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:: --dump-input-on-failure 81 82 When the check fails, dump all of the original input. 83 84.. option:: --enable-var-scope 85 86 Enables scope for regex variables. 87 88 Variables with names that start with ``$`` are considered global and 89 remain set throughout the file. 90 91 All other variables get undefined after each encountered ``CHECK-LABEL``. 92 93.. option:: -D<VAR=VALUE> 94 95 Sets a filecheck variable ``VAR`` with value ``VALUE`` that can be used in 96 ``CHECK:`` lines. 97 98.. option:: -version 99 100 Show the version number of this program. 101 102.. option:: -v 103 104 Print directive pattern matches. 105 106.. option:: -vv 107 108 Print information helpful in diagnosing internal FileCheck issues, such as 109 discarded overlapping ``CHECK-DAG:`` matches, implicit EOF pattern matches, 110 and ``CHECK-NOT:`` patterns that do not have matches. Implies ``-v``. 111 112.. option:: --allow-deprecated-dag-overlap 113 114 Enable overlapping among matches in a group of consecutive ``CHECK-DAG:`` 115 directives. This option is deprecated and is only provided for convenience 116 as old tests are migrated to the new non-overlapping ``CHECK-DAG:`` 117 implementation. 118 119EXIT STATUS 120----------- 121 122If :program:`FileCheck` verifies that the file matches the expected contents, 123it exits with 0. Otherwise, if not, or if an error occurs, it will exit with a 124non-zero value. 125 126TUTORIAL 127-------- 128 129FileCheck is typically used from LLVM regression tests, being invoked on the RUN 130line of the test. A simple example of using FileCheck from a RUN line looks 131like this: 132 133.. code-block:: llvm 134 135 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s 136 137This syntax says to pipe the current file ("``%s``") into ``llvm-as``, pipe 138that into ``llc``, then pipe the output of ``llc`` into ``FileCheck``. This 139means that FileCheck will be verifying its standard input (the llc output) 140against the filename argument specified (the original ``.ll`` file specified by 141"``%s``"). To see how this works, let's look at the rest of the ``.ll`` file 142(after the RUN line): 143 144.. code-block:: llvm 145 146 define void @sub1(i32* %p, i32 %v) { 147 entry: 148 ; CHECK: sub1: 149 ; CHECK: subl 150 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v) 151 ret void 152 } 153 154 define void @inc4(i64* %p) { 155 entry: 156 ; CHECK: inc4: 157 ; CHECK: incq 158 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1) 159 ret void 160 } 161 162Here you can see some "``CHECK:``" lines specified in comments. Now you can 163see how the file is piped into ``llvm-as``, then ``llc``, and the machine code 164output is what we are verifying. FileCheck checks the machine code output to 165verify that it matches what the "``CHECK:``" lines specify. 166 167The syntax of the "``CHECK:``" lines is very simple: they are fixed strings that 168must occur in order. FileCheck defaults to ignoring horizontal whitespace 169differences (e.g. a space is allowed to match a tab) but otherwise, the contents 170of the "``CHECK:``" line is required to match some thing in the test file exactly. 171 172One nice thing about FileCheck (compared to grep) is that it allows merging 173test cases together into logical groups. For example, because the test above 174is checking for the "``sub1:``" and "``inc4:``" labels, it will not match 175unless there is a "``subl``" in between those labels. If it existed somewhere 176else in the file, that would not count: "``grep subl``" matches if "``subl``" 177exists anywhere in the file. 178 179The FileCheck -check-prefix option 180~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 181 182The FileCheck `-check-prefix` option allows multiple test 183configurations to be driven from one `.ll` file. This is useful in many 184circumstances, for example, testing different architectural variants with 185:program:`llc`. Here's a simple example: 186 187.. code-block:: llvm 188 189 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \ 190 ; RUN: | FileCheck %s -check-prefix=X32 191 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \ 192 ; RUN: | FileCheck %s -check-prefix=X64 193 194 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind { 195 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1 196 ret <4 x i32> %tmp1 197 ; X32: pinsrd_1: 198 ; X32: pinsrd $1, 4(%esp), %xmm0 199 200 ; X64: pinsrd_1: 201 ; X64: pinsrd $1, %edi, %xmm0 202 } 203 204In this case, we're testing that we get the expected code generation with 205both 32-bit and 64-bit code generation. 