1FileCheck - Flexible pattern matching file verifier 2=================================================== 3 4.. program:: FileCheck 5 6SYNOPSIS 7-------- 8 9:program:`FileCheck` *match-filename* [*--check-prefix=XXX*] [*--strict-whitespace*] 10 11DESCRIPTION 12----------- 13 14:program:`FileCheck` reads two files (one from standard input, and one 15specified on the command line) and uses one to verify the other. This 16behavior is particularly useful for the testsuite, which wants to verify that 17the output of some tool (e.g. :program:`llc`) contains the expected information 18(for example, a movsd from esp or whatever is interesting). This is similar to 19using :program:`grep`, but it is optimized for matching multiple different 20inputs in one file in a specific order. 21 22The ``match-filename`` file specifies the file that contains the patterns to 23match. The file to verify is read from standard input unless the 24:option:`--input-file` option is used. 25 26OPTIONS 27------- 28 29Options are parsed from the environment variable ``FILECHECK_OPTS`` 30and from the command line. 31 32.. option:: -help 33 34 Print a summary of command line options. 35 36.. option:: --check-prefix prefix 37 38 FileCheck searches the contents of ``match-filename`` for patterns to 39 match. By default, these patterns are prefixed with "``CHECK:``". 40 If you'd like to use a different prefix (e.g. because the same input 41 file is checking multiple different tool or options), the 42 :option:`--check-prefix` argument allows you to specify (without the trailing 43 "``:``") one or more prefixes to match. Multiple prefixes are useful for tests 44 which might change for different run options, but most lines remain the same. 45 46 FileCheck does not permit duplicate prefixes, even if one is a check prefix 47 and one is a comment prefix (see :option:`--comment-prefixes` below). 48 49.. option:: --check-prefixes prefix1,prefix2,... 50 51 An alias of :option:`--check-prefix` that allows multiple prefixes to be 52 specified as a comma separated list. 53 54.. option:: --comment-prefixes prefix1,prefix2,... 55 56 By default, FileCheck ignores any occurrence in ``match-filename`` of any check 57 prefix if it is preceded on the same line by "``COM:``" or "``RUN:``". See the 58 section `The "COM:" directive`_ for usage details. 59 60 These default comment prefixes can be overridden by 61 :option:`--comment-prefixes` if they are not appropriate for your testing 62 environment. However, doing so is not recommended in LLVM's LIT-based test 63 suites, which should be easier to maintain if they all follow a consistent 64 comment style. In that case, consider proposing a change to the default 65 comment prefixes instead. 66 67.. option:: --allow-unused-prefixes 68 69 This option controls the behavior when using more than one prefix as specified 70 by :option:`--check-prefix` or :option:`--check-prefixes`, and some of these 71 prefixes are missing in the test file. If true, this is allowed, if false, 72 FileCheck will report an error, listing the missing prefixes. 73 74 It is currently, temporarily, true by default, and will be subsequently 75 switched to false. 76 77.. option:: --input-file filename 78 79 File to check (defaults to stdin). 80 81.. option:: --match-full-lines 82 83 By default, FileCheck allows matches of anywhere on a line. This 84 option will require all positive matches to cover an entire 85 line. Leading and trailing whitespace is ignored, unless 86 :option:`--strict-whitespace` is also specified. (Note: negative 87 matches from ``CHECK-NOT`` are not affected by this option!) 88 89 Passing this option is equivalent to inserting ``{{^ *}}`` or 90 ``{{^}}`` before, and ``{{ *$}}`` or ``{{$}}`` after every positive 91 check pattern. 92 93.. option:: --strict-whitespace 94 95 By default, FileCheck canonicalizes input horizontal whitespace (spaces and 96 tabs) which causes it to ignore these differences (a space will match a tab). 97 The :option:`--strict-whitespace` argument disables this behavior. End-of-line 98 sequences are canonicalized to UNIX-style ``\n`` in all modes. 99 100.. option:: --ignore-case 101 102 By default, FileCheck uses case-sensitive matching. This option causes 103 FileCheck to use case-insensitive matching. 104 105.. option:: --implicit-check-not check-pattern 106 107 Adds implicit negative checks for the specified patterns between positive 108 checks. The option allows writing stricter tests without stuffing them with 109 ``CHECK-NOT``\ s. 110 111 For example, "``--implicit-check-not warning:``" can be useful when testing 112 diagnostic messages from tools that don't have an option similar to ``clang 113 -verify``. With this option FileCheck will verify that input does not contain 114 warnings not covered by any ``CHECK:`` patterns. 