1FileCheck - Flexible pattern matching file verifier
2===================================================
3
4SYNOPSIS
5--------
6
7:program:`FileCheck` *match-filename* [*--check-prefix=XXX*] [*--strict-whitespace*]
8
9DESCRIPTION
10-----------
11
12:program:`FileCheck` reads two files (one from standard input, and one
13specified on the command line) and uses one to verify the other.  This
14behavior is particularly useful for the testsuite, which wants to verify that
15the 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
17using :program:`grep`, but it is optimized for matching multiple different
18inputs in one file in a specific order.
19
20The ``match-filename`` file specifies the file that contains the patterns to
21match.  The file to verify is read from standard input unless the
22:option:`--input-file` option is used.
23
24OPTIONS
25-------
26
27.. option:: -help
28
29 Print a summary of command line options.
30
31.. option:: --check-prefix prefix
32
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.
40
41.. option:: --check-prefixes prefix1,prefix2,...
42
43 An alias of :option:`--check-prefix` that allows multiple prefixes to be
44 specified as a comma separated list.
45
46.. option:: --input-file filename
47
48  File to check (defaults to stdin).
49
50.. option:: --match-full-lines
51
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!)
57
58 Passing this option is equivalent to inserting ``{{^ *}}`` or
59 ``{{^}}`` before, and ``{{ *$}}`` or ``{{$}}`` after every positive
60 check pattern.
61
62.. option:: --strict-whitespace
63
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.
68
69.. option:: --implicit-check-not check-pattern
70
71  Adds implicit negative checks for the specified patterns between positive
72  checks. The option allows writing stricter tests without stuffing them with
73  ``CHECK-NOT``\ s.
74
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.
79
80.. option:: --enable-var-scope
81
82  Enables scope for regex variables.
83
84  Variables with names that start with ``$`` are considered global and
85  remain set throughout the file.
86
87  All other variables get undefined after each encountered ``CHECK-LABEL``.
88
89.. option:: -version
90
91 Show the version number of this program.
92
93EXIT STATUS
94-----------
95
96If :program:`FileCheck` verifies that the file matches the expected contents,
97it exits with 0.  Otherwise, if not, or if an error occurs, it will exit with a
98non-zero value.
99
100TUTORIAL
101--------
102
103FileCheck is typically used from LLVM regression tests, being invoked on the RUN
104line of the test.  A simple example of using FileCheck from a RUN line looks
105like this:
106
107.. code-block:: llvm
108
109   ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s
110
111This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe
112that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``.  This
113means that FileCheck will be verifying its standard input (the llc output)
114against the filename argument specified (the original ``.ll`` file specified by
115"``%s``").  To see how this works, let's look at the rest of the ``.ll`` file
116(after the RUN line):
117
118.. code-block:: llvm
119
120   define void @sub1(i32* %p, i32 %v) {
121   entry:
122   ; CHECK: sub1:
123   ; CHECK: subl
124           %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
125           ret void
126   }
127
128   define void @inc4(i64* %p) {
129   entry:
130   ; CHECK: inc4:
131   ; CHECK: incq
132           %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
133           ret void
134   }
135
136Here you can see some "``CHECK:``" lines specified in comments.  Now you can
137see how the file is piped into ``llvm-as``, then ``llc``, and the machine code
138output is what we are verifying.  FileCheck checks the machine code output to
139verify that it matches what the "``CHECK:``" lines specify.
140
141The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that
142must occur in order.  FileCheck defaults to ignoring horizontal whitespace
143differences (e.g. a space is allowed to match a tab) but otherwise, the contents
144of the "``CHECK:``" line is required to match some thing in the test file exactly.
145
146One nice thing about FileCheck (compared to grep) is that it allows merging
147test cases together into logical groups.  For example, because the test above
148is checking for the "``sub1:``" and "``inc4:``" labels, it will not match
149unless there is a "``subl``" in between those labels.  If it existed somewhere
150else in the file, that would not count: "``grep subl``" matches if "``subl``"
151exists anywhere in the file.
