1 FileCheck - Flexible pattern matching file verifier 2 =================================================== 3 4 SYNOPSIS 5 -------- 6 7 :program:`FileCheck` *match-filename* [*--check-prefix=XXX*] [*--strict-whitespace*] 8 9 DESCRIPTION 10 ----------- 11 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. 19 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. 23 24 OPTIONS 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:: --input-file filename 42 43 File to check (defaults to stdin). 44 45 .. option:: --strict-whitespace 46 47 By default, FileCheck canonicalizes input horizontal whitespace (spaces and 48 tabs) which causes it to ignore these differences (a space will match a tab). 49 The :option:`--strict-whitespace` argument disables this behavior. End-of-line 50 sequences are canonicalized to UNIX-style ``\n`` in all modes. 51 52 .. option:: -version 53 54 Show the version number of this program. 55 56 EXIT STATUS 57 ----------- 58 59 If :program:`FileCheck` verifies that the file matches the expected contents, 60 it exits with 0. Otherwise, if not, or if an error occurs, it will exit with a 61 non-zero value. 62 63 TUTORIAL 64 -------- 65 66 FileCheck is typically used from LLVM regression tests, being invoked on the RUN 67 line of the test. A simple example of using FileCheck from a RUN line looks 68 like this: 69 70 .. code-block:: llvm 71 72 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s 73 74 This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe 75 that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``. This 76 means that FileCheck will be verifying its standard input (the llc output) 77 against the filename argument specified (the original ``.ll`` file specified by 78 "``%s``"). To see how this works, let's look at the rest of the ``.ll`` file 79 (after the RUN line): 80 81 .. code-block:: llvm 82 83 define void @sub1(i32* %p, i32 %v) { 84 entry: 85 ; CHECK: sub1: 86 ; CHECK: subl 87 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v) 88 ret void 89 } 90 91 define void @inc4(i64* %p) { 92 entry: 93 ; CHECK: inc4: 94 ; CHECK: incq 95 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1) 96 ret void 97 } 98 99 Here you can see some "``CHECK:``" lines specified in comments. Now you can 100 see how the file is piped into ``llvm-as``, then ``llc``, and the machine code 101 output is what we are verifying. FileCheck checks the machine code output to 102 verify that it matches what the "``CHECK:``" lines specify. 103 104 The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that 105 must occur in order. FileCheck defaults to ignoring horizontal whitespace 106 differences (e.g. a space is allowed to match a tab) but otherwise, the contents 107 of the "``CHECK:``" line is required to match some thing in the test file exactly. 108 109 One nice thing about FileCheck (compared to grep) is that it allows merging 110 test cases together into logical groups. For example, because the test above 111 is checking for the "``sub1:``" and "``inc4:``" labels, it will not match 112 unless there is a "``subl``" in between those labels. If it existed somewhere 113 else in the file, that would not count: "``grep subl``" matches if "``subl``" 114 exists anywhere in the file. 115 116 The FileCheck -check-prefix option 117 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 118 119 The FileCheck :option:`-check-prefix` option allows multiple test 120 configurations to be driven from one `.ll` file. This is useful in many 121 circumstances, for example, testing different architectural variants with 122 :program:`llc`. Here's a simple example: 123 124 .. code-block:: llvm 125 126 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \ 127 ; RUN: | FileCheck %s -check-prefix=X32 128 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \ 129 ; RUN: | FileCheck %s -check-prefix=X64 130 131 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind { 132 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1 133 ret <4 x i32> %tmp1 134 ; X32: pinsrd_1: 135 ; X32: pinsrd $1, 4(%esp), %xmm0 136 137 ; X64: pinsrd_1: 138 ; X64: pinsrd $1, %edi, %xmm0 139 } 140 141 In this case, we're testing that we get the expected code generation with 142 both 32-bit and 64-bit code generation. 143 144 The "CHECK-NEXT:" directive 145 ~~~~~~~~~~~~~~~~~~~~~~~~~~~ 146 147 Sometimes you want to match lines and would like to verify that matches 148 happen on exactly consecutive lines with no other lines in between them. In 149 this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify 150 this. If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``". 