115 116.. option:: --dump-input <value> 117 118 Dump input to stderr, adding annotations representing currently enabled 119 diagnostics. When there are multiple occurrences of this option, the 120 ``<value>`` that appears earliest in the list below has precedence. The 121 default is ``fail``. 122 123 * ``help`` - Explain input dump and quit 124 * ``always`` - Always dump input 125 * ``fail`` - Dump input on failure 126 * ``never`` - Never dump input 127 128.. option:: --dump-input-context <N> 129 130 In the dump requested by ``--dump-input``, print ``<N>`` input lines before 131 and ``<N>`` input lines after any lines specified by ``--dump-input-filter``. 132 When there are multiple occurrences of this option, the largest specified 133 ``<N>`` has precedence. The default is 5. 134 135.. option:: --dump-input-filter <value> 136 137 In the dump requested by ``--dump-input``, print only input lines of kind 138 ``<value>`` plus any context specified by ``--dump-input-context``. When 139 there are multiple occurrences of this option, the ``<value>`` that appears 140 earliest in the list below has precedence. The default is ``error`` when 141 ``--dump-input=fail``, and it's ``all`` when ``--dump-input=always``. 142 143 * ``all`` - All input lines 144 * ``annotation-full`` - Input lines with annotations 145 * ``annotation`` - Input lines with starting points of annotations 146 * ``error`` - Input lines with starting points of error annotations 147 148.. option:: --enable-var-scope 149 150 Enables scope for regex variables. 151 152 Variables with names that start with ``$`` are considered global and 153 remain set throughout the file. 154 155 All other variables get undefined after each encountered ``CHECK-LABEL``. 156 157.. option:: -D<VAR=VALUE> 158 159 Sets a filecheck pattern variable ``VAR`` with value ``VALUE`` that can be 160 used in ``CHECK:`` lines. 161 162.. option:: -D#<FMT>,<NUMVAR>=<NUMERIC EXPRESSION> 163 164 Sets a filecheck numeric variable ``NUMVAR`` of matching format ``FMT`` to 165 the result of evaluating ``<NUMERIC EXPRESSION>`` that can be used in 166 ``CHECK:`` lines. See section 167 ``FileCheck Numeric Variables and Expressions`` for details on supported 168 numeric expressions. 169 170.. option:: -version 171 172 Show the version number of this program. 173 174.. option:: -v 175 176 Print good directive pattern matches. However, if ``-dump-input=fail`` or 177 ``-dump-input=always``, add those matches as input annotations instead. 178 179.. option:: -vv 180 181 Print information helpful in diagnosing internal FileCheck issues, such as 182 discarded overlapping ``CHECK-DAG:`` matches, implicit EOF pattern matches, 183 and ``CHECK-NOT:`` patterns that do not have matches. Implies ``-v``. 184 However, if ``-dump-input=fail`` or ``-dump-input=always``, just add that 185 information as input annotations instead. 186 187.. option:: --allow-deprecated-dag-overlap 188 189 Enable overlapping among matches in a group of consecutive ``CHECK-DAG:`` 190 directives. This option is deprecated and is only provided for convenience 191 as old tests are migrated to the new non-overlapping ``CHECK-DAG:`` 192 implementation. 193 194.. option:: --allow-empty 195 196 Allow checking empty input. By default, empty input is rejected. 197 198.. option:: --color 199 200 Use colors in output (autodetected by default). 201 202EXIT STATUS 203----------- 204 205If :program:`FileCheck` verifies that the file matches the expected contents, 206it exits with 0. Otherwise, if not, or if an error occurs, it will exit with a 207non-zero value. 208 209TUTORIAL 210-------- 211 212FileCheck is typically used from LLVM regression tests, being invoked on the RUN 213line of the test. A simple example of using FileCheck from a RUN line looks 214like this: 215 216.. code-block:: llvm 217 218 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s 219 220This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe 221that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``. This 222means that FileCheck will be verifying its standard input (the llc output) 223against the filename argument specified (the original ``.ll`` file specified by 224"``%s``"). To see how this works, let's look at the rest of the ``.ll`` file 225(after the RUN line): 226 227.. code-block:: llvm 228 229 define void @sub1(i32* %p, i32 %v) { 230 entry: 231 ; CHECK: sub1: 232 ; CHECK: subl 233 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v) 234 ret void 235 } 236 237 define void @inc4(i64* %p) { 238 entry: 239 ; CHECK: inc4: 240 ; CHECK: incq 241 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1) 242 ret void 243 } 244 245Here you can see some "``CHECK:``" lines specified in comments. Now you can 246see how the file is piped into ``llvm-as``, then ``llc``, and the machine code 247output is what we are verifying. FileCheck checks the machine code output to 248verify that it matches what the "``CHECK:``" lines specify. 