^
4083 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
4088 The '``switch``' instruction is used to transfer control flow to one of
4089 several different places. It is a generalization of the '``br``'
4090 instruction, allowing a branch to occur to one of many possible
4096 The '``switch``' instruction uses three parameters: an integer
4097 comparison value '``value``', a default '``label``' destination, and an
4098 array of pairs of comparison value constants and '``label``'s. The table
4099 is not allowed to contain duplicate constant entries.
4104 The ``switch`` instruction specifies a table of values and destinations.
4105 When the '``switch``' instruction is executed, this table is searched
4106 for the given value. If the value is found, control flow is transferred
4107 to the corresponding destination; otherwise, control flow is transferred
4108 to the default destination.
4113 Depending on properties of the target machine and the particular
4114 ``switch`` instruction, this instruction may be code generated in
4115 different ways. For example, it could be generated as a series of
4116 chained conditional branches or with a lookup table.
4121 .. code-block:: llvm
4123 ; Emulate a conditional br instruction
4124 %Val = zext i1 %value to i32
4125 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
4127 ; Emulate an unconditional br instruction
4128 switch i32 0, label %dest [ ]
4130 ; Implement a jump table:
4131 switch i32 %val, label %otherwise [ i32 0, label %onzero
4133 i32 2, label %ontwo ]
4137 '``indirectbr``' Instruction
4138 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4145 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
4150 The '``indirectbr``' instruction implements an indirect branch to a
4151 label within the current function, whose address is specified by
4152 "``address``". Address must be derived from a
4153 :ref:`blockaddress <blockaddress>` constant.
4158 The '``address``' argument is the address of the label to jump to. The
4159 rest of the arguments indicate the full set of possible destinations
4160 that the address may point to. Blocks are allowed to occur multiple
4161 times in the destination list, though this isn't particularly useful.
4163 This destination list is required so that dataflow analysis has an
4164 accurate understanding of the CFG.
4169 Control transfers to the block specified in the address argument. All
4170 possible destination blocks must be listed in the label list, otherwise
4171 this instruction has undefined behavior. This implies that jumps to
4172 labels defined in other functions have undefined behavior as well.
4177 This is typically implemented with a jump through a register.
4182 .. code-block:: llvm
4184 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
4188 '``invoke``' Instruction
4189 ^^^^^^^^^^^^^^^^^^^^^^^^
4196 <result> = invoke [cconv] [ret attrs] <ptr to function ty> <function ptr val>(<function args>) [fn attrs]
4197 to label <normal label> unwind label <exception label>
4202 The '``invoke``' instruction causes control to transfer to a specified
4203 function, with the possibility of control flow transfer to either the
4204 '``normal``' label or the '``exception``' label. If the callee function
4205 returns with the "``ret``" instruction, control flow will return to the
4206 "normal" label. If the callee (or any indirect callees) returns via the
4207 ":ref:`resume <i_resume>`" instruction or other exception handling
4208 mechanism, control is interrupted and continued at the dynamically
4209 nearest "exception" label.
4211 The '``exception``' label is a `landing
4212 pad <ExceptionHandling.html#overview>`_ for the exception. As such,
4213 '``exception``' label is required to have the
4214 ":ref:`landingpad <i_landingpad>`" instruction, which contains the
4215 information about the behavior of the program after unwinding happens,
4216 as its first non-PHI instruction. The restrictions on the
4217 "``landingpad``" instruction's tightly couples it to the "``invoke``"
4218 instruction, so that the important information contained within the
4219 "``landingpad``" instruction can't be lost through normal code motion.
4224 This instruction requires several arguments:
4226 #. The optional "cconv" marker indicates which :ref:`calling
4227 convention <callingconv>` the call should use. If none is
4228 specified, the call defaults to using C calling conventions.
4229 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
4230 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
4232 #. '``ptr to function ty``': shall be the signature of the pointer to
4233 function value being invoked. In most cases, this is a direct
4234 function invocation, but indirect ``invoke``'s are just as possible,
4235 branching off an arbitrary pointer to function value.
4236 #. '``function ptr val``': An LLVM value containing a pointer to a
4237 function to be invoked.
4238 #. '``function args``': argument list whose types match the function
4239 signature argument types and parameter attributes. All arguments must
4240 be of :ref:`first class <t_firstclass>` type. If the function signature
4241 indicates the function accepts a variable number of arguments, the
4242 extra arguments can be specified.
4243 #. '``normal label``': the label reached when the called function
4244 executes a '``ret``' instruction.
4245 #. '``exception label``': the label reached when a callee returns via
4246 the :ref:`resume <i_resume>` instruction or other exception handling
4248 #. The optional :ref:`function attributes <fnattrs>` list. Only
4249 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
4250 attributes are valid here.
4255 This instruction is designed to operate as a standard '``call``'
4256 instruction in most regards. The primary difference is that it
4257 establishes an association with a label, which is used by the runtime
4258 library to unwind the stack.
4260 This instruction is used in languages with destructors to ensure that
4261 proper cleanup is performed in the case of either a ``longjmp`` or a
4262 thrown exception. Additionally, this is important for implementation of
4263 '``catch``' clauses in high-level languages that support them.
4265 For the purposes of the SSA form, the definition of the value returned
4266 by the '``invoke``' instruction is deemed to occur on the edge from the
4267 current block to the "normal" label. If the callee unwinds then no
4268 return value is available.
4273 .. code-block:: llvm
4275 %retval = invoke i32 @Test(i32 15) to label %Continue
4276 unwind label %TestCleanup ; i32:retval set
4277 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue
4278 unwind label %TestCleanup ; i32:retval set
4282 '``resume``' Instruction
4283 ^^^^^^^^^^^^^^^^^^^^^^^^
4290 resume <type> <value>
4295 The '``resume``' instruction is a terminator instruction that has no
4301 The '``resume``' instruction requires one argument, which must have the
4302 same type as the result of any '``landingpad``' instruction in the same
4308 The '``resume``' instruction resumes propagation of an existing
4309 (in-flight) exception whose unwinding was interrupted with a
4310 :ref:`landingpad <i_landingpad>` instruction.
4315 .. code-block:: llvm
4317 resume { i8*, i32 } %exn
4321 '``unreachable``' Instruction
4322 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4334 The '``unreachable``' instruction has no defined semantics. This
4335 instruction is used to inform the optimizer that a particular portion of
4336 the code is not reachable. This can be used to indicate that the code
4337 after a no-return function cannot be reached, and other facts.
4342 The '``unreachable``' instruction has no defined semantics.
4349 Binary operators are used to do most of the computation in a program.
4350 They require two operands of the same type, execute an operation on
4351 them, and produce a single value. The operands might represent multiple
4352 data, as is the case with the :ref:`vector <t_vector>` data type. The
4353 result value has the same type as its operands.
4355 There are several different binary operators:
4359 '``add``' Instruction
4360 ^^^^^^^^^^^^^^^^^^^^^
4367 <result> = add <ty> <op1>, <op2> ; yields ty:result
4368 <result> = add nuw <ty> <op1>, <op2> ; yields ty:result
4369 <result> = add nsw <ty> <op1>, <op2> ; yields ty:result
4370 <result> = add nuw nsw <ty> <op1>, <op2> ; yields ty:result
4375 The '``add``' instruction returns the sum of its two operands.
4380 The two arguments to the '``add``' instruction must be
4381 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4382 arguments must have identical types.
4387 The value produced is the integer sum of the two operands.
4389 If the sum has unsigned overflow, the result returned is the
4390 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
4393 Because LLVM integers use a two's complement representation, this
4394 instruction is appropriate for both signed and unsigned integers.
4396 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4397 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4398 result value of the ``add`` is a :ref:`poison value <poisonvalues>` if
4399 unsigned and/or signed overflow, respectively, occurs.
4404 .. code-block:: llvm
4406 <result> = add i32 4, %var ; yields i32:result = 4 + %var
4410 '``fadd``' Instruction
4411 ^^^^^^^^^^^^^^^^^^^^^^
4418 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4423 The '``fadd``' instruction returns the sum of its two operands.
4428 The two arguments to the '``fadd``' instruction must be :ref:`floating
4429 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4430 Both arguments must have identical types.
4435 The value produced is the floating point sum of the two operands. This
4436 instruction can also take any number of :ref:`fast-math flags <fastmath>`,
4437 which are optimization hints to enable otherwise unsafe floating point
4443 .. code-block:: llvm
4445 <result> = fadd float 4.0, %var ; yields float:result = 4.0 + %var
4447 '``sub``' Instruction
4448 ^^^^^^^^^^^^^^^^^^^^^
4455 <result> = sub <ty> <op1>, <op2> ; yields ty:result
4456 <result> = sub nuw <ty> <op1>, <op2> ; yields ty:result
4457 <result> = sub nsw <ty> <op1>, <op2> ; yields ty:result
4458 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields ty:result
4463 The '``sub``' instruction returns the difference of its two operands.
4465 Note that the '``sub``' instruction is used to represent the '``neg``'
4466 instruction present in most other intermediate representations.
4471 The two arguments to the '``sub``' instruction must be
4472 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4473 arguments must have identical types.
4478 The value produced is the integer difference of the two operands.
4480 If the difference has unsigned overflow, the result returned is the
4481 mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of
4484 Because LLVM integers use a two's complement representation, this
4485 instruction is appropriate for both signed and unsigned integers.
4487 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4488 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4489 result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if
4490 unsigned and/or signed overflow, respectively, occurs.
4495 .. code-block:: llvm
4497 <result> = sub i32 4, %var ; yields i32:result = 4 - %var
4498 <result> = sub i32 0, %val ; yields i32:result = -%var
4502 '``fsub``' Instruction
4503 ^^^^^^^^^^^^^^^^^^^^^^
4510 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4515 The '``fsub``' instruction returns the difference of its two operands.
4517 Note that the '``fsub``' instruction is used to represent the '``fneg``'
4518 instruction present in most other intermediate representations.
4523 The two arguments to the '``fsub``' instruction must be :ref:`floating
4524 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4525 Both arguments must have identical types.
4530 The value produced is the floating point difference of the two operands.
4531 This instruction can also take any number of :ref:`fast-math
4532 flags <fastmath>`, which are optimization hints to enable otherwise
4533 unsafe floating point optimizations:
4538 .. code-block:: llvm
4540 <result> = fsub float 4.0, %var ; yields float:result = 4.0 - %var
4541 <result> = fsub float -0.0, %val ; yields float:result = -%var
4543 '``mul``' Instruction
4544 ^^^^^^^^^^^^^^^^^^^^^
4551 <result> = mul <ty> <op1>, <op2> ; yields ty:result
4552 <result> = mul nuw <ty> <op1>, <op2> ; yields ty:result
4553 <result> = mul nsw <ty> <op1>, <op2> ; yields ty:result
4554 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields ty:result
4559 The '``mul``' instruction returns the product of its two operands.
4564 The two arguments to the '``mul``' instruction must be
4565 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4566 arguments must have identical types.
4571 The value produced is the integer product of the two operands.
4573 If the result of the multiplication has unsigned overflow, the result
4574 returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the
4575 bit width of the result.
4577 Because LLVM integers use a two's complement representation, and the
4578 result is the same width as the operands, this instruction returns the
4579 correct result for both signed and unsigned integers. If a full product
4580 (e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be
4581 sign-extended or zero-extended as appropriate to the width of the full
4584 ``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap",
4585 respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the
4586 result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if
4587 unsigned and/or signed overflow, respectively, occurs.
4592 .. code-block:: llvm
4594 <result> = mul i32 4, %var ; yields i32:result = 4 * %var
4598 '``fmul``' Instruction
4599 ^^^^^^^^^^^^^^^^^^^^^^
4606 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4611 The '``fmul``' instruction returns the product of its two operands.
4616 The two arguments to the '``fmul``' instruction must be :ref:`floating
4617 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4618 Both arguments must have identical types.
4623 The value produced is the floating point product of the two operands.
4624 This instruction can also take any number of :ref:`fast-math
4625 flags <fastmath>`, which are optimization hints to enable otherwise
4626 unsafe floating point optimizations:
4631 .. code-block:: llvm
4633 <result> = fmul float 4.0, %var ; yields float:result = 4.0 * %var
4635 '``udiv``' Instruction
4636 ^^^^^^^^^^^^^^^^^^^^^^
4643 <result> = udiv <ty> <op1>, <op2> ; yields ty:result
4644 <result> = udiv exact <ty> <op1>, <op2> ; yields ty:result
4649 The '``udiv``' instruction returns the quotient of its two operands.
4654 The two arguments to the '``udiv``' instruction must be
4655 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4656 arguments must have identical types.
4661 The value produced is the unsigned integer quotient of the two operands.
4663 Note that unsigned integer division and signed integer division are
4664 distinct operations; for signed integer division, use '``sdiv``'.
4666 Division by zero leads to undefined behavior.
4668 If the ``exact`` keyword is present, the result value of the ``udiv`` is
4669 a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as
4670 such, "((a udiv exact b) mul b) == a").
4675 .. code-block:: llvm
4677 <result> = udiv i32 4, %var ; yields i32:result = 4 / %var
4679 '``sdiv``' Instruction
4680 ^^^^^^^^^^^^^^^^^^^^^^
4687 <result> = sdiv <ty> <op1>, <op2> ; yields ty:result
4688 <result> = sdiv exact <ty> <op1>, <op2> ; yields ty:result
4693 The '``sdiv``' instruction returns the quotient of its two operands.
4698 The two arguments to the '``sdiv``' instruction must be
4699 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4700 arguments must have identical types.
4705 The value produced is the signed integer quotient of the two operands
4706 rounded towards zero.
4708 Note that signed integer division and unsigned integer division are
4709 distinct operations; for unsigned integer division, use '``udiv``'.
4711 Division by zero leads to undefined behavior. Overflow also leads to
4712 undefined behavior; this is a rare case, but can occur, for example, by
4713 doing a 32-bit division of -2147483648 by -1.
4715 If the ``exact`` keyword is present, the result value of the ``sdiv`` is
4716 a :ref:`poison value <poisonvalues>` if the result would be rounded.
4721 .. code-block:: llvm
4723 <result> = sdiv i32 4, %var ; yields i32:result = 4 / %var
4727 '``fdiv``' Instruction
4728 ^^^^^^^^^^^^^^^^^^^^^^
4735 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4740 The '``fdiv``' instruction returns the quotient of its two operands.
4745 The two arguments to the '``fdiv``' instruction must be :ref:`floating
4746 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4747 Both arguments must have identical types.
4752 The value produced is the floating point quotient of the two operands.
4753 This instruction can also take any number of :ref:`fast-math
4754 flags <fastmath>`, which are optimization hints to enable otherwise
4755 unsafe floating point optimizations:
4760 .. code-block:: llvm
4762 <result> = fdiv float 4.0, %var ; yields float:result = 4.0 / %var
4764 '``urem``' Instruction
4765 ^^^^^^^^^^^^^^^^^^^^^^
4772 <result> = urem <ty> <op1>, <op2> ; yields ty:result
4777 The '``urem``' instruction returns the remainder from the unsigned
4778 division of its two arguments.
4783 The two arguments to the '``urem``' instruction must be
4784 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4785 arguments must have identical types.
4790 This instruction returns the unsigned integer *remainder* of a division.
4791 This instruction always performs an unsigned division to get the
4794 Note that unsigned integer remainder and signed integer remainder are
4795 distinct operations; for signed integer remainder, use '``srem``'.
4797 Taking the remainder of a division by zero leads to undefined behavior.
4802 .. code-block:: llvm
4804 <result> = urem i32 4, %var ; yields i32:result = 4 % %var
4806 '``srem``' Instruction
4807 ^^^^^^^^^^^^^^^^^^^^^^
4814 <result> = srem <ty> <op1>, <op2> ; yields ty:result
4819 The '``srem``' instruction returns the remainder from the signed
4820 division of its two operands. This instruction can also take
4821 :ref:`vector <t_vector>` versions of the values in which case the elements
4827 The two arguments to the '``srem``' instruction must be
4828 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
4829 arguments must have identical types.
4834 This instruction returns the *remainder* of a division (where the result
4835 is either zero or has the same sign as the dividend, ``op1``), not the
4836 *modulo* operator (where the result is either zero or has the same sign
4837 as the divisor, ``op2``) of a value. For more information about the
4838 difference, see `The Math
4839 Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a
4840 table of how this is implemented in various languages, please see
4842 operation <http://en.wikipedia.org/wiki/Modulo_operation>`_.
4844 Note that signed integer remainder and unsigned integer remainder are
4845 distinct operations; for unsigned integer remainder, use '``urem``'.
4847 Taking the remainder of a division by zero leads to undefined behavior.
4848 Overflow also leads to undefined behavior; this is a rare case, but can
4849 occur, for example, by taking the remainder of a 32-bit division of
4850 -2147483648 by -1. (The remainder doesn't actually overflow, but this
4851 rule lets srem be implemented using instructions that return both the
4852 result of the division and the remainder.)
4857 .. code-block:: llvm
4859 <result> = srem i32 4, %var ; yields i32:result = 4 % %var
4863 '``frem``' Instruction
4864 ^^^^^^^^^^^^^^^^^^^^^^
4871 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result
4876 The '``frem``' instruction returns the remainder from the division of
4882 The two arguments to the '``frem``' instruction must be :ref:`floating
4883 point <t_floating>` or :ref:`vector <t_vector>` of floating point values.
4884 Both arguments must have identical types.
4889 This instruction returns the *remainder* of a division. The remainder
4890 has the same sign as the dividend. This instruction can also take any
4891 number of :ref:`fast-math flags <fastmath>`, which are optimization hints
4892 to enable otherwise unsafe floating point optimizations:
4897 .. code-block:: llvm
4899 <result> = frem float 4.0, %var ; yields float:result = 4.0 % %var
4903 Bitwise Binary Operations
4904 -------------------------
4906 Bitwise binary operators are used to do various forms of bit-twiddling
4907 in a program. They are generally very efficient instructions and can
4908 commonly be strength reduced from other instructions. They require two
4909 operands of the same type, execute an operation on them, and produce a
4910 single value. The resulting value is the same type as its operands.
