1 FileCheck - Flexible pattern matching file verifier
2 ===================================================
7 :program:`FileCheck` *match-filename* [*--check-prefix=XXX*] [*--strict-whitespace*]
12 :program:`FileCheck` reads two files (one from standard input, and one
13 specified on the command line) and uses one to verify the other. This
14 behavior is particularly useful for the testsuite, which wants to verify that
15 the output of some tool (e.g. :program:`llc`) contains the expected information
16 (for example, a movsd from esp or whatever is interesting). This is similar to
17 using :program:`grep`, but it is optimized for matching multiple different
18 inputs in one file in a specific order.
20 The ``match-filename`` file specifies the file that contains the patterns to
21 match. The file to verify is read from standard input unless the
22 :option:`--input-file` option is used.
29 Print a summary of command line options.
31 .. option:: --check-prefix prefix
33 FileCheck searches the contents of ``match-filename`` for patterns to
34 match. By default, these patterns are prefixed with "``CHECK:``".
35 If you'd like to use a different prefix (e.g. because the same input
36 file is checking multiple different tool or options), the
37 :option:`--check-prefix` argument allows you to specify one or more
38 prefixes to match. Multiple prefixes are useful for tests which might
39 change for different run options, but most lines remain the same.
41 .. option:: --check-prefixes prefix1,prefix2,...
43 An alias of :option:`--check-prefix` that allows multiple prefixes to be
44 specified as a comma separated list.
46 .. option:: --input-file filename
48 File to check (defaults to stdin).
50 .. option:: --match-full-lines
52 By default, FileCheck allows matches of anywhere on a line. This
53 option will require all positive matches to cover an entire
54 line. Leading and trailing whitespace is ignored, unless
55 :option:`--strict-whitespace` is also specified. (Note: negative
56 matches from ``CHECK-NOT`` are not affected by this option!)
58 Passing this option is equivalent to inserting ``{{^ *}}`` or
59 ``{{^}}`` before, and ``{{ *$}}`` or ``{{$}}`` after every positive
62 .. option:: --strict-whitespace
64 By default, FileCheck canonicalizes input horizontal whitespace (spaces and
65 tabs) which causes it to ignore these differences (a space will match a tab).
66 The :option:`--strict-whitespace` argument disables this behavior. End-of-line
67 sequences are canonicalized to UNIX-style ``\n`` in all modes.
69 .. option:: --implicit-check-not check-pattern
71 Adds implicit negative checks for the specified patterns between positive
72 checks. The option allows writing stricter tests without stuffing them with
75 For example, "``--implicit-check-not warning:``" can be useful when testing
76 diagnostic messages from tools that don't have an option similar to ``clang
77 -verify``. With this option FileCheck will verify that input does not contain
78 warnings not covered by any ``CHECK:`` patterns.
80 .. option:: --enable-var-scope
82 Enables scope for regex variables.
84 Variables with names that start with ``$`` are considered global and
85 remain set throughout the file.
87 All other variables get undefined after each encountered ``CHECK-LABEL``.
89 .. option:: -D<VAR=VALUE>
91 Sets a filecheck variable ``VAR`` with value ``VALUE`` that can be used in
96 Show the version number of this program.
98 .. option:: --allow-deprecated-dag-overlap
100 Enable overlapping among matches in a group of consecutive ``CHECK-DAG:``
101 directives. This option is deprecated and is only provided for convenience
102 as old tests are migrated to the new non-overlapping ``CHECK-DAG:``
108 If :program:`FileCheck` verifies that the file matches the expected contents,
109 it exits with 0. Otherwise, if not, or if an error occurs, it will exit with a
115 FileCheck is typically used from LLVM regression tests, being invoked on the RUN
116 line of the test. A simple example of using FileCheck from a RUN line looks
121 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s
123 This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe
124 that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``. This
125 means that FileCheck will be verifying its standard input (the llc output)
126 against the filename argument specified (the original ``.ll`` file specified by
127 "``%s``"). To see how this works, let's look at the rest of the ``.ll`` file
128 (after the RUN line):
132 define void @sub1(i32* %p, i32 %v) {
136 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
140 define void @inc4(i64* %p) {
144 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
148 Here you can see some "``CHECK:``" lines specified in comments. Now you can
149 see how the file is piped into ``llvm-as``, then ``llc``, and the machine code
150 output is what we are verifying. FileCheck checks the machine code output to
151 verify that it matches what the "``CHECK:``" lines specify.
