1============================== 2LLVM Language Reference Manual 3============================== 4 5.. contents:: 6 :local: 7 :depth: 4 8 9Abstract 10======== 11 12This document is a reference manual for the LLVM assembly language. LLVM 13is a Static Single Assignment (SSA) based representation that provides 14type safety, low-level operations, flexibility, and the capability of 15representing 'all' high-level languages cleanly. It is the common code 16representation used throughout all phases of the LLVM compilation 17strategy. 18 19Introduction 20============ 21 22The LLVM code representation is designed to be used in three different 23forms: 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 25readable assembly language representation. This allows LLVM to provide a 26powerful intermediate representation for efficient compiler 27transformations and analysis, while providing a natural means to debug 28and visualize the transformations. The three different forms of LLVM are 29all equivalent. This document describes the human readable 30representation and notation. 31 32The LLVM representation aims to be light-weight and low-level while 33being expressive, typed, and extensible at the same time. It aims to be 34a "universal IR" of sorts, by being at a low enough level that 35high-level ideas may be cleanly mapped to it (similar to how 36microprocessors are "universal IR's", allowing many source languages to 37be mapped to them). By providing type information, LLVM can be used as 38the target of optimizations: for example, through pointer analysis, it 39can be proven that a C automatic variable is never accessed outside of 40the current function, allowing it to be promoted to a simple SSA value 41instead of a memory location. 42 43.. _wellformed: 44 45Well-Formedness 46--------------- 47 48It is important to note that this document describes 'well formed' LLVM 49assembly language. There is a difference between what the parser accepts 50and what is considered 'well formed'. For example, the following 51instruction is syntactically okay, but not well formed: 52 53.. code-block:: llvm 54 55 %x = add i32 1, %x 56 57because the definition of ``%x`` does not dominate all of its uses. The 58LLVM infrastructure provides a verification pass that may be used to 59verify that an LLVM module is well formed. This pass is automatically 60run by the parser after parsing input assembly and by the optimizer 61before it outputs bitcode. The violations pointed out by the verifier 62pass indicate bugs in transformation passes or input to the parser. 63 64.. _identifiers: 65 66Identifiers 67=========== 68 69LLVM identifiers come in two basic types: global and local. Global 70identifiers (functions, global variables) begin with the ``'@'`` 71character. Local identifiers (register names, types) begin with the 72``'%'`` character. Additionally, there are three different formats for 73identifiers, for different purposes: 74 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. 87 88LLVM requires that values start with a prefix for two reasons: Compilers 89don't need to worry about name clashes with reserved words, and the set 90of reserved words may be expanded in the future without penalty. 91Additionally, unnamed identifiers allow a compiler to quickly come up 92with a temporary variable without having to avoid symbol table 93conflicts. 94 95Reserved words in LLVM are very similar to reserved words in other 96languages. 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 99with variable names, because none of them start with a prefix character 100(``'%'`` or ``'@'``). 101 102Here is an example of LLVM code to multiply the integer variable 103'``%X``' by 8: 104 105The easy way: 106 107.. code-block:: llvm 108 109 %result = mul i32 %X, 8 110 111After strength reduction: 112 113.. code-block:: llvm 114 115 %result = shl i32 %X, 3 116 117And the hard way: 118 119.. code-block:: llvm 120 121 %0 = add i32 %X, %X ; yields i32:%0 122 %1 = add i32 %0, %0 ; yields i32:%1 123 %result = add i32 %1, %1 124 125This last way of multiplying ``%X`` by 8 illustrates several important 126lexical features of LLVM: 127 128#. 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. 136 137It also shows a convention that we follow in this document. When 138demonstrating instructions, we will follow an instruction with a comment 139that defines the type and name of value produced. 140 141High Level Structure 142==================== 143 144Module Structure 145---------------- 146 147LLVM programs are composed of ``Module``'s, each of which is a 148translation unit of the input programs. Each module consists of 149functions, global variables, and symbol table entries. Modules may be 150combined together with the LLVM linker, which merges function (and 151global variable) definitions, resolves forward declarations, and merges 152symbol table entries. Here is an example of the "hello world" module: 153 154.. code-block:: llvm 155 156 ; Declare the string constant as a global constant. 157 @.str = private unnamed_addr constant [13 x i8] c"hello world\0A\00" 158 159 ; External declaration of the puts function 160 declare i32 @puts(i8* nocapture) nounwind 161 162 ; 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 0 166 167 ; Call puts function to write out the string to stdout. 168 call i32 @puts(i8* %cast210) 169 ret i32 0 170 } 171 172 ; Named metadata 173 !0 = !{i32 42, null, !"string"} 174 !foo = !{!0} 175 176This 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``". 180 181In general, a module is made up of a list of global values (where both 182functions and global variables are global values). Global values are 183represented by a pointer to a memory location (in this case, a pointer 184to an array of char, and a pointer to a function), and have one of the 185following :ref:`linkage types <linkage>`. 186 187.. _linkage: 188 189Linkage Types 190------------- 191 192All Global Variables and Functions have one of the following types of 193linkage: 194 195``private`` 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. 202``internal`` 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, 214 not declarations. 215``linkonce`` 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. 228``weak`` 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" 232 in C source code. 233``common`` 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 241 common linkage. 242 243.. _linkage_appending: 244 245``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 251 .o files are linked. 252``extern_weak`` 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. 264``external`` 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. 268 269It is illegal for a function *declaration* to have any linkage type 270other than ``external`` or ``extern_weak``. 271 272.. _callingconv: 273 274Calling Conventions 275------------------- 276 277LLVM :ref:`functions <functionstructure>`, :ref:`calls <i_call>` and 278:ref:`invokes <i_invoke>` can all have an optional calling convention 279specified for the call. The calling convention of any pair of dynamic 280caller/callee must match, or the behavior of the program is undefined. 281The following calling conventions are supported by LLVM, and more may be 282added in the future: 283 284"``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 308 calls for inlining. 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 318 limitations: 319 320 - 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. 324 325 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 64 342 bit). 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. 364 365 - 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. 368 369 The idea behind this convention is to support calls to runtime functions 370 that have a hot path and a cold path. The hot path is usually a small piece 371 of code that doesn't many registers. The cold path might need to call out to 372 another function and therefore only needs to preserve the caller-saved 373 registers, which haven't already been saved by the caller. The 374 `PreserveMost` calling convention is very similar to the `cold` calling 375 convention in terms of caller/callee-saved registers, but they are used for 376 different types of function calls. `coldcc` is for function calls that are 377 rarely executed, whereas `preserve_mostcc` function calls are intended to be 378 on the hot path and definitely executed a lot. Furthermore `preserve_mostcc` 379 doesn't prevent the inliner from inlining the function call. 380 381 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 387 future. 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. 398 399 - 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). 402 403 The idea behind this convention is to support calls to runtime functions 404 that don't need to call out to any other functions. 405 406 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. 413 414More calling conventions can be added/defined on an as-needed basis, to 415support Pascal conventions or any other well-known target-independent 416convention. 417 418.. _visibilitystyles: 419 420Visibility Styles 421----------------- 422 423All Global Variables and Functions have one of the following visibility 424styles: 425 426"``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. 444 445A symbol with ``internal`` or ``private`` linkage must have ``default`` 446visibility. 447 448.. _dllstorageclass: 449 450DLL Storage Classes 451------------------- 452 453All Global Variables, Functions and Aliases can have one of the following 454DLL storage class: 455 456``dllimport`` 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. 461``dllexport`` 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. 468 469.. _tls_model: 470 471Thread Local Storage Models 472--------------------------- 473 474A variable may be defined as ``thread_local``, which means that it will 475not be shared by threads (each thread will have a separated copy of the 476variable). Not all targets support thread-local variables. Optionally, a 477TLS model may be specified: 478 479``localdynamic`` 480 For variables that are only used within the current shared library. 481``initialexec`` 482 For variables in modules that will not be loaded dynamically. 483``localexec`` 484 For variables defined in the executable and only used within it. 485 486If no explicit model is given, the "general dynamic" model is used. 487 488The models correspond to the ELF TLS models; see `ELF Handling For 489Thread-Local Storage <http://people.redhat.com/drepper/tls.pdf>`_ for 490more information on under which circumstances the different models may 491be used. The target may choose a different TLS model if the specified 492model is not supported, or if a better choice of model can be made. 493 494A model can also be specified in a alias, but then it only governs how 495the alias is accessed. It will not have any effect in the aliasee. 496 497.. _namedtypes: 498 499Structure Types 500--------------- 501 502LLVM IR allows you to specify both "identified" and "literal" :ref:`structure 503types <t_struct>`. Literal types are uniqued structurally, but identified types 504are never uniqued. An :ref:`opaque structural type <t_opaque>` can also be used 505to forward declare a type that is not yet available. 506 507An example of a identified structure specification is: 508 509.. code-block:: llvm 510 511 %mytype = type { %mytype*, i32 } 512 513Prior to the LLVM 3.0 release, identified types were structurally uniqued. Only 514literal types are uniqued in recent versions of LLVM. 515 516.. _globalvars: 517 518Global Variables 519---------------- 520 521Global variables define regions of memory allocated at compilation time 522instead of run-time. 523 524Global variables definitions must be initialized. 525 526Global variables in other translation units can also be declared, in which 527case they don't have an initializer. 528 529Either global variable definitions or declarations may have an explicit section 530to be placed in and may have an optional explicit alignment specified. 531 532A variable may be defined as a global ``constant``, which indicates that 533the contents of the variable will **never** be modified (enabling better 534optimization, allowing the global data to be placed in the read-only 535section of an executable, etc). Note that variables that need runtime 536initialization cannot be marked ``constant`` as there is a store to the 537variable. 538 539LLVM explicitly allows *declarations* of global variables to be marked 540constant, even if the final definition of the global is not. This 541capability can be used to enable slightly better optimization of the 542program, but requires the language definition to guarantee that 543optimizations based on the 'constantness' are valid for the translation 544units that do not include the definition. 545 546As SSA values, global variables define pointer values that are in scope 547(i.e. they dominate) all basic blocks in the program. Global variables 548always define a pointer to their "content" type because they describe a 549region of memory, and all memory objects in LLVM are accessed through 550pointers. 551 552Global variables can be marked with ``unnamed_addr`` which indicates 553that the address is not significant, only the content. Constants marked 554like this can be merged with other constants if they have the same 555initializer. Note that a constant with significant address *can* be 556merged with a ``unnamed_addr`` constant, the result being a constant 557whose address is significant. 558 559A global variable may be declared to reside in a target-specific 560numbered address space. For targets that support them, address spaces 561may affect how optimizations are performed and/or what target 562instructions are used to access the variable. The default address space 563is zero. The address space qualifier must precede any other attributes. 564 565LLVM allows an explicit section to be specified for globals. If the 566target supports it, it will emit globals to the section specified. 567Additionally, the global can placed in a comdat if the target has the necessary 568support. 569 570By default, global initializers are optimized by assuming that global 571variables defined within the module are not modified from their 572initial values before the start of the global initializer. This is 573true even for variables potentially accessible from outside the 574module, including those with external linkage or appearing in 575``@llvm.used`` or dllexported variables. This assumption may be suppressed 576by marking the variable with ``externally_initialized``. 577 578An explicit alignment may be specified for a global, which must be a 579power of 2. If not present, or if the alignment is set to zero, the 580alignment of the global is set by the target to whatever it feels 581convenient. If an explicit alignment is specified, the global is forced 582to have exactly that alignment. Targets and optimizers are not allowed 583to over-align the global if the global has an assigned section. In this 584case, the extra alignment could be observable: for example, code could 585assume that the globals are densely packed in their section and try to 586iterate over them as an array, alignment padding would break this 587iteration. The maximum alignment is ``1 << 29``. 588 589Globals can also have a :ref:`DLL storage class <dllstorageclass>`. 590 591Variables and aliasaes can have a 592:ref:`Thread Local Storage Model <tls_model>`. 593 594Syntax:: 595 596 [@<GlobalVarName> =] [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal] 597 [unnamed_addr] [AddrSpace] [ExternallyInitialized] 598 <global | constant> <Type> [<InitializerConstant>] 599 [, section "name"] [, comdat [($name)]] 600 [, align <Alignment>] 601 602For example, the following defines a global in a numbered address space 603with an initializer, section, and alignment: 604 605.. code-block:: llvm 606 607 @G = addrspace(5) constant float 1.0, section "foo", align 4 608 609The following example just declares a global variable 610 611.. code-block:: llvm 612 613 @G = external global i32 614 615The following example defines a thread-local global with the 616``initialexec`` TLS model: 617 618.. code-block:: llvm 619 620 @G = thread_local(initialexec) global i32 0, align 4 621 622.. _functionstructure: 623 624Functions 625--------- 626 627LLVM function definitions consist of the "``define``" keyword, an 628optional :ref:`linkage type <linkage>`, an optional :ref:`visibility 629style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`, 630an optional :ref:`calling convention <callingconv>`, 631an optional ``unnamed_addr`` attribute, a return type, an optional 632:ref:`parameter attribute <paramattrs>` for the return type, a function 633name, a (possibly empty) argument list (each with optional :ref:`parameter 634attributes <paramattrs>`), optional :ref:`function attributes <fnattrs>`, 635an optional section, an optional alignment, 636an optional :ref:`comdat <langref_comdats>`, 637an optional :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`, 638an optional :ref:`prologue <prologuedata>`, an opening 639curly brace, a list of basic blocks, and a closing curly brace. 640 641LLVM function declarations consist of the "``declare``" keyword, an 642optional :ref:`linkage type <linkage>`, an optional :ref:`visibility 643style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`, 644an optional :ref:`calling convention <callingconv>`, 645an optional ``unnamed_addr`` attribute, a return type, an optional 646:ref:`parameter attribute <paramattrs>` for the return type, a function 647name, a possibly empty list of arguments, an optional alignment, an optional 648:ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`, 649and an optional :ref:`prologue <prologuedata>`. 650 651A function definition contains a list of basic blocks, forming the CFG (Control 652Flow 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, 654and ends with a :ref:`terminator <terminators>` instruction (such as a branch or 655function return). If an explicit label is not provided, a block is assigned an 656implicit numbered label, using the next value from the same counter as used for 657unnamed temporaries (:ref:`see above<identifiers>`). For example, if a function 658entry block does not have an explicit label, it will be assigned label "%0", 659then the first unnamed temporary in that block will be "%1", etc. 660 661The first basic block in a function is special in two ways: it is 662immediately executed on entrance to the function, and it is not allowed 663to have predecessor basic blocks (i.e. there can not be any branches to 664the entry block of a function). Because the block can have no 665predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`. 666 667LLVM allows an explicit section to be specified for functions. If the 668target supports it, it will emit functions to the section specified. 669Additionally, the function can placed in a COMDAT. 670 671An explicit alignment may be specified for a function. If not present, 672or if the alignment is set to zero, the alignment of the function is set 673by the target to whatever it feels convenient. If an explicit alignment 674is specified, the function is forced to have at least that much 675alignment. All alignments must be a power of 2. 676 677If the ``unnamed_addr`` attribute is given, the address is know to not 678be significant and two identical functions can be merged. 679 680Syntax:: 681 682 define [linkage] [visibility] [DLLStorageClass] 683 [cconv] [ret attrs] 684 <ResultType> @<FunctionName> ([argument list]) 685 [unnamed_addr] [fn Attrs] [section "name"] [comdat [($name)]] 686 [align N] [gc] [prefix Constant] [prologue Constant] { ... } 687 688The argument list is a comma seperated sequence of arguments where each 689argument is of the following form 690 691Syntax:: 692 693 <type> [parameter Attrs] [name] 694 695 696.. _langref_aliases: 697 698Aliases 699------- 700 701Aliases, unlike function or variables, don't create any new data. They 702are just a new symbol and metadata for an existing position. 703 704Aliases have a name and an aliasee that is either a global value or a 705constant expression. 706 707Aliases 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>`. 710 711Syntax:: 712 713 @<Name> = [Linkage] [Visibility] [DLLStorageClass] [ThreadLocal] [unnamed_addr] alias <AliaseeTy> @<Aliasee> 714 715The linkage must be one of ``private``, ``internal``, ``linkonce``, ``weak``, 716``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers 717might not correctly handle dropping a weak symbol that is aliased. 718 719Alias that are not ``unnamed_addr`` are guaranteed to have the same address as 720the aliasee expression. ``unnamed_addr`` ones are only guaranteed to point 721to the same content. 722 723Since aliases are only a second name, some restrictions apply, of which 724some can only be checked when producing an object file: 725 726* The expression defining the aliasee must be computable at assembly 727 time. Since it is just a name, no relocations can be used. 728 729* No alias in the expression can be weak as the possibility of the 730 intermediate alias being overridden cannot be represented in an 731 object file. 732 733* No global value in the expression can be a declaration, since that 734 would require a relocation, which is not possible. 735 736.. _langref_comdats: 737 738Comdats 739------- 740 741Comdat IR provides access to COFF and ELF object file COMDAT functionality. 742 743Comdats have a name which represents the COMDAT key. All global objects that 744specify this key will only end up in the final object file if the linker chooses 745that key over some other key. Aliases are placed in the same COMDAT that their 746aliasee computes to, if any. 747 748Comdats have a selection kind to provide input on how the linker should 749choose between keys in two different object files. 750 751Syntax:: 752 753 $<Name> = comdat SelectionKind 754 755The selection kind must be one of the following: 756 757``any`` 758 The linker may choose any COMDAT key, the choice is arbitrary. 759``exactmatch`` 760 The linker may choose any COMDAT key but the sections must contain the 761 same data. 762``largest`` 763 The linker will choose the section containing the largest COMDAT key. 764``noduplicates`` 765 The linker requires that only section with this COMDAT key exist. 766``samesize`` 767 The linker may choose any COMDAT key but the sections must contain the 768 same amount of data. 769 770Note that the Mach-O platform doesn't support COMDATs and ELF only supports 771``any`` as a selection kind. 772 773Here is an example of a COMDAT group where a function will only be selected if 774the COMDAT key's section is the largest: 775 776.. code-block:: llvm 777 778 $foo = comdat largest 779 @foo = global i32 2, comdat($foo) 780 781 define void @bar() comdat($foo) { 782 ret void 783 } 784 785As a syntactic sugar the ``$name`` can be omitted if the name is the same as 786the global name: 787 788.. code-block:: llvm 789 790 $foo = comdat any 791 @foo = global i32 2, comdat 792 793 794In 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 796and another COMDAT section with selection kind 797``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT 798section and contains the contents of the ``@bar`` symbol. 799 800There are some restrictions on the properties of the global object. 801It, or an alias to it, must have the same name as the COMDAT group when 802targeting COFF. 803The contents and size of this object may be used during link-time to determine 804which COMDAT groups get selected depending on the selection kind. 805Because the name of the object must match the name of the COMDAT group, the 806linkage of the global object must not be local; local symbols can get renamed 807if a collision occurs in the symbol table. 808 809The combined use of COMDATS and section attributes may yield surprising results. 810For example: 811 812.. code-block:: llvm 813 814 $foo = comdat any 815 $bar = comdat any 816 @g1 = global i32 42, section "sec", comdat($foo) 817 @g2 = global i32 42, section "sec", comdat($bar) 818 819From the object file perspective, this requires the creation of two sections 820with the same name. This is necessary because both globals belong to different 821COMDAT groups and COMDATs, at the object file level, are represented by 822sections. 823 824Note that certain IR constructs like global variables and functions may create 825COMDATs in the object file in addition to any which are specified using COMDAT 826IR. This arises, for example, when a global variable has linkonce_odr linkage. 827 828.. _namedmetadatastructure: 829 830Named Metadata 831-------------- 832 833Named metadata is a collection of metadata. :ref:`Metadata 834nodes <metadata>` (but not metadata strings) are the only valid 835operands for a named metadata. 836 837Syntax:: 838 839 ; Some unnamed metadata nodes, which are referenced by the named metadata. 840 !0 = !{!"zero"} 841 !1 = !{!"one"} 842 !2 = !{!"two"} 843 ; A named metadata. 844 !name = !{!0, !1, !2} 845 846.. _paramattrs: 847 848Parameter Attributes 849-------------------- 850 851The return type and each parameter of a function type may have a set of 852*parameter attributes* associated with them. Parameter attributes are 853used to communicate additional information about the result or 854parameters of a function. Parameter attributes are considered to be part 855of the function, not of the function type, so functions with different 856parameter attributes can have the same function type. 857 858Parameter attributes are simple keywords that follow the type specified. 859If multiple parameter attributes are needed, they are space separated. 860For example: 861 862.. code-block:: llvm 863 864 declare i32 @printf(i8* noalias nocapture, ...) 865 declare i32 @atoi(i8 zeroext) 866 declare signext i8 @returns_signed_char() 867 868Note that any attributes for the function result (``nounwind``, 869``readonly``) come immediately after the argument list. 870 871Currently, only the following parameter attributes are defined: 872 873``zeroext`` 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). 878``signext`` 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). 883``inreg`` 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 889 target-specific. 890``byval`` 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 900 values. 901 902 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. 907 908.. _attr_inalloca: 909 910``inalloca`` 911 912 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. 918 919 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. 926 927 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 932 <int_stackrestore>`. 933 934 See :doc:`InAlloca` for more information on how to use this 935 attribute. 936 937``sret`` 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 944 values. 945 946``align <n>`` 947 This indicates that the pointer value may be assumed by the optimizer to 948 have the specified alignment. 949 950 Note that this attribute has additional semantics when combined with the 951 ``byval`` attribute. 952 953.. _noalias: 954 955``noalias`` 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>`. 964 965 Note that this definition of ``noalias`` is intentionally similar 966 to the definition of ``restrict`` in C99 for function arguments. 967 968 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. 975 976``nocapture`` 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. 980 981.. _nest: 982 983``nest`` 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. 987 988``returned`` 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. 996 997``nonnull`` 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 1002 is non-null. 1003 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). 1014 1015.. _gc: 1016 1017Garbage Collector Names 1018----------------------- 1019 1020Each function may specify a garbage collector name, which is simply a 1021string: 1022 1023.. code-block:: llvm 1024 1025 define void @f() gc "name" { ... } 1026 1027The compiler declares the supported values of *name*. Specifying a 1028collector will cause the compiler to alter its output in order to 1029support the named garbage collection algorithm. 1030 1031.. _prefixdata: 1032 1033Prefix Data 1034----------- 1035 1036Prefix data is data associated with a function which the code 1037generator will emit immediately before the function's entrypoint. 1038The purpose of this feature is to allow frontends to associate 1039language-specific runtime metadata with specific functions and make it 1040available through the function pointer while still allowing the 1041function pointer to be called. 1042 1043To access the data for a given function, a program may bitcast the 1044function pointer to a pointer to the constant's type and dereference 1045index -1. This implies that the IR symbol points just past the end of 1046the prefix data. For instance, take the example of a function annotated 1047with a single ``i32``, 1048 1049.. code-block:: llvm 1050 1051 define void @f() prefix i32 123 { ... } 1052 1053The prefix data can be referenced as, 1054 1055.. code-block:: llvm 1056 1057 %0 = bitcast *void () @f to *i32 1058 %a = getelementptr inbounds *i32 %0, i32 -1 1059 %b = load i32* %a 1060 1061Prefix data is laid out as if it were an initializer for a global variable 1062of the prefix data's type. The function will be placed such that the 1063beginning of the prefix data is aligned. This means that if the size 1064of the prefix data is not a multiple of the alignment size, the 1065function's entrypoint will not be aligned. If alignment of the 1066function's entrypoint is desired, padding must be added to the prefix 1067data. 1068 1069A function may have prefix data but no body. This has similar semantics 1070to the ``available_externally`` linkage in that the data may be used by the 1071optimizers but will not be emitted in the object file. 1072 1073.. _prologuedata: 1074 1075Prologue Data 1076------------- 1077 1078The ``prologue`` attribute allows arbitrary code (encoded as bytes) to 1079be inserted prior to the function body. This can be used for enabling 1080function hot-patching and instrumentation. 1081 1082To maintain the semantics of ordinary function calls, the prologue data must 1083have a particular format. Specifically, it must begin with a sequence of 1084bytes which decode to a sequence of machine instructions, valid for the 1085module's target, which transfer control to the point immediately succeeding 1086the prologue data, without performing any other visible action. This allows 1087the inliner and other passes to reason about the semantics of the function 1088definition without needing to reason about the prologue data. Obviously this 1089makes the format of the prologue data highly target dependent. 1090 1091A trivial example of valid prologue data for the x86 architecture is ``i8 144``, 1092which encodes the ``nop`` instruction: 1093 1094.. code-block:: llvm 1095 1096 define void @f() prologue i8 144 { ... } 1097 1098Generally prologue data can be formed by encoding a relative branch instruction 1099which skips the metadata, as in this example of valid prologue data for the 1100x86_64 architecture, where the first two bytes encode ``jmp .+10``: 1101 1102.. code-block:: llvm 1103 1104 %0 = type <{ i8, i8, i8* }> 1105 1106 define void @f() prologue %0 <{ i8 235, i8 8, i8* @md}> { ... } 1107 1108A function may have prologue data but no body. This has similar semantics 1109to the ``available_externally`` linkage in that the data may be used by the 1110optimizers but will not be emitted in the object file. 1111 1112.. _attrgrp: 1113 1114Attribute Groups 1115---------------- 1116 1117Attribute groups are groups of attributes that are referenced by objects within 1118the IR. They are important for keeping ``.ll`` files readable, because a lot of 1119functions will use the same set of attributes. In the degenerative case of a 1120``.ll`` file that corresponds to a single ``.c`` file, the single attribute 1121group will capture the important command line flags used to build that file. 1122 1123An attribute group is a module-level object. To use an attribute group, an 1124object references the attribute group's ID (e.g. ``#37``). An object may refer 1125to more than one attribute group. In that situation, the attributes from the 1126different groups are merged. 1127 1128Here is an example of attribute groups for a function that should always be 1129inlined, has a stack alignment of 4, and which shouldn't use SSE instructions: 1130 1131.. code-block:: llvm 1132 1133 ; Target-independent attributes: 1134 attributes #0 = { alwaysinline alignstack=4 } 1135 1136 ; Target-dependent attributes: 1137 attributes #1 = { "no-sse" } 1138 1139 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse". 1140 define void @f() #0 #1 { ... } 1141 1142.. _fnattrs: 1143 1144Function Attributes 1145------------------- 1146 1147Function attributes are set to communicate additional information about 1148a function. Function attributes are considered to be part of the 1149function, not of the function type, so functions with different function 1150attributes can have the same function type. 1151 1152Function attributes are simple keywords that follow the type specified. 1153If multiple attributes are needed, they are space separated. For 1154example: 1155 1156.. code-block:: llvm 1157 1158 define void @f() noinline { ... } 1159 define void @f() alwaysinline { ... } 1160 define void @f() alwaysinline optsize { ... } 1161 define void @f() optsize { ... } 1162 1163``alignstack(<n>)`` 1164 This attribute indicates that, when emitting the prologue and 1165 epilogue, the backend should forcibly align the stack pointer. 1166 Specify the desired alignment, which must be a power of two, in 1167 parentheses. 1168``alwaysinline`` 1169 This attribute indicates that the inliner should attempt to inline 1170 this function into callers whenever possible, ignoring any active 1171 inlining size threshold for this caller. 1172``builtin`` 1173 This indicates that the callee function at a call site should be 1174 recognized as a built-in function, even though the function's declaration 1175 uses the ``nobuiltin`` attribute. This is only valid at call sites for 1176 direct calls to functions that are declared with the ``nobuiltin`` 1177 attribute. 1178``cold`` 1179 This attribute indicates that this function is rarely called. When 1180 computing edge weights, basic blocks post-dominated by a cold 1181 function call are also considered to be cold; and, thus, given low 1182 weight. 1183``inlinehint`` 1184 This attribute indicates that the source code contained a hint that 1185 inlining this function is desirable (such as the "inline" keyword in 1186 C/C++). It is just a hint; it imposes no requirements on the 1187 inliner. 1188``jumptable`` 1189 This attribute indicates that the function should be added to a 1190 jump-instruction table at code-generation time, and that all address-taken 1191 references to this function should be replaced with a reference to the 1192 appropriate jump-instruction-table function pointer. Note that this creates 1193 a new pointer for the original function, which means that code that depends 1194 on function-pointer identity can break. So, any function annotated with 1195 ``jumptable`` must also be ``unnamed_addr``. 1196``minsize`` 1197 This attribute suggests that optimization passes and code generator 1198 passes make choices that keep the code size of this function as small 1199 as possible and perform optimizations that may sacrifice runtime 1200 performance in order to minimize the size of the generated code. 1201``naked`` 1202 This attribute disables prologue / epilogue emission for the 1203 function. This can have very system-specific consequences. 1204``nobuiltin`` 1205 This indicates that the callee function at a call site is not recognized as 1206 a built-in function. LLVM will retain the original call and not replace it 1207 with equivalent code based on the semantics of the built-in function, unless 1208 the call site uses the ``builtin`` attribute. This is valid at call sites 1209 and on function declarations and definitions. 1210``noduplicate`` 1211 This attribute indicates that calls to the function cannot be 1212 duplicated. A call to a ``noduplicate`` function may be moved 1213 within its parent function, but may not be duplicated within 1214 its parent function. 1215 1216 A function containing a ``noduplicate`` call may still 1217 be an inlining candidate, provided that the call is not 1218 duplicated by inlining. That implies that the function has 1219 internal linkage and only has one call site, so the original 1220 call is dead after inlining. 1221``noimplicitfloat`` 1222 This attributes disables implicit floating point instructions. 1223``noinline`` 1224 This attribute indicates that the inliner should never inline this 1225 function in any situation. This attribute may not be used together 1226 with the ``alwaysinline`` attribute. 1227``nonlazybind`` 1228 This attribute suppresses lazy symbol binding for the function. This 1229 may make calls to the function faster, at the cost of extra program 1230 startup time if the function is not called during program startup. 1231``noredzone`` 1232 This attribute indicates that the code generator should not use a 1233 red zone, even if the target-specific ABI normally permits it. 1234``noreturn`` 1235 This function attribute indicates that the function never returns 1236 normally. This produces undefined behavior at runtime if the 1237 function ever does dynamically return. 1238``nounwind`` 1239 This function attribute indicates that the function never returns 1240 with an unwind or exceptional control flow. If the function does 1241 unwind, its runtime behavior is undefined. 1242``optnone`` 1243 This function attribute indicates that the function is not optimized 1244 by any optimization or code generator passes with the 1245 exception of interprocedural optimization passes. 1246 This attribute cannot be used together with the ``alwaysinline`` 1247 attribute; this attribute is also incompatible 1248 with the ``minsize`` attribute and the ``optsize`` attribute. 1249 1250 This attribute requires the ``noinline`` attribute to be specified on 1251 the function as well, so the function is never inlined into any caller. 1252 Only functions with the ``alwaysinline`` attribute are valid 1253 candidates for inlining into the body of this function. 1254``optsize`` 1255 This attribute suggests that optimization passes and code generator 1256 passes make choices that keep the code size of this function low, 1257 and otherwise do optimizations specifically to reduce code size as 1258 long as they do not significantly impact runtime performance. 1259``readnone`` 1260 On a function, this attribute indicates that the function computes its 1261 result (or decides to unwind an exception) based strictly on its arguments, 1262 without dereferencing any pointer arguments or otherwise accessing 1263 any mutable state (e.g. memory, control registers, etc) visible to 1264 caller functions. It does not write through any pointer arguments 1265 (including ``byval`` arguments) and never changes any state visible 1266 to callers. This means that it cannot unwind exceptions by calling 1267 the ``C++`` exception throwing methods. 1268 1269 On an argument, this attribute indicates that the function does not 1270 dereference that pointer argument, even though it may read or write the 1271 memory that the pointer points to if accessed through other pointers. 1272``readonly`` 1273 On a function, this attribute indicates that the function does not write 1274 through any pointer arguments (including ``byval`` arguments) or otherwise 1275 modify any state (e.g. memory, control registers, etc) visible to 1276 caller functions. It may dereference pointer arguments and read 1277 state that may be set in the caller. A readonly function always 1278 returns the same value (or unwinds an exception identically) when 1279 called with the same set of arguments and global state. It cannot 1280 unwind an exception by calling the ``C++`` exception throwing 1281 methods. 1282 1283 On an argument, this attribute indicates that the function does not write 1284 through this pointer argument, even though it may write to the memory that 1285 the pointer points to. 1286``returns_twice`` 1287 This attribute indicates that this function can return twice. The C 1288 ``setjmp`` is an example of such a function. The compiler disables 1289 some optimizations (like tail calls) in the caller of these 1290 functions. 1291``sanitize_address`` 1292 This attribute indicates that AddressSanitizer checks 1293 (dynamic address safety analysis) are enabled for this function. 1294``sanitize_memory`` 1295 This attribute indicates that MemorySanitizer checks (dynamic detection 1296 of accesses to uninitialized memory) are enabled for this function. 1297``sanitize_thread`` 1298 This attribute indicates that ThreadSanitizer checks 1299 (dynamic thread safety analysis) are enabled for this function. 1300``ssp`` 1301 This attribute indicates that the function should emit a stack 1302 smashing protector. It is in the form of a "canary" --- a random value 1303 placed on the stack before the local variables that's checked upon 1304 return from the function to see if it has been overwritten. A 1305 heuristic is used to determine if a function needs stack protectors 1306 or not. The heuristic used will enable protectors for functions with: 1307 1308 - Character arrays larger than ``ssp-buffer-size`` (default 8). 1309 - Aggregates containing character arrays larger than ``ssp-buffer-size``. 1310 - Calls to alloca() with variable sizes or constant sizes greater than 1311 ``ssp-buffer-size``. 1312 1313 Variables that are identified as requiring a protector will be arranged 1314 on the stack such that they are adjacent to the stack protector guard. 1315 1316 If a function that has an ``ssp`` attribute is inlined into a 1317 function that doesn't have an ``ssp`` attribute, then the resulting 1318 function will have an ``ssp`` attribute. 1319``sspreq`` 1320 This attribute indicates that the function should *always* emit a 1321 stack smashing protector. This overrides the ``ssp`` function 1322 attribute. 1323 1324 Variables that are identified as requiring a protector will be arranged 1325 on the stack such that they are adjacent to the stack protector guard. 1326 The specific layout rules are: 1327 1328 #. Large arrays and structures containing large arrays 1329 (``>= ssp-buffer-size``) are closest to the stack protector. 1330 #. Small arrays and structures containing small arrays 1331 (``< ssp-buffer-size``) are 2nd closest to the protector. 1332 #. Variables that have had their address taken are 3rd closest to the 1333 protector. 