4912 '``shl``' Instruction
4913 ^^^^^^^^^^^^^^^^^^^^^
4920 <result> = shl <ty> <op1>, <op2> ; yields ty:result
4921 <result> = shl nuw <ty> <op1>, <op2> ; yields ty:result
4922 <result> = shl nsw <ty> <op1>, <op2> ; yields ty:result
4923 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields ty:result
4928 The '``shl``' instruction returns the first operand shifted to the left
4929 a specified number of bits.
4934 Both arguments to the '``shl``' instruction must be the same
4935 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4936 '``op2``' is treated as an unsigned value.
4941 The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`,
4942 where ``n`` is the width of the result. If ``op2`` is (statically or
4943 dynamically) negative or equal to or larger than the number of bits in
4944 ``op1``, the result is undefined. If the arguments are vectors, each
4945 vector element of ``op1`` is shifted by the corresponding shift amount
4948 If the ``nuw`` keyword is present, then the shift produces a :ref:`poison
4949 value <poisonvalues>` if it shifts out any non-zero bits. If the
4950 ``nsw`` keyword is present, then the shift produces a :ref:`poison
4951 value <poisonvalues>` if it shifts out any bits that disagree with the
4952 resultant sign bit. As such, NUW/NSW have the same semantics as they
4953 would if the shift were expressed as a mul instruction with the same
4954 nsw/nuw bits in (mul %op1, (shl 1, %op2)).
4959 .. code-block:: llvm
4961 <result> = shl i32 4, %var ; yields i32: 4 << %var
4962 <result> = shl i32 4, 2 ; yields i32: 16
4963 <result> = shl i32 1, 10 ; yields i32: 1024
4964 <result> = shl i32 1, 32 ; undefined
4965 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4>
4967 '``lshr``' Instruction
4968 ^^^^^^^^^^^^^^^^^^^^^^
4975 <result> = lshr <ty> <op1>, <op2> ; yields ty:result
4976 <result> = lshr exact <ty> <op1>, <op2> ; yields ty:result
4981 The '``lshr``' instruction (logical shift right) returns the first
4982 operand shifted to the right a specified number of bits with zero fill.
4987 Both arguments to the '``lshr``' instruction must be the same
4988 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
4989 '``op2``' is treated as an unsigned value.
4994 This instruction always performs a logical shift right operation. The
4995 most significant bits of the result will be filled with zero bits after
4996 the shift. If ``op2`` is (statically or dynamically) equal to or larger
4997 than the number of bits in ``op1``, the result is undefined. If the
4998 arguments are vectors, each vector element of ``op1`` is shifted by the
4999 corresponding shift amount in ``op2``.
5001 If the ``exact`` keyword is present, the result value of the ``lshr`` is
5002 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
5008 .. code-block:: llvm
5010 <result> = lshr i32 4, 1 ; yields i32:result = 2
5011 <result> = lshr i32 4, 2 ; yields i32:result = 1
5012 <result> = lshr i8 4, 3 ; yields i8:result = 0
5013 <result> = lshr i8 -2, 1 ; yields i8:result = 0x7F
5014 <result> = lshr i32 1, 32 ; undefined
5015 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1>
5017 '``ashr``' Instruction
5018 ^^^^^^^^^^^^^^^^^^^^^^
5025 <result> = ashr <ty> <op1>, <op2> ; yields ty:result
5026 <result> = ashr exact <ty> <op1>, <op2> ; yields ty:result
5031 The '``ashr``' instruction (arithmetic shift right) returns the first
5032 operand shifted to the right a specified number of bits with sign
5038 Both arguments to the '``ashr``' instruction must be the same
5039 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type.
5040 '``op2``' is treated as an unsigned value.
5045 This instruction always performs an arithmetic shift right operation,
5046 The most significant bits of the result will be filled with the sign bit
5047 of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger
5048 than the number of bits in ``op1``, the result is undefined. If the
5049 arguments are vectors, each vector element of ``op1`` is shifted by the
5050 corresponding shift amount in ``op2``.
5052 If the ``exact`` keyword is present, the result value of the ``ashr`` is
5053 a :ref:`poison value <poisonvalues>` if any of the bits shifted out are
5059 .. code-block:: llvm
5061 <result> = ashr i32 4, 1 ; yields i32:result = 2
5062 <result> = ashr i32 4, 2 ; yields i32:result = 1
5063 <result> = ashr i8 4, 3 ; yields i8:result = 0
5064 <result> = ashr i8 -2, 1 ; yields i8:result = -1
5065 <result> = ashr i32 1, 32 ; undefined
5066 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0>
5068 '``and``' Instruction
5069 ^^^^^^^^^^^^^^^^^^^^^
5076 <result> = and <ty> <op1>, <op2> ; yields ty:result
5081 The '``and``' instruction returns the bitwise logical and of its two
5087 The two arguments to the '``and``' instruction must be
5088 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5089 arguments must have identical types.
5094 The truth table used for the '``and``' instruction is:
5111 .. code-block:: llvm
5113 <result> = and i32 4, %var ; yields i32:result = 4 & %var
5114 <result> = and i32 15, 40 ; yields i32:result = 8
5115 <result> = and i32 4, 8 ; yields i32:result = 0
5117 '``or``' Instruction
5118 ^^^^^^^^^^^^^^^^^^^^
5125 <result> = or <ty> <op1>, <op2> ; yields ty:result
5130 The '``or``' instruction returns the bitwise logical inclusive or of its
5136 The two arguments to the '``or``' instruction must be
5137 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5138 arguments must have identical types.
5143 The truth table used for the '``or``' instruction is:
5162 <result> = or i32 4, %var ; yields i32:result = 4 | %var
5163 <result> = or i32 15, 40 ; yields i32:result = 47
5164 <result> = or i32 4, 8 ; yields i32:result = 12
5166 '``xor``' Instruction
5167 ^^^^^^^^^^^^^^^^^^^^^
5174 <result> = xor <ty> <op1>, <op2> ; yields ty:result
5179 The '``xor``' instruction returns the bitwise logical exclusive or of
5180 its two operands. The ``xor`` is used to implement the "one's
5181 complement" operation, which is the "~" operator in C.
5186 The two arguments to the '``xor``' instruction must be
5187 :ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both
5188 arguments must have identical types.
5193 The truth table used for the '``xor``' instruction is:
5210 .. code-block:: llvm
5212 <result> = xor i32 4, %var ; yields i32:result = 4 ^ %var
5213 <result> = xor i32 15, 40 ; yields i32:result = 39
5214 <result> = xor i32 4, 8 ; yields i32:result = 12
5215 <result> = xor i32 %V, -1 ; yields i32:result = ~%V
5220 LLVM supports several instructions to represent vector operations in a
5221 target-independent manner. These instructions cover the element-access
5222 and vector-specific operations needed to process vectors effectively.
5223 While LLVM does directly support these vector operations, many
5224 sophisticated algorithms will want to use target-specific intrinsics to
5225 take full advantage of a specific target.
5227 .. _i_extractelement:
5229 '``extractelement``' Instruction
5230 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5237 <result> = extractelement <n x <ty>> <val>, <ty2> <idx> ; yields <ty>
5242 The '``extractelement``' instruction extracts a single scalar element
5243 from a vector at a specified index.
5248 The first operand of an '``extractelement``' instruction is a value of
5249 :ref:`vector <t_vector>` type. The second operand is an index indicating
5250 the position from which to extract the element. The index may be a
5251 variable of any integer type.
5256 The result is a scalar of the same type as the element type of ``val``.
5257 Its value is the value at position ``idx`` of ``val``. If ``idx``
5258 exceeds the length of ``val``, the results are undefined.
5263 .. code-block:: llvm
5265 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32
5267 .. _i_insertelement:
5269 '``insertelement``' Instruction
5270 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5277 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <n x <ty>>
5282 The '``insertelement``' instruction inserts a scalar element into a
5283 vector at a specified index.
5288 The first operand of an '``insertelement``' instruction is a value of
5289 :ref:`vector <t_vector>` type. The second operand is a scalar value whose
5290 type must equal the element type of the first operand. The third operand
5291 is an index indicating the position at which to insert the value. The
5292 index may be a variable of any integer type.
5297 The result is a vector of the same type as ``val``. Its element values
5298 are those of ``val`` except at position ``idx``, where it gets the value
5299 ``elt``. If ``idx`` exceeds the length of ``val``, the results are
5305 .. code-block:: llvm
5307 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32>
5309 .. _i_shufflevector:
5311 '``shufflevector``' Instruction
5312 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5319 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>>
5324 The '``shufflevector``' instruction constructs a permutation of elements
5325 from two input vectors, returning a vector with the same element type as
5326 the input and length that is the same as the shuffle mask.
5331 The first two operands of a '``shufflevector``' instruction are vectors
5332 with the same type. The third argument is a shuffle mask whose element
5333 type is always 'i32'. The result of the instruction is a vector whose
5334 length is the same as the shuffle mask and whose element type is the
5335 same as the element type of the first two operands.
5337 The shuffle mask operand is required to be a constant vector with either
5338 constant integer or undef values.
5343 The elements of the two input vectors are numbered from left to right
5344 across both of the vectors. The shuffle mask operand specifies, for each
5345 element of the result vector, which element of the two input vectors the
5346 result element gets. The element selector may be undef (meaning "don't
5347 care") and the second operand may be undef if performing a shuffle from
5353 .. code-block:: llvm
5355 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
5356 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32>
5357 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
5358 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle.
5359 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
5360 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32>
5361 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
5362 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32>
5364 Aggregate Operations
5365 --------------------
5367 LLVM supports several instructions for working with
5368 :ref:`aggregate <t_aggregate>` values.
5372 '``extractvalue``' Instruction
5373 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5380 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
5385 The '``extractvalue``' instruction extracts the value of a member field
5386 from an :ref:`aggregate <t_aggregate>` value.
5391 The first operand of an '``extractvalue``' instruction is a value of
5392 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The operands are
5393 constant indices to specify which value to extract in a similar manner
5394 as indices in a '``getelementptr``' instruction.
5396 The major differences to ``getelementptr`` indexing are:
5398 - Since the value being indexed is not a pointer, the first index is
5399 omitted and assumed to be zero.
5400 - At least one index must be specified.
5401 - Not only struct indices but also array indices must be in bounds.
5406 The result is the value at the position in the aggregate specified by
5412 .. code-block:: llvm
5414 <result> = extractvalue {i32, float} %agg, 0 ; yields i32
5418 '``insertvalue``' Instruction
5419 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5426 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type>
5431 The '``insertvalue``' instruction inserts a value into a member field in
5432 an :ref:`aggregate <t_aggregate>` value.
5437 The first operand of an '``insertvalue``' instruction is a value of
5438 :ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is
5439 a first-class value to insert. The following operands are constant
5440 indices indicating the position at which to insert the value in a
5441 similar manner as indices in a '``extractvalue``' instruction. The value
5442 to insert must have the same type as the value identified by the
5448 The result is an aggregate of the same type as ``val``. Its value is
5449 that of ``val`` except that the value at the position specified by the
5450 indices is that of ``elt``.
5455 .. code-block:: llvm
5457 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef}
5458 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val}
5459 %agg3 = insertvalue {i32, {float}} undef, float %val, 1, 0 ; yields {i32 undef, {float %val}}
5463 Memory Access and Addressing Operations
5464 ---------------------------------------
5466 A key design point of an SSA-based representation is how it represents
5467 memory. In LLVM, no memory locations are in SSA form, which makes things
5468 very simple. This section describes how to read, write, and allocate
5473 '``alloca``' Instruction
5474 ^^^^^^^^^^^^^^^^^^^^^^^^
5481 <result> = alloca [inalloca] <type> [, <ty> <NumElements>] [, align <alignment>] ; yields type*:result
5486 The '``alloca``' instruction allocates memory on the stack frame of the
5487 currently executing function, to be automatically released when this
5488 function returns to its caller. The object is always allocated in the
5489 generic address space (address space zero).
5494 The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements``
5495 bytes of memory on the runtime stack, returning a pointer of the
5496 appropriate type to the program. If "NumElements" is specified, it is
5497 the number of elements allocated, otherwise "NumElements" is defaulted
5498 to be one. If a constant alignment is specified, the value result of the
5499 allocation is guaranteed to be aligned to at least that boundary. The
5500 alignment may not be greater than ``1 << 29``. If not specified, or if
5501 zero, the target can choose to align the allocation on any convenient
5502 boundary compatible with the type.
5504 '``type``' may be any sized type.
5509 Memory is allocated; a pointer is returned. The operation is undefined
5510 if there is insufficient stack space for the allocation. '``alloca``'d
5511 memory is automatically released when the function returns. The
5512 '``alloca``' instruction is commonly used to represent automatic
5513 variables that must have an address available. When the function returns
5514 (either with the ``ret`` or ``resume`` instructions), the memory is
5515 reclaimed. Allocating zero bytes is legal, but the result is undefined.
5516 The order in which memory is allocated (ie., which way the stack grows)
5522 .. code-block:: llvm
5524 %ptr = alloca i32 ; yields i32*:ptr
5525 %ptr = alloca i32, i32 4 ; yields i32*:ptr
5526 %ptr = alloca i32, i32 4, align 1024 ; yields i32*:ptr
5527 %ptr = alloca i32, align 1024 ; yields i32*:ptr
5531 '``load``' Instruction
5532 ^^^^^^^^^^^^^^^^^^^^^^
5539 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>][, !nonnull !<index>]
5540 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
5541 !<index> = !{ i32 1 }
5546 The '``load``' instruction is used to read from memory.
5551 The argument to the ``load`` instruction specifies the memory address
5552 from which to load. The pointer must point to a :ref:`first
5553 class <t_firstclass>` type. If the ``load`` is marked as ``volatile``,
5554 then the optimizer is not allowed to modify the number or order of
5555 execution of this ``load`` with other :ref:`volatile
5556 operations <volatile>`.
5558 If the ``load`` is marked as ``atomic``, it takes an extra
5559 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
5560 ``release`` and ``acq_rel`` orderings are not valid on ``load``
5561 instructions. Atomic loads produce :ref:`defined <memmodel>` results
5562 when they may see multiple atomic stores. The type of the pointee must
5563 be an integer type whose bit width is a power of two greater than or
5564 equal to eight and less than or equal to a target-specific size limit.
5565 ``align`` must be explicitly specified on atomic loads, and the load has
5566 undefined behavior if the alignment is not set to a value which is at
5567 least the size in bytes of the pointee. ``!nontemporal`` does not have
5568 any defined semantics for atomic loads.
5570 The optional constant ``align`` argument specifies the alignment of the
5571 operation (that is, the alignment of the memory address). A value of 0
5572 or an omitted ``align`` argument means that the operation has the ABI
5573 alignment for the target. It is the responsibility of the code emitter
5574 to ensure that the alignment information is correct. Overestimating the
5575 alignment results in undefined behavior. Underestimating the alignment
5576 may produce less efficient code. An alignment of 1 is always safe. The
5577 maximum possible alignment is ``1 << 29``.
5579 The optional ``!nontemporal`` metadata must reference a single
5580 metadata name ``<index>`` corresponding to a metadata node with one
5581 ``i32`` entry of value 1. The existence of the ``!nontemporal``
5582 metadata on the instruction tells the optimizer and code generator
5583 that this load is not expected to be reused in the cache. The code
5584 generator may select special instructions to save cache bandwidth, such
5585 as the ``MOVNT`` instruction on x86.
5587 The optional ``!invariant.load`` metadata must reference a single
5588 metadata name ``<index>`` corresponding to a metadata node with no
5589 entries. The existence of the ``!invariant.load`` metadata on the
5590 instruction tells the optimizer and code generator that the address
5591 operand to this load points to memory which can be assumed unchanged.
5592 Being invariant does not imply that a location is dereferenceable,
5593 but it does imply that once the location is known dereferenceable
5594 its value is henceforth unchanging.
5596 The optional ``!nonnull`` metadata must reference a single
5597 metadata name ``<index>`` corresponding to a metadata node with no
5598 entries. The existence of the ``!nonnull`` metadata on the
5599 instruction tells the optimizer that the value loaded is known to
5600 never be null. This is analogous to the ''nonnull'' attribute
5601 on parameters and return values. This metadata can only be applied
5602 to loads of a pointer type.
5607 The location of memory pointed to is loaded. If the value being loaded
5608 is of scalar type then the number of bytes read does not exceed the
5609 minimum number of bytes needed to hold all bits of the type. For
5610 example, loading an ``i24`` reads at most three bytes. When loading a
5611 value of a type like ``i20`` with a size that is not an integral number
5612 of bytes, the result is undefined if the value was not originally
5613 written using a store of the same type.
5618 .. code-block:: llvm
5620 %ptr = alloca i32 ; yields i32*:ptr
5621 store i32 3, i32* %ptr ; yields void
5622 %val = load i32* %ptr ; yields i32:val = i32 3
5626 '``store``' Instruction
5627 ^^^^^^^^^^^^^^^^^^^^^^^
5634 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] ; yields void
5635 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> ; yields void
5640 The '``store``' instruction is used to write to memory.
5645 There are two arguments to the ``store`` instruction: a value to store
5646 and an address at which to store it. The type of the ``<pointer>``
5647 operand must be a pointer to the :ref:`first class <t_firstclass>` type of
5648 the ``<value>`` operand. If the ``store`` is marked as ``volatile``,
5649 then the optimizer is not allowed to modify the number or order of
5650 execution of this ``store`` with other :ref:`volatile
5651 operations <volatile>`.
5653 If the ``store`` is marked as ``atomic``, it takes an extra
5654 :ref:`ordering <ordering>` and optional ``singlethread`` argument. The
5655 ``acquire`` and ``acq_rel`` orderings aren't valid on ``store``
5656 instructions. Atomic loads produce :ref:`defined <memmodel>` results
5657 when they may see multiple atomic stores. The type of the pointee must
5658 be an integer type whose bit width is a power of two greater than or
5659 equal to eight and less than or equal to a target-specific size limit.
5660 ``align`` must be explicitly specified on atomic stores, and the store
5661 has undefined behavior if the alignment is not set to a value which is
5662 at least the size in bytes of the pointee. ``!nontemporal`` does not
5663 have any defined semantics for atomic stores.
