1FILECHECK(1) LLVM FILECHECK(1)
2
6 FileCheck - Flexible pattern matching file verifier
7
9 FileCheck match-filename [--check-prefix=XXX] [--strict-whitespace]
10
12 FileCheck reads two files (one from standard input, and one specified
13 on the command line) and uses one to verify the other. This behavior
14 is particularly useful for the testsuite, which wants to verify that
15 the output of some tool (e.g. llc) contains the expected information
16 (for example, a movsd from esp or whatever is interesting). This is
17 similar to using grep, but it is optimized for matching multiple dif‐
18 ferent inputs in one file in a specific order.
19 The match-filename file specifies the file that contains the patterns
21 to match. The file to verify is read from standard input unless the
22 --input-file option is used.
23
25 Options are parsed from the environment variable FILECHECK_OPTS and
26 from the command line.
27 -help Print a summary of command line options.
29 --check-prefix prefix
31 FileCheck searches the contents of match-filename for patterns
32 to match. By default, these patterns are prefixed with
33 "CHECK:". If you'd like to use a different prefix (e.g. because
34 the same input file is checking multiple different tool or op‐
35 tions), the --check-prefix argument allows you to specify (with‐
36 out the trailing ":") one or more prefixes to match. Multiple
37 prefixes are useful for tests which might change for different
38 run options, but most lines remain the same.
39 FileCheck does not permit duplicate prefixes, even if one is a
41 check prefix and one is a comment prefix (see --comment-prefixes
42 below).
43 --check-prefixes prefix1,prefix2,...
45 An alias of --check-prefix that allows multiple prefixes to be
46 specified as a comma separated list.
47 --comment-prefixes prefix1,prefix2,...
49 By default, FileCheck ignores any occurrence in match-filename
50 of any check prefix if it is preceded on the same line by "COM:"
51 or "RUN:". See the section The "COM:" directive for usage de‐
52 tails.
53 These default comment prefixes can be overridden by
55 --comment-prefixes if they are not appropriate for your testing
56 environment. However, doing so is not recommended in LLVM's
57 LIT-based test suites, which should be easier to maintain if
58 they all follow a consistent comment style. In that case, con‐
59 sider proposing a change to the default comment prefixes in‐
60 stead.
61 --input-file filename
63 File to check (defaults to stdin).
64 --match-full-lines
66 By default, FileCheck allows matches of anywhere on a line. This
67 option will require all positive matches to cover an entire
68 line. Leading and trailing whitespace is ignored, unless
69 --strict-whitespace is also specified. (Note: negative matches
70 from CHECK-NOT are not affected by this option!)
71 Passing this option is equivalent to inserting {{^ *}} or {{^}}
73 before, and {{ *$}} or {{$}} after every positive check pattern.
74 --strict-whitespace
76 By default, FileCheck canonicalizes input horizontal whitespace
77 (spaces and tabs) which causes it to ignore these differences (a
78 space will match a tab). The --strict-whitespace argument dis‐
79 ables this behavior. End-of-line sequences are canonicalized to
80 UNIX-style \n in all modes.
81 --ignore-case
83 By default, FileCheck uses case-sensitive matching. This option
84 causes FileCheck to use case-insensitive matching.
85 --implicit-check-not check-pattern
87 Adds implicit negative checks for the specified patterns between
88 positive checks. The option allows writing stricter tests with‐
89 out stuffing them with CHECK-NOTs.
90 For example, "--implicit-check-not warning:" can be useful when
92 testing diagnostic messages from tools that don't have an option
93 similar to clang -verify. With this option FileCheck will verify
94 that input does not contain warnings not covered by any CHECK:
95 patterns.
96 --dump-input <value>
98 Dump input to stderr, adding annotations representing currently
99 enabled diagnostics. When there are multiple occurrences of
100 this option, the <value> that appears earliest in the list below
101 has precedence. The default is fail.
102 • help - Explain input dump and quit
104 • always - Always dump input
106 • fail - Dump input on failure
108 • never - Never dump input
110 --dump-input-context <N>
112 In the dump requested by --dump-input, print <N> input lines be‐
113 fore and <N> input lines after any lines specified by --dump-in‐
114 put-filter. When there are multiple occurrences of this option,
115 the largest specified <N> has precedence. The default is 5.
116 --dump-input-filter <value>
118 In the dump requested by --dump-input, print only input lines of
119 kind <value> plus any context specified by --dump-input-context.