1334 1335 If a function that has an ``sspreq`` attribute is inlined into a 1336 function that doesn't have an ``sspreq`` attribute or which has an 1337 ``ssp`` or ``sspstrong`` attribute, then the resulting function will have 1338 an ``sspreq`` attribute. 1339``sspstrong`` 1340 This attribute indicates that the function should emit a stack smashing 1341 protector. This attribute causes a strong heuristic to be used when 1342 determining if a function needs stack protectors. The strong heuristic 1343 will enable protectors for functions with: 1344 1345 - Arrays of any size and type 1346 - Aggregates containing an array of any size and type. 1347 - Calls to alloca(). 1348 - Local variables that have had their address taken. 1349 1350 Variables that are identified as requiring a protector will be arranged 1351 on the stack such that they are adjacent to the stack protector guard. 1352 The specific layout rules are: 1353 1354 #. Large arrays and structures containing large arrays 1355 (``>= ssp-buffer-size``) are closest to the stack protector. 1356 #. Small arrays and structures containing small arrays 1357 (``< ssp-buffer-size``) are 2nd closest to the protector. 1358 #. Variables that have had their address taken are 3rd closest to the 1359 protector. 1360 1361 This overrides the ``ssp`` function attribute. 1362 1363 If a function that has an ``sspstrong`` attribute is inlined into a 1364 function that doesn't have an ``sspstrong`` attribute, then the 1365 resulting function will have an ``sspstrong`` attribute. 1366``uwtable`` 1367 This attribute indicates that the ABI being targeted requires that 1368 an unwind table entry be produce for this function even if we can 1369 show that no exceptions passes by it. This is normally the case for 1370 the ELF x86-64 abi, but it can be disabled for some compilation 1371 units. 1372 1373.. _moduleasm: 1374 1375Module-Level Inline Assembly 1376---------------------------- 1377 1378Modules may contain "module-level inline asm" blocks, which corresponds 1379to the GCC "file scope inline asm" blocks. These blocks are internally 1380concatenated by LLVM and treated as a single unit, but may be separated 1381in the ``.ll`` file if desired. The syntax is very simple: 1382 1383.. code-block:: llvm 1384 1385 module asm "inline asm code goes here" 1386 module asm "more can go here" 1387 1388The strings can contain any character by escaping non-printable 1389characters. The escape sequence used is simply "\\xx" where "xx" is the 1390two digit hex code for the number. 1391 1392The inline asm code is simply printed to the machine code .s file when 1393assembly code is generated. 1394 1395.. _langref_datalayout: 1396 1397Data Layout 1398----------- 1399 1400A module may specify a target specific data layout string that specifies 1401how data is to be laid out in memory. The syntax for the data layout is 1402simply: 1403 1404.. code-block:: llvm 1405 1406 target datalayout = "layout specification" 1407 1408The *layout specification* consists of a list of specifications 1409separated by the minus sign character ('-'). Each specification starts 1410with a letter and may include other information after the letter to 1411define some aspect of the data layout. The specifications accepted are 1412as follows: 1413 1414``E`` 1415 Specifies that the target lays out data in big-endian form. That is, 1416 the bits with the most significance have the lowest address 1417 location. 1418``e`` 1419 Specifies that the target lays out data in little-endian form. That 1420 is, the bits with the least significance have the lowest address 1421 location. 1422``S<size>`` 1423 Specifies the natural alignment of the stack in bits. Alignment 1424 promotion of stack variables is limited to the natural stack 1425 alignment to avoid dynamic stack realignment. The stack alignment 1426 must be a multiple of 8-bits. If omitted, the natural stack 1427 alignment defaults to "unspecified", which does not prevent any 1428 alignment promotions. 1429``p[n]:<size>:<abi>:<pref>`` 1430 This specifies the *size* of a pointer and its ``<abi>`` and 1431 ``<pref>``\erred alignments for address space ``n``. All sizes are in 1432 bits. The address space, ``n`` is optional, and if not specified, 1433 denotes the default address space 0. The value of ``n`` must be 1434 in the range [1,2^23). 1435``i<size>:<abi>:<pref>`` 1436 This specifies the alignment for an integer type of a given bit 1437 ``<size>``. The value of ``<size>`` must be in the range [1,2^23). 1438``v<size>:<abi>:<pref>`` 1439 This specifies the alignment for a vector type of a given bit 1440 ``<size>``. 1441``f<size>:<abi>:<pref>`` 1442 This specifies the alignment for a floating point type of a given bit 1443 ``<size>``. Only values of ``<size>`` that are supported by the target 1444 will work. 32 (float) and 64 (double) are supported on all targets; 80 1445 or 128 (different flavors of long double) are also supported on some 1446 targets. 1447``a:<abi>:<pref>`` 1448 This specifies the alignment for an object of aggregate type. 1449``m:<mangling>`` 1450 If present, specifies that llvm names are mangled in the output. The 1451 options are 1452 1453 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix. 1454 * ``m``: Mips mangling: Private symbols get a ``$`` prefix. 1455 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other 1456 symbols get a ``_`` prefix. 1457 * ``w``: Windows COFF prefix: Similar to Mach-O, but stdcall and fastcall 1458 functions also get a suffix based on the frame size. 1459``n<size1>:<size2>:<size3>...`` 1460 This specifies a set of native integer widths for the target CPU in 1461 bits. For example, it might contain ``n32`` for 32-bit PowerPC, 1462 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of 1463 this set are considered to support most general arithmetic operations 1464 efficiently. 1465 1466On every specification that takes a ``<abi>:<pref>``, specifying the 1467``<pref>`` alignment is optional. If omitted, the preceding ``:`` 1468should be omitted too and ``<pref>`` will be equal to ``<abi>``. 1469 1470When constructing the data layout for a given target, LLVM starts with a 1471default set of specifications which are then (possibly) overridden by 1472the specifications in the ``datalayout`` keyword. The default 1473specifications are given in this list: 1474 1475- ``E`` - big endian 1476- ``p:64:64:64`` - 64-bit pointers with 64-bit alignment. 1477- ``p[n]:64:64:64`` - Other address spaces are assumed to be the 1478 same as the default address space. 1479- ``S0`` - natural stack alignment is unspecified 1480- ``i1:8:8`` - i1 is 8-bit (byte) aligned 1481- ``i8:8:8`` - i8 is 8-bit (byte) aligned 1482- ``i16:16:16`` - i16 is 16-bit aligned 1483- ``i32:32:32`` - i32 is 32-bit aligned 1484- ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred 1485 alignment of 64-bits 1486- ``f16:16:16`` - half is 16-bit aligned 1487- ``f32:32:32`` - float is 32-bit aligned 1488- ``f64:64:64`` - double is 64-bit aligned 1489- ``f128:128:128`` - quad is 128-bit aligned 1490- ``v64:64:64`` - 64-bit vector is 64-bit aligned 1491- ``v128:128:128`` - 128-bit vector is 128-bit aligned 1492- ``a:0:64`` - aggregates are 64-bit aligned 1493 1494When LLVM is determining the alignment for a given type, it uses the 1495following rules: 1496 1497#. If the type sought is an exact match for one of the specifications, 1498 that specification is used. 1499#. If no match is found, and the type sought is an integer type, then 1500 the smallest integer type that is larger than the bitwidth of the 1501 sought type is used. If none of the specifications are larger than 1502 the bitwidth then the largest integer type is used. For example, 1503 given the default specifications above, the i7 type will use the 1504 alignment of i8 (next largest) while both i65 and i256 will use the 1505 alignment of i64 (largest specified). 1506#. If no match is found, and the type sought is a vector type, then the 1507 largest vector type that is smaller than the sought vector type will 1508 be used as a fall back. This happens because <128 x double> can be 1509 implemented in terms of 64 <2 x double>, for example. 1510 1511The function of the data layout string may not be what you expect. 1512Notably, this is not a specification from the frontend of what alignment 1513the code generator should use. 1514 1515Instead, if specified, the target data layout is required to match what 1516the ultimate *code generator* expects. This string is used by the 1517mid-level optimizers to improve code, and this only works if it matches 1518what the ultimate code generator uses. If you would like to generate IR 1519that does not embed this target-specific detail into the IR, then you 1520don't have to specify the string. This will disable some optimizations 1521that require precise layout information, but this also prevents those 1522optimizations from introducing target specificity into the IR. 1523 1524.. _langref_triple: 1525 1526Target Triple 1527------------- 1528 1529A module may specify a target triple string that describes the target 1530host. The syntax for the target triple is simply: 1531 1532.. code-block:: llvm 1533 1534 target triple = "x86_64-apple-macosx10.7.0" 1535 1536The *target triple* string consists of a series of identifiers delimited 1537by the minus sign character ('-'). The canonical forms are: 1538 1539:: 1540 1541 ARCHITECTURE-VENDOR-OPERATING_SYSTEM 1542 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT 1543 1544This information is passed along to the backend so that it generates 1545code for the proper architecture. It's possible to override this on the 1546command line with the ``-mtriple`` command line option. 1547 1548.. _pointeraliasing: 1549 1550Pointer Aliasing Rules 1551---------------------- 1552 1553Any memory access must be done through a pointer value associated with 1554an address range of the memory access, otherwise the behavior is 1555undefined. Pointer values are associated with address ranges according 1556to the following rules: 1557 1558- A pointer value is associated with the addresses associated with any 1559 value it is *based* on. 1560- An address of a global variable is associated with the address range 1561 of the variable's storage. 1562- The result value of an allocation instruction is associated with the 1563 address range of the allocated storage. 1564- A null pointer in the default address-space is associated with no 1565 address. 1566- An integer constant other than zero or a pointer value returned from 1567 a function not defined within LLVM may be associated with address 1568 ranges allocated through mechanisms other than those provided by 1569 LLVM. Such ranges shall not overlap with any ranges of addresses 1570 allocated by mechanisms provided by LLVM. 1571 1572A pointer value is *based* on another pointer value according to the 1573following rules: 1574 1575- A pointer value formed from a ``getelementptr`` operation is *based* 1576 on the first operand of the ``getelementptr``. 1577- The result value of a ``bitcast`` is *based* on the operand of the 1578 ``bitcast``. 1579- A pointer value formed by an ``inttoptr`` is *based* on all pointer 1580 values that contribute (directly or indirectly) to the computation of 1581 the pointer's value. 1582- The "*based* on" relationship is transitive. 1583 1584Note that this definition of *"based"* is intentionally similar to the 1585definition of *"based"* in C99, though it is slightly weaker. 1586 1587LLVM IR does not associate types with memory. The result type of a 1588``load`` merely indicates the size and alignment of the memory from 1589which to load, as well as the interpretation of the value. The first 1590operand type of a ``store`` similarly only indicates the size and 1591alignment of the store. 1592 1593Consequently, type-based alias analysis, aka TBAA, aka 1594``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR. 1595:ref:`Metadata <metadata>` may be used to encode additional information 1596which specialized optimization passes may use to implement type-based 1597alias analysis. 1598 1599.. _volatile: 1600 1601Volatile Memory Accesses 1602------------------------ 1603 1604Certain memory accesses, such as :ref:`load <i_load>`'s, 1605:ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be 1606marked ``volatile``. The optimizers must not change the number of 1607volatile operations or change their order of execution relative to other 1608volatile operations. The optimizers *may* change the order of volatile 1609operations relative to non-volatile operations. This is not Java's 1610"volatile" and has no cross-thread synchronization behavior. 1611 1612IR-level volatile loads and stores cannot safely be optimized into 1613llvm.memcpy or llvm.memmove intrinsics even when those intrinsics are 1614flagged volatile. Likewise, the backend should never split or merge 1615target-legal volatile load/store instructions. 1616 1617.. admonition:: Rationale 1618 1619 Platforms may rely on volatile loads and stores of natively supported 1620 data width to be executed as single instruction. For example, in C 1621 this holds for an l-value of volatile primitive type with native 1622 hardware support, but not necessarily for aggregate types. The 1623 frontend upholds these expectations, which are intentionally 1624 unspecified in the IR. The rules above ensure that IR transformation 1625 do not violate the frontend's contract with the language. 1626 1627.. _memmodel: 1628 1629Memory Model for Concurrent Operations 1630-------------------------------------- 1631 1632The LLVM IR does not define any way to start parallel threads of 1633execution or to register signal handlers. Nonetheless, there are 1634platform-specific ways to create them, and we define LLVM IR's behavior 1635in their presence. This model is inspired by the C++0x memory model. 1636 1637For a more informal introduction to this model, see the :doc:`Atomics`. 1638 1639We define a *happens-before* partial order as the least partial order 1640that 1641 1642- Is a superset of single-thread program order, and 1643- When a *synchronizes-with* ``b``, includes an edge from ``a`` to 1644 ``b``. *Synchronizes-with* pairs are introduced by platform-specific 1645 techniques, like pthread locks, thread creation, thread joining, 1646 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering 1647 Constraints <ordering>`). 1648 1649Note that program order does not introduce *happens-before* edges 1650between a thread and signals executing inside that thread. 1651 1652Every (defined) read operation (load instructions, memcpy, atomic 1653loads/read-modify-writes, etc.) R reads a series of bytes written by 1654(defined) write operations (store instructions, atomic 1655stores/read-modify-writes, memcpy, etc.). For the purposes of this 1656section, initialized globals are considered to have a write of the 1657initializer which is atomic and happens before any other read or write 1658of the memory in question. For each byte of a read R, R\ :sub:`byte` 1659may see any write to the same byte, except: 1660 1661- If write\ :sub:`1` happens before write\ :sub:`2`, and 1662 write\ :sub:`2` happens before R\ :sub:`byte`, then 1663 R\ :sub:`byte` does not see write\ :sub:`1`. 1664- If R\ :sub:`byte` happens before write\ :sub:`3`, then 1665 R\ :sub:`byte` does not see write\ :sub:`3`. 1666 1667Given that definition, R\ :sub:`byte` is defined as follows: 1668 1669- If R is volatile, the result is target-dependent. (Volatile is 1670 supposed to give guarantees which can support ``sig_atomic_t`` in 1671 C/C++, and may be used for accesses to addresses that do not behave 1672 like normal memory. It does not generally provide cross-thread 1673 synchronization.) 1674- Otherwise, if there is no write to the same byte that happens before 1675 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte. 1676- Otherwise, if R\ :sub:`byte` may see exactly one write, 1677 R\ :sub:`byte` returns the value written by that write. 1678- Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may 1679 see are atomic, it chooses one of the values written. See the :ref:`Atomic 1680 Memory Ordering Constraints <ordering>` section for additional 1681 constraints on how the choice is made. 1682- Otherwise R\ :sub:`byte` returns ``undef``. 1683 1684R returns the value composed of the series of bytes it read. This 1685implies that some bytes within the value may be ``undef`` **without** 1686the entire value being ``undef``. Note that this only defines the 1687semantics of the operation; it doesn't mean that targets will emit more 1688than one instruction to read the series of bytes. 1689 1690Note that in cases where none of the atomic intrinsics are used, this 1691model places only one restriction on IR transformations on top of what 1692is required for single-threaded execution: introducing a store to a byte 1693which might not otherwise be stored is not allowed in general. 1694(Specifically, in the case where another thread might write to and read 1695from an address, introducing a store can change a load that may see 1696exactly one write into a load that may see multiple writes.) 1697 1698.. _ordering: 1699 1700Atomic Memory Ordering Constraints 1701---------------------------------- 1702 1703Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`, 1704:ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`, 1705:ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take 1706ordering parameters that determine which other atomic instructions on 1707the same address they *synchronize with*. These semantics are borrowed 1708from Java and C++0x, but are somewhat more colloquial. If these 1709descriptions aren't precise enough, check those specs (see spec 1710references in the :doc:`atomics guide <Atomics>`). 1711:ref:`fence <i_fence>` instructions treat these orderings somewhat 1712differently since they don't take an address. See that instruction's 1713documentation for details. 1714 1715For a simpler introduction to the ordering constraints, see the 1716:doc:`Atomics`. 1717 1718``unordered`` 1719 The set of values that can be read is governed by the happens-before 1720 partial order. A value cannot be read unless some operation wrote 1721 it. This is intended to provide a guarantee strong enough to model 1722 Java's non-volatile shared variables. This ordering cannot be 1723 specified for read-modify-write operations; it is not strong enough 1724 to make them atomic in any interesting way. 1725``monotonic`` 1726 In addition to the guarantees of ``unordered``, there is a single 1727 total order for modifications by ``monotonic`` operations on each 1728 address. All modification orders must be compatible with the 1729 happens-before order. There is no guarantee that the modification 1730 orders can be combined to a global total order for the whole program 1731 (and this often will not be possible). The read in an atomic 1732 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and 1733 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification 1734 order immediately before the value it writes. If one atomic read 1735 happens before another atomic read of the same address, the later 1736 read must see the same value or a later value in the address's 1737 modification order. This disallows reordering of ``monotonic`` (or 1738 stronger) operations on the same address. If an address is written 1739 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally 1740 read that address repeatedly, the other threads must eventually see 1741 the write. This corresponds to the C++0x/C1x 1742 ``memory_order_relaxed``. 1743``acquire`` 1744 In addition to the guarantees of ``monotonic``, a 1745 *synchronizes-with* edge may be formed with a ``release`` operation. 1746 This is intended to model C++'s ``memory_order_acquire``. 1747``release`` 1748 In addition to the guarantees of ``monotonic``, if this operation 1749 writes a value which is subsequently read by an ``acquire`` 1750 operation, it *synchronizes-with* that operation. (This isn't a 1751 complete description; see the C++0x definition of a release 1752 sequence.) This corresponds to the C++0x/C1x 1753 ``memory_order_release``. 1754``acq_rel`` (acquire+release) 1755 Acts as both an ``acquire`` and ``release`` operation on its 1756 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``. 1757``seq_cst`` (sequentially consistent) 1758 In addition to the guarantees of ``acq_rel`` (``acquire`` for an 1759 operation that only reads, ``release`` for an operation that only 1760 writes), there is a global total order on all 1761 sequentially-consistent operations on all addresses, which is 1762 consistent with the *happens-before* partial order and with the 1763 modification orders of all the affected addresses. Each 1764 sequentially-consistent read sees the last preceding write to the 1765 same address in this global order. This corresponds to the C++0x/C1x 1766 ``memory_order_seq_cst`` and Java volatile. 1767 1768.. _singlethread: 1769 1770If an atomic operation is marked ``singlethread``, it only *synchronizes 1771with* or participates in modification and seq\_cst total orderings with 1772other operations running in the same thread (for example, in signal 1773handlers). 1774 1775.. _fastmath: 1776 1777Fast-Math Flags 1778--------------- 1779 1780LLVM IR floating-point binary ops (:ref:`fadd <i_fadd>`, 1781:ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`, 1782:ref:`frem <i_frem>`) have the following flags that can set to enable 1783otherwise unsafe floating point operations 1784 1785``nnan`` 1786 No NaNs - Allow optimizations to assume the arguments and result are not 1787 NaN. Such optimizations are required to retain defined behavior over 1788 NaNs, but the value of the result is undefined. 1789 1790``ninf`` 1791 No Infs - Allow optimizations to assume the arguments and result are not 1792 +/-Inf. Such optimizations are required to retain defined behavior over 1793 +/-Inf, but the value of the result is undefined. 1794 1795``nsz`` 1796 No Signed Zeros - Allow optimizations to treat the sign of a zero 1797 argument or result as insignificant. 1798 1799``arcp`` 1800 Allow Reciprocal - Allow optimizations to use the reciprocal of an 1801 argument rather than perform division. 1802 1803``fast`` 1804 Fast - Allow algebraically equivalent transformations that may 1805 dramatically change results in floating point (e.g. reassociate). This 1806 flag implies all the others. 1807 1808.. _uselistorder: 1809 1810Use-list Order Directives 1811------------------------- 1812 1813Use-list directives encode the in-memory order of each use-list, allowing the 1814order to be recreated. ``<order-indexes>`` is a comma-separated list of 1815indexes that are assigned to the referenced value's uses. The referenced 1816value's use-list is immediately sorted by these indexes. 1817 1818Use-list directives may appear at function scope or global scope. They are not 1819instructions, and have no effect on the semantics of the IR. When they're at 1820function scope, they must appear after the terminator of the final basic block. 1821 1822If basic blocks have their address taken via ``blockaddress()`` expressions, 1823``uselistorder_bb`` can be used to reorder their use-lists from outside their 1824function's scope. 1825 1826:Syntax: 1827 1828:: 1829 1830 uselistorder <ty> <value>, { <order-indexes> } 1831 uselistorder_bb @function, %block { <order-indexes> } 1832 1833:Examples: 1834 1835:: 1836 1837 define void @foo(i32 %arg1, i32 %arg2) { 1838 entry: 1839 ; ... instructions ... 1840 bb: 1841 ; ... instructions ... 1842 1843 ; At function scope. 1844 uselistorder i32 %arg1, { 1, 0, 2 } 1845 uselistorder label %bb, { 1, 0 } 1846 } 1847 1848 ; At global scope. 1849 uselistorder i32* @global, { 1, 2, 0 } 1850 uselistorder i32 7, { 1, 0 } 1851 uselistorder i32 (i32) @bar, { 1, 0 } 1852 uselistorder_bb @foo, %bb, { 5, 1, 3, 2, 0, 4 } 1853 1854.. _typesystem: 1855 1856Type System 1857=========== 1858 1859The LLVM type system is one of the most important features of the 1860intermediate representation. Being typed enables a number of 1861optimizations to be performed on the intermediate representation 1862directly, without having to do extra analyses on the side before the 1863transformation. A strong type system makes it easier to read the 1864generated code and enables novel analyses and transformations that are 1865not feasible to perform on normal three address code representations. 1866 1867.. _t_void: 1868 1869Void Type 1870--------- 1871 1872:Overview: 1873 1874 1875The void type does not represent any value and has no size. 1876 1877:Syntax: 1878 1879 1880:: 1881 1882 void 1883 1884 1885.. _t_function: 1886 1887Function Type 1888------------- 1889 1890:Overview: 1891 1892 1893The function type can be thought of as a function signature. It consists of a 1894return type and a list of formal parameter types. The return type of a function 1895type is a void type or first class type --- except for :ref:`label <t_label>` 1896and :ref:`metadata <t_metadata>` types. 1897 1898:Syntax: 1899 1900:: 1901 1902 <returntype> (<parameter list>) 1903 1904...where '``<parameter list>``' is a comma-separated list of type 1905specifiers. Optionally, the parameter list may include a type ``...``, which 1906indicates that the function takes a variable number of arguments. Variable 1907argument functions can access their arguments with the :ref:`variable argument 1908handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type 1909except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`. 1910 1911:Examples: 1912 1913+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1914| ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` | 1915+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1916| ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. | 1917+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1918| ``i32 (i8*, ...)`` | A vararg function that takes at least one :ref:`pointer <t_pointer>` to ``i8`` (char in C), which returns an integer. This is the signature for ``printf`` in LLVM. | 1919+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1920| ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values | 1921+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 1922 1923.. _t_firstclass: 1924 1925First Class Types 1926----------------- 1927 1928The :ref:`first class <t_firstclass>` types are perhaps the most important. 1929Values of these types are the only ones which can be produced by 1930instructions. 1931 1932.. _t_single_value: 1933 1934Single Value Types 1935^^^^^^^^^^^^^^^^^^ 1936 1937These are the types that are valid in registers from CodeGen's perspective. 1938 1939.. _t_integer: 1940 1941Integer Type 1942"""""""""""" 1943 1944:Overview: 1945 1946The integer type is a very simple type that simply specifies an 1947arbitrary bit width for the integer type desired. Any bit width from 1 1948bit to 2\ :sup:`23`\ -1 (about 8 million) can be specified. 1949 1950:Syntax: 1951 1952:: 1953 1954 iN 1955 1956The number of bits the integer will occupy is specified by the ``N`` 1957value. 1958 1959Examples: 1960********* 1961 1962+----------------+------------------------------------------------+ 1963| ``i1`` | a single-bit integer. | 1964+----------------+------------------------------------------------+ 1965| ``i32`` | a 32-bit integer. | 1966+----------------+------------------------------------------------+ 1967| ``i1942652`` | a really big integer of over 1 million bits. | 1968+----------------+------------------------------------------------+ 1969 1970.. _t_floating: 1971 1972Floating Point Types 1973"""""""""""""""""""" 1974 1975.. list-table:: 1976 :header-rows: 1 1977 1978 * - Type 1979 - Description 1980 1981 * - ``half`` 1982 - 16-bit floating point value 1983 1984 * - ``float`` 1985 - 32-bit floating point value 1986 1987 * - ``double`` 1988 - 64-bit floating point value 1989 1990 * - ``fp128`` 1991 - 128-bit floating point value (112-bit mantissa) 1992 1993 * - ``x86_fp80`` 1994 - 80-bit floating point value (X87) 1995 1996 * - ``ppc_fp128`` 1997 - 128-bit floating point value (two 64-bits) 1998 1999X86_mmx Type 2000"""""""""""" 2001 2002:Overview: 2003 2004The x86_mmx type represents a value held in an MMX register on an x86 2005machine. The operations allowed on it are quite limited: parameters and 2006return values, load and store, and bitcast. User-specified MMX 2007instructions are represented as intrinsic or asm calls with arguments 2008and/or results of this type. There are no arrays, vectors or constants 2009of this type. 2010 2011:Syntax: 2012 2013:: 2014 2015 x86_mmx 2016 2017 2018.. _t_pointer: 2019 2020Pointer Type 2021"""""""""""" 2022 2023:Overview: 2024 2025The pointer type is used to specify memory locations. Pointers are 2026commonly used to reference objects in memory. 2027 2028Pointer types may have an optional address space attribute defining the 2029numbered address space where the pointed-to object resides. The default 2030address space is number zero. The semantics of non-zero address spaces 2031are target-specific. 2032 2033Note that LLVM does not permit pointers to void (``void*``) nor does it 2034permit pointers to labels (``label*``). Use ``i8*`` instead. 2035 2036:Syntax: 2037 2038:: 2039 2040 <type> * 2041 2042:Examples: 2043 2044+-------------------------+--------------------------------------------------------------------------------------------------------------+ 2045| ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. | 2046+-------------------------+--------------------------------------------------------------------------------------------------------------+ 2047| ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. | 2048+-------------------------+--------------------------------------------------------------------------------------------------------------+ 2049| ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space #5. | 2050+-------------------------+--------------------------------------------------------------------------------------------------------------+ 2051 2052.. _t_vector: 2053 2054Vector Type 2055""""""""""" 2056 2057:Overview: 2058 2059A vector type is a simple derived type that represents a vector of 2060elements. Vector types are used when multiple primitive data are 2061operated in parallel using a single instruction (SIMD). A vector type 2062requires a size (number of elements) and an underlying primitive data 2063type. Vector types are considered :ref:`first class <t_firstclass>`. 2064 2065:Syntax: 2066 2067:: 2068 2069 < <# elements> x <elementtype> > 2070 2071The number of elements is a constant integer value larger than 0; 2072elementtype may be any integer, floating point or pointer type. Vectors 2073of size zero are not allowed. 2074 2075:Examples: 2076 2077+-------------------+--------------------------------------------------+ 2078| ``<4 x i32>`` | Vector of 4 32-bit integer values. | 2079+-------------------+--------------------------------------------------+ 2080| ``<8 x float>`` | Vector of 8 32-bit floating-point values. | 2081+-------------------+--------------------------------------------------+ 2082| ``<2 x i64>`` | Vector of 2 64-bit integer values. | 2083+-------------------+--------------------------------------------------+ 2084| ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. | 2085+-------------------+--------------------------------------------------+ 2086 2087.. _t_label: 2088 2089Label Type 2090^^^^^^^^^^ 2091 2092:Overview: 2093 2094The label type represents code labels. 2095 2096:Syntax: 2097 2098:: 2099 2100 label 2101 2102.. _t_metadata: 2103 2104Metadata Type 2105^^^^^^^^^^^^^ 2106 2107:Overview: 2108 2109The metadata type represents embedded metadata. No derived types may be 2110created from metadata except for :ref:`function <t_function>` arguments. 2111 2112:Syntax: 2113 2114:: 2115 2116 metadata 2117 2118.. _t_aggregate: 2119 2120Aggregate Types 2121^^^^^^^^^^^^^^^ 2122 2123Aggregate Types are a subset of derived types that can contain multiple 2124member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are 2125aggregate types. :ref:`Vectors <t_vector>` are not considered to be 2126aggregate types. 2127 2128.. _t_array: 2129 2130Array Type 2131"""""""""" 2132 2133:Overview: 2134 2135The array type is a very simple derived type that arranges elements 2136sequentially in memory. The array type requires a size (number of 2137elements) and an underlying data type. 2138 2139:Syntax: 2140 2141:: 2142 2143 [<# elements> x <elementtype>] 2144 2145The number of elements is a constant integer value; ``elementtype`` may 2146be any type with a size. 2147 2148:Examples: 2149 2150+------------------+--------------------------------------+ 2151| ``[40 x i32]`` | Array of 40 32-bit integer values. | 2152+------------------+--------------------------------------+ 2153| ``[41 x i32]`` | Array of 41 32-bit integer values. | 2154+------------------+--------------------------------------+ 2155| ``[4 x i8]`` | Array of 4 8-bit integer values. | 2156+------------------+--------------------------------------+ 2157 2158Here are some examples of multidimensional arrays: 2159 2160+-----------------------------+----------------------------------------------------------+ 2161| ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. | 2162+-----------------------------+----------------------------------------------------------+ 2163| ``[12 x [10 x float]]`` | 12x10 array of single precision floating point values. | 2164+-----------------------------+----------------------------------------------------------+ 2165| ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. | 2166+-----------------------------+----------------------------------------------------------+ 2167 2168There is no restriction on indexing beyond the end of the array implied 2169by a static type (though there are restrictions on indexing beyond the 2170bounds of an allocated object in some cases). This means that 2171single-dimension 'variable sized array' addressing can be implemented in 2172LLVM with a zero length array type. An implementation of 'pascal style 2173arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for 2174example. 2175 2176.. _t_struct: 2177 2178Structure Type 2179"""""""""""""" 2180 2181:Overview: 2182 2183The structure type is used to represent a collection of data members 2184together in memory. The elements of a structure may be any type that has 2185a size. 2186 2187Structures in memory are accessed using '``load``' and '``store``' by 2188getting a pointer to a field with the '``getelementptr``' instruction. 2189Structures in registers are accessed using the '``extractvalue``' and 2190'``insertvalue``' instructions. 2191 2192Structures may optionally be "packed" structures, which indicate that 2193the alignment of the struct is one byte, and that there is no padding 2194between the elements. In non-packed structs, padding between field types 2195is inserted as defined by the DataLayout string in the module, which is 2196required to match what the underlying code generator expects. 2197 2198Structures can either be "literal" or "identified". A literal structure 2199is defined inline with other types (e.g. ``{i32, i32}*``) whereas 2200identified types are always defined at the top level with a name. 2201Literal types are uniqued by their contents and can never be recursive 2202or opaque since there is no way to write one. Identified types can be 2203recursive, can be opaqued, and are never uniqued. 2204 2205:Syntax: 2206 2207:: 2208 2209 %T1 = type { <type list> } ; Identified normal struct type 2210 %T2 = type <{ <type list> }> ; Identified packed struct type 2211 2212:Examples: 2213 2214+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 2215| ``{ i32, i32, i32 }`` | A triple of three ``i32`` values | 2216+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 2217| ``{ float, i32 (i32) * }`` | A pair, where the first element is a ``float`` and the second element is a :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32``, returning an ``i32``. | 2218+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 2219| ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. | 2220+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 2221 2222.. _t_opaque: 2223 2224Opaque Structure Types 2225"""""""""""""""""""""" 2226 2227:Overview: 2228 2229Opaque structure types are used to represent named structure types that 2230do not have a body specified. This corresponds (for example) to the C 2231notion of a forward declared structure. 2232 2233:Syntax: 2234 2235:: 2236 2237 %X = type opaque 2238 %52 = type opaque 2239 2240:Examples: 2241 2242+--------------+-------------------+ 2243| ``opaque`` | An opaque type. | 2244+--------------+-------------------+ 2245 2246.. _constants: 2247 2248Constants 2249========= 2250 2251LLVM has several different basic types of constants. This section 2252describes them all and their syntax. 2253 2254Simple Constants 2255---------------- 2256 2257**Boolean constants** 2258 The two strings '``true``' and '``false``' are both valid constants 2259 of the ``i1`` type. 2260**Integer constants** 2261 Standard integers (such as '4') are constants of the 2262 :ref:`integer <t_integer>` type. Negative numbers may be used with 2263 integer types. 2264**Floating point constants** 2265 Floating point constants use standard decimal notation (e.g. 2266 123.421), exponential notation (e.g. 1.23421e+2), or a more precise 2267 hexadecimal notation (see below). The assembler requires the exact 2268 decimal value of a floating-point constant. For example, the 2269 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating 2270 decimal in binary. Floating point constants must have a :ref:`floating 2271 point <t_floating>` type. 2272**Null pointer constants** 2273 The identifier '``null``' is recognized as a null pointer constant 2274 and must be of :ref:`pointer type <t_pointer>`. 2275 2276The one non-intuitive notation for constants is the hexadecimal form of 2277floating point constants. For example, the form 2278'``double 0x432ff973cafa8000``' is equivalent to (but harder to read 2279than) '``double 4.5e+15``'. The only time hexadecimal floating point 2280constants are required (and the only time that they are generated by the 2281disassembler) is when a floating point constant must be emitted but it 2282cannot be represented as a decimal floating point number in a reasonable 2283number of digits. For example, NaN's, infinities, and other special 2284values are represented in their IEEE hexadecimal format so that assembly 2285and disassembly do not cause any bits to change in the constants. 2286 2287When using the hexadecimal form, constants of types half, float, and 2288double are represented using the 16-digit form shown above (which 2289matches the IEEE754 representation for double); half and float values 2290must, however, be exactly representable as IEEE 754 half and single 2291precision, respectively. Hexadecimal format is always used for long 2292double, and there are three forms of long double. The 80-bit format used 2293by x86 is represented as ``0xK`` followed by 20 hexadecimal digits. The 2294128-bit format used by PowerPC (two adjacent doubles) is represented by 2295``0xM`` followed by 32 hexadecimal digits. The IEEE 128-bit format is 2296represented by ``0xL`` followed by 32 hexadecimal digits. Long doubles 2297will only work if they match the long double format on your target. 2298The IEEE 16-bit format (half precision) is represented by ``0xH`` 2299followed by 4 hexadecimal digits. All hexadecimal formats are big-endian 2300(sign bit at the left). 2301 2302There are no constants of type x86_mmx. 2303 2304.. _complexconstants: 2305 2306Complex Constants 2307----------------- 2308 2309Complex constants are a (potentially recursive) combination of simple 2310constants and smaller complex constants. 2311 2312**Structure constants** 2313 Structure constants are represented with notation similar to 2314 structure type definitions (a comma separated list of elements, 2315 surrounded by braces (``{}``)). For example: 2316 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as 2317 "``@G = external global i32``". Structure constants must have 2318 :ref:`structure type <t_struct>`, and the number and types of elements 2319 must match those specified by the type. 2320**Array constants** 2321 Array constants are represented with notation similar to array type 2322 definitions (a comma separated list of elements, surrounded by 2323 square brackets (``[]``)). For example: 2324 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have 2325 :ref:`array type <t_array>`, and the number and types of elements must 2326 match those specified by the type. As a special case, character array 2327 constants may also be represented as a double-quoted string using the ``c`` 2328 prefix. For example: "``c"Hello World\0A\00"``". 2329**Vector constants** 2330 Vector constants are represented with notation similar to vector 2331 type definitions (a comma separated list of elements, surrounded by 2332 less-than/greater-than's (``<>``)). For example: 2333 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants 2334 must have :ref:`vector type <t_vector>`, and the number and types of 2335 elements must match those specified by the type. 2336**Zero initialization** 2337 The string '``zeroinitializer``' can be used to zero initialize a 2338 value to zero of *any* type, including scalar and 2339 :ref:`aggregate <t_aggregate>` types. This is often used to avoid 2340 having to print large zero initializers (e.g. for large arrays) and 2341 is always exactly equivalent to using explicit zero initializers. 2342**Metadata node** 2343 A metadata node is a constant tuple without types. For example: 2344 "``!{!0, !{!2, !0}, !"test"}``". Metadata can reference constant values, 2345 for example: "``!{!0, i32 0, i8* @global, i64 (i64)* @function, !"str"}``". 2346 Unlike other typed constants that are meant to be interpreted as part of 2347 the instruction stream, metadata is a place to attach additional 2348 information such as debug info. 2349 2350Global Variable and Function Addresses 2351-------------------------------------- 2352 2353The addresses of :ref:`global variables <globalvars>` and 2354:ref:`functions <functionstructure>` are always implicitly valid 2355(link-time) constants. These constants are explicitly referenced when 2356the :ref:`identifier for the global <identifiers>` is used and always have 2357:ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM 2358file: 2359 2360.. code-block:: llvm 2361 2362 @X = global i32 17 2363 @Y = global i32 42 2364 @Z = global [2 x i32*] [ i32* @X, i32* @Y ] 2365 2366.. _undefvalues: 2367 2368Undefined Values 2369---------------- 2370 2371The string '``undef``' can be used anywhere a constant is expected, and 2372indicates that the user of the value may receive an unspecified 2373bit-pattern. Undefined values may be of any type (other than '``label``' 2374or '``void``') and be used anywhere a constant is permitted. 