5665 The optional constant ``align`` argument specifies the alignment of the
5666 operation (that is, the alignment of the memory address). A value of 0
5667 or an omitted ``align`` argument means that the operation has the ABI
5668 alignment for the target. It is the responsibility of the code emitter
5669 to ensure that the alignment information is correct. Overestimating the
5670 alignment results in undefined behavior. Underestimating the
5671 alignment may produce less efficient code. An alignment of 1 is always
5672 safe. The maximum possible alignment is ``1 << 29``.
5674 The optional ``!nontemporal`` metadata must reference a single metadata
5675 name ``<index>`` corresponding to a metadata node with one ``i32`` entry of
5676 value 1. The existence of the ``!nontemporal`` metadata on the instruction
5677 tells the optimizer and code generator that this load is not expected to
5678 be reused in the cache. The code generator may select special
5679 instructions to save cache bandwidth, such as the MOVNT instruction on
5685 The contents of memory are updated to contain ``<value>`` at the
5686 location specified by the ``<pointer>`` operand. If ``<value>`` is
5687 of scalar type then the number of bytes written does not exceed the
5688 minimum number of bytes needed to hold all bits of the type. For
5689 example, storing an ``i24`` writes at most three bytes. When writing a
5690 value of a type like ``i20`` with a size that is not an integral number
5691 of bytes, it is unspecified what happens to the extra bits that do not
5692 belong to the type, but they will typically be overwritten.
5697 .. code-block:: llvm
5699 %ptr = alloca i32 ; yields i32*:ptr
5700 store i32 3, i32* %ptr ; yields void
5701 %val = load i32* %ptr ; yields i32:val = i32 3
5705 '``fence``' Instruction
5706 ^^^^^^^^^^^^^^^^^^^^^^^
5713 fence [singlethread] <ordering> ; yields void
5718 The '``fence``' instruction is used to introduce happens-before edges
5724 '``fence``' instructions take an :ref:`ordering <ordering>` argument which
5725 defines what *synchronizes-with* edges they add. They can only be given
5726 ``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings.
5731 A fence A which has (at least) ``release`` ordering semantics
5732 *synchronizes with* a fence B with (at least) ``acquire`` ordering
5733 semantics if and only if there exist atomic operations X and Y, both
5734 operating on some atomic object M, such that A is sequenced before X, X
5735 modifies M (either directly or through some side effect of a sequence
5736 headed by X), Y is sequenced before B, and Y observes M. This provides a
5737 *happens-before* dependency between A and B. Rather than an explicit
5738 ``fence``, one (but not both) of the atomic operations X or Y might
5739 provide a ``release`` or ``acquire`` (resp.) ordering constraint and
5740 still *synchronize-with* the explicit ``fence`` and establish the
5741 *happens-before* edge.
5743 A ``fence`` which has ``seq_cst`` ordering, in addition to having both
5744 ``acquire`` and ``release`` semantics specified above, participates in
5745 the global program order of other ``seq_cst`` operations and/or fences.
5747 The optional ":ref:`singlethread <singlethread>`" argument specifies
5748 that the fence only synchronizes with other fences in the same thread.
5749 (This is useful for interacting with signal handlers.)
5754 .. code-block:: llvm
5756 fence acquire ; yields void
5757 fence singlethread seq_cst ; yields void
5761 '``cmpxchg``' Instruction
5762 ^^^^^^^^^^^^^^^^^^^^^^^^^
5769 cmpxchg [weak] [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <success ordering> <failure ordering> ; yields { ty, i1 }
5774 The '``cmpxchg``' instruction is used to atomically modify memory. It
5775 loads a value in memory and compares it to a given value. If they are
5776 equal, it tries to store a new value into the memory.
5781 There are three arguments to the '``cmpxchg``' instruction: an address
5782 to operate on, a value to compare to the value currently be at that
5783 address, and a new value to place at that address if the compared values
5784 are equal. The type of '<cmp>' must be an integer type whose bit width
5785 is a power of two greater than or equal to eight and less than or equal
5786 to a target-specific size limit. '<cmp>' and '<new>' must have the same
5787 type, and the type of '<pointer>' must be a pointer to that type. If the
5788 ``cmpxchg`` is marked as ``volatile``, then the optimizer is not allowed
5789 to modify the number or order of execution of this ``cmpxchg`` with
5790 other :ref:`volatile operations <volatile>`.
5792 The success and failure :ref:`ordering <ordering>` arguments specify how this
5793 ``cmpxchg`` synchronizes with other atomic operations. Both ordering parameters
5794 must be at least ``monotonic``, the ordering constraint on failure must be no
5795 stronger than that on success, and the failure ordering cannot be either
5796 ``release`` or ``acq_rel``.
5798 The optional "``singlethread``" argument declares that the ``cmpxchg``
5799 is only atomic with respect to code (usually signal handlers) running in
5800 the same thread as the ``cmpxchg``. Otherwise the cmpxchg is atomic with
5801 respect to all other code in the system.
5803 The pointer passed into cmpxchg must have alignment greater than or
5804 equal to the size in memory of the operand.
5809 The contents of memory at the location specified by the '``<pointer>``' operand
5810 is read and compared to '``<cmp>``'; if the read value is the equal, the
5811 '``<new>``' is written. The original value at the location is returned, together
5812 with a flag indicating success (true) or failure (false).
5814 If the cmpxchg operation is marked as ``weak`` then a spurious failure is
5815 permitted: the operation may not write ``<new>`` even if the comparison
5818 If the cmpxchg operation is strong (the default), the i1 value is 1 if and only
5819 if the value loaded equals ``cmp``.
5821 A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of
5822 identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic
5823 load with an ordering parameter determined the second ordering parameter.
5828 .. code-block:: llvm
5831 %orig = atomic load i32* %ptr unordered ; yields i32
5835 %cmp = phi i32 [ %orig, %entry ], [%old, %loop]
5836 %squared = mul i32 %cmp, %cmp
5837 %val_success = cmpxchg i32* %ptr, i32 %cmp, i32 %squared acq_rel monotonic ; yields { i32, i1 }
5838 %value_loaded = extractvalue { i32, i1 } %val_success, 0
5839 %success = extractvalue { i32, i1 } %val_success, 1
5840 br i1 %success, label %done, label %loop
5847 '``atomicrmw``' Instruction
5848 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
5855 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> ; yields ty
5860 The '``atomicrmw``' instruction is used to atomically modify memory.
5865 There are three arguments to the '``atomicrmw``' instruction: an
5866 operation to apply, an address whose value to modify, an argument to the
5867 operation. The operation must be one of the following keywords:
5881 The type of '<value>' must be an integer type whose bit width is a power
5882 of two greater than or equal to eight and less than or equal to a
5883 target-specific size limit. The type of the '``<pointer>``' operand must
5884 be a pointer to that type. If the ``atomicrmw`` is marked as
5885 ``volatile``, then the optimizer is not allowed to modify the number or
5886 order of execution of this ``atomicrmw`` with other :ref:`volatile
5887 operations <volatile>`.
5892 The contents of memory at the location specified by the '``<pointer>``'
5893 operand are atomically read, modified, and written back. The original
5894 value at the location is returned. The modification is specified by the
5897 - xchg: ``*ptr = val``
5898 - add: ``*ptr = *ptr + val``
5899 - sub: ``*ptr = *ptr - val``
5900 - and: ``*ptr = *ptr & val``
5901 - nand: ``*ptr = ~(*ptr & val)``
5902 - or: ``*ptr = *ptr | val``
5903 - xor: ``*ptr = *ptr ^ val``
5904 - max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison)
5905 - min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison)
5906 - umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned
5908 - umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned
5914 .. code-block:: llvm
5916 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields i32
5918 .. _i_getelementptr:
5920 '``getelementptr``' Instruction
5921 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5928 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
5929 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
5930 <result> = getelementptr <ptr vector> ptrval, <vector index type> idx
5935 The '``getelementptr``' instruction is used to get the address of a
5936 subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs
5937 address calculation only and does not access memory.
5942 The first argument is always a pointer or a vector of pointers, and
5943 forms the basis of the calculation. The remaining arguments are indices
5944 that indicate which of the elements of the aggregate object are indexed.
5945 The interpretation of each index is dependent on the type being indexed
5946 into. The first index always indexes the pointer value given as the
5947 first argument, the second index indexes a value of the type pointed to
5948 (not necessarily the value directly pointed to, since the first index
5949 can be non-zero), etc. The first type indexed into must be a pointer
5950 value, subsequent types can be arrays, vectors, and structs. Note that
5951 subsequent types being indexed into can never be pointers, since that
5952 would require loading the pointer before continuing calculation.
5954 The type of each index argument depends on the type it is indexing into.
5955 When indexing into a (optionally packed) structure, only ``i32`` integer
5956 **constants** are allowed (when using a vector of indices they must all
5957 be the **same** ``i32`` integer constant). When indexing into an array,
5958 pointer or vector, integers of any width are allowed, and they are not
5959 required to be constant. These integers are treated as signed values
5962 For example, let's consider a C code fragment and how it gets compiled
5978 int *foo(struct ST *s) {
5979 return &s[1].Z.B[5][13];
5982 The LLVM code generated by Clang is:
5984 .. code-block:: llvm
5986 %struct.RT = type { i8, [10 x [20 x i32]], i8 }
5987 %struct.ST = type { i32, double, %struct.RT }
5989 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
5991 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5998 In the example above, the first index is indexing into the
5999 '``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``'
6000 = '``{ i32, double, %struct.RT }``' type, a structure. The second index
6001 indexes into the third element of the structure, yielding a
6002 '``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another
6003 structure. The third index indexes into the second element of the
6004 structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two
6005 dimensions of the array are subscripted into, yielding an '``i32``'
6006 type. The '``getelementptr``' instruction returns a pointer to this
6007 element, thus computing a value of '``i32*``' type.
6009 Note that it is perfectly legal to index partially through a structure,
6010 returning a pointer to an inner element. Because of this, the LLVM code
6011 for the given testcase is equivalent to:
6013 .. code-block:: llvm
6015 define i32* @foo(%struct.ST* %s) {
6016 %t1 = getelementptr %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1
6017 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2
6018 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3
6019 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4
6020 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5
6024 If the ``inbounds`` keyword is present, the result value of the
6025 ``getelementptr`` is a :ref:`poison value <poisonvalues>` if the base
6026 pointer is not an *in bounds* address of an allocated object, or if any
6027 of the addresses that would be formed by successive addition of the
6028 offsets implied by the indices to the base address with infinitely
6029 precise signed arithmetic are not an *in bounds* address of that
6030 allocated object. The *in bounds* addresses for an allocated object are
6031 all the addresses that point into the object, plus the address one byte
6032 past the end. In cases where the base is a vector of pointers the
6033 ``inbounds`` keyword applies to each of the computations element-wise.
6035 If the ``inbounds`` keyword is not present, the offsets are added to the
6036 base address with silently-wrapping two's complement arithmetic. If the
6037 offsets have a different width from the pointer, they are sign-extended
6038 or truncated to the width of the pointer. The result value of the
6039 ``getelementptr`` may be outside the object pointed to by the base
6040 pointer. The result value may not necessarily be used to access memory
6041 though, even if it happens to point into allocated storage. See the
6042 :ref:`Pointer Aliasing Rules <pointeraliasing>` section for more
6045 The getelementptr instruction is often confusing. For some more insight
6046 into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`.
6051 .. code-block:: llvm
6053 ; yields [12 x i8]*:aptr
6054 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
6056 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
6058 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
6060 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
6062 In cases where the pointer argument is a vector of pointers, each index
6063 must be a vector with the same number of elements. For example:
6065 .. code-block:: llvm
6067 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
6069 Conversion Operations
6070 ---------------------
6072 The instructions in this category are the conversion instructions
6073 (casting) which all take a single operand and a type. They perform
6074 various bit conversions on the operand.
6076 '``trunc .. to``' Instruction
6077 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6084 <result> = trunc <ty> <value> to <ty2> ; yields ty2
6089 The '``trunc``' instruction truncates its operand to the type ``ty2``.
6094 The '``trunc``' instruction takes a value to trunc, and a type to trunc
6095 it to. Both types must be of :ref:`integer <t_integer>` types, or vectors
6096 of the same number of integers. The bit size of the ``value`` must be
6097 larger than the bit size of the destination type, ``ty2``. Equal sized
6098 types are not allowed.
6103 The '``trunc``' instruction truncates the high order bits in ``value``
6104 and converts the remaining bits to ``ty2``. Since the source size must
6105 be larger than the destination size, ``trunc`` cannot be a *no-op cast*.
6106 It will always truncate bits.
6111 .. code-block:: llvm
6113 %X = trunc i32 257 to i8 ; yields i8:1
6114 %Y = trunc i32 123 to i1 ; yields i1:true
6115 %Z = trunc i32 122 to i1 ; yields i1:false
6116 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7>
6118 '``zext .. to``' Instruction
6119 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6126 <result> = zext <ty> <value> to <ty2> ; yields ty2
6131 The '``zext``' instruction zero extends its operand to type ``ty2``.
6136 The '``zext``' instruction takes a value to cast, and a type to cast it
6137 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
6138 the same number of integers. The bit size of the ``value`` must be
6139 smaller than the bit size of the destination type, ``ty2``.
6144 The ``zext`` fills the high order bits of the ``value`` with zero bits
6145 until it reaches the size of the destination type, ``ty2``.
6147 When zero extending from i1, the result will always be either 0 or 1.
6152 .. code-block:: llvm
6154 %X = zext i32 257 to i64 ; yields i64:257
6155 %Y = zext i1 true to i32 ; yields i32:1
6156 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
6158 '``sext .. to``' Instruction
6159 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6166 <result> = sext <ty> <value> to <ty2> ; yields ty2
6171 The '``sext``' sign extends ``value`` to the type ``ty2``.
6176 The '``sext``' instruction takes a value to cast, and a type to cast it
6177 to. Both types must be of :ref:`integer <t_integer>` types, or vectors of
6178 the same number of integers. The bit size of the ``value`` must be
6179 smaller than the bit size of the destination type, ``ty2``.
6184 The '``sext``' instruction performs a sign extension by copying the sign
6185 bit (highest order bit) of the ``value`` until it reaches the bit size
6186 of the type ``ty2``.
6188 When sign extending from i1, the extension always results in -1 or 0.
6193 .. code-block:: llvm
6195 %X = sext i8 -1 to i16 ; yields i16 :65535
6196 %Y = sext i1 true to i32 ; yields i32:-1
6197 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7>
6199 '``fptrunc .. to``' Instruction
6200 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6207 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2
6212 The '``fptrunc``' instruction truncates ``value`` to type ``ty2``.
6217 The '``fptrunc``' instruction takes a :ref:`floating point <t_floating>`
6218 value to cast and a :ref:`floating point <t_floating>` type to cast it to.
6219 The size of ``value`` must be larger than the size of ``ty2``. This
6220 implies that ``fptrunc`` cannot be used to make a *no-op cast*.
6225 The '``fptrunc``' instruction truncates a ``value`` from a larger
6226 :ref:`floating point <t_floating>` type to a smaller :ref:`floating
6227 point <t_floating>` type. If the value cannot fit within the
6228 destination type, ``ty2``, then the results are undefined.
6233 .. code-block:: llvm
6235 %X = fptrunc double 123.0 to float ; yields float:123.0
6236 %Y = fptrunc double 1.0E+300 to float ; yields undefined
6238 '``fpext .. to``' Instruction
6239 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6246 <result> = fpext <ty> <value> to <ty2> ; yields ty2
6251 The '``fpext``' extends a floating point ``value`` to a larger floating
6257 The '``fpext``' instruction takes a :ref:`floating point <t_floating>`
6258 ``value`` to cast, and a :ref:`floating point <t_floating>` type to cast it
6259 to. The source type must be smaller than the destination type.
6264 The '``fpext``' instruction extends the ``value`` from a smaller
6265 :ref:`floating point <t_floating>` type to a larger :ref:`floating
6266 point <t_floating>` type. The ``fpext`` cannot be used to make a
6267 *no-op cast* because it always changes bits. Use ``bitcast`` to make a
6268 *no-op cast* for a floating point cast.
6273 .. code-block:: llvm
6275 %X = fpext float 3.125 to double ; yields double:3.125000e+00
6276 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000
6278 '``fptoui .. to``' Instruction
6279 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6286 <result> = fptoui <ty> <value> to <ty2> ; yields ty2
6291 The '``fptoui``' converts a floating point ``value`` to its unsigned
6292 integer equivalent of type ``ty2``.
6297 The '``fptoui``' instruction takes a value to cast, which must be a
6298 scalar or vector :ref:`floating point <t_floating>` value, and a type to
6299 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
6300 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
6301 type with the same number of elements as ``ty``
6306 The '``fptoui``' instruction converts its :ref:`floating
6307 point <t_floating>` operand into the nearest (rounding towards zero)
6308 unsigned integer value. If the value cannot fit in ``ty2``, the results
6314 .. code-block:: llvm
6316 %X = fptoui double 123.0 to i32 ; yields i32:123
6317 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1
6318 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1
6320 '``fptosi .. to``' Instruction
6321 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6328 <result> = fptosi <ty> <value> to <ty2> ; yields ty2
6333 The '``fptosi``' instruction converts :ref:`floating point <t_floating>`
6334 ``value`` to type ``ty2``.
6339 The '``fptosi``' instruction takes a value to cast, which must be a
6340 scalar or vector :ref:`floating point <t_floating>` value, and a type to
6341 cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If
6342 ``ty`` is a vector floating point type, ``ty2`` must be a vector integer
6343 type with the same number of elements as ``ty``
6348 The '``fptosi``' instruction converts its :ref:`floating
6349 point <t_floating>` operand into the nearest (rounding towards zero)
6350 signed integer value. If the value cannot fit in ``ty2``, the results
6356 .. code-block:: llvm
6358 %X = fptosi double -123.0 to i32 ; yields i32:-123
6359 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1
6360 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1
6362 '``uitofp .. to``' Instruction
6363 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6370 <result> = uitofp <ty> <value> to <ty2> ; yields ty2
6375 The '``uitofp``' instruction regards ``value`` as an unsigned integer
6376 and converts that value to the ``ty2`` type.