120 When there are multiple occurrences of this option, the <value>
121 that appears earliest in the list below has precedence. The de‐
122 fault is error when --dump-input=fail, and it's all when
123 --dump-input=always.
124 • all - All input lines
126 • annotation-full - Input lines with annotations
128 • annotation - Input lines with starting points of annota‐
130 tions
131 • error - Input lines with starting points of error
133 annotations
134 --enable-var-scope
136 Enables scope for regex variables.
137 Variables with names that start with $ are considered global and
139 remain set throughout the file.
140 All other variables get undefined after each encountered
142 CHECK-LABEL.
143 -D<VAR=VALUE>
145 Sets a filecheck pattern variable VAR with value VALUE that can
146 be used in CHECK: lines.
147 -D#<FMT>,<NUMVAR>=<NUMERIC EXPRESSION>
149 Sets a filecheck numeric variable NUMVAR of matching format FMT
150 to the result of evaluating <NUMERIC EXPRESSION> that can be
151 used in CHECK: lines. See section FileCheck Numeric Variables
152 and Expressions for details on supported numeric expressions.
153 -version
155 Show the version number of this program.
156 -v Print good directive pattern matches. However, if -dump-in‐
158 put=fail or -dump-input=always, add those matches as input anno‐
159 tations instead.
160 -vv Print information helpful in diagnosing internal FileCheck is‐
162 sues, such as discarded overlapping CHECK-DAG: matches, implicit
163 EOF pattern matches, and CHECK-NOT: patterns that do not have
164 matches. Implies -v. However, if -dump-input=fail or -dump-in‐
165 put=always, just add that information as input annotations in‐
166 stead.
167 --allow-deprecated-dag-overlap
169 Enable overlapping among matches in a group of consecutive
170 CHECK-DAG: directives. This option is deprecated and is only
171 provided for convenience as old tests are migrated to the new
172 non-overlapping CHECK-DAG: implementation.
173 --color
175 Use colors in output (autodetected by default).
176
178 If FileCheck verifies that the file matches the expected contents, it
179 exits with 0. Otherwise, if not, or if an error occurs, it will exit
180 with a non-zero value.
181
183 FileCheck is typically used from LLVM regression tests, being invoked
184 on the RUN line of the test. A simple example of using FileCheck from
185 a RUN line looks like this:
186 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s
188 This syntax says to pipe the current file ("%s") into llvm-as, pipe
190 that into llc, then pipe the output of llc into FileCheck. This means
191 that FileCheck will be verifying its standard input (the llc output)
192 against the filename argument specified (the original .ll file speci‐
193 fied by "%s"). To see how this works, let's look at the rest of the
194 .ll file (after the RUN line):
195 define void @sub1(i32* %p, i32 %v) {
197 entry:
198 ; CHECK: sub1:
199 ; CHECK: subl
200 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
201 ret void
202 }
203 define void @inc4(i64* %p) {
205 entry:
206 ; CHECK: inc4:
207 ; CHECK: incq
208 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
209 ret void
210 }
211 Here you can see some "CHECK:" lines specified in comments. Now you
213 can see how the file is piped into llvm-as, then llc, and the machine
214 code output is what we are verifying. FileCheck checks the machine
215 code output to verify that it matches what the "CHECK:" lines specify.
216 The syntax of the "CHECK:" lines is very simple: they are fixed strings
218 that must occur in order. FileCheck defaults to ignoring horizontal
219 whitespace differences (e.g. a space is allowed to match a tab) but
220 otherwise, the contents of the "CHECK:" line is required to match some
221 thing in the test file exactly.
222 One nice thing about FileCheck (compared to grep) is that it allows
224 merging test cases together into logical groups. For example, because
225 the test above is checking for the "sub1:" and "inc4:" labels, it will
226 not match unless there is a "subl" in between those labels. If it ex‐
227 isted somewhere else in the file, that would not count: "grep subl"
228 matches if "subl" exists anywhere in the file.
229 The FileCheck -check-prefix option
231 The FileCheck -check-prefix option allows multiple test configurations
232 to be driven from one .ll file. This is useful in many circumstances,
233 for example, testing different architectural variants with llc. Here's
234 a simple example:
235 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
237 ; RUN: | FileCheck %s -check-prefix=X32
238 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
239 ; RUN: | FileCheck %s -check-prefix=X64
240 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind {
242 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1
243 ret <4 x i32> %tmp1
244 ; X32: pinsrd_1:
245 ; X32: pinsrd $1, 4(%esp), %xmm0
246 ; X64: pinsrd_1:
248 ; X64: pinsrd $1, %edi, %xmm0
249 }
250 In this case, we're testing that we get the expected code generation
252 with both 32-bit and 64-bit code generation.