2375 2376Undefined values are useful because they indicate to the compiler that 2377the program is well defined no matter what value is used. This gives the 2378compiler more freedom to optimize. Here are some examples of 2379(potentially surprising) transformations that are valid (in pseudo IR): 2380 2381.. code-block:: llvm 2382 2383 %A = add %X, undef 2384 %B = sub %X, undef 2385 %C = xor %X, undef 2386 Safe: 2387 %A = undef 2388 %B = undef 2389 %C = undef 2390 2391This is safe because all of the output bits are affected by the undef 2392bits. Any output bit can have a zero or one depending on the input bits. 2393 2394.. code-block:: llvm 2395 2396 %A = or %X, undef 2397 %B = and %X, undef 2398 Safe: 2399 %A = -1 2400 %B = 0 2401 Unsafe: 2402 %A = undef 2403 %B = undef 2404 2405These logical operations have bits that are not always affected by the 2406input. For example, if ``%X`` has a zero bit, then the output of the 2407'``and``' operation will always be a zero for that bit, no matter what 2408the corresponding bit from the '``undef``' is. As such, it is unsafe to 2409optimize or assume that the result of the '``and``' is '``undef``'. 2410However, it is safe to assume that all bits of the '``undef``' could be 24110, and optimize the '``and``' to 0. Likewise, it is safe to assume that 2412all the bits of the '``undef``' operand to the '``or``' could be set, 2413allowing the '``or``' to be folded to -1. 2414 2415.. code-block:: llvm 2416 2417 %A = select undef, %X, %Y 2418 %B = select undef, 42, %Y 2419 %C = select %X, %Y, undef 2420 Safe: 2421 %A = %X (or %Y) 2422 %B = 42 (or %Y) 2423 %C = %Y 2424 Unsafe: 2425 %A = undef 2426 %B = undef 2427 %C = undef 2428 2429This set of examples shows that undefined '``select``' (and conditional 2430branch) conditions can go *either way*, but they have to come from one 2431of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were 2432both known to have a clear low bit, then ``%A`` would have to have a 2433cleared low bit. However, in the ``%C`` example, the optimizer is 2434allowed to assume that the '``undef``' operand could be the same as 2435``%Y``, allowing the whole '``select``' to be eliminated. 2436 2437.. code-block:: llvm 2438 2439 %A = xor undef, undef 2440 2441 %B = undef 2442 %C = xor %B, %B 2443 2444 %D = undef 2445 %E = icmp slt %D, 4 2446 %F = icmp gte %D, 4 2447 2448 Safe: 2449 %A = undef 2450 %B = undef 2451 %C = undef 2452 %D = undef 2453 %E = undef 2454 %F = undef 2455 2456This example points out that two '``undef``' operands are not 2457necessarily the same. This can be surprising to people (and also matches 2458C semantics) where they assume that "``X^X``" is always zero, even if 2459``X`` is undefined. This isn't true for a number of reasons, but the 2460short answer is that an '``undef``' "variable" can arbitrarily change 2461its value over its "live range". This is true because the variable 2462doesn't actually *have a live range*. Instead, the value is logically 2463read from arbitrary registers that happen to be around when needed, so 2464the value is not necessarily consistent over time. In fact, ``%A`` and 2465``%C`` need to have the same semantics or the core LLVM "replace all 2466uses with" concept would not hold. 2467 2468.. code-block:: llvm 2469 2470 %A = fdiv undef, %X 2471 %B = fdiv %X, undef 2472 Safe: 2473 %A = undef 2474 b: unreachable 2475 2476These examples show the crucial difference between an *undefined value* 2477and *undefined behavior*. An undefined value (like '``undef``') is 2478allowed to have an arbitrary bit-pattern. This means that the ``%A`` 2479operation can be constant folded to '``undef``', because the '``undef``' 2480could be an SNaN, and ``fdiv`` is not (currently) defined on SNaN's. 2481However, in the second example, we can make a more aggressive 2482assumption: because the ``undef`` is allowed to be an arbitrary value, 2483we are allowed to assume that it could be zero. Since a divide by zero 2484has *undefined behavior*, we are allowed to assume that the operation 2485does not execute at all. This allows us to delete the divide and all 2486code after it. Because the undefined operation "can't happen", the 2487optimizer can assume that it occurs in dead code. 2488 2489.. code-block:: llvm 2490 2491 a: store undef -> %X 2492 b: store %X -> undef 2493 Safe: 2494 a: <deleted> 2495 b: unreachable 2496 2497These examples reiterate the ``fdiv`` example: a store *of* an undefined 2498value can be assumed to not have any effect; we can assume that the 2499value is overwritten with bits that happen to match what was already 2500there. However, a store *to* an undefined location could clobber 2501arbitrary memory, therefore, it has undefined behavior. 2502 2503.. _poisonvalues: 2504 2505Poison Values 2506------------- 2507 2508Poison values are similar to :ref:`undef values <undefvalues>`, however 2509they also represent the fact that an instruction or constant expression 2510that cannot evoke side effects has nevertheless detected a condition 2511that results in undefined behavior. 2512 2513There is currently no way of representing a poison value in the IR; they 2514only exist when produced by operations such as :ref:`add <i_add>` with 2515the ``nsw`` flag. 2516 2517Poison value behavior is defined in terms of value *dependence*: 2518 2519- Values other than :ref:`phi <i_phi>` nodes depend on their operands. 2520- :ref:`Phi <i_phi>` nodes depend on the operand corresponding to 2521 their dynamic predecessor basic block. 2522- Function arguments depend on the corresponding actual argument values 2523 in the dynamic callers of their functions. 2524- :ref:`Call <i_call>` instructions depend on the :ref:`ret <i_ret>` 2525 instructions that dynamically transfer control back to them. 2526- :ref:`Invoke <i_invoke>` instructions depend on the 2527 :ref:`ret <i_ret>`, :ref:`resume <i_resume>`, or exception-throwing 2528 call instructions that dynamically transfer control back to them. 2529- Non-volatile loads and stores depend on the most recent stores to all 2530 of the referenced memory addresses, following the order in the IR 2531 (including loads and stores implied by intrinsics such as 2532 :ref:`@llvm.memcpy <int_memcpy>`.) 2533- An instruction with externally visible side effects depends on the 2534 most recent preceding instruction with externally visible side 2535 effects, following the order in the IR. (This includes :ref:`volatile 2536 operations <volatile>`.) 2537- An instruction *control-depends* on a :ref:`terminator 2538 instruction <terminators>` if the terminator instruction has 2539 multiple successors and the instruction is always executed when 2540 control transfers to one of the successors, and may not be executed 2541 when control is transferred to another. 2542- Additionally, an instruction also *control-depends* on a terminator 2543 instruction if the set of instructions it otherwise depends on would 2544 be different if the terminator had transferred control to a different 2545 successor. 2546- Dependence is transitive. 2547 2548Poison values have the same behavior as :ref:`undef values <undefvalues>`, 2549with the additional effect that any instruction that has a *dependence* 2550on a poison value has undefined behavior. 2551 2552Here are some examples: 2553 2554.. code-block:: llvm 2555 2556 entry: 2557 %poison = sub nuw i32 0, 1 ; Results in a poison value. 2558 %still_poison = and i32 %poison, 0 ; 0, but also poison. 2559 %poison_yet_again = getelementptr i32* @h, i32 %still_poison 2560 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned 2561 2562 store i32 %poison, i32* @g ; Poison value stored to memory. 2563 %poison2 = load i32* @g ; Poison value loaded back from memory. 2564 2565 store volatile i32 %poison, i32* @g ; External observation; undefined behavior. 2566 2567 %narrowaddr = bitcast i32* @g to i16* 2568 %wideaddr = bitcast i32* @g to i64* 2569 %poison3 = load i16* %narrowaddr ; Returns a poison value. 2570 %poison4 = load i64* %wideaddr ; Returns a poison value. 2571 2572 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value. 2573 br i1 %cmp, label %true, label %end ; Branch to either destination. 2574 2575 true: 2576 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so 2577 ; it has undefined behavior. 2578 br label %end 2579 2580 end: 2581 %p = phi i32 [ 0, %entry ], [ 1, %true ] 2582 ; Both edges into this PHI are 2583 ; control-dependent on %cmp, so this 2584 ; always results in a poison value. 2585 2586 store volatile i32 0, i32* @g ; This would depend on the store in %true 2587 ; if %cmp is true, or the store in %entry 2588 ; otherwise, so this is undefined behavior. 2589 2590 br i1 %cmp, label %second_true, label %second_end 2591 ; The same branch again, but this time the 2592 ; true block doesn't have side effects. 2593 2594 second_true: 2595 ; No side effects! 2596 ret void 2597 2598 second_end: 2599 store volatile i32 0, i32* @g ; This time, the instruction always depends 2600 ; on the store in %end. Also, it is 2601 ; control-equivalent to %end, so this is 2602 ; well-defined (ignoring earlier undefined 2603 ; behavior in this example). 2604 2605.. _blockaddress: 2606 2607Addresses of Basic Blocks 2608------------------------- 2609 2610``blockaddress(@function, %block)`` 2611 2612The '``blockaddress``' constant computes the address of the specified 2613basic block in the specified function, and always has an ``i8*`` type. 2614Taking the address of the entry block is illegal. 2615 2616This value only has defined behavior when used as an operand to the 2617':ref:`indirectbr <i_indirectbr>`' instruction, or for comparisons 2618against null. Pointer equality tests between labels addresses results in 2619undefined behavior --- though, again, comparison against null is ok, and 2620no label is equal to the null pointer. This may be passed around as an 2621opaque pointer sized value as long as the bits are not inspected. This 2622allows ``ptrtoint`` and arithmetic to be performed on these values so 2623long as the original value is reconstituted before the ``indirectbr`` 2624instruction. 2625 2626Finally, some targets may provide defined semantics when using the value 2627as the operand to an inline assembly, but that is target specific. 2628 2629.. _constantexprs: 2630 2631Constant Expressions 2632-------------------- 2633 2634Constant expressions are used to allow expressions involving other 2635constants to be used as constants. Constant expressions may be of any 2636:ref:`first class <t_firstclass>` type and may involve any LLVM operation 2637that does not have side effects (e.g. load and call are not supported). 2638The following is the syntax for constant expressions: 2639 2640``trunc (CST to TYPE)`` 2641 Truncate a constant to another type. The bit size of CST must be 2642 larger than the bit size of TYPE. Both types must be integers. 2643``zext (CST to TYPE)`` 2644 Zero extend a constant to another type. The bit size of CST must be 2645 smaller than the bit size of TYPE. Both types must be integers. 2646``sext (CST to TYPE)`` 2647 Sign extend a constant to another type. The bit size of CST must be 2648 smaller than the bit size of TYPE. Both types must be integers. 2649``fptrunc (CST to TYPE)`` 2650 Truncate a floating point constant to another floating point type. 2651 The size of CST must be larger than the size of TYPE. Both types 2652 must be floating point. 2653``fpext (CST to TYPE)`` 2654 Floating point extend a constant to another type. The size of CST 2655 must be smaller or equal to the size of TYPE. Both types must be 2656 floating point. 2657``fptoui (CST to TYPE)`` 2658 Convert a floating point constant to the corresponding unsigned 2659 integer constant. TYPE must be a scalar or vector integer type. CST 2660 must be of scalar or vector floating point type. Both CST and TYPE 2661 must be scalars, or vectors of the same number of elements. If the 2662 value won't fit in the integer type, the results are undefined. 2663``fptosi (CST to TYPE)`` 2664 Convert a floating point constant to the corresponding signed 2665 integer constant. TYPE must be a scalar or vector integer type. CST 2666 must be of scalar or vector floating point type. Both CST and TYPE 2667 must be scalars, or vectors of the same number of elements. If the 2668 value won't fit in the integer type, the results are undefined. 2669``uitofp (CST to TYPE)`` 2670 Convert an unsigned integer constant to the corresponding floating 2671 point constant. TYPE must be a scalar or vector floating point type. 2672 CST must be of scalar or vector integer type. Both CST and TYPE must 2673 be scalars, or vectors of the same number of elements. If the value 2674 won't fit in the floating point type, the results are undefined. 2675``sitofp (CST to TYPE)`` 2676 Convert a signed integer constant to the corresponding floating 2677 point constant. TYPE must be a scalar or vector floating point type. 2678 CST must be of scalar or vector integer type. Both CST and TYPE must 2679 be scalars, or vectors of the same number of elements. If the value 2680 won't fit in the floating point type, the results are undefined. 2681``ptrtoint (CST to TYPE)`` 2682 Convert a pointer typed constant to the corresponding integer 2683 constant. ``TYPE`` must be an integer type. ``CST`` must be of 2684 pointer type. The ``CST`` value is zero extended, truncated, or 2685 unchanged to make it fit in ``TYPE``. 2686``inttoptr (CST to TYPE)`` 2687 Convert an integer constant to a pointer constant. TYPE must be a 2688 pointer type. CST must be of integer type. The CST value is zero 2689 extended, truncated, or unchanged to make it fit in a pointer size. 2690 This one is *really* dangerous! 2691``bitcast (CST to TYPE)`` 2692 Convert a constant, CST, to another TYPE. The constraints of the 2693 operands are the same as those for the :ref:`bitcast 2694 instruction <i_bitcast>`. 2695``addrspacecast (CST to TYPE)`` 2696 Convert a constant pointer or constant vector of pointer, CST, to another 2697 TYPE in a different address space. The constraints of the operands are the 2698 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`. 2699``getelementptr (CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)`` 2700 Perform the :ref:`getelementptr operation <i_getelementptr>` on 2701 constants. As with the :ref:`getelementptr <i_getelementptr>` 2702 instruction, the index list may have zero or more indexes, which are 2703 required to make sense for the type of "CSTPTR". 2704``select (COND, VAL1, VAL2)`` 2705 Perform the :ref:`select operation <i_select>` on constants. 2706``icmp COND (VAL1, VAL2)`` 2707 Performs the :ref:`icmp operation <i_icmp>` on constants. 2708``fcmp COND (VAL1, VAL2)`` 2709 Performs the :ref:`fcmp operation <i_fcmp>` on constants. 2710``extractelement (VAL, IDX)`` 2711 Perform the :ref:`extractelement operation <i_extractelement>` on 2712 constants. 2713``insertelement (VAL, ELT, IDX)`` 2714 Perform the :ref:`insertelement operation <i_insertelement>` on 2715 constants. 2716``shufflevector (VEC1, VEC2, IDXMASK)`` 2717 Perform the :ref:`shufflevector operation <i_shufflevector>` on 2718 constants. 2719``extractvalue (VAL, IDX0, IDX1, ...)`` 2720 Perform the :ref:`extractvalue operation <i_extractvalue>` on 2721 constants. The index list is interpreted in a similar manner as 2722 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At 2723 least one index value must be specified. 2724``insertvalue (VAL, ELT, IDX0, IDX1, ...)`` 2725 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants. 2726 The index list is interpreted in a similar manner as indices in a 2727 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index 2728 value must be specified. 2729``OPCODE (LHS, RHS)`` 2730 Perform the specified operation of the LHS and RHS constants. OPCODE 2731 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise 2732 binary <bitwiseops>` operations. The constraints on operands are 2733 the same as those for the corresponding instruction (e.g. no bitwise 2734 operations on floating point values are allowed). 2735 2736Other Values 2737============ 2738 2739.. _inlineasmexprs: 2740 2741Inline Assembler Expressions 2742---------------------------- 2743 2744LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level 2745Inline Assembly <moduleasm>`) through the use of a special value. This 2746value represents the inline assembler as a string (containing the 2747instructions to emit), a list of operand constraints (stored as a 2748string), a flag that indicates whether or not the inline asm expression 2749has side effects, and a flag indicating whether the function containing 2750the asm needs to align its stack conservatively. An example inline 2751assembler expression is: 2752 2753.. code-block:: llvm 2754 2755 i32 (i32) asm "bswap $0", "=r,r" 2756 2757Inline assembler expressions may **only** be used as the callee operand 2758of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction. 2759Thus, typically we have: 2760 2761.. code-block:: llvm 2762 2763 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y) 2764 2765Inline asms with side effects not visible in the constraint list must be 2766marked as having side effects. This is done through the use of the 2767'``sideeffect``' keyword, like so: 2768 2769.. code-block:: llvm 2770 2771 call void asm sideeffect "eieio", ""() 2772 2773In some cases inline asms will contain code that will not work unless 2774the stack is aligned in some way, such as calls or SSE instructions on 2775x86, yet will not contain code that does that alignment within the asm. 2776The compiler should make conservative assumptions about what the asm 2777might contain and should generate its usual stack alignment code in the 2778prologue if the '``alignstack``' keyword is present: 2779 2780.. code-block:: llvm 2781 2782 call void asm alignstack "eieio", ""() 2783 2784Inline asms also support using non-standard assembly dialects. The 2785assumed dialect is ATT. When the '``inteldialect``' keyword is present, 2786the inline asm is using the Intel dialect. Currently, ATT and Intel are 2787the only supported dialects. An example is: 2788 2789.. code-block:: llvm 2790 2791 call void asm inteldialect "eieio", ""() 2792 2793If multiple keywords appear the '``sideeffect``' keyword must come 2794first, the '``alignstack``' keyword second and the '``inteldialect``' 2795keyword last. 2796 2797Inline Asm Metadata 2798^^^^^^^^^^^^^^^^^^^ 2799 2800The call instructions that wrap inline asm nodes may have a 2801"``!srcloc``" MDNode attached to it that contains a list of constant 2802integers. If present, the code generator will use the integer as the 2803location cookie value when report errors through the ``LLVMContext`` 2804error reporting mechanisms. This allows a front-end to correlate backend 2805errors that occur with inline asm back to the source code that produced 2806it. For example: 2807 2808.. code-block:: llvm 2809 2810 call void asm sideeffect "something bad", ""(), !srcloc !42 2811 ... 2812 !42 = !{ i32 1234567 } 2813 2814It is up to the front-end to make sense of the magic numbers it places 2815in the IR. If the MDNode contains multiple constants, the code generator 2816will use the one that corresponds to the line of the asm that the error 2817occurs on. 2818 2819.. _metadata: 2820 2821Metadata 2822======== 2823 2824LLVM IR allows metadata to be attached to instructions in the program 2825that can convey extra information about the code to the optimizers and 2826code generator. One example application of metadata is source-level 2827debug information. There are two metadata primitives: strings and nodes. 2828 2829Metadata does not have a type, and is not a value. If referenced from a 2830``call`` instruction, it uses the ``metadata`` type. 2831 2832All metadata are identified in syntax by a exclamation point ('``!``'). 2833 2834Metadata Nodes and Metadata Strings 2835----------------------------------- 2836 2837A metadata string is a string surrounded by double quotes. It can 2838contain any character by escaping non-printable characters with 2839"``\xx``" where "``xx``" is the two digit hex code. For example: 2840"``!"test\00"``". 2841 2842Metadata nodes are represented with notation similar to structure 2843constants (a comma separated list of elements, surrounded by braces and 2844preceded by an exclamation point). Metadata nodes can have any values as 2845their operand. For example: 2846 2847.. code-block:: llvm 2848 2849 !{ !"test\00", i32 10} 2850 2851Metadata nodes that aren't uniqued use the ``distinct`` keyword. For example: 2852 2853.. code-block:: llvm 2854 2855 !0 = distinct !{!"test\00", i32 10} 2856 2857``distinct`` nodes are useful when nodes shouldn't be merged based on their 2858content. They can also occur when transformations cause uniquing collisions 2859when metadata operands change. 2860 2861A :ref:`named metadata <namedmetadatastructure>` is a collection of 2862metadata nodes, which can be looked up in the module symbol table. For 2863example: 2864 2865.. code-block:: llvm 2866 2867 !foo = !{!4, !3} 2868 2869Metadata can be used as function arguments. Here ``llvm.dbg.value`` 2870function is using two metadata arguments: 2871 2872.. code-block:: llvm 2873 2874 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25) 2875 2876Metadata can be attached with an instruction. Here metadata ``!21`` is 2877attached to the ``add`` instruction using the ``!dbg`` identifier: 2878 2879.. code-block:: llvm 2880 2881 %indvar.next = add i64 %indvar, 1, !dbg !21 2882 2883More information about specific metadata nodes recognized by the 2884optimizers and code generator is found below. 2885 2886Specialized Metadata Nodes 2887^^^^^^^^^^^^^^^^^^^^^^^^^^ 2888 2889Specialized metadata nodes are custom data structures in metadata (as opposed 2890to generic tuples). Their fields are labelled, and can be specified in any 2891order. 2892 2893MDLocation 2894"""""""""" 2895 2896``MDLocation`` nodes represent source debug locations. The ``scope:`` field is 2897mandatory. 2898 2899.. code-block:: llvm 2900 2901 !0 = !MDLocation(line: 2900, column: 42, scope: !1, inlinedAt: !2) 2902 2903'``tbaa``' Metadata 2904^^^^^^^^^^^^^^^^^^^ 2905 2906In LLVM IR, memory does not have types, so LLVM's own type system is not 2907suitable for doing TBAA. Instead, metadata is added to the IR to 2908describe a type system of a higher level language. This can be used to 2909implement typical C/C++ TBAA, but it can also be used to implement 2910custom alias analysis behavior for other languages. 2911 2912The current metadata format is very simple. TBAA metadata nodes have up 2913to three fields, e.g.: 2914 2915.. code-block:: llvm 2916 2917 !0 = !{ !"an example type tree" } 2918 !1 = !{ !"int", !0 } 2919 !2 = !{ !"float", !0 } 2920 !3 = !{ !"const float", !2, i64 1 } 2921 2922The first field is an identity field. It can be any value, usually a 2923metadata string, which uniquely identifies the type. The most important 2924name in the tree is the name of the root node. Two trees with different 2925root node names are entirely disjoint, even if they have leaves with 2926common names. 2927 2928The second field identifies the type's parent node in the tree, or is 2929null or omitted for a root node. A type is considered to alias all of 2930its descendants and all of its ancestors in the tree. Also, a type is 2931considered to alias all types in other trees, so that bitcode produced 2932from multiple front-ends is handled conservatively. 2933 2934If the third field is present, it's an integer which if equal to 1 2935indicates that the type is "constant" (meaning 2936``pointsToConstantMemory`` should return true; see `other useful 2937AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_). 2938 2939'``tbaa.struct``' Metadata 2940^^^^^^^^^^^^^^^^^^^^^^^^^^ 2941 2942The :ref:`llvm.memcpy <int_memcpy>` is often used to implement 2943aggregate assignment operations in C and similar languages, however it 2944is defined to copy a contiguous region of memory, which is more than 2945strictly necessary for aggregate types which contain holes due to 2946padding. Also, it doesn't contain any TBAA information about the fields 2947of the aggregate. 2948 2949``!tbaa.struct`` metadata can describe which memory subregions in a 2950memcpy are padding and what the TBAA tags of the struct are. 2951 2952The current metadata format is very simple. ``!tbaa.struct`` metadata 2953nodes are a list of operands which are in conceptual groups of three. 2954For each group of three, the first operand gives the byte offset of a 2955field in bytes, the second gives its size in bytes, and the third gives 2956its tbaa tag. e.g.: 2957 2958.. code-block:: llvm 2959 2960 !4 = !{ i64 0, i64 4, !1, i64 8, i64 4, !2 } 2961 2962This describes a struct with two fields. The first is at offset 0 bytes 2963with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes 2964and has size 4 bytes and has tbaa tag !2. 2965 2966Note that the fields need not be contiguous. In this example, there is a 29674 byte gap between the two fields. This gap represents padding which 2968does not carry useful data and need not be preserved. 2969 2970'``noalias``' and '``alias.scope``' Metadata 2971^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2972 2973``noalias`` and ``alias.scope`` metadata provide the ability to specify generic 2974noalias memory-access sets. This means that some collection of memory access 2975instructions (loads, stores, memory-accessing calls, etc.) that carry 2976``noalias`` metadata can specifically be specified not to alias with some other 2977collection of memory access instructions that carry ``alias.scope`` metadata. 2978Each type of metadata specifies a list of scopes where each scope has an id and 2979a domain. When evaluating an aliasing query, if for some some domain, the set 2980of scopes with that domain in one instruction's ``alias.scope`` list is a 2981subset of (or qual to) the set of scopes for that domain in another 2982instruction's ``noalias`` list, then the two memory accesses are assumed not to 2983alias. 2984 2985The metadata identifying each domain is itself a list containing one or two 2986entries. The first entry is the name of the domain. Note that if the name is a 2987string then it can be combined accross functions and translation units. A 2988self-reference can be used to create globally unique domain names. A 2989descriptive string may optionally be provided as a second list entry. 2990 2991The metadata identifying each scope is also itself a list containing two or 2992three entries. The first entry is the name of the scope. Note that if the name 2993is a string then it can be combined accross functions and translation units. A 2994self-reference can be used to create globally unique scope names. A metadata 2995reference to the scope's domain is the second entry. A descriptive string may 2996optionally be provided as a third list entry. 2997 2998For example, 2999 3000.. code-block:: llvm 3001 3002 ; Two scope domains: 3003 !0 = !{!0} 3004 !1 = !{!1} 3005 3006 ; Some scopes in these domains: 3007 !2 = !{!2, !0} 3008 !3 = !{!3, !0} 3009 !4 = !{!4, !1} 3010 3011 ; Some scope lists: 3012 !5 = !{!4} ; A list containing only scope !4 3013 !6 = !{!4, !3, !2} 3014 !7 = !{!3} 3015 3016 ; These two instructions don't alias: 3017 %0 = load float* %c, align 4, !alias.scope !5 3018 store float %0, float* %arrayidx.i, align 4, !noalias !5 3019 3020 ; These two instructions also don't alias (for domain !1, the set of scopes 3021 ; in the !alias.scope equals that in the !noalias list): 3022 %2 = load float* %c, align 4, !alias.scope !5 3023 store float %2, float* %arrayidx.i2, align 4, !noalias !6 3024 3025 ; These two instructions don't alias (for domain !0, the set of scopes in 3026 ; the !noalias list is not a superset of, or equal to, the scopes in the 3027 ; !alias.scope list): 3028 %2 = load float* %c, align 4, !alias.scope !6 3029 store float %0, float* %arrayidx.i, align 4, !noalias !7 3030 3031'``fpmath``' Metadata 3032^^^^^^^^^^^^^^^^^^^^^ 3033 3034``fpmath`` metadata may be attached to any instruction of floating point 3035type. It can be used to express the maximum acceptable error in the 3036result of that instruction, in ULPs, thus potentially allowing the 3037compiler to use a more efficient but less accurate method of computing 3038it. ULP is defined as follows: 3039 3040 If ``x`` is a real number that lies between two finite consecutive 3041 floating-point numbers ``a`` and ``b``, without being equal to one 3042 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the 3043 distance between the two non-equal finite floating-point numbers 3044 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``. 3045 3046The metadata node shall consist of a single positive floating point 3047number representing the maximum relative error, for example: 3048 3049.. code-block:: llvm 3050 3051 !0 = !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs 3052 3053'``range``' Metadata 3054^^^^^^^^^^^^^^^^^^^^ 3055 3056``range`` metadata may be attached only to ``load``, ``call`` and ``invoke`` of 3057integer types. It expresses the possible ranges the loaded value or the value 3058returned by the called function at this call site is in. The ranges are 3059represented with a flattened list of integers. The loaded value or the value 3060returned is known to be in the union of the ranges defined by each consecutive 3061pair. Each pair has the following properties: 3062 3063- The type must match the type loaded by the instruction. 3064- The pair ``a,b`` represents the range ``[a,b)``. 3065- Both ``a`` and ``b`` are constants. 3066- The range is allowed to wrap. 3067- The range should not represent the full or empty set. That is, 3068 ``a!=b``. 3069 3070In addition, the pairs must be in signed order of the lower bound and 3071they must be non-contiguous. 3072 3073Examples: 3074 3075.. code-block:: llvm 3076 3077 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1 3078 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1 3079 %c = call i8 @foo(), !range !2 ; Can only be 0, 1, 3, 4 or 5 3080 %d = invoke i8 @bar() to label %cont 3081 unwind label %lpad, !range !3 ; Can only be -2, -1, 3, 4 or 5 3082 ... 3083 !0 = !{ i8 0, i8 2 } 3084 !1 = !{ i8 255, i8 2 } 3085 !2 = !{ i8 0, i8 2, i8 3, i8 6 } 3086 !3 = !{ i8 -2, i8 0, i8 3, i8 6 } 3087 3088'``llvm.loop``' 3089^^^^^^^^^^^^^^^ 3090 3091It is sometimes useful to attach information to loop constructs. Currently, 3092loop metadata is implemented as metadata attached to the branch instruction 3093in the loop latch block. This type of metadata refer to a metadata node that is 3094guaranteed to be separate for each loop. The loop identifier metadata is 3095specified with the name ``llvm.loop``. 3096 3097The loop identifier metadata is implemented using a metadata that refers to 3098itself to avoid merging it with any other identifier metadata, e.g., 3099during module linkage or function inlining. That is, each loop should refer 3100to their own identification metadata even if they reside in separate functions. 3101The following example contains loop identifier metadata for two separate loop 3102constructs: 3103 3104.. code-block:: llvm 3105 3106 !0 = !{!0} 3107 !1 = !{!1} 3108 3109The loop identifier metadata can be used to specify additional 3110per-loop metadata. Any operands after the first operand can be treated 3111as user-defined metadata. For example the ``llvm.loop.unroll.count`` 3112suggests an unroll factor to the loop unroller: 3113 3114.. code-block:: llvm 3115 3116 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0 3117 ... 3118 !0 = !{!0, !1} 3119 !1 = !{!"llvm.loop.unroll.count", i32 4} 3120 3121'``llvm.loop.vectorize``' and '``llvm.loop.interleave``' 3122^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3123 3124Metadata prefixed with ``llvm.loop.vectorize`` or ``llvm.loop.interleave`` are 3125used to control per-loop vectorization and interleaving parameters such as 3126vectorization width and interleave count. These metadata should be used in 3127conjunction with ``llvm.loop`` loop identification metadata. The 3128``llvm.loop.vectorize`` and ``llvm.loop.interleave`` metadata are only 3129optimization hints and the optimizer will only interleave and vectorize loops if 3130it believes it is safe to do so. The ``llvm.mem.parallel_loop_access`` metadata 3131which contains information about loop-carried memory dependencies can be helpful 3132in determining the safety of these transformations. 3133 3134'``llvm.loop.interleave.count``' Metadata 3135^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3136 3137This metadata suggests an interleave count to the loop interleaver. 3138The first operand is the string ``llvm.loop.interleave.count`` and the 3139second operand is an integer specifying the interleave count. For 3140example: 3141 3142.. code-block:: llvm 3143 3144 !0 = !{!"llvm.loop.interleave.count", i32 4} 3145 3146Note that setting ``llvm.loop.interleave.count`` to 1 disables interleaving 3147multiple iterations of the loop. If ``llvm.loop.interleave.count`` is set to 0 3148then the interleave count will be determined automatically. 3149 3150'``llvm.loop.vectorize.enable``' Metadata 3151^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3152 3153This metadata selectively enables or disables vectorization for the loop. The 3154first operand is the string ``llvm.loop.vectorize.enable`` and the second operand 3155is a bit. If the bit operand value is 1 vectorization is enabled. A value of 31560 disables vectorization: 3157 3158.. code-block:: llvm 3159 3160 !0 = !{!"llvm.loop.vectorize.enable", i1 0} 3161 !1 = !{!"llvm.loop.vectorize.enable", i1 1} 3162 3163'``llvm.loop.vectorize.width``' Metadata 3164^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3165 3166This metadata sets the target width of the vectorizer. The first 3167operand is the string ``llvm.loop.vectorize.width`` and the second 3168operand is an integer specifying the width. For example: 3169 3170.. code-block:: llvm 3171 3172 !0 = !{!"llvm.loop.vectorize.width", i32 4} 3173 3174Note that setting ``llvm.loop.vectorize.width`` to 1 disables 3175vectorization of the loop. If ``llvm.loop.vectorize.width`` is set to 31760 or if the loop does not have this metadata the width will be 3177determined automatically. 3178 3179'``llvm.loop.unroll``' 3180^^^^^^^^^^^^^^^^^^^^^^ 3181 3182Metadata prefixed with ``llvm.loop.unroll`` are loop unrolling 3183optimization hints such as the unroll factor. ``llvm.loop.unroll`` 3184metadata should be used in conjunction with ``llvm.loop`` loop 3185identification metadata. The ``llvm.loop.unroll`` metadata are only 3186optimization hints and the unrolling will only be performed if the 3187optimizer believes it is safe to do so. 3188 3189'``llvm.loop.unroll.count``' Metadata 3190^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3191 3192This metadata suggests an unroll factor to the loop unroller. The 3193first operand is the string ``llvm.loop.unroll.count`` and the second 3194operand is a positive integer specifying the unroll factor. For 3195example: 3196 3197.. code-block:: llvm 3198 3199 !0 = !{!"llvm.loop.unroll.count", i32 4} 3200 3201If the trip count of the loop is less than the unroll count the loop 3202will be partially unrolled. 3203 3204'``llvm.loop.unroll.disable``' Metadata 3205^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3206 3207This metadata either disables loop unrolling. The metadata has a single operand 3208which is the string ``llvm.loop.unroll.disable``. For example: 3209 3210.. code-block:: llvm 3211 3212 !0 = !{!"llvm.loop.unroll.disable"} 3213 3214'``llvm.loop.unroll.full``' Metadata 3215^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3216 3217This metadata either suggests that the loop should be unrolled fully. The 3218metadata has a single operand which is the string ``llvm.loop.unroll.disable``. 3219For example: 3220 3221.. code-block:: llvm 3222 3223 !0 = !{!"llvm.loop.unroll.full"} 3224 3225'``llvm.mem``' 3226^^^^^^^^^^^^^^^ 3227 3228Metadata types used to annotate memory accesses with information helpful 3229for optimizations are prefixed with ``llvm.mem``. 3230 3231'``llvm.mem.parallel_loop_access``' Metadata 3232^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3233 3234The ``llvm.mem.parallel_loop_access`` metadata refers to a loop identifier, 3235or metadata containing a list of loop identifiers for nested loops. 3236The metadata is attached to memory accessing instructions and denotes that 3237no loop carried memory dependence exist between it and other instructions denoted 3238with the same loop identifier. 3239 3240Precisely, given two instructions ``m1`` and ``m2`` that both have the 3241``llvm.mem.parallel_loop_access`` metadata, with ``L1`` and ``L2`` being the 3242set of loops associated with that metadata, respectively, then there is no loop 3243carried dependence between ``m1`` and ``m2`` for loops in both ``L1`` and 3244``L2``. 3245 3246As a special case, if all memory accessing instructions in a loop have 3247``llvm.mem.parallel_loop_access`` metadata that refers to that loop, then the 3248loop has no loop carried memory dependences and is considered to be a parallel 3249loop. 3250 3251Note that if not all memory access instructions have such metadata referring to 3252the loop, then the loop is considered not being trivially parallel. Additional 3253memory dependence analysis is required to make that determination. As a fail 3254safe mechanism, this causes loops that were originally parallel to be considered 3255sequential (if optimization passes that are unaware of the parallel semantics 3256insert new memory instructions into the loop body). 3257 3258Example of a loop that is considered parallel due to its correct use of 3259both ``llvm.loop`` and ``llvm.mem.parallel_loop_access`` 3260metadata types that refer to the same loop identifier metadata. 3261 3262.. code-block:: llvm 3263 3264 for.body: 3265 ... 3266 %val0 = load i32* %arrayidx, !llvm.mem.parallel_loop_access !0 3267 ... 3268 store i32 %val0, i32* %arrayidx1, !llvm.mem.parallel_loop_access !0 3269 ... 3270 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0 3271 3272 for.end: 3273 ... 3274 !0 = !{!0} 3275 3276It is also possible to have nested parallel loops. In that case the 3277memory accesses refer to a list of loop identifier metadata nodes instead of 3278the loop identifier metadata node directly: 3279 3280.. code-block:: llvm 3281 3282 outer.for.body: 3283 ... 3284 %val1 = load i32* %arrayidx3, !llvm.mem.parallel_loop_access !2 3285 ... 3286 br label %inner.for.body 3287 3288 inner.for.body: 3289 ... 3290 %val0 = load i32* %arrayidx1, !llvm.mem.parallel_loop_access !0 3291 ... 3292 store i32 %val0, i32* %arrayidx2, !llvm.mem.parallel_loop_access !0 3293 ... 3294 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1 3295 3296 inner.for.end: 3297 ... 3298 store i32 %val1, i32* %arrayidx4, !llvm.mem.parallel_loop_access !2 3299 ... 3300 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2 3301 3302 outer.for.end: ; preds = %for.body 3303 ... 3304 !0 = !{!1, !2} ; a list of loop identifiers 3305 !1 = !{!1} ; an identifier for the inner loop 3306 !2 = !{!2} ; an identifier for the outer loop 3307 3308Module Flags Metadata 3309===================== 3310 3311Information about the module as a whole is difficult to convey to LLVM's 3312subsystems. The LLVM IR isn't sufficient to transmit this information. 3313The ``llvm.module.flags`` named metadata exists in order to facilitate 3314this. These flags are in the form of key / value pairs --- much like a 3315dictionary --- making it easy for any subsystem who cares about a flag to 3316look it up. 3317 3318The ``llvm.module.flags`` metadata contains a list of metadata triplets. 3319Each triplet has the following form: 3320 3321- The first element is a *behavior* flag, which specifies the behavior 3322 when two (or more) modules are merged together, and it encounters two 3323 (or more) metadata with the same ID. The supported behaviors are 3324 described below. 3325- The second element is a metadata string that is a unique ID for the 3326 metadata. Each module may only have one flag entry for each unique ID (not 3327 including entries with the **Require** behavior). 3328- The third element is the value of the flag. 3329 3330When two (or more) modules are merged together, the resulting 3331``llvm.module.flags`` metadata is the union of the modules' flags. That is, for 3332each unique metadata ID string, there will be exactly one entry in the merged 3333modules ``llvm.module.flags`` metadata table, and the value for that entry will 3334be determined by the merge behavior flag, as described below. The only exception 3335is that entries with the *Require* behavior are always preserved. 3336 3337The following behaviors are supported: 3338 3339.. list-table:: 3340 :header-rows: 1 3341 :widths: 10 90 3342 3343 * - Value 3344 - Behavior 3345 3346 * - 1 3347 - **Error** 3348 Emits an error if two values disagree, otherwise the resulting value 3349 is that of the operands. 3350 3351 * - 2 3352 - **Warning** 3353 Emits a warning if two values disagree. The result value will be the 3354 operand for the flag from the first module being linked. 3355 3356 * - 3 3357 - **Require** 3358 Adds a requirement that another module flag be present and have a 3359 specified value after linking is performed. The value must be a 3360 metadata pair, where the first element of the pair is the ID of the 3361 module flag to be restricted, and the second element of the pair is 3362 the value the module flag should be restricted to. This behavior can 3363 be used to restrict the allowable results (via triggering of an 3364 error) of linking IDs with the **Override** behavior. 3365 3366 * - 4 3367 - **Override** 3368 Uses the specified value, regardless of the behavior or value of the 3369 other module. If both modules specify **Override**, but the values 3370 differ, an error will be emitted. 3371 3372 * - 5 3373 - **Append** 3374 Appends the two values, which are required to be metadata nodes. 3375 3376 * - 6 3377 - **AppendUnique** 3378 Appends the two values, which are required to be metadata 3379 nodes. However, duplicate entries in the second list are dropped 3380 during the append operation. 3381 3382It is an error for a particular unique flag ID to have multiple behaviors, 3383except in the case of **Require** (which adds restrictions on another metadata 3384value) or **Override**. 3385 3386An example of module flags: 3387 3388.. code-block:: llvm 3389 3390 !0 = !{ i32 1, !"foo", i32 1 } 3391 !1 = !{ i32 4, !"bar", i32 37 } 3392 !2 = !{ i32 2, !"qux", i32 42 } 3393 !3 = !{ i32 3, !"qux", 3394 !{ 3395 !"foo", i32 1 3396 } 3397 } 3398 !llvm.module.flags = !{ !0, !1, !2, !3 } 3399 3400- Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior 3401 if two or more ``!"foo"`` flags are seen is to emit an error if their 3402 values are not equal. 3403 3404- Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The 3405 behavior if two or more ``!"bar"`` flags are seen is to use the value 3406 '37'. 