153 The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that
154 must occur in order. FileCheck defaults to ignoring horizontal whitespace
155 differences (e.g. a space is allowed to match a tab) but otherwise, the contents
156 of the "``CHECK:``" line is required to match some thing in the test file exactly.
158 One nice thing about FileCheck (compared to grep) is that it allows merging
159 test cases together into logical groups. For example, because the test above
160 is checking for the "``sub1:``" and "``inc4:``" labels, it will not match
161 unless there is a "``subl``" in between those labels. If it existed somewhere
162 else in the file, that would not count: "``grep subl``" matches if "``subl``"
163 exists anywhere in the file.
165 The FileCheck -check-prefix option
166 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
168 The FileCheck `-check-prefix` option allows multiple test
169 configurations to be driven from one `.ll` file. This is useful in many
170 circumstances, for example, testing different architectural variants with
171 :program:`llc`. Here's a simple example:
175 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
176 ; RUN: | FileCheck %s -check-prefix=X32
177 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
178 ; RUN: | FileCheck %s -check-prefix=X64
180 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind {
181 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1
184 ; X32: pinsrd $1, 4(%esp), %xmm0
187 ; X64: pinsrd $1, %edi, %xmm0
190 In this case, we're testing that we get the expected code generation with
191 both 32-bit and 64-bit code generation.
193 The "CHECK-NEXT:" directive
194 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
196 Sometimes you want to match lines and would like to verify that matches
197 happen on exactly consecutive lines with no other lines in between them. In
198 this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify
199 this. If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``".
200 For example, something like this works as you'd expect:
204 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) {
205 %tmp3 = load <2 x double>* %A, align 16
206 %tmp7 = insertelement <2 x double> undef, double %B, i32 0
207 %tmp9 = shufflevector <2 x double> %tmp3,
209 <2 x i32> < i32 0, i32 2 >
210 store <2 x double> %tmp9, <2 x double>* %r, align 16
214 ; CHECK: movl 8(%esp), %eax
215 ; CHECK-NEXT: movapd (%eax), %xmm0
216 ; CHECK-NEXT: movhpd 12(%esp), %xmm0
217 ; CHECK-NEXT: movl 4(%esp), %eax
218 ; CHECK-NEXT: movapd %xmm0, (%eax)
222 "``CHECK-NEXT:``" directives reject the input unless there is exactly one
223 newline between it and the previous directive. A "``CHECK-NEXT:``" cannot be
224 the first directive in a file.
226 The "CHECK-SAME:" directive
227 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
229 Sometimes you want to match lines and would like to verify that matches happen
230 on the same line as the previous match. In this case, you can use "``CHECK:``"
231 and "``CHECK-SAME:``" directives to specify this. If you specified a custom
232 check prefix, just use "``<PREFIX>-SAME:``".
234 "``CHECK-SAME:``" is particularly powerful in conjunction with "``CHECK-NOT:``"
237 For example, the following works like you'd expect:
241 !0 = !DILocation(line: 5, scope: !1, inlinedAt: !2)
243 ; CHECK: !DILocation(line: 5,
245 ; CHECK-SAME: scope: ![[SCOPE:[0-9]+]]
247 "``CHECK-SAME:``" directives reject the input if there are any newlines between
248 it and the previous directive. A "``CHECK-SAME:``" cannot be the first
251 The "CHECK-EMPTY:" directive
252 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
254 If you need to check that the next line has nothing on it, not even whitespace,
255 you can use the "``CHECK-EMPTY:``" directive.
266 Just like "``CHECK-NEXT:``" the directive will fail if there is more than one
267 newline before it finds the next blank line, and it cannot be the first
270 The "CHECK-NOT:" directive
271 ~~~~~~~~~~~~~~~~~~~~~~~~~~
273 The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur
274 between two matches (or before the first match, or after the last match). For
275 example, to verify that a load is removed by a transformation, a test like this
280 define i8 @coerce_offset0(i32 %V, i32* %P) {
281 store i32 %V, i32* %P
283 %P2 = bitcast i32* %P to i8*
284 %P3 = getelementptr i8* %P2, i32 2
288 ; CHECK: @coerce_offset0
293 The "CHECK-DAG:" directive
294 ~~~~~~~~~~~~~~~~~~~~~~~~~~
296 If it's necessary to match strings that don't occur in a strictly sequential
297 order, "``CHECK-DAG:``" could be used to verify them between two matches (or
298 before the first match, or after the last match). For example, clang emits
299 vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks
300 in the natural order:
304 // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s
306 struct Foo { virtual void method(); };
307 Foo f; // emit vtable
308 // CHECK-DAG: @_ZTV3Foo =
310 struct Bar { virtual void method(); };
312 // CHECK-DAG: @_ZTV3Bar =
314 ``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to
315 exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result,
316 the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all
317 occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind
318 occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example,
326 This case will reject input strings where ``BEFORE`` occurs after ``AFTER``.
328 With captured variables, ``CHECK-DAG:`` is able to match valid topological
329 orderings of a DAG with edges from the definition of a variable to its use.
