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 always read from standard input.
22
23OPTIONS
24-------
25
26.. option:: -help
27
28 Print a summary of command line options.
29
30.. option:: --check-prefix prefix
31
32 FileCheck searches the contents of ``match-filename`` for patterns to match.
33 By default, these patterns are prefixed with "``CHECK:``".  If you'd like to
34 use a different prefix (e.g. because the same input file is checking multiple
35 different tool or options), the :option:`--check-prefix` argument allows you
36 to specify a specific prefix to match.
37
38.. option:: --input-file filename
39
40  File to check (defaults to stdin).
41
42.. option:: --strict-whitespace
43
44 By default, FileCheck canonicalizes input horizontal whitespace (spaces and
45 tabs) which causes it to ignore these differences (a space will match a tab).
46 The :option:`--strict-whitespace` argument disables this behavior. End-of-line
47 sequences are canonicalized to UNIX-style ``\n`` in all modes.
48
49.. option:: -version
50
51 Show the version number of this program.
52
53EXIT STATUS
54-----------
55
56If :program:`FileCheck` verifies that the file matches the expected contents,
57it exits with 0.  Otherwise, if not, or if an error occurs, it will exit with a
58non-zero value.
59
60TUTORIAL
61--------
62
63FileCheck is typically used from LLVM regression tests, being invoked on the RUN
64line of the test.  A simple example of using FileCheck from a RUN line looks
65like this:
66
67.. code-block:: llvm
68
69   ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s
70
71This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe
72that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``.  This
73means that FileCheck will be verifying its standard input (the llc output)
74against the filename argument specified (the original ``.ll`` file specified by
75"``%s``").  To see how this works, let's look at the rest of the ``.ll`` file
76(after the RUN line):
77
78.. code-block:: llvm
79
80   define void @sub1(i32* %p, i32 %v) {
81   entry:
82   ; CHECK: sub1:
83   ; CHECK: subl
84           %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
85           ret void
86   }
87
88   define void @inc4(i64* %p) {
89   entry:
90   ; CHECK: inc4:
91   ; CHECK: incq
92           %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
93           ret void
94   }
95
96Here you can see some "``CHECK:``" lines specified in comments.  Now you can
97see how the file is piped into ``llvm-as``, then ``llc``, and the machine code
98output is what we are verifying.  FileCheck checks the machine code output to
99verify that it matches what the "``CHECK:``" lines specify.
100
101The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that
102must occur in order.  FileCheck defaults to ignoring horizontal whitespace
103differences (e.g. a space is allowed to match a tab) but otherwise, the contents
104of the "``CHECK:``" line is required to match some thing in the test file exactly.
105
106One nice thing about FileCheck (compared to grep) is that it allows merging
107test cases together into logical groups.  For example, because the test above
108is checking for the "``sub1:``" and "``inc4:``" labels, it will not match
109unless there is a "``subl``" in between those labels.  If it existed somewhere
110else in the file, that would not count: "``grep subl``" matches if "``subl``"
111exists anywhere in the file.
112
113The FileCheck -check-prefix option
114~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
115
116The FileCheck :option:`-check-prefix` option allows multiple test
117configurations to be driven from one `.ll` file.  This is useful in many
118circumstances, for example, testing different architectural variants with
119:program:`llc`.  Here's a simple example:
120
121.. code-block:: llvm
122
123   ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
124   ; RUN:              | FileCheck %s -check-prefix=X32
125   ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
126   ; RUN:              | FileCheck %s -check-prefix=X64
127
128   define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind {
129           %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1
130           ret <4 x i32> %tmp1
131   ; X32: pinsrd_1:
132   ; X32:    pinsrd $1, 4(%esp), %xmm0
133
134   ; X64: pinsrd_1:
135   ; X64:    pinsrd $1, %edi, %xmm0
136   }
137
138In this case, we're testing that we get the expected code generation with
139both 32-bit and 64-bit code generation.
140
141The "CHECK-NEXT:" directive
142~~~~~~~~~~~~~~~~~~~~~~~~~~~
143
144Sometimes you want to match lines and would like to verify that matches
145happen on exactly consecutive lines with no other lines in between them.  In
146this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify
147this.  If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``".
148For example, something like this works as you'd expect:
149
150.. code-block:: llvm
151
152   define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) {
153 	%tmp3 = load <2 x double>* %A, align 16
154 	%tmp7 = insertelement <2 x double> undef, double %B, i32 0
155 	%tmp9 = shufflevector <2 x double> %tmp3,
156                               <2 x double> %tmp7,
157                               <2 x i32> < i32 0, i32 2 >
158 	store <2 x double> %tmp9, <2 x double>* %r, align 16
159 	ret void
160
161   ; CHECK:          t2:
162   ; CHECK: 	        movl	8(%esp), %eax
163   ; CHECK-NEXT: 	movapd	(%eax), %xmm0
164   ; CHECK-NEXT: 	movhpd	12(%esp), %xmm0
165   ; CHECK-NEXT: 	movl	4(%esp), %eax
166   ; CHECK-NEXT: 	movapd	%xmm0, (%eax)
167   ; CHECK-NEXT: 	ret
168   }
169
170"``CHECK-NEXT:``" directives reject the input unless there is exactly one
171newline between it and the previous directive.  A "``CHECK-NEXT:``" cannot be
172the first directive in a file.