152
153The FileCheck -check-prefix option
154~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
155
156The FileCheck `-check-prefix` option allows multiple test
157configurations to be driven from one `.ll` file.  This is useful in many
158circumstances, for example, testing different architectural variants with
159:program:`llc`.  Here's a simple example:
160
161.. code-block:: llvm
162
163   ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
164   ; RUN:              | FileCheck %s -check-prefix=X32
165   ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
166   ; RUN:              | FileCheck %s -check-prefix=X64
167
168   define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind {
169           %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1
170           ret <4 x i32> %tmp1
171   ; X32: pinsrd_1:
172   ; X32:    pinsrd $1, 4(%esp), %xmm0
173
174   ; X64: pinsrd_1:
175   ; X64:    pinsrd $1, %edi, %xmm0
176   }
177
178In this case, we're testing that we get the expected code generation with
179both 32-bit and 64-bit code generation.
180
181The "CHECK-NEXT:" directive
182~~~~~~~~~~~~~~~~~~~~~~~~~~~
183
184Sometimes you want to match lines and would like to verify that matches
185happen on exactly consecutive lines with no other lines in between them.  In
186this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify
187this.  If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``".
188For example, something like this works as you'd expect:
189
190.. code-block:: llvm
191
192   define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) {
193 	%tmp3 = load <2 x double>* %A, align 16
194 	%tmp7 = insertelement <2 x double> undef, double %B, i32 0
195 	%tmp9 = shufflevector <2 x double> %tmp3,
196                               <2 x double> %tmp7,
197                               <2 x i32> < i32 0, i32 2 >
198 	store <2 x double> %tmp9, <2 x double>* %r, align 16
199 	ret void
200
201   ; CHECK:          t2:
202   ; CHECK: 	        movl	8(%esp), %eax
203   ; CHECK-NEXT: 	movapd	(%eax), %xmm0
204   ; CHECK-NEXT: 	movhpd	12(%esp), %xmm0
205   ; CHECK-NEXT: 	movl	4(%esp), %eax
206   ; CHECK-NEXT: 	movapd	%xmm0, (%eax)
207   ; CHECK-NEXT: 	ret
208   }
209
210"``CHECK-NEXT:``" directives reject the input unless there is exactly one
211newline between it and the previous directive.  A "``CHECK-NEXT:``" cannot be
212the first directive in a file.
213
214The "CHECK-SAME:" directive
215~~~~~~~~~~~~~~~~~~~~~~~~~~~
216
217Sometimes you want to match lines and would like to verify that matches happen
218on the same line as the previous match.  In this case, you can use "``CHECK:``"
219and "``CHECK-SAME:``" directives to specify this.  If you specified a custom
220check prefix, just use "``<PREFIX>-SAME:``".
221
222"``CHECK-SAME:``" is particularly powerful in conjunction with "``CHECK-NOT:``"
223(described below).
224
225For example, the following works like you'd expect:
226
227.. code-block:: llvm
228
229   !0 = !DILocation(line: 5, scope: !1, inlinedAt: !2)
230
231   ; CHECK:       !DILocation(line: 5,
232   ; CHECK-NOT:               column:
233   ; CHECK-SAME:              scope: ![[SCOPE:[0-9]+]]
234
235"``CHECK-SAME:``" directives reject the input if there are any newlines between
236it and the previous directive.  A "``CHECK-SAME:``" cannot be the first
237directive in a file.
238
239The "CHECK-NOT:" directive
240~~~~~~~~~~~~~~~~~~~~~~~~~~
241
242The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur
243between two matches (or before the first match, or after the last match).  For
244example, to verify that a load is removed by a transformation, a test like this
245can be used:
246
247.. code-block:: llvm
248
249   define i8 @coerce_offset0(i32 %V, i32* %P) {
250     store i32 %V, i32* %P
251
252     %P2 = bitcast i32* %P to i8*
253     %P3 = getelementptr i8* %P2, i32 2
254
255     %A = load i8* %P3
256     ret i8 %A
257   ; CHECK: @coerce_offset0
258   ; CHECK-NOT: load
259   ; CHECK: ret i8
260   }
261
262The "CHECK-DAG:" directive
263~~~~~~~~~~~~~~~~~~~~~~~~~~
264
265If it's necessary to match strings that don't occur in a strictly sequential
266order, "``CHECK-DAG:``" could be used to verify them between two matches (or
267before the first match, or after the last match). For example, clang emits
268vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks
269in the natural order:
270
271.. code-block:: c++
272
273    // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s
274
275    struct Foo { virtual void method(); };
276    Foo f;  // emit vtable
277    // CHECK-DAG: @_ZTV3Foo =
278
279    struct Bar { virtual void method(); };
280    Bar b;
281    // CHECK-DAG: @_ZTV3Bar =
282
283``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to
284exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result,
285the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all
286occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind
287occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example,
288
289.. code-block:: llvm
290
291   ; CHECK-DAG: BEFORE
292   ; CHECK-NOT: NOT
293   ; CHECK-DAG: AFTER
294
295This case will reject input strings where ``BEFORE`` occurs after ``AFTER``.