330 It's useful, e.g., when your test cases need to match different output
331 sequences from the instruction scheduler. For example,
335 ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2
336 ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4
337 ; CHECK: mul r5, [[REG1]], [[REG2]]
339 In this case, any order of that two ``add`` instructions will be allowed.
341 If you are defining `and` using variables in the same ``CHECK-DAG:`` block,
342 be aware that the definition rule can match `after` its use.
344 So, for instance, the code below will pass:
348 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
349 ; CHECK-DAG: vmov.32 [[REG2]][1]
353 While this other code, will not:
357 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
358 ; CHECK-DAG: vmov.32 [[REG2]][1]
362 While this can be very useful, it's also dangerous, because in the case of
363 register sequence, you must have a strong order (read before write, copy before
364 use, etc). If the definition your test is looking for doesn't match (because
365 of a bug in the compiler), it may match further away from the use, and mask
368 In those cases, to enforce the order, use a non-DAG directive between DAG-blocks.
370 A ``CHECK-DAG:`` directive skips matches that overlap the matches of any
371 preceding ``CHECK-DAG:`` directives in the same ``CHECK-DAG:`` block. Not only
372 is this non-overlapping behavior consistent with other directives, but it's
373 also necessary to handle sets of non-unique strings or patterns. For example,
374 the following directives look for unordered log entries for two tasks in a
375 parallel program, such as the OpenMP runtime:
379 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin
380 // CHECK-DAG: [[THREAD_ID]]: task_end
382 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin
383 // CHECK-DAG: [[THREAD_ID]]: task_end
385 The second pair of directives is guaranteed not to match the same log entries
386 as the first pair even though the patterns are identical and even if the text
387 of the log entries is identical because the thread ID manages to be reused.
389 The "CHECK-LABEL:" directive
390 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
392 Sometimes in a file containing multiple tests divided into logical blocks, one
393 or more ``CHECK:`` directives may inadvertently succeed by matching lines in a
394 later block. While an error will usually eventually be generated, the check
395 flagged as causing the error may not actually bear any relationship to the
396 actual source of the problem.
398 In order to produce better error messages in these cases, the "``CHECK-LABEL:``"
399 directive can be used. It is treated identically to a normal ``CHECK``
400 directive except that FileCheck makes an additional assumption that a line
401 matched by the directive cannot also be matched by any other check present in
402 ``match-filename``; this is intended to be used for lines containing labels or
403 other unique identifiers. Conceptually, the presence of ``CHECK-LABEL`` divides
404 the input stream into separate blocks, each of which is processed independently,
405 preventing a ``CHECK:`` directive in one block matching a line in another block.
406 If ``--enable-var-scope`` is in effect, all local variables are cleared at the
407 beginning of the block.
413 define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) {
415 ; CHECK-LABEL: C_ctor_base:
416 ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0
417 ; CHECK: bl A_ctor_base
418 ; CHECK: mov r0, [[SAVETHIS]]
419 %0 = bitcast %struct.C* %this to %struct.A*
420 %call = tail call %struct.A* @A_ctor_base(%struct.A* %0)
421 %1 = bitcast %struct.C* %this to %struct.B*
422 %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x)
426 define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) {
428 ; CHECK-LABEL: D_ctor_base:
430 The use of ``CHECK-LABEL:`` directives in this case ensures that the three
431 ``CHECK:`` directives only accept lines corresponding to the body of the
432 ``@C_ctor_base`` function, even if the patterns match lines found later in
433 the file. Furthermore, if one of these three ``CHECK:`` directives fail,
434 FileCheck will recover by continuing to the next block, allowing multiple test
435 failures to be detected in a single invocation.
437 There is no requirement that ``CHECK-LABEL:`` directives contain strings that
438 correspond to actual syntactic labels in a source or output language: they must
439 simply uniquely match a single line in the file being verified.
