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 values 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], [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 a 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 into 208 the object file corresponding to the LLVM module. From the linker's 209 perspective, an ``available_externally`` global is equivalent to 210 an external declaration. They exist to allow inlining and other 211 optimizations to take place given knowledge of the definition of the 212 global, which is known to be somewhere outside the module. Globals 213 with ``available_externally`` linkage are allowed to be discarded at 214 will, and allow inlining and other optimizations. This linkage type is 215 only allowed on definitions, not declarations. 216``linkonce`` 217 Globals with "``linkonce``" linkage are merged with other globals of 218 the same name when linkage occurs. This can be used to implement 219 some forms of inline functions, templates, or other code which must 220 be generated in each translation unit that uses it, but where the 221 body may be overridden with a more definitive definition later. 222 Unreferenced ``linkonce`` globals are allowed to be discarded. Note 223 that ``linkonce`` linkage does not actually allow the optimizer to 224 inline the body of this function into callers because it doesn't 225 know if this definition of the function is the definitive definition 226 within the program or whether it will be overridden by a stronger 227 definition. To enable inlining and other optimizations, use 228 "``linkonce_odr``" linkage. 229``weak`` 230 "``weak``" linkage has the same merging semantics as ``linkonce`` 231 linkage, except that unreferenced globals with ``weak`` linkage may 232 not be discarded. This is used for globals that are declared "weak" 233 in C source code. 234``common`` 235 "``common``" linkage is most similar to "``weak``" linkage, but they 236 are used for tentative definitions in C, such as "``int X;``" at 237 global scope. Symbols with "``common``" linkage are merged in the 238 same way as ``weak symbols``, and they may not be deleted if 239 unreferenced. ``common`` symbols may not have an explicit section, 240 must have a zero initializer, and may not be marked 241 ':ref:`constant <globalvars>`'. Functions and aliases may not have 242 common linkage. 243 244.. _linkage_appending: 245 246``appending`` 247 "``appending``" linkage may only be applied to global variables of 248 pointer to array type. When two global variables with appending 249 linkage are linked together, the two global arrays are appended 250 together. This is the LLVM, typesafe, equivalent of having the 251 system linker append together "sections" with identical names when 252 .o files are linked. 253 254 Unfortunately this doesn't correspond to any feature in .o files, so it 255 can only be used for variables like ``llvm.global_ctors`` which llvm 256 interprets specially. 257 258``extern_weak`` 259 The semantics of this linkage follow the ELF object file model: the 260 symbol is weak until linked, if not linked, the symbol becomes null 261 instead of being an undefined reference. 262``linkonce_odr``, ``weak_odr`` 263 Some languages allow differing globals to be merged, such as two 264 functions with different semantics. Other languages, such as 265 ``C++``, ensure that only equivalent globals are ever merged (the 266 "one definition rule" --- "ODR"). Such languages can use the 267 ``linkonce_odr`` and ``weak_odr`` linkage types to indicate that the 268 global will only be merged with equivalent globals. These linkage 269 types are otherwise the same as their non-``odr`` versions. 270``external`` 271 If none of the above identifiers are used, the global is externally 272 visible, meaning that it participates in linkage and can be used to 273 resolve external symbol references. 274 275It is illegal for a global variable or function *declaration* to have any 276linkage type other than ``external`` or ``extern_weak``. 277 278.. _callingconv: 279 280Calling Conventions 281------------------- 282 283LLVM :ref:`functions <functionstructure>`, :ref:`calls <i_call>` and 284:ref:`invokes <i_invoke>` can all have an optional calling convention 285specified for the call. The calling convention of any pair of dynamic 286caller/callee must match, or the behavior of the program is undefined. 287The following calling conventions are supported by LLVM, and more may be 288added in the future: 289 290"``ccc``" - The C calling convention 291 This calling convention (the default if no other calling convention 292 is specified) matches the target C calling conventions. This calling 293 convention supports varargs function calls and tolerates some 294 mismatch in the declared prototype and implemented declaration of 295 the function (as does normal C). 296"``fastcc``" - The fast calling convention 297 This calling convention attempts to make calls as fast as possible 298 (e.g. by passing things in registers). This calling convention 299 allows the target to use whatever tricks it wants to produce fast 300 code for the target, without having to conform to an externally 301 specified ABI (Application Binary Interface). `Tail calls can only 302 be optimized when this, the tailcc, the GHC or the HiPE convention is 303 used. <CodeGenerator.html#tail-call-optimization>`_ This calling 304 convention does not support varargs and requires the prototype of all 305 callees to exactly match the prototype of the function definition. 306"``coldcc``" - The cold calling convention 307 This calling convention attempts to make code in the caller as 308 efficient as possible under the assumption that the call is not 309 commonly executed. As such, these calls often preserve all registers 310 so that the call does not break any live ranges in the caller side. 311 This calling convention does not support varargs and requires the 312 prototype of all callees to exactly match the prototype of the 313 function definition. Furthermore the inliner doesn't consider such function 314 calls for inlining. 315"``cc 10``" - GHC convention 316 This calling convention has been implemented specifically for use by 317 the `Glasgow Haskell Compiler (GHC) <http://www.haskell.org/ghc>`_. 318 It passes everything in registers, going to extremes to achieve this 319 by disabling callee save registers. This calling convention should 320 not be used lightly but only for specific situations such as an 321 alternative to the *register pinning* performance technique often 322 used when implementing functional programming languages. At the 323 moment only X86 supports this convention and it has the following 324 limitations: 325 326 - On *X86-32* only supports up to 4 bit type parameters. No 327 floating-point types are supported. 328 - On *X86-64* only supports up to 10 bit type parameters and 6 329 floating-point parameters. 330 331 This calling convention supports `tail call 332 optimization <CodeGenerator.html#tail-call-optimization>`_ but requires 333 both the caller and callee are using it. 334"``cc 11``" - The HiPE calling convention 335 This calling convention has been implemented specifically for use by 336 the `High-Performance Erlang 337 (HiPE) <http://www.it.uu.se/research/group/hipe/>`_ compiler, *the* 338 native code compiler of the `Ericsson's Open Source Erlang/OTP 339 system <http://www.erlang.org/download.shtml>`_. It uses more 340 registers for argument passing than the ordinary C calling 341 convention and defines no callee-saved registers. The calling 342 convention properly supports `tail call 343 optimization <CodeGenerator.html#tail-call-optimization>`_ but requires 344 that both the caller and the callee use it. It uses a *register pinning* 345 mechanism, similar to GHC's convention, for keeping frequently 346 accessed runtime components pinned to specific hardware registers. 347 At the moment only X86 supports this convention (both 32 and 64 348 bit). 349"``webkit_jscc``" - WebKit's JavaScript calling convention 350 This calling convention has been implemented for `WebKit FTL JIT 351 <https://trac.webkit.org/wiki/FTLJIT>`_. It passes arguments on the 352 stack right to left (as cdecl does), and returns a value in the 353 platform's customary return register. 354"``anyregcc``" - Dynamic calling convention for code patching 355 This is a special convention that supports patching an arbitrary code 356 sequence in place of a call site. This convention forces the call 357 arguments into registers but allows them to be dynamically 358 allocated. This can currently only be used with calls to 359 llvm.experimental.patchpoint because only this intrinsic records 360 the location of its arguments in a side table. See :doc:`StackMaps`. 361"``preserve_mostcc``" - The `PreserveMost` calling convention 362 This calling convention attempts to make the code in the caller as 363 unintrusive as possible. This convention behaves identically to the `C` 364 calling convention on how arguments and return values are passed, but it 365 uses a different set of caller/callee-saved registers. This alleviates the 366 burden of saving and recovering a large register set before and after the 367 call in the caller. If the arguments are passed in callee-saved registers, 368 then they will be preserved by the callee across the call. This doesn't 369 apply for values returned in callee-saved registers. 370 371 - On X86-64 the callee preserves all general purpose registers, except for 372 R11. R11 can be used as a scratch register. Floating-point registers 373 (XMMs/YMMs) are not preserved and need to be saved by the caller. 374 375 The idea behind this convention is to support calls to runtime functions 376 that have a hot path and a cold path. The hot path is usually a small piece 377 of code that doesn't use many registers. The cold path might need to call out to 378 another function and therefore only needs to preserve the caller-saved 379 registers, which haven't already been saved by the caller. The 380 `PreserveMost` calling convention is very similar to the `cold` calling 381 convention in terms of caller/callee-saved registers, but they are used for 382 different types of function calls. `coldcc` is for function calls that are 383 rarely executed, whereas `preserve_mostcc` function calls are intended to be 384 on the hot path and definitely executed a lot. Furthermore `preserve_mostcc` 385 doesn't prevent the inliner from inlining the function call. 386 387 This calling convention will be used by a future version of the ObjectiveC 388 runtime and should therefore still be considered experimental at this time. 389 Although this convention was created to optimize certain runtime calls to 390 the ObjectiveC runtime, it is not limited to this runtime and might be used 391 by other runtimes in the future too. The current implementation only 392 supports X86-64, but the intention is to support more architectures in the 393 future. 394"``preserve_allcc``" - The `PreserveAll` calling convention 395 This calling convention attempts to make the code in the caller even less 396 intrusive than the `PreserveMost` calling convention. This calling 397 convention also behaves identical to the `C` calling convention on how 398 arguments and return values are passed, but it uses a different set of 399 caller/callee-saved registers. This removes the burden of saving and 400 recovering a large register set before and after the call in the caller. If 401 the arguments are passed in callee-saved registers, then they will be 402 preserved by the callee across the call. This doesn't apply for values 403 returned in callee-saved registers. 404 405 - On X86-64 the callee preserves all general purpose registers, except for 406 R11. R11 can be used as a scratch register. Furthermore it also preserves 407 all floating-point registers (XMMs/YMMs). 408 409 The idea behind this convention is to support calls to runtime functions 410 that don't need to call out to any other functions. 411 412 This calling convention, like the `PreserveMost` calling convention, will be 413 used by a future version of the ObjectiveC runtime and should be considered 414 experimental at this time. 415"``cxx_fast_tlscc``" - The `CXX_FAST_TLS` calling convention for access functions 416 Clang generates an access function to access C++-style TLS. The access 417 function generally has an entry block, an exit block and an initialization 418 block that is run at the first time. The entry and exit blocks can access 419 a few TLS IR variables, each access will be lowered to a platform-specific 420 sequence. 421 422 This calling convention aims to minimize overhead in the caller by 423 preserving as many registers as possible (all the registers that are 424 preserved on the fast path, composed of the entry and exit blocks). 425 426 This calling convention behaves identical to the `C` calling convention on 427 how arguments and return values are passed, but it uses a different set of 428 caller/callee-saved registers. 429 430 Given that each platform has its own lowering sequence, hence its own set 431 of preserved registers, we can't use the existing `PreserveMost`. 432 433 - On X86-64 the callee preserves all general purpose registers, except for 434 RDI and RAX. 435"``tailcc``" - Tail callable calling convention 436 This calling convention ensures that calls in tail position will always be 437 tail call optimized. This calling convention is equivalent to fastcc, 438 except for an additional guarantee that tail calls will be produced 439 whenever possible. `Tail calls can only be optimized when this, the fastcc, 440 the GHC or the HiPE convention is used. <CodeGenerator.html#tail-call-optimization>`_ 441 This calling convention does not support varargs and requires the prototype of 442 all callees to exactly match the prototype of the function definition. 443"``swiftcc``" - This calling convention is used for Swift language. 444 - On X86-64 RCX and R8 are available for additional integer returns, and 445 XMM2 and XMM3 are available for additional FP/vector returns. 446 - On iOS platforms, we use AAPCS-VFP calling convention. 447"``swifttailcc``" 448 This calling convention is like ``swiftcc`` in most respects, but also the 449 callee pops the argument area of the stack so that mandatory tail calls are 450 possible as in ``tailcc``. 451"``cfguard_checkcc``" - Windows Control Flow Guard (Check mechanism) 452 This calling convention is used for the Control Flow Guard check function, 453 calls to which can be inserted before indirect calls to check that the call 454 target is a valid function address. The check function has no return value, 455 but it will trigger an OS-level error if the address is not a valid target. 456 The set of registers preserved by the check function, and the register 457 containing the target address are architecture-specific. 458 459 - On X86 the target address is passed in ECX. 460 - On ARM the target address is passed in R0. 461 - On AArch64 the target address is passed in X15. 462"``cc <n>``" - Numbered convention 463 Any calling convention may be specified by number, allowing 464 target-specific calling conventions to be used. Target specific 465 calling conventions start at 64. 466 467More calling conventions can be added/defined on an as-needed basis, to 468support Pascal conventions or any other well-known target-independent 469convention. 470 471.. _visibilitystyles: 472 473Visibility Styles 474----------------- 475 476All Global Variables and Functions have one of the following visibility 477styles: 478 479"``default``" - Default style 480 On targets that use the ELF object file format, default visibility 481 means that the declaration is visible to other modules and, in 482 shared libraries, means that the declared entity may be overridden. 483 On Darwin, default visibility means that the declaration is visible 484 to other modules. On XCOFF, default visibility means no explicit 485 visibility bit will be set and whether the symbol is visible 486 (i.e "exported") to other modules depends primarily on export lists 487 provided to the linker. Default visibility corresponds to "external 488 linkage" in the language. 489"``hidden``" - Hidden style 490 Two declarations of an object with hidden visibility refer to the 491 same object if they are in the same shared object. Usually, hidden 492 visibility indicates that the symbol will not be placed into the 493 dynamic symbol table, so no other module (executable or shared 494 library) can reference it directly. 495"``protected``" - Protected style 496 On ELF, protected visibility indicates that the symbol will be 497 placed in the dynamic symbol table, but that references within the 498 defining module will bind to the local symbol. That is, the symbol 499 cannot be overridden by another module. 500 501A symbol with ``internal`` or ``private`` linkage must have ``default`` 502visibility. 503 504.. _dllstorageclass: 505 506DLL Storage Classes 507------------------- 508 509All Global Variables, Functions and Aliases can have one of the following 510DLL storage class: 511 512``dllimport`` 513 "``dllimport``" causes the compiler to reference a function or variable via 514 a global pointer to a pointer that is set up by the DLL exporting the 515 symbol. On Microsoft Windows targets, the pointer name is formed by 516 combining ``__imp_`` and the function or variable name. 517``dllexport`` 518 On Microsoft Windows targets, "``dllexport``" causes the compiler to provide 519 a global pointer to a pointer in a DLL, so that it can be referenced with the 520 ``dllimport`` attribute. the pointer name is formed by combining ``__imp_`` 521 and the function or variable name. On XCOFF targets, ``dllexport`` indicates 522 that the symbol will be made visible to other modules using "exported" 523 visibility and thus placed by the linker in the loader section symbol table. 524 Since this storage class exists for defining a dll interface, the compiler, 525 assembler and linker know it is externally referenced and must refrain from 526 deleting the symbol. 527 528.. _tls_model: 529 530Thread Local Storage Models 531--------------------------- 532 533A variable may be defined as ``thread_local``, which means that it will 534not be shared by threads (each thread will have a separated copy of the 535variable). Not all targets support thread-local variables. Optionally, a 536TLS model may be specified: 537 538``localdynamic`` 539 For variables that are only used within the current shared library. 540``initialexec`` 541 For variables in modules that will not be loaded dynamically. 542``localexec`` 543 For variables defined in the executable and only used within it. 544 545If no explicit model is given, the "general dynamic" model is used. 546 547The models correspond to the ELF TLS models; see `ELF Handling For 548Thread-Local Storage <http://people.redhat.com/drepper/tls.pdf>`_ for 549more information on under which circumstances the different models may 550be used. The target may choose a different TLS model if the specified 551model is not supported, or if a better choice of model can be made. 552 553A model can also be specified in an alias, but then it only governs how 554the alias is accessed. It will not have any effect in the aliasee. 555 556For platforms without linker support of ELF TLS model, the -femulated-tls 557flag can be used to generate GCC compatible emulated TLS code. 558 559.. _runtime_preemption_model: 560 561Runtime Preemption Specifiers 562----------------------------- 563 564Global variables, functions and aliases may have an optional runtime preemption 565specifier. If a preemption specifier isn't given explicitly, then a 566symbol is assumed to be ``dso_preemptable``. 567 568``dso_preemptable`` 569 Indicates that the function or variable may be replaced by a symbol from 570 outside the linkage unit at runtime. 571 572``dso_local`` 573 The compiler may assume that a function or variable marked as ``dso_local`` 574 will resolve to a symbol within the same linkage unit. Direct access will 575 be generated even if the definition is not within this compilation unit. 576 577.. _namedtypes: 578 579Structure Types 580--------------- 581 582LLVM IR allows you to specify both "identified" and "literal" :ref:`structure 583types <t_struct>`. Literal types are uniqued structurally, but identified types 584are never uniqued. An :ref:`opaque structural type <t_opaque>` can also be used 585to forward declare a type that is not yet available. 586 587An example of an identified structure specification is: 588 589.. code-block:: llvm 590 591 %mytype = type { %mytype*, i32 } 592 593Prior to the LLVM 3.0 release, identified types were structurally uniqued. Only 594literal types are uniqued in recent versions of LLVM. 595 596.. _nointptrtype: 597 598Non-Integral Pointer Type 599------------------------- 600 601Note: non-integral pointer types are a work in progress, and they should be 602considered experimental at this time. 603 604LLVM IR optionally allows the frontend to denote pointers in certain address 605spaces as "non-integral" via the :ref:`datalayout string<langref_datalayout>`. 606Non-integral pointer types represent pointers that have an *unspecified* bitwise 607representation; that is, the integral representation may be target dependent or 608unstable (not backed by a fixed integer). 609 610``inttoptr`` and ``ptrtoint`` instructions have the same semantics as for 611integral (i.e. normal) pointers in that they convert integers to and from 612corresponding pointer types, but there are additional implications to be 613aware of. Because the bit-representation of a non-integral pointer may 614not be stable, two identical casts of the same operand may or may not 615return the same value. Said differently, the conversion to or from the 616non-integral type depends on environmental state in an implementation 617defined manner. 618 619If the frontend wishes to observe a *particular* value following a cast, the 620generated IR must fence with the underlying environment in an implementation 621defined manner. (In practice, this tends to require ``noinline`` routines for 622such operations.) 623 624From the perspective of the optimizer, ``inttoptr`` and ``ptrtoint`` for 625non-integral types are analogous to ones on integral types with one 626key exception: the optimizer may not, in general, insert new dynamic 627occurrences of such casts. If a new cast is inserted, the optimizer would 628need to either ensure that a) all possible values are valid, or b) 629appropriate fencing is inserted. Since the appropriate fencing is 630implementation defined, the optimizer can't do the latter. The former is 631challenging as many commonly expected properties, such as 632``ptrtoint(v)-ptrtoint(v) == 0``, don't hold for non-integral types. 633 634.. _globalvars: 635 636Global Variables 637---------------- 638 639Global variables define regions of memory allocated at compilation time 640instead of run-time. 641 642Global variable definitions must be initialized. 643 644Global variables in other translation units can also be declared, in which 645case they don't have an initializer. 646 647Global variables can optionally specify a :ref:`linkage type <linkage>`. 648 649Either global variable definitions or declarations may have an explicit section 650to be placed in and may have an optional explicit alignment specified. If there 651is a mismatch between the explicit or inferred section information for the 652variable declaration and its definition the resulting behavior is undefined. 653 654A variable may be defined as a global ``constant``, which indicates that 655the contents of the variable will **never** be modified (enabling better 656optimization, allowing the global data to be placed in the read-only 657section of an executable, etc). Note that variables that need runtime 658initialization cannot be marked ``constant`` as there is a store to the 659variable. 660 661LLVM explicitly allows *declarations* of global variables to be marked 662constant, even if the final definition of the global is not. This 663capability can be used to enable slightly better optimization of the 664program, but requires the language definition to guarantee that 665optimizations based on the 'constantness' are valid for the translation 666units that do not include the definition. 667 668As SSA values, global variables define pointer values that are in scope 669(i.e. they dominate) all basic blocks in the program. Global variables 670always define a pointer to their "content" type because they describe a 671region of memory, and all memory objects in LLVM are accessed through 672pointers. 673 674Global variables can be marked with ``unnamed_addr`` which indicates 675that the address is not significant, only the content. Constants marked 676like this can be merged with other constants if they have the same 677initializer. Note that a constant with significant address *can* be 678merged with a ``unnamed_addr`` constant, the result being a constant 679whose address is significant. 680 681If the ``local_unnamed_addr`` attribute is given, the address is known to 682not be significant within the module. 683 684A global variable may be declared to reside in a target-specific 685numbered address space. For targets that support them, address spaces 686may affect how optimizations are performed and/or what target 687instructions are used to access the variable. The default address space 688is zero. The address space qualifier must precede any other attributes. 689 690LLVM allows an explicit section to be specified for globals. If the 691target supports it, it will emit globals to the section specified. 692Additionally, the global can placed in a comdat if the target has the necessary 693support. 694 695External declarations may have an explicit section specified. Section 696information is retained in LLVM IR for targets that make use of this 697information. Attaching section information to an external declaration is an 698assertion that its definition is located in the specified section. If the 699definition is located in a different section, the behavior is undefined. 700 701By default, global initializers are optimized by assuming that global 702variables defined within the module are not modified from their 703initial values before the start of the global initializer. This is 704true even for variables potentially accessible from outside the 705module, including those with external linkage or appearing in 706``@llvm.used`` or dllexported variables. This assumption may be suppressed 707by marking the variable with ``externally_initialized``. 708 709An explicit alignment may be specified for a global, which must be a 710power of 2. If not present, or if the alignment is set to zero, the 711alignment of the global is set by the target to whatever it feels 712convenient. If an explicit alignment is specified, the global is forced 713to have exactly that alignment. Targets and optimizers are not allowed 714to over-align the global if the global has an assigned section. In this 715case, the extra alignment could be observable: for example, code could 716assume that the globals are densely packed in their section and try to 717iterate over them as an array, alignment padding would break this 718iteration. The maximum alignment is ``1 << 32``. 719 720For global variables declarations, as well as definitions that may be 721replaced at link time (``linkonce``, ``weak``, ``extern_weak`` and ``common`` 722linkage types), LLVM makes no assumptions about the allocation size of the 723variables, except that they may not overlap. The alignment of a global variable 724declaration or replaceable definition must not be greater than the alignment of 725the definition it resolves to. 726 727Globals can also have a :ref:`DLL storage class <dllstorageclass>`, 728an optional :ref:`runtime preemption specifier <runtime_preemption_model>`, 729an optional :ref:`global attributes <glattrs>` and 730an optional list of attached :ref:`metadata <metadata>`. 731 732Variables and aliases can have a 733:ref:`Thread Local Storage Model <tls_model>`. 734 735:ref:`Scalable vectors <t_vector>` cannot be global variables or members of 736arrays because their size is unknown at compile time. They are allowed in 737structs to facilitate intrinsics returning multiple values. Structs containing 738scalable vectors cannot be used in loads, stores, allocas, or GEPs. 739 740Syntax:: 741 742 @<GlobalVarName> = [Linkage] [PreemptionSpecifier] [Visibility] 743 [DLLStorageClass] [ThreadLocal] 744 [(unnamed_addr|local_unnamed_addr)] [AddrSpace] 745 [ExternallyInitialized] 746 <global | constant> <Type> [<InitializerConstant>] 747 [, section "name"] [, partition "name"] 748 [, comdat [($name)]] [, align <Alignment>] 749 [, no_sanitize_address] [, no_sanitize_hwaddress] 750 [, sanitize_address_dyninit] [, sanitize_memtag] 751 (, !name !N)* 752 753For example, the following defines a global in a numbered address space 754with an initializer, section, and alignment: 755 756.. code-block:: llvm 757 758 @G = addrspace(5) constant float 1.0, section "foo", align 4 759 760The following example just declares a global variable 761 762.. code-block:: llvm 763 764 @G = external global i32 765 766The following example defines a thread-local global with the 767``initialexec`` TLS model: 768 769.. code-block:: llvm 770 771 @G = thread_local(initialexec) global i32 0, align 4 772 773.. _functionstructure: 774 775Functions 776--------- 777 778LLVM function definitions consist of the "``define``" keyword, an 779optional :ref:`linkage type <linkage>`, an optional :ref:`runtime preemption 780specifier <runtime_preemption_model>`, an optional :ref:`visibility 781style <visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`, 782an optional :ref:`calling convention <callingconv>`, 783an optional ``unnamed_addr`` attribute, a return type, an optional 784:ref:`parameter attribute <paramattrs>` for the return type, a function 785name, a (possibly empty) argument list (each with optional :ref:`parameter 786attributes <paramattrs>`), optional :ref:`function attributes <fnattrs>`, 787an optional address space, an optional section, an optional partition, 788an optional alignment, an optional :ref:`comdat <langref_comdats>`, 789an optional :ref:`garbage collector name <gc>`, an optional :ref:`prefix <prefixdata>`, 790an optional :ref:`prologue <prologuedata>`, 791an optional :ref:`personality <personalityfn>`, 792an optional list of attached :ref:`metadata <metadata>`, 793an opening curly brace, a list of basic blocks, and a closing curly brace. 794 795Syntax:: 796 797 define [linkage] [PreemptionSpecifier] [visibility] [DLLStorageClass] 798 [cconv] [ret attrs] 799 <ResultType> @<FunctionName> ([argument list]) 800 [(unnamed_addr|local_unnamed_addr)] [AddrSpace] [fn Attrs] 801 [section "name"] [partition "name"] [comdat [($name)]] [align N] 802 [gc] [prefix Constant] [prologue Constant] [personality Constant] 803 (!name !N)* { ... } 804 805The argument list is a comma separated sequence of arguments where each 806argument is of the following form: 807 808Syntax:: 809 810 <type> [parameter Attrs] [name] 811 812LLVM function declarations consist of the "``declare``" keyword, an 813optional :ref:`linkage type <linkage>`, an optional :ref:`visibility style 814<visibility>`, an optional :ref:`DLL storage class <dllstorageclass>`, an 815optional :ref:`calling convention <callingconv>`, an optional ``unnamed_addr`` 816or ``local_unnamed_addr`` attribute, an optional address space, a return type, 817an optional :ref:`parameter attribute <paramattrs>` for the return type, a function name, a possibly 818empty list of arguments, an optional alignment, an optional :ref:`garbage 819collector name <gc>`, an optional :ref:`prefix <prefixdata>`, and an optional 820:ref:`prologue <prologuedata>`. 821 822Syntax:: 823 824 declare [linkage] [visibility] [DLLStorageClass] 825 [cconv] [ret attrs] 826 <ResultType> @<FunctionName> ([argument list]) 827 [(unnamed_addr|local_unnamed_addr)] [align N] [gc] 828 [prefix Constant] [prologue Constant] 829 830A function definition contains a list of basic blocks, forming the CFG (Control 831Flow Graph) for the function. Each basic block may optionally start with a label 832(giving the basic block a symbol table entry), contains a list of instructions, 833and ends with a :ref:`terminator <terminators>` instruction (such as a branch or 834function return). If an explicit label name is not provided, a block is assigned 835an implicit numbered label, using the next value from the same counter as used 836for unnamed temporaries (:ref:`see above<identifiers>`). For example, if a 837function entry block does not have an explicit label, it will be assigned label 838"%0", then the first unnamed temporary in that block will be "%1", etc. If a 839numeric label is explicitly specified, it must match the numeric label that 840would be used implicitly. 841 842The first basic block in a function is special in two ways: it is 843immediately executed on entrance to the function, and it is not allowed 844to have predecessor basic blocks (i.e. there can not be any branches to 845the entry block of a function). Because the block can have no 846predecessors, it also cannot have any :ref:`PHI nodes <i_phi>`. 847 848LLVM allows an explicit section to be specified for functions. If the 849target supports it, it will emit functions to the section specified. 850Additionally, the function can be placed in a COMDAT. 851 852An explicit alignment may be specified for a function. If not present, 853or if the alignment is set to zero, the alignment of the function is set 854by the target to whatever it feels convenient. If an explicit alignment 855is specified, the function is forced to have at least that much 856alignment. All alignments must be a power of 2. 857 858If the ``unnamed_addr`` attribute is given, the address is known to not 859be significant and two identical functions can be merged. 860 861If the ``local_unnamed_addr`` attribute is given, the address is known to 862not be significant within the module. 863 864If an explicit address space is not given, it will default to the program 865address space from the :ref:`datalayout string<langref_datalayout>`. 866 867.. _langref_aliases: 868 869Aliases 870------- 871 872Aliases, unlike function or variables, don't create any new data. They 873are just a new symbol and metadata for an existing position. 874 875Aliases have a name and an aliasee that is either a global value or a 876constant expression. 877 878Aliases may have an optional :ref:`linkage type <linkage>`, an optional 879:ref:`runtime preemption specifier <runtime_preemption_model>`, an optional 880:ref:`visibility style <visibility>`, an optional :ref:`DLL storage class 881<dllstorageclass>` and an optional :ref:`tls model <tls_model>`. 882 883Syntax:: 884 885 @<Name> = [Linkage] [PreemptionSpecifier] [Visibility] [DLLStorageClass] [ThreadLocal] [(unnamed_addr|local_unnamed_addr)] alias <AliaseeTy>, <AliaseeTy>* @<Aliasee> 886 [, partition "name"] 887 888The linkage must be one of ``private``, ``internal``, ``linkonce``, ``weak``, 889``linkonce_odr``, ``weak_odr``, ``external``. Note that some system linkers 890might not correctly handle dropping a weak symbol that is aliased. 891 892Aliases that are not ``unnamed_addr`` are guaranteed to have the same address as 893the aliasee expression. ``unnamed_addr`` ones are only guaranteed to point 894to the same content. 895 896If the ``local_unnamed_addr`` attribute is given, the address is known to 897not be significant within the module. 898 899Since aliases are only a second name, some restrictions apply, of which 900some can only be checked when producing an object file: 901 902* The expression defining the aliasee must be computable at assembly 903 time. Since it is just a name, no relocations can be used. 904 905* No alias in the expression can be weak as the possibility of the 906 intermediate alias being overridden cannot be represented in an 907 object file. 908 909* No global value in the expression can be a declaration, since that 910 would require a relocation, which is not possible. 911 912* If either the alias or the aliasee may be replaced by a symbol outside the 913 module at link time or runtime, any optimization cannot replace the alias with 914 the aliasee, since the behavior may be different. The alias may be used as a 915 name guaranteed to point to the content in the current module. 916 917.. _langref_ifunc: 918 919IFuncs 920------- 921 922IFuncs, like as aliases, don't create any new data or func. They are just a new 923symbol that dynamic linker resolves at runtime by calling a resolver function. 924 925IFuncs have a name and a resolver that is a function called by dynamic linker 926that returns address of another function associated with the name. 927 928IFunc may have an optional :ref:`linkage type <linkage>` and an optional 929:ref:`visibility style <visibility>`. 930 931Syntax:: 932 933 @<Name> = [Linkage] [PreemptionSpecifier] [Visibility] ifunc <IFuncTy>, <ResolverTy>* @<Resolver> 934 [, partition "name"] 935 936 937.. _langref_comdats: 938 939Comdats 940------- 941 942Comdat IR provides access to object file COMDAT/section group functionality 943which represents interrelated sections. 944 945Comdats have a name which represents the COMDAT key and a selection kind to 946provide input on how the linker deduplicates comdats with the same key in two 947different object files. A comdat must be included or omitted as a unit. 948Discarding the whole comdat is allowed but discarding a subset is not. 949 950A global object may be a member of at most one comdat. Aliases are placed in the 951same COMDAT that their aliasee computes to, if any. 952 953Syntax:: 954 955 $<Name> = comdat SelectionKind 956 957For selection kinds other than ``nodeduplicate``, only one of the duplicate 958comdats may be retained by the linker and the members of the remaining comdats 959must be discarded. The following selection kinds are supported: 960 961``any`` 962 The linker may choose any COMDAT key, the choice is arbitrary. 963``exactmatch`` 964 The linker may choose any COMDAT key but the sections must contain the 965 same data. 966``largest`` 967 The linker will choose the section containing the largest COMDAT key. 968``nodeduplicate`` 969 No deduplication is performed. 970``samesize`` 971 The linker may choose any COMDAT key but the sections must contain the 972 same amount of data. 973 974- XCOFF and Mach-O don't support COMDATs. 975- COFF supports all selection kinds. Non-``nodeduplicate`` selection kinds need 976 a non-local linkage COMDAT symbol. 977- ELF supports ``any`` and ``nodeduplicate``. 978- WebAssembly only supports ``any``. 979 980Here is an example of a COFF COMDAT where a function will only be selected if 981the COMDAT key's section is the largest: 982 983.. code-block:: text 984 985 $foo = comdat largest 986 @foo = global i32 2, comdat($foo) 987 988 define void @bar() comdat($foo) { 989 ret void 990 } 991 992In a COFF object file, this will create a COMDAT section with selection kind 993``IMAGE_COMDAT_SELECT_LARGEST`` containing the contents of the ``@foo`` symbol 994and another COMDAT section with selection kind 995``IMAGE_COMDAT_SELECT_ASSOCIATIVE`` which is associated with the first COMDAT 996section and contains the contents of the ``@bar`` symbol. 997 998As a syntactic sugar the ``$name`` can be omitted if the name is the same as 999the global name: 1000 1001.. code-block:: llvm 1002 1003 $foo = comdat any 1004 @foo = global i32 2, comdat 1005 @bar = global i32 3, comdat($foo) 1006 1007There are some restrictions on the properties of the global object. 1008It, or an alias to it, must have the same name as the COMDAT group when 1009targeting COFF. 1010The contents and size of this object may be used during link-time to determine 1011which COMDAT groups get selected depending on the selection kind. 1012Because the name of the object must match the name of the COMDAT group, the 1013linkage of the global object must not be local; local symbols can get renamed 1014if a collision occurs in the symbol table. 1015 1016The combined use of COMDATS and section attributes may yield surprising results. 1017For example: 1018 1019.. code-block:: llvm 1020 1021 $foo = comdat any 1022 $bar = comdat any 1023 @g1 = global i32 42, section "sec", comdat($foo) 1024 @g2 = global i32 42, section "sec", comdat($bar) 1025 1026From the object file perspective, this requires the creation of two sections 1027with the same name. This is necessary because both globals belong to different 1028COMDAT groups and COMDATs, at the object file level, are represented by 1029sections. 1030 1031Note that certain IR constructs like global variables and functions may 1032create COMDATs in the object file in addition to any which are specified using 1033COMDAT IR. This arises when the code generator is configured to emit globals 1034in individual sections (e.g. when `-data-sections` or `-function-sections` 1035is supplied to `llc`). 1036 1037.. _namedmetadatastructure: 1038 1039Named Metadata 1040-------------- 1041 1042Named metadata is a collection of metadata. :ref:`Metadata 1043nodes <metadata>` (but not metadata strings) are the only valid 1044operands for a named metadata. 1045 1046#. Named metadata are represented as a string of characters with the 1047 metadata prefix. The rules for metadata names are the same as for 1048 identifiers, but quoted names are not allowed. ``"\xx"`` type escapes 1049 are still valid, which allows any character to be part of a name. 1050 1051Syntax:: 1052 1053 ; Some unnamed metadata nodes, which are referenced by the named metadata. 1054 !0 = !{!"zero"} 1055 !1 = !{!"one"} 1056 !2 = !{!"two"} 1057 ; A named metadata. 1058 !name = !{!0, !1, !2} 1059 1060.. _paramattrs: 1061 1062Parameter Attributes 1063-------------------- 1064 1065The return type and each parameter of a function type may have a set of 1066*parameter attributes* associated with them. Parameter attributes are 1067used to communicate additional information about the result or 1068parameters of a function. Parameter attributes are considered to be part 1069of the function, not of the function type, so functions with different 1070parameter attributes can have the same function type. 1071 1072Parameter attributes are simple keywords that follow the type specified. 1073If multiple parameter attributes are needed, they are space separated. 1074For example: 1075 1076.. code-block:: llvm 1077 1078 declare i32 @printf(i8* noalias nocapture, ...) 1079 declare i32 @atoi(i8 zeroext) 1080 declare signext i8 @returns_signed_char() 1081 1082Note that any attributes for the function result (``nounwind``, 1083``readonly``) come immediately after the argument list. 1084 1085Currently, only the following parameter attributes are defined: 1086 1087``zeroext`` 1088 This indicates to the code generator that the parameter or return 1089 value should be zero-extended to the extent required by the target's 1090 ABI by the caller (for a parameter) or the callee (for a return value). 1091``signext`` 1092 This indicates to the code generator that the parameter or return 1093 value should be sign-extended to the extent required by the target's 1094 ABI (which is usually 32-bits) by the caller (for a parameter) or 1095 the callee (for a return value). 1096``inreg`` 1097 This indicates that this parameter or return value should be treated 1098 in a special target-dependent fashion while emitting code for 1099 a function call or return (usually, by putting it in a register as 1100 opposed to memory, though some targets use it to distinguish between 1101 two different kinds of registers). Use of this attribute is 1102 target-specific. 1103``byval(<ty>)`` 1104 This indicates that the pointer parameter should really be passed by 1105 value to the function. The attribute implies that a hidden copy of 1106 the pointee is made between the caller and the callee, so the callee 1107 is unable to modify the value in the caller. This attribute is only 1108 valid on LLVM pointer arguments. It is generally used to pass 1109 structs and arrays by value, but is also valid on pointers to 1110 scalars. The copy is considered to belong to the caller not the 1111 callee (for example, ``readonly`` functions should not write to 1112 ``byval`` parameters). This is not a valid attribute for return 1113 values. 1114 1115 The byval type argument indicates the in-memory value type, and 1116 must be the same as the pointee type of the argument. 1117 1118 The byval attribute also supports specifying an alignment with the 1119 align attribute. It indicates the alignment of the stack slot to 1120 form and the known alignment of the pointer specified to the call 1121 site. If the alignment is not specified, then the code generator 1122 makes a target-specific assumption. 1123 1124.. _attr_byref: 1125 1126``byref(<ty>)`` 1127 1128 The ``byref`` argument attribute allows specifying the pointee 1129 memory type of an argument. This is similar to ``byval``, but does 1130 not imply a copy is made anywhere, or that the argument is passed 1131 on the stack. This implies the pointer is dereferenceable up to 1132 the storage size of the type. 1133 1134 It is not generally permissible to introduce a write to an 1135 ``byref`` pointer. The pointer may have any address space and may 1136 be read only. 1137 1138 This is not a valid attribute for return values. 1139 1140 The alignment for an ``byref`` parameter can be explicitly 1141 specified by combining it with the ``align`` attribute, similar to 1142 ``byval``. If the alignment is not specified, then the code generator 1143 makes a target-specific assumption. 1144 1145 This is intended for representing ABI constraints, and is not 1146 intended to be inferred for optimization use. 1147 1148.. _attr_preallocated: 1149 1150``preallocated(<ty>)`` 1151 This indicates that the pointer parameter should really be passed by 1152 value to the function, and that the pointer parameter's pointee has 1153 already been initialized before the call instruction. This attribute 1154 is only valid on LLVM pointer arguments. The argument must be the value 1155 returned by the appropriate 1156 :ref:`llvm.call.preallocated.arg<int_call_preallocated_arg>` on non 1157 ``musttail`` calls, or the corresponding caller parameter in ``musttail`` 1158 calls, although it is ignored during codegen. 1159 1160 A non ``musttail`` function call with a ``preallocated`` attribute in 1161 any parameter must have a ``"preallocated"`` operand bundle. A ``musttail`` 1162 function call cannot have a ``"preallocated"`` operand bundle. 1163 1164 The preallocated attribute requires a type argument, which must be 1165 the same as the pointee type of the argument. 1166 1167 The preallocated attribute also supports specifying an alignment with the 1168 align attribute. It indicates the alignment of the stack slot to 1169 form and the known alignment of the pointer specified to the call 1170 site. If the alignment is not specified, then the code generator 1171 makes a target-specific assumption. 1172 1173.. _attr_inalloca: 1174 1175``inalloca(<ty>)`` 1176 1177 The ``inalloca`` argument attribute allows the caller to take the 1178 address of outgoing stack arguments. An ``inalloca`` argument must 1179 be a pointer to stack memory produced by an ``alloca`` instruction. 1180 The alloca, or argument allocation, must also be tagged with the 1181 inalloca keyword. Only the last argument may have the ``inalloca`` 1182 attribute, and that argument is guaranteed to be passed in memory. 1183 1184 An argument allocation may be used by a call at most once because 1185 the call may deallocate it. The ``inalloca`` attribute cannot be 1186 used in conjunction with other attributes that affect argument 1187 storage, like ``inreg``, ``nest``, ``sret``, or ``byval``. The 1188 ``inalloca`` attribute also disables LLVM's implicit lowering of 1189 large aggregate return values, which means that frontend authors 1190 must lower them with ``sret`` pointers. 1191 1192 When the call site is reached, the argument allocation must have 1193 been the most recent stack allocation that is still live, or the 1194 behavior is undefined. It is possible to allocate additional stack 1195 space after an argument allocation and before its call site, but it 1196 must be cleared off with :ref:`llvm.stackrestore 1197 <int_stackrestore>`. 1198 1199 The inalloca attribute requires a type argument, which must be the 1200 same as the pointee type of the argument. 1201 1202 See :doc:`InAlloca` for more information on how to use this 1203 attribute. 1204 1205``sret(<ty>)`` 1206 This indicates that the pointer parameter specifies the address of a 1207 structure that is the return value of the function in the source 1208 program. This pointer must be guaranteed by the caller to be valid: 1209 loads and stores to the structure may be assumed by the callee not 1210 to trap and to be properly aligned. This is not a valid attribute 1211 for return values. 1212 1213 The sret type argument specifies the in memory type, which must be 1214 the same as the pointee type of the argument. 1215 1216.. _attr_elementtype: 1217 1218``elementtype(<ty>)`` 1219 1220 The ``elementtype`` argument attribute can be used to specify a pointer 1221 element type in a way that is compatible with `opaque pointers 1222 <OpaquePointers.html>`__. 1223 1224 The ``elementtype`` attribute by itself does not carry any specific 1225 semantics. However, certain intrinsics may require this attribute to be 1226 present and assign it particular semantics. This will be documented on 1227 individual intrinsics. 1228 1229 The attribute may only be applied to pointer typed arguments of intrinsic 1230 calls. It cannot be applied to non-intrinsic calls, and cannot be applied 1231 to parameters on function declarations. For non-opaque pointers, the type 1232 passed to ``elementtype`` must match the pointer element type. 1233 1234.. _attr_align: 1235 1236``align <n>`` or ``align(<n>)`` 1237 This indicates that the pointer value or vector of pointers has the 1238 specified alignment. If applied to a vector of pointers, *all* pointers 1239 (elements) have the specified alignment. If the pointer value does not have 1240 the specified alignment, :ref:`poison value <poisonvalues>` is returned or 1241 passed instead. The ``align`` attribute should be combined with the 1242 ``noundef`` attribute to ensure a pointer is aligned, or otherwise the 1243 behavior is undefined. Note that ``align 1`` has no effect on non-byval, 1244 non-preallocated arguments. 1245 1246 Note that this attribute has additional semantics when combined with the 1247 ``byval`` or ``preallocated`` attribute, which are documented there. 1248 1249.. _noalias: 1250 1251``noalias`` 1252 This indicates that memory locations accessed via pointer values 1253 :ref:`based <pointeraliasing>` on the argument or return value are not also 1254 accessed, during the execution of the function, via pointer values not 1255 *based* on the argument or return value. This guarantee only holds for 1256 memory locations that are *modified*, by any means, during the execution of 1257 the function. The attribute on a return value also has additional semantics 1258 described below. The caller shares the responsibility with the callee for 1259 ensuring that these requirements are met. For further details, please see 1260 the discussion of the NoAlias response in :ref:`alias analysis <Must, May, 1261 or No>`. 1262 1263 Note that this definition of ``noalias`` is intentionally similar 1264 to the definition of ``restrict`` in C99 for function arguments. 1265 1266 For function return values, C99's ``restrict`` is not meaningful, 1267 while LLVM's ``noalias`` is. Furthermore, the semantics of the ``noalias`` 1268 attribute on return values are stronger than the semantics of the attribute 1269 when used on function arguments. On function return values, the ``noalias`` 1270 attribute indicates that the function acts like a system memory allocation 1271 function, returning a pointer to allocated storage disjoint from the 1272 storage for any other object accessible to the caller. 1273 1274.. _nocapture: 1275 1276``nocapture`` 1277 This indicates that the callee does not :ref:`capture <pointercapture>` the 1278 pointer. This is not a valid attribute for return values. 1279 This attribute applies only to the particular copy of the pointer passed in 1280 this argument. A caller could pass two copies of the same pointer with one 1281 being annotated nocapture and the other not, and the callee could validly 1282 capture through the non annotated parameter. 1283 1284.. code-block:: llvm 1285 1286 define void @f(i8* nocapture %a, i8* %b) { 1287 ; (capture %b) 1288 } 1289 1290 call void @f(i8* @glb, i8* @glb) ; well-defined 1291 1292``nofree`` 1293 This indicates that callee does not free the pointer argument. This is not 1294 a valid attribute for return values. 1295 1296.. _nest: 1297 1298``nest`` 1299 This indicates that the pointer parameter can be excised using the 1300 :ref:`trampoline intrinsics <int_trampoline>`. This is not a valid 1301 attribute for return values and can only be applied to one parameter. 1302 1303``returned`` 1304 This indicates that the function always returns the argument as its return 1305 value. This is a hint to the optimizer and code generator used when 1306 generating the caller, allowing value propagation, tail call optimization, 1307 and omission of register saves and restores in some cases; it is not 1308 checked or enforced when generating the callee. The parameter and the 1309 function return type must be valid operands for the 1310 :ref:`bitcast instruction <i_bitcast>`. This is not a valid attribute for 1311 return values and can only be applied to one parameter. 1312 1313``nonnull`` 1314 This indicates that the parameter or return pointer is not null. This 1315 attribute may only be applied to pointer typed parameters. This is not 1316 checked or enforced by LLVM; if the parameter or return pointer is null, 1317 :ref:`poison value <poisonvalues>` is returned or passed instead. 1318 The ``nonnull`` attribute should be combined with the ``noundef`` attribute 1319 to ensure a pointer is not null or otherwise the behavior is undefined. 1320 1321``dereferenceable(<n>)`` 1322 This indicates that the parameter or return pointer is dereferenceable. This 1323 attribute may only be applied to pointer typed parameters. A pointer that 1324 is dereferenceable can be loaded from speculatively without a risk of 1325 trapping. The number of bytes known to be dereferenceable must be provided 1326 in parentheses. It is legal for the number of bytes to be less than the 1327 size of the pointee type. The ``nonnull`` attribute does not imply 1328 dereferenceability (consider a pointer to one element past the end of an 1329 array), however ``dereferenceable(<n>)`` does imply ``nonnull`` in 1330 ``addrspace(0)`` (which is the default address space), except if the 1331 ``null_pointer_is_valid`` function attribute is present. 1332 ``n`` should be a positive number. The pointer should be well defined, 1333 otherwise it is undefined behavior. This means ``dereferenceable(<n>)`` 1334 implies ``noundef``. 1335 1336``dereferenceable_or_null(<n>)`` 1337 This indicates that the parameter or return value isn't both 1338 non-null and non-dereferenceable (up to ``<n>`` bytes) at the same 1339 time. All non-null pointers tagged with 1340 ``dereferenceable_or_null(<n>)`` are ``dereferenceable(<n>)``. 1341 For address space 0 ``dereferenceable_or_null(<n>)`` implies that 1342 a pointer is exactly one of ``dereferenceable(<n>)`` or ``null``, 1343 and in other address spaces ``dereferenceable_or_null(<n>)`` 1344 implies that a pointer is at least one of ``dereferenceable(<n>)`` 1345 or ``null`` (i.e. it may be both ``null`` and 1346 ``dereferenceable(<n>)``). This attribute may only be applied to 1347 pointer typed parameters. 1348 1349``swiftself`` 1350 This indicates that the parameter is the self/context parameter. This is not 1351 a valid attribute for return values and can only be applied to one 1352 parameter. 1353 1354``swiftasync`` 1355 This indicates that the parameter is the asynchronous context parameter and 1356 triggers the creation of a target-specific extended frame record to store 1357 this pointer. This is not a valid attribute for return values and can only 1358 be applied to one parameter. 1359 1360``swifterror`` 1361 This attribute is motivated to model and optimize Swift error handling. It 1362 can be applied to a parameter with pointer to pointer type or a 1363 pointer-sized alloca. At the call site, the actual argument that corresponds 1364 to a ``swifterror`` parameter has to come from a ``swifterror`` alloca or 1365 the ``swifterror`` parameter of the caller. A ``swifterror`` value (either 1366 the parameter or the alloca) can only be loaded and stored from, or used as 1367 a ``swifterror`` argument. This is not a valid attribute for return values 1368 and can only be applied to one parameter. 1369 1370 These constraints allow the calling convention to optimize access to 1371 ``swifterror`` variables by associating them with a specific register at 1372 call boundaries rather than placing them in memory. Since this does change 1373 the calling convention, a function which uses the ``swifterror`` attribute 1374 on a parameter is not ABI-compatible with one which does not. 1375 1376 These constraints also allow LLVM to assume that a ``swifterror`` argument 1377 does not alias any other memory visible within a function and that a 1378 ``swifterror`` alloca passed as an argument does not escape. 1379 1380``immarg`` 1381 This indicates the parameter is required to be an immediate 1382 value. This must be a trivial immediate integer or floating-point 1383 constant. Undef or constant expressions are not valid. This is 1384 only valid on intrinsic declarations and cannot be applied to a 1385 call site or arbitrary function. 1386 1387``noundef`` 1388 This attribute applies to parameters and return values. If the value 1389 representation contains any undefined or poison bits, the behavior is 1390 undefined. Note that this does not refer to padding introduced by the 1391 type's storage representation. 1392 1393``alignstack(<n>)`` 1394 This indicates the alignment that should be considered by the backend when 1395 assigning this parameter to a stack slot during calling convention 1396 lowering. The enforcement of the specified alignment is target-dependent, 1397 as target-specific calling convention rules may override this value. This 1398 attribute serves the purpose of carrying language specific alignment 1399 information that is not mapped to base types in the backend (for example, 1400 over-alignment specification through language attributes). 1401 1402``allocalign`` 1403 The function parameter marked with this attribute is is the alignment in bytes of the 1404 newly allocated block returned by this function. The returned value must either have 1405 the specified alignment or be the null pointer. The return value MAY be more aligned 1406 than the requested alignment, but not less aligned. Invalid (e.g. non-power-of-2) 1407 alignments are permitted for the allocalign parameter, so long as the returned pointer 1408 is null. This attribute may only be applied to integer parameters. 1409 1410``allocptr`` 1411 The function parameter marked with this attribute is the pointer 1412 that will be manipulated by the allocator. For a realloc-like 1413 function the pointer will be invalidated upon success (but the 1414 same address may be returned), for a free-like function the 1415 pointer will always be invalidated. 1416 1417.. _gc: 1418 1419Garbage Collector Strategy Names 1420-------------------------------- 1421 1422Each function may specify a garbage collector strategy name, which is simply a 1423string: 1424 1425.. code-block:: llvm 1426 1427 define void @f() gc "name" { ... } 1428 1429The supported values of *name* includes those :ref:`built in to LLVM 1430<builtin-gc-strategies>` and any provided by loaded plugins. Specifying a GC 1431strategy will cause the compiler to alter its output in order to support the 1432named garbage collection algorithm. Note that LLVM itself does not contain a 1433garbage collector, this functionality is restricted to generating machine code 1434which can interoperate with a collector provided externally. 1435 1436.. _prefixdata: 1437 1438Prefix Data 1439----------- 1440 1441Prefix data is data associated with a function which the code 1442generator will emit immediately before the function's entrypoint. 1443The purpose of this feature is to allow frontends to associate 1444language-specific runtime metadata with specific functions and make it 1445available through the function pointer while still allowing the 1446function pointer to be called. 1447 1448To access the data for a given function, a program may bitcast the 1449function pointer to a pointer to the constant's type and dereference 1450index -1. This implies that the IR symbol points just past the end of 1451the prefix data. For instance, take the example of a function annotated 1452with a single ``i32``, 1453 1454.. code-block:: llvm 1455 1456 define void @f() prefix i32 123 { ... } 1457 1458The prefix data can be referenced as, 1459 1460.. code-block:: llvm 1461 1462 %0 = bitcast void* () @f to i32* 1463 %a = getelementptr inbounds i32, i32* %0, i32 -1 1464 %b = load i32, i32* %a 1465 1466Prefix data is laid out as if it were an initializer for a global variable 1467of the prefix data's type. The function will be placed such that the 1468beginning of the prefix data is aligned. This means that if the size 1469of the prefix data is not a multiple of the alignment size, the 1470function's entrypoint will not be aligned. If alignment of the 1471function's entrypoint is desired, padding must be added to the prefix 1472data. 1473 1474A function may have prefix data but no body. This has similar semantics 1475to the ``available_externally`` linkage in that the data may be used by the 1476optimizers but will not be emitted in the object file. 1477 1478.. _prologuedata: 1479 1480Prologue Data 1481------------- 1482 1483The ``prologue`` attribute allows arbitrary code (encoded as bytes) to 1484be inserted prior to the function body. This can be used for enabling 1485function hot-patching and instrumentation. 1486 1487To maintain the semantics of ordinary function calls, the prologue data must 1488have a particular format. Specifically, it must begin with a sequence of 1489bytes which decode to a sequence of machine instructions, valid for the 1490module's target, which transfer control to the point immediately succeeding 1491the prologue data, without performing any other visible action. This allows 1492the inliner and other passes to reason about the semantics of the function 1493definition without needing to reason about the prologue data. Obviously this 1494makes the format of the prologue data highly target dependent. 1495 1496A trivial example of valid prologue data for the x86 architecture is ``i8 144``, 1497which encodes the ``nop`` instruction: 1498 1499.. code-block:: text 1500 1501 define void @f() prologue i8 144 { ... } 1502 1503Generally prologue data can be formed by encoding a relative branch instruction 1504which skips the metadata, as in this example of valid prologue data for the 1505x86_64 architecture, where the first two bytes encode ``jmp .+10``: 1506 1507.. code-block:: text 1508 1509 %0 = type <{ i8, i8, i8* }> 1510 1511 define void @f() prologue %0 <{ i8 235, i8 8, i8* @md}> { ... } 1512 1513A function may have prologue data but no body. This has similar semantics 1514to the ``available_externally`` linkage in that the data may be used by the 1515optimizers but will not be emitted in the object file. 1516 1517.. _personalityfn: 1518 1519Personality Function 1520-------------------- 1521 1522The ``personality`` attribute permits functions to specify what function 1523to use for exception handling. 1524 1525.. _attrgrp: 1526 1527Attribute Groups 1528---------------- 1529 1530Attribute groups are groups of attributes that are referenced by objects within 1531the IR. They are important for keeping ``.ll`` files readable, because a lot of 1532functions will use the same set of attributes. In the degenerative case of a 1533``.ll`` file that corresponds to a single ``.c`` file, the single attribute 1534group will capture the important command line flags used to build that file. 1535 1536An attribute group is a module-level object. To use an attribute group, an 1537object references the attribute group's ID (e.g. ``#37``). An object may refer 1538to more than one attribute group. In that situation, the attributes from the 1539different groups are merged. 1540 1541Here is an example of attribute groups for a function that should always be 1542inlined, has a stack alignment of 4, and which shouldn't use SSE instructions: 1543 1544.. code-block:: llvm 1545 1546 ; Target-independent attributes: 1547 attributes #0 = { alwaysinline alignstack=4 } 1548 1549 ; Target-dependent attributes: 1550 attributes #1 = { "no-sse" } 1551 1552 ; Function @f has attributes: alwaysinline, alignstack=4, and "no-sse". 1553 define void @f() #0 #1 { ... } 1554 1555.. _fnattrs: 1556 1557Function Attributes 1558------------------- 1559 1560Function attributes are set to communicate additional information about 1561a function. Function attributes are considered to be part of the 1562function, not of the function type, so functions with different function 1563attributes can have the same function type. 1564 1565Function attributes are simple keywords that follow the type specified. 1566If multiple attributes are needed, they are space separated. For 1567example: 1568 1569.. code-block:: llvm 1570 1571 define void @f() noinline { ... } 1572 define void @f() alwaysinline { ... } 1573 define void @f() alwaysinline optsize { ... } 1574 define void @f() optsize { ... } 1575 1576``alignstack(<n>)`` 1577 This attribute indicates that, when emitting the prologue and 1578 epilogue, the backend should forcibly align the stack pointer. 1579 Specify the desired alignment, which must be a power of two, in 1580 parentheses. 1581``"alloc-family"="FAMILY"`` 1582 This indicates which "family" an allocator function is part of. To avoid 1583 collisions, the family name should match the mangled name of the primary 1584 allocator function, that is "malloc" for malloc/calloc/realloc/free, 1585 "_Znwm" for ``::operator::new`` and ``::operator::delete``, and 1586 "_ZnwmSt11align_val_t" for aligned ``::operator::new`` and 1587 ``::operator::delete``. Matching malloc/realloc/free calls within a family 1588 can be optimized, but mismatched ones will be left alone. 1589``allockind("KIND")`` 1590 Describes the behavior of an allocation function. The KIND string contains comma 1591 separated entries from the following options: 1592 1593 * "alloc": the function returns a new block of memory or null. 1594 * "realloc": the function returns a new block of memory or null. If the 1595 result is non-null the memory contents from the start of the block up to 1596 the smaller of the original allocation size and the new allocation size 1597 will match that of the ``allocptr`` argument and the ``allocptr`` 1598 argument is invalidated, even if the function returns the same address. 1599 * "free": the function frees the block of memory specified by ``allocptr``. 1600 * "uninitialized": Any newly-allocated memory (either a new block from 1601 a "alloc" function or the enlarged capacity from a "realloc" function) 1602 will be uninitialized. 1603 * "zeroed": Any newly-allocated memory (either a new block from a "alloc" 1604 function or the enlarged capacity from a "realloc" function) will be 1605 zeroed. 1606 * "aligned": the function returns memory aligned according to the 1607 ``allocalign`` parameter. 1608 1609 The first three options are mutually exclusive, and the remaining options 1610 describe more details of how the function behaves. The remaining options 1611 are invalid for "free"-type functions. 1612``allocsize(<EltSizeParam>[, <NumEltsParam>])`` 1613 This attribute indicates that the annotated function will always return at 1614 least a given number of bytes (or null). Its arguments are zero-indexed 1615 parameter numbers; if one argument is provided, then it's assumed that at 1616 least ``CallSite.Args[EltSizeParam]`` bytes will be available at the 1617 returned pointer. If two are provided, then it's assumed that 1618 ``CallSite.Args[EltSizeParam] * CallSite.Args[NumEltsParam]`` bytes are 1619 available. The referenced parameters must be integer types. No assumptions 1620 are made about the contents of the returned block of memory. 1621``alwaysinline`` 1622 This attribute indicates that the inliner should attempt to inline 1623 this function into callers whenever possible, ignoring any active 1624 inlining size threshold for this caller. 1625``builtin`` 1626 This indicates that the callee function at a call site should be 1627 recognized as a built-in function, even though the function's declaration 1628 uses the ``nobuiltin`` attribute. This is only valid at call sites for 1629 direct calls to functions that are declared with the ``nobuiltin`` 1630 attribute. 1631``cold`` 1632 This attribute indicates that this function is rarely called. When 1633 computing edge weights, basic blocks post-dominated by a cold 1634 function call are also considered to be cold; and, thus, given low 1635 weight. 1636``convergent`` 1637 In some parallel execution models, there exist operations that cannot be 1638 made control-dependent on any additional values. We call such operations 1639 ``convergent``, and mark them with this attribute. 1640 1641 The ``convergent`` attribute may appear on functions or call/invoke 1642 instructions. When it appears on a function, it indicates that calls to 1643 this function should not be made control-dependent on additional values. 1644 For example, the intrinsic ``llvm.nvvm.barrier0`` is ``convergent``, so 1645 calls to this intrinsic cannot be made control-dependent on additional 1646 values. 1647 1648 When it appears on a call/invoke, the ``convergent`` attribute indicates 1649 that we should treat the call as though we're calling a convergent 1650 function. This is particularly useful on indirect calls; without this we 1651 may treat such calls as though the target is non-convergent. 1652 1653 The optimizer may remove the ``convergent`` attribute on functions when it 1654 can prove that the function does not execute any convergent operations. 1655 Similarly, the optimizer may remove ``convergent`` on calls/invokes when it 1656 can prove that the call/invoke cannot call a convergent function. 1657``disable_sanitizer_instrumentation`` 1658 When instrumenting code with sanitizers, it can be important to skip certain 1659 functions to ensure no instrumentation is applied to them. 1660 1661 This attribute is not always similar to absent ``sanitize_<name>`` 1662 attributes: depending on the specific sanitizer, code can be inserted into 1663 functions regardless of the ``sanitize_<name>`` attribute to prevent false 1664 positive reports. 1665 1666 ``disable_sanitizer_instrumentation`` disables all kinds of instrumentation, 1667 taking precedence over the ``sanitize_<name>`` attributes and other compiler 1668 flags. 1669``"dontcall-error"`` 1670 This attribute denotes that an error diagnostic should be emitted when a 1671 call of a function with this attribute is not eliminated via optimization. 1672 Front ends can provide optional ``srcloc`` metadata nodes on call sites of 1673 such callees to attach information about where in the source language such a 1674 call came from. A string value can be provided as a note. 1675``"dontcall-warn"`` 1676 This attribute denotes that a warning diagnostic should be emitted when a 1677 call of a function with this attribute is not eliminated via optimization. 1678 Front ends can provide optional ``srcloc`` metadata nodes on call sites of 1679 such callees to attach information about where in the source language such a 1680 call came from. A string value can be provided as a note. 1681``fn_ret_thunk_extern`` 1682 This attribute tells the code generator that returns from functions should 1683 be replaced with jumps to externally-defined architecture-specific symbols. 1684 For X86, this symbol's identifier is ``__x86_return_thunk``. 1685``"frame-pointer"`` 1686 This attribute tells the code generator whether the function 1687 should keep the frame pointer. The code generator may emit the frame pointer 1688 even if this attribute says the frame pointer can be eliminated. 1689 The allowed string values are: 1690 1691 * ``"none"`` (default) - the frame pointer can be eliminated. 1692 * ``"non-leaf"`` - the frame pointer should be kept if the function calls 1693 other functions. 1694 * ``"all"`` - the frame pointer should be kept. 1695``hot`` 1696 This attribute indicates that this function is a hot spot of the program 1697 execution. The function will be optimized more aggressively and will be 1698 placed into special subsection of the text section to improving locality. 1699 1700 When profile feedback is enabled, this attribute has the precedence over 1701 the profile information. By marking a function ``hot``, users can work 1702 around the cases where the training input does not have good coverage 1703 on all the hot functions. 1704``inaccessiblememonly`` 1705 This attribute indicates that the function may only access memory that 1706 is not accessible by the module being compiled before return from the 1707 function. This is a weaker form of ``readnone``. If the function reads 1708 or writes other memory, the behavior is undefined. 1709 1710 For clarity, note that such functions are allowed to return new memory 1711 which is ``noalias`` with respect to memory already accessible from 1712 the module. That is, a function can be both ``inaccessiblememonly`` and 1713 have a ``noalias`` return which introduces a new, potentially initialized, 1714 allocation. 1715``inaccessiblemem_or_argmemonly`` 1716 This attribute indicates that the function may only access memory that is 1717 either not accessible by the module being compiled, or is pointed to 1718 by its pointer arguments. This is a weaker form of ``argmemonly``. If the 1719 function reads or writes other memory, the behavior is undefined. 1720``inlinehint`` 1721 This attribute indicates that the source code contained a hint that 1722 inlining this function is desirable (such as the "inline" keyword in 1723 C/C++). It is just a hint; it imposes no requirements on the 1724 inliner. 1725``jumptable`` 1726 This attribute indicates that the function should be added to a 1727 jump-instruction table at code-generation time, and that all address-taken 1728 references to this function should be replaced with a reference to the 1729 appropriate jump-instruction-table function pointer. Note that this creates 1730 a new pointer for the original function, which means that code that depends 1731 on function-pointer identity can break. So, any function annotated with 1732 ``jumptable`` must also be ``unnamed_addr``. 1733``minsize`` 1734 This attribute suggests that optimization passes and code generator 1735 passes make choices that keep the code size of this function as small 1736 as possible and perform optimizations that may sacrifice runtime 1737 performance in order to minimize the size of the generated code. 1738``naked`` 1739 This attribute disables prologue / epilogue emission for the 1740 function. This can have very system-specific consequences. 1741``"no-inline-line-tables"`` 1742 When this attribute is set to true, the inliner discards source locations 1743 when inlining code and instead uses the source location of the call site. 1744 Breakpoints set on code that was inlined into the current function will 1745 not fire during the execution of the inlined call sites. If the debugger 1746 stops inside an inlined call site, it will appear to be stopped at the 1747 outermost inlined call site. 1748``no-jump-tables`` 1749 When this attribute is set to true, the jump tables and lookup tables that 1750 can be generated from a switch case lowering are disabled. 1751``nobuiltin`` 1752 This indicates that the callee function at a call site is not recognized as 1753 a built-in function. LLVM will retain the original call and not replace it 1754 with equivalent code based on the semantics of the built-in function, unless 1755 the call site uses the ``builtin`` attribute. This is valid at call sites 1756 and on function declarations and definitions. 1757``noduplicate`` 1758 This attribute indicates that calls to the function cannot be 1759 duplicated. A call to a ``noduplicate`` function may be moved 1760 within its parent function, but may not be duplicated within 1761 its parent function. 1762 1763 A function containing a ``noduplicate`` call may still 1764 be an inlining candidate, provided that the call is not 1765 duplicated by inlining. That implies that the function has 1766 internal linkage and only has one call site, so the original 1767 call is dead after inlining. 1768``nofree`` 1769 This function attribute indicates that the function does not, directly or 1770 transitively, call a memory-deallocation function (``free``, for example) 1771 on a memory allocation which existed before the call. 1772 1773 As a result, uncaptured pointers that are known to be dereferenceable 1774 prior to a call to a function with the ``nofree`` attribute are still 1775 known to be dereferenceable after the call. The capturing condition is 1776 necessary in environments where the function might communicate the 1777 pointer to another thread which then deallocates the memory. Alternatively, 1778 ``nosync`` would ensure such communication cannot happen and even captured 1779 pointers cannot be freed by the function. 1780 1781 A ``nofree`` function is explicitly allowed to free memory which it 1782 allocated or (if not ``nosync``) arrange for another thread to free 1783 memory on it's behalf. As a result, perhaps surprisingly, a ``nofree`` 1784 function can return a pointer to a previously deallocated memory object. 1785``noimplicitfloat`` 1786 Disallows implicit floating-point code. This inhibits optimizations that 1787 use floating-point code and floating-point/SIMD/vector registers for 1788 operations that are not nominally floating-point. LLVM instructions that 1789 perform floating-point operations or require access to floating-point 1790 registers may still cause floating-point code to be generated. 1791``noinline`` 1792 This attribute indicates that the inliner should never inline this 1793 function in any situation. This attribute may not be used together 1794 with the ``alwaysinline`` attribute. 1795``nomerge`` 1796 This attribute indicates that calls to this function should never be merged 1797 during optimization. For example, it will prevent tail merging otherwise 1798 identical code sequences that raise an exception or terminate the program. 1799 Tail merging normally reduces the precision of source location information, 1800 making stack traces less useful for debugging. This attribute gives the 1801 user control over the tradeoff between code size and debug information 1802 precision. 1803``nonlazybind`` 1804 This attribute suppresses lazy symbol binding for the function. This 1805 may make calls to the function faster, at the cost of extra program 1806 startup time if the function is not called during program startup. 1807``noprofile`` 1808 This function attribute prevents instrumentation based profiling, used for 1809 coverage or profile based optimization, from being added to a function, 1810 even when inlined. 1811``noredzone`` 1812 This attribute indicates that the code generator should not use a 1813 red zone, even if the target-specific ABI normally permits it. 1814``indirect-tls-seg-refs`` 1815 This attribute indicates that the code generator should not use 1816 direct TLS access through segment registers, even if the 1817 target-specific ABI normally permits it. 1818``noreturn`` 1819 This function attribute indicates that the function never returns 1820 normally, hence through a return instruction. This produces undefined 1821 behavior at runtime if the function ever does dynamically return. Annotated 1822 functions may still raise an exception, i.a., ``nounwind`` is not implied. 1823``norecurse`` 1824 This function attribute indicates that the function does not call itself 1825 either directly or indirectly down any possible call path. This produces 1826 undefined behavior at runtime if the function ever does recurse. 1827 1828.. _langref_willreturn: 1829 1830``willreturn`` 1831 This function attribute indicates that a call of this function will 1832 either exhibit undefined behavior or comes back and continues execution 1833 at a point in the existing call stack that includes the current invocation. 1834 Annotated functions may still raise an exception, i.a., ``nounwind`` is not implied. 1835 If an invocation of an annotated function does not return control back 1836 to a point in the call stack, the behavior is undefined. 1837``nosync`` 1838 This function attribute indicates that the function does not communicate 1839 (synchronize) with another thread through memory or other well-defined means. 1840 Synchronization is considered possible in the presence of `atomic` accesses 1841 that enforce an order, thus not "unordered" and "monotonic", `volatile` accesses, 1842 as well as `convergent` function calls. Note that through `convergent` function calls 1843 non-memory communication, e.g., cross-lane operations, are possible and are also 1844 considered synchronization. However `convergent` does not contradict `nosync`. 1845 If an annotated function does ever synchronize with another thread, 1846 the behavior is undefined. 1847``nounwind`` 1848 This function attribute indicates that the function never raises an 1849 exception. If the function does raise an exception, its runtime 1850 behavior is undefined. However, functions marked nounwind may still 1851 trap or generate asynchronous exceptions. Exception handling schemes 1852 that are recognized by LLVM to handle asynchronous exceptions, such 1853 as SEH, will still provide their implementation defined semantics. 1854``nosanitize_bounds`` 1855 This attribute indicates that bounds checking sanitizer instrumentation 1856 is disabled for this function. 1857``nosanitize_coverage`` 1858 This attribute indicates that SanitizerCoverage instrumentation is disabled 1859 for this function. 1860``null_pointer_is_valid`` 1861 If ``null_pointer_is_valid`` is set, then the ``null`` address 1862 in address-space 0 is considered to be a valid address for memory loads and 1863 stores. Any analysis or optimization should not treat dereferencing a 1864 pointer to ``null`` as undefined behavior in this function. 1865 Note: Comparing address of a global variable to ``null`` may still 1866 evaluate to false because of a limitation in querying this attribute inside 1867 constant expressions. 1868``optforfuzzing`` 1869 This attribute indicates that this function should be optimized 1870 for maximum fuzzing signal. 1871``optnone`` 1872 This function attribute indicates that most optimization passes will skip 1873 this function, with the exception of interprocedural optimization passes. 1874 Code generation defaults to the "fast" instruction selector. 1875 This attribute cannot be used together with the ``alwaysinline`` 1876 attribute; this attribute is also incompatible 1877 with the ``minsize`` attribute and the ``optsize`` attribute. 1878 1879 This attribute requires the ``noinline`` attribute to be specified on 1880 the function as well, so the function is never inlined into any caller. 1881 Only functions with the ``alwaysinline`` attribute are valid 1882 candidates for inlining into the body of this function. 1883``optsize`` 1884 This attribute suggests that optimization passes and code generator 1885 passes make choices that keep the code size of this function low, 1886 and otherwise do optimizations specifically to reduce code size as 1887 long as they do not significantly impact runtime performance. 1888``"patchable-function"`` 1889 This attribute tells the code generator that the code 1890 generated for this function needs to follow certain conventions that 1891 make it possible for a runtime function to patch over it later. 1892 The exact effect of this attribute depends on its string value, 1893 for which there currently is one legal possibility: 1894 1895 * ``"prologue-short-redirect"`` - This style of patchable 1896 function is intended to support patching a function prologue to 1897 redirect control away from the function in a thread safe 1898 manner. It guarantees that the first instruction of the 1899 function will be large enough to accommodate a short jump 1900 instruction, and will be sufficiently aligned to allow being 1901 fully changed via an atomic compare-and-swap instruction. 1902 While the first requirement can be satisfied by inserting large 1903 enough NOP, LLVM can and will try to re-purpose an existing 1904 instruction (i.e. one that would have to be emitted anyway) as 1905 the patchable instruction larger than a short jump. 1906 1907 ``"prologue-short-redirect"`` is currently only supported on 1908 x86-64. 1909 1910 This attribute by itself does not imply restrictions on 1911 inter-procedural optimizations. All of the semantic effects the 1912 patching may have to be separately conveyed via the linkage type. 1913``"probe-stack"`` 1914 This attribute indicates that the function will trigger a guard region 1915 in the end of the stack. It ensures that accesses to the stack must be 1916 no further apart than the size of the guard region to a previous 1917 access of the stack. It takes one required string value, the name of 1918 the stack probing function that will be called. 1919 1920 If a function that has a ``"probe-stack"`` attribute is inlined into 1921 a function with another ``"probe-stack"`` attribute, the resulting 1922 function has the ``"probe-stack"`` attribute of the caller. If a 1923 function that has a ``"probe-stack"`` attribute is inlined into a 1924 function that has no ``"probe-stack"`` attribute at all, the resulting 1925 function has the ``"probe-stack"`` attribute of the callee. 1926``readnone`` 1927 On a function, this attribute indicates that the function computes its 1928 result (or decides to unwind an exception) based strictly on its arguments, 1929 without dereferencing any pointer arguments or otherwise accessing 1930 any mutable state (e.g. memory, control registers, etc) visible outside the 1931 ``readnone`` function. It does not write through any pointer arguments 1932 (including ``byval`` arguments) and never changes any state visible to 1933 callers. This means while it cannot unwind exceptions by calling the ``C++`` 1934 exception throwing methods (since they write to memory), there may be 1935 non-``C++`` mechanisms that throw exceptions without writing to LLVM visible 1936 memory. 1937 1938 On an argument, this attribute indicates that the function does not 1939 dereference that pointer argument, even though it may read or write the 1940 memory that the pointer points to if accessed through other pointers. 1941 1942 If a readnone function reads or writes memory visible outside the function, 1943 or has other side-effects, the behavior is undefined. If a 1944 function reads from or writes to a readnone pointer argument, the behavior 1945 is undefined. 1946``readonly`` 1947 On a function, this attribute indicates that the function does not write 1948 through any pointer arguments (including ``byval`` arguments) or otherwise 1949 modify any state (e.g. memory, control registers, etc) visible outside the 1950 ``readonly`` function. It may dereference pointer arguments and read 1951 state that may be set in the caller. A readonly function always 1952 returns the same value (or unwinds an exception identically) when 1953 called with the same set of arguments and global state. This means while it 1954 cannot unwind exceptions by calling the ``C++`` exception throwing methods 1955 (since they write to memory), there may be non-``C++`` mechanisms that throw 1956 exceptions without writing to LLVM visible memory. 1957 1958 On an argument, this attribute indicates that the function does not write 1959 through this pointer argument, even though it may write to the memory that 1960 the pointer points to. 1961 1962 If a readonly function writes memory visible outside the function, or has 1963 other side-effects, the behavior is undefined. If a function writes to a 1964 readonly pointer argument, the behavior is undefined. 1965``"stack-probe-size"`` 1966 This attribute controls the behavior of stack probes: either 1967 the ``"probe-stack"`` attribute, or ABI-required stack probes, if any. 1968 It defines the size of the guard region. It ensures that if the function 1969 may use more stack space than the size of the guard region, stack probing 1970 sequence will be emitted. It takes one required integer value, which 1971 is 4096 by default. 1972 1973 If a function that has a ``"stack-probe-size"`` attribute is inlined into 1974 a function with another ``"stack-probe-size"`` attribute, the resulting 1975 function has the ``"stack-probe-size"`` attribute that has the lower 1976 numeric value. If a function that has a ``"stack-probe-size"`` attribute is 1977 inlined into a function that has no ``"stack-probe-size"`` attribute 1978 at all, the resulting function has the ``"stack-probe-size"`` attribute 1979 of the callee. 1980``"no-stack-arg-probe"`` 1981 This attribute disables ABI-required stack probes, if any. 1982``writeonly`` 1983 On a function, this attribute indicates that the function may write to but 1984 does not read from memory visible outside the ``writeonly`` function. 1985 1986 On an argument, this attribute indicates that the function may write to but 1987 does not read through this pointer argument (even though it may read from 1988 the memory that the pointer points to). 1989 1990 If a writeonly function reads memory visible outside the function or has 1991 other side-effects, the behavior is undefined. If a function reads 1992 from a writeonly pointer argument, the behavior is undefined. 1993``argmemonly`` 1994 This attribute indicates that the only memory accesses inside function are 1995 loads and stores from objects pointed to by its pointer-typed arguments, 1996 with arbitrary offsets. Or in other words, all memory operations in the 1997 function can refer to memory only using pointers based on its function 1998 arguments. 1999 2000 Note that ``argmemonly`` can be used together with ``readonly`` attribute 2001 in order to specify that function reads only from its arguments. 2002 2003 If an argmemonly function reads or writes memory other than the pointer 2004 arguments, or has other side-effects, the behavior is undefined. 2005``returns_twice`` 2006 This attribute indicates that this function can return twice. The C 2007 ``setjmp`` is an example of such a function. The compiler disables 2008 some optimizations (like tail calls) in the caller of these 2009 functions. 2010``safestack`` 2011 This attribute indicates that 2012 `SafeStack <https://clang.llvm.org/docs/SafeStack.html>`_ 2013 protection is enabled for this function. 2014 2015 If a function that has a ``safestack`` attribute is inlined into a 2016 function that doesn't have a ``safestack`` attribute or which has an 2017 ``ssp``, ``sspstrong`` or ``sspreq`` attribute, then the resulting 2018 function will have a ``safestack`` attribute. 2019``sanitize_address`` 2020 This attribute indicates that AddressSanitizer checks 2021 (dynamic address safety analysis) are enabled for this function. 2022``sanitize_memory`` 2023 This attribute indicates that MemorySanitizer checks (dynamic detection 2024 of accesses to uninitialized memory) are enabled for this function. 2025``sanitize_thread`` 2026 This attribute indicates that ThreadSanitizer checks 2027 (dynamic thread safety analysis) are enabled for this function. 2028``sanitize_hwaddress`` 2029 This attribute indicates that HWAddressSanitizer checks 2030 (dynamic address safety analysis based on tagged pointers) are enabled for 2031 this function. 2032``sanitize_memtag`` 2033 This attribute indicates that MemTagSanitizer checks 2034 (dynamic address safety analysis based on Armv8 MTE) are enabled for 2035 this function. 2036``speculative_load_hardening`` 2037 This attribute indicates that 2038 `Speculative Load Hardening <https://llvm.org/docs/SpeculativeLoadHardening.html>`_ 2039 should be enabled for the function body. 2040 2041 Speculative Load Hardening is a best-effort mitigation against 2042 information leak attacks that make use of control flow 2043 miss-speculation - specifically miss-speculation of whether a branch 2044 is taken or not. Typically vulnerabilities enabling such attacks are 2045 classified as "Spectre variant #1". Notably, this does not attempt to 2046 mitigate against miss-speculation of branch target, classified as 2047 "Spectre variant #2" vulnerabilities. 2048 2049 When inlining, the attribute is sticky. Inlining a function that carries 2050 this attribute will cause the caller to gain the attribute. This is intended 2051 to provide a maximally conservative model where the code in a function 2052 annotated with this attribute will always (even after inlining) end up 2053 hardened. 2054``speculatable`` 2055 This function attribute indicates that the function does not have any 2056 effects besides calculating its result and does not have undefined behavior. 2057 Note that ``speculatable`` is not enough to conclude that along any 2058 particular execution path the number of calls to this function will not be 2059 externally observable. This attribute is only valid on functions 2060 and declarations, not on individual call sites. If a function is 2061 incorrectly marked as speculatable and really does exhibit 2062 undefined behavior, the undefined behavior may be observed even 2063 if the call site is dead code. 2064 2065``ssp`` 2066 This attribute indicates that the function should emit a stack 2067 smashing protector. It is in the form of a "canary" --- a random value 2068 placed on the stack before the local variables that's checked upon 2069 return from the function to see if it has been overwritten. A 2070 heuristic is used to determine if a function needs stack protectors 2071 or not. The heuristic used will enable protectors for functions with: 2072 2073 - Character arrays larger than ``ssp-buffer-size`` (default 8). 2074 - Aggregates containing character arrays larger than ``ssp-buffer-size``. 2075 - Calls to alloca() with variable sizes or constant sizes greater than 2076 ``ssp-buffer-size``. 2077 2078 Variables that are identified as requiring a protector will be arranged 2079 on the stack such that they are adjacent to the stack protector guard. 2080 2081 If a function with an ``ssp`` attribute is inlined into a calling function, 2082 the attribute is not carried over to the calling function. 2083 2084``sspstrong`` 2085 This attribute indicates that the function should emit a stack smashing 2086 protector. This attribute causes a strong heuristic to be used when 2087 determining if a function needs stack protectors. The strong heuristic 2088 will enable protectors for functions with: 2089 2090 - Arrays of any size and type 2091 - Aggregates containing an array of any size and type. 2092 - Calls to alloca(). 2093 - Local variables that have had their address taken. 2094 2095 Variables that are identified as requiring a protector will be arranged 2096 on the stack such that they are adjacent to the stack protector guard. 2097 The specific layout rules are: 2098 2099 #. Large arrays and structures containing large arrays 2100 (``>= ssp-buffer-size``) are closest to the stack protector. 2101 #. Small arrays and structures containing small arrays 2102 (``< ssp-buffer-size``) are 2nd closest to the protector. 2103 #. Variables that have had their address taken are 3rd closest to the 2104 protector. 2105 2106 This overrides the ``ssp`` function attribute. 2107 2108 If a function with an ``sspstrong`` attribute is inlined into a calling 2109 function which has an ``ssp`` attribute, the calling function's attribute 2110 will be upgraded to ``sspstrong``. 2111 2112``sspreq`` 2113 This attribute indicates that the function should *always* emit a stack 2114 smashing protector. This overrides the ``ssp`` and ``sspstrong`` function 2115 attributes. 2116 2117 Variables that are identified as requiring a protector will be arranged 2118 on the stack such that they are adjacent to the stack protector guard. 2119 The specific layout rules are: 2120 2121 #. Large arrays and structures containing large arrays 2122 (``>= ssp-buffer-size``) are closest to the stack protector. 2123 #. Small arrays and structures containing small arrays 2124 (``< ssp-buffer-size``) are 2nd closest to the protector. 2125 #. Variables that have had their address taken are 3rd closest to the 2126 protector. 2127 2128 If a function with an ``sspreq`` attribute is inlined into a calling 2129 function which has an ``ssp`` or ``sspstrong`` attribute, the calling 2130 function's attribute will be upgraded to ``sspreq``. 2131 2132``strictfp`` 2133 This attribute indicates that the function was called from a scope that 2134 requires strict floating-point semantics. LLVM will not attempt any 2135 optimizations that require assumptions about the floating-point rounding 2136 mode or that might alter the state of floating-point status flags that 2137 might otherwise be set or cleared by calling this function. LLVM will 2138 not introduce any new floating-point instructions that may trap. 2139 2140``"denormal-fp-math"`` 2141 This indicates the denormal (subnormal) handling that may be 2142 assumed for the default floating-point environment. This is a 2143 comma separated pair. The elements may be one of ``"ieee"``, 2144 ``"preserve-sign"``, or ``"positive-zero"``. The first entry 2145 indicates the flushing mode for the result of floating point 2146 operations. The second indicates the handling of denormal inputs 2147 to floating point instructions. For compatibility with older 2148 bitcode, if the second value is omitted, both input and output 2149 modes will assume the same mode. 2150 2151 If this is attribute is not specified, the default is 2152 ``"ieee,ieee"``. 2153 2154 If the output mode is ``"preserve-sign"``, or ``"positive-zero"``, 2155 denormal outputs may be flushed to zero by standard floating-point 2156 operations. It is not mandated that flushing to zero occurs, but if 2157 a denormal output is flushed to zero, it must respect the sign 2158 mode. Not all targets support all modes. While this indicates the 2159 expected floating point mode the function will be executed with, 2160 this does not make any attempt to ensure the mode is 2161 consistent. User or platform code is expected to set the floating 2162 point mode appropriately before function entry. 2163 2164 If the input mode is ``"preserve-sign"``, or ``"positive-zero"``, a 2165 floating-point operation must treat any input denormal value as 2166 zero. In some situations, if an instruction does not respect this 2167 mode, the input may need to be converted to 0 as if by 2168 ``@llvm.canonicalize`` during lowering for correctness. 2169 2170``"denormal-fp-math-f32"`` 2171 Same as ``"denormal-fp-math"``, but only controls the behavior of 2172 the 32-bit float type (or vectors of 32-bit floats). If both are 2173 are present, this overrides ``"denormal-fp-math"``. Not all targets 2174 support separately setting the denormal mode per type, and no 2175 attempt is made to diagnose unsupported uses. Currently this 2176 attribute is respected by the AMDGPU and NVPTX backends. 2177 2178``"thunk"`` 2179 This attribute indicates that the function will delegate to some other 2180 function with a tail call. The prototype of a thunk should not be used for 2181 optimization purposes. The caller is expected to cast the thunk prototype to 2182 match the thunk target prototype. 2183 2184``"tls-load-hoist"`` 2185 This attribute indicates that the function will try to reduce redundant 2186 tls address calculation by hoisting tls variable. 2187 2188``uwtable[(sync|async)]`` 2189 This attribute indicates that the ABI being targeted requires that 2190 an unwind table entry be produced for this function even if we can 2191 show that no exceptions passes by it. This is normally the case for 2192 the ELF x86-64 abi, but it can be disabled for some compilation 2193 units. The optional parameter describes what kind of unwind tables 2194 to generate: ``sync`` for normal unwind tables, ``async`` for asynchronous 2195 (instruction precise) unwind tables. Without the parameter, the attribute 2196 ``uwtable`` is equivalent to ``uwtable(async)``. 2197``nocf_check`` 2198 This attribute indicates that no control-flow check will be performed on 2199 the attributed entity. It disables -fcf-protection=<> for a specific 2200 entity to fine grain the HW control flow protection mechanism. The flag 2201 is target independent and currently appertains to a function or function 2202 pointer. 2203``shadowcallstack`` 2204 This attribute indicates that the ShadowCallStack checks are enabled for 2205 the function. The instrumentation checks that the return address for the 2206 function has not changed between the function prolog and epilog. It is 2207 currently x86_64-specific. 2208 2209.. _langref_mustprogress: 2210 2211``mustprogress`` 2212 This attribute indicates that the function is required to return, unwind, 2213 or interact with the environment in an observable way e.g. via a volatile 2214 memory access, I/O, or other synchronization. The ``mustprogress`` 2215 attribute is intended to model the requirements of the first section of 2216 [intro.progress] of the C++ Standard. As a consequence, a loop in a 2217 function with the `mustprogress` attribute can be assumed to terminate if 2218 it does not interact with the environment in an observable way, and 2219 terminating loops without side-effects can be removed. If a `mustprogress` 2220 function does not satisfy this contract, the behavior is undefined. This 2221 attribute does not apply transitively to callees, but does apply to call 2222 sites within the function. Note that `willreturn` implies `mustprogress`. 2223``"warn-stack-size"="<threshold>"`` 2224 This attribute sets a threshold to emit diagnostics once the frame size is 2225 known should the frame size exceed the specified value. It takes one 2226 required integer value, which should be a non-negative integer, and less 2227 than `UINT_MAX`. It's unspecified which threshold will be used when 2228 duplicate definitions are linked together with differing values. 2229``vscale_range(<min>[, <max>])`` 2230 This attribute indicates the minimum and maximum vscale value for the given 2231 function. The min must be greater than 0. A maximum value of 0 means 2232 unbounded. If the optional max value is omitted then max is set to the 2233 value of min. If the attribute is not present, no assumptions are made 2234 about the range of vscale. 2235``"min-legal-vector-width"="<size>"`` 2236 This attribute indicates the minimum legal vector width required by the 2237 calling conversion. It is the maximum width of vector arguments and 2238 returnings in the function and functions called by this function. Because 2239 all the vectors are supposed to be legal type for compatibility. 2240 Backends are free to ignore the attribute if they don't need to support 2241 different maximum legal vector types or such information can be inferred by 2242 other attributes. 2243 2244Call Site Attributes 2245---------------------- 2246 2247In addition to function attributes the following call site only 2248attributes are supported: 2249 2250``vector-function-abi-variant`` 2251 This attribute can be attached to a :ref:`call <i_call>` to list 2252 the vector functions associated to the function. Notice that the 2253 attribute cannot be attached to a :ref:`invoke <i_invoke>` or a 2254 :ref:`callbr <i_callbr>` instruction. The attribute consists of a 2255 comma separated list of mangled names. The order of the list does 2256 not imply preference (it is logically a set). The compiler is free 2257 to pick any listed vector function of its choosing. 2258 2259 The syntax for the mangled names is as follows::: 2260 2261 _ZGV<isa><mask><vlen><parameters>_<scalar_name>[(<vector_redirection>)] 2262 2263 When present, the attribute informs the compiler that the function 2264 ``<scalar_name>`` has a corresponding vector variant that can be 2265 used to perform the concurrent invocation of ``<scalar_name>`` on 2266 vectors. The shape of the vector function is described by the 2267 tokens between the prefix ``_ZGV`` and the ``<scalar_name>`` 2268 token. The standard name of the vector function is 2269 ``_ZGV<isa><mask><vlen><parameters>_<scalar_name>``. When present, 2270 the optional token ``(<vector_redirection>)`` informs the compiler 2271 that a custom name is provided in addition to the standard one 2272 (custom names can be provided for example via the use of ``declare 2273 variant`` in OpenMP 5.0). The declaration of the variant must be 2274 present in the IR Module. The signature of the vector variant is 2275 determined by the rules of the Vector Function ABI (VFABI) 2276 specifications of the target. For Arm and X86, the VFABI can be 2277 found at https://github.com/ARM-software/abi-aa and 2278 https://software.intel.com/content/www/us/en/develop/download/vector-simd-function-abi.html, 2279 respectively. 2280 2281 For X86 and Arm targets, the values of the tokens in the standard 2282 name are those that are defined in the VFABI. LLVM has an internal 2283 ``<isa>`` token that can be used to create scalar-to-vector 2284 mappings for functions that are not directly associated to any of 2285 the target ISAs (for example, some of the mappings stored in the 2286 TargetLibraryInfo). Valid values for the ``<isa>`` token are::: 2287 2288 <isa>:= b | c | d | e -> X86 SSE, AVX, AVX2, AVX512 2289 | n | s -> Armv8 Advanced SIMD, SVE 2290 | __LLVM__ -> Internal LLVM Vector ISA 2291 2292 For all targets currently supported (x86, Arm and Internal LLVM), 2293 the remaining tokens can have the following values::: 2294 2295 <mask>:= M | N -> mask | no mask 2296 2297 <vlen>:= number -> number of lanes 2298 | x -> VLA (Vector Length Agnostic) 2299 2300 <parameters>:= v -> vector 2301 | l | l <number> -> linear 2302 | R | R <number> -> linear with ref modifier 2303 | L | L <number> -> linear with val modifier 2304 | U | U <number> -> linear with uval modifier 2305 | ls <pos> -> runtime linear 2306 | Rs <pos> -> runtime linear with ref modifier 2307 | Ls <pos> -> runtime linear with val modifier 2308 | Us <pos> -> runtime linear with uval modifier 2309 | u -> uniform 2310 2311 <scalar_name>:= name of the scalar function 2312 2313 <vector_redirection>:= optional, custom name of the vector function 2314 2315``preallocated(<ty>)`` 2316 This attribute is required on calls to ``llvm.call.preallocated.arg`` 2317 and cannot be used on any other call. See 2318 :ref:`llvm.call.preallocated.arg<int_call_preallocated_arg>` for more 2319 details. 2320 2321.. _glattrs: 2322 2323Global Attributes 2324----------------- 2325 2326Attributes may be set to communicate additional information about a global variable. 2327Unlike :ref:`function attributes <fnattrs>`, attributes on a global variable 2328are grouped into a single :ref:`attribute group <attrgrp>`. 2329 2330``no_sanitize_address`` 2331 This attribute indicates that the global variable should not have 2332 AddressSanitizer instrumentation applied to it, because it was annotated 2333 with `__attribute__((no_sanitize("address")))`, 2334 `__attribute__((disable_sanitizer_instrumentation))`, or included in the 2335 `-fsanitize-ignorelist` file. 2336``no_sanitize_hwaddress`` 2337 This attribute indicates that the global variable should not have 2338 HWAddressSanitizer instrumentation applied to it, because it was annotated 2339 with `__attribute__((no_sanitize("hwaddress")))`, 2340 `__attribute__((disable_sanitizer_instrumentation))`, or included in the 2341 `-fsanitize-ignorelist` file. 2342``sanitize_memtag`` 2343 This attribute indicates that the global variable should have AArch64 memory 2344 tags (MTE) instrumentation applied to it. This attribute causes the 2345 suppression of certain optimisations, like GlobalMerge, as well as ensuring 2346 extra directives are emitted in the assembly and extra bits of metadata are 2347 placed in the object file so that the linker can ensure the accesses are 2348 protected by MTE. This attribute is added by clang when 2349 `-fsanitize=memtag-globals` is provided, as long as the global is not marked 2350 with `__attribute__((no_sanitize("memtag")))`, 2351 `__attribute__((disable_sanitizer_instrumentation))`, or included in the 2352 `-fsanitize-ignorelist` file. The AArch64 Globals Tagging pass may remove 2353 this attribute when it's not possible to tag the global (e.g. it's a TLS 2354 variable). 2355``sanitize_address_dyninit`` 2356 This attribute indicates that the global variable, when instrumented with 2357 AddressSanitizer, should be checked for ODR violations. This attribute is 2358 applied to global variables that are dynamically initialized according to 2359 C++ rules. 2360 2361.. _opbundles: 2362 2363Operand Bundles 2364--------------- 2365 2366Operand bundles are tagged sets of SSA values that can be associated 2367with certain LLVM instructions (currently only ``call`` s and 2368``invoke`` s). In a way they are like metadata, but dropping them is 2369incorrect and will change program semantics. 2370 2371Syntax:: 2372 2373 operand bundle set ::= '[' operand bundle (, operand bundle )* ']' 2374 operand bundle ::= tag '(' [ bundle operand ] (, bundle operand )* ')' 2375 bundle operand ::= SSA value 2376 tag ::= string constant 2377 2378Operand bundles are **not** part of a function's signature, and a 2379given function may be called from multiple places with different kinds 2380of operand bundles. This reflects the fact that the operand bundles 2381are conceptually a part of the ``call`` (or ``invoke``), not the 2382callee being dispatched to. 2383 2384Operand bundles are a generic mechanism intended to support 2385runtime-introspection-like functionality for managed languages. While 2386the exact semantics of an operand bundle depend on the bundle tag, 2387there are certain limitations to how much the presence of an operand 2388bundle can influence the semantics of a program. These restrictions 2389are described as the semantics of an "unknown" operand bundle. As 2390long as the behavior of an operand bundle is describable within these 2391restrictions, LLVM does not need to have special knowledge of the 2392operand bundle to not miscompile programs containing it. 2393 2394- The bundle operands for an unknown operand bundle escape in unknown 2395 ways before control is transferred to the callee or invokee. 2396- Calls and invokes with operand bundles have unknown read / write 2397 effect on the heap on entry and exit (even if the call target is 2398 ``readnone`` or ``readonly``), unless they're overridden with 2399 callsite specific attributes. 2400- An operand bundle at a call site cannot change the implementation 2401 of the called function. Inter-procedural optimizations work as 2402 usual as long as they take into account the first two properties. 2403 2404More specific types of operand bundles are described below. 2405 2406.. _deopt_opbundles: 2407 2408Deoptimization Operand Bundles 2409^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2410 2411Deoptimization operand bundles are characterized by the ``"deopt"`` 2412operand bundle tag. These operand bundles represent an alternate 2413"safe" continuation for the call site they're attached to, and can be 2414used by a suitable runtime to deoptimize the compiled frame at the 2415specified call site. There can be at most one ``"deopt"`` operand 2416bundle attached to a call site. Exact details of deoptimization is 2417out of scope for the language reference, but it usually involves 2418rewriting a compiled frame into a set of interpreted frames. 2419 2420From the compiler's perspective, deoptimization operand bundles make 2421the call sites they're attached to at least ``readonly``. They read 2422through all of their pointer typed operands (even if they're not 2423otherwise escaped) and the entire visible heap. Deoptimization 2424operand bundles do not capture their operands except during 2425deoptimization, in which case control will not be returned to the 2426compiled frame. 2427 2428The inliner knows how to inline through calls that have deoptimization 2429operand bundles. Just like inlining through a normal call site 2430involves composing the normal and exceptional continuations, inlining 2431through a call site with a deoptimization operand bundle needs to 2432appropriately compose the "safe" deoptimization continuation. The 2433inliner does this by prepending the parent's deoptimization 2434continuation to every deoptimization continuation in the inlined body. 2435E.g. inlining ``@f`` into ``@g`` in the following example 2436 2437.. code-block:: llvm 2438 2439 define void @f() { 2440 call void @x() ;; no deopt state 2441 call void @y() [ "deopt"(i32 10) ] 2442 call void @y() [ "deopt"(i32 10), "unknown"(i8* null) ] 2443 ret void 2444 } 2445 2446 define void @g() { 2447 call void @f() [ "deopt"(i32 20) ] 2448 ret void 2449 } 2450 2451will result in 2452 2453.. code-block:: llvm 2454 2455 define void @g() { 2456 call void @x() ;; still no deopt state 2457 call void @y() [ "deopt"(i32 20, i32 10) ] 2458 call void @y() [ "deopt"(i32 20, i32 10), "unknown"(i8* null) ] 2459 ret void 2460 } 2461 2462It is the frontend's responsibility to structure or encode the 2463deoptimization state in a way that syntactically prepending the 2464caller's deoptimization state to the callee's deoptimization state is 2465semantically equivalent to composing the caller's deoptimization 2466continuation after the callee's deoptimization continuation. 2467 2468.. _ob_funclet: 2469 2470Funclet Operand Bundles 2471^^^^^^^^^^^^^^^^^^^^^^^ 2472 2473Funclet operand bundles are characterized by the ``"funclet"`` 2474operand bundle tag. These operand bundles indicate that a call site 2475is within a particular funclet. There can be at most one 2476``"funclet"`` operand bundle attached to a call site and it must have 2477exactly one bundle operand. 2478 2479If any funclet EH pads have been "entered" but not "exited" (per the 2480`description in the EH doc\ <ExceptionHandling.html#wineh-constraints>`_), 2481it is undefined behavior to execute a ``call`` or ``invoke`` which: 2482 2483* does not have a ``"funclet"`` bundle and is not a ``call`` to a nounwind 2484 intrinsic, or 2485* has a ``"funclet"`` bundle whose operand is not the most-recently-entered 2486 not-yet-exited funclet EH pad. 2487 2488Similarly, if no funclet EH pads have been entered-but-not-yet-exited, 2489executing a ``call`` or ``invoke`` with a ``"funclet"`` bundle is undefined behavior. 2490 2491GC Transition Operand Bundles 2492^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2493 2494GC transition operand bundles are characterized by the 2495``"gc-transition"`` operand bundle tag. These operand bundles mark a 2496call as a transition between a function with one GC strategy to a 2497function with a different GC strategy. If coordinating the transition 2498between GC strategies requires additional code generation at the call 2499site, these bundles may contain any values that are needed by the 2500generated code. For more details, see :ref:`GC Transitions 2501<gc_transition_args>`. 2502 2503The bundle contain an arbitrary list of Values which need to be passed 2504to GC transition code. They will be lowered and passed as operands to 2505the appropriate GC_TRANSITION nodes in the selection DAG. It is assumed 2506that these arguments must be available before and after (but not 2507necessarily during) the execution of the callee. 2508 2509.. _assume_opbundles: 2510 2511Assume Operand Bundles 2512^^^^^^^^^^^^^^^^^^^^^^ 2513 2514Operand bundles on an :ref:`llvm.assume <int_assume>` allows representing 2515assumptions that a :ref:`parameter attribute <paramattrs>` or a 2516:ref:`function attribute <fnattrs>` holds for a certain value at a certain 2517location. Operand bundles enable assumptions that are either hard or impossible 2518to represent as a boolean argument of an :ref:`llvm.assume <int_assume>`. 2519 2520An assume operand bundle has the form: 2521 2522:: 2523 2524 "<tag>"([ <holds for value> [, <attribute argument>] ]) 2525 2526* The tag of the operand bundle is usually the name of attribute that can be 2527 assumed to hold. It can also be `ignore`, this tag doesn't contain any 2528 information and should be ignored. 2529* The first argument if present is the value for which the attribute hold. 2530* The second argument if present is an argument of the attribute. 2531 2532If there are no arguments the attribute is a property of the call location. 2533 2534For example: 2535 2536.. code-block:: llvm 2537 2538 call void @llvm.assume(i1 true) ["align"(i32* %val, i32 8)] 2539 2540allows the optimizer to assume that at location of call to 2541:ref:`llvm.assume <int_assume>` ``%val`` has an alignment of at least 8. 2542 2543.. code-block:: llvm 2544 2545 call void @llvm.assume(i1 %cond) ["cold"(), "nonnull"(i64* %val)] 2546 2547allows the optimizer to assume that the :ref:`llvm.assume <int_assume>` 2548call location is cold and that ``%val`` may not be null. 2549 2550Just like for the argument of :ref:`llvm.assume <int_assume>`, if any of the 2551provided guarantees are violated at runtime the behavior is undefined. 2552 2553While attributes expect constant arguments, assume operand bundles may be 2554provided a dynamic value, for example: 2555 2556.. code-block:: llvm 2557 2558 call void @llvm.assume(i1 true) ["align"(i32* %val, i32 %align)] 2559 2560If the operand bundle value violates any requirements on the attribute value, 2561the behavior is undefined, unless one of the following exceptions applies: 2562 2563* ``"assume"`` operand bundles may specify a non-power-of-two alignment 2564 (including a zero alignment). If this is the case, then the pointer value 2565 must be a null pointer, otherwise the behavior is undefined. 2566 2567Even if the assumed property can be encoded as a boolean value, like 2568``nonnull``, using operand bundles to express the property can still have 2569benefits: 2570 2571* Attributes that can be expressed via operand bundles are directly the 2572 property that the optimizer uses and cares about. Encoding attributes as 2573 operand bundles removes the need for an instruction sequence that represents 2574 the property (e.g., `icmp ne i32* %p, null` for `nonnull`) and for the 2575 optimizer to deduce the property from that instruction sequence. 2576* Expressing the property using operand bundles makes it easy to identify the 2577 use of the value as a use in an :ref:`llvm.assume <int_assume>`. This then 2578 simplifies and improves heuristics, e.g., for use "use-sensitive" 2579 optimizations. 2580 2581.. _ob_preallocated: 2582 2583Preallocated Operand Bundles 2584^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2585 2586Preallocated operand bundles are characterized by the ``"preallocated"`` 2587operand bundle tag. These operand bundles allow separation of the allocation 2588of the call argument memory from the call site. This is necessary to pass 2589non-trivially copyable objects by value in a way that is compatible with MSVC 2590on some targets. There can be at most one ``"preallocated"`` operand bundle 2591attached to a call site and it must have exactly one bundle operand, which is 2592a token generated by ``@llvm.call.preallocated.setup``. A call with this 2593operand bundle should not adjust the stack before entering the function, as 2594that will have been done by one of the ``@llvm.call.preallocated.*`` intrinsics. 2595 2596.. code-block:: llvm 2597 2598 %foo = type { i64, i32 } 2599 2600 ... 2601 2602 %t = call token @llvm.call.preallocated.setup(i32 1) 2603 %a = call i8* @llvm.call.preallocated.arg(token %t, i32 0) preallocated(%foo) 2604 %b = bitcast i8* %a to %foo* 2605 ; initialize %b 2606 call void @bar(i32 42, %foo* preallocated(%foo) %b) ["preallocated"(token %t)] 2607 2608.. _ob_gc_live: 2609 2610GC Live Operand Bundles 2611^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2612 2613A "gc-live" operand bundle is only valid on a :ref:`gc.statepoint <gc_statepoint>` 2614intrinsic. The operand bundle must contain every pointer to a garbage collected 2615object which potentially needs to be updated by the garbage collector. 2616 2617When lowered, any relocated value will be recorded in the corresponding 2618:ref:`stackmap entry <statepoint-stackmap-format>`. See the intrinsic description 2619for further details. 2620 2621ObjC ARC Attached Call Operand Bundles 2622^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2623 2624A ``"clang.arc.attachedcall"`` operand bundle on a call indicates the call is 2625implicitly followed by a marker instruction and a call to an ObjC runtime 2626function that uses the result of the call. The operand bundle takes a mandatory 2627pointer to the runtime function (``@objc_retainAutoreleasedReturnValue`` or 2628``@objc_unsafeClaimAutoreleasedReturnValue``). 2629The return value of a call with this bundle is used by a call to 2630``@llvm.objc.clang.arc.noop.use`` unless the called function's return type is 2631void, in which case the operand bundle is ignored. 2632 2633.. code-block:: llvm 2634 2635 ; The marker instruction and a runtime function call are inserted after the call 2636 ; to @foo. 2637 call i8* @foo() [ "clang.arc.attachedcall"(i8* (i8*)* @objc_retainAutoreleasedReturnValue) ] 2638 call i8* @foo() [ "clang.arc.attachedcall"(i8* (i8*)* @objc_unsafeClaimAutoreleasedReturnValue) ] 2639 2640The operand bundle is needed to ensure the call is immediately followed by the 2641marker instruction and the ObjC runtime call in the final output. 