253 The COM: directive
255 Sometimes you want to disable a FileCheck directive without removing it
256 entirely, or you want to write comments that mention a directive by
257 name. The "COM:" directive makes it easy to do this. For example, you
258 might have:
259 ; X32: pinsrd_1:
261 ; X32: pinsrd $1, 4(%esp), %xmm0
262 ; COM: FIXME: X64 isn't working correctly yet for this part of codegen, but
264 ; COM: X64 will have something similar to X32:
265 ; COM:
266 ; COM: X64: pinsrd_1:
267 ; COM: X64: pinsrd $1, %edi, %xmm0
268 Without "COM:", you would need to use some combination of rewording and
270 directive syntax mangling to prevent FileCheck from recognizing the
271 commented occurrences of "X32:" and "X64:" above as directives. More‐
272 over, FileCheck diagnostics have been proposed that might complain
273 about the above occurrences of "X64" that don't have the trailing ":"
274 because they look like directive typos. Dodging all these problems can
275 be tedious for a test author, and directive syntax mangling can make
276 the purpose of test code unclear. "COM:" avoids all these problems.
277 A few important usage notes:
279 • "COM:" within another directive's pattern does not comment out the
281 remainder of the pattern. For example:
282 ; X32: pinsrd $1, 4(%esp), %xmm0 COM: This is part of the X32 pattern!
284 If you need to temporarily comment out part of a directive's pattern,
286 move it to another line. The reason is that FileCheck parses "COM:"
287 in the same manner as any other directive: only the first directive
288 on the line is recognized as a directive.
289 • For the sake of LIT, FileCheck treats "RUN:" just like "COM:". If
291 this is not suitable for your test environment, see
292 --comment-prefixes.
293 • FileCheck does not recognize "COM", "RUN", or any user-defined com‐
295 ment prefix as a comment directive if it's combined with one of the
296 usual check directive suffixes, such as "-NEXT:" or "-NOT:", dis‐
297 cussed below. FileCheck treats such a combination as plain text in‐
298 stead. If it needs to act as a comment directive for your test envi‐
299 ronment, define it as such with --comment-prefixes.
300 The CHECK-NEXT: directive
302 Sometimes you want to match lines and would like to verify that matches
303 happen on exactly consecutive lines with no other lines in between
304 them. In this case, you can use "CHECK:" and "CHECK-NEXT:" directives
305 to specify this. If you specified a custom check prefix, just use
306 "<PREFIX>-NEXT:". For example, something like this works as you'd ex‐
307 pect:
308 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) {
310 %tmp3 = load <2 x double>* %A, align 16
311 %tmp7 = insertelement <2 x double> undef, double %B, i32 0
312 %tmp9 = shufflevector <2 x double> %tmp3,
313 <2 x double> %tmp7,
314 <2 x i32> < i32 0, i32 2 >
315 store <2 x double> %tmp9, <2 x double>* %r, align 16
316 ret void
317 ; CHECK: t2:
319 ; CHECK: movl 8(%esp), %eax
320 ; CHECK-NEXT: movapd (%eax), %xmm0
321 ; CHECK-NEXT: movhpd 12(%esp), %xmm0
322 ; CHECK-NEXT: movl 4(%esp), %eax
323 ; CHECK-NEXT: movapd %xmm0, (%eax)
324 ; CHECK-NEXT: ret
325 }
326 "CHECK-NEXT:" directives reject the input unless there is exactly one
328 newline between it and the previous directive. A "CHECK-NEXT:" cannot
329 be the first directive in a file.
330 The CHECK-SAME: directive
332 Sometimes you want to match lines and would like to verify that matches
333 happen on the same line as the previous match. In this case, you can
334 use "CHECK:" and "CHECK-SAME:" directives to specify this. If you
335 specified a custom check prefix, just use "<PREFIX>-SAME:".
336 "CHECK-SAME:" is particularly powerful in conjunction with "CHECK-NOT:"
338 (described below).