296
297With captured variables, ``CHECK-DAG:`` is able to match valid topological
298orderings of a DAG with edges from the definition of a variable to its use.
299It's useful, e.g., when your test cases need to match different output
300sequences from the instruction scheduler. For example,
301
302.. code-block:: llvm
303
304   ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2
305   ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4
306   ; CHECK:     mul r5, [[REG1]], [[REG2]]
307
308In this case, any order of that two ``add`` instructions will be allowed.
309
310If you are defining `and` using variables in the same ``CHECK-DAG:`` block,
311be aware that the definition rule can match `after` its use.
312
313So, for instance, the code below will pass:
314
315.. code-block:: text
316
317  ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
318  ; CHECK-DAG: vmov.32 [[REG2]][1]
319  vmov.32 d0[1]
320  vmov.32 d0[0]
321
322While this other code, will not:
323
324.. code-block:: text
325
326  ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0]
327  ; CHECK-DAG: vmov.32 [[REG2]][1]
328  vmov.32 d1[1]
329  vmov.32 d0[0]
330
331While this can be very useful, it's also dangerous, because in the case of
332register sequence, you must have a strong order (read before write, copy before
333use, etc). If the definition your test is looking for doesn't match (because
334of a bug in the compiler), it may match further away from the use, and mask
335real bugs away.
336
337In those cases, to enforce the order, use a non-DAG directive between DAG-blocks.
338
339The "CHECK-LABEL:" directive
340~~~~~~~~~~~~~~~~~~~~~~~~~~~~
341
342Sometimes in a file containing multiple tests divided into logical blocks, one
343or more ``CHECK:`` directives may inadvertently succeed by matching lines in a
344later block. While an error will usually eventually be generated, the check
345flagged as causing the error may not actually bear any relationship to the
346actual source of the problem.
347
348In order to produce better error messages in these cases, the "``CHECK-LABEL:``"
349directive can be used. It is treated identically to a normal ``CHECK``
350directive except that FileCheck makes an additional assumption that a line
351matched by the directive cannot also be matched by any other check present in
352``match-filename``; this is intended to be used for lines containing labels or
353other unique identifiers. Conceptually, the presence of ``CHECK-LABEL`` divides
354the input stream into separate blocks, each of which is processed independently,
355preventing a ``CHECK:`` directive in one block matching a line in another block.
356If ``--enable-var-scope`` is in effect, all local variables are cleared at the
357beginning of the block.
358
359For example,
360
361.. code-block:: llvm
362
363  define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) {
364  entry:
365  ; CHECK-LABEL: C_ctor_base:
366  ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0
367  ; CHECK: bl A_ctor_base
368  ; CHECK: mov r0, [[SAVETHIS]]
369    %0 = bitcast %struct.C* %this to %struct.A*
370    %call = tail call %struct.A* @A_ctor_base(%struct.A* %0)
371    %1 = bitcast %struct.C* %this to %struct.B*
372    %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x)
373    ret %struct.C* %this
374  }
375
376  define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) {
377  entry:
378  ; CHECK-LABEL: D_ctor_base:
379
380The use of ``CHECK-LABEL:`` directives in this case ensures that the three
381``CHECK:`` directives only accept lines corresponding to the body of the
382``@C_ctor_base`` function, even if the patterns match lines found later in
383the file. Furthermore, if one of these three ``CHECK:`` directives fail,
384FileCheck will recover by continuing to the next block, allowing multiple test
385failures to be detected in a single invocation.
386
387There is no requirement that ``CHECK-LABEL:`` directives contain strings that
388correspond to actual syntactic labels in a source or output language: they must
389simply uniquely match a single line in the file being verified.
390
391``CHECK-LABEL:`` directives cannot contain variable definitions or uses.
392
393FileCheck Pattern Matching Syntax
394~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
395
396All FileCheck directives take a pattern to match.