3407 3408- Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The 3409 behavior if two or more ``!"qux"`` flags are seen is to emit a 3410 warning if their values are not equal. 3411 3412- Metadata ``!3`` has the ID ``!"qux"`` and the value: 3413 3414 :: 3415 3416 !{ !"foo", i32 1 } 3417 3418 The behavior is to emit an error if the ``llvm.module.flags`` does not 3419 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is 3420 performed. 3421 3422Objective-C Garbage Collection Module Flags Metadata 3423---------------------------------------------------- 3424 3425On the Mach-O platform, Objective-C stores metadata about garbage 3426collection in a special section called "image info". The metadata 3427consists of a version number and a bitmask specifying what types of 3428garbage collection are supported (if any) by the file. If two or more 3429modules are linked together their garbage collection metadata needs to 3430be merged rather than appended together. 3431 3432The Objective-C garbage collection module flags metadata consists of the 3433following key-value pairs: 3434 3435.. list-table:: 3436 :header-rows: 1 3437 :widths: 30 70 3438 3439 * - Key 3440 - Value 3441 3442 * - ``Objective-C Version`` 3443 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2. 3444 3445 * - ``Objective-C Image Info Version`` 3446 - **[Required]** --- The version of the image info section. Currently 3447 always 0. 3448 3449 * - ``Objective-C Image Info Section`` 3450 - **[Required]** --- The section to place the metadata. Valid values are 3451 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and 3452 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for 3453 Objective-C ABI version 2. 3454 3455 * - ``Objective-C Garbage Collection`` 3456 - **[Required]** --- Specifies whether garbage collection is supported or 3457 not. Valid values are 0, for no garbage collection, and 2, for garbage 3458 collection supported. 3459 3460 * - ``Objective-C GC Only`` 3461 - **[Optional]** --- Specifies that only garbage collection is supported. 3462 If present, its value must be 6. This flag requires that the 3463 ``Objective-C Garbage Collection`` flag have the value 2. 3464 3465Some important flag interactions: 3466 3467- If a module with ``Objective-C Garbage Collection`` set to 0 is 3468 merged with a module with ``Objective-C Garbage Collection`` set to 3469 2, then the resulting module has the 3470 ``Objective-C Garbage Collection`` flag set to 0. 3471- A module with ``Objective-C Garbage Collection`` set to 0 cannot be 3472 merged with a module with ``Objective-C GC Only`` set to 6. 3473 3474Automatic Linker Flags Module Flags Metadata 3475-------------------------------------------- 3476 3477Some targets support embedding flags to the linker inside individual object 3478files. Typically this is used in conjunction with language extensions which 3479allow source files to explicitly declare the libraries they depend on, and have 3480these automatically be transmitted to the linker via object files. 3481 3482These flags are encoded in the IR using metadata in the module flags section, 3483using the ``Linker Options`` key. The merge behavior for this flag is required 3484to be ``AppendUnique``, and the value for the key is expected to be a metadata 3485node which should be a list of other metadata nodes, each of which should be a 3486list of metadata strings defining linker options. 3487 3488For example, the following metadata section specifies two separate sets of 3489linker options, presumably to link against ``libz`` and the ``Cocoa`` 3490framework:: 3491 3492 !0 = !{ i32 6, !"Linker Options", 3493 !{ 3494 !{ !"-lz" }, 3495 !{ !"-framework", !"Cocoa" } } } 3496 !llvm.module.flags = !{ !0 } 3497 3498The metadata encoding as lists of lists of options, as opposed to a collapsed 3499list of options, is chosen so that the IR encoding can use multiple option 3500strings to specify e.g., a single library, while still having that specifier be 3501preserved as an atomic element that can be recognized by a target specific 3502assembly writer or object file emitter. 3503 3504Each individual option is required to be either a valid option for the target's 3505linker, or an option that is reserved by the target specific assembly writer or 3506object file emitter. No other aspect of these options is defined by the IR. 3507 3508C type width Module Flags Metadata 3509---------------------------------- 3510 3511The ARM backend emits a section into each generated object file describing the 3512options that it was compiled with (in a compiler-independent way) to prevent 3513linking incompatible objects, and to allow automatic library selection. Some 3514of these options are not visible at the IR level, namely wchar_t width and enum 3515width. 3516 3517To pass this information to the backend, these options are encoded in module 3518flags metadata, using the following key-value pairs: 3519 3520.. list-table:: 3521 :header-rows: 1 3522 :widths: 30 70 3523 3524 * - Key 3525 - Value 3526 3527 * - short_wchar 3528 - * 0 --- sizeof(wchar_t) == 4 3529 * 1 --- sizeof(wchar_t) == 2 3530 3531 * - short_enum 3532 - * 0 --- Enums are at least as large as an ``int``. 3533 * 1 --- Enums are stored in the smallest integer type which can 3534 represent all of its values. 3535 3536For example, the following metadata section specifies that the module was 3537compiled with a ``wchar_t`` width of 4 bytes, and the underlying type of an 3538enum is the smallest type which can represent all of its values:: 3539 3540 !llvm.module.flags = !{!0, !1} 3541 !0 = !{i32 1, !"short_wchar", i32 1} 3542 !1 = !{i32 1, !"short_enum", i32 0} 3543 3544.. _intrinsicglobalvariables: 3545 3546Intrinsic Global Variables 3547========================== 3548 3549LLVM has a number of "magic" global variables that contain data that 3550affect code generation or other IR semantics. These are documented here. 3551All globals of this sort should have a section specified as 3552"``llvm.metadata``". This section and all globals that start with 3553"``llvm.``" are reserved for use by LLVM. 3554 3555.. _gv_llvmused: 3556 3557The '``llvm.used``' Global Variable 3558----------------------------------- 3559 3560The ``@llvm.used`` global is an array which has 3561:ref:`appending linkage <linkage_appending>`. This array contains a list of 3562pointers to named global variables, functions and aliases which may optionally 3563have a pointer cast formed of bitcast or getelementptr. For example, a legal 3564use of it is: 3565 3566.. code-block:: llvm 3567 3568 @X = global i8 4 3569 @Y = global i32 123 3570 3571 @llvm.used = appending global [2 x i8*] [ 3572 i8* @X, 3573 i8* bitcast (i32* @Y to i8*) 3574 ], section "llvm.metadata" 3575 3576If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler, 3577and linker are required to treat the symbol as if there is a reference to the 3578symbol that it cannot see (which is why they have to be named). For example, if 3579a variable has internal linkage and no references other than that from the 3580``@llvm.used`` list, it cannot be deleted. This is commonly used to represent 3581references from inline asms and other things the compiler cannot "see", and 3582corresponds to "``attribute((used))``" in GNU C. 3583 3584On some targets, the code generator must emit a directive to the 3585assembler or object file to prevent the assembler and linker from 3586molesting the symbol. 3587 3588.. _gv_llvmcompilerused: 3589 3590The '``llvm.compiler.used``' Global Variable 3591-------------------------------------------- 3592 3593The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used`` 3594directive, except that it only prevents the compiler from touching the 3595symbol. On targets that support it, this allows an intelligent linker to 3596optimize references to the symbol without being impeded as it would be 3597by ``@llvm.used``. 3598 3599This is a rare construct that should only be used in rare circumstances, 3600and should not be exposed to source languages. 3601 3602.. _gv_llvmglobalctors: 3603 3604The '``llvm.global_ctors``' Global Variable 3605------------------------------------------- 3606 3607.. code-block:: llvm 3608 3609 %0 = type { i32, void ()*, i8* } 3610 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor, i8* @data }] 3611 3612The ``@llvm.global_ctors`` array contains a list of constructor 3613functions, priorities, and an optional associated global or function. 3614The functions referenced by this array will be called in ascending order 3615of priority (i.e. lowest first) when the module is loaded. The order of 3616functions with the same priority is not defined. 3617 3618If the third field is present, non-null, and points to a global variable 3619or function, the initializer function will only run if the associated 3620data from the current module is not discarded. 3621 3622.. _llvmglobaldtors: 3623 3624The '``llvm.global_dtors``' Global Variable 3625------------------------------------------- 3626 3627.. code-block:: llvm 3628 3629 %0 = type { i32, void ()*, i8* } 3630 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor, i8* @data }] 3631 3632The ``@llvm.global_dtors`` array contains a list of destructor 3633functions, priorities, and an optional associated global or function. 3634The functions referenced by this array will be called in descending 3635order of priority (i.e. highest first) when the module is unloaded. The 3636order of functions with the same priority is not defined. 3637 3638If the third field is present, non-null, and points to a global variable 3639or function, the destructor function will only run if the associated 3640data from the current module is not discarded. 3641 3642Instruction Reference 3643===================== 3644 3645The LLVM instruction set consists of several different classifications 3646of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary 3647instructions <binaryops>`, :ref:`bitwise binary 3648instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and 3649:ref:`other instructions <otherops>`. 3650 3651.. _terminators: 3652 3653Terminator Instructions 3654----------------------- 3655 3656As mentioned :ref:`previously <functionstructure>`, every basic block in a 3657program ends with a "Terminator" instruction, which indicates which 3658block should be executed after the current block is finished. These 3659terminator instructions typically yield a '``void``' value: they produce 3660control flow, not values (the one exception being the 3661':ref:`invoke <i_invoke>`' instruction). 3662 3663The terminator instructions are: ':ref:`ret <i_ret>`', 3664':ref:`br <i_br>`', ':ref:`switch <i_switch>`', 3665':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`', 3666':ref:`resume <i_resume>`', and ':ref:`unreachable <i_unreachable>`'. 3667 3668.. _i_ret: 3669 3670'``ret``' Instruction 3671^^^^^^^^^^^^^^^^^^^^^ 3672 3673Syntax: 3674""""""" 3675 3676:: 3677 3678 ret <type> <value> ; Return a value from a non-void function 3679 ret void ; Return from void function 3680 3681Overview: 3682""""""""" 3683 3684The '``ret``' instruction is used to return control flow (and optionally 3685a value) from a function back to the caller. 3686 3687There are two forms of the '``ret``' instruction: one that returns a 3688value and then causes control flow, and one that just causes control 3689flow to occur. 3690 3691Arguments: 3692"""""""""" 3693 3694The '``ret``' instruction optionally accepts a single argument, the 3695return value. The type of the return value must be a ':ref:`first 3696class <t_firstclass>`' type. 3697 3698A function is not :ref:`well formed <wellformed>` if it it has a non-void 3699return type and contains a '``ret``' instruction with no return value or 3700a return value with a type that does not match its type, or if it has a 3701void return type and contains a '``ret``' instruction with a return 3702value. 3703 3704Semantics: 3705"""""""""" 3706 3707When the '``ret``' instruction is executed, control flow returns back to 3708the calling function's context. If the caller is a 3709":ref:`call <i_call>`" instruction, execution continues at the 3710instruction after the call. If the caller was an 3711":ref:`invoke <i_invoke>`" instruction, execution continues at the 3712beginning of the "normal" destination block. If the instruction returns 3713a value, that value shall set the call or invoke instruction's return 3714value. 3715 3716Example: 3717"""""""" 3718 3719.. code-block:: llvm 3720 3721 ret i32 5 ; Return an integer value of 5 3722 ret void ; Return from a void function 3723 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2 3724 3725.. _i_br: 3726 3727'``br``' Instruction 3728^^^^^^^^^^^^^^^^^^^^ 3729 3730Syntax: 3731""""""" 3732 3733:: 3734 3735 br i1 <cond>, label <iftrue>, label <iffalse> 3736 br label <dest> ; Unconditional branch 3737 3738Overview: 3739""""""""" 3740 3741The '``br``' instruction is used to cause control flow to transfer to a 3742different basic block in the current function. There are two forms of 3743this instruction, corresponding to a conditional branch and an 3744unconditional branch. 3745 3746Arguments: 3747"""""""""" 3748 3749The conditional branch form of the '``br``' instruction takes a single 3750'``i1``' value and two '``label``' values. The unconditional form of the 3751'``br``' instruction takes a single '``label``' value as a target. 3752 3753Semantics: 3754"""""""""" 3755 3756Upon execution of a conditional '``br``' instruction, the '``i1``' 3757argument is evaluated. If the value is ``true``, control flows to the 3758'``iftrue``' ``label`` argument. If "cond" is ``false``, control flows 3759to the '``iffalse``' ``label`` argument. 3760 3761Example: 3762"""""""" 3763 3764.. code-block:: llvm 3765 3766 Test: 3767 %cond = icmp eq i32 %a, %b 3768 br i1 %cond, label %IfEqual, label %IfUnequal 3769 IfEqual: 3770 ret i32 1 3771 IfUnequal: 3772 ret i32 0 3773 3774.. _i_switch: 3775 3776'``switch``' Instruction 3777^^^^^^^^^^^^^^^^^^^^^^^^ 3778 3779Syntax: 3780""""""" 3781 3782:: 3783 3784 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ] 3785 3786Overview: 3787""""""""" 3788 3789The '``switch``' instruction is used to transfer control flow to one of 3790several different places. It is a generalization of the '``br``' 3791instruction, allowing a branch to occur to one of many possible 3792destinations. 3793 3794Arguments: 3795"""""""""" 3796 3797The '``switch``' instruction uses three parameters: an integer 3798comparison value '``value``', a default '``label``' destination, and an 3799array of pairs of comparison value constants and '``label``'s. The table 3800is not allowed to contain duplicate constant entries. 3801 3802Semantics: 3803"""""""""" 3804 3805The ``switch`` instruction specifies a table of values and destinations. 3806When the '``switch``' instruction is executed, this table is searched 3807for the given value. If the value is found, control flow is transferred 3808to the corresponding destination; otherwise, control flow is transferred 3809to the default destination. 3810 3811Implementation: 3812""""""""""""""" 3813 3814Depending on properties of the target machine and the particular 3815``switch`` instruction, this instruction may be code generated in 3816different ways. For example, it could be generated as a series of 3817chained conditional branches or with a lookup table. 3818 3819Example: 3820"""""""" 3821 3822.. code-block:: llvm 3823 3824 ; Emulate a conditional br instruction 3825 %Val = zext i1 %value to i32 3826 switch i32 %Val, label %truedest [ i32 0, label %falsedest ] 3827 3828 ; Emulate an unconditional br instruction 3829 switch i32 0, label %dest [ ] 3830 3831 ; Implement a jump table: 3832 switch i32 %val, label %otherwise [ i32 0, label %onzero 3833 i32 1, label %onone 3834 i32 2, label %ontwo ] 3835 3836.. _i_indirectbr: 3837 3838'``indirectbr``' Instruction 3839^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3840 3841Syntax: 3842""""""" 3843 3844:: 3845 3846 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ] 3847 3848Overview: 3849""""""""" 3850 3851The '``indirectbr``' instruction implements an indirect branch to a 3852label within the current function, whose address is specified by 3853"``address``". Address must be derived from a 3854:ref:`blockaddress <blockaddress>` constant. 3855 3856Arguments: 3857"""""""""" 3858 3859The '``address``' argument is the address of the label to jump to. The 3860rest of the arguments indicate the full set of possible destinations 3861that the address may point to. Blocks are allowed to occur multiple 3862times in the destination list, though this isn't particularly useful. 3863 3864This destination list is required so that dataflow analysis has an 3865accurate understanding of the CFG. 3866 3867Semantics: 3868"""""""""" 3869 3870Control transfers to the block specified in the address argument. All 3871possible destination blocks must be listed in the label list, otherwise 3872this instruction has undefined behavior. This implies that jumps to 3873labels defined in other functions have undefined behavior as well. 3874 3875Implementation: 3876""""""""""""""" 3877 3878This is typically implemented with a jump through a register. 3879 3880Example: 3881"""""""" 3882 3883.. code-block:: llvm 3884 3885 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ] 3886 3887.. _i_invoke: 3888 3889'``invoke``' Instruction 3890^^^^^^^^^^^^^^^^^^^^^^^^ 3891 3892Syntax: 3893""""""" 3894 3895:: 3896 3897 <result> = invoke [cconv] [ret attrs] <ptr to function ty> <function ptr val>(<function args>) [fn attrs] 3898 to label <normal label> unwind label <exception label> 3899 3900Overview: 3901""""""""" 3902 3903The '``invoke``' instruction causes control to transfer to a specified 3904function, with the possibility of control flow transfer to either the 3905'``normal``' label or the '``exception``' label. If the callee function 3906returns with the "``ret``" instruction, control flow will return to the 3907"normal" label. If the callee (or any indirect callees) returns via the 3908":ref:`resume <i_resume>`" instruction or other exception handling 3909mechanism, control is interrupted and continued at the dynamically 3910nearest "exception" label. 3911 3912The '``exception``' label is a `landing 3913pad <ExceptionHandling.html#overview>`_ for the exception. As such, 3914'``exception``' label is required to have the 3915":ref:`landingpad <i_landingpad>`" instruction, which contains the 3916information about the behavior of the program after unwinding happens, 3917as its first non-PHI instruction. The restrictions on the 3918"``landingpad``" instruction's tightly couples it to the "``invoke``" 3919instruction, so that the important information contained within the 3920"``landingpad``" instruction can't be lost through normal code motion. 3921 3922Arguments: 3923"""""""""" 3924 3925This instruction requires several arguments: 3926 3927#. The optional "cconv" marker indicates which :ref:`calling 3928 convention <callingconv>` the call should use. If none is 3929 specified, the call defaults to using C calling conventions. 3930#. The optional :ref:`Parameter Attributes <paramattrs>` list for return 3931 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes 3932 are valid here. 3933#. '``ptr to function ty``': shall be the signature of the pointer to 3934 function value being invoked. In most cases, this is a direct 3935 function invocation, but indirect ``invoke``'s are just as possible, 3936 branching off an arbitrary pointer to function value. 3937#. '``function ptr val``': An LLVM value containing a pointer to a 3938 function to be invoked. 3939#. '``function args``': argument list whose types match the function 3940 signature argument types and parameter attributes. All arguments must 3941 be of :ref:`first class <t_firstclass>` type. If the function signature 3942 indicates the function accepts a variable number of arguments, the 3943 extra arguments can be specified. 3944#. '``normal label``': the label reached when the called function 3945 executes a '``ret``' instruction. 3946#. '``exception label``': the label reached when a callee returns via 3947 the :ref:`resume <i_resume>` instruction or other exception handling 3948 mechanism. 3949#. The optional :ref:`function attributes <fnattrs>` list. Only 3950 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``' 3951 attributes are valid here. 3952 3953Semantics: 3954"""""""""" 3955 3956This instruction is designed to operate as a standard '``call``' 3957instruction in most regards. The primary difference is that it 3958establishes an association with a label, which is used by the runtime 3959library to unwind the stack. 3960 3961This instruction is used in languages with destructors to ensure that 3962proper cleanup is performed in the case of either a ``longjmp`` or a 3963thrown exception. Additionally, this is important for implementation of 3964'``catch``' clauses in high-level languages that support them. 3965 3966For the purposes of the SSA form, the definition of the value returned 3967by the '``invoke``' instruction is deemed to occur on the edge from the 3968current block to the "normal" label. If the callee unwinds then no 3969return value is available. 3970 3971Example: 3972"""""""" 3973 3974.. code-block:: llvm 3975 3976 %retval = invoke i32 @Test(i32 15) to label %Continue 3977 unwind label %TestCleanup ; i32:retval set 3978 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue 3979 unwind label %TestCleanup ; i32:retval set 3980 3981.. _i_resume: 3982 3983'``resume``' Instruction 3984^^^^^^^^^^^^^^^^^^^^^^^^ 3985 3986Syntax: 3987""""""" 3988 3989:: 3990 3991 resume <type> <value> 3992 3993Overview: 3994""""""""" 3995 3996The '``resume``' instruction is a terminator instruction that has no 3997successors. 3998 3999Arguments: 4000"""""""""" 4001 4002The '``resume``' instruction requires one argument, which must have the 4003same type as the result of any '``landingpad``' instruction in the same 4004function. 4005 4006Semantics: 4007"""""""""" 4008 4009The '``resume``' instruction resumes propagation of an existing 4010(in-flight) exception whose unwinding was interrupted with a 4011:ref:`landingpad <i_landingpad>` instruction. 4012 4013Example: 4014"""""""" 4015 4016.. code-block:: llvm 4017 4018 resume { i8*, i32 } %exn 4019 4020.. _i_unreachable: 4021 4022'``unreachable``' Instruction 4023^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 4024 4025Syntax: 4026""""""" 4027 4028:: 4029 4030 unreachable 4031 4032Overview: 4033""""""""" 4034 4035The '``unreachable``' instruction has no defined semantics. This 4036instruction is used to inform the optimizer that a particular portion of 4037the code is not reachable. This can be used to indicate that the code 4038after a no-return function cannot be reached, and other facts. 4039 4040Semantics: 4041"""""""""" 4042 4043The '``unreachable``' instruction has no defined semantics. 4044 4045.. _binaryops: 4046 4047Binary Operations 4048----------------- 4049 4050Binary operators are used to do most of the computation in a program. 4051They require two operands of the same type, execute an operation on 4052them, and produce a single value. The operands might represent multiple 4053data, as is the case with the :ref:`vector <t_vector>` data type. The 4054result value has the same type as its operands. 4055 4056There are several different binary operators: 4057 4058.. _i_add: 4059 4060'``add``' Instruction 4061^^^^^^^^^^^^^^^^^^^^^ 4062 4063Syntax: 4064""""""" 4065 4066:: 4067 4068 <result> = add <ty> <op1>, <op2> ; yields ty:result 4069 <result> = add nuw <ty> <op1>, <op2> ; yields ty:result 4070 <result> = add nsw <ty> <op1>, <op2> ; yields ty:result 4071 <result> = add nuw nsw <ty> <op1>, <op2> ; yields ty:result 4072 4073Overview: 4074""""""""" 4075 4076The '``add``' instruction returns the sum of its two operands. 4077 4078Arguments: 4079"""""""""" 4080 4081The two arguments to the '``add``' instruction must be 4082:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4083arguments must have identical types. 4084 4085Semantics: 4086"""""""""" 4087 4088The value produced is the integer sum of the two operands. 4089 4090If the sum has unsigned overflow, the result returned is the 4091mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of 4092the result. 4093 4094Because LLVM integers use a two's complement representation, this 4095instruction is appropriate for both signed and unsigned integers. 4096 4097``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 4098respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 4099result value of the ``add`` is a :ref:`poison value <poisonvalues>` if 4100unsigned and/or signed overflow, respectively, occurs. 4101 4102Example: 4103"""""""" 4104 4105.. code-block:: llvm 4106 4107 <result> = add i32 4, %var ; yields i32:result = 4 + %var 4108 4109.. _i_fadd: 4110 4111'``fadd``' Instruction 4112^^^^^^^^^^^^^^^^^^^^^^ 4113 4114Syntax: 4115""""""" 4116 4117:: 4118 4119 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 4120 4121Overview: 4122""""""""" 4123 4124The '``fadd``' instruction returns the sum of its two operands. 4125 4126Arguments: 4127"""""""""" 4128 4129The two arguments to the '``fadd``' instruction must be :ref:`floating 4130point <t_floating>` or :ref:`vector <t_vector>` of floating point values. 4131Both arguments must have identical types. 4132 4133Semantics: 4134"""""""""" 4135 4136The value produced is the floating point sum of the two operands. This 4137instruction can also take any number of :ref:`fast-math flags <fastmath>`, 4138which are optimization hints to enable otherwise unsafe floating point 4139optimizations: 4140 4141Example: 4142"""""""" 4143 4144.. code-block:: llvm 4145 4146 <result> = fadd float 4.0, %var ; yields float:result = 4.0 + %var 4147 4148'``sub``' Instruction 4149^^^^^^^^^^^^^^^^^^^^^ 4150 4151Syntax: 4152""""""" 4153 4154:: 4155 4156 <result> = sub <ty> <op1>, <op2> ; yields ty:result 4157 <result> = sub nuw <ty> <op1>, <op2> ; yields ty:result 4158 <result> = sub nsw <ty> <op1>, <op2> ; yields ty:result 4159 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields ty:result 4160 4161Overview: 4162""""""""" 4163 4164The '``sub``' instruction returns the difference of its two operands. 4165 4166Note that the '``sub``' instruction is used to represent the '``neg``' 4167instruction present in most other intermediate representations. 4168 4169Arguments: 4170"""""""""" 4171 4172The two arguments to the '``sub``' instruction must be 4173:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4174arguments must have identical types. 4175 4176Semantics: 4177"""""""""" 4178 4179The value produced is the integer difference of the two operands. 4180 4181If the difference has unsigned overflow, the result returned is the 4182mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of 4183the result. 4184 4185Because LLVM integers use a two's complement representation, this 4186instruction is appropriate for both signed and unsigned integers. 4187 4188``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 4189respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 4190result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if 4191unsigned and/or signed overflow, respectively, occurs. 4192 4193Example: 4194"""""""" 4195 4196.. code-block:: llvm 4197 4198 <result> = sub i32 4, %var ; yields i32:result = 4 - %var 4199 <result> = sub i32 0, %val ; yields i32:result = -%var 4200 4201.. _i_fsub: 4202 4203'``fsub``' Instruction 4204^^^^^^^^^^^^^^^^^^^^^^ 4205 4206Syntax: 4207""""""" 4208 4209:: 4210 4211 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 4212 4213Overview: 4214""""""""" 4215 4216The '``fsub``' instruction returns the difference of its two operands. 4217 4218Note that the '``fsub``' instruction is used to represent the '``fneg``' 4219instruction present in most other intermediate representations. 4220 4221Arguments: 4222"""""""""" 4223 4224The two arguments to the '``fsub``' instruction must be :ref:`floating 4225point <t_floating>` or :ref:`vector <t_vector>` of floating point values. 4226Both arguments must have identical types. 4227 4228Semantics: 4229"""""""""" 4230 4231The value produced is the floating point difference of the two operands. 4232This instruction can also take any number of :ref:`fast-math 4233flags <fastmath>`, which are optimization hints to enable otherwise 4234unsafe floating point optimizations: 4235 4236Example: 4237"""""""" 4238 4239.. code-block:: llvm 4240 4241 <result> = fsub float 4.0, %var ; yields float:result = 4.0 - %var 4242 <result> = fsub float -0.0, %val ; yields float:result = -%var 4243 4244'``mul``' Instruction 4245^^^^^^^^^^^^^^^^^^^^^ 4246 4247Syntax: 4248""""""" 4249 4250:: 4251 4252 <result> = mul <ty> <op1>, <op2> ; yields ty:result 4253 <result> = mul nuw <ty> <op1>, <op2> ; yields ty:result 4254 <result> = mul nsw <ty> <op1>, <op2> ; yields ty:result 4255 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields ty:result 4256 4257Overview: 4258""""""""" 4259 4260The '``mul``' instruction returns the product of its two operands. 4261 4262Arguments: 4263"""""""""" 4264 4265The two arguments to the '``mul``' instruction must be 4266:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4267arguments must have identical types. 4268 4269Semantics: 4270"""""""""" 4271 4272The value produced is the integer product of the two operands. 4273 4274If the result of the multiplication has unsigned overflow, the result 4275returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the 4276bit width of the result. 4277 4278Because LLVM integers use a two's complement representation, and the 4279result is the same width as the operands, this instruction returns the 4280correct result for both signed and unsigned integers. If a full product 4281(e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be 4282sign-extended or zero-extended as appropriate to the width of the full 4283product. 4284 4285``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 4286respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 4287result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if 4288unsigned and/or signed overflow, respectively, occurs. 4289 4290Example: 4291"""""""" 4292 4293.. code-block:: llvm 4294 4295 <result> = mul i32 4, %var ; yields i32:result = 4 * %var 4296 4297.. _i_fmul: 4298 4299'``fmul``' Instruction 4300^^^^^^^^^^^^^^^^^^^^^^ 4301 4302Syntax: 4303""""""" 4304 4305:: 4306 4307 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 4308 4309Overview: 4310""""""""" 4311 4312The '``fmul``' instruction returns the product of its two operands. 4313 4314Arguments: 4315"""""""""" 4316 4317The two arguments to the '``fmul``' instruction must be :ref:`floating 4318point <t_floating>` or :ref:`vector <t_vector>` of floating point values. 4319Both arguments must have identical types. 4320 4321Semantics: 4322"""""""""" 4323 4324The value produced is the floating point product of the two operands. 4325This instruction can also take any number of :ref:`fast-math 4326flags <fastmath>`, which are optimization hints to enable otherwise 4327unsafe floating point optimizations: 4328 4329Example: 4330"""""""" 4331 4332.. code-block:: llvm 4333 4334 <result> = fmul float 4.0, %var ; yields float:result = 4.0 * %var 4335 4336'``udiv``' Instruction 4337^^^^^^^^^^^^^^^^^^^^^^ 4338 4339Syntax: 4340""""""" 4341 4342:: 4343 4344 <result> = udiv <ty> <op1>, <op2> ; yields ty:result 4345 <result> = udiv exact <ty> <op1>, <op2> ; yields ty:result 4346 4347Overview: 4348""""""""" 4349 4350The '``udiv``' instruction returns the quotient of its two operands. 4351 4352Arguments: 4353"""""""""" 4354 4355The two arguments to the '``udiv``' instruction must be 4356:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4357arguments must have identical types. 4358 4359Semantics: 4360"""""""""" 4361 4362The value produced is the unsigned integer quotient of the two operands. 4363 4364Note that unsigned integer division and signed integer division are 4365distinct operations; for signed integer division, use '``sdiv``'. 4366 4367Division by zero leads to undefined behavior. 4368 4369If the ``exact`` keyword is present, the result value of the ``udiv`` is 4370a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as 4371such, "((a udiv exact b) mul b) == a"). 4372 4373Example: 4374"""""""" 4375 4376.. code-block:: llvm 4377 4378 <result> = udiv i32 4, %var ; yields i32:result = 4 / %var 4379 4380'``sdiv``' Instruction 4381^^^^^^^^^^^^^^^^^^^^^^ 4382 4383Syntax: 4384""""""" 4385 4386:: 4387 4388 <result> = sdiv <ty> <op1>, <op2> ; yields ty:result 4389 <result> = sdiv exact <ty> <op1>, <op2> ; yields ty:result 4390 4391Overview: 4392""""""""" 4393 4394The '``sdiv``' instruction returns the quotient of its two operands. 4395 4396Arguments: 4397"""""""""" 4398 4399The two arguments to the '``sdiv``' instruction must be 4400:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4401arguments must have identical types. 4402 4403Semantics: 4404"""""""""" 4405 4406The value produced is the signed integer quotient of the two operands 4407rounded towards zero. 4408 4409Note that signed integer division and unsigned integer division are 4410distinct operations; for unsigned integer division, use '``udiv``'. 4411 4412Division by zero leads to undefined behavior. Overflow also leads to 4413undefined behavior; this is a rare case, but can occur, for example, by 4414doing a 32-bit division of -2147483648 by -1. 4415 4416If the ``exact`` keyword is present, the result value of the ``sdiv`` is 4417a :ref:`poison value <poisonvalues>` if the result would be rounded. 4418 4419Example: 4420"""""""" 4421 4422.. code-block:: llvm 4423 4424 <result> = sdiv i32 4, %var ; yields i32:result = 4 / %var 4425 4426.. _i_fdiv: 4427 4428'``fdiv``' Instruction 4429^^^^^^^^^^^^^^^^^^^^^^ 4430 4431Syntax: 4432""""""" 4433 4434:: 4435 4436 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 4437 4438Overview: 4439""""""""" 4440 4441The '``fdiv``' instruction returns the quotient of its two operands. 4442 4443Arguments: 4444"""""""""" 4445 4446The two arguments to the '``fdiv``' instruction must be :ref:`floating 4447point <t_floating>` or :ref:`vector <t_vector>` of floating point values. 4448Both arguments must have identical types. 4449 4450Semantics: 4451"""""""""" 4452 4453The value produced is the floating point quotient of the two operands. 4454This instruction can also take any number of :ref:`fast-math 4455flags <fastmath>`, which are optimization hints to enable otherwise 4456unsafe floating point optimizations: 4457 4458Example: 4459"""""""" 4460 4461.. code-block:: llvm 4462 4463 <result> = fdiv float 4.0, %var ; yields float:result = 4.0 / %var 4464 4465'``urem``' Instruction 4466^^^^^^^^^^^^^^^^^^^^^^ 4467 4468Syntax: 4469""""""" 4470 4471:: 4472 4473 <result> = urem <ty> <op1>, <op2> ; yields ty:result 4474 4475Overview: 4476""""""""" 4477 4478The '``urem``' instruction returns the remainder from the unsigned 4479division of its two arguments. 4480 4481Arguments: 4482"""""""""" 4483 4484The two arguments to the '``urem``' instruction must be 4485:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4486arguments must have identical types. 4487 4488Semantics: 4489"""""""""" 4490 4491This instruction returns the unsigned integer *remainder* of a division. 4492This instruction always performs an unsigned division to get the 4493remainder. 4494 4495Note that unsigned integer remainder and signed integer remainder are 4496distinct operations; for signed integer remainder, use '``srem``'. 4497 4498Taking the remainder of a division by zero leads to undefined behavior. 4499 4500Example: 4501"""""""" 4502 4503.. code-block:: llvm 4504 4505 <result> = urem i32 4, %var ; yields i32:result = 4 % %var 4506 4507'``srem``' Instruction 4508^^^^^^^^^^^^^^^^^^^^^^ 4509 4510Syntax: 4511""""""" 4512 4513:: 4514 4515 <result> = srem <ty> <op1>, <op2> ; yields ty:result 4516 4517Overview: 4518""""""""" 4519 4520The '``srem``' instruction returns the remainder from the signed 4521division of its two operands. This instruction can also take 4522:ref:`vector <t_vector>` versions of the values in which case the elements 4523must be integers. 4524 4525Arguments: 4526"""""""""" 4527 4528The two arguments to the '``srem``' instruction must be 4529:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4530arguments must have identical types. 4531 4532Semantics: 4533"""""""""" 4534 4535This instruction returns the *remainder* of a division (where the result 4536is either zero or has the same sign as the dividend, ``op1``), not the 4537*modulo* operator (where the result is either zero or has the same sign 4538as the divisor, ``op2``) of a value. For more information about the 4539difference, see `The Math 4540Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a 4541table of how this is implemented in various languages, please see 4542`Wikipedia: modulo 4543operation <http://en.wikipedia.org/wiki/Modulo_operation>`_. 4544 4545Note that signed integer remainder and unsigned integer remainder are 4546distinct operations; for unsigned integer remainder, use '``urem``'. 4547 4548Taking the remainder of a division by zero leads to undefined behavior. 4549Overflow also leads to undefined behavior; this is a rare case, but can 4550occur, for example, by taking the remainder of a 32-bit division of 4551-2147483648 by -1. (The remainder doesn't actually overflow, but this 4552rule lets srem be implemented using instructions that return both the 4553result of the division and the remainder.) 4554 4555Example: 4556"""""""" 4557 4558.. code-block:: llvm 4559 4560 <result> = srem i32 4, %var ; yields i32:result = 4 % %var 4561 4562.. _i_frem: 4563 4564'``frem``' Instruction 4565^^^^^^^^^^^^^^^^^^^^^^ 4566 4567Syntax: 4568""""""" 4569 4570:: 4571 4572 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 4573 4574Overview: 4575""""""""" 4576 4577The '``frem``' instruction returns the remainder from the division of 4578its two operands. 4579 4580Arguments: 4581"""""""""" 4582 4583The two arguments to the '``frem``' instruction must be :ref:`floating 4584point <t_floating>` or :ref:`vector <t_vector>` of floating point values. 4585Both arguments must have identical types. 4586 4587Semantics: 4588"""""""""" 4589 4590This instruction returns the *remainder* of a division. The remainder 4591has the same sign as the dividend. This instruction can also take any 4592number of :ref:`fast-math flags <fastmath>`, which are optimization hints 4593to enable otherwise unsafe floating point optimizations: 4594 4595Example: 4596"""""""" 4597 4598.. code-block:: llvm 4599 4600 <result> = frem float 4.0, %var ; yields float:result = 4.0 % %var 4601 4602.. _bitwiseops: 4603 4604Bitwise Binary Operations 4605------------------------- 4606 4607Bitwise binary operators are used to do various forms of bit-twiddling 4608in a program. They are generally very efficient instructions and can 4609commonly be strength reduced from other instructions. They require two 4610operands of the same type, execute an operation on them, and produce a 4611single value. The resulting value is the same type as its operands. 4612 4613'``shl``' Instruction 4614^^^^^^^^^^^^^^^^^^^^^ 4615 4616Syntax: 4617""""""" 4618 4619:: 4620 4621 <result> = shl <ty> <op1>, <op2> ; yields ty:result 4622 <result> = shl nuw <ty> <op1>, <op2> ; yields ty:result 4623 <result> = shl nsw <ty> <op1>, <op2> ; yields ty:result 4624 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields ty:result 4625 4626Overview: 4627""""""""" 4628 4629The '``shl``' instruction returns the first operand shifted to the left 4630a specified number of bits. 4631 4632Arguments: 4633"""""""""" 4634 4635Both arguments to the '``shl``' instruction must be the same 4636:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 4637'``op2``' is treated as an unsigned value. 4638 4639Semantics: 4640"""""""""" 4641 4642The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`, 4643where ``n`` is the width of the result. If ``op2`` is (statically or 4644dynamically) negative or equal to or larger than the number of bits in 4645``op1``, the result is undefined. If the arguments are vectors, each 4646vector element of ``op1`` is shifted by the corresponding shift amount 4647in ``op2``. 4648 4649If the ``nuw`` keyword is present, then the shift produces a :ref:`poison 4650value <poisonvalues>` if it shifts out any non-zero bits. If the 4651``nsw`` keyword is present, then the shift produces a :ref:`poison 4652value <poisonvalues>` if it shifts out any bits that disagree with the 4653resultant sign bit. As such, NUW/NSW have the same semantics as they 4654would if the shift were expressed as a mul instruction with the same 4655nsw/nuw bits in (mul %op1, (shl 1, %op2)). 4656 4657Example: 4658"""""""" 4659 4660.. code-block:: llvm 4661 4662 <result> = shl i32 4, %var ; yields i32: 4 << %var 4663 <result> = shl i32 4, 2 ; yields i32: 16 4664 <result> = shl i32 1, 10 ; yields i32: 1024 4665 <result> = shl i32 1, 32 ; undefined 4666 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4> 4667 4668'``lshr``' Instruction 4669^^^^^^^^^^^^^^^^^^^^^^ 4670 4671Syntax: 4672""""""" 4673 4674:: 4675 4676 <result> = lshr <ty> <op1>, <op2> ; yields ty:result 4677 <result> = lshr exact <ty> <op1>, <op2> ; yields ty:result 4678 4679Overview: 4680""""""""" 4681 4682The '``lshr``' instruction (logical shift right) returns the first 4683operand shifted to the right a specified number of bits with zero fill. 4684 4685Arguments: 4686"""""""""" 4687 4688Both arguments to the '``lshr``' instruction must be the same 4689:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 4690'``op2``' is treated as an unsigned value. 4691 4692Semantics: 4693"""""""""" 4694 4695This instruction always performs a logical shift right operation. The 4696most significant bits of the result will be filled with zero bits after 4697the shift. If ``op2`` is (statically or dynamically) equal to or larger 4698than the number of bits in ``op1``, the result is undefined. If the 4699arguments are vectors, each vector element of ``op1`` is shifted by the 4700corresponding shift amount in ``op2``. 4701 4702If the ``exact`` keyword is present, the result value of the ``lshr`` is 4703a :ref:`poison value <poisonvalues>` if any of the bits shifted out are 4704non-zero. 4705 4706Example: 4707"""""""" 4708 4709.. code-block:: llvm 4710 4711 <result> = lshr i32 4, 1 ; yields i32:result = 2 4712 <result> = lshr i32 4, 2 ; yields i32:result = 1 4713 <result> = lshr i8 4, 3 ; yields i8:result = 0 4714 <result> = lshr i8 -2, 1 ; yields i8:result = 0x7F 4715 <result> = lshr i32 1, 32 ; undefined 4716 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1> 4717 4718'``ashr``' Instruction 4719^^^^^^^^^^^^^^^^^^^^^^ 4720 4721Syntax: 4722""""""" 4723 4724:: 4725 4726 <result> = ashr <ty> <op1>, <op2> ; yields ty:result 4727 <result> = ashr exact <ty> <op1>, <op2> ; yields ty:result 4728 4729Overview: 4730""""""""" 4731 4732The '``ashr``' instruction (arithmetic shift right) returns the first 4733operand shifted to the right a specified number of bits with sign 4734extension. 