6381 The '``uitofp``' instruction takes a value to cast, which must be a
6382 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
6383 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
6384 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
6385 type with the same number of elements as ``ty``
6390 The '``uitofp``' instruction interprets its operand as an unsigned
6391 integer quantity and converts it to the corresponding floating point
6392 value. If the value cannot fit in the floating point value, the results
6398 .. code-block:: llvm
6400 %X = uitofp i32 257 to float ; yields float:257.0
6401 %Y = uitofp i8 -1 to double ; yields double:255.0
6403 '``sitofp .. to``' Instruction
6404 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6411 <result> = sitofp <ty> <value> to <ty2> ; yields ty2
6416 The '``sitofp``' instruction regards ``value`` as a signed integer and
6417 converts that value to the ``ty2`` type.
6422 The '``sitofp``' instruction takes a value to cast, which must be a
6423 scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to
6424 ``ty2``, which must be an :ref:`floating point <t_floating>` type. If
6425 ``ty`` is a vector integer type, ``ty2`` must be a vector floating point
6426 type with the same number of elements as ``ty``
6431 The '``sitofp``' instruction interprets its operand as a signed integer
6432 quantity and converts it to the corresponding floating point value. If
6433 the value cannot fit in the floating point value, the results are
6439 .. code-block:: llvm
6441 %X = sitofp i32 257 to float ; yields float:257.0
6442 %Y = sitofp i8 -1 to double ; yields double:-1.0
6446 '``ptrtoint .. to``' Instruction
6447 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6454 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2
6459 The '``ptrtoint``' instruction converts the pointer or a vector of
6460 pointers ``value`` to the integer (or vector of integers) type ``ty2``.
6465 The '``ptrtoint``' instruction takes a ``value`` to cast, which must be
6466 a a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a
6467 type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or
6468 a vector of integers type.
6473 The '``ptrtoint``' instruction converts ``value`` to integer type
6474 ``ty2`` by interpreting the pointer value as an integer and either
6475 truncating or zero extending that value to the size of the integer type.
6476 If ``value`` is smaller than ``ty2`` then a zero extension is done. If
6477 ``value`` is larger than ``ty2`` then a truncation is done. If they are
6478 the same size, then nothing is done (*no-op cast*) other than a type
6484 .. code-block:: llvm
6486 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture
6487 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture
6488 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture
6492 '``inttoptr .. to``' Instruction
6493 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6500 <result> = inttoptr <ty> <value> to <ty2> ; yields ty2
6505 The '``inttoptr``' instruction converts an integer ``value`` to a
6506 pointer type, ``ty2``.
6511 The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to
6512 cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>`
6518 The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by
6519 applying either a zero extension or a truncation depending on the size
6520 of the integer ``value``. If ``value`` is larger than the size of a
6521 pointer then a truncation is done. If ``value`` is smaller than the size
6522 of a pointer then a zero extension is done. If they are the same size,
6523 nothing is done (*no-op cast*).
6528 .. code-block:: llvm
6530 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture
6531 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture
6532 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture
6533 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers
6537 '``bitcast .. to``' Instruction
6538 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6545 <result> = bitcast <ty> <value> to <ty2> ; yields ty2
6550 The '``bitcast``' instruction converts ``value`` to type ``ty2`` without
6556 The '``bitcast``' instruction takes a value to cast, which must be a
6557 non-aggregate first class value, and a type to cast it to, which must
6558 also be a non-aggregate :ref:`first class <t_firstclass>` type. The
6559 bit sizes of ``value`` and the destination type, ``ty2``, must be
6560 identical. If the source type is a pointer, the destination type must
6561 also be a pointer of the same size. This instruction supports bitwise
6562 conversion of vectors to integers and to vectors of other types (as
6563 long as they have the same size).
6568 The '``bitcast``' instruction converts ``value`` to type ``ty2``. It
6569 is always a *no-op cast* because no bits change with this
6570 conversion. The conversion is done as if the ``value`` had been stored
6571 to memory and read back as type ``ty2``. Pointer (or vector of
6572 pointers) types may only be converted to other pointer (or vector of
6573 pointers) types with the same address space through this instruction.
6574 To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>`
6575 or :ref:`ptrtoint <i_ptrtoint>` instructions first.
6580 .. code-block:: llvm
6582 %X = bitcast i8 255 to i8 ; yields i8 :-1
6583 %Y = bitcast i32* %x to sint* ; yields sint*:%x
6584 %Z = bitcast <2 x int> %V to i64; ; yields i64: %V
6585 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*>
6587 .. _i_addrspacecast:
6589 '``addrspacecast .. to``' Instruction
6590 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6597 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2
6602 The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in
6603 address space ``n`` to type ``pty2`` in address space ``m``.
6608 The '``addrspacecast``' instruction takes a pointer or vector of pointer value
6609 to cast and a pointer type to cast it to, which must have a different
6615 The '``addrspacecast``' instruction converts the pointer value
6616 ``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex
6617 value modification, depending on the target and the address space
6618 pair. Pointer conversions within the same address space must be
6619 performed with the ``bitcast`` instruction. Note that if the address space
6620 conversion is legal then both result and operand refer to the same memory
6626 .. code-block:: llvm
6628 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x
6629 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y
6630 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z
6637 The instructions in this category are the "miscellaneous" instructions,
6638 which defy better classification.
6642 '``icmp``' Instruction
6643 ^^^^^^^^^^^^^^^^^^^^^^
6650 <result> = icmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
6655 The '``icmp``' instruction returns a boolean value or a vector of
6656 boolean values based on comparison of its two integer, integer vector,
6657 pointer, or pointer vector operands.
6662 The '``icmp``' instruction takes three operands. The first operand is
6663 the condition code indicating the kind of comparison to perform. It is
6664 not a value, just a keyword. The possible condition code are:
6667 #. ``ne``: not equal
6668 #. ``ugt``: unsigned greater than
6669 #. ``uge``: unsigned greater or equal
6670 #. ``ult``: unsigned less than
6671 #. ``ule``: unsigned less or equal
6672 #. ``sgt``: signed greater than
6673 #. ``sge``: signed greater or equal
6674 #. ``slt``: signed less than
6675 #. ``sle``: signed less or equal
6677 The remaining two arguments must be :ref:`integer <t_integer>` or
6678 :ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They
6679 must also be identical types.
6684 The '``icmp``' compares ``op1`` and ``op2`` according to the condition
6685 code given as ``cond``. The comparison performed always yields either an
6686 :ref:`i1 <t_integer>` or vector of ``i1`` result, as follows:
6688 #. ``eq``: yields ``true`` if the operands are equal, ``false``
6689 otherwise. No sign interpretation is necessary or performed.
6690 #. ``ne``: yields ``true`` if the operands are unequal, ``false``
6691 otherwise. No sign interpretation is necessary or performed.
6692 #. ``ugt``: interprets the operands as unsigned values and yields
6693 ``true`` if ``op1`` is greater than ``op2``.
6694 #. ``uge``: interprets the operands as unsigned values and yields
6695 ``true`` if ``op1`` is greater than or equal to ``op2``.
6696 #. ``ult``: interprets the operands as unsigned values and yields
6697 ``true`` if ``op1`` is less than ``op2``.
6698 #. ``ule``: interprets the operands as unsigned values and yields
6699 ``true`` if ``op1`` is less than or equal to ``op2``.
6700 #. ``sgt``: interprets the operands as signed values and yields ``true``
6701 if ``op1`` is greater than ``op2``.
6702 #. ``sge``: interprets the operands as signed values and yields ``true``
6703 if ``op1`` is greater than or equal to ``op2``.
6704 #. ``slt``: interprets the operands as signed values and yields ``true``
6705 if ``op1`` is less than ``op2``.
6706 #. ``sle``: interprets the operands as signed values and yields ``true``
6707 if ``op1`` is less than or equal to ``op2``.
6709 If the operands are :ref:`pointer <t_pointer>` typed, the pointer values
6710 are compared as if they were integers.
6712 If the operands are integer vectors, then they are compared element by
6713 element. The result is an ``i1`` vector with the same number of elements
6714 as the values being compared. Otherwise, the result is an ``i1``.
6719 .. code-block:: llvm
6721 <result> = icmp eq i32 4, 5 ; yields: result=false
6722 <result> = icmp ne float* %X, %X ; yields: result=false
6723 <result> = icmp ult i16 4, 5 ; yields: result=true
6724 <result> = icmp sgt i16 4, 5 ; yields: result=false
6725 <result> = icmp ule i16 -4, 5 ; yields: result=false
6726 <result> = icmp sge i16 4, 5 ; yields: result=false
6728 Note that the code generator does not yet support vector types with the
6729 ``icmp`` instruction.
6733 '``fcmp``' Instruction
6734 ^^^^^^^^^^^^^^^^^^^^^^
6741 <result> = fcmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result
6746 The '``fcmp``' instruction returns a boolean value or vector of boolean
6747 values based on comparison of its operands.
6749 If the operands are floating point scalars, then the result type is a
6750 boolean (:ref:`i1 <t_integer>`).
6752 If the operands are floating point vectors, then the result type is a
6753 vector of boolean with the same number of elements as the operands being
6759 The '``fcmp``' instruction takes three operands. The first operand is
6760 the condition code indicating the kind of comparison to perform. It is
6761 not a value, just a keyword. The possible condition code are:
6763 #. ``false``: no comparison, always returns false
6764 #. ``oeq``: ordered and equal
6765 #. ``ogt``: ordered and greater than
6766 #. ``oge``: ordered and greater than or equal
6767 #. ``olt``: ordered and less than
6768 #. ``ole``: ordered and less than or equal
6769 #. ``one``: ordered and not equal
6770 #. ``ord``: ordered (no nans)
6771 #. ``ueq``: unordered or equal
6772 #. ``ugt``: unordered or greater than
6773 #. ``uge``: unordered or greater than or equal
6774 #. ``ult``: unordered or less than
6775 #. ``ule``: unordered or less than or equal
6776 #. ``une``: unordered or not equal
6777 #. ``uno``: unordered (either nans)
6778 #. ``true``: no comparison, always returns true
6780 *Ordered* means that neither operand is a QNAN while *unordered* means
6781 that either operand may be a QNAN.
6783 Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating
6784 point <t_floating>` type or a :ref:`vector <t_vector>` of floating point
6785 type. They must have identical types.
6790 The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the
6791 condition code given as ``cond``. If the operands are vectors, then the
6792 vectors are compared element by element. Each comparison performed
6793 always yields an :ref:`i1 <t_integer>` result, as follows:
6795 #. ``false``: always yields ``false``, regardless of operands.
6796 #. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1``
6797 is equal to ``op2``.
6798 #. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1``
6799 is greater than ``op2``.
6800 #. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1``
6801 is greater than or equal to ``op2``.
6802 #. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1``
6803 is less than ``op2``.
6804 #. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1``
6805 is less than or equal to ``op2``.
6806 #. ``one``: yields ``true`` if both operands are not a QNAN and ``op1``
6807 is not equal to ``op2``.
6808 #. ``ord``: yields ``true`` if both operands are not a QNAN.
6809 #. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is
6811 #. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is
6812 greater than ``op2``.
6813 #. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is
6814 greater than or equal to ``op2``.
6815 #. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is
6817 #. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is
6818 less than or equal to ``op2``.
6819 #. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is
6820 not equal to ``op2``.
6821 #. ``uno``: yields ``true`` if either operand is a QNAN.
6822 #. ``true``: always yields ``true``, regardless of operands.
6827 .. code-block:: llvm
6829 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false
6830 <result> = fcmp one float 4.0, 5.0 ; yields: result=true
6831 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true
6832 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false
6834 Note that the code generator does not yet support vector types with the
6835 ``fcmp`` instruction.
6839 '``phi``' Instruction
6840 ^^^^^^^^^^^^^^^^^^^^^
6847 <result> = phi <ty> [ <val0>, <label0>], ...
6852 The '``phi``' instruction is used to implement the φ node in the SSA
6853 graph representing the function.
6858 The type of the incoming values is specified with the first type field.
6859 After this, the '``phi``' instruction takes a list of pairs as
6860 arguments, with one pair for each predecessor basic block of the current
6861 block. Only values of :ref:`first class <t_firstclass>` type may be used as
6862 the value arguments to the PHI node. Only labels may be used as the
6865 There must be no non-phi instructions between the start of a basic block
6866 and the PHI instructions: i.e. PHI instructions must be first in a basic
6869 For the purposes of the SSA form, the use of each incoming value is
6870 deemed to occur on the edge from the corresponding predecessor block to
6871 the current block (but after any definition of an '``invoke``'
6872 instruction's return value on the same edge).
6877 At runtime, the '``phi``' instruction logically takes on the value
6878 specified by the pair corresponding to the predecessor basic block that
6879 executed just prior to the current block.
6884 .. code-block:: llvm
6886 Loop: ; Infinite loop that counts from 0 on up...
6887 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6888 %nextindvar = add i32 %indvar, 1
6893 '``select``' Instruction
6894 ^^^^^^^^^^^^^^^^^^^^^^^^
6901 <result> = select selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty
6903 selty is either i1 or {<N x i1>}
6908 The '``select``' instruction is used to choose one value based on a
6909 condition, without IR-level branching.
6914 The '``select``' instruction requires an 'i1' value or a vector of 'i1'
6915 values indicating the condition, and two values of the same :ref:`first
6916 class <t_firstclass>` type. If the val1/val2 are vectors and the
6917 condition is a scalar, then entire vectors are selected, not individual
6923 If the condition is an i1 and it evaluates to 1, the instruction returns
6924 the first value argument; otherwise, it returns the second value
6927 If the condition is a vector of i1, then the value arguments must be
6928 vectors of the same size, and the selection is done element by element.
6933 .. code-block:: llvm
6935 %X = select i1 true, i8 17, i8 42 ; yields i8:17
6939 '``call``' Instruction
6940 ^^^^^^^^^^^^^^^^^^^^^^
6947 <result> = [tail | musttail] call [cconv] [ret attrs] <ty> [<fnty>*] <fnptrval>(<function args>) [fn attrs]
6952 The '``call``' instruction represents a simple function call.
6957 This instruction requires several arguments:
6959 #. The optional ``tail`` and ``musttail`` markers indicate that the optimizers
6960 should perform tail call optimization. The ``tail`` marker is a hint that
6961 `can be ignored <CodeGenerator.html#sibcallopt>`_. The ``musttail`` marker
6962 means that the call must be tail call optimized in order for the program to
6963 be correct. The ``musttail`` marker provides these guarantees:
6965 #. The call will not cause unbounded stack growth if it is part of a
6966 recursive cycle in the call graph.
6967 #. Arguments with the :ref:`inalloca <attr_inalloca>` attribute are
6970 Both markers imply that the callee does not access allocas or varargs from
6971 the caller. Calls marked ``musttail`` must obey the following additional
6974 - The call must immediately precede a :ref:`ret <i_ret>` instruction,
6975 or a pointer bitcast followed by a ret instruction.
6976 - The ret instruction must return the (possibly bitcasted) value
6977 produced by the call or void.
6978 - The caller and callee prototypes must match. Pointer types of
6979 parameters or return types may differ in pointee type, but not
6981 - The calling conventions of the caller and callee must match.
6982 - All ABI-impacting function attributes, such as sret, byval, inreg,
6983 returned, and inalloca, must match.
6984 - The callee must be varargs iff the caller is varargs. Bitcasting a
6985 non-varargs function to the appropriate varargs type is legal so
6986 long as the non-varargs prefixes obey the other rules.
6988 Tail call optimization for calls marked ``tail`` is guaranteed to occur if
6989 the following conditions are met:
6991 - Caller and callee both have the calling convention ``fastcc``.
6992 - The call is in tail position (ret immediately follows call and ret
6993 uses value of call or is void).
6994 - Option ``-tailcallopt`` is enabled, or
6995 ``llvm::GuaranteedTailCallOpt`` is ``true``.
6996 - `Platform-specific constraints are
6997 met. <CodeGenerator.html#tailcallopt>`_
6999 #. The optional "cconv" marker indicates which :ref:`calling
7000 convention <callingconv>` the call should use. If none is
7001 specified, the call defaults to using C calling conventions. The
7002 calling convention of the call must match the calling convention of
7003 the target function, or else the behavior is undefined.
7004 #. The optional :ref:`Parameter Attributes <paramattrs>` list for return
7005 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes
7007 #. '``ty``': the type of the call instruction itself which is also the
7008 type of the return value. Functions that return no value are marked
7010 #. '``fnty``': shall be the signature of the pointer to function value
7011 being invoked. The argument types must match the types implied by
7012 this signature. This type can be omitted if the function is not
7013 varargs and if the function type does not return a pointer to a
7015 #. '``fnptrval``': An LLVM value containing a pointer to a function to
7016 be invoked. In most cases, this is a direct function invocation, but
7017 indirect ``call``'s are just as possible, calling an arbitrary pointer
7019 #. '``function args``': argument list whose types match the function
7020 signature argument types and parameter attributes. All arguments must
7021 be of :ref:`first class <t_firstclass>` type. If the function signature
7022 indicates the function accepts a variable number of arguments, the
7023 extra arguments can be specified.
7024 #. The optional :ref:`function attributes <fnattrs>` list. Only
7025 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``'
7026 attributes are valid here.
7031 The '``call``' instruction is used to cause control flow to transfer to
7032 a specified function, with its incoming arguments bound to the specified
7033 values. Upon a '``ret``' instruction in the called function, control
7034 flow continues with the instruction after the function call, and the
7035 return value of the function is bound to the result argument.
7040 .. code-block:: llvm
7042 %retval = call i32 @test(i32 %argc)
7043 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32
7044 %X = tail call i32 @foo() ; yields i32
7045 %Y = tail call fastcc i32 @foo() ; yields i32
7046 call void %foo(i8 97 signext)
7048 %struct.A = type { i32, i8 }
7049 %r = call %struct.A @foo() ; yields { i32, i8 }
7050 %gr = extractvalue %struct.A %r, 0 ; yields i32
7051 %gr1 = extractvalue %struct.A %r, 1 ; yields i8
7052 %Z = call void @foo() noreturn ; indicates that %foo never returns normally
7053 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended
7055 llvm treats calls to some functions with names and arguments that match
7056 the standard C99 library as being the C99 library functions, and may
7057 perform optimizations or generate code for them under that assumption.