249 250The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that 251must occur in order. FileCheck defaults to ignoring horizontal whitespace 252differences (e.g. a space is allowed to match a tab) but otherwise, the contents 253of the "``CHECK:``" line is required to match some thing in the test file exactly. 254 255One nice thing about FileCheck (compared to grep) is that it allows merging 256test cases together into logical groups. For example, because the test above 257is checking for the "``sub1:``" and "``inc4:``" labels, it will not match 258unless there is a "``subl``" in between those labels. If it existed somewhere 259else in the file, that would not count: "``grep subl``" matches if "``subl``" 260exists anywhere in the file. 261 262The FileCheck -check-prefix option 263~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 264 265The FileCheck `-check-prefix` option allows multiple test 266configurations to be driven from one `.ll` file. This is useful in many 267circumstances, for example, testing different architectural variants with 268:program:`llc`. Here's a simple example: 269 270.. code-block:: llvm 271 272 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \ 273 ; RUN: | FileCheck %s -check-prefix=X32 274 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \ 275 ; RUN: | FileCheck %s -check-prefix=X64 276 277 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind { 278 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1 279 ret <4 x i32> %tmp1 280 ; X32: pinsrd_1: 281 ; X32: pinsrd $1, 4(%esp), %xmm0 282 283 ; X64: pinsrd_1: 284 ; X64: pinsrd $1, %edi, %xmm0 285 } 286 287In this case, we're testing that we get the expected code generation with 288both 32-bit and 64-bit code generation. 289 290The "COM:" directive 291~~~~~~~~~~~~~~~~~~~~ 292 293Sometimes you want to disable a FileCheck directive without removing it 294entirely, or you want to write comments that mention a directive by name. The 295"``COM:``" directive makes it easy to do this. For example, you might have: 296 297.. code-block:: llvm 298 299 ; X32: pinsrd_1: 300 ; X32: pinsrd $1, 4(%esp), %xmm0 301 302 ; COM: FIXME: X64 isn't working correctly yet for this part of codegen, but 303 ; COM: X64 will have something similar to X32: 304 ; COM: 305 ; COM: X64: pinsrd_1: 306 ; COM: X64: pinsrd $1, %edi, %xmm0 307 308Without "``COM:``", you would need to use some combination of rewording and 309directive syntax mangling to prevent FileCheck from recognizing the commented 310occurrences of "``X32:``" and "``X64:``" above as directives. Moreover, 311FileCheck diagnostics have been proposed that might complain about the above 312occurrences of "``X64``" that don't have the trailing "``:``" because they look 313like directive typos. Dodging all these problems can be tedious for a test 314author, and directive syntax mangling can make the purpose of test code unclear. 315"``COM:``" avoids all these problems. 316 317A few important usage notes: 318 319* "``COM:``" within another directive's pattern does *not* comment out the 320 remainder of the pattern. For example: 321 322 .. code-block:: llvm 323 324 ; X32: pinsrd $1, 4(%esp), %xmm0 COM: This is part of the X32 pattern! 325 326 If you need to temporarily comment out part of a directive's pattern, move it 327 to another line. The reason is that FileCheck parses "``COM:``" in the same 328 manner as any other directive: only the first directive on the line is 329 recognized as a directive. 330 331* For the sake of LIT, FileCheck treats "``RUN:``" just like "``COM:``". If this 332 is not suitable for your test environment, see :option:`--comment-prefixes`. 333 334* FileCheck does not recognize "``COM``", "``RUN``", or any user-defined comment 335 prefix as a comment directive if it's combined with one of the usual check 336 directive suffixes, such as "``-NEXT:``" or "``-NOT:``", discussed below. 337 FileCheck treats such a combination as plain text instead. If it needs to act 338 as a comment directive for your test environment, define it as such with 339 :option:`--comment-prefixes`. 