151 For example, something like this works as you'd expect: 152 153 .. code-block:: llvm 154 155 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) { 156 %tmp3 = load <2 x double>* %A, align 16 157 %tmp7 = insertelement <2 x double> undef, double %B, i32 0 158 %tmp9 = shufflevector <2 x double> %tmp3, 159 <2 x double> %tmp7, 160 <2 x i32> < i32 0, i32 2 > 161 store <2 x double> %tmp9, <2 x double>* %r, align 16 162 ret void 163 164 ; CHECK: t2: 165 ; CHECK: movl 8(%esp), %eax 166 ; CHECK-NEXT: movapd (%eax), %xmm0 167 ; CHECK-NEXT: movhpd 12(%esp), %xmm0 168 ; CHECK-NEXT: movl 4(%esp), %eax 169 ; CHECK-NEXT: movapd %xmm0, (%eax) 170 ; CHECK-NEXT: ret 171 } 172 173 "``CHECK-NEXT:``" directives reject the input unless there is exactly one 174 newline between it and the previous directive. A "``CHECK-NEXT:``" cannot be 175 the first directive in a file. 176 177 The "CHECK-NOT:" directive 178 ~~~~~~~~~~~~~~~~~~~~~~~~~~ 179 180 The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur 181 between two matches (or before the first match, or after the last match). For 182 example, to verify that a load is removed by a transformation, a test like this 183 can be used: 184 185 .. code-block:: llvm 186 187 define i8 @coerce_offset0(i32 %V, i32* %P) { 188 store i32 %V, i32* %P 189 190 %P2 = bitcast i32* %P to i8* 191 %P3 = getelementptr i8* %P2, i32 2 192 193 %A = load i8* %P3 194 ret i8 %A 195 ; CHECK: @coerce_offset0 196 ; CHECK-NOT: load 197 ; CHECK: ret i8 198 } 199 200 The "CHECK-DAG:" directive 201 ~~~~~~~~~~~~~~~~~~~~~~~~~~ 202 203 If it's necessary to match strings that don't occur in a strictly sequential 204 order, "``CHECK-DAG:``" could be used to verify them between two matches (or 205 before the first match, or after the last match). For example, clang emits 206 vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks 207 in the natural order: 208 209 .. code-block:: c++ 210 211 // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s 212 213 struct Foo { virtual void method(); }; 214 Foo f; // emit vtable 215 // CHECK-DAG: @_ZTV3Foo = 216 217 struct Bar { virtual void method(); }; 218 Bar b; 219 // CHECK-DAG: @_ZTV3Bar = 220 221 ``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to 222 exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result, 223 the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all 224 occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind 225 occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example, 226 227 .. code-block:: llvm 228 229 ; CHECK-DAG: BEFORE 230 ; CHECK-NOT: NOT 231 ; CHECK-DAG: AFTER 232 233 This case will reject input strings where ``BEFORE`` occurs after ``AFTER``. 234 235 With captured variables, ``CHECK-DAG:`` is able to match valid topological 236 orderings of a DAG with edges from the definition of a variable to its use. 237 It's useful, e.g., when your test cases need to match different output 238 sequences from the instruction scheduler. For example, 239 240 .. code-block:: llvm 241 242 ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2 243 ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4 244 ; CHECK: mul r5, [[REG1]], [[REG2]] 245 246 In this case, any order of that two ``add`` instructions will be allowed. 247 248 If you are defining `and` using variables in the same ``CHECK-DAG:`` block, 249 be aware that the definition rule can match `after` its use. 250 251 So, for instance, the code below will pass: 252 253 .. code-block:: llvm 254 255 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 256 ; CHECK-DAG: vmov.32 [[REG2]][1] 257 vmov.32 d0[1] 258 vmov.32 d0[0] 259 260 While this other code, will not: 261 262 .. code-block:: llvm 263 264 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 265 ; CHECK-DAG: vmov.32 [[REG2]][1] 266 vmov.32 d1[1] 267 vmov.32 d0[0] 268 269 While this can be very useful, it's also dangerous, because in the case of 270 register sequence, you must have a strong order (read before write, copy before 271 use, etc). If the definition your test is looking for doesn't match (because 272 of a bug in the compiler), it may match further away from the use, and mask 273 real bugs away. 274 275 In those cases, to enforce the order, use a non-DAG directive between DAG-blocks. 276 277 The "CHECK-LABEL:" directive 278 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 279 280 Sometimes in a file containing multiple tests divided into logical blocks, one 281 or more ``CHECK:`` directives may inadvertently succeed by matching lines in a 282 later block. While an error will usually eventually be generated, the check 283 flagged as causing the error may not actually bear any relationship to the 284 actual source of the problem. 285 286 In order to produce better error messages in these cases, the "``CHECK-LABEL:``" 287 directive can be used. It is treated identically to a normal ``CHECK`` 288 directive except that FileCheck makes an additional assumption that a line 289 matched by the directive cannot also be matched by any other check present in 290 ``match-filename``; this is intended to be used for lines containing labels or 291 other unique identifiers. Conceptually, the presence of ``CHECK-LABEL`` divides 292 the input stream into separate blocks, each of which is processed independently, 293 preventing a ``CHECK:`` directive in one block matching a line in another block. 