4735 4736Arguments: 4737"""""""""" 4738 4739Both arguments to the '``ashr``' instruction must be the same 4740:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 4741'``op2``' is treated as an unsigned value. 4742 4743Semantics: 4744"""""""""" 4745 4746This instruction always performs an arithmetic shift right operation, 4747The most significant bits of the result will be filled with the sign bit 4748of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger 4749than the number of bits in ``op1``, the result is undefined. If the 4750arguments are vectors, each vector element of ``op1`` is shifted by the 4751corresponding shift amount in ``op2``. 4752 4753If the ``exact`` keyword is present, the result value of the ``ashr`` is 4754a :ref:`poison value <poisonvalues>` if any of the bits shifted out are 4755non-zero. 4756 4757Example: 4758"""""""" 4759 4760.. code-block:: llvm 4761 4762 <result> = ashr i32 4, 1 ; yields i32:result = 2 4763 <result> = ashr i32 4, 2 ; yields i32:result = 1 4764 <result> = ashr i8 4, 3 ; yields i8:result = 0 4765 <result> = ashr i8 -2, 1 ; yields i8:result = -1 4766 <result> = ashr i32 1, 32 ; undefined 4767 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0> 4768 4769'``and``' Instruction 4770^^^^^^^^^^^^^^^^^^^^^ 4771 4772Syntax: 4773""""""" 4774 4775:: 4776 4777 <result> = and <ty> <op1>, <op2> ; yields ty:result 4778 4779Overview: 4780""""""""" 4781 4782The '``and``' instruction returns the bitwise logical and of its two 4783operands. 4784 4785Arguments: 4786"""""""""" 4787 4788The two arguments to the '``and``' instruction must be 4789:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4790arguments must have identical types. 4791 4792Semantics: 4793"""""""""" 4794 4795The truth table used for the '``and``' instruction is: 4796 4797+-----+-----+-----+ 4798| In0 | In1 | Out | 4799+-----+-----+-----+ 4800| 0 | 0 | 0 | 4801+-----+-----+-----+ 4802| 0 | 1 | 0 | 4803+-----+-----+-----+ 4804| 1 | 0 | 0 | 4805+-----+-----+-----+ 4806| 1 | 1 | 1 | 4807+-----+-----+-----+ 4808 4809Example: 4810"""""""" 4811 4812.. code-block:: llvm 4813 4814 <result> = and i32 4, %var ; yields i32:result = 4 & %var 4815 <result> = and i32 15, 40 ; yields i32:result = 8 4816 <result> = and i32 4, 8 ; yields i32:result = 0 4817 4818'``or``' Instruction 4819^^^^^^^^^^^^^^^^^^^^ 4820 4821Syntax: 4822""""""" 4823 4824:: 4825 4826 <result> = or <ty> <op1>, <op2> ; yields ty:result 4827 4828Overview: 4829""""""""" 4830 4831The '``or``' instruction returns the bitwise logical inclusive or of its 4832two operands. 4833 4834Arguments: 4835"""""""""" 4836 4837The two arguments to the '``or``' instruction must be 4838:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4839arguments must have identical types. 4840 4841Semantics: 4842"""""""""" 4843 4844The truth table used for the '``or``' instruction is: 4845 4846+-----+-----+-----+ 4847| In0 | In1 | Out | 4848+-----+-----+-----+ 4849| 0 | 0 | 0 | 4850+-----+-----+-----+ 4851| 0 | 1 | 1 | 4852+-----+-----+-----+ 4853| 1 | 0 | 1 | 4854+-----+-----+-----+ 4855| 1 | 1 | 1 | 4856+-----+-----+-----+ 4857 4858Example: 4859"""""""" 4860 4861:: 4862 4863 <result> = or i32 4, %var ; yields i32:result = 4 | %var 4864 <result> = or i32 15, 40 ; yields i32:result = 47 4865 <result> = or i32 4, 8 ; yields i32:result = 12 4866 4867'``xor``' Instruction 4868^^^^^^^^^^^^^^^^^^^^^ 4869 4870Syntax: 4871""""""" 4872 4873:: 4874 4875 <result> = xor <ty> <op1>, <op2> ; yields ty:result 4876 4877Overview: 4878""""""""" 4879 4880The '``xor``' instruction returns the bitwise logical exclusive or of 4881its two operands. The ``xor`` is used to implement the "one's 4882complement" operation, which is the "~" operator in C. 4883 4884Arguments: 4885"""""""""" 4886 4887The two arguments to the '``xor``' instruction must be 4888:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 4889arguments must have identical types. 4890 4891Semantics: 4892"""""""""" 4893 4894The truth table used for the '``xor``' instruction is: 4895 4896+-----+-----+-----+ 4897| In0 | In1 | Out | 4898+-----+-----+-----+ 4899| 0 | 0 | 0 | 4900+-----+-----+-----+ 4901| 0 | 1 | 1 | 4902+-----+-----+-----+ 4903| 1 | 0 | 1 | 4904+-----+-----+-----+ 4905| 1 | 1 | 0 | 4906+-----+-----+-----+ 4907 4908Example: 4909"""""""" 4910 4911.. code-block:: llvm 4912 4913 <result> = xor i32 4, %var ; yields i32:result = 4 ^ %var 4914 <result> = xor i32 15, 40 ; yields i32:result = 39 4915 <result> = xor i32 4, 8 ; yields i32:result = 12 4916 <result> = xor i32 %V, -1 ; yields i32:result = ~%V 4917 4918Vector Operations 4919----------------- 4920 4921LLVM supports several instructions to represent vector operations in a 4922target-independent manner. These instructions cover the element-access 4923and vector-specific operations needed to process vectors effectively. 4924While LLVM does directly support these vector operations, many 4925sophisticated algorithms will want to use target-specific intrinsics to 4926take full advantage of a specific target. 4927 4928.. _i_extractelement: 4929 4930'``extractelement``' Instruction 4931^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 4932 4933Syntax: 4934""""""" 4935 4936:: 4937 4938 <result> = extractelement <n x <ty>> <val>, <ty2> <idx> ; yields <ty> 4939 4940Overview: 4941""""""""" 4942 4943The '``extractelement``' instruction extracts a single scalar element 4944from a vector at a specified index. 4945 4946Arguments: 4947"""""""""" 4948 4949The first operand of an '``extractelement``' instruction is a value of 4950:ref:`vector <t_vector>` type. The second operand is an index indicating 4951the position from which to extract the element. The index may be a 4952variable of any integer type. 4953 4954Semantics: 4955"""""""""" 4956 4957The result is a scalar of the same type as the element type of ``val``. 4958Its value is the value at position ``idx`` of ``val``. If ``idx`` 4959exceeds the length of ``val``, the results are undefined. 4960 4961Example: 4962"""""""" 4963 4964.. code-block:: llvm 4965 4966 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32 4967 4968.. _i_insertelement: 4969 4970'``insertelement``' Instruction 4971^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 4972 4973Syntax: 4974""""""" 4975 4976:: 4977 4978 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <n x <ty>> 4979 4980Overview: 4981""""""""" 4982 4983The '``insertelement``' instruction inserts a scalar element into a 4984vector at a specified index. 4985 4986Arguments: 4987"""""""""" 4988 4989The first operand of an '``insertelement``' instruction is a value of 4990:ref:`vector <t_vector>` type. The second operand is a scalar value whose 4991type must equal the element type of the first operand. The third operand 4992is an index indicating the position at which to insert the value. The 4993index may be a variable of any integer type. 4994 4995Semantics: 4996"""""""""" 4997 4998The result is a vector of the same type as ``val``. Its element values 4999are those of ``val`` except at position ``idx``, where it gets the value 5000``elt``. If ``idx`` exceeds the length of ``val``, the results are 5001undefined. 5002 5003Example: 5004"""""""" 5005 5006.. code-block:: llvm 5007 5008 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32> 5009 5010.. _i_shufflevector: 5011 5012'``shufflevector``' Instruction 5013^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5014 5015Syntax: 5016""""""" 5017 5018:: 5019 5020 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>> 5021 5022Overview: 5023""""""""" 5024 5025The '``shufflevector``' instruction constructs a permutation of elements 5026from two input vectors, returning a vector with the same element type as 5027the input and length that is the same as the shuffle mask. 5028 5029Arguments: 5030"""""""""" 5031 5032The first two operands of a '``shufflevector``' instruction are vectors 5033with the same type. The third argument is a shuffle mask whose element 5034type is always 'i32'. The result of the instruction is a vector whose 5035length is the same as the shuffle mask and whose element type is the 5036same as the element type of the first two operands. 5037 5038The shuffle mask operand is required to be a constant vector with either 5039constant integer or undef values. 5040 5041Semantics: 5042"""""""""" 5043 5044The elements of the two input vectors are numbered from left to right 5045across both of the vectors. The shuffle mask operand specifies, for each 5046element of the result vector, which element of the two input vectors the 5047result element gets. The element selector may be undef (meaning "don't 5048care") and the second operand may be undef if performing a shuffle from 5049only one vector. 5050 5051Example: 5052"""""""" 5053 5054.. code-block:: llvm 5055 5056 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2, 5057 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32> 5058 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef, 5059 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle. 5060 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef, 5061 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> 5062 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2, 5063 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32> 5064 5065Aggregate Operations 5066-------------------- 5067 5068LLVM supports several instructions for working with 5069:ref:`aggregate <t_aggregate>` values. 5070 5071.. _i_extractvalue: 5072 5073'``extractvalue``' Instruction 5074^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5075 5076Syntax: 5077""""""" 5078 5079:: 5080 5081 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}* 5082 5083Overview: 5084""""""""" 5085 5086The '``extractvalue``' instruction extracts the value of a member field 5087from an :ref:`aggregate <t_aggregate>` value. 5088 5089Arguments: 5090"""""""""" 5091 5092The first operand of an '``extractvalue``' instruction is a value of 5093:ref:`struct <t_struct>` or :ref:`array <t_array>` type. The operands are 5094constant indices to specify which value to extract in a similar manner 5095as indices in a '``getelementptr``' instruction. 5096 5097The major differences to ``getelementptr`` indexing are: 5098 5099- Since the value being indexed is not a pointer, the first index is 5100 omitted and assumed to be zero. 5101- At least one index must be specified. 5102- Not only struct indices but also array indices must be in bounds. 5103 5104Semantics: 5105"""""""""" 5106 5107The result is the value at the position in the aggregate specified by 5108the index operands. 5109 5110Example: 5111"""""""" 5112 5113.. code-block:: llvm 5114 5115 <result> = extractvalue {i32, float} %agg, 0 ; yields i32 5116 5117.. _i_insertvalue: 5118 5119'``insertvalue``' Instruction 5120^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5121 5122Syntax: 5123""""""" 5124 5125:: 5126 5127 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type> 5128 5129Overview: 5130""""""""" 5131 5132The '``insertvalue``' instruction inserts a value into a member field in 5133an :ref:`aggregate <t_aggregate>` value. 5134 5135Arguments: 5136"""""""""" 5137 5138The first operand of an '``insertvalue``' instruction is a value of 5139:ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is 5140a first-class value to insert. The following operands are constant 5141indices indicating the position at which to insert the value in a 5142similar manner as indices in a '``extractvalue``' instruction. The value 5143to insert must have the same type as the value identified by the 5144indices. 5145 5146Semantics: 5147"""""""""" 5148 5149The result is an aggregate of the same type as ``val``. Its value is 5150that of ``val`` except that the value at the position specified by the 5151indices is that of ``elt``. 5152 5153Example: 5154"""""""" 5155 5156.. code-block:: llvm 5157 5158 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef} 5159 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val} 5160 %agg3 = insertvalue {i32, {float}} undef, float %val, 1, 0 ; yields {i32 undef, {float %val}} 5161 5162.. _memoryops: 5163 5164Memory Access and Addressing Operations 5165--------------------------------------- 5166 5167A key design point of an SSA-based representation is how it represents 5168memory. In LLVM, no memory locations are in SSA form, which makes things 5169very simple. This section describes how to read, write, and allocate 5170memory in LLVM. 5171 5172.. _i_alloca: 5173 5174'``alloca``' Instruction 5175^^^^^^^^^^^^^^^^^^^^^^^^ 5176 5177Syntax: 5178""""""" 5179 5180:: 5181 5182 <result> = alloca [inalloca] <type> [, <ty> <NumElements>] [, align <alignment>] ; yields type*:result 5183 5184Overview: 5185""""""""" 5186 5187The '``alloca``' instruction allocates memory on the stack frame of the 5188currently executing function, to be automatically released when this 5189function returns to its caller. The object is always allocated in the 5190generic address space (address space zero). 5191 5192Arguments: 5193"""""""""" 5194 5195The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements`` 5196bytes of memory on the runtime stack, returning a pointer of the 5197appropriate type to the program. If "NumElements" is specified, it is 5198the number of elements allocated, otherwise "NumElements" is defaulted 5199to be one. If a constant alignment is specified, the value result of the 5200allocation is guaranteed to be aligned to at least that boundary. The 5201alignment may not be greater than ``1 << 29``. If not specified, or if 5202zero, the target can choose to align the allocation on any convenient 5203boundary compatible with the type. 5204 5205'``type``' may be any sized type. 5206 5207Semantics: 5208"""""""""" 5209 5210Memory is allocated; a pointer is returned. The operation is undefined 5211if there is insufficient stack space for the allocation. '``alloca``'d 5212memory is automatically released when the function returns. The 5213'``alloca``' instruction is commonly used to represent automatic 5214variables that must have an address available. When the function returns 5215(either with the ``ret`` or ``resume`` instructions), the memory is 5216reclaimed. Allocating zero bytes is legal, but the result is undefined. 5217The order in which memory is allocated (ie., which way the stack grows) 5218is not specified. 5219 5220Example: 5221"""""""" 5222 5223.. code-block:: llvm 5224 5225 %ptr = alloca i32 ; yields i32*:ptr 5226 %ptr = alloca i32, i32 4 ; yields i32*:ptr 5227 %ptr = alloca i32, i32 4, align 1024 ; yields i32*:ptr 5228 %ptr = alloca i32, align 1024 ; yields i32*:ptr 5229 5230.. _i_load: 5231 5232'``load``' Instruction 5233^^^^^^^^^^^^^^^^^^^^^^ 5234 5235Syntax: 5236""""""" 5237 5238:: 5239 5240 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>][, !nonnull !<index>] 5241 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment> 5242 !<index> = !{ i32 1 } 5243 5244Overview: 5245""""""""" 5246 5247The '``load``' instruction is used to read from memory. 5248 5249Arguments: 5250"""""""""" 5251 5252The argument to the ``load`` instruction specifies the memory address 5253from which to load. The pointer must point to a :ref:`first 5254class <t_firstclass>` type. If the ``load`` is marked as ``volatile``, 5255then the optimizer is not allowed to modify the number or order of 5256execution of this ``load`` with other :ref:`volatile 5257operations <volatile>`. 5258 5259If the ``load`` is marked as ``atomic``, it takes an extra 5260:ref:`ordering <ordering>` and optional ``singlethread`` argument. The 5261``release`` and ``acq_rel`` orderings are not valid on ``load`` 5262instructions. Atomic loads produce :ref:`defined <memmodel>` results 5263when they may see multiple atomic stores. The type of the pointee must 5264be an integer type whose bit width is a power of two greater than or 5265equal to eight and less than or equal to a target-specific size limit. 5266``align`` must be explicitly specified on atomic loads, and the load has 5267undefined behavior if the alignment is not set to a value which is at 5268least the size in bytes of the pointee. ``!nontemporal`` does not have 5269any defined semantics for atomic loads. 5270 5271The optional constant ``align`` argument specifies the alignment of the 5272operation (that is, the alignment of the memory address). A value of 0 5273or an omitted ``align`` argument means that the operation has the ABI 5274alignment for the target. It is the responsibility of the code emitter 5275to ensure that the alignment information is correct. Overestimating the 5276alignment results in undefined behavior. Underestimating the alignment 5277may produce less efficient code. An alignment of 1 is always safe. The 5278maximum possible alignment is ``1 << 29``. 5279 5280The optional ``!nontemporal`` metadata must reference a single 5281metadata name ``<index>`` corresponding to a metadata node with one 5282``i32`` entry of value 1. The existence of the ``!nontemporal`` 5283metadata on the instruction tells the optimizer and code generator 5284that this load is not expected to be reused in the cache. The code 5285generator may select special instructions to save cache bandwidth, such 5286as the ``MOVNT`` instruction on x86. 5287 5288The optional ``!invariant.load`` metadata must reference a single 5289metadata name ``<index>`` corresponding to a metadata node with no 5290entries. The existence of the ``!invariant.load`` metadata on the 5291instruction tells the optimizer and code generator that the address 5292operand to this load points to memory which can be assumed unchanged. 5293Being invariant does not imply that a location is dereferenceable, 5294but it does imply that once the location is known dereferenceable 5295its value is henceforth unchanging. 5296 5297The optional ``!nonnull`` metadata must reference a single 5298metadata name ``<index>`` corresponding to a metadata node with no 5299entries. The existence of the ``!nonnull`` metadata on the 5300instruction tells the optimizer that the value loaded is known to 5301never be null. This is analogous to the ''nonnull'' attribute 5302on parameters and return values. This metadata can only be applied 5303to loads of a pointer type. 5304 5305Semantics: 5306"""""""""" 5307 5308The location of memory pointed to is loaded. If the value being loaded 5309is of scalar type then the number of bytes read does not exceed the 5310minimum number of bytes needed to hold all bits of the type. For 5311example, loading an ``i24`` reads at most three bytes. When loading a 5312value of a type like ``i20`` with a size that is not an integral number 5313of bytes, the result is undefined if the value was not originally 5314written using a store of the same type. 5315 5316Examples: 5317""""""""" 5318 5319.. code-block:: llvm 5320 5321 %ptr = alloca i32 ; yields i32*:ptr 5322 store i32 3, i32* %ptr ; yields void 5323 %val = load i32* %ptr ; yields i32:val = i32 3 5324 5325.. _i_store: 5326 5327'``store``' Instruction 5328^^^^^^^^^^^^^^^^^^^^^^^ 5329 5330Syntax: 5331""""""" 5332 5333:: 5334 5335 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] ; yields void 5336 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> ; yields void 5337 5338Overview: 5339""""""""" 5340 5341The '``store``' instruction is used to write to memory. 5342 5343Arguments: 5344"""""""""" 5345 5346There are two arguments to the ``store`` instruction: a value to store 5347and an address at which to store it. The type of the ``<pointer>`` 5348operand must be a pointer to the :ref:`first class <t_firstclass>` type of 5349the ``<value>`` operand. If the ``store`` is marked as ``volatile``, 5350then the optimizer is not allowed to modify the number or order of 5351execution of this ``store`` with other :ref:`volatile 5352operations <volatile>`. 5353 5354If the ``store`` is marked as ``atomic``, it takes an extra 5355:ref:`ordering <ordering>` and optional ``singlethread`` argument. The 5356``acquire`` and ``acq_rel`` orderings aren't valid on ``store`` 5357instructions. Atomic loads produce :ref:`defined <memmodel>` results 5358when they may see multiple atomic stores. The type of the pointee must 5359be an integer type whose bit width is a power of two greater than or 5360equal to eight and less than or equal to a target-specific size limit. 5361``align`` must be explicitly specified on atomic stores, and the store 5362has undefined behavior if the alignment is not set to a value which is 5363at least the size in bytes of the pointee. ``!nontemporal`` does not 5364have any defined semantics for atomic stores. 5365 5366The optional constant ``align`` argument specifies the alignment of the 5367operation (that is, the alignment of the memory address). A value of 0 5368or an omitted ``align`` argument means that the operation has the ABI 5369alignment for the target. It is the responsibility of the code emitter 5370to ensure that the alignment information is correct. Overestimating the 5371alignment results in undefined behavior. Underestimating the 5372alignment may produce less efficient code. An alignment of 1 is always 5373safe. The maximum possible alignment is ``1 << 29``. 5374 5375The optional ``!nontemporal`` metadata must reference a single metadata 5376name ``<index>`` corresponding to a metadata node with one ``i32`` entry of 5377value 1. The existence of the ``!nontemporal`` metadata on the instruction 5378tells the optimizer and code generator that this load is not expected to 5379be reused in the cache. The code generator may select special 5380instructions to save cache bandwidth, such as the MOVNT instruction on 5381x86. 5382 5383Semantics: 5384"""""""""" 5385 5386The contents of memory are updated to contain ``<value>`` at the 5387location specified by the ``<pointer>`` operand. If ``<value>`` is 5388of scalar type then the number of bytes written does not exceed the 5389minimum number of bytes needed to hold all bits of the type. For 5390example, storing an ``i24`` writes at most three bytes. When writing a 5391value of a type like ``i20`` with a size that is not an integral number 5392of bytes, it is unspecified what happens to the extra bits that do not 5393belong to the type, but they will typically be overwritten. 5394 5395Example: 5396"""""""" 5397 5398.. code-block:: llvm 5399 5400 %ptr = alloca i32 ; yields i32*:ptr 5401 store i32 3, i32* %ptr ; yields void 5402 %val = load i32* %ptr ; yields i32:val = i32 3 5403 5404.. _i_fence: 5405 5406'``fence``' Instruction 5407^^^^^^^^^^^^^^^^^^^^^^^ 5408 5409Syntax: 5410""""""" 5411 5412:: 5413 5414 fence [singlethread] <ordering> ; yields void 5415 5416Overview: 5417""""""""" 5418 5419The '``fence``' instruction is used to introduce happens-before edges 5420between operations. 5421 5422Arguments: 5423"""""""""" 5424 5425'``fence``' instructions take an :ref:`ordering <ordering>` argument which 5426defines what *synchronizes-with* edges they add. They can only be given 5427``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings. 5428 5429Semantics: 5430"""""""""" 5431 5432A fence A which has (at least) ``release`` ordering semantics 5433*synchronizes with* a fence B with (at least) ``acquire`` ordering 5434semantics if and only if there exist atomic operations X and Y, both 5435operating on some atomic object M, such that A is sequenced before X, X 5436modifies M (either directly or through some side effect of a sequence 5437headed by X), Y is sequenced before B, and Y observes M. This provides a 5438*happens-before* dependency between A and B. Rather than an explicit 5439``fence``, one (but not both) of the atomic operations X or Y might 5440provide a ``release`` or ``acquire`` (resp.) ordering constraint and 5441still *synchronize-with* the explicit ``fence`` and establish the 5442*happens-before* edge. 5443 5444A ``fence`` which has ``seq_cst`` ordering, in addition to having both 5445``acquire`` and ``release`` semantics specified above, participates in 5446the global program order of other ``seq_cst`` operations and/or fences. 5447 5448The optional ":ref:`singlethread <singlethread>`" argument specifies 5449that the fence only synchronizes with other fences in the same thread. 5450(This is useful for interacting with signal handlers.) 5451 5452Example: 5453"""""""" 5454 5455.. code-block:: llvm 5456 5457 fence acquire ; yields void 5458 fence singlethread seq_cst ; yields void 5459 5460.. _i_cmpxchg: 5461 5462'``cmpxchg``' Instruction 5463^^^^^^^^^^^^^^^^^^^^^^^^^ 5464 5465Syntax: 5466""""""" 5467 5468:: 5469 5470 cmpxchg [weak] [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <success ordering> <failure ordering> ; yields { ty, i1 } 5471 5472Overview: 5473""""""""" 5474 5475The '``cmpxchg``' instruction is used to atomically modify memory. It 5476loads a value in memory and compares it to a given value. If they are 5477equal, it tries to store a new value into the memory. 5478 5479Arguments: 5480"""""""""" 5481 5482There are three arguments to the '``cmpxchg``' instruction: an address 5483to operate on, a value to compare to the value currently be at that 5484address, and a new value to place at that address if the compared values 5485are equal. The type of '<cmp>' must be an integer type whose bit width 5486is a power of two greater than or equal to eight and less than or equal 5487to a target-specific size limit. '<cmp>' and '<new>' must have the same 5488type, and the type of '<pointer>' must be a pointer to that type. If the 5489``cmpxchg`` is marked as ``volatile``, then the optimizer is not allowed 5490to modify the number or order of execution of this ``cmpxchg`` with 5491other :ref:`volatile operations <volatile>`. 5492 5493The success and failure :ref:`ordering <ordering>` arguments specify how this 5494``cmpxchg`` synchronizes with other atomic operations. Both ordering parameters 5495must be at least ``monotonic``, the ordering constraint on failure must be no 5496stronger than that on success, and the failure ordering cannot be either 5497``release`` or ``acq_rel``. 5498 5499The optional "``singlethread``" argument declares that the ``cmpxchg`` 5500is only atomic with respect to code (usually signal handlers) running in 5501the same thread as the ``cmpxchg``. Otherwise the cmpxchg is atomic with 5502respect to all other code in the system. 5503 5504The pointer passed into cmpxchg must have alignment greater than or 5505equal to the size in memory of the operand. 5506 5507Semantics: 5508"""""""""" 5509 5510The contents of memory at the location specified by the '``<pointer>``' operand 5511is read and compared to '``<cmp>``'; if the read value is the equal, the 5512'``<new>``' is written. The original value at the location is returned, together 5513with a flag indicating success (true) or failure (false). 5514 5515If the cmpxchg operation is marked as ``weak`` then a spurious failure is 5516permitted: the operation may not write ``<new>`` even if the comparison 5517matched. 5518 5519If the cmpxchg operation is strong (the default), the i1 value is 1 if and only 5520if the value loaded equals ``cmp``. 5521 5522A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of 5523identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic 5524load with an ordering parameter determined the second ordering parameter. 5525 5526Example: 5527"""""""" 5528 5529.. code-block:: llvm 5530 5531 entry: 5532 %orig = atomic load i32* %ptr unordered ; yields i32 5533 br label %loop 5534 5535 loop: 5536 %cmp = phi i32 [ %orig, %entry ], [%old, %loop] 5537 %squared = mul i32 %cmp, %cmp 5538 %val_success = cmpxchg i32* %ptr, i32 %cmp, i32 %squared acq_rel monotonic ; yields { i32, i1 } 5539 %value_loaded = extractvalue { i32, i1 } %val_success, 0 5540 %success = extractvalue { i32, i1 } %val_success, 1 5541 br i1 %success, label %done, label %loop 5542 5543 done: 5544 ... 5545 5546.. _i_atomicrmw: 5547 5548'``atomicrmw``' Instruction 5549^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5550 5551Syntax: 5552""""""" 5553 5554:: 5555 5556 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> ; yields ty 5557 5558Overview: 5559""""""""" 5560 5561The '``atomicrmw``' instruction is used to atomically modify memory. 5562 5563Arguments: 5564"""""""""" 5565 5566There are three arguments to the '``atomicrmw``' instruction: an 5567operation to apply, an address whose value to modify, an argument to the 5568operation. The operation must be one of the following keywords: 5569 5570- xchg 5571- add 5572- sub 5573- and 5574- nand 5575- or 5576- xor 5577- max 5578- min 5579- umax 5580- umin 5581 5582The type of '<value>' must be an integer type whose bit width is a power 5583of two greater than or equal to eight and less than or equal to a 5584target-specific size limit. The type of the '``<pointer>``' operand must 5585be a pointer to that type. If the ``atomicrmw`` is marked as 5586``volatile``, then the optimizer is not allowed to modify the number or 5587order of execution of this ``atomicrmw`` with other :ref:`volatile 5588operations <volatile>`. 5589 5590Semantics: 5591"""""""""" 5592 5593The contents of memory at the location specified by the '``<pointer>``' 5594operand are atomically read, modified, and written back. The original 5595value at the location is returned. The modification is specified by the 5596operation argument: 5597 5598- xchg: ``*ptr = val`` 5599- add: ``*ptr = *ptr + val`` 5600- sub: ``*ptr = *ptr - val`` 5601- and: ``*ptr = *ptr & val`` 5602- nand: ``*ptr = ~(*ptr & val)`` 5603- or: ``*ptr = *ptr | val`` 5604- xor: ``*ptr = *ptr ^ val`` 5605- max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison) 5606- min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison) 5607- umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned 5608 comparison) 5609- umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned 5610 comparison) 5611 5612Example: 5613"""""""" 5614 5615.. code-block:: llvm 5616 5617 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields i32 5618 5619.. _i_getelementptr: 5620 5621'``getelementptr``' Instruction 5622^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5623 5624Syntax: 5625""""""" 5626 5627:: 5628 5629 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}* 5630 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}* 5631 <result> = getelementptr <ptr vector> ptrval, <vector index type> idx 5632 5633Overview: 5634""""""""" 5635 5636The '``getelementptr``' instruction is used to get the address of a 5637subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs 5638address calculation only and does not access memory. 5639 5640Arguments: 5641"""""""""" 5642 5643The first argument is always a pointer or a vector of pointers, and 5644forms the basis of the calculation. The remaining arguments are indices 5645that indicate which of the elements of the aggregate object are indexed. 5646The interpretation of each index is dependent on the type being indexed 5647into. The first index always indexes the pointer value given as the 5648first argument, the second index indexes a value of the type pointed to 5649(not necessarily the value directly pointed to, since the first index 5650can be non-zero), etc. The first type indexed into must be a pointer 5651value, subsequent types can be arrays, vectors, and structs. Note that 5652subsequent types being indexed into can never be pointers, since that 5653would require loading the pointer before continuing calculation. 5654 5655The type of each index argument depends on the type it is indexing into. 5656When indexing into a (optionally packed) structure, only ``i32`` integer 5657**constants** are allowed (when using a vector of indices they must all 5658be the **same** ``i32`` integer constant). When indexing into an array, 5659pointer or vector, integers of any width are allowed, and they are not 5660required to be constant. These integers are treated as signed values 5661where relevant. 5662 5663For example, let's consider a C code fragment and how it gets compiled 5664to LLVM: 5665 5666.. code-block:: c 5667 5668 struct RT { 5669 char A; 5670 int B[10][20]; 5671 char C; 5672 }; 5673 struct ST { 5674 int X; 5675 double Y; 5676 struct RT Z; 5677 }; 5678 5679 int *foo(struct ST *s) { 5680 return &s[1].Z.B[5][13]; 5681 } 5682 5683The LLVM code generated by Clang is: 5684 5685.. code-block:: llvm 5686 5687 %struct.RT = type { i8, [10 x [20 x i32]], i8 } 5688 %struct.ST = type { i32, double, %struct.RT } 5689 5690 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp { 5691 entry: 5692 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13 5693 ret i32* %arrayidx 5694 } 5695 5696Semantics: 5697"""""""""" 5698 5699In the example above, the first index is indexing into the 5700'``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``' 5701= '``{ i32, double, %struct.RT }``' type, a structure. The second index 5702indexes into the third element of the structure, yielding a 5703'``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another 5704structure. The third index indexes into the second element of the 5705structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two 5706dimensions of the array are subscripted into, yielding an '``i32``' 5707type. The '``getelementptr``' instruction returns a pointer to this 5708element, thus computing a value of '``i32*``' type. 5709 5710Note that it is perfectly legal to index partially through a structure, 5711returning a pointer to an inner element. Because of this, the LLVM code 5712for the given testcase is equivalent to: 5713 5714.. code-block:: llvm 5715 5716 define i32* @foo(%struct.ST* %s) { 5717 %t1 = getelementptr %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1 5718 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2 5719 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3 5720 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4 5721 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5 5722 ret i32* %t5 5723 } 5724 5725If the ``inbounds`` keyword is present, the result value of the 5726``getelementptr`` is a :ref:`poison value <poisonvalues>` if the base 5727pointer is not an *in bounds* address of an allocated object, or if any 5728of the addresses that would be formed by successive addition of the 5729offsets implied by the indices to the base address with infinitely 5730precise signed arithmetic are not an *in bounds* address of that 5731allocated object. The *in bounds* addresses for an allocated object are 5732all the addresses that point into the object, plus the address one byte 5733past the end. In cases where the base is a vector of pointers the 5734``inbounds`` keyword applies to each of the computations element-wise. 5735 5736If the ``inbounds`` keyword is not present, the offsets are added to the 5737base address with silently-wrapping two's complement arithmetic. If the 5738offsets have a different width from the pointer, they are sign-extended 5739or truncated to the width of the pointer. The result value of the 5740``getelementptr`` may be outside the object pointed to by the base 5741pointer. The result value may not necessarily be used to access memory 5742though, even if it happens to point into allocated storage. See the 5743:ref:`Pointer Aliasing Rules <pointeraliasing>` section for more 5744information. 5745 5746The getelementptr instruction is often confusing. For some more insight 5747into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`. 5748 5749Example: 5750"""""""" 5751 5752.. code-block:: llvm 5753 5754 ; yields [12 x i8]*:aptr 5755 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1 5756 ; yields i8*:vptr 5757 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1 5758 ; yields i8*:eptr 5759 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1 5760 ; yields i32*:iptr 5761 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0 5762 5763In cases where the pointer argument is a vector of pointers, each index 5764must be a vector with the same number of elements. For example: 5765 5766.. code-block:: llvm 5767 5768 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets, 5769 5770Conversion Operations 5771--------------------- 5772 5773The instructions in this category are the conversion instructions 5774(casting) which all take a single operand and a type. They perform 5775various bit conversions on the operand. 5776 5777'``trunc .. to``' Instruction 5778^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5779 5780Syntax: 5781""""""" 5782 5783:: 5784 5785 <result> = trunc <ty> <value> to <ty2> ; yields ty2 5786 5787Overview: 5788""""""""" 5789 5790The '``trunc``' instruction truncates its operand to the type ``ty2``. 5791 5792Arguments: 5793"""""""""" 5794 5795The '``trunc``' instruction takes a value to trunc, and a type to trunc 5796it to. Both types must be of :ref:`integer <t_integer>` types, or vectors 5797of the same number of integers. The bit size of the ``value`` must be 5798larger than the bit size of the destination type, ``ty2``. Equal sized 5799types are not allowed. 5800 5801Semantics: 5802"""""""""" 5803 5804The '``trunc``' instruction truncates the high order bits in ``value`` 5805and converts the remaining bits to ``ty2``. Since the source size must 5806be larger than the destination size, ``trunc`` cannot be a *no-op cast*. 5807It will always truncate bits. 5808 5809Example: 5810"""""""" 5811 5812.. code-block:: llvm 5813 5814 %X = trunc i32 257 to i8 ; yields i8:1 5815 %Y = trunc i32 123 to i1 ; yields i1:true 5816 %Z = trunc i32 122 to i1 ; yields i1:false 5817 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7> 5818 5819'``zext .. to``' Instruction 5820^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5821 5822Syntax: 5823""""""" 5824 5825:: 5826 5827 <result> = zext <ty> <value> to <ty2> ; yields ty2 5828 5829Overview: 5830""""""""" 5831 5832The '``zext``' instruction zero extends its operand to type ``ty2``. 5833 5834Arguments: 5835"""""""""" 5836 5837The '``zext``' instruction takes a value to cast, and a type to cast it 5838to. Both types must be of :ref:`integer <t_integer>` types, or vectors of 5839the same number of integers. The bit size of the ``value`` must be 5840smaller than the bit size of the destination type, ``ty2``. 5841 5842Semantics: 5843"""""""""" 5844 5845The ``zext`` fills the high order bits of the ``value`` with zero bits 5846until it reaches the size of the destination type, ``ty2``. 5847 5848When zero extending from i1, the result will always be either 0 or 1. 5849 5850Example: 5851"""""""" 5852 5853.. code-block:: llvm 5854 5855 %X = zext i32 257 to i64 ; yields i64:257 5856 %Y = zext i1 true to i32 ; yields i32:1 5857 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7> 5858 5859'``sext .. to``' Instruction 5860^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5861 5862Syntax: 5863""""""" 5864 5865:: 5866 5867 <result> = sext <ty> <value> to <ty2> ; yields ty2 5868 5869Overview: 5870""""""""" 5871 5872The '``sext``' sign extends ``value`` to the type ``ty2``. 5873 5874Arguments: 5875"""""""""" 5876 5877The '``sext``' instruction takes a value to cast, and a type to cast it 5878to. Both types must be of :ref:`integer <t_integer>` types, or vectors of 5879the same number of integers. The bit size of the ``value`` must be 5880smaller than the bit size of the destination type, ``ty2``. 5881 5882Semantics: 5883"""""""""" 5884 5885The '``sext``' instruction performs a sign extension by copying the sign 5886bit (highest order bit) of the ``value`` until it reaches the bit size 5887of the type ``ty2``. 5888 5889When sign extending from i1, the extension always results in -1 or 0. 5890 5891Example: 5892"""""""" 5893 5894.. code-block:: llvm 5895 5896 %X = sext i8 -1 to i16 ; yields i16 :65535 5897 %Y = sext i1 true to i32 ; yields i32:-1 5898 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7> 5899 5900'``fptrunc .. to``' Instruction 5901^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5902 5903Syntax: 5904""""""" 5905 5906:: 5907 5908 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2 5909 5910Overview: 5911""""""""" 5912 5913The '``fptrunc``' instruction truncates ``value`` to type ``ty2``. 5914 5915Arguments: 5916"""""""""" 5917 5918The '``fptrunc``' instruction takes a :ref:`floating point <t_floating>` 5919value to cast and a :ref:`floating point <t_floating>` type to cast it to. 5920The size of ``value`` must be larger than the size of ``ty2``. This 5921implies that ``fptrunc`` cannot be used to make a *no-op cast*. 5922 5923Semantics: 5924"""""""""" 5925 5926The '``fptrunc``' instruction truncates a ``value`` from a larger 5927:ref:`floating point <t_floating>` type to a smaller :ref:`floating 5928point <t_floating>` type. If the value cannot fit within the 5929destination type, ``ty2``, then the results are undefined. 5930 5931Example: 5932"""""""" 5933 5934.. code-block:: llvm 5935 5936 %X = fptrunc double 123.0 to float ; yields float:123.0 5937 %Y = fptrunc double 1.0E+300 to float ; yields undefined 5938 5939'``fpext .. to``' Instruction 5940^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5941 5942Syntax: 5943""""""" 5944 5945:: 5946 5947 <result> = fpext <ty> <value> to <ty2> ; yields ty2 5948 5949Overview: 5950""""""""" 5951 5952The '``fpext``' extends a floating point ``value`` to a larger floating 5953point value. 5954 5955Arguments: 5956"""""""""" 5957 5958The '``fpext``' instruction takes a :ref:`floating point <t_floating>` 5959``value`` to cast, and a :ref:`floating point <t_floating>` type to cast it 5960to. The source type must be smaller than the destination type. 5961 5962Semantics: 5963"""""""""" 5964 5965The '``fpext``' instruction extends the ``value`` from a smaller 5966:ref:`floating point <t_floating>` type to a larger :ref:`floating 5967point <t_floating>` type. The ``fpext`` cannot be used to make a 5968*no-op cast* because it always changes bits. Use ``bitcast`` to make a 5969*no-op cast* for a floating point cast. 5970 5971Example: 5972"""""""" 5973 5974.. code-block:: llvm 5975 5976 %X = fpext float 3.125 to double ; yields double:3.125000e+00 5977 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000 5978 5979'``fptoui .. to``' Instruction 5980^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5981 5982Syntax: 5983""""""" 5984 5985:: 5986 5987 <result> = fptoui <ty> <value> to <ty2> ; yields ty2 5988 5989Overview: 5990""""""""" 5991 5992The '``fptoui``' converts a floating point ``value`` to its unsigned 5993integer equivalent of type ``ty2``. 5994 5995Arguments: 5996"""""""""" 5997 5998The '``fptoui``' instruction takes a value to cast, which must be a 5999scalar or vector :ref:`floating point <t_floating>` value, and a type to 6000cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If 6001``ty`` is a vector floating point type, ``ty2`` must be a vector integer 6002type with the same number of elements as ``ty`` 6003 6004Semantics: 6005"""""""""" 6006 6007The '``fptoui``' instruction converts its :ref:`floating 6008point <t_floating>` operand into the nearest (rounding towards zero) 6009unsigned integer value. If the value cannot fit in ``ty2``, the results 6010are undefined. 