7058 This is something we'd like to change in the future to provide better
7059 support for freestanding environments and non-C-based languages.
7063 '``va_arg``' Instruction
7064 ^^^^^^^^^^^^^^^^^^^^^^^^
7071 <resultval> = va_arg <va_list*> <arglist>, <argty>
7076 The '``va_arg``' instruction is used to access arguments passed through
7077 the "variable argument" area of a function call. It is used to implement
7078 the ``va_arg`` macro in C.
7083 This instruction takes a ``va_list*`` value and the type of the
7084 argument. It returns a value of the specified argument type and
7085 increments the ``va_list`` to point to the next argument. The actual
7086 type of ``va_list`` is target specific.
7091 The '``va_arg``' instruction loads an argument of the specified type
7092 from the specified ``va_list`` and causes the ``va_list`` to point to
7093 the next argument. For more information, see the variable argument
7094 handling :ref:`Intrinsic Functions <int_varargs>`.
7096 It is legal for this instruction to be called in a function which does
7097 not take a variable number of arguments, for example, the ``vfprintf``
7100 ``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic
7101 function <intrinsics>` because it takes a type as an argument.
7106 See the :ref:`variable argument processing <int_varargs>` section.
7108 Note that the code generator does not yet fully support va\_arg on many
7109 targets. Also, it does not currently support va\_arg with aggregate
7110 types on any target.
7114 '``landingpad``' Instruction
7115 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7122 <resultval> = landingpad <resultty> personality <type> <pers_fn> <clause>+
7123 <resultval> = landingpad <resultty> personality <type> <pers_fn> cleanup <clause>*
7125 <clause> := catch <type> <value>
7126 <clause> := filter <array constant type> <array constant>
7131 The '``landingpad``' instruction is used by `LLVM's exception handling
7132 system <ExceptionHandling.html#overview>`_ to specify that a basic block
7133 is a landing pad --- one where the exception lands, and corresponds to the
7134 code found in the ``catch`` portion of a ``try``/``catch`` sequence. It
7135 defines values supplied by the personality function (``pers_fn``) upon
7136 re-entry to the function. The ``resultval`` has the type ``resultty``.
7141 This instruction takes a ``pers_fn`` value. This is the personality
7142 function associated with the unwinding mechanism. The optional
7143 ``cleanup`` flag indicates that the landing pad block is a cleanup.
7145 A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and
7146 contains the global variable representing the "type" that may be caught
7147 or filtered respectively. Unlike the ``catch`` clause, the ``filter``
7148 clause takes an array constant as its argument. Use
7149 "``[0 x i8**] undef``" for a filter which cannot throw. The
7150 '``landingpad``' instruction must contain *at least* one ``clause`` or
7151 the ``cleanup`` flag.
7156 The '``landingpad``' instruction defines the values which are set by the
7157 personality function (``pers_fn``) upon re-entry to the function, and
7158 therefore the "result type" of the ``landingpad`` instruction. As with
7159 calling conventions, how the personality function results are
7160 represented in LLVM IR is target specific.
7162 The clauses are applied in order from top to bottom. If two
7163 ``landingpad`` instructions are merged together through inlining, the
7164 clauses from the calling function are appended to the list of clauses.
7165 When the call stack is being unwound due to an exception being thrown,
7166 the exception is compared against each ``clause`` in turn. If it doesn't
7167 match any of the clauses, and the ``cleanup`` flag is not set, then
7168 unwinding continues further up the call stack.
7170 The ``landingpad`` instruction has several restrictions:
7172 - A landing pad block is a basic block which is the unwind destination
7173 of an '``invoke``' instruction.
7174 - A landing pad block must have a '``landingpad``' instruction as its
7175 first non-PHI instruction.
7176 - There can be only one '``landingpad``' instruction within the landing
7178 - A basic block that is not a landing pad block may not include a
7179 '``landingpad``' instruction.
7180 - All '``landingpad``' instructions in a function must have the same
7181 personality function.
7186 .. code-block:: llvm
7188 ;; A landing pad which can catch an integer.
7189 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
7191 ;; A landing pad that is a cleanup.
7192 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
7194 ;; A landing pad which can catch an integer and can only throw a double.
7195 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
7197 filter [1 x i8**] [@_ZTId]
7204 LLVM supports the notion of an "intrinsic function". These functions
7205 have well known names and semantics and are required to follow certain
7206 restrictions. Overall, these intrinsics represent an extension mechanism
7207 for the LLVM language that does not require changing all of the
7208 transformations in LLVM when adding to the language (or the bitcode
7209 reader/writer, the parser, etc...).
7211 Intrinsic function names must all start with an "``llvm.``" prefix. This
7212 prefix is reserved in LLVM for intrinsic names; thus, function names may
7213 not begin with this prefix. Intrinsic functions must always be external
7214 functions: you cannot define the body of intrinsic functions. Intrinsic
7215 functions may only be used in call or invoke instructions: it is illegal
7216 to take the address of an intrinsic function. Additionally, because
7217 intrinsic functions are part of the LLVM language, it is required if any
7218 are added that they be documented here.
7220 Some intrinsic functions can be overloaded, i.e., the intrinsic
7221 represents a family of functions that perform the same operation but on
7222 different data types. Because LLVM can represent over 8 million
7223 different integer types, overloading is used commonly to allow an
7224 intrinsic function to operate on any integer type. One or more of the
7225 argument types or the result type can be overloaded to accept any
7226 integer type. Argument types may also be defined as exactly matching a
7227 previous argument's type or the result type. This allows an intrinsic
7228 function which accepts multiple arguments, but needs all of them to be
7229 of the same type, to only be overloaded with respect to a single
7230 argument or the result.
7232 Overloaded intrinsics will have the names of its overloaded argument
7233 types encoded into its function name, each preceded by a period. Only
7234 those types which are overloaded result in a name suffix. Arguments
7235 whose type is matched against another type do not. For example, the
7236 ``llvm.ctpop`` function can take an integer of any width and returns an
7237 integer of exactly the same integer width. This leads to a family of
7238 functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and
7239 ``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is
7240 overloaded, and only one type suffix is required. Because the argument's
7241 type is matched against the return type, it does not require its own
7244 To learn how to add an intrinsic function, please see the `Extending
7245 LLVM Guide <ExtendingLLVM.html>`_.
7249 Variable Argument Handling Intrinsics
7250 -------------------------------------
7252 Variable argument support is defined in LLVM with the
7253 :ref:`va_arg <i_va_arg>` instruction and these three intrinsic
7254 functions. These functions are related to the similarly named macros
7255 defined in the ``<stdarg.h>`` header file.
7257 All of these functions operate on arguments that use a target-specific
7258 value type "``va_list``". The LLVM assembly language reference manual
7259 does not define what this type is, so all transformations should be
7260 prepared to handle these functions regardless of the type used.
7262 This example shows how the :ref:`va_arg <i_va_arg>` instruction and the
7263 variable argument handling intrinsic functions are used.
7265 .. code-block:: llvm
7267 ; This struct is different for every platform. For most platforms,
7268 ; it is merely an i8*.
7269 %struct.va_list = type { i8* }
7271 ; For Unix x86_64 platforms, va_list is the following struct:
7272 ; %struct.va_list = type { i32, i32, i8*, i8* }
7274 define i32 @test(i32 %X, ...) {
7275 ; Initialize variable argument processing
7276 %ap = alloca %struct.va_list
7277 %ap2 = bitcast %struct.va_list* %ap to i8*
7278 call void @llvm.va_start(i8* %ap2)
7280 ; Read a single integer argument
7281 %tmp = va_arg i8* %ap2, i32
7283 ; Demonstrate usage of llvm.va_copy and llvm.va_end
7285 %aq2 = bitcast i8** %aq to i8*
7286 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
7287 call void @llvm.va_end(i8* %aq2)
7289 ; Stop processing of arguments.
7290 call void @llvm.va_end(i8* %ap2)
7294 declare void @llvm.va_start(i8*)
7295 declare void @llvm.va_copy(i8*, i8*)
7296 declare void @llvm.va_end(i8*)
7300 '``llvm.va_start``' Intrinsic
7301 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7308 declare void @llvm.va_start(i8* <arglist>)
7313 The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for
7314 subsequent use by ``va_arg``.
7319 The argument is a pointer to a ``va_list`` element to initialize.
7324 The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro
7325 available in C. In a target-dependent way, it initializes the
7326 ``va_list`` element to which the argument points, so that the next call
7327 to ``va_arg`` will produce the first variable argument passed to the
7328 function. Unlike the C ``va_start`` macro, this intrinsic does not need
7329 to know the last argument of the function as the compiler can figure
7332 '``llvm.va_end``' Intrinsic
7333 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7340 declare void @llvm.va_end(i8* <arglist>)
7345 The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been
7346 initialized previously with ``llvm.va_start`` or ``llvm.va_copy``.
7351 The argument is a pointer to a ``va_list`` to destroy.
7356 The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro
7357 available in C. In a target-dependent way, it destroys the ``va_list``
7358 element to which the argument points. Calls to
7359 :ref:`llvm.va_start <int_va_start>` and
7360 :ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to
7365 '``llvm.va_copy``' Intrinsic
7366 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7373 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
7378 The '``llvm.va_copy``' intrinsic copies the current argument position
7379 from the source argument list to the destination argument list.
7384 The first argument is a pointer to a ``va_list`` element to initialize.
7385 The second argument is a pointer to a ``va_list`` element to copy from.
7390 The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro
7391 available in C. In a target-dependent way, it copies the source
7392 ``va_list`` element into the destination ``va_list`` element. This
7393 intrinsic is necessary because the `` llvm.va_start`` intrinsic may be
7394 arbitrarily complex and require, for example, memory allocation.
7396 Accurate Garbage Collection Intrinsics
7397 --------------------------------------
7399 LLVM's support for `Accurate Garbage Collection <GarbageCollection.html>`_
7400 (GC) requires the frontend to generate code containing appropriate intrinsic
7401 calls and select an appropriate GC strategy which knows how to lower these
7402 intrinsics in a manner which is appropriate for the target collector.
7404 These intrinsics allow identification of :ref:`GC roots on the
7405 stack <int_gcroot>`, as well as garbage collector implementations that
7406 require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers.
7407 Frontends for type-safe garbage collected languages should generate
7408 these intrinsics to make use of the LLVM garbage collectors. For more
7409 details, see `Garbage Collection with LLVM <GarbageCollection.html>`_.
7411 Experimental Statepoint Intrinsics
7412 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7414 LLVM provides an second experimental set of intrinsics for describing garbage
7415 collection safepoints in compiled code. These intrinsics are an alternative
7416 to the ``llvm.gcroot`` intrinsics, but are compatible with the ones for
7417 :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers. The
7418 differences in approach are covered in the `Garbage Collection with LLVM
7419 <GarbageCollection.html>`_ documentation. The intrinsics themselves are
7420 described in :doc:`Statepoints`.
7424 '``llvm.gcroot``' Intrinsic
7425 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7432 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
7437 The '``llvm.gcroot``' intrinsic declares the existence of a GC root to
7438 the code generator, and allows some metadata to be associated with it.
7443 The first argument specifies the address of a stack object that contains
7444 the root pointer. The second pointer (which must be either a constant or
7445 a global value address) contains the meta-data to be associated with the
7451 At runtime, a call to this intrinsic stores a null pointer into the
7452 "ptrloc" location. At compile-time, the code generator generates
7453 information to allow the runtime to find the pointer at GC safe points.
7454 The '``llvm.gcroot``' intrinsic may only be used in a function which
7455 :ref:`specifies a GC algorithm <gc>`.
7459 '``llvm.gcread``' Intrinsic
7460 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7467 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
7472 The '``llvm.gcread``' intrinsic identifies reads of references from heap
7473 locations, allowing garbage collector implementations that require read
7479 The second argument is the address to read from, which should be an
7480 address allocated from the garbage collector. The first object is a
7481 pointer to the start of the referenced object, if needed by the language
7482 runtime (otherwise null).
7487 The '``llvm.gcread``' intrinsic has the same semantics as a load
7488 instruction, but may be replaced with substantially more complex code by
7489 the garbage collector runtime, as needed. The '``llvm.gcread``'
7490 intrinsic may only be used in a function which :ref:`specifies a GC
7495 '``llvm.gcwrite``' Intrinsic
7496 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7503 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
7508 The '``llvm.gcwrite``' intrinsic identifies writes of references to heap
7509 locations, allowing garbage collector implementations that require write
7510 barriers (such as generational or reference counting collectors).
7515 The first argument is the reference to store, the second is the start of
7516 the object to store it to, and the third is the address of the field of
7517 Obj to store to. If the runtime does not require a pointer to the
7518 object, Obj may be null.
7523 The '``llvm.gcwrite``' intrinsic has the same semantics as a store
7524 instruction, but may be replaced with substantially more complex code by
7525 the garbage collector runtime, as needed. The '``llvm.gcwrite``'
7526 intrinsic may only be used in a function which :ref:`specifies a GC
7529 Code Generator Intrinsics
7530 -------------------------
7532 These intrinsics are provided by LLVM to expose special features that
7533 may only be implemented with code generator support.
7535 '``llvm.returnaddress``' Intrinsic
7536 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7543 declare i8 *@llvm.returnaddress(i32 <level>)
7548 The '``llvm.returnaddress``' intrinsic attempts to compute a
7549 target-specific value indicating the return address of the current
7550 function or one of its callers.
7555 The argument to this intrinsic indicates which function to return the
7556 address for. Zero indicates the calling function, one indicates its
7557 caller, etc. The argument is **required** to be a constant integer
7563 The '``llvm.returnaddress``' intrinsic either returns a pointer
7564 indicating the return address of the specified call frame, or zero if it
7565 cannot be identified. The value returned by this intrinsic is likely to
7566 be incorrect or 0 for arguments other than zero, so it should only be
7567 used for debugging purposes.
7569 Note that calling this intrinsic does not prevent function inlining or
7570 other aggressive transformations, so the value returned may not be that
7571 of the obvious source-language caller.
7573 '``llvm.frameaddress``' Intrinsic
7574 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7581 declare i8* @llvm.frameaddress(i32 <level>)
7586 The '``llvm.frameaddress``' intrinsic attempts to return the
7587 target-specific frame pointer value for the specified stack frame.
7592 The argument to this intrinsic indicates which function to return the
7593 frame pointer for. Zero indicates the calling function, one indicates
7594 its caller, etc. The argument is **required** to be a constant integer
7600 The '``llvm.frameaddress``' intrinsic either returns a pointer
7601 indicating the frame address of the specified call frame, or zero if it
7602 cannot be identified. The value returned by this intrinsic is likely to
7603 be incorrect or 0 for arguments other than zero, so it should only be
7604 used for debugging purposes.
7606 Note that calling this intrinsic does not prevent function inlining or
7607 other aggressive transformations, so the value returned may not be that
7608 of the obvious source-language caller.
7610 '``llvm.frameallocate``' and '``llvm.framerecover``' Intrinsics
7611 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7618 declare i8* @llvm.frameallocate(i32 %size)
7619 declare i8* @llvm.framerecover(i8* %func, i8* %fp)
7624 The '``llvm.frameallocate``' intrinsic allocates stack memory at some fixed
7625 offset from the frame pointer, and the '``llvm.framerecover``'
7626 intrinsic applies that offset to a live frame pointer to recover the address of
7627 the allocation. The offset is computed during frame layout of the caller of
7628 ``llvm.frameallocate``.
7633 The ``size`` argument to '``llvm.frameallocate``' must be a constant integer
7634 indicating the amount of stack memory to allocate. As with allocas, allocating
7635 zero bytes is legal, but the result is undefined.
7637 The ``func`` argument to '``llvm.framerecover``' must be a constant
7638 bitcasted pointer to a function defined in the current module. The code
7639 generator cannot determine the frame allocation offset of functions defined in
7642 The ``fp`` argument to '``llvm.framerecover``' must be a frame
7643 pointer of a call frame that is currently live. The return value of
7644 '``llvm.frameaddress``' is one way to produce such a value, but most platforms
7645 also expose the frame pointer through stack unwinding mechanisms.
7650 These intrinsics allow a group of functions to access one stack memory
7651 allocation in an ancestor stack frame. The memory returned from
7652 '``llvm.frameallocate``' may be allocated prior to stack realignment, so the
7653 memory is only aligned to the ABI-required stack alignment. Each function may
7654 only call '``llvm.frameallocate``' one or zero times from the function entry
7655 block. The frame allocation intrinsic inhibits inlining, as any frame
7656 allocations in the inlined function frame are likely to be at a different
7657 offset from the one used by '``llvm.framerecover``' called with the
7660 .. _int_read_register:
7661 .. _int_write_register:
7663 '``llvm.read_register``' and '``llvm.write_register``' Intrinsics
7664 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7671 declare i32 @llvm.read_register.i32(metadata)
7672 declare i64 @llvm.read_register.i64(metadata)
7673 declare void @llvm.write_register.i32(metadata, i32 @value)
7674 declare void @llvm.write_register.i64(metadata, i64 @value)
7680 The '``llvm.read_register``' and '``llvm.write_register``' intrinsics
7681 provides access to the named register. The register must be valid on
7682 the architecture being compiled to. The type needs to be compatible
7683 with the register being read.
7688 The '``llvm.read_register``' intrinsic returns the current value of the
7689 register, where possible. The '``llvm.write_register``' intrinsic sets
7690 the current value of the register, where possible.
7692 This is useful to implement named register global variables that need
7693 to always be mapped to a specific register, as is common practice on
7694 bare-metal programs including OS kernels.
7696 The compiler doesn't check for register availability or use of the used
7697 register in surrounding code, including inline assembly. Because of that,
7698 allocatable registers are not supported.