340 341The "CHECK-NEXT:" directive 342~~~~~~~~~~~~~~~~~~~~~~~~~~~ 343 344Sometimes you want to match lines and would like to verify that matches 345happen on exactly consecutive lines with no other lines in between them. In 346this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify 347this. If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``". 348For example, something like this works as you'd expect: 349 350.. code-block:: llvm 351 352 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) { 353 %tmp3 = load <2 x double>* %A, align 16 354 %tmp7 = insertelement <2 x double> undef, double %B, i32 0 355 %tmp9 = shufflevector <2 x double> %tmp3, 356 <2 x double> %tmp7, 357 <2 x i32> < i32 0, i32 2 > 358 store <2 x double> %tmp9, <2 x double>* %r, align 16 359 ret void 360 361 ; CHECK: t2: 362 ; CHECK: movl 8(%esp), %eax 363 ; CHECK-NEXT: movapd (%eax), %xmm0 364 ; CHECK-NEXT: movhpd 12(%esp), %xmm0 365 ; CHECK-NEXT: movl 4(%esp), %eax 366 ; CHECK-NEXT: movapd %xmm0, (%eax) 367 ; CHECK-NEXT: ret 368 } 369 370"``CHECK-NEXT:``" directives reject the input unless there is exactly one 371newline between it and the previous directive. A "``CHECK-NEXT:``" cannot be 372the first directive in a file. 373 374The "CHECK-SAME:" directive 375~~~~~~~~~~~~~~~~~~~~~~~~~~~ 376 377Sometimes you want to match lines and would like to verify that matches happen 378on the same line as the previous match. In this case, you can use "``CHECK:``" 379and "``CHECK-SAME:``" directives to specify this. If you specified a custom 380check prefix, just use "``<PREFIX>-SAME:``". 381 382"``CHECK-SAME:``" is particularly powerful in conjunction with "``CHECK-NOT:``" 383(described below). 384 385For example, the following works like you'd expect: 386 387.. code-block:: llvm 388 389 !0 = !DILocation(line: 5, scope: !1, inlinedAt: !2) 390 391 ; CHECK: !DILocation(line: 5, 392 ; CHECK-NOT: column: 393 ; CHECK-SAME: scope: ![[SCOPE:[0-9]+]] 394 395"``CHECK-SAME:``" directives reject the input if there are any newlines between 396it and the previous directive. 397 398"``CHECK-SAME:``" is also useful to avoid writing matchers for irrelevant 399fields. For example, suppose you're writing a test which parses a tool that 400generates output like this: 401 402.. code-block:: text 403 404 Name: foo 405 Field1: ... 406 Field2: ... 407 Field3: ... 408 Value: 1 409 410 Name: bar 411 Field1: ... 412 Field2: ... 413 Field3: ... 414 Value: 2 415 416 Name: baz 417 Field1: ... 418 Field2: ... 419 Field3: ... 420 Value: 1 421 422To write a test that verifies ``foo`` has the value ``1``, you might first 423write this: 424 425.. code-block:: text 426 427 CHECK: Name: foo 428 CHECK: Value: 1{{$}} 429 430However, this would be a bad test: if the value for ``foo`` changes, the test 431would still pass because the "``CHECK: Value: 1``" line would match the value 432from ``baz``. To fix this, you could add ``CHECK-NEXT`` matchers for every 433``FieldN:`` line, but that would be verbose, and need to be updated when 434``Field4`` is added. A more succint way to write the test using the 435"``CHECK-SAME:``" matcher would be as follows: 436 437.. code-block:: text 438 439 CHECK: Name: foo 440 CHECK: Value: 441 CHECK-SAME: {{ 1$}} 442 443This verifies that the *next* time "``Value:``" appears in the output, it has 444the value ``1``. 445 446Note: a "``CHECK-SAME:``" cannot be the first directive in a file. 447 448The "CHECK-EMPTY:" directive 449~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 450 451If you need to check that the next line has nothing on it, not even whitespace, 452you can use the "``CHECK-EMPTY:``" directive. 453 454.. code-block:: llvm 455 456 declare void @foo() 457 458 declare void @bar() 459 ; CHECK: foo 460 ; CHECK-EMPTY: 461 ; CHECK-NEXT: bar 462 463Just like "``CHECK-NEXT:``" the directive will fail if there is more than one 464newline before it finds the next blank line, and it cannot be the first 465directive in a file. 466 467The "CHECK-NOT:" directive 468~~~~~~~~~~~~~~~~~~~~~~~~~~ 469 470The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur 471between two matches (or before the first match, or after the last match). For 472example, to verify that a load is removed by a transformation, a test like this 473can be used: 474 475.. code-block:: llvm 476 477 define i8 @coerce_offset0(i32 %V, i32* %P) { 478 store i32 %V, i32* %P 479 480 %P2 = bitcast i32* %P to i8* 481 %P3 = getelementptr i8* %P2, i32 2 482 483 %A = load i8* %P3 484 ret i8 %A 485 ; CHECK: @coerce_offset0 486 ; CHECK-NOT: load 487 ; CHECK: ret i8 488 } 489 490The "CHECK-COUNT:" directive 491~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 492 493If you need to match multiple lines with the same pattern over and over again 494you can repeat a plain ``CHECK:`` as many times as needed. If that looks too 495boring you can instead use a counted check "``CHECK-COUNT-<num>:``", where 496``<num>`` is a positive decimal number. It will match the pattern exactly 497``<num>`` times, no more and no less. If you specified a custom check prefix, 498just use "``<PREFIX>-COUNT-<num>:``" for the same effect. 