339 For example, the following works like you'd expect:
341 !0 = !DILocation(line: 5, scope: !1, inlinedAt: !2)
343 ; CHECK: !DILocation(line: 5,
345 ; CHECK-NOT: column:
346 ; CHECK-SAME: scope: ![[SCOPE:[0-9]+]]
347 "CHECK-SAME:" directives reject the input if there are any newlines be‐
349 tween it and the previous directive. A "CHECK-SAME:" cannot be the
350 first directive in a file.
351 The CHECK-EMPTY: directive
353 If you need to check that the next line has nothing on it, not even
354 whitespace, you can use the "CHECK-EMPTY:" directive.
355 declare void @foo()
357 declare void @bar()
359 ; CHECK: foo
360 ; CHECK-EMPTY:
361 ; CHECK-NEXT: bar
362 Just like "CHECK-NEXT:" the directive will fail if there is more than
364 one newline before it finds the next blank line, and it cannot be the
365 first directive in a file.
366 The CHECK-NOT: directive
368 The "CHECK-NOT:" directive is used to verify that a string doesn't oc‐
369 cur between two matches (or before the first match, or after the last
370 match). For example, to verify that a load is removed by a transforma‐
371 tion, a test like this can be used:
372 define i8 @coerce_offset0(i32 %V, i32* %P) {
374 store i32 %V, i32* %P
375 %P2 = bitcast i32* %P to i8*
377 %P3 = getelementptr i8* %P2, i32 2
378 %A = load i8* %P3
380 ret i8 %A
381 ; CHECK: @coerce_offset0
382 ; CHECK-NOT: load
383 ; CHECK: ret i8
384 }
385 The CHECK-COUNT: directive
387 If you need to match multiple lines with the same pattern over and over
388 again you can repeat a plain CHECK: as many times as needed. If that
389 looks too boring you can instead use a counted check
390 "CHECK-COUNT-<num>:", where <num> is a positive decimal number. It will
391 match the pattern exactly <num> times, no more and no less. If you
392 specified a custom check prefix, just use "<PREFIX>-COUNT-<num>:" for
393 the same effect. Here is a simple example:
394 Loop at depth 1
396 Loop at depth 1
397 Loop at depth 1
398 Loop at depth 1
399 Loop at depth 2
400 Loop at depth 3
401 ; CHECK-COUNT-6: Loop at depth {{[0-9]+}}
403 ; CHECK-NOT: Loop at depth {{[0-9]+}}
404 The CHECK-DAG: directive
406 If it's necessary to match strings that don't occur in a strictly se‐
407 quential order, "CHECK-DAG:" could be used to verify them between two
408 matches (or before the first match, or after the last match). For exam‐
409 ple, clang emits vtable globals in reverse order. Using CHECK-DAG:, we
410 can keep the checks in the natural order:
411 // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s
413 struct Foo { virtual void method(); };
415 Foo f; // emit vtable
416 // CHECK-DAG: @_ZTV3Foo =
417 struct Bar { virtual void method(); };
419 Bar b;
420 // CHECK-DAG: @_ZTV3Bar =
421 CHECK-NOT: directives could be mixed with CHECK-DAG: directives to ex‐
423 clude strings between the surrounding CHECK-DAG: directives. As a re‐
424 sult, the surrounding CHECK-DAG: directives cannot be reordered, i.e.
425 all occurrences matching CHECK-DAG: before CHECK-NOT: must not fall be‐
426 hind occurrences matching CHECK-DAG: after CHECK-NOT:. For example,
427 ; CHECK-DAG: BEFORE
429 ; CHECK-NOT: NOT
430 ; CHECK-DAG: AFTER
431 This case will reject input strings where BEFORE occurs after AFTER.
433 With captured variables, CHECK-DAG: is able to match valid topological
435 orderings of a DAG with edges from the definition of a variable to its
436 use. It's useful, e.g., when your test cases need to match different
437 output sequences from the instruction scheduler. For example,
438 ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2
440 ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4
441 ; CHECK: mul r5, [[REG1]], [[REG2]]
442 In this case, any order of that two add instructions will be allowed.
444 If you are defining and using variables in the same CHECK-DAG: block,
446 be aware that the definition rule can match after its use.