2642 2643.. _ob_ptrauth: 2644 2645Pointer Authentication Operand Bundles 2646^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2647 2648Pointer Authentication operand bundles are characterized by the 2649``"ptrauth"`` operand bundle tag. They are described in the 2650`Pointer Authentication <PointerAuth.html#operand-bundle>`__ document. 2651 2652.. _moduleasm: 2653 2654Module-Level Inline Assembly 2655---------------------------- 2656 2657Modules may contain "module-level inline asm" blocks, which corresponds 2658to the GCC "file scope inline asm" blocks. These blocks are internally 2659concatenated by LLVM and treated as a single unit, but may be separated 2660in the ``.ll`` file if desired. The syntax is very simple: 2661 2662.. code-block:: llvm 2663 2664 module asm "inline asm code goes here" 2665 module asm "more can go here" 2666 2667The strings can contain any character by escaping non-printable 2668characters. The escape sequence used is simply "\\xx" where "xx" is the 2669two digit hex code for the number. 2670 2671Note that the assembly string *must* be parseable by LLVM's integrated assembler 2672(unless it is disabled), even when emitting a ``.s`` file. 2673 2674.. _langref_datalayout: 2675 2676Data Layout 2677----------- 2678 2679A module may specify a target specific data layout string that specifies 2680how data is to be laid out in memory. The syntax for the data layout is 2681simply: 2682 2683.. code-block:: llvm 2684 2685 target datalayout = "layout specification" 2686 2687The *layout specification* consists of a list of specifications 2688separated by the minus sign character ('-'). Each specification starts 2689with a letter and may include other information after the letter to 2690define some aspect of the data layout. The specifications accepted are 2691as follows: 2692 2693``E`` 2694 Specifies that the target lays out data in big-endian form. That is, 2695 the bits with the most significance have the lowest address 2696 location. 2697``e`` 2698 Specifies that the target lays out data in little-endian form. That 2699 is, the bits with the least significance have the lowest address 2700 location. 2701``S<size>`` 2702 Specifies the natural alignment of the stack in bits. Alignment 2703 promotion of stack variables is limited to the natural stack 2704 alignment to avoid dynamic stack realignment. The stack alignment 2705 must be a multiple of 8-bits. If omitted, the natural stack 2706 alignment defaults to "unspecified", which does not prevent any 2707 alignment promotions. 2708``P<address space>`` 2709 Specifies the address space that corresponds to program memory. 2710 Harvard architectures can use this to specify what space LLVM 2711 should place things such as functions into. If omitted, the 2712 program memory space defaults to the default address space of 0, 2713 which corresponds to a Von Neumann architecture that has code 2714 and data in the same space. 2715``G<address space>`` 2716 Specifies the address space to be used by default when creating global 2717 variables. If omitted, the globals address space defaults to the default 2718 address space 0. 2719 Note: variable declarations without an address space are always created in 2720 address space 0, this property only affects the default value to be used 2721 when creating globals without additional contextual information (e.g. in 2722 LLVM passes). 2723``A<address space>`` 2724 Specifies the address space of objects created by '``alloca``'. 2725 Defaults to the default address space of 0. 2726``p[n]:<size>:<abi>[:<pref>][:<idx>]`` 2727 This specifies the *size* of a pointer and its ``<abi>`` and 2728 ``<pref>``\erred alignments for address space ``n``. ``<pref>`` is optional 2729 and defaults to ``<abi>``. The fourth parameter ``<idx>`` is the size of the 2730 index that used for address calculation. If not 2731 specified, the default index size is equal to the pointer size. All sizes 2732 are in bits. The address space, ``n``, is optional, and if not specified, 2733 denotes the default address space 0. The value of ``n`` must be 2734 in the range [1,2^23). 2735``i<size>:<abi>[:<pref>]`` 2736 This specifies the alignment for an integer type of a given bit 2737 ``<size>``. The value of ``<size>`` must be in the range [1,2^23). 2738 ``<pref>`` is optional and defaults to ``<abi>``. 2739``v<size>:<abi>[:<pref>]`` 2740 This specifies the alignment for a vector type of a given bit 2741 ``<size>``. The value of ``<size>`` must be in the range [1,2^23). 2742 ``<pref>`` is optional and defaults to ``<abi>``. 2743``f<size>:<abi>[:<pref>]`` 2744 This specifies the alignment for a floating-point type of a given bit 2745 ``<size>``. Only values of ``<size>`` that are supported by the target 2746 will work. 32 (float) and 64 (double) are supported on all targets; 80 2747 or 128 (different flavors of long double) are also supported on some 2748 targets. The value of ``<size>`` must be in the range [1,2^23). 2749 ``<pref>`` is optional and defaults to ``<abi>``. 2750``a:<abi>[:<pref>]`` 2751 This specifies the alignment for an object of aggregate type. 2752 ``<pref>`` is optional and defaults to ``<abi>``. 2753``F<type><abi>`` 2754 This specifies the alignment for function pointers. 2755 The options for ``<type>`` are: 2756 2757 * ``i``: The alignment of function pointers is independent of the alignment 2758 of functions, and is a multiple of ``<abi>``. 2759 * ``n``: The alignment of function pointers is a multiple of the explicit 2760 alignment specified on the function, and is a multiple of ``<abi>``. 2761``m:<mangling>`` 2762 If present, specifies that llvm names are mangled in the output. Symbols 2763 prefixed with the mangling escape character ``\01`` are passed through 2764 directly to the assembler without the escape character. The mangling style 2765 options are 2766 2767 * ``e``: ELF mangling: Private symbols get a ``.L`` prefix. 2768 * ``l``: GOFF mangling: Private symbols get a ``@`` prefix. 2769 * ``m``: Mips mangling: Private symbols get a ``$`` prefix. 2770 * ``o``: Mach-O mangling: Private symbols get ``L`` prefix. Other 2771 symbols get a ``_`` prefix. 2772 * ``x``: Windows x86 COFF mangling: Private symbols get the usual prefix. 2773 Regular C symbols get a ``_`` prefix. Functions with ``__stdcall``, 2774 ``__fastcall``, and ``__vectorcall`` have custom mangling that appends 2775 ``@N`` where N is the number of bytes used to pass parameters. C++ symbols 2776 starting with ``?`` are not mangled in any way. 2777 * ``w``: Windows COFF mangling: Similar to ``x``, except that normal C 2778 symbols do not receive a ``_`` prefix. 2779 * ``a``: XCOFF mangling: Private symbols get a ``L..`` prefix. 2780``n<size1>:<size2>:<size3>...`` 2781 This specifies a set of native integer widths for the target CPU in 2782 bits. For example, it might contain ``n32`` for 32-bit PowerPC, 2783 ``n32:64`` for PowerPC 64, or ``n8:16:32:64`` for X86-64. Elements of 2784 this set are considered to support most general arithmetic operations 2785 efficiently. 2786``ni:<address space0>:<address space1>:<address space2>...`` 2787 This specifies pointer types with the specified address spaces 2788 as :ref:`Non-Integral Pointer Type <nointptrtype>` s. The ``0`` 2789 address space cannot be specified as non-integral. 2790 2791On every specification that takes a ``<abi>:<pref>``, specifying the 2792``<pref>`` alignment is optional. If omitted, the preceding ``:`` 2793should be omitted too and ``<pref>`` will be equal to ``<abi>``. 2794 2795When constructing the data layout for a given target, LLVM starts with a 2796default set of specifications which are then (possibly) overridden by 2797the specifications in the ``datalayout`` keyword. The default 2798specifications are given in this list: 2799 2800- ``e`` - little endian 2801- ``p:64:64:64`` - 64-bit pointers with 64-bit alignment. 2802- ``p[n]:64:64:64`` - Other address spaces are assumed to be the 2803 same as the default address space. 2804- ``S0`` - natural stack alignment is unspecified 2805- ``i1:8:8`` - i1 is 8-bit (byte) aligned 2806- ``i8:8:8`` - i8 is 8-bit (byte) aligned 2807- ``i16:16:16`` - i16 is 16-bit aligned 2808- ``i32:32:32`` - i32 is 32-bit aligned 2809- ``i64:32:64`` - i64 has ABI alignment of 32-bits but preferred 2810 alignment of 64-bits 2811- ``f16:16:16`` - half is 16-bit aligned 2812- ``f32:32:32`` - float is 32-bit aligned 2813- ``f64:64:64`` - double is 64-bit aligned 2814- ``f128:128:128`` - quad is 128-bit aligned 2815- ``v64:64:64`` - 64-bit vector is 64-bit aligned 2816- ``v128:128:128`` - 128-bit vector is 128-bit aligned 2817- ``a:0:64`` - aggregates are 64-bit aligned 2818 2819When LLVM is determining the alignment for a given type, it uses the 2820following rules: 2821 2822#. If the type sought is an exact match for one of the specifications, 2823 that specification is used. 2824#. If no match is found, and the type sought is an integer type, then 2825 the smallest integer type that is larger than the bitwidth of the 2826 sought type is used. If none of the specifications are larger than 2827 the bitwidth then the largest integer type is used. For example, 2828 given the default specifications above, the i7 type will use the 2829 alignment of i8 (next largest) while both i65 and i256 will use the 2830 alignment of i64 (largest specified). 2831 2832The function of the data layout string may not be what you expect. 2833Notably, this is not a specification from the frontend of what alignment 2834the code generator should use. 2835 2836Instead, if specified, the target data layout is required to match what 2837the ultimate *code generator* expects. This string is used by the 2838mid-level optimizers to improve code, and this only works if it matches 2839what the ultimate code generator uses. There is no way to generate IR 2840that does not embed this target-specific detail into the IR. If you 2841don't specify the string, the default specifications will be used to 2842generate a Data Layout and the optimization phases will operate 2843accordingly and introduce target specificity into the IR with respect to 2844these default specifications. 2845 2846.. _langref_triple: 2847 2848Target Triple 2849------------- 2850 2851A module may specify a target triple string that describes the target 2852host. The syntax for the target triple is simply: 2853 2854.. code-block:: llvm 2855 2856 target triple = "x86_64-apple-macosx10.7.0" 2857 2858The *target triple* string consists of a series of identifiers delimited 2859by the minus sign character ('-'). The canonical forms are: 2860 2861:: 2862 2863 ARCHITECTURE-VENDOR-OPERATING_SYSTEM 2864 ARCHITECTURE-VENDOR-OPERATING_SYSTEM-ENVIRONMENT 2865 2866This information is passed along to the backend so that it generates 2867code for the proper architecture. It's possible to override this on the 2868command line with the ``-mtriple`` command line option. 2869 2870.. _objectlifetime: 2871 2872Object Lifetime 2873---------------------- 2874 2875A memory object, or simply object, is a region of a memory space that is 2876reserved by a memory allocation such as :ref:`alloca <i_alloca>`, heap 2877allocation calls, and global variable definitions. 2878Once it is allocated, the bytes stored in the region can only be read or written 2879through a pointer that is :ref:`based on <pointeraliasing>` the allocation 2880value. 2881If a pointer that is not based on the object tries to read or write to the 2882object, it is undefined behavior. 2883 2884A lifetime of a memory object is a property that decides its accessibility. 2885Unless stated otherwise, a memory object is alive since its allocation, and 2886dead after its deallocation. 2887It is undefined behavior to access a memory object that isn't alive, but 2888operations that don't dereference it such as 2889:ref:`getelementptr <i_getelementptr>`, :ref:`ptrtoint <i_ptrtoint>` and 2890:ref:`icmp <i_icmp>` return a valid result. 2891This explains code motion of these instructions across operations that 2892impact the object's lifetime. 2893A stack object's lifetime can be explicitly specified using 2894:ref:`llvm.lifetime.start <int_lifestart>` and 2895:ref:`llvm.lifetime.end <int_lifeend>` intrinsic function calls. 2896 2897.. _pointeraliasing: 2898 2899Pointer Aliasing Rules 2900---------------------- 2901 2902Any memory access must be done through a pointer value associated with 2903an address range of the memory access, otherwise the behavior is 2904undefined. Pointer values are associated with address ranges according 2905to the following rules: 2906 2907- A pointer value is associated with the addresses associated with any 2908 value it is *based* on. 2909- An address of a global variable is associated with the address range 2910 of the variable's storage. 2911- The result value of an allocation instruction is associated with the 2912 address range of the allocated storage. 2913- A null pointer in the default address-space is associated with no 2914 address. 2915- An :ref:`undef value <undefvalues>` in *any* address-space is 2916 associated with no address. 2917- An integer constant other than zero or a pointer value returned from 2918 a function not defined within LLVM may be associated with address 2919 ranges allocated through mechanisms other than those provided by 2920 LLVM. Such ranges shall not overlap with any ranges of addresses 2921 allocated by mechanisms provided by LLVM. 2922 2923A pointer value is *based* on another pointer value according to the 2924following rules: 2925 2926- A pointer value formed from a scalar ``getelementptr`` operation is *based* on 2927 the pointer-typed operand of the ``getelementptr``. 2928- The pointer in lane *l* of the result of a vector ``getelementptr`` operation 2929 is *based* on the pointer in lane *l* of the vector-of-pointers-typed operand 2930 of the ``getelementptr``. 2931- The result value of a ``bitcast`` is *based* on the operand of the 2932 ``bitcast``. 2933- A pointer value formed by an ``inttoptr`` is *based* on all pointer 2934 values that contribute (directly or indirectly) to the computation of 2935 the pointer's value. 2936- The "*based* on" relationship is transitive. 2937 2938Note that this definition of *"based"* is intentionally similar to the 2939definition of *"based"* in C99, though it is slightly weaker. 2940 2941LLVM IR does not associate types with memory. The result type of a 2942``load`` merely indicates the size and alignment of the memory from 2943which to load, as well as the interpretation of the value. The first 2944operand type of a ``store`` similarly only indicates the size and 2945alignment of the store. 2946 2947Consequently, type-based alias analysis, aka TBAA, aka 2948``-fstrict-aliasing``, is not applicable to general unadorned LLVM IR. 2949:ref:`Metadata <metadata>` may be used to encode additional information 2950which specialized optimization passes may use to implement type-based 2951alias analysis. 2952 2953.. _pointercapture: 2954 2955Pointer Capture 2956--------------- 2957 2958Given a function call and a pointer that is passed as an argument or stored in 2959the memory before the call, a pointer is *captured* by the call if it makes a 2960copy of any part of the pointer that outlives the call. 2961To be precise, a pointer is captured if one or more of the following conditions 2962hold: 2963 29641. The call stores any bit of the pointer carrying information into a place, 2965 and the stored bits can be read from the place by the caller after this call 2966 exits. 2967 2968.. code-block:: llvm 2969 2970 @glb = global i8* null 2971 @glb2 = global i8* null 2972 @glb3 = global i8* null 2973 @glbi = global i32 0 2974 2975 define i8* @f(i8* %a, i8* %b, i8* %c, i8* %d, i8* %e) { 2976 store i8* %a, i8** @glb ; %a is captured by this call 2977 2978 store i8* %b, i8** @glb2 ; %b isn't captured because the stored value is overwritten by the store below 2979 store i8* null, i8** @glb2 2980 2981 store i8* %c, i8** @glb3 2982 call void @g() ; If @g makes a copy of %c that outlives this call (@f), %c is captured 2983 store i8* null, i8** @glb3 2984 2985 %i = ptrtoint i8* %d to i64 2986 %j = trunc i64 %i to i32 2987 store i32 %j, i32* @glbi ; %d is captured 2988 2989 ret i8* %e ; %e is captured 2990 } 2991 29922. The call stores any bit of the pointer carrying information into a place, 2993 and the stored bits can be safely read from the place by another thread via 2994 synchronization. 2995 2996.. code-block:: llvm 2997 2998 @lock = global i1 true 2999 3000 define void @f(i8* %a) { 3001 store i8* %a, i8** @glb 3002 store atomic i1 false, i1* @lock release ; %a is captured because another thread can safely read @glb 3003 store i8* null, i8** @glb 3004 ret void 3005 } 3006 30073. The call's behavior depends on any bit of the pointer carrying information. 3008 3009.. code-block:: llvm 3010 3011 @glb = global i8 0 3012 3013 define void @f(i8* %a) { 3014 %c = icmp eq i8* %a, @glb 3015 br i1 %c, label %BB_EXIT, label %BB_CONTINUE ; escapes %a 3016 BB_EXIT: 3017 call void @exit() 3018 unreachable 3019 BB_CONTINUE: 3020 ret void 3021 } 3022 30234. The pointer is used in a volatile access as its address. 3024 3025 3026.. _volatile: 3027 3028Volatile Memory Accesses 3029------------------------ 3030 3031Certain memory accesses, such as :ref:`load <i_load>`'s, 3032:ref:`store <i_store>`'s, and :ref:`llvm.memcpy <int_memcpy>`'s may be 3033marked ``volatile``. The optimizers must not change the number of 3034volatile operations or change their order of execution relative to other 3035volatile operations. The optimizers *may* change the order of volatile 3036operations relative to non-volatile operations. This is not Java's 3037"volatile" and has no cross-thread synchronization behavior. 3038 3039A volatile load or store may have additional target-specific semantics. 3040Any volatile operation can have side effects, and any volatile operation 3041can read and/or modify state which is not accessible via a regular load 3042or store in this module. Volatile operations may use addresses which do 3043not point to memory (like MMIO registers). This means the compiler may 3044not use a volatile operation to prove a non-volatile access to that 3045address has defined behavior. 3046 3047The allowed side-effects for volatile accesses are limited. If a 3048non-volatile store to a given address would be legal, a volatile 3049operation may modify the memory at that address. A volatile operation 3050may not modify any other memory accessible by the module being compiled. 3051A volatile operation may not call any code in the current module. 3052 3053The compiler may assume execution will continue after a volatile operation, 3054so operations which modify memory or may have undefined behavior can be 3055hoisted past a volatile operation. 3056 3057As an exception to the preceding rule, the compiler may not assume execution 3058will continue after a volatile store operation. This restriction is necessary 3059to support the somewhat common pattern in C of intentionally storing to an 3060invalid pointer to crash the program. In the future, it might make sense to 3061allow frontends to control this behavior. 3062 3063IR-level volatile loads and stores cannot safely be optimized into llvm.memcpy 3064or llvm.memmove intrinsics even when those intrinsics are flagged volatile. 3065Likewise, the backend should never split or merge target-legal volatile 3066load/store instructions. Similarly, IR-level volatile loads and stores cannot 3067change from integer to floating-point or vice versa. 3068 3069.. admonition:: Rationale 3070 3071 Platforms may rely on volatile loads and stores of natively supported 3072 data width to be executed as single instruction. For example, in C 3073 this holds for an l-value of volatile primitive type with native 3074 hardware support, but not necessarily for aggregate types. The 3075 frontend upholds these expectations, which are intentionally 3076 unspecified in the IR. The rules above ensure that IR transformations 3077 do not violate the frontend's contract with the language. 3078 3079.. _memmodel: 3080 3081Memory Model for Concurrent Operations 3082-------------------------------------- 3083 3084The LLVM IR does not define any way to start parallel threads of 3085execution or to register signal handlers. Nonetheless, there are 3086platform-specific ways to create them, and we define LLVM IR's behavior 3087in their presence. This model is inspired by the C++0x memory model. 3088 3089For a more informal introduction to this model, see the :doc:`Atomics`. 3090 3091We define a *happens-before* partial order as the least partial order 3092that 3093 3094- Is a superset of single-thread program order, and 3095- When a *synchronizes-with* ``b``, includes an edge from ``a`` to 3096 ``b``. *Synchronizes-with* pairs are introduced by platform-specific 3097 techniques, like pthread locks, thread creation, thread joining, 3098 etc., and by atomic instructions. (See also :ref:`Atomic Memory Ordering 3099 Constraints <ordering>`). 3100 3101Note that program order does not introduce *happens-before* edges 3102between a thread and signals executing inside that thread. 3103 3104Every (defined) read operation (load instructions, memcpy, atomic 3105loads/read-modify-writes, etc.) R reads a series of bytes written by 3106(defined) write operations (store instructions, atomic 3107stores/read-modify-writes, memcpy, etc.). For the purposes of this 3108section, initialized globals are considered to have a write of the 3109initializer which is atomic and happens before any other read or write 3110of the memory in question. For each byte of a read R, R\ :sub:`byte` 3111may see any write to the same byte, except: 3112 3113- If write\ :sub:`1` happens before write\ :sub:`2`, and 3114 write\ :sub:`2` happens before R\ :sub:`byte`, then 3115 R\ :sub:`byte` does not see write\ :sub:`1`. 3116- If R\ :sub:`byte` happens before write\ :sub:`3`, then 3117 R\ :sub:`byte` does not see write\ :sub:`3`. 3118 3119Given that definition, R\ :sub:`byte` is defined as follows: 3120 3121- If R is volatile, the result is target-dependent. (Volatile is 3122 supposed to give guarantees which can support ``sig_atomic_t`` in 3123 C/C++, and may be used for accesses to addresses that do not behave 3124 like normal memory. It does not generally provide cross-thread 3125 synchronization.) 3126- Otherwise, if there is no write to the same byte that happens before 3127 R\ :sub:`byte`, R\ :sub:`byte` returns ``undef`` for that byte. 3128- Otherwise, if R\ :sub:`byte` may see exactly one write, 3129 R\ :sub:`byte` returns the value written by that write. 3130- Otherwise, if R is atomic, and all the writes R\ :sub:`byte` may 3131 see are atomic, it chooses one of the values written. See the :ref:`Atomic 3132 Memory Ordering Constraints <ordering>` section for additional 3133 constraints on how the choice is made. 3134- Otherwise R\ :sub:`byte` returns ``undef``. 3135 3136R returns the value composed of the series of bytes it read. This 3137implies that some bytes within the value may be ``undef`` **without** 3138the entire value being ``undef``. Note that this only defines the 3139semantics of the operation; it doesn't mean that targets will emit more 3140than one instruction to read the series of bytes. 3141 3142Note that in cases where none of the atomic intrinsics are used, this 3143model places only one restriction on IR transformations on top of what 3144is required for single-threaded execution: introducing a store to a byte 3145which might not otherwise be stored is not allowed in general. 3146(Specifically, in the case where another thread might write to and read 3147from an address, introducing a store can change a load that may see 3148exactly one write into a load that may see multiple writes.) 3149 3150.. _ordering: 3151 3152Atomic Memory Ordering Constraints 3153---------------------------------- 3154 3155Atomic instructions (:ref:`cmpxchg <i_cmpxchg>`, 3156:ref:`atomicrmw <i_atomicrmw>`, :ref:`fence <i_fence>`, 3157:ref:`atomic load <i_load>`, and :ref:`atomic store <i_store>`) take 3158ordering parameters that determine which other atomic instructions on 3159the same address they *synchronize with*. These semantics are borrowed 3160from Java and C++0x, but are somewhat more colloquial. If these 3161descriptions aren't precise enough, check those specs (see spec 3162references in the :doc:`atomics guide <Atomics>`). 3163:ref:`fence <i_fence>` instructions treat these orderings somewhat 3164differently since they don't take an address. See that instruction's 3165documentation for details. 3166 3167For a simpler introduction to the ordering constraints, see the 3168:doc:`Atomics`. 3169 3170``unordered`` 3171 The set of values that can be read is governed by the happens-before 3172 partial order. A value cannot be read unless some operation wrote 3173 it. This is intended to provide a guarantee strong enough to model 3174 Java's non-volatile shared variables. This ordering cannot be 3175 specified for read-modify-write operations; it is not strong enough 3176 to make them atomic in any interesting way. 3177``monotonic`` 3178 In addition to the guarantees of ``unordered``, there is a single 3179 total order for modifications by ``monotonic`` operations on each 3180 address. All modification orders must be compatible with the 3181 happens-before order. There is no guarantee that the modification 3182 orders can be combined to a global total order for the whole program 3183 (and this often will not be possible). The read in an atomic 3184 read-modify-write operation (:ref:`cmpxchg <i_cmpxchg>` and 3185 :ref:`atomicrmw <i_atomicrmw>`) reads the value in the modification 3186 order immediately before the value it writes. If one atomic read 3187 happens before another atomic read of the same address, the later 3188 read must see the same value or a later value in the address's 3189 modification order. This disallows reordering of ``monotonic`` (or 3190 stronger) operations on the same address. If an address is written 3191 ``monotonic``-ally by one thread, and other threads ``monotonic``-ally 3192 read that address repeatedly, the other threads must eventually see 3193 the write. This corresponds to the C++0x/C1x 3194 ``memory_order_relaxed``. 3195``acquire`` 3196 In addition to the guarantees of ``monotonic``, a 3197 *synchronizes-with* edge may be formed with a ``release`` operation. 3198 This is intended to model C++'s ``memory_order_acquire``. 3199``release`` 3200 In addition to the guarantees of ``monotonic``, if this operation 3201 writes a value which is subsequently read by an ``acquire`` 3202 operation, it *synchronizes-with* that operation. (This isn't a 3203 complete description; see the C++0x definition of a release 3204 sequence.) This corresponds to the C++0x/C1x 3205 ``memory_order_release``. 3206``acq_rel`` (acquire+release) 3207 Acts as both an ``acquire`` and ``release`` operation on its 3208 address. This corresponds to the C++0x/C1x ``memory_order_acq_rel``. 3209``seq_cst`` (sequentially consistent) 3210 In addition to the guarantees of ``acq_rel`` (``acquire`` for an 3211 operation that only reads, ``release`` for an operation that only 3212 writes), there is a global total order on all 3213 sequentially-consistent operations on all addresses, which is 3214 consistent with the *happens-before* partial order and with the 3215 modification orders of all the affected addresses. Each 3216 sequentially-consistent read sees the last preceding write to the 3217 same address in this global order. This corresponds to the C++0x/C1x 3218 ``memory_order_seq_cst`` and Java volatile. 3219 3220.. _syncscope: 3221 3222If an atomic operation is marked ``syncscope("singlethread")``, it only 3223*synchronizes with* and only participates in the seq\_cst total orderings of 3224other operations running in the same thread (for example, in signal handlers). 3225 3226If an atomic operation is marked ``syncscope("<target-scope>")``, where 3227``<target-scope>`` is a target specific synchronization scope, then it is target 3228dependent if it *synchronizes with* and participates in the seq\_cst total 3229orderings of other operations. 3230 3231Otherwise, an atomic operation that is not marked ``syncscope("singlethread")`` 3232or ``syncscope("<target-scope>")`` *synchronizes with* and participates in the 3233seq\_cst total orderings of other operations that are not marked 3234``syncscope("singlethread")`` or ``syncscope("<target-scope>")``. 3235 3236.. _floatenv: 3237 3238Floating-Point Environment 3239-------------------------- 3240 3241The default LLVM floating-point environment assumes that floating-point 3242instructions do not have side effects. Results assume the round-to-nearest 3243rounding mode. No floating-point exception state is maintained in this 3244environment. Therefore, there is no attempt to create or preserve invalid 3245operation (SNaN) or division-by-zero exceptions. 3246 3247The benefit of this exception-free assumption is that floating-point 3248operations may be speculated freely without any other fast-math relaxations 3249to the floating-point model. 3250 3251Code that requires different behavior than this should use the 3252:ref:`Constrained Floating-Point Intrinsics <constrainedfp>`. 3253 3254.. _fastmath: 3255 3256Fast-Math Flags 3257--------------- 3258 3259LLVM IR floating-point operations (:ref:`fneg <i_fneg>`, :ref:`fadd <i_fadd>`, 3260:ref:`fsub <i_fsub>`, :ref:`fmul <i_fmul>`, :ref:`fdiv <i_fdiv>`, 3261:ref:`frem <i_frem>`, :ref:`fcmp <i_fcmp>`), :ref:`phi <i_phi>`, 3262:ref:`select <i_select>` and :ref:`call <i_call>` 3263may use the following flags to enable otherwise unsafe 3264floating-point transformations. 3265 3266``nnan`` 3267 No NaNs - Allow optimizations to assume the arguments and result are not 3268 NaN. If an argument is a nan, or the result would be a nan, it produces 3269 a :ref:`poison value <poisonvalues>` instead. 3270 3271``ninf`` 3272 No Infs - Allow optimizations to assume the arguments and result are not 3273 +/-Inf. If an argument is +/-Inf, or the result would be +/-Inf, it 3274 produces a :ref:`poison value <poisonvalues>` instead. 3275 3276``nsz`` 3277 No Signed Zeros - Allow optimizations to treat the sign of a zero 3278 argument or result as insignificant. This does not imply that -0.0 3279 is poison and/or guaranteed to not exist in the operation. 3280 3281``arcp`` 3282 Allow Reciprocal - Allow optimizations to use the reciprocal of an 3283 argument rather than perform division. 3284 3285``contract`` 3286 Allow floating-point contraction (e.g. fusing a multiply followed by an 3287 addition into a fused multiply-and-add). This does not enable reassociating 3288 to form arbitrary contractions. For example, ``(a*b) + (c*d) + e`` can not 3289 be transformed into ``(a*b) + ((c*d) + e)`` to create two fma operations. 3290 3291``afn`` 3292 Approximate functions - Allow substitution of approximate calculations for 3293 functions (sin, log, sqrt, etc). See floating-point intrinsic definitions 3294 for places where this can apply to LLVM's intrinsic math functions. 3295 3296``reassoc`` 3297 Allow reassociation transformations for floating-point instructions. 3298 This may dramatically change results in floating-point. 3299 3300``fast`` 3301 This flag implies all of the others. 3302 3303.. _uselistorder: 3304 3305Use-list Order Directives 3306------------------------- 3307 3308Use-list directives encode the in-memory order of each use-list, allowing the 3309order to be recreated. ``<order-indexes>`` is a comma-separated list of 3310indexes that are assigned to the referenced value's uses. The referenced 3311value's use-list is immediately sorted by these indexes. 3312 3313Use-list directives may appear at function scope or global scope. They are not 3314instructions, and have no effect on the semantics of the IR. When they're at 3315function scope, they must appear after the terminator of the final basic block. 3316 3317If basic blocks have their address taken via ``blockaddress()`` expressions, 3318``uselistorder_bb`` can be used to reorder their use-lists from outside their 3319function's scope. 3320 3321:Syntax: 3322 3323:: 3324 3325 uselistorder <ty> <value>, { <order-indexes> } 3326 uselistorder_bb @function, %block { <order-indexes> } 3327 3328:Examples: 3329 3330:: 3331 3332 define void @foo(i32 %arg1, i32 %arg2) { 3333 entry: 3334 ; ... instructions ... 3335 bb: 3336 ; ... instructions ... 3337 3338 ; At function scope. 3339 uselistorder i32 %arg1, { 1, 0, 2 } 3340 uselistorder label %bb, { 1, 0 } 3341 } 3342 3343 ; At global scope. 3344 uselistorder i32* @global, { 1, 2, 0 } 3345 uselistorder i32 7, { 1, 0 } 3346 uselistorder i32 (i32) @bar, { 1, 0 } 3347 uselistorder_bb @foo, %bb, { 5, 1, 3, 2, 0, 4 } 3348 3349.. _source_filename: 3350 3351Source Filename 3352--------------- 3353 3354The *source filename* string is set to the original module identifier, 3355which will be the name of the compiled source file when compiling from 3356source through the clang front end, for example. It is then preserved through 3357the IR and bitcode. 3358 3359This is currently necessary to generate a consistent unique global 3360identifier for local functions used in profile data, which prepends the 3361source file name to the local function name. 3362 3363The syntax for the source file name is simply: 3364 3365.. code-block:: text 3366 3367 source_filename = "/path/to/source.c" 3368 3369.. _typesystem: 3370 3371Type System 3372=========== 3373 3374The LLVM type system is one of the most important features of the 3375intermediate representation. Being typed enables a number of 3376optimizations to be performed on the intermediate representation 3377directly, without having to do extra analyses on the side before the 3378transformation. A strong type system makes it easier to read the 3379generated code and enables novel analyses and transformations that are 3380not feasible to perform on normal three address code representations. 3381 3382.. _t_void: 3383 3384Void Type 3385--------- 3386 3387:Overview: 3388 3389 3390The void type does not represent any value and has no size. 3391 3392:Syntax: 3393 3394 3395:: 3396 3397 void 3398 3399 3400.. _t_function: 3401 3402Function Type 3403------------- 3404 3405:Overview: 3406 3407 3408The function type can be thought of as a function signature. It consists of a 3409return type and a list of formal parameter types. The return type of a function 3410type is a void type or first class type --- except for :ref:`label <t_label>` 3411and :ref:`metadata <t_metadata>` types. 3412 3413:Syntax: 3414 3415:: 3416 3417 <returntype> (<parameter list>) 3418 3419...where '``<parameter list>``' is a comma-separated list of type 3420specifiers. Optionally, the parameter list may include a type ``...``, which 3421indicates that the function takes a variable number of arguments. Variable 3422argument functions can access their arguments with the :ref:`variable argument 3423handling intrinsic <int_varargs>` functions. '``<returntype>``' is any type 3424except :ref:`label <t_label>` and :ref:`metadata <t_metadata>`. 3425 3426:Examples: 3427 3428+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3429| ``i32 (i32)`` | function taking an ``i32``, returning an ``i32`` | 3430+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3431| ``float (i16, i32 *) *`` | :ref:`Pointer <t_pointer>` to a function that takes an ``i16`` and a :ref:`pointer <t_pointer>` to ``i32``, returning ``float``. | 3432+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3433| ``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. | 3434+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3435| ``{i32, i32} (i32)`` | A function taking an ``i32``, returning a :ref:`structure <t_struct>` containing two ``i32`` values | 3436+---------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3437 3438.. _t_firstclass: 3439 3440First Class Types 3441----------------- 3442 3443The :ref:`first class <t_firstclass>` types are perhaps the most important. 3444Values of these types are the only ones which can be produced by 3445instructions. 3446 3447.. _t_single_value: 3448 3449Single Value Types 3450^^^^^^^^^^^^^^^^^^ 3451 3452These are the types that are valid in registers from CodeGen's perspective. 3453 3454.. _t_integer: 3455 3456Integer Type 3457"""""""""""" 3458 3459:Overview: 3460 3461The integer type is a very simple type that simply specifies an 3462arbitrary bit width for the integer type desired. Any bit width from 1 3463bit to 2\ :sup:`23`\ (about 8 million) can be specified. 3464 3465:Syntax: 3466 3467:: 3468 3469 iN 3470 3471The number of bits the integer will occupy is specified by the ``N`` 3472value. 3473 3474Examples: 3475********* 3476 3477+----------------+------------------------------------------------+ 3478| ``i1`` | a single-bit integer. | 3479+----------------+------------------------------------------------+ 3480| ``i32`` | a 32-bit integer. | 3481+----------------+------------------------------------------------+ 3482| ``i1942652`` | a really big integer of over 1 million bits. | 3483+----------------+------------------------------------------------+ 3484 3485.. _t_floating: 3486 3487Floating-Point Types 3488"""""""""""""""""""" 3489 3490.. list-table:: 3491 :header-rows: 1 3492 3493 * - Type 3494 - Description 3495 3496 * - ``half`` 3497 - 16-bit floating-point value 3498 3499 * - ``bfloat`` 3500 - 16-bit "brain" floating-point value (7-bit significand). Provides the 3501 same number of exponent bits as ``float``, so that it matches its dynamic 3502 range, but with greatly reduced precision. Used in Intel's AVX-512 BF16 3503 extensions and Arm's ARMv8.6-A extensions, among others. 3504 3505 * - ``float`` 3506 - 32-bit floating-point value 3507 3508 * - ``double`` 3509 - 64-bit floating-point value 3510 3511 * - ``fp128`` 3512 - 128-bit floating-point value (113-bit significand) 3513 3514 * - ``x86_fp80`` 3515 - 80-bit floating-point value (X87) 3516 3517 * - ``ppc_fp128`` 3518 - 128-bit floating-point value (two 64-bits) 3519 3520The binary format of half, float, double, and fp128 correspond to the 3521IEEE-754-2008 specifications for binary16, binary32, binary64, and binary128 3522respectively. 3523 3524X86_amx Type 3525"""""""""""" 3526 3527:Overview: 3528 3529The x86_amx type represents a value held in an AMX tile register on an x86 3530machine. The operations allowed on it are quite limited. Only few intrinsics 3531are allowed: stride load and store, zero and dot product. No instruction is 3532allowed for this type. There are no arguments, arrays, pointers, vectors 3533or constants of this type. 3534 3535:Syntax: 3536 3537:: 3538 3539 x86_amx 3540 3541 3542X86_mmx Type 3543"""""""""""" 3544 3545:Overview: 3546 3547The x86_mmx type represents a value held in an MMX register on an x86 3548machine. The operations allowed on it are quite limited: parameters and 3549return values, load and store, and bitcast. User-specified MMX 3550instructions are represented as intrinsic or asm calls with arguments 3551and/or results of this type. There are no arrays, vectors or constants 3552of this type. 3553 3554:Syntax: 3555 3556:: 3557 3558 x86_mmx 3559 3560 3561.. _t_pointer: 3562 3563Pointer Type 3564"""""""""""" 3565 3566:Overview: 3567 3568The pointer type is used to specify memory locations. Pointers are 3569commonly used to reference objects in memory. 3570 3571Pointer types may have an optional address space attribute defining the 3572numbered address space where the pointed-to object resides. The default 3573address space is number zero. The semantics of non-zero address spaces 3574are target-specific. 3575 3576Note that LLVM does not permit pointers to void (``void*``) nor does it 3577permit pointers to labels (``label*``). Use ``i8*`` instead. 3578 3579LLVM is in the process of transitioning to 3580`opaque pointers <OpaquePointers.html#opaque-pointers>`_. 3581Opaque pointers do not have a pointee type. Rather, instructions 3582interacting through pointers specify the type of the underlying memory 3583they are interacting with. Opaque pointers are still in the process of 3584being worked on and are not complete. 3585 3586:Syntax: 3587 3588:: 3589 3590 <type> * 3591 ptr 3592 3593:Examples: 3594 3595+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3596| ``[4 x i32]*`` | A :ref:`pointer <t_pointer>` to :ref:`array <t_array>` of four ``i32`` values. | 3597+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3598| ``i32 (i32*) *`` | A :ref:`pointer <t_pointer>` to a :ref:`function <t_function>` that takes an ``i32*``, returning an ``i32``. | 3599+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3600| ``i32 addrspace(5)*`` | A :ref:`pointer <t_pointer>` to an ``i32`` value that resides in address space 5. | 3601+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3602| ``ptr`` | An opaque pointer type to a value that resides in address space 0. | 3603+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3604| ``ptr addrspace(5)`` | An opaque pointer type to a value that resides in address space 5. | 3605+-------------------------+--------------------------------------------------------------------------------------------------------------+ 3606 3607.. _t_vector: 3608 3609Vector Type 3610""""""""""" 3611 3612:Overview: 3613 3614A vector type is a simple derived type that represents a vector of 3615elements. Vector types are used when multiple primitive data are 3616operated in parallel using a single instruction (SIMD). A vector type 3617requires a size (number of elements), an underlying primitive data type, 3618and a scalable property to represent vectors where the exact hardware 3619vector length is unknown at compile time. Vector types are considered 3620:ref:`first class <t_firstclass>`. 3621 3622:Memory Layout: 3623 3624In general vector elements are laid out in memory in the same way as 3625:ref:`array types <t_array>`. Such an analogy works fine as long as the vector 3626elements are byte sized. However, when the elements of the vector aren't byte 3627sized it gets a bit more complicated. One way to describe the layout is by 3628describing what happens when a vector such as <N x iM> is bitcasted to an 3629integer type with N*M bits, and then following the rules for storing such an 3630integer to memory. 3631 3632A bitcast from a vector type to a scalar integer type will see the elements 3633being packed together (without padding). The order in which elements are 3634inserted in the integer depends on endianess. For little endian element zero 3635is put in the least significant bits of the integer, and for big endian 3636element zero is put in the most significant bits. 3637 3638Using a vector such as ``<i4 1, i4 2, i4 3, i4 5>`` as an example, together 3639with the analogy that we can replace a vector store by a bitcast followed by 3640an integer store, we get this for big endian: 3641 3642.. code-block:: llvm 3643 3644 %val = bitcast <4 x i4> <i4 1, i4 2, i4 3, i4 5> to i16 3645 3646 ; Bitcasting from a vector to an integral type can be seen as 3647 ; concatenating the values: 3648 ; %val now has the hexadecimal value 0x1235. 3649 3650 store i16 %val, i16* %ptr 3651 3652 ; In memory the content will be (8-bit addressing): 3653 ; 3654 ; [%ptr + 0]: 00010010 (0x12) 3655 ; [%ptr + 1]: 00110101 (0x35) 3656 3657The same example for little endian: 3658 3659.. code-block:: llvm 3660 3661 %val = bitcast <4 x i4> <i4 1, i4 2, i4 3, i4 5> to i16 3662 3663 ; Bitcasting from a vector to an integral type can be seen as 3664 ; concatenating the values: 3665 ; %val now has the hexadecimal value 0x5321. 3666 3667 store i16 %val, i16* %ptr 3668 3669 ; In memory the content will be (8-bit addressing): 3670 ; 3671 ; [%ptr + 0]: 01010011 (0x53) 3672 ; [%ptr + 1]: 00100001 (0x21) 3673 3674When ``<N*M>`` isn't evenly divisible by the byte size the exact memory layout 3675is unspecified (just like it is for an integral type of the same size). This 3676is because different targets could put the padding at different positions when 3677the type size is smaller than the type's store size. 3678 3679:Syntax: 3680 3681:: 3682 3683 < <# elements> x <elementtype> > ; Fixed-length vector 3684 < vscale x <# elements> x <elementtype> > ; Scalable vector 3685 3686The number of elements is a constant integer value larger than 0; 3687elementtype may be any integer, floating-point or pointer type. Vectors 3688of size zero are not allowed. For scalable vectors, the total number of 3689elements is a constant multiple (called vscale) of the specified number 3690of elements; vscale is a positive integer that is unknown at compile time 3691and the same hardware-dependent constant for all scalable vectors at run 3692time. The size of a specific scalable vector type is thus constant within 3693IR, even if the exact size in bytes cannot be determined until run time. 3694 3695:Examples: 3696 3697+------------------------+----------------------------------------------------+ 3698| ``<4 x i32>`` | Vector of 4 32-bit integer values. | 3699+------------------------+----------------------------------------------------+ 3700| ``<8 x float>`` | Vector of 8 32-bit floating-point values. | 3701+------------------------+----------------------------------------------------+ 3702| ``<2 x i64>`` | Vector of 2 64-bit integer values. | 3703+------------------------+----------------------------------------------------+ 3704| ``<4 x i64*>`` | Vector of 4 pointers to 64-bit integer values. | 3705+------------------------+----------------------------------------------------+ 3706| ``<vscale x 4 x i32>`` | Vector with a multiple of 4 32-bit integer values. | 3707+------------------------+----------------------------------------------------+ 3708 3709.. _t_label: 3710 3711Label Type 3712^^^^^^^^^^ 3713 3714:Overview: 3715 3716The label type represents code labels. 3717 3718:Syntax: 3719 3720:: 3721 3722 label 3723 3724.. _t_token: 3725 3726Token Type 3727^^^^^^^^^^ 3728 3729:Overview: 3730 3731The token type is used when a value is associated with an instruction 3732but all uses of the value must not attempt to introspect or obscure it. 3733As such, it is not appropriate to have a :ref:`phi <i_phi>` or 3734:ref:`select <i_select>` of type token. 3735 3736:Syntax: 3737 3738:: 3739 3740 token 3741 3742 3743 3744.. _t_metadata: 3745 3746Metadata Type 3747^^^^^^^^^^^^^ 3748 3749:Overview: 3750 3751The metadata type represents embedded metadata. No derived types may be 3752created from metadata except for :ref:`function <t_function>` arguments. 3753 3754:Syntax: 3755 3756:: 3757 3758 metadata 3759 3760.. _t_aggregate: 3761 3762Aggregate Types 3763^^^^^^^^^^^^^^^ 3764 3765Aggregate Types are a subset of derived types that can contain multiple 3766member types. :ref:`Arrays <t_array>` and :ref:`structs <t_struct>` are 3767aggregate types. :ref:`Vectors <t_vector>` are not considered to be 3768aggregate types. 3769 3770.. _t_array: 3771 3772Array Type 3773"""""""""" 3774 3775:Overview: 3776 3777The array type is a very simple derived type that arranges elements 3778sequentially in memory. The array type requires a size (number of 3779elements) and an underlying data type. 3780 3781:Syntax: 3782 3783:: 3784 3785 [<# elements> x <elementtype>] 3786 3787The number of elements is a constant integer value; ``elementtype`` may 3788be any type with a size. 3789 3790:Examples: 3791 3792+------------------+--------------------------------------+ 3793| ``[40 x i32]`` | Array of 40 32-bit integer values. | 3794+------------------+--------------------------------------+ 3795| ``[41 x i32]`` | Array of 41 32-bit integer values. | 3796+------------------+--------------------------------------+ 3797| ``[4 x i8]`` | Array of 4 8-bit integer values. | 3798+------------------+--------------------------------------+ 3799 3800Here are some examples of multidimensional arrays: 3801 3802+-----------------------------+----------------------------------------------------------+ 3803| ``[3 x [4 x i32]]`` | 3x4 array of 32-bit integer values. | 3804+-----------------------------+----------------------------------------------------------+ 3805| ``[12 x [10 x float]]`` | 12x10 array of single precision floating-point values. | 3806+-----------------------------+----------------------------------------------------------+ 3807| ``[2 x [3 x [4 x i16]]]`` | 2x3x4 array of 16-bit integer values. | 3808+-----------------------------+----------------------------------------------------------+ 3809 3810There is no restriction on indexing beyond the end of the array implied 3811by a static type (though there are restrictions on indexing beyond the 3812bounds of an allocated object in some cases). This means that 3813single-dimension 'variable sized array' addressing can be implemented in 3814LLVM with a zero length array type. An implementation of 'pascal style 3815arrays' in LLVM could use the type "``{ i32, [0 x float]}``", for 3816example. 3817 3818.. _t_struct: 3819 3820Structure Type 3821"""""""""""""" 3822 3823:Overview: 3824 3825The structure type is used to represent a collection of data members 3826together in memory. The elements of a structure may be any type that has 3827a size. 3828 3829Structures in memory are accessed using '``load``' and '``store``' by 3830getting a pointer to a field with the '``getelementptr``' instruction. 3831Structures in registers are accessed using the '``extractvalue``' and 3832'``insertvalue``' instructions. 3833 3834Structures may optionally be "packed" structures, which indicate that 3835the alignment of the struct is one byte, and that there is no padding 3836between the elements. In non-packed structs, padding between field types 3837is inserted as defined by the DataLayout string in the module, which is 3838required to match what the underlying code generator expects. 3839 3840Structures can either be "literal" or "identified". A literal structure 3841is defined inline with other types (e.g. ``{i32, i32}*``) whereas 3842identified types are always defined at the top level with a name. 3843Literal types are uniqued by their contents and can never be recursive 3844or opaque since there is no way to write one. Identified types can be 3845recursive, can be opaqued, and are never uniqued. 3846 3847:Syntax: 3848 3849:: 3850 3851 %T1 = type { <type list> } ; Identified normal struct type 3852 %T2 = type <{ <type list> }> ; Identified packed struct type 3853 3854:Examples: 3855 3856+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3857| ``{ i32, i32, i32 }`` | A triple of three ``i32`` values | 3858+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3859| ``{ 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``. | 3860+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3861| ``<{ i8, i32 }>`` | A packed struct known to be 5 bytes in size. | 3862+------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3863 3864.. _t_opaque: 3865 3866Opaque Structure Types 3867"""""""""""""""""""""" 3868 3869:Overview: 3870 3871Opaque structure types are used to represent structure types that 3872do not have a body specified. This corresponds (for example) to the C 3873notion of a forward declared structure. They can be named (``%X``) or 3874unnamed (``%52``). 3875 3876:Syntax: 3877 3878:: 3879 3880 %X = type opaque 3881 %52 = type opaque 3882 3883:Examples: 3884 3885+--------------+-------------------+ 3886| ``opaque`` | An opaque type. | 3887+--------------+-------------------+ 3888 3889.. _constants: 3890 3891Constants 3892========= 3893 3894LLVM has several different basic types of constants. This section 3895describes them all and their syntax. 3896 3897Simple Constants 3898---------------- 3899 3900**Boolean constants** 3901 The two strings '``true``' and '``false``' are both valid constants 3902 of the ``i1`` type. 3903**Integer constants** 3904 Standard integers (such as '4') are constants of the 3905 :ref:`integer <t_integer>` type. Negative numbers may be used with 3906 integer types. 3907**Floating-point constants** 3908 Floating-point constants use standard decimal notation (e.g. 3909 123.421), exponential notation (e.g. 1.23421e+2), or a more precise 3910 hexadecimal notation (see below). The assembler requires the exact 3911 decimal value of a floating-point constant. For example, the 3912 assembler accepts 1.25 but rejects 1.3 because 1.3 is a repeating 3913 decimal in binary. Floating-point constants must have a 3914 :ref:`floating-point <t_floating>` type. 3915**Null pointer constants** 3916 The identifier '``null``' is recognized as a null pointer constant 3917 and must be of :ref:`pointer type <t_pointer>`. 3918**Token constants** 3919 The identifier '``none``' is recognized as an empty token constant 3920 and must be of :ref:`token type <t_token>`. 3921 3922The one non-intuitive notation for constants is the hexadecimal form of 3923floating-point constants. For example, the form 3924'``double 0x432ff973cafa8000``' is equivalent to (but harder to read 3925than) '``double 4.5e+15``'. The only time hexadecimal floating-point 3926constants are required (and the only time that they are generated by the 3927disassembler) is when a floating-point constant must be emitted but it 3928cannot be represented as a decimal floating-point number in a reasonable 3929number of digits. For example, NaN's, infinities, and other special 3930values are represented in their IEEE hexadecimal format so that assembly 3931and disassembly do not cause any bits to change in the constants. 3932 3933When using the hexadecimal form, constants of types bfloat, half, float, and 3934double are represented using the 16-digit form shown above (which matches the 3935IEEE754 representation for double); bfloat, half and float values must, however, 3936be exactly representable as bfloat, IEEE 754 half, and IEEE 754 single 3937precision respectively. Hexadecimal format is always used for long double, and 3938there are three forms of long double. The 80-bit format used by x86 is 3939represented as ``0xK`` followed by 20 hexadecimal digits. The 128-bit format 3940used by PowerPC (two adjacent doubles) is represented by ``0xM`` followed by 32 3941hexadecimal digits. The IEEE 128-bit format is represented by ``0xL`` followed 3942by 32 hexadecimal digits. Long doubles will only work if they match the long 3943double format on your target. The IEEE 16-bit format (half precision) is 3944represented by ``0xH`` followed by 4 hexadecimal digits. The bfloat 16-bit 3945format is represented by ``0xR`` followed by 4 hexadecimal digits. All 3946hexadecimal formats are big-endian (sign bit at the left). 3947 3948There are no constants of type x86_mmx and x86_amx. 3949 3950.. _complexconstants: 3951 3952Complex Constants 3953----------------- 3954 3955Complex constants are a (potentially recursive) combination of simple 3956constants and smaller complex constants. 3957 3958**Structure constants** 3959 Structure constants are represented with notation similar to 3960 structure type definitions (a comma separated list of elements, 3961 surrounded by braces (``{}``)). For example: 3962 "``{ i32 4, float 17.0, i32* @G }``", where "``@G``" is declared as 3963 "``@G = external global i32``". Structure constants must have 3964 :ref:`structure type <t_struct>`, and the number and types of elements 3965 must match those specified by the type. 3966**Array constants** 3967 Array constants are represented with notation similar to array type 3968 definitions (a comma separated list of elements, surrounded by 3969 square brackets (``[]``)). For example: 3970 "``[ i32 42, i32 11, i32 74 ]``". Array constants must have 3971 :ref:`array type <t_array>`, and the number and types of elements must 3972 match those specified by the type. As a special case, character array 3973 constants may also be represented as a double-quoted string using the ``c`` 3974 prefix. For example: "``c"Hello World\0A\00"``". 3975**Vector constants** 3976 Vector constants are represented with notation similar to vector 3977 type definitions (a comma separated list of elements, surrounded by 3978 less-than/greater-than's (``<>``)). For example: 3979 "``< i32 42, i32 11, i32 74, i32 100 >``". Vector constants 3980 must have :ref:`vector type <t_vector>`, and the number and types of 3981 elements must match those specified by the type. 3982**Zero initialization** 3983 The string '``zeroinitializer``' can be used to zero initialize a 3984 value to zero of *any* type, including scalar and 3985 :ref:`aggregate <t_aggregate>` types. This is often used to avoid 3986 having to print large zero initializers (e.g. for large arrays) and 3987 is always exactly equivalent to using explicit zero initializers. 3988**Metadata node** 3989 A metadata node is a constant tuple without types. For example: 3990 "``!{!0, !{!2, !0}, !"test"}``". Metadata can reference constant values, 3991 for example: "``!{!0, i32 0, i8* @global, i64 (i64)* @function, !"str"}``". 3992 Unlike other typed constants that are meant to be interpreted as part of 3993 the instruction stream, metadata is a place to attach additional 3994 information such as debug info. 3995 3996Global Variable and Function Addresses 3997-------------------------------------- 3998 3999The addresses of :ref:`global variables <globalvars>` and 4000:ref:`functions <functionstructure>` are always implicitly valid 4001(link-time) constants. These constants are explicitly referenced when 4002the :ref:`identifier for the global <identifiers>` is used and always have 4003:ref:`pointer <t_pointer>` type. For example, the following is a legal LLVM 4004file: 4005 4006.. code-block:: llvm 4007 4008 @X = global i32 17 4009 @Y = global i32 42 4010 @Z = global [2 x i32*] [ i32* @X, i32* @Y ] 4011 4012.. _undefvalues: 4013 4014Undefined Values 4015---------------- 4016 4017The string '``undef``' can be used anywhere a constant is expected, and 4018indicates that the user of the value may receive an unspecified 4019bit-pattern. Undefined values may be of any type (other than '``label``' 4020or '``void``') and be used anywhere a constant is permitted. 4021 4022.. note:: 4023 4024 A '``poison``' value (decribed in the next section) should be used instead of 4025 '``undef``' whenever possible. Poison values are stronger than undef, and 4026 enable more optimizations. Just the existence of '``undef``' blocks certain 4027 optimizations (see the examples below). 4028 4029Undefined values are useful because they indicate to the compiler that 4030the program is well defined no matter what value is used. This gives the 4031compiler more freedom to optimize. Here are some examples of 4032(potentially surprising) transformations that are valid (in pseudo IR): 4033 4034.. code-block:: llvm 4035 4036 %A = add %X, undef 4037 %B = sub %X, undef 4038 %C = xor %X, undef 4039 Safe: 4040 %A = undef 4041 %B = undef 4042 %C = undef 4043 4044This is safe because all of the output bits are affected by the undef 4045bits. Any output bit can have a zero or one depending on the input bits. 4046 4047.. code-block:: llvm 4048 4049 %A = or %X, undef 4050 %B = and %X, undef 4051 Safe: 4052 %A = -1 4053 %B = 0 4054 Safe: 4055 %A = %X ;; By choosing undef as 0 4056 %B = %X ;; By choosing undef as -1 4057 Unsafe: 4058 %A = undef 4059 %B = undef 4060 4061These logical operations have bits that are not always affected by the 4062input. For example, if ``%X`` has a zero bit, then the output of the 4063'``and``' operation will always be a zero for that bit, no matter what 4064the corresponding bit from the '``undef``' is. As such, it is unsafe to 4065optimize or assume that the result of the '``and``' is '``undef``'. 4066However, it is safe to assume that all bits of the '``undef``' could be 40670, and optimize the '``and``' to 0. Likewise, it is safe to assume that 4068all the bits of the '``undef``' operand to the '``or``' could be set, 4069allowing the '``or``' to be folded to -1. 4070 4071.. code-block:: llvm 4072 4073 %A = select undef, %X, %Y 4074 %B = select undef, 42, %Y 4075 %C = select %X, %Y, undef 4076 Safe: 4077 %A = %X (or %Y) 4078 %B = 42 (or %Y) 4079 %C = %Y (if %Y is provably not poison; unsafe otherwise) 4080 Unsafe: 4081 %A = undef 4082 %B = undef 4083 %C = undef 4084 4085This set of examples shows that undefined '``select``' (and conditional 4086branch) conditions can go *either way*, but they have to come from one 4087of the two operands. In the ``%A`` example, if ``%X`` and ``%Y`` were 4088both known to have a clear low bit, then ``%A`` would have to have a 4089cleared low bit. However, in the ``%C`` example, the optimizer is 4090allowed to assume that the '``undef``' operand could be the same as 4091``%Y`` if ``%Y`` is provably not '``poison``', allowing the whole '``select``' 4092to be eliminated. This is because '``poison``' is stronger than '``undef``'. 4093 4094.. code-block:: llvm 4095 4096 %A = xor undef, undef 4097 4098 %B = undef 4099 %C = xor %B, %B 4100 4101 %D = undef 4102 %E = icmp slt %D, 4 4103 %F = icmp gte %D, 4 4104 4105 Safe: 4106 %A = undef 4107 %B = undef 4108 %C = undef 4109 %D = undef 4110 %E = undef 4111 %F = undef 4112 4113This example points out that two '``undef``' operands are not 4114necessarily the same. This can be surprising to people (and also matches 4115C semantics) where they assume that "``X^X``" is always zero, even if 4116``X`` is undefined. This isn't true for a number of reasons, but the 4117short answer is that an '``undef``' "variable" can arbitrarily change 4118its value over its "live range". This is true because the variable 4119doesn't actually *have a live range*. Instead, the value is logically 4120read from arbitrary registers that happen to be around when needed, so 4121the value is not necessarily consistent over time. In fact, ``%A`` and 4122``%C`` need to have the same semantics or the core LLVM "replace all 4123uses with" concept would not hold. 4124 4125To ensure all uses of a given register observe the same value (even if 4126'``undef``'), the :ref:`freeze instruction <i_freeze>` can be used. 4127 4128.. code-block:: llvm 4129 4130 %A = sdiv undef, %X 4131 %B = sdiv %X, undef 4132 Safe: 4133 %A = 0 4134 b: unreachable 4135 4136These examples show the crucial difference between an *undefined value* 4137and *undefined behavior*. An undefined value (like '``undef``') is 4138allowed to have an arbitrary bit-pattern. This means that the ``%A`` 4139operation can be constant folded to '``0``', because the '``undef``' 4140could be zero, and zero divided by any value is zero. 4141However, in the second example, we can make a more aggressive 4142assumption: because the ``undef`` is allowed to be an arbitrary value, 4143we are allowed to assume that it could be zero. Since a divide by zero 4144has *undefined behavior*, we are allowed to assume that the operation 4145does not execute at all. This allows us to delete the divide and all 4146code after it. Because the undefined operation "can't happen", the 4147optimizer can assume that it occurs in dead code. 4148 4149.. code-block:: text 4150 4151 a: store undef -> %X 4152 b: store %X -> undef 4153 Safe: 4154 a: <deleted> (if the stored value in %X is provably not poison) 4155 b: unreachable 4156 4157A store *of* an undefined value can be assumed to not have any effect; 4158we can assume that the value is overwritten with bits that happen to 4159match what was already there. This argument is only valid if the stored value 4160is provably not ``poison``. However, a store *to* an undefined 4161location could clobber arbitrary memory, therefore, it has undefined 4162behavior. 4163 4164Branching on an undefined value is undefined behavior. 4165This explains optimizations that depend on branch conditions to construct 4166predicates, such as Correlated Value Propagation and Global Value Numbering. 4167In case of switch instruction, the branch condition should be frozen, otherwise 4168it is undefined behavior. 4169 4170.. code-block:: llvm 4171 4172 Unsafe: 4173 br undef, BB1, BB2 ; UB 4174 4175 %X = and i32 undef, 255 4176 switch %X, label %ret [ .. ] ; UB 4177 4178 store undef, i8* %ptr 4179 %X = load i8* %ptr ; %X is undef 4180 switch i8 %X, label %ret [ .. ] ; UB 4181 4182 Safe: 4183 %X = or i8 undef, 255 ; always 255 4184 switch i8 %X, label %ret [ .. ] ; Well-defined 4185 4186 %X = freeze i1 undef 4187 br %X, BB1, BB2 ; Well-defined (non-deterministic jump) 4188 4189 4190 4191.. _poisonvalues: 4192 4193Poison Values 4194------------- 4195 4196A poison value is a result of an erroneous operation. 4197In order to facilitate speculative execution, many instructions do not 4198invoke immediate undefined behavior when provided with illegal operands, 4199and return a poison value instead. 4200The string '``poison``' can be used anywhere a constant is expected, and 4201operations such as :ref:`add <i_add>` with the ``nsw`` flag can produce 4202a poison value. 4203 4204Most instructions return '``poison``' when one of their arguments is 4205'``poison``'. A notable exception is the :ref:`select instruction <i_select>`. 4206Propagation of poison can be stopped with the 4207:ref:`freeze instruction <i_freeze>`. 4208 4209It is correct to replace a poison value with an 4210:ref:`undef value <undefvalues>` or any value of the type. 4211 4212This means that immediate undefined behavior occurs if a poison value is 4213used as an instruction operand that has any values that trigger undefined 4214behavior. Notably this includes (but is not limited to): 4215 4216- The pointer operand of a :ref:`load <i_load>`, :ref:`store <i_store>` or 4217 any other pointer dereferencing instruction (independent of address 4218 space). 4219- The divisor operand of a ``udiv``, ``sdiv``, ``urem`` or ``srem`` 4220 instruction. 4221- The condition operand of a :ref:`br <i_br>` instruction. 4222- The callee operand of a :ref:`call <i_call>` or :ref:`invoke <i_invoke>` 4223 instruction. 4224- The parameter operand of a :ref:`call <i_call>` or :ref:`invoke <i_invoke>` 4225 instruction, when the function or invoking call site has a ``noundef`` 4226 attribute in the corresponding position. 4227- The operand of a :ref:`ret <i_ret>` instruction if the function or invoking 4228 call site has a `noundef` attribute in the return value position. 4229 4230Here are some examples: 4231 4232.. code-block:: llvm 4233 4234 entry: 4235 %poison = sub nuw i32 0, 1 ; Results in a poison value. 4236 %poison2 = sub i32 poison, 1 ; Also results in a poison value. 4237 %still_poison = and i32 %poison, 0 ; 0, but also poison. 4238 %poison_yet_again = getelementptr i32, i32* @h, i32 %still_poison 4239 store i32 0, i32* %poison_yet_again ; Undefined behavior due to 4240 ; store to poison. 4241 4242 store i32 %poison, i32* @g ; Poison value stored to memory. 4243 %poison3 = load i32, i32* @g ; Poison value loaded back from memory. 4244 4245 %narrowaddr = bitcast i32* @g to i16* 4246 %wideaddr = bitcast i32* @g to i64* 4247 %poison4 = load i16, i16* %narrowaddr ; Returns a poison value. 4248 %poison5 = load i64, i64* %wideaddr ; Returns a poison value. 4249 4250 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value. 4251 br i1 %cmp, label %end, label %end ; undefined behavior 4252 4253 end: 4254 4255.. _welldefinedvalues: 4256 4257Well-Defined Values 4258------------------- 4259 4260Given a program execution, a value is *well defined* if the value does not 4261have an undef bit and is not poison in the execution. 4262An aggregate value or vector is well defined if its elements are well defined. 4263The padding of an aggregate isn't considered, since it isn't visible 4264without storing it into memory and loading it with a different type. 4265 4266A constant of a :ref:`single value <t_single_value>`, non-vector type is well 4267defined if it is neither '``undef``' constant nor '``poison``' constant. 4268The result of :ref:`freeze instruction <i_freeze>` is well defined regardless 4269of its operand. 4270 4271.. _blockaddress: 4272 4273Addresses of Basic Blocks 4274------------------------- 4275 4276``blockaddress(@function, %block)`` 4277 4278The '``blockaddress``' constant computes the address of the specified 4279basic block in the specified function. 4280 4281It always has an ``i8 addrspace(P)*`` type, where ``P`` is the address space 4282of the function containing ``%block`` (usually ``addrspace(0)``). 4283 4284Taking the address of the entry block is illegal. 4285 4286This value only has defined behavior when used as an operand to the 4287':ref:`indirectbr <i_indirectbr>`' or ':ref:`callbr <i_callbr>`'instruction, or 4288for comparisons against null. Pointer equality tests between labels addresses 4289results in undefined behavior --- though, again, comparison against null is ok, 4290and no label is equal to the null pointer. This may be passed around as an 4291opaque pointer sized value as long as the bits are not inspected. This 4292allows ``ptrtoint`` and arithmetic to be performed on these values so 4293long as the original value is reconstituted before the ``indirectbr`` or 4294``callbr`` instruction. 4295 4296Finally, some targets may provide defined semantics when using the value 4297as the operand to an inline assembly, but that is target specific. 4298 4299.. _dso_local_equivalent: 4300 4301DSO Local Equivalent 4302-------------------- 4303 4304``dso_local_equivalent @func`` 4305 4306A '``dso_local_equivalent``' constant represents a function which is 4307functionally equivalent to a given function, but is always defined in the 4308current linkage unit. The resulting pointer has the same type as the underlying 4309function. The resulting pointer is permitted, but not required, to be different 4310from a pointer to the function, and it may have different values in different 4311translation units. 4312 4313The target function may not have ``extern_weak`` linkage. 4314 4315``dso_local_equivalent`` can be implemented as such: 4316 4317- If the function has local linkage, hidden visibility, or is 4318 ``dso_local``, ``dso_local_equivalent`` can be implemented as simply a pointer 4319 to the function. 4320- ``dso_local_equivalent`` can be implemented with a stub that tail-calls the 4321 function. Many targets support relocations that resolve at link time to either 4322 a function or a stub for it, depending on if the function is defined within the 4323 linkage unit; LLVM will use this when available. (This is commonly called a 4324 "PLT stub".) On other targets, the stub may need to be emitted explicitly. 4325 4326This can be used wherever a ``dso_local`` instance of a function is needed without 4327needing to explicitly make the original function ``dso_local``. An instance where 4328this can be used is for static offset calculations between a function and some other 4329``dso_local`` symbol. This is especially useful for the Relative VTables C++ ABI, 4330where dynamic relocations for function pointers in VTables can be replaced with 4331static relocations for offsets between the VTable and virtual functions which 4332may not be ``dso_local``. 4333 4334This is currently only supported for ELF binary formats. 4335 4336.. _no_cfi: 4337 4338No CFI 4339------ 4340 4341``no_cfi @func`` 4342 4343With `Control-Flow Integrity (CFI) 4344<https://clang.llvm.org/docs/ControlFlowIntegrity.html>`_, a '``no_cfi``' 4345constant represents a function reference that does not get replaced with a 4346reference to the CFI jump table in the ``LowerTypeTests`` pass. These constants 4347may be useful in low-level programs, such as operating system kernels, which 4348need to refer to the actual function body. 4349 4350.. _constantexprs: 4351 4352Constant Expressions 4353-------------------- 4354 4355Constant expressions are used to allow expressions involving other 4356constants to be used as constants. Constant expressions may be of any 4357:ref:`first class <t_firstclass>` type and may involve any LLVM operation 4358that does not have side effects (e.g. load and call are not supported). 4359The following is the syntax for constant expressions: 4360 4361``trunc (CST to TYPE)`` 4362 Perform the :ref:`trunc operation <i_trunc>` on constants. 4363``zext (CST to TYPE)`` 4364 Perform the :ref:`zext operation <i_zext>` on constants. 4365``sext (CST to TYPE)`` 4366 Perform the :ref:`sext operation <i_sext>` on constants. 4367``fptrunc (CST to TYPE)`` 4368 Truncate a floating-point constant to another floating-point type. 4369 The size of CST must be larger than the size of TYPE. Both types 4370 must be floating-point. 4371``fpext (CST to TYPE)`` 4372 Floating-point extend a constant to another type. The size of CST 4373 must be smaller or equal to the size of TYPE. Both types must be 4374 floating-point. 4375``fptoui (CST to TYPE)`` 4376 Convert a floating-point constant to the corresponding unsigned 4377 integer constant. TYPE must be a scalar or vector integer type. CST 4378 must be of scalar or vector floating-point type. Both CST and TYPE 4379 must be scalars, or vectors of the same number of elements. If the 4380 value won't fit in the integer type, the result is a 4381 :ref:`poison value <poisonvalues>`. 4382``fptosi (CST to TYPE)`` 4383 Convert a floating-point constant to the corresponding signed 4384 integer constant. TYPE must be a scalar or vector integer type. CST 4385 must be of scalar or vector floating-point type. Both CST and TYPE 4386 must be scalars, or vectors of the same number of elements. If the 4387 value won't fit in the integer type, the result is a 4388 :ref:`poison value <poisonvalues>`. 4389``uitofp (CST to TYPE)`` 4390 Convert an unsigned integer constant to the corresponding 4391 floating-point constant. TYPE must be a scalar or vector floating-point 4392 type. CST must be of scalar or vector integer type. Both CST and TYPE must 4393 be scalars, or vectors of the same number of elements. 4394``sitofp (CST to TYPE)`` 4395 Convert a signed integer constant to the corresponding floating-point 4396 constant. TYPE must be a scalar or vector floating-point type. 4397 CST must be of scalar or vector integer type. Both CST and TYPE must 4398 be scalars, or vectors of the same number of elements. 4399``ptrtoint (CST to TYPE)`` 4400 Perform the :ref:`ptrtoint operation <i_ptrtoint>` on constants. 4401``inttoptr (CST to TYPE)`` 4402 Perform the :ref:`inttoptr operation <i_inttoptr>` on constants. 4403 This one is *really* dangerous! 4404``bitcast (CST to TYPE)`` 4405 Convert a constant, CST, to another TYPE. 4406 The constraints of the operands are the same as those for the 4407 :ref:`bitcast instruction <i_bitcast>`. 4408``addrspacecast (CST to TYPE)`` 4409 Convert a constant pointer or constant vector of pointer, CST, to another 4410 TYPE in a different address space. The constraints of the operands are the 4411 same as those for the :ref:`addrspacecast instruction <i_addrspacecast>`. 4412``getelementptr (TY, CSTPTR, IDX0, IDX1, ...)``, ``getelementptr inbounds (TY, CSTPTR, IDX0, IDX1, ...)`` 4413 Perform the :ref:`getelementptr operation <i_getelementptr>` on 4414 constants. As with the :ref:`getelementptr <i_getelementptr>` 4415 instruction, the index list may have one or more indexes, which are 4416 required to make sense for the type of "pointer to TY". 4417``select (COND, VAL1, VAL2)`` 4418 Perform the :ref:`select operation <i_select>` on constants. 4419``icmp COND (VAL1, VAL2)`` 4420 Perform the :ref:`icmp operation <i_icmp>` on constants. 4421``fcmp COND (VAL1, VAL2)`` 4422 Perform the :ref:`fcmp operation <i_fcmp>` on constants. 4423``extractelement (VAL, IDX)`` 4424 Perform the :ref:`extractelement operation <i_extractelement>` on 4425 constants. 4426``insertelement (VAL, ELT, IDX)`` 4427 Perform the :ref:`insertelement operation <i_insertelement>` on 4428 constants. 4429``shufflevector (VEC1, VEC2, IDXMASK)`` 4430 Perform the :ref:`shufflevector operation <i_shufflevector>` on 4431 constants. 4432``extractvalue (VAL, IDX0, IDX1, ...)`` 4433 Perform the :ref:`extractvalue operation <i_extractvalue>` on 4434 constants. The index list is interpreted in a similar manner as 4435 indices in a ':ref:`getelementptr <i_getelementptr>`' operation. At 4436 least one index value must be specified. 4437``insertvalue (VAL, ELT, IDX0, IDX1, ...)`` 4438 Perform the :ref:`insertvalue operation <i_insertvalue>` on constants. 4439 The index list is interpreted in a similar manner as indices in a 4440 ':ref:`getelementptr <i_getelementptr>`' operation. At least one index 4441 value must be specified. 4442``OPCODE (LHS, RHS)`` 4443 Perform the specified operation of the LHS and RHS constants. OPCODE 4444 may be any of the :ref:`binary <binaryops>` or :ref:`bitwise 4445 binary <bitwiseops>` operations. The constraints on operands are 4446 the same as those for the corresponding instruction (e.g. no bitwise 4447 operations on floating-point values are allowed). 4448 4449Other Values 4450============ 4451 4452.. _inlineasmexprs: 4453 4454Inline Assembler Expressions 4455---------------------------- 4456 4457LLVM supports inline assembler expressions (as opposed to :ref:`Module-Level 4458Inline Assembly <moduleasm>`) through the use of a special value. This value 4459represents the inline assembler as a template string (containing the 4460instructions to emit), a list of operand constraints (stored as a string), a 4461flag that indicates whether or not the inline asm expression has side effects, 4462and a flag indicating whether the function containing the asm needs to align its 4463stack conservatively. 4464 4465The template string supports argument substitution of the operands using "``$``" 4466followed by a number, to indicate substitution of the given register/memory 4467location, as specified by the constraint string. "``${NUM:MODIFIER}``" may also 4468be used, where ``MODIFIER`` is a target-specific annotation for how to print the 4469operand (See :ref:`inline-asm-modifiers`). 4470 4471A literal "``$``" may be included by using "``$$``" in the template. To include 4472other special characters into the output, the usual "``\XX``" escapes may be 4473used, just as in other strings. Note that after template substitution, the 4474resulting assembly string is parsed by LLVM's integrated assembler unless it is 4475disabled -- even when emitting a ``.s`` file -- and thus must contain assembly 4476syntax known to LLVM. 4477 4478LLVM also supports a few more substitutions useful for writing inline assembly: 4479 4480- ``${:uid}``: Expands to a decimal integer unique to this inline assembly blob. 4481 This substitution is useful when declaring a local label. Many standard 4482 compiler optimizations, such as inlining, may duplicate an inline asm blob. 4483 Adding a blob-unique identifier ensures that the two labels will not conflict 4484 during assembly. This is used to implement `GCC's %= special format 4485 string <https://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html>`_. 4486- ``${:comment}``: Expands to the comment character of the current target's 4487 assembly dialect. This is usually ``#``, but many targets use other strings, 4488 such as ``;``, ``//``, or ``!``. 4489- ``${:private}``: Expands to the assembler private label prefix. Labels with 4490 this prefix will not appear in the symbol table of the assembled object. 4491 Typically the prefix is ``L``, but targets may use other strings. ``.L`` is 4492 relatively popular. 4493 4494LLVM's support for inline asm is modeled closely on the requirements of Clang's 4495GCC-compatible inline-asm support. Thus, the feature-set and the constraint and 4496modifier codes listed here are similar or identical to those in GCC's inline asm 4497support. However, to be clear, the syntax of the template and constraint strings 4498described here is *not* the same as the syntax accepted by GCC and Clang, and, 4499while most constraint letters are passed through as-is by Clang, some get 4500translated to other codes when converting from the C source to the LLVM 4501assembly. 4502 4503An example inline assembler expression is: 4504 4505.. code-block:: llvm 4506 4507 i32 (i32) asm "bswap $0", "=r,r" 4508 4509Inline assembler expressions may **only** be used as the callee operand 4510of a :ref:`call <i_call>` or an :ref:`invoke <i_invoke>` instruction. 4511Thus, typically we have: 4512 4513.. code-block:: llvm 4514 4515 %X = call i32 asm "bswap $0", "=r,r"(i32 %Y) 4516 4517Inline asms with side effects not visible in the constraint list must be 4518marked as having side effects. This is done through the use of the 4519'``sideeffect``' keyword, like so: 4520 4521.. code-block:: llvm 4522 4523 call void asm sideeffect "eieio", ""() 4524 4525In some cases inline asms will contain code that will not work unless 4526the stack is aligned in some way, such as calls or SSE instructions on 4527x86, yet will not contain code that does that alignment within the asm. 4528The compiler should make conservative assumptions about what the asm 4529might contain and should generate its usual stack alignment code in the 4530prologue if the '``alignstack``' keyword is present: 4531 4532.. code-block:: llvm 4533 4534 call void asm alignstack "eieio", ""() 4535 4536Inline asms also support using non-standard assembly dialects. The 4537assumed dialect is ATT. When the '``inteldialect``' keyword is present, 4538the inline asm is using the Intel dialect. Currently, ATT and Intel are 4539the only supported dialects. An example is: 4540 4541.. code-block:: llvm 4542 4543 call void asm inteldialect "eieio", ""() 4544 4545In the case that the inline asm might unwind the stack, 4546the '``unwind``' keyword must be used, so that the compiler emits 4547unwinding information: 4548 4549.. code-block:: llvm 4550 4551 call void asm unwind "call func", ""() 4552 4553If the inline asm unwinds the stack and isn't marked with 4554the '``unwind``' keyword, the behavior is undefined. 4555 4556If multiple keywords appear, the '``sideeffect``' keyword must come 4557first, the '``alignstack``' keyword second, the '``inteldialect``' keyword 4558third and the '``unwind``' keyword last. 4559 4560Inline Asm Constraint String 4561^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 4562 4563The constraint list is a comma-separated string, each element containing one or 4564more constraint codes. 4565 4566For each element in the constraint list an appropriate register or memory 4567operand will be chosen, and it will be made available to assembly template 4568string expansion as ``$0`` for the first constraint in the list, ``$1`` for the 4569second, etc. 4570 4571There are three different types of constraints, which are distinguished by a 4572prefix symbol in front of the constraint code: Output, Input, and Clobber. The 4573constraints must always be given in that order: outputs first, then inputs, then 4574clobbers. They cannot be intermingled. 4575 4576There are also three different categories of constraint codes: 4577 4578- Register constraint. This is either a register class, or a fixed physical 4579 register. This kind of constraint will allocate a register, and if necessary, 4580 bitcast the argument or result to the appropriate type. 4581- Memory constraint. This kind of constraint is for use with an instruction 4582 taking a memory operand. Different constraints allow for different addressing 4583 modes used by the target. 4584- Immediate value constraint. This kind of constraint is for an integer or other 4585 immediate value which can be rendered directly into an instruction. The 4586 various target-specific constraints allow the selection of a value in the 4587 proper range for the instruction you wish to use it with. 4588 4589Output constraints 4590"""""""""""""""""" 4591 4592Output constraints are specified by an "``=``" prefix (e.g. "``=r``"). This 4593indicates that the assembly will write to this operand, and the operand will 4594then be made available as a return value of the ``asm`` expression. Output 4595constraints do not consume an argument from the call instruction. (Except, see 4596below about indirect outputs). 4597 4598Normally, it is expected that no output locations are written to by the assembly 4599expression until *all* of the inputs have been read. As such, LLVM may assign 4600the same register to an output and an input. If this is not safe (e.g. if the 4601assembly contains two instructions, where the first writes to one output, and 4602the second reads an input and writes to a second output), then the "``&``" 4603modifier must be used (e.g. "``=&r``") to specify that the output is an 4604"early-clobber" output. Marking an output as "early-clobber" ensures that LLVM 4605will not use the same register for any inputs (other than an input tied to this 4606output). 4607 4608Input constraints 4609""""""""""""""""" 4610 4611Input constraints do not have a prefix -- just the constraint codes. Each input 4612constraint will consume one argument from the call instruction. It is not 4613permitted for the asm to write to any input register or memory location (unless 4614that input is tied to an output). Note also that multiple inputs may all be 4615assigned to the same register, if LLVM can determine that they necessarily all 4616contain the same value. 4617 4618Instead of providing a Constraint Code, input constraints may also "tie" 4619themselves to an output constraint, by providing an integer as the constraint 4620string. Tied inputs still consume an argument from the call instruction, and 4621take up a position in the asm template numbering as is usual -- they will simply 4622be constrained to always use the same register as the output they've been tied 4623to. For example, a constraint string of "``=r,0``" says to assign a register for 4624output, and use that register as an input as well (it being the 0'th 4625constraint). 4626 4627It is permitted to tie an input to an "early-clobber" output. In that case, no 4628*other* input may share the same register as the input tied to the early-clobber 4629(even when the other input has the same value). 4630 4631You may only tie an input to an output which has a register constraint, not a 4632memory constraint. Only a single input may be tied to an output. 4633 4634There is also an "interesting" feature which deserves a bit of explanation: if a 4635register class constraint allocates a register which is too small for the value 4636type operand provided as input, the input value will be split into multiple 4637registers, and all of them passed to the inline asm. 4638 4639However, this feature is often not as useful as you might think. 4640 4641Firstly, the registers are *not* guaranteed to be consecutive. So, on those 4642architectures that have instructions which operate on multiple consecutive 4643instructions, this is not an appropriate way to support them. (e.g. the 32-bit 4644SparcV8 has a 64-bit load, which instruction takes a single 32-bit register. The 4645hardware then loads into both the named register, and the next register. This 4646feature of inline asm would not be useful to support that.) 4647 4648A few of the targets provide a template string modifier allowing explicit access 4649to the second register of a two-register operand (e.g. MIPS ``L``, ``M``, and 4650``D``). On such an architecture, you can actually access the second allocated 4651register (yet, still, not any subsequent ones). But, in that case, you're still 4652probably better off simply splitting the value into two separate operands, for 4653clarity. (e.g. see the description of the ``A`` constraint on X86, which, 4654despite existing only for use with this feature, is not really a good idea to 4655use) 4656 4657Indirect inputs and outputs 4658""""""""""""""""""""""""""" 4659 4660Indirect output or input constraints can be specified by the "``*``" modifier 4661(which goes after the "``=``" in case of an output). This indicates that the asm 4662will write to or read from the contents of an *address* provided as an input 4663argument. (Note that in this way, indirect outputs act more like an *input* than 4664an output: just like an input, they consume an argument of the call expression, 4665rather than producing a return value. An indirect output constraint is an 4666"output" only in that the asm is expected to write to the contents of the input 4667memory location, instead of just read from it). 4668 4669This is most typically used for memory constraint, e.g. "``=*m``", to pass the 4670address of a variable as a value. 4671 4672It is also possible to use an indirect *register* constraint, but only on output 4673(e.g. "``=*r``"). This will cause LLVM to allocate a register for an output 4674value normally, and then, separately emit a store to the address provided as 4675input, after the provided inline asm. (It's not clear what value this 4676functionality provides, compared to writing the store explicitly after the asm 4677statement, and it can only produce worse code, since it bypasses many 4678optimization passes. I would recommend not using it.) 4679 4680Call arguments for indirect constraints must have pointer type and must specify 4681the :ref:`elementtype <attr_elementtype>` attribute to indicate the pointer 4682element type. 4683 4684Clobber constraints 4685""""""""""""""""""" 4686 4687A clobber constraint is indicated by a "``~``" prefix. A clobber does not 4688consume an input operand, nor generate an output. Clobbers cannot use any of the 4689general constraint code letters -- they may use only explicit register 4690constraints, e.g. "``~{eax}``". The one exception is that a clobber string of 4691"``~{memory}``" indicates that the assembly writes to arbitrary undeclared 4692memory locations -- not only the memory pointed to by a declared indirect 4693output. 4694 4695Note that clobbering named registers that are also present in output 4696constraints is not legal. 4697 4698Label constraints 4699""""""""""""""""" 4700 4701A label constraint is indicated by a "``!``" prefix and typically used in the 4702form ``"!i"``. Instead of consuming call arguments, label constraints consume 4703indirect destination labels of ``callbr`` instructions. 4704 4705Label constraints can only be used in conjunction with ``callbr`` and the 4706number of label constraints must match the number of indirect destination 4707labels in the ``callbr`` instruction. 4708 4709 4710Constraint Codes 4711"""""""""""""""" 4712After a potential prefix comes constraint code, or codes. 4713 4714A Constraint Code is either a single letter (e.g. "``r``"), a "``^``" character 4715followed by two letters (e.g. "``^wc``"), or "``{``" register-name "``}``" 4716(e.g. "``{eax}``"). 4717 4718The one and two letter constraint codes are typically chosen to be the same as 4719GCC's constraint codes. 4720 4721A single constraint may include one or more than constraint code in it, leaving 4722it up to LLVM to choose which one to use. This is included mainly for 4723compatibility with the translation of GCC inline asm coming from clang. 4724 4725There are two ways to specify alternatives, and either or both may be used in an 4726inline asm constraint list: 4727 47281) Append the codes to each other, making a constraint code set. E.g. "``im``" 4729 or "``{eax}m``". This means "choose any of the options in the set". The 4730 choice of constraint is made independently for each constraint in the 4731 constraint list. 4732 47332) Use "``|``" between constraint code sets, creating alternatives. Every 4734 constraint in the constraint list must have the same number of alternative 4735 sets. With this syntax, the same alternative in *all* of the items in the 4736 constraint list will be chosen together. 4737 4738Putting those together, you might have a two operand constraint string like 4739``"rm|r,ri|rm"``. This indicates that if operand 0 is ``r`` or ``m``, then 4740operand 1 may be one of ``r`` or ``i``. If operand 0 is ``r``, then operand 1 4741may be one of ``r`` or ``m``. But, operand 0 and 1 cannot both be of type m. 4742 4743However, the use of either of the alternatives features is *NOT* recommended, as 4744LLVM is not able to make an intelligent choice about which one to use. (At the 4745point it currently needs to choose, not enough information is available to do so 4746in a smart way.) Thus, it simply tries to make a choice that's most likely to 4747compile, not one that will be optimal performance. (e.g., given "``rm``", it'll 4748always choose to use memory, not registers). And, if given multiple registers, 4749or multiple register classes, it will simply choose the first one. (In fact, it 4750doesn't currently even ensure explicitly specified physical registers are 4751unique, so specifying multiple physical registers as alternatives, like 4752``{r11}{r12},{r11}{r12}``, will assign r11 to both operands, not at all what was 4753intended.) 4754 4755Supported Constraint Code List 4756"""""""""""""""""""""""""""""" 4757 4758The constraint codes are, in general, expected to behave the same way they do in 4759GCC. LLVM's support is often implemented on an 'as-needed' basis, to support C 4760inline asm code which was supported by GCC. A mismatch in behavior between LLVM 4761and GCC likely indicates a bug in LLVM. 4762 4763Some constraint codes are typically supported by all targets: 4764 4765- ``r``: A register in the target's general purpose register class. 4766- ``m``: A memory address operand. It is target-specific what addressing modes 4767 are supported, typical examples are register, or register + register offset, 4768 or register + immediate offset (of some target-specific size). 4769- ``p``: An address operand. Similar to ``m``, but used by "load address" 4770 type instructions without touching memory. 4771- ``i``: An integer constant (of target-specific width). Allows either a simple 4772 immediate, or a relocatable value. 4773- ``n``: An integer constant -- *not* including relocatable values. 4774- ``s``: An integer constant, but allowing *only* relocatable values. 4775- ``X``: Allows an operand of any kind, no constraint whatsoever. Typically 4776 useful to pass a label for an asm branch or call. 4777 4778 .. FIXME: but that surely isn't actually okay to jump out of an asm 4779 block without telling llvm about the control transfer???) 4780 4781- ``{register-name}``: Requires exactly the named physical register. 4782 4783Other constraints are target-specific: 4784 4785AArch64: 4786 4787- ``z``: An immediate integer 0. Outputs ``WZR`` or ``XZR``, as appropriate. 4788- ``I``: An immediate integer valid for an ``ADD`` or ``SUB`` instruction, 4789 i.e. 0 to 4095 with optional shift by 12. 4790- ``J``: An immediate integer that, when negated, is valid for an ``ADD`` or 4791 ``SUB`` instruction, i.e. -1 to -4095 with optional left shift by 12. 4792- ``K``: An immediate integer that is valid for the 'bitmask immediate 32' of a 4793 logical instruction like ``AND``, ``EOR``, or ``ORR`` with a 32-bit register. 4794- ``L``: An immediate integer that is valid for the 'bitmask immediate 64' of a 4795 logical instruction like ``AND``, ``EOR``, or ``ORR`` with a 64-bit register. 4796- ``M``: An immediate integer for use with the ``MOV`` assembly alias on a 4797 32-bit register. This is a superset of ``K``: in addition to the bitmask 4798 immediate, also allows immediate integers which can be loaded with a single 4799 ``MOVZ`` or ``MOVL`` instruction. 4800- ``N``: An immediate integer for use with the ``MOV`` assembly alias on a 4801 64-bit register. This is a superset of ``L``. 4802- ``Q``: Memory address operand must be in a single register (no 4803 offsets). (However, LLVM currently does this for the ``m`` constraint as 4804 well.) 4805- ``r``: A 32 or 64-bit integer register (W* or X*). 4806- ``w``: A 32, 64, or 128-bit floating-point, SIMD or SVE vector register. 4807- ``x``: Like w, but restricted to registers 0 to 15 inclusive. 4808- ``y``: Like w, but restricted to SVE vector registers Z0 to Z7 inclusive. 4809- ``Upl``: One of the low eight SVE predicate registers (P0 to P7) 4810- ``Upa``: Any of the SVE predicate registers (P0 to P15) 4811 4812AMDGPU: 4813 4814- ``r``: A 32 or 64-bit integer register. 4815- ``[0-9]v``: The 32-bit VGPR register, number 0-9. 4816- ``[0-9]s``: The 32-bit SGPR register, number 0-9. 4817- ``[0-9]a``: The 32-bit AGPR register, number 0-9. 4818- ``I``: An integer inline constant in the range from -16 to 64. 4819- ``J``: A 16-bit signed integer constant. 4820- ``A``: An integer or a floating-point inline constant. 4821- ``B``: A 32-bit signed integer constant. 4822- ``C``: A 32-bit unsigned integer constant or an integer inline constant in the range from -16 to 64. 4823- ``DA``: A 64-bit constant that can be split into two "A" constants. 4824- ``DB``: A 64-bit constant that can be split into two "B" constants. 4825 4826All ARM modes: 4827 4828- ``Q``, ``Um``, ``Un``, ``Uq``, ``Us``, ``Ut``, ``Uv``, ``Uy``: Memory address 4829 operand. Treated the same as operand ``m``, at the moment. 4830- ``Te``: An even general-purpose 32-bit integer register: ``r0,r2,...,r12,r14`` 4831- ``To``: An odd general-purpose 32-bit integer register: ``r1,r3,...,r11`` 4832 4833ARM and ARM's Thumb2 mode: 4834 4835- ``j``: An immediate integer between 0 and 65535 (valid for ``MOVW``) 4836- ``I``: An immediate integer valid for a data-processing instruction. 4837- ``J``: An immediate integer between -4095 and 4095. 4838- ``K``: An immediate integer whose bitwise inverse is valid for a 4839 data-processing instruction. (Can be used with template modifier "``B``" to 4840 print the inverted value). 4841- ``L``: An immediate integer whose negation is valid for a data-processing 4842 instruction. (Can be used with template modifier "``n``" to print the negated 4843 value). 4844- ``M``: A power of two or an integer between 0 and 32. 4845- ``N``: Invalid immediate constraint. 4846- ``O``: Invalid immediate constraint. 4847- ``r``: A general-purpose 32-bit integer register (``r0-r15``). 4848- ``l``: In Thumb2 mode, low 32-bit GPR registers (``r0-r7``). In ARM mode, same 4849 as ``r``. 4850- ``h``: In Thumb2 mode, a high 32-bit GPR register (``r8-r15``). In ARM mode, 4851 invalid. 4852- ``w``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4853 ``s0-s31``, ``d0-d31``, or ``q0-q15``, respectively. 4854- ``t``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4855 ``s0-s31``, ``d0-d15``, or ``q0-q7``, respectively. 4856- ``x``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4857 ``s0-s15``, ``d0-d7``, or ``q0-q3``, respectively. 4858 4859ARM's Thumb1 mode: 4860 4861- ``I``: An immediate integer between 0 and 255. 4862- ``J``: An immediate integer between -255 and -1. 4863- ``K``: An immediate integer between 0 and 255, with optional left-shift by 4864 some amount. 4865- ``L``: An immediate integer between -7 and 7. 4866- ``M``: An immediate integer which is a multiple of 4 between 0 and 1020. 4867- ``N``: An immediate integer between 0 and 31. 4868- ``O``: An immediate integer which is a multiple of 4 between -508 and 508. 4869- ``r``: A low 32-bit GPR register (``r0-r7``). 4870- ``l``: A low 32-bit GPR register (``r0-r7``). 4871- ``h``: A high GPR register (``r0-r7``). 4872- ``w``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4873 ``s0-s31``, ``d0-d31``, or ``q0-q15``, respectively. 4874- ``t``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4875 ``s0-s31``, ``d0-d15``, or ``q0-q7``, respectively. 4876- ``x``: A 32, 64, or 128-bit floating-point/SIMD register in the ranges 4877 ``s0-s15``, ``d0-d7``, or ``q0-q3``, respectively. 4878 4879 4880Hexagon: 4881 4882- ``o``, ``v``: A memory address operand, treated the same as constraint ``m``, 4883 at the moment. 4884- ``r``: A 32 or 64-bit register. 4885 4886MSP430: 4887 4888- ``r``: An 8 or 16-bit register. 4889 4890MIPS: 4891 4892- ``I``: An immediate signed 16-bit integer. 4893- ``J``: An immediate integer zero. 4894- ``K``: An immediate unsigned 16-bit integer. 4895- ``L``: An immediate 32-bit integer, where the lower 16 bits are 0. 4896- ``N``: An immediate integer between -65535 and -1. 4897- ``O``: An immediate signed 15-bit integer. 4898- ``P``: An immediate integer between 1 and 65535. 4899- ``m``: A memory address operand. In MIPS-SE mode, allows a base address 4900 register plus 16-bit immediate offset. In MIPS mode, just a base register. 4901- ``R``: A memory address operand. In MIPS-SE mode, allows a base address 4902 register plus a 9-bit signed offset. In MIPS mode, the same as constraint 4903 ``m``. 4904- ``ZC``: A memory address operand, suitable for use in a ``pref``, ``ll``, or 4905 ``sc`` instruction on the given subtarget (details vary). 4906- ``r``, ``d``, ``y``: A 32 or 64-bit GPR register. 4907- ``f``: A 32 or 64-bit FPU register (``F0-F31``), or a 128-bit MSA register 4908 (``W0-W31``). In the case of MSA registers, it is recommended to use the ``w`` 4909 argument modifier for compatibility with GCC. 4910- ``c``: A 32-bit or 64-bit GPR register suitable for indirect jump (always 4911 ``25``). 4912- ``l``: The ``lo`` register, 32 or 64-bit. 4913- ``x``: Invalid. 4914 4915NVPTX: 4916 4917- ``b``: A 1-bit integer register. 4918- ``c`` or ``h``: A 16-bit integer register. 4919- ``r``: A 32-bit integer register. 4920- ``l`` or ``N``: A 64-bit integer register. 4921- ``f``: A 32-bit float register. 4922- ``d``: A 64-bit float register. 4923 4924 4925PowerPC: 4926 4927- ``I``: An immediate signed 16-bit integer. 4928- ``J``: An immediate unsigned 16-bit integer, shifted left 16 bits. 4929- ``K``: An immediate unsigned 16-bit integer. 4930- ``L``: An immediate signed 16-bit integer, shifted left 16 bits. 4931- ``M``: An immediate integer greater than 31. 4932- ``N``: An immediate integer that is an exact power of 2. 4933- ``O``: The immediate integer constant 0. 4934- ``P``: An immediate integer constant whose negation is a signed 16-bit 4935 constant. 4936- ``es``, ``o``, ``Q``, ``Z``, ``Zy``: A memory address operand, currently 4937 treated the same as ``m``. 4938- ``r``: A 32 or 64-bit integer register. 4939- ``b``: A 32 or 64-bit integer register, excluding ``R0`` (that is: 4940 ``R1-R31``). 4941- ``f``: A 32 or 64-bit float register (``F0-F31``), 4942- ``v``: For ``4 x f32`` or ``4 x f64`` types, a 128-bit altivec vector 4943 register (``V0-V31``). 4944 4945- ``y``: Condition register (``CR0-CR7``). 4946- ``wc``: An individual CR bit in a CR register. 4947- ``wa``, ``wd``, ``wf``: Any 128-bit VSX vector register, from the full VSX 4948 register set (overlapping both the floating-point and vector register files). 4949- ``ws``: A 32 or 64-bit floating-point register, from the full VSX register 4950 set. 4951 4952RISC-V: 4953 4954- ``A``: An address operand (using a general-purpose register, without an 4955 offset). 4956- ``I``: A 12-bit signed integer immediate operand. 4957- ``J``: A zero integer immediate operand. 4958- ``K``: A 5-bit unsigned integer immediate operand. 4959- ``f``: A 32- or 64-bit floating-point register (requires F or D extension). 4960- ``r``: A 32- or 64-bit general-purpose register (depending on the platform 4961 ``XLEN``). 4962- ``vr``: A vector register. (requires V extension). 4963- ``vm``: A vector register for masking operand. (requires V extension). 4964 4965Sparc: 4966 4967- ``I``: An immediate 13-bit signed integer. 4968- ``r``: A 32-bit integer register. 4969- ``f``: Any floating-point register on SparcV8, or a floating-point 4970 register in the "low" half of the registers on SparcV9. 4971- ``e``: Any floating-point register. (Same as ``f`` on SparcV8.) 4972 4973SystemZ: 4974 4975- ``I``: An immediate unsigned 8-bit integer. 4976- ``J``: An immediate unsigned 12-bit integer. 4977- ``K``: An immediate signed 16-bit integer. 4978- ``L``: An immediate signed 20-bit integer. 4979- ``M``: An immediate integer 0x7fffffff. 4980- ``Q``: A memory address operand with a base address and a 12-bit immediate 4981 unsigned displacement. 4982- ``R``: A memory address operand with a base address, a 12-bit immediate 4983 unsigned displacement, and an index register. 4984- ``S``: A memory address operand with a base address and a 20-bit immediate 4985 signed displacement. 4986- ``T``: A memory address operand with a base address, a 20-bit immediate 4987 signed displacement, and an index register. 4988- ``r`` or ``d``: A 32, 64, or 128-bit integer register. 4989- ``a``: A 32, 64, or 128-bit integer address register (excludes R0, which in an 4990 address context evaluates as zero). 4991- ``h``: A 32-bit value in the high part of a 64bit data register 4992 (LLVM-specific) 4993- ``f``: A 32, 64, or 128-bit floating-point register. 4994 4995X86: 4996 4997- ``I``: An immediate integer between 0 and 31. 4998- ``J``: An immediate integer between 0 and 64. 4999- ``K``: An immediate signed 8-bit integer. 5000- ``L``: An immediate integer, 0xff or 0xffff or (in 64-bit mode only) 5001 0xffffffff. 5002- ``M``: An immediate integer between 0 and 3. 5003- ``N``: An immediate unsigned 8-bit integer. 5004- ``O``: An immediate integer between 0 and 127. 5005- ``e``: An immediate 32-bit signed integer. 5006- ``Z``: An immediate 32-bit unsigned integer. 5007- ``o``, ``v``: Treated the same as ``m``, at the moment. 5008- ``q``: An 8, 16, 32, or 64-bit register which can be accessed as an 8-bit 5009 ``l`` integer register. On X86-32, this is the ``a``, ``b``, ``c``, and ``d`` 5010 registers, and on X86-64, it is all of the integer registers. 5011- ``Q``: An 8, 16, 32, or 64-bit register which can be accessed as an 8-bit 5012 ``h`` integer register. This is the ``a``, ``b``, ``c``, and ``d`` registers. 5013- ``r`` or ``l``: An 8, 16, 32, or 64-bit integer register. 5014- ``R``: An 8, 16, 32, or 64-bit "legacy" integer register -- one which has 5015 existed since i386, and can be accessed without the REX prefix. 5016- ``f``: A 32, 64, or 80-bit '387 FPU stack pseudo-register. 5017- ``y``: A 64-bit MMX register, if MMX is enabled. 5018- ``x``: If SSE is enabled: a 32 or 64-bit scalar operand, or 128-bit vector 5019 operand in a SSE register. If AVX is also enabled, can also be a 256-bit 5020 vector operand in an AVX register. If AVX-512 is also enabled, can also be a 5021 512-bit vector operand in an AVX512 register, Otherwise, an error. 5022- ``Y``: The same as ``x``, if *SSE2* is enabled, otherwise an error. 5023- ``A``: Special case: allocates EAX first, then EDX, for a single operand (in 5024 32-bit mode, a 64-bit integer operand will get split into two registers). It 5025 is not recommended to use this constraint, as in 64-bit mode, the 64-bit 5026 operand will get allocated only to RAX -- if two 32-bit operands are needed, 5027 you're better off splitting it yourself, before passing it to the asm 5028 statement. 5029 5030XCore: 5031 5032- ``r``: A 32-bit integer register. 5033 5034 5035.. _inline-asm-modifiers: 5036 5037Asm template argument modifiers 5038^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 5039 5040In the asm template string, modifiers can be used on the operand reference, like 5041"``${0:n}``". 5042 5043The modifiers are, in general, expected to behave the same way they do in 5044GCC. LLVM's support is often implemented on an 'as-needed' basis, to support C 5045inline asm code which was supported by GCC. A mismatch in behavior between LLVM 5046and GCC likely indicates a bug in LLVM. 5047 5048Target-independent: 5049 5050- ``c``: Print an immediate integer constant unadorned, without 5051 the target-specific immediate punctuation (e.g. no ``$`` prefix). 5052- ``n``: Negate and print immediate integer constant unadorned, without the 5053 target-specific immediate punctuation (e.g. no ``$`` prefix). 5054- ``l``: Print as an unadorned label, without the target-specific label 5055 punctuation (e.g. no ``$`` prefix). 5056 5057AArch64: 5058 5059- ``w``: Print a GPR register with a ``w*`` name instead of ``x*`` name. E.g., 5060 instead of ``x30``, print ``w30``. 5061- ``x``: Print a GPR register with a ``x*`` name. (this is the default, anyhow). 5062- ``b``, ``h``, ``s``, ``d``, ``q``: Print a floating-point/SIMD register with a 5063 ``b*``, ``h*``, ``s*``, ``d*``, or ``q*`` name, rather than the default of 5064 ``v*``. 5065 5066AMDGPU: 5067 5068- ``r``: No effect. 5069 5070ARM: 5071 5072- ``a``: Print an operand as an address (with ``[`` and ``]`` surrounding a 5073 register). 5074- ``P``: No effect. 5075- ``q``: No effect. 5076- ``y``: Print a VFP single-precision register as an indexed double (e.g. print 5077 as ``d4[1]`` instead of ``s9``) 5078- ``B``: Bitwise invert and print an immediate integer constant without ``#`` 5079 prefix. 5080- ``L``: Print the low 16-bits of an immediate integer constant. 5081- ``M``: Print as a register set suitable for ldm/stm. Also prints *all* 5082 register operands subsequent to the specified one (!), so use carefully. 5083- ``Q``: Print the low-order register of a register-pair, or the low-order 5084 register of a two-register operand. 5085- ``R``: Print the high-order register of a register-pair, or the high-order 5086 register of a two-register operand. 5087- ``H``: Print the second register of a register-pair. (On a big-endian system, 5088 ``H`` is equivalent to ``Q``, and on little-endian system, ``H`` is equivalent 5089 to ``R``.) 5090 5091 .. FIXME: H doesn't currently support printing the second register 5092 of a two-register operand. 5093 5094- ``e``: Print the low doubleword register of a NEON quad register. 5095- ``f``: Print the high doubleword register of a NEON quad register. 5096- ``m``: Print the base register of a memory operand without the ``[`` and ``]`` 5097 adornment. 5098 5099Hexagon: 5100 5101- ``L``: Print the second register of a two-register operand. Requires that it 5102 has been allocated consecutively to the first. 5103 5104 .. FIXME: why is it restricted to consecutive ones? And there's 5105 nothing that ensures that happens, is there? 5106 5107- ``I``: Print the letter 'i' if the operand is an integer constant, otherwise 5108 nothing. Used to print 'addi' vs 'add' instructions. 5109 5110MSP430: 5111 5112No additional modifiers. 5113 5114MIPS: 5115 5116- ``X``: Print an immediate integer as hexadecimal 5117- ``x``: Print the low 16 bits of an immediate integer as hexadecimal. 5118- ``d``: Print an immediate integer as decimal. 5119- ``m``: Subtract one and print an immediate integer as decimal. 5120- ``z``: Print $0 if an immediate zero, otherwise print normally. 5121- ``L``: Print the low-order register of a two-register operand, or prints the 5122 address of the low-order word of a double-word memory operand. 5123 5124 .. FIXME: L seems to be missing memory operand support. 5125 5126- ``M``: Print the high-order register of a two-register operand, or prints the 5127 address of the high-order word of a double-word memory operand. 5128 5129 .. FIXME: M seems to be missing memory operand support. 5130 5131- ``D``: Print the second register of a two-register operand, or prints the 5132 second word of a double-word memory operand. (On a big-endian system, ``D`` is 5133 equivalent to ``L``, and on little-endian system, ``D`` is equivalent to 5134 ``M``.) 5135- ``w``: No effect. Provided for compatibility with GCC which requires this 5136 modifier in order to print MSA registers (``W0-W31``) with the ``f`` 5137 constraint. 5138 5139NVPTX: 5140 5141- ``r``: No effect. 5142 5143PowerPC: 5144 5145- ``L``: Print the second register of a two-register operand. Requires that it 5146 has been allocated consecutively to the first. 5147 5148 .. FIXME: why is it restricted to consecutive ones? And there's 5149 nothing that ensures that happens, is there? 5150 5151- ``I``: Print the letter 'i' if the operand is an integer constant, otherwise 5152 nothing. Used to print 'addi' vs 'add' instructions. 5153- ``y``: For a memory operand, prints formatter for a two-register X-form 5154 instruction. (Currently always prints ``r0,OPERAND``). 5155- ``U``: Prints 'u' if the memory operand is an update form, and nothing 5156 otherwise. (NOTE: LLVM does not support update form, so this will currently 5157 always print nothing) 5158- ``X``: Prints 'x' if the memory operand is an indexed form. (NOTE: LLVM does 5159 not support indexed form, so this will currently always print nothing) 5160 5161RISC-V: 5162 5163- ``i``: Print the letter 'i' if the operand is not a register, otherwise print 5164 nothing. Used to print 'addi' vs 'add' instructions, etc. 5165- ``z``: Print the register ``zero`` if an immediate zero, otherwise print 5166 normally. 5167 5168Sparc: 5169 5170- ``r``: No effect. 5171 5172SystemZ: 5173 5174SystemZ implements only ``n``, and does *not* support any of the other 5175target-independent modifiers. 5176 5177X86: 5178 5179- ``c``: Print an unadorned integer or symbol name. (The latter is 5180 target-specific behavior for this typically target-independent modifier). 5181- ``A``: Print a register name with a '``*``' before it. 5182- ``b``: Print an 8-bit register name (e.g. ``al``); do nothing on a memory 5183 operand. 5184- ``h``: Print the upper 8-bit register name (e.g. ``ah``); do nothing on a 5185 memory operand. 5186- ``w``: Print the 16-bit register name (e.g. ``ax``); do nothing on a memory 5187 operand. 5188- ``k``: Print the 32-bit register name (e.g. ``eax``); do nothing on a memory 5189 operand. 5190- ``q``: Print the 64-bit register name (e.g. ``rax``), if 64-bit registers are 5191 available, otherwise the 32-bit register name; do nothing on a memory operand. 5192- ``n``: Negate and print an unadorned integer, or, for operands other than an 5193 immediate integer (e.g. a relocatable symbol expression), print a '-' before 5194 the operand. (The behavior for relocatable symbol expressions is a 5195 target-specific behavior for this typically target-independent modifier) 5196- ``H``: Print a memory reference with additional offset +8. 5197- ``P``: Print a memory reference used as the argument of a call instruction or 5198 used with explicit base reg and index reg as its offset. So it can not use 5199 additional regs to present the memory reference. (E.g. omit ``(rip)``, even 5200 though it's PC-relative.) 5201 5202XCore: 5203 5204No additional modifiers. 5205 5206 5207Inline Asm Metadata 5208^^^^^^^^^^^^^^^^^^^ 5209 5210The call instructions that wrap inline asm nodes may have a 5211"``!srcloc``" MDNode attached to it that contains a list of constant 5212integers. If present, the code generator will use the integer as the 5213location cookie value when report errors through the ``LLVMContext`` 5214error reporting mechanisms. This allows a front-end to correlate backend 5215errors that occur with inline asm back to the source code that produced 5216it. For example: 5217 5218.. code-block:: llvm 5219 5220 call void asm sideeffect "something bad", ""(), !srcloc !42 5221 ... 5222 !42 = !{ i32 1234567 } 5223 5224It is up to the front-end to make sense of the magic numbers it places 5225in the IR. If the MDNode contains multiple constants, the code generator 5226will use the one that corresponds to the line of the asm that the error 5227occurs on. 5228 5229.. _metadata: 5230 5231Metadata 5232======== 5233 5234LLVM IR allows metadata to be attached to instructions and global objects in the 5235program that can convey extra information about the code to the optimizers and 5236code generator. One example application of metadata is source-level 5237debug information. There are two metadata primitives: strings and nodes. 5238 5239Metadata does not have a type, and is not a value. If referenced from a 5240``call`` instruction, it uses the ``metadata`` type. 5241 5242All metadata are identified in syntax by an exclamation point ('``!``'). 5243 5244.. _metadata-string: 5245 5246Metadata Nodes and Metadata Strings 5247----------------------------------- 5248 5249A metadata string is a string surrounded by double quotes. It can 5250contain any character by escaping non-printable characters with 5251"``\xx``" where "``xx``" is the two digit hex code. For example: 5252"``!"test\00"``". 5253 5254Metadata nodes are represented with notation similar to structure 5255constants (a comma separated list of elements, surrounded by braces and 5256preceded by an exclamation point). Metadata nodes can have any values as 5257their operand. For example: 5258 5259.. code-block:: llvm 5260 5261 !{ !"test\00", i32 10} 5262 5263Metadata nodes that aren't uniqued use the ``distinct`` keyword. For example: 5264 5265.. code-block:: text 5266 5267 !0 = distinct !{!"test\00", i32 10} 5268 5269``distinct`` nodes are useful when nodes shouldn't be merged based on their 5270content. They can also occur when transformations cause uniquing collisions 5271when metadata operands change. 5272 5273A :ref:`named metadata <namedmetadatastructure>` is a collection of 5274metadata nodes, which can be looked up in the module symbol table. For 5275example: 5276 5277.. code-block:: llvm 5278 5279 !foo = !{!4, !3} 5280 5281Metadata can be used as function arguments. Here the ``llvm.dbg.value`` 5282intrinsic is using three metadata arguments: 5283 5284.. code-block:: llvm 5285 5286 call void @llvm.dbg.value(metadata !24, metadata !25, metadata !26) 5287 5288Metadata can be attached to an instruction. Here metadata ``!21`` is attached 5289to the ``add`` instruction using the ``!dbg`` identifier: 5290 5291.. code-block:: llvm 5292 5293 %indvar.next = add i64 %indvar, 1, !dbg !21 5294 5295Instructions may not have multiple metadata attachments with the same 5296identifier. 5297 5298Metadata can also be attached to a function or a global variable. Here metadata 5299``!22`` is attached to the ``f1`` and ``f2`` functions, and the globals ``g1`` 5300and ``g2`` using the ``!dbg`` identifier: 5301 5302.. code-block:: llvm 5303 5304 declare !dbg !22 void @f1() 5305 define void @f2() !dbg !22 { 5306 ret void 5307 } 5308 5309 @g1 = global i32 0, !dbg !22 5310 @g2 = external global i32, !dbg !22 5311 5312Unlike instructions, global objects (functions and global variables) may have 5313multiple metadata attachments with the same identifier. 5314 5315A transformation is required to drop any metadata attachment that it does not 5316know or know it can't preserve. Currently there is an exception for metadata 5317attachment to globals for ``!func_sanitize``, ``!type`` and ``!absolute_symbol`` which can't be 5318unconditionally dropped unless the global is itself deleted. 5319 5320Metadata attached to a module using named metadata may not be dropped, with 5321the exception of debug metadata (named metadata with the name ``!llvm.dbg.*``). 5322 5323More information about specific metadata nodes recognized by the 5324optimizers and code generator is found below. 5325 5326.. _specialized-metadata: 5327 5328Specialized Metadata Nodes 5329^^^^^^^^^^^^^^^^^^^^^^^^^^ 5330 5331Specialized metadata nodes are custom data structures in metadata (as opposed 5332to generic tuples). Their fields are labelled, and can be specified in any 5333order. 5334 5335These aren't inherently debug info centric, but currently all the specialized 5336metadata nodes are related to debug info. 5337 5338.. _DICompileUnit: 5339 5340DICompileUnit 5341""""""""""""" 5342 5343``DICompileUnit`` nodes represent a compile unit. The ``enums:``, 5344``retainedTypes:``, ``globals:``, ``imports:`` and ``macros:`` fields are tuples 5345containing the debug info to be emitted along with the compile unit, regardless 5346of code optimizations (some nodes are only emitted if there are references to 5347them from instructions). The ``debugInfoForProfiling:`` field is a boolean 5348indicating whether or not line-table discriminators are updated to provide 5349more-accurate debug info for profiling results. 5350 5351.. code-block:: text 5352 5353 !0 = !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang", 5354 isOptimized: true, flags: "-O2", runtimeVersion: 2, 5355 splitDebugFilename: "abc.debug", emissionKind: FullDebug, 5356 enums: !2, retainedTypes: !3, globals: !4, imports: !5, 5357 macros: !6, dwoId: 0x0abcd) 5358 5359Compile unit descriptors provide the root scope for objects declared in a 5360specific compilation unit. File descriptors are defined using this scope. These 5361descriptors are collected by a named metadata node ``!llvm.dbg.cu``. They keep 5362track of global variables, type information, and imported entities (declarations 5363and namespaces). 5364 5365.. _DIFile: 5366 5367DIFile 5368"""""" 5369 5370``DIFile`` nodes represent files. The ``filename:`` can include slashes. 5371 5372.. code-block:: none 5373 5374 !0 = !DIFile(filename: "path/to/file", directory: "/path/to/dir", 5375 checksumkind: CSK_MD5, 5376 checksum: "000102030405060708090a0b0c0d0e0f") 5377 5378Files are sometimes used in ``scope:`` fields, and are the only valid target 5379for ``file:`` fields. 5380Valid values for ``checksumkind:`` field are: {CSK_None, CSK_MD5, CSK_SHA1, CSK_SHA256} 5381 5382.. _DIBasicType: 5383 5384DIBasicType 5385""""""""""" 5386 5387``DIBasicType`` nodes represent primitive types, such as ``int``, ``bool`` and 5388``float``. ``tag:`` defaults to ``DW_TAG_base_type``. 5389 5390.. code-block:: text 5391 5392 !0 = !DIBasicType(name: "unsigned char", size: 8, align: 8, 5393 encoding: DW_ATE_unsigned_char) 5394 !1 = !DIBasicType(tag: DW_TAG_unspecified_type, name: "decltype(nullptr)") 5395 5396The ``encoding:`` describes the details of the type. Usually it's one of the 5397following: 5398 5399.. code-block:: text 5400 5401 DW_ATE_address = 1 5402 DW_ATE_boolean = 2 5403 DW_ATE_float = 4 5404 DW_ATE_signed = 5 5405 DW_ATE_signed_char = 6 5406 DW_ATE_unsigned = 7 5407 DW_ATE_unsigned_char = 8 5408 5409.. _DISubroutineType: 5410 5411DISubroutineType 5412"""""""""""""""" 5413 5414``DISubroutineType`` nodes represent subroutine types. Their ``types:`` field 5415refers to a tuple; the first operand is the return type, while the rest are the 5416types of the formal arguments in order. If the first operand is ``null``, that 5417represents a function with no return value (such as ``void foo() {}`` in C++). 5418 5419.. code-block:: text 5420 5421 !0 = !BasicType(name: "int", size: 32, align: 32, DW_ATE_signed) 5422 !1 = !BasicType(name: "char", size: 8, align: 8, DW_ATE_signed_char) 5423 !2 = !DISubroutineType(types: !{null, !0, !1}) ; void (int, char) 5424 5425.. _DIDerivedType: 5426 5427DIDerivedType 5428""""""""""""" 5429 5430``DIDerivedType`` nodes represent types derived from other types, such as 5431qualified types. 