499Here is a simple example: 500 501.. code-block:: text 502 503 Loop at depth 1 504 Loop at depth 1 505 Loop at depth 1 506 Loop at depth 1 507 Loop at depth 2 508 Loop at depth 3 509 510 ; CHECK-COUNT-6: Loop at depth {{[0-9]+}} 511 ; CHECK-NOT: Loop at depth {{[0-9]+}} 512 513The "CHECK-DAG:" directive 514~~~~~~~~~~~~~~~~~~~~~~~~~~ 515 516If it's necessary to match strings that don't occur in a strictly sequential 517order, "``CHECK-DAG:``" could be used to verify them between two matches (or 518before the first match, or after the last match). For example, clang emits 519vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks 520in the natural order: 521 522.. code-block:: c++ 523 524 // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s 525 526 struct Foo { virtual void method(); }; 527 Foo f; // emit vtable 528 // CHECK-DAG: @_ZTV3Foo = 529 530 struct Bar { virtual void method(); }; 531 Bar b; 532 // CHECK-DAG: @_ZTV3Bar = 533 534``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to 535exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result, 536the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all 537occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind 538occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example, 539 540.. code-block:: llvm 541 542 ; CHECK-DAG: BEFORE 543 ; CHECK-NOT: NOT 544 ; CHECK-DAG: AFTER 545 546This case will reject input strings where ``BEFORE`` occurs after ``AFTER``. 547 548With captured variables, ``CHECK-DAG:`` is able to match valid topological 549orderings of a DAG with edges from the definition of a variable to its use. 550It's useful, e.g., when your test cases need to match different output 551sequences from the instruction scheduler. For example, 552 553.. code-block:: llvm 554 555 ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2 556 ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4 557 ; CHECK: mul r5, [[REG1]], [[REG2]] 558 559In this case, any order of that two ``add`` instructions will be allowed. 560 561If you are defining `and` using variables in the same ``CHECK-DAG:`` block, 562be aware that the definition rule can match `after` its use. 563 564So, for instance, the code below will pass: 565 566.. code-block:: text 567 568 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 569 ; CHECK-DAG: vmov.32 [[REG2]][1] 570 vmov.32 d0[1] 571 vmov.32 d0[0] 572 573While this other code, will not: 574 575.. code-block:: text 576 577 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 578 ; CHECK-DAG: vmov.32 [[REG2]][1] 579 vmov.32 d1[1] 580 vmov.32 d0[0] 581 582While this can be very useful, it's also dangerous, because in the case of 583register sequence, you must have a strong order (read before write, copy before 584use, etc). If the definition your test is looking for doesn't match (because 585of a bug in the compiler), it may match further away from the use, and mask 586real bugs away. 587 588In those cases, to enforce the order, use a non-DAG directive between DAG-blocks. 589 590A ``CHECK-DAG:`` directive skips matches that overlap the matches of any 591preceding ``CHECK-DAG:`` directives in the same ``CHECK-DAG:`` block. Not only 592is this non-overlapping behavior consistent with other directives, but it's 593also necessary to handle sets of non-unique strings or patterns. For example, 594the following directives look for unordered log entries for two tasks in a 595parallel program, such as the OpenMP runtime: 596 597.. code-block:: text 598 599 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin 600 // CHECK-DAG: [[THREAD_ID]]: task_end 601 // 602 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin 603 // CHECK-DAG: [[THREAD_ID]]: task_end 604 605The second pair of directives is guaranteed not to match the same log entries 606as the first pair even though the patterns are identical and even if the text 607of the log entries is identical because the thread ID manages to be reused. 608 609The "CHECK-LABEL:" directive 610~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 611 612Sometimes in a file containing multiple tests divided into logical blocks, one 613or more ``CHECK:`` directives may inadvertently succeed by matching lines in a 614later block. While an error will usually eventually be generated, the check 615flagged as causing the error may not actually bear any relationship to the 616actual source of the problem. 