447 So, for instance, the code below will pass:
449 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
451 ; CHECK-DAG: vmov.32 [[REG2]][1]
452 vmov.32 d0[1]
453 vmov.32 d0[0]
454 While this other code, will not:
456 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
458 ; CHECK-DAG: vmov.32 [[REG2]][1]
459 vmov.32 d1[1]
460 vmov.32 d0[0]
461 While this can be very useful, it's also dangerous, because in the case
463 of register sequence, you must have a strong order (read before write,
464 copy before use, etc). If the definition your test is looking for
465 doesn't match (because of a bug in the compiler), it may match further
466 away from the use, and mask real bugs away.
467 In those cases, to enforce the order, use a non-DAG directive between
469 DAG-blocks.
470 A CHECK-DAG: directive skips matches that overlap the matches of any
472 preceding CHECK-DAG: directives in the same CHECK-DAG: block. Not only
473 is this non-overlapping behavior consistent with other directives, but
474 it's also necessary to handle sets of non-unique strings or patterns.
475 For example, the following directives look for unordered log entries
476 for two tasks in a parallel program, such as the OpenMP runtime:
477 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin
479 // CHECK-DAG: [[THREAD_ID]]: task_end
480 //
481 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin
482 // CHECK-DAG: [[THREAD_ID]]: task_end
483 The second pair of directives is guaranteed not to match the same log
485 entries as the first pair even though the patterns are identical and
486 even if the text of the log entries is identical because the thread ID
487 manages to be reused.
488 The CHECK-LABEL: directive
490 Sometimes in a file containing multiple tests divided into logical
491 blocks, one or more CHECK: directives may inadvertently succeed by
492 matching lines in a later block. While an error will usually eventually
493 be generated, the check flagged as causing the error may not actually
494 bear any relationship to the actual source of the problem.
495 In order to produce better error messages in these cases, the
497 "CHECK-LABEL:" directive can be used. It is treated identically to a
498 normal CHECK directive except that FileCheck makes an additional as‐
499 sumption that a line matched by the directive cannot also be matched by
500 any other check present in match-filename; this is intended to be used
501 for lines containing labels or other unique identifiers. Conceptually,
502 the presence of CHECK-LABEL divides the input stream into separate
503 blocks, each of which is processed independently, preventing a CHECK:
504 directive in one block matching a line in another block. If --en‐
505 able-var-scope is in effect, all local variables are cleared at the be‐
506 ginning of the block.
507 For example,
509 define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) {
511 entry:
512 ; CHECK-LABEL: C_ctor_base:
513 ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0
514 ; CHECK: bl A_ctor_base
515 ; CHECK: mov r0, [[SAVETHIS]]
516 %0 = bitcast %struct.C* %this to %struct.A*
517 %call = tail call %struct.A* @A_ctor_base(%struct.A* %0)
518 %1 = bitcast %struct.C* %this to %struct.B*
519 %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x)
520 ret %struct.C* %this
521 }
522 define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) {
524 entry:
525 ; CHECK-LABEL: D_ctor_base:
526 The use of CHECK-LABEL: directives in this case ensures that the three
528 CHECK: directives only accept lines corresponding to the body of the
529 @C_ctor_base function, even if the patterns match lines found later in
530 the file. Furthermore, if one of these three CHECK: directives fail,
531 FileCheck will recover by continuing to the next block, allowing multi‐
532 ple test failures to be detected in a single invocation.
533 There is no requirement that CHECK-LABEL: directives contain strings
535 that correspond to actual syntactic labels in a source or output lan‐
536 guage: they must simply uniquely match a single line in the file being
537 verified.
538 CHECK-LABEL: directives cannot contain variable definitions or uses.
540 FileCheck Regex Matching Syntax
542 All FileCheck directives take a pattern to match. For most uses of
543 FileCheck, fixed string matching is perfectly sufficient. For some
544 things, a more flexible form of matching is desired. To support this,
545 FileCheck allows you to specify regular expressions in matching
546 strings, surrounded by double braces: {{yourregex}}. FileCheck imple‐
547 ments a POSIX regular expression matcher; it supports Extended POSIX
548 regular expressions (ERE). Because we want to use fixed string matching
549 for a majority of what we do, FileCheck has been designed to support
550 mixing and matching fixed string matching with regular expressions.
551 This allows you to write things like this:
552 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}}
554 In this case, any offset from the ESP register will be allowed, and any
556 xmm register will be allowed.