5432 5433.. code-block:: text 5434 5435 !0 = !DIBasicType(name: "unsigned char", size: 8, align: 8, 5436 encoding: DW_ATE_unsigned_char) 5437 !1 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !0, size: 32, 5438 align: 32) 5439 5440The following ``tag:`` values are valid: 5441 5442.. code-block:: text 5443 5444 DW_TAG_member = 13 5445 DW_TAG_pointer_type = 15 5446 DW_TAG_reference_type = 16 5447 DW_TAG_typedef = 22 5448 DW_TAG_inheritance = 28 5449 DW_TAG_ptr_to_member_type = 31 5450 DW_TAG_const_type = 38 5451 DW_TAG_friend = 42 5452 DW_TAG_volatile_type = 53 5453 DW_TAG_restrict_type = 55 5454 DW_TAG_atomic_type = 71 5455 DW_TAG_immutable_type = 75 5456 5457.. _DIDerivedTypeMember: 5458 5459``DW_TAG_member`` is used to define a member of a :ref:`composite type 5460<DICompositeType>`. The type of the member is the ``baseType:``. The 5461``offset:`` is the member's bit offset. If the composite type has an ODR 5462``identifier:`` and does not set ``flags: DIFwdDecl``, then the member is 5463uniqued based only on its ``name:`` and ``scope:``. 5464 5465``DW_TAG_inheritance`` and ``DW_TAG_friend`` are used in the ``elements:`` 5466field of :ref:`composite types <DICompositeType>` to describe parents and 5467friends. 5468 5469``DW_TAG_typedef`` is used to provide a name for the ``baseType:``. 5470 5471``DW_TAG_pointer_type``, ``DW_TAG_reference_type``, ``DW_TAG_const_type``, 5472``DW_TAG_volatile_type``, ``DW_TAG_restrict_type``, ``DW_TAG_atomic_type`` and 5473``DW_TAG_immutable_type`` are used to qualify the ``baseType:``. 5474 5475Note that the ``void *`` type is expressed as a type derived from NULL. 5476 5477.. _DICompositeType: 5478 5479DICompositeType 5480""""""""""""""" 5481 5482``DICompositeType`` nodes represent types composed of other types, like 5483structures and unions. ``elements:`` points to a tuple of the composed types. 5484 5485If the source language supports ODR, the ``identifier:`` field gives the unique 5486identifier used for type merging between modules. When specified, 5487:ref:`subprogram declarations <DISubprogramDeclaration>` and :ref:`member 5488derived types <DIDerivedTypeMember>` that reference the ODR-type in their 5489``scope:`` change uniquing rules. 5490 5491For a given ``identifier:``, there should only be a single composite type that 5492does not have ``flags: DIFlagFwdDecl`` set. LLVM tools that link modules 5493together will unique such definitions at parse time via the ``identifier:`` 5494field, even if the nodes are ``distinct``. 5495 5496.. code-block:: text 5497 5498 !0 = !DIEnumerator(name: "SixKind", value: 7) 5499 !1 = !DIEnumerator(name: "SevenKind", value: 7) 5500 !2 = !DIEnumerator(name: "NegEightKind", value: -8) 5501 !3 = !DICompositeType(tag: DW_TAG_enumeration_type, name: "Enum", file: !12, 5502 line: 2, size: 32, align: 32, identifier: "_M4Enum", 5503 elements: !{!0, !1, !2}) 5504 5505The following ``tag:`` values are valid: 5506 5507.. code-block:: text 5508 5509 DW_TAG_array_type = 1 5510 DW_TAG_class_type = 2 5511 DW_TAG_enumeration_type = 4 5512 DW_TAG_structure_type = 19 5513 DW_TAG_union_type = 23 5514 5515For ``DW_TAG_array_type``, the ``elements:`` should be :ref:`subrange 5516descriptors <DISubrange>`, each representing the range of subscripts at that 5517level of indexing. The ``DIFlagVector`` flag to ``flags:`` indicates that an 5518array type is a native packed vector. The optional ``dataLocation`` is a 5519DIExpression that describes how to get from an object's address to the actual 5520raw data, if they aren't equivalent. This is only supported for array types, 5521particularly to describe Fortran arrays, which have an array descriptor in 5522addition to the array data. Alternatively it can also be DIVariable which 5523has the address of the actual raw data. The Fortran language supports pointer 5524arrays which can be attached to actual arrays, this attachment between pointer 5525and pointee is called association. The optional ``associated`` is a 5526DIExpression that describes whether the pointer array is currently associated. 5527The optional ``allocated`` is a DIExpression that describes whether the 5528allocatable array is currently allocated. The optional ``rank`` is a 5529DIExpression that describes the rank (number of dimensions) of fortran assumed 5530rank array (rank is known at runtime). 5531 5532For ``DW_TAG_enumeration_type``, the ``elements:`` should be :ref:`enumerator 5533descriptors <DIEnumerator>`, each representing the definition of an enumeration 5534value for the set. All enumeration type descriptors are collected in the 5535``enums:`` field of the :ref:`compile unit <DICompileUnit>`. 5536 5537For ``DW_TAG_structure_type``, ``DW_TAG_class_type``, and 5538``DW_TAG_union_type``, the ``elements:`` should be :ref:`derived types 5539<DIDerivedType>` with ``tag: DW_TAG_member``, ``tag: DW_TAG_inheritance``, or 5540``tag: DW_TAG_friend``; or :ref:`subprograms <DISubprogram>` with 5541``isDefinition: false``. 5542 5543.. _DISubrange: 5544 5545DISubrange 5546"""""""""" 5547 5548``DISubrange`` nodes are the elements for ``DW_TAG_array_type`` variants of 5549:ref:`DICompositeType`. 5550 5551- ``count: -1`` indicates an empty array. 5552- ``count: !10`` describes the count with a :ref:`DILocalVariable`. 5553- ``count: !12`` describes the count with a :ref:`DIGlobalVariable`. 5554 5555.. code-block:: text 5556 5557 !0 = !DISubrange(count: 5, lowerBound: 0) ; array counting from 0 5558 !1 = !DISubrange(count: 5, lowerBound: 1) ; array counting from 1 5559 !2 = !DISubrange(count: -1) ; empty array. 5560 5561 ; Scopes used in rest of example 5562 !6 = !DIFile(filename: "vla.c", directory: "/path/to/file") 5563 !7 = distinct !DICompileUnit(language: DW_LANG_C99, file: !6) 5564 !8 = distinct !DISubprogram(name: "foo", scope: !7, file: !6, line: 5) 5565 5566 ; Use of local variable as count value 5567 !9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed) 5568 !10 = !DILocalVariable(name: "count", scope: !8, file: !6, line: 42, type: !9) 5569 !11 = !DISubrange(count: !10, lowerBound: 0) 5570 5571 ; Use of global variable as count value 5572 !12 = !DIGlobalVariable(name: "count", scope: !8, file: !6, line: 22, type: !9) 5573 !13 = !DISubrange(count: !12, lowerBound: 0) 5574 5575.. _DIEnumerator: 5576 5577DIEnumerator 5578"""""""""""" 5579 5580``DIEnumerator`` nodes are the elements for ``DW_TAG_enumeration_type`` 5581variants of :ref:`DICompositeType`. 5582 5583.. code-block:: text 5584 5585 !0 = !DIEnumerator(name: "SixKind", value: 7) 5586 !1 = !DIEnumerator(name: "SevenKind", value: 7) 5587 !2 = !DIEnumerator(name: "NegEightKind", value: -8) 5588 5589DITemplateTypeParameter 5590""""""""""""""""""""""" 5591 5592``DITemplateTypeParameter`` nodes represent type parameters to generic source 5593language constructs. They are used (optionally) in :ref:`DICompositeType` and 5594:ref:`DISubprogram` ``templateParams:`` fields. 5595 5596.. code-block:: text 5597 5598 !0 = !DITemplateTypeParameter(name: "Ty", type: !1) 5599 5600DITemplateValueParameter 5601"""""""""""""""""""""""" 5602 5603``DITemplateValueParameter`` nodes represent value parameters to generic source 5604language constructs. ``tag:`` defaults to ``DW_TAG_template_value_parameter``, 5605but if specified can also be set to ``DW_TAG_GNU_template_template_param`` or 5606``DW_TAG_GNU_template_param_pack``. They are used (optionally) in 5607:ref:`DICompositeType` and :ref:`DISubprogram` ``templateParams:`` fields. 5608 5609.. code-block:: text 5610 5611 !0 = !DITemplateValueParameter(name: "Ty", type: !1, value: i32 7) 5612 5613DINamespace 5614""""""""""" 5615 5616``DINamespace`` nodes represent namespaces in the source language. 5617 5618.. code-block:: text 5619 5620 !0 = !DINamespace(name: "myawesomeproject", scope: !1, file: !2, line: 7) 5621 5622.. _DIGlobalVariable: 5623 5624DIGlobalVariable 5625"""""""""""""""" 5626 5627``DIGlobalVariable`` nodes represent global variables in the source language. 5628 5629.. code-block:: text 5630 5631 @foo = global i32, !dbg !0 5632 !0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression()) 5633 !1 = !DIGlobalVariable(name: "foo", linkageName: "foo", scope: !2, 5634 file: !3, line: 7, type: !4, isLocal: true, 5635 isDefinition: false, declaration: !5) 5636 5637 5638DIGlobalVariableExpression 5639"""""""""""""""""""""""""" 5640 5641``DIGlobalVariableExpression`` nodes tie a :ref:`DIGlobalVariable` together 5642with a :ref:`DIExpression`. 5643 5644.. code-block:: text 5645 5646 @lower = global i32, !dbg !0 5647 @upper = global i32, !dbg !1 5648 !0 = !DIGlobalVariableExpression( 5649 var: !2, 5650 expr: !DIExpression(DW_OP_LLVM_fragment, 0, 32) 5651 ) 5652 !1 = !DIGlobalVariableExpression( 5653 var: !2, 5654 expr: !DIExpression(DW_OP_LLVM_fragment, 32, 32) 5655 ) 5656 !2 = !DIGlobalVariable(name: "split64", linkageName: "split64", scope: !3, 5657 file: !4, line: 8, type: !5, declaration: !6) 5658 5659All global variable expressions should be referenced by the `globals:` field of 5660a :ref:`compile unit <DICompileUnit>`. 5661 5662.. _DISubprogram: 5663 5664DISubprogram 5665"""""""""""" 5666 5667``DISubprogram`` nodes represent functions from the source language. A distinct 5668``DISubprogram`` may be attached to a function definition using ``!dbg`` 5669metadata. A unique ``DISubprogram`` may be attached to a function declaration 5670used for call site debug info. The ``retainedNodes:`` field is a list of 5671:ref:`variables <DILocalVariable>` and :ref:`labels <DILabel>` that must be 5672retained, even if their IR counterparts are optimized out of the IR. The 5673``type:`` field must point at an :ref:`DISubroutineType`. 5674 5675.. _DISubprogramDeclaration: 5676 5677When ``isDefinition: false``, subprograms describe a declaration in the type 5678tree as opposed to a definition of a function. If the scope is a composite 5679type with an ODR ``identifier:`` and that does not set ``flags: DIFwdDecl``, 5680then the subprogram declaration is uniqued based only on its ``linkageName:`` 5681and ``scope:``. 5682 5683.. code-block:: text 5684 5685 define void @_Z3foov() !dbg !0 { 5686 ... 5687 } 5688 5689 !0 = distinct !DISubprogram(name: "foo", linkageName: "_Zfoov", scope: !1, 5690 file: !2, line: 7, type: !3, isLocal: true, 5691 isDefinition: true, scopeLine: 8, 5692 containingType: !4, 5693 virtuality: DW_VIRTUALITY_pure_virtual, 5694 virtualIndex: 10, flags: DIFlagPrototyped, 5695 isOptimized: true, unit: !5, templateParams: !6, 5696 declaration: !7, retainedNodes: !8, 5697 thrownTypes: !9) 5698 5699.. _DILexicalBlock: 5700 5701DILexicalBlock 5702"""""""""""""" 5703 5704``DILexicalBlock`` nodes describe nested blocks within a :ref:`subprogram 5705<DISubprogram>`. The line number and column numbers are used to distinguish 5706two lexical blocks at same depth. They are valid targets for ``scope:`` 5707fields. 5708 5709.. code-block:: text 5710 5711 !0 = distinct !DILexicalBlock(scope: !1, file: !2, line: 7, column: 35) 5712 5713Usually lexical blocks are ``distinct`` to prevent node merging based on 5714operands. 5715 5716.. _DILexicalBlockFile: 5717 5718DILexicalBlockFile 5719"""""""""""""""""" 5720 5721``DILexicalBlockFile`` nodes are used to discriminate between sections of a 5722:ref:`lexical block <DILexicalBlock>`. The ``file:`` field can be changed to 5723indicate textual inclusion, or the ``discriminator:`` field can be used to 5724discriminate between control flow within a single block in the source language. 5725 5726.. code-block:: text 5727 5728 !0 = !DILexicalBlock(scope: !3, file: !4, line: 7, column: 35) 5729 !1 = !DILexicalBlockFile(scope: !0, file: !4, discriminator: 0) 5730 !2 = !DILexicalBlockFile(scope: !0, file: !4, discriminator: 1) 5731 5732.. _DILocation: 5733 5734DILocation 5735"""""""""" 5736 5737``DILocation`` nodes represent source debug locations. The ``scope:`` field is 5738mandatory, and points at an :ref:`DILexicalBlockFile`, an 5739:ref:`DILexicalBlock`, or an :ref:`DISubprogram`. 5740 5741.. code-block:: text 5742 5743 !0 = !DILocation(line: 2900, column: 42, scope: !1, inlinedAt: !2) 5744 5745.. _DILocalVariable: 5746 5747DILocalVariable 5748""""""""""""""" 5749 5750``DILocalVariable`` nodes represent local variables in the source language. If 5751the ``arg:`` field is set to non-zero, then this variable is a subprogram 5752parameter, and it will be included in the ``retainedNodes:`` field of its 5753:ref:`DISubprogram`. 5754 5755.. code-block:: text 5756 5757 !0 = !DILocalVariable(name: "this", arg: 1, scope: !3, file: !2, line: 7, 5758 type: !3, flags: DIFlagArtificial) 5759 !1 = !DILocalVariable(name: "x", arg: 2, scope: !4, file: !2, line: 7, 5760 type: !3) 5761 !2 = !DILocalVariable(name: "y", scope: !5, file: !2, line: 7, type: !3) 5762 5763.. _DIExpression: 5764 5765DIExpression 5766"""""""""""" 5767 5768``DIExpression`` nodes represent expressions that are inspired by the DWARF 5769expression language. They are used in :ref:`debug intrinsics<dbg_intrinsics>` 5770(such as ``llvm.dbg.declare`` and ``llvm.dbg.value``) to describe how the 5771referenced LLVM variable relates to the source language variable. Debug 5772intrinsics are interpreted left-to-right: start by pushing the value/address 5773operand of the intrinsic onto a stack, then repeatedly push and evaluate 5774opcodes from the DIExpression until the final variable description is produced. 5775 5776The current supported opcode vocabulary is limited: 5777 5778- ``DW_OP_deref`` dereferences the top of the expression stack. 5779- ``DW_OP_plus`` pops the last two entries from the expression stack, adds 5780 them together and appends the result to the expression stack. 5781- ``DW_OP_minus`` pops the last two entries from the expression stack, subtracts 5782 the last entry from the second last entry and appends the result to the 5783 expression stack. 5784- ``DW_OP_plus_uconst, 93`` adds ``93`` to the working expression. 5785- ``DW_OP_LLVM_fragment, 16, 8`` specifies the offset and size (``16`` and ``8`` 5786 here, respectively) of the variable fragment from the working expression. Note 5787 that contrary to DW_OP_bit_piece, the offset is describing the location 5788 within the described source variable. 5789- ``DW_OP_LLVM_convert, 16, DW_ATE_signed`` specifies a bit size and encoding 5790 (``16`` and ``DW_ATE_signed`` here, respectively) to which the top of the 5791 expression stack is to be converted. Maps into a ``DW_OP_convert`` operation 5792 that references a base type constructed from the supplied values. 5793- ``DW_OP_LLVM_tag_offset, tag_offset`` specifies that a memory tag should be 5794 optionally applied to the pointer. The memory tag is derived from the 5795 given tag offset in an implementation-defined manner. 5796- ``DW_OP_swap`` swaps top two stack entries. 5797- ``DW_OP_xderef`` provides extended dereference mechanism. The entry at the top 5798 of the stack is treated as an address. The second stack entry is treated as an 5799 address space identifier. 5800- ``DW_OP_stack_value`` marks a constant value. 5801- ``DW_OP_LLVM_entry_value, N`` may only appear in MIR and at the 5802 beginning of a ``DIExpression``. In DWARF a ``DBG_VALUE`` 5803 instruction binding a ``DIExpression(DW_OP_LLVM_entry_value`` to a 5804 register is lowered to a ``DW_OP_entry_value [reg]``, pushing the 5805 value the register had upon function entry onto the stack. The next 5806 ``(N - 1)`` operations will be part of the ``DW_OP_entry_value`` 5807 block argument. For example, ``!DIExpression(DW_OP_LLVM_entry_value, 5808 1, DW_OP_plus_uconst, 123, DW_OP_stack_value)`` specifies an 5809 expression where the entry value of the debug value instruction's 5810 value/address operand is pushed to the stack, and is added 5811 with 123. Due to framework limitations ``N`` can currently only 5812 be 1. 5813 5814 The operation is introduced by the ``LiveDebugValues`` pass, which 5815 applies it only to function parameters that are unmodified 5816 throughout the function. Support is limited to simple register 5817 location descriptions, or as indirect locations (e.g., when a struct 5818 is passed-by-value to a callee via a pointer to a temporary copy 5819 made in the caller). The entry value op is also introduced by the 5820 ``AsmPrinter`` pass when a call site parameter value 5821 (``DW_AT_call_site_parameter_value``) is represented as entry value 5822 of the parameter. 5823- ``DW_OP_LLVM_arg, N`` is used in debug intrinsics that refer to more than one 5824 value, such as one that calculates the sum of two registers. This is always 5825 used in combination with an ordered list of values, such that 5826 ``DW_OP_LLVM_arg, N`` refers to the ``N``th element in that list. For 5827 example, ``!DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_minus, 5828 DW_OP_stack_value)`` used with the list ``(%reg1, %reg2)`` would evaluate to 5829 ``%reg1 - reg2``. This list of values should be provided by the containing 5830 intrinsic/instruction. 5831- ``DW_OP_breg`` (or ``DW_OP_bregx``) represents a content on the provided 5832 signed offset of the specified register. The opcode is only generated by the 5833 ``AsmPrinter`` pass to describe call site parameter value which requires an 5834 expression over two registers. 5835- ``DW_OP_push_object_address`` pushes the address of the object which can then 5836 serve as a descriptor in subsequent calculation. This opcode can be used to 5837 calculate bounds of fortran allocatable array which has array descriptors. 5838- ``DW_OP_over`` duplicates the entry currently second in the stack at the top 5839 of the stack. This opcode can be used to calculate bounds of fortran assumed 5840 rank array which has rank known at run time and current dimension number is 5841 implicitly first element of the stack. 5842- ``DW_OP_LLVM_implicit_pointer`` It specifies the dereferenced value. It can 5843 be used to represent pointer variables which are optimized out but the value 5844 it points to is known. This operator is required as it is different than DWARF 5845 operator DW_OP_implicit_pointer in representation and specification (number 5846 and types of operands) and later can not be used as multiple level. 5847 5848.. code-block:: text 5849 5850 IR for "*ptr = 4;" 5851 -------------- 5852 call void @llvm.dbg.value(metadata i32 4, metadata !17, metadata !20) 5853 !17 = !DILocalVariable(name: "ptr1", scope: !12, file: !3, line: 5, 5854 type: !18) 5855 !18 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !19, size: 64) 5856 !19 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed) 5857 !20 = !DIExpression(DW_OP_LLVM_implicit_pointer)) 5858 5859 IR for "**ptr = 4;" 5860 -------------- 5861 call void @llvm.dbg.value(metadata i32 4, metadata !17, metadata !21) 5862 !17 = !DILocalVariable(name: "ptr1", scope: !12, file: !3, line: 5, 5863 type: !18) 5864 !18 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !19, size: 64) 5865 !19 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !20, size: 64) 5866 !20 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed) 5867 !21 = !DIExpression(DW_OP_LLVM_implicit_pointer, 5868 DW_OP_LLVM_implicit_pointer)) 5869 5870DWARF specifies three kinds of simple location descriptions: Register, memory, 5871and implicit location descriptions. Note that a location description is 5872defined over certain ranges of a program, i.e the location of a variable may 5873change over the course of the program. Register and memory location 5874descriptions describe the *concrete location* of a source variable (in the 5875sense that a debugger might modify its value), whereas *implicit locations* 5876describe merely the actual *value* of a source variable which might not exist 5877in registers or in memory (see ``DW_OP_stack_value``). 5878 5879A ``llvm.dbg.addr`` or ``llvm.dbg.declare`` intrinsic describes an indirect 5880value (the address) of a source variable. The first operand of the intrinsic 5881must be an address of some kind. A DIExpression attached to the intrinsic 5882refines this address to produce a concrete location for the source variable. 5883 5884A ``llvm.dbg.value`` intrinsic describes the direct value of a source variable. 5885The first operand of the intrinsic may be a direct or indirect value. A 5886DIExpression attached to the intrinsic refines the first operand to produce a 5887direct value. For example, if the first operand is an indirect value, it may be 5888necessary to insert ``DW_OP_deref`` into the DIExpression in order to produce a 5889valid debug intrinsic. 5890 5891.. note:: 5892 5893 A DIExpression is interpreted in the same way regardless of which kind of 5894 debug intrinsic it's attached to. 5895 5896.. code-block:: text 5897 5898 !0 = !DIExpression(DW_OP_deref) 5899 !1 = !DIExpression(DW_OP_plus_uconst, 3) 5900 !1 = !DIExpression(DW_OP_constu, 3, DW_OP_plus) 5901 !2 = !DIExpression(DW_OP_bit_piece, 3, 7) 5902 !3 = !DIExpression(DW_OP_deref, DW_OP_constu, 3, DW_OP_plus, DW_OP_LLVM_fragment, 3, 7) 5903 !4 = !DIExpression(DW_OP_constu, 2, DW_OP_swap, DW_OP_xderef) 5904 !5 = !DIExpression(DW_OP_constu, 42, DW_OP_stack_value) 5905 5906DIArgList 5907"""""""""""" 5908 5909``DIArgList`` nodes hold a list of constant or SSA value references. These are 5910used in :ref:`debug intrinsics<dbg_intrinsics>` (currently only in 5911``llvm.dbg.value``) in combination with a ``DIExpression`` that uses the 5912``DW_OP_LLVM_arg`` operator. Because a DIArgList may refer to local values 5913within a function, it must only be used as a function argument, must always be 5914inlined, and cannot appear in named metadata. 5915 5916.. code-block:: text 5917 5918 llvm.dbg.value(metadata !DIArgList(i32 %a, i32 %b), 5919 metadata !16, 5920 metadata !DIExpression(DW_OP_LLVM_arg, 0, DW_OP_LLVM_arg, 1, DW_OP_plus)) 5921 5922DIFlags 5923""""""""""""""" 5924 5925These flags encode various properties of DINodes. 5926 5927The `ExportSymbols` flag marks a class, struct or union whose members 5928may be referenced as if they were defined in the containing class or 5929union. This flag is used to decide whether the DW_AT_export_symbols can 5930be used for the structure type. 5931 5932DIObjCProperty 5933"""""""""""""" 5934 5935``DIObjCProperty`` nodes represent Objective-C property nodes. 5936 5937.. code-block:: text 5938 5939 !3 = !DIObjCProperty(name: "foo", file: !1, line: 7, setter: "setFoo", 5940 getter: "getFoo", attributes: 7, type: !2) 5941 5942DIImportedEntity 5943"""""""""""""""" 5944 5945``DIImportedEntity`` nodes represent entities (such as modules) imported into a 5946compile unit. The ``elements`` field is a list of renamed entities (such as 5947variables and subprograms) in the imported entity (such as module). 5948 5949.. code-block:: text 5950 5951 !2 = !DIImportedEntity(tag: DW_TAG_imported_module, name: "foo", scope: !0, 5952 entity: !1, line: 7, elements: !3) 5953 !3 = !{!4} 5954 !4 = !DIImportedEntity(tag: DW_TAG_imported_declaration, name: "bar", scope: !0, 5955 entity: !5, line: 7) 5956 5957DIMacro 5958""""""" 5959 5960``DIMacro`` nodes represent definition or undefinition of a macro identifiers. 5961The ``name:`` field is the macro identifier, followed by macro parameters when 5962defining a function-like macro, and the ``value`` field is the token-string 5963used to expand the macro identifier. 5964 5965.. code-block:: text 5966 5967 !2 = !DIMacro(macinfo: DW_MACINFO_define, line: 7, name: "foo(x)", 5968 value: "((x) + 1)") 5969 !3 = !DIMacro(macinfo: DW_MACINFO_undef, line: 30, name: "foo") 5970 5971DIMacroFile 5972""""""""""" 5973 5974``DIMacroFile`` nodes represent inclusion of source files. 5975The ``nodes:`` field is a list of ``DIMacro`` and ``DIMacroFile`` nodes that 5976appear in the included source file. 5977 5978.. code-block:: text 5979 5980 !2 = !DIMacroFile(macinfo: DW_MACINFO_start_file, line: 7, file: !2, 5981 nodes: !3) 5982 5983.. _DILabel: 5984 5985DILabel 5986""""""" 5987 5988``DILabel`` nodes represent labels within a :ref:`DISubprogram`. All fields of 5989a ``DILabel`` are mandatory. The ``scope:`` field must be one of either a 5990:ref:`DILexicalBlockFile`, a :ref:`DILexicalBlock`, or a :ref:`DISubprogram`. 5991The ``name:`` field is the label identifier. The ``file:`` field is the 5992:ref:`DIFile` the label is present in. The ``line:`` field is the source line 5993within the file where the label is declared. 5994 5995.. code-block:: text 5996 5997 !2 = !DILabel(scope: !0, name: "foo", file: !1, line: 7) 5998 5999'``tbaa``' Metadata 6000^^^^^^^^^^^^^^^^^^^ 6001 6002In LLVM IR, memory does not have types, so LLVM's own type system is not 6003suitable for doing type based alias analysis (TBAA). Instead, metadata is 6004added to the IR to describe a type system of a higher level language. This 6005can be used to implement C/C++ strict type aliasing rules, but it can also 6006be used to implement custom alias analysis behavior for other languages. 6007 6008This description of LLVM's TBAA system is broken into two parts: 6009:ref:`Semantics<tbaa_node_semantics>` talks about high level issues, and 6010:ref:`Representation<tbaa_node_representation>` talks about the metadata 6011encoding of various entities. 6012 6013It is always possible to trace any TBAA node to a "root" TBAA node (details 6014in the :ref:`Representation<tbaa_node_representation>` section). TBAA 6015nodes with different roots have an unknown aliasing relationship, and LLVM 6016conservatively infers ``MayAlias`` between them. The rules mentioned in 6017this section only pertain to TBAA nodes living under the same root. 6018 6019.. _tbaa_node_semantics: 6020 6021Semantics 6022""""""""" 6023 6024The TBAA metadata system, referred to as "struct path TBAA" (not to be 6025confused with ``tbaa.struct``), consists of the following high level 6026concepts: *Type Descriptors*, further subdivided into scalar type 6027descriptors and struct type descriptors; and *Access Tags*. 6028 6029**Type descriptors** describe the type system of the higher level language 6030being compiled. **Scalar type descriptors** describe types that do not 6031contain other types. Each scalar type has a parent type, which must also 6032be a scalar type or the TBAA root. Via this parent relation, scalar types 6033within a TBAA root form a tree. **Struct type descriptors** denote types 6034that contain a sequence of other type descriptors, at known offsets. These 6035contained type descriptors can either be struct type descriptors themselves 6036or scalar type descriptors. 6037 6038**Access tags** are metadata nodes attached to load and store instructions. 6039Access tags use type descriptors to describe the *location* being accessed 6040in terms of the type system of the higher level language. Access tags are 6041tuples consisting of a base type, an access type and an offset. The base 6042type is a scalar type descriptor or a struct type descriptor, the access 6043type is a scalar type descriptor, and the offset is a constant integer. 6044 6045The access tag ``(BaseTy, AccessTy, Offset)`` can describe one of two 6046things: 6047 6048 * If ``BaseTy`` is a struct type, the tag describes a memory access (load 6049 or store) of a value of type ``AccessTy`` contained in the struct type 6050 ``BaseTy`` at offset ``Offset``. 6051 6052 * If ``BaseTy`` is a scalar type, ``Offset`` must be 0 and ``BaseTy`` and 6053 ``AccessTy`` must be the same; and the access tag describes a scalar 6054 access with scalar type ``AccessTy``. 6055 6056We first define an ``ImmediateParent`` relation on ``(BaseTy, Offset)`` 6057tuples this way: 6058 6059 * If ``BaseTy`` is a scalar type then ``ImmediateParent(BaseTy, 0)`` is 6060 ``(ParentTy, 0)`` where ``ParentTy`` is the parent of the scalar type as 6061 described in the TBAA metadata. ``ImmediateParent(BaseTy, Offset)`` is 6062 undefined if ``Offset`` is non-zero. 6063 6064 * If ``BaseTy`` is a struct type then ``ImmediateParent(BaseTy, Offset)`` 6065 is ``(NewTy, NewOffset)`` where ``NewTy`` is the type contained in 6066 ``BaseTy`` at offset ``Offset`` and ``NewOffset`` is ``Offset`` adjusted 6067 to be relative within that inner type. 6068 6069A memory access with an access tag ``(BaseTy1, AccessTy1, Offset1)`` 6070aliases a memory access with an access tag ``(BaseTy2, AccessTy2, 6071Offset2)`` if either ``(BaseTy1, Offset1)`` is reachable from ``(Base2, 6072Offset2)`` via the ``Parent`` relation or vice versa. 6073 6074As a concrete example, the type descriptor graph for the following program 6075 6076.. code-block:: c 6077 6078 struct Inner { 6079 int i; // offset 0 6080 float f; // offset 4 6081 }; 6082 6083 struct Outer { 6084 float f; // offset 0 6085 double d; // offset 4 6086 struct Inner inner_a; // offset 12 6087 }; 6088 6089 void f(struct Outer* outer, struct Inner* inner, float* f, int* i, char* c) { 6090 outer->f = 0; // tag0: (OuterStructTy, FloatScalarTy, 0) 6091 outer->inner_a.i = 0; // tag1: (OuterStructTy, IntScalarTy, 12) 6092 outer->inner_a.f = 0.0; // tag2: (OuterStructTy, FloatScalarTy, 16) 6093 *f = 0.0; // tag3: (FloatScalarTy, FloatScalarTy, 0) 6094 } 6095 6096is (note that in C and C++, ``char`` can be used to access any arbitrary 6097type): 6098 6099.. code-block:: text 6100 6101 Root = "TBAA Root" 6102 CharScalarTy = ("char", Root, 0) 6103 FloatScalarTy = ("float", CharScalarTy, 0) 6104 DoubleScalarTy = ("double", CharScalarTy, 0) 6105 IntScalarTy = ("int", CharScalarTy, 0) 6106 InnerStructTy = {"Inner" (IntScalarTy, 0), (FloatScalarTy, 4)} 6107 OuterStructTy = {"Outer", (FloatScalarTy, 0), (DoubleScalarTy, 4), 6108 (InnerStructTy, 12)} 6109 6110 6111with (e.g.) ``ImmediateParent(OuterStructTy, 12)`` = ``(InnerStructTy, 61120)``, ``ImmediateParent(InnerStructTy, 0)`` = ``(IntScalarTy, 0)``, and 6113``ImmediateParent(IntScalarTy, 0)`` = ``(CharScalarTy, 0)``. 6114 6115.. _tbaa_node_representation: 6116 6117Representation 6118"""""""""""""" 6119 6120The root node of a TBAA type hierarchy is an ``MDNode`` with 0 operands or 6121with exactly one ``MDString`` operand. 6122 6123Scalar type descriptors are represented as an ``MDNode`` s with two 6124operands. The first operand is an ``MDString`` denoting the name of the 6125struct type. LLVM does not assign meaning to the value of this operand, it 6126only cares about it being an ``MDString``. The second operand is an 6127``MDNode`` which points to the parent for said scalar type descriptor, 6128which is either another scalar type descriptor or the TBAA root. Scalar 6129type descriptors can have an optional third argument, but that must be the 6130constant integer zero. 6131 6132Struct type descriptors are represented as ``MDNode`` s with an odd number 6133of operands greater than 1. The first operand is an ``MDString`` denoting 6134the name of the struct type. Like in scalar type descriptors the actual 6135value of this name operand is irrelevant to LLVM. After the name operand, 6136the struct type descriptors have a sequence of alternating ``MDNode`` and 6137``ConstantInt`` operands. With N starting from 1, the 2N - 1 th operand, 6138an ``MDNode``, denotes a contained field, and the 2N th operand, a 6139``ConstantInt``, is the offset of the said contained field. The offsets 6140must be in non-decreasing order. 6141 6142Access tags are represented as ``MDNode`` s with either 3 or 4 operands. 6143The first operand is an ``MDNode`` pointing to the node representing the 6144base type. The second operand is an ``MDNode`` pointing to the node 6145representing the access type. The third operand is a ``ConstantInt`` that 6146states the offset of the access. If a fourth field is present, it must be 6147a ``ConstantInt`` valued at 0 or 1. If it is 1 then the access tag states 6148that the location being accessed is "constant" (meaning 6149``pointsToConstantMemory`` should return true; see `other useful 6150AliasAnalysis methods <AliasAnalysis.html#OtherItfs>`_). The TBAA root of 6151the access type and the base type of an access tag must be the same, and 6152that is the TBAA root of the access tag. 6153 6154'``tbaa.struct``' Metadata 6155^^^^^^^^^^^^^^^^^^^^^^^^^^ 6156 6157The :ref:`llvm.memcpy <int_memcpy>` is often used to implement 6158aggregate assignment operations in C and similar languages, however it 6159is defined to copy a contiguous region of memory, which is more than 6160strictly necessary for aggregate types which contain holes due to 6161padding. Also, it doesn't contain any TBAA information about the fields 6162of the aggregate. 6163 6164``!tbaa.struct`` metadata can describe which memory subregions in a 6165memcpy are padding and what the TBAA tags of the struct are. 6166 6167The current metadata format is very simple. ``!tbaa.struct`` metadata 6168nodes are a list of operands which are in conceptual groups of three. 6169For each group of three, the first operand gives the byte offset of a 6170field in bytes, the second gives its size in bytes, and the third gives 6171its tbaa tag. e.g.: 6172 6173.. code-block:: llvm 6174 6175 !4 = !{ i64 0, i64 4, !1, i64 8, i64 4, !2 } 6176 6177This describes a struct with two fields. The first is at offset 0 bytes 6178with size 4 bytes, and has tbaa tag !1. The second is at offset 8 bytes 6179and has size 4 bytes and has tbaa tag !2. 6180 6181Note that the fields need not be contiguous. In this example, there is a 61824 byte gap between the two fields. This gap represents padding which 6183does not carry useful data and need not be preserved. 6184 6185'``noalias``' and '``alias.scope``' Metadata 6186^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6187 6188``noalias`` and ``alias.scope`` metadata provide the ability to specify generic 6189noalias memory-access sets. This means that some collection of memory access 6190instructions (loads, stores, memory-accessing calls, etc.) that carry 6191``noalias`` metadata can specifically be specified not to alias with some other 6192collection of memory access instructions that carry ``alias.scope`` metadata. 6193Each type of metadata specifies a list of scopes where each scope has an id and 6194a domain. 6195 6196When evaluating an aliasing query, if for some domain, the set 6197of scopes with that domain in one instruction's ``alias.scope`` list is a 6198subset of (or equal to) the set of scopes for that domain in another 6199instruction's ``noalias`` list, then the two memory accesses are assumed not to 6200alias. 6201 6202Because scopes in one domain don't affect scopes in other domains, separate 6203domains can be used to compose multiple independent noalias sets. This is 6204used for example during inlining. As the noalias function parameters are 6205turned into noalias scope metadata, a new domain is used every time the 6206function is inlined. 6207 6208The metadata identifying each domain is itself a list containing one or two 6209entries. The first entry is the name of the domain. Note that if the name is a 6210string then it can be combined across functions and translation units. A 6211self-reference can be used to create globally unique domain names. A 6212descriptive string may optionally be provided as a second list entry. 6213 6214The metadata identifying each scope is also itself a list containing two or 6215three entries. The first entry is the name of the scope. Note that if the name 6216is a string then it can be combined across functions and translation units. A 6217self-reference can be used to create globally unique scope names. A metadata 6218reference to the scope's domain is the second entry. A descriptive string may 6219optionally be provided as a third list entry. 6220 6221For example, 6222 6223.. code-block:: llvm 6224 6225 ; Two scope domains: 6226 !0 = !{!0} 6227 !1 = !{!1} 6228 6229 ; Some scopes in these domains: 6230 !2 = !{!2, !0} 6231 !3 = !{!3, !0} 6232 !4 = !{!4, !1} 6233 6234 ; Some scope lists: 6235 !5 = !{!4} ; A list containing only scope !4 6236 !6 = !{!4, !3, !2} 6237 !7 = !{!3} 6238 6239 ; These two instructions don't alias: 6240 %0 = load float, float* %c, align 4, !alias.scope !5 6241 store float %0, float* %arrayidx.i, align 4, !noalias !5 6242 6243 ; These two instructions also don't alias (for domain !1, the set of scopes 6244 ; in the !alias.scope equals that in the !noalias list): 6245 %2 = load float, float* %c, align 4, !alias.scope !5 6246 store float %2, float* %arrayidx.i2, align 4, !noalias !6 6247 6248 ; These two instructions may alias (for domain !0, the set of scopes in 6249 ; the !noalias list is not a superset of, or equal to, the scopes in the 6250 ; !alias.scope list): 6251 %2 = load float, float* %c, align 4, !alias.scope !6 6252 store float %0, float* %arrayidx.i, align 4, !noalias !7 6253 6254'``fpmath``' Metadata 6255^^^^^^^^^^^^^^^^^^^^^ 6256 6257``fpmath`` metadata may be attached to any instruction of floating-point 6258type. It can be used to express the maximum acceptable error in the 6259result of that instruction, in ULPs, thus potentially allowing the 6260compiler to use a more efficient but less accurate method of computing 6261it. ULP is defined as follows: 6262 6263 If ``x`` is a real number that lies between two finite consecutive 6264 floating-point numbers ``a`` and ``b``, without being equal to one 6265 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the 6266 distance between the two non-equal finite floating-point numbers 6267 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``. 6268 6269The metadata node shall consist of a single positive float type number 6270representing the maximum relative error, for example: 6271 6272.. code-block:: llvm 6273 6274 !0 = !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs 6275 6276.. _range-metadata: 6277 6278'``range``' Metadata 6279^^^^^^^^^^^^^^^^^^^^ 6280 6281``range`` metadata may be attached only to ``load``, ``call`` and ``invoke`` of 6282integer types. It expresses the possible ranges the loaded value or the value 6283returned by the called function at this call site is in. If the loaded or 6284returned value is not in the specified range, the behavior is undefined. The 6285ranges are represented with a flattened list of integers. The loaded value or 6286the value returned is known to be in the union of the ranges defined by each 6287consecutive pair. Each pair has the following properties: 6288 6289- The type must match the type loaded by the instruction. 6290- The pair ``a,b`` represents the range ``[a,b)``. 6291- Both ``a`` and ``b`` are constants. 6292- The range is allowed to wrap. 6293- The range should not represent the full or empty set. That is, 6294 ``a!=b``. 6295 6296In addition, the pairs must be in signed order of the lower bound and 6297they must be non-contiguous. 6298 6299Examples: 6300 6301.. code-block:: llvm 6302 6303 %a = load i8, i8* %x, align 1, !range !0 ; Can only be 0 or 1 6304 %b = load i8, i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1 6305 %c = call i8 @foo(), !range !2 ; Can only be 0, 1, 3, 4 or 5 6306 %d = invoke i8 @bar() to label %cont 6307 unwind label %lpad, !range !3 ; Can only be -2, -1, 3, 4 or 5 6308 ... 6309 !0 = !{ i8 0, i8 2 } 6310 !1 = !{ i8 255, i8 2 } 6311 !2 = !{ i8 0, i8 2, i8 3, i8 6 } 6312 !3 = !{ i8 -2, i8 0, i8 3, i8 6 } 6313 6314'``absolute_symbol``' Metadata 6315^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6316 6317``absolute_symbol`` metadata may be attached to a global variable 6318declaration. It marks the declaration as a reference to an absolute symbol, 6319which causes the backend to use absolute relocations for the symbol even 6320in position independent code, and expresses the possible ranges that the 6321global variable's *address* (not its value) is in, in the same format as 6322``range`` metadata, with the extension that the pair ``all-ones,all-ones`` 6323may be used to represent the full set. 6324 6325Example (assuming 64-bit pointers): 6326 6327.. code-block:: llvm 6328 6329 @a = external global i8, !absolute_symbol !0 ; Absolute symbol in range [0,256) 6330 @b = external global i8, !absolute_symbol !1 ; Absolute symbol in range [0,2^64) 6331 6332 ... 6333 !0 = !{ i64 0, i64 256 } 6334 !1 = !{ i64 -1, i64 -1 } 6335 6336'``callees``' Metadata 6337^^^^^^^^^^^^^^^^^^^^^^ 6338 6339``callees`` metadata may be attached to indirect call sites. If ``callees`` 6340metadata is attached to a call site, and any callee is not among the set of 6341functions provided by the metadata, the behavior is undefined. The intent of 6342this metadata is to facilitate optimizations such as indirect-call promotion. 6343For example, in the code below, the call instruction may only target the 6344``add`` or ``sub`` functions: 6345 6346.. code-block:: llvm 6347 6348 %result = call i64 %binop(i64 %x, i64 %y), !callees !0 6349 6350 ... 6351 !0 = !{i64 (i64, i64)* @add, i64 (i64, i64)* @sub} 6352 6353'``callback``' Metadata 6354^^^^^^^^^^^^^^^^^^^^^^^ 6355 6356``callback`` metadata may be attached to a function declaration, or definition. 6357(Call sites are excluded only due to the lack of a use case.) For ease of 6358exposition, we'll refer to the function annotated w/ metadata as a broker 6359function. The metadata describes how the arguments of a call to the broker are 6360in turn passed to the callback function specified by the metadata. Thus, the 6361``callback`` metadata provides a partial description of a call site inside the 6362broker function with regards to the arguments of a call to the broker. The only 6363semantic restriction on the broker function itself is that it is not allowed to 6364inspect or modify arguments referenced in the ``callback`` metadata as 6365pass-through to the callback function. 6366 6367The broker is not required to actually invoke the callback function at runtime. 6368However, the assumptions about not inspecting or modifying arguments that would 6369be passed to the specified callback function still hold, even if the callback 6370function is not dynamically invoked. The broker is allowed to invoke the 6371callback function more than once per invocation of the broker. The broker is 6372also allowed to invoke (directly or indirectly) the function passed as a 6373callback through another use. Finally, the broker is also allowed to relay the 6374callback callee invocation to a different thread. 6375 6376The metadata is structured as follows: At the outer level, ``callback`` 6377metadata is a list of ``callback`` encodings. Each encoding starts with a 6378constant ``i64`` which describes the argument position of the callback function 6379in the call to the broker. The following elements, except the last, describe 6380what arguments are passed to the callback function. Each element is again an 6381``i64`` constant identifying the argument of the broker that is passed through, 6382or ``i64 -1`` to indicate an unknown or inspected argument. The order in which 6383they are listed has to be the same in which they are passed to the callback 6384callee. The last element of the encoding is a boolean which specifies how 6385variadic arguments of the broker are handled. If it is true, all variadic 6386arguments of the broker are passed through to the callback function *after* the 6387arguments encoded explicitly before. 6388 6389In the code below, the ``pthread_create`` function is marked as a broker 6390through the ``!callback !1`` metadata. In the example, there is only one 6391callback encoding, namely ``!2``, associated with the broker. This encoding 6392identifies the callback function as the second argument of the broker (``i64 63932``) and the sole argument of the callback function as the third one of the 6394broker function (``i64 3``). 6395 6396.. FIXME why does the llvm-sphinx-docs builder give a highlighting 6397 error if the below is set to highlight as 'llvm', despite that we 6398 have misc.highlighting_failure set? 6399 6400.. code-block:: text 6401 6402 declare !callback !1 dso_local i32 @pthread_create(i64*, %union.pthread_attr_t*, i8* (i8*)*, i8*) 6403 6404 ... 6405 !2 = !{i64 2, i64 3, i1 false} 6406 !1 = !{!2} 6407 6408Another example is shown below. The callback callee is the second argument of 6409the ``__kmpc_fork_call`` function (``i64 2``). The callee is given two unknown 6410values (each identified by a ``i64 -1``) and afterwards all 6411variadic arguments that are passed to the ``__kmpc_fork_call`` call (due to the 6412final ``i1 true``). 6413 6414.. FIXME why does the llvm-sphinx-docs builder give a highlighting 6415 error if the below is set to highlight as 'llvm', despite that we 6416 have misc.highlighting_failure set? 6417 6418.. code-block:: text 6419 6420 declare !callback !0 dso_local void @__kmpc_fork_call(%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) 6421 6422 ... 6423 !1 = !{i64 2, i64 -1, i64 -1, i1 true} 6424 !0 = !{!1} 6425 6426'``exclude``' Metadata 6427^^^^^^^^^^^^^^^^^^^^^^ 6428 6429``exclude`` metadata may be attached to a global variable to signify that its 6430section should not be included in the final executable or shared library. This 6431option is only valid for global variables with an explicit section targeting ELF 6432or COFF. This is done using the ``SHF_EXCLUDE`` flag on ELF targets and the 6433``IMAGE_SCN_LNK_REMOVE`` and ``IMAGE_SCN_MEM_DISCARDABLE`` flags for COFF 6434targets. Additionally, this metadata is only used as a flag, so the associated 6435node must be empty. The explicit section should not conflict with any other 6436sections that the user does not want removed after linking. 6437 6438.. code-block:: text 6439 6440 @object = private constant [1 x i8] c"\00", section ".foo" !exclude !0 6441 6442 ... 6443 !0 = !{} 6444 6445'``unpredictable``' Metadata 6446^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6447 6448``unpredictable`` metadata may be attached to any branch or switch 6449instruction. It can be used to express the unpredictability of control 6450flow. Similar to the llvm.expect intrinsic, it may be used to alter 6451optimizations related to compare and branch instructions. The metadata 6452is treated as a boolean value; if it exists, it signals that the branch 6453or switch that it is attached to is completely unpredictable. 6454 6455.. _md_dereferenceable: 6456 6457'``dereferenceable``' Metadata 6458^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6459 6460The existence of the ``!dereferenceable`` metadata on the instruction 6461tells the optimizer that the value loaded is known to be dereferenceable. 6462The number of bytes known to be dereferenceable is specified by the integer 6463value in the metadata node. This is analogous to the ''dereferenceable'' 6464attribute on parameters and return values. 6465 6466.. _md_dereferenceable_or_null: 6467 6468'``dereferenceable_or_null``' Metadata 6469^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6470 6471The existence of the ``!dereferenceable_or_null`` metadata on the 6472instruction tells the optimizer that the value loaded is known to be either 6473dereferenceable or null. 6474The number of bytes known to be dereferenceable is specified by the integer 6475value in the metadata node. This is analogous to the ''dereferenceable_or_null'' 6476attribute on parameters and return values. 6477 6478.. _llvm.loop: 6479 6480'``llvm.loop``' 6481^^^^^^^^^^^^^^^ 6482 6483It is sometimes useful to attach information to loop constructs. Currently, 6484loop metadata is implemented as metadata attached to the branch instruction 6485in the loop latch block. The loop metadata node is a list of 6486other metadata nodes, each representing a property of the loop. Usually, 6487the first item of the property node is a string. For example, the 6488``llvm.loop.unroll.count`` suggests an unroll factor to the loop 6489unroller: 6490 6491.. code-block:: llvm 6492 6493 br i1 %exitcond, label %._crit_edge, label %.lr.ph, !llvm.loop !0 6494 ... 6495 !0 = !{!0, !1, !2} 6496 !1 = !{!"llvm.loop.unroll.enable"} 6497 !2 = !{!"llvm.loop.unroll.count", i32 4} 6498 6499For legacy reasons, the first item of a loop metadata node must be a 6500reference to itself. Before the advent of the 'distinct' keyword, this 6501forced the preservation of otherwise identical metadata nodes. Since 6502the loop-metadata node can be attached to multiple nodes, the 'distinct' 6503keyword has become unnecessary. 6504 6505Prior to the property nodes, one or two ``DILocation`` (debug location) 6506nodes can be present in the list. The first, if present, identifies the 6507source-code location where the loop begins. The second, if present, 6508identifies the source-code location where the loop ends. 6509 6510Loop metadata nodes cannot be used as unique identifiers. They are 6511neither persistent for the same loop through transformations nor 6512necessarily unique to just one loop. 6513 6514'``llvm.loop.disable_nonforced``' 6515^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6516 6517This metadata disables all optional loop transformations unless 6518explicitly instructed using other transformation metadata such as 6519``llvm.loop.unroll.enable``. That is, no heuristic will try to determine 6520whether a transformation is profitable. The purpose is to avoid that the 6521loop is transformed to a different loop before an explicitly requested 6522(forced) transformation is applied. For instance, loop fusion can make 6523other transformations impossible. Mandatory loop canonicalizations such 6524as loop rotation are still applied. 6525 6526It is recommended to use this metadata in addition to any llvm.loop.* 6527transformation directive. Also, any loop should have at most one 6528directive applied to it (and a sequence of transformations built using 6529followup-attributes). Otherwise, which transformation will be applied 6530depends on implementation details such as the pass pipeline order. 6531 6532See :ref:`transformation-metadata` for details. 6533 6534'``llvm.loop.vectorize``' and '``llvm.loop.interleave``' 6535^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6536 6537Metadata prefixed with ``llvm.loop.vectorize`` or ``llvm.loop.interleave`` are 6538used to control per-loop vectorization and interleaving parameters such as 6539vectorization width and interleave count. These metadata should be used in 6540conjunction with ``llvm.loop`` loop identification metadata. The 6541``llvm.loop.vectorize`` and ``llvm.loop.interleave`` metadata are only 6542optimization hints and the optimizer will only interleave and vectorize loops if 6543it believes it is safe to do so. The ``llvm.loop.parallel_accesses`` metadata 6544which contains information about loop-carried memory dependencies can be helpful 6545in determining the safety of these transformations. 6546 6547'``llvm.loop.interleave.count``' Metadata 6548^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6549 6550This metadata suggests an interleave count to the loop interleaver. 6551The first operand is the string ``llvm.loop.interleave.count`` and the 6552second operand is an integer specifying the interleave count. For 6553example: 6554 6555.. code-block:: llvm 6556 6557 !0 = !{!"llvm.loop.interleave.count", i32 4} 6558 6559Note that setting ``llvm.loop.interleave.count`` to 1 disables interleaving 6560multiple iterations of the loop. If ``llvm.loop.interleave.count`` is set to 0 6561then the interleave count will be determined automatically. 6562 6563'``llvm.loop.vectorize.enable``' Metadata 6564^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6565 6566This metadata selectively enables or disables vectorization for the loop. The 6567first operand is the string ``llvm.loop.vectorize.enable`` and the second operand 6568is a bit. If the bit operand value is 1 vectorization is enabled. A value of 65690 disables vectorization: 6570 6571.. code-block:: llvm 6572 6573 !0 = !{!"llvm.loop.vectorize.enable", i1 0} 6574 !1 = !{!"llvm.loop.vectorize.enable", i1 1} 6575 6576'``llvm.loop.vectorize.predicate.enable``' Metadata 6577^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6578 6579This metadata selectively enables or disables creating predicated instructions 6580for the loop, which can enable folding of the scalar epilogue loop into the 6581main loop. The first operand is the string 6582``llvm.loop.vectorize.predicate.enable`` and the second operand is a bit. If 6583the bit operand value is 1 vectorization is enabled. A value of 0 disables 6584vectorization: 6585 6586.. code-block:: llvm 6587 6588 !0 = !{!"llvm.loop.vectorize.predicate.enable", i1 0} 6589 !1 = !{!"llvm.loop.vectorize.predicate.enable", i1 1} 6590 6591'``llvm.loop.vectorize.scalable.enable``' Metadata 6592^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6593 6594This metadata selectively enables or disables scalable vectorization for the 6595loop, and only has any effect if vectorization for the loop is already enabled. 6596The first operand is the string ``llvm.loop.vectorize.scalable.enable`` 6597and the second operand is a bit. If the bit operand value is 1 scalable 6598vectorization is enabled, whereas a value of 0 reverts to the default fixed 6599width vectorization: 6600 6601.. code-block:: llvm 6602 6603 !0 = !{!"llvm.loop.vectorize.scalable.enable", i1 0} 6604 !1 = !{!"llvm.loop.vectorize.scalable.enable", i1 1} 6605 6606'``llvm.loop.vectorize.width``' Metadata 6607^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6608 6609This metadata sets the target width of the vectorizer. The first 6610operand is the string ``llvm.loop.vectorize.width`` and the second 6611operand is an integer specifying the width. For example: 6612 6613.. code-block:: llvm 6614 6615 !0 = !{!"llvm.loop.vectorize.width", i32 4} 6616 6617Note that setting ``llvm.loop.vectorize.width`` to 1 disables 6618vectorization of the loop. If ``llvm.loop.vectorize.width`` is set to 66190 or if the loop does not have this metadata the width will be 6620determined automatically. 6621 6622'``llvm.loop.vectorize.followup_vectorized``' Metadata 6623^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6624 6625This metadata defines which loop attributes the vectorized loop will 6626have. See :ref:`transformation-metadata` for details. 6627 6628'``llvm.loop.vectorize.followup_epilogue``' Metadata 6629^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6630 6631This metadata defines which loop attributes the epilogue will have. The 6632epilogue is not vectorized and is executed when either the vectorized 6633loop is not known to preserve semantics (because e.g., it processes two 6634arrays that are found to alias by a runtime check) or for the last 6635iterations that do not fill a complete set of vector lanes. See 6636:ref:`Transformation Metadata <transformation-metadata>` for details. 6637 6638'``llvm.loop.vectorize.followup_all``' Metadata 6639^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6640 6641Attributes in the metadata will be added to both the vectorized and 6642epilogue loop. 6643See :ref:`Transformation Metadata <transformation-metadata>` for details. 6644 6645'``llvm.loop.unroll``' 6646^^^^^^^^^^^^^^^^^^^^^^ 6647 6648Metadata prefixed with ``llvm.loop.unroll`` are loop unrolling 6649optimization hints such as the unroll factor. ``llvm.loop.unroll`` 6650metadata should be used in conjunction with ``llvm.loop`` loop 6651identification metadata. The ``llvm.loop.unroll`` metadata are only 6652optimization hints and the unrolling will only be performed if the 6653optimizer believes it is safe to do so. 6654 6655'``llvm.loop.unroll.count``' Metadata 6656^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6657 6658This metadata suggests an unroll factor to the loop unroller. The 6659first operand is the string ``llvm.loop.unroll.count`` and the second 6660operand is a positive integer specifying the unroll factor. For 6661example: 6662 6663.. code-block:: llvm 6664 6665 !0 = !{!"llvm.loop.unroll.count", i32 4} 6666 6667If the trip count of the loop is less than the unroll count the loop 6668will be partially unrolled. 6669 6670'``llvm.loop.unroll.disable``' Metadata 6671^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6672 6673This metadata disables loop unrolling. The metadata has a single operand 6674which is the string ``llvm.loop.unroll.disable``. For example: 6675 6676.. code-block:: llvm 6677 6678 !0 = !{!"llvm.loop.unroll.disable"} 6679 6680'``llvm.loop.unroll.runtime.disable``' Metadata 6681^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6682 6683This metadata disables runtime loop unrolling. The metadata has a single 6684operand which is the string ``llvm.loop.unroll.runtime.disable``. For example: 6685 6686.. code-block:: llvm 6687 6688 !0 = !{!"llvm.loop.unroll.runtime.disable"} 6689 6690'``llvm.loop.unroll.enable``' Metadata 6691^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6692 6693This metadata suggests that the loop should be fully unrolled if the trip count 6694is known at compile time and partially unrolled if the trip count is not known 6695at compile time. The metadata has a single operand which is the string 6696``llvm.loop.unroll.enable``. For example: 6697 6698.. code-block:: llvm 6699 6700 !0 = !{!"llvm.loop.unroll.enable"} 6701 6702'``llvm.loop.unroll.full``' Metadata 6703^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6704 6705This metadata suggests that the loop should be unrolled fully. The 6706metadata has a single operand which is the string ``llvm.loop.unroll.full``. 6707For example: 6708 6709.. code-block:: llvm 6710 6711 !0 = !{!"llvm.loop.unroll.full"} 6712 6713'``llvm.loop.unroll.followup``' Metadata 6714^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6715 6716This metadata defines which loop attributes the unrolled loop will have. 6717See :ref:`Transformation Metadata <transformation-metadata>` for details. 6718 6719'``llvm.loop.unroll.followup_remainder``' Metadata 6720^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6721 6722This metadata defines which loop attributes the remainder loop after 6723partial/runtime unrolling will have. See 6724:ref:`Transformation Metadata <transformation-metadata>` for details. 6725 6726'``llvm.loop.unroll_and_jam``' 6727^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6728 6729This metadata is treated very similarly to the ``llvm.loop.unroll`` metadata 6730above, but affect the unroll and jam pass. In addition any loop with 6731``llvm.loop.unroll`` metadata but no ``llvm.loop.unroll_and_jam`` metadata will 6732disable unroll and jam (so ``llvm.loop.unroll`` metadata will be left to the 6733unroller, plus ``llvm.loop.unroll.disable`` metadata will disable unroll and jam 6734too.) 6735 6736The metadata for unroll and jam otherwise is the same as for ``unroll``. 6737``llvm.loop.unroll_and_jam.enable``, ``llvm.loop.unroll_and_jam.disable`` and 6738``llvm.loop.unroll_and_jam.count`` do the same as for unroll. 6739``llvm.loop.unroll_and_jam.full`` is not supported. Again these are only hints 6740and the normal safety checks will still be performed. 6741 6742'``llvm.loop.unroll_and_jam.count``' Metadata 6743^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6744 6745This metadata suggests an unroll and jam factor to use, similarly to 6746``llvm.loop.unroll.count``. The first operand is the string 6747``llvm.loop.unroll_and_jam.count`` and the second operand is a positive integer 6748specifying the unroll factor. For example: 6749 6750.. code-block:: llvm 6751 6752 !0 = !{!"llvm.loop.unroll_and_jam.count", i32 4} 6753 6754If the trip count of the loop is less than the unroll count the loop 6755will be partially unroll and jammed. 6756 6757'``llvm.loop.unroll_and_jam.disable``' Metadata 6758^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6759 6760This metadata disables loop unroll and jamming. The metadata has a single 6761operand which is the string ``llvm.loop.unroll_and_jam.disable``. For example: 6762 6763.. code-block:: llvm 6764 6765 !0 = !{!"llvm.loop.unroll_and_jam.disable"} 6766 6767'``llvm.loop.unroll_and_jam.enable``' Metadata 6768^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6769 6770This metadata suggests that the loop should be fully unroll and jammed if the 6771trip count is known at compile time and partially unrolled if the trip count is 6772not known at compile time. The metadata has a single operand which is the 6773string ``llvm.loop.unroll_and_jam.enable``. For example: 6774 6775.. code-block:: llvm 6776 6777 !0 = !{!"llvm.loop.unroll_and_jam.enable"} 6778 6779'``llvm.loop.unroll_and_jam.followup_outer``' Metadata 6780^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6781 6782This metadata defines which loop attributes the outer unrolled loop will 6783have. See :ref:`Transformation Metadata <transformation-metadata>` for 6784details. 6785 6786'``llvm.loop.unroll_and_jam.followup_inner``' Metadata 6787^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6788 6789This metadata defines which loop attributes the inner jammed loop will 6790have. See :ref:`Transformation Metadata <transformation-metadata>` for 6791details. 6792 6793'``llvm.loop.unroll_and_jam.followup_remainder_outer``' Metadata 6794^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6795 6796This metadata defines which attributes the epilogue of the outer loop 6797will have. This loop is usually unrolled, meaning there is no such 6798loop. This attribute will be ignored in this case. See 6799:ref:`Transformation Metadata <transformation-metadata>` for details. 6800 6801'``llvm.loop.unroll_and_jam.followup_remainder_inner``' Metadata 6802^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6803 6804This metadata defines which attributes the inner loop of the epilogue 6805will have. The outer epilogue will usually be unrolled, meaning there 6806can be multiple inner remainder loops. See 6807:ref:`Transformation Metadata <transformation-metadata>` for details. 6808 6809'``llvm.loop.unroll_and_jam.followup_all``' Metadata 6810^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6811 6812Attributes specified in the metadata is added to all 6813``llvm.loop.unroll_and_jam.*`` loops. See 6814:ref:`Transformation Metadata <transformation-metadata>` for details. 6815 6816'``llvm.loop.licm_versioning.disable``' Metadata 6817^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6818 6819This metadata indicates that the loop should not be versioned for the purpose 6820of enabling loop-invariant code motion (LICM). The metadata has a single operand 6821which is the string ``llvm.loop.licm_versioning.disable``. For example: 6822 6823.. code-block:: llvm 6824 6825 !0 = !{!"llvm.loop.licm_versioning.disable"} 6826 6827'``llvm.loop.distribute.enable``' Metadata 6828^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6829 6830Loop distribution allows splitting a loop into multiple loops. Currently, 6831this is only performed if the entire loop cannot be vectorized due to unsafe 6832memory dependencies. The transformation will attempt to isolate the unsafe 6833dependencies into their own loop. 6834 6835This metadata can be used to selectively enable or disable distribution of the 6836loop. The first operand is the string ``llvm.loop.distribute.enable`` and the 6837second operand is a bit. If the bit operand value is 1 distribution is 6838enabled. A value of 0 disables distribution: 6839 6840.. code-block:: llvm 6841 6842 !0 = !{!"llvm.loop.distribute.enable", i1 0} 6843 !1 = !{!"llvm.loop.distribute.enable", i1 1} 6844 6845This metadata should be used in conjunction with ``llvm.loop`` loop 6846identification metadata. 6847 6848'``llvm.loop.distribute.followup_coincident``' Metadata 6849^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6850 6851This metadata defines which attributes extracted loops with no cyclic 6852dependencies will have (i.e. can be vectorized). See 6853:ref:`Transformation Metadata <transformation-metadata>` for details. 6854 6855'``llvm.loop.distribute.followup_sequential``' Metadata 6856^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6857 6858This metadata defines which attributes the isolated loops with unsafe 6859memory dependencies will have. See 6860:ref:`Transformation Metadata <transformation-metadata>` for details. 6861 6862'``llvm.loop.distribute.followup_fallback``' Metadata 6863^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6864 6865If loop versioning is necessary, this metadata defined the attributes 6866the non-distributed fallback version will have. See 6867:ref:`Transformation Metadata <transformation-metadata>` for details. 6868 6869'``llvm.loop.distribute.followup_all``' Metadata 6870^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6871 6872The attributes in this metadata is added to all followup loops of the 6873loop distribution pass. See 6874:ref:`Transformation Metadata <transformation-metadata>` for details. 6875 6876'``llvm.licm.disable``' Metadata 6877^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6878 6879This metadata indicates that loop-invariant code motion (LICM) should not be 6880performed on this loop. The metadata has a single operand which is the string 6881``llvm.licm.disable``. For example: 6882 6883.. code-block:: llvm 6884 6885 !0 = !{!"llvm.licm.disable"} 6886 6887Note that although it operates per loop it isn't given the llvm.loop prefix 6888as it is not affected by the ``llvm.loop.disable_nonforced`` metadata. 6889 6890'``llvm.access.group``' Metadata 6891^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6892 6893``llvm.access.group`` metadata can be attached to any instruction that 6894potentially accesses memory. It can point to a single distinct metadata 6895node, which we call access group. This node represents all memory access 6896instructions referring to it via ``llvm.access.group``. When an 6897instruction belongs to multiple access groups, it can also point to a 6898list of accesses groups, illustrated by the following example. 6899 6900.. code-block:: llvm 6901 6902 %val = load i32, i32* %arrayidx, !llvm.access.group !0 6903 ... 6904 !0 = !{!1, !2} 6905 !1 = distinct !{} 6906 !2 = distinct !{} 6907 6908It is illegal for the list node to be empty since it might be confused 6909with an access group. 6910 6911The access group metadata node must be 'distinct' to avoid collapsing 6912multiple access groups by content. A access group metadata node must 6913always be empty which can be used to distinguish an access group 6914metadata node from a list of access groups. Being empty avoids the 6915situation that the content must be updated which, because metadata is 6916immutable by design, would required finding and updating all references 6917to the access group node. 6918 6919The access group can be used to refer to a memory access instruction 6920without pointing to it directly (which is not possible in global 6921metadata). Currently, the only metadata making use of it is 6922``llvm.loop.parallel_accesses``. 6923 6924'``llvm.loop.parallel_accesses``' Metadata 6925^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 6926 6927The ``llvm.loop.parallel_accesses`` metadata refers to one or more 6928access group metadata nodes (see ``llvm.access.group``). It denotes that 6929no loop-carried memory dependence exist between it and other instructions 6930in the loop with this metadata. 6931 6932Let ``m1`` and ``m2`` be two instructions that both have the 6933``llvm.access.group`` metadata to the access group ``g1``, respectively 6934``g2`` (which might be identical). If a loop contains both access groups 6935in its ``llvm.loop.parallel_accesses`` metadata, then the compiler can 6936assume that there is no dependency between ``m1`` and ``m2`` carried by 6937this loop. Instructions that belong to multiple access groups are 6938considered having this property if at least one of the access groups 6939matches the ``llvm.loop.parallel_accesses`` list. 6940 6941If all memory-accessing instructions in a loop have 6942``llvm.access.group`` metadata that each refer to one of the access 6943groups of a loop's ``llvm.loop.parallel_accesses`` metadata, then the 6944loop has no loop carried memory dependences and is considered to be a 6945parallel loop. 6946 6947Note that if not all memory access instructions belong to an access 6948group referred to by ``llvm.loop.parallel_accesses``, then the loop must 6949not be considered trivially parallel. Additional 6950memory dependence analysis is required to make that determination. As a fail 6951safe mechanism, this causes loops that were originally parallel to be considered 6952sequential (if optimization passes that are unaware of the parallel semantics 6953insert new memory instructions into the loop body). 6954 6955Example of a loop that is considered parallel due to its correct use of 6956both ``llvm.access.group`` and ``llvm.loop.parallel_accesses`` 6957metadata types. 6958 6959.. code-block:: llvm 6960 6961 for.body: 6962 ... 6963 %val0 = load i32, i32* %arrayidx, !llvm.access.group !1 6964 ... 6965 store i32 %val0, i32* %arrayidx1, !llvm.access.group !1 6966 ... 6967 br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !0 6968 6969 for.end: 6970 ... 6971 !0 = distinct !{!0, !{!"llvm.loop.parallel_accesses", !1}} 6972 !1 = distinct !{} 6973 6974It is also possible to have nested parallel loops: 6975 6976.. code-block:: llvm 6977 6978 outer.for.body: 6979 ... 6980 %val1 = load i32, i32* %arrayidx3, !llvm.access.group !4 6981 ... 6982 br label %inner.for.body 6983 6984 inner.for.body: 6985 ... 6986 %val0 = load i32, i32* %arrayidx1, !llvm.access.group !3 6987 ... 6988 store i32 %val0, i32* %arrayidx2, !llvm.access.group !3 6989 ... 6990 br i1 %exitcond, label %inner.for.end, label %inner.for.body, !llvm.loop !1 6991 6992 inner.for.end: 6993 ... 6994 store i32 %val1, i32* %arrayidx4, !llvm.access.group !4 6995 ... 6996 br i1 %exitcond, label %outer.for.end, label %outer.for.body, !llvm.loop !2 6997 6998 outer.for.end: ; preds = %for.body 6999 ... 7000 !1 = distinct !{!1, !{!"llvm.loop.parallel_accesses", !3}} ; metadata for the inner loop 7001 !2 = distinct !{!2, !{!"llvm.loop.parallel_accesses", !3, !4}} ; metadata for the outer loop 7002 !3 = distinct !{} ; access group for instructions in the inner loop (which are implicitly contained in outer loop as well) 7003 !4 = distinct !{} ; access group for instructions in the outer, but not the inner loop 7004 7005.. _langref_llvm_loop_mustprogress: 7006 7007'``llvm.loop.mustprogress``' Metadata 7008^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7009 7010The ``llvm.loop.mustprogress`` metadata indicates that this loop is required to 7011terminate, unwind, or interact with the environment in an observable way e.g. 7012via a volatile memory access, I/O, or other synchronization. If such a loop is 7013not found to interact with the environment in an observable way, the loop may 7014be removed. This corresponds to the ``mustprogress`` function attribute. 7015 7016'``irr_loop``' Metadata 7017^^^^^^^^^^^^^^^^^^^^^^^ 7018 7019``irr_loop`` metadata may be attached to the terminator instruction of a basic 7020block that's an irreducible loop header (note that an irreducible loop has more 7021than once header basic blocks.) If ``irr_loop`` metadata is attached to the 7022terminator instruction of a basic block that is not really an irreducible loop 7023header, the behavior is undefined. The intent of this metadata is to improve the 7024accuracy of the block frequency propagation. For example, in the code below, the 7025block ``header0`` may have a loop header weight (relative to the other headers of 7026the irreducible loop) of 100: 7027 7028.. code-block:: llvm 7029 7030 header0: 7031 ... 7032 br i1 %cmp, label %t1, label %t2, !irr_loop !0 7033 7034 ... 7035 !0 = !{"loop_header_weight", i64 100} 7036 7037Irreducible loop header weights are typically based on profile data. 7038 7039.. _md_invariant.group: 7040 7041'``invariant.group``' Metadata 7042^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7043 7044The experimental ``invariant.group`` metadata may be attached to 7045``load``/``store`` instructions referencing a single metadata with no entries. 7046The existence of the ``invariant.group`` metadata on the instruction tells 7047the optimizer that every ``load`` and ``store`` to the same pointer operand 7048can be assumed to load or store the same 7049value (but see the ``llvm.launder.invariant.group`` intrinsic which affects 7050when two pointers are considered the same). Pointers returned by bitcast or 7051getelementptr with only zero indices are considered the same. 7052 7053Examples: 7054 7055.. code-block:: llvm 7056 7057 @unknownPtr = external global i8 7058 ... 7059 %ptr = alloca i8 7060 store i8 42, i8* %ptr, !invariant.group !0 7061 call void @foo(i8* %ptr) 7062 7063 %a = load i8, i8* %ptr, !invariant.group !0 ; Can assume that value under %ptr didn't change 7064 call void @foo(i8* %ptr) 7065 7066 %newPtr = call i8* @getPointer(i8* %ptr) 7067 %c = load i8, i8* %newPtr, !invariant.group !0 ; Can't assume anything, because we only have information about %ptr 7068 7069 %unknownValue = load i8, i8* @unknownPtr 7070 store i8 %unknownValue, i8* %ptr, !invariant.group !0 ; Can assume that %unknownValue == 42 7071 7072 call void @foo(i8* %ptr) 7073 %newPtr2 = call i8* @llvm.launder.invariant.group(i8* %ptr) 7074 %d = load i8, i8* %newPtr2, !invariant.group !0 ; Can't step through launder.invariant.group to get value of %ptr 7075 7076 ... 7077 declare void @foo(i8*) 7078 declare i8* @getPointer(i8*) 7079 declare i8* @llvm.launder.invariant.group(i8*) 7080 7081 !0 = !{} 7082 7083The invariant.group metadata must be dropped when replacing one pointer by 7084another based on aliasing information. This is because invariant.group is tied 7085to the SSA value of the pointer operand. 7086 7087.. code-block:: llvm 7088 7089 %v = load i8, i8* %x, !invariant.group !0 7090 ; if %x mustalias %y then we can replace the above instruction with 7091 %v = load i8, i8* %y 7092 7093Note that this is an experimental feature, which means that its semantics might 7094change in the future. 7095 7096'``type``' Metadata 7097^^^^^^^^^^^^^^^^^^^ 7098 7099See :doc:`TypeMetadata`. 7100 7101'``associated``' Metadata 7102^^^^^^^^^^^^^^^^^^^^^^^^^ 7103 7104The ``associated`` metadata may be attached to a global variable definition with 7105a single argument that references a global object (optionally through an alias). 7106 7107This metadata lowers to the ELF section flag ``SHF_LINK_ORDER`` which prevents 7108discarding of the global variable in linker GC unless the referenced object is 7109also discarded. The linker support for this feature is spotty. For best 7110compatibility, globals carrying this metadata should: 7111 7112- Be in ``@llvm.compiler.used``. 7113- If the referenced global variable is in a comdat, be in the same comdat. 7114 7115``!associated`` can not express many-to-one relationship. A global variable with 7116the metadata should generally not be referenced by a function: the function may 7117be inlined into other functions, leading to more references to the metadata. 7118Ideally we would want to keep metadata alive as long as any inline location is 7119alive, but this many-to-one relationship is not representable. Moreover, if the 7120metadata is retained while the function is discarded, the linker will report an 7121error of a relocation referencing a discarded section. 7122 7123The metadata is often used with an explicit section consisting of valid C 7124identifiers so that the runtime can find the metadata section with 7125linker-defined encapsulation symbols ``__start_<section_name>`` and 7126``__stop_<section_name>``. 7127 7128It does not have any effect on non-ELF targets. 7129 7130Example: 7131 7132.. code-block:: text 7133 7134 $a = comdat any 7135 @a = global i32 1, comdat $a 7136 @b = internal global i32 2, comdat $a, section "abc", !associated !0 7137 !0 = !{i32* @a} 7138 7139 7140'``prof``' Metadata 7141^^^^^^^^^^^^^^^^^^^ 7142 7143The ``prof`` metadata is used to record profile data in the IR. 7144The first operand of the metadata node indicates the profile metadata 7145type. There are currently 3 types: 7146:ref:`branch_weights<prof_node_branch_weights>`, 7147:ref:`function_entry_count<prof_node_function_entry_count>`, and 7148:ref:`VP<prof_node_VP>`. 7149 7150.. _prof_node_branch_weights: 7151 7152branch_weights 7153"""""""""""""" 7154 7155Branch weight metadata attached to a branch, select, switch or call instruction 7156represents the likeliness of the associated branch being taken. 7157For more information, see :doc:`BranchWeightMetadata`. 7158 7159.. _prof_node_function_entry_count: 7160 7161function_entry_count 7162"""""""""""""""""""" 7163 7164Function entry count metadata can be attached to function definitions 7165to record the number of times the function is called. Used with BFI 7166information, it is also used to derive the basic block profile count. 7167For more information, see :doc:`BranchWeightMetadata`. 7168 7169.. _prof_node_VP: 7170 7171VP 7172"" 7173 7174VP (value profile) metadata can be attached to instructions that have 7175value profile information. Currently this is indirect calls (where it 7176records the hottest callees) and calls to memory intrinsics such as memcpy, 7177memmove, and memset (where it records the hottest byte lengths). 7178 7179Each VP metadata node contains "VP" string, then a uint32_t value for the value 7180profiling kind, a uint64_t value for the total number of times the instruction 7181is executed, followed by uint64_t value and execution count pairs. 7182The value profiling kind is 0 for indirect call targets and 1 for memory 7183operations. For indirect call targets, each profile value is a hash 7184of the callee function name, and for memory operations each value is the 7185byte length. 7186 7187Note that the value counts do not need to add up to the total count 7188listed in the third operand (in practice only the top hottest values 7189are tracked and reported). 7190 7191Indirect call example: 7192 7193.. code-block:: llvm 7194 7195 call void %f(), !prof !1 7196 !1 = !{!"VP", i32 0, i64 1600, i64 7651369219802541373, i64 1030, i64 -4377547752858689819, i64 410} 7197 7198Note that the VP type is 0 (the second operand), which indicates this is 7199an indirect call value profile data. The third operand indicates that the 7200indirect call executed 1600 times. The 4th and 6th operands give the 7201hashes of the 2 hottest target functions' names (this is the same hash used 7202to represent function names in the profile database), and the 5th and 7th 7203operands give the execution count that each of the respective prior target 7204functions was called. 7205 7206.. _md_annotation: 7207 7208'``annotation``' Metadata 7209^^^^^^^^^^^^^^^^^^^^^^^^^ 7210 7211The ``annotation`` metadata can be used to attach a tuple of annotation strings 7212to any instruction. This metadata does not impact the semantics of the program 7213and may only be used to provide additional insight about the program and 7214transformations to users. 7215 7216Example: 7217 7218.. code-block:: text 7219 7220 %a.addr = alloca float*, align 8, !annotation !0 7221 !0 = !{!"auto-init"} 7222 7223'``func_sanitize``' Metadata 7224^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7225 7226The ``func_sanitize`` metadata is used to attach two values for the function 7227sanitizer instrumentation. The first value is the ubsan function signature. 7228The second value is the address of the proxy variable which stores the address 7229of the RTTI descriptor. If :ref:`prologue <prologuedata>` and '``func_sanitize``' 7230are used at the same time, :ref:`prologue <prologuedata>` is emitted before 7231'``func_sanitize``' in the output. 7232 7233Example: 7234 7235.. code-block:: text 7236 7237 @__llvm_rtti_proxy = private unnamed_addr constant i8* bitcast ({ i8*, i8* }* @_ZTIFvvE to i8*) 7238 define void @_Z3funv() !func_sanitize !0 { 7239 return void 7240 } 7241 !0 = !{i32 846595819, i8** @__llvm_rtti_proxy} 7242 7243Module Flags Metadata 7244===================== 7245 7246Information about the module as a whole is difficult to convey to LLVM's 7247subsystems. The LLVM IR isn't sufficient to transmit this information. 7248The ``llvm.module.flags`` named metadata exists in order to facilitate 7249this. These flags are in the form of key / value pairs --- much like a 7250dictionary --- making it easy for any subsystem who cares about a flag to 7251look it up. 7252 7253The ``llvm.module.flags`` metadata contains a list of metadata triplets. 7254Each triplet has the following form: 7255 7256- The first element is a *behavior* flag, which specifies the behavior 7257 when two (or more) modules are merged together, and it encounters two 7258 (or more) metadata with the same ID. The supported behaviors are 7259 described below. 7260- The second element is a metadata string that is a unique ID for the 7261 metadata. Each module may only have one flag entry for each unique ID (not 7262 including entries with the **Require** behavior). 7263- The third element is the value of the flag. 7264 7265When two (or more) modules are merged together, the resulting 7266``llvm.module.flags`` metadata is the union of the modules' flags. That is, for 7267each unique metadata ID string, there will be exactly one entry in the merged 7268modules ``llvm.module.flags`` metadata table, and the value for that entry will 7269be determined by the merge behavior flag, as described below. The only exception 7270is that entries with the *Require* behavior are always preserved. 7271 7272The following behaviors are supported: 7273 7274.. list-table:: 7275 :header-rows: 1 7276 :widths: 10 90 7277 7278 * - Value 7279 - Behavior 7280 7281 * - 1 7282 - **Error** 7283 Emits an error if two values disagree, otherwise the resulting value 7284 is that of the operands. 7285 7286 * - 2 7287 - **Warning** 7288 Emits a warning if two values disagree. The result value will be the 7289 operand for the flag from the first module being linked, or the max 7290 if the other module uses **Max** (in which case the resulting flag 7291 will be **Max**). 7292 7293 * - 3 7294 - **Require** 7295 Adds a requirement that another module flag be present and have a 7296 specified value after linking is performed. The value must be a 7297 metadata pair, where the first element of the pair is the ID of the 7298 module flag to be restricted, and the second element of the pair is 7299 the value the module flag should be restricted to. This behavior can 7300 be used to restrict the allowable results (via triggering of an 7301 error) of linking IDs with the **Override** behavior. 7302 7303 * - 4 7304 - **Override** 7305 Uses the specified value, regardless of the behavior or value of the 7306 other module. If both modules specify **Override**, but the values 7307 differ, an error will be emitted. 7308 7309 * - 5 7310 - **Append** 7311 Appends the two values, which are required to be metadata nodes. 7312 7313 * - 6 7314 - **AppendUnique** 7315 Appends the two values, which are required to be metadata 7316 nodes. However, duplicate entries in the second list are dropped 7317 during the append operation. 7318 7319 * - 7 7320 - **Max** 7321 Takes the max of the two values, which are required to be integers. 7322 7323It is an error for a particular unique flag ID to have multiple behaviors, 7324except in the case of **Require** (which adds restrictions on another metadata 7325value) or **Override**. 7326 7327An example of module flags: 7328 7329.. code-block:: llvm 7330 7331 !0 = !{ i32 1, !"foo", i32 1 } 7332 !1 = !{ i32 4, !"bar", i32 37 } 7333 !2 = !{ i32 2, !"qux", i32 42 } 7334 !3 = !{ i32 3, !"qux", 7335 !{ 7336 !"foo", i32 1 7337 } 7338 } 7339 !llvm.module.flags = !{ !0, !1, !2, !3 } 7340 7341- Metadata ``!0`` has the ID ``!"foo"`` and the value '1'. The behavior 7342 if two or more ``!"foo"`` flags are seen is to emit an error if their 7343 values are not equal. 7344 7345- Metadata ``!1`` has the ID ``!"bar"`` and the value '37'. The 7346 behavior if two or more ``!"bar"`` flags are seen is to use the value 7347 '37'. 7348 7349- Metadata ``!2`` has the ID ``!"qux"`` and the value '42'. The 7350 behavior if two or more ``!"qux"`` flags are seen is to emit a 7351 warning if their values are not equal. 7352 7353- Metadata ``!3`` has the ID ``!"qux"`` and the value: 7354 7355 :: 7356 7357 !{ !"foo", i32 1 } 7358 7359 The behavior is to emit an error if the ``llvm.module.flags`` does not 7360 contain a flag with the ID ``!"foo"`` that has the value '1' after linking is 7361 performed. 7362 7363Synthesized Functions Module Flags Metadata 7364------------------------------------------- 7365 7366These metadata specify the default attributes synthesized functions should have. 7367These metadata are currently respected by a few instrumentation passes, such as 7368sanitizers. 7369 7370These metadata correspond to a few function attributes with significant code 7371generation behaviors. Function attributes with just optimization purposes 7372should not be listed because the performance impact of these synthesized 7373functions is small. 7374 7375- "frame-pointer": **Max**. The value can be 0, 1, or 2. A synthesized function 7376 will get the "frame-pointer" function attribute, with value being "none", 7377 "non-leaf", or "all", respectively. 7378- "function_return_thunk_extern": The synthesized function will get the 7379 ``fn_return_thunk_extern`` function attribute. 7380- "uwtable": **Max**. The value can be 0, 1, or 2. If the value is 1, a synthesized 7381 function will get the ``uwtable(sync)`` function attribute, if the value is 2, 7382 a synthesized function will get the ``uwtable(async)`` function attribute. 7383 7384Objective-C Garbage Collection Module Flags Metadata 7385---------------------------------------------------- 7386 7387On the Mach-O platform, Objective-C stores metadata about garbage 7388collection in a special section called "image info". The metadata 7389consists of a version number and a bitmask specifying what types of 7390garbage collection are supported (if any) by the file. If two or more 7391modules are linked together their garbage collection metadata needs to 7392be merged rather than appended together. 7393 7394The Objective-C garbage collection module flags metadata consists of the 7395following key-value pairs: 7396 7397.. list-table:: 7398 :header-rows: 1 7399 :widths: 30 70 7400 7401 * - Key 7402 - Value 7403 7404 * - ``Objective-C Version`` 7405 - **[Required]** --- The Objective-C ABI version. Valid values are 1 and 2. 7406 7407 * - ``Objective-C Image Info Version`` 7408 - **[Required]** --- The version of the image info section. Currently 7409 always 0. 7410 7411 * - ``Objective-C Image Info Section`` 7412 - **[Required]** --- The section to place the metadata. Valid values are 7413 ``"__OBJC, __image_info, regular"`` for Objective-C ABI version 1, and 7414 ``"__DATA,__objc_imageinfo, regular, no_dead_strip"`` for 7415 Objective-C ABI version 2. 7416 7417 * - ``Objective-C Garbage Collection`` 7418 - **[Required]** --- Specifies whether garbage collection is supported or 7419 not. Valid values are 0, for no garbage collection, and 2, for garbage 7420 collection supported. 7421 7422 * - ``Objective-C GC Only`` 7423 - **[Optional]** --- Specifies that only garbage collection is supported. 7424 If present, its value must be 6. This flag requires that the 7425 ``Objective-C Garbage Collection`` flag have the value 2. 7426 7427Some important flag interactions: 7428 7429- If a module with ``Objective-C Garbage Collection`` set to 0 is 7430 merged with a module with ``Objective-C Garbage Collection`` set to 7431 2, then the resulting module has the 7432 ``Objective-C Garbage Collection`` flag set to 0. 7433- A module with ``Objective-C Garbage Collection`` set to 0 cannot be 7434 merged with a module with ``Objective-C GC Only`` set to 6. 7435 7436C type width Module Flags Metadata 7437---------------------------------- 7438 7439The ARM backend emits a section into each generated object file describing the 7440options that it was compiled with (in a compiler-independent way) to prevent 7441linking incompatible objects, and to allow automatic library selection. Some 7442of these options are not visible at the IR level, namely wchar_t width and enum 7443width. 7444 7445To pass this information to the backend, these options are encoded in module 7446flags metadata, using the following key-value pairs: 7447 7448.. list-table:: 7449 :header-rows: 1 7450 :widths: 30 70 7451 7452 * - Key 7453 - Value 7454 7455 * - short_wchar 7456 - * 0 --- sizeof(wchar_t) == 4 7457 * 1 --- sizeof(wchar_t) == 2 7458 7459 * - short_enum 7460 - * 0 --- Enums are at least as large as an ``int``. 7461 * 1 --- Enums are stored in the smallest integer type which can 7462 represent all of its values. 7463 7464For example, the following metadata section specifies that the module was 7465compiled with a ``wchar_t`` width of 4 bytes, and the underlying type of an 7466enum is the smallest type which can represent all of its values:: 7467 7468 !llvm.module.flags = !{!0, !1} 7469 !0 = !{i32 1, !"short_wchar", i32 1} 7470 !1 = !{i32 1, !"short_enum", i32 0} 7471 7472LTO Post-Link Module Flags Metadata 7473----------------------------------- 7474 7475Some optimisations are only when the entire LTO unit is present in the current 7476module. This is represented by the ``LTOPostLink`` module flags metadata, which 7477will be created with a value of ``1`` when LTO linking occurs. 7478 7479Embedded Objects Names Metadata 7480=============================== 7481 7482Offloading compilations need to embed device code into the host section table to 7483create a fat binary. This metadata node references each global that will be 7484embedded in the module. The primary use for this is to make referencing these 7485globals more efficient in the IR. The metadata references nodes containing 7486pointers to the global to be embedded followed by the section name it will be 7487stored at:: 7488 7489 !llvm.embedded.objects = !{!0} 7490 !0 = !{ptr @object, !".section"} 7491 7492Automatic Linker Flags Named Metadata 7493===================================== 7494 7495Some targets support embedding of flags to the linker inside individual object 7496files. Typically this is used in conjunction with language extensions which 7497allow source files to contain linker command line options, and have these 7498automatically be transmitted to the linker via object files. 7499 7500These flags are encoded in the IR using named metadata with the name 7501``!llvm.linker.options``. Each operand is expected to be a metadata node 7502which should be a list of other metadata nodes, each of which should be a 7503list of metadata strings defining linker options. 7504 7505For example, the following metadata section specifies two separate sets of 7506linker options, presumably to link against ``libz`` and the ``Cocoa`` 7507framework:: 7508 7509 !0 = !{ !"-lz" } 7510 !1 = !{ !"-framework", !"Cocoa" } 7511 !llvm.linker.options = !{ !0, !1 } 7512 7513The metadata encoding as lists of lists of options, as opposed to a collapsed 7514list of options, is chosen so that the IR encoding can use multiple option 7515strings to specify e.g., a single library, while still having that specifier be 7516preserved as an atomic element that can be recognized by a target specific 7517assembly writer or object file emitter. 7518 7519Each individual option is required to be either a valid option for the target's 7520linker, or an option that is reserved by the target specific assembly writer or 7521object file emitter. No other aspect of these options is defined by the IR. 7522 7523Dependent Libs Named Metadata 7524============================= 7525 7526Some targets support embedding of strings into object files to indicate 7527a set of libraries to add to the link. Typically this is used in conjunction 7528with language extensions which allow source files to explicitly declare the 7529libraries they depend on, and have these automatically be transmitted to the 7530linker via object files. 7531 7532The list is encoded in the IR using named metadata with the name 7533``!llvm.dependent-libraries``. Each operand is expected to be a metadata node 7534which should contain a single string operand. 7535 7536For example, the following metadata section contains two library specifiers:: 7537 7538 !0 = !{!"a library specifier"} 7539 !1 = !{!"another library specifier"} 7540 !llvm.dependent-libraries = !{ !0, !1 } 7541 7542Each library specifier will be handled independently by the consuming linker. 7543The effect of the library specifiers are defined by the consuming linker. 7544 7545.. _summary: 7546 7547ThinLTO Summary 7548=============== 7549 7550Compiling with `ThinLTO <https://clang.llvm.org/docs/ThinLTO.html>`_ 7551causes the building of a compact summary of the module that is emitted into 7552the bitcode. The summary is emitted into the LLVM assembly and identified 7553in syntax by a caret ('``^``'). 7554 7555The summary is parsed into a bitcode output, along with the Module 7556IR, via the "``llvm-as``" tool. Tools that parse the Module IR for the purposes 7557of optimization (e.g. "``clang -x ir``" and "``opt``"), will ignore the 7558summary entries (just as they currently ignore summary entries in a bitcode 7559input file). 7560 7561Eventually, the summary will be parsed into a ModuleSummaryIndex object under 7562the same conditions where summary index is currently built from bitcode. 7563Specifically, tools that test the Thin Link portion of a ThinLTO compile 7564(i.e. llvm-lto and llvm-lto2), or when parsing a combined index 7565for a distributed ThinLTO backend via clang's "``-fthinlto-index=<>``" flag 7566(this part is not yet implemented, use llvm-as to create a bitcode object 7567before feeding into thin link tools for now). 7568 7569There are currently 3 types of summary entries in the LLVM assembly: 7570:ref:`module paths<module_path_summary>`, 7571:ref:`global values<gv_summary>`, and 7572:ref:`type identifiers<typeid_summary>`. 7573 7574.. _module_path_summary: 7575 7576Module Path Summary Entry 7577------------------------- 7578 7579Each module path summary entry lists a module containing global values included 7580in the summary. For a single IR module there will be one such entry, but 7581in a combined summary index produced during the thin link, there will be 7582one module path entry per linked module with summary. 7583 7584Example: 7585 7586.. code-block:: text 7587 7588 ^0 = module: (path: "/path/to/file.o", hash: (2468601609, 1329373163, 1565878005, 638838075, 3148790418)) 7589 7590The ``path`` field is a string path to the bitcode file, and the ``hash`` 7591field is the 160-bit SHA-1 hash of the IR bitcode contents, used for 7592incremental builds and caching. 7593 7594.. _gv_summary: 7595 7596Global Value Summary Entry 7597-------------------------- 7598 7599Each global value summary entry corresponds to a global value defined or 7600referenced by a summarized module. 7601 7602Example: 7603 7604.. code-block:: text 7605 7606 ^4 = gv: (name: "f"[, summaries: (Summary)[, (Summary)]*]?) ; guid = 14740650423002898831 7607 7608For declarations, there will not be a summary list. For definitions, a 7609global value will contain a list of summaries, one per module containing 7610a definition. There can be multiple entries in a combined summary index 7611for symbols with weak linkage. 7612 7613Each ``Summary`` format will depend on whether the global value is a 7614:ref:`function<function_summary>`, :ref:`variable<variable_summary>`, or 7615:ref:`alias<alias_summary>`. 7616 7617.. _function_summary: 7618 7619Function Summary 7620^^^^^^^^^^^^^^^^ 7621 7622If the global value is a function, the ``Summary`` entry will look like: 7623 7624.. code-block:: text 7625 7626 function: (module: ^0, flags: (linkage: external, notEligibleToImport: 0, live: 0, dsoLocal: 0), insts: 2[, FuncFlags]?[, Calls]?[, TypeIdInfo]?[, Params]?[, Refs]? 7627 7628The ``module`` field includes the summary entry id for the module containing 7629this definition, and the ``flags`` field contains information such as 7630the linkage type, a flag indicating whether it is legal to import the 7631definition, whether it is globally live and whether the linker resolved it 7632to a local definition (the latter two are populated during the thin link). 7633The ``insts`` field contains the number of IR instructions in the function. 7634Finally, there are several optional fields: :ref:`FuncFlags<funcflags_summary>`, 7635:ref:`Calls<calls_summary>`, :ref:`TypeIdInfo<typeidinfo_summary>`, 7636:ref:`Params<params_summary>`, :ref:`Refs<refs_summary>`. 7637 7638.. _variable_summary: 7639 7640Global Variable Summary 7641^^^^^^^^^^^^^^^^^^^^^^^ 7642 7643If the global value is a variable, the ``Summary`` entry will look like: 7644 7645.. code-block:: text 7646 7647 variable: (module: ^0, flags: (linkage: external, notEligibleToImport: 0, live: 0, dsoLocal: 0)[, Refs]? 7648 7649The variable entry contains a subset of the fields in a 7650:ref:`function summary <function_summary>`, see the descriptions there. 7651 7652.. _alias_summary: 7653 7654Alias Summary 7655^^^^^^^^^^^^^ 7656 7657If the global value is an alias, the ``Summary`` entry will look like: 7658 7659.. code-block:: text 7660 7661 alias: (module: ^0, flags: (linkage: external, notEligibleToImport: 0, live: 0, dsoLocal: 0), aliasee: ^2) 7662 7663The ``module`` and ``flags`` fields are as described for a 7664:ref:`function summary <function_summary>`. The ``aliasee`` field 7665contains a reference to the global value summary entry of the aliasee. 7666 7667.. _funcflags_summary: 7668 7669Function Flags 7670^^^^^^^^^^^^^^ 7671 7672The optional ``FuncFlags`` field looks like: 7673 7674.. code-block:: text 7675 7676 funcFlags: (readNone: 0, readOnly: 0, noRecurse: 0, returnDoesNotAlias: 0, noInline: 0, alwaysInline: 0, noUnwind: 1, mayThrow: 0, hasUnknownCall: 0) 7677 7678If unspecified, flags are assumed to hold the conservative ``false`` value of 7679``0``. 7680 7681.. _calls_summary: 7682 7683Calls 7684^^^^^ 7685 7686The optional ``Calls`` field looks like: 7687 7688.. code-block:: text 7689 7690 calls: ((Callee)[, (Callee)]*) 7691 7692where each ``Callee`` looks like: 7693 7694.. code-block:: text 7695 7696 callee: ^1[, hotness: None]?[, relbf: 0]? 7697 7698The ``callee`` refers to the summary entry id of the callee. At most one 7699of ``hotness`` (which can take the values ``Unknown``, ``Cold``, ``None``, 7700``Hot``, and ``Critical``), and ``relbf`` (which holds the integer 7701branch frequency relative to the entry frequency, scaled down by 2^8) 7702may be specified. The defaults are ``Unknown`` and ``0``, respectively. 7703 7704.. _params_summary: 7705 7706Params 7707^^^^^^ 7708 7709The optional ``Params`` is used by ``StackSafety`` and looks like: 7710 7711.. code-block:: text 7712 7713 Params: ((Param)[, (Param)]*) 7714 7715where each ``Param`` describes pointer parameter access inside of the 7716function and looks like: 7717 7718.. code-block:: text 7719 7720 param: 4, offset: [0, 5][, calls: ((Callee)[, (Callee)]*)]? 7721 7722where the first ``param`` is the number of the parameter it describes, 7723``offset`` is the inclusive range of offsets from the pointer parameter to bytes 7724which can be accessed by the function. This range does not include accesses by 7725function calls from ``calls`` list. 7726 7727where each ``Callee`` describes how parameter is forwarded into other 7728functions and looks like: 7729 7730.. code-block:: text 7731 7732 callee: ^3, param: 5, offset: [-3, 3] 7733 7734The ``callee`` refers to the summary entry id of the callee, ``param`` is 7735the number of the callee parameter which points into the callers parameter 7736with offset known to be inside of the ``offset`` range. ``calls`` will be 7737consumed and removed by thin link stage to update ``Param::offset`` so it 7738covers all accesses possible by ``calls``. 7739 7740Pointer parameter without corresponding ``Param`` is considered unsafe and we 7741assume that access with any offset is possible. 7742 7743Example: 7744 7745If we have the following function: 7746 7747.. code-block:: text 7748 7749 define i64 @foo(i64* %0, i32* %1, i8* %2, i8 %3) { 7750 store i32* %1, i32** @x 7751 %5 = getelementptr inbounds i8, i8* %2, i64 5 7752 %6 = load i8, i8* %5 7753 %7 = getelementptr inbounds i8, i8* %2, i8 %3 7754 tail call void @bar(i8 %3, i8* %7) 7755 %8 = load i64, i64* %0 7756 ret i64 %8 7757 } 7758 7759We can expect the record like this: 7760 7761.. code-block:: text 7762 7763 params: ((param: 0, offset: [0, 7]),(param: 2, offset: [5, 5], calls: ((callee: ^3, param: 1, offset: [-128, 127])))) 7764 7765The function may access just 8 bytes of the parameter %0 . ``calls`` is empty, 7766so the parameter is either not used for function calls or ``offset`` already 7767covers all accesses from nested function calls. 7768Parameter %1 escapes, so access is unknown. 7769The function itself can access just a single byte of the parameter %2. Additional 7770access is possible inside of the ``@bar`` or ``^3``. The function adds signed 7771offset to the pointer and passes the result as the argument %1 into ``^3``. 7772This record itself does not tell us how ``^3`` will access the parameter. 7773Parameter %3 is not a pointer. 7774 7775.. _refs_summary: 7776 7777Refs 7778^^^^ 7779 7780The optional ``Refs`` field looks like: 7781 7782.. code-block:: text 7783 7784 refs: ((Ref)[, (Ref)]*) 7785 7786where each ``Ref`` contains a reference to the summary id of the referenced 7787value (e.g. ``^1``). 7788 7789.. _typeidinfo_summary: 7790 7791TypeIdInfo 7792^^^^^^^^^^ 7793 7794The optional ``TypeIdInfo`` field, used for 7795`Control Flow Integrity <https://clang.llvm.org/docs/ControlFlowIntegrity.html>`_, 7796looks like: 7797 7798.. code-block:: text 7799 7800 typeIdInfo: [(TypeTests)]?[, (TypeTestAssumeVCalls)]?[, (TypeCheckedLoadVCalls)]?[, (TypeTestAssumeConstVCalls)]?[, (TypeCheckedLoadConstVCalls)]? 7801 7802These optional fields have the following forms: 7803 7804TypeTests 7805""""""""" 7806 7807.. code-block:: text 7808 7809 typeTests: (TypeIdRef[, TypeIdRef]*) 7810 7811Where each ``TypeIdRef`` refers to a :ref:`type id<typeid_summary>` 7812by summary id or ``GUID``. 7813 7814TypeTestAssumeVCalls 7815"""""""""""""""""""" 7816 7817.. code-block:: text 7818 7819 typeTestAssumeVCalls: (VFuncId[, VFuncId]*) 7820 7821Where each VFuncId has the format: 7822 7823.. code-block:: text 7824 7825 vFuncId: (TypeIdRef, offset: 16) 7826 7827Where each ``TypeIdRef`` refers to a :ref:`type id<typeid_summary>` 7828by summary id or ``GUID`` preceded by a ``guid:`` tag. 7829 7830TypeCheckedLoadVCalls 7831""""""""""""""""""""" 7832 7833.. code-block:: text 7834 7835 typeCheckedLoadVCalls: (VFuncId[, VFuncId]*) 7836 7837Where each VFuncId has the format described for ``TypeTestAssumeVCalls``. 7838 7839TypeTestAssumeConstVCalls 7840""""""""""""""""""""""""" 7841 7842.. code-block:: text 7843 7844 typeTestAssumeConstVCalls: (ConstVCall[, ConstVCall]*) 7845 7846Where each ConstVCall has the format: 7847 7848.. code-block:: text 7849 7850 (VFuncId, args: (Arg[, Arg]*)) 7851 7852and where each VFuncId has the format described for ``TypeTestAssumeVCalls``, 7853and each Arg is an integer argument number. 7854 7855TypeCheckedLoadConstVCalls 7856"""""""""""""""""""""""""" 7857 7858.. code-block:: text 7859 7860 typeCheckedLoadConstVCalls: (ConstVCall[, ConstVCall]*) 7861 7862Where each ConstVCall has the format described for 7863``TypeTestAssumeConstVCalls``. 7864 7865.. _typeid_summary: 7866 7867Type ID Summary Entry 7868--------------------- 7869 7870Each type id summary entry corresponds to a type identifier resolution 7871which is generated during the LTO link portion of the compile when building 7872with `Control Flow Integrity <https://clang.llvm.org/docs/ControlFlowIntegrity.html>`_, 7873so these are only present in a combined summary index. 7874 7875Example: 7876 7877.. code-block:: text 7878 7879 ^4 = typeid: (name: "_ZTS1A", summary: (typeTestRes: (kind: allOnes, sizeM1BitWidth: 7[, alignLog2: 0]?[, sizeM1: 0]?[, bitMask: 0]?[, inlineBits: 0]?)[, WpdResolutions]?)) ; guid = 7004155349499253778 7880 7881The ``typeTestRes`` gives the type test resolution ``kind`` (which may 7882be ``unsat``, ``byteArray``, ``inline``, ``single``, or ``allOnes``), and 7883the ``size-1`` bit width. It is followed by optional flags, which default to 0, 7884and an optional WpdResolutions (whole program devirtualization resolution) 7885field that looks like: 7886 7887.. code-block:: text 7888 7889 wpdResolutions: ((offset: 0, WpdRes)[, (offset: 1, WpdRes)]* 7890 7891where each entry is a mapping from the given byte offset to the whole-program 7892devirtualization resolution WpdRes, that has one of the following formats: 7893 7894.. code-block:: text 7895 7896 wpdRes: (kind: branchFunnel) 7897 wpdRes: (kind: singleImpl, singleImplName: "_ZN1A1nEi") 7898 wpdRes: (kind: indir) 7899 7900Additionally, each wpdRes has an optional ``resByArg`` field, which 7901describes the resolutions for calls with all constant integer arguments: 7902 7903.. code-block:: text 7904 7905 resByArg: (ResByArg[, ResByArg]*) 7906 7907where ResByArg is: 7908 7909.. code-block:: text 7910 7911 args: (Arg[, Arg]*), byArg: (kind: UniformRetVal[, info: 0][, byte: 0][, bit: 0]) 7912 7913Where the ``kind`` can be ``Indir``, ``UniformRetVal``, ``UniqueRetVal`` 7914or ``VirtualConstProp``. The ``info`` field is only used if the kind 7915is ``UniformRetVal`` (indicates the uniform return value), or 7916``UniqueRetVal`` (holds the return value associated with the unique vtable 7917(0 or 1)). The ``byte`` and ``bit`` fields are only used if the target does 7918not support the use of absolute symbols to store constants. 7919 7920.. _intrinsicglobalvariables: 7921 7922Intrinsic Global Variables 7923========================== 7924 7925LLVM has a number of "magic" global variables that contain data that 7926affect code generation or other IR semantics. These are documented here. 7927All globals of this sort should have a section specified as 7928"``llvm.metadata``". This section and all globals that start with 7929"``llvm.``" are reserved for use by LLVM. 7930 7931.. _gv_llvmused: 7932 7933The '``llvm.used``' Global Variable 7934----------------------------------- 7935 7936The ``@llvm.used`` global is an array which has 7937:ref:`appending linkage <linkage_appending>`. This array contains a list of 7938pointers to named global variables, functions and aliases which may optionally 7939have a pointer cast formed of bitcast or getelementptr. For example, a legal 7940use of it is: 7941 7942.. code-block:: llvm 7943 7944 @X = global i8 4 7945 @Y = global i32 123 7946 7947 @llvm.used = appending global [2 x i8*] [ 7948 i8* @X, 7949 i8* bitcast (i32* @Y to i8*) 7950 ], section "llvm.metadata" 7951 7952If a symbol appears in the ``@llvm.used`` list, then the compiler, assembler, 7953and linker are required to treat the symbol as if there is a reference to the 7954symbol that it cannot see (which is why they have to be named). For example, if 7955a variable has internal linkage and no references other than that from the 7956``@llvm.used`` list, it cannot be deleted. This is commonly used to represent 7957references from inline asms and other things the compiler cannot "see", and 7958corresponds to "``attribute((used))``" in GNU C. 7959 7960On some targets, the code generator must emit a directive to the 7961assembler or object file to prevent the assembler and linker from 7962removing the symbol. 7963 7964.. _gv_llvmcompilerused: 7965 7966The '``llvm.compiler.used``' Global Variable 7967-------------------------------------------- 7968 7969The ``@llvm.compiler.used`` directive is the same as the ``@llvm.used`` 7970directive, except that it only prevents the compiler from touching the 7971symbol. On targets that support it, this allows an intelligent linker to 7972optimize references to the symbol without being impeded as it would be 7973by ``@llvm.used``. 7974 7975This is a rare construct that should only be used in rare circumstances, 7976and should not be exposed to source languages. 7977 7978.. _gv_llvmglobalctors: 7979 7980The '``llvm.global_ctors``' Global Variable 7981------------------------------------------- 7982 7983.. code-block:: llvm 7984 7985 %0 = type { i32, void ()*, i8* } 7986 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor, i8* @data }] 7987 7988The ``@llvm.global_ctors`` array contains a list of constructor 7989functions, priorities, and an associated global or function. 7990The functions referenced by this array will be called in ascending order 7991of priority (i.e. lowest first) when the module is loaded. The order of 7992functions with the same priority is not defined. 7993 7994If the third field is non-null, and points to a global variable 7995or function, the initializer function will only run if the associated 7996data from the current module is not discarded. 7997On ELF the referenced global variable or function must be in a comdat. 7998 7999.. _llvmglobaldtors: 8000 8001The '``llvm.global_dtors``' Global Variable 8002------------------------------------------- 8003 8004.. code-block:: llvm 8005 8006 %0 = type { i32, void ()*, i8* } 8007 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor, i8* @data }] 8008 8009The ``@llvm.global_dtors`` array contains a list of destructor 8010functions, priorities, and an associated global or function. 8011The functions referenced by this array will be called in descending 8012order of priority (i.e. highest first) when the module is unloaded. The 8013order of functions with the same priority is not defined. 8014 8015If the third field is non-null, and points to a global variable 8016or function, the destructor function will only run if the associated 8017data from the current module is not discarded. 8018On ELF the referenced global variable or function must be in a comdat. 8019 8020Instruction Reference 8021===================== 8022 8023The LLVM instruction set consists of several different classifications 8024of instructions: :ref:`terminator instructions <terminators>`, :ref:`binary 8025instructions <binaryops>`, :ref:`bitwise binary 8026instructions <bitwiseops>`, :ref:`memory instructions <memoryops>`, and 8027:ref:`other instructions <otherops>`. 8028 8029.. _terminators: 8030 8031Terminator Instructions 8032----------------------- 8033 8034As mentioned :ref:`previously <functionstructure>`, every basic block in a 8035program ends with a "Terminator" instruction, which indicates which 8036block should be executed after the current block is finished. These 8037terminator instructions typically yield a '``void``' value: they produce 8038control flow, not values (the one exception being the 8039':ref:`invoke <i_invoke>`' instruction). 8040 8041The terminator instructions are: ':ref:`ret <i_ret>`', 8042':ref:`br <i_br>`', ':ref:`switch <i_switch>`', 8043':ref:`indirectbr <i_indirectbr>`', ':ref:`invoke <i_invoke>`', 8044':ref:`callbr <i_callbr>`' 8045':ref:`resume <i_resume>`', ':ref:`catchswitch <i_catchswitch>`', 8046':ref:`catchret <i_catchret>`', 8047':ref:`cleanupret <i_cleanupret>`', 8048and ':ref:`unreachable <i_unreachable>`'. 8049 8050.. _i_ret: 8051 8052'``ret``' Instruction 8053^^^^^^^^^^^^^^^^^^^^^ 8054 8055Syntax: 8056""""""" 8057 8058:: 8059 8060 ret <type> <value> ; Return a value from a non-void function 8061 ret void ; Return from void function 8062 8063Overview: 8064""""""""" 8065 8066The '``ret``' instruction is used to return control flow (and optionally 8067a value) from a function back to the caller. 8068 8069There are two forms of the '``ret``' instruction: one that returns a 8070value and then causes control flow, and one that just causes control 8071flow to occur. 8072 8073Arguments: 8074"""""""""" 8075 8076The '``ret``' instruction optionally accepts a single argument, the 8077return value. The type of the return value must be a ':ref:`first 8078class <t_firstclass>`' type. 8079 8080A function is not :ref:`well formed <wellformed>` if it has a non-void 8081return type and contains a '``ret``' instruction with no return value or 8082a return value with a type that does not match its type, or if it has a 8083void return type and contains a '``ret``' instruction with a return 8084value. 8085 8086Semantics: 8087"""""""""" 8088 8089When the '``ret``' instruction is executed, control flow returns back to 8090the calling function's context. If the caller is a 8091":ref:`call <i_call>`" instruction, execution continues at the 8092instruction after the call. If the caller was an 8093":ref:`invoke <i_invoke>`" instruction, execution continues at the 8094beginning of the "normal" destination block. If the instruction returns 8095a value, that value shall set the call or invoke instruction's return 8096value. 8097 8098Example: 8099"""""""" 8100 8101.. code-block:: llvm 8102 8103 ret i32 5 ; Return an integer value of 5 8104 ret void ; Return from a void function 8105 ret { i32, i8 } { i32 4, i8 2 } ; Return a struct of values 4 and 2 8106 8107.. _i_br: 8108 8109'``br``' Instruction 8110^^^^^^^^^^^^^^^^^^^^ 8111 8112Syntax: 8113""""""" 8114 8115:: 8116 8117 br i1 <cond>, label <iftrue>, label <iffalse> 8118 br label <dest> ; Unconditional branch 8119 8120Overview: 8121""""""""" 8122 8123The '``br``' instruction is used to cause control flow to transfer to a 8124different basic block in the current function. There are two forms of 8125this instruction, corresponding to a conditional branch and an 8126unconditional branch. 8127 8128Arguments: 8129"""""""""" 8130 8131The conditional branch form of the '``br``' instruction takes a single 8132'``i1``' value and two '``label``' values. The unconditional form of the 8133'``br``' instruction takes a single '``label``' value as a target. 8134 8135Semantics: 8136"""""""""" 8137 8138Upon execution of a conditional '``br``' instruction, the '``i1``' 8139argument is evaluated. If the value is ``true``, control flows to the 8140'``iftrue``' ``label`` argument. If "cond" is ``false``, control flows 8141to the '``iffalse``' ``label`` argument. 8142If '``cond``' is ``poison`` or ``undef``, this instruction has undefined 8143behavior. 8144 8145Example: 8146"""""""" 8147 8148.. code-block:: llvm 8149 8150 Test: 8151 %cond = icmp eq i32 %a, %b 8152 br i1 %cond, label %IfEqual, label %IfUnequal 8153 IfEqual: 8154 ret i32 1 8155 IfUnequal: 8156 ret i32 0 8157 8158.. _i_switch: 8159 8160'``switch``' Instruction 8161^^^^^^^^^^^^^^^^^^^^^^^^ 8162 8163Syntax: 8164""""""" 8165 8166:: 8167 8168 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ] 8169 8170Overview: 8171""""""""" 8172 8173The '``switch``' instruction is used to transfer control flow to one of 8174several different places. It is a generalization of the '``br``' 8175instruction, allowing a branch to occur to one of many possible 8176destinations. 8177 8178Arguments: 8179"""""""""" 8180 8181The '``switch``' instruction uses three parameters: an integer 8182comparison value '``value``', a default '``label``' destination, and an 8183array of pairs of comparison value constants and '``label``'s. The table 8184is not allowed to contain duplicate constant entries. 8185 8186Semantics: 8187"""""""""" 8188 8189The ``switch`` instruction specifies a table of values and destinations. 8190When the '``switch``' instruction is executed, this table is searched 8191for the given value. If the value is found, control flow is transferred 8192to the corresponding destination; otherwise, control flow is transferred 8193to the default destination. 8194If '``value``' is ``poison`` or ``undef``, this instruction has undefined 8195behavior. 8196 8197Implementation: 8198""""""""""""""" 8199 8200Depending on properties of the target machine and the particular 8201``switch`` instruction, this instruction may be code generated in 8202different ways. For example, it could be generated as a series of 8203chained conditional branches or with a lookup table. 8204 8205Example: 8206"""""""" 8207 8208.. code-block:: llvm 8209 8210 ; Emulate a conditional br instruction 8211 %Val = zext i1 %value to i32 8212 switch i32 %Val, label %truedest [ i32 0, label %falsedest ] 8213 8214 ; Emulate an unconditional br instruction 8215 switch i32 0, label %dest [ ] 8216 8217 ; Implement a jump table: 8218 switch i32 %val, label %otherwise [ i32 0, label %onzero 8219 i32 1, label %onone 8220 i32 2, label %ontwo ] 8221 8222.. _i_indirectbr: 8223 8224'``indirectbr``' Instruction 8225^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8226 8227Syntax: 8228""""""" 8229 8230:: 8231 8232 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ] 8233 8234Overview: 8235""""""""" 8236 8237The '``indirectbr``' instruction implements an indirect branch to a 8238label within the current function, whose address is specified by 8239"``address``". Address must be derived from a 8240:ref:`blockaddress <blockaddress>` constant. 8241 8242Arguments: 8243"""""""""" 8244 8245The '``address``' argument is the address of the label to jump to. The 8246rest of the arguments indicate the full set of possible destinations 8247that the address may point to. Blocks are allowed to occur multiple 8248times in the destination list, though this isn't particularly useful. 8249 8250This destination list is required so that dataflow analysis has an 8251accurate understanding of the CFG. 8252 8253Semantics: 8254"""""""""" 8255 8256Control transfers to the block specified in the address argument. All 8257possible destination blocks must be listed in the label list, otherwise 8258this instruction has undefined behavior. This implies that jumps to 8259labels defined in other functions have undefined behavior as well. 8260If '``address``' is ``poison`` or ``undef``, this instruction has undefined 8261behavior. 8262 8263Implementation: 8264""""""""""""""" 8265 8266This is typically implemented with a jump through a register. 8267 8268Example: 8269"""""""" 8270 8271.. code-block:: llvm 8272 8273 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ] 8274 8275.. _i_invoke: 8276 8277'``invoke``' Instruction 8278^^^^^^^^^^^^^^^^^^^^^^^^ 8279 8280Syntax: 8281""""""" 8282 8283:: 8284 8285 <result> = invoke [cconv] [ret attrs] [addrspace(<num>)] <ty>|<fnty> <fnptrval>(<function args>) [fn attrs] 8286 [operand bundles] to label <normal label> unwind label <exception label> 8287 8288Overview: 8289""""""""" 8290 8291The '``invoke``' instruction causes control to transfer to a specified 8292function, with the possibility of control flow transfer to either the 8293'``normal``' label or the '``exception``' label. If the callee function 8294returns with the "``ret``" instruction, control flow will return to the 8295"normal" label. If the callee (or any indirect callees) returns via the 8296":ref:`resume <i_resume>`" instruction or other exception handling 8297mechanism, control is interrupted and continued at the dynamically 8298nearest "exception" label. 8299 8300The '``exception``' label is a `landing 8301pad <ExceptionHandling.html#overview>`_ for the exception. As such, 8302'``exception``' label is required to have the 8303":ref:`landingpad <i_landingpad>`" instruction, which contains the 8304information about the behavior of the program after unwinding happens, 8305as its first non-PHI instruction. The restrictions on the 8306"``landingpad``" instruction's tightly couples it to the "``invoke``" 8307instruction, so that the important information contained within the 8308"``landingpad``" instruction can't be lost through normal code motion. 8309 8310Arguments: 8311"""""""""" 8312 8313This instruction requires several arguments: 8314 8315#. The optional "cconv" marker indicates which :ref:`calling 8316 convention <callingconv>` the call should use. If none is 8317 specified, the call defaults to using C calling conventions. 8318#. The optional :ref:`Parameter Attributes <paramattrs>` list for return 8319 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes 8320 are valid here. 8321#. The optional addrspace attribute can be used to indicate the address space 8322 of the called function. If it is not specified, the program address space 8323 from the :ref:`datalayout string<langref_datalayout>` will be used. 8324#. '``ty``': the type of the call instruction itself which is also the 8325 type of the return value. Functions that return no value are marked 8326 ``void``. 8327#. '``fnty``': shall be the signature of the function being invoked. The 8328 argument types must match the types implied by this signature. This 8329 type can be omitted if the function is not varargs. 8330#. '``fnptrval``': An LLVM value containing a pointer to a function to 8331 be invoked. In most cases, this is a direct function invocation, but 8332 indirect ``invoke``'s are just as possible, calling an arbitrary pointer 8333 to function value. 8334#. '``function args``': argument list whose types match the function 8335 signature argument types and parameter attributes. All arguments must 8336 be of :ref:`first class <t_firstclass>` type. If the function signature 8337 indicates the function accepts a variable number of arguments, the 8338 extra arguments can be specified. 8339#. '``normal label``': the label reached when the called function 8340 executes a '``ret``' instruction. 8341#. '``exception label``': the label reached when a callee returns via 8342 the :ref:`resume <i_resume>` instruction or other exception handling 8343 mechanism. 8344#. The optional :ref:`function attributes <fnattrs>` list. 8345#. The optional :ref:`operand bundles <opbundles>` list. 8346 8347Semantics: 8348"""""""""" 8349 8350This instruction is designed to operate as a standard '``call``' 8351instruction in most regards. The primary difference is that it 8352establishes an association with a label, which is used by the runtime 8353library to unwind the stack. 8354 8355This instruction is used in languages with destructors to ensure that 8356proper cleanup is performed in the case of either a ``longjmp`` or a 8357thrown exception. Additionally, this is important for implementation of 8358'``catch``' clauses in high-level languages that support them. 8359 8360For the purposes of the SSA form, the definition of the value returned 8361by the '``invoke``' instruction is deemed to occur on the edge from the 8362current block to the "normal" label. If the callee unwinds then no 8363return value is available. 