617 618In order to produce better error messages in these cases, the "``CHECK-LABEL:``" 619directive can be used. It is treated identically to a normal ``CHECK`` 620directive except that FileCheck makes an additional assumption that a line 621matched by the directive cannot also be matched by any other check present in 622``match-filename``; this is intended to be used for lines containing labels or 623other unique identifiers. Conceptually, the presence of ``CHECK-LABEL`` divides 624the input stream into separate blocks, each of which is processed independently, 625preventing a ``CHECK:`` directive in one block matching a line in another block. 626If ``--enable-var-scope`` is in effect, all local variables are cleared at the 627beginning of the block. 628 629For example, 630 631.. code-block:: llvm 632 633 define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) { 634 entry: 635 ; CHECK-LABEL: C_ctor_base: 636 ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0 637 ; CHECK: bl A_ctor_base 638 ; CHECK: mov r0, [[SAVETHIS]] 639 %0 = bitcast %struct.C* %this to %struct.A* 640 %call = tail call %struct.A* @A_ctor_base(%struct.A* %0) 641 %1 = bitcast %struct.C* %this to %struct.B* 642 %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x) 643 ret %struct.C* %this 644 } 645 646 define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) { 647 entry: 648 ; CHECK-LABEL: D_ctor_base: 649 650The use of ``CHECK-LABEL:`` directives in this case ensures that the three 651``CHECK:`` directives only accept lines corresponding to the body of the 652``@C_ctor_base`` function, even if the patterns match lines found later in 653the file. Furthermore, if one of these three ``CHECK:`` directives fail, 654FileCheck will recover by continuing to the next block, allowing multiple test 655failures to be detected in a single invocation. 656 657There is no requirement that ``CHECK-LABEL:`` directives contain strings that 658correspond to actual syntactic labels in a source or output language: they must 659simply uniquely match a single line in the file being verified. 660 661``CHECK-LABEL:`` directives cannot contain variable definitions or uses. 662 663FileCheck Regex Matching Syntax 664~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 665 666All FileCheck directives take a pattern to match. 667For most uses of FileCheck, fixed string matching is perfectly sufficient. For 668some things, a more flexible form of matching is desired. To support this, 669FileCheck allows you to specify regular expressions in matching strings, 670surrounded by double braces: ``{{yourregex}}``. FileCheck implements a POSIX 671regular expression matcher; it supports Extended POSIX regular expressions 672(ERE). Because we want to use fixed string matching for a majority of what we 673do, FileCheck has been designed to support mixing and matching fixed string 674matching with regular expressions. This allows you to write things like this: 675 676.. code-block:: llvm 677 678 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}} 679 680In this case, any offset from the ESP register will be allowed, and any xmm 681register will be allowed. 682 683Because regular expressions are enclosed with double braces, they are 684visually distinct, and you don't need to use escape characters within the double 685braces like you would in C. In the rare case that you want to match double 686braces explicitly from the input, you can use something ugly like 687``{{[}][}]}}`` as your pattern. Or if you are using the repetition count 688syntax, for example ``[[:xdigit:]]{8}`` to match exactly 8 hex digits, you 689would need to add parentheses like this ``{{([[:xdigit:]]{8})}}`` to avoid 690confusion with FileCheck's closing double-brace. 691 692FileCheck String Substitution Blocks 693~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 694 695It is often useful to match a pattern and then verify that it occurs again 696later in the file. For codegen tests, this can be useful to allow any 697register, but verify that that register is used consistently later. To do 698this, :program:`FileCheck` supports string substitution blocks that allow 699string variables to be defined and substituted into patterns. Here is a simple 700example: 701 702.. code-block:: llvm 703 704 ; CHECK: test5: 705 ; CHECK: notw [[REGISTER:%[a-z]+]] 706 ; CHECK: andw {{.