6011 6012Example: 6013"""""""" 6014 6015.. code-block:: llvm 6016 6017 %X = fptoui double 123.0 to i32 ; yields i32:123 6018 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1 6019 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1 6020 6021'``fptosi .. to``' Instruction 6022^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6023 6024Syntax: 6025""""""" 6026 6027:: 6028 6029 <result> = fptosi <ty> <value> to <ty2> ; yields ty2 6030 6031Overview: 6032""""""""" 6033 6034The '``fptosi``' instruction converts :ref:`floating point <t_floating>` 6035``value`` to type ``ty2``. 6036 6037Arguments: 6038"""""""""" 6039 6040The '``fptosi``' instruction takes a value to cast, which must be a 6041scalar or vector :ref:`floating point <t_floating>` value, and a type to 6042cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If 6043``ty`` is a vector floating point type, ``ty2`` must be a vector integer 6044type with the same number of elements as ``ty`` 6045 6046Semantics: 6047"""""""""" 6048 6049The '``fptosi``' instruction converts its :ref:`floating 6050point <t_floating>` operand into the nearest (rounding towards zero) 6051signed integer value. If the value cannot fit in ``ty2``, the results 6052are undefined. 6053 6054Example: 6055"""""""" 6056 6057.. code-block:: llvm 6058 6059 %X = fptosi double -123.0 to i32 ; yields i32:-123 6060 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1 6061 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1 6062 6063'``uitofp .. to``' Instruction 6064^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6065 6066Syntax: 6067""""""" 6068 6069:: 6070 6071 <result> = uitofp <ty> <value> to <ty2> ; yields ty2 6072 6073Overview: 6074""""""""" 6075 6076The '``uitofp``' instruction regards ``value`` as an unsigned integer 6077and converts that value to the ``ty2`` type. 6078 6079Arguments: 6080"""""""""" 6081 6082The '``uitofp``' instruction takes a value to cast, which must be a 6083scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to 6084``ty2``, which must be an :ref:`floating point <t_floating>` type. If 6085``ty`` is a vector integer type, ``ty2`` must be a vector floating point 6086type with the same number of elements as ``ty`` 6087 6088Semantics: 6089"""""""""" 6090 6091The '``uitofp``' instruction interprets its operand as an unsigned 6092integer quantity and converts it to the corresponding floating point 6093value. If the value cannot fit in the floating point value, the results 6094are undefined. 6095 6096Example: 6097"""""""" 6098 6099.. code-block:: llvm 6100 6101 %X = uitofp i32 257 to float ; yields float:257.0 6102 %Y = uitofp i8 -1 to double ; yields double:255.0 6103 6104'``sitofp .. to``' Instruction 6105^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6106 6107Syntax: 6108""""""" 6109 6110:: 6111 6112 <result> = sitofp <ty> <value> to <ty2> ; yields ty2 6113 6114Overview: 6115""""""""" 6116 6117The '``sitofp``' instruction regards ``value`` as a signed integer and 6118converts that value to the ``ty2`` type. 6119 6120Arguments: 6121"""""""""" 6122 6123The '``sitofp``' instruction takes a value to cast, which must be a 6124scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to 6125``ty2``, which must be an :ref:`floating point <t_floating>` type. If 6126``ty`` is a vector integer type, ``ty2`` must be a vector floating point 6127type with the same number of elements as ``ty`` 6128 6129Semantics: 6130"""""""""" 6131 6132The '``sitofp``' instruction interprets its operand as a signed integer 6133quantity and converts it to the corresponding floating point value. If 6134the value cannot fit in the floating point value, the results are 6135undefined. 6136 6137Example: 6138"""""""" 6139 6140.. code-block:: llvm 6141 6142 %X = sitofp i32 257 to float ; yields float:257.0 6143 %Y = sitofp i8 -1 to double ; yields double:-1.0 6144 6145.. _i_ptrtoint: 6146 6147'``ptrtoint .. to``' Instruction 6148^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6149 6150Syntax: 6151""""""" 6152 6153:: 6154 6155 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2 6156 6157Overview: 6158""""""""" 6159 6160The '``ptrtoint``' instruction converts the pointer or a vector of 6161pointers ``value`` to the integer (or vector of integers) type ``ty2``. 6162 6163Arguments: 6164"""""""""" 6165 6166The '``ptrtoint``' instruction takes a ``value`` to cast, which must be 6167a a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a 6168type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or 6169a vector of integers type. 6170 6171Semantics: 6172"""""""""" 6173 6174The '``ptrtoint``' instruction converts ``value`` to integer type 6175``ty2`` by interpreting the pointer value as an integer and either 6176truncating or zero extending that value to the size of the integer type. 6177If ``value`` is smaller than ``ty2`` then a zero extension is done. If 6178``value`` is larger than ``ty2`` then a truncation is done. If they are 6179the same size, then nothing is done (*no-op cast*) other than a type 6180change. 6181 6182Example: 6183"""""""" 6184 6185.. code-block:: llvm 6186 6187 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture 6188 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture 6189 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture 6190 6191.. _i_inttoptr: 6192 6193'``inttoptr .. to``' Instruction 6194^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6195 6196Syntax: 6197""""""" 6198 6199:: 6200 6201 <result> = inttoptr <ty> <value> to <ty2> ; yields ty2 6202 6203Overview: 6204""""""""" 6205 6206The '``inttoptr``' instruction converts an integer ``value`` to a 6207pointer type, ``ty2``. 6208 6209Arguments: 6210"""""""""" 6211 6212The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to 6213cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>` 6214type. 6215 6216Semantics: 6217"""""""""" 6218 6219The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by 6220applying either a zero extension or a truncation depending on the size 6221of the integer ``value``. If ``value`` is larger than the size of a 6222pointer then a truncation is done. If ``value`` is smaller than the size 6223of a pointer then a zero extension is done. If they are the same size, 6224nothing is done (*no-op cast*). 6225 6226Example: 6227"""""""" 6228 6229.. code-block:: llvm 6230 6231 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture 6232 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture 6233 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture 6234 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers 6235 6236.. _i_bitcast: 6237 6238'``bitcast .. to``' Instruction 6239^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6240 6241Syntax: 6242""""""" 6243 6244:: 6245 6246 <result> = bitcast <ty> <value> to <ty2> ; yields ty2 6247 6248Overview: 6249""""""""" 6250 6251The '``bitcast``' instruction converts ``value`` to type ``ty2`` without 6252changing any bits. 6253 6254Arguments: 6255"""""""""" 6256 6257The '``bitcast``' instruction takes a value to cast, which must be a 6258non-aggregate first class value, and a type to cast it to, which must 6259also be a non-aggregate :ref:`first class <t_firstclass>` type. The 6260bit sizes of ``value`` and the destination type, ``ty2``, must be 6261identical. If the source type is a pointer, the destination type must 6262also be a pointer of the same size. This instruction supports bitwise 6263conversion of vectors to integers and to vectors of other types (as 6264long as they have the same size). 6265 6266Semantics: 6267"""""""""" 6268 6269The '``bitcast``' instruction converts ``value`` to type ``ty2``. It 6270is always a *no-op cast* because no bits change with this 6271conversion. The conversion is done as if the ``value`` had been stored 6272to memory and read back as type ``ty2``. Pointer (or vector of 6273pointers) types may only be converted to other pointer (or vector of 6274pointers) types with the same address space through this instruction. 6275To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>` 6276or :ref:`ptrtoint <i_ptrtoint>` instructions first. 6277 6278Example: 6279"""""""" 6280 6281.. code-block:: llvm 6282 6283 %X = bitcast i8 255 to i8 ; yields i8 :-1 6284 %Y = bitcast i32* %x to sint* ; yields sint*:%x 6285 %Z = bitcast <2 x int> %V to i64; ; yields i64: %V 6286 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*> 6287 6288.. _i_addrspacecast: 6289 6290'``addrspacecast .. to``' Instruction 6291^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6292 6293Syntax: 6294""""""" 6295 6296:: 6297 6298 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2 6299 6300Overview: 6301""""""""" 6302 6303The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in 6304address space ``n`` to type ``pty2`` in address space ``m``. 6305 6306Arguments: 6307"""""""""" 6308 6309The '``addrspacecast``' instruction takes a pointer or vector of pointer value 6310to cast and a pointer type to cast it to, which must have a different 6311address space. 6312 6313Semantics: 6314"""""""""" 6315 6316The '``addrspacecast``' instruction converts the pointer value 6317``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex 6318value modification, depending on the target and the address space 6319pair. Pointer conversions within the same address space must be 6320performed with the ``bitcast`` instruction. Note that if the address space 6321conversion is legal then both result and operand refer to the same memory 6322location. 6323 6324Example: 6325"""""""" 6326 6327.. code-block:: llvm 6328 6329 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x 6330 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y 6331 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z 6332 6333.. _otherops: 6334 6335Other Operations 6336---------------- 6337 6338The instructions in this category are the "miscellaneous" instructions, 6339which defy better classification. 6340 6341.. _i_icmp: 6342 6343'``icmp``' Instruction 6344^^^^^^^^^^^^^^^^^^^^^^ 6345 6346Syntax: 6347""""""" 6348 6349:: 6350 6351 <result> = icmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result 6352 6353Overview: 6354""""""""" 6355 6356The '``icmp``' instruction returns a boolean value or a vector of 6357boolean values based on comparison of its two integer, integer vector, 6358pointer, or pointer vector operands. 6359 6360Arguments: 6361"""""""""" 6362 6363The '``icmp``' instruction takes three operands. The first operand is 6364the condition code indicating the kind of comparison to perform. It is 6365not a value, just a keyword. The possible condition code are: 6366 6367#. ``eq``: equal 6368#. ``ne``: not equal 6369#. ``ugt``: unsigned greater than 6370#. ``uge``: unsigned greater or equal 6371#. ``ult``: unsigned less than 6372#. ``ule``: unsigned less or equal 6373#. ``sgt``: signed greater than 6374#. ``sge``: signed greater or equal 6375#. ``slt``: signed less than 6376#. ``sle``: signed less or equal 6377 6378The remaining two arguments must be :ref:`integer <t_integer>` or 6379:ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They 6380must also be identical types. 6381 6382Semantics: 6383"""""""""" 6384 6385The '``icmp``' compares ``op1`` and ``op2`` according to the condition 6386code given as ``cond``. The comparison performed always yields either an 6387:ref:`i1 <t_integer>` or vector of ``i1`` result, as follows: 6388 6389#. ``eq``: yields ``true`` if the operands are equal, ``false`` 6390 otherwise. No sign interpretation is necessary or performed. 6391#. ``ne``: yields ``true`` if the operands are unequal, ``false`` 6392 otherwise. No sign interpretation is necessary or performed. 6393#. ``ugt``: interprets the operands as unsigned values and yields 6394 ``true`` if ``op1`` is greater than ``op2``. 6395#. ``uge``: interprets the operands as unsigned values and yields 6396 ``true`` if ``op1`` is greater than or equal to ``op2``. 6397#. ``ult``: interprets the operands as unsigned values and yields 6398 ``true`` if ``op1`` is less than ``op2``. 6399#. ``ule``: interprets the operands as unsigned values and yields 6400 ``true`` if ``op1`` is less than or equal to ``op2``. 6401#. ``sgt``: interprets the operands as signed values and yields ``true`` 6402 if ``op1`` is greater than ``op2``. 6403#. ``sge``: interprets the operands as signed values and yields ``true`` 6404 if ``op1`` is greater than or equal to ``op2``. 6405#. ``slt``: interprets the operands as signed values and yields ``true`` 6406 if ``op1`` is less than ``op2``. 6407#. ``sle``: interprets the operands as signed values and yields ``true`` 6408 if ``op1`` is less than or equal to ``op2``. 6409 6410If the operands are :ref:`pointer <t_pointer>` typed, the pointer values 6411are compared as if they were integers. 6412 6413If the operands are integer vectors, then they are compared element by 6414element. The result is an ``i1`` vector with the same number of elements 6415as the values being compared. Otherwise, the result is an ``i1``. 6416 6417Example: 6418"""""""" 6419 6420.. code-block:: llvm 6421 6422 <result> = icmp eq i32 4, 5 ; yields: result=false 6423 <result> = icmp ne float* %X, %X ; yields: result=false 6424 <result> = icmp ult i16 4, 5 ; yields: result=true 6425 <result> = icmp sgt i16 4, 5 ; yields: result=false 6426 <result> = icmp ule i16 -4, 5 ; yields: result=false 6427 <result> = icmp sge i16 4, 5 ; yields: result=false 6428 6429Note that the code generator does not yet support vector types with the 6430``icmp`` instruction. 6431 6432.. _i_fcmp: 6433 6434'``fcmp``' Instruction 6435^^^^^^^^^^^^^^^^^^^^^^ 6436 6437Syntax: 6438""""""" 6439 6440:: 6441 6442 <result> = fcmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result 6443 6444Overview: 6445""""""""" 6446 6447The '``fcmp``' instruction returns a boolean value or vector of boolean 6448values based on comparison of its operands. 6449 6450If the operands are floating point scalars, then the result type is a 6451boolean (:ref:`i1 <t_integer>`). 6452 6453If the operands are floating point vectors, then the result type is a 6454vector of boolean with the same number of elements as the operands being 6455compared. 6456 6457Arguments: 6458"""""""""" 6459 6460The '``fcmp``' instruction takes three operands. The first operand is 6461the condition code indicating the kind of comparison to perform. It is 6462not a value, just a keyword. The possible condition code are: 6463 6464#. ``false``: no comparison, always returns false 6465#. ``oeq``: ordered and equal 6466#. ``ogt``: ordered and greater than 6467#. ``oge``: ordered and greater than or equal 6468#. ``olt``: ordered and less than 6469#. ``ole``: ordered and less than or equal 6470#. ``one``: ordered and not equal 6471#. ``ord``: ordered (no nans) 6472#. ``ueq``: unordered or equal 6473#. ``ugt``: unordered or greater than 6474#. ``uge``: unordered or greater than or equal 6475#. ``ult``: unordered or less than 6476#. ``ule``: unordered or less than or equal 6477#. ``une``: unordered or not equal 6478#. ``uno``: unordered (either nans) 6479#. ``true``: no comparison, always returns true 6480 6481*Ordered* means that neither operand is a QNAN while *unordered* means 6482that either operand may be a QNAN. 6483 6484Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating 6485point <t_floating>` type or a :ref:`vector <t_vector>` of floating point 6486type. They must have identical types. 6487 6488Semantics: 6489"""""""""" 6490 6491The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the 6492condition code given as ``cond``. If the operands are vectors, then the 6493vectors are compared element by element. Each comparison performed 6494always yields an :ref:`i1 <t_integer>` result, as follows: 6495 6496#. ``false``: always yields ``false``, regardless of operands. 6497#. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1`` 6498 is equal to ``op2``. 6499#. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1`` 6500 is greater than ``op2``. 6501#. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1`` 6502 is greater than or equal to ``op2``. 6503#. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1`` 6504 is less than ``op2``. 6505#. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1`` 6506 is less than or equal to ``op2``. 6507#. ``one``: yields ``true`` if both operands are not a QNAN and ``op1`` 6508 is not equal to ``op2``. 6509#. ``ord``: yields ``true`` if both operands are not a QNAN. 6510#. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is 6511 equal to ``op2``. 6512#. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is 6513 greater than ``op2``. 6514#. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is 6515 greater than or equal to ``op2``. 6516#. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is 6517 less than ``op2``. 6518#. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is 6519 less than or equal to ``op2``. 6520#. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is 6521 not equal to ``op2``. 6522#. ``uno``: yields ``true`` if either operand is a QNAN. 6523#. ``true``: always yields ``true``, regardless of operands. 6524 6525Example: 6526"""""""" 6527 6528.. code-block:: llvm 6529 6530 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false 6531 <result> = fcmp one float 4.0, 5.0 ; yields: result=true 6532 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true 6533 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false 6534 6535Note that the code generator does not yet support vector types with the 6536``fcmp`` instruction. 6537 6538.. _i_phi: 6539 6540'``phi``' Instruction 6541^^^^^^^^^^^^^^^^^^^^^ 6542 6543Syntax: 6544""""""" 6545 6546:: 6547 6548 <result> = phi <ty> [ <val0>, <label0>], ... 6549 6550Overview: 6551""""""""" 6552 6553The '``phi``' instruction is used to implement the φ node in the SSA 6554graph representing the function. 6555 6556Arguments: 6557"""""""""" 6558 6559The type of the incoming values is specified with the first type field. 6560After this, the '``phi``' instruction takes a list of pairs as 6561arguments, with one pair for each predecessor basic block of the current 6562block. Only values of :ref:`first class <t_firstclass>` type may be used as 6563the value arguments to the PHI node. Only labels may be used as the 6564label arguments. 6565 6566There must be no non-phi instructions between the start of a basic block 6567and the PHI instructions: i.e. PHI instructions must be first in a basic 6568block. 6569 6570For the purposes of the SSA form, the use of each incoming value is 6571deemed to occur on the edge from the corresponding predecessor block to 6572the current block (but after any definition of an '``invoke``' 6573instruction's return value on the same edge). 6574 6575Semantics: 6576"""""""""" 6577 6578At runtime, the '``phi``' instruction logically takes on the value 6579specified by the pair corresponding to the predecessor basic block that 6580executed just prior to the current block. 6581 6582Example: 6583"""""""" 6584 6585.. code-block:: llvm 6586 6587 Loop: ; Infinite loop that counts from 0 on up... 6588 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ] 6589 %nextindvar = add i32 %indvar, 1 6590 br label %Loop 6591 6592.. _i_select: 6593 6594'``select``' Instruction 6595^^^^^^^^^^^^^^^^^^^^^^^^ 6596 6597Syntax: 6598""""""" 6599 6600:: 6601 6602 <result> = select selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty 6603 6604 selty is either i1 or {<N x i1>} 6605 6606Overview: 6607""""""""" 6608 6609The '``select``' instruction is used to choose one value based on a 6610condition, without IR-level branching. 6611 6612Arguments: 6613"""""""""" 6614 6615The '``select``' instruction requires an 'i1' value or a vector of 'i1' 6616values indicating the condition, and two values of the same :ref:`first 6617class <t_firstclass>` type. If the val1/val2 are vectors and the 6618condition is a scalar, then entire vectors are selected, not individual 6619elements. 6620 6621Semantics: 6622"""""""""" 6623 6624If the condition is an i1 and it evaluates to 1, the instruction returns 6625the first value argument; otherwise, it returns the second value 6626argument. 6627 6628If the condition is a vector of i1, then the value arguments must be 6629vectors of the same size, and the selection is done element by element. 6630 6631Example: 6632"""""""" 6633 6634.. code-block:: llvm 6635 6636 %X = select i1 true, i8 17, i8 42 ; yields i8:17 6637 6638.. _i_call: 6639 6640'``call``' Instruction 6641^^^^^^^^^^^^^^^^^^^^^^ 6642 6643Syntax: 6644""""""" 6645 6646:: 6647 6648 <result> = [tail | musttail] call [cconv] [ret attrs] <ty> [<fnty>*] <fnptrval>(<function args>) [fn attrs] 6649 6650Overview: 6651""""""""" 6652 6653The '``call``' instruction represents a simple function call. 6654 6655Arguments: 6656"""""""""" 6657 6658This instruction requires several arguments: 6659 6660#. The optional ``tail`` and ``musttail`` markers indicate that the optimizers 6661 should perform tail call optimization. The ``tail`` marker is a hint that 6662 `can be ignored <CodeGenerator.html#sibcallopt>`_. The ``musttail`` marker 6663 means that the call must be tail call optimized in order for the program to 6664 be correct. The ``musttail`` marker provides these guarantees: 6665 6666 #. The call will not cause unbounded stack growth if it is part of a 6667 recursive cycle in the call graph. 6668 #. Arguments with the :ref:`inalloca <attr_inalloca>` attribute are 6669 forwarded in place. 6670 6671 Both markers imply that the callee does not access allocas or varargs from 6672 the caller. Calls marked ``musttail`` must obey the following additional 6673 rules: 6674 6675 - The call must immediately precede a :ref:`ret <i_ret>` instruction, 6676 or a pointer bitcast followed by a ret instruction. 6677 - The ret instruction must return the (possibly bitcasted) value 6678 produced by the call or void. 6679 - The caller and callee prototypes must match. Pointer types of 6680 parameters or return types may differ in pointee type, but not 6681 in address space. 6682 - The calling conventions of the caller and callee must match. 6683 - All ABI-impacting function attributes, such as sret, byval, inreg, 6684 returned, and inalloca, must match. 6685 - The callee must be varargs iff the caller is varargs. Bitcasting a 6686 non-varargs function to the appropriate varargs type is legal so 6687 long as the non-varargs prefixes obey the other rules. 6688 6689 Tail call optimization for calls marked ``tail`` is guaranteed to occur if 6690 the following conditions are met: 6691 6692 - Caller and callee both have the calling convention ``fastcc``. 6693 - The call is in tail position (ret immediately follows call and ret 6694 uses value of call or is void). 6695 - Option ``-tailcallopt`` is enabled, or 6696 ``llvm::GuaranteedTailCallOpt`` is ``true``. 6697 - `Platform-specific constraints are 6698 met. <CodeGenerator.html#tailcallopt>`_ 6699 6700#. The optional "cconv" marker indicates which :ref:`calling 6701 convention <callingconv>` the call should use. If none is 6702 specified, the call defaults to using C calling conventions. The 6703 calling convention of the call must match the calling convention of 6704 the target function, or else the behavior is undefined. 6705#. The optional :ref:`Parameter Attributes <paramattrs>` list for return 6706 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes 6707 are valid here. 6708#. '``ty``': the type of the call instruction itself which is also the 6709 type of the return value. Functions that return no value are marked 6710 ``void``. 6711#. '``fnty``': shall be the signature of the pointer to function value 6712 being invoked. The argument types must match the types implied by 6713 this signature. This type can be omitted if the function is not 6714 varargs and if the function type does not return a pointer to a 6715 function. 6716#. '``fnptrval``': An LLVM value containing a pointer to a function to 6717 be invoked. In most cases, this is a direct function invocation, but 6718 indirect ``call``'s are just as possible, calling an arbitrary pointer 6719 to function value. 6720#. '``function args``': argument list whose types match the function 6721 signature argument types and parameter attributes. All arguments must 6722 be of :ref:`first class <t_firstclass>` type. If the function signature 6723 indicates the function accepts a variable number of arguments, the 6724 extra arguments can be specified. 6725#. The optional :ref:`function attributes <fnattrs>` list. Only 6726 '``noreturn``', '``nounwind``', '``readonly``' and '``readnone``' 6727 attributes are valid here. 6728 6729Semantics: 6730"""""""""" 6731 6732The '``call``' instruction is used to cause control flow to transfer to 6733a specified function, with its incoming arguments bound to the specified 6734values. Upon a '``ret``' instruction in the called function, control 6735flow continues with the instruction after the function call, and the 6736return value of the function is bound to the result argument. 6737 6738Example: 6739"""""""" 6740 6741.. code-block:: llvm 6742 6743 %retval = call i32 @test(i32 %argc) 6744 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32 6745 %X = tail call i32 @foo() ; yields i32 6746 %Y = tail call fastcc i32 @foo() ; yields i32 6747 call void %foo(i8 97 signext) 6748 6749 %struct.A = type { i32, i8 } 6750 %r = call %struct.A @foo() ; yields { i32, i8 } 6751 %gr = extractvalue %struct.A %r, 0 ; yields i32 6752 %gr1 = extractvalue %struct.A %r, 1 ; yields i8 6753 %Z = call void @foo() noreturn ; indicates that %foo never returns normally 6754 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended 6755 6756llvm treats calls to some functions with names and arguments that match 6757the standard C99 library as being the C99 library functions, and may 6758perform optimizations or generate code for them under that assumption. 6759This is something we'd like to change in the future to provide better 6760support for freestanding environments and non-C-based languages. 6761 6762.. _i_va_arg: 6763 6764'``va_arg``' Instruction 6765^^^^^^^^^^^^^^^^^^^^^^^^ 6766 6767Syntax: 6768""""""" 6769 6770:: 6771 6772 <resultval> = va_arg <va_list*> <arglist>, <argty> 6773 6774Overview: 6775""""""""" 6776 6777The '``va_arg``' instruction is used to access arguments passed through 6778the "variable argument" area of a function call. It is used to implement 6779the ``va_arg`` macro in C. 6780 6781Arguments: 6782"""""""""" 6783 6784This instruction takes a ``va_list*`` value and the type of the 6785argument. It returns a value of the specified argument type and 6786increments the ``va_list`` to point to the next argument. The actual 6787type of ``va_list`` is target specific. 6788 6789Semantics: 6790"""""""""" 6791 6792The '``va_arg``' instruction loads an argument of the specified type 6793from the specified ``va_list`` and causes the ``va_list`` to point to 6794the next argument. For more information, see the variable argument 6795handling :ref:`Intrinsic Functions <int_varargs>`. 6796 6797It is legal for this instruction to be called in a function which does 6798not take a variable number of arguments, for example, the ``vfprintf`` 6799function. 6800 6801``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic 6802function <intrinsics>` because it takes a type as an argument. 6803 6804Example: 6805"""""""" 6806 6807See the :ref:`variable argument processing <int_varargs>` section. 6808 6809Note that the code generator does not yet fully support va\_arg on many 6810targets. Also, it does not currently support va\_arg with aggregate 6811types on any target. 6812 6813.. _i_landingpad: 6814 6815'``landingpad``' Instruction 6816^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6817 6818Syntax: 6819""""""" 6820 6821:: 6822 6823 <resultval> = landingpad <resultty> personality <type> <pers_fn> <clause>+ 6824 <resultval> = landingpad <resultty> personality <type> <pers_fn> cleanup <clause>* 6825 6826 <clause> := catch <type> <value> 6827 <clause> := filter <array constant type> <array constant> 6828 6829Overview: 6830""""""""" 6831 6832The '``landingpad``' instruction is used by `LLVM's exception handling 6833system <ExceptionHandling.html#overview>`_ to specify that a basic block 6834is a landing pad --- one where the exception lands, and corresponds to the 6835code found in the ``catch`` portion of a ``try``/``catch`` sequence. It 6836defines values supplied by the personality function (``pers_fn``) upon 6837re-entry to the function. The ``resultval`` has the type ``resultty``. 6838 6839Arguments: 6840"""""""""" 6841 6842This instruction takes a ``pers_fn`` value. This is the personality 6843function associated with the unwinding mechanism. The optional 6844``cleanup`` flag indicates that the landing pad block is a cleanup. 6845 6846A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and 6847contains the global variable representing the "type" that may be caught 6848or filtered respectively. Unlike the ``catch`` clause, the ``filter`` 6849clause takes an array constant as its argument. Use 6850"``[0 x i8**] undef``" for a filter which cannot throw. The 6851'``landingpad``' instruction must contain *at least* one ``clause`` or 6852the ``cleanup`` flag. 6853 6854Semantics: 6855"""""""""" 6856 6857The '``landingpad``' instruction defines the values which are set by the 6858personality function (``pers_fn``) upon re-entry to the function, and 6859therefore the "result type" of the ``landingpad`` instruction. As with 6860calling conventions, how the personality function results are 6861represented in LLVM IR is target specific. 6862 6863The clauses are applied in order from top to bottom. If two 6864``landingpad`` instructions are merged together through inlining, the 6865clauses from the calling function are appended to the list of clauses. 6866When the call stack is being unwound due to an exception being thrown, 6867the exception is compared against each ``clause`` in turn. If it doesn't 6868match any of the clauses, and the ``cleanup`` flag is not set, then 6869unwinding continues further up the call stack. 6870 6871The ``landingpad`` instruction has several restrictions: 6872 6873- A landing pad block is a basic block which is the unwind destination 6874 of an '``invoke``' instruction. 6875- A landing pad block must have a '``landingpad``' instruction as its 6876 first non-PHI instruction. 6877- There can be only one '``landingpad``' instruction within the landing 6878 pad block. 6879- A basic block that is not a landing pad block may not include a 6880 '``landingpad``' instruction. 6881- All '``landingpad``' instructions in a function must have the same 6882 personality function. 6883 6884Example: 6885"""""""" 6886 6887.. code-block:: llvm 6888 6889 ;; A landing pad which can catch an integer. 6890 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 6891 catch i8** @_ZTIi 6892 ;; A landing pad that is a cleanup. 6893 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 6894 cleanup 6895 ;; A landing pad which can catch an integer and can only throw a double. 6896 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 6897 catch i8** @_ZTIi 6898 filter [1 x i8**] [@_ZTId] 6899 6900.. _intrinsics: 6901 6902Intrinsic Functions 6903=================== 6904 6905LLVM supports the notion of an "intrinsic function". These functions 6906have well known names and semantics and are required to follow certain 6907restrictions. Overall, these intrinsics represent an extension mechanism 6908for the LLVM language that does not require changing all of the 6909transformations in LLVM when adding to the language (or the bitcode 6910reader/writer, the parser, etc...). 6911 6912Intrinsic function names must all start with an "``llvm.``" prefix. This 6913prefix is reserved in LLVM for intrinsic names; thus, function names may 6914not begin with this prefix. Intrinsic functions must always be external 6915functions: you cannot define the body of intrinsic functions. Intrinsic 6916functions may only be used in call or invoke instructions: it is illegal 6917to take the address of an intrinsic function. Additionally, because 6918intrinsic functions are part of the LLVM language, it is required if any 6919are added that they be documented here. 6920 6921Some intrinsic functions can be overloaded, i.e., the intrinsic 6922represents a family of functions that perform the same operation but on 6923different data types. Because LLVM can represent over 8 million 6924different integer types, overloading is used commonly to allow an 6925intrinsic function to operate on any integer type. One or more of the 6926argument types or the result type can be overloaded to accept any 6927integer type. Argument types may also be defined as exactly matching a 6928previous argument's type or the result type. This allows an intrinsic 6929function which accepts multiple arguments, but needs all of them to be 6930of the same type, to only be overloaded with respect to a single 6931argument or the result. 6932 6933Overloaded intrinsics will have the names of its overloaded argument 6934types encoded into its function name, each preceded by a period. Only 6935those types which are overloaded result in a name suffix. Arguments 6936whose type is matched against another type do not. For example, the 6937``llvm.ctpop`` function can take an integer of any width and returns an 6938integer of exactly the same integer width. This leads to a family of 6939functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and 6940``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is 6941overloaded, and only one type suffix is required. Because the argument's 6942type is matched against the return type, it does not require its own 6943name suffix. 6944 6945To learn how to add an intrinsic function, please see the `Extending 6946LLVM Guide <ExtendingLLVM.html>`_. 6947 6948.. _int_varargs: 6949 6950Variable Argument Handling Intrinsics 6951------------------------------------- 6952 6953Variable argument support is defined in LLVM with the 6954:ref:`va_arg <i_va_arg>` instruction and these three intrinsic 6955functions. These functions are related to the similarly named macros 6956defined in the ``<stdarg.h>`` header file. 6957 6958All of these functions operate on arguments that use a target-specific 6959value type "``va_list``". The LLVM assembly language reference manual 6960does not define what this type is, so all transformations should be 6961prepared to handle these functions regardless of the type used. 6962 6963This example shows how the :ref:`va_arg <i_va_arg>` instruction and the 6964variable argument handling intrinsic functions are used. 6965 6966.. code-block:: llvm 6967 6968 ; This struct is different for every platform. For most platforms, 6969 ; it is merely an i8*. 6970 %struct.va_list = type { i8* } 6971 6972 ; For Unix x86_64 platforms, va_list is the following struct: 6973 ; %struct.va_list = type { i32, i32, i8*, i8* } 6974 6975 define i32 @test(i32 %X, ...) { 6976 ; Initialize variable argument processing 6977 %ap = alloca %struct.va_list 6978 %ap2 = bitcast %struct.va_list* %ap to i8* 6979 call void @llvm.va_start(i8* %ap2) 6980 6981 ; Read a single integer argument 6982 %tmp = va_arg i8* %ap2, i32 6983 6984 ; Demonstrate usage of llvm.va_copy and llvm.va_end 6985 %aq = alloca i8* 6986 %aq2 = bitcast i8** %aq to i8* 6987 call void @llvm.va_copy(i8* %aq2, i8* %ap2) 6988 call void @llvm.va_end(i8* %aq2) 6989 6990 ; Stop processing of arguments. 6991 call void @llvm.va_end(i8* %ap2) 6992 ret i32 %tmp 6993 } 6994 6995 declare void @llvm.va_start(i8*) 6996 declare void @llvm.va_copy(i8*, i8*) 6997 declare void @llvm.va_end(i8*) 6998 6999.. _int_va_start: 7000 7001'``llvm.va_start``' Intrinsic 7002^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7003 7004Syntax: 7005""""""" 7006 7007:: 7008 7009 declare void @llvm.va_start(i8* <arglist>) 7010 7011Overview: 7012""""""""" 7013 7014The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for 7015subsequent use by ``va_arg``. 7016 7017Arguments: 7018"""""""""" 7019 7020The argument is a pointer to a ``va_list`` element to initialize. 7021 7022Semantics: 7023"""""""""" 7024 7025The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro 7026available in C. In a target-dependent way, it initializes the 7027``va_list`` element to which the argument points, so that the next call 7028to ``va_arg`` will produce the first variable argument passed to the 7029function. Unlike the C ``va_start`` macro, this intrinsic does not need 7030to know the last argument of the function as the compiler can figure 7031that out. 7032 7033'``llvm.va_end``' Intrinsic 7034^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7035 7036Syntax: 7037""""""" 7038 7039:: 7040 7041 declare void @llvm.va_end(i8* <arglist>) 7042 7043Overview: 7044""""""""" 7045 7046The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been 7047initialized previously with ``llvm.va_start`` or ``llvm.va_copy``. 7048 7049Arguments: 7050"""""""""" 7051 7052The argument is a pointer to a ``va_list`` to destroy. 7053 7054Semantics: 7055"""""""""" 7056 7057The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro 7058available in C. In a target-dependent way, it destroys the ``va_list`` 7059element to which the argument points. Calls to 7060:ref:`llvm.va_start <int_va_start>` and 7061:ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to 7062``llvm.va_end``. 7063 7064.. _int_va_copy: 7065 7066'``llvm.va_copy``' Intrinsic 7067^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7068 7069Syntax: 7070""""""" 7071 7072:: 7073 7074 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>) 7075 7076Overview: 7077""""""""" 7078 7079The '``llvm.va_copy``' intrinsic copies the current argument position 7080from the source argument list to the destination argument list. 7081 7082Arguments: 7083"""""""""" 7084 7085The first argument is a pointer to a ``va_list`` element to initialize. 7086The second argument is a pointer to a ``va_list`` element to copy from. 7087 7088Semantics: 7089"""""""""" 7090 7091The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro 7092available in C. In a target-dependent way, it copies the source 7093``va_list`` element into the destination ``va_list`` element. This 7094intrinsic is necessary because the `` llvm.va_start`` intrinsic may be 7095arbitrarily complex and require, for example, memory allocation. 7096 7097Accurate Garbage Collection Intrinsics 7098-------------------------------------- 7099 7100LLVM support for `Accurate Garbage Collection <GarbageCollection.html>`_ 7101(GC) requires the implementation and generation of these intrinsics. 7102These intrinsics allow identification of :ref:`GC roots on the 7103stack <int_gcroot>`, as well as garbage collector implementations that 7104require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers. 7105Front-ends for type-safe garbage collected languages should generate 7106these intrinsics to make use of the LLVM garbage collectors. For more 7107details, see `Accurate Garbage Collection with 7108LLVM <GarbageCollection.html>`_. 7109 7110The garbage collection intrinsics only operate on objects in the generic 7111address space (address space zero). 7112 7113.. _int_gcroot: 7114 7115'``llvm.gcroot``' Intrinsic 7116^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7117 7118Syntax: 7119""""""" 7120 7121:: 7122 7123 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata) 7124 7125Overview: 7126""""""""" 7127 7128The '``llvm.gcroot``' intrinsic declares the existence of a GC root to 7129the code generator, and allows some metadata to be associated with it. 7130 7131Arguments: 7132"""""""""" 7133 7134The first argument specifies the address of a stack object that contains 7135the root pointer. The second pointer (which must be either a constant or 7136a global value address) contains the meta-data to be associated with the 7137root. 7138 7139Semantics: 7140"""""""""" 7141 7142At runtime, a call to this intrinsic stores a null pointer into the 7143"ptrloc" location. At compile-time, the code generator generates 7144information to allow the runtime to find the pointer at GC safe points. 7145The '``llvm.gcroot``' intrinsic may only be used in a function which 7146:ref:`specifies a GC algorithm <gc>`. 7147 7148.. _int_gcread: 7149 7150'``llvm.gcread``' Intrinsic 7151^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7152 7153Syntax: 7154""""""" 7155 7156:: 7157 7158 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr) 7159 7160Overview: 7161""""""""" 7162 7163The '``llvm.gcread``' intrinsic identifies reads of references from heap 7164locations, allowing garbage collector implementations that require read 7165barriers. 7166 7167Arguments: 7168"""""""""" 7169 7170The second argument is the address to read from, which should be an 7171address allocated from the garbage collector. The first object is a 7172pointer to the start of the referenced object, if needed by the language 7173runtime (otherwise null). 7174 7175Semantics: 7176"""""""""" 7177 7178The '``llvm.gcread``' intrinsic has the same semantics as a load 7179instruction, but may be replaced with substantially more complex code by 7180the garbage collector runtime, as needed. The '``llvm.gcread``' 7181intrinsic may only be used in a function which :ref:`specifies a GC 7182algorithm <gc>`. 7183 7184.. _int_gcwrite: 7185 7186'``llvm.gcwrite``' Intrinsic 7187^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7188 7189Syntax: 7190""""""" 7191 7192:: 7193 7194 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2) 7195 7196Overview: 7197""""""""" 7198 7199The '``llvm.gcwrite``' intrinsic identifies writes of references to heap 7200locations, allowing garbage collector implementations that require write 7201barriers (such as generational or reference counting collectors). 7202 7203Arguments: 7204"""""""""" 7205 7206The first argument is the reference to store, the second is the start of 7207the object to store it to, and the third is the address of the field of 7208Obj to store to. If the runtime does not require a pointer to the 7209object, Obj may be null. 7210 7211Semantics: 7212"""""""""" 7213 7214The '``llvm.gcwrite``' intrinsic has the same semantics as a store 7215instruction, but may be replaced with substantially more complex code by 7216the garbage collector runtime, as needed. The '``llvm.gcwrite``' 7217intrinsic may only be used in a function which :ref:`specifies a GC 7218algorithm <gc>`. 7219 7220Code Generator Intrinsics 7221------------------------- 7222 7223These intrinsics are provided by LLVM to expose special features that 7224may only be implemented with code generator support. 7225 7226'``llvm.returnaddress``' Intrinsic 7227^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7228 7229Syntax: 7230""""""" 7231 7232:: 7233 7234 declare i8 *@llvm.returnaddress(i32 <level>) 7235 7236Overview: 7237""""""""" 7238 7239The '``llvm.returnaddress``' intrinsic attempts to compute a 7240target-specific value indicating the return address of the current 7241function or one of its callers. 