8364 8365Example: 8366"""""""" 8367 8368.. code-block:: llvm 8369 8370 %retval = invoke i32 @Test(i32 15) to label %Continue 8371 unwind label %TestCleanup ; i32:retval set 8372 %retval = invoke coldcc i32 %Testfnptr(i32 15) to label %Continue 8373 unwind label %TestCleanup ; i32:retval set 8374 8375.. _i_callbr: 8376 8377'``callbr``' Instruction 8378^^^^^^^^^^^^^^^^^^^^^^^^ 8379 8380Syntax: 8381""""""" 8382 8383:: 8384 8385 <result> = callbr [cconv] [ret attrs] [addrspace(<num>)] <ty>|<fnty> <fnptrval>(<function args>) [fn attrs] 8386 [operand bundles] to label <fallthrough label> [indirect labels] 8387 8388Overview: 8389""""""""" 8390 8391The '``callbr``' instruction causes control to transfer to a specified 8392function, with the possibility of control flow transfer to either the 8393'``fallthrough``' label or one of the '``indirect``' labels. 8394 8395This instruction should only be used to implement the "goto" feature of gcc 8396style inline assembly. Any other usage is an error in the IR verifier. 8397 8398Arguments: 8399"""""""""" 8400 8401This instruction requires several arguments: 8402 8403#. The optional "cconv" marker indicates which :ref:`calling 8404 convention <callingconv>` the call should use. If none is 8405 specified, the call defaults to using C calling conventions. 8406#. The optional :ref:`Parameter Attributes <paramattrs>` list for return 8407 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes 8408 are valid here. 8409#. The optional addrspace attribute can be used to indicate the address space 8410 of the called function. If it is not specified, the program address space 8411 from the :ref:`datalayout string<langref_datalayout>` will be used. 8412#. '``ty``': the type of the call instruction itself which is also the 8413 type of the return value. Functions that return no value are marked 8414 ``void``. 8415#. '``fnty``': shall be the signature of the function being called. The 8416 argument types must match the types implied by this signature. This 8417 type can be omitted if the function is not varargs. 8418#. '``fnptrval``': An LLVM value containing a pointer to a function to 8419 be called. In most cases, this is a direct function call, but 8420 other ``callbr``'s are just as possible, calling an arbitrary pointer 8421 to function value. 8422#. '``function args``': argument list whose types match the function 8423 signature argument types and parameter attributes. All arguments must 8424 be of :ref:`first class <t_firstclass>` type. If the function signature 8425 indicates the function accepts a variable number of arguments, the 8426 extra arguments can be specified. 8427#. '``fallthrough label``': the label reached when the inline assembly's 8428 execution exits the bottom. 8429#. '``indirect labels``': the labels reached when a callee transfers control 8430 to a location other than the '``fallthrough label``'. Label constraints 8431 refer to these destinations. 8432#. The optional :ref:`function attributes <fnattrs>` list. 8433#. The optional :ref:`operand bundles <opbundles>` list. 8434 8435Semantics: 8436"""""""""" 8437 8438This instruction is designed to operate as a standard '``call``' 8439instruction in most regards. The primary difference is that it 8440establishes an association with additional labels to define where control 8441flow goes after the call. 8442 8443The output values of a '``callbr``' instruction are available only to 8444the '``fallthrough``' block, not to any '``indirect``' blocks(s). 8445 8446The only use of this today is to implement the "goto" feature of gcc inline 8447assembly where additional labels can be provided as locations for the inline 8448assembly to jump to. 8449 8450Example: 8451"""""""" 8452 8453.. code-block:: llvm 8454 8455 ; "asm goto" without output constraints. 8456 callbr void asm "", "r,!i"(i32 %x) 8457 to label %fallthrough [label %indirect] 8458 8459 ; "asm goto" with output constraints. 8460 <result> = callbr i32 asm "", "=r,r,!i"(i32 %x) 8461 to label %fallthrough [label %indirect] 8462 8463.. _i_resume: 8464 8465'``resume``' Instruction 8466^^^^^^^^^^^^^^^^^^^^^^^^ 8467 8468Syntax: 8469""""""" 8470 8471:: 8472 8473 resume <type> <value> 8474 8475Overview: 8476""""""""" 8477 8478The '``resume``' instruction is a terminator instruction that has no 8479successors. 8480 8481Arguments: 8482"""""""""" 8483 8484The '``resume``' instruction requires one argument, which must have the 8485same type as the result of any '``landingpad``' instruction in the same 8486function. 8487 8488Semantics: 8489"""""""""" 8490 8491The '``resume``' instruction resumes propagation of an existing 8492(in-flight) exception whose unwinding was interrupted with a 8493:ref:`landingpad <i_landingpad>` instruction. 8494 8495Example: 8496"""""""" 8497 8498.. code-block:: llvm 8499 8500 resume { i8*, i32 } %exn 8501 8502.. _i_catchswitch: 8503 8504'``catchswitch``' Instruction 8505^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8506 8507Syntax: 8508""""""" 8509 8510:: 8511 8512 <resultval> = catchswitch within <parent> [ label <handler1>, label <handler2>, ... ] unwind to caller 8513 <resultval> = catchswitch within <parent> [ label <handler1>, label <handler2>, ... ] unwind label <default> 8514 8515Overview: 8516""""""""" 8517 8518The '``catchswitch``' instruction is used by `LLVM's exception handling system 8519<ExceptionHandling.html#overview>`_ to describe the set of possible catch handlers 8520that may be executed by the :ref:`EH personality routine <personalityfn>`. 8521 8522Arguments: 8523"""""""""" 8524 8525The ``parent`` argument is the token of the funclet that contains the 8526``catchswitch`` instruction. If the ``catchswitch`` is not inside a funclet, 8527this operand may be the token ``none``. 8528 8529The ``default`` argument is the label of another basic block beginning with 8530either a ``cleanuppad`` or ``catchswitch`` instruction. This unwind destination 8531must be a legal target with respect to the ``parent`` links, as described in 8532the `exception handling documentation\ <ExceptionHandling.html#wineh-constraints>`_. 8533 8534The ``handlers`` are a nonempty list of successor blocks that each begin with a 8535:ref:`catchpad <i_catchpad>` instruction. 8536 8537Semantics: 8538"""""""""" 8539 8540Executing this instruction transfers control to one of the successors in 8541``handlers``, if appropriate, or continues to unwind via the unwind label if 8542present. 8543 8544The ``catchswitch`` is both a terminator and a "pad" instruction, meaning that 8545it must be both the first non-phi instruction and last instruction in the basic 8546block. Therefore, it must be the only non-phi instruction in the block. 8547 8548Example: 8549"""""""" 8550 8551.. code-block:: text 8552 8553 dispatch1: 8554 %cs1 = catchswitch within none [label %handler0, label %handler1] unwind to caller 8555 dispatch2: 8556 %cs2 = catchswitch within %parenthandler [label %handler0] unwind label %cleanup 8557 8558.. _i_catchret: 8559 8560'``catchret``' Instruction 8561^^^^^^^^^^^^^^^^^^^^^^^^^^ 8562 8563Syntax: 8564""""""" 8565 8566:: 8567 8568 catchret from <token> to label <normal> 8569 8570Overview: 8571""""""""" 8572 8573The '``catchret``' instruction is a terminator instruction that has a 8574single successor. 8575 8576 8577Arguments: 8578"""""""""" 8579 8580The first argument to a '``catchret``' indicates which ``catchpad`` it 8581exits. It must be a :ref:`catchpad <i_catchpad>`. 8582The second argument to a '``catchret``' specifies where control will 8583transfer to next. 8584 8585Semantics: 8586"""""""""" 8587 8588The '``catchret``' instruction ends an existing (in-flight) exception whose 8589unwinding was interrupted with a :ref:`catchpad <i_catchpad>` instruction. The 8590:ref:`personality function <personalityfn>` gets a chance to execute arbitrary 8591code to, for example, destroy the active exception. Control then transfers to 8592``normal``. 8593 8594The ``token`` argument must be a token produced by a ``catchpad`` instruction. 8595If the specified ``catchpad`` is not the most-recently-entered not-yet-exited 8596funclet pad (as described in the `EH documentation\ <ExceptionHandling.html#wineh-constraints>`_), 8597the ``catchret``'s behavior is undefined. 8598 8599Example: 8600"""""""" 8601 8602.. code-block:: text 8603 8604 catchret from %catch to label %continue 8605 8606.. _i_cleanupret: 8607 8608'``cleanupret``' Instruction 8609^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8610 8611Syntax: 8612""""""" 8613 8614:: 8615 8616 cleanupret from <value> unwind label <continue> 8617 cleanupret from <value> unwind to caller 8618 8619Overview: 8620""""""""" 8621 8622The '``cleanupret``' instruction is a terminator instruction that has 8623an optional successor. 8624 8625 8626Arguments: 8627"""""""""" 8628 8629The '``cleanupret``' instruction requires one argument, which indicates 8630which ``cleanuppad`` it exits, and must be a :ref:`cleanuppad <i_cleanuppad>`. 8631If the specified ``cleanuppad`` is not the most-recently-entered not-yet-exited 8632funclet pad (as described in the `EH documentation\ <ExceptionHandling.html#wineh-constraints>`_), 8633the ``cleanupret``'s behavior is undefined. 8634 8635The '``cleanupret``' instruction also has an optional successor, ``continue``, 8636which must be the label of another basic block beginning with either a 8637``cleanuppad`` or ``catchswitch`` instruction. This unwind destination must 8638be a legal target with respect to the ``parent`` links, as described in the 8639`exception handling documentation\ <ExceptionHandling.html#wineh-constraints>`_. 8640 8641Semantics: 8642"""""""""" 8643 8644The '``cleanupret``' instruction indicates to the 8645:ref:`personality function <personalityfn>` that one 8646:ref:`cleanuppad <i_cleanuppad>` it transferred control to has ended. 8647It transfers control to ``continue`` or unwinds out of the function. 8648 8649Example: 8650"""""""" 8651 8652.. code-block:: text 8653 8654 cleanupret from %cleanup unwind to caller 8655 cleanupret from %cleanup unwind label %continue 8656 8657.. _i_unreachable: 8658 8659'``unreachable``' Instruction 8660^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 8661 8662Syntax: 8663""""""" 8664 8665:: 8666 8667 unreachable 8668 8669Overview: 8670""""""""" 8671 8672The '``unreachable``' instruction has no defined semantics. This 8673instruction is used to inform the optimizer that a particular portion of 8674the code is not reachable. This can be used to indicate that the code 8675after a no-return function cannot be reached, and other facts. 8676 8677Semantics: 8678"""""""""" 8679 8680The '``unreachable``' instruction has no defined semantics. 8681 8682.. _unaryops: 8683 8684Unary Operations 8685----------------- 8686 8687Unary operators require a single operand, execute an operation on 8688it, and produce a single value. The operand might represent multiple 8689data, as is the case with the :ref:`vector <t_vector>` data type. The 8690result value has the same type as its operand. 8691 8692.. _i_fneg: 8693 8694'``fneg``' Instruction 8695^^^^^^^^^^^^^^^^^^^^^^ 8696 8697Syntax: 8698""""""" 8699 8700:: 8701 8702 <result> = fneg [fast-math flags]* <ty> <op1> ; yields ty:result 8703 8704Overview: 8705""""""""" 8706 8707The '``fneg``' instruction returns the negation of its operand. 8708 8709Arguments: 8710"""""""""" 8711 8712The argument to the '``fneg``' instruction must be a 8713:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 8714floating-point values. 8715 8716Semantics: 8717"""""""""" 8718 8719The value produced is a copy of the operand with its sign bit flipped. 8720This instruction can also take any number of :ref:`fast-math 8721flags <fastmath>`, which are optimization hints to enable otherwise 8722unsafe floating-point optimizations: 8723 8724Example: 8725"""""""" 8726 8727.. code-block:: text 8728 8729 <result> = fneg float %val ; yields float:result = -%var 8730 8731.. _binaryops: 8732 8733Binary Operations 8734----------------- 8735 8736Binary operators are used to do most of the computation in a program. 8737They require two operands of the same type, execute an operation on 8738them, and produce a single value. The operands might represent multiple 8739data, as is the case with the :ref:`vector <t_vector>` data type. The 8740result value has the same type as its operands. 8741 8742There are several different binary operators: 8743 8744.. _i_add: 8745 8746'``add``' Instruction 8747^^^^^^^^^^^^^^^^^^^^^ 8748 8749Syntax: 8750""""""" 8751 8752:: 8753 8754 <result> = add <ty> <op1>, <op2> ; yields ty:result 8755 <result> = add nuw <ty> <op1>, <op2> ; yields ty:result 8756 <result> = add nsw <ty> <op1>, <op2> ; yields ty:result 8757 <result> = add nuw nsw <ty> <op1>, <op2> ; yields ty:result 8758 8759Overview: 8760""""""""" 8761 8762The '``add``' instruction returns the sum of its two operands. 8763 8764Arguments: 8765"""""""""" 8766 8767The two arguments to the '``add``' instruction must be 8768:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 8769arguments must have identical types. 8770 8771Semantics: 8772"""""""""" 8773 8774The value produced is the integer sum of the two operands. 8775 8776If the sum has unsigned overflow, the result returned is the 8777mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of 8778the result. 8779 8780Because LLVM integers use a two's complement representation, this 8781instruction is appropriate for both signed and unsigned integers. 8782 8783``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 8784respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 8785result value of the ``add`` is a :ref:`poison value <poisonvalues>` if 8786unsigned and/or signed overflow, respectively, occurs. 8787 8788Example: 8789"""""""" 8790 8791.. code-block:: text 8792 8793 <result> = add i32 4, %var ; yields i32:result = 4 + %var 8794 8795.. _i_fadd: 8796 8797'``fadd``' Instruction 8798^^^^^^^^^^^^^^^^^^^^^^ 8799 8800Syntax: 8801""""""" 8802 8803:: 8804 8805 <result> = fadd [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 8806 8807Overview: 8808""""""""" 8809 8810The '``fadd``' instruction returns the sum of its two operands. 8811 8812Arguments: 8813"""""""""" 8814 8815The two arguments to the '``fadd``' instruction must be 8816:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 8817floating-point values. Both arguments must have identical types. 8818 8819Semantics: 8820"""""""""" 8821 8822The value produced is the floating-point sum of the two operands. 8823This instruction is assumed to execute in the default :ref:`floating-point 8824environment <floatenv>`. 8825This instruction can also take any number of :ref:`fast-math 8826flags <fastmath>`, which are optimization hints to enable otherwise 8827unsafe floating-point optimizations: 8828 8829Example: 8830"""""""" 8831 8832.. code-block:: text 8833 8834 <result> = fadd float 4.0, %var ; yields float:result = 4.0 + %var 8835 8836.. _i_sub: 8837 8838'``sub``' Instruction 8839^^^^^^^^^^^^^^^^^^^^^ 8840 8841Syntax: 8842""""""" 8843 8844:: 8845 8846 <result> = sub <ty> <op1>, <op2> ; yields ty:result 8847 <result> = sub nuw <ty> <op1>, <op2> ; yields ty:result 8848 <result> = sub nsw <ty> <op1>, <op2> ; yields ty:result 8849 <result> = sub nuw nsw <ty> <op1>, <op2> ; yields ty:result 8850 8851Overview: 8852""""""""" 8853 8854The '``sub``' instruction returns the difference of its two operands. 8855 8856Note that the '``sub``' instruction is used to represent the '``neg``' 8857instruction present in most other intermediate representations. 8858 8859Arguments: 8860"""""""""" 8861 8862The two arguments to the '``sub``' instruction must be 8863:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 8864arguments must have identical types. 8865 8866Semantics: 8867"""""""""" 8868 8869The value produced is the integer difference of the two operands. 8870 8871If the difference has unsigned overflow, the result returned is the 8872mathematical result modulo 2\ :sup:`n`\ , where n is the bit width of 8873the result. 8874 8875Because LLVM integers use a two's complement representation, this 8876instruction is appropriate for both signed and unsigned integers. 8877 8878``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 8879respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 8880result value of the ``sub`` is a :ref:`poison value <poisonvalues>` if 8881unsigned and/or signed overflow, respectively, occurs. 8882 8883Example: 8884"""""""" 8885 8886.. code-block:: text 8887 8888 <result> = sub i32 4, %var ; yields i32:result = 4 - %var 8889 <result> = sub i32 0, %val ; yields i32:result = -%var 8890 8891.. _i_fsub: 8892 8893'``fsub``' Instruction 8894^^^^^^^^^^^^^^^^^^^^^^ 8895 8896Syntax: 8897""""""" 8898 8899:: 8900 8901 <result> = fsub [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 8902 8903Overview: 8904""""""""" 8905 8906The '``fsub``' instruction returns the difference of its two operands. 8907 8908Arguments: 8909"""""""""" 8910 8911The two arguments to the '``fsub``' instruction must be 8912:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 8913floating-point values. Both arguments must have identical types. 8914 8915Semantics: 8916"""""""""" 8917 8918The value produced is the floating-point difference of the two operands. 8919This instruction is assumed to execute in the default :ref:`floating-point 8920environment <floatenv>`. 8921This instruction can also take any number of :ref:`fast-math 8922flags <fastmath>`, which are optimization hints to enable otherwise 8923unsafe floating-point optimizations: 8924 8925Example: 8926"""""""" 8927 8928.. code-block:: text 8929 8930 <result> = fsub float 4.0, %var ; yields float:result = 4.0 - %var 8931 <result> = fsub float -0.0, %val ; yields float:result = -%var 8932 8933.. _i_mul: 8934 8935'``mul``' Instruction 8936^^^^^^^^^^^^^^^^^^^^^ 8937 8938Syntax: 8939""""""" 8940 8941:: 8942 8943 <result> = mul <ty> <op1>, <op2> ; yields ty:result 8944 <result> = mul nuw <ty> <op1>, <op2> ; yields ty:result 8945 <result> = mul nsw <ty> <op1>, <op2> ; yields ty:result 8946 <result> = mul nuw nsw <ty> <op1>, <op2> ; yields ty:result 8947 8948Overview: 8949""""""""" 8950 8951The '``mul``' instruction returns the product of its two operands. 8952 8953Arguments: 8954"""""""""" 8955 8956The two arguments to the '``mul``' instruction must be 8957:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 8958arguments must have identical types. 8959 8960Semantics: 8961"""""""""" 8962 8963The value produced is the integer product of the two operands. 8964 8965If the result of the multiplication has unsigned overflow, the result 8966returned is the mathematical result modulo 2\ :sup:`n`\ , where n is the 8967bit width of the result. 8968 8969Because LLVM integers use a two's complement representation, and the 8970result is the same width as the operands, this instruction returns the 8971correct result for both signed and unsigned integers. If a full product 8972(e.g. ``i32`` * ``i32`` -> ``i64``) is needed, the operands should be 8973sign-extended or zero-extended as appropriate to the width of the full 8974product. 8975 8976``nuw`` and ``nsw`` stand for "No Unsigned Wrap" and "No Signed Wrap", 8977respectively. If the ``nuw`` and/or ``nsw`` keywords are present, the 8978result value of the ``mul`` is a :ref:`poison value <poisonvalues>` if 8979unsigned and/or signed overflow, respectively, occurs. 8980 8981Example: 8982"""""""" 8983 8984.. code-block:: text 8985 8986 <result> = mul i32 4, %var ; yields i32:result = 4 * %var 8987 8988.. _i_fmul: 8989 8990'``fmul``' Instruction 8991^^^^^^^^^^^^^^^^^^^^^^ 8992 8993Syntax: 8994""""""" 8995 8996:: 8997 8998 <result> = fmul [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 8999 9000Overview: 9001""""""""" 9002 9003The '``fmul``' instruction returns the product of its two operands. 9004 9005Arguments: 9006"""""""""" 9007 9008The two arguments to the '``fmul``' instruction must be 9009:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 9010floating-point values. Both arguments must have identical types. 9011 9012Semantics: 9013"""""""""" 9014 9015The value produced is the floating-point product of the two operands. 9016This instruction is assumed to execute in the default :ref:`floating-point 9017environment <floatenv>`. 9018This instruction can also take any number of :ref:`fast-math 9019flags <fastmath>`, which are optimization hints to enable otherwise 9020unsafe floating-point optimizations: 9021 9022Example: 9023"""""""" 9024 9025.. code-block:: text 9026 9027 <result> = fmul float 4.0, %var ; yields float:result = 4.0 * %var 9028 9029.. _i_udiv: 9030 9031'``udiv``' Instruction 9032^^^^^^^^^^^^^^^^^^^^^^ 9033 9034Syntax: 9035""""""" 9036 9037:: 9038 9039 <result> = udiv <ty> <op1>, <op2> ; yields ty:result 9040 <result> = udiv exact <ty> <op1>, <op2> ; yields ty:result 9041 9042Overview: 9043""""""""" 9044 9045The '``udiv``' instruction returns the quotient of its two operands. 9046 9047Arguments: 9048"""""""""" 9049 9050The two arguments to the '``udiv``' instruction must be 9051:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9052arguments must have identical types. 9053 9054Semantics: 9055"""""""""" 9056 9057The value produced is the unsigned integer quotient of the two operands. 9058 9059Note that unsigned integer division and signed integer division are 9060distinct operations; for signed integer division, use '``sdiv``'. 9061 9062Division by zero is undefined behavior. For vectors, if any element 9063of the divisor is zero, the operation has undefined behavior. 9064 9065 9066If the ``exact`` keyword is present, the result value of the ``udiv`` is 9067a :ref:`poison value <poisonvalues>` if %op1 is not a multiple of %op2 (as 9068such, "((a udiv exact b) mul b) == a"). 9069 9070Example: 9071"""""""" 9072 9073.. code-block:: text 9074 9075 <result> = udiv i32 4, %var ; yields i32:result = 4 / %var 9076 9077.. _i_sdiv: 9078 9079'``sdiv``' Instruction 9080^^^^^^^^^^^^^^^^^^^^^^ 9081 9082Syntax: 9083""""""" 9084 9085:: 9086 9087 <result> = sdiv <ty> <op1>, <op2> ; yields ty:result 9088 <result> = sdiv exact <ty> <op1>, <op2> ; yields ty:result 9089 9090Overview: 9091""""""""" 9092 9093The '``sdiv``' instruction returns the quotient of its two operands. 9094 9095Arguments: 9096"""""""""" 9097 9098The two arguments to the '``sdiv``' instruction must be 9099:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9100arguments must have identical types. 9101 9102Semantics: 9103"""""""""" 9104 9105The value produced is the signed integer quotient of the two operands 9106rounded towards zero. 9107 9108Note that signed integer division and unsigned integer division are 9109distinct operations; for unsigned integer division, use '``udiv``'. 9110 9111Division by zero is undefined behavior. For vectors, if any element 9112of the divisor is zero, the operation has undefined behavior. 9113Overflow also leads to undefined behavior; this is a rare case, but can 9114occur, for example, by doing a 32-bit division of -2147483648 by -1. 9115 9116If the ``exact`` keyword is present, the result value of the ``sdiv`` is 9117a :ref:`poison value <poisonvalues>` if the result would be rounded. 9118 9119Example: 9120"""""""" 9121 9122.. code-block:: text 9123 9124 <result> = sdiv i32 4, %var ; yields i32:result = 4 / %var 9125 9126.. _i_fdiv: 9127 9128'``fdiv``' Instruction 9129^^^^^^^^^^^^^^^^^^^^^^ 9130 9131Syntax: 9132""""""" 9133 9134:: 9135 9136 <result> = fdiv [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 9137 9138Overview: 9139""""""""" 9140 9141The '``fdiv``' instruction returns the quotient of its two operands. 9142 9143Arguments: 9144"""""""""" 9145 9146The two arguments to the '``fdiv``' instruction must be 9147:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 9148floating-point values. Both arguments must have identical types. 9149 9150Semantics: 9151"""""""""" 9152 9153The value produced is the floating-point quotient of the two operands. 9154This instruction is assumed to execute in the default :ref:`floating-point 9155environment <floatenv>`. 9156This instruction can also take any number of :ref:`fast-math 9157flags <fastmath>`, which are optimization hints to enable otherwise 9158unsafe floating-point optimizations: 9159 9160Example: 9161"""""""" 9162 9163.. code-block:: text 9164 9165 <result> = fdiv float 4.0, %var ; yields float:result = 4.0 / %var 9166 9167.. _i_urem: 9168 9169'``urem``' Instruction 9170^^^^^^^^^^^^^^^^^^^^^^ 9171 9172Syntax: 9173""""""" 9174 9175:: 9176 9177 <result> = urem <ty> <op1>, <op2> ; yields ty:result 9178 9179Overview: 9180""""""""" 9181 9182The '``urem``' instruction returns the remainder from the unsigned 9183division of its two arguments. 9184 9185Arguments: 9186"""""""""" 9187 9188The two arguments to the '``urem``' instruction must be 9189:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9190arguments must have identical types. 9191 9192Semantics: 9193"""""""""" 9194 9195This instruction returns the unsigned integer *remainder* of a division. 9196This instruction always performs an unsigned division to get the 9197remainder. 9198 9199Note that unsigned integer remainder and signed integer remainder are 9200distinct operations; for signed integer remainder, use '``srem``'. 9201 9202Taking the remainder of a division by zero is undefined behavior. 9203For vectors, if any element of the divisor is zero, the operation has 9204undefined behavior. 9205 9206Example: 9207"""""""" 9208 9209.. code-block:: text 9210 9211 <result> = urem i32 4, %var ; yields i32:result = 4 % %var 9212 9213.. _i_srem: 9214 9215'``srem``' Instruction 9216^^^^^^^^^^^^^^^^^^^^^^ 9217 9218Syntax: 9219""""""" 9220 9221:: 9222 9223 <result> = srem <ty> <op1>, <op2> ; yields ty:result 9224 9225Overview: 9226""""""""" 9227 9228The '``srem``' instruction returns the remainder from the signed 9229division of its two operands. This instruction can also take 9230:ref:`vector <t_vector>` versions of the values in which case the elements 9231must be integers. 9232 9233Arguments: 9234"""""""""" 9235 9236The two arguments to the '``srem``' instruction must be 9237:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9238arguments must have identical types. 9239 9240Semantics: 9241"""""""""" 9242 9243This instruction returns the *remainder* of a division (where the result 9244is either zero or has the same sign as the dividend, ``op1``), not the 9245*modulo* operator (where the result is either zero or has the same sign 9246as the divisor, ``op2``) of a value. For more information about the 9247difference, see `The Math 9248Forum <http://mathforum.org/dr.math/problems/anne.4.28.99.html>`_. For a 9249table of how this is implemented in various languages, please see 9250`Wikipedia: modulo 9251operation <http://en.wikipedia.org/wiki/Modulo_operation>`_. 9252 9253Note that signed integer remainder and unsigned integer remainder are 9254distinct operations; for unsigned integer remainder, use '``urem``'. 9255 9256Taking the remainder of a division by zero is undefined behavior. 9257For vectors, if any element of the divisor is zero, the operation has 9258undefined behavior. 9259Overflow also leads to undefined behavior; this is a rare case, but can 9260occur, for example, by taking the remainder of a 32-bit division of 9261-2147483648 by -1. (The remainder doesn't actually overflow, but this 9262rule lets srem be implemented using instructions that return both the 9263result of the division and the remainder.) 9264 9265Example: 9266"""""""" 9267 9268.. code-block:: text 9269 9270 <result> = srem i32 4, %var ; yields i32:result = 4 % %var 9271 9272.. _i_frem: 9273 9274'``frem``' Instruction 9275^^^^^^^^^^^^^^^^^^^^^^ 9276 9277Syntax: 9278""""""" 9279 9280:: 9281 9282 <result> = frem [fast-math flags]* <ty> <op1>, <op2> ; yields ty:result 9283 9284Overview: 9285""""""""" 9286 9287The '``frem``' instruction returns the remainder from the division of 9288its two operands. 9289 9290Arguments: 9291"""""""""" 9292 9293The two arguments to the '``frem``' instruction must be 9294:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` of 9295floating-point values. Both arguments must have identical types. 9296 9297Semantics: 9298"""""""""" 9299 9300The value produced is the floating-point remainder of the two operands. 9301This is the same output as a libm '``fmod``' function, but without any 9302possibility of setting ``errno``. The remainder has the same sign as the 9303dividend. 9304This instruction is assumed to execute in the default :ref:`floating-point 9305environment <floatenv>`. 9306This instruction can also take any number of :ref:`fast-math 9307flags <fastmath>`, which are optimization hints to enable otherwise 9308unsafe floating-point optimizations: 9309 9310Example: 9311"""""""" 9312 9313.. code-block:: text 9314 9315 <result> = frem float 4.0, %var ; yields float:result = 4.0 % %var 9316 9317.. _bitwiseops: 9318 9319Bitwise Binary Operations 9320------------------------- 9321 9322Bitwise binary operators are used to do various forms of bit-twiddling 9323in a program. They are generally very efficient instructions and can 9324commonly be strength reduced from other instructions. They require two 9325operands of the same type, execute an operation on them, and produce a 9326single value. The resulting value is the same type as its operands. 9327 9328.. _i_shl: 9329 9330'``shl``' Instruction 9331^^^^^^^^^^^^^^^^^^^^^ 9332 9333Syntax: 9334""""""" 9335 9336:: 9337 9338 <result> = shl <ty> <op1>, <op2> ; yields ty:result 9339 <result> = shl nuw <ty> <op1>, <op2> ; yields ty:result 9340 <result> = shl nsw <ty> <op1>, <op2> ; yields ty:result 9341 <result> = shl nuw nsw <ty> <op1>, <op2> ; yields ty:result 9342 9343Overview: 9344""""""""" 9345 9346The '``shl``' instruction returns the first operand shifted to the left 9347a specified number of bits. 9348 9349Arguments: 9350"""""""""" 9351 9352Both arguments to the '``shl``' instruction must be the same 9353:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 9354'``op2``' is treated as an unsigned value. 9355 9356Semantics: 9357"""""""""" 9358 9359The value produced is ``op1`` \* 2\ :sup:`op2` mod 2\ :sup:`n`, 9360where ``n`` is the width of the result. If ``op2`` is (statically or 9361dynamically) equal to or larger than the number of bits in 9362``op1``, this instruction returns a :ref:`poison value <poisonvalues>`. 9363If the arguments are vectors, each vector element of ``op1`` is shifted 9364by the corresponding shift amount in ``op2``. 9365 9366If the ``nuw`` keyword is present, then the shift produces a poison 9367value if it shifts out any non-zero bits. 9368If the ``nsw`` keyword is present, then the shift produces a poison 9369value if it shifts out any bits that disagree with the resultant sign bit. 9370 9371Example: 9372"""""""" 9373 9374.. code-block:: text 9375 9376 <result> = shl i32 4, %var ; yields i32: 4 << %var 9377 <result> = shl i32 4, 2 ; yields i32: 16 9378 <result> = shl i32 1, 10 ; yields i32: 1024 9379 <result> = shl i32 1, 32 ; undefined 9380 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 2, i32 4> 9381 9382.. _i_lshr: 9383 9384 9385'``lshr``' Instruction 9386^^^^^^^^^^^^^^^^^^^^^^ 9387 9388Syntax: 9389""""""" 9390 9391:: 9392 9393 <result> = lshr <ty> <op1>, <op2> ; yields ty:result 9394 <result> = lshr exact <ty> <op1>, <op2> ; yields ty:result 9395 9396Overview: 9397""""""""" 9398 9399The '``lshr``' instruction (logical shift right) returns the first 9400operand shifted to the right a specified number of bits with zero fill. 9401 9402Arguments: 9403"""""""""" 9404 9405Both arguments to the '``lshr``' instruction must be the same 9406:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 9407'``op2``' is treated as an unsigned value. 9408 9409Semantics: 9410"""""""""" 9411 9412This instruction always performs a logical shift right operation. The 9413most significant bits of the result will be filled with zero bits after 9414the shift. If ``op2`` is (statically or dynamically) equal to or larger 9415than the number of bits in ``op1``, this instruction returns a :ref:`poison 9416value <poisonvalues>`. If the arguments are vectors, each vector element 9417of ``op1`` is shifted by the corresponding shift amount in ``op2``. 9418 9419If the ``exact`` keyword is present, the result value of the ``lshr`` is 9420a poison value if any of the bits shifted out are non-zero. 9421 9422Example: 9423"""""""" 9424 9425.. code-block:: text 9426 9427 <result> = lshr i32 4, 1 ; yields i32:result = 2 9428 <result> = lshr i32 4, 2 ; yields i32:result = 1 9429 <result> = lshr i8 4, 3 ; yields i8:result = 0 9430 <result> = lshr i8 -2, 1 ; yields i8:result = 0x7F 9431 <result> = lshr i32 1, 32 ; undefined 9432 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> ; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1> 9433 9434.. _i_ashr: 9435 9436'``ashr``' Instruction 9437^^^^^^^^^^^^^^^^^^^^^^ 9438 9439Syntax: 9440""""""" 9441 9442:: 9443 9444 <result> = ashr <ty> <op1>, <op2> ; yields ty:result 9445 <result> = ashr exact <ty> <op1>, <op2> ; yields ty:result 9446 9447Overview: 9448""""""""" 9449 9450The '``ashr``' instruction (arithmetic shift right) returns the first 9451operand shifted to the right a specified number of bits with sign 9452extension. 9453 9454Arguments: 9455"""""""""" 9456 9457Both arguments to the '``ashr``' instruction must be the same 9458:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer type. 9459'``op2``' is treated as an unsigned value. 9460 9461Semantics: 9462"""""""""" 9463 9464This instruction always performs an arithmetic shift right operation, 9465The most significant bits of the result will be filled with the sign bit 9466of ``op1``. If ``op2`` is (statically or dynamically) equal to or larger 9467than the number of bits in ``op1``, this instruction returns a :ref:`poison 9468value <poisonvalues>`. If the arguments are vectors, each vector element 9469of ``op1`` is shifted by the corresponding shift amount in ``op2``. 9470 9471If the ``exact`` keyword is present, the result value of the ``ashr`` is 9472a poison value if any of the bits shifted out are non-zero. 9473 9474Example: 9475"""""""" 9476 9477.. code-block:: text 9478 9479 <result> = ashr i32 4, 1 ; yields i32:result = 2 9480 <result> = ashr i32 4, 2 ; yields i32:result = 1 9481 <result> = ashr i8 4, 3 ; yields i8:result = 0 9482 <result> = ashr i8 -2, 1 ; yields i8:result = -1 9483 <result> = ashr i32 1, 32 ; undefined 9484 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> ; yields: result=<2 x i32> < i32 -1, i32 0> 9485 9486.. _i_and: 9487 9488'``and``' Instruction 9489^^^^^^^^^^^^^^^^^^^^^ 9490 9491Syntax: 9492""""""" 9493 9494:: 9495 9496 <result> = and <ty> <op1>, <op2> ; yields ty:result 9497 9498Overview: 9499""""""""" 9500 9501The '``and``' instruction returns the bitwise logical and of its two 9502operands. 9503 9504Arguments: 9505"""""""""" 9506 9507The two arguments to the '``and``' instruction must be 9508:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9509arguments must have identical types. 9510 9511Semantics: 9512"""""""""" 9513 9514The truth table used for the '``and``' instruction is: 9515 9516+-----+-----+-----+ 9517| In0 | In1 | Out | 9518+-----+-----+-----+ 9519| 0 | 0 | 0 | 9520+-----+-----+-----+ 9521| 0 | 1 | 0 | 9522+-----+-----+-----+ 9523| 1 | 0 | 0 | 9524+-----+-----+-----+ 9525| 1 | 1 | 1 | 9526+-----+-----+-----+ 9527 9528Example: 9529"""""""" 9530 9531.. code-block:: text 9532 9533 <result> = and i32 4, %var ; yields i32:result = 4 & %var 9534 <result> = and i32 15, 40 ; yields i32:result = 8 9535 <result> = and i32 4, 8 ; yields i32:result = 0 9536 9537.. _i_or: 9538 9539'``or``' Instruction 9540^^^^^^^^^^^^^^^^^^^^ 9541 9542Syntax: 9543""""""" 9544 9545:: 9546 9547 <result> = or <ty> <op1>, <op2> ; yields ty:result 9548 9549Overview: 9550""""""""" 9551 9552The '``or``' instruction returns the bitwise logical inclusive or of its 9553two operands. 9554 9555Arguments: 9556"""""""""" 9557 9558The two arguments to the '``or``' instruction must be 9559:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9560arguments must have identical types. 9561 9562Semantics: 9563"""""""""" 9564 9565The truth table used for the '``or``' instruction is: 9566 9567+-----+-----+-----+ 9568| In0 | In1 | Out | 9569+-----+-----+-----+ 9570| 0 | 0 | 0 | 9571+-----+-----+-----+ 9572| 0 | 1 | 1 | 9573+-----+-----+-----+ 9574| 1 | 0 | 1 | 9575+-----+-----+-----+ 9576| 1 | 1 | 1 | 9577+-----+-----+-----+ 9578 9579Example: 9580"""""""" 9581 9582:: 9583 9584 <result> = or i32 4, %var ; yields i32:result = 4 | %var 9585 <result> = or i32 15, 40 ; yields i32:result = 47 9586 <result> = or i32 4, 8 ; yields i32:result = 12 9587 9588.. _i_xor: 9589 9590'``xor``' Instruction 9591^^^^^^^^^^^^^^^^^^^^^ 9592 9593Syntax: 9594""""""" 9595 9596:: 9597 9598 <result> = xor <ty> <op1>, <op2> ; yields ty:result 9599 9600Overview: 9601""""""""" 9602 9603The '``xor``' instruction returns the bitwise logical exclusive or of 9604its two operands. The ``xor`` is used to implement the "one's 9605complement" operation, which is the "~" operator in C. 9606 9607Arguments: 9608"""""""""" 9609 9610The two arguments to the '``xor``' instruction must be 9611:ref:`integer <t_integer>` or :ref:`vector <t_vector>` of integer values. Both 9612arguments must have identical types. 9613 9614Semantics: 9615"""""""""" 9616 9617The truth table used for the '``xor``' instruction is: 9618 9619+-----+-----+-----+ 9620| In0 | In1 | Out | 9621+-----+-----+-----+ 9622| 0 | 0 | 0 | 9623+-----+-----+-----+ 9624| 0 | 1 | 1 | 9625+-----+-----+-----+ 9626| 1 | 0 | 1 | 9627+-----+-----+-----+ 9628| 1 | 1 | 0 | 9629+-----+-----+-----+ 9630 9631Example: 9632"""""""" 9633 9634.. code-block:: text 9635 9636 <result> = xor i32 4, %var ; yields i32:result = 4 ^ %var 9637 <result> = xor i32 15, 40 ; yields i32:result = 39 9638 <result> = xor i32 4, 8 ; yields i32:result = 12 9639 <result> = xor i32 %V, -1 ; yields i32:result = ~%V 9640 9641Vector Operations 9642----------------- 9643 9644LLVM supports several instructions to represent vector operations in a 9645target-independent manner. These instructions cover the element-access 9646and vector-specific operations needed to process vectors effectively. 9647While LLVM does directly support these vector operations, many 9648sophisticated algorithms will want to use target-specific intrinsics to 9649take full advantage of a specific target. 9650 9651.. _i_extractelement: 9652 9653'``extractelement``' Instruction 9654^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9655 9656Syntax: 9657""""""" 9658 9659:: 9660 9661 <result> = extractelement <n x <ty>> <val>, <ty2> <idx> ; yields <ty> 9662 <result> = extractelement <vscale x n x <ty>> <val>, <ty2> <idx> ; yields <ty> 9663 9664Overview: 9665""""""""" 9666 9667The '``extractelement``' instruction extracts a single scalar element 9668from a vector at a specified index. 9669 9670Arguments: 9671"""""""""" 9672 9673The first operand of an '``extractelement``' instruction is a value of 9674:ref:`vector <t_vector>` type. The second operand is an index indicating 9675the position from which to extract the element. The index may be a 9676variable of any integer type. 9677 9678Semantics: 9679"""""""""" 9680 9681The result is a scalar of the same type as the element type of ``val``. 9682Its value is the value at position ``idx`` of ``val``. If ``idx`` 9683exceeds the length of ``val`` for a fixed-length vector, the result is a 9684:ref:`poison value <poisonvalues>`. For a scalable vector, if the value 9685of ``idx`` exceeds the runtime length of the vector, the result is a 9686:ref:`poison value <poisonvalues>`. 9687 9688Example: 9689"""""""" 9690 9691.. code-block:: text 9692 9693 <result> = extractelement <4 x i32> %vec, i32 0 ; yields i32 9694 9695.. _i_insertelement: 9696 9697'``insertelement``' Instruction 9698^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9699 9700Syntax: 9701""""""" 9702 9703:: 9704 9705 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <n x <ty>> 9706 <result> = insertelement <vscale x n x <ty>> <val>, <ty> <elt>, <ty2> <idx> ; yields <vscale x n x <ty>> 9707 9708Overview: 9709""""""""" 9710 9711The '``insertelement``' instruction inserts a scalar element into a 9712vector at a specified index. 9713 9714Arguments: 9715"""""""""" 9716 9717The first operand of an '``insertelement``' instruction is a value of 9718:ref:`vector <t_vector>` type. The second operand is a scalar value whose 9719type must equal the element type of the first operand. The third operand 9720is an index indicating the position at which to insert the value. The 9721index may be a variable of any integer type. 9722 9723Semantics: 9724"""""""""" 9725 9726The result is a vector of the same type as ``val``. Its element values 9727are those of ``val`` except at position ``idx``, where it gets the value 9728``elt``. If ``idx`` exceeds the length of ``val`` for a fixed-length vector, 9729the result is a :ref:`poison value <poisonvalues>`. For a scalable vector, 9730if the value of ``idx`` exceeds the runtime length of the vector, the result 9731is a :ref:`poison value <poisonvalues>`. 9732 9733Example: 9734"""""""" 9735 9736.. code-block:: text 9737 9738 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 ; yields <4 x i32> 9739 9740.. _i_shufflevector: 9741 9742'``shufflevector``' Instruction 9743^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9744 9745Syntax: 9746""""""" 9747 9748:: 9749 9750 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> ; yields <m x <ty>> 9751 <result> = shufflevector <vscale x n x <ty>> <v1>, <vscale x n x <ty>> v2, <vscale x m x i32> <mask> ; yields <vscale x m x <ty>> 9752 9753Overview: 9754""""""""" 9755 9756The '``shufflevector``' instruction constructs a permutation of elements 9757from two input vectors, returning a vector with the same element type as 9758the input and length that is the same as the shuffle mask. 9759 9760Arguments: 9761"""""""""" 9762 9763The first two operands of a '``shufflevector``' instruction are vectors 9764with the same type. The third argument is a shuffle mask vector constant 9765whose element type is ``i32``. The mask vector elements must be constant 9766integers or ``undef`` values. The result of the instruction is a vector 9767whose length is the same as the shuffle mask and whose element type is the 9768same as the element type of the first two operands. 9769 9770Semantics: 9771"""""""""" 9772 9773The elements of the two input vectors are numbered from left to right 9774across both of the vectors. For each element of the result vector, the 9775shuffle mask selects an element from one of the input vectors to copy 9776to the result. Non-negative elements in the mask represent an index 9777into the concatenated pair of input vectors. 9778 9779If the shuffle mask is undefined, the result vector is undefined. If 9780the shuffle mask selects an undefined element from one of the input 9781vectors, the resulting element is undefined. An undefined element 9782in the mask vector specifies that the resulting element is undefined. 9783An undefined element in the mask vector prevents a poisoned vector 9784element from propagating. 9785 9786For scalable vectors, the only valid mask values at present are 9787``zeroinitializer`` and ``undef``, since we cannot write all indices as 9788literals for a vector with a length unknown at compile time. 9789 9790Example: 9791"""""""" 9792 9793.. code-block:: text 9794 9795 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2, 9796 <4 x i32> <i32 0, i32 4, i32 1, i32 5> ; yields <4 x i32> 9797 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef, 9798 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> - Identity shuffle. 9799 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef, 9800 <4 x i32> <i32 0, i32 1, i32 2, i32 3> ; yields <4 x i32> 9801 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2, 9802 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > ; yields <8 x i32> 9803 9804Aggregate Operations 9805-------------------- 9806 9807LLVM supports several instructions for working with 9808:ref:`aggregate <t_aggregate>` values. 9809 9810.. _i_extractvalue: 9811 9812'``extractvalue``' Instruction 9813^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9814 9815Syntax: 9816""""""" 9817 9818:: 9819 9820 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}* 9821 9822Overview: 9823""""""""" 9824 9825The '``extractvalue``' instruction extracts the value of a member field 9826from an :ref:`aggregate <t_aggregate>` value. 9827 9828Arguments: 9829"""""""""" 9830 9831The first operand of an '``extractvalue``' instruction is a value of 9832:ref:`struct <t_struct>` or :ref:`array <t_array>` type. The other operands are 9833constant indices to specify which value to extract in a similar manner 9834as indices in a '``getelementptr``' instruction. 9835 9836The major differences to ``getelementptr`` indexing are: 9837 9838- Since the value being indexed is not a pointer, the first index is 9839 omitted and assumed to be zero. 9840- At least one index must be specified. 9841- Not only struct indices but also array indices must be in bounds. 9842 9843Semantics: 9844"""""""""" 9845 9846The result is the value at the position in the aggregate specified by 9847the index operands. 9848 9849Example: 9850"""""""" 9851 9852.. code-block:: text 9853 9854 <result> = extractvalue {i32, float} %agg, 0 ; yields i32 9855 9856.. _i_insertvalue: 9857 9858'``insertvalue``' Instruction 9859^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 9860 9861Syntax: 9862""""""" 9863 9864:: 9865 9866 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* ; yields <aggregate type> 9867 9868Overview: 9869""""""""" 9870 9871The '``insertvalue``' instruction inserts a value into a member field in 9872an :ref:`aggregate <t_aggregate>` value. 9873 9874Arguments: 9875"""""""""" 9876 9877The first operand of an '``insertvalue``' instruction is a value of 9878:ref:`struct <t_struct>` or :ref:`array <t_array>` type. The second operand is 9879a first-class value to insert. The following operands are constant 9880indices indicating the position at which to insert the value in a 9881similar manner as indices in a '``extractvalue``' instruction. The value 9882to insert must have the same type as the value identified by the 9883indices. 9884 9885Semantics: 9886"""""""""" 9887 9888The result is an aggregate of the same type as ``val``. Its value is 9889that of ``val`` except that the value at the position specified by the 9890indices is that of ``elt``. 9891 9892Example: 9893"""""""" 9894 9895.. code-block:: llvm 9896 9897 %agg1 = insertvalue {i32, float} undef, i32 1, 0 ; yields {i32 1, float undef} 9898 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 ; yields {i32 1, float %val} 9899 %agg3 = insertvalue {i32, {float}} undef, float %val, 1, 0 ; yields {i32 undef, {float %val}} 9900 9901.. _memoryops: 9902 9903Memory Access and Addressing Operations 9904--------------------------------------- 9905 9906A key design point of an SSA-based representation is how it represents 9907memory. In LLVM, no memory locations are in SSA form, which makes things 9908very simple. This section describes how to read, write, and allocate 9909memory in LLVM. 9910 9911.. _i_alloca: 9912 9913'``alloca``' Instruction 9914^^^^^^^^^^^^^^^^^^^^^^^^ 9915 9916Syntax: 9917""""""" 9918 9919:: 9920 9921 <result> = alloca [inalloca] <type> [, <ty> <NumElements>] [, align <alignment>] [, addrspace(<num>)] ; yields type addrspace(num)*:result 9922 9923Overview: 9924""""""""" 9925 9926The '``alloca``' instruction allocates memory on the stack frame of the 9927currently executing function, to be automatically released when this 9928function returns to its caller. If the address space is not explicitly 9929specified, the object is allocated in the alloca address space from the 9930:ref:`datalayout string<langref_datalayout>`. 9931 9932Arguments: 9933"""""""""" 9934 9935The '``alloca``' instruction allocates ``sizeof(<type>)*NumElements`` 9936bytes of memory on the runtime stack, returning a pointer of the 9937appropriate type to the program. If "NumElements" is specified, it is 9938the number of elements allocated, otherwise "NumElements" is defaulted 9939to be one. If a constant alignment is specified, the value result of the 9940allocation is guaranteed to be aligned to at least that boundary. The 9941alignment may not be greater than ``1 << 32``. If not specified, or if 9942zero, the target can choose to align the allocation on any convenient 9943boundary compatible with the type. 9944 9945'``type``' may be any sized type. 9946 9947Semantics: 9948"""""""""" 9949 9950Memory is allocated; a pointer is returned. The allocated memory is 9951uninitialized, and loading from uninitialized memory produces an undefined 9952value. The operation itself is undefined if there is insufficient stack 9953space for the allocation.'``alloca``'d memory is automatically released 9954when the function returns. The '``alloca``' instruction is commonly used 9955to represent automatic variables that must have an address available. When 9956the function returns (either with the ``ret`` or ``resume`` instructions), 9957the memory is reclaimed. Allocating zero bytes is legal, but the returned 9958pointer may not be unique. The order in which memory is allocated (ie., 9959which way the stack grows) is not specified. 9960 9961Note that '``alloca``' outside of the alloca address space from the 9962:ref:`datalayout string<langref_datalayout>` is meaningful only if the 9963target has assigned it a semantics. 9964 9965If the returned pointer is used by :ref:`llvm.lifetime.start <int_lifestart>`, 9966the returned object is initially dead. 9967See :ref:`llvm.lifetime.start <int_lifestart>` and 9968:ref:`llvm.lifetime.end <int_lifeend>` for the precise semantics of 9969lifetime-manipulating intrinsics. 9970 9971Example: 9972"""""""" 9973 9974.. code-block:: llvm 9975 9976 %ptr = alloca i32 ; yields i32*:ptr 9977 %ptr = alloca i32, i32 4 ; yields i32*:ptr 9978 %ptr = alloca i32, i32 4, align 1024 ; yields i32*:ptr 9979 %ptr = alloca i32, align 1024 ; yields i32*:ptr 9980 9981.. _i_load: 9982 9983'``load``' Instruction 9984^^^^^^^^^^^^^^^^^^^^^^ 9985 9986Syntax: 9987""""""" 9988 9989:: 9990 9991 <result> = load [volatile] <ty>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<nontemp_node>][, !invariant.load !<empty_node>][, !invariant.group !<empty_node>][, !nonnull !<empty_node>][, !dereferenceable !<deref_bytes_node>][, !dereferenceable_or_null !<deref_bytes_node>][, !align !<align_node>][, !noundef !<empty_node>] 9992 <result> = load atomic [volatile] <ty>, <ty>* <pointer> [syncscope("<target-scope>")] <ordering>, align <alignment> [, !invariant.group !<empty_node>] 9993 !<nontemp_node> = !{ i32 1 } 9994 !<empty_node> = !{} 9995 !<deref_bytes_node> = !{ i64 <dereferenceable_bytes> } 9996 !<align_node> = !{ i64 <value_alignment> } 9997 9998Overview: 9999""""""""" 10000 10001The '``load``' instruction is used to read from memory. 10002 10003Arguments: 10004"""""""""" 10005 10006The argument to the ``load`` instruction specifies the memory address from which 10007to load. The type specified must be a :ref:`first class <t_firstclass>` type of 10008known size (i.e. not containing an :ref:`opaque structural type <t_opaque>`). If 10009the ``load`` is marked as ``volatile``, then the optimizer is not allowed to 10010modify the number or order of execution of this ``load`` with other 10011:ref:`volatile operations <volatile>`. 10012 10013If the ``load`` is marked as ``atomic``, it takes an extra :ref:`ordering 10014<ordering>` and optional ``syncscope("<target-scope>")`` argument. The 10015``release`` and ``acq_rel`` orderings are not valid on ``load`` instructions. 10016Atomic loads produce :ref:`defined <memmodel>` results when they may see 10017multiple atomic stores. The type of the pointee must be an integer, pointer, or 10018floating-point type whose bit width is a power of two greater than or equal to 10019eight and less than or equal to a target-specific size limit. ``align`` must be 10020explicitly specified on atomic loads, and the load has undefined behavior if the 10021alignment is not set to a value which is at least the size in bytes of the 10022pointee. ``!nontemporal`` does not have any defined semantics for atomic loads. 10023 10024The optional constant ``align`` argument specifies the alignment of the 10025operation (that is, the alignment of the memory address). A value of 0 10026or an omitted ``align`` argument means that the operation has the ABI 10027alignment for the target. It is the responsibility of the code emitter 10028to ensure that the alignment information is correct. Overestimating the 10029alignment results in undefined behavior. Underestimating the alignment 10030may produce less efficient code. An alignment of 1 is always safe. The 10031maximum possible alignment is ``1 << 32``. An alignment value higher 10032than the size of the loaded type implies memory up to the alignment 10033value bytes can be safely loaded without trapping in the default 10034address space. Access of the high bytes can interfere with debugging 10035tools, so should not be accessed if the function has the 10036``sanitize_thread`` or ``sanitize_address`` attributes. 10037 10038The optional ``!nontemporal`` metadata must reference a single 10039metadata name ``<nontemp_node>`` corresponding to a metadata node with one 10040``i32`` entry of value 1. The existence of the ``!nontemporal`` 10041metadata on the instruction tells the optimizer and code generator 10042that this load is not expected to be reused in the cache. The code 10043generator may select special instructions to save cache bandwidth, such 10044as the ``MOVNT`` instruction on x86. 10045 10046The optional ``!invariant.load`` metadata must reference a single 10047metadata name ``<empty_node>`` corresponding to a metadata node with no 10048entries. If a load instruction tagged with the ``!invariant.load`` 10049metadata is executed, the memory location referenced by the load has 10050to contain the same value at all points in the program where the 10051memory location is dereferenceable; otherwise, the behavior is 10052undefined. 10053 10054The optional ``!invariant.group`` metadata must reference a single metadata name 10055 ``<empty_node>`` corresponding to a metadata node with no entries. 10056 See ``invariant.group`` metadata :ref:`invariant.group <md_invariant.group>`. 10057 10058The optional ``!nonnull`` metadata must reference a single 10059metadata name ``<empty_node>`` corresponding to a metadata node with no 10060entries. The existence of the ``!nonnull`` metadata on the 10061instruction tells the optimizer that the value loaded is known to 10062never be null. If the value is null at runtime, the behavior is undefined. 10063This is analogous to the ``nonnull`` attribute on parameters and return 10064values. This metadata can only be applied to loads of a pointer type. 10065 10066The optional ``!dereferenceable`` metadata must reference a single metadata 10067name ``<deref_bytes_node>`` corresponding to a metadata node with one ``i64`` 10068entry. 10069See ``dereferenceable`` metadata :ref:`dereferenceable <md_dereferenceable>`. 10070 10071The optional ``!dereferenceable_or_null`` metadata must reference a single 10072metadata name ``<deref_bytes_node>`` corresponding to a metadata node with one 10073``i64`` entry. 10074See ``dereferenceable_or_null`` metadata :ref:`dereferenceable_or_null 10075<md_dereferenceable_or_null>`. 10076 10077The optional ``!align`` metadata must reference a single metadata name 10078``<align_node>`` corresponding to a metadata node with one ``i64`` entry. 10079The existence of the ``!align`` metadata on the instruction tells the 10080optimizer that the value loaded is known to be aligned to a boundary specified 10081by the integer value in the metadata node. The alignment must be a power of 2. 10082This is analogous to the ''align'' attribute on parameters and return values. 10083This metadata can only be applied to loads of a pointer type. If the returned 10084value is not appropriately aligned at runtime, the behavior is undefined. 10085 10086The optional ``!noundef`` metadata must reference a single metadata name 10087``<empty_node>`` corresponding to a node with no entries. The existence of 10088``!noundef`` metadata on the instruction tells the optimizer that the value 10089loaded is known to be :ref:`well defined <welldefinedvalues>`. 10090If the value isn't well defined, the behavior is undefined. 10091 10092Semantics: 10093"""""""""" 10094 10095The location of memory pointed to is loaded. If the value being loaded 10096is of scalar type then the number of bytes read does not exceed the 10097minimum number of bytes needed to hold all bits of the type. For 10098example, loading an ``i24`` reads at most three bytes. When loading a 10099value of a type like ``i20`` with a size that is not an integral number 10100of bytes, the result is undefined if the value was not originally 10101written using a store of the same type. 10102If the value being loaded is of aggregate type, the bytes that correspond to 10103padding may be accessed but are ignored, because it is impossible to observe 10104padding from the loaded aggregate value. 10105If ``<pointer>`` is not a well-defined value, the behavior is undefined. 10106 10107Examples: 10108""""""""" 10109 10110.. code-block:: llvm 10111 10112 %ptr = alloca i32 ; yields i32*:ptr 10113 store i32 3, i32* %ptr ; yields void 10114 %val = load i32, i32* %ptr ; yields i32:val = i32 3 10115 10116.. _i_store: 10117 10118'``store``' Instruction 10119^^^^^^^^^^^^^^^^^^^^^^^ 10120 10121Syntax: 10122""""""" 10123 10124:: 10125 10126 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<nontemp_node>][, !invariant.group !<empty_node>] ; yields void 10127 store atomic [volatile] <ty> <value>, <ty>* <pointer> [syncscope("<target-scope>")] <ordering>, align <alignment> [, !invariant.group !<empty_node>] ; yields void 10128 !<nontemp_node> = !{ i32 1 } 10129 !<empty_node> = !{} 10130 10131Overview: 10132""""""""" 10133 10134The '``store``' instruction is used to write to memory. 10135 10136Arguments: 10137"""""""""" 10138 10139There are two arguments to the ``store`` instruction: a value to store and an 10140address at which to store it. The type of the ``<pointer>`` operand must be a 10141pointer to the :ref:`first class <t_firstclass>` type of the ``<value>`` 10142operand. If the ``store`` is marked as ``volatile``, then the optimizer is not 10143allowed to modify the number or order of execution of this ``store`` with other 10144:ref:`volatile operations <volatile>`. Only values of :ref:`first class 10145<t_firstclass>` types of known size (i.e. not containing an :ref:`opaque 10146structural type <t_opaque>`) can be stored. 10147 10148If the ``store`` is marked as ``atomic``, it takes an extra :ref:`ordering 10149<ordering>` and optional ``syncscope("<target-scope>")`` argument. The 10150``acquire`` and ``acq_rel`` orderings aren't valid on ``store`` instructions. 10151Atomic loads produce :ref:`defined <memmodel>` results when they may see 10152multiple atomic stores. The type of the pointee must be an integer, pointer, or 10153floating-point type whose bit width is a power of two greater than or equal to 10154eight and less than or equal to a target-specific size limit. ``align`` must be 10155explicitly specified on atomic stores, and the store has undefined behavior if 10156the alignment is not set to a value which is at least the size in bytes of the 10157pointee. ``!nontemporal`` does not have any defined semantics for atomic stores. 10158 10159The optional constant ``align`` argument specifies the alignment of the 10160operation (that is, the alignment of the memory address). A value of 0 10161or an omitted ``align`` argument means that the operation has the ABI 10162alignment for the target. It is the responsibility of the code emitter 10163to ensure that the alignment information is correct. Overestimating the 10164alignment results in undefined behavior. Underestimating the 10165alignment may produce less efficient code. An alignment of 1 is always 10166safe. The maximum possible alignment is ``1 << 32``. An alignment 10167value higher than the size of the stored type implies memory up to the 10168alignment value bytes can be stored to without trapping in the default 10169address space. Storing to the higher bytes however may result in data 10170races if another thread can access the same address. Introducing a 10171data race is not allowed. Storing to the extra bytes is not allowed 10172even in situations where a data race is known to not exist if the 10173function has the ``sanitize_address`` attribute. 10174 10175The optional ``!nontemporal`` metadata must reference a single metadata 10176name ``<nontemp_node>`` corresponding to a metadata node with one ``i32`` entry 10177of value 1. The existence of the ``!nontemporal`` metadata on the instruction 10178tells the optimizer and code generator that this load is not expected to 10179be reused in the cache. The code generator may select special 10180instructions to save cache bandwidth, such as the ``MOVNT`` instruction on 10181x86. 10182 10183The optional ``!invariant.group`` metadata must reference a 10184single metadata name ``<empty_node>``. See ``invariant.group`` metadata. 10185 10186Semantics: 10187"""""""""" 10188 10189The contents of memory are updated to contain ``<value>`` at the 10190location specified by the ``<pointer>`` operand. If ``<value>`` is 10191of scalar type then the number of bytes written does not exceed the 10192minimum number of bytes needed to hold all bits of the type. For 10193example, storing an ``i24`` writes at most three bytes. When writing a 10194value of a type like ``i20`` with a size that is not an integral number 10195of bytes, it is unspecified what happens to the extra bits that do not 10196belong to the type, but they will typically be overwritten. 10197If ``<value>`` is of aggregate type, padding is filled with 10198:ref:`undef <undefvalues>`. 10199If ``<pointer>`` is not a well-defined value, the behavior is undefined. 10200 10201Example: 10202"""""""" 10203 10204.. code-block:: llvm 10205 10206 %ptr = alloca i32 ; yields i32*:ptr 10207 store i32 3, i32* %ptr ; yields void 10208 %val = load i32, i32* %ptr ; yields i32:val = i32 3 10209 10210.. _i_fence: 10211 10212'``fence``' Instruction 10213^^^^^^^^^^^^^^^^^^^^^^^ 10214 10215Syntax: 10216""""""" 10217 10218:: 10219 10220 fence [syncscope("<target-scope>")] <ordering> ; yields void 10221 10222Overview: 10223""""""""" 10224 10225The '``fence``' instruction is used to introduce happens-before edges 10226between operations. 10227 10228Arguments: 10229"""""""""" 10230 10231'``fence``' instructions take an :ref:`ordering <ordering>` argument which 10232defines what *synchronizes-with* edges they add. They can only be given 10233``acquire``, ``release``, ``acq_rel``, and ``seq_cst`` orderings. 10234 10235Semantics: 10236"""""""""" 10237 10238A fence A which has (at least) ``release`` ordering semantics 10239*synchronizes with* a fence B with (at least) ``acquire`` ordering 10240semantics if and only if there exist atomic operations X and Y, both 10241operating on some atomic object M, such that A is sequenced before X, X 10242modifies M (either directly or through some side effect of a sequence 10243headed by X), Y is sequenced before B, and Y observes M. This provides a 10244*happens-before* dependency between A and B. Rather than an explicit 10245``fence``, one (but not both) of the atomic operations X or Y might 10246provide a ``release`` or ``acquire`` (resp.) ordering constraint and 10247still *synchronize-with* the explicit ``fence`` and establish the 10248*happens-before* edge. 10249 10250A ``fence`` which has ``seq_cst`` ordering, in addition to having both 10251``acquire`` and ``release`` semantics specified above, participates in 10252the global program order of other ``seq_cst`` operations and/or fences. 10253 10254A ``fence`` instruction can also take an optional 10255":ref:`syncscope <syncscope>`" argument. 10256 10257Example: 10258"""""""" 10259 10260.. code-block:: text 10261 10262 fence acquire ; yields void 10263 fence syncscope("singlethread") seq_cst ; yields void 10264 fence syncscope("agent") seq_cst ; yields void 10265 10266.. _i_cmpxchg: 10267 10268'``cmpxchg``' Instruction 10269^^^^^^^^^^^^^^^^^^^^^^^^^ 10270 10271Syntax: 10272""""""" 10273 10274:: 10275 10276 cmpxchg [weak] [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [syncscope("<target-scope>")] <success ordering> <failure ordering>[, align <alignment>] ; yields { ty, i1 } 10277 10278Overview: 10279""""""""" 10280 10281The '``cmpxchg``' instruction is used to atomically modify memory. It 10282loads a value in memory and compares it to a given value. If they are 10283equal, it tries to store a new value into the memory. 10284 10285Arguments: 10286"""""""""" 10287 10288There are three arguments to the '``cmpxchg``' instruction: an address 10289to operate on, a value to compare to the value currently be at that 10290address, and a new value to place at that address if the compared values 10291are equal. The type of '<cmp>' must be an integer or pointer type whose 10292bit width is a power of two greater than or equal to eight and less 10293than or equal to a target-specific size limit. '<cmp>' and '<new>' must 10294have the same type, and the type of '<pointer>' must be a pointer to 10295that type. If the ``cmpxchg`` is marked as ``volatile``, then the 10296optimizer is not allowed to modify the number or order of execution of 10297this ``cmpxchg`` with other :ref:`volatile operations <volatile>`. 10298 10299The success and failure :ref:`ordering <ordering>` arguments specify how this 10300``cmpxchg`` synchronizes with other atomic operations. Both ordering parameters 10301must be at least ``monotonic``, the failure ordering cannot be either 10302``release`` or ``acq_rel``. 10303 10304A ``cmpxchg`` instruction can also take an optional 10305":ref:`syncscope <syncscope>`" argument. 10306 10307The instruction can take an optional ``align`` attribute. 10308The alignment must be a power of two greater or equal to the size of the 10309`<value>` type. If unspecified, the alignment is assumed to be equal to the 10310size of the '<value>' type. Note that this default alignment assumption is 10311different from the alignment used for the load/store instructions when align 10312isn't specified. 10313 10314The pointer passed into cmpxchg must have alignment greater than or 10315equal to the size in memory of the operand. 10316 10317Semantics: 10318"""""""""" 10319 10320The contents of memory at the location specified by the '``<pointer>``' operand 10321is read and compared to '``<cmp>``'; if the values are equal, '``<new>``' is 10322written to the location. The original value at the location is returned, 10323together with a flag indicating success (true) or failure (false). 10324 10325If the cmpxchg operation is marked as ``weak`` then a spurious failure is 10326permitted: the operation may not write ``<new>`` even if the comparison 10327matched. 10328 10329If the cmpxchg operation is strong (the default), the i1 value is 1 if and only 10330if the value loaded equals ``cmp``. 10331 10332A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of 10333identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic 10334load with an ordering parameter determined the second ordering parameter. 10335 10336Example: 10337"""""""" 10338 10339.. code-block:: llvm 10340 10341 entry: 10342 %orig = load atomic i32, i32* %ptr unordered, align 4 ; yields i32 10343 br label %loop 10344 10345 loop: 10346 %cmp = phi i32 [ %orig, %entry ], [%value_loaded, %loop] 10347 %squared = mul i32 %cmp, %cmp 10348 %val_success = cmpxchg i32* %ptr, i32 %cmp, i32 %squared acq_rel monotonic ; yields { i32, i1 } 10349 %value_loaded = extractvalue { i32, i1 } %val_success, 0 10350 %success = extractvalue { i32, i1 } %val_success, 1 10351 br i1 %success, label %done, label %loop 10352 10353 done: 10354 ... 10355 10356.. _i_atomicrmw: 10357 10358'``atomicrmw``' Instruction 10359^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10360 10361Syntax: 10362""""""" 10363 10364:: 10365 10366 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [syncscope("<target-scope>")] <ordering>[, align <alignment>] ; yields ty 10367 10368Overview: 10369""""""""" 10370 10371The '``atomicrmw``' instruction is used to atomically modify memory. 10372 10373Arguments: 10374"""""""""" 10375 10376There are three arguments to the '``atomicrmw``' instruction: an 10377operation to apply, an address whose value to modify, an argument to the 10378operation. The operation must be one of the following keywords: 10379 10380- xchg 10381- add 10382- sub 10383- and 10384- nand 10385- or 10386- xor 10387- max 10388- min 10389- umax 10390- umin 10391- fadd 10392- fsub 10393- fmax 10394- fmin 10395 10396For most of these operations, the type of '<value>' must be an integer 10397type whose bit width is a power of two greater than or equal to eight 10398and less than or equal to a target-specific size limit. For xchg, this 10399may also be a floating point or a pointer type with the same size constraints 10400as integers. For fadd/fsub/fmax/fmin, this must be a floating point type. The 10401type of the '``<pointer>``' operand must be a pointer to that type. If 10402the ``atomicrmw`` is marked as ``volatile``, then the optimizer is not 10403allowed to modify the number or order of execution of this 10404``atomicrmw`` with other :ref:`volatile operations <volatile>`. 10405 10406The instruction can take an optional ``align`` attribute. 10407The alignment must be a power of two greater or equal to the size of the 10408`<value>` type. If unspecified, the alignment is assumed to be equal to the 10409size of the '<value>' type. Note that this default alignment assumption is 10410different from the alignment used for the load/store instructions when align 10411isn't specified. 10412 10413A ``atomicrmw`` instruction can also take an optional 10414":ref:`syncscope <syncscope>`" argument. 10415 10416Semantics: 10417"""""""""" 10418 10419The contents of memory at the location specified by the '``<pointer>``' 10420operand are atomically read, modified, and written back. The original 10421value at the location is returned. The modification is specified by the 10422operation argument: 10423 10424- xchg: ``*ptr = val`` 10425- add: ``*ptr = *ptr + val`` 10426- sub: ``*ptr = *ptr - val`` 10427- and: ``*ptr = *ptr & val`` 10428- nand: ``*ptr = ~(*ptr & val)`` 10429- or: ``*ptr = *ptr | val`` 10430- xor: ``*ptr = *ptr ^ val`` 10431- max: ``*ptr = *ptr > val ? *ptr : val`` (using a signed comparison) 10432- min: ``*ptr = *ptr < val ? *ptr : val`` (using a signed comparison) 10433- umax: ``*ptr = *ptr > val ? *ptr : val`` (using an unsigned comparison) 10434- umin: ``*ptr = *ptr < val ? *ptr : val`` (using an unsigned comparison) 10435- fadd: ``*ptr = *ptr + val`` (using floating point arithmetic) 10436- fsub: ``*ptr = *ptr - val`` (using floating point arithmetic) 10437- fmax: ``*ptr = maxnum(*ptr, val)`` (match the `llvm.maxnum.*`` intrinsic) 10438- fmin: ``*ptr = minnum(*ptr, val)`` (match the `llvm.minnum.*`` intrinsic) 10439 10440Example: 10441"""""""" 10442 10443.. code-block:: llvm 10444 10445 %old = atomicrmw add i32* %ptr, i32 1 acquire ; yields i32 10446 10447.. _i_getelementptr: 10448 10449'``getelementptr``' Instruction 10450^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10451 10452Syntax: 10453""""""" 10454 10455:: 10456 10457 <result> = getelementptr <ty>, <ty>* <ptrval>{, [inrange] <ty> <idx>}* 10458 <result> = getelementptr inbounds <ty>, <ty>* <ptrval>{, [inrange] <ty> <idx>}* 10459 <result> = getelementptr <ty>, <ptr vector> <ptrval>, [inrange] <vector index type> <idx> 10460 10461Overview: 10462""""""""" 10463 10464The '``getelementptr``' instruction is used to get the address of a 10465subelement of an :ref:`aggregate <t_aggregate>` data structure. It performs 10466address calculation only and does not access memory. The instruction can also 10467be used to calculate a vector of such addresses. 10468 10469Arguments: 10470"""""""""" 10471 10472The first argument is always a type used as the basis for the calculations. 10473The second argument is always a pointer or a vector of pointers, and is the 10474base address to start from. The remaining arguments are indices 10475that indicate which of the elements of the aggregate object are indexed. 10476The interpretation of each index is dependent on the type being indexed 10477into. The first index always indexes the pointer value given as the 10478second argument, the second index indexes a value of the type pointed to 10479(not necessarily the value directly pointed to, since the first index 10480can be non-zero), etc. The first type indexed into must be a pointer 10481value, subsequent types can be arrays, vectors, and structs. Note that 10482subsequent types being indexed into can never be pointers, since that 10483would require loading the pointer before continuing calculation. 10484 10485The type of each index argument depends on the type it is indexing into. 10486When indexing into a (optionally packed) structure, only ``i32`` integer 10487**constants** are allowed (when using a vector of indices they must all 10488be the **same** ``i32`` integer constant). When indexing into an array, 10489pointer or vector, integers of any width are allowed, and they are not 10490required to be constant. These integers are treated as signed values 10491where relevant. 10492 10493For example, let's consider a C code fragment and how it gets compiled 10494to LLVM: 10495 10496.. code-block:: c 10497 10498 struct RT { 10499 char A; 10500 int B[10][20]; 10501 char C; 10502 }; 10503 struct ST { 10504 int X; 10505 double Y; 10506 struct RT Z; 10507 }; 10508 10509 int *foo(struct ST *s) { 10510 return &s[1].Z.B[5][13]; 10511 } 10512 10513The LLVM code generated by Clang is: 10514 10515.. code-block:: llvm 10516 10517 %struct.RT = type { i8, [10 x [20 x i32]], i8 } 10518 %struct.ST = type { i32, double, %struct.RT } 10519 10520 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp { 10521 entry: 10522 %arrayidx = getelementptr inbounds %struct.ST, %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13 10523 ret i32* %arrayidx 10524 } 10525 10526Semantics: 10527"""""""""" 10528 10529In the example above, the first index is indexing into the 10530'``%struct.ST*``' type, which is a pointer, yielding a '``%struct.ST``' 10531= '``{ i32, double, %struct.RT }``' type, a structure. The second index 10532indexes into the third element of the structure, yielding a 10533'``%struct.RT``' = '``{ i8 , [10 x [20 x i32]], i8 }``' type, another 10534structure. The third index indexes into the second element of the 10535structure, yielding a '``[10 x [20 x i32]]``' type, an array. The two 10536dimensions of the array are subscripted into, yielding an '``i32``' 10537type. The '``getelementptr``' instruction returns a pointer to this 10538element, thus computing a value of '``i32*``' type. 10539 10540Note that it is perfectly legal to index partially through a structure, 10541returning a pointer to an inner element. Because of this, the LLVM code 10542for the given testcase is equivalent to: 10543 10544.. code-block:: llvm 10545 10546 define i32* @foo(%struct.ST* %s) { 10547 %t1 = getelementptr %struct.ST, %struct.ST* %s, i32 1 ; yields %struct.ST*:%t1 10548 %t2 = getelementptr %struct.ST, %struct.ST* %t1, i32 0, i32 2 ; yields %struct.RT*:%t2 10549 %t3 = getelementptr %struct.RT, %struct.RT* %t2, i32 0, i32 1 ; yields [10 x [20 x i32]]*:%t3 10550 %t4 = getelementptr [10 x [20 x i32]], [10 x [20 x i32]]* %t3, i32 0, i32 5 ; yields [20 x i32]*:%t4 10551 %t5 = getelementptr [20 x i32], [20 x i32]* %t4, i32 0, i32 13 ; yields i32*:%t5 10552 ret i32* %t5 10553 } 10554 10555If the ``inbounds`` keyword is present, the result value of the 10556``getelementptr`` is a :ref:`poison value <poisonvalues>` if one of the 10557following rules is violated: 10558 10559* The base pointer has an *in bounds* address of an allocated object, which 10560 means that it points into an allocated object, or to its end. The only 10561 *in bounds* address for a null pointer in the default address-space is the 10562 null pointer itself. 10563* If the type of an index is larger than the pointer index type, the 10564 truncation to the pointer index type preserves the signed value. 10565* The multiplication of an index by the type size does not wrap the pointer 10566 index type in a signed sense (``nsw``). 10567* The successive addition of offsets (without adding the base address) does 10568 not wrap the pointer index type in a signed sense (``nsw``). 10569* The successive addition of the current address, interpreted as an unsigned 10570 number, and an offset, interpreted as a signed number, does not wrap the 10571 unsigned address space and remains *in bounds* of the allocated object. 10572 As a corollary, if the added offset is non-negative, the addition does not 10573 wrap in an unsigned sense (``nuw``). 10574* In cases where the base is a vector of pointers, the ``inbounds`` keyword 10575 applies to each of the computations element-wise. 10576 10577These rules are based on the assumption that no allocated object may cross 10578the unsigned address space boundary, and no allocated object may be larger 10579than half the pointer index type space. 10580 10581If the ``inbounds`` keyword is not present, the offsets are added to the 10582base address with silently-wrapping two's complement arithmetic. If the 10583offsets have a different width from the pointer, they are sign-extended 10584or truncated to the width of the pointer. The result value of the 10585``getelementptr`` may be outside the object pointed to by the base 10586pointer. The result value may not necessarily be used to access memory 10587though, even if it happens to point into allocated storage. See the 10588:ref:`Pointer Aliasing Rules <pointeraliasing>` section for more 10589information. 10590 10591If the ``inrange`` keyword is present before any index, loading from or 10592storing to any pointer derived from the ``getelementptr`` has undefined 10593behavior if the load or store would access memory outside of the bounds of 10594the element selected by the index marked as ``inrange``. The result of a 10595pointer comparison or ``ptrtoint`` (including ``ptrtoint``-like operations 10596involving memory) involving a pointer derived from a ``getelementptr`` with 10597the ``inrange`` keyword is undefined, with the exception of comparisons 10598in the case where both operands are in the range of the element selected 10599by the ``inrange`` keyword, inclusive of the address one past the end of 10600that element. Note that the ``inrange`` keyword is currently only allowed 10601in constant ``getelementptr`` expressions. 10602 10603The getelementptr instruction is often confusing. For some more insight 10604into how it works, see :doc:`the getelementptr FAQ <GetElementPtr>`. 10605 10606Example: 10607"""""""" 10608 10609.. code-block:: llvm 10610 10611 ; yields [12 x i8]*:aptr 10612 %aptr = getelementptr {i32, [12 x i8]}, {i32, [12 x i8]}* %saptr, i64 0, i32 1 10613 ; yields i8*:vptr 10614 %vptr = getelementptr {i32, <2 x i8>}, {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1 10615 ; yields i8*:eptr 10616 %eptr = getelementptr [12 x i8], [12 x i8]* %aptr, i64 0, i32 1 10617 ; yields i32*:iptr 10618 %iptr = getelementptr [10 x i32], [10 x i32]* @arr, i16 0, i16 0 10619 10620Vector of pointers: 10621""""""""""""""""""" 10622 10623The ``getelementptr`` returns a vector of pointers, instead of a single address, 10624when one or more of its arguments is a vector. In such cases, all vector 10625arguments should have the same number of elements, and every scalar argument 10626will be effectively broadcast into a vector during address calculation. 10627 10628.. code-block:: llvm 10629 10630 ; All arguments are vectors: 10631 ; A[i] = ptrs[i] + offsets[i]*sizeof(i8) 10632 %A = getelementptr i8, <4 x i8*> %ptrs, <4 x i64> %offsets 10633 10634 ; Add the same scalar offset to each pointer of a vector: 10635 ; A[i] = ptrs[i] + offset*sizeof(i8) 10636 %A = getelementptr i8, <4 x i8*> %ptrs, i64 %offset 10637 10638 ; Add distinct offsets to the same pointer: 10639 ; A[i] = ptr + offsets[i]*sizeof(i8) 10640 %A = getelementptr i8, i8* %ptr, <4 x i64> %offsets 10641 10642 ; In all cases described above the type of the result is <4 x i8*> 10643 10644The two following instructions are equivalent: 10645 10646.. code-block:: llvm 10647 10648 getelementptr %struct.ST, <4 x %struct.ST*> %s, <4 x i64> %ind1, 10649 <4 x i32> <i32 2, i32 2, i32 2, i32 2>, 10650 <4 x i32> <i32 1, i32 1, i32 1, i32 1>, 10651 <4 x i32> %ind4, 10652 <4 x i64> <i64 13, i64 13, i64 13, i64 13> 10653 10654 getelementptr %struct.ST, <4 x %struct.ST*> %s, <4 x i64> %ind1, 10655 i32 2, i32 1, <4 x i32> %ind4, i64 13 10656 10657Let's look at the C code, where the vector version of ``getelementptr`` 10658makes sense: 10659 10660.. code-block:: c 10661 10662 // Let's assume that we vectorize the following loop: 10663 double *A, *B; int *C; 10664 for (int i = 0; i < size; ++i) { 10665 A[i] = B[C[i]]; 10666 } 10667 10668.. code-block:: llvm 10669 10670 ; get pointers for 8 elements from array B 10671 %ptrs = getelementptr double, double* %B, <8 x i32> %C 10672 ; load 8 elements from array B into A 10673 %A = call <8 x double> @llvm.masked.gather.v8f64.v8p0f64(<8 x double*> %ptrs, 10674 i32 8, <8 x i1> %mask, <8 x double> %passthru) 10675 10676Conversion Operations 10677--------------------- 10678 10679The instructions in this category are the conversion instructions 10680(casting) which all take a single operand and a type. They perform 10681various bit conversions on the operand. 10682 10683.. _i_trunc: 10684 10685'``trunc .. to``' Instruction 10686^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10687 10688Syntax: 10689""""""" 10690 10691:: 10692 10693 <result> = trunc <ty> <value> to <ty2> ; yields ty2 10694 10695Overview: 10696""""""""" 10697 10698The '``trunc``' instruction truncates its operand to the type ``ty2``. 10699 10700Arguments: 10701"""""""""" 10702 10703The '``trunc``' instruction takes a value to trunc, and a type to trunc 10704it to. Both types must be of :ref:`integer <t_integer>` types, or vectors 10705of the same number of integers. The bit size of the ``value`` must be 10706larger than the bit size of the destination type, ``ty2``. Equal sized 10707types are not allowed. 10708 10709Semantics: 10710"""""""""" 10711 10712The '``trunc``' instruction truncates the high order bits in ``value`` 10713and converts the remaining bits to ``ty2``. Since the source size must 10714be larger than the destination size, ``trunc`` cannot be a *no-op cast*. 10715It will always truncate bits. 10716 10717Example: 10718"""""""" 10719 10720.. code-block:: llvm 10721 10722 %X = trunc i32 257 to i8 ; yields i8:1 10723 %Y = trunc i32 123 to i1 ; yields i1:true 10724 %Z = trunc i32 122 to i1 ; yields i1:false 10725 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> ; yields <i8 8, i8 7> 10726 10727.. _i_zext: 10728 10729'``zext .. to``' Instruction 10730^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10731 10732Syntax: 10733""""""" 10734 10735:: 10736 10737 <result> = zext <ty> <value> to <ty2> ; yields ty2 10738 10739Overview: 10740""""""""" 10741 10742The '``zext``' instruction zero extends its operand to type ``ty2``. 10743 10744Arguments: 10745"""""""""" 10746 10747The '``zext``' instruction takes a value to cast, and a type to cast it 10748to. Both types must be of :ref:`integer <t_integer>` types, or vectors of 10749the same number of integers. The bit size of the ``value`` must be 10750smaller than the bit size of the destination type, ``ty2``. 10751 10752Semantics: 10753"""""""""" 10754 10755The ``zext`` fills the high order bits of the ``value`` with zero bits 10756until it reaches the size of the destination type, ``ty2``. 10757 10758When zero extending from i1, the result will always be either 0 or 1. 10759 10760Example: 10761"""""""" 10762 10763.. code-block:: llvm 10764 10765 %X = zext i32 257 to i64 ; yields i64:257 10766 %Y = zext i1 true to i32 ; yields i32:1 10767 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7> 10768 10769.. _i_sext: 10770 10771'``sext .. to``' Instruction 10772^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10773 10774Syntax: 10775""""""" 10776 10777:: 10778 10779 <result> = sext <ty> <value> to <ty2> ; yields ty2 10780 10781Overview: 10782""""""""" 10783 10784The '``sext``' sign extends ``value`` to the type ``ty2``. 10785 10786Arguments: 10787"""""""""" 10788 10789The '``sext``' instruction takes a value to cast, and a type to cast it 10790to. Both types must be of :ref:`integer <t_integer>` types, or vectors of 10791the same number of integers. The bit size of the ``value`` must be 10792smaller than the bit size of the destination type, ``ty2``. 10793 10794Semantics: 10795"""""""""" 10796 10797The '``sext``' instruction performs a sign extension by copying the sign 10798bit (highest order bit) of the ``value`` until it reaches the bit size 10799of the type ``ty2``. 10800 10801When sign extending from i1, the extension always results in -1 or 0. 10802 10803Example: 10804"""""""" 10805 10806.. code-block:: llvm 10807 10808 %X = sext i8 -1 to i16 ; yields i16 :65535 10809 %Y = sext i1 true to i32 ; yields i32:-1 10810 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> ; yields <i32 8, i32 7> 10811 10812'``fptrunc .. to``' Instruction 10813^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10814 10815Syntax: 10816""""""" 10817 10818:: 10819 10820 <result> = fptrunc <ty> <value> to <ty2> ; yields ty2 10821 10822Overview: 10823""""""""" 10824 10825The '``fptrunc``' instruction truncates ``value`` to type ``ty2``. 10826 10827Arguments: 10828"""""""""" 10829 10830The '``fptrunc``' instruction takes a :ref:`floating-point <t_floating>` 10831value to cast and a :ref:`floating-point <t_floating>` type to cast it to. 10832The size of ``value`` must be larger than the size of ``ty2``. This 10833implies that ``fptrunc`` cannot be used to make a *no-op cast*. 10834 10835Semantics: 10836"""""""""" 10837 10838The '``fptrunc``' instruction casts a ``value`` from a larger 10839:ref:`floating-point <t_floating>` type to a smaller :ref:`floating-point 10840<t_floating>` type. 10841This instruction is assumed to execute in the default :ref:`floating-point 10842environment <floatenv>`. 10843 10844Example: 10845"""""""" 10846 10847.. code-block:: llvm 10848 10849 %X = fptrunc double 16777217.0 to float ; yields float:16777216.0 10850 %Y = fptrunc double 1.0E+300 to half ; yields half:+infinity 10851 10852'``fpext .. to``' Instruction 10853^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10854 10855Syntax: 10856""""""" 10857 10858:: 10859 10860 <result> = fpext <ty> <value> to <ty2> ; yields ty2 10861 10862Overview: 10863""""""""" 10864 10865The '``fpext``' extends a floating-point ``value`` to a larger floating-point 10866value. 10867 10868Arguments: 10869"""""""""" 10870 10871The '``fpext``' instruction takes a :ref:`floating-point <t_floating>` 10872``value`` to cast, and a :ref:`floating-point <t_floating>` type to cast it 10873to. The source type must be smaller than the destination type. 10874 10875Semantics: 10876"""""""""" 10877 10878The '``fpext``' instruction extends the ``value`` from a smaller 10879:ref:`floating-point <t_floating>` type to a larger :ref:`floating-point 10880<t_floating>` type. The ``fpext`` cannot be used to make a 10881*no-op cast* because it always changes bits. Use ``bitcast`` to make a 10882*no-op cast* for a floating-point cast. 10883 10884Example: 10885"""""""" 10886 10887.. code-block:: llvm 10888 10889 %X = fpext float 3.125 to double ; yields double:3.125000e+00 10890 %Y = fpext double %X to fp128 ; yields fp128:0xL00000000000000004000900000000000 10891 10892'``fptoui .. to``' Instruction 10893^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10894 10895Syntax: 10896""""""" 10897 10898:: 10899 10900 <result> = fptoui <ty> <value> to <ty2> ; yields ty2 10901 10902Overview: 10903""""""""" 10904 10905The '``fptoui``' converts a floating-point ``value`` to its unsigned 10906integer equivalent of type ``ty2``. 10907 10908Arguments: 10909"""""""""" 10910 10911The '``fptoui``' instruction takes a value to cast, which must be a 10912scalar or vector :ref:`floating-point <t_floating>` value, and a type to 10913cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If 10914``ty`` is a vector floating-point type, ``ty2`` must be a vector integer 10915type with the same number of elements as ``ty`` 10916 10917Semantics: 10918"""""""""" 10919 10920The '``fptoui``' instruction converts its :ref:`floating-point 10921<t_floating>` operand into the nearest (rounding towards zero) 10922unsigned integer value. If the value cannot fit in ``ty2``, the result 10923is a :ref:`poison value <poisonvalues>`. 10924 10925Example: 10926"""""""" 10927 10928.. code-block:: llvm 10929 10930 %X = fptoui double 123.0 to i32 ; yields i32:123 10931 %Y = fptoui float 1.0E+300 to i1 ; yields undefined:1 10932 %Z = fptoui float 1.04E+17 to i8 ; yields undefined:1 10933 10934'``fptosi .. to``' Instruction 10935^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10936 10937Syntax: 10938""""""" 10939 10940:: 10941 10942 <result> = fptosi <ty> <value> to <ty2> ; yields ty2 10943 10944Overview: 10945""""""""" 10946 10947The '``fptosi``' instruction converts :ref:`floating-point <t_floating>` 10948``value`` to type ``ty2``. 10949 10950Arguments: 10951"""""""""" 10952 10953The '``fptosi``' instruction takes a value to cast, which must be a 10954scalar or vector :ref:`floating-point <t_floating>` value, and a type to 10955cast it to ``ty2``, which must be an :ref:`integer <t_integer>` type. If 10956``ty`` is a vector floating-point type, ``ty2`` must be a vector integer 10957type with the same number of elements as ``ty`` 10958 10959Semantics: 10960"""""""""" 10961 10962The '``fptosi``' instruction converts its :ref:`floating-point 10963<t_floating>` operand into the nearest (rounding towards zero) 10964signed integer value. If the value cannot fit in ``ty2``, the result 10965is a :ref:`poison value <poisonvalues>`. 10966 10967Example: 10968"""""""" 10969 10970.. code-block:: llvm 10971 10972 %X = fptosi double -123.0 to i32 ; yields i32:-123 10973 %Y = fptosi float 1.0E-247 to i1 ; yields undefined:1 10974 %Z = fptosi float 1.04E+17 to i8 ; yields undefined:1 10975 10976'``uitofp .. to``' Instruction 10977^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 10978 10979Syntax: 10980""""""" 10981 10982:: 10983 10984 <result> = uitofp <ty> <value> to <ty2> ; yields ty2 10985 10986Overview: 10987""""""""" 10988 10989The '``uitofp``' instruction regards ``value`` as an unsigned integer 10990and converts that value to the ``ty2`` type. 10991 10992Arguments: 10993"""""""""" 10994 10995The '``uitofp``' instruction takes a value to cast, which must be a 10996scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to 10997``ty2``, which must be an :ref:`floating-point <t_floating>` type. If 10998``ty`` is a vector integer type, ``ty2`` must be a vector floating-point 10999type with the same number of elements as ``ty`` 11000 11001Semantics: 11002"""""""""" 11003 11004The '``uitofp``' instruction interprets its operand as an unsigned 11005integer quantity and converts it to the corresponding floating-point 11006value. If the value cannot be exactly represented, it is rounded using 11007the default rounding mode. 11008 11009 11010Example: 11011"""""""" 11012 11013.. code-block:: llvm 11014 11015 %X = uitofp i32 257 to float ; yields float:257.0 11016 %Y = uitofp i8 -1 to double ; yields double:255.0 11017 11018'``sitofp .. to``' Instruction 11019^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11020 11021Syntax: 11022""""""" 11023 11024:: 11025 11026 <result> = sitofp <ty> <value> to <ty2> ; yields ty2 11027 11028Overview: 11029""""""""" 11030 11031The '``sitofp``' instruction regards ``value`` as a signed integer and 11032converts that value to the ``ty2`` type. 11033 11034Arguments: 11035"""""""""" 11036 11037The '``sitofp``' instruction takes a value to cast, which must be a 11038scalar or vector :ref:`integer <t_integer>` value, and a type to cast it to 11039``ty2``, which must be an :ref:`floating-point <t_floating>` type. If 11040``ty`` is a vector integer type, ``ty2`` must be a vector floating-point 11041type with the same number of elements as ``ty`` 11042 11043Semantics: 11044"""""""""" 11045 11046The '``sitofp``' instruction interprets its operand as a signed integer 11047quantity and converts it to the corresponding floating-point value. If the 11048value cannot be exactly represented, it is rounded using the default rounding 11049mode. 11050 11051Example: 11052"""""""" 11053 11054.. code-block:: llvm 11055 11056 %X = sitofp i32 257 to float ; yields float:257.0 11057 %Y = sitofp i8 -1 to double ; yields double:-1.0 11058 11059.. _i_ptrtoint: 11060 11061'``ptrtoint .. to``' Instruction 11062^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11063 11064Syntax: 11065""""""" 11066 11067:: 11068 11069 <result> = ptrtoint <ty> <value> to <ty2> ; yields ty2 11070 11071Overview: 11072""""""""" 11073 11074The '``ptrtoint``' instruction converts the pointer or a vector of 11075pointers ``value`` to the integer (or vector of integers) type ``ty2``. 11076 11077Arguments: 11078"""""""""" 11079 11080The '``ptrtoint``' instruction takes a ``value`` to cast, which must be 11081a value of type :ref:`pointer <t_pointer>` or a vector of pointers, and a 11082type to cast it to ``ty2``, which must be an :ref:`integer <t_integer>` or 11083a vector of integers type. 11084 11085Semantics: 11086"""""""""" 11087 11088The '``ptrtoint``' instruction converts ``value`` to integer type 11089``ty2`` by interpreting the pointer value as an integer and either 11090truncating or zero extending that value to the size of the integer type. 11091If ``value`` is smaller than ``ty2`` then a zero extension is done. If 11092``value`` is larger than ``ty2`` then a truncation is done. If they are 11093the same size, then nothing is done (*no-op cast*) other than a type 11094change. 11095 11096Example: 11097"""""""" 11098 11099.. code-block:: llvm 11100 11101 %X = ptrtoint i32* %P to i8 ; yields truncation on 32-bit architecture 11102 %Y = ptrtoint i32* %P to i64 ; yields zero extension on 32-bit architecture 11103 %Z = ptrtoint <4 x i32*> %P to <4 x i64>; yields vector zero extension for a vector of addresses on 32-bit architecture 11104 11105.. _i_inttoptr: 11106 11107'``inttoptr .. to``' Instruction 11108^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11109 11110Syntax: 11111""""""" 11112 11113:: 11114 11115 <result> = inttoptr <ty> <value> to <ty2>[, !dereferenceable !<deref_bytes_node>][, !dereferenceable_or_null !<deref_bytes_node>] ; yields ty2 11116 11117Overview: 11118""""""""" 11119 11120The '``inttoptr``' instruction converts an integer ``value`` to a 11121pointer type, ``ty2``. 11122 11123Arguments: 11124"""""""""" 11125 11126The '``inttoptr``' instruction takes an :ref:`integer <t_integer>` value to 11127cast, and a type to cast it to, which must be a :ref:`pointer <t_pointer>` 11128type. 11129 11130The optional ``!dereferenceable`` metadata must reference a single metadata 11131name ``<deref_bytes_node>`` corresponding to a metadata node with one ``i64`` 11132entry. 11133See ``dereferenceable`` metadata. 11134 11135The optional ``!dereferenceable_or_null`` metadata must reference a single 11136metadata name ``<deref_bytes_node>`` corresponding to a metadata node with one 11137``i64`` entry. 11138See ``dereferenceable_or_null`` metadata. 11139 11140Semantics: 11141"""""""""" 11142 11143The '``inttoptr``' instruction converts ``value`` to type ``ty2`` by 11144applying either a zero extension or a truncation depending on the size 11145of the integer ``value``. If ``value`` is larger than the size of a 11146pointer then a truncation is done. If ``value`` is smaller than the size 11147of a pointer then a zero extension is done. If they are the same size, 11148nothing is done (*no-op cast*). 11149 11150Example: 11151"""""""" 11152 11153.. code-block:: llvm 11154 11155 %X = inttoptr i32 255 to i32* ; yields zero extension on 64-bit architecture 11156 %Y = inttoptr i32 255 to i32* ; yields no-op on 32-bit architecture 11157 %Z = inttoptr i64 0 to i32* ; yields truncation on 32-bit architecture 11158 %Z = inttoptr <4 x i32> %G to <4 x i8*>; yields truncation of vector G to four pointers 11159 11160.. _i_bitcast: 11161 11162'``bitcast .. to``' Instruction 11163^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11164 11165Syntax: 11166""""""" 11167 11168:: 11169 11170 <result> = bitcast <ty> <value> to <ty2> ; yields ty2 11171 11172Overview: 11173""""""""" 11174 11175The '``bitcast``' instruction converts ``value`` to type ``ty2`` without 11176changing any bits. 11177 11178Arguments: 11179"""""""""" 11180 11181The '``bitcast``' instruction takes a value to cast, which must be a 11182non-aggregate first class value, and a type to cast it to, which must 11183also be a non-aggregate :ref:`first class <t_firstclass>` type. The 11184bit sizes of ``value`` and the destination type, ``ty2``, must be 11185identical. If the source type is a pointer, the destination type must 11186also be a pointer of the same size. This instruction supports bitwise 11187conversion of vectors to integers and to vectors of other types (as 11188long as they have the same size). 11189 11190Semantics: 11191"""""""""" 11192 11193The '``bitcast``' instruction converts ``value`` to type ``ty2``. It 11194is always a *no-op cast* because no bits change with this 11195conversion. The conversion is done as if the ``value`` had been stored 11196to memory and read back as type ``ty2``. Pointer (or vector of 11197pointers) types may only be converted to other pointer (or vector of 11198pointers) types with the same address space through this instruction. 11199To convert pointers to other types, use the :ref:`inttoptr <i_inttoptr>` 11200or :ref:`ptrtoint <i_ptrtoint>` instructions first. 11201 11202There is a caveat for bitcasts involving vector types in relation to 11203endianess. For example ``bitcast <2 x i8> <value> to i16`` puts element zero 11204of the vector in the least significant bits of the i16 for little-endian while 11205element zero ends up in the most significant bits for big-endian. 11206 11207Example: 11208"""""""" 11209 11210.. code-block:: text 11211 11212 %X = bitcast i8 255 to i8 ; yields i8 :-1 11213 %Y = bitcast i32* %x to i16* ; yields i16*:%x 11214 %Z = bitcast <2 x i32> %V to i64; ; yields i64: %V (depends on endianess) 11215 %Z = bitcast <2 x i32*> %V to <2 x i64*> ; yields <2 x i64*> 11216 11217.. _i_addrspacecast: 11218 11219'``addrspacecast .. to``' Instruction 11220^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11221 11222Syntax: 11223""""""" 11224 11225:: 11226 11227 <result> = addrspacecast <pty> <ptrval> to <pty2> ; yields pty2 11228 11229Overview: 11230""""""""" 11231 11232The '``addrspacecast``' instruction converts ``ptrval`` from ``pty`` in 11233address space ``n`` to type ``pty2`` in address space ``m``. 11234 11235Arguments: 11236"""""""""" 11237 11238The '``addrspacecast``' instruction takes a pointer or vector of pointer value 11239to cast and a pointer type to cast it to, which must have a different 11240address space. 11241 11242Semantics: 11243"""""""""" 11244 11245The '``addrspacecast``' instruction converts the pointer value 11246``ptrval`` to type ``pty2``. It can be a *no-op cast* or a complex 11247value modification, depending on the target and the address space 11248pair. Pointer conversions within the same address space must be 11249performed with the ``bitcast`` instruction. Note that if the address space 11250conversion is legal then both result and operand refer to the same memory 11251location. 11252 11253Example: 11254"""""""" 11255 11256.. code-block:: llvm 11257 11258 %X = addrspacecast i32* %x to i32 addrspace(1)* ; yields i32 addrspace(1)*:%x 11259 %Y = addrspacecast i32 addrspace(1)* %y to i64 addrspace(2)* ; yields i64 addrspace(2)*:%y 11260 %Z = addrspacecast <4 x i32*> %z to <4 x float addrspace(3)*> ; yields <4 x float addrspace(3)*>:%z 11261 11262.. _otherops: 11263 11264Other Operations 11265---------------- 11266 11267The instructions in this category are the "miscellaneous" instructions, 11268which defy better classification. 11269 11270.. _i_icmp: 11271 11272'``icmp``' Instruction 11273^^^^^^^^^^^^^^^^^^^^^^ 11274 11275Syntax: 11276""""""" 11277 11278:: 11279 11280 <result> = icmp <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result 11281 11282Overview: 11283""""""""" 11284 11285The '``icmp``' instruction returns a boolean value or a vector of 11286boolean values based on comparison of its two integer, integer vector, 11287pointer, or pointer vector operands. 11288 11289Arguments: 11290"""""""""" 11291 11292The '``icmp``' instruction takes three operands. The first operand is 11293the condition code indicating the kind of comparison to perform. It is 11294not a value, just a keyword. The possible condition codes are: 11295 11296.. _icmp_md_cc: 11297 11298#. ``eq``: equal 11299#. ``ne``: not equal 11300#. ``ugt``: unsigned greater than 11301#. ``uge``: unsigned greater or equal 11302#. ``ult``: unsigned less than 11303#. ``ule``: unsigned less or equal 11304#. ``sgt``: signed greater than 11305#. ``sge``: signed greater or equal 11306#. ``slt``: signed less than 11307#. ``sle``: signed less or equal 11308 11309The remaining two arguments must be :ref:`integer <t_integer>` or 11310:ref:`pointer <t_pointer>` or integer :ref:`vector <t_vector>` typed. They 11311must also be identical types. 11312 11313Semantics: 11314"""""""""" 11315 11316The '``icmp``' compares ``op1`` and ``op2`` according to the condition 11317code given as ``cond``. The comparison performed always yields either an 11318:ref:`i1 <t_integer>` or vector of ``i1`` result, as follows: 11319 11320.. _icmp_md_cc_sem: 11321 11322#. ``eq``: yields ``true`` if the operands are equal, ``false`` 11323 otherwise. No sign interpretation is necessary or performed. 11324#. ``ne``: yields ``true`` if the operands are unequal, ``false`` 11325 otherwise. No sign interpretation is necessary or performed. 11326#. ``ugt``: interprets the operands as unsigned values and yields 11327 ``true`` if ``op1`` is greater than ``op2``. 11328#. ``uge``: interprets the operands as unsigned values and yields 11329 ``true`` if ``op1`` is greater than or equal to ``op2``. 11330#. ``ult``: interprets the operands as unsigned values and yields 11331 ``true`` if ``op1`` is less than ``op2``. 11332#. ``ule``: interprets the operands as unsigned values and yields 11333 ``true`` if ``op1`` is less than or equal to ``op2``. 11334#. ``sgt``: interprets the operands as signed values and yields ``true`` 11335 if ``op1`` is greater than ``op2``. 11336#. ``sge``: interprets the operands as signed values and yields ``true`` 11337 if ``op1`` is greater than or equal to ``op2``. 11338#. ``slt``: interprets the operands as signed values and yields ``true`` 11339 if ``op1`` is less than ``op2``. 11340#. ``sle``: interprets the operands as signed values and yields ``true`` 11341 if ``op1`` is less than or equal to ``op2``. 11342 11343If the operands are :ref:`pointer <t_pointer>` typed, the pointer values 11344are compared as if they were integers. 11345 11346If the operands are integer vectors, then they are compared element by 11347element. The result is an ``i1`` vector with the same number of elements 11348as the values being compared. Otherwise, the result is an ``i1``. 11349 11350Example: 11351"""""""" 11352 11353.. code-block:: text 11354 11355 <result> = icmp eq i32 4, 5 ; yields: result=false 11356 <result> = icmp ne float* %X, %X ; yields: result=false 11357 <result> = icmp ult i16 4, 5 ; yields: result=true 11358 <result> = icmp sgt i16 4, 5 ; yields: result=false 11359 <result> = icmp ule i16 -4, 5 ; yields: result=false 11360 <result> = icmp sge i16 4, 5 ; yields: result=false 11361 11362.. _i_fcmp: 11363 11364'``fcmp``' Instruction 11365^^^^^^^^^^^^^^^^^^^^^^ 11366 11367Syntax: 11368""""""" 11369 11370:: 11371 11372 <result> = fcmp [fast-math flags]* <cond> <ty> <op1>, <op2> ; yields i1 or <N x i1>:result 11373 11374Overview: 11375""""""""" 11376 11377The '``fcmp``' instruction returns a boolean value or vector of boolean 11378values based on comparison of its operands. 11379 11380If the operands are floating-point scalars, then the result type is a 11381boolean (:ref:`i1 <t_integer>`). 11382 11383If the operands are floating-point vectors, then the result type is a 11384vector of boolean with the same number of elements as the operands being 11385compared. 11386 11387Arguments: 11388"""""""""" 11389 11390The '``fcmp``' instruction takes three operands. The first operand is 11391the condition code indicating the kind of comparison to perform. It is 11392not a value, just a keyword. The possible condition codes are: 11393 11394#. ``false``: no comparison, always returns false 11395#. ``oeq``: ordered and equal 11396#. ``ogt``: ordered and greater than 11397#. ``oge``: ordered and greater than or equal 11398#. ``olt``: ordered and less than 11399#. ``ole``: ordered and less than or equal 11400#. ``one``: ordered and not equal 11401#. ``ord``: ordered (no nans) 11402#. ``ueq``: unordered or equal 11403#. ``ugt``: unordered or greater than 11404#. ``uge``: unordered or greater than or equal 11405#. ``ult``: unordered or less than 11406#. ``ule``: unordered or less than or equal 11407#. ``une``: unordered or not equal 11408#. ``uno``: unordered (either nans) 11409#. ``true``: no comparison, always returns true 11410 11411*Ordered* means that neither operand is a QNAN while *unordered* means 11412that either operand may be a QNAN. 11413 11414Each of ``val1`` and ``val2`` arguments must be either a :ref:`floating-point 11415<t_floating>` type or a :ref:`vector <t_vector>` of floating-point type. 11416They must have identical types. 11417 11418Semantics: 11419"""""""""" 11420 11421The '``fcmp``' instruction compares ``op1`` and ``op2`` according to the 11422condition code given as ``cond``. If the operands are vectors, then the 11423vectors are compared element by element. Each comparison performed 11424always yields an :ref:`i1 <t_integer>` result, as follows: 11425 11426#. ``false``: always yields ``false``, regardless of operands. 11427#. ``oeq``: yields ``true`` if both operands are not a QNAN and ``op1`` 11428 is equal to ``op2``. 11429#. ``ogt``: yields ``true`` if both operands are not a QNAN and ``op1`` 11430 is greater than ``op2``. 11431#. ``oge``: yields ``true`` if both operands are not a QNAN and ``op1`` 11432 is greater than or equal to ``op2``. 11433#. ``olt``: yields ``true`` if both operands are not a QNAN and ``op1`` 11434 is less than ``op2``. 11435#. ``ole``: yields ``true`` if both operands are not a QNAN and ``op1`` 11436 is less than or equal to ``op2``. 11437#. ``one``: yields ``true`` if both operands are not a QNAN and ``op1`` 11438 is not equal to ``op2``. 11439#. ``ord``: yields ``true`` if both operands are not a QNAN. 11440#. ``ueq``: yields ``true`` if either operand is a QNAN or ``op1`` is 11441 equal to ``op2``. 11442#. ``ugt``: yields ``true`` if either operand is a QNAN or ``op1`` is 11443 greater than ``op2``. 11444#. ``uge``: yields ``true`` if either operand is a QNAN or ``op1`` is 11445 greater than or equal to ``op2``. 11446#. ``ult``: yields ``true`` if either operand is a QNAN or ``op1`` is 11447 less than ``op2``. 11448#. ``ule``: yields ``true`` if either operand is a QNAN or ``op1`` is 11449 less than or equal to ``op2``. 11450#. ``une``: yields ``true`` if either operand is a QNAN or ``op1`` is 11451 not equal to ``op2``. 11452#. ``uno``: yields ``true`` if either operand is a QNAN. 11453#. ``true``: always yields ``true``, regardless of operands. 11454 11455The ``fcmp`` instruction can also optionally take any number of 11456:ref:`fast-math flags <fastmath>`, which are optimization hints to enable 11457otherwise unsafe floating-point optimizations. 11458 11459Any set of fast-math flags are legal on an ``fcmp`` instruction, but the 11460only flags that have any effect on its semantics are those that allow 11461assumptions to be made about the values of input arguments; namely 11462``nnan``, ``ninf``, and ``reassoc``. See :ref:`fastmath` for more information. 11463 11464Example: 11465"""""""" 11466 11467.. code-block:: text 11468 11469 <result> = fcmp oeq float 4.0, 5.0 ; yields: result=false 11470 <result> = fcmp one float 4.0, 5.0 ; yields: result=true 11471 <result> = fcmp olt float 4.0, 5.0 ; yields: result=true 11472 <result> = fcmp ueq double 1.0, 2.0 ; yields: result=false 11473 11474.. _i_phi: 11475 11476'``phi``' Instruction 11477^^^^^^^^^^^^^^^^^^^^^ 11478 11479Syntax: 11480""""""" 11481 11482:: 11483 11484 <result> = phi [fast-math-flags] <ty> [ <val0>, <label0>], ... 11485 11486Overview: 11487""""""""" 11488 11489The '``phi``' instruction is used to implement the φ node in the SSA 11490graph representing the function. 11491 11492Arguments: 11493"""""""""" 11494 11495The type of the incoming values is specified with the first type field. 11496After this, the '``phi``' instruction takes a list of pairs as 11497arguments, with one pair for each predecessor basic block of the current 11498block. Only values of :ref:`first class <t_firstclass>` type may be used as 11499the value arguments to the PHI node. Only labels may be used as the 11500label arguments. 11501 11502There must be no non-phi instructions between the start of a basic block 11503and the PHI instructions: i.e. PHI instructions must be first in a basic 11504block. 11505 11506For the purposes of the SSA form, the use of each incoming value is 11507deemed to occur on the edge from the corresponding predecessor block to 11508the current block (but after any definition of an '``invoke``' 11509instruction's return value on the same edge). 11510 11511The optional ``fast-math-flags`` marker indicates that the phi has one 11512or more :ref:`fast-math-flags <fastmath>`. These are optimization hints 11513to enable otherwise unsafe floating-point optimizations. Fast-math-flags 11514are only valid for phis that return a floating-point scalar or vector 11515type, or an array (nested to any depth) of floating-point scalar or vector 11516types. 11517 11518Semantics: 11519"""""""""" 11520 11521At runtime, the '``phi``' instruction logically takes on the value 11522specified by the pair corresponding to the predecessor basic block that 11523executed just prior to the current block. 11524 11525Example: 11526"""""""" 11527 11528.. code-block:: llvm 11529 11530 Loop: ; Infinite loop that counts from 0 on up... 11531 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ] 11532 %nextindvar = add i32 %indvar, 1 11533 br label %Loop 11534 11535.. _i_select: 11536 11537'``select``' Instruction 11538^^^^^^^^^^^^^^^^^^^^^^^^ 11539 11540Syntax: 11541""""""" 11542 11543:: 11544 11545 <result> = select [fast-math flags] selty <cond>, <ty> <val1>, <ty> <val2> ; yields ty 11546 11547 selty is either i1 or {<N x i1>} 11548 11549Overview: 11550""""""""" 11551 11552The '``select``' instruction is used to choose one value based on a 11553condition, without IR-level branching. 11554 11555Arguments: 11556"""""""""" 11557 11558The '``select``' instruction requires an 'i1' value or a vector of 'i1' 11559values indicating the condition, and two values of the same :ref:`first 11560class <t_firstclass>` type. 11561 11562#. The optional ``fast-math flags`` marker indicates that the select has one or more 11563 :ref:`fast-math flags <fastmath>`. These are optimization hints to enable 11564 otherwise unsafe floating-point optimizations. Fast-math flags are only valid 11565 for selects that return a floating-point scalar or vector type, or an array 11566 (nested to any depth) of floating-point scalar or vector types. 11567 11568Semantics: 11569"""""""""" 11570 11571If the condition is an i1 and it evaluates to 1, the instruction returns 11572the first value argument; otherwise, it returns the second value 11573argument. 11574 11575If the condition is a vector of i1, then the value arguments must be 11576vectors of the same size, and the selection is done element by element. 11577 11578If the condition is an i1 and the value arguments are vectors of the 11579same size, then an entire vector is selected. 11580 11581Example: 11582"""""""" 11583 11584.. code-block:: llvm 11585 11586 %X = select i1 true, i8 17, i8 42 ; yields i8:17 11587 11588 11589.. _i_freeze: 11590 11591'``freeze``' Instruction 11592^^^^^^^^^^^^^^^^^^^^^^^^ 11593 11594Syntax: 11595""""""" 11596 11597:: 11598 11599 <result> = freeze ty <val> ; yields ty:result 11600 11601Overview: 11602""""""""" 11603 11604The '``freeze``' instruction is used to stop propagation of 11605:ref:`undef <undefvalues>` and :ref:`poison <poisonvalues>` values. 11606 11607Arguments: 11608"""""""""" 11609 11610The '``freeze``' instruction takes a single argument. 11611 11612Semantics: 11613"""""""""" 11614 11615If the argument is ``undef`` or ``poison``, '``freeze``' returns an 11616arbitrary, but fixed, value of type '``ty``'. 11617Otherwise, this instruction is a no-op and returns the input argument. 11618All uses of a value returned by the same '``freeze``' instruction are 11619guaranteed to always observe the same value, while different '``freeze``' 11620instructions may yield different values. 11621 11622While ``undef`` and ``poison`` pointers can be frozen, the result is a 11623non-dereferenceable pointer. See the 11624:ref:`Pointer Aliasing Rules <pointeraliasing>` section for more information. 11625If an aggregate value or vector is frozen, the operand is frozen element-wise. 11626The padding of an aggregate isn't considered, since it isn't visible 11627without storing it into memory and loading it with a different type. 11628 11629 11630Example: 11631"""""""" 11632 11633.. code-block:: text 11634 11635 %w = i32 undef 11636 %x = freeze i32 %w 11637 %y = add i32 %w, %w ; undef 11638 %z = add i32 %x, %x ; even number because all uses of %x observe 11639 ; the same value 11640 %x2 = freeze i32 %w 11641 %cmp = icmp eq i32 %x, %x2 ; can be true or false 11642 11643 ; example with vectors 11644 %v = <2 x i32> <i32 undef, i32 poison> 11645 %a = extractelement <2 x i32> %v, i32 0 ; undef 11646 %b = extractelement <2 x i32> %v, i32 1 ; poison 11647 %add = add i32 %a, %a ; undef 11648 11649 %v.fr = freeze <2 x i32> %v ; element-wise freeze 11650 %d = extractelement <2 x i32> %v.fr, i32 0 ; not undef 11651 %add.f = add i32 %d, %d ; even number 11652 11653 ; branching on frozen value 11654 %poison = add nsw i1 %k, undef ; poison 11655 %c = freeze i1 %poison 11656 br i1 %c, label %foo, label %bar ; non-deterministic branch to %foo or %bar 11657 11658 11659.. _i_call: 11660 11661'``call``' Instruction 11662^^^^^^^^^^^^^^^^^^^^^^ 11663 11664Syntax: 11665""""""" 11666 11667:: 11668 11669 <result> = [tail | musttail | notail ] call [fast-math flags] [cconv] [ret attrs] [addrspace(<num>)] 11670 <ty>|<fnty> <fnptrval>(<function args>) [fn attrs] [ operand bundles ] 11671 11672Overview: 11673""""""""" 11674 11675The '``call``' instruction represents a simple function call. 11676 11677Arguments: 11678"""""""""" 11679 11680This instruction requires several arguments: 11681 11682#. The optional ``tail`` and ``musttail`` markers indicate that the optimizers 11683 should perform tail call optimization. The ``tail`` marker is a hint that 11684 `can be ignored <CodeGenerator.html#sibcallopt>`_. The ``musttail`` marker 11685 means that the call must be tail call optimized in order for the program to 11686 be correct. The ``musttail`` marker provides these guarantees: 11687 11688 #. The call will not cause unbounded stack growth if it is part of a 11689 recursive cycle in the call graph. 11690 #. Arguments with the :ref:`inalloca <attr_inalloca>` or 11691 :ref:`preallocated <attr_preallocated>` attribute are forwarded in place. 11692 #. If the musttail call appears in a function with the ``"thunk"`` attribute 11693 and the caller and callee both have varargs, than any unprototyped 11694 arguments in register or memory are forwarded to the callee. Similarly, 11695 the return value of the callee is returned to the caller's caller, even 11696 if a void return type is in use. 11697 11698 Both markers imply that the callee does not access allocas from the caller. 11699 The ``tail`` marker additionally implies that the callee does not access 11700 varargs from the caller. Calls marked ``musttail`` must obey the following 11701 additional rules: 11702 11703 - The call must immediately precede a :ref:`ret <i_ret>` instruction, 11704 or a pointer bitcast followed by a ret instruction. 11705 - The ret instruction must return the (possibly bitcasted) value 11706 produced by the call, undef, or void. 11707 - The calling conventions of the caller and callee must match. 11708 - The callee must be varargs iff the caller is varargs. Bitcasting a 11709 non-varargs function to the appropriate varargs type is legal so 11710 long as the non-varargs prefixes obey the other rules. 11711 - The return type must not undergo automatic conversion to an `sret` pointer. 11712 11713 In addition, if the calling convention is not `swifttailcc` or `tailcc`: 11714 11715 - All ABI-impacting function attributes, such as sret, byval, inreg, 11716 returned, and inalloca, must match. 11717 - The caller and callee prototypes must match. Pointer types of parameters 11718 or return types may differ in pointee type, but not in address space. 11719 11720 On the other hand, if the calling convention is `swifttailcc` or `swiftcc`: 11721 11722 - Only these ABI-impacting attributes attributes are allowed: sret, byval, 11723 swiftself, and swiftasync. 11724 - Prototypes are not required to match. 11725 11726 Tail call optimization for calls marked ``tail`` is guaranteed to occur if 11727 the following conditions are met: 11728 11729 - Caller and callee both have the calling convention ``fastcc`` or ``tailcc``. 11730 - The call is in tail position (ret immediately follows call and ret 11731 uses value of call or is void). 11732 - Option ``-tailcallopt`` is enabled, 11733 ``llvm::GuaranteedTailCallOpt`` is ``true``, or the calling convention 11734 is ``tailcc`` 11735 - `Platform-specific constraints are 11736 met. <CodeGenerator.html#tailcallopt>`_ 11737 11738#. The optional ``notail`` marker indicates that the optimizers should not add 11739 ``tail`` or ``musttail`` markers to the call. It is used to prevent tail 11740 call optimization from being performed on the call. 11741 11742#. The optional ``fast-math flags`` marker indicates that the call has one or more 11743 :ref:`fast-math flags <fastmath>`, which are optimization hints to enable 11744 otherwise unsafe floating-point optimizations. Fast-math flags are only valid 11745 for calls that return a floating-point scalar or vector type, or an array 11746 (nested to any depth) of floating-point scalar or vector types. 11747 11748#. The optional "cconv" marker indicates which :ref:`calling 11749 convention <callingconv>` the call should use. If none is 11750 specified, the call defaults to using C calling conventions. The 11751 calling convention of the call must match the calling convention of 11752 the target function, or else the behavior is undefined. 11753#. The optional :ref:`Parameter Attributes <paramattrs>` list for return 11754 values. Only '``zeroext``', '``signext``', and '``inreg``' attributes 11755 are valid here. 11756#. The optional addrspace attribute can be used to indicate the address space 11757 of the called function. If it is not specified, the program address space 11758 from the :ref:`datalayout string<langref_datalayout>` will be used. 11759#. '``ty``': the type of the call instruction itself which is also the 11760 type of the return value. Functions that return no value are marked 11761 ``void``. 11762#. '``fnty``': shall be the signature of the function being called. The 11763 argument types must match the types implied by this signature. This 11764 type can be omitted if the function is not varargs. 11765#. '``fnptrval``': An LLVM value containing a pointer to a function to 11766 be called. In most cases, this is a direct function call, but 11767 indirect ``call``'s are just as possible, calling an arbitrary pointer 11768 to function value. 11769#. '``function args``': argument list whose types match the function 11770 signature argument types and parameter attributes. All arguments must 11771 be of :ref:`first class <t_firstclass>` type. If the function signature 11772 indicates the function accepts a variable number of arguments, the 11773 extra arguments can be specified. 11774#. The optional :ref:`function attributes <fnattrs>` list. 11775#. The optional :ref:`operand bundles <opbundles>` list. 11776 11777Semantics: 11778"""""""""" 11779 11780The '``call``' instruction is used to cause control flow to transfer to 11781a specified function, with its incoming arguments bound to the specified 11782values. Upon a '``ret``' instruction in the called function, control 11783flow continues with the instruction after the function call, and the 11784return value of the function is bound to the result argument. 11785 11786Example: 11787"""""""" 11788 11789.. code-block:: llvm 11790 11791 %retval = call i32 @test(i32 %argc) 11792 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) ; yields i32 11793 %X = tail call i32 @foo() ; yields i32 11794 %Y = tail call fastcc i32 @foo() ; yields i32 11795 call void %foo(i8 signext 97) 11796 11797 %struct.A = type { i32, i8 } 11798 %r = call %struct.A @foo() ; yields { i32, i8 } 11799 %gr = extractvalue %struct.A %r, 0 ; yields i32 11800 %gr1 = extractvalue %struct.A %r, 1 ; yields i8 11801 %Z = call void @foo() noreturn ; indicates that %foo never returns normally 11802 %ZZ = call zeroext i32 @bar() ; Return value is %zero extended 11803 11804llvm treats calls to some functions with names and arguments that match 11805the standard C99 library as being the C99 library functions, and may 11806perform optimizations or generate code for them under that assumption. 11807This is something we'd like to change in the future to provide better 11808support for freestanding environments and non-C-based languages. 