*}}[[REGISTER]] 707 708The first check line matches a regex ``%[a-z]+`` and captures it into the 709string variable ``REGISTER``. The second line verifies that whatever is in 710``REGISTER`` occurs later in the file after an "``andw``". :program:`FileCheck` 711string substitution blocks are always contained in ``[[ ]]`` pairs, and string 712variable names can be formed with the regex ``[a-zA-Z_][a-zA-Z0-9_]*``. If a 713colon follows the name, then it is a definition of the variable; otherwise, it 714is a substitution. 715 716:program:`FileCheck` variables can be defined multiple times, and substitutions 717always get the latest value. Variables can also be substituted later on the 718same line they were defined on. For example: 719 720.. code-block:: llvm 721 722 ; CHECK: op [[REG:r[0-9]+]], [[REG]] 723 724Can be useful if you want the operands of ``op`` to be the same register, 725and don't care exactly which register it is. 726 727If ``--enable-var-scope`` is in effect, variables with names that 728start with ``$`` are considered to be global. All others variables are 729local. All local variables get undefined at the beginning of each 730CHECK-LABEL block. Global variables are not affected by CHECK-LABEL. 731This makes it easier to ensure that individual tests are not affected 732by variables set in preceding tests. 733 734FileCheck Numeric Substitution Blocks 735~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 736 737:program:`FileCheck` also supports numeric substitution blocks that allow 738defining numeric variables and checking for numeric values that satisfy a 739numeric expression constraint based on those variables via a numeric 740substitution. This allows ``CHECK:`` directives to verify a numeric relation 741between two numbers, such as the need for consecutive registers to be used. 742 743The syntax to capture a numeric value is 744``[[#%<fmtspec>,<NUMVAR>:]]`` where: 745 746* ``%<fmtspec>,`` is an optional format specifier to indicate what number 747 format to match and the minimum number of digits to expect. 748 749* ``<NUMVAR>:`` is an optional definition of variable ``<NUMVAR>`` from the 750 captured value. 751 752The syntax of ``<fmtspec>`` is: ``.<precision><conversion specifier>`` where: 753 754* ``.<precision>`` is an optional printf-style precision specifier in which 755 ``<precision>`` indicates the minimum number of digits that the value matched 756 must have, expecting leading zeros if needed. 757 758* ``<conversion specifier>`` is an optional scanf-style conversion specifier 759 to indicate what number format to match (e.g. hex number). Currently 760 accepted format specifiers are ``%u``, ``%d``, ``%x`` and ``%X``. If absent, 761 the format specifier defaults to ``%u``. 762 763 764For example: 765 766.. code-block:: llvm 767 768 ; CHECK: mov r[[#REG:]], 0x[[#%.8X,ADDR:]] 769 770would match ``mov r5, 0x0000FEFE`` and set ``REG`` to the value ``5`` and 771``ADDR`` to the value ``0xFEFE``. Note that due to the precision it would fail 772to match ``mov r5, 0xFEFE``. 773 774As a result of the numeric variable definition being optional, it is possible 775to only check that a numeric value is present in a given format. This can be 776useful when the value itself is not useful, for instance: 777 778.. code-block:: gas 779 780 ; CHECK-NOT: mov r0, r[[#]] 781 782to check that a value is synthesized rather than moved around. 783 784 785The syntax of a numeric substitution is 786``[[#%<fmtspec>, <constraint> <expr>]]`` where: 787 788* ``<fmtspec>`` is the same format specifier as for defining a variable but 789 in this context indicating how a numeric expression value should be matched 790 against. If absent, both components of the format specifier are inferred from 791 the matching format of the numeric variable(s) used by the expression 792 constraint if any, and defaults to ``%u`` if no numeric variable is used, 793 denoting that the value should be unsigned with no leading zeros. In case of 794 conflict between format specifiers of several numeric variables, the 795 conversion specifier becomes mandatory but the precision specifier remains 796 optional. 797 798* ``<constraint>`` is the constraint describing how the value to match must 799 relate to the value of the numeric expression. The only currently accepted 800 constraint is ``==`` for an exact match and is the default if 801 ``<constraint>`` is not provided. No matching constraint must be specified 802 when the ``<expr>`` is empty. 803 804* ``<expr>`` is an expression. An expression is in turn recursively defined 805 as: 806 807 * a numeric operand, or 808 * an expression followed by an operator and a numeric operand. 