173
174The "CHECK-NOT:" directive
175~~~~~~~~~~~~~~~~~~~~~~~~~~
176
177The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur
178between two matches (or before the first match, or after the last match).  For
179example, to verify that a load is removed by a transformation, a test like this
180can be used:
181
182.. code-block:: llvm
183
184   define i8 @coerce_offset0(i32 %V, i32* %P) {
185     store i32 %V, i32* %P
186
187     %P2 = bitcast i32* %P to i8*
188     %P3 = getelementptr i8* %P2, i32 2
189
190     %A = load i8* %P3
191     ret i8 %A
192   ; CHECK: @coerce_offset0
193   ; CHECK-NOT: load
194   ; CHECK: ret i8
195   }
196
197The "CHECK-DAG:" directive
198~~~~~~~~~~~~~~~~~~~~~~~~~~
199
200If it's necessary to match strings that don't occur in a strictly sequential
201order, "``CHECK-DAG:``" could be used to verify them between two matches (or
202before the first match, or after the last match). For example, clang emits
203vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks
204in the natural order:
205
206.. code-block:: c++
207
208    // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s
209
210    struct Foo { virtual void method(); };
211    Foo f;  // emit vtable
212    // CHECK-DAG: @_ZTV3Foo =
213
214    struct Bar { virtual void method(); };
215    Bar b;
216    // CHECK-DAG: @_ZTV3Bar =
217
218
219With captured variables, ``CHECK-DAG:`` is able to match valid topological
220orderings of a DAG with edges from the definition of a variable to its use.
221It's useful, e.g., when your test cases need to match different output
222sequences from the instruction scheduler. For example,
223
224.. code-block:: llvm
225
226   ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2
227   ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4
228   ; CHECK:     mul r5, [[REG1]], [[REG2]]
229
230In this case, any order of that two ``add`` instructions will be allowed.
231
232``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to
233exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result,
234the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all
235occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind
236occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example,
237
238.. code-block:: llvm
239
240   ; CHECK-DAG: BEFORE
241   ; CHECK-NOT: NOT
242   ; CHECK-DAG: AFTER
243
244This case will reject input strings where ``BEFORE`` occurs after ``AFTER``.
245
246FileCheck Pattern Matching Syntax
247~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
248
249The "``CHECK:``" and "``CHECK-NOT:``" directives both take a pattern to match.
250For most uses of FileCheck, fixed string matching is perfectly sufficient.  For
251some things, a more flexible form of matching is desired.  To support this,
252FileCheck allows you to specify regular expressions in matching strings,
253surrounded by double braces: ``{{yourregex}}``.  Because we want to use fixed
254string matching for a majority of what we do, FileCheck has been designed to
255support mixing and matching fixed string matching with regular expressions.
256This allows you to write things like this:
257
258.. code-block:: llvm
259
260   ; CHECK: movhpd	{{[0-9]+}}(%esp), {{%xmm[0-7]}}
261
262In this case, any offset from the ESP register will be allowed, and any xmm
263register will be allowed.
264
265Because regular expressions are enclosed with double braces, they are
266visually distinct, and you don't need to use escape characters within the double
267braces like you would in C.  In the rare case that you want to match double
268braces explicitly from the input, you can use something ugly like
269``{{[{][{]}}`` as your pattern.
270
271FileCheck Variables
272~~~~~~~~~~~~~~~~~~~
273
274It is often useful to match a pattern and then verify that it occurs again
275later in the file.  For codegen tests, this can be useful to allow any register,
276but verify that that register is used consistently later.  To do this,
277:program:`FileCheck` allows named variables to be defined and substituted into
278patterns.  Here is a simple example:
279
280.. code-block:: llvm
281
282   ; CHECK: test5:
283   ; CHECK:    notw	[[REGISTER:%[a-z]+]]
284   ; CHECK:    andw	{{.*}}[[REGISTER]]
285
286The first check line matches a regex ``%[a-z]+`` and captures it into the
287variable ``REGISTER``.  The second line verifies that whatever is in
288``REGISTER`` occurs later in the file after an "``andw``".  :program:`FileCheck`
289variable references are always contained in ``[[ ]]`` pairs, and their names can
290be formed with the regex ``[a-zA-Z][a-zA-Z0-9]*``.  If a colon follows the name,
291then it is a definition of the variable; otherwise, it is a use.
292
293:program:`FileCheck` variables can be defined multiple times, and uses always
294get the latest value.  Variables can also be used later on the same line they
295were defined on. For example:
296
297.. code-block:: llvm
298
299    ; CHECK: op [[REG:r[0-9]+]], [[REG]]
300
301Can be useful if you want the operands of ``op`` to be the same register,
302and don't care exactly which register it is.
303
304FileCheck Expressions
305~~~~~~~~~~~~~~~~~~~~~
306
307Sometimes there's a need to verify output which refers line numbers of the
308match file, e.g. when testing compiler diagnostics.  This introduces a certain
309fragility of the match file structure, as "``CHECK:``" lines contain absolute
310line numbers in the same file, which have to be updated whenever line numbers
311change due to text addition or deletion.
312
313To support this case, FileCheck allows using ``[[@LINE]]``,
314``[[@LINE+<offset>]]``, ``[[@LINE-<offset>]]`` expressions in patterns. These
315expressions expand to a number of the line where a pattern is located (with an
316optional integer offset).
317
318This way match patterns can be put near the relevant test lines and include
319relative line number references, for example:
320
321.. code-block:: c++
322
323   // CHECK: test.cpp:[[@LINE+4]]:6: error: expected ';' after top level declarator
324   // CHECK-NEXT: {{^int a}}
325   // CHECK-NEXT: {{^     \^}}
326   // CHECK-NEXT: {{^     ;}}
327   int a
328
329