7700 Warning: So far it only works with the stack pointer on selected
7701 architectures (ARM, AArch64, PowerPC and x86_64). Significant amount of
7702 work is needed to support other registers and even more so, allocatable
7707 '``llvm.stacksave``' Intrinsic
7708 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7715 declare i8* @llvm.stacksave()
7720 The '``llvm.stacksave``' intrinsic is used to remember the current state
7721 of the function stack, for use with
7722 :ref:`llvm.stackrestore <int_stackrestore>`. This is useful for
7723 implementing language features like scoped automatic variable sized
7729 This intrinsic returns a opaque pointer value that can be passed to
7730 :ref:`llvm.stackrestore <int_stackrestore>`. When an
7731 ``llvm.stackrestore`` intrinsic is executed with a value saved from
7732 ``llvm.stacksave``, it effectively restores the state of the stack to
7733 the state it was in when the ``llvm.stacksave`` intrinsic executed. In
7734 practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that
7735 were allocated after the ``llvm.stacksave`` was executed.
7737 .. _int_stackrestore:
7739 '``llvm.stackrestore``' Intrinsic
7740 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7747 declare void @llvm.stackrestore(i8* %ptr)
7752 The '``llvm.stackrestore``' intrinsic is used to restore the state of
7753 the function stack to the state it was in when the corresponding
7754 :ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is
7755 useful for implementing language features like scoped automatic variable
7756 sized arrays in C99.
7761 See the description for :ref:`llvm.stacksave <int_stacksave>`.
7763 '``llvm.prefetch``' Intrinsic
7764 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7771 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
7776 The '``llvm.prefetch``' intrinsic is a hint to the code generator to
7777 insert a prefetch instruction if supported; otherwise, it is a noop.
7778 Prefetches have no effect on the behavior of the program but can change
7779 its performance characteristics.
7784 ``address`` is the address to be prefetched, ``rw`` is the specifier
7785 determining if the fetch should be for a read (0) or write (1), and
7786 ``locality`` is a temporal locality specifier ranging from (0) - no
7787 locality, to (3) - extremely local keep in cache. The ``cache type``
7788 specifies whether the prefetch is performed on the data (1) or
7789 instruction (0) cache. The ``rw``, ``locality`` and ``cache type``
7790 arguments must be constant integers.
7795 This intrinsic does not modify the behavior of the program. In
7796 particular, prefetches cannot trap and do not produce a value. On
7797 targets that support this intrinsic, the prefetch can provide hints to
7798 the processor cache for better performance.
7800 '``llvm.pcmarker``' Intrinsic
7801 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7808 declare void @llvm.pcmarker(i32 <id>)
7813 The '``llvm.pcmarker``' intrinsic is a method to export a Program
7814 Counter (PC) in a region of code to simulators and other tools. The
7815 method is target specific, but it is expected that the marker will use
7816 exported symbols to transmit the PC of the marker. The marker makes no
7817 guarantees that it will remain with any specific instruction after
7818 optimizations. It is possible that the presence of a marker will inhibit
7819 optimizations. The intended use is to be inserted after optimizations to
7820 allow correlations of simulation runs.
7825 ``id`` is a numerical id identifying the marker.
7830 This intrinsic does not modify the behavior of the program. Backends
7831 that do not support this intrinsic may ignore it.
7833 '``llvm.readcyclecounter``' Intrinsic
7834 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7841 declare i64 @llvm.readcyclecounter()
7846 The '``llvm.readcyclecounter``' intrinsic provides access to the cycle
7847 counter register (or similar low latency, high accuracy clocks) on those
7848 targets that support it. On X86, it should map to RDTSC. On Alpha, it
7849 should map to RPCC. As the backing counters overflow quickly (on the
7850 order of 9 seconds on alpha), this should only be used for small
7856 When directly supported, reading the cycle counter should not modify any
7857 memory. Implementations are allowed to either return a application
7858 specific value or a system wide value. On backends without support, this
7859 is lowered to a constant 0.
7861 Note that runtime support may be conditional on the privilege-level code is
7862 running at and the host platform.
7864 '``llvm.clear_cache``' Intrinsic
7865 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7872 declare void @llvm.clear_cache(i8*, i8*)
7877 The '``llvm.clear_cache``' intrinsic ensures visibility of modifications
7878 in the specified range to the execution unit of the processor. On
7879 targets with non-unified instruction and data cache, the implementation
7880 flushes the instruction cache.
7885 On platforms with coherent instruction and data caches (e.g. x86), this
7886 intrinsic is a nop. On platforms with non-coherent instruction and data
7887 cache (e.g. ARM, MIPS), the intrinsic is lowered either to appropriate
7888 instructions or a system call, if cache flushing requires special
7891 The default behavior is to emit a call to ``__clear_cache`` from the run
7894 This instrinsic does *not* empty the instruction pipeline. Modifications
7895 of the current function are outside the scope of the intrinsic.
7897 '``llvm.instrprof_increment``' Intrinsic
7898 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7905 declare void @llvm.instrprof_increment(i8* <name>, i64 <hash>,
7906 i32 <num-counters>, i32 <index>)
7911 The '``llvm.instrprof_increment``' intrinsic can be emitted by a
7912 frontend for use with instrumentation based profiling. These will be
7913 lowered by the ``-instrprof`` pass to generate execution counts of a
7919 The first argument is a pointer to a global variable containing the
7920 name of the entity being instrumented. This should generally be the
7921 (mangled) function name for a set of counters.
7923 The second argument is a hash value that can be used by the consumer
7924 of the profile data to detect changes to the instrumented source, and
7925 the third is the number of counters associated with ``name``. It is an
7926 error if ``hash`` or ``num-counters`` differ between two instances of
7927 ``instrprof_increment`` that refer to the same name.
7929 The last argument refers to which of the counters for ``name`` should
7930 be incremented. It should be a value between 0 and ``num-counters``.
7935 This intrinsic represents an increment of a profiling counter. It will
7936 cause the ``-instrprof`` pass to generate the appropriate data
7937 structures and the code to increment the appropriate value, in a
7938 format that can be written out by a compiler runtime and consumed via
7939 the ``llvm-profdata`` tool.
7941 Standard C Library Intrinsics
7942 -----------------------------
7944 LLVM provides intrinsics for a few important standard C library
7945 functions. These intrinsics allow source-language front-ends to pass
7946 information about the alignment of the pointer arguments to the code
7947 generator, providing opportunity for more efficient code generation.
7951 '``llvm.memcpy``' Intrinsic
7952 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
7957 This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any
7958 integer bit width and for different address spaces. Not all targets
7959 support all bit widths however.
7963 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
7964 i32 <len>, i32 <align>, i1 <isvolatile>)
7965 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
7966 i64 <len>, i32 <align>, i1 <isvolatile>)
7971 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
7972 source location to the destination location.
7974 Note that, unlike the standard libc function, the ``llvm.memcpy.*``
7975 intrinsics do not return a value, takes extra alignment/isvolatile
7976 arguments and the pointers can be in specified address spaces.
7981 The first argument is a pointer to the destination, the second is a
7982 pointer to the source. The third argument is an integer argument
7983 specifying the number of bytes to copy, the fourth argument is the
7984 alignment of the source and destination locations, and the fifth is a
7985 boolean indicating a volatile access.
7987 If the call to this intrinsic has an alignment value that is not 0 or 1,
7988 then the caller guarantees that both the source and destination pointers
7989 are aligned to that boundary.
7991 If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is
7992 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
7993 very cleanly specified and it is unwise to depend on it.
7998 The '``llvm.memcpy.*``' intrinsics copy a block of memory from the
7999 source location to the destination location, which are not allowed to
8000 overlap. It copies "len" bytes of memory over. If the argument is known
8001 to be aligned to some boundary, this can be specified as the fourth
8002 argument, otherwise it should be set to 0 or 1 (both meaning no alignment).
8004 '``llvm.memmove``' Intrinsic
8005 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8010 This is an overloaded intrinsic. You can use llvm.memmove on any integer
8011 bit width and for different address space. Not all targets support all
8016 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
8017 i32 <len>, i32 <align>, i1 <isvolatile>)
8018 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
8019 i64 <len>, i32 <align>, i1 <isvolatile>)
8024 The '``llvm.memmove.*``' intrinsics move a block of memory from the
8025 source location to the destination location. It is similar to the
8026 '``llvm.memcpy``' intrinsic but allows the two memory locations to
8029 Note that, unlike the standard libc function, the ``llvm.memmove.*``
8030 intrinsics do not return a value, takes extra alignment/isvolatile
8031 arguments and the pointers can be in specified address spaces.
8036 The first argument is a pointer to the destination, the second is a
8037 pointer to the source. The third argument is an integer argument
8038 specifying the number of bytes to copy, the fourth argument is the
8039 alignment of the source and destination locations, and the fifth is a
8040 boolean indicating a volatile access.
8042 If the call to this intrinsic has an alignment value that is not 0 or 1,
8043 then the caller guarantees that the source and destination pointers are
8044 aligned to that boundary.
8046 If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call
8047 is a :ref:`volatile operation <volatile>`. The detailed access behavior is
8048 not very cleanly specified and it is unwise to depend on it.
8053 The '``llvm.memmove.*``' intrinsics copy a block of memory from the
8054 source location to the destination location, which may overlap. It
8055 copies "len" bytes of memory over. If the argument is known to be
8056 aligned to some boundary, this can be specified as the fourth argument,
8057 otherwise it should be set to 0 or 1 (both meaning no alignment).
8059 '``llvm.memset.*``' Intrinsics
8060 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8065 This is an overloaded intrinsic. You can use llvm.memset on any integer
8066 bit width and for different address spaces. However, not all targets
8067 support all bit widths.
8071 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
8072 i32 <len>, i32 <align>, i1 <isvolatile>)
8073 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
8074 i64 <len>, i32 <align>, i1 <isvolatile>)
8079 The '``llvm.memset.*``' intrinsics fill a block of memory with a
8080 particular byte value.
8082 Note that, unlike the standard libc function, the ``llvm.memset``
8083 intrinsic does not return a value and takes extra alignment/volatile
8084 arguments. Also, the destination can be in an arbitrary address space.
8089 The first argument is a pointer to the destination to fill, the second
8090 is the byte value with which to fill it, the third argument is an
8091 integer argument specifying the number of bytes to fill, and the fourth
8092 argument is the known alignment of the destination location.
8094 If the call to this intrinsic has an alignment value that is not 0 or 1,
8095 then the caller guarantees that the destination pointer is aligned to
8098 If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is
8099 a :ref:`volatile operation <volatile>`. The detailed access behavior is not
8100 very cleanly specified and it is unwise to depend on it.
8105 The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting
8106 at the destination location. If the argument is known to be aligned to
8107 some boundary, this can be specified as the fourth argument, otherwise
8108 it should be set to 0 or 1 (both meaning no alignment).
8110 '``llvm.sqrt.*``' Intrinsic
8111 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8116 This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any
8117 floating point or vector of floating point type. Not all targets support
8122 declare float @llvm.sqrt.f32(float %Val)
8123 declare double @llvm.sqrt.f64(double %Val)
8124 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
8125 declare fp128 @llvm.sqrt.f128(fp128 %Val)
8126 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
8131 The '``llvm.sqrt``' intrinsics return the sqrt of the specified operand,
8132 returning the same value as the libm '``sqrt``' functions would. Unlike
8133 ``sqrt`` in libm, however, ``llvm.sqrt`` has undefined behavior for
8134 negative numbers other than -0.0 (which allows for better optimization,
8135 because there is no need to worry about errno being set).
8136 ``llvm.sqrt(-0.0)`` is defined to return -0.0 like IEEE sqrt.
8141 The argument and return value are floating point numbers of the same
8147 This function returns the sqrt of the specified operand if it is a
8148 nonnegative floating point number.
8150 '``llvm.powi.*``' Intrinsic
8151 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8156 This is an overloaded intrinsic. You can use ``llvm.powi`` on any
8157 floating point or vector of floating point type. Not all targets support
8162 declare float @llvm.powi.f32(float %Val, i32 %power)
8163 declare double @llvm.powi.f64(double %Val, i32 %power)
8164 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
8165 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
8166 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
8171 The '``llvm.powi.*``' intrinsics return the first operand raised to the
8172 specified (positive or negative) power. The order of evaluation of
8173 multiplications is not defined. When a vector of floating point type is
8174 used, the second argument remains a scalar integer value.
8179 The second argument is an integer power, and the first is a value to
8180 raise to that power.
8185 This function returns the first value raised to the second power with an
8186 unspecified sequence of rounding operations.
8188 '``llvm.sin.*``' Intrinsic
8189 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8194 This is an overloaded intrinsic. You can use ``llvm.sin`` on any
8195 floating point or vector of floating point type. Not all targets support
8200 declare float @llvm.sin.f32(float %Val)
8201 declare double @llvm.sin.f64(double %Val)
8202 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
8203 declare fp128 @llvm.sin.f128(fp128 %Val)
8204 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
8209 The '``llvm.sin.*``' intrinsics return the sine of the operand.
8214 The argument and return value are floating point numbers of the same
8220 This function returns the sine of the specified operand, returning the
8221 same values as the libm ``sin`` functions would, and handles error
8222 conditions in the same way.
8224 '``llvm.cos.*``' Intrinsic
8225 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8230 This is an overloaded intrinsic. You can use ``llvm.cos`` on any
8231 floating point or vector of floating point type. Not all targets support
8236 declare float @llvm.cos.f32(float %Val)
8237 declare double @llvm.cos.f64(double %Val)
8238 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
8239 declare fp128 @llvm.cos.f128(fp128 %Val)
8240 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
8245 The '``llvm.cos.*``' intrinsics return the cosine of the operand.
8250 The argument and return value are floating point numbers of the same
8256 This function returns the cosine of the specified operand, returning the
8257 same values as the libm ``cos`` functions would, and handles error
8258 conditions in the same way.
8260 '``llvm.pow.*``' Intrinsic
8261 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8266 This is an overloaded intrinsic. You can use ``llvm.pow`` on any
8267 floating point or vector of floating point type. Not all targets support
8272 declare float @llvm.pow.f32(float %Val, float %Power)
8273 declare double @llvm.pow.f64(double %Val, double %Power)
8274 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
8275 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
8276 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
8281 The '``llvm.pow.*``' intrinsics return the first operand raised to the
8282 specified (positive or negative) power.
8287 The second argument is a floating point power, and the first is a value
8288 to raise to that power.
8293 This function returns the first value raised to the second power,
8294 returning the same values as the libm ``pow`` functions would, and
8295 handles error conditions in the same way.
8297 '``llvm.exp.*``' Intrinsic
8298 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8303 This is an overloaded intrinsic. You can use ``llvm.exp`` on any
8304 floating point or vector of floating point type. Not all targets support
8309 declare float @llvm.exp.f32(float %Val)
8310 declare double @llvm.exp.f64(double %Val)
8311 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
8312 declare fp128 @llvm.exp.f128(fp128 %Val)
8313 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
8318 The '``llvm.exp.*``' intrinsics perform the exp function.
8323 The argument and return value are floating point numbers of the same
8329 This function returns the same values as the libm ``exp`` functions
8330 would, and handles error conditions in the same way.
8332 '``llvm.exp2.*``' Intrinsic
8333 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8338 This is an overloaded intrinsic. You can use ``llvm.exp2`` on any
8339 floating point or vector of floating point type. Not all targets support
8344 declare float @llvm.exp2.f32(float %Val)
8345 declare double @llvm.exp2.f64(double %Val)
8346 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val)
8347 declare fp128 @llvm.exp2.f128(fp128 %Val)
8348 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val)
8353 The '``llvm.exp2.*``' intrinsics perform the exp2 function.
8358 The argument and return value are floating point numbers of the same
8364 This function returns the same values as the libm ``exp2`` functions
8365 would, and handles error conditions in the same way.
8367 '``llvm.log.*``' Intrinsic
8368 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8373 This is an overloaded intrinsic. You can use ``llvm.log`` on any
8374 floating point or vector of floating point type. Not all targets support
8379 declare float @llvm.log.f32(float %Val)
8380 declare double @llvm.log.f64(double %Val)
8381 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
8382 declare fp128 @llvm.log.f128(fp128 %Val)
8383 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
8388 The '``llvm.log.*``' intrinsics perform the log function.
8393 The argument and return value are floating point numbers of the same
8399 This function returns the same values as the libm ``log`` functions
8400 would, and handles error conditions in the same way.
8402 '``llvm.log10.*``' Intrinsic
8403 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8408 This is an overloaded intrinsic. You can use ``llvm.log10`` on any
8409 floating point or vector of floating point type. Not all targets support
8414 declare float @llvm.log10.f32(float %Val)
8415 declare double @llvm.log10.f64(double %Val)
8416 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val)
8417 declare fp128 @llvm.log10.f128(fp128 %Val)
8418 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val)
8423 The '``llvm.log10.*``' intrinsics perform the log10 function.
8428 The argument and return value are floating point numbers of the same
8434 This function returns the same values as the libm ``log10`` functions
8435 would, and handles error conditions in the same way.
8437 '``llvm.log2.*``' Intrinsic
8438 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8443 This is an overloaded intrinsic. You can use ``llvm.log2`` on any
8444 floating point or vector of floating point type. Not all targets support
8449 declare float @llvm.log2.f32(float %Val)
8450 declare double @llvm.log2.f64(double %Val)
8451 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val)
8452 declare fp128 @llvm.log2.f128(fp128 %Val)
8453 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val)
8458 The '``llvm.log2.*``' intrinsics perform the log2 function.
8463 The argument and return value are floating point numbers of the same
8469 This function returns the same values as the libm ``log2`` functions
8470 would, and handles error conditions in the same way.
8472 '``llvm.fma.*``' Intrinsic
8473 ^^^^^^^^^^^^^^^^^^^^^^^^^^
8478 This is an overloaded intrinsic. You can use ``llvm.fma`` on any
8479 floating point or vector of floating point type. Not all targets support
8484 declare float @llvm.fma.f32(float %a, float %b, float %c)
8485 declare double @llvm.fma.f64(double %a, double %b, double %c)
8486 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
8487 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
8488 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
8493 The '``llvm.fma.*``' intrinsics perform the fused multiply-add
8499 The argument and return value are floating point numbers of the same
8505 This function returns the same values as the libm ``fma`` functions
8506 would, and does not set errno.