11809 11810.. _i_va_arg: 11811 11812'``va_arg``' Instruction 11813^^^^^^^^^^^^^^^^^^^^^^^^ 11814 11815Syntax: 11816""""""" 11817 11818:: 11819 11820 <resultval> = va_arg <va_list*> <arglist>, <argty> 11821 11822Overview: 11823""""""""" 11824 11825The '``va_arg``' instruction is used to access arguments passed through 11826the "variable argument" area of a function call. It is used to implement 11827the ``va_arg`` macro in C. 11828 11829Arguments: 11830"""""""""" 11831 11832This instruction takes a ``va_list*`` value and the type of the 11833argument. It returns a value of the specified argument type and 11834increments the ``va_list`` to point to the next argument. The actual 11835type of ``va_list`` is target specific. 11836 11837Semantics: 11838"""""""""" 11839 11840The '``va_arg``' instruction loads an argument of the specified type 11841from the specified ``va_list`` and causes the ``va_list`` to point to 11842the next argument. For more information, see the variable argument 11843handling :ref:`Intrinsic Functions <int_varargs>`. 11844 11845It is legal for this instruction to be called in a function which does 11846not take a variable number of arguments, for example, the ``vfprintf`` 11847function. 11848 11849``va_arg`` is an LLVM instruction instead of an :ref:`intrinsic 11850function <intrinsics>` because it takes a type as an argument. 11851 11852Example: 11853"""""""" 11854 11855See the :ref:`variable argument processing <int_varargs>` section. 11856 11857Note that the code generator does not yet fully support va\_arg on many 11858targets. Also, it does not currently support va\_arg with aggregate 11859types on any target. 11860 11861.. _i_landingpad: 11862 11863'``landingpad``' Instruction 11864^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11865 11866Syntax: 11867""""""" 11868 11869:: 11870 11871 <resultval> = landingpad <resultty> <clause>+ 11872 <resultval> = landingpad <resultty> cleanup <clause>* 11873 11874 <clause> := catch <type> <value> 11875 <clause> := filter <array constant type> <array constant> 11876 11877Overview: 11878""""""""" 11879 11880The '``landingpad``' instruction is used by `LLVM's exception handling 11881system <ExceptionHandling.html#overview>`_ to specify that a basic block 11882is a landing pad --- one where the exception lands, and corresponds to the 11883code found in the ``catch`` portion of a ``try``/``catch`` sequence. It 11884defines values supplied by the :ref:`personality function <personalityfn>` upon 11885re-entry to the function. The ``resultval`` has the type ``resultty``. 11886 11887Arguments: 11888"""""""""" 11889 11890The optional 11891``cleanup`` flag indicates that the landing pad block is a cleanup. 11892 11893A ``clause`` begins with the clause type --- ``catch`` or ``filter`` --- and 11894contains the global variable representing the "type" that may be caught 11895or filtered respectively. Unlike the ``catch`` clause, the ``filter`` 11896clause takes an array constant as its argument. Use 11897"``[0 x i8**] undef``" for a filter which cannot throw. The 11898'``landingpad``' instruction must contain *at least* one ``clause`` or 11899the ``cleanup`` flag. 11900 11901Semantics: 11902"""""""""" 11903 11904The '``landingpad``' instruction defines the values which are set by the 11905:ref:`personality function <personalityfn>` upon re-entry to the function, and 11906therefore the "result type" of the ``landingpad`` instruction. As with 11907calling conventions, how the personality function results are 11908represented in LLVM IR is target specific. 11909 11910The clauses are applied in order from top to bottom. If two 11911``landingpad`` instructions are merged together through inlining, the 11912clauses from the calling function are appended to the list of clauses. 11913When the call stack is being unwound due to an exception being thrown, 11914the exception is compared against each ``clause`` in turn. If it doesn't 11915match any of the clauses, and the ``cleanup`` flag is not set, then 11916unwinding continues further up the call stack. 11917 11918The ``landingpad`` instruction has several restrictions: 11919 11920- A landing pad block is a basic block which is the unwind destination 11921 of an '``invoke``' instruction. 11922- A landing pad block must have a '``landingpad``' instruction as its 11923 first non-PHI instruction. 11924- There can be only one '``landingpad``' instruction within the landing 11925 pad block. 11926- A basic block that is not a landing pad block may not include a 11927 '``landingpad``' instruction. 11928 11929Example: 11930"""""""" 11931 11932.. code-block:: llvm 11933 11934 ;; A landing pad which can catch an integer. 11935 %res = landingpad { i8*, i32 } 11936 catch i8** @_ZTIi 11937 ;; A landing pad that is a cleanup. 11938 %res = landingpad { i8*, i32 } 11939 cleanup 11940 ;; A landing pad which can catch an integer and can only throw a double. 11941 %res = landingpad { i8*, i32 } 11942 catch i8** @_ZTIi 11943 filter [1 x i8**] [i8** @_ZTId] 11944 11945.. _i_catchpad: 11946 11947'``catchpad``' Instruction 11948^^^^^^^^^^^^^^^^^^^^^^^^^^ 11949 11950Syntax: 11951""""""" 11952 11953:: 11954 11955 <resultval> = catchpad within <catchswitch> [<args>*] 11956 11957Overview: 11958""""""""" 11959 11960The '``catchpad``' instruction is used by `LLVM's exception handling 11961system <ExceptionHandling.html#overview>`_ to specify that a basic block 11962begins a catch handler --- one where a personality routine attempts to transfer 11963control to catch an exception. 11964 11965Arguments: 11966"""""""""" 11967 11968The ``catchswitch`` operand must always be a token produced by a 11969:ref:`catchswitch <i_catchswitch>` instruction in a predecessor block. This 11970ensures that each ``catchpad`` has exactly one predecessor block, and it always 11971terminates in a ``catchswitch``. 11972 11973The ``args`` correspond to whatever information the personality routine 11974requires to know if this is an appropriate handler for the exception. Control 11975will transfer to the ``catchpad`` if this is the first appropriate handler for 11976the exception. 11977 11978The ``resultval`` has the type :ref:`token <t_token>` and is used to match the 11979``catchpad`` to corresponding :ref:`catchrets <i_catchret>` and other nested EH 11980pads. 11981 11982Semantics: 11983"""""""""" 11984 11985When the call stack is being unwound due to an exception being thrown, the 11986exception is compared against the ``args``. If it doesn't match, control will 11987not reach the ``catchpad`` instruction. The representation of ``args`` is 11988entirely target and personality function-specific. 11989 11990Like the :ref:`landingpad <i_landingpad>` instruction, the ``catchpad`` 11991instruction must be the first non-phi of its parent basic block. 11992 11993The meaning of the tokens produced and consumed by ``catchpad`` and other "pad" 11994instructions is described in the 11995`Windows exception handling documentation\ <ExceptionHandling.html#wineh>`_. 11996 11997When a ``catchpad`` has been "entered" but not yet "exited" (as 11998described in the `EH documentation\ <ExceptionHandling.html#wineh-constraints>`_), 11999it is undefined behavior to execute a :ref:`call <i_call>` or :ref:`invoke <i_invoke>` 12000that does not carry an appropriate :ref:`"funclet" bundle <ob_funclet>`. 12001 12002Example: 12003"""""""" 12004 12005.. code-block:: text 12006 12007 dispatch: 12008 %cs = catchswitch within none [label %handler0] unwind to caller 12009 ;; A catch block which can catch an integer. 12010 handler0: 12011 %tok = catchpad within %cs [i8** @_ZTIi] 12012 12013.. _i_cleanuppad: 12014 12015'``cleanuppad``' Instruction 12016^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12017 12018Syntax: 12019""""""" 12020 12021:: 12022 12023 <resultval> = cleanuppad within <parent> [<args>*] 12024 12025Overview: 12026""""""""" 12027 12028The '``cleanuppad``' instruction is used by `LLVM's exception handling 12029system <ExceptionHandling.html#overview>`_ to specify that a basic block 12030is a cleanup block --- one where a personality routine attempts to 12031transfer control to run cleanup actions. 12032The ``args`` correspond to whatever additional 12033information the :ref:`personality function <personalityfn>` requires to 12034execute the cleanup. 12035The ``resultval`` has the type :ref:`token <t_token>` and is used to 12036match the ``cleanuppad`` to corresponding :ref:`cleanuprets <i_cleanupret>`. 12037The ``parent`` argument is the token of the funclet that contains the 12038``cleanuppad`` instruction. If the ``cleanuppad`` is not inside a funclet, 12039this operand may be the token ``none``. 12040 12041Arguments: 12042"""""""""" 12043 12044The instruction takes a list of arbitrary values which are interpreted 12045by the :ref:`personality function <personalityfn>`. 12046 12047Semantics: 12048"""""""""" 12049 12050When the call stack is being unwound due to an exception being thrown, 12051the :ref:`personality function <personalityfn>` transfers control to the 12052``cleanuppad`` with the aid of the personality-specific arguments. 12053As with calling conventions, how the personality function results are 12054represented in LLVM IR is target specific. 12055 12056The ``cleanuppad`` instruction has several restrictions: 12057 12058- A cleanup block is a basic block which is the unwind destination of 12059 an exceptional instruction. 12060- A cleanup block must have a '``cleanuppad``' instruction as its 12061 first non-PHI instruction. 12062- There can be only one '``cleanuppad``' instruction within the 12063 cleanup block. 12064- A basic block that is not a cleanup block may not include a 12065 '``cleanuppad``' instruction. 12066 12067When a ``cleanuppad`` has been "entered" but not yet "exited" (as 12068described in the `EH documentation\ <ExceptionHandling.html#wineh-constraints>`_), 12069it is undefined behavior to execute a :ref:`call <i_call>` or :ref:`invoke <i_invoke>` 12070that does not carry an appropriate :ref:`"funclet" bundle <ob_funclet>`. 12071 12072Example: 12073"""""""" 12074 12075.. code-block:: text 12076 12077 %tok = cleanuppad within %cs [] 12078 12079.. _intrinsics: 12080 12081Intrinsic Functions 12082=================== 12083 12084LLVM supports the notion of an "intrinsic function". These functions 12085have well known names and semantics and are required to follow certain 12086restrictions. Overall, these intrinsics represent an extension mechanism 12087for the LLVM language that does not require changing all of the 12088transformations in LLVM when adding to the language (or the bitcode 12089reader/writer, the parser, etc...). 12090 12091Intrinsic function names must all start with an "``llvm.``" prefix. This 12092prefix is reserved in LLVM for intrinsic names; thus, function names may 12093not begin with this prefix. Intrinsic functions must always be external 12094functions: you cannot define the body of intrinsic functions. Intrinsic 12095functions may only be used in call or invoke instructions: it is illegal 12096to take the address of an intrinsic function. Additionally, because 12097intrinsic functions are part of the LLVM language, it is required if any 12098are added that they be documented here. 12099 12100Some intrinsic functions can be overloaded, i.e., the intrinsic 12101represents a family of functions that perform the same operation but on 12102different data types. Because LLVM can represent over 8 million 12103different integer types, overloading is used commonly to allow an 12104intrinsic function to operate on any integer type. One or more of the 12105argument types or the result type can be overloaded to accept any 12106integer type. Argument types may also be defined as exactly matching a 12107previous argument's type or the result type. This allows an intrinsic 12108function which accepts multiple arguments, but needs all of them to be 12109of the same type, to only be overloaded with respect to a single 12110argument or the result. 12111 12112Overloaded intrinsics will have the names of its overloaded argument 12113types encoded into its function name, each preceded by a period. Only 12114those types which are overloaded result in a name suffix. Arguments 12115whose type is matched against another type do not. For example, the 12116``llvm.ctpop`` function can take an integer of any width and returns an 12117integer of exactly the same integer width. This leads to a family of 12118functions such as ``i8 @llvm.ctpop.i8(i8 %val)`` and 12119``i29 @llvm.ctpop.i29(i29 %val)``. Only one type, the return type, is 12120overloaded, and only one type suffix is required. Because the argument's 12121type is matched against the return type, it does not require its own 12122name suffix. 12123 12124:ref:`Unnamed types <t_opaque>` are encoded as ``s_s``. Overloaded intrinsics 12125that depend on an unnamed type in one of its overloaded argument types get an 12126additional ``.<number>`` suffix. This allows differentiating intrinsics with 12127different unnamed types as arguments. (For example: 12128``llvm.ssa.copy.p0s_s.2(%42*)``) The number is tracked in the LLVM module and 12129it ensures unique names in the module. While linking together two modules, it is 12130still possible to get a name clash. In that case one of the names will be 12131changed by getting a new number. 12132 12133For target developers who are defining intrinsics for back-end code 12134generation, any intrinsic overloads based solely the distinction between 12135integer or floating point types should not be relied upon for correct 12136code generation. In such cases, the recommended approach for target 12137maintainers when defining intrinsics is to create separate integer and 12138FP intrinsics rather than rely on overloading. For example, if different 12139codegen is required for ``llvm.target.foo(<4 x i32>)`` and 12140``llvm.target.foo(<4 x float>)`` then these should be split into 12141different intrinsics. 12142 12143To learn how to add an intrinsic function, please see the `Extending 12144LLVM Guide <ExtendingLLVM.html>`_. 12145 12146.. _int_varargs: 12147 12148Variable Argument Handling Intrinsics 12149------------------------------------- 12150 12151Variable argument support is defined in LLVM with the 12152:ref:`va_arg <i_va_arg>` instruction and these three intrinsic 12153functions. These functions are related to the similarly named macros 12154defined in the ``<stdarg.h>`` header file. 12155 12156All of these functions operate on arguments that use a target-specific 12157value type "``va_list``". The LLVM assembly language reference manual 12158does not define what this type is, so all transformations should be 12159prepared to handle these functions regardless of the type used. 12160 12161This example shows how the :ref:`va_arg <i_va_arg>` instruction and the 12162variable argument handling intrinsic functions are used. 12163 12164.. code-block:: llvm 12165 12166 ; This struct is different for every platform. For most platforms, 12167 ; it is merely an i8*. 12168 %struct.va_list = type { i8* } 12169 12170 ; For Unix x86_64 platforms, va_list is the following struct: 12171 ; %struct.va_list = type { i32, i32, i8*, i8* } 12172 12173 define i32 @test(i32 %X, ...) { 12174 ; Initialize variable argument processing 12175 %ap = alloca %struct.va_list 12176 %ap2 = bitcast %struct.va_list* %ap to i8* 12177 call void @llvm.va_start(i8* %ap2) 12178 12179 ; Read a single integer argument 12180 %tmp = va_arg i8* %ap2, i32 12181 12182 ; Demonstrate usage of llvm.va_copy and llvm.va_end 12183 %aq = alloca i8* 12184 %aq2 = bitcast i8** %aq to i8* 12185 call void @llvm.va_copy(i8* %aq2, i8* %ap2) 12186 call void @llvm.va_end(i8* %aq2) 12187 12188 ; Stop processing of arguments. 12189 call void @llvm.va_end(i8* %ap2) 12190 ret i32 %tmp 12191 } 12192 12193 declare void @llvm.va_start(i8*) 12194 declare void @llvm.va_copy(i8*, i8*) 12195 declare void @llvm.va_end(i8*) 12196 12197.. _int_va_start: 12198 12199'``llvm.va_start``' Intrinsic 12200^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12201 12202Syntax: 12203""""""" 12204 12205:: 12206 12207 declare void @llvm.va_start(i8* <arglist>) 12208 12209Overview: 12210""""""""" 12211 12212The '``llvm.va_start``' intrinsic initializes ``*<arglist>`` for 12213subsequent use by ``va_arg``. 12214 12215Arguments: 12216"""""""""" 12217 12218The argument is a pointer to a ``va_list`` element to initialize. 12219 12220Semantics: 12221"""""""""" 12222 12223The '``llvm.va_start``' intrinsic works just like the ``va_start`` macro 12224available in C. In a target-dependent way, it initializes the 12225``va_list`` element to which the argument points, so that the next call 12226to ``va_arg`` will produce the first variable argument passed to the 12227function. Unlike the C ``va_start`` macro, this intrinsic does not need 12228to know the last argument of the function as the compiler can figure 12229that out. 12230 12231'``llvm.va_end``' Intrinsic 12232^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12233 12234Syntax: 12235""""""" 12236 12237:: 12238 12239 declare void @llvm.va_end(i8* <arglist>) 12240 12241Overview: 12242""""""""" 12243 12244The '``llvm.va_end``' intrinsic destroys ``*<arglist>``, which has been 12245initialized previously with ``llvm.va_start`` or ``llvm.va_copy``. 12246 12247Arguments: 12248"""""""""" 12249 12250The argument is a pointer to a ``va_list`` to destroy. 12251 12252Semantics: 12253"""""""""" 12254 12255The '``llvm.va_end``' intrinsic works just like the ``va_end`` macro 12256available in C. In a target-dependent way, it destroys the ``va_list`` 12257element to which the argument points. Calls to 12258:ref:`llvm.va_start <int_va_start>` and 12259:ref:`llvm.va_copy <int_va_copy>` must be matched exactly with calls to 12260``llvm.va_end``. 12261 12262.. _int_va_copy: 12263 12264'``llvm.va_copy``' Intrinsic 12265^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12266 12267Syntax: 12268""""""" 12269 12270:: 12271 12272 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>) 12273 12274Overview: 12275""""""""" 12276 12277The '``llvm.va_copy``' intrinsic copies the current argument position 12278from the source argument list to the destination argument list. 12279 12280Arguments: 12281"""""""""" 12282 12283The first argument is a pointer to a ``va_list`` element to initialize. 12284The second argument is a pointer to a ``va_list`` element to copy from. 12285 12286Semantics: 12287"""""""""" 12288 12289The '``llvm.va_copy``' intrinsic works just like the ``va_copy`` macro 12290available in C. In a target-dependent way, it copies the source 12291``va_list`` element into the destination ``va_list`` element. This 12292intrinsic is necessary because the `` llvm.va_start`` intrinsic may be 12293arbitrarily complex and require, for example, memory allocation. 12294 12295Accurate Garbage Collection Intrinsics 12296-------------------------------------- 12297 12298LLVM's support for `Accurate Garbage Collection <GarbageCollection.html>`_ 12299(GC) requires the frontend to generate code containing appropriate intrinsic 12300calls and select an appropriate GC strategy which knows how to lower these 12301intrinsics in a manner which is appropriate for the target collector. 12302 12303These intrinsics allow identification of :ref:`GC roots on the 12304stack <int_gcroot>`, as well as garbage collector implementations that 12305require :ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers. 12306Frontends for type-safe garbage collected languages should generate 12307these intrinsics to make use of the LLVM garbage collectors. For more 12308details, see `Garbage Collection with LLVM <GarbageCollection.html>`_. 12309 12310LLVM provides an second experimental set of intrinsics for describing garbage 12311collection safepoints in compiled code. These intrinsics are an alternative 12312to the ``llvm.gcroot`` intrinsics, but are compatible with the ones for 12313:ref:`read <int_gcread>` and :ref:`write <int_gcwrite>` barriers. The 12314differences in approach are covered in the `Garbage Collection with LLVM 12315<GarbageCollection.html>`_ documentation. The intrinsics themselves are 12316described in :doc:`Statepoints`. 12317 12318.. _int_gcroot: 12319 12320'``llvm.gcroot``' Intrinsic 12321^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12322 12323Syntax: 12324""""""" 12325 12326:: 12327 12328 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata) 12329 12330Overview: 12331""""""""" 12332 12333The '``llvm.gcroot``' intrinsic declares the existence of a GC root to 12334the code generator, and allows some metadata to be associated with it. 12335 12336Arguments: 12337"""""""""" 12338 12339The first argument specifies the address of a stack object that contains 12340the root pointer. The second pointer (which must be either a constant or 12341a global value address) contains the meta-data to be associated with the 12342root. 12343 12344Semantics: 12345"""""""""" 12346 12347At runtime, a call to this intrinsic stores a null pointer into the 12348"ptrloc" location. At compile-time, the code generator generates 12349information to allow the runtime to find the pointer at GC safe points. 12350The '``llvm.gcroot``' intrinsic may only be used in a function which 12351:ref:`specifies a GC algorithm <gc>`. 12352 12353.. _int_gcread: 12354 12355'``llvm.gcread``' Intrinsic 12356^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12357 12358Syntax: 12359""""""" 12360 12361:: 12362 12363 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr) 12364 12365Overview: 12366""""""""" 12367 12368The '``llvm.gcread``' intrinsic identifies reads of references from heap 12369locations, allowing garbage collector implementations that require read 12370barriers. 12371 12372Arguments: 12373"""""""""" 12374 12375The second argument is the address to read from, which should be an 12376address allocated from the garbage collector. The first object is a 12377pointer to the start of the referenced object, if needed by the language 12378runtime (otherwise null). 12379 12380Semantics: 12381"""""""""" 12382 12383The '``llvm.gcread``' intrinsic has the same semantics as a load 12384instruction, but may be replaced with substantially more complex code by 12385the garbage collector runtime, as needed. The '``llvm.gcread``' 12386intrinsic may only be used in a function which :ref:`specifies a GC 12387algorithm <gc>`. 12388 12389.. _int_gcwrite: 12390 12391'``llvm.gcwrite``' Intrinsic 12392^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12393 12394Syntax: 12395""""""" 12396 12397:: 12398 12399 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2) 12400 12401Overview: 12402""""""""" 12403 12404The '``llvm.gcwrite``' intrinsic identifies writes of references to heap 12405locations, allowing garbage collector implementations that require write 12406barriers (such as generational or reference counting collectors). 12407 12408Arguments: 12409"""""""""" 12410 12411The first argument is the reference to store, the second is the start of 12412the object to store it to, and the third is the address of the field of 12413Obj to store to. If the runtime does not require a pointer to the 12414object, Obj may be null. 12415 12416Semantics: 12417"""""""""" 12418 12419The '``llvm.gcwrite``' intrinsic has the same semantics as a store 12420instruction, but may be replaced with substantially more complex code by 12421the garbage collector runtime, as needed. The '``llvm.gcwrite``' 12422intrinsic may only be used in a function which :ref:`specifies a GC 12423algorithm <gc>`. 12424 12425 12426.. _gc_statepoint: 12427 12428'llvm.experimental.gc.statepoint' Intrinsic 12429^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12430 12431Syntax: 12432""""""" 12433 12434:: 12435 12436 declare token 12437 @llvm.experimental.gc.statepoint(i64 <id>, i32 <num patch bytes>, 12438 func_type* elementtype(func_type) <target>, 12439 i64 <#call args>, i64 <flags>, 12440 ... (call parameters), 12441 i64 0, i64 0) 12442 12443Overview: 12444""""""""" 12445 12446The statepoint intrinsic represents a call which is parse-able by the 12447runtime. 12448 12449Operands: 12450""""""""" 12451 12452The 'id' operand is a constant integer that is reported as the ID 12453field in the generated stackmap. LLVM does not interpret this 12454parameter in any way and its meaning is up to the statepoint user to 12455decide. Note that LLVM is free to duplicate code containing 12456statepoint calls, and this may transform IR that had a unique 'id' per 12457lexical call to statepoint to IR that does not. 12458 12459If 'num patch bytes' is non-zero then the call instruction 12460corresponding to the statepoint is not emitted and LLVM emits 'num 12461patch bytes' bytes of nops in its place. LLVM will emit code to 12462prepare the function arguments and retrieve the function return value 12463in accordance to the calling convention; the former before the nop 12464sequence and the latter after the nop sequence. It is expected that 12465the user will patch over the 'num patch bytes' bytes of nops with a 12466calling sequence specific to their runtime before executing the 12467generated machine code. There are no guarantees with respect to the 12468alignment of the nop sequence. Unlike :doc:`StackMaps` statepoints do 12469not have a concept of shadow bytes. Note that semantically the 12470statepoint still represents a call or invoke to 'target', and the nop 12471sequence after patching is expected to represent an operation 12472equivalent to a call or invoke to 'target'. 12473 12474The 'target' operand is the function actually being called. The operand 12475must have an :ref:`elementtype <attr_elementtype>` attribute specifying 12476the function type of the target. The target can be specified as either 12477a symbolic LLVM function, or as an arbitrary Value of pointer type. Note 12478that the function type must match the signature of the callee and the 12479types of the 'call parameters' arguments. 12480 12481The '#call args' operand is the number of arguments to the actual 12482call. It must exactly match the number of arguments passed in the 12483'call parameters' variable length section. 12484 12485The 'flags' operand is used to specify extra information about the 12486statepoint. This is currently only used to mark certain statepoints 12487as GC transitions. This operand is a 64-bit integer with the following 12488layout, where bit 0 is the least significant bit: 12489 12490 +-------+---------------------------------------------------+ 12491 | Bit # | Usage | 12492 +=======+===================================================+ 12493 | 0 | Set if the statepoint is a GC transition, cleared | 12494 | | otherwise. | 12495 +-------+---------------------------------------------------+ 12496 | 1-63 | Reserved for future use; must be cleared. | 12497 +-------+---------------------------------------------------+ 12498 12499The 'call parameters' arguments are simply the arguments which need to 12500be passed to the call target. They will be lowered according to the 12501specified calling convention and otherwise handled like a normal call 12502instruction. The number of arguments must exactly match what is 12503specified in '# call args'. The types must match the signature of 12504'target'. 12505 12506The 'call parameter' attributes must be followed by two 'i64 0' constants. 12507These were originally the length prefixes for 'gc transition parameter' and 12508'deopt parameter' arguments, but the role of these parameter sets have been 12509entirely replaced with the corresponding operand bundles. In a future 12510revision, these now redundant arguments will be removed. 12511 12512Semantics: 12513"""""""""" 12514 12515A statepoint is assumed to read and write all memory. As a result, 12516memory operations can not be reordered past a statepoint. It is 12517illegal to mark a statepoint as being either 'readonly' or 'readnone'. 12518 12519Note that legal IR can not perform any memory operation on a 'gc 12520pointer' argument of the statepoint in a location statically reachable 12521from the statepoint. Instead, the explicitly relocated value (from a 12522``gc.relocate``) must be used. 12523 12524'llvm.experimental.gc.result' Intrinsic 12525^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12526 12527Syntax: 12528""""""" 12529 12530:: 12531 12532 declare type* 12533 @llvm.experimental.gc.result(token %statepoint_token) 12534 12535Overview: 12536""""""""" 12537 12538``gc.result`` extracts the result of the original call instruction 12539which was replaced by the ``gc.statepoint``. The ``gc.result`` 12540intrinsic is actually a family of three intrinsics due to an 12541implementation limitation. Other than the type of the return value, 12542the semantics are the same. 12543 12544Operands: 12545""""""""" 12546 12547The first and only argument is the ``gc.statepoint`` which starts 12548the safepoint sequence of which this ``gc.result`` is a part. 12549Despite the typing of this as a generic token, *only* the value defined 12550by a ``gc.statepoint`` is legal here. 12551 12552Semantics: 12553"""""""""" 12554 12555The ``gc.result`` represents the return value of the call target of 12556the ``statepoint``. The type of the ``gc.result`` must exactly match 12557the type of the target. If the call target returns void, there will 12558be no ``gc.result``. 12559 12560A ``gc.result`` is modeled as a 'readnone' pure function. It has no 12561side effects since it is just a projection of the return value of the 12562previous call represented by the ``gc.statepoint``. 12563 12564'llvm.experimental.gc.relocate' Intrinsic 12565^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12566 12567Syntax: 12568""""""" 12569 12570:: 12571 12572 declare <pointer type> 12573 @llvm.experimental.gc.relocate(token %statepoint_token, 12574 i32 %base_offset, 12575 i32 %pointer_offset) 12576 12577Overview: 12578""""""""" 12579 12580A ``gc.relocate`` returns the potentially relocated value of a pointer 12581at the safepoint. 12582 12583Operands: 12584""""""""" 12585 12586The first argument is the ``gc.statepoint`` which starts the 12587safepoint sequence of which this ``gc.relocation`` is a part. 12588Despite the typing of this as a generic token, *only* the value defined 12589by a ``gc.statepoint`` is legal here. 12590 12591The second and third arguments are both indices into operands of the 12592corresponding statepoint's :ref:`gc-live <ob_gc_live>` operand bundle. 12593 12594The second argument is an index which specifies the allocation for the pointer 12595being relocated. The associated value must be within the object with which the 12596pointer being relocated is associated. The optimizer is free to change *which* 12597interior derived pointer is reported, provided that it does not replace an 12598actual base pointer with another interior derived pointer. Collectors are 12599allowed to rely on the base pointer operand remaining an actual base pointer if 12600so constructed. 12601 12602The third argument is an index which specify the (potentially) derived pointer 12603being relocated. It is legal for this index to be the same as the second 12604argument if-and-only-if a base pointer is being relocated. 12605 12606Semantics: 12607"""""""""" 12608 12609The return value of ``gc.relocate`` is the potentially relocated value 12610of the pointer specified by its arguments. It is unspecified how the 12611value of the returned pointer relates to the argument to the 12612``gc.statepoint`` other than that a) it points to the same source 12613language object with the same offset, and b) the 'based-on' 12614relationship of the newly relocated pointers is a projection of the 12615unrelocated pointers. In particular, the integer value of the pointer 12616returned is unspecified. 12617 12618A ``gc.relocate`` is modeled as a ``readnone`` pure function. It has no 12619side effects since it is just a way to extract information about work 12620done during the actual call modeled by the ``gc.statepoint``. 12621 12622.. _gc.get.pointer.base: 12623 12624'llvm.experimental.gc.get.pointer.base' Intrinsic 12625^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12626 12627Syntax: 12628""""""" 12629 12630:: 12631 12632 declare <pointer type> 12633 @llvm.experimental.gc.get.pointer.base( 12634 <pointer type> readnone nocapture %derived_ptr) 12635 nounwind readnone willreturn 12636 12637Overview: 12638""""""""" 12639 12640``gc.get.pointer.base`` for a derived pointer returns its base pointer. 12641 12642Operands: 12643""""""""" 12644 12645The only argument is a pointer which is based on some object with 12646an unknown offset from the base of said object. 12647 12648Semantics: 12649"""""""""" 12650 12651This intrinsic is used in the abstract machine model for GC to represent 12652the base pointer for an arbitrary derived pointer. 12653 12654This intrinsic is inlined by the :ref:`RewriteStatepointsForGC` pass by 12655replacing all uses of this callsite with the offset of a derived pointer from 12656its base pointer value. The replacement is done as part of the lowering to the 12657explicit statepoint model. 12658 12659The return pointer type must be the same as the type of the parameter. 12660 12661 12662'llvm.experimental.gc.get.pointer.offset' Intrinsic 12663^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12664 12665Syntax: 12666""""""" 12667 12668:: 12669 12670 declare i64 12671 @llvm.experimental.gc.get.pointer.offset( 12672 <pointer type> readnone nocapture %derived_ptr) 12673 nounwind readnone willreturn 12674 12675Overview: 12676""""""""" 12677 12678``gc.get.pointer.offset`` for a derived pointer returns the offset from its 12679base pointer. 12680 12681Operands: 12682""""""""" 12683 12684The only argument is a pointer which is based on some object with 12685an unknown offset from the base of said object. 12686 12687Semantics: 12688"""""""""" 12689 12690This intrinsic is used in the abstract machine model for GC to represent 12691the offset of an arbitrary derived pointer from its base pointer. 12692 12693This intrinsic is inlined by the :ref:`RewriteStatepointsForGC` pass by 12694replacing all uses of this callsite with the offset of a derived pointer from 12695its base pointer value. The replacement is done as part of the lowering to the 12696explicit statepoint model. 12697 12698Basically this call calculates difference between the derived pointer and its 12699base pointer (see :ref:`gc.get.pointer.base`) both ptrtoint casted. But 12700this cast done outside the :ref:`RewriteStatepointsForGC` pass could result 12701in the pointers lost for further lowering from the abstract model to the 12702explicit physical one. 12703 12704Code Generator Intrinsics 12705------------------------- 12706 12707These intrinsics are provided by LLVM to expose special features that 12708may only be implemented with code generator support. 12709 12710'``llvm.returnaddress``' Intrinsic 12711^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12712 12713Syntax: 12714""""""" 12715 12716:: 12717 12718 declare i8* @llvm.returnaddress(i32 <level>) 12719 12720Overview: 12721""""""""" 12722 12723The '``llvm.returnaddress``' intrinsic attempts to compute a 12724target-specific value indicating the return address of the current 12725function or one of its callers. 12726 12727Arguments: 12728"""""""""" 12729 12730The argument to this intrinsic indicates which function to return the 12731address for. Zero indicates the calling function, one indicates its 12732caller, etc. The argument is **required** to be a constant integer 12733value. 12734 12735Semantics: 12736"""""""""" 12737 12738The '``llvm.returnaddress``' intrinsic either returns a pointer 12739indicating the return address of the specified call frame, or zero if it 12740cannot be identified. The value returned by this intrinsic is likely to 12741be incorrect or 0 for arguments other than zero, so it should only be 12742used for debugging purposes. 12743 12744Note that calling this intrinsic does not prevent function inlining or 12745other aggressive transformations, so the value returned may not be that 12746of the obvious source-language caller. 12747 12748'``llvm.addressofreturnaddress``' Intrinsic 12749^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12750 12751Syntax: 12752""""""" 12753 12754:: 12755 12756 declare i8* @llvm.addressofreturnaddress() 12757 12758Overview: 12759""""""""" 12760 12761The '``llvm.addressofreturnaddress``' intrinsic returns a target-specific 12762pointer to the place in the stack frame where the return address of the 12763current function is stored. 12764 12765Semantics: 12766"""""""""" 12767 12768Note that calling this intrinsic does not prevent function inlining or 12769other aggressive transformations, so the value returned may not be that 12770of the obvious source-language caller. 12771 12772This intrinsic is only implemented for x86 and aarch64. 12773 12774'``llvm.sponentry``' Intrinsic 12775^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12776 12777Syntax: 12778""""""" 12779 12780:: 12781 12782 declare i8* @llvm.sponentry() 12783 12784Overview: 12785""""""""" 12786 12787The '``llvm.sponentry``' intrinsic returns the stack pointer value at 12788the entry of the current function calling this intrinsic. 12789 12790Semantics: 12791"""""""""" 12792 12793Note this intrinsic is only verified on AArch64 and ARM. 12794 12795'``llvm.frameaddress``' Intrinsic 12796^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12797 12798Syntax: 12799""""""" 12800 12801:: 12802 12803 declare i8* @llvm.frameaddress(i32 <level>) 12804 12805Overview: 12806""""""""" 12807 12808The '``llvm.frameaddress``' intrinsic attempts to return the 12809target-specific frame pointer value for the specified stack frame. 12810 12811Arguments: 12812"""""""""" 12813 12814The argument to this intrinsic indicates which function to return the 12815frame pointer for. Zero indicates the calling function, one indicates 12816its caller, etc. The argument is **required** to be a constant integer 12817value. 12818 12819Semantics: 12820"""""""""" 12821 12822The '``llvm.frameaddress``' intrinsic either returns a pointer 12823indicating the frame address of the specified call frame, or zero if it 12824cannot be identified. The value returned by this intrinsic is likely to 12825be incorrect or 0 for arguments other than zero, so it should only be 12826used for debugging purposes. 12827 12828Note that calling this intrinsic does not prevent function inlining or 12829other aggressive transformations, so the value returned may not be that 12830of the obvious source-language caller. 12831 12832'``llvm.swift.async.context.addr``' Intrinsic 12833^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12834 12835Syntax: 12836""""""" 12837 12838:: 12839 12840 declare i8** @llvm.swift.async.context.addr() 12841 12842Overview: 12843""""""""" 12844 12845The '``llvm.swift.async.context.addr``' intrinsic returns a pointer to 12846the part of the extended frame record containing the asynchronous 12847context of a Swift execution. 12848 12849Semantics: 12850"""""""""" 12851 12852If the caller has a ``swiftasync`` parameter, that argument will initially 12853be stored at the returned address. If not, it will be initialized to null. 12854 12855'``llvm.localescape``' and '``llvm.localrecover``' Intrinsics 12856^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12857 12858Syntax: 12859""""""" 12860 12861:: 12862 12863 declare void @llvm.localescape(...) 12864 declare i8* @llvm.localrecover(i8* %func, i8* %fp, i32 %idx) 12865 12866Overview: 12867""""""""" 12868 12869The '``llvm.localescape``' intrinsic escapes offsets of a collection of static 12870allocas, and the '``llvm.localrecover``' intrinsic applies those offsets to a 12871live frame pointer to recover the address of the allocation. The offset is 12872computed during frame layout of the caller of ``llvm.localescape``. 12873 12874Arguments: 12875"""""""""" 12876 12877All arguments to '``llvm.localescape``' must be pointers to static allocas or 12878casts of static allocas. Each function can only call '``llvm.localescape``' 12879once, and it can only do so from the entry block. 12880 12881The ``func`` argument to '``llvm.localrecover``' must be a constant 12882bitcasted pointer to a function defined in the current module. The code 12883generator cannot determine the frame allocation offset of functions defined in 12884other modules. 12885 12886The ``fp`` argument to '``llvm.localrecover``' must be a frame pointer of a 12887call frame that is currently live. The return value of '``llvm.localaddress``' 12888is one way to produce such a value, but various runtimes also expose a suitable 12889pointer in platform-specific ways. 12890 12891The ``idx`` argument to '``llvm.localrecover``' indicates which alloca passed to 12892'``llvm.localescape``' to recover. It is zero-indexed. 12893 12894Semantics: 12895"""""""""" 12896 12897These intrinsics allow a group of functions to share access to a set of local 12898stack allocations of a one parent function. The parent function may call the 12899'``llvm.localescape``' intrinsic once from the function entry block, and the 12900child functions can use '``llvm.localrecover``' to access the escaped allocas. 12901The '``llvm.localescape``' intrinsic blocks inlining, as inlining changes where 12902the escaped allocas are allocated, which would break attempts to use 12903'``llvm.localrecover``'. 12904 12905'``llvm.seh.try.begin``' and '``llvm.seh.try.end``' Intrinsics 12906^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12907 12908Syntax: 12909""""""" 12910 12911:: 12912 12913 declare void @llvm.seh.try.begin() 12914 declare void @llvm.seh.try.end() 12915 12916Overview: 12917""""""""" 12918 12919The '``llvm.seh.try.begin``' and '``llvm.seh.try.end``' intrinsics mark 12920the boundary of a _try region for Windows SEH Asynchrous Exception Handling. 12921 12922Semantics: 12923"""""""""" 12924 12925When a C-function is compiled with Windows SEH Asynchrous Exception option, 12926-feh_asynch (aka MSVC -EHa), these two intrinsics are injected to mark _try 12927boundary and to prevent potential exceptions from being moved across boundary. 12928Any set of operations can then be confined to the region by reading their leaf 12929inputs via volatile loads and writing their root outputs via volatile stores. 12930 12931'``llvm.seh.scope.begin``' and '``llvm.seh.scope.end``' Intrinsics 12932^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12933 12934Syntax: 12935""""""" 12936 12937:: 12938 12939 declare void @llvm.seh.scope.begin() 12940 declare void @llvm.seh.scope.end() 12941 12942Overview: 12943""""""""" 12944 12945The '``llvm.seh.scope.begin``' and '``llvm.seh.scope.end``' intrinsics mark 12946the boundary of a CPP object lifetime for Windows SEH Asynchrous Exception 12947Handling (MSVC option -EHa). 12948 12949Semantics: 12950"""""""""" 12951 12952LLVM's ordinary exception-handling representation associates EH cleanups and 12953handlers only with ``invoke``s, which normally correspond only to call sites. To 12954support arbitrary faulting instructions, it must be possible to recover the current 12955EH scope for any instruction. Turning every operation in LLVM that could fault 12956into an ``invoke`` of a new, potentially-throwing intrinsic would require adding a 12957large number of intrinsics, impede optimization of those operations, and make 12958compilation slower by introducing many extra basic blocks. These intrinsics can 12959be used instead to mark the region protected by a cleanup, such as for a local 12960C++ object with a non-trivial destructor. ``llvm.seh.scope.begin`` is used to mark 12961the start of the region; it is always called with ``invoke``, with the unwind block 12962being the desired unwind destination for any potentially-throwing instructions 12963within the region. `llvm.seh.scope.end` is used to mark when the scope ends 12964and the EH cleanup is no longer required (e.g. because the destructor is being 12965called). 12966 12967.. _int_read_register: 12968.. _int_read_volatile_register: 12969.. _int_write_register: 12970 12971'``llvm.read_register``', '``llvm.read_volatile_register``', and '``llvm.write_register``' Intrinsics 12972^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 12973 12974Syntax: 12975""""""" 12976 12977:: 12978 12979 declare i32 @llvm.read_register.i32(metadata) 12980 declare i64 @llvm.read_register.i64(metadata) 12981 declare i32 @llvm.read_volatile_register.i32(metadata) 12982 declare i64 @llvm.read_volatile_register.i64(metadata) 12983 declare void @llvm.write_register.i32(metadata, i32 @value) 12984 declare void @llvm.write_register.i64(metadata, i64 @value) 12985 !0 = !{!"sp\00"} 12986 12987Overview: 12988""""""""" 12989 12990The '``llvm.read_register``', '``llvm.read_volatile_register``', and 12991'``llvm.write_register``' intrinsics provide access to the named register. 12992The register must be valid on the architecture being compiled to. The type 12993needs to be compatible with the register being read. 12994 12995Semantics: 12996"""""""""" 12997 12998The '``llvm.read_register``' and '``llvm.read_volatile_register``' intrinsics 12999return the current value of the register, where possible. The 13000'``llvm.write_register``' intrinsic sets the current value of the register, 13001where possible. 13002 13003A call to '``llvm.read_volatile_register``' is assumed to have side-effects 13004and possibly return a different value each time (e.g. for a timer register). 13005 13006This is useful to implement named register global variables that need 13007to always be mapped to a specific register, as is common practice on 13008bare-metal programs including OS kernels. 13009 13010The compiler doesn't check for register availability or use of the used 13011register in surrounding code, including inline assembly. Because of that, 13012allocatable registers are not supported. 13013 13014Warning: So far it only works with the stack pointer on selected 13015architectures (ARM, AArch64, PowerPC and x86_64). Significant amount of 13016work is needed to support other registers and even more so, allocatable 13017registers. 13018 13019.. _int_stacksave: 13020 13021'``llvm.stacksave``' Intrinsic 13022^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13023 13024Syntax: 13025""""""" 13026 13027:: 13028 13029 declare i8* @llvm.stacksave() 13030 13031Overview: 13032""""""""" 13033 13034The '``llvm.stacksave``' intrinsic is used to remember the current state 13035of the function stack, for use with 13036:ref:`llvm.stackrestore <int_stackrestore>`. This is useful for 13037implementing language features like scoped automatic variable sized 13038arrays in C99. 13039 13040Semantics: 13041"""""""""" 13042 13043This intrinsic returns an opaque pointer value that can be passed to 13044:ref:`llvm.stackrestore <int_stackrestore>`. When an 13045``llvm.stackrestore`` intrinsic is executed with a value saved from 13046``llvm.stacksave``, it effectively restores the state of the stack to 13047the state it was in when the ``llvm.stacksave`` intrinsic executed. In 13048practice, this pops any :ref:`alloca <i_alloca>` blocks from the stack that 13049were allocated after the ``llvm.stacksave`` was executed. 13050 13051.. _int_stackrestore: 13052 13053'``llvm.stackrestore``' Intrinsic 13054^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13055 13056Syntax: 13057""""""" 13058 13059:: 13060 13061 declare void @llvm.stackrestore(i8* %ptr) 13062 13063Overview: 13064""""""""" 13065 13066The '``llvm.stackrestore``' intrinsic is used to restore the state of 13067the function stack to the state it was in when the corresponding 13068:ref:`llvm.stacksave <int_stacksave>` intrinsic executed. This is 13069useful for implementing language features like scoped automatic variable 13070sized arrays in C99. 13071 13072Semantics: 13073"""""""""" 13074 13075See the description for :ref:`llvm.stacksave <int_stacksave>`. 13076 13077.. _int_get_dynamic_area_offset: 13078 13079'``llvm.get.dynamic.area.offset``' Intrinsic 13080^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13081 13082Syntax: 13083""""""" 13084 13085:: 13086 13087 declare i32 @llvm.get.dynamic.area.offset.i32() 13088 declare i64 @llvm.get.dynamic.area.offset.i64() 13089 13090Overview: 13091""""""""" 13092 13093 The '``llvm.get.dynamic.area.offset.*``' intrinsic family is used to 13094 get the offset from native stack pointer to the address of the most 13095 recent dynamic alloca on the caller's stack. These intrinsics are 13096 intended for use in combination with 13097 :ref:`llvm.stacksave <int_stacksave>` to get a 13098 pointer to the most recent dynamic alloca. This is useful, for example, 13099 for AddressSanitizer's stack unpoisoning routines. 13100 13101Semantics: 13102"""""""""" 13103 13104 These intrinsics return a non-negative integer value that can be used to 13105 get the address of the most recent dynamic alloca, allocated by :ref:`alloca <i_alloca>` 13106 on the caller's stack. In particular, for targets where stack grows downwards, 13107 adding this offset to the native stack pointer would get the address of the most 13108 recent dynamic alloca. For targets where stack grows upwards, the situation is a bit more 13109 complicated, because subtracting this value from stack pointer would get the address 13110 one past the end of the most recent dynamic alloca. 13111 13112 Although for most targets `llvm.get.dynamic.area.offset <int_get_dynamic_area_offset>` 13113 returns just a zero, for others, such as PowerPC and PowerPC64, it returns a 13114 compile-time-known constant value. 13115 13116 The return value type of :ref:`llvm.get.dynamic.area.offset <int_get_dynamic_area_offset>` 13117 must match the target's default address space's (address space 0) pointer type. 13118 13119'``llvm.prefetch``' Intrinsic 13120^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13121 13122Syntax: 13123""""""" 13124 13125:: 13126 13127 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>) 13128 13129Overview: 13130""""""""" 13131 13132The '``llvm.prefetch``' intrinsic is a hint to the code generator to 13133insert a prefetch instruction if supported; otherwise, it is a noop. 13134Prefetches have no effect on the behavior of the program but can change 13135its performance characteristics. 13136 13137Arguments: 13138"""""""""" 13139 13140``address`` is the address to be prefetched, ``rw`` is the specifier 13141determining if the fetch should be for a read (0) or write (1), and 13142``locality`` is a temporal locality specifier ranging from (0) - no 13143locality, to (3) - extremely local keep in cache. The ``cache type`` 13144specifies whether the prefetch is performed on the data (1) or 13145instruction (0) cache. The ``rw``, ``locality`` and ``cache type`` 13146arguments must be constant integers. 13147 13148Semantics: 13149"""""""""" 13150 13151This intrinsic does not modify the behavior of the program. In 13152particular, prefetches cannot trap and do not produce a value. On 13153targets that support this intrinsic, the prefetch can provide hints to 13154the processor cache for better performance. 13155 13156'``llvm.pcmarker``' Intrinsic 13157^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13158 13159Syntax: 13160""""""" 13161 13162:: 13163 13164 declare void @llvm.pcmarker(i32 <id>) 13165 13166Overview: 13167""""""""" 13168 13169The '``llvm.pcmarker``' intrinsic is a method to export a Program 13170Counter (PC) in a region of code to simulators and other tools. The 13171method is target specific, but it is expected that the marker will use 13172exported symbols to transmit the PC of the marker. The marker makes no 13173guarantees that it will remain with any specific instruction after 13174optimizations. It is possible that the presence of a marker will inhibit 13175optimizations. The intended use is to be inserted after optimizations to 13176allow correlations of simulation runs. 13177 13178Arguments: 13179"""""""""" 13180 13181``id`` is a numerical id identifying the marker. 13182 13183Semantics: 13184"""""""""" 13185 13186This intrinsic does not modify the behavior of the program. Backends 13187that do not support this intrinsic may ignore it. 13188 13189'``llvm.readcyclecounter``' Intrinsic 13190^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13191 13192Syntax: 13193""""""" 13194 13195:: 13196 13197 declare i64 @llvm.readcyclecounter() 13198 13199Overview: 13200""""""""" 13201 13202The '``llvm.readcyclecounter``' intrinsic provides access to the cycle 13203counter register (or similar low latency, high accuracy clocks) on those 13204targets that support it. On X86, it should map to RDTSC. On Alpha, it 13205should map to RPCC. As the backing counters overflow quickly (on the 13206order of 9 seconds on alpha), this should only be used for small 13207timings. 13208 13209Semantics: 13210"""""""""" 13211 13212When directly supported, reading the cycle counter should not modify any 13213memory. Implementations are allowed to either return an application 13214specific value or a system wide value. On backends without support, this 13215is lowered to a constant 0. 13216 13217Note that runtime support may be conditional on the privilege-level code is 13218running at and the host platform. 13219 13220'``llvm.clear_cache``' Intrinsic 13221^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13222 13223Syntax: 13224""""""" 13225 13226:: 13227 13228 declare void @llvm.clear_cache(i8*, i8*) 13229 13230Overview: 13231""""""""" 13232 13233The '``llvm.clear_cache``' intrinsic ensures visibility of modifications 13234in the specified range to the execution unit of the processor. On 13235targets with non-unified instruction and data cache, the implementation 13236flushes the instruction cache. 13237 13238Semantics: 13239"""""""""" 13240 13241On platforms with coherent instruction and data caches (e.g. x86), this 13242intrinsic is a nop. On platforms with non-coherent instruction and data 13243cache (e.g. ARM, MIPS), the intrinsic is lowered either to appropriate 13244instructions or a system call, if cache flushing requires special 13245privileges. 13246 13247The default behavior is to emit a call to ``__clear_cache`` from the run 13248time library. 13249 13250This intrinsic does *not* empty the instruction pipeline. Modifications 13251of the current function are outside the scope of the intrinsic. 13252 13253'``llvm.instrprof.increment``' Intrinsic 13254^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13255 13256Syntax: 13257""""""" 13258 13259:: 13260 13261 declare void @llvm.instrprof.increment(i8* <name>, i64 <hash>, 13262 i32 <num-counters>, i32 <index>) 13263 13264Overview: 13265""""""""" 13266 13267The '``llvm.instrprof.increment``' intrinsic can be emitted by a 13268frontend for use with instrumentation based profiling. These will be 13269lowered by the ``-instrprof`` pass to generate execution counts of a 13270program at runtime. 13271 13272Arguments: 13273"""""""""" 13274 13275The first argument is a pointer to a global variable containing the 13276name of the entity being instrumented. This should generally be the 13277(mangled) function name for a set of counters. 13278 13279The second argument is a hash value that can be used by the consumer 13280of the profile data to detect changes to the instrumented source, and 13281the third is the number of counters associated with ``name``. It is an 13282error if ``hash`` or ``num-counters`` differ between two instances of 13283``instrprof.increment`` that refer to the same name. 13284 13285The last argument refers to which of the counters for ``name`` should 13286be incremented. It should be a value between 0 and ``num-counters``. 13287 13288Semantics: 13289"""""""""" 13290 13291This intrinsic represents an increment of a profiling counter. It will 13292cause the ``-instrprof`` pass to generate the appropriate data 13293structures and the code to increment the appropriate value, in a 13294format that can be written out by a compiler runtime and consumed via 13295the ``llvm-profdata`` tool. 13296 13297'``llvm.instrprof.increment.step``' Intrinsic 13298^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13299 13300Syntax: 13301""""""" 13302 13303:: 13304 13305 declare void @llvm.instrprof.increment.step(i8* <name>, i64 <hash>, 13306 i32 <num-counters>, 13307 i32 <index>, i64 <step>) 13308 13309Overview: 13310""""""""" 13311 13312The '``llvm.instrprof.increment.step``' intrinsic is an extension to 13313the '``llvm.instrprof.increment``' intrinsic with an additional fifth 13314argument to specify the step of the increment. 13315 13316Arguments: 13317"""""""""" 13318The first four arguments are the same as '``llvm.instrprof.increment``' 13319intrinsic. 13320 13321The last argument specifies the value of the increment of the counter variable. 13322 13323Semantics: 13324"""""""""" 13325See description of '``llvm.instrprof.increment``' intrinsic. 13326 13327'``llvm.instrprof.cover``' Intrinsic 13328^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13329 13330Syntax: 13331""""""" 13332 13333:: 13334 13335 declare void @llvm.instrprof.cover(i8* <name>, i64 <hash>, 13336 i32 <num-counters>, i32 <index>) 13337 13338Overview: 13339""""""""" 13340 13341The '``llvm.instrprof.cover``' intrinsic is used to implement coverage 13342instrumentation. 13343 13344Arguments: 13345"""""""""" 13346The arguments are the same as the first four arguments of 13347'``llvm.instrprof.increment``'. 13348 13349Semantics: 13350"""""""""" 13351Similar to the '``llvm.instrprof.increment``' intrinsic, but it stores zero to 13352the profiling variable to signify that the function has been covered. We store 13353zero because this is more efficient on some targets. 13354 13355'``llvm.instrprof.value.profile``' Intrinsic 13356^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13357 13358Syntax: 13359""""""" 13360 13361:: 13362 13363 declare void @llvm.instrprof.value.profile(i8* <name>, i64 <hash>, 13364 i64 <value>, i32 <value_kind>, 13365 i32 <index>) 13366 13367Overview: 13368""""""""" 13369 13370The '``llvm.instrprof.value.profile``' intrinsic can be emitted by a 13371frontend for use with instrumentation based profiling. This will be 13372lowered by the ``-instrprof`` pass to find out the target values, 13373instrumented expressions take in a program at runtime. 13374 13375Arguments: 13376"""""""""" 13377 13378The first argument is a pointer to a global variable containing the 13379name of the entity being instrumented. ``name`` should generally be the 13380(mangled) function name for a set of counters. 13381 13382The second argument is a hash value that can be used by the consumer 13383of the profile data to detect changes to the instrumented source. It 13384is an error if ``hash`` differs between two instances of 13385``llvm.instrprof.*`` that refer to the same name. 13386 13387The third argument is the value of the expression being profiled. The profiled 13388expression's value should be representable as an unsigned 64-bit value. The 13389fourth argument represents the kind of value profiling that is being done. The 13390supported value profiling kinds are enumerated through the 13391``InstrProfValueKind`` type declared in the 13392``<include/llvm/ProfileData/InstrProf.h>`` header file. The last argument is the 13393index of the instrumented expression within ``name``. It should be >= 0. 13394 13395Semantics: 13396"""""""""" 13397 13398This intrinsic represents the point where a call to a runtime routine 13399should be inserted for value profiling of target expressions. ``-instrprof`` 13400pass will generate the appropriate data structures and replace the 13401``llvm.instrprof.value.profile`` intrinsic with the call to the profile 13402runtime library with proper arguments. 13403 13404'``llvm.thread.pointer``' Intrinsic 13405^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13406 13407Syntax: 13408""""""" 13409 13410:: 13411 13412 declare i8* @llvm.thread.pointer() 13413 13414Overview: 13415""""""""" 13416 13417The '``llvm.thread.pointer``' intrinsic returns the value of the thread 13418pointer. 13419 13420Semantics: 13421"""""""""" 13422 13423The '``llvm.thread.pointer``' intrinsic returns a pointer to the TLS area 13424for the current thread. The exact semantics of this value are target 13425specific: it may point to the start of TLS area, to the end, or somewhere 13426in the middle. Depending on the target, this intrinsic may read a register, 13427call a helper function, read from an alternate memory space, or perform 13428other operations necessary to locate the TLS area. Not all targets support 13429this intrinsic. 13430 13431'``llvm.call.preallocated.setup``' Intrinsic 13432^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13433 13434Syntax: 13435""""""" 13436 13437:: 13438 13439 declare token @llvm.call.preallocated.setup(i32 %num_args) 13440 13441Overview: 13442""""""""" 13443 13444The '``llvm.call.preallocated.setup``' intrinsic returns a token which can 13445be used with a call's ``"preallocated"`` operand bundle to indicate that 13446certain arguments are allocated and initialized before the call. 13447 13448Semantics: 13449"""""""""" 13450 13451The '``llvm.call.preallocated.setup``' intrinsic returns a token which is 13452associated with at most one call. The token can be passed to 13453'``@llvm.call.preallocated.arg``' to get a pointer to get that 13454corresponding argument. The token must be the parameter to a 13455``"preallocated"`` operand bundle for the corresponding call. 13456 13457Nested calls to '``llvm.call.preallocated.setup``' are allowed, but must 13458be properly nested. e.g. 13459 13460:: code-block:: llvm 13461 13462 %t1 = call token @llvm.call.preallocated.setup(i32 0) 13463 %t2 = call token @llvm.call.preallocated.setup(i32 0) 13464 call void foo() ["preallocated"(token %t2)] 13465 call void foo() ["preallocated"(token %t1)] 13466 13467is allowed, but not 13468 13469:: code-block:: llvm 13470 13471 %t1 = call token @llvm.call.preallocated.setup(i32 0) 13472 %t2 = call token @llvm.call.preallocated.setup(i32 0) 13473 call void foo() ["preallocated"(token %t1)] 13474 call void foo() ["preallocated"(token %t2)] 13475 13476.. _int_call_preallocated_arg: 13477 13478'``llvm.call.preallocated.arg``' Intrinsic 13479^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13480 13481Syntax: 13482""""""" 13483 13484:: 13485 13486 declare i8* @llvm.call.preallocated.arg(token %setup_token, i32 %arg_index) 13487 13488Overview: 13489""""""""" 13490 13491The '``llvm.call.preallocated.arg``' intrinsic returns a pointer to the 13492corresponding preallocated argument for the preallocated call. 13493 13494Semantics: 13495"""""""""" 13496 13497The '``llvm.call.preallocated.arg``' intrinsic returns a pointer to the 13498``%arg_index``th argument with the ``preallocated`` attribute for 13499the call associated with the ``%setup_token``, which must be from 13500'``llvm.call.preallocated.setup``'. 13501 13502A call to '``llvm.call.preallocated.arg``' must have a call site 13503``preallocated`` attribute. The type of the ``preallocated`` attribute must 13504match the type used by the ``preallocated`` attribute of the corresponding 13505argument at the preallocated call. The type is used in the case that an 13506``llvm.call.preallocated.setup`` does not have a corresponding call (e.g. due 13507to DCE), where otherwise we cannot know how large the arguments are. 13508 13509It is undefined behavior if this is called with a token from an 13510'``llvm.call.preallocated.setup``' if another 13511'``llvm.call.preallocated.setup``' has already been called or if the 13512preallocated call corresponding to the '``llvm.call.preallocated.setup``' 13513has already been called. 13514 13515.. _int_call_preallocated_teardown: 13516 13517'``llvm.call.preallocated.teardown``' Intrinsic 13518^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13519 13520Syntax: 13521""""""" 13522 13523:: 13524 13525 declare i8* @llvm.call.preallocated.teardown(token %setup_token) 13526 13527Overview: 13528""""""""" 13529 13530The '``llvm.call.preallocated.teardown``' intrinsic cleans up the stack 13531created by a '``llvm.call.preallocated.setup``'. 13532 13533Semantics: 13534"""""""""" 13535 13536The token argument must be a '``llvm.call.preallocated.setup``'. 13537 13538The '``llvm.call.preallocated.teardown``' intrinsic cleans up the stack 13539allocated by the corresponding '``llvm.call.preallocated.setup``'. Exactly 13540one of this or the preallocated call must be called to prevent stack leaks. 13541It is undefined behavior to call both a '``llvm.call.preallocated.teardown``' 13542and the preallocated call for a given '``llvm.call.preallocated.setup``'. 13543 13544For example, if the stack is allocated for a preallocated call by a 13545'``llvm.call.preallocated.setup``', then an initializer function called on an 13546allocated argument throws an exception, there should be a 13547'``llvm.call.preallocated.teardown``' in the exception handler to prevent 13548stack leaks. 13549 13550Following the nesting rules in '``llvm.call.preallocated.setup``', nested 13551calls to '``llvm.call.preallocated.setup``' and 13552'``llvm.call.preallocated.teardown``' are allowed but must be properly 13553nested. 13554 13555Example: 13556"""""""" 13557 13558.. code-block:: llvm 13559 13560 %cs = call token @llvm.call.preallocated.setup(i32 1) 13561 %x = call i8* @llvm.call.preallocated.arg(token %cs, i32 0) preallocated(i32) 13562 %y = bitcast i8* %x to i32* 13563 invoke void @constructor(i32* %y) to label %conta unwind label %contb 13564 conta: 13565 call void @foo1(i32* preallocated(i32) %y) ["preallocated"(token %cs)] 13566 ret void 13567 contb: 13568 %s = catchswitch within none [label %catch] unwind to caller 13569 catch: 13570 %p = catchpad within %s [] 13571 call void @llvm.call.preallocated.teardown(token %cs) 13572 ret void 13573 13574Standard C/C++ Library Intrinsics 13575--------------------------------- 13576 13577LLVM provides intrinsics for a few important standard C/C++ library 13578functions. These intrinsics allow source-language front-ends to pass 13579information about the alignment of the pointer arguments to the code 13580generator, providing opportunity for more efficient code generation. 13581 13582 13583'``llvm.abs.*``' Intrinsic 13584^^^^^^^^^^^^^^^^^^^^^^^^^^ 13585 13586Syntax: 13587""""""" 13588 13589This is an overloaded intrinsic. You can use ``llvm.abs`` on any 13590integer bit width or any vector of integer elements. 13591 13592:: 13593 13594 declare i32 @llvm.abs.i32(i32 <src>, i1 <is_int_min_poison>) 13595 declare <4 x i32> @llvm.abs.v4i32(<4 x i32> <src>, i1 <is_int_min_poison>) 13596 13597Overview: 13598""""""""" 13599 13600The '``llvm.abs``' family of intrinsic functions returns the absolute value 13601of an argument. 13602 13603Arguments: 13604"""""""""" 13605 13606The first argument is the value for which the absolute value is to be returned. 13607This argument may be of any integer type or a vector with integer element type. 13608The return type must match the first argument type. 13609 13610The second argument must be a constant and is a flag to indicate whether the 13611result value of the '``llvm.abs``' intrinsic is a 13612:ref:`poison value <poisonvalues>` if the argument is statically or dynamically 13613an ``INT_MIN`` value. 13614 13615Semantics: 13616"""""""""" 13617 13618The '``llvm.abs``' intrinsic returns the magnitude (always positive) of the 13619argument or each element of a vector argument.". If the argument is ``INT_MIN``, 13620then the result is also ``INT_MIN`` if ``is_int_min_poison == 0`` and 13621``poison`` otherwise. 13622 13623 13624'``llvm.smax.*``' Intrinsic 13625^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13626 13627Syntax: 13628""""""" 13629 13630This is an overloaded intrinsic. You can use ``@llvm.smax`` on any 13631integer bit width or any vector of integer elements. 13632 13633:: 13634 13635 declare i32 @llvm.smax.i32(i32 %a, i32 %b) 13636 declare <4 x i32> @llvm.smax.v4i32(<4 x i32> %a, <4 x i32> %b) 13637 13638Overview: 13639""""""""" 13640 13641Return the larger of ``%a`` and ``%b`` comparing the values as signed integers. 13642Vector intrinsics operate on a per-element basis. The larger element of ``%a`` 13643and ``%b`` at a given index is returned for that index. 13644 13645Arguments: 13646"""""""""" 13647 13648The arguments (``%a`` and ``%b``) may be of any integer type or a vector with 13649integer element type. The argument types must match each other, and the return 13650type must match the argument type. 13651 13652 13653'``llvm.smin.*``' Intrinsic 13654^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13655 13656Syntax: 13657""""""" 13658 13659This is an overloaded intrinsic. You can use ``@llvm.smin`` on any 13660integer bit width or any vector of integer elements. 13661 13662:: 13663 13664 declare i32 @llvm.smin.i32(i32 %a, i32 %b) 13665 declare <4 x i32> @llvm.smin.v4i32(<4 x i32> %a, <4 x i32> %b) 13666 13667Overview: 13668""""""""" 13669 13670Return the smaller of ``%a`` and ``%b`` comparing the values as signed integers. 13671Vector intrinsics operate on a per-element basis. The smaller element of ``%a`` 13672and ``%b`` at a given index is returned for that index. 13673 13674Arguments: 13675"""""""""" 13676 13677The arguments (``%a`` and ``%b``) may be of any integer type or a vector with 13678integer element type. The argument types must match each other, and the return 13679type must match the argument type. 13680 13681 13682'``llvm.umax.*``' Intrinsic 13683^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13684 13685Syntax: 13686""""""" 13687 13688This is an overloaded intrinsic. You can use ``@llvm.umax`` on any 13689integer bit width or any vector of integer elements. 13690 13691:: 13692 13693 declare i32 @llvm.umax.i32(i32 %a, i32 %b) 13694 declare <4 x i32> @llvm.umax.v4i32(<4 x i32> %a, <4 x i32> %b) 13695 13696Overview: 13697""""""""" 13698 13699Return the larger of ``%a`` and ``%b`` comparing the values as unsigned 13700integers. Vector intrinsics operate on a per-element basis. The larger element 13701of ``%a`` and ``%b`` at a given index is returned for that index. 13702 13703Arguments: 13704"""""""""" 13705 13706The arguments (``%a`` and ``%b``) may be of any integer type or a vector with 13707integer element type. The argument types must match each other, and the return 13708type must match the argument type. 13709 13710 13711'``llvm.umin.*``' Intrinsic 13712^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13713 13714Syntax: 13715""""""" 13716 13717This is an overloaded intrinsic. You can use ``@llvm.umin`` on any 13718integer bit width or any vector of integer elements. 13719 13720:: 13721 13722 declare i32 @llvm.umin.i32(i32 %a, i32 %b) 13723 declare <4 x i32> @llvm.umin.v4i32(<4 x i32> %a, <4 x i32> %b) 13724 13725Overview: 13726""""""""" 13727 13728Return the smaller of ``%a`` and ``%b`` comparing the values as unsigned 13729integers. Vector intrinsics operate on a per-element basis. The smaller element 13730of ``%a`` and ``%b`` at a given index is returned for that index. 13731 13732Arguments: 13733"""""""""" 13734 13735The arguments (``%a`` and ``%b``) may be of any integer type or a vector with 13736integer element type. The argument types must match each other, and the return 13737type must match the argument type. 13738 13739 13740.. _int_memcpy: 13741 13742'``llvm.memcpy``' Intrinsic 13743^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13744 13745Syntax: 13746""""""" 13747 13748This is an overloaded intrinsic. You can use ``llvm.memcpy`` on any 13749integer bit width and for different address spaces. Not all targets 13750support all bit widths however. 13751 13752:: 13753 13754 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>, 13755 i32 <len>, i1 <isvolatile>) 13756 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>, 13757 i64 <len>, i1 <isvolatile>) 13758 13759Overview: 13760""""""""" 13761 13762The '``llvm.memcpy.*``' intrinsics copy a block of memory from the 13763source location to the destination location. 13764 13765Note that, unlike the standard libc function, the ``llvm.memcpy.*`` 13766intrinsics do not return a value, takes extra isvolatile 13767arguments and the pointers can be in specified address spaces. 13768 13769Arguments: 13770"""""""""" 13771 13772The first argument is a pointer to the destination, the second is a 13773pointer to the source. The third argument is an integer argument 13774specifying the number of bytes to copy, and the fourth is a 13775boolean indicating a volatile access. 13776 13777The :ref:`align <attr_align>` parameter attribute can be provided 13778for the first and second arguments. 13779 13780If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy`` call is 13781a :ref:`volatile operation <volatile>`. The detailed access behavior is not 13782very cleanly specified and it is unwise to depend on it. 13783 13784Semantics: 13785"""""""""" 13786 13787The '``llvm.memcpy.*``' intrinsics copy a block of memory from the source 13788location to the destination location, which must either be equal or 13789non-overlapping. It copies "len" bytes of memory over. If the argument is known 13790to be aligned to some boundary, this can be specified as an attribute on the 13791argument. 13792 13793If ``<len>`` is 0, it is no-op modulo the behavior of attributes attached to 13794the arguments. 13795If ``<len>`` is not a well-defined value, the behavior is undefined. 13796If ``<len>`` is not zero, both ``<dest>`` and ``<src>`` should be well-defined, 13797otherwise the behavior is undefined. 13798 13799.. _int_memcpy_inline: 13800 13801'``llvm.memcpy.inline``' Intrinsic 13802^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13803 13804Syntax: 13805""""""" 13806 13807This is an overloaded intrinsic. You can use ``llvm.memcpy.inline`` on any 13808integer bit width and for different address spaces. Not all targets 13809support all bit widths however. 13810 13811:: 13812 13813 declare void @llvm.memcpy.inline.p0i8.p0i8.i32(i8* <dest>, i8* <src>, 13814 i32 <len>, i1 <isvolatile>) 13815 declare void @llvm.memcpy.inline.p0i8.p0i8.i64(i8* <dest>, i8* <src>, 13816 i64 <len>, i1 <isvolatile>) 13817 13818Overview: 13819""""""""" 13820 13821The '``llvm.memcpy.inline.*``' intrinsics copy a block of memory from the 13822source location to the destination location and guarantees that no external 13823functions are called. 13824 13825Note that, unlike the standard libc function, the ``llvm.memcpy.inline.*`` 13826intrinsics do not return a value, takes extra isvolatile 13827arguments and the pointers can be in specified address spaces. 13828 13829Arguments: 13830"""""""""" 13831 13832The first argument is a pointer to the destination, the second is a 13833pointer to the source. The third argument is a constant integer argument 13834specifying the number of bytes to copy, and the fourth is a 13835boolean indicating a volatile access. 13836 13837The :ref:`align <attr_align>` parameter attribute can be provided 13838for the first and second arguments. 13839 13840If the ``isvolatile`` parameter is ``true``, the ``llvm.memcpy.inline`` call is 13841a :ref:`volatile operation <volatile>`. The detailed access behavior is not 13842very cleanly specified and it is unwise to depend on it. 13843 13844Semantics: 13845"""""""""" 13846 13847The '``llvm.memcpy.inline.*``' intrinsics copy a block of memory from the 13848source location to the destination location, which are not allowed to 13849overlap. It copies "len" bytes of memory over. If the argument is known 13850to be aligned to some boundary, this can be specified as an attribute on 13851the argument. 13852The behavior of '``llvm.memcpy.inline.*``' is equivalent to the behavior of 13853'``llvm.memcpy.*``', but the generated code is guaranteed not to call any 13854external functions. 13855 13856.. _int_memmove: 13857 13858'``llvm.memmove``' Intrinsic 13859^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13860 13861Syntax: 13862""""""" 13863 13864This is an overloaded intrinsic. You can use llvm.memmove on any integer 13865bit width and for different address space. Not all targets support all 13866bit widths however. 13867 13868:: 13869 13870 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>, 13871 i32 <len>, i1 <isvolatile>) 13872 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>, 13873 i64 <len>, i1 <isvolatile>) 13874 13875Overview: 13876""""""""" 13877 13878The '``llvm.memmove.*``' intrinsics move a block of memory from the 13879source location to the destination location. It is similar to the 13880'``llvm.memcpy``' intrinsic but allows the two memory locations to 13881overlap. 13882 13883Note that, unlike the standard libc function, the ``llvm.memmove.*`` 13884intrinsics do not return a value, takes an extra isvolatile 13885argument and the pointers can be in specified address spaces. 13886 13887Arguments: 13888"""""""""" 13889 13890The first argument is a pointer to the destination, the second is a 13891pointer to the source. The third argument is an integer argument 13892specifying the number of bytes to copy, and the fourth is a 13893boolean indicating a volatile access. 13894 13895The :ref:`align <attr_align>` parameter attribute can be provided 13896for the first and second arguments. 13897 13898If the ``isvolatile`` parameter is ``true``, the ``llvm.memmove`` call 13899is a :ref:`volatile operation <volatile>`. The detailed access behavior is 13900not very cleanly specified and it is unwise to depend on it. 13901 13902Semantics: 13903"""""""""" 13904 13905The '``llvm.memmove.*``' intrinsics copy a block of memory from the 13906source location to the destination location, which may overlap. It 13907copies "len" bytes of memory over. If the argument is known to be 13908aligned to some boundary, this can be specified as an attribute on 13909the argument. 13910 13911If ``<len>`` is 0, it is no-op modulo the behavior of attributes attached to 13912the arguments. 13913If ``<len>`` is not a well-defined value, the behavior is undefined. 13914If ``<len>`` is not zero, both ``<dest>`` and ``<src>`` should be well-defined, 13915otherwise the behavior is undefined. 13916 13917.. _int_memset: 13918 13919'``llvm.memset.*``' Intrinsics 13920^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13921 13922Syntax: 13923""""""" 13924 13925This is an overloaded intrinsic. You can use llvm.memset on any integer 13926bit width and for different address spaces. However, not all targets 13927support all bit widths. 13928 13929:: 13930 13931 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>, 13932 i32 <len>, i1 <isvolatile>) 13933 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>, 13934 i64 <len>, i1 <isvolatile>) 13935 13936Overview: 13937""""""""" 13938 13939The '``llvm.memset.*``' intrinsics fill a block of memory with a 13940particular byte value. 13941 13942Note that, unlike the standard libc function, the ``llvm.memset`` 13943intrinsic does not return a value and takes an extra volatile 13944argument. Also, the destination can be in an arbitrary address space. 13945 13946Arguments: 13947"""""""""" 13948 13949The first argument is a pointer to the destination to fill, the second 13950is the byte value with which to fill it, the third argument is an 13951integer argument specifying the number of bytes to fill, and the fourth 13952is a boolean indicating a volatile access. 13953 13954The :ref:`align <attr_align>` parameter attribute can be provided 13955for the first arguments. 13956 13957If the ``isvolatile`` parameter is ``true``, the ``llvm.memset`` call is 13958a :ref:`volatile operation <volatile>`. The detailed access behavior is not 13959very cleanly specified and it is unwise to depend on it. 13960 13961Semantics: 13962"""""""""" 13963 13964The '``llvm.memset.*``' intrinsics fill "len" bytes of memory starting 13965at the destination location. If the argument is known to be 13966aligned to some boundary, this can be specified as an attribute on 13967the argument. 13968 13969If ``<len>`` is 0, it is no-op modulo the behavior of attributes attached to 13970the arguments. 13971If ``<len>`` is not a well-defined value, the behavior is undefined. 13972If ``<len>`` is not zero, ``<dest>`` should be well-defined, otherwise the 13973behavior is undefined. 13974 13975.. _int_memset_inline: 13976 13977'``llvm.memset.inline``' Intrinsic 13978^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 13979 13980Syntax: 13981""""""" 13982 13983This is an overloaded intrinsic. You can use ``llvm.memset.inline`` on any 13984integer bit width and for different address spaces. Not all targets 13985support all bit widths however. 13986 13987:: 13988 13989 declare void @llvm.memset.inline.p0i8.p0i8.i32(i8* <dest>, i8 <val>, 13990 i32 <len>, 13991 i1 <isvolatile>) 13992 declare void @llvm.memset.inline.p0i8.p0i8.i64(i8* <dest>, i8 <val>, 13993 i64 <len>, 13994 i1 <isvolatile>) 13995 13996Overview: 13997""""""""" 13998 13999The '``llvm.memset.inline.*``' intrinsics fill a block of memory with a 14000particular byte value and guarantees that no external functions are called. 14001 14002Note that, unlike the standard libc function, the ``llvm.memset.inline.*`` 14003intrinsics do not return a value, take an extra isvolatile argument and the 14004pointer can be in specified address spaces. 14005 14006Arguments: 14007"""""""""" 14008 14009The first argument is a pointer to the destination to fill, the second 14010is the byte value with which to fill it, the third argument is a constant 14011integer argument specifying the number of bytes to fill, and the fourth 14012is a boolean indicating a volatile access. 14013 14014The :ref:`align <attr_align>` parameter attribute can be provided 14015for the first argument. 14016 14017If the ``isvolatile`` parameter is ``true``, the ``llvm.memset.inline`` call is 14018a :ref:`volatile operation <volatile>`. The detailed access behavior is not 14019very cleanly specified and it is unwise to depend on it. 14020 14021Semantics: 14022"""""""""" 14023 14024The '``llvm.memset.inline.*``' intrinsics fill "len" bytes of memory starting 14025at the destination location. If the argument is known to be 14026aligned to some boundary, this can be specified as an attribute on 14027the argument. 14028 14029``len`` must be a constant expression. 14030If ``<len>`` is 0, it is no-op modulo the behavior of attributes attached to 14031the arguments. 14032If ``<len>`` is not a well-defined value, the behavior is undefined. 14033If ``<len>`` is not zero, ``<dest>`` should be well-defined, otherwise the 14034behavior is undefined. 14035 14036The behavior of '``llvm.memset.inline.*``' is equivalent to the behavior of 14037'``llvm.memset.*``', but the generated code is guaranteed not to call any 14038external functions. 14039 14040'``llvm.sqrt.*``' Intrinsic 14041^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14042 14043Syntax: 14044""""""" 14045 14046This is an overloaded intrinsic. You can use ``llvm.sqrt`` on any 14047floating-point or vector of floating-point type. Not all targets support 14048all types however. 14049 14050:: 14051 14052 declare float @llvm.sqrt.f32(float %Val) 14053 declare double @llvm.sqrt.f64(double %Val) 14054 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val) 14055 declare fp128 @llvm.sqrt.f128(fp128 %Val) 14056 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val) 14057 14058Overview: 14059""""""""" 14060 14061The '``llvm.sqrt``' intrinsics return the square root of the specified value. 14062 14063Arguments: 14064"""""""""" 14065 14066The argument and return value are floating-point numbers of the same type. 14067 14068Semantics: 14069"""""""""" 14070 14071Return the same value as a corresponding libm '``sqrt``' function but without 14072trapping or setting ``errno``. For types specified by IEEE-754, the result 14073matches a conforming libm implementation. 14074 14075When specified with the fast-math-flag 'afn', the result may be approximated 14076using a less accurate calculation. 14077 14078'``llvm.powi.*``' Intrinsic 14079^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14080 14081Syntax: 14082""""""" 14083 14084This is an overloaded intrinsic. You can use ``llvm.powi`` on any 14085floating-point or vector of floating-point type. Not all targets support 14086all types however. 14087 14088Generally, the only supported type for the exponent is the one matching 14089with the C type ``int``. 14090 14091:: 14092 14093 declare float @llvm.powi.f32.i32(float %Val, i32 %power) 14094 declare double @llvm.powi.f64.i16(double %Val, i16 %power) 14095 declare x86_fp80 @llvm.powi.f80.i32(x86_fp80 %Val, i32 %power) 14096 declare fp128 @llvm.powi.f128.i32(fp128 %Val, i32 %power) 14097 declare ppc_fp128 @llvm.powi.ppcf128.i32(ppc_fp128 %Val, i32 %power) 14098 14099Overview: 14100""""""""" 14101 14102The '``llvm.powi.*``' intrinsics return the first operand raised to the 14103specified (positive or negative) power. The order of evaluation of 14104multiplications is not defined. When a vector of floating-point type is 14105used, the second argument remains a scalar integer value. 14106 14107Arguments: 14108"""""""""" 14109 14110The second argument is an integer power, and the first is a value to 14111raise to that power. 14112 14113Semantics: 14114"""""""""" 14115 14116This function returns the first value raised to the second power with an 14117unspecified sequence of rounding operations. 14118 14119'``llvm.sin.*``' Intrinsic 14120^^^^^^^^^^^^^^^^^^^^^^^^^^ 14121 14122Syntax: 14123""""""" 14124 14125This is an overloaded intrinsic. You can use ``llvm.sin`` on any 14126floating-point or vector of floating-point type. Not all targets support 14127all types however. 14128 14129:: 14130 14131 declare float @llvm.sin.f32(float %Val) 14132 declare double @llvm.sin.f64(double %Val) 14133 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val) 14134 declare fp128 @llvm.sin.f128(fp128 %Val) 14135 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val) 14136 14137Overview: 14138""""""""" 14139 14140The '``llvm.sin.*``' intrinsics return the sine of the operand. 14141 14142Arguments: 14143"""""""""" 14144 14145The argument and return value are floating-point numbers of the same type. 14146 14147Semantics: 14148"""""""""" 14149 14150Return the same value as a corresponding libm '``sin``' function but without 14151trapping or setting ``errno``. 14152 14153When specified with the fast-math-flag 'afn', the result may be approximated 14154using a less accurate calculation. 14155 14156'``llvm.cos.*``' Intrinsic 14157^^^^^^^^^^^^^^^^^^^^^^^^^^ 14158 14159Syntax: 14160""""""" 14161 14162This is an overloaded intrinsic. You can use ``llvm.cos`` on any 14163floating-point or vector of floating-point type. Not all targets support 14164all types however. 14165 14166:: 14167 14168 declare float @llvm.cos.f32(float %Val) 14169 declare double @llvm.cos.f64(double %Val) 14170 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val) 14171 declare fp128 @llvm.cos.f128(fp128 %Val) 14172 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val) 14173 14174Overview: 14175""""""""" 14176 14177The '``llvm.cos.*``' intrinsics return the cosine of the operand. 14178 14179Arguments: 14180"""""""""" 14181 14182The argument and return value are floating-point numbers of the same type. 14183 14184Semantics: 14185"""""""""" 14186 14187Return the same value as a corresponding libm '``cos``' function but without 14188trapping or setting ``errno``. 14189 14190When specified with the fast-math-flag 'afn', the result may be approximated 14191using a less accurate calculation. 14192 14193'``llvm.pow.*``' Intrinsic 14194^^^^^^^^^^^^^^^^^^^^^^^^^^ 14195 14196Syntax: 14197""""""" 14198 14199This is an overloaded intrinsic. You can use ``llvm.pow`` on any 14200floating-point or vector of floating-point type. Not all targets support 14201all types however. 14202 14203:: 14204 14205 declare float @llvm.pow.f32(float %Val, float %Power) 14206 declare double @llvm.pow.f64(double %Val, double %Power) 14207 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power) 14208 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power) 14209 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power) 14210 14211Overview: 14212""""""""" 14213 14214The '``llvm.pow.*``' intrinsics return the first operand raised to the 14215specified (positive or negative) power. 14216 14217Arguments: 14218"""""""""" 14219 14220The arguments and return value are floating-point numbers of the same type. 14221 14222Semantics: 14223"""""""""" 14224 14225Return the same value as a corresponding libm '``pow``' function but without 14226trapping or setting ``errno``. 14227 14228When specified with the fast-math-flag 'afn', the result may be approximated 14229using a less accurate calculation. 14230 14231'``llvm.exp.*``' Intrinsic 14232^^^^^^^^^^^^^^^^^^^^^^^^^^ 14233 14234Syntax: 14235""""""" 14236 14237This is an overloaded intrinsic. You can use ``llvm.exp`` on any 14238floating-point or vector of floating-point type. Not all targets support 14239all types however. 14240 14241:: 14242 14243 declare float @llvm.exp.f32(float %Val) 14244 declare double @llvm.exp.f64(double %Val) 14245 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val) 14246 declare fp128 @llvm.exp.f128(fp128 %Val) 14247 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val) 14248 14249Overview: 14250""""""""" 14251 14252The '``llvm.exp.*``' intrinsics compute the base-e exponential of the specified 14253value. 14254 14255Arguments: 14256"""""""""" 14257 14258The argument and return value are floating-point numbers of the same type. 14259 14260Semantics: 14261"""""""""" 14262 14263Return the same value as a corresponding libm '``exp``' function but without 14264trapping or setting ``errno``. 14265 14266When specified with the fast-math-flag 'afn', the result may be approximated 14267using a less accurate calculation. 14268 14269'``llvm.exp2.*``' Intrinsic 14270^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14271 14272Syntax: 14273""""""" 14274 14275This is an overloaded intrinsic. You can use ``llvm.exp2`` on any 14276floating-point or vector of floating-point type. Not all targets support 14277all types however. 14278 14279:: 14280 14281 declare float @llvm.exp2.f32(float %Val) 14282 declare double @llvm.exp2.f64(double %Val) 14283 declare x86_fp80 @llvm.exp2.f80(x86_fp80 %Val) 14284 declare fp128 @llvm.exp2.f128(fp128 %Val) 14285 declare ppc_fp128 @llvm.exp2.ppcf128(ppc_fp128 %Val) 14286 14287Overview: 14288""""""""" 14289 14290The '``llvm.exp2.*``' intrinsics compute the base-2 exponential of the 14291specified value. 14292 14293Arguments: 14294"""""""""" 14295 14296The argument and return value are floating-point numbers of the same type. 14297 14298Semantics: 14299"""""""""" 14300 14301Return the same value as a corresponding libm '``exp2``' function but without 14302trapping or setting ``errno``. 14303 14304When specified with the fast-math-flag 'afn', the result may be approximated 14305using a less accurate calculation. 14306 14307'``llvm.log.*``' Intrinsic 14308^^^^^^^^^^^^^^^^^^^^^^^^^^ 14309 14310Syntax: 14311""""""" 14312 14313This is an overloaded intrinsic. You can use ``llvm.log`` on any 14314floating-point or vector of floating-point type. Not all targets support 14315all types however. 14316 14317:: 14318 14319 declare float @llvm.log.f32(float %Val) 14320 declare double @llvm.log.f64(double %Val) 14321 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val) 14322 declare fp128 @llvm.log.f128(fp128 %Val) 14323 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val) 14324 14325Overview: 14326""""""""" 14327 14328The '``llvm.log.*``' intrinsics compute the base-e logarithm of the specified 14329value. 14330 14331Arguments: 14332"""""""""" 14333 14334The argument and return value are floating-point numbers of the same type. 14335 14336Semantics: 14337"""""""""" 14338 14339Return the same value as a corresponding libm '``log``' function but without 14340trapping or setting ``errno``. 14341 14342When specified with the fast-math-flag 'afn', the result may be approximated 14343using a less accurate calculation. 14344 14345'``llvm.log10.*``' Intrinsic 14346^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14347 14348Syntax: 14349""""""" 14350 14351This is an overloaded intrinsic. You can use ``llvm.log10`` on any 14352floating-point or vector of floating-point type. Not all targets support 14353all types however. 14354 14355:: 14356 14357 declare float @llvm.log10.f32(float %Val) 14358 declare double @llvm.log10.f64(double %Val) 14359 declare x86_fp80 @llvm.log10.f80(x86_fp80 %Val) 14360 declare fp128 @llvm.log10.f128(fp128 %Val) 14361 declare ppc_fp128 @llvm.log10.ppcf128(ppc_fp128 %Val) 14362 14363Overview: 14364""""""""" 14365 14366The '``llvm.log10.*``' intrinsics compute the base-10 logarithm of the 14367specified value. 14368 14369Arguments: 14370"""""""""" 14371 14372The argument and return value are floating-point numbers of the same type. 14373 14374Semantics: 14375"""""""""" 14376 14377Return the same value as a corresponding libm '``log10``' function but without 14378trapping or setting ``errno``. 14379 14380When specified with the fast-math-flag 'afn', the result may be approximated 14381using a less accurate calculation. 14382 14383'``llvm.log2.*``' Intrinsic 14384^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14385 14386Syntax: 14387""""""" 14388 14389This is an overloaded intrinsic. You can use ``llvm.log2`` on any 14390floating-point or vector of floating-point type. Not all targets support 14391all types however. 14392 14393:: 14394 14395 declare float @llvm.log2.f32(float %Val) 14396 declare double @llvm.log2.f64(double %Val) 14397 declare x86_fp80 @llvm.log2.f80(x86_fp80 %Val) 14398 declare fp128 @llvm.log2.f128(fp128 %Val) 14399 declare ppc_fp128 @llvm.log2.ppcf128(ppc_fp128 %Val) 14400 14401Overview: 14402""""""""" 14403 14404The '``llvm.log2.*``' intrinsics compute the base-2 logarithm of the specified 14405value. 14406 14407Arguments: 14408"""""""""" 14409 14410The argument and return value are floating-point numbers of the same type. 14411 14412Semantics: 14413"""""""""" 14414 14415Return the same value as a corresponding libm '``log2``' function but without 14416trapping or setting ``errno``. 14417 14418When specified with the fast-math-flag 'afn', the result may be approximated 14419using a less accurate calculation. 14420 14421.. _int_fma: 14422 14423'``llvm.fma.*``' Intrinsic 14424^^^^^^^^^^^^^^^^^^^^^^^^^^ 14425 14426Syntax: 14427""""""" 14428 14429This is an overloaded intrinsic. You can use ``llvm.fma`` on any 14430floating-point or vector of floating-point type. Not all targets support 14431all types however. 14432 14433:: 14434 14435 declare float @llvm.fma.f32(float %a, float %b, float %c) 14436 declare double @llvm.fma.f64(double %a, double %b, double %c) 14437 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c) 14438 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c) 14439 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c) 14440 14441Overview: 14442""""""""" 14443 14444The '``llvm.fma.*``' intrinsics perform the fused multiply-add operation. 14445 14446Arguments: 14447"""""""""" 14448 14449The arguments and return value are floating-point numbers of the same type. 14450 14451Semantics: 14452"""""""""" 14453 14454Return the same value as a corresponding libm '``fma``' function but without 14455trapping or setting ``errno``. 14456 14457When specified with the fast-math-flag 'afn', the result may be approximated 14458using a less accurate calculation. 14459 14460'``llvm.fabs.*``' Intrinsic 14461^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14462 14463Syntax: 14464""""""" 14465 14466This is an overloaded intrinsic. You can use ``llvm.fabs`` on any 14467floating-point or vector of floating-point type. Not all targets support 14468all types however. 14469 14470:: 14471 14472 declare float @llvm.fabs.f32(float %Val) 14473 declare double @llvm.fabs.f64(double %Val) 14474 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val) 14475 declare fp128 @llvm.fabs.f128(fp128 %Val) 14476 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val) 14477 14478Overview: 14479""""""""" 14480 14481The '``llvm.fabs.*``' intrinsics return the absolute value of the 14482operand. 14483 14484Arguments: 14485"""""""""" 14486 14487The argument and return value are floating-point numbers of the same 14488type. 14489 14490Semantics: 14491"""""""""" 14492 14493This function returns the same values as the libm ``fabs`` functions 14494would, and handles error conditions in the same way. 14495 14496'``llvm.minnum.*``' Intrinsic 14497^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14498 14499Syntax: 14500""""""" 14501 14502This is an overloaded intrinsic. You can use ``llvm.minnum`` on any 14503floating-point or vector of floating-point type. Not all targets support 14504all types however. 14505 14506:: 14507 14508 declare float @llvm.minnum.f32(float %Val0, float %Val1) 14509 declare double @llvm.minnum.f64(double %Val0, double %Val1) 14510 declare x86_fp80 @llvm.minnum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 14511 declare fp128 @llvm.minnum.f128(fp128 %Val0, fp128 %Val1) 14512 declare ppc_fp128 @llvm.minnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 14513 14514Overview: 14515""""""""" 14516 14517The '``llvm.minnum.*``' intrinsics return the minimum of the two 14518arguments. 14519 14520 14521Arguments: 14522"""""""""" 14523 14524The arguments and return value are floating-point numbers of the same 14525type. 14526 14527Semantics: 14528"""""""""" 14529 14530Follows the IEEE-754 semantics for minNum, except for handling of 14531signaling NaNs. This match's the behavior of libm's fmin. 14532 14533If either operand is a NaN, returns the other non-NaN operand. Returns 14534NaN only if both operands are NaN. The returned NaN is always 14535quiet. If the operands compare equal, returns a value that compares 14536equal to both operands. This means that fmin(+/-0.0, +/-0.0) could 14537return either -0.0 or 0.0. 14538 14539Unlike the IEEE-754 2008 behavior, this does not distinguish between 14540signaling and quiet NaN inputs. If a target's implementation follows 14541the standard and returns a quiet NaN if either input is a signaling 14542NaN, the intrinsic lowering is responsible for quieting the inputs to 14543correctly return the non-NaN input (e.g. by using the equivalent of 14544``llvm.canonicalize``). 14545 14546 14547'``llvm.maxnum.*``' Intrinsic 14548^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14549 14550Syntax: 14551""""""" 14552 14553This is an overloaded intrinsic. You can use ``llvm.maxnum`` on any 14554floating-point or vector of floating-point type. Not all targets support 14555all types however. 14556 14557:: 14558 14559 declare float @llvm.maxnum.f32(float %Val0, float %Val1) 14560 declare double @llvm.maxnum.f64(double %Val0, double %Val1) 14561 declare x86_fp80 @llvm.maxnum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 14562 declare fp128 @llvm.maxnum.f128(fp128 %Val0, fp128 %Val1) 14563 declare ppc_fp128 @llvm.maxnum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 14564 14565Overview: 14566""""""""" 14567 14568The '``llvm.maxnum.*``' intrinsics return the maximum of the two 14569arguments. 14570 14571 14572Arguments: 14573"""""""""" 14574 14575The arguments and return value are floating-point numbers of the same 14576type. 14577 14578Semantics: 14579"""""""""" 14580Follows the IEEE-754 semantics for maxNum except for the handling of 14581signaling NaNs. This matches the behavior of libm's fmax. 14582 14583If either operand is a NaN, returns the other non-NaN operand. Returns 14584NaN only if both operands are NaN. The returned NaN is always 14585quiet. If the operands compare equal, returns a value that compares 14586equal to both operands. This means that fmax(+/-0.0, +/-0.0) could 14587return either -0.0 or 0.0. 14588 14589Unlike the IEEE-754 2008 behavior, this does not distinguish between 14590signaling and quiet NaN inputs. If a target's implementation follows 14591the standard and returns a quiet NaN if either input is a signaling 14592NaN, the intrinsic lowering is responsible for quieting the inputs to 14593correctly return the non-NaN input (e.g. by using the equivalent of 14594``llvm.canonicalize``). 14595 14596'``llvm.minimum.*``' Intrinsic 14597^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14598 14599Syntax: 14600""""""" 14601 14602This is an overloaded intrinsic. You can use ``llvm.minimum`` on any 14603floating-point or vector of floating-point type. Not all targets support 14604all types however. 14605 14606:: 14607 14608 declare float @llvm.minimum.f32(float %Val0, float %Val1) 14609 declare double @llvm.minimum.f64(double %Val0, double %Val1) 14610 declare x86_fp80 @llvm.minimum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 14611 declare fp128 @llvm.minimum.f128(fp128 %Val0, fp128 %Val1) 14612 declare ppc_fp128 @llvm.minimum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 14613 14614Overview: 14615""""""""" 14616 14617The '``llvm.minimum.*``' intrinsics return the minimum of the two 14618arguments, propagating NaNs and treating -0.0 as less than +0.0. 14619 14620 14621Arguments: 14622"""""""""" 14623 14624The arguments and return value are floating-point numbers of the same 14625type. 14626 14627Semantics: 14628"""""""""" 14629If either operand is a NaN, returns NaN. Otherwise returns the lesser 14630of the two arguments. -0.0 is considered to be less than +0.0 for this 14631intrinsic. Note that these are the semantics specified in the draft of 14632IEEE 754-2018. 14633 14634'``llvm.maximum.*``' Intrinsic 14635^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14636 14637Syntax: 14638""""""" 14639 14640This is an overloaded intrinsic. You can use ``llvm.maximum`` on any 14641floating-point or vector of floating-point type. Not all targets support 14642all types however. 14643 14644:: 14645 14646 declare float @llvm.maximum.f32(float %Val0, float %Val1) 14647 declare double @llvm.maximum.f64(double %Val0, double %Val1) 14648 declare x86_fp80 @llvm.maximum.f80(x86_fp80 %Val0, x86_fp80 %Val1) 14649 declare fp128 @llvm.maximum.f128(fp128 %Val0, fp128 %Val1) 14650 declare ppc_fp128 @llvm.maximum.ppcf128(ppc_fp128 %Val0, ppc_fp128 %Val1) 14651 14652Overview: 14653""""""""" 14654 14655The '``llvm.maximum.*``' intrinsics return the maximum of the two 14656arguments, propagating NaNs and treating -0.0 as less than +0.0. 14657 14658 14659Arguments: 14660"""""""""" 14661 14662The arguments and return value are floating-point numbers of the same 14663type. 14664 14665Semantics: 14666"""""""""" 14667If either operand is a NaN, returns NaN. Otherwise returns the greater 14668of the two arguments. -0.0 is considered to be less than +0.0 for this 14669intrinsic. Note that these are the semantics specified in the draft of 14670IEEE 754-2018. 14671 14672'``llvm.copysign.*``' Intrinsic 14673^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14674 14675Syntax: 14676""""""" 14677 14678This is an overloaded intrinsic. You can use ``llvm.copysign`` on any 14679floating-point or vector of floating-point type. Not all targets support 14680all types however. 14681 14682:: 14683 14684 declare float @llvm.copysign.f32(float %Mag, float %Sgn) 14685 declare double @llvm.copysign.f64(double %Mag, double %Sgn) 14686 declare x86_fp80 @llvm.copysign.f80(x86_fp80 %Mag, x86_fp80 %Sgn) 14687 declare fp128 @llvm.copysign.f128(fp128 %Mag, fp128 %Sgn) 14688 declare ppc_fp128 @llvm.copysign.ppcf128(ppc_fp128 %Mag, ppc_fp128 %Sgn) 14689 14690Overview: 14691""""""""" 14692 14693The '``llvm.copysign.*``' intrinsics return a value with the magnitude of the 14694first operand and the sign of the second operand. 14695 14696Arguments: 14697"""""""""" 14698 14699The arguments and return value are floating-point numbers of the same 14700type. 14701 14702Semantics: 14703"""""""""" 14704 14705This function returns the same values as the libm ``copysign`` 14706functions would, and handles error conditions in the same way. 14707 14708'``llvm.floor.*``' Intrinsic 14709^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14710 14711Syntax: 14712""""""" 14713 14714This is an overloaded intrinsic. You can use ``llvm.floor`` on any 14715floating-point or vector of floating-point type. Not all targets support 14716all types however. 14717 14718:: 14719 14720 declare float @llvm.floor.f32(float %Val) 14721 declare double @llvm.floor.f64(double %Val) 14722 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val) 14723 declare fp128 @llvm.floor.f128(fp128 %Val) 14724 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val) 14725 14726Overview: 14727""""""""" 14728 14729The '``llvm.floor.*``' intrinsics return the floor of the operand. 14730 14731Arguments: 14732"""""""""" 14733 14734The argument and return value are floating-point numbers of the same 14735type. 14736 14737Semantics: 14738"""""""""" 14739 14740This function returns the same values as the libm ``floor`` functions 14741would, and handles error conditions in the same way. 14742 14743'``llvm.ceil.*``' Intrinsic 14744^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14745 14746Syntax: 14747""""""" 14748 14749This is an overloaded intrinsic. You can use ``llvm.ceil`` on any 14750floating-point or vector of floating-point type. Not all targets support 14751all types however. 14752 14753:: 14754 14755 declare float @llvm.ceil.f32(float %Val) 14756 declare double @llvm.ceil.f64(double %Val) 14757 declare x86_fp80 @llvm.ceil.f80(x86_fp80 %Val) 14758 declare fp128 @llvm.ceil.f128(fp128 %Val) 14759 declare ppc_fp128 @llvm.ceil.ppcf128(ppc_fp128 %Val) 14760 14761Overview: 14762""""""""" 14763 14764The '``llvm.ceil.*``' intrinsics return the ceiling of the operand. 14765 14766Arguments: 14767"""""""""" 14768 14769The argument and return value are floating-point numbers of the same 14770type. 14771 14772Semantics: 14773"""""""""" 14774 14775This function returns the same values as the libm ``ceil`` functions 14776would, and handles error conditions in the same way. 14777 14778'``llvm.trunc.*``' Intrinsic 14779^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14780 14781Syntax: 14782""""""" 14783 14784This is an overloaded intrinsic. You can use ``llvm.trunc`` on any 14785floating-point or vector of floating-point type. Not all targets support 14786all types however. 14787 14788:: 14789 14790 declare float @llvm.trunc.f32(float %Val) 14791 declare double @llvm.trunc.f64(double %Val) 14792 declare x86_fp80 @llvm.trunc.f80(x86_fp80 %Val) 14793 declare fp128 @llvm.trunc.f128(fp128 %Val) 14794 declare ppc_fp128 @llvm.trunc.ppcf128(ppc_fp128 %Val) 14795 14796Overview: 14797""""""""" 14798 14799The '``llvm.trunc.*``' intrinsics returns the operand rounded to the 14800nearest integer not larger in magnitude than the operand. 14801 14802Arguments: 14803"""""""""" 14804 14805The argument and return value are floating-point numbers of the same 14806type. 14807 14808Semantics: 14809"""""""""" 14810 14811This function returns the same values as the libm ``trunc`` functions 14812would, and handles error conditions in the same way. 14813 14814'``llvm.rint.*``' Intrinsic 14815^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14816 14817Syntax: 14818""""""" 14819 14820This is an overloaded intrinsic. You can use ``llvm.rint`` on any 14821floating-point or vector of floating-point type. Not all targets support 14822all types however. 14823 14824:: 14825 14826 declare float @llvm.rint.f32(float %Val) 14827 declare double @llvm.rint.f64(double %Val) 14828 declare x86_fp80 @llvm.rint.f80(x86_fp80 %Val) 14829 declare fp128 @llvm.rint.f128(fp128 %Val) 14830 declare ppc_fp128 @llvm.rint.ppcf128(ppc_fp128 %Val) 14831 14832Overview: 14833""""""""" 14834 14835The '``llvm.rint.*``' intrinsics returns the operand rounded to the 14836nearest integer. It may raise an inexact floating-point exception if the 14837operand isn't an integer. 14838 14839Arguments: 14840"""""""""" 14841 14842The argument and return value are floating-point numbers of the same 14843type. 14844 14845Semantics: 14846"""""""""" 14847 14848This function returns the same values as the libm ``rint`` functions 14849would, and handles error conditions in the same way. 14850 14851'``llvm.nearbyint.*``' Intrinsic 14852^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14853 14854Syntax: 14855""""""" 14856 14857This is an overloaded intrinsic. You can use ``llvm.nearbyint`` on any 14858floating-point or vector of floating-point type. Not all targets support 14859all types however. 14860 14861:: 14862 14863 declare float @llvm.nearbyint.f32(float %Val) 14864 declare double @llvm.nearbyint.f64(double %Val) 14865 declare x86_fp80 @llvm.nearbyint.f80(x86_fp80 %Val) 14866 declare fp128 @llvm.nearbyint.f128(fp128 %Val) 14867 declare ppc_fp128 @llvm.nearbyint.ppcf128(ppc_fp128 %Val) 14868 14869Overview: 14870""""""""" 14871 14872The '``llvm.nearbyint.*``' intrinsics returns the operand rounded to the 14873nearest integer. 14874 14875Arguments: 14876"""""""""" 14877 14878The argument and return value are floating-point numbers of the same 14879type. 14880 14881Semantics: 14882"""""""""" 14883 14884This function returns the same values as the libm ``nearbyint`` 14885functions would, and handles error conditions in the same way. 14886 14887'``llvm.round.*``' Intrinsic 14888^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14889 14890Syntax: 14891""""""" 14892 14893This is an overloaded intrinsic. You can use ``llvm.round`` on any 14894floating-point or vector of floating-point type. Not all targets support 14895all types however. 14896 14897:: 14898 14899 declare float @llvm.round.f32(float %Val) 14900 declare double @llvm.round.f64(double %Val) 14901 declare x86_fp80 @llvm.round.f80(x86_fp80 %Val) 14902 declare fp128 @llvm.round.f128(fp128 %Val) 14903 declare ppc_fp128 @llvm.round.ppcf128(ppc_fp128 %Val) 14904 14905Overview: 14906""""""""" 14907 14908The '``llvm.round.*``' intrinsics returns the operand rounded to the 14909nearest integer. 14910 14911Arguments: 14912"""""""""" 14913 14914The argument and return value are floating-point numbers of the same 14915type. 14916 14917Semantics: 14918"""""""""" 14919 14920This function returns the same values as the libm ``round`` 14921functions would, and handles error conditions in the same way. 14922 14923'``llvm.roundeven.*``' Intrinsic 14924^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14925 14926Syntax: 14927""""""" 14928 14929This is an overloaded intrinsic. You can use ``llvm.roundeven`` on any 14930floating-point or vector of floating-point type. Not all targets support 14931all types however. 14932 14933:: 14934 14935 declare float @llvm.roundeven.f32(float %Val) 14936 declare double @llvm.roundeven.f64(double %Val) 14937 declare x86_fp80 @llvm.roundeven.f80(x86_fp80 %Val) 14938 declare fp128 @llvm.roundeven.f128(fp128 %Val) 14939 declare ppc_fp128 @llvm.roundeven.ppcf128(ppc_fp128 %Val) 14940 14941Overview: 14942""""""""" 14943 14944The '``llvm.roundeven.*``' intrinsics returns the operand rounded to the nearest 14945integer in floating-point format rounding halfway cases to even (that is, to the 14946nearest value that is an even integer). 14947 14948Arguments: 14949"""""""""" 14950 14951The argument and return value are floating-point numbers of the same type. 14952 14953Semantics: 14954"""""""""" 14955 14956This function implements IEEE-754 operation ``roundToIntegralTiesToEven``. It 14957also behaves in the same way as C standard function ``roundeven``, except that 14958it does not raise floating point exceptions. 14959 14960 14961'``llvm.lround.*``' Intrinsic 14962^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 14963 14964Syntax: 14965""""""" 14966 14967This is an overloaded intrinsic. You can use ``llvm.lround`` on any 14968floating-point type. Not all targets support all types however. 14969 14970:: 14971 14972 declare i32 @llvm.lround.i32.f32(float %Val) 14973 declare i32 @llvm.lround.i32.f64(double %Val) 14974 declare i32 @llvm.lround.i32.f80(float %Val) 14975 declare i32 @llvm.lround.i32.f128(double %Val) 14976 declare i32 @llvm.lround.i32.ppcf128(double %Val) 14977 14978 declare i64 @llvm.lround.i64.f32(float %Val) 14979 declare i64 @llvm.lround.i64.f64(double %Val) 14980 declare i64 @llvm.lround.i64.f80(float %Val) 14981 declare i64 @llvm.lround.i64.f128(double %Val) 14982 declare i64 @llvm.lround.i64.ppcf128(double %Val) 14983 14984Overview: 14985""""""""" 14986 14987The '``llvm.lround.*``' intrinsics return the operand rounded to the nearest 14988integer with ties away from zero. 14989 14990 14991Arguments: 14992"""""""""" 14993 14994The argument is a floating-point number and the return value is an integer 14995type. 14996 14997Semantics: 14998"""""""""" 14999 15000This function returns the same values as the libm ``lround`` 15001functions would, but without setting errno. 15002 15003'``llvm.llround.*``' Intrinsic 15004^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15005 15006Syntax: 15007""""""" 15008 15009This is an overloaded intrinsic. You can use ``llvm.llround`` on any 15010floating-point type. Not all targets support all types however. 15011 15012:: 15013 15014 declare i64 @llvm.lround.i64.f32(float %Val) 15015 declare i64 @llvm.lround.i64.f64(double %Val) 15016 declare i64 @llvm.lround.i64.f80(float %Val) 15017 declare i64 @llvm.lround.i64.f128(double %Val) 15018 declare i64 @llvm.lround.i64.ppcf128(double %Val) 15019 15020Overview: 15021""""""""" 15022 15023The '``llvm.llround.*``' intrinsics return the operand rounded to the nearest 15024integer with ties away from zero. 15025 15026Arguments: 15027"""""""""" 15028 15029The argument is a floating-point number and the return value is an integer 15030type. 15031 15032Semantics: 15033"""""""""" 15034 15035This function returns the same values as the libm ``llround`` 15036functions would, but without setting errno. 15037 15038'``llvm.lrint.*``' Intrinsic 15039^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15040 15041Syntax: 15042""""""" 15043 15044This is an overloaded intrinsic. You can use ``llvm.lrint`` on any 15045floating-point type. Not all targets support all types however. 15046 15047:: 15048 15049 declare i32 @llvm.lrint.i32.f32(float %Val) 15050 declare i32 @llvm.lrint.i32.f64(double %Val) 15051 declare i32 @llvm.lrint.i32.f80(float %Val) 15052 declare i32 @llvm.lrint.i32.f128(double %Val) 15053 declare i32 @llvm.lrint.i32.ppcf128(double %Val) 15054 15055 declare i64 @llvm.lrint.i64.f32(float %Val) 15056 declare i64 @llvm.lrint.i64.f64(double %Val) 15057 declare i64 @llvm.lrint.i64.f80(float %Val) 15058 declare i64 @llvm.lrint.i64.f128(double %Val) 15059 declare i64 @llvm.lrint.i64.ppcf128(double %Val) 15060 15061Overview: 15062""""""""" 15063 15064The '``llvm.lrint.*``' intrinsics return the operand rounded to the nearest 15065integer. 15066 15067 15068Arguments: 15069"""""""""" 15070 15071The argument is a floating-point number and the return value is an integer 15072type. 15073 15074Semantics: 15075"""""""""" 15076 15077This function returns the same values as the libm ``lrint`` 15078functions would, but without setting errno. 15079 15080'``llvm.llrint.*``' Intrinsic 15081^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15082 15083Syntax: 15084""""""" 15085 15086This is an overloaded intrinsic. You can use ``llvm.llrint`` on any 15087floating-point type. Not all targets support all types however. 15088 15089:: 15090 15091 declare i64 @llvm.llrint.i64.f32(float %Val) 15092 declare i64 @llvm.llrint.i64.f64(double %Val) 15093 declare i64 @llvm.llrint.i64.f80(float %Val) 15094 declare i64 @llvm.llrint.i64.f128(double %Val) 15095 declare i64 @llvm.llrint.i64.ppcf128(double %Val) 15096 15097Overview: 15098""""""""" 15099 15100The '``llvm.llrint.*``' intrinsics return the operand rounded to the nearest 15101integer. 15102 15103Arguments: 15104"""""""""" 15105 15106The argument is a floating-point number and the return value is an integer 15107type. 15108 15109Semantics: 15110"""""""""" 15111 15112This function returns the same values as the libm ``llrint`` 15113functions would, but without setting errno. 15114 15115Bit Manipulation Intrinsics 15116--------------------------- 15117 15118LLVM provides intrinsics for a few important bit manipulation 15119operations. These allow efficient code generation for some algorithms. 15120 15121'``llvm.bitreverse.*``' Intrinsics 15122^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15123 15124Syntax: 15125""""""" 15126 15127This is an overloaded intrinsic function. You can use bitreverse on any 15128integer type. 15129 15130:: 15131 15132 declare i16 @llvm.bitreverse.i16(i16 <id>) 15133 declare i32 @llvm.bitreverse.i32(i32 <id>) 15134 declare i64 @llvm.bitreverse.i64(i64 <id>) 15135 declare <4 x i32> @llvm.bitreverse.v4i32(<4 x i32> <id>) 15136 15137Overview: 15138""""""""" 15139 15140The '``llvm.bitreverse``' family of intrinsics is used to reverse the 15141bitpattern of an integer value or vector of integer values; for example 15142``0b10110110`` becomes ``0b01101101``. 15143 15144Semantics: 15145"""""""""" 15146 15147The ``llvm.bitreverse.iN`` intrinsic returns an iN value that has bit 15148``M`` in the input moved to bit ``N-M-1`` in the output. The vector 15149intrinsics, such as ``llvm.bitreverse.v4i32``, operate on a per-element 15150basis and the element order is not affected. 15151 15152'``llvm.bswap.*``' Intrinsics 15153^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15154 15155Syntax: 15156""""""" 15157 15158This is an overloaded intrinsic function. You can use bswap on any 15159integer type that is an even number of bytes (i.e. BitWidth % 16 == 0). 15160 15161:: 15162 15163 declare i16 @llvm.bswap.i16(i16 <id>) 15164 declare i32 @llvm.bswap.i32(i32 <id>) 15165 declare i64 @llvm.bswap.i64(i64 <id>) 15166 declare <4 x i32> @llvm.bswap.v4i32(<4 x i32> <id>) 15167 15168Overview: 15169""""""""" 15170 15171The '``llvm.bswap``' family of intrinsics is used to byte swap an integer 15172value or vector of integer values with an even number of bytes (positive 15173multiple of 16 bits). 15174 15175Semantics: 15176"""""""""" 15177 15178The ``llvm.bswap.i16`` intrinsic returns an i16 value that has the high 15179and low byte of the input i16 swapped. Similarly, the ``llvm.bswap.i32`` 15180intrinsic returns an i32 value that has the four bytes of the input i32 15181swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the 15182returned i32 will have its bytes in 3, 2, 1, 0 order. The 15183``llvm.bswap.i48``, ``llvm.bswap.i64`` and other intrinsics extend this 15184concept to additional even-byte lengths (6 bytes, 8 bytes and more, 15185respectively). The vector intrinsics, such as ``llvm.bswap.v4i32``, 15186operate on a per-element basis and the element order is not affected. 15187 15188'``llvm.ctpop.*``' Intrinsic 15189^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15190 15191Syntax: 15192""""""" 15193 15194This is an overloaded intrinsic. You can use llvm.ctpop on any integer 15195bit width, or on any vector with integer elements. Not all targets 15196support all bit widths or vector types, however. 15197 15198:: 15199 15200 declare i8 @llvm.ctpop.i8(i8 <src>) 15201 declare i16 @llvm.ctpop.i16(i16 <src>) 15202 declare i32 @llvm.ctpop.i32(i32 <src>) 15203 declare i64 @llvm.ctpop.i64(i64 <src>) 15204 declare i256 @llvm.ctpop.i256(i256 <src>) 15205 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>) 15206 15207Overview: 15208""""""""" 15209 15210The '``llvm.ctpop``' family of intrinsics counts the number of bits set 15211in a value. 15212 15213Arguments: 15214"""""""""" 15215 15216The only argument is the value to be counted. The argument may be of any 15217integer type, or a vector with integer elements. The return type must 15218match the argument type. 15219 15220Semantics: 15221"""""""""" 15222 15223The '``llvm.ctpop``' intrinsic counts the 1's in a variable, or within 15224each element of a vector. 15225 15226'``llvm.ctlz.*``' Intrinsic 15227^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15228 15229Syntax: 15230""""""" 15231 15232This is an overloaded intrinsic. You can use ``llvm.ctlz`` on any 15233integer bit width, or any vector whose elements are integers. Not all 15234targets support all bit widths or vector types, however. 15235 15236:: 15237 15238 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_poison>) 15239 declare <2 x i37> @llvm.ctlz.v2i37(<2 x i37> <src>, i1 <is_zero_poison>) 15240 15241Overview: 15242""""""""" 15243 15244The '``llvm.ctlz``' family of intrinsic functions counts the number of 15245leading zeros in a variable. 15246 15247Arguments: 15248"""""""""" 15249 15250The first argument is the value to be counted. This argument may be of 15251any integer type, or a vector with integer element type. The return 15252type must match the first argument type. 15253 15254The second argument is a constant flag that indicates whether the intrinsic 15255returns a valid result if the first argument is zero. If the first 15256argument is zero and the second argument is true, the result is poison. 15257Historically some architectures did not provide a defined result for zero 15258values as efficiently, and many algorithms are now predicated on avoiding 15259zero-value inputs. 15260 15261Semantics: 15262"""""""""" 15263 15264The '``llvm.ctlz``' intrinsic counts the leading (most significant) 15265zeros in a variable, or within each element of the vector. If 15266``src == 0`` then the result is the size in bits of the type of ``src`` 15267if ``is_zero_poison == 0`` and ``poison`` otherwise. For example, 15268``llvm.ctlz(i32 2) = 30``. 15269 15270'``llvm.cttz.*``' Intrinsic 15271^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15272 15273Syntax: 15274""""""" 15275 15276This is an overloaded intrinsic. You can use ``llvm.cttz`` on any 15277integer bit width, or any vector of integer elements. Not all targets 15278support all bit widths or vector types, however. 15279 15280:: 15281 15282 declare i42 @llvm.cttz.i42 (i42 <src>, i1 <is_zero_poison>) 15283 declare <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_poison>) 15284 15285Overview: 15286""""""""" 15287 15288The '``llvm.cttz``' family of intrinsic functions counts the number of 15289trailing zeros. 