809 810 A numeric operand is a previously defined numeric variable, an integer 811 literal, or a function. Spaces are accepted before, after and between any of 812 these elements. Numeric operands have 64-bit precision. Overflow and underflow 813 are rejected. There is no support for operator precedence, but parentheses 814 can be used to change the evaluation order. 815 816The supported operators are: 817 818 * ``+`` - Returns the sum of its two operands. 819 * ``-`` - Returns the difference of its two operands. 820 821The syntax of a function call is ``<name>(<arguments>)`` where: 822 823* ``name`` is a predefined string literal. Accepted values are: 824 825 * add - Returns the sum of its two operands. 826 * div - Returns the quotient of its two operands. 827 * max - Returns the largest of its two operands. 828 * min - Returns the smallest of its two operands. 829 * mul - Returns the product of its two operands. 830 * sub - Returns the difference of its two operands. 831 832* ``<arguments>`` is a comma separated list of expressions. 833 834For example: 835 836.. code-block:: llvm 837 838 ; CHECK: load r[[#REG:]], [r0] 839 ; CHECK: load r[[#REG+1]], [r1] 840 ; CHECK: Loading from 0x[[#%x,ADDR:]] 841 ; CHECK-SAME: to 0x[[#ADDR + 7]] 842 843The above example would match the text: 844 845.. code-block:: gas 846 847 load r5, [r0] 848 load r6, [r1] 849 Loading from 0xa0463440 to 0xa0463447 850 851but would not match the text: 852 853.. code-block:: gas 854 855 load r5, [r0] 856 load r7, [r1] 857 Loading from 0xa0463440 to 0xa0463443 858 859Due to ``7`` being unequal to ``5 + 1`` and ``a0463443`` being unequal to 860``a0463440 + 7``. 861 862 863A numeric variable can also be defined to the result of a numeric expression, 864in which case the numeric expression constraint is checked and if verified the 865variable is assigned to the value. The unified syntax for both checking a 866numeric expression and capturing its value into a numeric variable is thus 867``[[#%<fmtspec>,<NUMVAR>: <constraint> <expr>]]`` with each element as 868described previously. One can use this syntax to make a testcase more 869self-describing by using variables instead of values: 870 871.. code-block:: gas 872 873 ; CHECK: mov r[[#REG_OFFSET:]], 0x[[#%X,FIELD_OFFSET:12]] 874 ; CHECK-NEXT: load r[[#]], [r[[#REG_BASE:]], r[[#REG_OFFSET]]] 875 876which would match: 877 878.. code-block:: gas 879 880 mov r4, 0xC 881 load r6, [r5, r4] 882 883The ``--enable-var-scope`` option has the same effect on numeric variables as 884on string variables. 885 886Important note: In its current implementation, an expression cannot use a 887numeric variable defined earlier in the same CHECK directive. 888 889FileCheck Pseudo Numeric Variables 890~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 891 892Sometimes there's a need to verify output that contains line numbers of the 893match file, e.g. when testing compiler diagnostics. This introduces a certain 894fragility of the match file structure, as "``CHECK:``" lines contain absolute 895line numbers in the same file, which have to be updated whenever line numbers 896change due to text addition or deletion. 897 898To support this case, FileCheck expressions understand the ``@LINE`` pseudo 899numeric variable which evaluates to the line number of the CHECK pattern where 900it is found. 901 902This way match patterns can be put near the relevant test lines and include 903relative line number references, for example: 904 905.. code-block:: c++ 906 907 // CHECK: test.cpp:[[# @LINE + 4]]:6: error: expected ';' after top level declarator 908 // CHECK-NEXT: {{^int a}} 909 // CHECK-NEXT: {{^ \^}} 910 // CHECK-NEXT: {{^ ;}} 911 int a 912 913To support legacy uses of ``@LINE`` as a special string variable, 914:program:`FileCheck` also accepts the following uses of ``@LINE`` with string 915substitution block syntax: ``[[@LINE]]``, ``[[@LINE+<offset>]]`` and 916``[[@LINE-<offset>]]`` without any spaces inside the brackets and where 917``offset`` is an integer. 918 919Matching Newline Characters 920~~~~~~~~~~~~~~~~~~~~~~~~~~~ 921 922To match newline characters in regular expressions the character class 923``[[:space:]]`` can be used. For example, the following pattern: 924 925.. code-block:: c++ 926 927 // CHECK: DW_AT_location [DW_FORM_sec_offset] ([[DLOC:0x[0-9a-f]+]]){{[[:space:]].*}}"intd" 928 929matches output of the form (from llvm-dwarfdump): 930 931.. code-block:: text 932 933 DW_AT_location [DW_FORM_sec_offset] (0x00000233) 934 DW_AT_name [DW_FORM_strp] ( .debug_str[0x000000c9] = "intd") 935 936letting us set the :program:`FileCheck` variable ``DLOC`` to the desired value 937``0x00000233``, extracted from the line immediately preceding "``intd``". 938