441 ``CHECK-LABEL:`` directives cannot contain variable definitions or uses.
443 FileCheck Pattern Matching Syntax
444 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
446 All FileCheck directives take a pattern to match.
447 For most uses of FileCheck, fixed string matching is perfectly sufficient. For
448 some things, a more flexible form of matching is desired. To support this,
449 FileCheck allows you to specify regular expressions in matching strings,
450 surrounded by double braces: ``{{yourregex}}``. FileCheck implements a POSIX
451 regular expression matcher; it supports Extended POSIX regular expressions
452 (ERE). Because we want to use fixed string matching for a majority of what we
453 do, FileCheck has been designed to support mixing and matching fixed string
454 matching with regular expressions. This allows you to write things like this:
458 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}}
460 In this case, any offset from the ESP register will be allowed, and any xmm
461 register will be allowed.
463 Because regular expressions are enclosed with double braces, they are
464 visually distinct, and you don't need to use escape characters within the double
465 braces like you would in C. In the rare case that you want to match double
466 braces explicitly from the input, you can use something ugly like
467 ``{{[{][{]}}`` as your pattern.
472 It is often useful to match a pattern and then verify that it occurs again
473 later in the file. For codegen tests, this can be useful to allow any register,
474 but verify that that register is used consistently later. To do this,
475 :program:`FileCheck` allows named variables to be defined and substituted into
476 patterns. Here is a simple example:
481 ; CHECK: notw [[REGISTER:%[a-z]+]]
482 ; CHECK: andw {{.*}}[[REGISTER]]
484 The first check line matches a regex ``%[a-z]+`` and captures it into the
485 variable ``REGISTER``. The second line verifies that whatever is in
486 ``REGISTER`` occurs later in the file after an "``andw``". :program:`FileCheck`
487 variable references are always contained in ``[[ ]]`` pairs, and their names can
488 be formed with the regex ``[a-zA-Z_][a-zA-Z0-9_]*``. If a colon follows the name,
489 then it is a definition of the variable; otherwise, it is a use.
491 :program:`FileCheck` variables can be defined multiple times, and uses always
492 get the latest value. Variables can also be used later on the same line they
493 were defined on. For example:
497 ; CHECK: op [[REG:r[0-9]+]], [[REG]]
499 Can be useful if you want the operands of ``op`` to be the same register,
500 and don't care exactly which register it is.
502 If ``--enable-var-scope`` is in effect, variables with names that
503 start with ``$`` are considered to be global. All others variables are
504 local. All local variables get undefined at the beginning of each
505 CHECK-LABEL block. Global variables are not affected by CHECK-LABEL.
506 This makes it easier to ensure that individual tests are not affected
507 by variables set in preceding tests.
509 FileCheck Expressions
510 ~~~~~~~~~~~~~~~~~~~~~
512 Sometimes there's a need to verify output which refers line numbers of the
513 match file, e.g. when testing compiler diagnostics. This introduces a certain
514 fragility of the match file structure, as "``CHECK:``" lines contain absolute
515 line numbers in the same file, which have to be updated whenever line numbers
516 change due to text addition or deletion.
518 To support this case, FileCheck allows using ``[[@LINE]]``,
519 ``[[@LINE+<offset>]]``, ``[[@LINE-<offset>]]`` expressions in patterns. These
520 expressions expand to a number of the line where a pattern is located (with an
521 optional integer offset).
523 This way match patterns can be put near the relevant test lines and include
524 relative line number references, for example:
528 // CHECK: test.cpp:[[@LINE+4]]:6: error: expected ';' after top level declarator
529 // CHECK-NEXT: {{^int a}}
530 // CHECK-NEXT: {{^ \^}}
531 // CHECK-NEXT: {{^ ;}}
534 Matching Newline Characters
535 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
537 To match newline characters in regular expressions the character class
538 ``[[:space:]]`` can be used. For example, the following pattern:
542 // CHECK: DW_AT_location [DW_FORM_sec_offset] ([[DLOC:0x[0-9a-f]+]]){{[[:space:]].*}}"intd"
544 matches output of the form (from llvm-dwarfdump):
548 DW_AT_location [DW_FORM_sec_offset] (0x00000233)
549 DW_AT_name [DW_FORM_strp] ( .debug_str[0x000000c9] = "intd")
551 letting us set the :program:`FileCheck` variable ``DLOC`` to the desired value
552 ``0x00000233``, extracted from the line immediately preceding "``intd``".