15290 15291Arguments: 15292"""""""""" 15293 15294The first argument is the value to be counted. This argument may be of 15295any integer type, or a vector with integer element type. The return 15296type must match the first argument type. 15297 15298The second argument is a constant flag that indicates whether the intrinsic 15299returns a valid result if the first argument is zero. If the first 15300argument is zero and the second argument is true, the result is poison. 15301Historically some architectures did not provide a defined result for zero 15302values as efficiently, and many algorithms are now predicated on avoiding 15303zero-value inputs. 15304 15305Semantics: 15306"""""""""" 15307 15308The '``llvm.cttz``' intrinsic counts the trailing (least significant) 15309zeros in a variable, or within each element of a vector. If ``src == 0`` 15310then the result is the size in bits of the type of ``src`` if 15311``is_zero_poison == 0`` and ``poison`` otherwise. For example, 15312``llvm.cttz(2) = 1``. 15313 15314.. _int_overflow: 15315 15316'``llvm.fshl.*``' Intrinsic 15317^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15318 15319Syntax: 15320""""""" 15321 15322This is an overloaded intrinsic. You can use ``llvm.fshl`` on any 15323integer bit width or any vector of integer elements. Not all targets 15324support all bit widths or vector types, however. 15325 15326:: 15327 15328 declare i8 @llvm.fshl.i8 (i8 %a, i8 %b, i8 %c) 15329 declare i67 @llvm.fshl.i67(i67 %a, i67 %b, i67 %c) 15330 declare <2 x i32> @llvm.fshl.v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c) 15331 15332Overview: 15333""""""""" 15334 15335The '``llvm.fshl``' family of intrinsic functions performs a funnel shift left: 15336the first two values are concatenated as { %a : %b } (%a is the most significant 15337bits of the wide value), the combined value is shifted left, and the most 15338significant bits are extracted to produce a result that is the same size as the 15339original arguments. If the first 2 arguments are identical, this is equivalent 15340to a rotate left operation. For vector types, the operation occurs for each 15341element of the vector. The shift argument is treated as an unsigned amount 15342modulo the element size of the arguments. 15343 15344Arguments: 15345"""""""""" 15346 15347The first two arguments are the values to be concatenated. The third 15348argument is the shift amount. The arguments may be any integer type or a 15349vector with integer element type. All arguments and the return value must 15350have the same type. 15351 15352Example: 15353"""""""" 15354 15355.. code-block:: text 15356 15357 %r = call i8 @llvm.fshl.i8(i8 %x, i8 %y, i8 %z) ; %r = i8: msb_extract((concat(x, y) << (z % 8)), 8) 15358 %r = call i8 @llvm.fshl.i8(i8 255, i8 0, i8 15) ; %r = i8: 128 (0b10000000) 15359 %r = call i8 @llvm.fshl.i8(i8 15, i8 15, i8 11) ; %r = i8: 120 (0b01111000) 15360 %r = call i8 @llvm.fshl.i8(i8 0, i8 255, i8 8) ; %r = i8: 0 (0b00000000) 15361 15362'``llvm.fshr.*``' Intrinsic 15363^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15364 15365Syntax: 15366""""""" 15367 15368This is an overloaded intrinsic. You can use ``llvm.fshr`` on any 15369integer bit width or any vector of integer elements. Not all targets 15370support all bit widths or vector types, however. 15371 15372:: 15373 15374 declare i8 @llvm.fshr.i8 (i8 %a, i8 %b, i8 %c) 15375 declare i67 @llvm.fshr.i67(i67 %a, i67 %b, i67 %c) 15376 declare <2 x i32> @llvm.fshr.v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c) 15377 15378Overview: 15379""""""""" 15380 15381The '``llvm.fshr``' family of intrinsic functions performs a funnel shift right: 15382the first two values are concatenated as { %a : %b } (%a is the most significant 15383bits of the wide value), the combined value is shifted right, and the least 15384significant bits are extracted to produce a result that is the same size as the 15385original arguments. If the first 2 arguments are identical, this is equivalent 15386to a rotate right operation. For vector types, the operation occurs for each 15387element of the vector. The shift argument is treated as an unsigned amount 15388modulo the element size of the arguments. 15389 15390Arguments: 15391"""""""""" 15392 15393The first two arguments are the values to be concatenated. The third 15394argument is the shift amount. The arguments may be any integer type or a 15395vector with integer element type. All arguments and the return value must 15396have the same type. 15397 15398Example: 15399"""""""" 15400 15401.. code-block:: text 15402 15403 %r = call i8 @llvm.fshr.i8(i8 %x, i8 %y, i8 %z) ; %r = i8: lsb_extract((concat(x, y) >> (z % 8)), 8) 15404 %r = call i8 @llvm.fshr.i8(i8 255, i8 0, i8 15) ; %r = i8: 254 (0b11111110) 15405 %r = call i8 @llvm.fshr.i8(i8 15, i8 15, i8 11) ; %r = i8: 225 (0b11100001) 15406 %r = call i8 @llvm.fshr.i8(i8 0, i8 255, i8 8) ; %r = i8: 255 (0b11111111) 15407 15408Arithmetic with Overflow Intrinsics 15409----------------------------------- 15410 15411LLVM provides intrinsics for fast arithmetic overflow checking. 15412 15413Each of these intrinsics returns a two-element struct. The first 15414element of this struct contains the result of the corresponding 15415arithmetic operation modulo 2\ :sup:`n`\ , where n is the bit width of 15416the result. Therefore, for example, the first element of the struct 15417returned by ``llvm.sadd.with.overflow.i32`` is always the same as the 15418result of a 32-bit ``add`` instruction with the same operands, where 15419the ``add`` is *not* modified by an ``nsw`` or ``nuw`` flag. 15420 15421The second element of the result is an ``i1`` that is 1 if the 15422arithmetic operation overflowed and 0 otherwise. An operation 15423overflows if, for any values of its operands ``A`` and ``B`` and for 15424any ``N`` larger than the operands' width, ``ext(A op B) to iN`` is 15425not equal to ``(ext(A) to iN) op (ext(B) to iN)`` where ``ext`` is 15426``sext`` for signed overflow and ``zext`` for unsigned overflow, and 15427``op`` is the underlying arithmetic operation. 15428 15429The behavior of these intrinsics is well-defined for all argument 15430values. 15431 15432'``llvm.sadd.with.overflow.*``' Intrinsics 15433^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15434 15435Syntax: 15436""""""" 15437 15438This is an overloaded intrinsic. You can use ``llvm.sadd.with.overflow`` 15439on any integer bit width or vectors of integers. 15440 15441:: 15442 15443 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b) 15444 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) 15445 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b) 15446 declare {<4 x i32>, <4 x i1>} @llvm.sadd.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15447 15448Overview: 15449""""""""" 15450 15451The '``llvm.sadd.with.overflow``' family of intrinsic functions perform 15452a signed addition of the two arguments, and indicate whether an overflow 15453occurred during the signed summation. 15454 15455Arguments: 15456"""""""""" 15457 15458The arguments (%a and %b) and the first element of the result structure 15459may be of integer types of any bit width, but they must have the same 15460bit width. The second element of the result structure must be of type 15461``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 15462addition. 15463 15464Semantics: 15465"""""""""" 15466 15467The '``llvm.sadd.with.overflow``' family of intrinsic functions perform 15468a signed addition of the two variables. They return a structure --- the 15469first element of which is the signed summation, and the second element 15470of which is a bit specifying if the signed summation resulted in an 15471overflow. 15472 15473Examples: 15474""""""""" 15475 15476.. code-block:: llvm 15477 15478 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b) 15479 %sum = extractvalue {i32, i1} %res, 0 15480 %obit = extractvalue {i32, i1} %res, 1 15481 br i1 %obit, label %overflow, label %normal 15482 15483'``llvm.uadd.with.overflow.*``' Intrinsics 15484^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15485 15486Syntax: 15487""""""" 15488 15489This is an overloaded intrinsic. You can use ``llvm.uadd.with.overflow`` 15490on any integer bit width or vectors of integers. 15491 15492:: 15493 15494 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b) 15495 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b) 15496 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b) 15497 declare {<4 x i32>, <4 x i1>} @llvm.uadd.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15498 15499Overview: 15500""""""""" 15501 15502The '``llvm.uadd.with.overflow``' family of intrinsic functions perform 15503an unsigned addition of the two arguments, and indicate whether a carry 15504occurred during the unsigned summation. 15505 15506Arguments: 15507"""""""""" 15508 15509The arguments (%a and %b) and the first element of the result structure 15510may be of integer types of any bit width, but they must have the same 15511bit width. The second element of the result structure must be of type 15512``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 15513addition. 15514 15515Semantics: 15516"""""""""" 15517 15518The '``llvm.uadd.with.overflow``' family of intrinsic functions perform 15519an unsigned addition of the two arguments. They return a structure --- the 15520first element of which is the sum, and the second element of which is a 15521bit specifying if the unsigned summation resulted in a carry. 15522 15523Examples: 15524""""""""" 15525 15526.. code-block:: llvm 15527 15528 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b) 15529 %sum = extractvalue {i32, i1} %res, 0 15530 %obit = extractvalue {i32, i1} %res, 1 15531 br i1 %obit, label %carry, label %normal 15532 15533'``llvm.ssub.with.overflow.*``' Intrinsics 15534^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15535 15536Syntax: 15537""""""" 15538 15539This is an overloaded intrinsic. You can use ``llvm.ssub.with.overflow`` 15540on any integer bit width or vectors of integers. 15541 15542:: 15543 15544 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b) 15545 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b) 15546 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b) 15547 declare {<4 x i32>, <4 x i1>} @llvm.ssub.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15548 15549Overview: 15550""""""""" 15551 15552The '``llvm.ssub.with.overflow``' family of intrinsic functions perform 15553a signed subtraction of the two arguments, and indicate whether an 15554overflow occurred during the signed subtraction. 15555 15556Arguments: 15557"""""""""" 15558 15559The arguments (%a and %b) and the first element of the result structure 15560may be of integer types of any bit width, but they must have the same 15561bit width. The second element of the result structure must be of type 15562``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 15563subtraction. 15564 15565Semantics: 15566"""""""""" 15567 15568The '``llvm.ssub.with.overflow``' family of intrinsic functions perform 15569a signed subtraction of the two arguments. They return a structure --- the 15570first element of which is the subtraction, and the second element of 15571which is a bit specifying if the signed subtraction resulted in an 15572overflow. 15573 15574Examples: 15575""""""""" 15576 15577.. code-block:: llvm 15578 15579 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b) 15580 %sum = extractvalue {i32, i1} %res, 0 15581 %obit = extractvalue {i32, i1} %res, 1 15582 br i1 %obit, label %overflow, label %normal 15583 15584'``llvm.usub.with.overflow.*``' Intrinsics 15585^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15586 15587Syntax: 15588""""""" 15589 15590This is an overloaded intrinsic. You can use ``llvm.usub.with.overflow`` 15591on any integer bit width or vectors of integers. 15592 15593:: 15594 15595 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b) 15596 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b) 15597 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b) 15598 declare {<4 x i32>, <4 x i1>} @llvm.usub.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15599 15600Overview: 15601""""""""" 15602 15603The '``llvm.usub.with.overflow``' family of intrinsic functions perform 15604an unsigned subtraction of the two arguments, and indicate whether an 15605overflow occurred during the unsigned subtraction. 15606 15607Arguments: 15608"""""""""" 15609 15610The arguments (%a and %b) and the first element of the result structure 15611may be of integer types of any bit width, but they must have the same 15612bit width. The second element of the result structure must be of type 15613``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 15614subtraction. 15615 15616Semantics: 15617"""""""""" 15618 15619The '``llvm.usub.with.overflow``' family of intrinsic functions perform 15620an unsigned subtraction of the two arguments. They return a structure --- 15621the first element of which is the subtraction, and the second element of 15622which is a bit specifying if the unsigned subtraction resulted in an 15623overflow. 15624 15625Examples: 15626""""""""" 15627 15628.. code-block:: llvm 15629 15630 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b) 15631 %sum = extractvalue {i32, i1} %res, 0 15632 %obit = extractvalue {i32, i1} %res, 1 15633 br i1 %obit, label %overflow, label %normal 15634 15635'``llvm.smul.with.overflow.*``' Intrinsics 15636^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15637 15638Syntax: 15639""""""" 15640 15641This is an overloaded intrinsic. You can use ``llvm.smul.with.overflow`` 15642on any integer bit width or vectors of integers. 15643 15644:: 15645 15646 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b) 15647 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b) 15648 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b) 15649 declare {<4 x i32>, <4 x i1>} @llvm.smul.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15650 15651Overview: 15652""""""""" 15653 15654The '``llvm.smul.with.overflow``' family of intrinsic functions perform 15655a signed multiplication of the two arguments, and indicate whether an 15656overflow occurred during the signed multiplication. 15657 15658Arguments: 15659"""""""""" 15660 15661The arguments (%a and %b) and the first element of the result structure 15662may be of integer types of any bit width, but they must have the same 15663bit width. The second element of the result structure must be of type 15664``i1``. ``%a`` and ``%b`` are the two values that will undergo signed 15665multiplication. 15666 15667Semantics: 15668"""""""""" 15669 15670The '``llvm.smul.with.overflow``' family of intrinsic functions perform 15671a signed multiplication of the two arguments. They return a structure --- 15672the first element of which is the multiplication, and the second element 15673of which is a bit specifying if the signed multiplication resulted in an 15674overflow. 15675 15676Examples: 15677""""""""" 15678 15679.. code-block:: llvm 15680 15681 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b) 15682 %sum = extractvalue {i32, i1} %res, 0 15683 %obit = extractvalue {i32, i1} %res, 1 15684 br i1 %obit, label %overflow, label %normal 15685 15686'``llvm.umul.with.overflow.*``' Intrinsics 15687^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15688 15689Syntax: 15690""""""" 15691 15692This is an overloaded intrinsic. You can use ``llvm.umul.with.overflow`` 15693on any integer bit width or vectors of integers. 15694 15695:: 15696 15697 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b) 15698 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b) 15699 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b) 15700 declare {<4 x i32>, <4 x i1>} @llvm.umul.with.overflow.v4i32(<4 x i32> %a, <4 x i32> %b) 15701 15702Overview: 15703""""""""" 15704 15705The '``llvm.umul.with.overflow``' family of intrinsic functions perform 15706a unsigned multiplication of the two arguments, and indicate whether an 15707overflow occurred during the unsigned multiplication. 15708 15709Arguments: 15710"""""""""" 15711 15712The arguments (%a and %b) and the first element of the result structure 15713may be of integer types of any bit width, but they must have the same 15714bit width. The second element of the result structure must be of type 15715``i1``. ``%a`` and ``%b`` are the two values that will undergo unsigned 15716multiplication. 15717 15718Semantics: 15719"""""""""" 15720 15721The '``llvm.umul.with.overflow``' family of intrinsic functions perform 15722an unsigned multiplication of the two arguments. They return a structure --- 15723the first element of which is the multiplication, and the second 15724element of which is a bit specifying if the unsigned multiplication 15725resulted in an overflow. 15726 15727Examples: 15728""""""""" 15729 15730.. code-block:: llvm 15731 15732 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b) 15733 %sum = extractvalue {i32, i1} %res, 0 15734 %obit = extractvalue {i32, i1} %res, 1 15735 br i1 %obit, label %overflow, label %normal 15736 15737Saturation Arithmetic Intrinsics 15738--------------------------------- 15739 15740Saturation arithmetic is a version of arithmetic in which operations are 15741limited to a fixed range between a minimum and maximum value. If the result of 15742an operation is greater than the maximum value, the result is set (or 15743"clamped") to this maximum. If it is below the minimum, it is clamped to this 15744minimum. 15745 15746 15747'``llvm.sadd.sat.*``' Intrinsics 15748^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15749 15750Syntax 15751""""""" 15752 15753This is an overloaded intrinsic. You can use ``llvm.sadd.sat`` 15754on any integer bit width or vectors of integers. 15755 15756:: 15757 15758 declare i16 @llvm.sadd.sat.i16(i16 %a, i16 %b) 15759 declare i32 @llvm.sadd.sat.i32(i32 %a, i32 %b) 15760 declare i64 @llvm.sadd.sat.i64(i64 %a, i64 %b) 15761 declare <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 15762 15763Overview 15764""""""""" 15765 15766The '``llvm.sadd.sat``' family of intrinsic functions perform signed 15767saturating addition on the 2 arguments. 15768 15769Arguments 15770"""""""""" 15771 15772The arguments (%a and %b) and the result may be of integer types of any bit 15773width, but they must have the same bit width. ``%a`` and ``%b`` are the two 15774values that will undergo signed addition. 15775 15776Semantics: 15777"""""""""" 15778 15779The maximum value this operation can clamp to is the largest signed value 15780representable by the bit width of the arguments. The minimum value is the 15781smallest signed value representable by this bit width. 15782 15783 15784Examples 15785""""""""" 15786 15787.. code-block:: llvm 15788 15789 %res = call i4 @llvm.sadd.sat.i4(i4 1, i4 2) ; %res = 3 15790 %res = call i4 @llvm.sadd.sat.i4(i4 5, i4 6) ; %res = 7 15791 %res = call i4 @llvm.sadd.sat.i4(i4 -4, i4 2) ; %res = -2 15792 %res = call i4 @llvm.sadd.sat.i4(i4 -4, i4 -5) ; %res = -8 15793 15794 15795'``llvm.uadd.sat.*``' Intrinsics 15796^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15797 15798Syntax 15799""""""" 15800 15801This is an overloaded intrinsic. You can use ``llvm.uadd.sat`` 15802on any integer bit width or vectors of integers. 15803 15804:: 15805 15806 declare i16 @llvm.uadd.sat.i16(i16 %a, i16 %b) 15807 declare i32 @llvm.uadd.sat.i32(i32 %a, i32 %b) 15808 declare i64 @llvm.uadd.sat.i64(i64 %a, i64 %b) 15809 declare <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 15810 15811Overview 15812""""""""" 15813 15814The '``llvm.uadd.sat``' family of intrinsic functions perform unsigned 15815saturating addition on the 2 arguments. 15816 15817Arguments 15818"""""""""" 15819 15820The arguments (%a and %b) and the result may be of integer types of any bit 15821width, but they must have the same bit width. ``%a`` and ``%b`` are the two 15822values that will undergo unsigned addition. 15823 15824Semantics: 15825"""""""""" 15826 15827The maximum value this operation can clamp to is the largest unsigned value 15828representable by the bit width of the arguments. Because this is an unsigned 15829operation, the result will never saturate towards zero. 15830 15831 15832Examples 15833""""""""" 15834 15835.. code-block:: llvm 15836 15837 %res = call i4 @llvm.uadd.sat.i4(i4 1, i4 2) ; %res = 3 15838 %res = call i4 @llvm.uadd.sat.i4(i4 5, i4 6) ; %res = 11 15839 %res = call i4 @llvm.uadd.sat.i4(i4 8, i4 8) ; %res = 15 15840 15841 15842'``llvm.ssub.sat.*``' Intrinsics 15843^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15844 15845Syntax 15846""""""" 15847 15848This is an overloaded intrinsic. You can use ``llvm.ssub.sat`` 15849on any integer bit width or vectors of integers. 15850 15851:: 15852 15853 declare i16 @llvm.ssub.sat.i16(i16 %a, i16 %b) 15854 declare i32 @llvm.ssub.sat.i32(i32 %a, i32 %b) 15855 declare i64 @llvm.ssub.sat.i64(i64 %a, i64 %b) 15856 declare <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 15857 15858Overview 15859""""""""" 15860 15861The '``llvm.ssub.sat``' family of intrinsic functions perform signed 15862saturating subtraction on the 2 arguments. 15863 15864Arguments 15865"""""""""" 15866 15867The arguments (%a and %b) and the result may be of integer types of any bit 15868width, but they must have the same bit width. ``%a`` and ``%b`` are the two 15869values that will undergo signed subtraction. 15870 15871Semantics: 15872"""""""""" 15873 15874The maximum value this operation can clamp to is the largest signed value 15875representable by the bit width of the arguments. The minimum value is the 15876smallest signed value representable by this bit width. 15877 15878 15879Examples 15880""""""""" 15881 15882.. code-block:: llvm 15883 15884 %res = call i4 @llvm.ssub.sat.i4(i4 2, i4 1) ; %res = 1 15885 %res = call i4 @llvm.ssub.sat.i4(i4 2, i4 6) ; %res = -4 15886 %res = call i4 @llvm.ssub.sat.i4(i4 -4, i4 5) ; %res = -8 15887 %res = call i4 @llvm.ssub.sat.i4(i4 4, i4 -5) ; %res = 7 15888 15889 15890'``llvm.usub.sat.*``' Intrinsics 15891^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15892 15893Syntax 15894""""""" 15895 15896This is an overloaded intrinsic. You can use ``llvm.usub.sat`` 15897on any integer bit width or vectors of integers. 15898 15899:: 15900 15901 declare i16 @llvm.usub.sat.i16(i16 %a, i16 %b) 15902 declare i32 @llvm.usub.sat.i32(i32 %a, i32 %b) 15903 declare i64 @llvm.usub.sat.i64(i64 %a, i64 %b) 15904 declare <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 15905 15906Overview 15907""""""""" 15908 15909The '``llvm.usub.sat``' family of intrinsic functions perform unsigned 15910saturating subtraction on the 2 arguments. 15911 15912Arguments 15913"""""""""" 15914 15915The arguments (%a and %b) and the result may be of integer types of any bit 15916width, but they must have the same bit width. ``%a`` and ``%b`` are the two 15917values that will undergo unsigned subtraction. 15918 15919Semantics: 15920"""""""""" 15921 15922The minimum value this operation can clamp to is 0, which is the smallest 15923unsigned value representable by the bit width of the unsigned arguments. 15924Because this is an unsigned operation, the result will never saturate towards 15925the largest possible value representable by this bit width. 15926 15927 15928Examples 15929""""""""" 15930 15931.. code-block:: llvm 15932 15933 %res = call i4 @llvm.usub.sat.i4(i4 2, i4 1) ; %res = 1 15934 %res = call i4 @llvm.usub.sat.i4(i4 2, i4 6) ; %res = 0 15935 15936 15937'``llvm.sshl.sat.*``' Intrinsics 15938^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15939 15940Syntax 15941""""""" 15942 15943This is an overloaded intrinsic. You can use ``llvm.sshl.sat`` 15944on integers or vectors of integers of any bit width. 15945 15946:: 15947 15948 declare i16 @llvm.sshl.sat.i16(i16 %a, i16 %b) 15949 declare i32 @llvm.sshl.sat.i32(i32 %a, i32 %b) 15950 declare i64 @llvm.sshl.sat.i64(i64 %a, i64 %b) 15951 declare <4 x i32> @llvm.sshl.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 15952 15953Overview 15954""""""""" 15955 15956The '``llvm.sshl.sat``' family of intrinsic functions perform signed 15957saturating left shift on the first argument. 15958 15959Arguments 15960"""""""""" 15961 15962The arguments (``%a`` and ``%b``) and the result may be of integer types of any 15963bit width, but they must have the same bit width. ``%a`` is the value to be 15964shifted, and ``%b`` is the amount to shift by. If ``b`` is (statically or 15965dynamically) equal to or larger than the integer bit width of the arguments, 15966the result is a :ref:`poison value <poisonvalues>`. If the arguments are 15967vectors, each vector element of ``a`` is shifted by the corresponding shift 15968amount in ``b``. 15969 15970 15971Semantics: 15972"""""""""" 15973 15974The maximum value this operation can clamp to is the largest signed value 15975representable by the bit width of the arguments. The minimum value is the 15976smallest signed value representable by this bit width. 15977 15978 15979Examples 15980""""""""" 15981 15982.. code-block:: llvm 15983 15984 %res = call i4 @llvm.sshl.sat.i4(i4 2, i4 1) ; %res = 4 15985 %res = call i4 @llvm.sshl.sat.i4(i4 2, i4 2) ; %res = 7 15986 %res = call i4 @llvm.sshl.sat.i4(i4 -5, i4 1) ; %res = -8 15987 %res = call i4 @llvm.sshl.sat.i4(i4 -1, i4 1) ; %res = -2 15988 15989 15990'``llvm.ushl.sat.*``' Intrinsics 15991^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 15992 15993Syntax 15994""""""" 15995 15996This is an overloaded intrinsic. You can use ``llvm.ushl.sat`` 15997on integers or vectors of integers of any bit width. 15998 15999:: 16000 16001 declare i16 @llvm.ushl.sat.i16(i16 %a, i16 %b) 16002 declare i32 @llvm.ushl.sat.i32(i32 %a, i32 %b) 16003 declare i64 @llvm.ushl.sat.i64(i64 %a, i64 %b) 16004 declare <4 x i32> @llvm.ushl.sat.v4i32(<4 x i32> %a, <4 x i32> %b) 16005 16006Overview 16007""""""""" 16008 16009The '``llvm.ushl.sat``' family of intrinsic functions perform unsigned 16010saturating left shift on the first argument. 16011 16012Arguments 16013"""""""""" 16014 16015The arguments (``%a`` and ``%b``) and the result may be of integer types of any 16016bit width, but they must have the same bit width. ``%a`` is the value to be 16017shifted, and ``%b`` is the amount to shift by. If ``b`` is (statically or 16018dynamically) equal to or larger than the integer bit width of the arguments, 16019the result is a :ref:`poison value <poisonvalues>`. If the arguments are 16020vectors, each vector element of ``a`` is shifted by the corresponding shift 16021amount in ``b``. 16022 16023Semantics: 16024"""""""""" 16025 16026The maximum value this operation can clamp to is the largest unsigned value 16027representable by the bit width of the arguments. 16028 16029 16030Examples 16031""""""""" 16032 16033.. code-block:: llvm 16034 16035 %res = call i4 @llvm.ushl.sat.i4(i4 2, i4 1) ; %res = 4 16036 %res = call i4 @llvm.ushl.sat.i4(i4 3, i4 3) ; %res = 15 16037 16038 16039Fixed Point Arithmetic Intrinsics 16040--------------------------------- 16041 16042A fixed point number represents a real data type for a number that has a fixed 16043number of digits after a radix point (equivalent to the decimal point '.'). 16044The number of digits after the radix point is referred as the `scale`. These 16045are useful for representing fractional values to a specific precision. The 16046following intrinsics perform fixed point arithmetic operations on 2 operands 16047of the same scale, specified as the third argument. 16048 16049The ``llvm.*mul.fix`` family of intrinsic functions represents a multiplication 16050of fixed point numbers through scaled integers. Therefore, fixed point 16051multiplication can be represented as 16052 16053.. code-block:: llvm 16054 16055 %result = call i4 @llvm.smul.fix.i4(i4 %a, i4 %b, i32 %scale) 16056 16057 ; Expands to 16058 %a2 = sext i4 %a to i8 16059 %b2 = sext i4 %b to i8 16060 %mul = mul nsw nuw i8 %a2, %b2 16061 %scale2 = trunc i32 %scale to i8 16062 %r = ashr i8 %mul, i8 %scale2 ; this is for a target rounding down towards negative infinity 16063 %result = trunc i8 %r to i4 16064 16065The ``llvm.*div.fix`` family of intrinsic functions represents a division of 16066fixed point numbers through scaled integers. Fixed point division can be 16067represented as: 16068 16069.. code-block:: llvm 16070 16071 %result call i4 @llvm.sdiv.fix.i4(i4 %a, i4 %b, i32 %scale) 16072 16073 ; Expands to 16074 %a2 = sext i4 %a to i8 16075 %b2 = sext i4 %b to i8 16076 %scale2 = trunc i32 %scale to i8 16077 %a3 = shl i8 %a2, %scale2 16078 %r = sdiv i8 %a3, %b2 ; this is for a target rounding towards zero 16079 %result = trunc i8 %r to i4 16080 16081For each of these functions, if the result cannot be represented exactly with 16082the provided scale, the result is rounded. Rounding is unspecified since 16083preferred rounding may vary for different targets. Rounding is specified 16084through a target hook. Different pipelines should legalize or optimize this 16085using the rounding specified by this hook if it is provided. Operations like 16086constant folding, instruction combining, KnownBits, and ValueTracking should 16087also use this hook, if provided, and not assume the direction of rounding. A 16088rounded result must always be within one unit of precision from the true 16089result. That is, the error between the returned result and the true result must 16090be less than 1/2^(scale). 16091 16092 16093'``llvm.smul.fix.*``' Intrinsics 16094^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16095 16096Syntax 16097""""""" 16098 16099This is an overloaded intrinsic. You can use ``llvm.smul.fix`` 16100on any integer bit width or vectors of integers. 16101 16102:: 16103 16104 declare i16 @llvm.smul.fix.i16(i16 %a, i16 %b, i32 %scale) 16105 declare i32 @llvm.smul.fix.i32(i32 %a, i32 %b, i32 %scale) 16106 declare i64 @llvm.smul.fix.i64(i64 %a, i64 %b, i32 %scale) 16107 declare <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16108 16109Overview 16110""""""""" 16111 16112The '``llvm.smul.fix``' family of intrinsic functions perform signed 16113fixed point multiplication on 2 arguments of the same scale. 16114 16115Arguments 16116"""""""""" 16117 16118The arguments (%a and %b) and the result may be of integer types of any bit 16119width, but they must have the same bit width. The arguments may also work with 16120int vectors of the same length and int size. ``%a`` and ``%b`` are the two 16121values that will undergo signed fixed point multiplication. The argument 16122``%scale`` represents the scale of both operands, and must be a constant 16123integer. 16124 16125Semantics: 16126"""""""""" 16127 16128This operation performs fixed point multiplication on the 2 arguments of a 16129specified scale. The result will also be returned in the same scale specified 16130in the third argument. 16131 16132If the result value cannot be precisely represented in the given scale, the 16133value is rounded up or down to the closest representable value. The rounding 16134direction is unspecified. 16135 16136It is undefined behavior if the result value does not fit within the range of 16137the fixed point type. 16138 16139 16140Examples 16141""""""""" 16142 16143.. code-block:: llvm 16144 16145 %res = call i4 @llvm.smul.fix.i4(i4 3, i4 2, i32 0) ; %res = 6 (2 x 3 = 6) 16146 %res = call i4 @llvm.smul.fix.i4(i4 3, i4 2, i32 1) ; %res = 3 (1.5 x 1 = 1.5) 16147 %res = call i4 @llvm.smul.fix.i4(i4 3, i4 -2, i32 1) ; %res = -3 (1.5 x -1 = -1.5) 16148 16149 ; The result in the following could be rounded up to -2 or down to -2.5 16150 %res = call i4 @llvm.smul.fix.i4(i4 3, i4 -3, i32 1) ; %res = -5 (or -4) (1.5 x -1.5 = -2.25) 16151 16152 16153'``llvm.umul.fix.*``' Intrinsics 16154^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16155 16156Syntax 16157""""""" 16158 16159This is an overloaded intrinsic. You can use ``llvm.umul.fix`` 16160on any integer bit width or vectors of integers. 16161 16162:: 16163 16164 declare i16 @llvm.umul.fix.i16(i16 %a, i16 %b, i32 %scale) 16165 declare i32 @llvm.umul.fix.i32(i32 %a, i32 %b, i32 %scale) 16166 declare i64 @llvm.umul.fix.i64(i64 %a, i64 %b, i32 %scale) 16167 declare <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16168 16169Overview 16170""""""""" 16171 16172The '``llvm.umul.fix``' family of intrinsic functions perform unsigned 16173fixed point multiplication on 2 arguments of the same scale. 16174 16175Arguments 16176"""""""""" 16177 16178The arguments (%a and %b) and the result may be of integer types of any bit 16179width, but they must have the same bit width. The arguments may also work with 16180int vectors of the same length and int size. ``%a`` and ``%b`` are the two 16181values that will undergo unsigned fixed point multiplication. The argument 16182``%scale`` represents the scale of both operands, and must be a constant 16183integer. 16184 16185Semantics: 16186"""""""""" 16187 16188This operation performs unsigned fixed point multiplication on the 2 arguments of a 16189specified scale. The result will also be returned in the same scale specified 16190in the third argument. 16191 16192If the result value cannot be precisely represented in the given scale, the 16193value is rounded up or down to the closest representable value. The rounding 16194direction is unspecified. 16195 16196It is undefined behavior if the result value does not fit within the range of 16197the fixed point type. 16198 16199 16200Examples 16201""""""""" 16202 16203.. code-block:: llvm 16204 16205 %res = call i4 @llvm.umul.fix.i4(i4 3, i4 2, i32 0) ; %res = 6 (2 x 3 = 6) 16206 %res = call i4 @llvm.umul.fix.i4(i4 3, i4 2, i32 1) ; %res = 3 (1.5 x 1 = 1.5) 16207 16208 ; The result in the following could be rounded down to 3.5 or up to 4 16209 %res = call i4 @llvm.umul.fix.i4(i4 15, i4 1, i32 1) ; %res = 7 (or 8) (7.5 x 0.5 = 3.75) 16210 16211 16212'``llvm.smul.fix.sat.*``' Intrinsics 16213^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16214 16215Syntax 16216""""""" 16217 16218This is an overloaded intrinsic. You can use ``llvm.smul.fix.sat`` 16219on any integer bit width or vectors of integers. 16220 16221:: 16222 16223 declare i16 @llvm.smul.fix.sat.i16(i16 %a, i16 %b, i32 %scale) 16224 declare i32 @llvm.smul.fix.sat.i32(i32 %a, i32 %b, i32 %scale) 16225 declare i64 @llvm.smul.fix.sat.i64(i64 %a, i64 %b, i32 %scale) 16226 declare <4 x i32> @llvm.smul.fix.sat.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16227 16228Overview 16229""""""""" 16230 16231The '``llvm.smul.fix.sat``' family of intrinsic functions perform signed 16232fixed point saturating multiplication on 2 arguments of the same scale. 16233 16234Arguments 16235"""""""""" 16236 16237The arguments (%a and %b) and the result may be of integer types of any bit 16238width, but they must have the same bit width. ``%a`` and ``%b`` are the two 16239values that will undergo signed fixed point multiplication. The argument 16240``%scale`` represents the scale of both operands, and must be a constant 16241integer. 16242 16243Semantics: 16244"""""""""" 16245 16246This operation performs fixed point multiplication on the 2 arguments of a 16247specified scale. The result will also be returned in the same scale specified 16248in the third argument. 16249 16250If the result value cannot be precisely represented in the given scale, the 16251value is rounded up or down to the closest representable value. The rounding 16252direction is unspecified. 16253 16254The maximum value this operation can clamp to is the largest signed value 16255representable by the bit width of the first 2 arguments. The minimum value is the 16256smallest signed value representable by this bit width. 16257 16258 16259Examples 16260""""""""" 16261 16262.. code-block:: llvm 16263 16264 %res = call i4 @llvm.smul.fix.sat.i4(i4 3, i4 2, i32 0) ; %res = 6 (2 x 3 = 6) 16265 %res = call i4 @llvm.smul.fix.sat.i4(i4 3, i4 2, i32 1) ; %res = 3 (1.5 x 1 = 1.5) 16266 %res = call i4 @llvm.smul.fix.sat.i4(i4 3, i4 -2, i32 1) ; %res = -3 (1.5 x -1 = -1.5) 16267 16268 ; The result in the following could be rounded up to -2 or down to -2.5 16269 %res = call i4 @llvm.smul.fix.sat.i4(i4 3, i4 -3, i32 1) ; %res = -5 (or -4) (1.5 x -1.5 = -2.25) 16270 16271 ; Saturation 16272 %res = call i4 @llvm.smul.fix.sat.i4(i4 7, i4 2, i32 0) ; %res = 7 16273 %res = call i4 @llvm.smul.fix.sat.i4(i4 7, i4 4, i32 2) ; %res = 7 16274 %res = call i4 @llvm.smul.fix.sat.i4(i4 -8, i4 5, i32 2) ; %res = -8 16275 %res = call i4 @llvm.smul.fix.sat.i4(i4 -8, i4 -2, i32 1) ; %res = 7 16276 16277 ; Scale can affect the saturation result 16278 %res = call i4 @llvm.smul.fix.sat.i4(i4 2, i4 4, i32 0) ; %res = 7 (2 x 4 -> clamped to 7) 16279 %res = call i4 @llvm.smul.fix.sat.i4(i4 2, i4 4, i32 1) ; %res = 4 (1 x 2 = 2) 16280 16281 16282'``llvm.umul.fix.sat.*``' Intrinsics 16283^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16284 16285Syntax 16286""""""" 16287 16288This is an overloaded intrinsic. You can use ``llvm.umul.fix.sat`` 16289on any integer bit width or vectors of integers. 16290 16291:: 16292 16293 declare i16 @llvm.umul.fix.sat.i16(i16 %a, i16 %b, i32 %scale) 16294 declare i32 @llvm.umul.fix.sat.i32(i32 %a, i32 %b, i32 %scale) 16295 declare i64 @llvm.umul.fix.sat.i64(i64 %a, i64 %b, i32 %scale) 16296 declare <4 x i32> @llvm.umul.fix.sat.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16297 16298Overview 16299""""""""" 16300 16301The '``llvm.umul.fix.sat``' family of intrinsic functions perform unsigned 16302fixed point saturating multiplication on 2 arguments of the same scale. 16303 16304Arguments 16305"""""""""" 16306 16307The arguments (%a and %b) and the result may be of integer types of any bit 16308width, but they must have the same bit width. ``%a`` and ``%b`` are the two 16309values that will undergo unsigned fixed point multiplication. The argument 16310``%scale`` represents the scale of both operands, and must be a constant 16311integer. 16312 16313Semantics: 16314"""""""""" 16315 16316This operation performs fixed point multiplication on the 2 arguments of a 16317specified scale. The result will also be returned in the same scale specified 16318in the third argument. 16319 16320If the result value cannot be precisely represented in the given scale, the 16321value is rounded up or down to the closest representable value. The rounding 16322direction is unspecified. 16323 16324The maximum value this operation can clamp to is the largest unsigned value 16325representable by the bit width of the first 2 arguments. The minimum value is the 16326smallest unsigned value representable by this bit width (zero). 16327 16328 16329Examples 16330""""""""" 16331 16332.. code-block:: llvm 16333 16334 %res = call i4 @llvm.umul.fix.sat.i4(i4 3, i4 2, i32 0) ; %res = 6 (2 x 3 = 6) 16335 %res = call i4 @llvm.umul.fix.sat.i4(i4 3, i4 2, i32 1) ; %res = 3 (1.5 x 1 = 1.5) 16336 16337 ; The result in the following could be rounded down to 2 or up to 2.5 16338 %res = call i4 @llvm.umul.fix.sat.i4(i4 3, i4 3, i32 1) ; %res = 4 (or 5) (1.5 x 1.5 = 2.25) 16339 16340 ; Saturation 16341 %res = call i4 @llvm.umul.fix.sat.i4(i4 8, i4 2, i32 0) ; %res = 15 (8 x 2 -> clamped to 15) 16342 %res = call i4 @llvm.umul.fix.sat.i4(i4 8, i4 8, i32 2) ; %res = 15 (2 x 2 -> clamped to 3.75) 16343 16344 ; Scale can affect the saturation result 16345 %res = call i4 @llvm.umul.fix.sat.i4(i4 2, i4 4, i32 0) ; %res = 7 (2 x 4 -> clamped to 7) 16346 %res = call i4 @llvm.umul.fix.sat.i4(i4 2, i4 4, i32 1) ; %res = 4 (1 x 2 = 2) 16347 16348 16349'``llvm.sdiv.fix.*``' Intrinsics 16350^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16351 16352Syntax 16353""""""" 16354 16355This is an overloaded intrinsic. You can use ``llvm.sdiv.fix`` 16356on any integer bit width or vectors of integers. 16357 16358:: 16359 16360 declare i16 @llvm.sdiv.fix.i16(i16 %a, i16 %b, i32 %scale) 16361 declare i32 @llvm.sdiv.fix.i32(i32 %a, i32 %b, i32 %scale) 16362 declare i64 @llvm.sdiv.fix.i64(i64 %a, i64 %b, i32 %scale) 16363 declare <4 x i32> @llvm.sdiv.fix.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16364 16365Overview 16366""""""""" 16367 16368The '``llvm.sdiv.fix``' family of intrinsic functions perform signed 16369fixed point division on 2 arguments of the same scale. 16370 16371Arguments 16372"""""""""" 16373 16374The arguments (%a and %b) and the result may be of integer types of any bit 16375width, but they must have the same bit width. The arguments may also work with 16376int vectors of the same length and int size. ``%a`` and ``%b`` are the two 16377values that will undergo signed fixed point division. The argument 16378``%scale`` represents the scale of both operands, and must be a constant 16379integer. 16380 16381Semantics: 16382"""""""""" 16383 16384This operation performs fixed point division on the 2 arguments of a 16385specified scale. The result will also be returned in the same scale specified 16386in the third argument. 16387 16388If the result value cannot be precisely represented in the given scale, the 16389value is rounded up or down to the closest representable value. The rounding 16390direction is unspecified. 16391 16392It is undefined behavior if the result value does not fit within the range of 16393the fixed point type, or if the second argument is zero. 16394 16395 16396Examples 16397""""""""" 16398 16399.. code-block:: llvm 16400 16401 %res = call i4 @llvm.sdiv.fix.i4(i4 6, i4 2, i32 0) ; %res = 3 (6 / 2 = 3) 16402 %res = call i4 @llvm.sdiv.fix.i4(i4 6, i4 4, i32 1) ; %res = 3 (3 / 2 = 1.5) 16403 %res = call i4 @llvm.sdiv.fix.i4(i4 3, i4 -2, i32 1) ; %res = -3 (1.5 / -1 = -1.5) 16404 16405 ; The result in the following could be rounded up to 1 or down to 0.5 16406 %res = call i4 @llvm.sdiv.fix.i4(i4 3, i4 4, i32 1) ; %res = 2 (or 1) (1.5 / 2 = 0.75) 16407 16408 16409'``llvm.udiv.fix.*``' Intrinsics 16410^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16411 16412Syntax 16413""""""" 16414 16415This is an overloaded intrinsic. You can use ``llvm.udiv.fix`` 16416on any integer bit width or vectors of integers. 16417 16418:: 16419 16420 declare i16 @llvm.udiv.fix.i16(i16 %a, i16 %b, i32 %scale) 16421 declare i32 @llvm.udiv.fix.i32(i32 %a, i32 %b, i32 %scale) 16422 declare i64 @llvm.udiv.fix.i64(i64 %a, i64 %b, i32 %scale) 16423 declare <4 x i32> @llvm.udiv.fix.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16424 16425Overview 16426""""""""" 16427 16428The '``llvm.udiv.fix``' family of intrinsic functions perform unsigned 16429fixed point division on 2 arguments of the same scale. 16430 16431Arguments 16432"""""""""" 16433 16434The arguments (%a and %b) and the result may be of integer types of any bit 16435width, but they must have the same bit width. The arguments may also work with 16436int vectors of the same length and int size. ``%a`` and ``%b`` are the two 16437values that will undergo unsigned fixed point division. The argument 16438``%scale`` represents the scale of both operands, and must be a constant 16439integer. 16440 16441Semantics: 16442"""""""""" 16443 16444This operation performs fixed point division on the 2 arguments of a 16445specified scale. The result will also be returned in the same scale specified 16446in the third argument. 16447 16448If the result value cannot be precisely represented in the given scale, the 16449value is rounded up or down to the closest representable value. The rounding 16450direction is unspecified. 16451 16452It is undefined behavior if the result value does not fit within the range of 16453the fixed point type, or if the second argument is zero. 16454 16455 16456Examples 16457""""""""" 16458 16459.. code-block:: llvm 16460 16461 %res = call i4 @llvm.udiv.fix.i4(i4 6, i4 2, i32 0) ; %res = 3 (6 / 2 = 3) 16462 %res = call i4 @llvm.udiv.fix.i4(i4 6, i4 4, i32 1) ; %res = 3 (3 / 2 = 1.5) 16463 %res = call i4 @llvm.udiv.fix.i4(i4 1, i4 -8, i32 4) ; %res = 2 (0.0625 / 0.5 = 0.125) 16464 16465 ; The result in the following could be rounded up to 1 or down to 0.5 16466 %res = call i4 @llvm.udiv.fix.i4(i4 3, i4 4, i32 1) ; %res = 2 (or 1) (1.5 / 2 = 0.75) 16467 16468 16469'``llvm.sdiv.fix.sat.*``' Intrinsics 16470^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16471 16472Syntax 16473""""""" 16474 16475This is an overloaded intrinsic. You can use ``llvm.sdiv.fix.sat`` 16476on any integer bit width or vectors of integers. 16477 16478:: 16479 16480 declare i16 @llvm.sdiv.fix.sat.i16(i16 %a, i16 %b, i32 %scale) 16481 declare i32 @llvm.sdiv.fix.sat.i32(i32 %a, i32 %b, i32 %scale) 16482 declare i64 @llvm.sdiv.fix.sat.i64(i64 %a, i64 %b, i32 %scale) 16483 declare <4 x i32> @llvm.sdiv.fix.sat.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16484 16485Overview 16486""""""""" 16487 16488The '``llvm.sdiv.fix.sat``' family of intrinsic functions perform signed 16489fixed point saturating division on 2 arguments of the same scale. 16490 16491Arguments 16492"""""""""" 16493 16494The arguments (%a and %b) and the result may be of integer types of any bit 16495width, but they must have the same bit width. ``%a`` and ``%b`` are the two 16496values that will undergo signed fixed point division. The argument 16497``%scale`` represents the scale of both operands, and must be a constant 16498integer. 16499 16500Semantics: 16501"""""""""" 16502 16503This operation performs fixed point division on the 2 arguments of a 16504specified scale. The result will also be returned in the same scale specified 16505in the third argument. 16506 16507If the result value cannot be precisely represented in the given scale, the 16508value is rounded up or down to the closest representable value. The rounding 16509direction is unspecified. 16510 16511The maximum value this operation can clamp to is the largest signed value 16512representable by the bit width of the first 2 arguments. The minimum value is the 16513smallest signed value representable by this bit width. 16514 16515It is undefined behavior if the second argument is zero. 16516 16517 16518Examples 16519""""""""" 16520 16521.. code-block:: llvm 16522 16523 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 6, i4 2, i32 0) ; %res = 3 (6 / 2 = 3) 16524 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 6, i4 4, i32 1) ; %res = 3 (3 / 2 = 1.5) 16525 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 3, i4 -2, i32 1) ; %res = -3 (1.5 / -1 = -1.5) 16526 16527 ; The result in the following could be rounded up to 1 or down to 0.5 16528 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 3, i4 4, i32 1) ; %res = 2 (or 1) (1.5 / 2 = 0.75) 16529 16530 ; Saturation 16531 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 -8, i4 -1, i32 0) ; %res = 7 (-8 / -1 = 8 => 7) 16532 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 4, i4 2, i32 2) ; %res = 7 (1 / 0.5 = 2 => 1.75) 16533 %res = call i4 @llvm.sdiv.fix.sat.i4(i4 -4, i4 1, i32 2) ; %res = -8 (-1 / 0.25 = -4 => -2) 16534 16535 16536'``llvm.udiv.fix.sat.*``' Intrinsics 16537^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16538 16539Syntax 16540""""""" 16541 16542This is an overloaded intrinsic. You can use ``llvm.udiv.fix.sat`` 16543on any integer bit width or vectors of integers. 16544 16545:: 16546 16547 declare i16 @llvm.udiv.fix.sat.i16(i16 %a, i16 %b, i32 %scale) 16548 declare i32 @llvm.udiv.fix.sat.i32(i32 %a, i32 %b, i32 %scale) 16549 declare i64 @llvm.udiv.fix.sat.i64(i64 %a, i64 %b, i32 %scale) 16550 declare <4 x i32> @llvm.udiv.fix.sat.v4i32(<4 x i32> %a, <4 x i32> %b, i32 %scale) 16551 16552Overview 16553""""""""" 16554 16555The '``llvm.udiv.fix.sat``' family of intrinsic functions perform unsigned 16556fixed point saturating division on 2 arguments of the same scale. 16557 16558Arguments 16559"""""""""" 16560 16561The arguments (%a and %b) and the result may be of integer types of any bit 16562width, but they must have the same bit width. ``%a`` and ``%b`` are the two 16563values that will undergo unsigned fixed point division. The argument 16564``%scale`` represents the scale of both operands, and must be a constant 16565integer. 16566 16567Semantics: 16568"""""""""" 16569 16570This operation performs fixed point division on the 2 arguments of a 16571specified scale. The result will also be returned in the same scale specified 16572in the third argument. 16573 16574If the result value cannot be precisely represented in the given scale, the 16575value is rounded up or down to the closest representable value. The rounding 16576direction is unspecified. 16577 16578The maximum value this operation can clamp to is the largest unsigned value 16579representable by the bit width of the first 2 arguments. The minimum value is the 16580smallest unsigned value representable by this bit width (zero). 16581 16582It is undefined behavior if the second argument is zero. 16583 16584Examples 16585""""""""" 16586 16587.. code-block:: llvm 16588 16589 %res = call i4 @llvm.udiv.fix.sat.i4(i4 6, i4 2, i32 0) ; %res = 3 (6 / 2 = 3) 16590 %res = call i4 @llvm.udiv.fix.sat.i4(i4 6, i4 4, i32 1) ; %res = 3 (3 / 2 = 1.5) 16591 16592 ; The result in the following could be rounded down to 0.5 or up to 1 16593 %res = call i4 @llvm.udiv.fix.sat.i4(i4 3, i4 4, i32 1) ; %res = 1 (or 2) (1.5 / 2 = 0.75) 16594 16595 ; Saturation 16596 %res = call i4 @llvm.udiv.fix.sat.i4(i4 8, i4 2, i32 2) ; %res = 15 (2 / 0.5 = 4 => 3.75) 16597 16598 16599Specialised Arithmetic Intrinsics 16600--------------------------------- 16601 16602.. _i_intr_llvm_canonicalize: 16603 16604'``llvm.canonicalize.*``' Intrinsic 16605^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16606 16607Syntax: 16608""""""" 16609 16610:: 16611 16612 declare float @llvm.canonicalize.f32(float %a) 16613 declare double @llvm.canonicalize.f64(double %b) 16614 16615Overview: 16616""""""""" 16617 16618The '``llvm.canonicalize.*``' intrinsic returns the platform specific canonical 16619encoding of a floating-point number. This canonicalization is useful for 16620implementing certain numeric primitives such as frexp. The canonical encoding is 16621defined by IEEE-754-2008 to be: 16622 16623:: 16624 16625 2.1.8 canonical encoding: The preferred encoding of a floating-point 16626 representation in a format. Applied to declets, significands of finite 16627 numbers, infinities, and NaNs, especially in decimal formats. 16628 16629This operation can also be considered equivalent to the IEEE-754-2008 16630conversion of a floating-point value to the same format. NaNs are handled 16631according to section 6.2. 16632 16633Examples of non-canonical encodings: 16634 16635- x87 pseudo denormals, pseudo NaNs, pseudo Infinity, Unnormals. These are 16636 converted to a canonical representation per hardware-specific protocol. 16637- Many normal decimal floating-point numbers have non-canonical alternative 16638 encodings. 16639- Some machines, like GPUs or ARMv7 NEON, do not support subnormal values. 16640 These are treated as non-canonical encodings of zero and will be flushed to 16641 a zero of the same sign by this operation. 16642 16643Note that per IEEE-754-2008 6.2, systems that support signaling NaNs with 16644default exception handling must signal an invalid exception, and produce a 16645quiet NaN result. 16646 16647This function should always be implementable as multiplication by 1.0, provided 16648that the compiler does not constant fold the operation. Likewise, division by 166491.0 and ``llvm.minnum(x, x)`` are possible implementations. Addition with 16650-0.0 is also sufficient provided that the rounding mode is not -Infinity. 16651 16652``@llvm.canonicalize`` must preserve the equality relation. That is: 16653 16654- ``(@llvm.canonicalize(x) == x)`` is equivalent to ``(x == x)`` 16655- ``(@llvm.canonicalize(x) == @llvm.canonicalize(y))`` is equivalent to 16656 to ``(x == y)`` 16657 16658Additionally, the sign of zero must be conserved: 16659``@llvm.canonicalize(-0.0) = -0.0`` and ``@llvm.canonicalize(+0.0) = +0.0`` 16660 16661The payload bits of a NaN must be conserved, with two exceptions. 16662First, environments which use only a single canonical representation of NaN 16663must perform said canonicalization. Second, SNaNs must be quieted per the 16664usual methods. 16665 16666The canonicalization operation may be optimized away if: 16667 16668- The input is known to be canonical. For example, it was produced by a 16669 floating-point operation that is required by the standard to be canonical. 16670- The result is consumed only by (or fused with) other floating-point 16671 operations. That is, the bits of the floating-point value are not examined. 16672 16673'``llvm.fmuladd.*``' Intrinsic 16674^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16675 16676Syntax: 16677""""""" 16678 16679:: 16680 16681 declare float @llvm.fmuladd.f32(float %a, float %b, float %c) 16682 declare double @llvm.fmuladd.f64(double %a, double %b, double %c) 16683 16684Overview: 16685""""""""" 16686 16687The '``llvm.fmuladd.*``' intrinsic functions represent multiply-add 16688expressions that can be fused if the code generator determines that (a) the 16689target instruction set has support for a fused operation, and (b) that the 16690fused operation is more efficient than the equivalent, separate pair of mul 16691and add instructions. 16692 16693Arguments: 16694"""""""""" 16695 16696The '``llvm.fmuladd.*``' intrinsics each take three arguments: two 16697multiplicands, a and b, and an addend c. 16698 16699Semantics: 16700"""""""""" 16701 16702The expression: 16703 16704:: 16705 16706 %0 = call float @llvm.fmuladd.f32(%a, %b, %c) 16707 16708is equivalent to the expression a \* b + c, except that it is unspecified 16709whether rounding will be performed between the multiplication and addition 16710steps. Fusion is not guaranteed, even if the target platform supports it. 16711If a fused multiply-add is required, the corresponding 16712:ref:`llvm.fma <int_fma>` intrinsic function should be used instead. 16713This never sets errno, just as '``llvm.fma.*``'. 16714 16715Examples: 16716""""""""" 16717 16718.. code-block:: llvm 16719 16720 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields float:r2 = (a * b) + c 16721 16722 16723Hardware-Loop Intrinsics 16724------------------------ 16725 16726LLVM support several intrinsics to mark a loop as a hardware-loop. They are 16727hints to the backend which are required to lower these intrinsics further to target 16728specific instructions, or revert the hardware-loop to a normal loop if target 16729specific restriction are not met and a hardware-loop can't be generated. 16730 16731These intrinsics may be modified in the future and are not intended to be used 16732outside the backend. Thus, front-end and mid-level optimizations should not be 16733generating these intrinsics. 16734 16735 16736'``llvm.set.loop.iterations.*``' Intrinsic 16737^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16738 16739Syntax: 16740""""""" 16741 16742This is an overloaded intrinsic. 16743 16744:: 16745 16746 declare void @llvm.set.loop.iterations.i32(i32) 16747 declare void @llvm.set.loop.iterations.i64(i64) 16748 16749Overview: 16750""""""""" 16751 16752The '``llvm.set.loop.iterations.*``' intrinsics are used to specify the 16753hardware-loop trip count. They are placed in the loop preheader basic block and 16754are marked as ``IntrNoDuplicate`` to avoid optimizers duplicating these 16755instructions. 16756 16757Arguments: 16758"""""""""" 16759 16760The integer operand is the loop trip count of the hardware-loop, and thus 16761not e.g. the loop back-edge taken count. 16762 16763Semantics: 16764"""""""""" 16765 16766The '``llvm.set.loop.iterations.*``' intrinsics do not perform any arithmetic 16767on their operand. It's a hint to the backend that can use this to set up the 16768hardware-loop count with a target specific instruction, usually a move of this 16769value to a special register or a hardware-loop instruction. 16770 16771 16772'``llvm.start.loop.iterations.*``' Intrinsic 16773^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16774 16775Syntax: 16776""""""" 16777 16778This is an overloaded intrinsic. 16779 16780:: 16781 16782 declare i32 @llvm.start.loop.iterations.i32(i32) 16783 declare i64 @llvm.start.loop.iterations.i64(i64) 16784 16785Overview: 16786""""""""" 16787 16788The '``llvm.start.loop.iterations.*``' intrinsics are similar to the 16789'``llvm.set.loop.iterations.*``' intrinsics, used to specify the 16790hardware-loop trip count but also produce a value identical to the input 16791that can be used as the input to the loop. They are placed in the loop 16792preheader basic block and the output is expected to be the input to the 16793phi for the induction variable of the loop, decremented by the 16794'``llvm.loop.decrement.reg.*``'. 16795 16796Arguments: 16797"""""""""" 16798 16799The integer operand is the loop trip count of the hardware-loop, and thus 16800not e.g. the loop back-edge taken count. 16801 16802Semantics: 16803"""""""""" 16804 16805The '``llvm.start.loop.iterations.*``' intrinsics do not perform any arithmetic 16806on their operand. It's a hint to the backend that can use this to set up the 16807hardware-loop count with a target specific instruction, usually a move of this 16808value to a special register or a hardware-loop instruction. 16809 16810'``llvm.test.set.loop.iterations.*``' Intrinsic 16811^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16812 16813Syntax: 16814""""""" 16815 16816This is an overloaded intrinsic. 16817 16818:: 16819 16820 declare i1 @llvm.test.set.loop.iterations.i32(i32) 16821 declare i1 @llvm.test.set.loop.iterations.i64(i64) 16822 16823Overview: 16824""""""""" 16825 16826The '``llvm.test.set.loop.iterations.*``' intrinsics are used to specify the 16827the loop trip count, and also test that the given count is not zero, allowing 16828it to control entry to a while-loop. They are placed in the loop preheader's 16829predecessor basic block, and are marked as ``IntrNoDuplicate`` to avoid 16830optimizers duplicating these instructions. 16831 16832Arguments: 16833"""""""""" 16834 16835The integer operand is the loop trip count of the hardware-loop, and thus 16836not e.g. the loop back-edge taken count. 16837 16838Semantics: 16839"""""""""" 16840 16841The '``llvm.test.set.loop.iterations.*``' intrinsics do not perform any 16842arithmetic on their operand. It's a hint to the backend that can use this to 16843set up the hardware-loop count with a target specific instruction, usually a 16844move of this value to a special register or a hardware-loop instruction. 16845The result is the conditional value of whether the given count is not zero. 16846 16847 16848'``llvm.test.start.loop.iterations.*``' Intrinsic 16849^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16850 16851Syntax: 16852""""""" 16853 16854This is an overloaded intrinsic. 16855 16856:: 16857 16858 declare {i32, i1} @llvm.test.start.loop.iterations.i32(i32) 16859 declare {i64, i1} @llvm.test.start.loop.iterations.i64(i64) 16860 16861Overview: 16862""""""""" 16863 16864The '``llvm.test.start.loop.iterations.*``' intrinsics are similar to the 16865'``llvm.test.set.loop.iterations.*``' and '``llvm.start.loop.iterations.*``' 16866intrinsics, used to specify the hardware-loop trip count, but also produce a 16867value identical to the input that can be used as the input to the loop. The 16868second i1 output controls entry to a while-loop. 16869 16870Arguments: 16871"""""""""" 16872 16873The integer operand is the loop trip count of the hardware-loop, and thus 16874not e.g. the loop back-edge taken count. 16875 16876Semantics: 16877"""""""""" 16878 16879The '``llvm.test.start.loop.iterations.*``' intrinsics do not perform any 16880arithmetic on their operand. It's a hint to the backend that can use this to 16881set up the hardware-loop count with a target specific instruction, usually a 16882move of this value to a special register or a hardware-loop instruction. 16883The result is a pair of the input and a conditional value of whether the 16884given count is not zero. 16885 16886 16887'``llvm.loop.decrement.reg.*``' Intrinsic 16888^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16889 16890Syntax: 16891""""""" 16892 16893This is an overloaded intrinsic. 16894 16895:: 16896 16897 declare i32 @llvm.loop.decrement.reg.i32(i32, i32) 16898 declare i64 @llvm.loop.decrement.reg.i64(i64, i64) 16899 16900Overview: 16901""""""""" 16902 16903The '``llvm.loop.decrement.reg.*``' intrinsics are used to lower the loop 16904iteration counter and return an updated value that will be used in the next 16905loop test check. 16906 16907Arguments: 16908"""""""""" 16909 16910Both arguments must have identical integer types. The first operand is the 16911loop iteration counter. The second operand is the maximum number of elements 16912processed in an iteration. 16913 16914Semantics: 16915"""""""""" 16916 16917The '``llvm.loop.decrement.reg.*``' intrinsics do an integer ``SUB`` of its 16918two operands, which is not allowed to wrap. They return the remaining number of 16919iterations still to be executed, and can be used together with a ``PHI``, 16920``ICMP`` and ``BR`` to control the number of loop iterations executed. Any 16921optimisations are allowed to treat it is a ``SUB``, and it is supported by 16922SCEV, so it's the backends responsibility to handle cases where it may be 16923optimised. These intrinsics are marked as ``IntrNoDuplicate`` to avoid 16924optimizers duplicating these instructions. 16925 16926 16927'``llvm.loop.decrement.*``' Intrinsic 16928^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16929 16930Syntax: 16931""""""" 16932 16933This is an overloaded intrinsic. 16934 16935:: 16936 16937 declare i1 @llvm.loop.decrement.i32(i32) 16938 declare i1 @llvm.loop.decrement.i64(i64) 16939 16940Overview: 16941""""""""" 16942 16943The HardwareLoops pass allows the loop decrement value to be specified with an 16944option. It defaults to a loop decrement value of 1, but it can be an unsigned 16945integer value provided by this option. The '``llvm.loop.decrement.*``' 16946intrinsics decrement the loop iteration counter with this value, and return a 16947false predicate if the loop should exit, and true otherwise. 16948This is emitted if the loop counter is not updated via a ``PHI`` node, which 16949can also be controlled with an option. 16950 16951Arguments: 16952"""""""""" 16953 16954The integer argument is the loop decrement value used to decrement the loop 16955iteration counter. 16956 16957Semantics: 16958"""""""""" 16959 16960The '``llvm.loop.decrement.*``' intrinsics do a ``SUB`` of the loop iteration 16961counter with the given loop decrement value, and return false if the loop 16962should exit, this ``SUB`` is not allowed to wrap. The result is a condition 16963that is used by the conditional branch controlling the loop. 16964 16965 16966Vector Reduction Intrinsics 16967--------------------------- 16968 16969Horizontal reductions of vectors can be expressed using the following 16970intrinsics. Each one takes a vector operand as an input and applies its 16971respective operation across all elements of the vector, returning a single 16972scalar result of the same element type. 16973 16974.. _int_vector_reduce_add: 16975 16976'``llvm.vector.reduce.add.*``' Intrinsic 16977^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 16978 16979Syntax: 16980""""""" 16981 16982:: 16983 16984 declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a) 16985 declare i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a) 16986 16987Overview: 16988""""""""" 16989 16990The '``llvm.vector.reduce.add.*``' intrinsics do an integer ``ADD`` 16991reduction of a vector, returning the result as a scalar. The return type matches 16992the element-type of the vector input. 16993 16994Arguments: 16995"""""""""" 16996The argument to this intrinsic must be a vector of integer values. 16997 16998.. _int_vector_reduce_fadd: 16999 17000'``llvm.vector.reduce.fadd.*``' Intrinsic 17001^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17002 17003Syntax: 17004""""""" 17005 17006:: 17007 17008 declare float @llvm.vector.reduce.fadd.v4f32(float %start_value, <4 x float> %a) 17009 declare double @llvm.vector.reduce.fadd.v2f64(double %start_value, <2 x double> %a) 17010 17011Overview: 17012""""""""" 17013 17014The '``llvm.vector.reduce.fadd.*``' intrinsics do a floating-point 17015``ADD`` reduction of a vector, returning the result as a scalar. The return type 17016matches the element-type of the vector input. 17017 17018If the intrinsic call has the 'reassoc' flag set, then the reduction will not 17019preserve the associativity of an equivalent scalarized counterpart. Otherwise 17020the reduction will be *sequential*, thus implying that the operation respects 17021the associativity of a scalarized reduction. That is, the reduction begins with 17022the start value and performs an fadd operation with consecutively increasing 17023vector element indices. See the following pseudocode: 17024 17025:: 17026 17027 float sequential_fadd(start_value, input_vector) 17028 result = start_value 17029 for i = 0 to length(input_vector) 17030 result = result + input_vector[i] 17031 return result 17032 17033 17034Arguments: 17035"""""""""" 17036The first argument to this intrinsic is a scalar start value for the reduction. 17037The type of the start value matches the element-type of the vector input. 17038The second argument must be a vector of floating-point values. 17039 17040To ignore the start value, negative zero (``-0.0``) can be used, as it is 17041the neutral value of floating point addition. 17042 17043Examples: 17044""""""""" 17045 17046:: 17047 17048 %unord = call reassoc float @llvm.vector.reduce.fadd.v4f32(float -0.0, <4 x float> %input) ; relaxed reduction 17049 %ord = call float @llvm.vector.reduce.fadd.v4f32(float %start_value, <4 x float> %input) ; sequential reduction 17050 17051 17052.. _int_vector_reduce_mul: 17053 17054'``llvm.vector.reduce.mul.*``' Intrinsic 17055^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17056 17057Syntax: 17058""""""" 17059 17060:: 17061 17062 declare i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %a) 17063 declare i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> %a) 17064 17065Overview: 17066""""""""" 17067 17068The '``llvm.vector.reduce.mul.*``' intrinsics do an integer ``MUL`` 17069reduction of a vector, returning the result as a scalar. The return type matches 17070the element-type of the vector input. 17071 17072Arguments: 17073"""""""""" 17074The argument to this intrinsic must be a vector of integer values. 17075 17076.. _int_vector_reduce_fmul: 17077 17078'``llvm.vector.reduce.fmul.*``' Intrinsic 17079^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17080 17081Syntax: 17082""""""" 17083 17084:: 17085 17086 declare float @llvm.vector.reduce.fmul.v4f32(float %start_value, <4 x float> %a) 17087 declare double @llvm.vector.reduce.fmul.v2f64(double %start_value, <2 x double> %a) 17088 17089Overview: 17090""""""""" 17091 17092The '``llvm.vector.reduce.fmul.*``' intrinsics do a floating-point 17093``MUL`` reduction of a vector, returning the result as a scalar. The return type 17094matches the element-type of the vector input. 17095 17096If the intrinsic call has the 'reassoc' flag set, then the reduction will not 17097preserve the associativity of an equivalent scalarized counterpart. Otherwise 17098the reduction will be *sequential*, thus implying that the operation respects 17099the associativity of a scalarized reduction. That is, the reduction begins with 17100the start value and performs an fmul operation with consecutively increasing 17101vector element indices. See the following pseudocode: 17102 17103:: 17104 17105 float sequential_fmul(start_value, input_vector) 17106 result = start_value 17107 for i = 0 to length(input_vector) 17108 result = result * input_vector[i] 17109 return result 17110 17111 17112Arguments: 17113"""""""""" 17114The first argument to this intrinsic is a scalar start value for the reduction. 17115The type of the start value matches the element-type of the vector input. 17116The second argument must be a vector of floating-point values. 17117 17118To ignore the start value, one (``1.0``) can be used, as it is the neutral 17119value of floating point multiplication. 17120 17121Examples: 17122""""""""" 17123 17124:: 17125 17126 %unord = call reassoc float @llvm.vector.reduce.fmul.v4f32(float 1.0, <4 x float> %input) ; relaxed reduction 17127 %ord = call float @llvm.vector.reduce.fmul.v4f32(float %start_value, <4 x float> %input) ; sequential reduction 17128 17129.. _int_vector_reduce_and: 17130 17131'``llvm.vector.reduce.and.*``' Intrinsic 17132^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17133 17134Syntax: 17135""""""" 17136 17137:: 17138 17139 declare i32 @llvm.vector.reduce.and.v4i32(<4 x i32> %a) 17140 17141Overview: 17142""""""""" 17143 17144The '``llvm.vector.reduce.and.*``' intrinsics do a bitwise ``AND`` 17145reduction of a vector, returning the result as a scalar. The return type matches 17146the element-type of the vector input. 17147 17148Arguments: 17149"""""""""" 17150The argument to this intrinsic must be a vector of integer values. 17151 17152.. _int_vector_reduce_or: 17153 17154'``llvm.vector.reduce.or.*``' Intrinsic 17155^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17156 17157Syntax: 17158""""""" 17159 17160:: 17161 17162 declare i32 @llvm.vector.reduce.or.v4i32(<4 x i32> %a) 17163 17164Overview: 17165""""""""" 17166 17167The '``llvm.vector.reduce.or.*``' intrinsics do a bitwise ``OR`` reduction 17168of a vector, returning the result as a scalar. The return type matches the 17169element-type of the vector input. 17170 17171Arguments: 17172"""""""""" 17173The argument to this intrinsic must be a vector of integer values. 17174 17175.. _int_vector_reduce_xor: 17176 17177'``llvm.vector.reduce.xor.*``' Intrinsic 17178^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17179 17180Syntax: 17181""""""" 17182 17183:: 17184 17185 declare i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> %a) 17186 17187Overview: 17188""""""""" 17189 17190The '``llvm.vector.reduce.xor.*``' intrinsics do a bitwise ``XOR`` 17191reduction of a vector, returning the result as a scalar. The return type matches 17192the element-type of the vector input. 17193 17194Arguments: 17195"""""""""" 17196The argument to this intrinsic must be a vector of integer values. 17197 17198.. _int_vector_reduce_smax: 17199 17200'``llvm.vector.reduce.smax.*``' Intrinsic 17201^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17202 17203Syntax: 17204""""""" 17205 17206:: 17207 17208 declare i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> %a) 17209 17210Overview: 17211""""""""" 17212 17213The '``llvm.vector.reduce.smax.*``' intrinsics do a signed integer 17214``MAX`` reduction of a vector, returning the result as a scalar. The return type 17215matches the element-type of the vector input. 17216 17217Arguments: 17218"""""""""" 17219The argument to this intrinsic must be a vector of integer values. 17220 17221.. _int_vector_reduce_smin: 17222 17223'``llvm.vector.reduce.smin.*``' Intrinsic 17224^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17225 17226Syntax: 17227""""""" 17228 17229:: 17230 17231 declare i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> %a) 17232 17233Overview: 17234""""""""" 17235 17236The '``llvm.vector.reduce.smin.*``' intrinsics do a signed integer 17237``MIN`` reduction of a vector, returning the result as a scalar. The return type 17238matches the element-type of the vector input. 17239 17240Arguments: 17241"""""""""" 17242The argument to this intrinsic must be a vector of integer values. 17243 17244.. _int_vector_reduce_umax: 17245 17246'``llvm.vector.reduce.umax.*``' Intrinsic 17247^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17248 17249Syntax: 17250""""""" 17251 17252:: 17253 17254 declare i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> %a) 17255 17256Overview: 17257""""""""" 17258 17259The '``llvm.vector.reduce.umax.*``' intrinsics do an unsigned 17260integer ``MAX`` reduction of a vector, returning the result as a scalar. The 17261return type matches the element-type of the vector input. 17262 17263Arguments: 17264"""""""""" 17265The argument to this intrinsic must be a vector of integer values. 17266 17267.. _int_vector_reduce_umin: 17268 17269'``llvm.vector.reduce.umin.*``' Intrinsic 17270^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17271 17272Syntax: 17273""""""" 17274 17275:: 17276 17277 declare i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> %a) 17278 17279Overview: 17280""""""""" 17281 17282The '``llvm.vector.reduce.umin.*``' intrinsics do an unsigned 17283integer ``MIN`` reduction of a vector, returning the result as a scalar. The 17284return type matches the element-type of the vector input. 17285 17286Arguments: 17287"""""""""" 17288The argument to this intrinsic must be a vector of integer values. 17289 17290.. _int_vector_reduce_fmax: 17291 17292'``llvm.vector.reduce.fmax.*``' Intrinsic 17293^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17294 17295Syntax: 17296""""""" 17297 17298:: 17299 17300 declare float @llvm.vector.reduce.fmax.v4f32(<4 x float> %a) 17301 declare double @llvm.vector.reduce.fmax.v2f64(<2 x double> %a) 17302 17303Overview: 17304""""""""" 17305 17306The '``llvm.vector.reduce.fmax.*``' intrinsics do a floating-point 17307``MAX`` reduction of a vector, returning the result as a scalar. The return type 17308matches the element-type of the vector input. 17309 17310This instruction has the same comparison semantics as the '``llvm.maxnum.*``' 17311intrinsic. That is, the result will always be a number unless all elements of 17312the vector are NaN. For a vector with maximum element magnitude 0.0 and 17313containing both +0.0 and -0.0 elements, the sign of the result is unspecified. 17314 17315If the intrinsic call has the ``nnan`` fast-math flag, then the operation can 17316assume that NaNs are not present in the input vector. 17317 17318Arguments: 17319"""""""""" 17320The argument to this intrinsic must be a vector of floating-point values. 17321 17322.. _int_vector_reduce_fmin: 17323 17324'``llvm.vector.reduce.fmin.*``' Intrinsic 17325^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17326 17327Syntax: 17328""""""" 17329This is an overloaded intrinsic. 17330 17331:: 17332 17333 declare float @llvm.vector.reduce.fmin.v4f32(<4 x float> %a) 17334 declare double @llvm.vector.reduce.fmin.v2f64(<2 x double> %a) 17335 17336Overview: 17337""""""""" 17338 17339The '``llvm.vector.reduce.fmin.*``' intrinsics do a floating-point 17340``MIN`` reduction of a vector, returning the result as a scalar. The return type 17341matches the element-type of the vector input. 17342 17343This instruction has the same comparison semantics as the '``llvm.minnum.*``' 17344intrinsic. That is, the result will always be a number unless all elements of 17345the vector are NaN. For a vector with minimum element magnitude 0.0 and 17346containing both +0.0 and -0.0 elements, the sign of the result is unspecified. 17347 17348If the intrinsic call has the ``nnan`` fast-math flag, then the operation can 17349assume that NaNs are not present in the input vector. 17350 17351Arguments: 17352"""""""""" 17353The argument to this intrinsic must be a vector of floating-point values. 17354 17355'``llvm.vector.insert``' Intrinsic 17356^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17357 17358Syntax: 17359""""""" 17360This is an overloaded intrinsic. 17361 17362:: 17363 17364 ; Insert fixed type into scalable type 17365 declare <vscale x 4 x float> @llvm.vector.insert.nxv4f32.v4f32(<vscale x 4 x float> %vec, <4 x float> %subvec, i64 <idx>) 17366 declare <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v2f64(<vscale x 2 x double> %vec, <2 x double> %subvec, i64 <idx>) 17367 17368 ; Insert scalable type into scalable type 17369 declare <vscale x 4 x float> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x float> %vec, <vscale x 2 x float> %subvec, i64 <idx>) 17370 17371 ; Insert fixed type into fixed type 17372 declare <4 x double> @llvm.vector.insert.v4f64.v2f64(<4 x double> %vec, <2 x double> %subvec, i64 <idx>) 17373 17374Overview: 17375""""""""" 17376 17377The '``llvm.vector.insert.*``' intrinsics insert a vector into another vector 17378starting from a given index. The return type matches the type of the vector we 17379insert into. Conceptually, this can be used to build a scalable vector out of 17380non-scalable vectors, however this intrinsic can also be used on purely fixed 17381types. 17382 17383Scalable vectors can only be inserted into other scalable vectors. 17384 17385Arguments: 17386"""""""""" 17387 17388The ``vec`` is the vector which ``subvec`` will be inserted into. 17389The ``subvec`` is the vector that will be inserted. 17390 17391``idx`` represents the starting element number at which ``subvec`` will be 17392inserted. ``idx`` must be a constant multiple of ``subvec``'s known minimum 17393vector length. If ``subvec`` is a scalable vector, ``idx`` is first scaled by 17394the runtime scaling factor of ``subvec``. The elements of ``vec`` starting at 17395``idx`` are overwritten with ``subvec``. Elements ``idx`` through (``idx`` + 17396num_elements(``subvec``) - 1) must be valid ``vec`` indices. If this condition 17397cannot be determined statically but is false at runtime, then the result vector 17398is a :ref:`poison value <poisonvalues>`. 17399 17400 17401'``llvm.vector.extract``' Intrinsic 17402^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17403 17404Syntax: 17405""""""" 17406This is an overloaded intrinsic. 17407 17408:: 17409 17410 ; Extract fixed type from scalable type 17411 declare <4 x float> @llvm.vector.extract.v4f32.nxv4f32(<vscale x 4 x float> %vec, i64 <idx>) 17412 declare <2 x double> @llvm.vector.extract.v2f64.nxv2f64(<vscale x 2 x double> %vec, i64 <idx>) 17413 17414 ; Extract scalable type from scalable type 17415 declare <vscale x 2 x float> @llvm.vector.extract.nxv2f32.nxv4f32(<vscale x 4 x float> %vec, i64 <idx>) 17416 17417 ; Extract fixed type from fixed type 17418 declare <2 x double> @llvm.vector.extract.v2f64.v4f64(<4 x double> %vec, i64 <idx>) 17419 17420Overview: 17421""""""""" 17422 17423The '``llvm.vector.extract.*``' intrinsics extract a vector from within another 17424vector starting from a given index. The return type must be explicitly 17425specified. Conceptually, this can be used to decompose a scalable vector into 17426non-scalable parts, however this intrinsic can also be used on purely fixed 17427types. 17428 17429Scalable vectors can only be extracted from other scalable vectors. 17430 17431Arguments: 17432"""""""""" 17433 17434The ``vec`` is the vector from which we will extract a subvector. 17435 17436The ``idx`` specifies the starting element number within ``vec`` from which a 17437subvector is extracted. ``idx`` must be a constant multiple of the known-minimum 17438vector length of the result type. If the result type is a scalable vector, 17439``idx`` is first scaled by the result type's runtime scaling factor. Elements 17440``idx`` through (``idx`` + num_elements(result_type) - 1) must be valid vector 17441indices. If this condition cannot be determined statically but is false at 17442runtime, then the result vector is a :ref:`poison value <poisonvalues>`. The 17443``idx`` parameter must be a vector index constant type (for most targets this 17444will be an integer pointer type). 17445 17446'``llvm.experimental.vector.reverse``' Intrinsic 17447^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17448 17449Syntax: 17450""""""" 17451This is an overloaded intrinsic. 17452 17453:: 17454 17455 declare <2 x i8> @llvm.experimental.vector.reverse.v2i8(<2 x i8> %a) 17456 declare <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> %a) 17457 17458Overview: 17459""""""""" 17460 17461The '``llvm.experimental.vector.reverse.*``' intrinsics reverse a vector. 17462The intrinsic takes a single vector and returns a vector of matching type but 17463with the original lane order reversed. These intrinsics work for both fixed 17464and scalable vectors. While this intrinsic is marked as experimental the 17465recommended way to express reverse operations for fixed-width vectors is still 17466to use a shufflevector, as that may allow for more optimization opportunities. 17467 17468Arguments: 17469"""""""""" 17470 17471The argument to this intrinsic must be a vector. 17472 17473'``llvm.experimental.vector.splice``' Intrinsic 17474^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17475 17476Syntax: 17477""""""" 17478This is an overloaded intrinsic. 17479 17480:: 17481 17482 declare <2 x double> @llvm.experimental.vector.splice.v2f64(<2 x double> %vec1, <2 x double> %vec2, i32 %imm) 17483 declare <vscale x 4 x i32> @llvm.experimental.vector.splice.nxv4i32(<vscale x 4 x i32> %vec1, <vscale x 4 x i32> %vec2, i32 %imm) 17484 17485Overview: 17486""""""""" 17487 17488The '``llvm.experimental.vector.splice.*``' intrinsics construct a vector by 17489concatenating elements from the first input vector with elements of the second 17490input vector, returning a vector of the same type as the input vectors. The 17491signed immediate, modulo the number of elements in the vector, is the index 17492into the first vector from which to extract the result value. This means 17493conceptually that for a positive immediate, a vector is extracted from 17494``concat(%vec1, %vec2)`` starting at index ``imm``, whereas for a negative 17495immediate, it extracts ``-imm`` trailing elements from the first vector, and 17496the remaining elements from ``%vec2``. 17497 17498These intrinsics work for both fixed and scalable vectors. While this intrinsic 17499is marked as experimental, the recommended way to express this operation for 17500fixed-width vectors is still to use a shufflevector, as that may allow for more 17501optimization opportunities. 17502 17503For example: 17504 17505.. code-block:: text 17506 17507 llvm.experimental.vector.splice(<A,B,C,D>, <E,F,G,H>, 1) ==> <B, C, D, E> ; index 17508 llvm.experimental.vector.splice(<A,B,C,D>, <E,F,G,H>, -3) ==> <B, C, D, E> ; trailing elements 17509 17510 17511Arguments: 17512"""""""""" 17513 17514The first two operands are vectors with the same type. The start index is imm 17515modulo the runtime number of elements in the source vector. For a fixed-width 17516vector <N x eltty>, imm is a signed integer constant in the range 17517-N <= imm < N. For a scalable vector <vscale x N x eltty>, imm is a signed 17518integer constant in the range -X <= imm < X where X=vscale_range_min * N. 17519 17520'``llvm.experimental.stepvector``' Intrinsic 17521^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17522 17523This is an overloaded intrinsic. You can use ``llvm.experimental.stepvector`` 17524to generate a vector whose lane values comprise the linear sequence 17525<0, 1, 2, ...>. It is primarily intended for scalable vectors. 17526 17527:: 17528 17529 declare <vscale x 4 x i32> @llvm.experimental.stepvector.nxv4i32() 17530 declare <vscale x 8 x i16> @llvm.experimental.stepvector.nxv8i16() 17531 17532The '``llvm.experimental.stepvector``' intrinsics are used to create vectors 17533of integers whose elements contain a linear sequence of values starting from 0 17534with a step of 1. This experimental intrinsic can only be used for vectors 17535with integer elements that are at least 8 bits in size. If the sequence value 17536exceeds the allowed limit for the element type then the result for that lane is 17537undefined. 17538 17539These intrinsics work for both fixed and scalable vectors. While this intrinsic 17540is marked as experimental, the recommended way to express this operation for 17541fixed-width vectors is still to generate a constant vector instead. 17542 17543 17544Arguments: 17545"""""""""" 17546 17547None. 17548 17549 17550Matrix Intrinsics 17551----------------- 17552 17553Operations on matrixes requiring shape information (like number of rows/columns 17554or the memory layout) can be expressed using the matrix intrinsics. These 17555intrinsics require matrix dimensions to be passed as immediate arguments, and 17556matrixes are passed and returned as vectors. This means that for a ``R`` x 17557``C`` matrix, element ``i`` of column ``j`` is at index ``j * R + i`` in the 17558corresponding vector, with indices starting at 0. Currently column-major layout 17559is assumed. The intrinsics support both integer and floating point matrixes. 17560 17561 17562'``llvm.matrix.transpose.*``' Intrinsic 17563^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17564 17565Syntax: 17566""""""" 17567This is an overloaded intrinsic. 17568 17569:: 17570 17571 declare vectorty @llvm.matrix.transpose.*(vectorty %In, i32 <Rows>, i32 <Cols>) 17572 17573Overview: 17574""""""""" 17575 17576The '``llvm.matrix.transpose.*``' intrinsics treat ``%In`` as a ``<Rows> x 17577<Cols>`` matrix and return the transposed matrix in the result vector. 17578 17579Arguments: 17580"""""""""" 17581 17582The first argument ``%In`` is a vector that corresponds to a ``<Rows> x 17583<Cols>`` matrix. Thus, arguments ``<Rows>`` and ``<Cols>`` correspond to the 17584number of rows and columns, respectively, and must be positive, constant 17585integers. The returned vector must have ``<Rows> * <Cols>`` elements, and have 17586the same float or integer element type as ``%In``. 17587 17588'``llvm.matrix.multiply.*``' Intrinsic 17589^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17590 17591Syntax: 17592""""""" 17593This is an overloaded intrinsic. 17594 17595:: 17596 17597 declare vectorty @llvm.matrix.multiply.*(vectorty %A, vectorty %B, i32 <OuterRows>, i32 <Inner>, i32 <OuterColumns>) 17598 17599Overview: 17600""""""""" 17601 17602The '``llvm.matrix.multiply.*``' intrinsics treat ``%A`` as a ``<OuterRows> x 17603<Inner>`` matrix, ``%B`` as a ``<Inner> x <OuterColumns>`` matrix, and 17604multiplies them. The result matrix is returned in the result vector. 17605 17606Arguments: 17607"""""""""" 17608 17609The first vector argument ``%A`` corresponds to a matrix with ``<OuterRows> * 17610<Inner>`` elements, and the second argument ``%B`` to a matrix with 17611``<Inner> * <OuterColumns>`` elements. Arguments ``<OuterRows>``, 17612``<Inner>`` and ``<OuterColumns>`` must be positive, constant integers. The 17613returned vector must have ``<OuterRows> * <OuterColumns>`` elements. 17614Vectors ``%A``, ``%B``, and the returned vector all have the same float or 17615integer element type. 17616 17617 17618'``llvm.matrix.column.major.load.*``' Intrinsic 17619^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17620 17621Syntax: 17622""""""" 17623This is an overloaded intrinsic. 17624 17625:: 17626 17627 declare vectorty @llvm.matrix.column.major.load.*( 17628 ptrty %Ptr, i64 %Stride, i1 <IsVolatile>, i32 <Rows>, i32 <Cols>) 17629 17630Overview: 17631""""""""" 17632 17633The '``llvm.matrix.column.major.load.*``' intrinsics load a ``<Rows> x <Cols>`` 17634matrix using a stride of ``%Stride`` to compute the start address of the 17635different columns. The offset is computed using ``%Stride``'s bitwidth. This 17636allows for convenient loading of sub matrixes. If ``<IsVolatile>`` is true, the 17637intrinsic is considered a :ref:`volatile memory access <volatile>`. The result 17638matrix is returned in the result vector. If the ``%Ptr`` argument is known to 17639be aligned to some boundary, this can be specified as an attribute on the 17640argument. 17641 17642Arguments: 17643"""""""""" 17644 17645The first argument ``%Ptr`` is a pointer type to the returned vector type, and 17646corresponds to the start address to load from. The second argument ``%Stride`` 17647is a positive, constant integer with ``%Stride >= <Rows>``. ``%Stride`` is used 17648to compute the column memory addresses. I.e., for a column ``C``, its start 17649memory addresses is calculated with ``%Ptr + C * %Stride``. The third Argument 17650``<IsVolatile>`` is a boolean value. The fourth and fifth arguments, 17651``<Rows>`` and ``<Cols>``, correspond to the number of rows and columns, 17652respectively, and must be positive, constant integers. The returned vector must 17653have ``<Rows> * <Cols>`` elements. 17654 17655The :ref:`align <attr_align>` parameter attribute can be provided for the 17656``%Ptr`` arguments. 17657 17658 17659'``llvm.matrix.column.major.store.*``' Intrinsic 17660^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17661 17662Syntax: 17663""""""" 17664 17665:: 17666 17667 declare void @llvm.matrix.column.major.store.*( 17668 vectorty %In, ptrty %Ptr, i64 %Stride, i1 <IsVolatile>, i32 <Rows>, i32 <Cols>) 17669 17670Overview: 17671""""""""" 17672 17673The '``llvm.matrix.column.major.store.*``' intrinsics store the ``<Rows> x 17674<Cols>`` matrix in ``%In`` to memory using a stride of ``%Stride`` between 17675columns. The offset is computed using ``%Stride``'s bitwidth. If 17676``<IsVolatile>`` is true, the intrinsic is considered a 17677:ref:`volatile memory access <volatile>`. 17678 17679If the ``%Ptr`` argument is known to be aligned to some boundary, this can be 17680specified as an attribute on the argument. 17681 17682Arguments: 17683"""""""""" 17684 17685The first argument ``%In`` is a vector that corresponds to a ``<Rows> x 17686<Cols>`` matrix to be stored to memory. The second argument ``%Ptr`` is a 17687pointer to the vector type of ``%In``, and is the start address of the matrix 17688in memory. The third argument ``%Stride`` is a positive, constant integer with 17689``%Stride >= <Rows>``. ``%Stride`` is used to compute the column memory 17690addresses. I.e., for a column ``C``, its start memory addresses is calculated 17691with ``%Ptr + C * %Stride``. The fourth argument ``<IsVolatile>`` is a boolean 17692value. The arguments ``<Rows>`` and ``<Cols>`` correspond to the number of rows 17693and columns, respectively, and must be positive, constant integers. 17694 17695The :ref:`align <attr_align>` parameter attribute can be provided 17696for the ``%Ptr`` arguments. 17697 17698 17699Half Precision Floating-Point Intrinsics 17700---------------------------------------- 17701 17702For most target platforms, half precision floating-point is a 17703storage-only format. This means that it is a dense encoding (in memory) 17704but does not support computation in the format. 17705 17706This means that code must first load the half-precision floating-point 17707value as an i16, then convert it to float with 17708:ref:`llvm.convert.from.fp16 <int_convert_from_fp16>`. Computation can 17709then be performed on the float value (including extending to double 17710etc). To store the value back to memory, it is first converted to float 17711if needed, then converted to i16 with 17712:ref:`llvm.convert.to.fp16 <int_convert_to_fp16>`, then storing as an 17713i16 value. 17714 17715.. _int_convert_to_fp16: 17716 17717'``llvm.convert.to.fp16``' Intrinsic 17718^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17719 17720Syntax: 17721""""""" 17722 17723:: 17724 17725 declare i16 @llvm.convert.to.fp16.f32(float %a) 17726 declare i16 @llvm.convert.to.fp16.f64(double %a) 17727 17728Overview: 17729""""""""" 17730 17731The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a 17732conventional floating-point type to half precision floating-point format. 17733 17734Arguments: 17735"""""""""" 17736 17737The intrinsic function contains single argument - the value to be 17738converted. 17739 17740Semantics: 17741"""""""""" 17742 17743The '``llvm.convert.to.fp16``' intrinsic function performs a conversion from a 17744conventional floating-point format to half precision floating-point format. The 17745return value is an ``i16`` which contains the converted number. 17746 17747Examples: 17748""""""""" 17749 17750.. code-block:: llvm 17751 17752 %res = call i16 @llvm.convert.to.fp16.f32(float %a) 17753 store i16 %res, i16* @x, align 2 17754 17755.. _int_convert_from_fp16: 17756 17757'``llvm.convert.from.fp16``' Intrinsic 17758^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17759 17760Syntax: 17761""""""" 17762 17763:: 17764 17765 declare float @llvm.convert.from.fp16.f32(i16 %a) 17766 declare double @llvm.convert.from.fp16.f64(i16 %a) 17767 17768Overview: 17769""""""""" 17770 17771The '``llvm.convert.from.fp16``' intrinsic function performs a 17772conversion from half precision floating-point format to single precision 17773floating-point format. 17774 17775Arguments: 17776"""""""""" 17777 17778The intrinsic function contains single argument - the value to be 17779converted. 17780 17781Semantics: 17782"""""""""" 17783 17784The '``llvm.convert.from.fp16``' intrinsic function performs a 17785conversion from half single precision floating-point format to single 17786precision floating-point format. The input half-float value is 17787represented by an ``i16`` value. 17788 17789Examples: 17790""""""""" 17791 17792.. code-block:: llvm 17793 17794 %a = load i16, i16* @x, align 2 17795 %res = call float @llvm.convert.from.fp16(i16 %a) 17796 17797Saturating floating-point to integer conversions 17798------------------------------------------------ 17799 17800The ``fptoui`` and ``fptosi`` instructions return a 17801:ref:`poison value <poisonvalues>` if the rounded-towards-zero value is not 17802representable by the result type. These intrinsics provide an alternative 17803conversion, which will saturate towards the smallest and largest representable 17804integer values instead. 17805 17806'``llvm.fptoui.sat.*``' Intrinsic 17807^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17808 17809Syntax: 17810""""""" 17811 17812This is an overloaded intrinsic. You can use ``llvm.fptoui.sat`` on any 17813floating-point argument type and any integer result type, or vectors thereof. 17814Not all targets may support all types, however. 17815 17816:: 17817 17818 declare i32 @llvm.fptoui.sat.i32.f32(float %f) 17819 declare i19 @llvm.fptoui.sat.i19.f64(double %f) 17820 declare <4 x i100> @llvm.fptoui.sat.v4i100.v4f128(<4 x fp128> %f) 17821 17822Overview: 17823""""""""" 17824 17825This intrinsic converts the argument into an unsigned integer using saturating 17826semantics. 17827 17828Arguments: 17829"""""""""" 17830 17831The argument may be any floating-point or vector of floating-point type. The 17832return value may be any integer or vector of integer type. The number of vector 17833elements in argument and return must be the same. 17834 17835Semantics: 17836"""""""""" 17837 17838The conversion to integer is performed subject to the following rules: 17839 17840- If the argument is any NaN, zero is returned. 17841- If the argument is smaller than zero (this includes negative infinity), 17842 zero is returned. 17843- If the argument is larger than the largest representable unsigned integer of 17844 the result type (this includes positive infinity), the largest representable 17845 unsigned integer is returned. 17846- Otherwise, the result of rounding the argument towards zero is returned. 17847 17848Example: 17849"""""""" 17850 17851.. code-block:: text 17852 17853 %a = call i8 @llvm.fptoui.sat.i8.f32(float 123.9) ; yields i8: 123 17854 %b = call i8 @llvm.fptoui.sat.i8.f32(float -5.7) ; yields i8: 0 17855 %c = call i8 @llvm.fptoui.sat.i8.f32(float 377.0) ; yields i8: 255 17856 %d = call i8 @llvm.fptoui.sat.i8.f32(float 0xFFF8000000000000) ; yields i8: 0 17857 17858'``llvm.fptosi.sat.*``' Intrinsic 17859^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17860 17861Syntax: 17862""""""" 17863 17864This is an overloaded intrinsic. You can use ``llvm.fptosi.sat`` on any 17865floating-point argument type and any integer result type, or vectors thereof. 17866Not all targets may support all types, however. 17867 17868:: 17869 17870 declare i32 @llvm.fptosi.sat.i32.f32(float %f) 17871 declare i19 @llvm.fptosi.sat.i19.f64(double %f) 17872 declare <4 x i100> @llvm.fptosi.sat.v4i100.v4f128(<4 x fp128> %f) 17873 17874Overview: 17875""""""""" 17876 17877This intrinsic converts the argument into a signed integer using saturating 17878semantics. 17879 17880Arguments: 17881"""""""""" 17882 17883The argument may be any floating-point or vector of floating-point type. The 17884return value may be any integer or vector of integer type. The number of vector 17885elements in argument and return must be the same. 17886 17887Semantics: 17888"""""""""" 17889 17890The conversion to integer is performed subject to the following rules: 17891 17892- If the argument is any NaN, zero is returned. 17893- If the argument is smaller than the smallest representable signed integer of 17894 the result type (this includes negative infinity), the smallest 17895 representable signed integer is returned. 17896- If the argument is larger than the largest representable signed integer of 17897 the result type (this includes positive infinity), the largest representable 17898 signed integer is returned. 17899- Otherwise, the result of rounding the argument towards zero is returned. 17900 17901Example: 17902"""""""" 17903 17904.. code-block:: text 17905 17906 %a = call i8 @llvm.fptosi.sat.i8.f32(float 23.9) ; yields i8: 23 17907 %b = call i8 @llvm.fptosi.sat.i8.f32(float -130.8) ; yields i8: -128 17908 %c = call i8 @llvm.fptosi.sat.i8.f32(float 999.0) ; yields i8: 127 17909 %d = call i8 @llvm.fptosi.sat.i8.f32(float 0xFFF8000000000000) ; yields i8: 0 17910 17911.. _dbg_intrinsics: 17912 17913Debugger Intrinsics 17914------------------- 17915 17916The LLVM debugger intrinsics (which all start with ``llvm.dbg.`` 17917prefix), are described in the `LLVM Source Level 17918Debugging <SourceLevelDebugging.html#format-common-intrinsics>`_ 17919document. 17920 17921Exception Handling Intrinsics 17922----------------------------- 17923 17924The LLVM exception handling intrinsics (which all start with 17925``llvm.eh.`` prefix), are described in the `LLVM Exception 17926Handling <ExceptionHandling.html#format-common-intrinsics>`_ document. 17927 17928Pointer Authentication Intrinsics 17929--------------------------------- 17930 17931The LLVM pointer authentication intrinsics (which all start with 17932``llvm.ptrauth.`` prefix), are described in the `Pointer Authentication 17933<PointerAuth.html#intrinsics>`_ document. 17934 17935.. _int_trampoline: 17936 17937Trampoline Intrinsics 17938--------------------- 17939 17940These intrinsics make it possible to excise one parameter, marked with 17941the :ref:`nest <nest>` attribute, from a function. The result is a 17942callable function pointer lacking the nest parameter - the caller does 17943not need to provide a value for it. Instead, the value to use is stored 17944in advance in a "trampoline", a block of memory usually allocated on the 17945stack, which also contains code to splice the nest value into the 17946argument list. This is used to implement the GCC nested function address 17947extension. 17948 17949For example, if the function is ``i32 f(i8* nest %c, i32 %x, i32 %y)`` 17950then the resulting function pointer has signature ``i32 (i32, i32)*``. 17951It can be created as follows: 17952 17953.. code-block:: llvm 17954 17955 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86 17956 %tramp1 = getelementptr [10 x i8], [10 x i8]* %tramp, i32 0, i32 0 17957 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval) 17958 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1) 17959 %fp = bitcast i8* %p to i32 (i32, i32)* 17960 17961The call ``%val = call i32 %fp(i32 %x, i32 %y)`` is then equivalent to 17962``%val = call i32 %f(i8* %nval, i32 %x, i32 %y)``. 17963 17964.. _int_it: 17965 17966'``llvm.init.trampoline``' Intrinsic 17967^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 17968 17969Syntax: 17970""""""" 17971 17972:: 17973 17974 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>) 17975 17976Overview: 17977""""""""" 17978 17979This fills the memory pointed to by ``tramp`` with executable code, 17980turning it into a trampoline. 17981 17982Arguments: 17983"""""""""" 17984 17985The ``llvm.init.trampoline`` intrinsic takes three arguments, all 17986pointers. The ``tramp`` argument must point to a sufficiently large and 17987sufficiently aligned block of memory; this memory is written to by the 17988intrinsic. Note that the size and the alignment are target-specific - 17989LLVM currently provides no portable way of determining them, so a 17990front-end that generates this intrinsic needs to have some 17991target-specific knowledge. The ``func`` argument must hold a function 17992bitcast to an ``i8*``. 17993 17994Semantics: 17995"""""""""" 17996 17997The block of memory pointed to by ``tramp`` is filled with target 17998dependent code, turning it into a function. Then ``tramp`` needs to be 17999passed to :ref:`llvm.adjust.trampoline <int_at>` to get a pointer which can 18000be :ref:`bitcast (to a new function) and called <int_trampoline>`. The new 18001function's signature is the same as that of ``func`` with any arguments 18002marked with the ``nest`` attribute removed. At most one such ``nest`` 18003argument is allowed, and it must be of pointer type. Calling the new 18004function is equivalent to calling ``func`` with the same argument list, 18005but with ``nval`` used for the missing ``nest`` argument. If, after 18006calling ``llvm.init.trampoline``, the memory pointed to by ``tramp`` is 18007modified, then the effect of any later call to the returned function 18008pointer is undefined. 18009 18010.. _int_at: 18011 18012'``llvm.adjust.trampoline``' Intrinsic 18013^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18014 18015Syntax: 18016""""""" 18017 18018:: 18019 18020 declare i8* @llvm.adjust.trampoline(i8* <tramp>) 18021 18022Overview: 18023""""""""" 18024 18025This performs any required machine-specific adjustment to the address of 18026a trampoline (passed as ``tramp``). 18027 18028Arguments: 18029"""""""""" 18030 18031``tramp`` must point to a block of memory which already has trampoline 18032code filled in by a previous call to 18033:ref:`llvm.init.trampoline <int_it>`. 18034 18035Semantics: 18036"""""""""" 18037 18038On some architectures the address of the code to be executed needs to be 18039different than the address where the trampoline is actually stored. This 18040intrinsic returns the executable address corresponding to ``tramp`` 18041after performing the required machine specific adjustments. The pointer 18042returned can then be :ref:`bitcast and executed <int_trampoline>`. 18043 18044 18045.. _int_vp: 18046 18047Vector Predication Intrinsics 18048----------------------------- 18049VP intrinsics are intended for predicated SIMD/vector code. A typical VP 18050operation takes a vector mask and an explicit vector length parameter as in: 18051 18052:: 18053 18054 <W x T> llvm.vp.<opcode>.*(<W x T> %x, <W x T> %y, <W x i1> %mask, i32 %evl) 18055 18056The vector mask parameter (%mask) always has a vector of `i1` type, for example 18057`<32 x i1>`. The explicit vector length parameter always has the type `i32` and 18058is an unsigned integer value. The explicit vector length parameter (%evl) is in 18059the range: 18060 18061:: 18062 18063 0 <= %evl <= W, where W is the number of vector elements 18064 18065Note that for :ref:`scalable vector types <t_vector>` ``W`` is the runtime 18066length of the vector. 18067 18068The VP intrinsic has undefined behavior if ``%evl > W``. The explicit vector 18069length (%evl) creates a mask, %EVLmask, with all elements ``0 <= i < %evl`` set 18070to True, and all other lanes ``%evl <= i < W`` to False. A new mask %M is 18071calculated with an element-wise AND from %mask and %EVLmask: 18072 18073:: 18074 18075 M = %mask AND %EVLmask 18076 18077A vector operation ``<opcode>`` on vectors ``A`` and ``B`` calculates: 18078 18079:: 18080 18081 A <opcode> B = { A[i] <opcode> B[i] M[i] = True, and 18082 { undef otherwise 18083 18084Optimization Hint 18085^^^^^^^^^^^^^^^^^ 18086 18087Some targets, such as AVX512, do not support the %evl parameter in hardware. 18088The use of an effective %evl is discouraged for those targets. The function 18089``TargetTransformInfo::hasActiveVectorLength()`` returns true when the target 18090has native support for %evl. 18091 18092.. _int_vp_select: 18093 18094'``llvm.vp.select.*``' Intrinsics 18095^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18096 18097Syntax: 18098""""""" 18099This is an overloaded intrinsic. 18100 18101:: 18102 18103 declare <16 x i32> @llvm.vp.select.v16i32 (<16 x i1> <condition>, <16 x i32> <on_true>, <16 x i32> <on_false>, i32 <evl>) 18104 declare <vscale x 4 x i64> @llvm.vp.select.nxv4i64 (<vscale x 4 x i1> <condition>, <vscale x 4 x i64> <on_true>, <vscale x 4 x i64> <on_false>, i32 <evl>) 18105 18106Overview: 18107""""""""" 18108 18109The '``llvm.vp.select``' intrinsic is used to choose one value based on a 18110condition vector, without IR-level branching. 18111 18112Arguments: 18113"""""""""" 18114 18115The first operand is a vector of ``i1`` and indicates the condition. The 18116second operand is the value that is selected where the condition vector is 18117true. The third operand is the value that is selected where the condition 18118vector is false. The vectors must be of the same size. The fourth operand is 18119the explicit vector length. 18120 18121#. The optional ``fast-math flags`` marker indicates that the select has one or 18122 more :ref:`fast-math flags <fastmath>`. These are optimization hints to 18123 enable otherwise unsafe floating-point optimizations. Fast-math flags are 18124 only valid for selects that return a floating-point scalar or vector type, 18125 or an array (nested to any depth) of floating-point scalar or vector types. 18126 18127Semantics: 18128"""""""""" 18129 18130The intrinsic selects lanes from the second and third operand depending on a 18131condition vector. 18132 18133All result lanes at positions greater or equal than ``%evl`` are undefined. 18134For all lanes below ``%evl`` where the condition vector is true the lane is 18135taken from the second operand. Otherwise, the lane is taken from the third 18136operand. 18137 18138Example: 18139"""""""" 18140 18141.. code-block:: llvm 18142 18143 %r = call <4 x i32> @llvm.vp.select.v4i32(<4 x i1> %cond, <4 x i32> %on_true, <4 x i32> %on_false, i32 %evl) 18144 18145 ;;; Expansion. 18146 ;; Any result is legal on lanes at and above %evl. 18147 %also.r = select <4 x i1> %cond, <4 x i32> %on_true, <4 x i32> %on_false 18148 18149 18150.. _int_vp_merge: 18151 18152'``llvm.vp.merge.*``' Intrinsics 18153^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18154 18155Syntax: 18156""""""" 18157This is an overloaded intrinsic. 18158 18159:: 18160 18161 declare <16 x i32> @llvm.vp.merge.v16i32 (<16 x i1> <condition>, <16 x i32> <on_true>, <16 x i32> <on_false>, i32 <pivot>) 18162 declare <vscale x 4 x i64> @llvm.vp.merge.nxv4i64 (<vscale x 4 x i1> <condition>, <vscale x 4 x i64> <on_true>, <vscale x 4 x i64> <on_false>, i32 <pivot>) 18163 18164Overview: 18165""""""""" 18166 18167The '``llvm.vp.merge``' intrinsic is used to choose one value based on a 18168condition vector and an index operand, without IR-level branching. 18169 18170Arguments: 18171"""""""""" 18172 18173The first operand is a vector of ``i1`` and indicates the condition. The 18174second operand is the value that is merged where the condition vector is true. 18175The third operand is the value that is selected where the condition vector is 18176false or the lane position is greater equal than the pivot. The fourth operand 18177is the pivot. 18178 18179#. The optional ``fast-math flags`` marker indicates that the merge has one or 18180 more :ref:`fast-math flags <fastmath>`. These are optimization hints to 18181 enable otherwise unsafe floating-point optimizations. Fast-math flags are 18182 only valid for merges that return a floating-point scalar or vector type, 18183 or an array (nested to any depth) of floating-point scalar or vector types. 18184 18185Semantics: 18186"""""""""" 18187 18188The intrinsic selects lanes from the second and third operand depending on a 18189condition vector and pivot value. 18190 18191For all lanes where the condition vector is true and the lane position is less 18192than ``%pivot`` the lane is taken from the second operand. Otherwise, the lane 18193is taken from the third operand. 18194 18195Example: 18196"""""""" 18197 18198.. code-block:: llvm 18199 18200 %r = call <4 x i32> @llvm.vp.merge.v4i32(<4 x i1> %cond, <4 x i32> %on_true, <4 x i32> %on_false, i32 %pivot) 18201 18202 ;;; Expansion. 18203 ;; Lanes at and above %pivot are taken from %on_false 18204 %atfirst = insertelement <4 x i32> undef, i32 %pivot, i32 0 18205 %splat = shufflevector <4 x i32> %atfirst, <4 x i32> poison, <4 x i32> zeroinitializer 18206 %pivotmask = icmp ult <4 x i32> <i32 0, i32 1, i32 2, i32 3>, <4 x i32> %splat 18207 %mergemask = and <4 x i1> %cond, <4 x i1> %pivotmask 18208 %also.r = select <4 x i1> %mergemask, <4 x i32> %on_true, <4 x i32> %on_false 18209 18210 18211 18212.. _int_vp_add: 18213 18214'``llvm.vp.add.*``' Intrinsics 18215^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18216 18217Syntax: 18218""""""" 18219This is an overloaded intrinsic. 18220 18221:: 18222 18223 declare <16 x i32> @llvm.vp.add.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18224 declare <vscale x 4 x i32> @llvm.vp.add.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18225 declare <256 x i64> @llvm.vp.add.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18226 18227Overview: 18228""""""""" 18229 18230Predicated integer addition of two vectors of integers. 18231 18232 18233Arguments: 18234"""""""""" 18235 18236The first two operands and the result have the same vector of integer type. The 18237third operand is the vector mask and has the same number of elements as the 18238result vector type. The fourth operand is the explicit vector length of the 18239operation. 18240 18241Semantics: 18242"""""""""" 18243 18244The '``llvm.vp.add``' intrinsic performs integer addition (:ref:`add <i_add>`) 18245of the first and second vector operand on each enabled lane. The result on 18246disabled lanes is undefined. 18247 18248Examples: 18249""""""""" 18250 18251.. code-block:: llvm 18252 18253 %r = call <4 x i32> @llvm.vp.add.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18254 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18255 18256 %t = add <4 x i32> %a, %b 18257 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18258 18259.. _int_vp_sub: 18260 18261'``llvm.vp.sub.*``' Intrinsics 18262^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18263 18264Syntax: 18265""""""" 18266This is an overloaded intrinsic. 18267 18268:: 18269 18270 declare <16 x i32> @llvm.vp.sub.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18271 declare <vscale x 4 x i32> @llvm.vp.sub.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18272 declare <256 x i64> @llvm.vp.sub.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18273 18274Overview: 18275""""""""" 18276 18277Predicated integer subtraction of two vectors of integers. 18278 18279 18280Arguments: 18281"""""""""" 18282 18283The first two operands and the result have the same vector of integer type. The 18284third operand is the vector mask and has the same number of elements as the 18285result vector type. The fourth operand is the explicit vector length of the 18286operation. 18287 18288Semantics: 18289"""""""""" 18290 18291The '``llvm.vp.sub``' intrinsic performs integer subtraction 18292(:ref:`sub <i_sub>`) of the first and second vector operand on each enabled 18293lane. The result on disabled lanes is undefined. 18294 18295Examples: 18296""""""""" 18297 18298.. code-block:: llvm 18299 18300 %r = call <4 x i32> @llvm.vp.sub.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18301 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18302 18303 %t = sub <4 x i32> %a, %b 18304 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18305 18306 18307 18308.. _int_vp_mul: 18309 18310'``llvm.vp.mul.*``' Intrinsics 18311^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18312 18313Syntax: 18314""""""" 18315This is an overloaded intrinsic. 18316 18317:: 18318 18319 declare <16 x i32> @llvm.vp.mul.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18320 declare <vscale x 4 x i32> @llvm.vp.mul.nxv46i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18321 declare <256 x i64> @llvm.vp.mul.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18322 18323Overview: 18324""""""""" 18325 18326Predicated integer multiplication of two vectors of integers. 18327 18328 18329Arguments: 18330"""""""""" 18331 18332The first two operands and the result have the same vector of integer type. The 18333third operand is the vector mask and has the same number of elements as the 18334result vector type. The fourth operand is the explicit vector length of the 18335operation. 18336 18337Semantics: 18338"""""""""" 18339The '``llvm.vp.mul``' intrinsic performs integer multiplication 18340(:ref:`mul <i_mul>`) of the first and second vector operand on each enabled 18341lane. The result on disabled lanes is undefined. 18342 18343Examples: 18344""""""""" 18345 18346.. code-block:: llvm 18347 18348 %r = call <4 x i32> @llvm.vp.mul.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18349 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18350 18351 %t = mul <4 x i32> %a, %b 18352 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18353 18354 18355.. _int_vp_sdiv: 18356 18357'``llvm.vp.sdiv.*``' Intrinsics 18358^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18359 18360Syntax: 18361""""""" 18362This is an overloaded intrinsic. 18363 18364:: 18365 18366 declare <16 x i32> @llvm.vp.sdiv.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18367 declare <vscale x 4 x i32> @llvm.vp.sdiv.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18368 declare <256 x i64> @llvm.vp.sdiv.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18369 18370Overview: 18371""""""""" 18372 18373Predicated, signed division of two vectors of integers. 18374 18375 18376Arguments: 18377"""""""""" 18378 18379The first two operands and the result have the same vector of integer type. The 18380third operand is the vector mask and has the same number of elements as the 18381result vector type. The fourth operand is the explicit vector length of the 18382operation. 18383 18384Semantics: 18385"""""""""" 18386 18387The '``llvm.vp.sdiv``' intrinsic performs signed division (:ref:`sdiv <i_sdiv>`) 18388of the first and second vector operand on each enabled lane. The result on 18389disabled lanes is undefined. 18390 18391Examples: 18392""""""""" 18393 18394.. code-block:: llvm 18395 18396 %r = call <4 x i32> @llvm.vp.sdiv.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18397 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18398 18399 %t = sdiv <4 x i32> %a, %b 18400 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18401 18402 18403.. _int_vp_udiv: 18404 18405'``llvm.vp.udiv.*``' Intrinsics 18406^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18407 18408Syntax: 18409""""""" 18410This is an overloaded intrinsic. 18411 18412:: 18413 18414 declare <16 x i32> @llvm.vp.udiv.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18415 declare <vscale x 4 x i32> @llvm.vp.udiv.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18416 declare <256 x i64> @llvm.vp.udiv.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18417 18418Overview: 18419""""""""" 18420 18421Predicated, unsigned division of two vectors of integers. 18422 18423 18424Arguments: 18425"""""""""" 18426 18427The first two operands and the result have the same vector of integer type. The third operand is the vector mask and has the same number of elements as the result vector type. The fourth operand is the explicit vector length of the operation. 18428 18429Semantics: 18430"""""""""" 18431 18432The '``llvm.vp.udiv``' intrinsic performs unsigned division 18433(:ref:`udiv <i_udiv>`) of the first and second vector operand on each enabled 18434lane. The result on disabled lanes is undefined. 18435 18436Examples: 18437""""""""" 18438 18439.. code-block:: llvm 18440 18441 %r = call <4 x i32> @llvm.vp.udiv.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18442 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18443 18444 %t = udiv <4 x i32> %a, %b 18445 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18446 18447 18448 18449.. _int_vp_srem: 18450 18451'``llvm.vp.srem.*``' Intrinsics 18452^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18453 18454Syntax: 18455""""""" 18456This is an overloaded intrinsic. 18457 18458:: 18459 18460 declare <16 x i32> @llvm.vp.srem.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18461 declare <vscale x 4 x i32> @llvm.vp.srem.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18462 declare <256 x i64> @llvm.vp.srem.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18463 18464Overview: 18465""""""""" 18466 18467Predicated computations of the signed remainder of two integer vectors. 18468 18469 18470Arguments: 18471"""""""""" 18472 18473The first two operands and the result have the same vector of integer type. The 18474third operand is the vector mask and has the same number of elements as the 18475result vector type. The fourth operand is the explicit vector length of the 18476operation. 18477 18478Semantics: 18479"""""""""" 18480 18481The '``llvm.vp.srem``' intrinsic computes the remainder of the signed division 18482(:ref:`srem <i_srem>`) of the first and second vector operand on each enabled 18483lane. The result on disabled lanes is undefined. 18484 18485Examples: 18486""""""""" 18487 18488.. code-block:: llvm 18489 18490 %r = call <4 x i32> @llvm.vp.srem.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18491 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18492 18493 %t = srem <4 x i32> %a, %b 18494 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18495 18496 18497 18498.. _int_vp_urem: 18499 18500'``llvm.vp.urem.*``' Intrinsics 18501^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18502 18503Syntax: 18504""""""" 18505This is an overloaded intrinsic. 18506 18507:: 18508 18509 declare <16 x i32> @llvm.vp.urem.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18510 declare <vscale x 4 x i32> @llvm.vp.urem.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18511 declare <256 x i64> @llvm.vp.urem.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18512 18513Overview: 18514""""""""" 18515 18516Predicated computation of the unsigned remainder of two integer vectors. 18517 18518 18519Arguments: 18520"""""""""" 18521 18522The first two operands and the result have the same vector of integer type. The 18523third operand is the vector mask and has the same number of elements as the 18524result vector type. The fourth operand is the explicit vector length of the 18525operation. 18526 18527Semantics: 18528"""""""""" 18529 18530The '``llvm.vp.urem``' intrinsic computes the remainder of the unsigned division 18531(:ref:`urem <i_urem>`) of the first and second vector operand on each enabled 18532lane. The result on disabled lanes is undefined. 18533 18534Examples: 18535""""""""" 18536 18537.. code-block:: llvm 18538 18539 %r = call <4 x i32> @llvm.vp.urem.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18540 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18541 18542 %t = urem <4 x i32> %a, %b 18543 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18544 18545 18546.. _int_vp_ashr: 18547 18548'``llvm.vp.ashr.*``' Intrinsics 18549^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18550 18551Syntax: 18552""""""" 18553This is an overloaded intrinsic. 18554 18555:: 18556 18557 declare <16 x i32> @llvm.vp.ashr.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18558 declare <vscale x 4 x i32> @llvm.vp.ashr.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18559 declare <256 x i64> @llvm.vp.ashr.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18560 18561Overview: 18562""""""""" 18563 18564Vector-predicated arithmetic right-shift. 18565 18566 18567Arguments: 18568"""""""""" 18569 18570The first two operands and the result have the same vector of integer type. The 18571third operand is the vector mask and has the same number of elements as the 18572result vector type. The fourth operand is the explicit vector length of the 18573operation. 18574 18575Semantics: 18576"""""""""" 18577 18578The '``llvm.vp.ashr``' intrinsic computes the arithmetic right shift 18579(:ref:`ashr <i_ashr>`) of the first operand by the second operand on each 18580enabled lane. The result on disabled lanes is undefined. 18581 18582Examples: 18583""""""""" 18584 18585.. code-block:: llvm 18586 18587 %r = call <4 x i32> @llvm.vp.ashr.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18588 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18589 18590 %t = ashr <4 x i32> %a, %b 18591 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18592 18593 18594.. _int_vp_lshr: 18595 18596 18597'``llvm.vp.lshr.*``' Intrinsics 18598^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18599 18600Syntax: 18601""""""" 18602This is an overloaded intrinsic. 18603 18604:: 18605 18606 declare <16 x i32> @llvm.vp.lshr.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18607 declare <vscale x 4 x i32> @llvm.vp.lshr.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18608 declare <256 x i64> @llvm.vp.lshr.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18609 18610Overview: 18611""""""""" 18612 18613Vector-predicated logical right-shift. 18614 18615 18616Arguments: 18617"""""""""" 18618 18619The first two operands and the result have the same vector of integer type. The 18620third operand is the vector mask and has the same number of elements as the 18621result vector type. The fourth operand is the explicit vector length of the 18622operation. 18623 18624Semantics: 18625"""""""""" 18626 18627The '``llvm.vp.lshr``' intrinsic computes the logical right shift 18628(:ref:`lshr <i_lshr>`) of the first operand by the second operand on each 18629enabled lane. The result on disabled lanes is undefined. 18630 18631Examples: 18632""""""""" 18633 18634.. code-block:: llvm 18635 18636 %r = call <4 x i32> @llvm.vp.lshr.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18637 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18638 18639 %t = lshr <4 x i32> %a, %b 18640 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18641 18642 18643.. _int_vp_shl: 18644 18645'``llvm.vp.shl.