7242 7243Arguments: 7244"""""""""" 7245 7246The argument to this intrinsic indicates which function to return the 7247address for. Zero indicates the calling function, one indicates its 7248caller, etc. The argument is **required** to be a constant integer 7249value. 7250 7251Semantics: 7252"""""""""" 7253 7254The '``llvm.returnaddress``' intrinsic either returns a pointer 7255indicating the return address of the specified call frame, or zero if it 7256cannot be identified. The value returned by this intrinsic is likely to 7257be incorrect or 0 for arguments other than zero, so it should only be 7258used for debugging purposes. 7259 7260Note that calling this intrinsic does not prevent function inlining or 7261other aggressive transformations, so the value returned may not be that 7262of the obvious source-language caller. 7263 7264'``llvm.frameaddress``' Intrinsic 7265^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7266 7267Syntax: 7268""""""" 7269 7270:: 7271 7272 declare i8* @llvm.frameaddress(i32 <level>) 7273 7274Overview: 7275""""""""" 7276 7277The '``llvm.frameaddress``' intrinsic attempts to return the 7278target-specific frame pointer value for the specified stack frame. 7279 7280Arguments: 7281"""""""""" 7282 7283The argument to this intrinsic indicates which function to return the 7284frame pointer for. Zero indicates the calling function, one indicates 7285its caller, etc. The argument is **required** to be a constant integer 7286value. 7287 7288Semantics: 7289"""""""""" 7290 7291The '``llvm.frameaddress``' intrinsic either returns a pointer 7292indicating the frame address of the specified call frame, or zero if it 7293cannot be identified. The value returned by this intrinsic is likely to 7294be incorrect or 0 for arguments other than zero, so it should only be 7295used for debugging purposes. 7296 7297Note that calling this intrinsic does not prevent function inlining or 7298other aggressive transformations, so the value returned may not be that 7299of the obvious source-language caller. 7300 7301'``llvm.frameallocate``' and '``llvm.framerecover``' Intrinsics 7302^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7303 7304Syntax: 7305""""""" 7306 7307:: 7308 7309 declare i8* @llvm.frameallocate(i32 %size) 7310 declare i8* @llvm.framerecover(i8* %func, i8* %fp) 7311 7312Overview: 7313""""""""" 7314 7315The '``llvm.frameallocate``' intrinsic allocates stack memory at some fixed 7316offset from the frame pointer, and the '``llvm.framerecover``' 7317intrinsic applies that offset to a live frame pointer to recover the address of 7318the allocation. The offset is computed during frame layout of the caller of 7319``llvm.frameallocate``. 7320 7321Arguments: 7322"""""""""" 7323 7324The ``size`` argument to '``llvm.frameallocate``' must be a constant integer 7325indicating the amount of stack memory to allocate. As with allocas, allocating 7326zero bytes is legal, but the result is undefined. 7327 7328The ``func`` argument to '``llvm.framerecover``' must be a constant 7329bitcasted pointer to a function defined in the current module. The code 7330generator cannot determine the frame allocation offset of functions defined in 7331other modules. 7332 7333The ``fp`` argument to '``llvm.framerecover``' must be a frame 7334pointer of a call frame that is currently live. The return value of 7335'``llvm.frameaddress``' is one way to produce such a value, but most platforms 7336also expose the frame pointer through stack unwinding mechanisms. 7337 7338Semantics: 7339"""""""""" 7340 7341These intrinsics allow a group of functions to access one stack memory 7342allocation in an ancestor stack frame. The memory returned from 7343'``llvm.frameallocate``' may be allocated prior to stack realignment, so the 7344memory is only aligned to the ABI-required stack alignment. Each function may 7345only call '``llvm.frameallocate``' one or zero times from the function entry 7346block. The frame allocation intrinsic inhibits inlining, as any frame 7347allocations in the inlined function frame are likely to be at a different 7348offset from the one used by '``llvm.framerecover``' called with the 7349uninlined function. 7350 7351.. _int_read_register: 7352.. _int_write_register: 7353 7354'``llvm.read_register``' and '``llvm.write_register``' Intrinsics 7355^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7356 7357Syntax: 7358""""""" 7359 7360:: 7361 7362 declare i32 @llvm.read_register.i32(metadata) 7363 declare i64 @llvm.read_register.i64(metadata) 7364 declare void @llvm.write_register.i32(metadata, i32 @value) 7365 declare void @llvm.write_register.i64(metadata, i64 @value) 7366 !0 = !{!"sp\00"} 7367 7368Overview: 7369""""""""" 7370 7371The '``llvm.read_register``' and '``llvm.write_register``' intrinsics 7372provides access to the named register. The register must be valid on 7373the architecture being compiled to. The type needs to be compatible 7374with the register being read. 7375 7376Semantics: 7377"""""""""" 7378 7379The '``llvm.read_register``' intrinsic returns the current value of the 7380register, where possible. The '``llvm.write_register``' intrinsic sets 7381the current value of the register, where possible. 7382 7383This is useful to implement named register global variables that need 7384to always be mapped to a specific register, as is common practice on 7385bare-metal programs including OS kernels. 7386 7387The compiler doesn't check for register availability or use of the used 7388register in surrounding code, including inline assembly. Because of that, 7389allocatable registers are not supported. 7390 7391Warning: So far it only works with the stack pointer on selected 7392architectures (ARM, AArch64, PowerPC and x86_64). Significant amount of 7393work is needed to support other registers and even more so, allocatable 7394registers. 7395 7396.. _int_stacksave: 7397 7398'``llvm.stacksave``' Intrinsic 7399^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7400 7401Syntax: 7402""""""" 7403 7404:: 7405 7406 declare i8* @llvm.stacksave() 7407 7408Overview: 7409""""""""" 7410 7411The '``llvm.stacksave``' intrinsic is used to remember the current state 7412of the function stack, for use with 7413:ref:`llvm.stackrestore <int_stackrestore>`. This is useful for 7414implementing language features like scoped automatic variable sized 7415arrays in C99. 7416 7417Semantics: 7418"""""""""" 7419 7420This intrinsic returns a opaque pointer value that can be passed to 7421:ref:`llvm.stackrestore <int_stackrestore>`. When an 7422``llvm.stackrestore`` intrinsic is executed with a value saved from 7423``llvm.stacksave``, it effectively restores the state of the stack to 7424the state it was in when the ``llvm.stacksave`` intrinsic executed. In 7425practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that 7426were allocated after the ``llvm.stacksave`` was executed. 7427 7428.. _int_stackrestore: 7429 7430'``llvm.stackrestore``' Intrinsic 7431^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7432 7433Syntax: 7434""""""" 7435 7436:: 7437 7438 declare void @llvm.stackrestore(i8* %ptr) 7439 7440Overview: 7441""""""""" 7442 7443The '``llvm.stackrestore``' intrinsic is used to restore the state of 7444the function stack to the state it was in when the corresponding 7445:ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is 7446useful for implementing language features like scoped automatic variable 7447sized arrays in C99. 7448 7449Semantics: 7450"""""""""" 7451 7452See the description for :ref:`llvm.stacksave <int_stacksave>`. 7453 7454'``llvm.prefetch``' Intrinsic 7455^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7456 7457Syntax: 7458""""""" 7459 7460:: 7461 7462 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>) 7463 7464Overview: 7465""""""""" 7466 7467The '``llvm.prefetch``' intrinsic is a hint to the code generator to 7468insert a prefetch instruction if supported; otherwise, it is a noop. 7469Prefetches have no effect on the behavior of the program but can change 7470its performance characteristics. 7471 7472Arguments: 7473"""""""""" 7474 7475``address`` is the address to be prefetched, ``rw`` is the specifier 7476determining if the fetch should be for a read (0) or write (1), and 7477``locality`` is a temporal locality specifier ranging from (0) - no 7478locality, to (3) - extremely local keep in cache. The ``cache type`` 7479specifies whether the prefetch is performed on the data (1) or 7480instruction (0) cache. The ``rw``, ``locality`` and ``cache type`` 7481arguments must be constant integers. 7482 7483Semantics: 7484"""""""""" 7485 7486This intrinsic does not modify the behavior of the program. In 7487particular, prefetches cannot trap and do not produce a value. On 7488targets that support this intrinsic, the prefetch can provide hints to 7489the processor cache for better performance. 7490 7491'``llvm.pcmarker``' Intrinsic 7492^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7493 7494Syntax: 7495""""""" 7496 7497:: 7498 7499 declare void @llvm.pcmarker(i32 <id>) 7500 7501Overview: 7502""""""""" 7503 7504The '``llvm.pcmarker``' intrinsic is a method to export a Program 7505Counter (PC) in a region of code to simulators and other tools. The 7506method is target specific, but it is expected that the marker will use 7507exported symbols to transmit the PC of the marker. The marker makes no 7508guarantees that it will remain with any specific instruction after 7509optimizations. It is possible that the presence of a marker will inhibit 7510optimizations. The intended use is to be inserted after optimizations to 7511allow correlations of simulation runs. 7512 7513Arguments: 7514"""""""""" 7515 7516``id`` is a numerical id identifying the marker. 7517 7518Semantics: 7519"""""""""" 7520 7521This intrinsic does not modify the behavior of the program. Backends 7522that do not support this intrinsic may ignore it. 7523 7524'``llvm.readcyclecounter``' Intrinsic 7525^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7526 7527Syntax: 7528""""""" 7529 7530:: 7531 7532 declare i64 @llvm.readcyclecounter() 7533 7534Overview: 7535""""""""" 7536 7537The '``llvm.readcyclecounter``' intrinsic provides access to the cycle 7538counter register (or similar low latency, high accuracy clocks) on those 7539targets that support it. On X86, it should map to RDTSC. On Alpha, it 7540should map to RPCC. As the backing counters overflow quickly (on the 7541order of 9 seconds on alpha), this should only be used for small 7542timings. 7543 7544Semantics: 7545"""""""""" 7546 7547When directly supported, reading the cycle counter should not modify any 7548memory. Implementations are allowed to either return a application 7549specific value or a system wide value. On backends without support, this 7550is lowered to a constant 0. 7551 7552Note that runtime support may be conditional on the privilege-level code is 7553running at and the host platform. 7554 7555'``llvm.clear_cache``' Intrinsic 7556^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7557 7558Syntax: 7559""""""" 7560 7561:: 7562 7563 declare void @llvm.clear_cache(i8*, i8*) 7564 7565Overview: 7566""""""""" 7567 7568The '``llvm.clear_cache``' intrinsic ensures visibility of modifications 7569in the specified range to the execution unit of the processor. On 7570targets with non-unified instruction and data cache, the implementation 7571flushes the instruction cache. 7572 7573Semantics: 7574"""""""""" 7575 7576On platforms with coherent instruction and data caches (e.g. x86), this 7577intrinsic is a nop. On platforms with non-coherent instruction and data 7578cache (e.g. ARM, MIPS), the intrinsic is lowered either to appropriate 7579instructions or a system call, if cache flushing requires special 7580privileges. 7581 7582The default behavior is to emit a call to ``__clear_cache`` from the run 7583time library. 7584 7585This instrinsic does *not* empty the instruction pipeline. Modifications 7586of the current function are outside the scope of the intrinsic. 7587 7588'``llvm.instrprof_increment``' Intrinsic 7589^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7590 7591Syntax: 7592""""""" 7593 7594:: 7595 7596 declare void @llvm.instrprof_increment(i8* <name>, i64 <hash>, 7597 i32 <num-counters>, i32 <index>) 7598 7599Overview: 7600""""""""" 7601 7602The '``llvm.instrprof_increment``' intrinsic can be emitted by a 7603frontend for use with instrumentation based profiling. These will be 7604lowered by the ``-instrprof`` pass to generate execution counts of a 7605program at runtime. 7606 7607Arguments: 7608"""""""""" 7609 7610The first argument is a pointer to a global variable containing the 7611name of the entity being instrumented. This should generally be the 7612(mangled) function name for a set of counters. 7613 7614The second argument is a hash value that can be used by the consumer 7615of the profile data to detect changes to the instrumented source, and 7616the third is the number of counters associated with ``name``. It is an 7617error if ``hash`` or ``num-counters`` differ between two instances of 7618``instrprof_increment`` that refer to the same name. 7619 7620The last argument refers to which of the counters for ``name`` should 7621be incremented. It should be a value between 0 and ``num-counters``. 7622 7623Semantics: 7624"""""""""" 7625 7626This intrinsic represents an increment of a profiling counter. It will 7627cause the ``-instrprof`` pass to generate the appropriate data 7628structures and the code to increment the appropriate value, in a 7629format that can be written out by a compiler runtime and consumed via 7630the ``llvm-profdata`` tool. 7631 7632Standard C Library Intrinsics 7633----------------------------- 7634 7635LLVM provides intrinsics for a few important standard C library 7636functions. These intrinsics allow source-language front-ends to pass 7637information about the alignment of the pointer arguments to the code 7638generator, providing opportunity for more efficient code generation. 7639 7640.. _int_memcpy: 7641 7642'``llvm.memcpy``' Intrinsic 7643^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7644 7645Syntax: 7646""""""" 7647 7648This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any 7649integer bit width and for different address spaces. Not all targets 7650support all bit widths however. 7651 7652:: 7653 7654 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>, 7655 i32 <len>, i32 <align>, i1 <isvolatile>) 7656 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>, 7657 i64 <len>, i32 <align>, i1 <isvolatile>) 7658 7659Overview: 7660""""""""" 7661 7662The '``llvm.memcpy.*``' intrinsics copy a block of memory from the 7663source location to the destination location. 7664 7665Note that, unlike the standard libc function, the ``llvm.memcpy.*`` 7666intrinsics do not return a value, takes extra alignment/isvolatile 7667arguments and the pointers can be in specified address spaces. 7668 7669Arguments: 7670"""""""""" 7671 7672The first argument is a pointer to the destination, the second is a 7673pointer to the source. The third argument is an integer argument 7674specifying the number of bytes to copy, the fourth argument is the 7675alignment of the source and destination locations, and the fifth is a 7676boolean indicating a volatile access. 7677 7678If the call to this intrinsic has an alignment value that is not 0 or 1, 7679then the caller guarantees that both the source and destination pointers 7680are aligned to that boundary. 7681 7682If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is 7683a :ref:`volatile operation <volatile>`. The detailed access behavior is not 7684very cleanly specified and it is unwise to depend on it. 7685 7686Semantics: 7687"""""""""" 7688 7689The '``llvm.memcpy.*``' intrinsics copy a block of memory from the 7690source location to the destination location, which are not allowed to 7691overlap. It copies "len" bytes of memory over. If the argument is known 7692to be aligned to some boundary, this can be specified as the fourth 7693argument, otherwise it should be set to 0 or 1 (both meaning no alignment). 7694 7695'``llvm.memmove``' Intrinsic 7696^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7697 7698Syntax: 7699""""""" 7700 7701This is an overloaded intrinsic. You can use llvm.memmove on any integer 7702bit width and for different address space. Not all targets support all 7703bit widths however. 7704 7705:: 7706 7707 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>, 7708 i32 <len>, i32 <align>, i1 <isvolatile>) 7709 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>, 7710 i64 <len>, i32 <align>, i1 <isvolatile>) 7711 7712Overview: 7713""""""""" 7714 7715The '``llvm.memmove.*``' intrinsics move a block of memory from the 7716source location to the destination location. It is similar to the 7717'``llvm.memcpy``' intrinsic but allows the two memory locations to 7718overlap. 7719 7720Note that, unlike the standard libc function, the ``llvm.memmove.*`` 7721intrinsics do not return a value, takes extra alignment/isvolatile 7722arguments and the pointers can be in specified address spaces. 7723 7724Arguments: 7725"""""""""" 7726 7727The first argument is a pointer to the destination, the second is a 7728pointer to the source. The third argument is an integer argument 7729specifying the number of bytes to copy, the fourth argument is the 7730alignment of the source and destination locations, and the fifth is a 7731boolean indicating a volatile access. 7732 7733If the call to this intrinsic has an alignment value that is not 0 or 1, 7734then the caller guarantees that the source and destination pointers are 7735aligned to that boundary. 7736 7737If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call 7738is a :ref:`volatile operation <volatile>`. The detailed access behavior is 7739not very cleanly specified and it is unwise to depend on it. 7740 7741Semantics: 7742"""""""""" 7743 7744The '``llvm.memmove.*``' intrinsics copy a block of memory from the 7745source location to the destination location, which may overlap. It 7746copies "len" bytes of memory over. If the argument is known to be 7747aligned to some boundary, this can be specified as the fourth argument, 7748otherwise it should be set to 0 or 1 (both meaning no alignment). 7749 7750'``llvm.memset.*``' Intrinsics 7751^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7752 7753Syntax: 7754""""""" 7755 7756This is an overloaded intrinsic. You can use llvm.memset on any integer 7757bit width and for different address spaces. However, not all targets 7758support all bit widths. 7759 7760:: 7761 7762 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>, 7763 i32 <len>, i32 <align>, i1 <isvolatile>) 7764 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>, 7765 i64 <len>, i32 <align>, i1 <isvolatile>) 7766 7767Overview: 7768""""""""" 7769 7770The '``llvm.memset.*``' intrinsics fill a block of memory with a 7771particular byte value. 7772 7773Note that, unlike the standard libc function, the ``llvm.memset`` 7774intrinsic does not return a value and takes extra alignment/volatile 7775arguments. Also, the destination can be in an arbitrary address space. 7776 7777Arguments: 7778"""""""""" 7779 7780The first argument is a pointer to the destination to fill, the second 7781is the byte value with which to fill it, the third argument is an 7782integer argument specifying the number of bytes to fill, and the fourth 7783argument is the known alignment of the destination location. 7784 7785If the call to this intrinsic has an alignment value that is not 0 or 1, 7786then the caller guarantees that the destination pointer is aligned to 7787that boundary. 7788 7789If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is 7790a :ref:`volatile operation <volatile>`. The detailed access behavior is not 7791very cleanly specified and it is unwise to depend on it. 7792 7793Semantics: 7794"""""""""" 7795 7796The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting 7797at the destination location. If the argument is known to be aligned to 7798some boundary, this can be specified as the fourth argument, otherwise 7799it should be set to 0 or 1 (both meaning no alignment). 7800 7801'``llvm.sqrt.*``' Intrinsic 7802^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7803 7804Syntax: 7805""""""" 7806 7807This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any 7808floating point or vector of floating point type. Not all targets support 7809all types however. 7810 7811:: 7812 7813 declare float @llvm.sqrt.f32(float %Val) 7814 declare double @llvm.sqrt.f64(double %Val) 7815 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val) 7816 declare fp128 @llvm.sqrt.f128(fp128 %Val) 7817 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val) 7818 7819Overview: 7820""""""""" 7821 7822The '``llvm.sqrt``' intrinsics return the sqrt of the specified operand, 7823returning the same value as the libm '``sqrt``' functions would. Unlike 7824``sqrt`` in libm, however, ``llvm.sqrt`` has undefined behavior for 7825negative numbers other than -0.0 (which allows for better optimization, 7826because there is no need to worry about errno being set). 7827``llvm.sqrt(-0.0)`` is defined to return -0.0 like IEEE sqrt. 7828 7829Arguments: 7830"""""""""" 7831 7832The argument and return value are floating point numbers of the same 7833type. 7834 7835Semantics: 7836"""""""""" 7837 7838This function returns the sqrt of the specified operand if it is a 7839nonnegative floating point number. 7840 7841'``llvm.powi.*``' Intrinsic 7842^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7843 7844Syntax: 7845""""""" 7846 7847This is an overloaded intrinsic. You can use ``llvm.powi`` on any 7848floating point or vector of floating point type. Not all targets support 7849all types however. 7850 7851:: 7852 7853 declare float @llvm.powi.f32(float %Val, i32 %power) 7854 declare double @llvm.powi.f64(double %Val, i32 %power) 7855 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power) 7856 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power) 7857 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power) 7858 7859Overview: 7860""""""""" 7861 7862The '``llvm.powi.*``' intrinsics return the first operand raised to the 7863specified (positive or negative) power. The order of evaluation of 7864multiplications is not defined. When a vector of floating point type is 7865used, the second argument remains a scalar integer value. 7866 7867Arguments: 7868"""""""""" 7869 7870The second argument is an integer power, and the first is a value to 7871raise to that power. 7872 7873Semantics: 7874"""""""""" 7875 7876This function returns the first value raised to the second power with an 7877unspecified sequence of rounding operations. 7878 7879'``llvm.sin.*``' Intrinsic 7880^^^^^^^^^^^^^^^^^^^^^^^^^^ 7881 7882Syntax: 7883""""""" 7884 7885This is an overloaded intrinsic. You can use ``llvm.sin`` on any 7886floating point or vector of floating point type. Not all targets support 7887all types however. 7888 7889:: 7890 7891 declare float @llvm.sin.f32(float %Val) 7892 declare double @llvm.sin.f64(double %Val) 7893 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val) 7894 declare fp128 @llvm.sin.f128(fp128 %Val) 7895 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val) 7896 7897Overview: 7898""""""""" 7899 7900The '``llvm.sin.*``' intrinsics return the sine of the operand. 7901 7902Arguments: 7903"""""""""" 7904 7905The argument and return value are floating point numbers of the same 7906type. 7907 7908Semantics: 7909"""""""""" 7910 7911This function returns the sine of the specified operand, returning the 7912same values as the libm ``sin`` functions would, and handles error 7913conditions in the same way. 7914 7915'``llvm.cos.*``' Intrinsic 7916^^^^^^^^^^^^^^^^^^^^^^^^^^ 7917 7918Syntax: 7919""""""" 7920 7921This is an overloaded intrinsic. You can use ``llvm.cos`` on any 7922floating point or vector of floating point type. Not all targets support 7923all types however. 7924 7925:: 7926 7927 declare float @llvm.cos.f32(float %Val) 7928 declare double @llvm.cos.f64(double %Val) 7929 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val) 7930 declare fp128 @llvm.cos.f128(fp128 %Val) 7931 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val) 7932 7933Overview: 7934""""""""" 7935 7936The '``llvm.cos.*``' intrinsics return the cosine of the operand. 7937 7938Arguments: 7939"""""""""" 7940 7941The argument and return value are floating point numbers of the same 7942type. 7943 7944Semantics: 7945"""""""""" 7946 7947This function returns the cosine of the specified operand, returning the 7948same values as the libm ``cos`` functions would, and handles error 7949conditions in the same way. 7950 7951'``llvm.pow.*``' Intrinsic 7952^^^^^^^^^^^^^^^^^^^^^^^^^^ 7953 7954Syntax: 7955""""""" 7956 7957This is an overloaded intrinsic. You can use ``llvm.pow`` on any 7958floating point or vector of floating point type. Not all targets support 7959all types however. 7960 7961:: 7962 7963 declare float @llvm.pow.f32(float %Val, float %Power) 7964 declare double @llvm.pow.f64(double %Val, double %Power) 7965 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power) 7966 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power) 7967 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power) 7968 7969Overview: 7970""""""""" 7971 7972The '``llvm.pow.*``' intrinsics return the first operand raised to the 7973specified (positive or negative) power. 7974 7975Arguments: 7976"""""""""" 7977 7978The second argument is a floating point power, and the first is a value 7979to raise to that power. 7980 7981Semantics: 7982"""""""""" 7983 7984This function returns the first value raised to the second power, 7985returning the same values as the libm ``pow`` functions would, and 7986handles error conditions in the same way. 7987 7988'``llvm.exp.*``' Intrinsic 7989^^^^^^^^^^^^^^^^^^^^^^^^^^ 7990 7991Syntax: 7992""""""" 7993 7994This is an overloaded intrinsic. You can use ``llvm.exp`` on any 7995floating point or vector of floating point type. Not all targets support 7996all types however. 7997 7998:: 7999 8000 declare float @llvm.exp.f32(float %Val) 8001 declare double @llvm.exp.f64(double %Val) 8002 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val) 8003 declare fp128 @llvm.exp.f128(fp128 %Val) 8004 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val) 8005 8006Overview: 8007""""""""" 8008 8009The '``llvm.exp.*``' intrinsics perform the exp function. 8010 8011Arguments: 8012"""""""""" 8013 8014The argument and return value are floating point numbers of the same 8015type. 8016 8017Semantics: 8018"""""""""" 8019 8020This function returns the same values as the libm ``exp`` functions 8021would, and handles error conditions in the same way. 8022 8023'``llvm.exp2.*``' Intrinsic 8024^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8025 8026Syntax: 8027""""""" 8028 8029This is an overloaded intrinsic. You can use ``llvm.exp2`` on any 8030floating point or vector of floating point type. Not all targets support 8031all types however. 8032 8033:: 8034 8035 declare float @llvm.exp2.f32(float %Val) 8036 declare double @llvm.exp2.f64(double %Val) 8037 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val) 8038 declare fp128 @llvm.exp2.f128(fp128 %Val) 8039 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val) 8040 8041Overview: 8042""""""""" 8043 8044The '``llvm.exp2.*``' intrinsics perform the exp2 function. 8045 8046Arguments: 8047"""""""""" 8048 8049The argument and return value are floating point numbers of the same 8050type. 8051 8052Semantics: 8053"""""""""" 8054 8055This function returns the same values as the libm ``exp2`` functions 8056would, and handles error conditions in the same way. 8057 8058'``llvm.log.*``' Intrinsic 8059^^^^^^^^^^^^^^^^^^^^^^^^^^ 8060 8061Syntax: 8062""""""" 8063 8064This is an overloaded intrinsic. You can use ``llvm.log`` on any 8065floating point or vector of floating point type. Not all targets support 8066all types however. 8067 8068:: 8069 8070 declare float @llvm.log.f32(float %Val) 8071 declare double @llvm.log.f64(double %Val) 8072 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val) 8073 declare fp128 @llvm.log.f128(fp128 %Val) 8074 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val) 8075 8076Overview: 8077""""""""" 8078 8079The '``llvm.log.*``' intrinsics perform the log function. 8080 8081Arguments: 8082"""""""""" 8083 8084The argument and return value are floating point numbers of the same 8085type. 8086 8087Semantics: 8088"""""""""" 8089 8090This function returns the same values as the libm ``log`` functions 8091would, and handles error conditions in the same way. 8092 8093'``llvm.log10.*``' Intrinsic 8094^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8095 8096Syntax: 8097""""""" 8098 8099This is an overloaded intrinsic. You can use ``llvm.log10`` on any 8100floating point or vector of floating point type. Not all targets support 8101all types however. 8102 8103:: 8104 8105 declare float @llvm.log10.f32(float %Val) 8106 declare double @llvm.log10.f64(double %Val) 8107 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val) 8108 declare fp128 @llvm.log10.f128(fp128 %Val) 8109 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val) 8110 8111Overview: 8112""""""""" 8113 8114The '``llvm.log10.*``' intrinsics perform the log10 function. 8115 8116Arguments: 8117"""""""""" 8118 8119The argument and return value are floating point numbers of the same 8120type. 8121 8122Semantics: 8123"""""""""" 8124 8125This function returns the same values as the libm ``log10`` functions 8126would, and handles error conditions in the same way. 8127 8128'``llvm.log2.*``' Intrinsic 8129^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8130 8131Syntax: 8132""""""" 8133 8134This is an overloaded intrinsic. You can use ``llvm.log2`` on any 8135floating point or vector of floating point type. Not all targets support 8136all types however. 8137 8138:: 8139 8140 declare float @llvm.log2.f32(float %Val) 8141 declare double @llvm.log2.f64(double %Val) 8142 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val) 8143 declare fp128 @llvm.log2.f128(fp128 %Val) 8144 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val) 8145 8146Overview: 8147""""""""" 8148 8149The '``llvm.log2.*``' intrinsics perform the log2 function. 8150 8151Arguments: 8152"""""""""" 8153 8154The argument and return value are floating point numbers of the same 8155type. 8156 8157Semantics: 8158"""""""""" 8159 8160This function returns the same values as the libm ``log2`` functions 8161would, and handles error conditions in the same way. 8162 8163'``llvm.fma.*``' Intrinsic 8164^^^^^^^^^^^^^^^^^^^^^^^^^^ 8165 8166Syntax: 8167""""""" 8168 8169This is an overloaded intrinsic. You can use ``llvm.fma`` on any 8170floating point or vector of floating point type. Not all targets support 8171all types however. 8172 8173:: 8174 8175 declare float @llvm.fma.f32(float %a, float %b, float %c) 8176 declare double @llvm.fma.f64(double %a, double %b, double %c) 8177 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c) 8178 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c) 8179 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c) 8180 8181Overview: 8182""""""""" 8183 8184The '``llvm.fma.*``' intrinsics perform the fused multiply-add 8185operation. 8186 8187Arguments: 8188"""""""""" 8189 8190The argument and return value are floating point numbers of the same 8191type. 8192 8193Semantics: 8194"""""""""" 8195 8196This function returns the same values as the libm ``fma`` functions 8197would, and does not set errno. 8198 8199'``llvm.fabs.*``' Intrinsic 8200^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8201 8202Syntax: 8203""""""" 8204 8205This is an overloaded intrinsic. You can use ``llvm.fabs`` on any 8206floating point or vector of floating point type. Not all targets support 8207all types however. 8208 8209:: 8210 8211 declare float @llvm.fabs.f32(float %Val) 8212 declare double @llvm.fabs.f64(double %Val) 8213 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val) 8214 declare fp128 @llvm.fabs.f128(fp128 %Val) 8215 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val) 8216 8217Overview: 8218""""""""" 8219 8220The '``llvm.fabs.*``' intrinsics return the absolute value of the 8221operand. 8222 8223Arguments: 8224"""""""""" 8225 8226The argument and return value are floating point numbers of the same 8227type. 8228 8229Semantics: 8230"""""""""" 8231 8232This function returns the same values as the libm ``fabs`` functions 8233would, and handles error conditions in the same way. 8234 8235'``llvm.minnum.*``' Intrinsic 8236^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8237 8238Syntax: 8239""""""" 8240 8241This is an overloaded intrinsic. You can use ``llvm.minnum`` on any 8242floating point or vector of floating point type. Not all targets support 8243all types however. 8244 8245:: 8246 8247 declare float @llvm.minnum.f32(float %Val0, float %Val1) 8248 declare double @llvm.minnum.f64(double %Val0, double %Val1) 8249 declare x86_fp80 @llvm.minnum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 8250 declare fp128 @llvm.minnum.f128(fp128 %Val0, fp128 %Val1) 8251 declare ppc_fp128 @llvm.minnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 8252 8253Overview: 8254""""""""" 8255 8256The '``llvm.minnum.*``' intrinsics return the minimum of the two 8257arguments. 8258 8259 8260Arguments: 8261"""""""""" 8262 8263The arguments and return value are floating point numbers of the same 8264type. 8265 8266Semantics: 8267"""""""""" 8268 8269Follows the IEEE-754 semantics for minNum, which also match for libm's 8270fmin. 8271 8272If either operand is a NaN, returns the other non-NaN operand. Returns 8273NaN only if both operands are NaN. If the operands compare equal, 8274returns a value that compares equal to both operands. This means that 8275fmin(+/-0.0, +/-0.0) could return either -0.0 or 0.0. 8276 8277'``llvm.maxnum.*``' Intrinsic 8278^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8279 8280Syntax: 8281""""""" 8282 8283This is an overloaded intrinsic. You can use ``llvm.maxnum`` on any 8284floating point or vector of floating point type. Not all targets support 8285all types however. 8286 8287:: 8288 8289 declare float @llvm.maxnum.f32(float %Val0, float %Val1l) 8290 declare double @llvm.maxnum.f64(double %Val0, double %Val1) 8291 declare x86_fp80 @llvm.maxnum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 8292 declare fp128 @llvm.maxnum.f128(fp128 %Val0, fp128 %Val1) 8293 declare ppc_fp128 @llvm.maxnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 8294 8295Overview: 8296""""""""" 8297 8298The '``llvm.maxnum.*``' intrinsics return the maximum of the two 8299arguments. 8300 8301 8302Arguments: 8303"""""""""" 8304 8305The arguments and return value are floating point numbers of the same 8306type. 8307 8308Semantics: 8309"""""""""" 8310Follows the IEEE-754 semantics for maxNum, which also match for libm's 8311fmax. 8312 8313If either operand is a NaN, returns the other non-NaN operand. Returns 8314NaN only if both operands are NaN. If the operands compare equal, 8315returns a value that compares equal to both operands. This means that 8316fmax(+/-0.0, +/-0.0) could return either -0.0 or 0.0. 8317 8318'``llvm.copysign.*``' Intrinsic 8319^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8320 8321Syntax: 8322""""""" 8323 8324This is an overloaded intrinsic. You can use ``llvm.copysign`` on any 8325floating point or vector of floating point type. Not all targets support 8326all types however. 8327 8328:: 8329 8330 declare float @llvm.copysign.f32(float %Mag, float %Sgn) 8331 declare double @llvm.copysign.f64(double %Mag, double %Sgn) 8332 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn) 8333 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn) 8334 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn) 8335 8336Overview: 8337""""""""" 8338 8339The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the 8340first operand and the sign of the second operand. 8341 8342Arguments: 8343"""""""""" 8344 8345The arguments and return value are floating point numbers of the same 8346type. 8347 8348Semantics: 8349"""""""""" 8350 8351This function returns the same values as the libm ``copysign`` 8352functions would, and handles error conditions in the same way. 8353 8354'``llvm.floor.*``' Intrinsic 8355^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8356 8357Syntax: 8358""""""" 8359 8360This is an overloaded intrinsic. You can use ``llvm.floor`` on any 8361floating point or vector of floating point type. Not all targets support 8362all types however. 8363 8364:: 8365 8366 declare float @llvm.floor.f32(float %Val) 8367 declare double @llvm.floor.f64(double %Val) 8368 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val) 8369 declare fp128 @llvm.floor.f128(fp128 %Val) 8370 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val) 8371 8372Overview: 8373""""""""" 8374 8375The '``llvm.floor.*``' intrinsics return the floor of the operand. 8376 8377Arguments: 8378"""""""""" 8379 8380The argument and return value are floating point numbers of the same 8381type. 8382 8383Semantics: 8384"""""""""" 8385 8386This function returns the same values as the libm ``floor`` functions 8387would, and handles error conditions in the same way. 8388 8389'``llvm.ceil.*``' Intrinsic 8390^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8391 8392Syntax: 8393""""""" 8394 8395This is an overloaded intrinsic. You can use ``llvm.ceil`` on any 8396floating point or vector of floating point type. Not all targets support 8397all types however. 8398 8399:: 8400 8401 declare float @llvm.ceil.f32(float %Val) 8402 declare double @llvm.ceil.f64(double %Val) 8403 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val) 8404 declare fp128 @llvm.ceil.f128(fp128 %Val) 8405 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val) 8406 8407Overview: 8408""""""""" 8409 8410The '``llvm.ceil.*``' intrinsics return the ceiling of the operand. 8411 8412Arguments: 8413"""""""""" 8414 8415The argument and return value are floating point numbers of the same 8416type. 8417 8418Semantics: 8419"""""""""" 8420 8421This function returns the same values as the libm ``ceil`` functions 8422would, and handles error conditions in the same way. 8423 8424'``llvm.trunc.*``' Intrinsic 8425^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8426 8427Syntax: 8428""""""" 8429 8430This is an overloaded intrinsic. You can use ``llvm.trunc`` on any 8431floating point or vector of floating point type. Not all targets support 8432all types however. 8433 8434:: 8435 8436 declare float @llvm.trunc.f32(float %Val) 8437 declare double @llvm.trunc.f64(double %Val) 8438 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val) 8439 declare fp128 @llvm.trunc.f128(fp128 %Val) 8440 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val) 8441 8442Overview: 8443""""""""" 8444 8445The '``llvm.trunc.*``' intrinsics returns the operand rounded to the 8446nearest integer not larger in magnitude than the operand. 8447 8448Arguments: 8449"""""""""" 8450 8451The argument and return value are floating point numbers of the same 8452type. 8453 8454Semantics: 8455"""""""""" 8456 8457This function returns the same values as the libm ``trunc`` functions 8458would, and handles error conditions in the same way. 8459 8460'``llvm.rint.*``' Intrinsic 8461^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8462 8463Syntax: 8464""""""" 8465 8466This is an overloaded intrinsic. You can use ``llvm.rint`` on any 8467floating point or vector of floating point type. Not all targets support 8468all types however. 8469 8470:: 8471 8472 declare float @llvm.rint.f32(float %Val) 8473 declare double @llvm.rint.f64(double %Val) 8474 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val) 8475 declare fp128 @llvm.rint.f128(fp128 %Val) 8476 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val) 8477 8478Overview: 8479""""""""" 8480 8481The '``llvm.rint.*``' intrinsics returns the operand rounded to the 8482nearest integer. It may raise an inexact floating-point exception if the 8483operand isn't an integer. 8484 8485Arguments: 8486"""""""""" 8487 8488The argument and return value are floating point numbers of the same 8489type. 8490 8491Semantics: 8492"""""""""" 8493 8494This function returns the same values as the libm ``rint`` functions 8495would, and handles error conditions in the same way. 8496 8497'``llvm.nearbyint.*``' Intrinsic 8498^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8499 8500Syntax: 8501""""""" 8502 8503This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any 8504floating point or vector of floating point type. Not all targets support 8505all types however. 8506 8507:: 8508 8509 declare float @llvm.nearbyint.f32(float %Val) 8510 declare double @llvm.nearbyint.f64(double %Val) 8511 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val) 8512 declare fp128 @llvm.nearbyint.f128(fp128 %Val) 8513 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val) 8514 8515Overview: 8516""""""""" 8517 8518The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the 8519nearest integer. 8520 8521Arguments: 8522"""""""""" 8523 8524The argument and return value are floating point numbers of the same 8525type. 8526 8527Semantics: 8528"""""""""" 8529 8530This function returns the same values as the libm ``nearbyint`` 8531functions would, and handles error conditions in the same way. 8532 8533'``llvm.round.*``' Intrinsic 8534^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8535 8536Syntax: 8537""""""" 8538 8539This is an overloaded intrinsic. You can use ``llvm.round`` on any 8540floating point or vector of floating point type. Not all targets support 8541all types however. 8542 8543:: 8544 8545 declare float @llvm.round.f32(float %Val) 8546 declare double @llvm.round.f64(double %Val) 8547 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val) 8548 declare fp128 @llvm.round.f128(fp128 %Val) 8549 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val) 8550 8551Overview: 8552""""""""" 8553 8554The '``llvm.round.*``' intrinsics returns the operand rounded to the 8555nearest integer. 8556 8557Arguments: 8558"""""""""" 8559 8560The argument and return value are floating point numbers of the same 8561type. 8562 8563Semantics: 8564"""""""""" 8565 8566This function returns the same values as the libm ``round`` 8567functions would, and handles error conditions in the same way. 8568 8569Bit Manipulation Intrinsics 8570--------------------------- 8571 8572LLVM provides intrinsics for a few important bit manipulation 8573operations. These allow efficient code generation for some algorithms. 8574 8575'``llvm.bswap.*``' Intrinsics 8576^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8577 8578Syntax: 8579""""""" 8580 8581This is an overloaded intrinsic function. You can use bswap on any 8582integer type that is an even number of bytes (i.e. BitWidth % 16 == 0). 