557 Because regular expressions are enclosed with double braces, they are
559 visually distinct, and you don't need to use escape characters within
560 the double braces like you would in C. In the rare case that you want
561 to match double braces explicitly from the input, you can use something
562 ugly like {{[}][}]}} as your pattern. Or if you are using the repeti‐
563 tion count syntax, for example [[:xdigit:]]{8} to match exactly 8 hex
564 digits, you would need to add parentheses like this
565 {{([[:xdigit:]]{8})}} to avoid confusion with FileCheck's closing dou‐
566 ble-brace.
567 FileCheck String Substitution Blocks
569 It is often useful to match a pattern and then verify that it occurs
570 again later in the file. For codegen tests, this can be useful to al‐
571 low any register, but verify that that register is used consistently
572 later. To do this, FileCheck supports string substitution blocks that
573 allow string variables to be defined and substituted into patterns.
574 Here is a simple example:
575 ; CHECK: test5:
577 ; CHECK: notw [[REGISTER:%[a-z]+]]
578 ; CHECK: andw {{.*}}[[REGISTER]]
579 The first check line matches a regex %[a-z]+ and captures it into the
581 string variable REGISTER. The second line verifies that whatever is in
582 REGISTER occurs later in the file after an "andw". FileCheck string
583 substitution blocks are always contained in [[ ]] pairs, and string
584 variable names can be formed with the regex [a-zA-Z_][a-zA-Z0-9_]*. If
585 a colon follows the name, then it is a definition of the variable; oth‐
586 erwise, it is a substitution.
587 FileCheck variables can be defined multiple times, and substitutions
589 always get the latest value. Variables can also be substituted later
590 on the same line they were defined on. For example:
591 ; CHECK: op [[REG:r[0-9]+]], [[REG]]
593 Can be useful if you want the operands of op to be the same register,
595 and don't care exactly which register it is.
596 If --enable-var-scope is in effect, variables with names that start
598 with $ are considered to be global. All others variables are local.
599 All local variables get undefined at the beginning of each CHECK-LABEL
600 block. Global variables are not affected by CHECK-LABEL. This makes it
601 easier to ensure that individual tests are not affected by variables
602 set in preceding tests.
603 FileCheck Numeric Substitution Blocks
605 FileCheck also supports numeric substitution blocks that allow defining
606 numeric variables and checking for numeric values that satisfy a nu‐
607 meric expression constraint based on those variables via a numeric sub‐
608 stitution. This allows CHECK: directives to verify a numeric relation
609 between two numbers, such as the need for consecutive registers to be
610 used.
611 The syntax to define a numeric variable is [[#%<fmtspec>,<NUMVAR>:]]
613 where:
614 • %<fmtspec> is an optional scanf-style matching format specifier to
616 indicate what number format to match (e.g. hex number). Currently
617 accepted format specifiers are %u, %d, %x and %X. If absent, the
618 format specifier defaults to %u.
619 • <NUMVAR> is the name of the numeric variable to define to the match‐
621 ing value.
622 For example:
624 ; CHECK: mov r[[#REG:]], 0x[[#%X,IMM:]]
626 would match mov r5, 0xF0F0 and set REG to the value 5 and IMM to the
628 value 0xF0F0.
629 The syntax of a numeric substitution is [[#%<fmtspec>: <constraint>
631 <expr>]] where:
632 • %<fmtspec> is the same matching format specifier as for defining nu‐
634 meric variables but acting as a printf-style format to indicate how a
635 numeric expression value should be matched against. If absent, the
636 format specifier is inferred from the matching format of the numeric
637 variable(s) used by the expression constraint if any, and defaults to
638 %u if no numeric variable is used. In case of conflict between
639 matching formats of several numeric variables the format specifier is
640 mandatory.
641 • <constraint> is the constraint describing how the value to match must
643 relate to the value of the numeric expression. The only currently ac‐
644 cepted constraint is == for an exact match and is the default if
645 <constraint> is not provided. No matching constraint must be speci‐
646 fied when the <expr> is empty.
647 • <expr> is an expression. An expression is in turn recursively defined
649 as:
650 • a numeric operand, or
652 • an expression followed by an operator and a numeric operand.
654 A numeric operand is a previously defined numeric variable, an inte‐
656 ger literal, or a function. Spaces are accepted before, after and be‐
657 tween any of these elements. Numeric operands have 64-bit precision.
658 Overflow and underflow are rejected. There is no support for operator
659 precedence, but parentheses can be used to change the evaluation or‐
660 der.
661 The supported operators are:
663 • + - Returns the sum of its two operands.
665 • - - Returns the difference of its two operands.