*``' Intrinsics 18646^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18647 18648Syntax: 18649""""""" 18650This is an overloaded intrinsic. 18651 18652:: 18653 18654 declare <16 x i32> @llvm.vp.shl.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18655 declare <vscale x 4 x i32> @llvm.vp.shl.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18656 declare <256 x i64> @llvm.vp.shl.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18657 18658Overview: 18659""""""""" 18660 18661Vector-predicated left shift. 18662 18663 18664Arguments: 18665"""""""""" 18666 18667The first two operands and the result have the same vector of integer type. The 18668third operand is the vector mask and has the same number of elements as the 18669result vector type. The fourth operand is the explicit vector length of the 18670operation. 18671 18672Semantics: 18673"""""""""" 18674 18675The '``llvm.vp.shl``' intrinsic computes the left shift (:ref:`shl <i_shl>`) of 18676the first operand by the second operand on each enabled lane. The result on 18677disabled lanes is undefined. 18678 18679Examples: 18680""""""""" 18681 18682.. code-block:: llvm 18683 18684 %r = call <4 x i32> @llvm.vp.shl.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18685 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18686 18687 %t = shl <4 x i32> %a, %b 18688 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18689 18690 18691.. _int_vp_or: 18692 18693'``llvm.vp.or.*``' Intrinsics 18694^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18695 18696Syntax: 18697""""""" 18698This is an overloaded intrinsic. 18699 18700:: 18701 18702 declare <16 x i32> @llvm.vp.or.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18703 declare <vscale x 4 x i32> @llvm.vp.or.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18704 declare <256 x i64> @llvm.vp.or.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18705 18706Overview: 18707""""""""" 18708 18709Vector-predicated or. 18710 18711 18712Arguments: 18713"""""""""" 18714 18715The first two operands and the result have the same vector of integer type. The 18716third operand is the vector mask and has the same number of elements as the 18717result vector type. The fourth operand is the explicit vector length of the 18718operation. 18719 18720Semantics: 18721"""""""""" 18722 18723The '``llvm.vp.or``' intrinsic performs a bitwise or (:ref:`or <i_or>`) of the 18724first two operands on each enabled lane. The result on disabled lanes is 18725undefined. 18726 18727Examples: 18728""""""""" 18729 18730.. code-block:: llvm 18731 18732 %r = call <4 x i32> @llvm.vp.or.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18733 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18734 18735 %t = or <4 x i32> %a, %b 18736 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18737 18738 18739.. _int_vp_and: 18740 18741'``llvm.vp.and.*``' Intrinsics 18742^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18743 18744Syntax: 18745""""""" 18746This is an overloaded intrinsic. 18747 18748:: 18749 18750 declare <16 x i32> @llvm.vp.and.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18751 declare <vscale x 4 x i32> @llvm.vp.and.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18752 declare <256 x i64> @llvm.vp.and.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18753 18754Overview: 18755""""""""" 18756 18757Vector-predicated and. 18758 18759 18760Arguments: 18761"""""""""" 18762 18763The first two operands and the result have the same vector of integer type. The 18764third operand is the vector mask and has the same number of elements as the 18765result vector type. The fourth operand is the explicit vector length of the 18766operation. 18767 18768Semantics: 18769"""""""""" 18770 18771The '``llvm.vp.and``' intrinsic performs a bitwise and (:ref:`and <i_or>`) of 18772the first two operands on each enabled lane. The result on disabled lanes is 18773undefined. 18774 18775Examples: 18776""""""""" 18777 18778.. code-block:: llvm 18779 18780 %r = call <4 x i32> @llvm.vp.and.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18781 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18782 18783 %t = and <4 x i32> %a, %b 18784 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18785 18786 18787.. _int_vp_xor: 18788 18789'``llvm.vp.xor.*``' Intrinsics 18790^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18791 18792Syntax: 18793""""""" 18794This is an overloaded intrinsic. 18795 18796:: 18797 18798 declare <16 x i32> @llvm.vp.xor.v16i32 (<16 x i32> <left_op>, <16 x i32> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18799 declare <vscale x 4 x i32> @llvm.vp.xor.nxv4i32 (<vscale x 4 x i32> <left_op>, <vscale x 4 x i32> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18800 declare <256 x i64> @llvm.vp.xor.v256i64 (<256 x i64> <left_op>, <256 x i64> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18801 18802Overview: 18803""""""""" 18804 18805Vector-predicated, bitwise xor. 18806 18807 18808Arguments: 18809"""""""""" 18810 18811The first two operands and the result have the same vector of integer type. The 18812third operand is the vector mask and has the same number of elements as the 18813result vector type. The fourth operand is the explicit vector length of the 18814operation. 18815 18816Semantics: 18817"""""""""" 18818 18819The '``llvm.vp.xor``' intrinsic performs a bitwise xor (:ref:`xor <i_xor>`) of 18820the first two operands on each enabled lane. 18821The result on disabled lanes is undefined. 18822 18823Examples: 18824""""""""" 18825 18826.. code-block:: llvm 18827 18828 %r = call <4 x i32> @llvm.vp.xor.v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i1> %mask, i32 %evl) 18829 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18830 18831 %t = xor <4 x i32> %a, %b 18832 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 18833 18834 18835.. _int_vp_fadd: 18836 18837'``llvm.vp.fadd.*``' Intrinsics 18838^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18839 18840Syntax: 18841""""""" 18842This is an overloaded intrinsic. 18843 18844:: 18845 18846 declare <16 x float> @llvm.vp.fadd.v16f32 (<16 x float> <left_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18847 declare <vscale x 4 x float> @llvm.vp.fadd.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18848 declare <256 x double> @llvm.vp.fadd.v256f64 (<256 x double> <left_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18849 18850Overview: 18851""""""""" 18852 18853Predicated floating-point addition of two vectors of floating-point values. 18854 18855 18856Arguments: 18857"""""""""" 18858 18859The first two operands and the result have the same vector of floating-point type. The 18860third operand is the vector mask and has the same number of elements as the 18861result vector type. The fourth operand is the explicit vector length of the 18862operation. 18863 18864Semantics: 18865"""""""""" 18866 18867The '``llvm.vp.fadd``' intrinsic performs floating-point addition (:ref:`fadd <i_fadd>`) 18868of the first and second vector operand on each enabled lane. The result on 18869disabled lanes is undefined. The operation is performed in the default 18870floating-point environment. 18871 18872Examples: 18873""""""""" 18874 18875.. code-block:: llvm 18876 18877 %r = call <4 x float> @llvm.vp.fadd.v4f32(<4 x float> %a, <4 x float> %b, <4 x i1> %mask, i32 %evl) 18878 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18879 18880 %t = fadd <4 x float> %a, %b 18881 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 18882 18883 18884.. _int_vp_fsub: 18885 18886'``llvm.vp.fsub.*``' Intrinsics 18887^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18888 18889Syntax: 18890""""""" 18891This is an overloaded intrinsic. 18892 18893:: 18894 18895 declare <16 x float> @llvm.vp.fsub.v16f32 (<16 x float> <left_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18896 declare <vscale x 4 x float> @llvm.vp.fsub.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18897 declare <256 x double> @llvm.vp.fsub.v256f64 (<256 x double> <left_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18898 18899Overview: 18900""""""""" 18901 18902Predicated floating-point subtraction of two vectors of floating-point values. 18903 18904 18905Arguments: 18906"""""""""" 18907 18908The first two operands and the result have the same vector of floating-point type. The 18909third operand is the vector mask and has the same number of elements as the 18910result vector type. The fourth operand is the explicit vector length of the 18911operation. 18912 18913Semantics: 18914"""""""""" 18915 18916The '``llvm.vp.fsub``' intrinsic performs floating-point subtraction (:ref:`fsub <i_fsub>`) 18917of the first and second vector operand on each enabled lane. The result on 18918disabled lanes is undefined. The operation is performed in the default 18919floating-point environment. 18920 18921Examples: 18922""""""""" 18923 18924.. code-block:: llvm 18925 18926 %r = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %a, <4 x float> %b, <4 x i1> %mask, i32 %evl) 18927 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18928 18929 %t = fsub <4 x float> %a, %b 18930 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 18931 18932 18933.. _int_vp_fmul: 18934 18935'``llvm.vp.fmul.*``' Intrinsics 18936^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18937 18938Syntax: 18939""""""" 18940This is an overloaded intrinsic. 18941 18942:: 18943 18944 declare <16 x float> @llvm.vp.fmul.v16f32 (<16 x float> <left_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18945 declare <vscale x 4 x float> @llvm.vp.fmul.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18946 declare <256 x double> @llvm.vp.fmul.v256f64 (<256 x double> <left_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18947 18948Overview: 18949""""""""" 18950 18951Predicated floating-point multiplication of two vectors of floating-point values. 18952 18953 18954Arguments: 18955"""""""""" 18956 18957The first two operands and the result have the same vector of floating-point type. The 18958third operand is the vector mask and has the same number of elements as the 18959result vector type. The fourth operand is the explicit vector length of the 18960operation. 18961 18962Semantics: 18963"""""""""" 18964 18965The '``llvm.vp.fmul``' intrinsic performs floating-point multiplication (:ref:`fmul <i_fmul>`) 18966of the first and second vector operand on each enabled lane. The result on 18967disabled lanes is undefined. The operation is performed in the default 18968floating-point environment. 18969 18970Examples: 18971""""""""" 18972 18973.. code-block:: llvm 18974 18975 %r = call <4 x float> @llvm.vp.fmul.v4f32(<4 x float> %a, <4 x float> %b, <4 x i1> %mask, i32 %evl) 18976 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 18977 18978 %t = fmul <4 x float> %a, %b 18979 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 18980 18981 18982.. _int_vp_fdiv: 18983 18984'``llvm.vp.fdiv.*``' Intrinsics 18985^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 18986 18987Syntax: 18988""""""" 18989This is an overloaded intrinsic. 18990 18991:: 18992 18993 declare <16 x float> @llvm.vp.fdiv.v16f32 (<16 x float> <left_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 18994 declare <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 18995 declare <256 x double> @llvm.vp.fdiv.v256f64 (<256 x double> <left_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 18996 18997Overview: 18998""""""""" 18999 19000Predicated floating-point division of two vectors of floating-point values. 19001 19002 19003Arguments: 19004"""""""""" 19005 19006The first two operands and the result have the same vector of floating-point type. The 19007third operand is the vector mask and has the same number of elements as the 19008result vector type. The fourth operand is the explicit vector length of the 19009operation. 19010 19011Semantics: 19012"""""""""" 19013 19014The '``llvm.vp.fdiv``' intrinsic performs floating-point division (:ref:`fdiv <i_fdiv>`) 19015of the first and second vector operand on each enabled lane. The result on 19016disabled lanes is undefined. The operation is performed in the default 19017floating-point environment. 19018 19019Examples: 19020""""""""" 19021 19022.. code-block:: llvm 19023 19024 %r = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %a, <4 x float> %b, <4 x i1> %mask, i32 %evl) 19025 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 19026 19027 %t = fdiv <4 x float> %a, %b 19028 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 19029 19030 19031.. _int_vp_frem: 19032 19033'``llvm.vp.frem.*``' Intrinsics 19034^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19035 19036Syntax: 19037""""""" 19038This is an overloaded intrinsic. 19039 19040:: 19041 19042 declare <16 x float> @llvm.vp.frem.v16f32 (<16 x float> <left_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 19043 declare <vscale x 4 x float> @llvm.vp.frem.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 19044 declare <256 x double> @llvm.vp.frem.v256f64 (<256 x double> <left_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 19045 19046Overview: 19047""""""""" 19048 19049Predicated floating-point remainder of two vectors of floating-point values. 19050 19051 19052Arguments: 19053"""""""""" 19054 19055The first two operands and the result have the same vector of floating-point type. The 19056third operand is the vector mask and has the same number of elements as the 19057result vector type. The fourth operand is the explicit vector length of the 19058operation. 19059 19060Semantics: 19061"""""""""" 19062 19063The '``llvm.vp.frem``' intrinsic performs floating-point remainder (:ref:`frem <i_frem>`) 19064of the first and second vector operand on each enabled lane. The result on 19065disabled lanes is undefined. The operation is performed in the default 19066floating-point environment. 19067 19068Examples: 19069""""""""" 19070 19071.. code-block:: llvm 19072 19073 %r = call <4 x float> @llvm.vp.frem.v4f32(<4 x float> %a, <4 x float> %b, <4 x i1> %mask, i32 %evl) 19074 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 19075 19076 %t = frem <4 x float> %a, %b 19077 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 19078 19079 19080.. _int_vp_fneg: 19081 19082'``llvm.vp.fneg.*``' Intrinsics 19083^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19084 19085Syntax: 19086""""""" 19087This is an overloaded intrinsic. 19088 19089:: 19090 19091 declare <16 x float> @llvm.vp.fneg.v16f32 (<16 x float> <op>, <16 x i1> <mask>, i32 <vector_length>) 19092 declare <vscale x 4 x float> @llvm.vp.fneg.nxv4f32 (<vscale x 4 x float> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 19093 declare <256 x double> @llvm.vp.fneg.v256f64 (<256 x double> <op>, <256 x i1> <mask>, i32 <vector_length>) 19094 19095Overview: 19096""""""""" 19097 19098Predicated floating-point negation of a vector of floating-point values. 19099 19100 19101Arguments: 19102"""""""""" 19103 19104The first operand and the result have the same vector of floating-point type. 19105The second operand is the vector mask and has the same number of elements as the 19106result vector type. The third operand is the explicit vector length of the 19107operation. 19108 19109Semantics: 19110"""""""""" 19111 19112The '``llvm.vp.fneg``' intrinsic performs floating-point negation (:ref:`fneg <i_fneg>`) 19113of the first vector operand on each enabled lane. The result on disabled lanes 19114is undefined. 19115 19116Examples: 19117""""""""" 19118 19119.. code-block:: llvm 19120 19121 %r = call <4 x float> @llvm.vp.fneg.v4f32(<4 x float> %a, <4 x i1> %mask, i32 %evl) 19122 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 19123 19124 %t = fneg <4 x float> %a 19125 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 19126 19127 19128.. _int_vp_fma: 19129 19130'``llvm.vp.fma.*``' Intrinsics 19131^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19132 19133Syntax: 19134""""""" 19135This is an overloaded intrinsic. 19136 19137:: 19138 19139 declare <16 x float> @llvm.vp.fma.v16f32 (<16 x float> <left_op>, <16 x float> <middle_op>, <16 x float> <right_op>, <16 x i1> <mask>, i32 <vector_length>) 19140 declare <vscale x 4 x float> @llvm.vp.fma.nxv4f32 (<vscale x 4 x float> <left_op>, <vscale x 4 x float> <middle_op>, <vscale x 4 x float> <right_op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 19141 declare <256 x double> @llvm.vp.fma.v256f64 (<256 x double> <left_op>, <256 x double> <middle_op>, <256 x double> <right_op>, <256 x i1> <mask>, i32 <vector_length>) 19142 19143Overview: 19144""""""""" 19145 19146Predicated floating-point fused multiply-add of two vectors of floating-point values. 19147 19148 19149Arguments: 19150"""""""""" 19151 19152The first three operands and the result have the same vector of floating-point type. The 19153fourth operand is the vector mask and has the same number of elements as the 19154result vector type. The fifth operand is the explicit vector length of the 19155operation. 19156 19157Semantics: 19158"""""""""" 19159 19160The '``llvm.vp.fma``' intrinsic performs floating-point fused multiply-add (:ref:`llvm.fma <int_fma>`) 19161of the first, second, and third vector operand on each enabled lane. The result on 19162disabled lanes is undefined. The operation is performed in the default 19163floating-point environment. 19164 19165Examples: 19166""""""""" 19167 19168.. code-block:: llvm 19169 19170 %r = call <4 x float> @llvm.vp.fma.v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x i1> %mask, i32 %evl) 19171 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 19172 19173 %t = call <4 x float> @llvm.fma(<4 x float> %a, <4 x float> %b, <4 x float> %c) 19174 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 19175 19176 19177.. _int_vp_reduce_add: 19178 19179'``llvm.vp.reduce.add.*``' Intrinsics 19180^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19181 19182Syntax: 19183""""""" 19184This is an overloaded intrinsic. 19185 19186:: 19187 19188 declare i32 @llvm.vp.reduce.add.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19189 declare i16 @llvm.vp.reduce.add.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19190 19191Overview: 19192""""""""" 19193 19194Predicated integer ``ADD`` reduction of a vector and a scalar starting value, 19195returning the result as a scalar. 19196 19197Arguments: 19198"""""""""" 19199 19200The first operand is the start value of the reduction, which must be a scalar 19201integer type equal to the result type. The second operand is the vector on 19202which the reduction is performed and must be a vector of integer values whose 19203element type is the result/start type. The third operand is the vector mask and 19204is a vector of boolean values with the same number of elements as the vector 19205operand. The fourth operand is the explicit vector length of the operation. 19206 19207Semantics: 19208"""""""""" 19209 19210The '``llvm.vp.reduce.add``' intrinsic performs the integer ``ADD`` reduction 19211(:ref:`llvm.vector.reduce.add <int_vector_reduce_add>`) of the vector operand 19212``val`` on each enabled lane, adding it to the scalar ``start_value``. Disabled 19213lanes are treated as containing the neutral value ``0`` (i.e. having no effect 19214on the reduction operation). If the vector length is zero, the result is equal 19215to ``start_value``. 19216 19217To ignore the start value, the neutral value can be used. 19218 19219Examples: 19220""""""""" 19221 19222.. code-block:: llvm 19223 19224 %r = call i32 @llvm.vp.reduce.add.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19225 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19226 ; are treated as though %mask were false for those lanes. 19227 19228 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> zeroinitializer 19229 %reduction = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %masked.a) 19230 %also.r = add i32 %reduction, %start 19231 19232 19233.. _int_vp_reduce_fadd: 19234 19235'``llvm.vp.reduce.fadd.*``' Intrinsics 19236^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19237 19238Syntax: 19239""""""" 19240This is an overloaded intrinsic. 19241 19242:: 19243 19244 declare float @llvm.vp.reduce.fadd.v4f32(float <start_value>, <4 x float> <val>, <4 x i1> <mask>, i32 <vector_length>) 19245 declare double @llvm.vp.reduce.fadd.nxv8f64(double <start_value>, <vscale x 8 x double> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19246 19247Overview: 19248""""""""" 19249 19250Predicated floating-point ``ADD`` reduction of a vector and a scalar starting 19251value, returning the result as a scalar. 19252 19253Arguments: 19254"""""""""" 19255 19256The first operand is the start value of the reduction, which must be a scalar 19257floating-point type equal to the result type. The second operand is the vector 19258on which the reduction is performed and must be a vector of floating-point 19259values whose element type is the result/start type. The third operand is the 19260vector mask and is a vector of boolean values with the same number of elements 19261as the vector operand. The fourth operand is the explicit vector length of the 19262operation. 19263 19264Semantics: 19265"""""""""" 19266 19267The '``llvm.vp.reduce.fadd``' intrinsic performs the floating-point ``ADD`` 19268reduction (:ref:`llvm.vector.reduce.fadd <int_vector_reduce_fadd>`) of the 19269vector operand ``val`` on each enabled lane, adding it to the scalar 19270``start_value``. Disabled lanes are treated as containing the neutral value 19271``-0.0`` (i.e. having no effect on the reduction operation). If no lanes are 19272enabled, the resulting value will be equal to ``start_value``. 19273 19274To ignore the start value, the neutral value can be used. 19275 19276See the unpredicated version (:ref:`llvm.vector.reduce.fadd 19277<int_vector_reduce_fadd>`) for more detail on the semantics of the reduction. 19278 19279Examples: 19280""""""""" 19281 19282.. code-block:: llvm 19283 19284 %r = call float @llvm.vp.reduce.fadd.v4f32(float %start, <4 x float> %a, <4 x i1> %mask, i32 %evl) 19285 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19286 ; are treated as though %mask were false for those lanes. 19287 19288 %masked.a = select <4 x i1> %mask, <4 x float> %a, <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0> 19289 %also.r = call float @llvm.vector.reduce.fadd.v4f32(float %start, <4 x float> %masked.a) 19290 19291 19292.. _int_vp_reduce_mul: 19293 19294'``llvm.vp.reduce.mul.*``' Intrinsics 19295^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19296 19297Syntax: 19298""""""" 19299This is an overloaded intrinsic. 19300 19301:: 19302 19303 declare i32 @llvm.vp.reduce.mul.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19304 declare i16 @llvm.vp.reduce.mul.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19305 19306Overview: 19307""""""""" 19308 19309Predicated integer ``MUL`` reduction of a vector and a scalar starting value, 19310returning the result as a scalar. 19311 19312 19313Arguments: 19314"""""""""" 19315 19316The first operand is the start value of the reduction, which must be a scalar 19317integer type equal to the result type. The second operand is the vector on 19318which the reduction is performed and must be a vector of integer values whose 19319element type is the result/start type. The third operand is the vector mask and 19320is a vector of boolean values with the same number of elements as the vector 19321operand. The fourth operand is the explicit vector length of the operation. 19322 19323Semantics: 19324"""""""""" 19325 19326The '``llvm.vp.reduce.mul``' intrinsic performs the integer ``MUL`` reduction 19327(:ref:`llvm.vector.reduce.mul <int_vector_reduce_mul>`) of the vector operand ``val`` 19328on each enabled lane, multiplying it by the scalar ``start_value``. Disabled 19329lanes are treated as containing the neutral value ``1`` (i.e. having no effect 19330on the reduction operation). If the vector length is zero, the result is the 19331start value. 19332 19333To ignore the start value, the neutral value can be used. 19334 19335Examples: 19336""""""""" 19337 19338.. code-block:: llvm 19339 19340 %r = call i32 @llvm.vp.reduce.mul.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19341 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19342 ; are treated as though %mask were false for those lanes. 19343 19344 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 1, i32 1, i32 1, i32 1> 19345 %reduction = call i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %masked.a) 19346 %also.r = mul i32 %reduction, %start 19347 19348.. _int_vp_reduce_fmul: 19349 19350'``llvm.vp.reduce.fmul.*``' Intrinsics 19351^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19352 19353Syntax: 19354""""""" 19355This is an overloaded intrinsic. 19356 19357:: 19358 19359 declare float @llvm.vp.reduce.fmul.v4f32(float <start_value>, <4 x float> <val>, <4 x i1> <mask>, i32 <vector_length>) 19360 declare double @llvm.vp.reduce.fmul.nxv8f64(double <start_value>, <vscale x 8 x double> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19361 19362Overview: 19363""""""""" 19364 19365Predicated floating-point ``MUL`` reduction of a vector and a scalar starting 19366value, returning the result as a scalar. 19367 19368 19369Arguments: 19370"""""""""" 19371 19372The first operand is the start value of the reduction, which must be a scalar 19373floating-point type equal to the result type. The second operand is the vector 19374on which the reduction is performed and must be a vector of floating-point 19375values whose element type is the result/start type. The third operand is the 19376vector mask and is a vector of boolean values with the same number of elements 19377as the vector operand. The fourth operand is the explicit vector length of the 19378operation. 19379 19380Semantics: 19381"""""""""" 19382 19383The '``llvm.vp.reduce.fmul``' intrinsic performs the floating-point ``MUL`` 19384reduction (:ref:`llvm.vector.reduce.fmul <int_vector_reduce_fmul>`) of the 19385vector operand ``val`` on each enabled lane, multiplying it by the scalar 19386`start_value``. Disabled lanes are treated as containing the neutral value 19387``1.0`` (i.e. having no effect on the reduction operation). If no lanes are 19388enabled, the resulting value will be equal to the starting value. 19389 19390To ignore the start value, the neutral value can be used. 19391 19392See the unpredicated version (:ref:`llvm.vector.reduce.fmul 19393<int_vector_reduce_fmul>`) for more detail on the semantics. 19394 19395Examples: 19396""""""""" 19397 19398.. code-block:: llvm 19399 19400 %r = call float @llvm.vp.reduce.fmul.v4f32(float %start, <4 x float> %a, <4 x i1> %mask, i32 %evl) 19401 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19402 ; are treated as though %mask were false for those lanes. 19403 19404 %masked.a = select <4 x i1> %mask, <4 x float> %a, <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0> 19405 %also.r = call float @llvm.vector.reduce.fmul.v4f32(float %start, <4 x float> %masked.a) 19406 19407 19408.. _int_vp_reduce_and: 19409 19410'``llvm.vp.reduce.and.*``' Intrinsics 19411^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19412 19413Syntax: 19414""""""" 19415This is an overloaded intrinsic. 19416 19417:: 19418 19419 declare i32 @llvm.vp.reduce.and.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19420 declare i16 @llvm.vp.reduce.and.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19421 19422Overview: 19423""""""""" 19424 19425Predicated integer ``AND`` reduction of a vector and a scalar starting value, 19426returning the result as a scalar. 19427 19428 19429Arguments: 19430"""""""""" 19431 19432The first operand is the start value of the reduction, which must be a scalar 19433integer type equal to the result type. The second operand is the vector on 19434which the reduction is performed and must be a vector of integer values whose 19435element type is the result/start type. The third operand is the vector mask and 19436is a vector of boolean values with the same number of elements as the vector 19437operand. The fourth operand is the explicit vector length of the operation. 19438 19439Semantics: 19440"""""""""" 19441 19442The '``llvm.vp.reduce.and``' intrinsic performs the integer ``AND`` reduction 19443(:ref:`llvm.vector.reduce.and <int_vector_reduce_and>`) of the vector operand 19444``val`` on each enabled lane, performing an '``and``' of that with with the 19445scalar ``start_value``. Disabled lanes are treated as containing the neutral 19446value ``UINT_MAX``, or ``-1`` (i.e. having no effect on the reduction 19447operation). If the vector length is zero, the result is the start value. 19448 19449To ignore the start value, the neutral value can be used. 19450 19451Examples: 19452""""""""" 19453 19454.. code-block:: llvm 19455 19456 %r = call i32 @llvm.vp.reduce.and.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19457 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19458 ; are treated as though %mask were false for those lanes. 19459 19460 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 -1, i32 -1, i32 -1, i32 -1> 19461 %reduction = call i32 @llvm.vector.reduce.and.v4i32(<4 x i32> %masked.a) 19462 %also.r = and i32 %reduction, %start 19463 19464 19465.. _int_vp_reduce_or: 19466 19467'``llvm.vp.reduce.or.*``' Intrinsics 19468^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19469 19470Syntax: 19471""""""" 19472This is an overloaded intrinsic. 19473 19474:: 19475 19476 declare i32 @llvm.vp.reduce.or.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19477 declare i16 @llvm.vp.reduce.or.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19478 19479Overview: 19480""""""""" 19481 19482Predicated integer ``OR`` reduction of a vector and a scalar starting value, 19483returning the result as a scalar. 19484 19485 19486Arguments: 19487"""""""""" 19488 19489The first operand is the start value of the reduction, which must be a scalar 19490integer type equal to the result type. The second operand is the vector on 19491which the reduction is performed and must be a vector of integer values whose 19492element type is the result/start type. The third operand is the vector mask and 19493is a vector of boolean values with the same number of elements as the vector 19494operand. The fourth operand is the explicit vector length of the operation. 19495 19496Semantics: 19497"""""""""" 19498 19499The '``llvm.vp.reduce.or``' intrinsic performs the integer ``OR`` reduction 19500(:ref:`llvm.vector.reduce.or <int_vector_reduce_or>`) of the vector operand 19501``val`` on each enabled lane, performing an '``or``' of that with the scalar 19502``start_value``. Disabled lanes are treated as containing the neutral value 19503``0`` (i.e. having no effect on the reduction operation). If the vector length 19504is zero, the result is the start value. 19505 19506To ignore the start value, the neutral value can be used. 19507 19508Examples: 19509""""""""" 19510 19511.. code-block:: llvm 19512 19513 %r = call i32 @llvm.vp.reduce.or.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19514 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19515 ; are treated as though %mask were false for those lanes. 19516 19517 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 0, i32 0, i32 0, i32 0> 19518 %reduction = call i32 @llvm.vector.reduce.or.v4i32(<4 x i32> %masked.a) 19519 %also.r = or i32 %reduction, %start 19520 19521.. _int_vp_reduce_xor: 19522 19523'``llvm.vp.reduce.xor.*``' Intrinsics 19524^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19525 19526Syntax: 19527""""""" 19528This is an overloaded intrinsic. 19529 19530:: 19531 19532 declare i32 @llvm.vp.reduce.xor.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19533 declare i16 @llvm.vp.reduce.xor.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19534 19535Overview: 19536""""""""" 19537 19538Predicated integer ``XOR`` reduction of a vector and a scalar starting value, 19539returning the result as a scalar. 19540 19541 19542Arguments: 19543"""""""""" 19544 19545The first operand is the start value of the reduction, which must be a scalar 19546integer type equal to the result type. The second operand is the vector on 19547which the reduction is performed and must be a vector of integer values whose 19548element type is the result/start type. The third operand is the vector mask and 19549is a vector of boolean values with the same number of elements as the vector 19550operand. The fourth operand is the explicit vector length of the operation. 19551 19552Semantics: 19553"""""""""" 19554 19555The '``llvm.vp.reduce.xor``' intrinsic performs the integer ``XOR`` reduction 19556(:ref:`llvm.vector.reduce.xor <int_vector_reduce_xor>`) of the vector operand 19557``val`` on each enabled lane, performing an '``xor``' of that with the scalar 19558``start_value``. Disabled lanes are treated as containing the neutral value 19559``0`` (i.e. having no effect on the reduction operation). If the vector length 19560is zero, the result is the start value. 19561 19562To ignore the start value, the neutral value can be used. 19563 19564Examples: 19565""""""""" 19566 19567.. code-block:: llvm 19568 19569 %r = call i32 @llvm.vp.reduce.xor.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19570 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19571 ; are treated as though %mask were false for those lanes. 19572 19573 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 0, i32 0, i32 0, i32 0> 19574 %reduction = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> %masked.a) 19575 %also.r = xor i32 %reduction, %start 19576 19577 19578.. _int_vp_reduce_smax: 19579 19580'``llvm.vp.reduce.smax.*``' Intrinsics 19581^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19582 19583Syntax: 19584""""""" 19585This is an overloaded intrinsic. 19586 19587:: 19588 19589 declare i32 @llvm.vp.reduce.smax.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19590 declare i16 @llvm.vp.reduce.smax.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19591 19592Overview: 19593""""""""" 19594 19595Predicated signed-integer ``MAX`` reduction of a vector and a scalar starting 19596value, returning the result as a scalar. 19597 19598 19599Arguments: 19600"""""""""" 19601 19602The first operand is the start value of the reduction, which must be a scalar 19603integer type equal to the result type. The second operand is the vector on 19604which the reduction is performed and must be a vector of integer values whose 19605element type is the result/start type. The third operand is the vector mask and 19606is a vector of boolean values with the same number of elements as the vector 19607operand. The fourth operand is the explicit vector length of the operation. 19608 19609Semantics: 19610"""""""""" 19611 19612The '``llvm.vp.reduce.smax``' intrinsic performs the signed-integer ``MAX`` 19613reduction (:ref:`llvm.vector.reduce.smax <int_vector_reduce_smax>`) of the 19614vector operand ``val`` on each enabled lane, and taking the maximum of that and 19615the scalar ``start_value``. Disabled lanes are treated as containing the 19616neutral value ``INT_MIN`` (i.e. having no effect on the reduction operation). 19617If the vector length is zero, the result is the start value. 19618 19619To ignore the start value, the neutral value can be used. 19620 19621Examples: 19622""""""""" 19623 19624.. code-block:: llvm 19625 19626 %r = call i8 @llvm.vp.reduce.smax.v4i8(i8 %start, <4 x i8> %a, <4 x i1> %mask, i32 %evl) 19627 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19628 ; are treated as though %mask were false for those lanes. 19629 19630 %masked.a = select <4 x i1> %mask, <4 x i8> %a, <4 x i8> <i8 -128, i8 -128, i8 -128, i8 -128> 19631 %reduction = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> %masked.a) 19632 %also.r = call i8 @llvm.smax.i8(i8 %reduction, i8 %start) 19633 19634 19635.. _int_vp_reduce_smin: 19636 19637'``llvm.vp.reduce.smin.*``' Intrinsics 19638^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19639 19640Syntax: 19641""""""" 19642This is an overloaded intrinsic. 19643 19644:: 19645 19646 declare i32 @llvm.vp.reduce.smin.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19647 declare i16 @llvm.vp.reduce.smin.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19648 19649Overview: 19650""""""""" 19651 19652Predicated signed-integer ``MIN`` reduction of a vector and a scalar starting 19653value, returning the result as a scalar. 19654 19655 19656Arguments: 19657"""""""""" 19658 19659The first operand is the start value of the reduction, which must be a scalar 19660integer type equal to the result type. The second operand is the vector on 19661which the reduction is performed and must be a vector of integer values whose 19662element type is the result/start type. The third operand is the vector mask and 19663is a vector of boolean values with the same number of elements as the vector 19664operand. The fourth operand is the explicit vector length of the operation. 19665 19666Semantics: 19667"""""""""" 19668 19669The '``llvm.vp.reduce.smin``' intrinsic performs the signed-integer ``MIN`` 19670reduction (:ref:`llvm.vector.reduce.smin <int_vector_reduce_smin>`) of the 19671vector operand ``val`` on each enabled lane, and taking the minimum of that and 19672the scalar ``start_value``. Disabled lanes are treated as containing the 19673neutral value ``INT_MAX`` (i.e. having no effect on the reduction operation). 19674If the vector length is zero, the result is the start value. 19675 19676To ignore the start value, the neutral value can be used. 19677 19678Examples: 19679""""""""" 19680 19681.. code-block:: llvm 19682 19683 %r = call i8 @llvm.vp.reduce.smin.v4i8(i8 %start, <4 x i8> %a, <4 x i1> %mask, i32 %evl) 19684 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19685 ; are treated as though %mask were false for those lanes. 19686 19687 %masked.a = select <4 x i1> %mask, <4 x i8> %a, <4 x i8> <i8 127, i8 127, i8 127, i8 127> 19688 %reduction = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> %masked.a) 19689 %also.r = call i8 @llvm.smin.i8(i8 %reduction, i8 %start) 19690 19691 19692.. _int_vp_reduce_umax: 19693 19694'``llvm.vp.reduce.umax.*``' Intrinsics 19695^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19696 19697Syntax: 19698""""""" 19699This is an overloaded intrinsic. 19700 19701:: 19702 19703 declare i32 @llvm.vp.reduce.umax.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19704 declare i16 @llvm.vp.reduce.umax.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19705 19706Overview: 19707""""""""" 19708 19709Predicated unsigned-integer ``MAX`` reduction of a vector and a scalar starting 19710value, returning the result as a scalar. 19711 19712 19713Arguments: 19714"""""""""" 19715 19716The first operand is the start value of the reduction, which must be a scalar 19717integer type equal to the result type. The second operand is the vector on 19718which the reduction is performed and must be a vector of integer values whose 19719element type is the result/start type. The third operand is the vector mask and 19720is a vector of boolean values with the same number of elements as the vector 19721operand. The fourth operand is the explicit vector length of the operation. 19722 19723Semantics: 19724"""""""""" 19725 19726The '``llvm.vp.reduce.umax``' intrinsic performs the unsigned-integer ``MAX`` 19727reduction (:ref:`llvm.vector.reduce.umax <int_vector_reduce_umax>`) of the 19728vector operand ``val`` on each enabled lane, and taking the maximum of that and 19729the scalar ``start_value``. Disabled lanes are treated as containing the 19730neutral value ``0`` (i.e. having no effect on the reduction operation). If the 19731vector length is zero, the result is the start value. 19732 19733To ignore the start value, the neutral value can be used. 19734 19735Examples: 19736""""""""" 19737 19738.. code-block:: llvm 19739 19740 %r = call i32 @llvm.vp.reduce.umax.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19741 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19742 ; are treated as though %mask were false for those lanes. 19743 19744 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 0, i32 0, i32 0, i32 0> 19745 %reduction = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> %masked.a) 19746 %also.r = call i32 @llvm.umax.i32(i32 %reduction, i32 %start) 19747 19748 19749.. _int_vp_reduce_umin: 19750 19751'``llvm.vp.reduce.umin.*``' Intrinsics 19752^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19753 19754Syntax: 19755""""""" 19756This is an overloaded intrinsic. 19757 19758:: 19759 19760 declare i32 @llvm.vp.reduce.umin.v4i32(i32 <start_value>, <4 x i32> <val>, <4 x i1> <mask>, i32 <vector_length>) 19761 declare i16 @llvm.vp.reduce.umin.nxv8i16(i16 <start_value>, <vscale x 8 x i16> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19762 19763Overview: 19764""""""""" 19765 19766Predicated unsigned-integer ``MIN`` reduction of a vector and a scalar starting 19767value, returning the result as a scalar. 19768 19769 19770Arguments: 19771"""""""""" 19772 19773The first operand is the start value of the reduction, which must be a scalar 19774integer type equal to the result type. The second operand is the vector on 19775which the reduction is performed and must be a vector of integer values whose 19776element type is the result/start type. The third operand is the vector mask and 19777is a vector of boolean values with the same number of elements as the vector 19778operand. The fourth operand is the explicit vector length of the operation. 19779 19780Semantics: 19781"""""""""" 19782 19783The '``llvm.vp.reduce.umin``' intrinsic performs the unsigned-integer ``MIN`` 19784reduction (:ref:`llvm.vector.reduce.umin <int_vector_reduce_umin>`) of the 19785vector operand ``val`` on each enabled lane, taking the minimum of that and the 19786scalar ``start_value``. Disabled lanes are treated as containing the neutral 19787value ``UINT_MAX``, or ``-1`` (i.e. having no effect on the reduction 19788operation). If the vector length is zero, the result is the start value. 19789 19790To ignore the start value, the neutral value can be used. 19791 19792Examples: 19793""""""""" 19794 19795.. code-block:: llvm 19796 19797 %r = call i32 @llvm.vp.reduce.umin.v4i32(i32 %start, <4 x i32> %a, <4 x i1> %mask, i32 %evl) 19798 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19799 ; are treated as though %mask were false for those lanes. 19800 19801 %masked.a = select <4 x i1> %mask, <4 x i32> %a, <4 x i32> <i32 -1, i32 -1, i32 -1, i32 -1> 19802 %reduction = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> %masked.a) 19803 %also.r = call i32 @llvm.umin.i32(i32 %reduction, i32 %start) 19804 19805 19806.. _int_vp_reduce_fmax: 19807 19808'``llvm.vp.reduce.fmax.*``' Intrinsics 19809^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19810 19811Syntax: 19812""""""" 19813This is an overloaded intrinsic. 19814 19815:: 19816 19817 declare float @llvm.vp.reduce.fmax.v4f32(float <start_value>, <4 x float> <val>, <4 x i1> <mask>, float <vector_length>) 19818 declare double @llvm.vp.reduce.fmax.nxv8f64(double <start_value>, <vscale x 8 x double> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19819 19820Overview: 19821""""""""" 19822 19823Predicated floating-point ``MAX`` reduction of a vector and a scalar starting 19824value, returning the result as a scalar. 19825 19826 19827Arguments: 19828"""""""""" 19829 19830The first operand is the start value of the reduction, which must be a scalar 19831floating-point type equal to the result type. The second operand is the vector 19832on which the reduction is performed and must be a vector of floating-point 19833values whose element type is the result/start type. The third operand is the 19834vector mask and is a vector of boolean values with the same number of elements 19835as the vector operand. The fourth operand is the explicit vector length of the 19836operation. 19837 19838Semantics: 19839"""""""""" 19840 19841The '``llvm.vp.reduce.fmax``' intrinsic performs the floating-point ``MAX`` 19842reduction (:ref:`llvm.vector.reduce.fmax <int_vector_reduce_fmax>`) of the 19843vector operand ``val`` on each enabled lane, taking the maximum of that and the 19844scalar ``start_value``. Disabled lanes are treated as containing the neutral 19845value (i.e. having no effect on the reduction operation). If the vector length 19846is zero, the result is the start value. 19847 19848The neutral value is dependent on the :ref:`fast-math flags <fastmath>`. If no 19849flags are set, the neutral value is ``-QNAN``. If ``nnan`` and ``ninf`` are 19850both set, then the neutral value is the smallest floating-point value for the 19851result type. If only ``nnan`` is set then the neutral value is ``-Infinity``. 19852 19853This instruction has the same comparison semantics as the 19854:ref:`llvm.vector.reduce.fmax <int_vector_reduce_fmax>` intrinsic (and thus the 19855'``llvm.maxnum.*``' intrinsic). That is, the result will always be a number 19856unless all elements of the vector and the starting value are ``NaN``. For a 19857vector with maximum element magnitude ``0.0`` and containing both ``+0.0`` and 19858``-0.0`` elements, the sign of the result is unspecified. 19859 19860To ignore the start value, the neutral value can be used. 19861 19862Examples: 19863""""""""" 19864 19865.. code-block:: llvm 19866 19867 %r = call float @llvm.vp.reduce.fmax.v4f32(float %float, <4 x float> %a, <4 x i1> %mask, i32 %evl) 19868 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19869 ; are treated as though %mask were false for those lanes. 19870 19871 %masked.a = select <4 x i1> %mask, <4 x float> %a, <4 x float> <float QNAN, float QNAN, float QNAN, float QNAN> 19872 %reduction = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> %masked.a) 19873 %also.r = call float @llvm.maxnum.f32(float %reduction, float %start) 19874 19875 19876.. _int_vp_reduce_fmin: 19877 19878'``llvm.vp.reduce.fmin.*``' Intrinsics 19879^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19880 19881Syntax: 19882""""""" 19883This is an overloaded intrinsic. 19884 19885:: 19886 19887 declare float @llvm.vp.reduce.fmin.v4f32(float <start_value>, <4 x float> <val>, <4 x i1> <mask>, float <vector_length>) 19888 declare double @llvm.vp.reduce.fmin.nxv8f64(double <start_value>, <vscale x 8 x double> <val>, <vscale x 8 x i1> <mask>, i32 <vector_length>) 19889 19890Overview: 19891""""""""" 19892 19893Predicated floating-point ``MIN`` reduction of a vector and a scalar starting 19894value, returning the result as a scalar. 19895 19896 19897Arguments: 19898"""""""""" 19899 19900The first operand is the start value of the reduction, which must be a scalar 19901floating-point type equal to the result type. The second operand is the vector 19902on which the reduction is performed and must be a vector of floating-point 19903values whose element type is the result/start type. The third operand is the 19904vector mask and is a vector of boolean values with the same number of elements 19905as the vector operand. The fourth operand is the explicit vector length of the 19906operation. 19907 19908Semantics: 19909"""""""""" 19910 19911The '``llvm.vp.reduce.fmin``' intrinsic performs the floating-point ``MIN`` 19912reduction (:ref:`llvm.vector.reduce.fmin <int_vector_reduce_fmin>`) of the 19913vector operand ``val`` on each enabled lane, taking the minimum of that and the 19914scalar ``start_value``. Disabled lanes are treated as containing the neutral 19915value (i.e. having no effect on the reduction operation). If the vector length 19916is zero, the result is the start value. 19917 19918The neutral value is dependent on the :ref:`fast-math flags <fastmath>`. If no 19919flags are set, the neutral value is ``+QNAN``. If ``nnan`` and ``ninf`` are 19920both set, then the neutral value is the largest floating-point value for the 19921result type. If only ``nnan`` is set then the neutral value is ``+Infinity``. 19922 19923This instruction has the same comparison semantics as the 19924:ref:`llvm.vector.reduce.fmin <int_vector_reduce_fmin>` intrinsic (and thus the 19925'``llvm.minnum.*``' intrinsic). That is, the result will always be a number 19926unless all elements of the vector and the starting value are ``NaN``. For a 19927vector with maximum element magnitude ``0.0`` and containing both ``+0.0`` and 19928``-0.0`` elements, the sign of the result is unspecified. 19929 19930To ignore the start value, the neutral value can be used. 19931 19932Examples: 19933""""""""" 19934 19935.. code-block:: llvm 19936 19937 %r = call float @llvm.vp.reduce.fmin.v4f32(float %start, <4 x float> %a, <4 x i1> %mask, i32 %evl) 19938 ; %r is equivalent to %also.r, where lanes greater than or equal to %evl 19939 ; are treated as though %mask were false for those lanes. 19940 19941 %masked.a = select <4 x i1> %mask, <4 x float> %a, <4 x float> <float QNAN, float QNAN, float QNAN, float QNAN> 19942 %reduction = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> %masked.a) 19943 %also.r = call float @llvm.minnum.f32(float %reduction, float %start) 19944 19945 19946.. _int_get_active_lane_mask: 19947 19948'``llvm.get.active.lane.mask.*``' Intrinsics 19949^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 19950 19951Syntax: 19952""""""" 19953This is an overloaded intrinsic. 19954 19955:: 19956 19957 declare <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 %base, i32 %n) 19958 declare <8 x i1> @llvm.get.active.lane.mask.v8i1.i64(i64 %base, i64 %n) 19959 declare <16 x i1> @llvm.get.active.lane.mask.v16i1.i64(i64 %base, i64 %n) 19960 declare <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 %base, i64 %n) 19961 19962 19963Overview: 19964""""""""" 19965 19966Create a mask representing active and inactive vector lanes. 19967 19968 19969Arguments: 19970"""""""""" 19971 19972Both operands have the same scalar integer type. The result is a vector with 19973the i1 element type. 19974 19975Semantics: 19976"""""""""" 19977 19978The '``llvm.get.active.lane.mask.*``' intrinsics are semantically equivalent 19979to: 19980 19981:: 19982 19983 %m[i] = icmp ult (%base + i), %n 19984 19985where ``%m`` is a vector (mask) of active/inactive lanes with its elements 19986indexed by ``i``, and ``%base``, ``%n`` are the two arguments to 19987``llvm.get.active.lane.mask.*``, ``%icmp`` is an integer compare and ``ult`` 19988the unsigned less-than comparison operator. Overflow cannot occur in 19989``(%base + i)`` and its comparison against ``%n`` as it is performed in integer 19990numbers and not in machine numbers. If ``%n`` is ``0``, then the result is a 19991poison value. The above is equivalent to: 19992 19993:: 19994 19995 %m = @llvm.get.active.lane.mask(%base, %n) 19996 19997This can, for example, be emitted by the loop vectorizer in which case 19998``%base`` is the first element of the vector induction variable (VIV) and 19999``%n`` is the loop tripcount. Thus, these intrinsics perform an element-wise 20000less than comparison of VIV with the loop tripcount, producing a mask of 20001true/false values representing active/inactive vector lanes, except if the VIV 20002overflows in which case they return false in the lanes where the VIV overflows. 20003The arguments are scalar types to accommodate scalable vector types, for which 20004it is unknown what the type of the step vector needs to be that enumerate its 20005lanes without overflow. 20006 20007This mask ``%m`` can e.g. be used in masked load/store instructions. These 20008intrinsics provide a hint to the backend. I.e., for a vector loop, the 20009back-edge taken count of the original scalar loop is explicit as the second 20010argument. 20011 20012 20013Examples: 20014""""""""" 20015 20016.. code-block:: llvm 20017 20018 %active.lane.mask = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i64(i64 %elem0, i64 429) 20019 %wide.masked.load = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* %3, i32 4, <4 x i1> %active.lane.mask, <4 x i32> undef) 20020 20021 20022.. _int_experimental_vp_splice: 20023 20024'``llvm.experimental.vp.splice``' Intrinsic 20025^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20026 20027Syntax: 20028""""""" 20029This is an overloaded intrinsic. 20030 20031:: 20032 20033 declare <2 x double> @llvm.experimental.vp.splice.v2f64(<2 x double> %vec1, <2 x double> %vec2, i32 %imm, <2 x i1> %mask, i32 %evl1, i32 %evl2) 20034 declare <vscale x 4 x i32> @llvm.experimental.vp.splice.nxv4i32(<vscale x 4 x i32> %vec1, <vscale x 4 x i32> %vec2, i32 %imm, <vscale x 4 x i1> %mask, i32 %evl1, i32 %evl2) 20035 20036Overview: 20037""""""""" 20038 20039The '``llvm.experimental.vp.splice.*``' intrinsic is the vector length 20040predicated version of the '``llvm.experimental.vector.splice.*``' intrinsic. 20041 20042Arguments: 20043"""""""""" 20044 20045The result and the first two arguments ``vec1`` and ``vec2`` are vectors with 20046the same type. The third argument ``imm`` is an immediate signed integer that 20047indicates the offset index. The fourth argument ``mask`` is a vector mask and 20048has the same number of elements as the result. The last two arguments ``evl1`` 20049and ``evl2`` are unsigned integers indicating the explicit vector lengths of 20050``vec1`` and ``vec2`` respectively. ``imm``, ``evl1`` and ``evl2`` should 20051respect the following constraints: ``-evl1 <= imm < evl1``, ``0 <= evl1 <= VL`` 20052and ``0 <= evl2 <= VL``, where ``VL`` is the runtime vector factor. If these 20053constraints are not satisfied the intrinsic has undefined behaviour. 20054 20055Semantics: 20056"""""""""" 20057 20058Effectively, this intrinsic concatenates ``vec1[0..evl1-1]`` and 20059``vec2[0..evl2-1]`` and creates the result vector by selecting the elements in a 20060window of size ``evl2``, starting at index ``imm`` (for a positive immediate) of 20061the concatenated vector. Elements in the result vector beyond ``evl2`` are 20062``undef``. If ``imm`` is negative the starting index is ``evl1 + imm``. The result 20063vector of active vector length ``evl2`` contains ``evl1 - imm`` (``-imm`` for 20064negative ``imm``) elements from indices ``[imm..evl1 - 1]`` 20065(``[evl1 + imm..evl1 -1]`` for negative ``imm``) of ``vec1`` followed by the 20066first ``evl2 - (evl1 - imm)`` (``evl2 + imm`` for negative ``imm``) elements of 20067``vec2``. If ``evl1 - imm`` (``-imm``) >= ``evl2``, only the first ``evl2`` 20068elements are considered and the remaining are ``undef``. The lanes in the result 20069vector disabled by ``mask`` are ``undef``. 20070 20071Examples: 20072""""""""" 20073 20074.. code-block:: text 20075 20076 llvm.experimental.vp.splice(<A,B,C,D>, <E,F,G,H>, 1, 2, 3) ==> <B, E, F, undef> ; index 20077 llvm.experimental.vp.splice(<A,B,C,D>, <E,F,G,H>, -2, 3, 2) ==> <B, C, undef, undef> ; trailing elements 20078 20079 20080.. _int_vp_load: 20081 20082'``llvm.vp.load``' Intrinsic 20083^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20084 20085Syntax: 20086""""""" 20087This is an overloaded intrinsic. 20088 20089:: 20090 20091 declare <4 x float> @llvm.vp.load.v4f32.p0v4f32(<4 x float>* %ptr, <4 x i1> %mask, i32 %evl) 20092 declare <vscale x 2 x i16> @llvm.vp.load.nxv2i16.p0nxv2i16(<vscale x 2 x i16>* %ptr, <vscale x 2 x i1> %mask, i32 %evl) 20093 declare <8 x float> @llvm.vp.load.v8f32.p1v8f32(<8 x float> addrspace(1)* %ptr, <8 x i1> %mask, i32 %evl) 20094 declare <vscale x 1 x i64> @llvm.vp.load.nxv1i64.p6nxv1i64(<vscale x 1 x i64> addrspace(6)* %ptr, <vscale x 1 x i1> %mask, i32 %evl) 20095 20096Overview: 20097""""""""" 20098 20099The '``llvm.vp.load.*``' intrinsic is the vector length predicated version of 20100the :ref:`llvm.masked.load <int_mload>` intrinsic. 20101 20102Arguments: 20103"""""""""" 20104 20105The first operand is the base pointer for the load. The second operand is a 20106vector of boolean values with the same number of elements as the return type. 20107The third is the explicit vector length of the operation. The return type and 20108underlying type of the base pointer are the same vector types. 20109 20110The :ref:`align <attr_align>` parameter attribute can be provided for the first 20111operand. 20112 20113Semantics: 20114"""""""""" 20115 20116The '``llvm.vp.load``' intrinsic reads a vector from memory in the same way as 20117the '``llvm.masked.load``' intrinsic, where the mask is taken from the 20118combination of the '``mask``' and '``evl``' operands in the usual VP way. 20119Certain '``llvm.masked.load``' operands do not have corresponding operands in 20120'``llvm.vp.load``': the '``passthru``' operand is implicitly ``undef``; the 20121'``alignment``' operand is taken as the ``align`` parameter attribute, if 20122provided. The default alignment is taken as the ABI alignment of the return 20123type as specified by the :ref:`datalayout string<langref_datalayout>`. 20124 20125Examples: 20126""""""""" 20127 20128.. code-block:: text 20129 20130 %r = call <8 x i8> @llvm.vp.load.v8i8.p0v8i8(<8 x i8>* align 2 %ptr, <8 x i1> %mask, i32 %evl) 20131 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20132 20133 %also.r = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* %ptr, i32 2, <8 x i1> %mask, <8 x i8> undef) 20134 20135 20136.. _int_vp_store: 20137 20138'``llvm.vp.store``' Intrinsic 20139^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20140 20141Syntax: 20142""""""" 20143This is an overloaded intrinsic. 20144 20145:: 20146 20147 declare void @llvm.vp.store.v4f32.p0v4f32(<4 x float> %val, <4 x float>* %ptr, <4 x i1> %mask, i32 %evl) 20148 declare void @llvm.vp.store.nxv2i16.p0nxv2i16(<vscale x 2 x i16> %val, <vscale x 2 x i16>* %ptr, <vscale x 2 x i1> %mask, i32 %evl) 20149 declare void @llvm.vp.store.v8f32.p1v8f32(<8 x float> %val, <8 x float> addrspace(1)* %ptr, <8 x i1> %mask, i32 %evl) 20150 declare void @llvm.vp.store.nxv1i64.p6nxv1i64(<vscale x 1 x i64> %val, <vscale x 1 x i64> addrspace(6)* %ptr, <vscale x 1 x i1> %mask, i32 %evl) 20151 20152Overview: 20153""""""""" 20154 20155The '``llvm.vp.store.*``' intrinsic is the vector length predicated version of 20156the :ref:`llvm.masked.store <int_mstore>` intrinsic. 20157 20158Arguments: 20159"""""""""" 20160 20161The first operand is the vector value to be written to memory. The second 20162operand is the base pointer for the store. It has the same underlying type as 20163the value operand. The third operand is a vector of boolean values with the 20164same number of elements as the return type. The fourth is the explicit vector 20165length of the operation. 20166 20167The :ref:`align <attr_align>` parameter attribute can be provided for the 20168second operand. 20169 20170Semantics: 20171"""""""""" 20172 20173The '``llvm.vp.store``' intrinsic reads a vector from memory in the same way as 20174the '``llvm.masked.store``' intrinsic, where the mask is taken from the 20175combination of the '``mask``' and '``evl``' operands in the usual VP way. The 20176alignment of the operation (corresponding to the '``alignment``' operand of 20177'``llvm.masked.store``') is specified by the ``align`` parameter attribute (see 20178above). If it is not provided then the ABI alignment of the type of the 20179'``value``' operand as specified by the :ref:`datalayout 20180string<langref_datalayout>` is used instead. 20181 20182Examples: 20183""""""""" 20184 20185.. code-block:: text 20186 20187 call void @llvm.vp.store.v8i8.p0v8i8(<8 x i8> %val, <8 x i8>* align 4 %ptr, <8 x i1> %mask, i32 %evl) 20188 ;; For all lanes below %evl, the call above is lane-wise equivalent to the call below. 20189 20190 call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> %val, <8 x i8>* %ptr, i32 4, <8 x i1> %mask) 20191 20192 20193.. _int_experimental_vp_strided_load: 20194 20195'``llvm.experimental.vp.strided.load``' Intrinsic 20196^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20197 20198Syntax: 20199""""""" 20200This is an overloaded intrinsic. 20201 20202:: 20203 20204 declare <4 x float> @llvm.experimental.vp.strided.load.v4f32.i64(float* %ptr, i64 %stride, <4 x i1> %mask, i32 %evl) 20205 declare <vscale x 2 x i16> @llvm.experimental.vp.strided.load.nxv2i16.i64(i16* %ptr, i64 %stride, <vscale x 2 x i1> %mask, i32 %evl) 20206 20207Overview: 20208""""""""" 20209 20210The '``llvm.experimental.vp.strided.load``' intrinsic loads, into a vector, scalar values from 20211memory locations evenly spaced apart by '``stride``' number of bytes, starting from '``ptr``'. 20212 20213Arguments: 20214"""""""""" 20215 20216The first operand is the base pointer for the load. The second operand is the stride 20217value expressed in bytes. The third operand is a vector of boolean values 20218with the same number of elements as the return type. The fourth is the explicit 20219vector length of the operation. The base pointer underlying type matches the type of the scalar 20220elements of the return operand. 20221 20222The :ref:`align <attr_align>` parameter attribute can be provided for the first 20223operand. 20224 20225Semantics: 20226"""""""""" 20227 20228The '``llvm.experimental.vp.strided.load``' intrinsic loads, into a vector, multiple scalar 20229values from memory in the same way as the :ref:`llvm.vp.gather <int_vp_gather>` intrinsic, 20230where the vector of pointers is in the form: 20231 20232 ``%ptrs = <%ptr, %ptr + %stride, %ptr + 2 * %stride, ... >``, 20233 20234with '``ptr``' previously casted to a pointer '``i8``', '``stride``' always interpreted as a signed 20235integer and all arithmetic occurring in the pointer type. 20236 20237Examples: 20238""""""""" 20239 20240.. code-block:: text 20241 20242 %r = call <8 x i64> @llvm.experimental.vp.strided.load.v8i64.i64(i64* %ptr, i64 %stride, <8 x i64> %mask, i32 %evl) 20243 ;; The operation can also be expressed like this: 20244 20245 %addr = bitcast i64* %ptr to i8* 20246 ;; Create a vector of pointers %addrs in the form: 20247 ;; %addrs = <%addr, %addr + %stride, %addr + 2 * %stride, ...> 20248 %ptrs = bitcast <8 x i8* > %addrs to <8 x i64* > 20249 %also.r = call <8 x i64> @llvm.vp.gather.v8i64.v8p0i64(<8 x i64* > %ptrs, <8 x i64> %mask, i32 %evl) 20250 20251 20252.. _int_experimental_vp_strided_store: 20253 20254'``llvm.experimental.vp.strided.store``' Intrinsic 20255^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20256 20257Syntax: 20258""""""" 20259This is an overloaded intrinsic. 20260 20261:: 20262 20263 declare void @llvm.experimental.vp.strided.store.v4f32.i64(<4 x float> %val, float* %ptr, i64 %stride, <4 x i1> %mask, i32 %evl) 20264 declare void @llvm.experimental.vp.strided.store.nxv2i16.i64(<vscale x 2 x i16> %val, i16* %ptr, i64 %stride, <vscale x 2 x i1> %mask, i32 %evl) 20265 20266Overview: 20267""""""""" 20268 20269The '``@llvm.experimental.vp.strided.store``' intrinsic stores the elements of 20270'``val``' into memory locations evenly spaced apart by '``stride``' number of 20271bytes, starting from '``ptr``'. 20272 20273Arguments: 20274"""""""""" 20275 20276The first operand is the vector value to be written to memory. The second 20277operand is the base pointer for the store. Its underlying type matches the 20278scalar element type of the value operand. The third operand is the stride value 20279expressed in bytes. The fourth operand is a vector of boolean values with the 20280same number of elements as the return type. The fifth is the explicit vector 20281length of the operation. 20282 20283The :ref:`align <attr_align>` parameter attribute can be provided for the 20284second operand. 20285 20286Semantics: 20287"""""""""" 20288 20289The '``llvm.experimental.vp.strided.store``' intrinsic stores the elements of 20290'``val``' in the same way as the :ref:`llvm.vp.scatter <int_vp_scatter>` intrinsic, 20291where the vector of pointers is in the form: 20292 20293 ``%ptrs = <%ptr, %ptr + %stride, %ptr + 2 * %stride, ... >``, 20294 20295with '``ptr``' previously casted to a pointer '``i8``', '``stride``' always interpreted as a signed 20296integer and all arithmetic occurring in the pointer type. 20297 20298Examples: 20299""""""""" 20300 20301.. code-block:: text 20302 20303 call void @llvm.experimental.vp.strided.store.v8i64.i64(<8 x i64> %val, i64* %ptr, i64 %stride, <8 x i1> %mask, i32 %evl) 20304 ;; The operation can also be expressed like this: 20305 20306 %addr = bitcast i64* %ptr to i8* 20307 ;; Create a vector of pointers %addrs in the form: 20308 ;; %addrs = <%addr, %addr + %stride, %addr + 2 * %stride, ...> 20309 %ptrs = bitcast <8 x i8* > %addrs to <8 x i64* > 20310 call void @llvm.vp.scatter.v8i64.v8p0i64(<8 x i64> %val, <8 x i64*> %ptrs, <8 x i1> %mask, i32 %evl) 20311 20312 20313.. _int_vp_gather: 20314 20315'``llvm.vp.gather``' Intrinsic 20316^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20317 20318Syntax: 20319""""""" 20320This is an overloaded intrinsic. 20321 20322:: 20323 20324 declare <4 x double> @llvm.vp.gather.v4f64.v4p0f64(<4 x double*> %ptrs, <4 x i1> %mask, i32 %evl) 20325 declare <vscale x 2 x i8> @llvm.vp.gather.nxv2i8.nxv2p0i8(<vscale x 2 x i8*> %ptrs, <vscale x 2 x i1> %mask, i32 %evl) 20326 declare <2 x float> @llvm.vp.gather.v2f32.v2p2f32(<2 x float addrspace(2)*> %ptrs, <2 x i1> %mask, i32 %evl) 20327 declare <vscale x 4 x i32> @llvm.vp.gather.nxv4i32.nxv4p4i32(<vscale x 4 x i32 addrspace(4)*> %ptrs, <vscale x 4 x i1> %mask, i32 %evl) 20328 20329Overview: 20330""""""""" 20331 20332The '``llvm.vp.gather.*``' intrinsic is the vector length predicated version of 20333the :ref:`llvm.masked.gather <int_mgather>` intrinsic. 20334 20335Arguments: 20336"""""""""" 20337 20338The first operand is a vector of pointers which holds all memory addresses to 20339read. The second operand is a vector of boolean values with the same number of 20340elements as the return type. The third is the explicit vector length of the 20341operation. The return type and underlying type of the vector of pointers are 20342the same vector types. 20343 20344The :ref:`align <attr_align>` parameter attribute can be provided for the first 20345operand. 20346 20347Semantics: 20348"""""""""" 20349 20350The '``llvm.vp.gather``' intrinsic reads multiple scalar values from memory in 20351the same way as the '``llvm.masked.gather``' intrinsic, where the mask is taken 20352from the combination of the '``mask``' and '``evl``' operands in the usual VP 20353way. Certain '``llvm.masked.gather``' operands do not have corresponding 20354operands in '``llvm.vp.gather``': the '``passthru``' operand is implicitly 20355``undef``; the '``alignment``' operand is taken as the ``align`` parameter, if 20356provided. The default alignment is taken as the ABI alignment of the source 20357addresses as specified by the :ref:`datalayout string<langref_datalayout>`. 20358 20359Examples: 20360""""""""" 20361 20362.. code-block:: text 20363 20364 %r = call <8 x i8> @llvm.vp.gather.v8i8.v8p0i8(<8 x i8*> align 8 %ptrs, <8 x i1> %mask, i32 %evl) 20365 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20366 20367 %also.r = call <8 x i8> @llvm.masked.gather.v8i8.v8p0i8(<8 x i8*> %ptrs, i32 8, <8 x i1> %mask, <8 x i8> undef) 20368 20369 20370.. _int_vp_scatter: 20371 20372'``llvm.vp.scatter``' Intrinsic 20373^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20374 20375Syntax: 20376""""""" 20377This is an overloaded intrinsic. 20378 20379:: 20380 20381 declare void @llvm.vp.scatter.v4f64.v4p0f64(<4 x double> %val, <4 x double*> %ptrs, <4 x i1> %mask, i32 %evl) 20382 declare void @llvm.vp.scatter.nxv2i8.nxv2p0i8(<vscale x 2 x i8> %val, <vscale x 2 x i8*> %ptrs, <vscale x 2 x i1> %mask, i32 %evl) 20383 declare void @llvm.vp.scatter.v2f32.v2p2f32(<2 x float> %val, <2 x float addrspace(2)*> %ptrs, <2 x i1> %mask, i32 %evl) 20384 declare void @llvm.vp.scatter.nxv4i32.nxv4p4i32(<vscale x 4 x i32> %val, <vscale x 4 x i32 addrspace(4)*> %ptrs, <vscale x 4 x i1> %mask, i32 %evl) 20385 20386Overview: 20387""""""""" 20388 20389The '``llvm.vp.scatter.*``' intrinsic is the vector length predicated version of 20390the :ref:`llvm.masked.scatter <int_mscatter>` intrinsic. 20391 20392Arguments: 20393"""""""""" 20394 20395The first operand is a vector value to be written to memory. The second operand 20396is a vector of pointers, pointing to where the value elements should be stored. 20397The third operand is a vector of boolean values with the same number of 20398elements as the return type. The fourth is the explicit vector length of the 20399operation. 20400 20401The :ref:`align <attr_align>` parameter attribute can be provided for the 20402second operand. 20403 20404Semantics: 20405"""""""""" 20406 20407The '``llvm.vp.scatter``' intrinsic writes multiple scalar values to memory in 20408the same way as the '``llvm.masked.scatter``' intrinsic, where the mask is 20409taken from the combination of the '``mask``' and '``evl``' operands in the 20410usual VP way. The '``alignment``' operand of the '``llvm.masked.scatter``' does 20411not have a corresponding operand in '``llvm.vp.scatter``': it is instead 20412provided via the optional ``align`` parameter attribute on the 20413vector-of-pointers operand. Otherwise it is taken as the ABI alignment of the 20414destination addresses as specified by the :ref:`datalayout 20415string<langref_datalayout>`. 20416 20417Examples: 20418""""""""" 20419 20420.. code-block:: text 20421 20422 call void @llvm.vp.scatter.v8i8.v8p0i8(<8 x i8> %val, <8 x i8*> align 1 %ptrs, <8 x i1> %mask, i32 %evl) 20423 ;; For all lanes below %evl, the call above is lane-wise equivalent to the call below. 20424 20425 call void @llvm.masked.scatter.v8i8.v8p0i8(<8 x i8> %val, <8 x i8*> %ptrs, i32 1, <8 x i1> %mask) 20426 20427 20428.. _int_vp_trunc: 20429 20430'``llvm.vp.trunc.*``' Intrinsics 20431^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20432 20433Syntax: 20434""""""" 20435This is an overloaded intrinsic. 20436 20437:: 20438 20439 declare <16 x i16> @llvm.vp.trunc.v16i16.v16i32 (<16 x i32> <op>, <16 x i1> <mask>, i32 <vector_length>) 20440 declare <vscale x 4 x i16> @llvm.vp.trunc.nxv4i16.nxv4i32 (<vscale x 4 x i32> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20441 20442Overview: 20443""""""""" 20444 20445The '``llvm.vp.trunc``' intrinsic truncates its first operand to the return 20446type. The operation has a mask and an explicit vector length parameter. 20447 20448 20449Arguments: 20450"""""""""" 20451 20452The '``llvm.vp.trunc``' intrinsic takes a value to cast as its first operand. 20453The return type is the type to cast the value to. Both types must be vector of 20454:ref:`integer <t_integer>` type. The bit size of the value must be larger than 20455the bit size of the return type. The second operand is the vector mask. The 20456return type, the value to cast, and the vector mask have the same number of 20457elements. The third operand is the explicit vector length of the operation. 20458 20459Semantics: 20460"""""""""" 20461 20462The '``llvm.vp.trunc``' intrinsic truncates the high order bits in value and 20463converts the remaining bits to return type. Since the source size must be larger 20464than the destination size, '``llvm.vp.trunc``' cannot be a *no-op cast*. It will 20465always truncate bits. The conversion is performed on lane positions below the 20466explicit vector length and where the vector mask is true. Masked-off lanes are 20467undefined. 20468 20469Examples: 20470""""""""" 20471 20472.. code-block:: llvm 20473 20474 %r = call <4 x i16> @llvm.vp.trunc.v4i16.v4i32(<4 x i32> %a, <4 x i1> %mask, i32 %evl) 20475 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20476 20477 %t = trunc <4 x i32> %a to <4 x i16> 20478 %also.r = select <4 x i1> %mask, <4 x i16> %t, <4 x i16> undef 20479 20480 20481.. _int_vp_zext: 20482 20483'``llvm.vp.zext.*``' Intrinsics 20484^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20485 20486Syntax: 20487""""""" 20488This is an overloaded intrinsic. 20489 20490:: 20491 20492 declare <16 x i32> @llvm.vp.zext.v16i32.v16i16 (<16 x i16> <op>, <16 x i1> <mask>, i32 <vector_length>) 20493 declare <vscale x 4 x i32> @llvm.vp.zext.nxv4i32.nxv4i16 (<vscale x 4 x i16> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20494 20495Overview: 20496""""""""" 20497 20498The '``llvm.vp.zext``' intrinsic zero extends its first operand to the return 20499type. The operation has a mask and an explicit vector length parameter. 20500 20501 20502Arguments: 20503"""""""""" 20504 20505The '``llvm.vp.zext``' intrinsic takes a value to cast as its first operand. 20506The return type is the type to cast the value to. Both types must be vectors of 20507:ref:`integer <t_integer>` type. The bit size of the value must be smaller than 20508the bit size of the return type. The second operand is the vector mask. The 20509return type, the value to cast, and the vector mask have the same number of 20510elements. The third operand is the explicit vector length of the operation. 20511 20512Semantics: 20513"""""""""" 20514 20515The '``llvm.vp.zext``' intrinsic fill the high order bits of the value with zero 20516bits until it reaches the size of the return type. When zero extending from i1, 20517the result will always be either 0 or 1. The conversion is performed on lane 20518positions below the explicit vector length and where the vector mask is true. 20519Masked-off lanes are undefined. 20520 20521Examples: 20522""""""""" 20523 20524.. code-block:: llvm 20525 20526 %r = call <4 x i32> @llvm.vp.zext.v4i32.v4i16(<4 x i16> %a, <4 x i1> %mask, i32 %evl) 20527 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20528 20529 %t = zext <4 x i16> %a to <4 x i32> 20530 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 20531 20532 20533.. _int_vp_sext: 20534 20535'``llvm.vp.sext.*``' Intrinsics 20536^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20537 20538Syntax: 20539""""""" 20540This is an overloaded intrinsic. 20541 20542:: 20543 20544 declare <16 x i32> @llvm.vp.sext.v16i32.v16i16 (<16 x i16> <op>, <16 x i1> <mask>, i32 <vector_length>) 20545 declare <vscale x 4 x i32> @llvm.vp.sext.nxv4i32.nxv4i16 (<vscale x 4 x i16> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20546 20547Overview: 20548""""""""" 20549 20550The '``llvm.vp.sext``' intrinsic sign extends its first operand to the return 20551type. The operation has a mask and an explicit vector length parameter. 20552 20553 20554Arguments: 20555"""""""""" 20556 20557The '``llvm.vp.sext``' intrinsic takes a value to cast as its first operand. 20558The return type is the type to cast the value to. Both types must be vectors of 20559:ref:`integer <t_integer>` type. The bit size of the value must be smaller than 20560the bit size of the return type. The second operand is the vector mask. The 20561return type, the value to cast, and the vector mask have the same number of 20562elements. The third operand is the explicit vector length of the operation. 20563 20564Semantics: 20565"""""""""" 20566 20567The '``llvm.vp.sext``' intrinsic performs a sign extension by copying the sign 20568bit (highest order bit) of the value until it reaches the size of the return 20569type. When zero extending from i1, the result will always be either -1 or 0. 20570The conversion is performed on lane positions below the explicit vector length 20571and where the vector mask is true. Masked-off lanes are undefined. 20572 20573Examples: 20574""""""""" 20575 20576.. code-block:: llvm 20577 20578 %r = call <4 x i32> @llvm.vp.sext.v4i32.v4i16(<4 x i16> %a, <4 x i1> %mask, i32 %evl) 20579 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20580 20581 %t = sext <4 x i16> %a to <4 x i32> 20582 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 20583 20584 20585.. _int_vp_fptrunc: 20586 20587'``llvm.vp.fptrunc.*``' Intrinsics 20588^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20589 20590Syntax: 20591""""""" 20592This is an overloaded intrinsic. 20593 20594:: 20595 20596 declare <16 x float> @llvm.vp.fptrunc.v16f32.v16f64 (<16 x double> <op>, <16 x i1> <mask>, i32 <vector_length>) 20597 declare <vscale x 4 x float> @llvm.vp.trunc.nxv4f32.nxv4f64 (<vscale x 4 x double> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20598 20599Overview: 20600""""""""" 20601 20602The '``llvm.vp.fptrunc``' intrinsic truncates its first operand to the return 20603type. The operation has a mask and an explicit vector length parameter. 20604 20605 20606Arguments: 20607"""""""""" 20608 20609The '``llvm.vp.fptrunc``' intrinsic takes a value to cast as its first operand. 20610The return type is the type to cast the value to. Both types must be vector of 20611:ref:`floating-point <t_floating>` type. The bit size of the value must be 20612larger than the bit size of the return type. This implies that 20613'``llvm.vp.fptrunc``' cannot be used to make a *no-op cast*. The second operand 20614is the vector mask. The return type, the value to cast, and the vector mask have 20615the same number of elements. The third operand is the explicit vector length of 20616the operation. 20617 20618Semantics: 20619"""""""""" 20620 20621The '``llvm.vp.fptrunc``' intrinsic casts a ``value`` from a larger 20622:ref:`floating-point <t_floating>` type to a smaller :ref:`floating-point 20623<t_floating>` type. 20624This instruction is assumed to execute in the default :ref:`floating-point 20625environment <floatenv>`. The conversion is performed on lane positions below the 20626explicit vector length and where the vector mask is true. Masked-off lanes are 20627undefined. 20628 20629Examples: 20630""""""""" 20631 20632.. code-block:: llvm 20633 20634 %r = call <4 x float> @llvm.vp.fptrunc.v4f32.v4f64(<4 x double> %a, <4 x i1> %mask, i32 %evl) 20635 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20636 20637 %t = fptrunc <4 x double> %a to <4 x float> 20638 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 20639 20640 20641.. _int_vp_fpext: 20642 20643'``llvm.vp.fpext.*``' Intrinsics 20644^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20645 20646Syntax: 20647""""""" 20648This is an overloaded intrinsic. 20649 20650:: 20651 20652 declare <16 x double> @llvm.vp.fpext.v16f64.v16f32 (<16 x float> <op>, <16 x i1> <mask>, i32 <vector_length>) 20653 declare <vscale x 4 x double> @llvm.vp.fpext.nxv4f64.nxv4f32 (<vscale x 4 x float> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20654 20655Overview: 20656""""""""" 20657 20658The '``llvm.vp.fpext``' intrinsic extends its first operand to the return 20659type. The operation has a mask and an explicit vector length parameter. 20660 20661 20662Arguments: 20663"""""""""" 20664 20665The '``llvm.vp.fpext``' intrinsic takes a value to cast as its first operand. 20666The return type is the type to cast the value to. Both types must be vector of 20667:ref:`floating-point <t_floating>` type. The bit size of the value must be 20668smaller than the bit size of the return type. This implies that 20669'``llvm.vp.fpext``' cannot be used to make a *no-op cast*. The second operand 20670is the vector mask. The return type, the value to cast, and the vector mask have 20671the same number of elements. The third operand is the explicit vector length of 20672the operation. 20673 20674Semantics: 20675"""""""""" 20676 20677The '``llvm.vp.fpext``' intrinsic extends the ``value`` from a smaller 20678:ref:`floating-point <t_floating>` type to a larger :ref:`floating-point 20679<t_floating>` type. The '``llvm.vp.fpext``' cannot be used to make a 20680*no-op cast* because it always changes bits. Use ``bitcast`` to make a 20681*no-op cast* for a floating-point cast. 20682The conversion is performed on lane positions below the explicit vector length 20683and where the vector mask is true. Masked-off lanes are undefined. 20684 20685Examples: 20686""""""""" 20687 20688.. code-block:: llvm 20689 20690 %r = call <4 x double> @llvm.vp.fpext.v4f64.v4f32(<4 x float> %a, <4 x i1> %mask, i32 %evl) 20691 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20692 20693 %t = fpext <4 x float> %a to <4 x double> 20694 %also.r = select <4 x i1> %mask, <4 x double> %t, <4 x double> undef 20695 20696 20697.. _int_vp_fptoui: 20698 20699'``llvm.vp.fptoui.*``' Intrinsics 20700^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20701 20702Syntax: 20703""""""" 20704This is an overloaded intrinsic. 20705 20706:: 20707 20708 declare <16 x i32> @llvm.vp.fptoui.v16i32.v16f32 (<16 x float> <op>, <16 x i1> <mask>, i32 <vector_length>) 20709 declare <vscale x 4 x i32> @llvm.vp.fptoui.nxv4i32.nxv4f32 (<vscale x 4 x float> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20710 declare <256 x i64> @llvm.vp.fptoui.v256i64.v256f64 (<256 x double> <op>, <256 x i1> <mask>, i32 <vector_length>) 20711 20712Overview: 20713""""""""" 20714 20715The '``llvm.vp.fptoui``' intrinsic converts the :ref:`floating-point 20716<t_floating>` operand to the unsigned integer return type. 20717The operation has a mask and an explicit vector length parameter. 20718 20719 20720Arguments: 20721"""""""""" 20722 20723The '``llvm.vp.fptoui``' intrinsic takes a value to cast as its first operand. 20724The value to cast must be a vector of :ref:`floating-point <t_floating>` type. 20725The return type is the type to cast the value to. The return type must be 20726vector of :ref:`integer <t_integer>` type. The second operand is the vector 20727mask. The return type, the value to cast, and the vector mask have the same 20728number of elements. The third operand is the explicit vector length of the 20729operation. 20730 20731Semantics: 20732"""""""""" 20733 20734The '``llvm.vp.fptoui``' intrinsic converts its :ref:`floating-point 20735<t_floating>` operand into the nearest (rounding towards zero) unsigned integer 20736value where the lane position is below the explicit vector length and the 20737vector mask is true. Masked-off lanes are undefined. On enabled lanes where 20738conversion takes place and the value cannot fit in the return type, the result 20739on that lane is a :ref:`poison value <poisonvalues>`. 20740 20741Examples: 20742""""""""" 20743 20744.. code-block:: llvm 20745 20746 %r = call <4 x i32> @llvm.vp.fptoui.v4i32.v4f32(<4 x float> %a, <4 x i1> %mask, i32 %evl) 20747 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20748 20749 %t = fptoui <4 x float> %a to <4 x i32> 20750 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 20751 20752 20753.. _int_vp_fptosi: 20754 20755'``llvm.vp.fptosi.*``' Intrinsics 20756^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20757 20758Syntax: 20759""""""" 20760This is an overloaded intrinsic. 20761 20762:: 20763 20764 declare <16 x i32> @llvm.vp.fptosi.v16i32.v16f32 (<16 x float> <op>, <16 x i1> <mask>, i32 <vector_length>) 20765 declare <vscale x 4 x i32> @llvm.vp.fptosi.nxv4i32.nxv4f32 (<vscale x 4 x float> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20766 declare <256 x i64> @llvm.vp.fptosi.v256i64.v256f64 (<256 x double> <op>, <256 x i1> <mask>, i32 <vector_length>) 20767 20768Overview: 20769""""""""" 20770 20771The '``llvm.vp.fptosi``' intrinsic converts the :ref:`floating-point 20772<t_floating>` operand to the signed integer return type. 20773The operation has a mask and an explicit vector length parameter. 20774 20775 20776Arguments: 20777"""""""""" 20778 20779The '``llvm.vp.fptosi``' intrinsic takes a value to cast as its first operand. 20780The value to cast must be a vector of :ref:`floating-point <t_floating>` type. 20781The return type is the type to cast the value to. The return type must be 20782vector of :ref:`integer <t_integer>` type. The second operand is the vector 20783mask. The return type, the value to cast, and the vector mask have the same 20784number of elements. The third operand is the explicit vector length of the 20785operation. 20786 20787Semantics: 20788"""""""""" 20789 20790The '``llvm.vp.fptosi``' intrinsic converts its :ref:`floating-point 20791<t_floating>` operand into the nearest (rounding towards zero) signed integer 20792value where the lane position is below the explicit vector length and the 20793vector mask is true. Masked-off lanes are undefined. On enabled lanes where 20794conversion takes place and the value cannot fit in the return type, the result 20795on that lane is a :ref:`poison value <poisonvalues>`. 20796 20797Examples: 20798""""""""" 20799 20800.. code-block:: llvm 20801 20802 %r = call <4 x i32> @llvm.vp.fptosi.v4i32.v4f32(<4 x float> %a, <4 x i1> %mask, i32 %evl) 20803 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20804 20805 %t = fptosi <4 x float> %a to <4 x i32> 20806 %also.r = select <4 x i1> %mask, <4 x i32> %t, <4 x i32> undef 20807 20808 20809.. _int_vp_uitofp: 20810 20811'``llvm.vp.uitofp.*``' Intrinsics 20812^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20813 20814Syntax: 20815""""""" 20816This is an overloaded intrinsic. 20817 20818:: 20819 20820 declare <16 x float> @llvm.vp.uitofp.v16f32.v16i32 (<16 x i32> <op>, <16 x i1> <mask>, i32 <vector_length>) 20821 declare <vscale x 4 x float> @llvm.vp.uitofp.nxv4f32.nxv4i32 (<vscale x 4 x i32> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20822 declare <256 x double> @llvm.vp.uitofp.v256f64.v256i64 (<256 x i64> <op>, <256 x i1> <mask>, i32 <vector_length>) 20823 20824Overview: 20825""""""""" 20826 20827The '``llvm.vp.uitofp``' intrinsic converts its unsigned integer operand to the 20828:ref:`floating-point <t_floating>` return type. The operation has a mask and 20829an explicit vector length parameter. 20830 20831 20832Arguments: 20833"""""""""" 20834 20835The '``llvm.vp.uitofp``' intrinsic takes a value to cast as its first operand. 20836The value to cast must be vector of :ref:`integer <t_integer>` type. The 20837return type is the type to cast the value to. The return type must be a vector 20838of :ref:`floating-point <t_floating>` type. The second operand is the vector 20839mask. The return type, the value to cast, and the vector mask have the same 20840number of elements. The third operand is the explicit vector length of the 20841operation. 20842 20843Semantics: 20844"""""""""" 20845 20846The '``llvm.vp.uitofp``' intrinsic interprets its first operand as an unsigned 20847integer quantity and converts it to the corresponding floating-point value. If 20848the value cannot be exactly represented, it is rounded using the default 20849rounding mode. The conversion is performed on lane positions below the 20850explicit vector length and where the vector mask is true. Masked-off lanes are 20851undefined. 20852 20853Examples: 20854""""""""" 20855 20856.. code-block:: llvm 20857 20858 %r = call <4 x float> @llvm.vp.uitofp.v4f32.v4i32(<4 x i32> %a, <4 x i1> %mask, i32 %evl) 20859 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20860 20861 %t = uitofp <4 x i32> %a to <4 x float> 20862 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 20863 20864 20865.. _int_vp_sitofp: 20866 20867'``llvm.vp.sitofp.*``' Intrinsics 20868^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20869 20870Syntax: 20871""""""" 20872This is an overloaded intrinsic. 20873 20874:: 20875 20876 declare <16 x float> @llvm.vp.sitofp.v16f32.v16i32 (<16 x i32> <op>, <16 x i1> <mask>, i32 <vector_length>) 20877 declare <vscale x 4 x float> @llvm.vp.sitofp.nxv4f32.nxv4i32 (<vscale x 4 x i32> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20878 declare <256 x double> @llvm.vp.sitofp.v256f64.v256i64 (<256 x i64> <op>, <256 x i1> <mask>, i32 <vector_length>) 20879 20880Overview: 20881""""""""" 20882 20883The '``llvm.vp.sitofp``' intrinsic converts its signed integer operand to the 20884:ref:`floating-point <t_floating>` return type. The operation has a mask and 20885an explicit vector length parameter. 20886 20887 20888Arguments: 20889"""""""""" 20890 20891The '``llvm.vp.sitofp``' intrinsic takes a value to cast as its first operand. 20892The value to cast must be vector of :ref:`integer <t_integer>` type. The 20893return type is the type to cast the value to. The return type must be a vector 20894of :ref:`floating-point <t_floating>` type. The second operand is the vector 20895mask. The return type, the value to cast, and the vector mask have the same 20896number of elements. The third operand is the explicit vector length of the 20897operation. 20898 20899Semantics: 20900"""""""""" 20901 20902The '``llvm.vp.sitofp``' intrinsic interprets its first operand as a signed 20903integer quantity and converts it to the corresponding floating-point value. If 20904the value cannot be exactly represented, it is rounded using the default 20905rounding mode. The conversion is performed on lane positions below the 20906explicit vector length and where the vector mask is true. Masked-off lanes are 20907undefined. 20908 20909Examples: 20910""""""""" 20911 20912.. code-block:: llvm 20913 20914 %r = call <4 x float> @llvm.vp.sitofp.v4f32.v4i32(<4 x i32> %a, <4 x i1> %mask, i32 %evl) 20915 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20916 20917 %t = sitofp <4 x i32> %a to <4 x float> 20918 %also.r = select <4 x i1> %mask, <4 x float> %t, <4 x float> undef 20919 20920 20921.. _int_vp_ptrtoint: 20922 20923'``llvm.vp.ptrtoint.*``' Intrinsics 20924^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20925 20926Syntax: 20927""""""" 20928This is an overloaded intrinsic. 20929 20930:: 20931 20932 declare <16 x i8> @llvm.vp.ptrtoint.v16i8.v16p0i32 (<16 x i32*> <op>, <16 x i1> <mask>, i32 <vector_length>) 20933 declare <vscale x 4 x i8> @llvm.vp.ptrtoint.nxv4i8.nxv4p0i32 (<vscale x 4 x i32*> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20934 declare <256 x i64> @llvm.vp.ptrtoint.v16i64.v16p0i32 (<256 x i32*> <op>, <256 x i1> <mask>, i32 <vector_length>) 20935 20936Overview: 20937""""""""" 20938 20939The '``llvm.vp.ptrtoint``' intrinsic converts its pointer to the integer return 20940type. The operation has a mask and an explicit vector length parameter. 20941 20942 20943Arguments: 20944"""""""""" 20945 20946The '``llvm.vp.ptrtoint``' intrinsic takes a value to cast as its first operand 20947, which must be a vector of pointers, and a type to cast it to return type, 20948which must be a vector of :ref:`integer <t_integer>` type. 20949The second operand is the vector mask. The return type, the value to cast, and 20950the vector mask have the same number of elements. 20951The third operand is the explicit vector length of the operation. 20952 20953Semantics: 20954"""""""""" 20955 20956The '``llvm.vp.ptrtoint``' intrinsic converts value to return type by 20957interpreting the pointer value as an integer and either truncating or zero 20958extending that value to the size of the integer type. 20959If ``value`` is smaller than return type, then a zero extension is done. If 20960``value`` is larger than return type, then a truncation is done. If they are 20961the same size, then nothing is done (*no-op cast*) other than a type 20962change. 20963The conversion is performed on lane positions below the explicit vector length 20964and where the vector mask is true. Masked-off lanes are undefined. 20965 20966Examples: 20967""""""""" 20968 20969.. code-block:: llvm 20970 20971 %r = call <4 x i8> @llvm.vp.ptrtoint.v4i8.v4p0i32(<4 x i32*> %a, <4 x i1> %mask, i32 %evl) 20972 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 20973 20974 %t = ptrtoint <4 x i32*> %a to <4 x i8> 20975 %also.r = select <4 x i1> %mask, <4 x i8> %t, <4 x i8> undef 20976 20977 20978.. _int_vp_inttoptr: 20979 20980'``llvm.vp.inttoptr.*``' Intrinsics 20981^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 20982 20983Syntax: 20984""""""" 20985This is an overloaded intrinsic. 20986 20987:: 20988 20989 declare <16 x i32*> @llvm.vp.inttoptr.v16p0i32.v16i32 (<16 x i32> <op>, <16 x i1> <mask>, i32 <vector_length>) 20990 declare <vscale x 4 x i32*> @llvm.vp.inttoptr.nxv4p0i32.nxv4i32 (<vscale x 4 x i32> <op>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 20991 declare <256 x i32*> @llvm.vp.inttoptr.v256p0i32.v256i32 (<256 x i32> <op>, <256 x i1> <mask>, i32 <vector_length>) 20992 20993Overview: 20994""""""""" 20995 20996The '``llvm.vp.inttoptr``' intrinsic converts its integer value to the point 20997return type. The operation has a mask and an explicit vector length parameter. 20998 20999 21000Arguments: 21001"""""""""" 21002 21003The '``llvm.vp.inttoptr``' intrinsic takes a value to cast as its first operand 21004, which must be a vector of :ref:`integer <t_integer>` type, and a type to cast 21005it to return type, which must be a vector of pointers type. 21006The second operand is the vector mask. The return type, the value to cast, and 21007the vector mask have the same number of elements. 21008The third operand is the explicit vector length of the operation. 21009 21010Semantics: 21011"""""""""" 21012 21013The '``llvm.vp.inttoptr``' intrinsic converts ``value`` to return type by 21014applying either a zero extension or a truncation depending on the size of the 21015integer ``value``. If ``value`` is larger than the size of a pointer, then a 21016truncation is done. If ``value`` is smaller than the size of a pointer, then a 21017zero extension is done. If they are the same size, nothing is done (*no-op cast*). 21018The conversion is performed on lane positions below the explicit vector length 21019and where the vector mask is true. Masked-off lanes are undefined. 21020 21021Examples: 21022""""""""" 21023 21024.. code-block:: llvm 21025 21026 %r = call <4 x i32*> @llvm.vp.inttoptr.v4p0i32.v4i32(<4 x i32> %a, <4 x i1> %mask, i32 %evl) 21027 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 21028 21029 %t = inttoptr <4 x i32> %a to <4 x i32*> 21030 %also.r = select <4 x i1> %mask, <4 x i32*> %t, <4 x i32*> undef 21031 21032 21033.. _int_vp_fcmp: 21034 21035'``llvm.vp.fcmp.*``' Intrinsics 21036^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21037 21038Syntax: 21039""""""" 21040This is an overloaded intrinsic. 21041 21042:: 21043 21044 declare <16 x i1> @llvm.vp.fcmp.v16f32(<16 x float> <left_op>, <16 x float> <right_op>, metadata <condition code>, <16 x i1> <mask>, i32 <vector_length>) 21045 declare <vscale x 4 x i1> @llvm.vp.fcmp.nxv4f32(<vscale x 4 x float> <left_op>, <vscale x 4 x float> <right_op>, metadata <condition code>, <vscale x 4 x i1> <mask>, i32 <vector_length>) 21046 declare <256 x i1> @llvm.vp.fcmp.v256f64(<256 x double> <left_op>, <256 x double> <right_op>, metadata <condition code>, <256 x i1> <mask>, i32 <vector_length>) 21047 21048Overview: 21049""""""""" 21050 21051The '``llvm.vp.fcmp``' intrinsic returns a vector of boolean values based on 21052the comparison of its operands. The operation has a mask and an explicit vector 21053length parameter. 21054 21055 21056Arguments: 21057"""""""""" 21058 21059The '``llvm.vp.fcmp``' intrinsic takes the two values to compare as its first 21060and second operands. These two values must be vectors of :ref:`floating-point 21061<t_floating>` types. 21062The return type is the result of the comparison. The return type must be a 21063vector of :ref:`i1 <t_integer>` type. The fourth operand is the vector mask. 21064The return type, the values to compare, and the vector mask have the same 21065number of elements. The third operand is the condition code indicating the kind 21066of comparison to perform. It must be a metadata string with :ref:`one of the 21067supported floating-point condition code values <fcmp_md_cc>`. The fifth operand 21068is the explicit vector length of the operation. 21069 21070Semantics: 21071"""""""""" 21072 21073The '``llvm.vp.fcmp``' compares its first two operands according to the 21074condition code given as the third operand. The operands are compared element by 21075element on each enabled lane, where the the semantics of the comparison are 21076defined :ref:`according to the condition code <fcmp_md_cc_sem>`. Masked-off 21077lanes are undefined. 21078 21079Examples: 21080""""""""" 21081 21082.. code-block:: llvm 21083 21084 %r = call <4 x i1> @llvm.vp.fcmp.v4f32(<4 x float> %a, <4 x float> %b, metadata !"oeq", <4 x i1> %mask, i32 %evl) 21085 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 21086 21087 %t = fcmp oeq <4 x float> %a, %b 21088 %also.r = select <4 x i1> %mask, <4 x i1> %t, <4 x i1> undef 21089 21090 21091.. _int_vp_icmp: 21092 21093'``llvm.vp.icmp.*``' Intrinsics 21094^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21095 21096Syntax: 21097""""""" 21098This is an overloaded intrinsic. 21099 21100:: 21101 21102 declare <32 x i1> @llvm.vp.icmp.v32i32(<32 x i32> <left_op>, <32 x i32> <right_op>, metadata <condition code>, <32 x i1> <mask>, i32 <vector_length>) 21103 declare <vscale x 2 x i1> @llvm.vp.icmp.nxv2i32(<vscale x 2 x i32> <left_op>, <vscale x 2 x i32> <right_op>, metadata <condition code>, <vscale x 2 x i1> <mask>, i32 <vector_length>) 21104 declare <128 x i1> @llvm.vp.icmp.v128i8(<128 x i8> <left_op>, <128 x i8> <right_op>, metadata <condition code>, <128 x i1> <mask>, i32 <vector_length>) 21105 21106Overview: 21107""""""""" 21108 21109The '``llvm.vp.icmp``' intrinsic returns a vector of boolean values based on 21110the comparison of its operands. The operation has a mask and an explicit vector 21111length parameter. 21112 21113 21114Arguments: 21115"""""""""" 21116 21117The '``llvm.vp.icmp``' intrinsic takes the two values to compare as its first 21118and second operands. These two values must be vectors of :ref:`integer 21119<t_integer>` types. 21120The return type is the result of the comparison. The return type must be a 21121vector of :ref:`i1 <t_integer>` type. The fourth operand is the vector mask. 21122The return type, the values to compare, and the vector mask have the same 21123number of elements. The third operand is the condition code indicating the kind 21124of comparison to perform. It must be a metadata string with :ref:`one of the 21125supported integer condition code values <icmp_md_cc>`. The fifth operand is the 21126explicit vector length of the operation. 21127 21128Semantics: 21129"""""""""" 21130 21131The '``llvm.vp.icmp``' compares its first two operands according to the 21132condition code given as the third operand. The operands are compared element by 21133element on each enabled lane, where the the semantics of the comparison are 21134defined :ref:`according to the condition code <icmp_md_cc_sem>`. Masked-off 21135lanes are undefined. 21136 21137Examples: 21138""""""""" 21139 21140.. code-block:: llvm 21141 21142 %r = call <4 x i1> @llvm.vp.icmp.v4i32(<4 x i32> %a, <4 x i32> %b, metadata !"ne", <4 x i1> %mask, i32 %evl) 21143 ;; For all lanes below %evl, %r is lane-wise equivalent to %also.r 21144 21145 %t = icmp ne <4 x i32> %a, %b 21146 %also.r = select <4 x i1> %mask, <4 x i1> %t, <4 x i1> undef 21147 21148 21149.. _int_mload_mstore: 21150 21151Masked Vector Load and Store Intrinsics 21152--------------------------------------- 21153 21154LLVM 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. 21155 21156.. _int_mload: 21157 21158'``llvm.masked.load.*``' Intrinsics 21159^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21160 21161Syntax: 21162""""""" 21163This is an overloaded intrinsic. The loaded data is a vector of any integer, floating-point or pointer data type. 21164 21165:: 21166 21167 declare <16 x float> @llvm.masked.load.v16f32.p0v16f32 (<16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>) 21168 declare <2 x double> @llvm.masked.load.v2f64.p0v2f64 (<2 x double>* <ptr>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>) 21169 ;; The data is a vector of pointers to double 21170 declare <8 x double*> @llvm.masked.load.v8p0f64.p0v8p0f64 (<8 x double*>* <ptr>, i32 <alignment>, <8 x i1> <mask>, <8 x double*> <passthru>) 21171 ;; The data is a vector of function pointers 21172 declare <8 x i32 ()*> @llvm.masked.load.v8p0f_i32f.p0v8p0f_i32f (<8 x i32 ()*>* <ptr>, i32 <alignment>, <8 x i1> <mask>, <8 x i32 ()*> <passthru>) 21173 21174Overview: 21175""""""""" 21176 21177Reads a vector from memory according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes. The masked-off lanes in the result vector are taken from the corresponding lanes of the '``passthru``' operand. 21178 21179 21180Arguments: 21181"""""""""" 21182 21183The first operand is the base pointer for the load. The second operand is the alignment of the source location. It must be a power of two constant integer value. The third operand, mask, is a vector of boolean values with the same number of elements as the return type. The fourth is a pass-through value that is used to fill the masked-off lanes of the result. The return type, underlying type of the base pointer and the type of the '``passthru``' operand are the same vector types. 21184 21185Semantics: 21186"""""""""" 21187 21188The '``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. 21189The 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. 21190 21191 21192:: 21193 21194 %res = call <16 x float> @llvm.masked.load.v16f32.p0v16f32 (<16 x float>* %ptr, i32 4, <16 x i1>%mask, <16 x float> %passthru) 21195 21196 ;; The result of the two following instructions is identical aside from potential memory access exception 21197 %loadlal = load <16 x float>, <16 x float>* %ptr, align 4 21198 %res = select <16 x i1> %mask, <16 x float> %loadlal, <16 x float> %passthru 21199 21200.. _int_mstore: 21201 21202'``llvm.masked.store.*``' Intrinsics 21203^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21204 21205Syntax: 21206""""""" 21207This is an overloaded intrinsic. The data stored in memory is a vector of any integer, floating-point or pointer data type. 21208 21209:: 21210 21211 declare void @llvm.masked.store.v8i32.p0v8i32 (<8 x i32> <value>, <8 x i32>* <ptr>, i32 <alignment>, <8 x i1> <mask>) 21212 declare void @llvm.masked.store.v16f32.p0v16f32 (<16 x float> <value>, <16 x float>* <ptr>, i32 <alignment>, <16 x i1> <mask>) 21213 ;; The data is a vector of pointers to double 21214 declare void @llvm.masked.store.v8p0f64.p0v8p0f64 (<8 x double*> <value>, <8 x double*>* <ptr>, i32 <alignment>, <8 x i1> <mask>) 21215 ;; The data is a vector of function pointers 21216 declare void @llvm.masked.store.v4p0f_i32f.p0v4p0f_i32f (<4 x i32 ()*> <value>, <4 x i32 ()*>* <ptr>, i32 <alignment>, <4 x i1> <mask>) 21217 21218Overview: 21219""""""""" 21220 21221Writes 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. 21222 21223Arguments: 21224"""""""""" 21225 21226The 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. It must be a power of two constant integer value. 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. 21227 21228 21229Semantics: 21230"""""""""" 21231 21232The '``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. 21233The 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. 21234 21235:: 21236 21237 call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> %value, <16 x float>* %ptr, i32 4, <16 x i1> %mask) 21238 21239 ;; The result of the following instructions is identical aside from potential data races and memory access exceptions 21240 %oldval = load <16 x float>, <16 x float>* %ptr, align 4 21241 %res = select <16 x i1> %mask, <16 x float> %value, <16 x float> %oldval 21242 store <16 x float> %res, <16 x float>* %ptr, align 4 21243 21244 21245Masked Vector Gather and Scatter Intrinsics 21246------------------------------------------- 21247 21248LLVM provides intrinsics for vector gather and scatter operations. They are similar to :ref:`Masked Vector Load and Store <int_mload_mstore>`, except they are designed for arbitrary memory accesses, rather than sequential memory accesses. Gather and scatter also employ a mask operand, which holds one bit per vector element, switching the associated vector lane on or off. The memory addresses corresponding to the "off" lanes are not accessed. When all bits are off, no memory is accessed. 21249 21250.. _int_mgather: 21251 21252'``llvm.masked.gather.*``' Intrinsics 21253^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21254 21255Syntax: 21256""""""" 21257This is an overloaded intrinsic. The loaded data are multiple scalar values of any integer, floating-point or pointer data type gathered together into one vector. 21258 21259:: 21260 21261 declare <16 x float> @llvm.masked.gather.v16f32.v16p0f32 (<16 x float*> <ptrs>, i32 <alignment>, <16 x i1> <mask>, <16 x float> <passthru>) 21262 declare <2 x double> @llvm.masked.gather.v2f64.v2p1f64 (<2 x double addrspace(1)*> <ptrs>, i32 <alignment>, <2 x i1> <mask>, <2 x double> <passthru>) 21263 declare <8 x float*> @llvm.masked.gather.v8p0f32.v8p0p0f32 (<8 x float**> <ptrs>, i32 <alignment>, <8 x i1> <mask>, <8 x float*> <passthru>) 21264 21265Overview: 21266""""""""" 21267 21268Reads scalar values from arbitrary memory locations and gathers them into one vector. The memory locations are provided in the vector of pointers '``ptrs``'. The memory is accessed according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes. The masked-off lanes in the result vector are taken from the corresponding lanes of the '``passthru``' operand. 21269 21270 21271Arguments: 21272"""""""""" 21273 21274The first operand is a vector of pointers which holds all memory addresses to read. The second operand is an alignment of the source addresses. It must be 0 or a power of two constant integer value. The third operand, mask, is a vector of boolean values with the same number of elements as the return type. The fourth is a pass-through value that is used to fill the masked-off lanes of the result. The return type, underlying type of the vector of pointers and the type of the '``passthru``' operand are the same vector types. 21275 21276Semantics: 21277"""""""""" 21278 21279The '``llvm.masked.gather``' intrinsic is designed for conditional reading of multiple scalar values from arbitrary memory locations in a single IR operation. It is useful for targets that support vector masked gathers and allows vectorizing basic blocks with data and control divergence. Other targets may support this intrinsic differently, for example by lowering it into a sequence of scalar load operations. 21280The semantics of this operation are equivalent to a sequence of conditional scalar loads with subsequent gathering all loaded values into a single vector. The mask restricts memory access to certain lanes and facilitates vectorization of predicated basic blocks. 21281 21282 21283:: 21284 21285 %res = call <4 x double> @llvm.masked.gather.v4f64.v4p0f64 (<4 x double*> %ptrs, i32 8, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x double> undef) 21286 21287 ;; The gather with all-true mask is equivalent to the following instruction sequence 21288 %ptr0 = extractelement <4 x double*> %ptrs, i32 0 21289 %ptr1 = extractelement <4 x double*> %ptrs, i32 1 21290 %ptr2 = extractelement <4 x double*> %ptrs, i32 2 21291 %ptr3 = extractelement <4 x double*> %ptrs, i32 3 21292 21293 %val0 = load double, double* %ptr0, align 8 21294 %val1 = load double, double* %ptr1, align 8 21295 %val2 = load double, double* %ptr2, align 8 21296 %val3 = load double, double* %ptr3, align 8 21297 21298 %vec0 = insertelement <4 x double>undef, %val0, 0 21299 %vec01 = insertelement <4 x double>%vec0, %val1, 1 21300 %vec012 = insertelement <4 x double>%vec01, %val2, 2 21301 %vec0123 = insertelement <4 x double>%vec012, %val3, 3 21302 21303.. _int_mscatter: 21304 21305'``llvm.masked.scatter.*``' Intrinsics 21306^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21307 21308Syntax: 21309""""""" 21310This is an overloaded intrinsic. The data stored in memory is a vector of any integer, floating-point or pointer data type. Each vector element is stored in an arbitrary memory address. Scatter with overlapping addresses is guaranteed to be ordered from least-significant to most-significant element. 21311 21312:: 21313 21314 declare void @llvm.masked.scatter.v8i32.v8p0i32 (<8 x i32> <value>, <8 x i32*> <ptrs>, i32 <alignment>, <8 x i1> <mask>) 21315 declare void @llvm.masked.scatter.v16f32.v16p1f32 (<16 x float> <value>, <16 x float addrspace(1)*> <ptrs>, i32 <alignment>, <16 x i1> <mask>) 21316 declare void @llvm.masked.scatter.v4p0f64.v4p0p0f64 (<4 x double*> <value>, <4 x double**> <ptrs>, i32 <alignment>, <4 x i1> <mask>) 21317 21318Overview: 21319""""""""" 21320 21321Writes each element from the value vector to the corresponding memory address. The memory addresses are represented as a vector of pointers. Writing is done according to the provided mask. The mask holds a bit for each vector lane, and is used to prevent memory accesses to the masked-off lanes. 21322 21323Arguments: 21324"""""""""" 21325 21326The first operand is a vector value to be written to memory. The second operand is a vector of pointers, pointing to where the value elements should be stored. It has the same underlying type as the value operand. The third operand is an alignment of the destination addresses. It must be 0 or a power of two constant integer value. 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. 21327 21328Semantics: 21329"""""""""" 21330 21331The '``llvm.masked.scatter``' intrinsics is designed for writing selected vector elements to arbitrary memory addresses in a single IR operation. The operation may be conditional, when not all bits in the mask are switched on. It is useful for targets that support vector masked scatter and allows vectorizing basic blocks with data and control divergence. Other targets may support this intrinsic differently, for example by lowering it into a sequence of branches that guard scalar store operations. 21332 21333:: 21334 21335 ;; This instruction unconditionally stores data vector in multiple addresses 21336 call @llvm.masked.scatter.v8i32.v8p0i32 (<8 x i32> %value, <8 x i32*> %ptrs, i32 4, <8 x i1> <true, true, .. true>) 21337 21338 ;; It is equivalent to a list of scalar stores 21339 %val0 = extractelement <8 x i32> %value, i32 0 21340 %val1 = extractelement <8 x i32> %value, i32 1 21341 .. 21342 %val7 = extractelement <8 x i32> %value, i32 7 21343 %ptr0 = extractelement <8 x i32*> %ptrs, i32 0 21344 %ptr1 = extractelement <8 x i32*> %ptrs, i32 1 21345 .. 21346 %ptr7 = extractelement <8 x i32*> %ptrs, i32 7 21347 ;; Note: the order of the following stores is important when they overlap: 21348 store i32 %val0, i32* %ptr0, align 4 21349 store i32 %val1, i32* %ptr1, align 4 21350 .. 21351 store i32 %val7, i32* %ptr7, align 4 21352 21353 21354Masked Vector Expanding Load and Compressing Store Intrinsics 21355------------------------------------------------------------- 21356 21357LLVM provides intrinsics for expanding load and compressing store operations. Data selected from a vector according to a mask is stored in consecutive memory addresses (compressed store), and vice-versa (expanding load). These operations effective map to "if (cond.i) a[j++] = v.i" and "if (cond.i) v.i = a[j++]" patterns, respectively. Note that when the mask starts with '1' bits followed by '0' bits, these operations are identical to :ref:`llvm.masked.store <int_mstore>` and :ref:`llvm.masked.load <int_mload>`. 21358 21359.. _int_expandload: 21360 21361'``llvm.masked.expandload.*``' Intrinsics 21362^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21363 21364Syntax: 21365""""""" 21366This is an overloaded intrinsic. Several values of integer, floating point or pointer data type are loaded from consecutive memory addresses and stored into the elements of a vector according to the mask. 21367 21368:: 21369 21370 declare <16 x float> @llvm.masked.expandload.v16f32 (float* <ptr>, <16 x i1> <mask>, <16 x float> <passthru>) 21371 declare <2 x i64> @llvm.masked.expandload.v2i64 (i64* <ptr>, <2 x i1> <mask>, <2 x i64> <passthru>) 21372 21373Overview: 21374""""""""" 21375 21376Reads a number of scalar values sequentially from memory location provided in '``ptr``' and spreads them in a vector. The '``mask``' holds a bit for each vector lane. The number of elements read from memory is equal to the number of '1' bits in the mask. The loaded elements are positioned in the destination vector according to the sequence of '1' and '0' bits in the mask. E.g., if the mask vector is '10010001', "expandload" reads 3 values from memory addresses ptr, ptr+1, ptr+2 and places them in lanes 0, 3 and 7 accordingly. The masked-off lanes are filled by elements from the corresponding lanes of the '``passthru``' operand. 21377 21378 21379Arguments: 21380"""""""""" 21381 21382The first operand is the base pointer for the load. It has the same underlying type as the element of the returned vector. The second operand, mask, is a vector of boolean values with the same number of elements as the return type. The third is a pass-through value that is used to fill the masked-off lanes of the result. The return type and the type of the '``passthru``' operand have the same vector type. 21383 21384Semantics: 21385"""""""""" 21386 21387The '``llvm.masked.expandload``' intrinsic is designed for reading multiple scalar values from adjacent memory addresses into possibly non-adjacent vector lanes. It is useful for targets that support vector expanding loads and allows vectorizing loop with cross-iteration dependency like in the following example: 21388 21389.. code-block:: c 21390 21391 // In this loop we load from B and spread the elements into array A. 21392 double *A, B; int *C; 21393 for (int i = 0; i < size; ++i) { 21394 if (C[i] != 0) 21395 A[i] = B[j++]; 21396 } 21397 21398 21399.. code-block:: llvm 21400 21401 ; Load several elements from array B and expand them in a vector. 21402 ; The number of loaded elements is equal to the number of '1' elements in the Mask. 21403 %Tmp = call <8 x double> @llvm.masked.expandload.v8f64(double* %Bptr, <8 x i1> %Mask, <8 x double> undef) 21404 ; Store the result in A 21405 call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> %Tmp, <8 x double>* %Aptr, i32 8, <8 x i1> %Mask) 21406 21407 ; %Bptr should be increased on each iteration according to the number of '1' elements in the Mask. 21408 %MaskI = bitcast <8 x i1> %Mask to i8 21409 %MaskIPopcnt = call i8 @llvm.ctpop.i8(i8 %MaskI) 21410 %MaskI64 = zext i8 %MaskIPopcnt to i64 21411 %BNextInd = add i64 %BInd, %MaskI64 21412 21413 21414Other targets may support this intrinsic differently, for example, by lowering it into a sequence of conditional scalar load operations and shuffles. 21415If all mask elements are '1', the intrinsic behavior is equivalent to the regular unmasked vector load. 21416 21417.. _int_compressstore: 21418 21419'``llvm.masked.compressstore.*``' Intrinsics 21420^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21421 21422Syntax: 21423""""""" 21424This is an overloaded intrinsic. A number of scalar values of integer, floating point or pointer data type are collected from an input vector and stored into adjacent memory addresses. A mask defines which elements to collect from the vector. 21425 21426:: 21427 21428 declare void @llvm.masked.compressstore.v8i32 (<8 x i32> <value>, i32* <ptr>, <8 x i1> <mask>) 21429 declare void @llvm.masked.compressstore.v16f32 (<16 x float> <value>, float* <ptr>, <16 x i1> <mask>) 21430 21431Overview: 21432""""""""" 21433 21434Selects elements from input vector '``value``' according to the '``mask``'. All selected elements are written into adjacent memory addresses starting at address '`ptr`', from lower to higher. The mask holds a bit for each vector lane, and is used to select elements to be stored. The number of elements to be stored is equal to the number of active bits in the mask. 21435 21436Arguments: 21437"""""""""" 21438 21439The first operand is the input vector, from which elements are collected and written to memory. The second operand is the base pointer for the store, it has the same underlying type as the element of the input vector operand. The third operand is the mask, a vector of boolean values. The mask and the input vector must have the same number of vector elements. 21440 21441 21442Semantics: 21443"""""""""" 21444 21445The '``llvm.masked.compressstore``' intrinsic is designed for compressing data in memory. It allows to collect elements from possibly non-adjacent lanes of a vector and store them contiguously in memory in one IR operation. It is useful for targets that support compressing store operations and allows vectorizing loops with cross-iteration dependences like in the following example: 21446 21447.. code-block:: c 21448 21449 // In this loop we load elements from A and store them consecutively in B 21450 double *A, B; int *C; 21451 for (int i = 0; i < size; ++i) { 21452 if (C[i] != 0) 21453 B[j++] = A[i] 21454 } 21455 21456 21457.. code-block:: llvm 21458 21459 ; Load elements from A. 21460 %Tmp = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* %Aptr, i32 8, <8 x i1> %Mask, <8 x double> undef) 21461 ; Store all selected elements consecutively in array B 21462 call <void> @llvm.masked.compressstore.v8f64(<8 x double> %Tmp, double* %Bptr, <8 x i1> %Mask) 21463 21464 ; %Bptr should be increased on each iteration according to the number of '1' elements in the Mask. 21465 %MaskI = bitcast <8 x i1> %Mask to i8 21466 %MaskIPopcnt = call i8 @llvm.ctpop.i8(i8 %MaskI) 21467 %MaskI64 = zext i8 %MaskIPopcnt to i64 21468 %BNextInd = add i64 %BInd, %MaskI64 21469 21470 21471Other targets may support this intrinsic differently, for example, by lowering it into a sequence of branches that guard scalar store operations. 21472 21473 21474Memory Use Markers 21475------------------ 21476 21477This class of intrinsics provides information about the 21478:ref:`lifetime of memory objects <objectlifetime>` and ranges where variables 21479are immutable. 21480 21481.. _int_lifestart: 21482 21483'``llvm.lifetime.start``' Intrinsic 21484^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21485 21486Syntax: 21487""""""" 21488 21489:: 21490 21491 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>) 21492 21493Overview: 21494""""""""" 21495 21496The '``llvm.lifetime.start``' intrinsic specifies the start of a memory 21497object's lifetime. 21498 21499Arguments: 21500"""""""""" 21501 21502The first argument is a constant integer representing the size of the 21503object, or -1 if it is variable sized. The second argument is a pointer 21504to the object. 21505 21506Semantics: 21507"""""""""" 21508 21509If ``ptr`` is a stack-allocated object and it points to the first byte of 21510the object, the object is initially marked as dead. 21511``ptr`` is conservatively considered as a non-stack-allocated object if 21512the stack coloring algorithm that is used in the optimization pipeline cannot 21513conclude that ``ptr`` is a stack-allocated object. 21514 21515After '``llvm.lifetime.start``', the stack object that ``ptr`` points is marked 21516as alive and has an uninitialized value. 21517The stack object is marked as dead when either 21518:ref:`llvm.lifetime.end <int_lifeend>` to the alloca is executed or the 21519function returns. 21520 21521After :ref:`llvm.lifetime.end <int_lifeend>` is called, 21522'``llvm.lifetime.start``' on the stack object can be called again. 21523The second '``llvm.lifetime.start``' call marks the object as alive, but it 21524does not change the address of the object. 21525 21526If ``ptr`` is a non-stack-allocated object, it does not point to the first 21527byte of the object or it is a stack object that is already alive, it simply 21528fills all bytes of the object with ``poison``. 21529 21530 21531.. _int_lifeend: 21532 21533'``llvm.lifetime.end``' Intrinsic 21534^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21535 21536Syntax: 21537""""""" 21538 21539:: 21540 21541 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>) 21542 21543Overview: 21544""""""""" 21545 21546The '``llvm.lifetime.end``' intrinsic specifies the end of a memory object's 21547lifetime. 21548 21549Arguments: 21550"""""""""" 21551 21552The first argument is a constant integer representing the size of the 21553object, or -1 if it is variable sized. The second argument is a pointer 21554to the object. 21555 21556Semantics: 21557"""""""""" 21558 21559If ``ptr`` is a stack-allocated object and it points to the first byte of the 21560object, the object is dead. 21561``ptr`` is conservatively considered as a non-stack-allocated object if 21562the stack coloring algorithm that is used in the optimization pipeline cannot 21563conclude that ``ptr`` is a stack-allocated object. 21564 21565Calling ``llvm.lifetime.end`` on an already dead alloca is no-op. 21566 21567If ``ptr`` is a non-stack-allocated object or it does not point to the first 21568byte of the object, it is equivalent to simply filling all bytes of the object 21569with ``poison``. 21570 21571 21572'``llvm.invariant.start``' Intrinsic 21573^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21574 21575Syntax: 21576""""""" 21577This is an overloaded intrinsic. The memory object can belong to any address space. 21578 21579:: 21580 21581 declare {}* @llvm.invariant.start.p0i8(i64 <size>, i8* nocapture <ptr>) 21582 21583Overview: 21584""""""""" 21585 21586The '``llvm.invariant.start``' intrinsic specifies that the contents of 21587a memory object will not change. 21588 21589Arguments: 21590"""""""""" 21591 21592The first argument is a constant integer representing the size of the 21593object, or -1 if it is variable sized. The second argument is a pointer 21594to the object. 21595 21596Semantics: 21597"""""""""" 21598 21599This intrinsic indicates that until an ``llvm.invariant.end`` that uses 21600the return value, the referenced memory location is constant and 21601unchanging. 21602 21603'``llvm.invariant.end``' Intrinsic 21604^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21605 21606Syntax: 21607""""""" 21608This is an overloaded intrinsic. The memory object can belong to any address space. 21609 21610:: 21611 21612 declare void @llvm.invariant.end.p0i8({}* <start>, i64 <size>, i8* nocapture <ptr>) 21613 21614Overview: 21615""""""""" 21616 21617The '``llvm.invariant.end``' intrinsic specifies that the contents of a 21618memory object are mutable. 21619 21620Arguments: 21621"""""""""" 21622 21623The first argument is the matching ``llvm.invariant.start`` intrinsic. 21624The second argument is a constant integer representing the size of the 21625object, or -1 if it is variable sized and the third argument is a 21626pointer to the object. 21627 21628Semantics: 21629"""""""""" 21630 21631This intrinsic indicates that the memory is mutable again. 21632 21633'``llvm.launder.invariant.group``' Intrinsic 21634^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21635 21636Syntax: 21637""""""" 21638This is an overloaded intrinsic. The memory object can belong to any address 21639space. The returned pointer must belong to the same address space as the 21640argument. 