8508 '``llvm.fabs.*``' Intrinsic
8509 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8514 This is an overloaded intrinsic. You can use ``llvm.fabs`` on any
8515 floating point or vector of floating point type. Not all targets support
8520 declare float @llvm.fabs.f32(float %Val)
8521 declare double @llvm.fabs.f64(double %Val)
8522 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val)
8523 declare fp128 @llvm.fabs.f128(fp128 %Val)
8524 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val)
8529 The '``llvm.fabs.*``' intrinsics return the absolute value of the
8535 The argument and return value are floating point numbers of the same
8541 This function returns the same values as the libm ``fabs`` functions
8542 would, and handles error conditions in the same way.
8544 '``llvm.minnum.*``' Intrinsic
8545 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8550 This is an overloaded intrinsic. You can use ``llvm.minnum`` on any
8551 floating point or vector of floating point type. Not all targets support
8556 declare float @llvm.minnum.f32(float %Val0, float %Val1)
8557 declare double @llvm.minnum.f64(double %Val0, double %Val1)
8558 declare x86_fp80 @llvm.minnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
8559 declare fp128 @llvm.minnum.f128(fp128 %Val0, fp128 %Val1)
8560 declare ppc_fp128 @llvm.minnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
8565 The '``llvm.minnum.*``' intrinsics return the minimum of the two
8572 The arguments and return value are floating point numbers of the same
8578 Follows the IEEE-754 semantics for minNum, which also match for libm's
8581 If either operand is a NaN, returns the other non-NaN operand. Returns
8582 NaN only if both operands are NaN. If the operands compare equal,
8583 returns a value that compares equal to both operands. This means that
8584 fmin(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
8586 '``llvm.maxnum.*``' Intrinsic
8587 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8592 This is an overloaded intrinsic. You can use ``llvm.maxnum`` on any
8593 floating point or vector of floating point type. Not all targets support
8598 declare float @llvm.maxnum.f32(float %Val0, float %Val1l)
8599 declare double @llvm.maxnum.f64(double %Val0, double %Val1)
8600 declare x86_fp80 @llvm.maxnum.f80(x86_fp80 %Val0, x86_fp80 %Val1)
8601 declare fp128 @llvm.maxnum.f128(fp128 %Val0, fp128 %Val1)
8602 declare ppc_fp128 @llvm.maxnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1)
8607 The '``llvm.maxnum.*``' intrinsics return the maximum of the two
8614 The arguments and return value are floating point numbers of the same
8619 Follows the IEEE-754 semantics for maxNum, which also match for libm's
8622 If either operand is a NaN, returns the other non-NaN operand. Returns
8623 NaN only if both operands are NaN. If the operands compare equal,
8624 returns a value that compares equal to both operands. This means that
8625 fmax(+/-0.0, +/-0.0) could return either -0.0 or 0.0.
8627 '``llvm.copysign.*``' Intrinsic
8628 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8633 This is an overloaded intrinsic. You can use ``llvm.copysign`` on any
8634 floating point or vector of floating point type. Not all targets support
8639 declare float @llvm.copysign.f32(float %Mag, float %Sgn)
8640 declare double @llvm.copysign.f64(double %Mag, double %Sgn)
8641 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn)
8642 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn)
8643 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn)
8648 The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the
8649 first operand and the sign of the second operand.
8654 The arguments and return value are floating point numbers of the same
8660 This function returns the same values as the libm ``copysign``
8661 functions would, and handles error conditions in the same way.
8663 '``llvm.floor.*``' Intrinsic
8664 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8669 This is an overloaded intrinsic. You can use ``llvm.floor`` on any
8670 floating point or vector of floating point type. Not all targets support
8675 declare float @llvm.floor.f32(float %Val)
8676 declare double @llvm.floor.f64(double %Val)
8677 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val)
8678 declare fp128 @llvm.floor.f128(fp128 %Val)
8679 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val)
8684 The '``llvm.floor.*``' intrinsics return the floor of the operand.
8689 The argument and return value are floating point numbers of the same
8695 This function returns the same values as the libm ``floor`` functions
8696 would, and handles error conditions in the same way.
8698 '``llvm.ceil.*``' Intrinsic
8699 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8704 This is an overloaded intrinsic. You can use ``llvm.ceil`` on any
8705 floating point or vector of floating point type. Not all targets support
8710 declare float @llvm.ceil.f32(float %Val)
8711 declare double @llvm.ceil.f64(double %Val)
8712 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val)
8713 declare fp128 @llvm.ceil.f128(fp128 %Val)
8714 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val)
8719 The '``llvm.ceil.*``' intrinsics return the ceiling of the operand.
8724 The argument and return value are floating point numbers of the same
8730 This function returns the same values as the libm ``ceil`` functions
8731 would, and handles error conditions in the same way.
8733 '``llvm.trunc.*``' Intrinsic
8734 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8739 This is an overloaded intrinsic. You can use ``llvm.trunc`` on any
8740 floating point or vector of floating point type. Not all targets support
8745 declare float @llvm.trunc.f32(float %Val)
8746 declare double @llvm.trunc.f64(double %Val)
8747 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val)
8748 declare fp128 @llvm.trunc.f128(fp128 %Val)
8749 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val)
8754 The '``llvm.trunc.*``' intrinsics returns the operand rounded to the
8755 nearest integer not larger in magnitude than the operand.
8760 The argument and return value are floating point numbers of the same
8766 This function returns the same values as the libm ``trunc`` functions
8767 would, and handles error conditions in the same way.
8769 '``llvm.rint.*``' Intrinsic
8770 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8775 This is an overloaded intrinsic. You can use ``llvm.rint`` on any
8776 floating point or vector of floating point type. Not all targets support
8781 declare float @llvm.rint.f32(float %Val)
8782 declare double @llvm.rint.f64(double %Val)
8783 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val)
8784 declare fp128 @llvm.rint.f128(fp128 %Val)
8785 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val)
8790 The '``llvm.rint.*``' intrinsics returns the operand rounded to the
8791 nearest integer. It may raise an inexact floating-point exception if the
8792 operand isn't an integer.
8797 The argument and return value are floating point numbers of the same
8803 This function returns the same values as the libm ``rint`` functions
8804 would, and handles error conditions in the same way.
8806 '``llvm.nearbyint.*``' Intrinsic
8807 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8812 This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any
8813 floating point or vector of floating point type. Not all targets support
8818 declare float @llvm.nearbyint.f32(float %Val)
8819 declare double @llvm.nearbyint.f64(double %Val)
8820 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val)
8821 declare fp128 @llvm.nearbyint.f128(fp128 %Val)
8822 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val)
8827 The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the
8833 The argument and return value are floating point numbers of the same
8839 This function returns the same values as the libm ``nearbyint``
8840 functions would, and handles error conditions in the same way.
8842 '``llvm.round.*``' Intrinsic
8843 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8848 This is an overloaded intrinsic. You can use ``llvm.round`` on any
8849 floating point or vector of floating point type. Not all targets support
8854 declare float @llvm.round.f32(float %Val)
8855 declare double @llvm.round.f64(double %Val)
8856 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val)
8857 declare fp128 @llvm.round.f128(fp128 %Val)
8858 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val)
8863 The '``llvm.round.*``' intrinsics returns the operand rounded to the
8869 The argument and return value are floating point numbers of the same
8875 This function returns the same values as the libm ``round``
8876 functions would, and handles error conditions in the same way.
8878 Bit Manipulation Intrinsics
8879 ---------------------------
8881 LLVM provides intrinsics for a few important bit manipulation
8882 operations. These allow efficient code generation for some algorithms.
8884 '``llvm.bswap.*``' Intrinsics
8885 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8890 This is an overloaded intrinsic function. You can use bswap on any
8891 integer type that is an even number of bytes (i.e. BitWidth % 16 == 0).
8895 declare i16 @llvm.bswap.i16(i16 <id>)
8896 declare i32 @llvm.bswap.i32(i32 <id>)
8897 declare i64 @llvm.bswap.i64(i64 <id>)
8902 The '``llvm.bswap``' family of intrinsics is used to byte swap integer
8903 values with an even number of bytes (positive multiple of 16 bits).
8904 These are useful for performing operations on data that is not in the
8905 target's native byte order.
8910 The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high
8911 and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32``
8912 intrinsic returns an i32 value that has the four bytes of the input i32
8913 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the
8914 returned i32 will have its bytes in 3, 2, 1, 0 order. The
8915 ``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this
8916 concept to additional even-byte lengths (6 bytes, 8 bytes and more,
8919 '``llvm.ctpop.*``' Intrinsic
8920 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8925 This is an overloaded intrinsic. You can use llvm.ctpop on any integer
8926 bit width, or on any vector with integer elements. Not all targets
8927 support all bit widths or vector types, however.
8931 declare i8 @llvm.ctpop.i8(i8 <src>)
8932 declare i16 @llvm.ctpop.i16(i16 <src>)
8933 declare i32 @llvm.ctpop.i32(i32 <src>)
8934 declare i64 @llvm.ctpop.i64(i64 <src>)
8935 declare i256 @llvm.ctpop.i256(i256 <src>)
8936 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
8941 The '``llvm.ctpop``' family of intrinsics counts the number of bits set
8947 The only argument is the value to be counted. The argument may be of any
8948 integer type, or a vector with integer elements. The return type must
8949 match the argument type.
8954 The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within
8955 each element of a vector.
8957 '``llvm.ctlz.*``' Intrinsic
8958 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
8963 This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any
8964 integer bit width, or any vector whose elements are integers. Not all
8965 targets support all bit widths or vector types, however.
8969 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>)
8970 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>)
8971 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>)
8972 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>)
8973 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>)
8974 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
8979 The '``llvm.ctlz``' family of intrinsic functions counts the number of
8980 leading zeros in a variable.
8985 The first argument is the value to be counted. This argument may be of
8986 any integer type, or a vector with integer element type. The return
8987 type must match the first argument type.
8989 The second argument must be a constant and is a flag to indicate whether
8990 the intrinsic should ensure that a zero as the first argument produces a
8991 defined result. Historically some architectures did not provide a
8992 defined result for zero values as efficiently, and many algorithms are
8993 now predicated on avoiding zero-value inputs.
8998 The '``llvm.ctlz``' intrinsic counts the leading (most significant)
8999 zeros in a variable, or within each element of the vector. If
9000 ``src == 0`` then the result is the size in bits of the type of ``src``
9001 if ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
9002 ``llvm.ctlz(i32 2) = 30``.
9004 '``llvm.cttz.*``' Intrinsic
9005 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
9010 This is an overloaded intrinsic. You can use ``llvm.cttz`` on any
9011 integer bit width, or any vector of integer elements. Not all targets
9012 support all bit widths or vector types, however.
9016 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>)
9017 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>)
9018 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>)
9019 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>)
9020 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>)
9021 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
9026 The '``llvm.cttz``' family of intrinsic functions counts the number of
9032 The first argument is the value to be counted. This argument may be of
9033 any integer type, or a vector with integer element type. The return
9034 type must match the first argument type.
9036 The second argument must be a constant and is a flag to indicate whether
9037 the intrinsic should ensure that a zero as the first argument produces a
9038 defined result. Historically some architectures did not provide a
9039 defined result for zero values as efficiently, and many algorithms are
9040 now predicated on avoiding zero-value inputs.
9045 The '``llvm.cttz``' intrinsic counts the trailing (least significant)
9046 zeros in a variable, or within each element of a vector. If ``src == 0``
9047 then the result is the size in bits of the type of ``src`` if
9048 ``is_zero_undef == 0`` and ``undef`` otherwise. For example,
9049 ``llvm.cttz(2) = 1``.
9051 Arithmetic with Overflow Intrinsics
9052 -----------------------------------
9054 LLVM provides intrinsics for some arithmetic with overflow operations.
9056 '``llvm.sadd.with.overflow.*``' Intrinsics
9057 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9062 This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow``
9063 on any integer bit width.
9067 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
9068 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
9069 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
9074 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
9075 a signed addition of the two arguments, and indicate whether an overflow
9076 occurred during the signed summation.
9081 The arguments (%a and %b) and the first element of the result structure
9082 may be of integer types of any bit width, but they must have the same
9083 bit width. The second element of the result structure must be of type
9084 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
9090 The '``llvm.sadd.with.overflow``' family of intrinsic functions perform
9091 a signed addition of the two variables. They return a structure --- the
9092 first element of which is the signed summation, and the second element
9093 of which is a bit specifying if the signed summation resulted in an
9099 .. code-block:: llvm
9101 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
9102 %sum = extractvalue {i32, i1} %res, 0
9103 %obit = extractvalue {i32, i1} %res, 1
9104 br i1 %obit, label %overflow, label %normal
9106 '``llvm.uadd.with.overflow.*``' Intrinsics
9107 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9112 This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow``
9113 on any integer bit width.
9117 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
9118 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
9119 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
9124 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
9125 an unsigned addition of the two arguments, and indicate whether a carry
9126 occurred during the unsigned summation.
9131 The arguments (%a and %b) and the first element of the result structure
9132 may be of integer types of any bit width, but they must have the same
9133 bit width. The second element of the result structure must be of type
9134 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
9140 The '``llvm.uadd.with.overflow``' family of intrinsic functions perform
9141 an unsigned addition of the two arguments. They return a structure --- the
9142 first element of which is the sum, and the second element of which is a
9143 bit specifying if the unsigned summation resulted in a carry.
9148 .. code-block:: llvm
9150 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
9151 %sum = extractvalue {i32, i1} %res, 0
9152 %obit = extractvalue {i32, i1} %res, 1
9153 br i1 %obit, label %carry, label %normal
9155 '``llvm.ssub.with.overflow.*``' Intrinsics
9156 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9161 This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow``
9162 on any integer bit width.
9166 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
9167 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
9168 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
9173 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
9174 a signed subtraction of the two arguments, and indicate whether an
9175 overflow occurred during the signed subtraction.
9180 The arguments (%a and %b) and the first element of the result structure
9181 may be of integer types of any bit width, but they must have the same
9182 bit width. The second element of the result structure must be of type
9183 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
9189 The '``llvm.ssub.with.overflow``' family of intrinsic functions perform
9190 a signed subtraction of the two arguments. They return a structure --- the
9191 first element of which is the subtraction, and the second element of
9192 which is a bit specifying if the signed subtraction resulted in an
9198 .. code-block:: llvm
9200 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
9201 %sum = extractvalue {i32, i1} %res, 0
9202 %obit = extractvalue {i32, i1} %res, 1
9203 br i1 %obit, label %overflow, label %normal
9205 '``llvm.usub.with.overflow.*``' Intrinsics
9206 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9211 This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow``
9212 on any integer bit width.
9216 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
9217 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
9218 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
9223 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
9224 an unsigned subtraction of the two arguments, and indicate whether an
9225 overflow occurred during the unsigned subtraction.
9230 The arguments (%a and %b) and the first element of the result structure
9231 may be of integer types of any bit width, but they must have the same
9232 bit width. The second element of the result structure must be of type
9233 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
9239 The '``llvm.usub.with.overflow``' family of intrinsic functions perform
9240 an unsigned subtraction of the two arguments. They return a structure ---
9241 the first element of which is the subtraction, and the second element of
9242 which is a bit specifying if the unsigned subtraction resulted in an
9248 .. code-block:: llvm
9250 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
9251 %sum = extractvalue {i32, i1} %res, 0
9252 %obit = extractvalue {i32, i1} %res, 1
9253 br i1 %obit, label %overflow, label %normal
9255 '``llvm.smul.with.overflow.*``' Intrinsics
9256 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9261 This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow``
9262 on any integer bit width.
9266 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
9267 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
9268 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
9273 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
9274 a signed multiplication of the two arguments, and indicate whether an
9275 overflow occurred during the signed multiplication.
9280 The arguments (%a and %b) and the first element of the result structure
9281 may be of integer types of any bit width, but they must have the same
9282 bit width. The second element of the result structure must be of type
9283 ``i1``. ``%a`` and ``%b`` are the two values that will undergo signed
9289 The '``llvm.smul.with.overflow``' family of intrinsic functions perform
9290 a signed multiplication of the two arguments. They return a structure ---
9291 the first element of which is the multiplication, and the second element
9292 of which is a bit specifying if the signed multiplication resulted in an
9298 .. code-block:: llvm
9300 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
9301 %sum = extractvalue {i32, i1} %res, 0
9302 %obit = extractvalue {i32, i1} %res, 1
9303 br i1 %obit, label %overflow, label %normal
9305 '``llvm.umul.with.overflow.*``' Intrinsics
9306 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9311 This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow``
9312 on any integer bit width.
9316 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
9317 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
9318 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
9323 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
9324 a unsigned multiplication of the two arguments, and indicate whether an
9325 overflow occurred during the unsigned multiplication.
9330 The arguments (%a and %b) and the first element of the result structure
9331 may be of integer types of any bit width, but they must have the same
9332 bit width. The second element of the result structure must be of type
9333 ``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned
9339 The '``llvm.umul.with.overflow``' family of intrinsic functions perform
9340 an unsigned multiplication of the two arguments. They return a structure ---
9341 the first element of which is the multiplication, and the second
9342 element of which is a bit specifying if the unsigned multiplication
9343 resulted in an overflow.
9348 .. code-block:: llvm
9350 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
9351 %sum = extractvalue {i32, i1} %res, 0
9352 %obit = extractvalue {i32, i1} %res, 1
9353 br i1 %obit, label %overflow, label %normal
9355 Specialised Arithmetic Intrinsics
9356 ---------------------------------
9358 '``llvm.fmuladd.*``' Intrinsic
9359 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9366 declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
9367 declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
9372 The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add
9373 expressions that can be fused if the code generator determines that (a) the
9374 target instruction set has support for a fused operation, and (b) that the
9375 fused operation is more efficient than the equivalent, separate pair of mul
9376 and add instructions.
9381 The '``llvm.fmuladd.*``' intrinsics each take three arguments: two
9382 multiplicands, a and b, and an addend c.
9391 %0 = call float @llvm.fmuladd.f32(%a, %b, %c)
9393 is equivalent to the expression a \* b + c, except that rounding will
9394 not be performed between the multiplication and addition steps if the
9395 code generator fuses the operations. Fusion is not guaranteed, even if
9396 the target platform supports it. If a fused multiply-add is required the
9397 corresponding llvm.fma.\* intrinsic function should be used
9398 instead. This never sets errno, just as '``llvm.fma.*``'.