397For most uses of FileCheck, fixed string matching is perfectly sufficient.  For
398some things, a more flexible form of matching is desired.  To support this,
399FileCheck allows you to specify regular expressions in matching strings,
400surrounded by double braces: ``{{yourregex}}``. FileCheck implements a POSIX
401regular expression matcher; it supports Extended POSIX regular expressions
402(ERE). Because we want to use fixed string matching for a majority of what we
403do, FileCheck has been designed to support mixing and matching fixed string
404matching with regular expressions.  This allows you to write things like this:
405
406.. code-block:: llvm
407
408   ; CHECK: movhpd	{{[0-9]+}}(%esp), {{%xmm[0-7]}}
409
410In this case, any offset from the ESP register will be allowed, and any xmm
411register will be allowed.
412
413Because regular expressions are enclosed with double braces, they are
414visually distinct, and you don't need to use escape characters within the double
415braces like you would in C.  In the rare case that you want to match double
416braces explicitly from the input, you can use something ugly like
417``{{[{][{]}}`` as your pattern.
418
419FileCheck Variables
420~~~~~~~~~~~~~~~~~~~
421
422It is often useful to match a pattern and then verify that it occurs again
423later in the file.  For codegen tests, this can be useful to allow any register,
424but verify that that register is used consistently later.  To do this,
425:program:`FileCheck` allows named variables to be defined and substituted into
426patterns.  Here is a simple example:
427
428.. code-block:: llvm
429
430   ; CHECK: test5:
431   ; CHECK:    notw	[[REGISTER:%[a-z]+]]
432   ; CHECK:    andw	{{.*}}[[REGISTER]]
433
434The first check line matches a regex ``%[a-z]+`` and captures it into the
435variable ``REGISTER``.  The second line verifies that whatever is in
436``REGISTER`` occurs later in the file after an "``andw``".  :program:`FileCheck`
437variable references are always contained in ``[[ ]]`` pairs, and their names can
438be formed with the regex ``[a-zA-Z_][a-zA-Z0-9_]*``.  If a colon follows the name,
439then it is a definition of the variable; otherwise, it is a use.
440
441:program:`FileCheck` variables can be defined multiple times, and uses always
442get the latest value.  Variables can also be used later on the same line they
443were defined on. For example:
444
445.. code-block:: llvm
446
447    ; CHECK: op [[REG:r[0-9]+]], [[REG]]
448
449Can be useful if you want the operands of ``op`` to be the same register,
450and don't care exactly which register it is.
451
452If ``--enable-var-scope`` is in effect, variables with names that
453start with ``$`` are considered to be global. All others variables are
454local.  All local variables get undefined at the beginning of each
455CHECK-LABEL block. Global variables are not affected by CHECK-LABEL.
456This makes it easier to ensure that individual tests are not affected
457by variables set in preceding tests.
458
459FileCheck Expressions
460~~~~~~~~~~~~~~~~~~~~~
461
462Sometimes there's a need to verify output which refers line numbers of the
463match file, e.g. when testing compiler diagnostics.  This introduces a certain
464fragility of the match file structure, as "``CHECK:``" lines contain absolute
465line numbers in the same file, which have to be updated whenever line numbers
466change due to text addition or deletion.
467
468To support this case, FileCheck allows using ``[[@LINE]]``,
469``[[@LINE+<offset>]]``, ``[[@LINE-<offset>]]`` expressions in patterns. These
470expressions expand to a number of the line where a pattern is located (with an
471optional integer offset).
472
473This way match patterns can be put near the relevant test lines and include
474relative line number references, for example:
475
476.. code-block:: c++
477
478   // CHECK: test.cpp:[[@LINE+4]]:6: error: expected ';' after top level declarator
479   // CHECK-NEXT: {{^int a}}
480   // CHECK-NEXT: {{^     \^}}
481   // CHECK-NEXT: {{^     ;}}
482   int a
483
484Matching Newline Characters
485~~~~~~~~~~~~~~~~~~~~~~~~~~~
486
487To match newline characters in regular expressions the character class
488``[[:space:]]`` can be used. For example, the following pattern:
489
490.. code-block:: c++
491
492   // CHECK: DW_AT_location [DW_FORM_sec_offset] ([[DLOC:0x[0-9a-f]+]]){{[[:space:]].*}}"intd"
493
494matches output of the form (from llvm-dwarfdump):
495
496.. code-block:: text
497
498       DW_AT_location [DW_FORM_sec_offset]   (0x00000233)
499       DW_AT_name [DW_FORM_strp]  ( .debug_str[0x000000c9] = "intd")
500
501letting us set the :program:`FileCheck` variable ``DLOC`` to the desired value
502``0x00000233``, extracted from the line immediately preceding "``intd``".
503