21641 21642:: 21643 21644 declare i8* @llvm.launder.invariant.group.p0i8(i8* <ptr>) 21645 21646Overview: 21647""""""""" 21648 21649The '``llvm.launder.invariant.group``' intrinsic can be used when an invariant 21650established by ``invariant.group`` metadata no longer holds, to obtain a new 21651pointer value that carries fresh invariant group information. It is an 21652experimental intrinsic, which means that its semantics might change in the 21653future. 21654 21655 21656Arguments: 21657"""""""""" 21658 21659The ``llvm.launder.invariant.group`` takes only one argument, which is a pointer 21660to the memory. 21661 21662Semantics: 21663"""""""""" 21664 21665Returns another pointer that aliases its argument but which is considered different 21666for the purposes of ``load``/``store`` ``invariant.group`` metadata. 21667It does not read any accessible memory and the execution can be speculated. 21668 21669'``llvm.strip.invariant.group``' Intrinsic 21670^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21671 21672Syntax: 21673""""""" 21674This is an overloaded intrinsic. The memory object can belong to any address 21675space. The returned pointer must belong to the same address space as the 21676argument. 21677 21678:: 21679 21680 declare i8* @llvm.strip.invariant.group.p0i8(i8* <ptr>) 21681 21682Overview: 21683""""""""" 21684 21685The '``llvm.strip.invariant.group``' intrinsic can be used when an invariant 21686established by ``invariant.group`` metadata no longer holds, to obtain a new pointer 21687value that does not carry the invariant information. It is an experimental 21688intrinsic, which means that its semantics might change in the future. 21689 21690 21691Arguments: 21692"""""""""" 21693 21694The ``llvm.strip.invariant.group`` takes only one argument, which is a pointer 21695to the memory. 21696 21697Semantics: 21698"""""""""" 21699 21700Returns another pointer that aliases its argument but which has no associated 21701``invariant.group`` metadata. 21702It does not read any memory and can be speculated. 21703 21704 21705 21706.. _constrainedfp: 21707 21708Constrained Floating-Point Intrinsics 21709------------------------------------- 21710 21711These intrinsics are used to provide special handling of floating-point 21712operations when specific rounding mode or floating-point exception behavior is 21713required. By default, LLVM optimization passes assume that the rounding mode is 21714round-to-nearest and that floating-point exceptions will not be monitored. 21715Constrained FP intrinsics are used to support non-default rounding modes and 21716accurately preserve exception behavior without compromising LLVM's ability to 21717optimize FP code when the default behavior is used. 21718 21719If any FP operation in a function is constrained then they all must be 21720constrained. This is required for correct LLVM IR. Optimizations that 21721move code around can create miscompiles if mixing of constrained and normal 21722operations is done. The correct way to mix constrained and less constrained 21723operations is to use the rounding mode and exception handling metadata to 21724mark constrained intrinsics as having LLVM's default behavior. 21725 21726Each of these intrinsics corresponds to a normal floating-point operation. The 21727data arguments and the return value are the same as the corresponding FP 21728operation. 21729 21730The rounding mode argument is a metadata string specifying what 21731assumptions, if any, the optimizer can make when transforming constant 21732values. Some constrained FP intrinsics omit this argument. If required 21733by the intrinsic, this argument must be one of the following strings: 21734 21735:: 21736 21737 "round.dynamic" 21738 "round.tonearest" 21739 "round.downward" 21740 "round.upward" 21741 "round.towardzero" 21742 "round.tonearestaway" 21743 21744If this argument is "round.dynamic" optimization passes must assume that the 21745rounding mode is unknown and may change at runtime. No transformations that 21746depend on rounding mode may be performed in this case. 21747 21748The other possible values for the rounding mode argument correspond to the 21749similarly named IEEE rounding modes. If the argument is any of these values 21750optimization passes may perform transformations as long as they are consistent 21751with the specified rounding mode. 21752 21753For example, 'x-0'->'x' is not a valid transformation if the rounding mode is 21754"round.downward" or "round.dynamic" because if the value of 'x' is +0 then 21755'x-0' should evaluate to '-0' when rounding downward. However, this 21756transformation is legal for all other rounding modes. 21757 21758For values other than "round.dynamic" optimization passes may assume that the 21759actual runtime rounding mode (as defined in a target-specific manner) matches 21760the specified rounding mode, but this is not guaranteed. Using a specific 21761non-dynamic rounding mode which does not match the actual rounding mode at 21762runtime results in undefined behavior. 21763 21764The exception behavior argument is a metadata string describing the floating 21765point exception semantics that required for the intrinsic. This argument 21766must be one of the following strings: 21767 21768:: 21769 21770 "fpexcept.ignore" 21771 "fpexcept.maytrap" 21772 "fpexcept.strict" 21773 21774If this argument is "fpexcept.ignore" optimization passes may assume that the 21775exception status flags will not be read and that floating-point exceptions will 21776be masked. This allows transformations to be performed that may change the 21777exception semantics of the original code. For example, FP operations may be 21778speculatively executed in this case whereas they must not be for either of the 21779other possible values of this argument. 21780 21781If the exception behavior argument is "fpexcept.maytrap" optimization passes 21782must avoid transformations that may raise exceptions that would not have been 21783raised by the original code (such as speculatively executing FP operations), but 21784passes are not required to preserve all exceptions that are implied by the 21785original code. For example, exceptions may be potentially hidden by constant 21786folding. 21787 21788If the exception behavior argument is "fpexcept.strict" all transformations must 21789strictly preserve the floating-point exception semantics of the original code. 21790Any FP exception that would have been raised by the original code must be raised 21791by the transformed code, and the transformed code must not raise any FP 21792exceptions that would not have been raised by the original code. This is the 21793exception behavior argument that will be used if the code being compiled reads 21794the FP exception status flags, but this mode can also be used with code that 21795unmasks FP exceptions. 21796 21797The number and order of floating-point exceptions is NOT guaranteed. For 21798example, a series of FP operations that each may raise exceptions may be 21799vectorized into a single instruction that raises each unique exception a single 21800time. 21801 21802Proper :ref:`function attributes <fnattrs>` usage is required for the 21803constrained intrinsics to function correctly. 21804 21805All function *calls* done in a function that uses constrained floating 21806point intrinsics must have the ``strictfp`` attribute. 21807 21808All function *definitions* that use constrained floating point intrinsics 21809must have the ``strictfp`` attribute. 21810 21811'``llvm.experimental.constrained.fadd``' Intrinsic 21812^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21813 21814Syntax: 21815""""""" 21816 21817:: 21818 21819 declare <type> 21820 @llvm.experimental.constrained.fadd(<type> <op1>, <type> <op2>, 21821 metadata <rounding mode>, 21822 metadata <exception behavior>) 21823 21824Overview: 21825""""""""" 21826 21827The '``llvm.experimental.constrained.fadd``' intrinsic returns the sum of its 21828two operands. 21829 21830 21831Arguments: 21832"""""""""" 21833 21834The first two arguments to the '``llvm.experimental.constrained.fadd``' 21835intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 21836of floating-point values. Both arguments must have identical types. 21837 21838The third and fourth arguments specify the rounding mode and exception 21839behavior as described above. 21840 21841Semantics: 21842"""""""""" 21843 21844The value produced is the floating-point sum of the two value operands and has 21845the same type as the operands. 21846 21847 21848'``llvm.experimental.constrained.fsub``' Intrinsic 21849^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21850 21851Syntax: 21852""""""" 21853 21854:: 21855 21856 declare <type> 21857 @llvm.experimental.constrained.fsub(<type> <op1>, <type> <op2>, 21858 metadata <rounding mode>, 21859 metadata <exception behavior>) 21860 21861Overview: 21862""""""""" 21863 21864The '``llvm.experimental.constrained.fsub``' intrinsic returns the difference 21865of its two operands. 21866 21867 21868Arguments: 21869"""""""""" 21870 21871The first two arguments to the '``llvm.experimental.constrained.fsub``' 21872intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 21873of floating-point values. Both arguments must have identical types. 21874 21875The third and fourth arguments specify the rounding mode and exception 21876behavior as described above. 21877 21878Semantics: 21879"""""""""" 21880 21881The value produced is the floating-point difference of the two value operands 21882and has the same type as the operands. 21883 21884 21885'``llvm.experimental.constrained.fmul``' Intrinsic 21886^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21887 21888Syntax: 21889""""""" 21890 21891:: 21892 21893 declare <type> 21894 @llvm.experimental.constrained.fmul(<type> <op1>, <type> <op2>, 21895 metadata <rounding mode>, 21896 metadata <exception behavior>) 21897 21898Overview: 21899""""""""" 21900 21901The '``llvm.experimental.constrained.fmul``' intrinsic returns the product of 21902its two operands. 21903 21904 21905Arguments: 21906"""""""""" 21907 21908The first two arguments to the '``llvm.experimental.constrained.fmul``' 21909intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 21910of floating-point values. Both arguments must have identical types. 21911 21912The third and fourth arguments specify the rounding mode and exception 21913behavior as described above. 21914 21915Semantics: 21916"""""""""" 21917 21918The value produced is the floating-point product of the two value operands and 21919has the same type as the operands. 21920 21921 21922'``llvm.experimental.constrained.fdiv``' Intrinsic 21923^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21924 21925Syntax: 21926""""""" 21927 21928:: 21929 21930 declare <type> 21931 @llvm.experimental.constrained.fdiv(<type> <op1>, <type> <op2>, 21932 metadata <rounding mode>, 21933 metadata <exception behavior>) 21934 21935Overview: 21936""""""""" 21937 21938The '``llvm.experimental.constrained.fdiv``' intrinsic returns the quotient of 21939its two operands. 21940 21941 21942Arguments: 21943"""""""""" 21944 21945The first two arguments to the '``llvm.experimental.constrained.fdiv``' 21946intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 21947of floating-point values. Both arguments must have identical types. 21948 21949The third and fourth arguments specify the rounding mode and exception 21950behavior as described above. 21951 21952Semantics: 21953"""""""""" 21954 21955The value produced is the floating-point quotient of the two value operands and 21956has the same type as the operands. 21957 21958 21959'``llvm.experimental.constrained.frem``' Intrinsic 21960^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 21961 21962Syntax: 21963""""""" 21964 21965:: 21966 21967 declare <type> 21968 @llvm.experimental.constrained.frem(<type> <op1>, <type> <op2>, 21969 metadata <rounding mode>, 21970 metadata <exception behavior>) 21971 21972Overview: 21973""""""""" 21974 21975The '``llvm.experimental.constrained.frem``' intrinsic returns the remainder 21976from the division of its two operands. 21977 21978 21979Arguments: 21980"""""""""" 21981 21982The first two arguments to the '``llvm.experimental.constrained.frem``' 21983intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 21984of floating-point values. Both arguments must have identical types. 21985 21986The third and fourth arguments specify the rounding mode and exception 21987behavior as described above. The rounding mode argument has no effect, since 21988the result of frem is never rounded, but the argument is included for 21989consistency with the other constrained floating-point intrinsics. 21990 21991Semantics: 21992"""""""""" 21993 21994The value produced is the floating-point remainder from the division of the two 21995value operands and has the same type as the operands. The remainder has the 21996same sign as the dividend. 21997 21998'``llvm.experimental.constrained.fma``' Intrinsic 21999^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22000 22001Syntax: 22002""""""" 22003 22004:: 22005 22006 declare <type> 22007 @llvm.experimental.constrained.fma(<type> <op1>, <type> <op2>, <type> <op3>, 22008 metadata <rounding mode>, 22009 metadata <exception behavior>) 22010 22011Overview: 22012""""""""" 22013 22014The '``llvm.experimental.constrained.fma``' intrinsic returns the result of a 22015fused-multiply-add operation on its operands. 22016 22017Arguments: 22018"""""""""" 22019 22020The first three arguments to the '``llvm.experimental.constrained.fma``' 22021intrinsic must be :ref:`floating-point <t_floating>` or :ref:`vector 22022<t_vector>` of floating-point values. All arguments must have identical types. 22023 22024The fourth and fifth arguments specify the rounding mode and exception behavior 22025as described above. 22026 22027Semantics: 22028"""""""""" 22029 22030The result produced is the product of the first two operands added to the third 22031operand computed with infinite precision, and then rounded to the target 22032precision. 22033 22034'``llvm.experimental.constrained.fptoui``' Intrinsic 22035^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22036 22037Syntax: 22038""""""" 22039 22040:: 22041 22042 declare <ty2> 22043 @llvm.experimental.constrained.fptoui(<type> <value>, 22044 metadata <exception behavior>) 22045 22046Overview: 22047""""""""" 22048 22049The '``llvm.experimental.constrained.fptoui``' intrinsic converts a 22050floating-point ``value`` to its unsigned integer equivalent of type ``ty2``. 22051 22052Arguments: 22053"""""""""" 22054 22055The first argument to the '``llvm.experimental.constrained.fptoui``' 22056intrinsic must be :ref:`floating point <t_floating>` or :ref:`vector 22057<t_vector>` of floating point values. 22058 22059The second argument specifies the exception behavior as described above. 22060 22061Semantics: 22062"""""""""" 22063 22064The result produced is an unsigned integer converted from the floating 22065point operand. The value is truncated, so it is rounded towards zero. 22066 22067'``llvm.experimental.constrained.fptosi``' Intrinsic 22068^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22069 22070Syntax: 22071""""""" 22072 22073:: 22074 22075 declare <ty2> 22076 @llvm.experimental.constrained.fptosi(<type> <value>, 22077 metadata <exception behavior>) 22078 22079Overview: 22080""""""""" 22081 22082The '``llvm.experimental.constrained.fptosi``' intrinsic converts 22083:ref:`floating-point <t_floating>` ``value`` to type ``ty2``. 22084 22085Arguments: 22086"""""""""" 22087 22088The first argument to the '``llvm.experimental.constrained.fptosi``' 22089intrinsic must be :ref:`floating point <t_floating>` or :ref:`vector 22090<t_vector>` of floating point values. 22091 22092The second argument specifies the exception behavior as described above. 22093 22094Semantics: 22095"""""""""" 22096 22097The result produced is a signed integer converted from the floating 22098point operand. The value is truncated, so it is rounded towards zero. 22099 22100'``llvm.experimental.constrained.uitofp``' Intrinsic 22101^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22102 22103Syntax: 22104""""""" 22105 22106:: 22107 22108 declare <ty2> 22109 @llvm.experimental.constrained.uitofp(<type> <value>, 22110 metadata <rounding mode>, 22111 metadata <exception behavior>) 22112 22113Overview: 22114""""""""" 22115 22116The '``llvm.experimental.constrained.uitofp``' intrinsic converts an 22117unsigned integer ``value`` to a floating-point of type ``ty2``. 22118 22119Arguments: 22120"""""""""" 22121 22122The first argument to the '``llvm.experimental.constrained.uitofp``' 22123intrinsic must be an :ref:`integer <t_integer>` or :ref:`vector 22124<t_vector>` of integer values. 22125 22126The second and third arguments specify the rounding mode and exception 22127behavior as described above. 22128 22129Semantics: 22130"""""""""" 22131 22132An inexact floating-point exception will be raised if rounding is required. 22133Any result produced is a floating point value converted from the input 22134integer operand. 22135 22136'``llvm.experimental.constrained.sitofp``' Intrinsic 22137^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22138 22139Syntax: 22140""""""" 22141 22142:: 22143 22144 declare <ty2> 22145 @llvm.experimental.constrained.sitofp(<type> <value>, 22146 metadata <rounding mode>, 22147 metadata <exception behavior>) 22148 22149Overview: 22150""""""""" 22151 22152The '``llvm.experimental.constrained.sitofp``' intrinsic converts a 22153signed integer ``value`` to a floating-point of type ``ty2``. 22154 22155Arguments: 22156"""""""""" 22157 22158The first argument to the '``llvm.experimental.constrained.sitofp``' 22159intrinsic must be an :ref:`integer <t_integer>` or :ref:`vector 22160<t_vector>` of integer values. 22161 22162The second and third arguments specify the rounding mode and exception 22163behavior as described above. 22164 22165Semantics: 22166"""""""""" 22167 22168An inexact floating-point exception will be raised if rounding is required. 22169Any result produced is a floating point value converted from the input 22170integer operand. 22171 22172'``llvm.experimental.constrained.fptrunc``' Intrinsic 22173^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22174 22175Syntax: 22176""""""" 22177 22178:: 22179 22180 declare <ty2> 22181 @llvm.experimental.constrained.fptrunc(<type> <value>, 22182 metadata <rounding mode>, 22183 metadata <exception behavior>) 22184 22185Overview: 22186""""""""" 22187 22188The '``llvm.experimental.constrained.fptrunc``' intrinsic truncates ``value`` 22189to type ``ty2``. 22190 22191Arguments: 22192"""""""""" 22193 22194The first argument to the '``llvm.experimental.constrained.fptrunc``' 22195intrinsic must be :ref:`floating point <t_floating>` or :ref:`vector 22196<t_vector>` of floating point values. This argument must be larger in size 22197than the result. 22198 22199The second and third arguments specify the rounding mode and exception 22200behavior as described above. 22201 22202Semantics: 22203"""""""""" 22204 22205The result produced is a floating point value truncated to be smaller in size 22206than the operand. 22207 22208'``llvm.experimental.constrained.fpext``' Intrinsic 22209^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22210 22211Syntax: 22212""""""" 22213 22214:: 22215 22216 declare <ty2> 22217 @llvm.experimental.constrained.fpext(<type> <value>, 22218 metadata <exception behavior>) 22219 22220Overview: 22221""""""""" 22222 22223The '``llvm.experimental.constrained.fpext``' intrinsic extends a 22224floating-point ``value`` to a larger floating-point value. 22225 22226Arguments: 22227"""""""""" 22228 22229The first argument to the '``llvm.experimental.constrained.fpext``' 22230intrinsic must be :ref:`floating point <t_floating>` or :ref:`vector 22231<t_vector>` of floating point values. This argument must be smaller in size 22232than the result. 22233 22234The second argument specifies the exception behavior as described above. 22235 22236Semantics: 22237"""""""""" 22238 22239The result produced is a floating point value extended to be larger in size 22240than the operand. All restrictions that apply to the fpext instruction also 22241apply to this intrinsic. 22242 22243'``llvm.experimental.constrained.fcmp``' and '``llvm.experimental.constrained.fcmps``' Intrinsics 22244^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22245 22246Syntax: 22247""""""" 22248 22249:: 22250 22251 declare <ty2> 22252 @llvm.experimental.constrained.fcmp(<type> <op1>, <type> <op2>, 22253 metadata <condition code>, 22254 metadata <exception behavior>) 22255 declare <ty2> 22256 @llvm.experimental.constrained.fcmps(<type> <op1>, <type> <op2>, 22257 metadata <condition code>, 22258 metadata <exception behavior>) 22259 22260Overview: 22261""""""""" 22262 22263The '``llvm.experimental.constrained.fcmp``' and 22264'``llvm.experimental.constrained.fcmps``' intrinsics return a boolean 22265value or vector of boolean values based on comparison of its operands. 22266 22267If the operands are floating-point scalars, then the result type is a 22268boolean (:ref:`i1 <t_integer>`). 22269 22270If the operands are floating-point vectors, then the result type is a 22271vector of boolean with the same number of elements as the operands being 22272compared. 22273 22274The '``llvm.experimental.constrained.fcmp``' intrinsic performs a quiet 22275comparison operation while the '``llvm.experimental.constrained.fcmps``' 22276intrinsic performs a signaling comparison operation. 22277 22278Arguments: 22279"""""""""" 22280 22281The first two arguments to the '``llvm.experimental.constrained.fcmp``' 22282and '``llvm.experimental.constrained.fcmps``' intrinsics must be 22283:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 22284of floating-point values. Both arguments must have identical types. 22285 22286The third argument is the condition code indicating the kind of comparison 22287to perform. It must be a metadata string with one of the following values: 22288 22289.. _fcmp_md_cc: 22290 22291- "``oeq``": ordered and equal 22292- "``ogt``": ordered and greater than 22293- "``oge``": ordered and greater than or equal 22294- "``olt``": ordered and less than 22295- "``ole``": ordered and less than or equal 22296- "``one``": ordered and not equal 22297- "``ord``": ordered (no nans) 22298- "``ueq``": unordered or equal 22299- "``ugt``": unordered or greater than 22300- "``uge``": unordered or greater than or equal 22301- "``ult``": unordered or less than 22302- "``ule``": unordered or less than or equal 22303- "``une``": unordered or not equal 22304- "``uno``": unordered (either nans) 22305 22306*Ordered* means that neither operand is a NAN while *unordered* means 22307that either operand may be a NAN. 22308 22309The fourth argument specifies the exception behavior as described above. 22310 22311Semantics: 22312"""""""""" 22313 22314``op1`` and ``op2`` are compared according to the condition code given 22315as the third argument. If the operands are vectors, then the 22316vectors are compared element by element. Each comparison performed 22317always yields an :ref:`i1 <t_integer>` result, as follows: 22318 22319.. _fcmp_md_cc_sem: 22320 22321- "``oeq``": yields ``true`` if both operands are not a NAN and ``op1`` 22322 is equal to ``op2``. 22323- "``ogt``": yields ``true`` if both operands are not a NAN and ``op1`` 22324 is greater than ``op2``. 22325- "``oge``": yields ``true`` if both operands are not a NAN and ``op1`` 22326 is greater than or equal to ``op2``. 22327- "``olt``": yields ``true`` if both operands are not a NAN and ``op1`` 22328 is less than ``op2``. 22329- "``ole``": yields ``true`` if both operands are not a NAN and ``op1`` 22330 is less than or equal to ``op2``. 22331- "``one``": yields ``true`` if both operands are not a NAN and ``op1`` 22332 is not equal to ``op2``. 22333- "``ord``": yields ``true`` if both operands are not a NAN. 22334- "``ueq``": yields ``true`` if either operand is a NAN or ``op1`` is 22335 equal to ``op2``. 22336- "``ugt``": yields ``true`` if either operand is a NAN or ``op1`` is 22337 greater than ``op2``. 22338- "``uge``": yields ``true`` if either operand is a NAN or ``op1`` is 22339 greater than or equal to ``op2``. 22340- "``ult``": yields ``true`` if either operand is a NAN or ``op1`` is 22341 less than ``op2``. 22342- "``ule``": yields ``true`` if either operand is a NAN or ``op1`` is 22343 less than or equal to ``op2``. 22344- "``une``": yields ``true`` if either operand is a NAN or ``op1`` is 22345 not equal to ``op2``. 22346- "``uno``": yields ``true`` if either operand is a NAN. 22347 22348The quiet comparison operation performed by 22349'``llvm.experimental.constrained.fcmp``' will only raise an exception 22350if either operand is a SNAN. The signaling comparison operation 22351performed by '``llvm.experimental.constrained.fcmps``' will raise an 22352exception if either operand is a NAN (QNAN or SNAN). Such an exception 22353does not preclude a result being produced (e.g. exception might only 22354set a flag), therefore the distinction between ordered and unordered 22355comparisons is also relevant for the 22356'``llvm.experimental.constrained.fcmps``' intrinsic. 22357 22358'``llvm.experimental.constrained.fmuladd``' Intrinsic 22359^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22360 22361Syntax: 22362""""""" 22363 22364:: 22365 22366 declare <type> 22367 @llvm.experimental.constrained.fmuladd(<type> <op1>, <type> <op2>, 22368 <type> <op3>, 22369 metadata <rounding mode>, 22370 metadata <exception behavior>) 22371 22372Overview: 22373""""""""" 22374 22375The '``llvm.experimental.constrained.fmuladd``' intrinsic represents 22376multiply-add expressions that can be fused if the code generator determines 22377that (a) the target instruction set has support for a fused operation, 22378and (b) that the fused operation is more efficient than the equivalent, 22379separate pair of mul and add instructions. 22380 22381Arguments: 22382"""""""""" 22383 22384The first three arguments to the '``llvm.experimental.constrained.fmuladd``' 22385intrinsic must be floating-point or vector of floating-point values. 22386All three arguments must have identical types. 22387 22388The fourth and fifth arguments specify the rounding mode and exception behavior 22389as described above. 22390 22391Semantics: 22392"""""""""" 22393 22394The expression: 22395 22396:: 22397 22398 %0 = call float @llvm.experimental.constrained.fmuladd.f32(%a, %b, %c, 22399 metadata <rounding mode>, 22400 metadata <exception behavior>) 22401 22402is equivalent to the expression: 22403 22404:: 22405 22406 %0 = call float @llvm.experimental.constrained.fmul.f32(%a, %b, 22407 metadata <rounding mode>, 22408 metadata <exception behavior>) 22409 %1 = call float @llvm.experimental.constrained.fadd.f32(%0, %c, 22410 metadata <rounding mode>, 22411 metadata <exception behavior>) 22412 22413except that it is unspecified whether rounding will be performed between the 22414multiplication and addition steps. Fusion is not guaranteed, even if the target 22415platform supports it. 22416If a fused multiply-add is required, the corresponding 22417:ref:`llvm.experimental.constrained.fma <int_fma>` intrinsic function should be 22418used instead. 22419This never sets errno, just as '``llvm.experimental.constrained.fma.*``'. 22420 22421Constrained libm-equivalent Intrinsics 22422-------------------------------------- 22423 22424In addition to the basic floating-point operations for which constrained 22425intrinsics are described above, there are constrained versions of various 22426operations which provide equivalent behavior to a corresponding libm function. 22427These intrinsics allow the precise behavior of these operations with respect to 22428rounding mode and exception behavior to be controlled. 22429 22430As with the basic constrained floating-point intrinsics, the rounding mode 22431and exception behavior arguments only control the behavior of the optimizer. 22432They do not change the runtime floating-point environment. 22433 22434 22435'``llvm.experimental.constrained.sqrt``' Intrinsic 22436^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22437 22438Syntax: 22439""""""" 22440 22441:: 22442 22443 declare <type> 22444 @llvm.experimental.constrained.sqrt(<type> <op1>, 22445 metadata <rounding mode>, 22446 metadata <exception behavior>) 22447 22448Overview: 22449""""""""" 22450 22451The '``llvm.experimental.constrained.sqrt``' intrinsic returns the square root 22452of the specified value, returning the same value as the libm '``sqrt``' 22453functions would, but without setting ``errno``. 22454 22455Arguments: 22456"""""""""" 22457 22458The first argument and the return type are floating-point numbers of the same 22459type. 22460 22461The second and third arguments specify the rounding mode and exception 22462behavior as described above. 22463 22464Semantics: 22465"""""""""" 22466 22467This function returns the nonnegative square root of the specified value. 22468If the value is less than negative zero, a floating-point exception occurs 22469and the return value is architecture specific. 22470 22471 22472'``llvm.experimental.constrained.pow``' Intrinsic 22473^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22474 22475Syntax: 22476""""""" 22477 22478:: 22479 22480 declare <type> 22481 @llvm.experimental.constrained.pow(<type> <op1>, <type> <op2>, 22482 metadata <rounding mode>, 22483 metadata <exception behavior>) 22484 22485Overview: 22486""""""""" 22487 22488The '``llvm.experimental.constrained.pow``' intrinsic returns the first operand 22489raised to the (positive or negative) power specified by the second operand. 22490 22491Arguments: 22492"""""""""" 22493 22494The first two arguments and the return value are floating-point numbers of the 22495same type. The second argument specifies the power to which the first argument 22496should be raised. 22497 22498The third and fourth arguments specify the rounding mode and exception 22499behavior as described above. 22500 22501Semantics: 22502"""""""""" 22503 22504This function returns the first value raised to the second power, 22505returning the same values as the libm ``pow`` functions would, and 22506handles error conditions in the same way. 22507 22508 22509'``llvm.experimental.constrained.powi``' Intrinsic 22510^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22511 22512Syntax: 22513""""""" 22514 22515:: 22516 22517 declare <type> 22518 @llvm.experimental.constrained.powi(<type> <op1>, i32 <op2>, 22519 metadata <rounding mode>, 22520 metadata <exception behavior>) 22521 22522Overview: 22523""""""""" 22524 22525The '``llvm.experimental.constrained.powi``' intrinsic returns the first operand 22526raised to the (positive or negative) power specified by the second operand. The 22527order of evaluation of multiplications is not defined. When a vector of 22528floating-point type is used, the second argument remains a scalar integer value. 22529 22530 22531Arguments: 22532"""""""""" 22533 22534The first argument and the return value are floating-point numbers of the same 22535type. The second argument is a 32-bit signed integer specifying the power to 22536which the first argument should be raised. 22537 22538The third and fourth arguments specify the rounding mode and exception 22539behavior as described above. 22540 22541Semantics: 22542"""""""""" 22543 22544This function returns the first value raised to the second power with an 22545unspecified sequence of rounding operations. 22546 22547 22548'``llvm.experimental.constrained.sin``' Intrinsic 22549^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22550 22551Syntax: 22552""""""" 22553 22554:: 22555 22556 declare <type> 22557 @llvm.experimental.constrained.sin(<type> <op1>, 22558 metadata <rounding mode>, 22559 metadata <exception behavior>) 22560 22561Overview: 22562""""""""" 22563 22564The '``llvm.experimental.constrained.sin``' intrinsic returns the sine of the 22565first operand. 22566 22567Arguments: 22568"""""""""" 22569 22570The first argument and the return type are floating-point numbers of the same 22571type. 22572 22573The second and third arguments specify the rounding mode and exception 22574behavior as described above. 22575 22576Semantics: 22577"""""""""" 22578 22579This function returns the sine of the specified operand, returning the 22580same values as the libm ``sin`` functions would, and handles error 22581conditions in the same way. 22582 22583 22584'``llvm.experimental.constrained.cos``' Intrinsic 22585^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22586 22587Syntax: 22588""""""" 22589 22590:: 22591 22592 declare <type> 22593 @llvm.experimental.constrained.cos(<type> <op1>, 22594 metadata <rounding mode>, 22595 metadata <exception behavior>) 22596 22597Overview: 22598""""""""" 22599 22600The '``llvm.experimental.constrained.cos``' intrinsic returns the cosine of the 22601first operand. 22602 22603Arguments: 22604"""""""""" 22605 22606The first argument and the return type are floating-point numbers of the same 22607type. 22608 22609The second and third arguments specify the rounding mode and exception 22610behavior as described above. 22611 22612Semantics: 22613"""""""""" 22614 22615This function returns the cosine of the specified operand, returning the 22616same values as the libm ``cos`` functions would, and handles error 22617conditions in the same way. 22618 22619 22620'``llvm.experimental.constrained.exp``' Intrinsic 22621^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22622 22623Syntax: 22624""""""" 22625 22626:: 22627 22628 declare <type> 22629 @llvm.experimental.constrained.exp(<type> <op1>, 22630 metadata <rounding mode>, 22631 metadata <exception behavior>) 22632 22633Overview: 22634""""""""" 22635 22636The '``llvm.experimental.constrained.exp``' intrinsic computes the base-e 22637exponential of the specified value. 22638 22639Arguments: 22640"""""""""" 22641 22642The first argument and the return value are floating-point numbers of the same 22643type. 22644 22645The second and third arguments specify the rounding mode and exception 22646behavior as described above. 22647 22648Semantics: 22649"""""""""" 22650 22651This function returns the same values as the libm ``exp`` functions 22652would, and handles error conditions in the same way. 22653 22654 22655'``llvm.experimental.constrained.exp2``' Intrinsic 22656^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22657 22658Syntax: 22659""""""" 22660 22661:: 22662 22663 declare <type> 22664 @llvm.experimental.constrained.exp2(<type> <op1>, 22665 metadata <rounding mode>, 22666 metadata <exception behavior>) 22667 22668Overview: 22669""""""""" 22670 22671The '``llvm.experimental.constrained.exp2``' intrinsic computes the base-2 22672exponential of the specified value. 22673 22674 22675Arguments: 22676"""""""""" 22677 22678The first argument and the return value are floating-point numbers of the same 22679type. 22680 22681The second and third arguments specify the rounding mode and exception 22682behavior as described above. 22683 22684Semantics: 22685"""""""""" 22686 22687This function returns the same values as the libm ``exp2`` functions 22688would, and handles error conditions in the same way. 22689 22690 22691'``llvm.experimental.constrained.log``' Intrinsic 22692^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22693 22694Syntax: 22695""""""" 22696 22697:: 22698 22699 declare <type> 22700 @llvm.experimental.constrained.log(<type> <op1>, 22701 metadata <rounding mode>, 22702 metadata <exception behavior>) 22703 22704Overview: 22705""""""""" 22706 22707The '``llvm.experimental.constrained.log``' intrinsic computes the base-e 22708logarithm of the specified value. 22709 22710Arguments: 22711"""""""""" 22712 22713The first argument and the return value are floating-point numbers of the same 22714type. 22715 22716The second and third arguments specify the rounding mode and exception 22717behavior as described above. 22718 22719 22720Semantics: 22721"""""""""" 22722 22723This function returns the same values as the libm ``log`` functions 22724would, and handles error conditions in the same way. 22725 22726 22727'``llvm.experimental.constrained.log10``' Intrinsic 22728^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22729 22730Syntax: 22731""""""" 22732 22733:: 22734 22735 declare <type> 22736 @llvm.experimental.constrained.log10(<type> <op1>, 22737 metadata <rounding mode>, 22738 metadata <exception behavior>) 22739 22740Overview: 22741""""""""" 22742 22743The '``llvm.experimental.constrained.log10``' intrinsic computes the base-10 22744logarithm of the specified value. 22745 22746Arguments: 22747"""""""""" 22748 22749The first argument and the return value are floating-point numbers of the same 22750type. 22751 22752The second and third arguments specify the rounding mode and exception 22753behavior as described above. 22754 22755Semantics: 22756"""""""""" 22757 22758This function returns the same values as the libm ``log10`` functions 22759would, and handles error conditions in the same way. 22760 22761 22762'``llvm.experimental.constrained.log2``' Intrinsic 22763^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22764 22765Syntax: 22766""""""" 22767 22768:: 22769 22770 declare <type> 22771 @llvm.experimental.constrained.log2(<type> <op1>, 22772 metadata <rounding mode>, 22773 metadata <exception behavior>) 22774 22775Overview: 22776""""""""" 22777 22778The '``llvm.experimental.constrained.log2``' intrinsic computes the base-2 22779logarithm of the specified value. 22780 22781Arguments: 22782"""""""""" 22783 22784The first argument and the return value are floating-point numbers of the same 22785type. 22786 22787The second and third arguments specify the rounding mode and exception 22788behavior as described above. 22789 22790Semantics: 22791"""""""""" 22792 22793This function returns the same values as the libm ``log2`` functions 22794would, and handles error conditions in the same way. 22795 22796 22797'``llvm.experimental.constrained.rint``' Intrinsic 22798^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22799 22800Syntax: 22801""""""" 22802 22803:: 22804 22805 declare <type> 22806 @llvm.experimental.constrained.rint(<type> <op1>, 22807 metadata <rounding mode>, 22808 metadata <exception behavior>) 22809 22810Overview: 22811""""""""" 22812 22813The '``llvm.experimental.constrained.rint``' intrinsic returns the first 22814operand rounded to the nearest integer. It may raise an inexact floating-point 22815exception if the operand is not an integer. 22816 22817Arguments: 22818"""""""""" 22819 22820The first argument and the return value are floating-point numbers of the same 22821type. 22822 22823The second and third arguments specify the rounding mode and exception 22824behavior as described above. 22825 22826Semantics: 22827"""""""""" 22828 22829This function returns the same values as the libm ``rint`` functions 22830would, and handles error conditions in the same way. The rounding mode is 22831described, not determined, by the rounding mode argument. The actual rounding 22832mode is determined by the runtime floating-point environment. The rounding 22833mode argument is only intended as information to the compiler. 22834 22835 22836'``llvm.experimental.constrained.lrint``' Intrinsic 22837^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22838 22839Syntax: 22840""""""" 22841 22842:: 22843 22844 declare <inttype> 22845 @llvm.experimental.constrained.lrint(<fptype> <op1>, 22846 metadata <rounding mode>, 22847 metadata <exception behavior>) 22848 22849Overview: 22850""""""""" 22851 22852The '``llvm.experimental.constrained.lrint``' intrinsic returns the first 22853operand rounded to the nearest integer. An inexact floating-point exception 22854will be raised if the operand is not an integer. An invalid exception is 22855raised if the result is too large to fit into a supported integer type, 22856and in this case the result is undefined. 22857 22858Arguments: 22859"""""""""" 22860 22861The first argument is a floating-point number. The return value is an 22862integer type. Not all types are supported on all targets. The supported 22863types are the same as the ``llvm.lrint`` intrinsic and the ``lrint`` 22864libm functions. 22865 22866The second and third arguments specify the rounding mode and exception 22867behavior as described above. 22868 22869Semantics: 22870"""""""""" 22871 22872This function returns the same values as the libm ``lrint`` functions 22873would, and handles error conditions in the same way. 22874 22875The rounding mode is described, not determined, by the rounding mode 22876argument. The actual rounding mode is determined by the runtime floating-point 22877environment. The rounding mode argument is only intended as information 22878to the compiler. 22879 22880If the runtime floating-point environment is using the default rounding mode 22881then the results will be the same as the llvm.lrint intrinsic. 22882 22883 22884'``llvm.experimental.constrained.llrint``' Intrinsic 22885^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22886 22887Syntax: 22888""""""" 22889 22890:: 22891 22892 declare <inttype> 22893 @llvm.experimental.constrained.llrint(<fptype> <op1>, 22894 metadata <rounding mode>, 22895 metadata <exception behavior>) 22896 22897Overview: 22898""""""""" 22899 22900The '``llvm.experimental.constrained.llrint``' intrinsic returns the first 22901operand rounded to the nearest integer. An inexact floating-point exception 22902will be raised if the operand is not an integer. An invalid exception is 22903raised if the result is too large to fit into a supported integer type, 22904and in this case the result is undefined. 22905 22906Arguments: 22907"""""""""" 22908 22909The first argument is a floating-point number. The return value is an 22910integer type. Not all types are supported on all targets. The supported 22911types are the same as the ``llvm.llrint`` intrinsic and the ``llrint`` 22912libm functions. 22913 22914The second and third arguments specify the rounding mode and exception 22915behavior as described above. 22916 22917Semantics: 22918"""""""""" 22919 22920This function returns the same values as the libm ``llrint`` functions 22921would, and handles error conditions in the same way. 22922 22923The rounding mode is described, not determined, by the rounding mode 22924argument. The actual rounding mode is determined by the runtime floating-point 22925environment. The rounding mode argument is only intended as information 22926to the compiler. 22927 22928If the runtime floating-point environment is using the default rounding mode 22929then the results will be the same as the llvm.llrint intrinsic. 22930 22931 22932'``llvm.experimental.constrained.nearbyint``' Intrinsic 22933^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22934 22935Syntax: 22936""""""" 22937 22938:: 22939 22940 declare <type> 22941 @llvm.experimental.constrained.nearbyint(<type> <op1>, 22942 metadata <rounding mode>, 22943 metadata <exception behavior>) 22944 22945Overview: 22946""""""""" 22947 22948The '``llvm.experimental.constrained.nearbyint``' intrinsic returns the first 22949operand rounded to the nearest integer. It will not raise an inexact 22950floating-point exception if the operand is not an integer. 22951 22952 22953Arguments: 22954"""""""""" 22955 22956The first argument and the return value are floating-point numbers of the same 22957type. 22958 22959The second and third arguments specify the rounding mode and exception 22960behavior as described above. 22961 22962Semantics: 22963"""""""""" 22964 22965This function returns the same values as the libm ``nearbyint`` functions 22966would, and handles error conditions in the same way. The rounding mode is 22967described, not determined, by the rounding mode argument. The actual rounding 22968mode is determined by the runtime floating-point environment. The rounding 22969mode argument is only intended as information to the compiler. 22970 22971 22972'``llvm.experimental.constrained.maxnum``' Intrinsic 22973^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 22974 22975Syntax: 22976""""""" 22977 22978:: 22979 22980 declare <type> 22981 @llvm.experimental.constrained.maxnum(<type> <op1>, <type> <op2> 22982 metadata <exception behavior>) 22983 22984Overview: 22985""""""""" 22986 22987The '``llvm.experimental.constrained.maxnum``' intrinsic returns the maximum 22988of the two arguments. 22989 22990Arguments: 22991"""""""""" 22992 22993The first two arguments and the return value are floating-point numbers 22994of the same type. 22995 22996The third argument specifies the exception behavior as described above. 22997 22998Semantics: 22999"""""""""" 23000 23001This function follows the IEEE-754 semantics for maxNum. 23002 23003 23004'``llvm.experimental.constrained.minnum``' Intrinsic 23005^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23006 23007Syntax: 23008""""""" 23009 23010:: 23011 23012 declare <type> 23013 @llvm.experimental.constrained.minnum(<type> <op1>, <type> <op2> 23014 metadata <exception behavior>) 23015 23016Overview: 23017""""""""" 23018 23019The '``llvm.experimental.constrained.minnum``' intrinsic returns the minimum 23020of the two arguments. 23021 23022Arguments: 23023"""""""""" 23024 23025The first two arguments and the return value are floating-point numbers 23026of the same type. 23027 23028The third argument specifies the exception behavior as described above. 23029 23030Semantics: 23031"""""""""" 23032 23033This function follows the IEEE-754 semantics for minNum. 23034 23035 23036'``llvm.experimental.constrained.maximum``' Intrinsic 23037^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23038 23039Syntax: 23040""""""" 23041 23042:: 23043 23044 declare <type> 23045 @llvm.experimental.constrained.maximum(<type> <op1>, <type> <op2> 23046 metadata <exception behavior>) 23047 23048Overview: 23049""""""""" 23050 23051The '``llvm.experimental.constrained.maximum``' intrinsic returns the maximum 23052of the two arguments, propagating NaNs and treating -0.0 as less than +0.0. 23053 23054Arguments: 23055"""""""""" 23056 23057The first two arguments and the return value are floating-point numbers 23058of the same type. 23059 23060The third argument specifies the exception behavior as described above. 23061 23062Semantics: 23063"""""""""" 23064 23065This function follows semantics specified in the draft of IEEE 754-2018. 23066 23067 23068'``llvm.experimental.constrained.minimum``' Intrinsic 23069^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23070 23071Syntax: 23072""""""" 23073 23074:: 23075 23076 declare <type> 23077 @llvm.experimental.constrained.minimum(<type> <op1>, <type> <op2> 23078 metadata <exception behavior>) 23079 23080Overview: 23081""""""""" 23082 23083The '``llvm.experimental.constrained.minimum``' intrinsic returns the minimum 23084of the two arguments, propagating NaNs and treating -0.0 as less than +0.0. 23085 23086Arguments: 23087"""""""""" 23088 23089The first two arguments and the return value are floating-point numbers 23090of the same type. 23091 23092The third argument specifies the exception behavior as described above. 23093 23094Semantics: 23095"""""""""" 23096 23097This function follows semantics specified in the draft of IEEE 754-2018. 23098 23099 23100'``llvm.experimental.constrained.ceil``' Intrinsic 23101^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23102 23103Syntax: 23104""""""" 23105 23106:: 23107 23108 declare <type> 23109 @llvm.experimental.constrained.ceil(<type> <op1>, 23110 metadata <exception behavior>) 23111 23112Overview: 23113""""""""" 23114 23115The '``llvm.experimental.constrained.ceil``' intrinsic returns the ceiling of the 23116first operand. 23117 23118Arguments: 23119"""""""""" 23120 23121The first argument and the return value are floating-point numbers of the same 23122type. 23123 23124The second argument specifies the exception behavior as described above. 23125 23126Semantics: 23127"""""""""" 23128 23129This function returns the same values as the libm ``ceil`` functions 23130would and handles error conditions in the same way. 23131 23132 23133'``llvm.experimental.constrained.floor``' Intrinsic 23134^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23135 23136Syntax: 23137""""""" 23138 23139:: 23140 23141 declare <type> 23142 @llvm.experimental.constrained.floor(<type> <op1>, 23143 metadata <exception behavior>) 23144 23145Overview: 23146""""""""" 23147 23148The '``llvm.experimental.constrained.floor``' intrinsic returns the floor of the 23149first operand. 23150 23151Arguments: 23152"""""""""" 23153 23154The first argument and the return value are floating-point numbers of the same 23155type. 23156 23157The second argument specifies the exception behavior as described above. 23158 23159Semantics: 23160"""""""""" 23161 23162This function returns the same values as the libm ``floor`` functions 23163would and handles error conditions in the same way. 23164 23165 23166'``llvm.experimental.constrained.round``' Intrinsic 23167^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23168 23169Syntax: 23170""""""" 23171 23172:: 23173 23174 declare <type> 23175 @llvm.experimental.constrained.round(<type> <op1>, 23176 metadata <exception behavior>) 23177 23178Overview: 23179""""""""" 23180 23181The '``llvm.experimental.constrained.round``' intrinsic returns the first 23182operand rounded to the nearest integer. 23183 23184Arguments: 23185"""""""""" 23186 23187The first argument and the return value are floating-point numbers of the same 23188type. 23189 23190The second argument specifies the exception behavior as described above. 23191 23192Semantics: 23193"""""""""" 23194 23195This function returns the same values as the libm ``round`` functions 23196would and handles error conditions in the same way. 23197 23198 23199'``llvm.experimental.constrained.roundeven``' Intrinsic 23200^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23201 23202Syntax: 23203""""""" 23204 23205:: 23206 23207 declare <type> 23208 @llvm.experimental.constrained.roundeven(<type> <op1>, 23209 metadata <exception behavior>) 23210 23211Overview: 23212""""""""" 23213 23214The '``llvm.experimental.constrained.roundeven``' intrinsic returns the first 23215operand rounded to the nearest integer in floating-point format, rounding 23216halfway cases to even (that is, to the nearest value that is an even integer), 23217regardless of the current rounding direction. 23218 23219Arguments: 23220"""""""""" 23221 23222The first argument and the return value are floating-point numbers of the same 23223type. 23224 23225The second argument specifies the exception behavior as described above. 23226 23227Semantics: 23228"""""""""" 23229 23230This function implements IEEE-754 operation ``roundToIntegralTiesToEven``. It 23231also behaves in the same way as C standard function ``roundeven`` and can signal 23232the invalid operation exception for a SNAN operand. 23233 23234 23235'``llvm.experimental.constrained.lround``' Intrinsic 23236^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23237 23238Syntax: 23239""""""" 23240 23241:: 23242 23243 declare <inttype> 23244 @llvm.experimental.constrained.lround(<fptype> <op1>, 23245 metadata <exception behavior>) 23246 23247Overview: 23248""""""""" 23249 23250The '``llvm.experimental.constrained.lround``' intrinsic returns the first 23251operand rounded to the nearest integer with ties away from zero. It will 23252raise an inexact floating-point exception if the operand is not an integer. 23253An invalid exception is raised if the result is too large to fit into a 23254supported integer type, and in this case the result is undefined. 23255 23256Arguments: 23257"""""""""" 23258 23259The first argument is a floating-point number. The return value is an 23260integer type. Not all types are supported on all targets. The supported 23261types are the same as the ``llvm.lround`` intrinsic and the ``lround`` 23262libm functions. 23263 23264The second argument specifies the exception behavior as described above. 23265 23266Semantics: 23267"""""""""" 23268 23269This function returns the same values as the libm ``lround`` functions 23270would and handles error conditions in the same way. 23271 23272 23273'``llvm.experimental.constrained.llround``' Intrinsic 23274^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23275 23276Syntax: 23277""""""" 23278 23279:: 23280 23281 declare <inttype> 23282 @llvm.experimental.constrained.llround(<fptype> <op1>, 23283 metadata <exception behavior>) 23284 23285Overview: 23286""""""""" 23287 23288The '``llvm.experimental.constrained.llround``' intrinsic returns the first 23289operand rounded to the nearest integer with ties away from zero. It will 23290raise an inexact floating-point exception if the operand is not an integer. 23291An invalid exception is raised if the result is too large to fit into a 23292supported integer type, and in this case the result is undefined. 23293 23294Arguments: 23295"""""""""" 23296 23297The first argument is a floating-point number. The return value is an 23298integer type. Not all types are supported on all targets. The supported 23299types are the same as the ``llvm.llround`` intrinsic and the ``llround`` 23300libm functions. 23301 23302The second argument specifies the exception behavior as described above. 23303 23304Semantics: 23305"""""""""" 23306 23307This function returns the same values as the libm ``llround`` functions 23308would and handles error conditions in the same way. 23309 23310 23311'``llvm.experimental.constrained.trunc``' Intrinsic 23312^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23313 23314Syntax: 23315""""""" 23316 23317:: 23318 23319 declare <type> 23320 @llvm.experimental.constrained.trunc(<type> <op1>, 23321 metadata <exception behavior>) 23322 23323Overview: 23324""""""""" 23325 23326The '``llvm.experimental.constrained.trunc``' intrinsic returns the first 23327operand rounded to the nearest integer not larger in magnitude than the 23328operand. 23329 23330Arguments: 23331"""""""""" 23332 23333The first argument and the return value are floating-point numbers of the same 23334type. 23335 23336The second argument specifies the exception behavior as described above. 23337 23338Semantics: 23339"""""""""" 23340 23341This function returns the same values as the libm ``trunc`` functions 23342would and handles error conditions in the same way. 23343 23344.. _int_experimental_noalias_scope_decl: 23345 23346'``llvm.experimental.noalias.scope.decl``' Intrinsic 23347^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23348 23349Syntax: 23350""""""" 23351 23352 23353:: 23354 23355 declare void @llvm.experimental.noalias.scope.decl(metadata !id.scope.list) 23356 23357Overview: 23358""""""""" 23359 23360The ``llvm.experimental.noalias.scope.decl`` intrinsic identifies where a 23361noalias scope is declared. When the intrinsic is duplicated, a decision must 23362also be made about the scope: depending on the reason of the duplication, 23363the scope might need to be duplicated as well. 23364 23365 23366Arguments: 23367"""""""""" 23368 23369The ``!id.scope.list`` argument is metadata that is a list of ``noalias`` 23370metadata references. The format is identical to that required for ``noalias`` 23371metadata. This list must have exactly one element. 23372 23373Semantics: 23374"""""""""" 23375 23376The ``llvm.experimental.noalias.scope.decl`` intrinsic identifies where a 23377noalias scope is declared. When the intrinsic is duplicated, a decision must 23378also be made about the scope: depending on the reason of the duplication, 23379the scope might need to be duplicated as well. 23380 23381For example, when the intrinsic is used inside a loop body, and that loop is 23382unrolled, the associated noalias scope must also be duplicated. Otherwise, the 23383noalias property it signifies would spill across loop iterations, whereas it 23384was only valid within a single iteration. 23385 23386.. code-block:: llvm 23387 23388 ; This examples shows two possible positions for noalias.decl and how they impact the semantics: 23389 ; If it is outside the loop (Version 1), then %a and %b are noalias across *all* iterations. 23390 ; If it is inside the loop (Version 2), then %a and %b are noalias only within *one* iteration. 23391 declare void @decl_in_loop(i8* %a.base, i8* %b.base) { 23392 entry: 23393 ; call void @llvm.experimental.noalias.scope.decl(metadata !2) ; Version 1: noalias decl outside loop 23394 br label %loop 23395 23396 loop: 23397 %a = phi i8* [ %a.base, %entry ], [ %a.inc, %loop ] 23398 %b = phi i8* [ %b.base, %entry ], [ %b.inc, %loop ] 23399 ; call void @llvm.experimental.noalias.scope.decl(metadata !2) ; Version 2: noalias decl inside loop 23400 %val = load i8, i8* %a, !alias.scope !2 23401 store i8 %val, i8* %b, !noalias !2 23402 %a.inc = getelementptr inbounds i8, i8* %a, i64 1 23403 %b.inc = getelementptr inbounds i8, i8* %b, i64 1 23404 %cond = call i1 @cond() 23405 br i1 %cond, label %loop, label %exit 23406 23407 exit: 23408 ret void 23409 } 23410 23411 !0 = !{!0} ; domain 23412 !1 = !{!1, !0} ; scope 23413 !2 = !{!1} ; scope list 23414 23415Multiple calls to `@llvm.experimental.noalias.scope.decl` for the same scope 23416are possible, but one should never dominate another. Violations are pointed out 23417by the verifier as they indicate a problem in either a transformation pass or 23418the input. 23419 23420 23421Floating Point Environment Manipulation intrinsics 23422-------------------------------------------------- 23423 23424These functions read or write floating point environment, such as rounding 23425mode or state of floating point exceptions. Altering the floating point 23426environment requires special care. See :ref:`Floating Point Environment <floatenv>`. 23427 23428'``llvm.flt.rounds``' Intrinsic 23429^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23430 23431Syntax: 23432""""""" 23433 23434:: 23435 23436 declare i32 @llvm.flt.rounds() 23437 23438Overview: 23439""""""""" 23440 23441The '``llvm.flt.rounds``' intrinsic reads the current rounding mode. 23442 23443Semantics: 23444"""""""""" 23445 23446The '``llvm.flt.rounds``' intrinsic returns the current rounding mode. 23447Encoding of the returned values is same as the result of ``FLT_ROUNDS``, 23448specified by C standard: 23449 23450:: 23451 23452 0 - toward zero 23453 1 - to nearest, ties to even 23454 2 - toward positive infinity 23455 3 - toward negative infinity 23456 4 - to nearest, ties away from zero 23457 23458Other values may be used to represent additional rounding modes, supported by a 23459target. These values are target-specific. 23460 23461 23462'``llvm.set.rounding``' Intrinsic 23463^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23464 23465Syntax: 23466""""""" 23467 23468:: 23469 23470 declare void @llvm.set.rounding(i32 <val>) 23471 23472Overview: 23473""""""""" 23474 23475The '``llvm.set.rounding``' intrinsic sets current rounding mode. 23476 23477Arguments: 23478"""""""""" 23479 23480The argument is the required rounding mode. Encoding of rounding mode is 23481the same as used by '``llvm.flt.rounds``'. 23482 23483Semantics: 23484"""""""""" 23485 23486The '``llvm.set.rounding``' intrinsic sets the current rounding mode. It is 23487similar to C library function 'fesetround', however this intrinsic does not 23488return any value and uses platform-independent representation of IEEE rounding 23489modes. 23490 23491 23492Floating-Point Test Intrinsics 23493------------------------------ 23494 23495These functions get properties of floating-point values. 23496 23497 23498'``llvm.is.fpclass``' Intrinsic 23499^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23500 23501Syntax: 23502""""""" 23503 23504:: 23505 23506 declare i1 @llvm.is.fpclass(<fptype> <op>, i32 <test>) 23507 declare <N x i1> @llvm.is.fpclass(<vector-fptype> <op>, i32 <test>) 23508 23509Overview: 23510""""""""" 23511 23512The '``llvm.is.fpclass``' intrinsic returns a boolean value or vector of boolean 23513values depending on whether the first argument satisfies the test specified by 23514the second argument. 23515 23516If the first argument is a floating-point scalar, then the result type is a 23517boolean (:ref:`i1 <t_integer>`). 23518 23519If the first argument is a floating-point vector, then the result type is a 23520vector of boolean with the same number of elements as the first argument. 23521 23522Arguments: 23523"""""""""" 23524 23525The first argument to the '``llvm.is.fpclass``' intrinsic must be 23526:ref:`floating-point <t_floating>` or :ref:`vector <t_vector>` 23527of floating-point values. 23528 23529The second argument specifies, which tests to perform. It must be a compile-time 23530integer constant, each bit in which specifies floating-point class: 23531 23532+-------+----------------------+ 23533| Bit # | floating-point class | 23534+=======+======================+ 23535| 0 | Signaling NaN | 23536+-------+----------------------+ 23537| 1 | Quiet NaN | 23538+-------+----------------------+ 23539| 2 | Negative infinity | 23540+-------+----------------------+ 23541| 3 | Negative normal | 23542+-------+----------------------+ 23543| 4 | Negative subnormal | 23544+-------+----------------------+ 23545| 5 | Negative zero | 23546+-------+----------------------+ 23547| 6 | Positive zero | 23548+-------+----------------------+ 23549| 7 | Positive subnormal | 23550+-------+----------------------+ 23551| 8 | Positive normal | 23552+-------+----------------------+ 23553| 9 | Positive infinity | 23554+-------+----------------------+ 23555 23556Semantics: 23557"""""""""" 23558 23559The function checks if ``op`` belongs to any of the floating-point classes 23560specified by ``test``. If ``op`` is a vector, then the check is made element by 23561element. Each check yields an :ref:`i1 <t_integer>` result, which is ``true``, 23562if the element value satisfies the specified test. The argument ``test`` is a 23563bit mask where each bit specifies floating-point class to test. For example, the 23564value 0x108 makes test for normal value, - bits 3 and 8 in it are set, which 23565means that the function returns ``true`` if ``op`` is a positive or negative 23566normal value. The function never raises floating-point exceptions. 23567 23568 23569General Intrinsics 23570------------------ 23571 23572This class of intrinsics is designed to be generic and has no specific 23573purpose. 23574 23575'``llvm.var.annotation``' Intrinsic 23576^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23577 23578Syntax: 23579""""""" 23580 23581:: 23582 23583 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>) 23584 23585Overview: 23586""""""""" 23587 23588The '``llvm.var.annotation``' intrinsic. 23589 23590Arguments: 23591"""""""""" 23592 23593The first argument is a pointer to a value, the second is a pointer to a 23594global string, the third is a pointer to a global string which is the 23595source file name, and the last argument is the line number. 23596 23597Semantics: 23598"""""""""" 23599 23600This intrinsic allows annotation of local variables with arbitrary 23601strings. This can be useful for special purpose optimizations that want 23602to look for these annotations. These have no other defined use; they are 23603ignored by code generation and optimization. 23604 23605'``llvm.ptr.annotation.*``' Intrinsic 23606^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23607 23608Syntax: 23609""""""" 23610 23611This is an overloaded intrinsic. You can use '``llvm.ptr.annotation``' on a 23612pointer to an integer of any width. *NOTE* you must specify an address space for 23613the pointer. The identifier for the default address space is the integer 23614'``0``'. 23615 23616:: 23617 23618 declare i8* @llvm.ptr.annotation.p<address space>i8(i8* <val>, i8* <str>, i8* <str>, i32 <int>) 23619 declare i16* @llvm.ptr.annotation.p<address space>i16(i16* <val>, i8* <str>, i8* <str>, i32 <int>) 23620 declare i32* @llvm.ptr.annotation.p<address space>i32(i32* <val>, i8* <str>, i8* <str>, i32 <int>) 23621 declare i64* @llvm.ptr.annotation.p<address space>i64(i64* <val>, i8* <str>, i8* <str>, i32 <int>) 23622 declare i256* @llvm.ptr.annotation.p<address space>i256(i256* <val>, i8* <str>, i8* <str>, i32 <int>) 23623 23624Overview: 23625""""""""" 23626 23627The '``llvm.ptr.annotation``' intrinsic. 23628 23629Arguments: 23630"""""""""" 23631 23632The first argument is a pointer to an integer value of arbitrary bitwidth 23633(result of some expression), the second is a pointer to a global string, the 23634third is a pointer to a global string which is the source file name, and the 23635last argument is the line number. It returns the value of the first argument. 23636 23637Semantics: 23638"""""""""" 23639 23640This intrinsic allows annotation of a pointer to an integer with arbitrary 23641strings. This can be useful for special purpose optimizations that want to look 23642for these annotations. These have no other defined use; transformations preserve 23643annotations on a best-effort basis but are allowed to replace the intrinsic with 23644its first argument without breaking semantics and the intrinsic is completely 23645dropped during instruction selection. 23646 23647'``llvm.annotation.*``' Intrinsic 23648^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23649 23650Syntax: 23651""""""" 23652 23653This is an overloaded intrinsic. You can use '``llvm.annotation``' on 23654any integer bit width. 23655 23656:: 23657 23658 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>) 23659 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>) 23660 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>) 23661 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>) 23662 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>) 23663 23664Overview: 23665""""""""" 23666 23667The '``llvm.annotation``' intrinsic. 23668 23669Arguments: 23670"""""""""" 23671 23672The first argument is an integer value (result of some expression), the 23673second is a pointer to a global string, the third is a pointer to a 23674global string which is the source file name, and the last argument is 23675the line number. It returns the value of the first argument. 23676 23677Semantics: 23678"""""""""" 23679 23680This intrinsic allows annotations to be put on arbitrary expressions with 23681arbitrary strings. This can be useful for special purpose optimizations that 23682want to look for these annotations. These have no other defined use; 23683transformations preserve annotations on a best-effort basis but are allowed to 23684replace the intrinsic with its first argument without breaking semantics and the 23685intrinsic is completely dropped during instruction selection. 23686 23687'``llvm.codeview.annotation``' Intrinsic 23688^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23689 23690Syntax: 23691""""""" 23692 23693This annotation emits a label at its program point and an associated 23694``S_ANNOTATION`` codeview record with some additional string metadata. This is 23695used to implement MSVC's ``__annotation`` intrinsic. It is marked 23696``noduplicate``, so calls to this intrinsic prevent inlining and should be 23697considered expensive. 23698 23699:: 23700 23701 declare void @llvm.codeview.annotation(metadata) 23702 23703Arguments: 23704"""""""""" 23705 23706The argument should be an MDTuple containing any number of MDStrings. 23707 23708'``llvm.trap``' Intrinsic 23709^^^^^^^^^^^^^^^^^^^^^^^^^ 23710 23711Syntax: 23712""""""" 23713 23714:: 23715 23716 declare void @llvm.trap() cold noreturn nounwind 23717 23718Overview: 23719""""""""" 23720 23721The '``llvm.trap``' intrinsic. 23722 23723Arguments: 23724"""""""""" 23725 23726None. 23727 23728Semantics: 23729"""""""""" 23730 23731This intrinsic is lowered to the target dependent trap instruction. If 23732the target does not have a trap instruction, this intrinsic will be 23733lowered to a call of the ``abort()`` function. 23734 23735'``llvm.debugtrap``' Intrinsic 23736^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23737 23738Syntax: 23739""""""" 23740 23741:: 23742 23743 declare void @llvm.debugtrap() nounwind 23744 23745Overview: 23746""""""""" 23747 23748The '``llvm.debugtrap``' intrinsic. 23749 23750Arguments: 23751"""""""""" 23752 23753None. 23754 23755Semantics: 23756"""""""""" 23757 23758This intrinsic is lowered to code which is intended to cause an 23759execution trap with the intention of requesting the attention of a 23760debugger. 23761 23762'``llvm.ubsantrap``' Intrinsic 23763^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23764 23765Syntax: 23766""""""" 23767 23768:: 23769 23770 declare void @llvm.ubsantrap(i8 immarg) cold noreturn nounwind 23771 23772Overview: 23773""""""""" 23774 23775The '``llvm.ubsantrap``' intrinsic. 23776 23777Arguments: 23778"""""""""" 23779 23780An integer describing the kind of failure detected. 23781 23782Semantics: 23783"""""""""" 23784 23785This intrinsic is lowered to code which is intended to cause an execution trap, 23786embedding the argument into encoding of that trap somehow to discriminate 23787crashes if possible. 23788 23789Equivalent to ``@llvm.trap`` for targets that do not support this behaviour. 23790 23791'``llvm.stackprotector``' Intrinsic 23792^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23793 23794Syntax: 23795""""""" 23796 23797:: 23798 23799 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>) 23800 23801Overview: 23802""""""""" 23803 23804The ``llvm.stackprotector`` intrinsic takes the ``guard`` and stores it 23805onto the stack at ``slot``. The stack slot is adjusted to ensure that it 23806is placed on the stack before local variables. 23807 23808Arguments: 23809"""""""""" 23810 23811The ``llvm.stackprotector`` intrinsic requires two pointer arguments. 23812The first argument is the value loaded from the stack guard 23813``@__stack_chk_guard``. The second variable is an ``alloca`` that has 23814enough space to hold the value of the guard. 23815 23816Semantics: 23817"""""""""" 23818 23819This intrinsic causes the prologue/epilogue inserter to force the position of 23820the ``AllocaInst`` stack slot to be before local variables on the stack. This is 23821to ensure that if a local variable on the stack is overwritten, it will destroy 23822the value of the guard. When the function exits, the guard on the stack is 23823checked against the original guard by ``llvm.stackprotectorcheck``. If they are 23824different, then ``llvm.stackprotectorcheck`` causes the program to abort by 23825calling the ``__stack_chk_fail()`` function. 23826 23827'``llvm.stackguard``' Intrinsic 23828^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23829 23830Syntax: 23831""""""" 23832 23833:: 23834 23835 declare i8* @llvm.stackguard() 23836 23837Overview: 23838""""""""" 23839 23840The ``llvm.stackguard`` intrinsic returns the system stack guard value. 23841 23842It should not be generated by frontends, since it is only for internal usage. 23843The reason why we create this intrinsic is that we still support IR form Stack 23844Protector in FastISel. 23845 23846Arguments: 23847"""""""""" 23848 23849None. 23850 23851Semantics: 23852"""""""""" 23853 23854On some platforms, the value returned by this intrinsic remains unchanged 23855between loads in the same thread. On other platforms, it returns the same 23856global variable value, if any, e.g. ``@__stack_chk_guard``. 23857 23858Currently some platforms have IR-level customized stack guard loading (e.g. 23859X86 Linux) that is not handled by ``llvm.stackguard()``, while they should be 23860in the future. 23861 23862'``llvm.objectsize``' Intrinsic 23863^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23864 23865Syntax: 23866""""""" 23867 23868:: 23869 23870 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <min>, i1 <nullunknown>, i1 <dynamic>) 23871 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <min>, i1 <nullunknown>, i1 <dynamic>) 23872 23873Overview: 23874""""""""" 23875 23876The ``llvm.objectsize`` intrinsic is designed to provide information to the 23877optimizer to determine whether a) an operation (like memcpy) will overflow a 23878buffer that corresponds to an object, or b) that a runtime check for overflow 23879isn't necessary. An object in this context means an allocation of a specific 23880class, structure, array, or other object. 23881 23882Arguments: 23883"""""""""" 23884 23885The ``llvm.objectsize`` intrinsic takes four arguments. The first argument is a 23886pointer to or into the ``object``. The second argument determines whether 23887``llvm.objectsize`` returns 0 (if true) or -1 (if false) when the object size is 23888unknown. The third argument controls how ``llvm.objectsize`` acts when ``null`` 23889in address space 0 is used as its pointer argument. If it's ``false``, 23890``llvm.objectsize`` reports 0 bytes available when given ``null``. Otherwise, if 23891the ``null`` is in a non-zero address space or if ``true`` is given for the 23892third argument of ``llvm.objectsize``, we assume its size is unknown. The fourth 23893argument to ``llvm.objectsize`` determines if the value should be evaluated at 23894runtime. 23895 23896The second, third, and fourth arguments only accept constants. 23897 23898Semantics: 23899"""""""""" 23900 23901The ``llvm.objectsize`` intrinsic is lowered to a value representing the size of 23902the object concerned. If the size cannot be determined, ``llvm.objectsize`` 23903returns ``i32/i64 -1 or 0`` (depending on the ``min`` argument). 23904 23905'``llvm.expect``' Intrinsic 23906^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23907 23908Syntax: 23909""""""" 23910 23911This is an overloaded intrinsic. You can use ``llvm.expect`` on any 23912integer bit width. 23913 23914:: 23915 23916 declare i1 @llvm.expect.i1(i1 <val>, i1 <expected_val>) 23917 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>) 23918 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>) 23919 23920Overview: 23921""""""""" 23922 23923The ``llvm.expect`` intrinsic provides information about expected (the 23924most probable) value of ``val``, which can be used by optimizers. 23925 23926Arguments: 23927"""""""""" 23928 23929The ``llvm.expect`` intrinsic takes two arguments. The first argument is 23930a value. The second argument is an expected value. 23931 23932Semantics: 23933"""""""""" 23934 23935This intrinsic is lowered to the ``val``. 23936 23937'``llvm.expect.with.probability``' Intrinsic 23938^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23939 23940Syntax: 23941""""""" 23942 23943This intrinsic is similar to ``llvm.expect``. This is an overloaded intrinsic. 23944You can use ``llvm.expect.with.probability`` on any integer bit width. 23945 23946:: 23947 23948 declare i1 @llvm.expect.with.probability.i1(i1 <val>, i1 <expected_val>, double <prob>) 23949 declare i32 @llvm.expect.with.probability.i32(i32 <val>, i32 <expected_val>, double <prob>) 23950 declare i64 @llvm.expect.with.probability.i64(i64 <val>, i64 <expected_val>, double <prob>) 23951 23952Overview: 23953""""""""" 23954 23955The ``llvm.expect.with.probability`` intrinsic provides information about 23956expected value of ``val`` with probability(or confidence) ``prob``, which can 23957be used by optimizers. 23958 23959Arguments: 23960"""""""""" 23961 23962The ``llvm.expect.with.probability`` intrinsic takes three arguments. The first 23963argument is a value. The second argument is an expected value. The third 23964argument is a probability. 23965 23966Semantics: 23967"""""""""" 23968 23969This intrinsic is lowered to the ``val``. 23970 23971.. _int_assume: 23972 23973'``llvm.assume``' Intrinsic 23974^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 23975 23976Syntax: 23977""""""" 23978 23979:: 23980 23981 declare void @llvm.assume(i1 %cond) 23982 23983Overview: 23984""""""""" 23985 23986The ``llvm.assume`` allows the optimizer to assume that the provided 23987condition is true. This information can then be used in simplifying other parts 23988of the code. 23989 23990More complex assumptions can be encoded as 23991:ref:`assume operand bundles <assume_opbundles>`. 23992 23993Arguments: 23994"""""""""" 23995 23996The argument of the call is the condition which the optimizer may assume is 23997always true. 23998 23999Semantics: 24000"""""""""" 24001 24002The intrinsic allows the optimizer to assume that the provided condition is 24003always true whenever the control flow reaches the intrinsic call. No code is 24004generated for this intrinsic, and instructions that contribute only to the 24005provided condition are not used for code generation. If the condition is 24006violated during execution, the behavior is undefined. 24007 24008Note that the optimizer might limit the transformations performed on values 24009used by the ``llvm.assume`` intrinsic in order to preserve the instructions 24010only used to form the intrinsic's input argument. This might prove undesirable 24011if the extra information provided by the ``llvm.assume`` intrinsic does not cause 24012sufficient overall improvement in code quality. For this reason, 24013``llvm.assume`` should not be used to document basic mathematical invariants 24014that the optimizer can otherwise deduce or facts that are of little use to the 24015optimizer. 24016 24017.. _int_ssa_copy: 24018 24019'``llvm.ssa.copy``' Intrinsic 24020^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24021 24022Syntax: 24023""""""" 24024 24025:: 24026 24027 declare type @llvm.ssa.copy(type %operand) returned(1) readnone 24028 24029Arguments: 24030"""""""""" 24031 24032The first argument is an operand which is used as the returned value. 24033 24034Overview: 24035"""""""""" 24036 24037The ``llvm.ssa.copy`` intrinsic can be used to attach information to 24038operations by copying them and giving them new names. For example, 24039the PredicateInfo utility uses it to build Extended SSA form, and 24040attach various forms of information to operands that dominate specific 24041uses. It is not meant for general use, only for building temporary 24042renaming forms that require value splits at certain points. 24043 24044.. _type.test: 24045 24046'``llvm.type.test``' Intrinsic 24047^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24048 24049Syntax: 24050""""""" 24051 24052:: 24053 24054 declare i1 @llvm.type.test(i8* %ptr, metadata %type) nounwind readnone 24055 24056 24057Arguments: 24058"""""""""" 24059 24060The first argument is a pointer to be tested. The second argument is a 24061metadata object representing a :doc:`type identifier <TypeMetadata>`. 24062 24063Overview: 24064""""""""" 24065 24066The ``llvm.type.test`` intrinsic tests whether the given pointer is associated 24067with the given type identifier. 24068 24069.. _type.checked.load: 24070 24071'``llvm.type.checked.load``' Intrinsic 24072^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24073 24074Syntax: 24075""""""" 24076 24077:: 24078 24079 declare {i8*, i1} @llvm.type.checked.load(i8* %ptr, i32 %offset, metadata %type) argmemonly nounwind readonly 24080 24081 24082Arguments: 24083"""""""""" 24084 24085The first argument is a pointer from which to load a function pointer. The 24086second argument is the byte offset from which to load the function pointer. The 24087third argument is a metadata object representing a :doc:`type identifier 24088<TypeMetadata>`. 24089 24090Overview: 24091""""""""" 24092 24093The ``llvm.type.checked.load`` intrinsic safely loads a function pointer from a 24094virtual table pointer using type metadata. This intrinsic is used to implement 24095control flow integrity in conjunction with virtual call optimization. The 24096virtual call optimization pass will optimize away ``llvm.type.checked.load`` 24097intrinsics associated with devirtualized calls, thereby removing the type 24098check in cases where it is not needed to enforce the control flow integrity 24099constraint. 24100 24101If the given pointer is associated with a type metadata identifier, this 24102function returns true as the second element of its return value. (Note that 24103the function may also return true if the given pointer is not associated 24104with a type metadata identifier.) If the function's return value's second 24105element is true, the following rules apply to the first element: 24106 24107- If the given pointer is associated with the given type metadata identifier, 24108 it is the function pointer loaded from the given byte offset from the given 24109 pointer. 24110 24111- If the given pointer is not associated with the given type metadata 24112 identifier, it is one of the following (the choice of which is unspecified): 24113 24114 1. The function pointer that would have been loaded from an arbitrarily chosen 24115 (through an unspecified mechanism) pointer associated with the type 24116 metadata. 24117 24118 2. If the function has a non-void return type, a pointer to a function that 24119 returns an unspecified value without causing side effects. 24120 24121If the function's return value's second element is false, the value of the 24122first element is undefined. 24123 24124 24125'``llvm.arithmetic.fence``' Intrinsic 24126^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24127 24128Syntax: 24129""""""" 24130 24131:: 24132 24133 declare <type> 24134 @llvm.arithmetic.fence(<type> <op>) 24135 24136Overview: 24137""""""""" 24138 24139The purpose of the ``llvm.arithmetic.fence`` intrinsic 24140is to prevent the optimizer from performing fast-math optimizations, 24141particularly reassociation, 24142between the argument and the expression that contains the argument. 24143It can be used to preserve the parentheses in the source language. 24144 24145Arguments: 24146"""""""""" 24147 24148The ``llvm.arithmetic.fence`` intrinsic takes only one argument. 24149The argument and the return value are floating-point numbers, 24150or vector floating-point numbers, of the same type. 24151 24152Semantics: 24153"""""""""" 24154 24155This intrinsic returns the value of its operand. The optimizer can optimize 24156the argument, but the optimizer cannot hoist any component of the operand 24157to the containing context, and the optimizer cannot move the calculation of 24158any expression in the containing context into the operand. 24159 24160 24161'``llvm.donothing``' Intrinsic 24162^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24163 24164Syntax: 24165""""""" 24166 24167:: 24168 24169 declare void @llvm.donothing() nounwind readnone 24170 24171Overview: 24172""""""""" 24173 24174The ``llvm.donothing`` intrinsic doesn't perform any operation. It's one of only 24175three intrinsics (besides ``llvm.experimental.patchpoint`` and 24176``llvm.experimental.gc.statepoint``) that can be called with an invoke 24177instruction. 24178 24179Arguments: 24180"""""""""" 24181 24182None. 24183 24184Semantics: 24185"""""""""" 24186 24187This intrinsic does nothing, and it's removed by optimizers and ignored 24188by codegen. 24189 24190'``llvm.experimental.deoptimize``' Intrinsic 24191^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24192 24193Syntax: 24194""""""" 24195 24196:: 24197 24198 declare type @llvm.experimental.deoptimize(...) [ "deopt"(...) ] 24199 24200Overview: 24201""""""""" 24202 24203This intrinsic, together with :ref:`deoptimization operand bundles 24204<deopt_opbundles>`, allow frontends to express transfer of control and 24205frame-local state from the currently executing (typically more specialized, 24206hence faster) version of a function into another (typically more generic, hence 24207slower) version. 24208 24209In languages with a fully integrated managed runtime like Java and JavaScript 24210this intrinsic can be used to implement "uncommon trap" or "side exit" like 24211functionality. In unmanaged languages like C and C++, this intrinsic can be 24212used to represent the slow paths of specialized functions. 24213 24214 24215Arguments: 24216"""""""""" 24217 24218The intrinsic takes an arbitrary number of arguments, whose meaning is 24219decided by the :ref:`lowering strategy<deoptimize_lowering>`. 24220 24221Semantics: 24222"""""""""" 24223 24224The ``@llvm.experimental.deoptimize`` intrinsic executes an attached 24225deoptimization continuation (denoted using a :ref:`deoptimization 24226operand bundle <deopt_opbundles>`) and returns the value returned by 24227the deoptimization continuation. Defining the semantic properties of 24228the continuation itself is out of scope of the language reference -- 24229as far as LLVM is concerned, the deoptimization continuation can 24230invoke arbitrary side effects, including reading from and writing to 24231the entire heap. 24232 24233Deoptimization continuations expressed using ``"deopt"`` operand bundles always 24234continue execution to the end of the physical frame containing them, so all 24235calls to ``@llvm.experimental.deoptimize`` must be in "tail position": 24236 24237 - ``@llvm.experimental.deoptimize`` cannot be invoked. 24238 - The call must immediately precede a :ref:`ret <i_ret>` instruction. 24239 - The ``ret`` instruction must return the value produced by the 24240 ``@llvm.experimental.deoptimize`` call if there is one, or void. 24241 24242Note that the above restrictions imply that the return type for a call to 24243``@llvm.experimental.deoptimize`` will match the return type of its immediate 24244caller. 24245 24246The inliner composes the ``"deopt"`` continuations of the caller into the 24247``"deopt"`` continuations present in the inlinee, and also updates calls to this 24248intrinsic to return directly from the frame of the function it inlined into. 24249 24250All declarations of ``@llvm.experimental.deoptimize`` must share the 24251same calling convention. 24252 24253.. _deoptimize_lowering: 24254 24255Lowering: 24256""""""""" 24257 24258Calls to ``@llvm.experimental.deoptimize`` are lowered to calls to the 24259symbol ``__llvm_deoptimize`` (it is the frontend's responsibility to 24260ensure that this symbol is defined). The call arguments to 24261``@llvm.experimental.deoptimize`` are lowered as if they were formal 24262arguments of the specified types, and not as varargs. 24263 24264 24265'``llvm.experimental.guard``' Intrinsic 24266^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24267 24268Syntax: 24269""""""" 24270 24271:: 24272 24273 declare void @llvm.experimental.guard(i1, ...) [ "deopt"(...) ] 24274 24275Overview: 24276""""""""" 24277 24278This intrinsic, together with :ref:`deoptimization operand bundles 24279<deopt_opbundles>`, allows frontends to express guards or checks on 24280optimistic assumptions made during compilation. The semantics of 24281``@llvm.experimental.guard`` is defined in terms of 24282``@llvm.experimental.deoptimize`` -- its body is defined to be 24283equivalent to: 24284 24285.. code-block:: text 24286 24287 define void @llvm.experimental.guard(i1 %pred, <args...>) { 24288 %realPred = and i1 %pred, undef 24289 br i1 %realPred, label %continue, label %leave [, !make.implicit !{}] 24290 24291 leave: 24292 call void @llvm.experimental.deoptimize(<args...>) [ "deopt"() ] 24293 ret void 24294 24295 continue: 24296 ret void 24297 } 24298 24299 24300with the optional ``[, !make.implicit !{}]`` present if and only if it 24301is present on the call site. For more details on ``!make.implicit``, 24302see :doc:`FaultMaps`. 24303 24304In words, ``@llvm.experimental.guard`` executes the attached 24305``"deopt"`` continuation if (but **not** only if) its first argument 24306is ``false``. Since the optimizer is allowed to replace the ``undef`` 24307with an arbitrary value, it can optimize guard to fail "spuriously", 24308i.e. without the original condition being false (hence the "not only 24309if"); and this allows for "check widening" type optimizations. 24310 24311``@llvm.experimental.guard`` cannot be invoked. 24312 24313After ``@llvm.experimental.guard`` was first added, a more general 24314formulation was found in ``@llvm.experimental.widenable.condition``. 24315Support for ``@llvm.experimental.guard`` is slowly being rephrased in 24316terms of this alternate. 24317 24318'``llvm.experimental.widenable.condition``' Intrinsic 24319^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24320 24321Syntax: 24322""""""" 24323 24324:: 24325 24326 declare i1 @llvm.experimental.widenable.condition() 24327 24328Overview: 24329""""""""" 24330 24331This intrinsic represents a "widenable condition" which is 24332boolean expressions with the following property: whether this 24333expression is `true` or `false`, the program is correct and 24334well-defined. 24335 24336Together with :ref:`deoptimization operand bundles <deopt_opbundles>`, 24337``@llvm.experimental.widenable.condition`` allows frontends to 24338express guards or checks on optimistic assumptions made during 24339compilation and represent them as branch instructions on special 24340conditions. 24341 24342While this may appear similar in semantics to `undef`, it is very 24343different in that an invocation produces a particular, singular 24344value. It is also intended to be lowered late, and remain available 24345for specific optimizations and transforms that can benefit from its 24346special properties. 24347 24348Arguments: 24349"""""""""" 24350 24351None. 24352 24353Semantics: 24354"""""""""" 24355 24356The intrinsic ``@llvm.experimental.widenable.condition()`` 24357returns either `true` or `false`. For each evaluation of a call 24358to this intrinsic, the program must be valid and correct both if 24359it returns `true` and if it returns `false`. This allows 24360transformation passes to replace evaluations of this intrinsic 24361with either value whenever one is beneficial. 24362 24363When used in a branch condition, it allows us to choose between 24364two alternative correct solutions for the same problem, like 24365in example below: 24366 24367.. code-block:: text 24368 24369 %cond = call i1 @llvm.experimental.widenable.condition() 24370 br i1 %cond, label %solution_1, label %solution_2 24371 24372 label %fast_path: 24373 ; Apply memory-consuming but fast solution for a task. 24374 24375 label %slow_path: 24376 ; Cheap in memory but slow solution. 24377 24378Whether the result of intrinsic's call is `true` or `false`, 24379it should be correct to pick either solution. We can switch 24380between them by replacing the result of 24381``@llvm.experimental.widenable.condition`` with different 24382`i1` expressions. 24383 24384This is how it can be used to represent guards as widenable branches: 24385 24386.. code-block:: text 24387 24388 block: 24389 ; Unguarded instructions 24390 call void @llvm.experimental.guard(i1 %cond, <args...>) ["deopt"(<deopt_args...>)] 24391 ; Guarded instructions 24392 24393Can be expressed in an alternative equivalent form of explicit branch using 24394``@llvm.experimental.widenable.condition``: 24395 24396.. code-block:: text 24397 24398 block: 24399 ; Unguarded instructions 24400 %widenable_condition = call i1 @llvm.experimental.widenable.condition() 24401 %guard_condition = and i1 %cond, %widenable_condition 24402 br i1 %guard_condition, label %guarded, label %deopt 24403 24404 guarded: 24405 ; Guarded instructions 24406 24407 deopt: 24408 call type @llvm.experimental.deoptimize(<args...>) [ "deopt"(<deopt_args...>) ] 24409 24410So the block `guarded` is only reachable when `%cond` is `true`, 24411and it should be valid to go to the block `deopt` whenever `%cond` 24412is `true` or `false`. 24413 24414``@llvm.experimental.widenable.condition`` will never throw, thus 24415it cannot be invoked. 24416 24417Guard widening: 24418""""""""""""""" 24419 24420When ``@llvm.experimental.widenable.condition()`` is used in 24421condition of a guard represented as explicit branch, it is 24422legal to widen the guard's condition with any additional 24423conditions. 24424 24425Guard widening looks like replacement of 24426 24427.. code-block:: text 24428 24429 %widenable_cond = call i1 @llvm.experimental.widenable.condition() 24430 %guard_cond = and i1 %cond, %widenable_cond 24431 br i1 %guard_cond, label %guarded, label %deopt 24432 24433with 24434 24435.. code-block:: text 24436 24437 %widenable_cond = call i1 @llvm.experimental.widenable.condition() 24438 %new_cond = and i1 %any_other_cond, %widenable_cond 24439 %new_guard_cond = and i1 %cond, %new_cond 24440 br i1 %new_guard_cond, label %guarded, label %deopt 24441 24442for this branch. Here `%any_other_cond` is an arbitrarily chosen 24443well-defined `i1` value. By making guard widening, we may 24444impose stricter conditions on `guarded` block and bail to the 24445deopt when the new condition is not met. 24446 24447Lowering: 24448""""""""" 24449 24450Default lowering strategy is replacing the result of 24451call of ``@llvm.experimental.widenable.condition`` with 24452constant `true`. However it is always correct to replace 24453it with any other `i1` value. Any pass can 24454freely do it if it can benefit from non-default lowering. 24455 24456 24457'``llvm.load.relative``' Intrinsic 24458^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24459 24460Syntax: 24461""""""" 24462 24463:: 24464 24465 declare i8* @llvm.load.relative.iN(i8* %ptr, iN %offset) argmemonly nounwind readonly 24466 24467Overview: 24468""""""""" 24469 24470This intrinsic loads a 32-bit value from the address ``%ptr + %offset``, 24471adds ``%ptr`` to that value and returns it. The constant folder specifically 24472recognizes the form of this intrinsic and the constant initializers it may 24473load from; if a loaded constant initializer is known to have the form 24474``i32 trunc(x - %ptr)``, the intrinsic call is folded to ``x``. 24475 24476LLVM provides that the calculation of such a constant initializer will 24477not overflow at link time under the medium code model if ``x`` is an 24478``unnamed_addr`` function. However, it does not provide this guarantee for 24479a constant initializer folded into a function body. This intrinsic can be 24480used to avoid the possibility of overflows when loading from such a constant. 24481 24482.. _llvm_sideeffect: 24483 24484'``llvm.sideeffect``' Intrinsic 24485^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24486 24487Syntax: 24488""""""" 24489 24490:: 24491 24492 declare void @llvm.sideeffect() inaccessiblememonly nounwind willreturn 24493 24494Overview: 24495""""""""" 24496 24497The ``llvm.sideeffect`` intrinsic doesn't perform any operation. Optimizers 24498treat it as having side effects, so it can be inserted into a loop to 24499indicate that the loop shouldn't be assumed to terminate (which could 24500potentially lead to the loop being optimized away entirely), even if it's 24501an infinite loop with no other side effects. 24502 24503Arguments: 24504"""""""""" 24505 24506None. 24507 24508Semantics: 24509"""""""""" 24510 24511This intrinsic actually does nothing, but optimizers must assume that it 24512has externally observable side effects. 24513 24514'``llvm.is.constant.*``' Intrinsic 24515^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24516 24517Syntax: 24518""""""" 24519 24520This is an overloaded intrinsic. You can use llvm.is.constant with any argument type. 24521 24522:: 24523 24524 declare i1 @llvm.is.constant.i32(i32 %operand) nounwind readnone 24525 declare i1 @llvm.is.constant.f32(float %operand) nounwind readnone 24526 declare i1 @llvm.is.constant.TYPENAME(TYPE %operand) nounwind readnone 24527 24528Overview: 24529""""""""" 24530 24531The '``llvm.is.constant``' intrinsic will return true if the argument 24532is known to be a manifest compile-time constant. It is guaranteed to 24533fold to either true or false before generating machine code. 24534 24535Semantics: 24536"""""""""" 24537 24538This intrinsic generates no code. If its argument is known to be a 24539manifest compile-time constant value, then the intrinsic will be 24540converted to a constant true value. Otherwise, it will be converted to 24541a constant false value. 24542 24543In particular, note that if the argument is a constant expression 24544which refers to a global (the address of which _is_ a constant, but 24545not manifest during the compile), then the intrinsic evaluates to 24546false. 24547 24548The result also intentionally depends on the result of optimization 24549passes -- e.g., the result can change depending on whether a 24550function gets inlined or not. A function's parameters are 24551obviously not constant. However, a call like 24552``llvm.is.constant.i32(i32 %param)`` *can* return true after the 24553function is inlined, if the value passed to the function parameter was 24554a constant. 24555 24556On the other hand, if constant folding is not run, it will never 24557evaluate to true, even in simple cases. 24558 24559.. _int_ptrmask: 24560 24561'``llvm.ptrmask``' Intrinsic 24562^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24563 24564Syntax: 24565""""""" 24566 24567:: 24568 24569 declare ptrty llvm.ptrmask(ptrty %ptr, intty %mask) readnone speculatable 24570 24571Arguments: 24572"""""""""" 24573 24574The first argument is a pointer. The second argument is an integer. 24575 24576Overview: 24577"""""""""" 24578 24579The ``llvm.ptrmask`` intrinsic masks out bits of the pointer according to a mask. 24580This allows stripping data from tagged pointers without converting them to an 24581integer (ptrtoint/inttoptr). As a consequence, we can preserve more information 24582to facilitate alias analysis and underlying-object detection. 24583 24584Semantics: 24585"""""""""" 24586 24587The result of ``ptrmask(ptr, mask)`` is equivalent to 24588``getelementptr ptr, (ptrtoint(ptr) & mask) - ptrtoint(ptr)``. Both the returned 24589pointer and the first argument are based on the same underlying object (for more 24590information on the *based on* terminology see 24591:ref:`the pointer aliasing rules <pointeraliasing>`). If the bitwidth of the 24592mask argument does not match the pointer size of the target, the mask is 24593zero-extended or truncated accordingly. 24594 24595.. _int_vscale: 24596 24597'``llvm.vscale``' Intrinsic 24598^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24599 24600Syntax: 24601""""""" 24602 24603:: 24604 24605 declare i32 llvm.vscale.i32() 24606 declare i64 llvm.vscale.i64() 24607 24608Overview: 24609""""""""" 24610 24611The ``llvm.vscale`` intrinsic returns the value for ``vscale`` in scalable 24612vectors such as ``<vscale x 16 x i8>``. 24613 24614Semantics: 24615"""""""""" 24616 24617``vscale`` is a positive value that is constant throughout program 24618execution, but is unknown at compile time. 24619If the result value does not fit in the result type, then the result is 24620a :ref:`poison value <poisonvalues>`. 24621 24622 24623Stack Map Intrinsics 24624-------------------- 24625 24626LLVM provides experimental intrinsics to support runtime patching 24627mechanisms commonly desired in dynamic language JITs. These intrinsics 24628are described in :doc:`StackMaps`. 24629 24630Element Wise Atomic Memory Intrinsics 24631------------------------------------- 24632 24633These intrinsics are similar to the standard library memory intrinsics except 24634that they perform memory transfer as a sequence of atomic memory accesses. 24635 24636.. _int_memcpy_element_unordered_atomic: 24637 24638'``llvm.memcpy.element.unordered.atomic``' Intrinsic 24639^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24640 24641Syntax: 24642""""""" 24643 24644This is an overloaded intrinsic. You can use ``llvm.memcpy.element.unordered.atomic`` on 24645any integer bit width and for different address spaces. Not all targets 24646support all bit widths however. 24647 24648:: 24649 24650 declare void @llvm.memcpy.element.unordered.atomic.p0i8.p0i8.i32(i8* <dest>, 24651 i8* <src>, 24652 i32 <len>, 24653 i32 <element_size>) 24654 declare void @llvm.memcpy.element.unordered.atomic.p0i8.p0i8.i64(i8* <dest>, 24655 i8* <src>, 24656 i64 <len>, 24657 i32 <element_size>) 24658 24659Overview: 24660""""""""" 24661 24662The '``llvm.memcpy.element.unordered.atomic.*``' intrinsic is a specialization of the 24663'``llvm.memcpy.*``' intrinsic. It differs in that the ``dest`` and ``src`` are treated 24664as arrays with elements that are exactly ``element_size`` bytes, and the copy between 24665buffers uses a sequence of :ref:`unordered atomic <ordering>` load/store operations 24666that are a positive integer multiple of the ``element_size`` in size. 24667 24668Arguments: 24669"""""""""" 24670 24671The first three arguments are the same as they are in the :ref:`@llvm.memcpy <int_memcpy>` 24672intrinsic, with the added constraint that ``len`` is required to be a positive integer 24673multiple of the ``element_size``. If ``len`` is not a positive integer multiple of 24674``element_size``, then the behaviour of the intrinsic is undefined. 24675 24676``element_size`` must be a compile-time constant positive power of two no greater than 24677target-specific atomic access size limit. 24678 24679For each of the input pointers ``align`` parameter attribute must be specified. It 24680must be a power of two no less than the ``element_size``. Caller guarantees that 24681both the source and destination pointers are aligned to that boundary. 24682 24683Semantics: 24684"""""""""" 24685 24686The '``llvm.memcpy.element.unordered.atomic.*``' intrinsic copies ``len`` bytes of 24687memory from the source location to the destination location. These locations are not 24688allowed to overlap. The memory copy is performed as a sequence of load/store operations 24689where each access is guaranteed to be a multiple of ``element_size`` bytes wide and 24690aligned at an ``element_size`` boundary. 24691 24692The order of the copy is unspecified. The same value may be read from the source 24693buffer many times, but only one write is issued to the destination buffer per 24694element. It is well defined to have concurrent reads and writes to both source and 24695destination provided those reads and writes are unordered atomic when specified. 24696 24697This intrinsic does not provide any additional ordering guarantees over those 24698provided by a set of unordered loads from the source location and stores to the 24699destination. 24700 24701Lowering: 24702""""""""" 24703 24704In the most general case call to the '``llvm.memcpy.element.unordered.atomic.*``' is 24705lowered to a call to the symbol ``__llvm_memcpy_element_unordered_atomic_*``. Where '*' 24706is replaced with an actual element size. See :ref:`RewriteStatepointsForGC intrinsic 24707lowering <RewriteStatepointsForGC_intrinsic_lowering>` for details on GC specific 24708lowering. 24709 24710Optimizer is allowed to inline memory copy when it's profitable to do so. 24711 24712'``llvm.memmove.element.unordered.atomic``' Intrinsic 24713^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24714 24715Syntax: 24716""""""" 24717 24718This is an overloaded intrinsic. You can use 24719``llvm.memmove.element.unordered.atomic`` on any integer bit width and for 24720different address spaces. Not all targets support all bit widths however. 24721 24722:: 24723 24724 declare void @llvm.memmove.element.unordered.atomic.p0i8.p0i8.i32(i8* <dest>, 24725 i8* <src>, 24726 i32 <len>, 24727 i32 <element_size>) 24728 declare void @llvm.memmove.element.unordered.atomic.p0i8.p0i8.i64(i8* <dest>, 24729 i8* <src>, 24730 i64 <len>, 24731 i32 <element_size>) 24732 24733Overview: 24734""""""""" 24735 24736The '``llvm.memmove.element.unordered.atomic.*``' intrinsic is a specialization 24737of the '``llvm.memmove.*``' intrinsic. It differs in that the ``dest`` and 24738``src`` are treated as arrays with elements that are exactly ``element_size`` 24739bytes, and the copy between buffers uses a sequence of 24740:ref:`unordered atomic <ordering>` load/store operations that are a positive 24741integer multiple of the ``element_size`` in size. 24742 24743Arguments: 24744"""""""""" 24745 24746The first three arguments are the same as they are in the 24747:ref:`@llvm.memmove <int_memmove>` intrinsic, with the added constraint that 24748``len`` is required to be a positive integer multiple of the ``element_size``. 24749If ``len`` is not a positive integer multiple of ``element_size``, then the 24750behaviour of the intrinsic is undefined. 24751 24752``element_size`` must be a compile-time constant positive power of two no 24753greater than a target-specific atomic access size limit. 24754 24755For each of the input pointers the ``align`` parameter attribute must be 24756specified. It must be a power of two no less than the ``element_size``. Caller 24757guarantees that both the source and destination pointers are aligned to that 24758boundary. 24759 24760Semantics: 24761"""""""""" 24762 24763The '``llvm.memmove.element.unordered.atomic.*``' intrinsic copies ``len`` bytes 24764of memory from the source location to the destination location. These locations 24765are allowed to overlap. The memory copy is performed as a sequence of load/store 24766operations where each access is guaranteed to be a multiple of ``element_size`` 24767bytes wide and aligned at an ``element_size`` boundary. 24768 24769The order of the copy is unspecified. The same value may be read from the source 24770buffer many times, but only one write is issued to the destination buffer per 24771element. It is well defined to have concurrent reads and writes to both source 24772and destination provided those reads and writes are unordered atomic when 24773specified. 24774 24775This intrinsic does not provide any additional ordering guarantees over those 24776provided by a set of unordered loads from the source location and stores to the 24777destination. 24778 24779Lowering: 24780""""""""" 24781 24782In the most general case call to the 24783'``llvm.memmove.element.unordered.atomic.*``' is lowered to a call to the symbol 24784``__llvm_memmove_element_unordered_atomic_*``. Where '*' is replaced with an 24785actual element size. See :ref:`RewriteStatepointsForGC intrinsic lowering 24786<RewriteStatepointsForGC_intrinsic_lowering>` for details on GC specific 24787lowering. 24788 24789The optimizer is allowed to inline the memory copy when it's profitable to do so. 24790 24791.. _int_memset_element_unordered_atomic: 24792 24793'``llvm.memset.element.unordered.atomic``' Intrinsic 24794^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24795 24796Syntax: 24797""""""" 24798 24799This is an overloaded intrinsic. You can use ``llvm.memset.element.unordered.atomic`` on 24800any integer bit width and for different address spaces. Not all targets 24801support all bit widths however. 24802 24803:: 24804 24805 declare void @llvm.memset.element.unordered.atomic.p0i8.i32(i8* <dest>, 24806 i8 <value>, 24807 i32 <len>, 24808 i32 <element_size>) 24809 declare void @llvm.memset.element.unordered.atomic.p0i8.i64(i8* <dest>, 24810 i8 <value>, 24811 i64 <len>, 24812 i32 <element_size>) 24813 24814Overview: 24815""""""""" 24816 24817The '``llvm.memset.element.unordered.atomic.*``' intrinsic is a specialization of the 24818'``llvm.memset.*``' intrinsic. It differs in that the ``dest`` is treated as an array 24819with elements that are exactly ``element_size`` bytes, and the assignment to that array 24820uses uses a sequence of :ref:`unordered atomic <ordering>` store operations 24821that are a positive integer multiple of the ``element_size`` in size. 24822 24823Arguments: 24824"""""""""" 24825 24826The first three arguments are the same as they are in the :ref:`@llvm.memset <int_memset>` 24827intrinsic, with the added constraint that ``len`` is required to be a positive integer 24828multiple of the ``element_size``. If ``len`` is not a positive integer multiple of 24829``element_size``, then the behaviour of the intrinsic is undefined. 24830 24831``element_size`` must be a compile-time constant positive power of two no greater than 24832target-specific atomic access size limit. 24833 24834The ``dest`` input pointer must have the ``align`` parameter attribute specified. It 24835must be a power of two no less than the ``element_size``. Caller guarantees that 24836the destination pointer is aligned to that boundary. 24837 24838Semantics: 24839"""""""""" 24840 24841The '``llvm.memset.element.unordered.atomic.*``' intrinsic sets the ``len`` bytes of 24842memory starting at the destination location to the given ``value``. The memory is 24843set with a sequence of store operations where each access is guaranteed to be a 24844multiple of ``element_size`` bytes wide and aligned at an ``element_size`` boundary. 24845 24846The order of the assignment is unspecified. Only one write is issued to the 24847destination buffer per element. It is well defined to have concurrent reads and 24848writes to the destination provided those reads and writes are unordered atomic 24849when specified. 24850 24851This intrinsic does not provide any additional ordering guarantees over those 24852provided by a set of unordered stores to the destination. 24853 24854Lowering: 24855""""""""" 24856 24857In the most general case call to the '``llvm.memset.element.unordered.atomic.*``' is 24858lowered to a call to the symbol ``__llvm_memset_element_unordered_atomic_*``. Where '*' 24859is replaced with an actual element size. 24860 24861The optimizer is allowed to inline the memory assignment when it's profitable to do so. 24862 24863Objective-C ARC Runtime Intrinsics 24864---------------------------------- 24865 24866LLVM provides intrinsics that lower to Objective-C ARC runtime entry points. 24867LLVM is aware of the semantics of these functions, and optimizes based on that 24868knowledge. You can read more about the details of Objective-C ARC `here 24869<https://clang.llvm.org/docs/AutomaticReferenceCounting.html>`_. 24870 24871'``llvm.objc.autorelease``' Intrinsic 24872^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24873 24874Syntax: 24875""""""" 24876:: 24877 24878 declare i8* @llvm.objc.autorelease(i8*) 24879 24880Lowering: 24881""""""""" 24882 24883Lowers to a call to `objc_autorelease <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autorelease>`_. 24884 24885'``llvm.objc.autoreleasePoolPop``' Intrinsic 24886^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24887 24888Syntax: 24889""""""" 24890:: 24891 24892 declare void @llvm.objc.autoreleasePoolPop(i8*) 24893 24894Lowering: 24895""""""""" 24896 24897Lowers to a call to `objc_autoreleasePoolPop <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-autoreleasepoolpop-void-pool>`_. 24898 24899'``llvm.objc.autoreleasePoolPush``' Intrinsic 24900^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24901 24902Syntax: 24903""""""" 24904:: 24905 24906 declare i8* @llvm.objc.autoreleasePoolPush() 24907 24908Lowering: 24909""""""""" 24910 24911Lowers to a call to `objc_autoreleasePoolPush <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-autoreleasepoolpush-void>`_. 24912 24913'``llvm.objc.autoreleaseReturnValue``' Intrinsic 24914^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24915 24916Syntax: 24917""""""" 24918:: 24919 24920 declare i8* @llvm.objc.autoreleaseReturnValue(i8*) 24921 24922Lowering: 24923""""""""" 24924 24925Lowers to a call to `objc_autoreleaseReturnValue <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-autoreleasereturnvalue>`_. 24926 24927'``llvm.objc.copyWeak``' Intrinsic 24928^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24929 24930Syntax: 24931""""""" 24932:: 24933 24934 declare void @llvm.objc.copyWeak(i8**, i8**) 24935 24936Lowering: 24937""""""""" 24938 24939Lowers to a call to `objc_copyWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-copyweak-id-dest-id-src>`_. 24940 24941'``llvm.objc.destroyWeak``' Intrinsic 24942^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24943 24944Syntax: 24945""""""" 24946:: 24947 24948 declare void @llvm.objc.destroyWeak(i8**) 24949 24950Lowering: 24951""""""""" 24952 24953Lowers to a call to `objc_destroyWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-destroyweak-id-object>`_. 24954 24955'``llvm.objc.initWeak``' Intrinsic 24956^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24957 24958Syntax: 24959""""""" 24960:: 24961 24962 declare i8* @llvm.objc.initWeak(i8**, i8*) 24963 24964Lowering: 24965""""""""" 24966 24967Lowers to a call to `objc_initWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-initweak>`_. 24968 24969'``llvm.objc.loadWeak``' Intrinsic 24970^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24971 24972Syntax: 24973""""""" 24974:: 24975 24976 declare i8* @llvm.objc.loadWeak(i8**) 24977 24978Lowering: 24979""""""""" 24980 24981Lowers to a call to `objc_loadWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-loadweak>`_. 24982 24983'``llvm.objc.loadWeakRetained``' Intrinsic 24984^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24985 24986Syntax: 24987""""""" 24988:: 24989 24990 declare i8* @llvm.objc.loadWeakRetained(i8**) 24991 24992Lowering: 24993""""""""" 24994 24995Lowers to a call to `objc_loadWeakRetained <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-loadweakretained>`_. 24996 24997'``llvm.objc.moveWeak``' Intrinsic 24998^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 24999 25000Syntax: 25001""""""" 25002:: 25003 25004 declare void @llvm.objc.moveWeak(i8**, i8**) 25005 25006Lowering: 25007""""""""" 25008 25009Lowers to a call to `objc_moveWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-moveweak-id-dest-id-src>`_. 25010 25011'``llvm.objc.release``' Intrinsic 25012^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25013 25014Syntax: 25015""""""" 25016:: 25017 25018 declare void @llvm.objc.release(i8*) 25019 25020Lowering: 25021""""""""" 25022 25023Lowers to a call to `objc_release <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-release-id-value>`_. 25024 25025'``llvm.objc.retain``' Intrinsic 25026^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25027 25028Syntax: 25029""""""" 25030:: 25031 25032 declare i8* @llvm.objc.retain(i8*) 25033 25034Lowering: 25035""""""""" 25036 25037Lowers to a call to `objc_retain <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-retain>`_. 25038 25039'``llvm.objc.retainAutorelease``' Intrinsic 25040^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25041 25042Syntax: 25043""""""" 25044:: 25045 25046 declare i8* @llvm.objc.retainAutorelease(i8*) 25047 25048Lowering: 25049""""""""" 25050 25051Lowers to a call to `objc_retainAutorelease <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-retainautorelease>`_. 25052 25053'``llvm.objc.retainAutoreleaseReturnValue``' Intrinsic 25054^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25055 25056Syntax: 25057""""""" 25058:: 25059 25060 declare i8* @llvm.objc.retainAutoreleaseReturnValue(i8*) 25061 25062Lowering: 25063""""""""" 25064 25065Lowers to a call to `objc_retainAutoreleaseReturnValue <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-retainautoreleasereturnvalue>`_. 25066 25067'``llvm.objc.retainAutoreleasedReturnValue``' Intrinsic 25068^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25069 25070Syntax: 25071""""""" 25072:: 25073 25074 declare i8* @llvm.objc.retainAutoreleasedReturnValue(i8*) 25075 25076Lowering: 25077""""""""" 25078 25079Lowers to a call to `objc_retainAutoreleasedReturnValue <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-retainautoreleasedreturnvalue>`_. 25080 25081'``llvm.objc.retainBlock``' Intrinsic 25082^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25083 25084Syntax: 25085""""""" 25086:: 25087 25088 declare i8* @llvm.objc.retainBlock(i8*) 25089 25090Lowering: 25091""""""""" 25092 25093Lowers to a call to `objc_retainBlock <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-retainblock>`_. 25094 25095'``llvm.objc.storeStrong``' Intrinsic 25096^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25097 25098Syntax: 25099""""""" 25100:: 25101 25102 declare void @llvm.objc.storeStrong(i8**, i8*) 25103 25104Lowering: 25105""""""""" 25106 25107Lowers to a call to `objc_storeStrong <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#void-objc-storestrong-id-object-id-value>`_. 25108 25109'``llvm.objc.storeWeak``' Intrinsic 25110^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25111 25112Syntax: 25113""""""" 25114:: 25115 25116 declare i8* @llvm.objc.storeWeak(i8**, i8*) 25117 25118Lowering: 25119""""""""" 25120 25121Lowers to a call to `objc_storeWeak <https://clang.llvm.org/docs/AutomaticReferenceCounting.html#arc-runtime-objc-storeweak>`_. 25122 25123Preserving Debug Information Intrinsics 25124^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25125 25126These intrinsics are used to carry certain debuginfo together with 25127IR-level operations. For example, it may be desirable to 25128know the structure/union name and the original user-level field 25129indices. Such information got lost in IR GetElementPtr instruction 25130since the IR types are different from debugInfo types and unions 25131are converted to structs in IR. 25132 25133'``llvm.preserve.array.access.index``' Intrinsic 25134^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25135 25136Syntax: 25137""""""" 25138:: 25139 25140 declare <ret_type> 25141 @llvm.preserve.array.access.index.p0s_union.anons.p0a10s_union.anons(<type> base, 25142 i32 dim, 25143 i32 index) 25144 25145Overview: 25146""""""""" 25147 25148The '``llvm.preserve.array.access.index``' intrinsic returns the getelementptr address 25149based on array base ``base``, array dimension ``dim`` and the last access index ``index`` 25150into the array. The return type ``ret_type`` is a pointer type to the array element. 25151The array ``dim`` and ``index`` are preserved which is more robust than 25152getelementptr instruction which may be subject to compiler transformation. 25153The ``llvm.preserve.access.index`` type of metadata is attached to this call instruction 25154to provide array or pointer debuginfo type. 25155The metadata is a ``DICompositeType`` or ``DIDerivedType`` representing the 25156debuginfo version of ``type``. 25157 25158Arguments: 25159"""""""""" 25160 25161The ``base`` is the array base address. The ``dim`` is the array dimension. 25162The ``base`` is a pointer if ``dim`` equals 0. 25163The ``index`` is the last access index into the array or pointer. 25164 25165The ``base`` argument must be annotated with an :ref:`elementtype 25166<attr_elementtype>` attribute at the call-site. This attribute specifies the 25167getelementptr element type. 25168 25169Semantics: 25170"""""""""" 25171 25172The '``llvm.preserve.array.access.index``' intrinsic produces the same result 25173as a getelementptr with base ``base`` and access operands ``{dim's 0's, index}``. 25174 25175'``llvm.preserve.union.access.index``' Intrinsic 25176^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25177 25178Syntax: 25179""""""" 25180:: 25181 25182 declare <type> 25183 @llvm.preserve.union.access.index.p0s_union.anons.p0s_union.anons(<type> base, 25184 i32 di_index) 25185 25186Overview: 25187""""""""" 25188 25189The '``llvm.preserve.union.access.index``' intrinsic carries the debuginfo field index 25190``di_index`` and returns the ``base`` address. 25191The ``llvm.preserve.access.index`` type of metadata is attached to this call instruction 25192to provide union debuginfo type. 25193The metadata is a ``DICompositeType`` representing the debuginfo version of ``type``. 25194The return type ``type`` is the same as the ``base`` type. 25195 25196Arguments: 25197"""""""""" 25198 25199The ``base`` is the union base address. The ``di_index`` is the field index in debuginfo. 25200 25201Semantics: 25202"""""""""" 25203 25204The '``llvm.preserve.union.access.index``' intrinsic returns the ``base`` address. 25205 25206'``llvm.preserve.struct.access.index``' Intrinsic 25207^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25208 25209Syntax: 25210""""""" 25211:: 25212 25213 declare <ret_type> 25214 @llvm.preserve.struct.access.index.p0i8.p0s_struct.anon.0s(<type> base, 25215 i32 gep_index, 25216 i32 di_index) 25217 25218Overview: 25219""""""""" 25220 25221The '``llvm.preserve.struct.access.index``' intrinsic returns the getelementptr address 25222based on struct base ``base`` and IR struct member index ``gep_index``. 25223The ``llvm.preserve.access.index`` type of metadata is attached to this call instruction 25224to provide struct debuginfo type. 25225The metadata is a ``DICompositeType`` representing the debuginfo version of ``type``. 25226The return type ``ret_type`` is a pointer type to the structure member. 25227 25228Arguments: 25229"""""""""" 25230 25231The ``base`` is the structure base address. The ``gep_index`` is the struct member index 25232based on IR structures. The ``di_index`` is the struct member index based on debuginfo. 25233 25234The ``base`` argument must be annotated with an :ref:`elementtype 25235<attr_elementtype>` attribute at the call-site. This attribute specifies the 25236getelementptr element type. 25237 25238Semantics: 25239"""""""""" 25240 25241The '``llvm.preserve.struct.access.index``' intrinsic produces the same result 25242as a getelementptr with base ``base`` and access operands ``{0, gep_index}``. 25243 25244'``llvm.fptrunc.round``' Intrinsic 25245^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 25246 25247Syntax: 25248""""""" 25249 25250:: 25251 25252 declare <ty2> 25253 @llvm.fptrunc.round(<type> <value>, metadata <rounding mode>) 25254 25255Overview: 25256""""""""" 25257 25258The '``llvm.fptrunc.round``' intrinsic truncates 25259:ref:`floating-point <t_floating>` ``value`` to type ``ty2`` 25260with a specified rounding mode. 25261 25262Arguments: 25263"""""""""" 25264 25265The '``llvm.fptrunc.round``' intrinsic takes a :ref:`floating-point 25266<t_floating>` value to cast and a :ref:`floating-point <t_floating>` type 25267to cast it to. This argument must be larger in size than the result. 25268 25269The second argument specifies the rounding mode as described in the constrained 25270intrinsics section. 25271For this intrinsic, the "round.dynamic" mode is not supported. 25272 25273Semantics: 25274"""""""""" 25275 25276The '``llvm.fptrunc.round``' intrinsic casts a ``value`` from a larger 25277:ref:`floating-point <t_floating>` type to a smaller :ref:`floating-point 25278<t_floating>` type. 25279This intrinsic is assumed to execute in the default :ref:`floating-point 25280environment <floatenv>` *except* for the rounding mode. 25281This intrinsic is not supported on all targets. Some targets may not support 25282all rounding modes. 25283