667 The syntax of a function call is <name>(<arguments>) where:
669 • name is a predefined string literal. Accepted values are:
671 • add - Returns the sum of its two operands.
673 • div - Returns the quotient of its two operands.
675 • max - Returns the largest of its two operands.
677 • min - Returns the smallest of its two operands.
679 • mul - Returns the product of its two operands.
681 • sub - Returns the difference of its two operands.
683 • <arguments> is a comma separated list of expressions.
685 For example:
687 ; CHECK: load r[[#REG:]], [r0]
689 ; CHECK: load r[[#REG+1]], [r1]
690 ; CHECK: Loading from 0x[[#%x,ADDR:]]
691 ; CHECK-SAME: to 0x[[#ADDR + 7]]
692 The above example would match the text:
694 load r5, [r0]
696 load r6, [r1]
697 Loading from 0xa0463440 to 0xa0463447
698 but would not match the text:
700 load r5, [r0]
702 load r7, [r1]
703 Loading from 0xa0463440 to 0xa0463443
704 Due to 7 being unequal to 5 + 1 and a0463443 being unequal to a0463440
706 + 7.
707 The syntax also supports an empty expression, equivalent to writing
709 {{[0-9]+}}, for cases where the input must contain a numeric value but
710 the value itself does not matter:
711 ; CHECK-NOT: mov r0, r[[#]]
713 to check that a value is synthesized rather than moved around.
715 A numeric variable can also be defined to the result of a numeric ex‐
717 pression, in which case the numeric expression constraint is checked
718 and if verified the variable is assigned to the value. The unified syn‐
719 tax for both defining numeric variables and checking a numeric expres‐
720 sion is thus [[#%<fmtspec>,<NUMVAR>: <constraint> <expr>]] with each
721 element as described previously. One can use this syntax to make a
722 testcase more self-describing by using variables instead of values:
723 ; CHECK: mov r[[#REG_OFFSET:]], 0x[[#%X,FIELD_OFFSET:12]]
725 ; CHECK-NEXT: load r[[#]], [r[[#REG_BASE:]], r[[#REG_OFFSET]]]
726 which would match:
728 mov r4, 0xC
730 load r6, [r5, r4]
731 The --enable-var-scope option has the same effect on numeric variables
733 as on string variables.
734 Important note: In its current implementation, an expression cannot use
736 a numeric variable defined earlier in the same CHECK directive.
737 FileCheck Pseudo Numeric Variables
739 Sometimes there's a need to verify output that contains line numbers of
740 the match file, e.g. when testing compiler diagnostics. This intro‐
741 duces a certain fragility of the match file structure, as "CHECK:"
742 lines contain absolute line numbers in the same file, which have to be
743 updated whenever line numbers change due to text addition or deletion.
744 To support this case, FileCheck expressions understand the @LINE pseudo
746 numeric variable which evaluates to the line number of the CHECK pat‐
747 tern where it is found.
748 This way match patterns can be put near the relevant test lines and in‐
750 clude relative line number references, for example:
751 // CHECK: test.cpp:[[# @LINE + 4]]:6: error: expected ';' after top level declarator
753 // CHECK-NEXT: {{^int a}}
754 // CHECK-NEXT: {{^ \^}}
755 // CHECK-NEXT: {{^ ;}}
756 int a
757 To support legacy uses of @LINE as a special string variable, FileCheck
759 also accepts the following uses of @LINE with string substitution block
760 syntax: [[@LINE]], [[@LINE+<offset>]] and [[@LINE-<offset>]] without
761 any spaces inside the brackets and where offset is an integer.
762 Matching Newline Characters
764 To match newline characters in regular expressions the character class
765 [[:space:]] can be used. For example, the following pattern:
766 // CHECK: DW_AT_location [DW_FORM_sec_offset] ([[DLOC:0x[0-9a-f]+]]){{[[:space:]].*}}"intd"
768 matches output of the form (from llvm-dwarfdump):
770 DW_AT_location [DW_FORM_sec_offset] (0x00000233)
772 DW_AT_name [DW_FORM_strp] ( .debug_str[0x000000c9] = "intd")
773 letting us set the FileCheck variable DLOC to the desired value
775 0x00000233, extracted from the line immediately preceding "intd".
776
778 Maintained by the LLVM Team (https://llvm.org/).
779
781 2003-2021, LLVM Project
78211 2021-03-23 FILECHECK(1)