294 For example, 295 296 .. code-block:: llvm 297 298 define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) { 299 entry: 300 ; CHECK-LABEL: C_ctor_base: 301 ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0 302 ; CHECK: bl A_ctor_base 303 ; CHECK: mov r0, [[SAVETHIS]] 304 %0 = bitcast %struct.C* %this to %struct.A* 305 %call = tail call %struct.A* @A_ctor_base(%struct.A* %0) 306 %1 = bitcast %struct.C* %this to %struct.B* 307 %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x) 308 ret %struct.C* %this 309 } 310 311 define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) { 312 entry: 313 ; CHECK-LABEL: D_ctor_base: 314 315 The use of ``CHECK-LABEL:`` directives in this case ensures that the three 316 ``CHECK:`` directives only accept lines corresponding to the body of the 317 ``@C_ctor_base`` function, even if the patterns match lines found later in 318 the file. Furthermore, if one of these three ``CHECK:`` directives fail, 319 FileCheck will recover by continuing to the next block, allowing multiple test 320 failures to be detected in a single invocation. 321 322 There is no requirement that ``CHECK-LABEL:`` directives contain strings that 323 correspond to actual syntactic labels in a source or output language: they must 324 simply uniquely match a single line in the file being verified. 325 326 ``CHECK-LABEL:`` directives cannot contain variable definitions or uses. 327 328 FileCheck Pattern Matching Syntax 329 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 330 331 The "``CHECK:``" and "``CHECK-NOT:``" directives both take a pattern to match. 332 For most uses of FileCheck, fixed string matching is perfectly sufficient. For 333 some things, a more flexible form of matching is desired. To support this, 334 FileCheck allows you to specify regular expressions in matching strings, 335 surrounded by double braces: ``{{yourregex}}``. Because we want to use fixed 336 string matching for a majority of what we do, FileCheck has been designed to 337 support mixing and matching fixed string matching with regular expressions. 338 This allows you to write things like this: 339 340 .. code-block:: llvm 341 342 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}} 343 344 In this case, any offset from the ESP register will be allowed, and any xmm 345 register will be allowed. 346 347 Because regular expressions are enclosed with double braces, they are 348 visually distinct, and you don't need to use escape characters within the double 349 braces like you would in C. In the rare case that you want to match double 350 braces explicitly from the input, you can use something ugly like 351 ``{{[{][{]}}`` as your pattern. 352 353 FileCheck Variables 354 ~~~~~~~~~~~~~~~~~~~ 355 356 It is often useful to match a pattern and then verify that it occurs again 357 later in the file. For codegen tests, this can be useful to allow any register, 358 but verify that that register is used consistently later. To do this, 359 :program:`FileCheck` allows named variables to be defined and substituted into 360 patterns. Here is a simple example: 361 362 .. code-block:: llvm 363 364 ; CHECK: test5: 365 ; CHECK: notw [[REGISTER:%[a-z]+]] 366 ; CHECK: andw {{.*}}[[REGISTER]] 367 368 The first check line matches a regex ``%[a-z]+`` and captures it into the 369 variable ``REGISTER``. The second line verifies that whatever is in 370 ``REGISTER`` occurs later in the file after an "``andw``". :program:`FileCheck` 371 variable references are always contained in ``[[ ]]`` pairs, and their names can 372 be formed with the regex ``[a-zA-Z][a-zA-Z0-9]*``. If a colon follows the name, 373 then it is a definition of the variable; otherwise, it is a use. 374 375 :program:`FileCheck` variables can be defined multiple times, and uses always 376 get the latest value. Variables can also be used later on the same line they 377 were defined on. For example: 378 379 .. code-block:: llvm 380 381 ; CHECK: op [[REG:r[0-9]+]], [[REG]] 382 383 Can be useful if you want the operands of ``op`` to be the same register, 384 and don't care exactly which register it is. 385 386 FileCheck Expressions 387 ~~~~~~~~~~~~~~~~~~~~~ 388 389 Sometimes there's a need to verify output which refers line numbers of the 390 match file, e.g. when testing compiler diagnostics. This introduces a certain 391 fragility of the match file structure, as "``CHECK:``" lines contain absolute 392 line numbers in the same file, which have to be updated whenever line numbers 393 change due to text addition or deletion. 394 395 To support this case, FileCheck allows using ``[[@LINE]]``, 396 ``[[@LINE+<offset>]]``, ``[[@LINE-<offset>]]`` expressions in patterns. These 397 expressions expand to a number of the line where a pattern is located (with an 398 optional integer offset). 399 400 This way match patterns can be put near the relevant test lines and include 401 relative line number references, for example: 402 403 .. code-block:: c++ 404 405 // CHECK: test.cpp:[[@LINE+4]]:6: error: expected ';' after top level declarator 406 // CHECK-NEXT: {{^int a}} 407 // CHECK-NEXT: {{^ \^}} 408 // CHECK-NEXT: {{^ ;}} 409 int a 410 411