8583 8584:: 8585 8586 declare i16 @llvm.bswap.i16(i16 <id>) 8587 declare i32 @llvm.bswap.i32(i32 <id>) 8588 declare i64 @llvm.bswap.i64(i64 <id>) 8589 8590Overview: 8591""""""""" 8592 8593The '``llvm.bswap``' family of intrinsics is used to byte swap integer 8594values with an even number of bytes (positive multiple of 16 bits). 8595These are useful for performing operations on data that is not in the 8596target's native byte order. 8597 8598Semantics: 8599"""""""""" 8600 8601The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high 8602and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32`` 8603intrinsic returns an i32 value that has the four bytes of the input i32 8604swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the 8605returned i32 will have its bytes in 3, 2, 1, 0 order. The 8606``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this 8607concept to additional even-byte lengths (6 bytes, 8 bytes and more, 8608respectively). 8609 8610'``llvm.ctpop.*``' Intrinsic 8611^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8612 8613Syntax: 8614""""""" 8615 8616This is an overloaded intrinsic. You can use llvm.ctpop on any integer 8617bit width, or on any vector with integer elements. Not all targets 8618support all bit widths or vector types, however. 8619 8620:: 8621 8622 declare i8 @llvm.ctpop.i8(i8 <src>) 8623 declare i16 @llvm.ctpop.i16(i16 <src>) 8624 declare i32 @llvm.ctpop.i32(i32 <src>) 8625 declare i64 @llvm.ctpop.i64(i64 <src>) 8626 declare i256 @llvm.ctpop.i256(i256 <src>) 8627 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>) 8628 8629Overview: 8630""""""""" 8631 8632The '``llvm.ctpop``' family of intrinsics counts the number of bits set 8633in a value. 8634 8635Arguments: 8636"""""""""" 8637 8638The only argument is the value to be counted. The argument may be of any 8639integer type, or a vector with integer elements. The return type must 8640match the argument type. 8641 8642Semantics: 8643"""""""""" 8644 8645The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within 8646each element of a vector. 8647 8648'``llvm.ctlz.*``' Intrinsic 8649^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8650 8651Syntax: 8652""""""" 8653 8654This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any 8655integer bit width, or any vector whose elements are integers. Not all 8656targets support all bit widths or vector types, however. 8657 8658:: 8659 8660 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>) 8661 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>) 8662 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>) 8663 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>) 8664 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>) 8665 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>) 8666 8667Overview: 8668""""""""" 8669 8670The '``llvm.ctlz``' family of intrinsic functions counts the number of 8671leading zeros in a variable. 8672 8673Arguments: 8674"""""""""" 8675 8676The first argument is the value to be counted. This argument may be of 8677any integer type, or a vector with integer element type. The return 8678type must match the first argument type. 8679 8680The second argument must be a constant and is a flag to indicate whether 8681the intrinsic should ensure that a zero as the first argument produces a 8682defined result. Historically some architectures did not provide a 8683defined result for zero values as efficiently, and many algorithms are 8684now predicated on avoiding zero-value inputs. 8685 8686Semantics: 8687"""""""""" 8688 8689The '``llvm.ctlz``' intrinsic counts the leading (most significant) 8690zeros in a variable, or within each element of the vector. If 8691``src == 0`` then the result is the size in bits of the type of ``src`` 8692if ``is_zero_undef == 0`` and ``undef`` otherwise. For example, 8693``llvm.ctlz(i32 2) = 30``. 8694 8695'``llvm.cttz.*``' Intrinsic 8696^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8697 8698Syntax: 8699""""""" 8700 8701This is an overloaded intrinsic. You can use ``llvm.cttz`` on any 8702integer bit width, or any vector of integer elements. Not all targets 8703support all bit widths or vector types, however. 8704 8705:: 8706 8707 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>) 8708 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>) 8709 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>) 8710 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>) 8711 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>) 8712 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>) 8713 8714Overview: 8715""""""""" 8716 8717The '``llvm.cttz``' family of intrinsic functions counts the number of 8718trailing zeros. 8719 8720Arguments: 8721"""""""""" 8722 8723The first argument is the value to be counted. This argument may be of 8724any integer type, or a vector with integer element type. The return 8725type must match the first argument type. 8726 8727The second argument must be a constant and is a flag to indicate whether 8728the intrinsic should ensure that a zero as the first argument produces a 8729defined result. Historically some architectures did not provide a 8730defined result for zero values as efficiently, and many algorithms are 8731now predicated on avoiding zero-value inputs. 8732 8733Semantics: 8734"""""""""" 8735 8736The '``llvm.cttz``' intrinsic counts the trailing (least significant) 8737zeros in a variable, or within each element of a vector. If ``src == 0`` 8738then the result is the size in bits of the type of ``src`` if 8739``is_zero_undef == 0`` and ``undef`` otherwise. For example, 8740``llvm.cttz(2) = 1``. 8741 8742Arithmetic with Overflow Intrinsics 8743----------------------------------- 8744 8745LLVM provides intrinsics for some arithmetic with overflow operations. 8746 8747'``llvm.sadd.with.overflow.*``' Intrinsics 8748^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8749 8750Syntax: 8751""""""" 8752 8753This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow`` 8754on any integer bit width. 8755 8756:: 8757 8758 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b) 8759 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) 8760 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b) 8761 8762Overview: 8763""""""""" 8764 8765The '``llvm.sadd.with.overflow``' family of intrinsic functions perform 8766a signed addition of the two arguments, and indicate whether an overflow 8767occurred during the signed summation. 8768 8769Arguments: 8770"""""""""" 8771 8772The arguments (%a and %b) and the first element of the result structure 8773may be of integer types of any bit width, but they must have the same 8774bit width. The second element of the result structure must be of type 8775``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 8776addition. 8777 8778Semantics: 8779"""""""""" 8780 8781The '``llvm.sadd.with.overflow``' family of intrinsic functions perform 8782a signed addition of the two variables. They return a structure --- the 8783first element of which is the signed summation, and the second element 8784of which is a bit specifying if the signed summation resulted in an 8785overflow. 8786 8787Examples: 8788""""""""" 8789 8790.. code-block:: llvm 8791 8792 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) 8793 %sum = extractvalue {i32, i1} %res, 0 8794 %obit = extractvalue {i32, i1} %res, 1 8795 br i1 %obit, label %overflow, label %normal 8796 8797'``llvm.uadd.with.overflow.*``' Intrinsics 8798^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8799 8800Syntax: 8801""""""" 8802 8803This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow`` 8804on any integer bit width. 8805 8806:: 8807 8808 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b) 8809 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b) 8810 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b) 8811 8812Overview: 8813""""""""" 8814 8815The '``llvm.uadd.with.overflow``' family of intrinsic functions perform 8816an unsigned addition of the two arguments, and indicate whether a carry 8817occurred during the unsigned summation. 8818 8819Arguments: 8820"""""""""" 8821 8822The arguments (%a and %b) and the first element of the result structure 8823may be of integer types of any bit width, but they must have the same 8824bit width. The second element of the result structure must be of type 8825``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 8826addition. 8827 8828Semantics: 8829"""""""""" 8830 8831The '``llvm.uadd.with.overflow``' family of intrinsic functions perform 8832an unsigned addition of the two arguments. They return a structure --- the 8833first element of which is the sum, and the second element of which is a 8834bit specifying if the unsigned summation resulted in a carry. 8835 8836Examples: 8837""""""""" 8838 8839.. code-block:: llvm 8840 8841 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b) 8842 %sum = extractvalue {i32, i1} %res, 0 8843 %obit = extractvalue {i32, i1} %res, 1 8844 br i1 %obit, label %carry, label %normal 8845 8846'``llvm.ssub.with.overflow.*``' Intrinsics 8847^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8848 8849Syntax: 8850""""""" 8851 8852This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow`` 8853on any integer bit width. 8854 8855:: 8856 8857 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b) 8858 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b) 8859 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b) 8860 8861Overview: 8862""""""""" 8863 8864The '``llvm.ssub.with.overflow``' family of intrinsic functions perform 8865a signed subtraction of the two arguments, and indicate whether an 8866overflow occurred during the signed subtraction. 8867 8868Arguments: 8869"""""""""" 8870 8871The arguments (%a and %b) and the first element of the result structure 8872may be of integer types of any bit width, but they must have the same 8873bit width. The second element of the result structure must be of type 8874``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 8875subtraction. 8876 8877Semantics: 8878"""""""""" 8879 8880The '``llvm.ssub.with.overflow``' family of intrinsic functions perform 8881a signed subtraction of the two arguments. They return a structure --- the 8882first element of which is the subtraction, and the second element of 8883which is a bit specifying if the signed subtraction resulted in an 8884overflow. 8885 8886Examples: 8887""""""""" 8888 8889.. code-block:: llvm 8890 8891 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b) 8892 %sum = extractvalue {i32, i1} %res, 0 8893 %obit = extractvalue {i32, i1} %res, 1 8894 br i1 %obit, label %overflow, label %normal 8895 8896'``llvm.usub.with.overflow.*``' Intrinsics 8897^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8898 8899Syntax: 8900""""""" 8901 8902This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow`` 8903on any integer bit width. 8904 8905:: 8906 8907 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b) 8908 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b) 8909 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b) 8910 8911Overview: 8912""""""""" 8913 8914The '``llvm.usub.with.overflow``' family of intrinsic functions perform 8915an unsigned subtraction of the two arguments, and indicate whether an 8916overflow occurred during the unsigned subtraction. 8917 8918Arguments: 8919"""""""""" 8920 8921The arguments (%a and %b) and the first element of the result structure 8922may be of integer types of any bit width, but they must have the same 8923bit width. The second element of the result structure must be of type 8924``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 8925subtraction. 8926 8927Semantics: 8928"""""""""" 8929 8930The '``llvm.usub.with.overflow``' family of intrinsic functions perform 8931an unsigned subtraction of the two arguments. They return a structure --- 8932the first element of which is the subtraction, and the second element of 8933which is a bit specifying if the unsigned subtraction resulted in an 8934overflow. 8935 8936Examples: 8937""""""""" 8938 8939.. code-block:: llvm 8940 8941 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b) 8942 %sum = extractvalue {i32, i1} %res, 0 8943 %obit = extractvalue {i32, i1} %res, 1 8944 br i1 %obit, label %overflow, label %normal 8945 8946'``llvm.smul.with.overflow.*``' Intrinsics 8947^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8948 8949Syntax: 8950""""""" 8951 8952This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow`` 8953on any integer bit width. 8954 8955:: 8956 8957 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b) 8958 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b) 8959 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b) 8960 8961Overview: 8962""""""""" 8963 8964The '``llvm.smul.with.overflow``' family of intrinsic functions perform 8965a signed multiplication of the two arguments, and indicate whether an 8966overflow occurred during the signed multiplication. 8967 8968Arguments: 8969"""""""""" 8970 8971The arguments (%a and %b) and the first element of the result structure 8972may be of integer types of any bit width, but they must have the same 8973bit width. The second element of the result structure must be of type 8974``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 8975multiplication. 8976 8977Semantics: 8978"""""""""" 8979 8980The '``llvm.smul.with.overflow``' family of intrinsic functions perform 8981a signed multiplication of the two arguments. They return a structure --- 8982the first element of which is the multiplication, and the second element 8983of which is a bit specifying if the signed multiplication resulted in an 8984overflow. 8985 8986Examples: 8987""""""""" 8988 8989.. code-block:: llvm 8990 8991 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b) 8992 %sum = extractvalue {i32, i1} %res, 0 8993 %obit = extractvalue {i32, i1} %res, 1 8994 br i1 %obit, label %overflow, label %normal 8995 8996'``llvm.umul.with.overflow.*``' Intrinsics 8997^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8998 8999Syntax: 9000""""""" 9001 9002This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow`` 9003on any integer bit width. 9004 9005:: 9006 9007 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b) 9008 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b) 9009 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b) 9010 9011Overview: 9012""""""""" 9013 9014The '``llvm.umul.with.overflow``' family of intrinsic functions perform 9015a unsigned multiplication of the two arguments, and indicate whether an 9016overflow occurred during the unsigned multiplication. 9017 9018Arguments: 9019"""""""""" 9020 9021The arguments (%a and %b) and the first element of the result structure 9022may be of integer types of any bit width, but they must have the same 9023bit width. The second element of the result structure must be of type 9024``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 9025multiplication. 9026 9027Semantics: 9028"""""""""" 9029 9030The '``llvm.umul.with.overflow``' family of intrinsic functions perform 9031an unsigned multiplication of the two arguments. They return a structure --- 9032the first element of which is the multiplication, and the second 9033element of which is a bit specifying if the unsigned multiplication 9034resulted in an overflow. 9035 9036Examples: 9037""""""""" 9038 9039.. code-block:: llvm 9040 9041 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b) 9042 %sum = extractvalue {i32, i1} %res, 0 9043 %obit = extractvalue {i32, i1} %res, 1 9044 br i1 %obit, label %overflow, label %normal 9045 9046Specialised Arithmetic Intrinsics 9047--------------------------------- 9048 9049'``llvm.fmuladd.*``' Intrinsic 9050^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9051 9052Syntax: 9053""""""" 9054 9055:: 9056 9057 declare float @llvm.fmuladd.f32(float %a, float %b, float %c) 9058 declare double @llvm.fmuladd.f64(double %a, double %b, double %c) 9059 9060Overview: 9061""""""""" 9062 9063The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add 9064expressions that can be fused if the code generator determines that (a) the 9065target instruction set has support for a fused operation, and (b) that the 9066fused operation is more efficient than the equivalent, separate pair of mul 9067and add instructions. 9068 9069Arguments: 9070"""""""""" 9071 9072The '``llvm.fmuladd.*``' intrinsics each take three arguments: two 9073multiplicands, a and b, and an addend c. 9074 9075Semantics: 9076"""""""""" 9077 9078The expression: 9079 9080:: 9081 9082 %0 = call float @llvm.fmuladd.f32(%a, %b, %c) 9083 9084is equivalent to the expression a \* b + c, except that rounding will 9085not be performed between the multiplication and addition steps if the 9086code generator fuses the operations. Fusion is not guaranteed, even if 9087the target platform supports it. If a fused multiply-add is required the 9088corresponding llvm.fma.\* intrinsic function should be used 9089instead. This never sets errno, just as '``llvm.fma.*``'. 9090 9091Examples: 9092""""""""" 9093 9094.. code-block:: llvm 9095 9096 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields float:r2 = (a * b) + c 9097 9098Half Precision Floating Point Intrinsics 9099---------------------------------------- 9100 9101For most target platforms, half precision floating point is a 9102storage-only format. This means that it is a dense encoding (in memory) 9103but does not support computation in the format. 9104 9105This means that code must first load the half-precision floating point 9106value as an i16, then convert it to float with 9107:ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can 9108then be performed on the float value (including extending to double 9109etc). To store the value back to memory, it is first converted to float 9110if needed, then converted to i16 with 9111:ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an 9112i16 value. 9113 9114.. _int_convert_to_fp16: 9115 9116'``llvm.convert.to.fp16``' Intrinsic 9117^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9118 9119Syntax: 9120""""""" 9121 9122:: 9123 9124 declare i16 @llvm.convert.to.fp16.f32(float %a) 9125 declare i16 @llvm.convert.to.fp16.f64(double %a) 9126 9127Overview: 9128""""""""" 9129 9130The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a 9131conventional floating point type to half precision floating point format. 9132 9133Arguments: 9134"""""""""" 9135 9136The intrinsic function contains single argument - the value to be 9137converted. 9138 9139Semantics: 9140"""""""""" 9141 9142The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a 9143conventional floating point format to half precision floating point format. The 9144return value is an ``i16`` which contains the converted number. 9145 9146Examples: 9147""""""""" 9148 9149.. code-block:: llvm 9150 9151 %res = call i16 @llvm.convert.to.fp16.f32(float %a) 9152 store i16 %res, i16* @x, align 2 9153 9154.. _int_convert_from_fp16: 9155 9156'``llvm.convert.from.fp16``' Intrinsic 9157^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9158 9159Syntax: 9160""""""" 9161 9162:: 9163 9164 declare float @llvm.convert.from.fp16.f32(i16 %a) 9165 declare double @llvm.convert.from.fp16.f64(i16 %a) 9166 9167Overview: 9168""""""""" 9169 9170The '``llvm.convert.from.fp16``' intrinsic function performs a 9171conversion from half precision floating point format to single precision 9172floating point format. 9173 9174Arguments: 9175"""""""""" 9176 9177The intrinsic function contains single argument - the value to be 9178converted. 9179 9180Semantics: 9181"""""""""" 9182 9183The '``llvm.convert.from.fp16``' intrinsic function performs a 9184conversion from half single precision floating point format to single 9185precision floating point format. The input half-float value is 9186represented by an ``i16`` value. 9187 9188Examples: 9189""""""""" 9190 9191.. code-block:: llvm 9192 9193 %a = load i16* @x, align 2 9194 %res = call float @llvm.convert.from.fp16(i16 %a) 9195 9196Debugger Intrinsics 9197------------------- 9198 9199The LLVM debugger intrinsics (which all start with ``llvm.dbg.`` 9200prefix), are described in the `LLVM Source Level 9201Debugging <SourceLevelDebugging.html#format_common_intrinsics>`_ 9202document. 9203 9204Exception Handling Intrinsics 9205----------------------------- 9206 9207The LLVM exception handling intrinsics (which all start with 9208``llvm.eh.`` prefix), are described in the `LLVM Exception 9209Handling <ExceptionHandling.html#format_common_intrinsics>`_ document. 9210 9211.. _int_trampoline: 9212 9213Trampoline Intrinsics 9214--------------------- 9215 9216These intrinsics make it possible to excise one parameter, marked with 9217the :ref:`nest <nest>` attribute, from a function. The result is a 9218callable function pointer lacking the nest parameter - the caller does 9219not need to provide a value for it. Instead, the value to use is stored 9220in advance in a "trampoline", a block of memory usually allocated on the 9221stack, which also contains code to splice the nest value into the 9222argument list. This is used to implement the GCC nested function address 9223extension. 9224 9225For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)`` 9226then the resulting function pointer has signature ``i32 (i32, i32)*``. 9227It can be created as follows: 9228 9229.. code-block:: llvm 9230 9231 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86 9232 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0 9233 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval) 9234 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1) 9235 %fp = bitcast i8* %p to i32 (i32, i32)* 9236 9237The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to 9238``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``. 9239 9240.. _int_it: 9241 9242'``llvm.init.trampoline``' Intrinsic 9243^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9244 9245Syntax: 9246""""""" 9247 9248:: 9249 9250 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>) 9251 9252Overview: 9253""""""""" 9254 9255This fills the memory pointed to by ``tramp`` with executable code, 9256turning it into a trampoline. 9257 9258Arguments: 9259"""""""""" 9260 9261The ``llvm.init.trampoline`` intrinsic takes three arguments, all 9262pointers. The ``tramp`` argument must point to a sufficiently large and 9263sufficiently aligned block of memory; this memory is written to by the 9264intrinsic. Note that the size and the alignment are target-specific - 9265LLVM currently provides no portable way of determining them, so a 9266front-end that generates this intrinsic needs to have some 9267target-specific knowledge. The ``func`` argument must hold a function 9268bitcast to an ``i8*``. 9269 9270Semantics: 9271"""""""""" 9272 9273The block of memory pointed to by ``tramp`` is filled with target 9274dependent code, turning it into a function. Then ``tramp`` needs to be 9275passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can 9276be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new 9277function's signature is the same as that of ``func`` with any arguments 9278marked with the ``nest`` attribute removed. At most one such ``nest`` 9279argument is allowed, and it must be of pointer type. Calling the new 9280function is equivalent to calling ``func`` with the same argument list, 9281but with ``nval`` used for the missing ``nest`` argument. If, after 9282calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is 9283modified, then the effect of any later call to the returned function 9284pointer is undefined. 9285 9286.. _int_at: 9287 9288'``llvm.adjust.trampoline``' Intrinsic 9289^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9290 9291Syntax: 9292""""""" 9293 9294:: 9295 9296 declare i8* @llvm.adjust.trampoline(i8* <tramp>) 9297 9298Overview: 9299""""""""" 9300 9301This performs any required machine-specific adjustment to the address of 9302a trampoline (passed as ``tramp``). 9303 9304Arguments: 9305"""""""""" 9306 9307``tramp`` must point to a block of memory which already has trampoline 9308code filled in by a previous call to 9309:ref:`llvm.init.trampoline <int_it>`. 9310 9311Semantics: 9312"""""""""" 9313 9314On some architectures the address of the code to be executed needs to be 9315different than the address where the trampoline is actually stored. This 9316intrinsic returns the executable address corresponding to ``tramp`` 9317after performing the required machine specific adjustments. The pointer 9318returned can then be :ref:`bitcast and executed <int_trampoline>`. 9319 9320Masked Vector Load and Store Intrinsics 9321--------------------------------------- 9322 9323LLVM 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. 9324 9325.. _int_mload: 9326 9327'``llvm.masked.load.*``' Intrinsics 9328^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9329 9330Syntax: 9331""""""" 9332This is an overloaded intrinsic. The loaded data is a vector of any integer or floating point data type. 9333 9334:: 9335 9336 declare <16 x float> @llvm.masked.load.v16f32 (<16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>) 9337 declare <2 x double> @llvm.masked.load.v2f64 (<2 x double>* <ptr>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>) 9338 9339Overview: 9340""""""""" 9341 9342Reads 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. 9343 9344 9345Arguments: 9346"""""""""" 9347 9348The 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. 9349 9350 9351Semantics: 9352"""""""""" 9353 9354The '``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. 9355The 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. 9356 9357 9358:: 9359 9360 %res = call <16 x float> @llvm.masked.load.v16f32 (<16 x float>* %ptr, i32 4, <16 x i1>%mask, <16 x float> %passthru) 9361 9362 ;; The result of the two following instructions is identical aside from potential memory access exception 9363 %loadlal = load <16 x float>* %ptr, align 4 9364 %res = select <16 x i1> %mask, <16 x float> %loadlal, <16 x float> %passthru 9365 9366.. _int_mstore: 9367 9368'``llvm.masked.store.*``' Intrinsics 9369^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9370 9371Syntax: 9372""""""" 9373This is an overloaded intrinsic. The data stored in memory is a vector of any integer or floating point data type. 9374 9375:: 9376 9377 declare void @llvm.masked.store.v8i32 (<8 x i32> <value>, <8 x i32> * <ptr>, i32 <alignment>, <8 x i1> <mask>) 9378 declare void @llvm.masked.store.v16f32(<16 x i32> <value>, <16 x i32>* <ptr>, i32 <alignment>, <16 x i1> <mask>) 9379 9380Overview: 9381""""""""" 9382 9383Writes 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. 9384 9385Arguments: 9386"""""""""" 9387 9388The 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. 9389 9390 9391Semantics: 9392"""""""""" 9393 9394The '``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. 9395The 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. 9396 9397:: 9398 9399 call void @llvm.masked.store.v16f32(<16 x float> %value, <16 x float>* %ptr, i32 4, <16 x i1> %mask) 9400 9401 ;; The result of the following instructions is identical aside from potential data races and memory access exceptions 9402 %oldval = load <16 x float>* %ptr, align 4 9403 %res = select <16 x i1> %mask, <16 x float> %value, <16 x float> %oldval 9404 store <16 x float> %res, <16 x float>* %ptr, align 4 9405 9406 9407Memory Use Markers 9408------------------ 9409 9410This class of intrinsics provides information about the lifetime of 9411memory objects and ranges where variables are immutable. 9412 9413.. _int_lifestart: 9414 9415'``llvm.lifetime.start``' Intrinsic 9416^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9417 9418Syntax: 9419""""""" 9420 9421:: 9422 9423 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>) 9424 9425Overview: 9426""""""""" 9427 9428The '``llvm.lifetime.start``' intrinsic specifies the start of a memory 9429object's lifetime. 9430 9431Arguments: 9432"""""""""" 9433 9434The first argument is a constant integer representing the size of the 9435object, or -1 if it is variable sized. The second argument is a pointer 9436to the object. 9437 9438Semantics: 9439"""""""""" 9440 9441This intrinsic indicates that before this point in the code, the value 9442of the memory pointed to by ``ptr`` is dead. This means that it is known 9443to never be used and has an undefined value. A load from the pointer 9444that precedes this intrinsic can be replaced with ``'undef'``. 9445 9446.. _int_lifeend: 9447 9448'``llvm.lifetime.end``' Intrinsic 9449^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9450 9451Syntax: 9452""""""" 9453 9454:: 9455 9456 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>) 9457 9458Overview: 9459""""""""" 9460 9461The '``llvm.lifetime.end``' intrinsic specifies the end of a memory 9462object's lifetime. 9463 9464Arguments: 9465"""""""""" 9466 9467The first argument is a constant integer representing the size of the 9468object, or -1 if it is variable sized. The second argument is a pointer 9469to the object. 9470 9471Semantics: 9472"""""""""" 9473 9474This intrinsic indicates that after this point in the code, the value of 9475the memory pointed to by ``ptr`` is dead. This means that it is known to 9476never be used and has an undefined value. Any stores into the memory 9477object following this intrinsic may be removed as dead. 9478 9479'``llvm.invariant.start``' Intrinsic 9480^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9481 9482Syntax: 9483""""""" 9484 9485:: 9486 9487 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) 9488 9489Overview: 9490""""""""" 9491 9492The '``llvm.invariant.start``' intrinsic specifies that the contents of 9493a memory object will not change. 9494 9495Arguments: 9496"""""""""" 9497 9498The first argument is a constant integer representing the size of the 9499object, or -1 if it is variable sized. The second argument is a pointer 9500to the object. 9501 9502Semantics: 9503"""""""""" 9504 9505This intrinsic indicates that until an ``llvm.invariant.end`` that uses 9506the return value, the referenced memory location is constant and 9507unchanging. 9508 9509'``llvm.invariant.end``' Intrinsic 9510^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9511 9512Syntax: 9513""""""" 9514 9515:: 9516 9517 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>) 9518 9519Overview: 9520""""""""" 9521 9522The '``llvm.invariant.end``' intrinsic specifies that the contents of a 9523memory object are mutable. 9524 9525Arguments: 9526"""""""""" 9527 9528The first argument is the matching ``llvm.invariant.start`` intrinsic. 9529The second argument is a constant integer representing the size of the 9530object, or -1 if it is variable sized and the third argument is a 9531pointer to the object. 9532 9533Semantics: 9534"""""""""" 9535 9536This intrinsic indicates that the memory is mutable again. 9537 9538General Intrinsics 9539------------------ 9540 9541This class of intrinsics is designed to be generic and has no specific 9542purpose. 9543 9544'``llvm.var.annotation``' Intrinsic 9545^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9546 9547Syntax: 9548""""""" 9549 9550:: 9551 9552 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>) 9553 9554Overview: 9555""""""""" 9556 9557The '``llvm.var.annotation``' intrinsic. 9558 9559Arguments: 9560"""""""""" 9561 9562The first argument is a pointer to a value, the second is a pointer to a 9563global string, the third is a pointer to a global string which is the 9564source file name, and the last argument is the line number. 9565 9566Semantics: 9567"""""""""" 9568 9569This intrinsic allows annotation of local variables with arbitrary 9570strings. This can be useful for special purpose optimizations that want 9571to look for these annotations. These have no other defined use; they are 9572ignored by code generation and optimization. 9573 9574'``llvm.ptr.annotation.*``' Intrinsic 9575^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9576 9577Syntax: 9578""""""" 9579 9580This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a 9581pointer to an integer of any width. *NOTE* you must specify an address space for 9582the pointer. The identifier for the default address space is the integer 9583'``0``'. 9584 9585:: 9586 9587 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>) 9588 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>) 9589 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>) 9590 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>) 9591 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>) 9592 9593Overview: 9594""""""""" 9595 9596The '``llvm.ptr.annotation``' intrinsic. 9597 9598Arguments: 9599"""""""""" 9600 9601The first argument is a pointer to an integer value of arbitrary bitwidth 9602(result of some expression), the second is a pointer to a global string, the 9603third is a pointer to a global string which is the source file name, and the 9604last argument is the line number. It returns the value of the first argument. 9605 9606Semantics: 9607"""""""""" 9608 9609This intrinsic allows annotation of a pointer to an integer with arbitrary 9610strings. This can be useful for special purpose optimizations that want to look 9611for these annotations. These have no other defined use; they are ignored by code 9612generation and optimization. 9613 9614'``llvm.annotation.*``' Intrinsic 9615^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9616 9617Syntax: 9618""""""" 9619 9620This is an overloaded intrinsic. You can use '``llvm.annotation``' on 9621any integer bit width. 9622 9623:: 9624 9625 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>) 9626 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>) 9627 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>) 9628 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>) 9629 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>) 9630 9631Overview: 9632""""""""" 9633 9634The '``llvm.annotation``' intrinsic. 9635 9636Arguments: 9637"""""""""" 9638 9639The first argument is an integer value (result of some expression), the 9640second is a pointer to a global string, the third is a pointer to a 9641global string which is the source file name, and the last argument is 9642the line number. It returns the value of the first argument. 9643 9644Semantics: 9645"""""""""" 9646 9647This intrinsic allows annotations to be put on arbitrary expressions 9648with arbitrary strings. This can be useful for special purpose 9649optimizations that want to look for these annotations. These have no 9650other defined use; they are ignored by code generation and optimization. 9651 9652'``llvm.trap``' Intrinsic 9653^^^^^^^^^^^^^^^^^^^^^^^^^ 9654 9655Syntax: 9656""""""" 9657 9658:: 9659 9660 declare void @llvm.trap() noreturn nounwind 9661 9662Overview: 9663""""""""" 9664 9665The '``llvm.trap``' intrinsic. 9666 9667Arguments: 9668"""""""""" 9669 9670None. 9671 9672Semantics: 9673"""""""""" 9674 9675This intrinsic is lowered to the target dependent trap instruction. If 9676the target does not have a trap instruction, this intrinsic will be 9677lowered to a call of the ``abort()`` function. 9678 9679'``llvm.debugtrap``' Intrinsic 9680^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9681 9682Syntax: 9683""""""" 9684 9685:: 9686 9687 declare void @llvm.debugtrap() nounwind 9688 9689Overview: 9690""""""""" 9691 9692The '``llvm.debugtrap``' intrinsic. 9693 9694Arguments: 9695"""""""""" 9696 9697None. 9698 9699Semantics: 9700"""""""""" 9701 9702This intrinsic is lowered to code which is intended to cause an 9703execution trap with the intention of requesting the attention of a 9704debugger. 9705 9706'``llvm.stackprotector``' Intrinsic 9707^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9708 9709Syntax: 9710""""""" 9711 9712:: 9713 9714 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>) 9715 9716Overview: 9717""""""""" 9718 9719The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it 9720onto the stack at ``slot``. The stack slot is adjusted to ensure that it 9721is placed on the stack before local variables. 9722 9723Arguments: 9724"""""""""" 9725 9726The ``llvm.stackprotector`` intrinsic requires two pointer arguments. 9727The first argument is the value loaded from the stack guard 9728``@__stack_chk_guard``. The second variable is an ``alloca`` that has 9729enough space to hold the value of the guard. 9730 9731Semantics: 9732"""""""""" 9733 9734This intrinsic causes the prologue/epilogue inserter to force the position of 9735the ``AllocaInst`` stack slot to be before local variables on the stack. This is 9736to ensure that if a local variable on the stack is overwritten, it will destroy 9737the value of the guard. When the function exits, the guard on the stack is 9738checked against the original guard by ``llvm.stackprotectorcheck``. If they are 9739different, then ``llvm.stackprotectorcheck`` causes the program to abort by 9740calling the ``__stack_chk_fail()`` function. 9741 9742'``llvm.stackprotectorcheck``' Intrinsic 9743^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9744 9745Syntax: 9746""""""" 9747 9748:: 9749 9750 declare void @llvm.stackprotectorcheck(i8** <guard>) 9751 9752Overview: 9753""""""""" 9754 9755The ``llvm.stackprotectorcheck`` intrinsic compares ``guard`` against an already 9756created stack protector and if they are not equal calls the 9757``__stack_chk_fail()`` function. 9758 9759Arguments: 9760"""""""""" 9761 9762The ``llvm.stackprotectorcheck`` intrinsic requires one pointer argument, the 9763the variable ``@__stack_chk_guard``. 9764 9765Semantics: 9766"""""""""" 9767 9768This intrinsic is provided to perform the stack protector check by comparing 9769``guard`` with the stack slot created by ``llvm.stackprotector`` and if the 9770values do not match call the ``__stack_chk_fail()`` function. 9771 9772The reason to provide this as an IR level intrinsic instead of implementing it 9773via other IR operations is that in order to perform this operation at the IR 9774level without an intrinsic, one would need to create additional basic blocks to 9775handle the success/failure cases. This makes it difficult to stop the stack 9776protector check from disrupting sibling tail calls in Codegen. With this 9777intrinsic, we are able to generate the stack protector basic blocks late in 9778codegen after the tail call decision has occurred. 9779 9780'``llvm.objectsize``' Intrinsic 9781^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9782 9783Syntax: 9784""""""" 9785 9786:: 9787 9788 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>) 9789 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>) 9790 9791Overview: 9792""""""""" 9793 9794The ``llvm.objectsize`` intrinsic is designed to provide information to 9795the optimizers to determine at compile time whether a) an operation 9796(like memcpy) will overflow a buffer that corresponds to an object, or 9797b) that a runtime check for overflow isn't necessary. An object in this 9798context means an allocation of a specific class, structure, array, or 9799other object. 9800 9801Arguments: 9802"""""""""" 9803 9804The ``llvm.objectsize`` intrinsic takes two arguments. The first 9805argument is a pointer to or into the ``object``. The second argument is 9806a boolean and determines whether ``llvm.objectsize`` returns 0 (if true) 9807or -1 (if false) when the object size is unknown. The second argument 9808only accepts constants. 9809 9810Semantics: 9811"""""""""" 9812 9813The ``llvm.objectsize`` intrinsic is lowered to a constant representing 9814the size of the object concerned. If the size cannot be determined at 9815compile time, ``llvm.objectsize`` returns ``i32/i64 -1 or 0`` (depending 9816on the ``min`` argument). 9817 9818'``llvm.expect``' Intrinsic 9819^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9820 9821Syntax: 9822""""""" 9823 9824This is an overloaded intrinsic. You can use ``llvm.expect`` on any 9825integer bit width. 9826 9827:: 9828 9829 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>) 9830 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>) 9831 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>) 9832 9833Overview: 9834""""""""" 9835 9836The ``llvm.expect`` intrinsic provides information about expected (the 9837most probable) value of ``val``, which can be used by optimizers. 9838 9839Arguments: 9840"""""""""" 9841 9842The ``llvm.expect`` intrinsic takes two arguments. The first argument is 9843a value. The second argument is an expected value, this needs to be a 9844constant value, variables are not allowed. 9845 9846Semantics: 9847"""""""""" 9848 9849This intrinsic is lowered to the ``val``. 9850 9851'``llvm.assume``' Intrinsic 9852^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9853 9854Syntax: 9855""""""" 9856 9857:: 9858 9859 declare void @llvm.assume(i1 %cond) 9860 9861Overview: 9862""""""""" 9863 9864The ``llvm.assume`` allows the optimizer to assume that the provided 9865condition is true. This information can then be used in simplifying other parts 9866of the code. 9867 9868Arguments: 9869"""""""""" 9870 9871The condition which the optimizer may assume is always true. 9872 9873Semantics: 9874"""""""""" 9875 9876The intrinsic allows the optimizer to assume that the provided condition is 9877always true whenever the control flow reaches the intrinsic call. No code is 9878generated for this intrinsic, and instructions that contribute only to the 9879provided condition are not used for code generation. If the condition is 9880violated during execution, the behavior is undefined. 9881 9882Note that the optimizer might limit the transformations performed on values 9883used by the ``llvm.assume`` intrinsic in order to preserve the instructions 9884only used to form the intrinsic's input argument. This might prove undesirable 9885if the extra information provided by the ``llvm.assume`` intrinsic does not cause 9886sufficient overall improvement in code quality. For this reason, 9887``llvm.assume`` should not be used to document basic mathematical invariants 9888that the optimizer can otherwise deduce or facts that are of little use to the 9889optimizer. 9890 9891'``llvm.donothing``' Intrinsic 9892^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9893 9894Syntax: 9895""""""" 9896 9897:: 9898 9899 declare void @llvm.donothing() nounwind readnone 9900 9901Overview: 9902""""""""" 9903 9904The ``llvm.donothing`` intrinsic doesn't perform any operation. It's one of only 9905two intrinsics (besides ``llvm.experimental.patchpoint``) that can be called 9906with an invoke instruction. 9907 9908Arguments: 9909"""""""""" 9910 9911None. 9912 9913Semantics: 9914"""""""""" 9915 9916This intrinsic does nothing, and it's removed by optimizers and ignored 9917by codegen. 9918 9919Stack Map Intrinsics 9920-------------------- 9921 9922LLVM provides experimental intrinsics to support runtime patching 9923mechanisms commonly desired in dynamic language JITs. These intrinsics 9924are described in :doc:`StackMaps`. 9925