9403 .. code-block:: llvm
9405 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields float:r2 = (a * b) + c
9407 Half Precision Floating Point Intrinsics
9408 ----------------------------------------
9410 For most target platforms, half precision floating point is a
9411 storage-only format. This means that it is a dense encoding (in memory)
9412 but does not support computation in the format.
9414 This means that code must first load the half-precision floating point
9415 value as an i16, then convert it to float with
9416 :ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can
9417 then be performed on the float value (including extending to double
9418 etc). To store the value back to memory, it is first converted to float
9419 if needed, then converted to i16 with
9420 :ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an
9423 .. _int_convert_to_fp16:
9425 '``llvm.convert.to.fp16``' Intrinsic
9426 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9433 declare i16 @llvm.convert.to.fp16.f32(float %a)
9434 declare i16 @llvm.convert.to.fp16.f64(double %a)
9439 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
9440 conventional floating point type to half precision floating point format.
9445 The intrinsic function contains single argument - the value to be
9451 The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a
9452 conventional floating point format to half precision floating point format. The
9453 return value is an ``i16`` which contains the converted number.
9458 .. code-block:: llvm
9460 %res = call i16 @llvm.convert.to.fp16.f32(float %a)
9461 store i16 %res, i16* @x, align 2
9463 .. _int_convert_from_fp16:
9465 '``llvm.convert.from.fp16``' Intrinsic
9466 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9473 declare float @llvm.convert.from.fp16.f32(i16 %a)
9474 declare double @llvm.convert.from.fp16.f64(i16 %a)
9479 The '``llvm.convert.from.fp16``' intrinsic function performs a
9480 conversion from half precision floating point format to single precision
9481 floating point format.
9486 The intrinsic function contains single argument - the value to be
9492 The '``llvm.convert.from.fp16``' intrinsic function performs a
9493 conversion from half single precision floating point format to single
9494 precision floating point format. The input half-float value is
9495 represented by an ``i16`` value.
9500 .. code-block:: llvm
9502 %a = load i16* @x, align 2
9503 %res = call float @llvm.convert.from.fp16(i16 %a)
9510 The LLVM debugger intrinsics (which all start with ``llvm.dbg.``
9511 prefix), are described in the `LLVM Source Level
9512 Debugging <SourceLevelDebugging.html#format_common_intrinsics>`_
9515 Exception Handling Intrinsics
9516 -----------------------------
9518 The LLVM exception handling intrinsics (which all start with
9519 ``llvm.eh.`` prefix), are described in the `LLVM Exception
9520 Handling <ExceptionHandling.html#format_common_intrinsics>`_ document.
9524 Trampoline Intrinsics
9525 ---------------------
9527 These intrinsics make it possible to excise one parameter, marked with
9528 the :ref:`nest <nest>` attribute, from a function. The result is a
9529 callable function pointer lacking the nest parameter - the caller does
9530 not need to provide a value for it. Instead, the value to use is stored
9531 in advance in a "trampoline", a block of memory usually allocated on the
9532 stack, which also contains code to splice the nest value into the
9533 argument list. This is used to implement the GCC nested function address
9536 For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)``
9537 then the resulting function pointer has signature ``i32 (i32, i32)*``.
9538 It can be created as follows:
9540 .. code-block:: llvm
9542 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
9543 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
9544 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
9545 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
9546 %fp = bitcast i8* %p to i32 (i32, i32)*
9548 The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to
9549 ``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``.
9553 '``llvm.init.trampoline``' Intrinsic
9554 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9561 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
9566 This fills the memory pointed to by ``tramp`` with executable code,
9567 turning it into a trampoline.
9572 The ``llvm.init.trampoline`` intrinsic takes three arguments, all
9573 pointers. The ``tramp`` argument must point to a sufficiently large and
9574 sufficiently aligned block of memory; this memory is written to by the
9575 intrinsic. Note that the size and the alignment are target-specific -
9576 LLVM currently provides no portable way of determining them, so a
9577 front-end that generates this intrinsic needs to have some
9578 target-specific knowledge. The ``func`` argument must hold a function
9579 bitcast to an ``i8*``.
9584 The block of memory pointed to by ``tramp`` is filled with target
9585 dependent code, turning it into a function. Then ``tramp`` needs to be
9586 passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can
9587 be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new
9588 function's signature is the same as that of ``func`` with any arguments
9589 marked with the ``nest`` attribute removed. At most one such ``nest``
9590 argument is allowed, and it must be of pointer type. Calling the new
9591 function is equivalent to calling ``func`` with the same argument list,
9592 but with ``nval`` used for the missing ``nest`` argument. If, after
9593 calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is
9594 modified, then the effect of any later call to the returned function
9595 pointer is undefined.
9599 '``llvm.adjust.trampoline``' Intrinsic
9600 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9607 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
9612 This performs any required machine-specific adjustment to the address of
9613 a trampoline (passed as ``tramp``).
9618 ``tramp`` must point to a block of memory which already has trampoline
9619 code filled in by a previous call to
9620 :ref:`llvm.init.trampoline <int_it>`.
9625 On some architectures the address of the code to be executed needs to be
9626 different than the address where the trampoline is actually stored. This
9627 intrinsic returns the executable address corresponding to ``tramp``
9628 after performing the required machine specific adjustments. The pointer
9629 returned can then be :ref:`bitcast and executed <int_trampoline>`.
9631 Masked Vector Load and Store Intrinsics
9632 ---------------------------------------
9634 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.
9638 '``llvm.masked.load.*``' Intrinsics
9639 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9643 This is an overloaded intrinsic. The loaded data is a vector of any integer or floating point data type.
9647 declare <16 x float> @llvm.masked.load.v16f32 (<16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>)
9648 declare <2 x double> @llvm.masked.load.v2f64 (<2 x double>* <ptr>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>)
9653 Reads a vector from memory according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes. The masked-off lanes in the result vector are taken from the corresponding lanes in the passthru operand.
9659 The first operand is the base pointer for the load. The second operand is the alignment of the source location. It must be a constant integer value. The third operand, mask, is a vector of boolean 'i1' values with the same number of elements as the return type. The fourth is a pass-through value that is used to fill the masked-off lanes of the result. The return type, underlying type of the base pointer and the type of passthru operand are the same vector types.
9665 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.
9666 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.
9671 %res = call <16 x float> @llvm.masked.load.v16f32 (<16 x float>* %ptr, i32 4, <16 x i1>%mask, <16 x float> %passthru)
9673 ;; The result of the two following instructions is identical aside from potential memory access exception
9674 %loadlal = load <16 x float>* %ptr, align 4
9675 %res = select <16 x i1> %mask, <16 x float> %loadlal, <16 x float> %passthru
9679 '``llvm.masked.store.*``' Intrinsics
9680 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9684 This is an overloaded intrinsic. The data stored in memory is a vector of any integer or floating point data type.
9688 declare void @llvm.masked.store.v8i32 (<8 x i32> <value>, <8 x i32> * <ptr>, i32 <alignment>, <8 x i1> <mask>)
9689 declare void @llvm.masked.store.v16f32(<16 x i32> <value>, <16 x i32>* <ptr>, i32 <alignment>, <16 x i1> <mask>)
9694 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.
9699 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.
9705 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.
9706 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.
9710 call void @llvm.masked.store.v16f32(<16 x float> %value, <16 x float>* %ptr, i32 4, <16 x i1> %mask)
9712 ;; The result of the following instructions is identical aside from potential data races and memory access exceptions
9713 %oldval = load <16 x float>* %ptr, align 4
9714 %res = select <16 x i1> %mask, <16 x float> %value, <16 x float> %oldval
9715 store <16 x float> %res, <16 x float>* %ptr, align 4
9721 This class of intrinsics provides information about the lifetime of
9722 memory objects and ranges where variables are immutable.
9726 '``llvm.lifetime.start``' Intrinsic
9727 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9734 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
9739 The '``llvm.lifetime.start``' intrinsic specifies the start of a memory
9745 The first argument is a constant integer representing the size of the
9746 object, or -1 if it is variable sized. The second argument is a pointer
9752 This intrinsic indicates that before this point in the code, the value
9753 of the memory pointed to by ``ptr`` is dead. This means that it is known
9754 to never be used and has an undefined value. A load from the pointer
9755 that precedes this intrinsic can be replaced with ``'undef'``.
9759 '``llvm.lifetime.end``' Intrinsic
9760 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9767 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
9772 The '``llvm.lifetime.end``' intrinsic specifies the end of a memory
9778 The first argument is a constant integer representing the size of the
9779 object, or -1 if it is variable sized. The second argument is a pointer
9785 This intrinsic indicates that after this point in the code, the value of
9786 the memory pointed to by ``ptr`` is dead. This means that it is known to
9787 never be used and has an undefined value. Any stores into the memory
9788 object following this intrinsic may be removed as dead.
9790 '``llvm.invariant.start``' Intrinsic
9791 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9798 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
9803 The '``llvm.invariant.start``' intrinsic specifies that the contents of
9804 a memory object will not change.
9809 The first argument is a constant integer representing the size of the
9810 object, or -1 if it is variable sized. The second argument is a pointer
9816 This intrinsic indicates that until an ``llvm.invariant.end`` that uses
9817 the return value, the referenced memory location is constant and
9820 '``llvm.invariant.end``' Intrinsic
9821 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9828 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
9833 The '``llvm.invariant.end``' intrinsic specifies that the contents of a
9834 memory object are mutable.
9839 The first argument is the matching ``llvm.invariant.start`` intrinsic.
9840 The second argument is a constant integer representing the size of the
9841 object, or -1 if it is variable sized and the third argument is a
9842 pointer to the object.
9847 This intrinsic indicates that the memory is mutable again.
9852 This class of intrinsics is designed to be generic and has no specific
9855 '``llvm.var.annotation``' Intrinsic
9856 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9863 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
9868 The '``llvm.var.annotation``' intrinsic.
9873 The first argument is a pointer to a value, the second is a pointer to a
9874 global string, the third is a pointer to a global string which is the
9875 source file name, and the last argument is the line number.
9880 This intrinsic allows annotation of local variables with arbitrary
9881 strings. This can be useful for special purpose optimizations that want
9882 to look for these annotations. These have no other defined use; they are
9883 ignored by code generation and optimization.
9885 '``llvm.ptr.annotation.*``' Intrinsic
9886 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9891 This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a
9892 pointer to an integer of any width. *NOTE* you must specify an address space for
9893 the pointer. The identifier for the default address space is the integer
9898 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
9899 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>)
9900 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>)
9901 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>)
9902 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>)
9907 The '``llvm.ptr.annotation``' intrinsic.
9912 The first argument is a pointer to an integer value of arbitrary bitwidth
9913 (result of some expression), the second is a pointer to a global string, the
9914 third is a pointer to a global string which is the source file name, and the
9915 last argument is the line number. It returns the value of the first argument.
9920 This intrinsic allows annotation of a pointer to an integer with arbitrary
9921 strings. This can be useful for special purpose optimizations that want to look
9922 for these annotations. These have no other defined use; they are ignored by code
9923 generation and optimization.
9925 '``llvm.annotation.*``' Intrinsic
9926 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9931 This is an overloaded intrinsic. You can use '``llvm.annotation``' on
9932 any integer bit width.
9936 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
9937 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
9938 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
9939 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
9940 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
9945 The '``llvm.annotation``' intrinsic.
9950 The first argument is an integer value (result of some expression), the
9951 second is a pointer to a global string, the third is a pointer to a
9952 global string which is the source file name, and the last argument is
9953 the line number. It returns the value of the first argument.
9958 This intrinsic allows annotations to be put on arbitrary expressions
9959 with arbitrary strings. This can be useful for special purpose
9960 optimizations that want to look for these annotations. These have no
9961 other defined use; they are ignored by code generation and optimization.
9963 '``llvm.trap``' Intrinsic
9964 ^^^^^^^^^^^^^^^^^^^^^^^^^
9971 declare void @llvm.trap() noreturn nounwind
9976 The '``llvm.trap``' intrinsic.
9986 This intrinsic is lowered to the target dependent trap instruction. If
9987 the target does not have a trap instruction, this intrinsic will be
9988 lowered to a call of the ``abort()`` function.
9990 '``llvm.debugtrap``' Intrinsic
9991 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9998 declare void @llvm.debugtrap() nounwind
10003 The '``llvm.debugtrap``' intrinsic.
10013 This intrinsic is lowered to code which is intended to cause an
10014 execution trap with the intention of requesting the attention of a
10017 '``llvm.stackprotector``' Intrinsic
10018 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10025 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
10030 The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it
10031 onto the stack at ``slot``. The stack slot is adjusted to ensure that it
10032 is placed on the stack before local variables.
10037 The ``llvm.stackprotector`` intrinsic requires two pointer arguments.
10038 The first argument is the value loaded from the stack guard
10039 ``@__stack_chk_guard``. The second variable is an ``alloca`` that has
10040 enough space to hold the value of the guard.
10045 This intrinsic causes the prologue/epilogue inserter to force the position of
10046 the ``AllocaInst`` stack slot to be before local variables on the stack. This is
10047 to ensure that if a local variable on the stack is overwritten, it will destroy
10048 the value of the guard. When the function exits, the guard on the stack is
10049 checked against the original guard by ``llvm.stackprotectorcheck``. If they are
10050 different, then ``llvm.stackprotectorcheck`` causes the program to abort by
10051 calling the ``__stack_chk_fail()`` function.
10053 '``llvm.stackprotectorcheck``' Intrinsic
10054 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10061 declare void @llvm.stackprotectorcheck(i8** <guard>)
10066 The ``llvm.stackprotectorcheck`` intrinsic compares ``guard`` against an already
10067 created stack protector and if they are not equal calls the
10068 ``__stack_chk_fail()`` function.
10073 The ``llvm.stackprotectorcheck`` intrinsic requires one pointer argument, the
10074 the variable ``@__stack_chk_guard``.
10079 This intrinsic is provided to perform the stack protector check by comparing
10080 ``guard`` with the stack slot created by ``llvm.stackprotector`` and if the
10081 values do not match call the ``__stack_chk_fail()`` function.
10083 The reason to provide this as an IR level intrinsic instead of implementing it
10084 via other IR operations is that in order to perform this operation at the IR
10085 level without an intrinsic, one would need to create additional basic blocks to
10086 handle the success/failure cases. This makes it difficult to stop the stack
10087 protector check from disrupting sibling tail calls in Codegen. With this
10088 intrinsic, we are able to generate the stack protector basic blocks late in
10089 codegen after the tail call decision has occurred.
10091 '``llvm.objectsize``' Intrinsic
10092 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10099 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>)
10100 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>)
10105 The ``llvm.objectsize`` intrinsic is designed to provide information to
10106 the optimizers to determine at compile time whether a) an operation
10107 (like memcpy) will overflow a buffer that corresponds to an object, or
10108 b) that a runtime check for overflow isn't necessary. An object in this
10109 context means an allocation of a specific class, structure, array, or
10115 The ``llvm.objectsize`` intrinsic takes two arguments. The first
10116 argument is a pointer to or into the ``object``. The second argument is
10117 a boolean and determines whether ``llvm.objectsize`` returns 0 (if true)
10118 or -1 (if false) when the object size is unknown. The second argument
10119 only accepts constants.
10124 The ``llvm.objectsize`` intrinsic is lowered to a constant representing
10125 the size of the object concerned. If the size cannot be determined at
10126 compile time, ``llvm.objectsize`` returns ``i32/i64 -1 or 0`` (depending
10127 on the ``min`` argument).
10129 '``llvm.expect``' Intrinsic
10130 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
10135 This is an overloaded intrinsic. You can use ``llvm.expect`` on any
10140 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>)
10141 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>)
10142 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>)
10147 The ``llvm.expect`` intrinsic provides information about expected (the
10148 most probable) value of ``val``, which can be used by optimizers.
10153 The ``llvm.expect`` intrinsic takes two arguments. The first argument is
10154 a value. The second argument is an expected value, this needs to be a
10155 constant value, variables are not allowed.
10160 This intrinsic is lowered to the ``val``.
10162 '``llvm.assume``' Intrinsic
10163 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10170 declare void @llvm.assume(i1 %cond)
10175 The ``llvm.assume`` allows the optimizer to assume that the provided
10176 condition is true. This information can then be used in simplifying other parts
10182 The condition which the optimizer may assume is always true.
10187 The intrinsic allows the optimizer to assume that the provided condition is
10188 always true whenever the control flow reaches the intrinsic call. No code is
10189 generated for this intrinsic, and instructions that contribute only to the
10190 provided condition are not used for code generation. If the condition is
10191 violated during execution, the behavior is undefined.
10193 Note that the optimizer might limit the transformations performed on values
10194 used by the ``llvm.assume`` intrinsic in order to preserve the instructions
10195 only used to form the intrinsic's input argument. This might prove undesirable
10196 if the extra information provided by the ``llvm.assume`` intrinsic does not cause
10197 sufficient overall improvement in code quality. For this reason,
10198 ``llvm.assume`` should not be used to document basic mathematical invariants
10199 that the optimizer can otherwise deduce or facts that are of little use to the
10204 '``llvm.bitset.test``' Intrinsic
10205 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10212 declare i1 @llvm.bitset.test(i8* %ptr, metadata %bitset) nounwind readnone
10218 The first argument is a pointer to be tested. The second argument is a
10219 metadata string containing the name of a :doc:`bitset <BitSets>`.
10224 The ``llvm.bitset.test`` intrinsic tests whether the given pointer is a
10225 member of the given bitset.
10227 '``llvm.donothing``' Intrinsic
10228 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10235 declare void @llvm.donothing() nounwind readnone
10240 The ``llvm.donothing`` intrinsic doesn't perform any operation. It's one of only
10241 two intrinsics (besides ``llvm.experimental.patchpoint``) that can be called
10242 with an invoke instruction.
10252 This intrinsic does nothing, and it's removed by optimizers and ignored
10255 Stack Map Intrinsics
10256 --------------------
10258 LLVM provides experimental intrinsics to support runtime patching
10259 mechanisms commonly desired in dynamic language JITs. These intrinsics
10260 are described in :doc:`StackMaps`.