206 207The "CHECK-NEXT:" directive 208~~~~~~~~~~~~~~~~~~~~~~~~~~~ 209 210Sometimes you want to match lines and would like to verify that matches 211happen on exactly consecutive lines with no other lines in between them. In 212this case, you can use "``CHECK:``" and "``CHECK-NEXT:``" directives to specify 213this. If you specified a custom check prefix, just use "``<PREFIX>-NEXT:``". 214For example, something like this works as you'd expect: 215 216.. code-block:: llvm 217 218 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) { 219 %tmp3 = load <2 x double>* %A, align 16 220 %tmp7 = insertelement <2 x double> undef, double %B, i32 0 221 %tmp9 = shufflevector <2 x double> %tmp3, 222 <2 x double> %tmp7, 223 <2 x i32> < i32 0, i32 2 > 224 store <2 x double> %tmp9, <2 x double>* %r, align 16 225 ret void 226 227 ; CHECK: t2: 228 ; CHECK: movl 8(%esp), %eax 229 ; CHECK-NEXT: movapd (%eax), %xmm0 230 ; CHECK-NEXT: movhpd 12(%esp), %xmm0 231 ; CHECK-NEXT: movl 4(%esp), %eax 232 ; CHECK-NEXT: movapd %xmm0, (%eax) 233 ; CHECK-NEXT: ret 234 } 235 236"``CHECK-NEXT:``" directives reject the input unless there is exactly one 237newline between it and the previous directive. A "``CHECK-NEXT:``" cannot be 238the first directive in a file. 239 240The "CHECK-SAME:" directive 241~~~~~~~~~~~~~~~~~~~~~~~~~~~ 242 243Sometimes you want to match lines and would like to verify that matches happen 244on the same line as the previous match. In this case, you can use "``CHECK:``" 245and "``CHECK-SAME:``" directives to specify this. If you specified a custom 246check prefix, just use "``<PREFIX>-SAME:``". 247 248"``CHECK-SAME:``" is particularly powerful in conjunction with "``CHECK-NOT:``" 249(described below). 250 251For example, the following works like you'd expect: 252 253.. code-block:: llvm 254 255 !0 = !DILocation(line: 5, scope: !1, inlinedAt: !2) 256 257 ; CHECK: !DILocation(line: 5, 258 ; CHECK-NOT: column: 259 ; CHECK-SAME: scope: ![[SCOPE:[0-9]+]] 260 261"``CHECK-SAME:``" directives reject the input if there are any newlines between 262it and the previous directive. A "``CHECK-SAME:``" cannot be the first 263directive in a file. 264 265The "CHECK-EMPTY:" directive 266~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 267 268If you need to check that the next line has nothing on it, not even whitespace, 269you can use the "``CHECK-EMPTY:``" directive. 270 271.. code-block:: llvm 272 273 foo 274 275 bar 276 ; CHECK: foo 277 ; CHECK-EMPTY: 278 ; CHECK-NEXT: bar 279 280Just like "``CHECK-NEXT:``" the directive will fail if there is more than one 281newline before it finds the next blank line, and it cannot be the first 282directive in a file. 283 284The "CHECK-NOT:" directive 285~~~~~~~~~~~~~~~~~~~~~~~~~~ 286 287The "``CHECK-NOT:``" directive is used to verify that a string doesn't occur 288between two matches (or before the first match, or after the last match). For 289example, to verify that a load is removed by a transformation, a test like this 290can be used: 291 292.. code-block:: llvm 293 294 define i8 @coerce_offset0(i32 %V, i32* %P) { 295 store i32 %V, i32* %P 296 297 %P2 = bitcast i32* %P to i8* 298 %P3 = getelementptr i8* %P2, i32 2 299 300 %A = load i8* %P3 301 ret i8 %A 302 ; CHECK: @coerce_offset0 303 ; CHECK-NOT: load 304 ; CHECK: ret i8 305 } 306 307The "CHECK-DAG:" directive 308~~~~~~~~~~~~~~~~~~~~~~~~~~ 309 310If it's necessary to match strings that don't occur in a strictly sequential 311order, "``CHECK-DAG:``" could be used to verify them between two matches (or 312before the first match, or after the last match). For example, clang emits 313vtable globals in reverse order. Using ``CHECK-DAG:``, we can keep the checks 314in the natural order: 315 316.. code-block:: c++ 317 318 // RUN: %clang_cc1 %s -emit-llvm -o - | FileCheck %s 319 320 struct Foo { virtual void method(); }; 321 Foo f; // emit vtable 322 // CHECK-DAG: @_ZTV3Foo = 323 324 struct Bar { virtual void method(); }; 325 Bar b; 326 // CHECK-DAG: @_ZTV3Bar = 327 328``CHECK-NOT:`` directives could be mixed with ``CHECK-DAG:`` directives to 329exclude strings between the surrounding ``CHECK-DAG:`` directives. As a result, 330the surrounding ``CHECK-DAG:`` directives cannot be reordered, i.e. all 331occurrences matching ``CHECK-DAG:`` before ``CHECK-NOT:`` must not fall behind 332occurrences matching ``CHECK-DAG:`` after ``CHECK-NOT:``. For example, 333 334.. code-block:: llvm 335 336 ; CHECK-DAG: BEFORE 337 ; CHECK-NOT: NOT 338 ; CHECK-DAG: AFTER 339 340This case will reject input strings where ``BEFORE`` occurs after ``AFTER``. 341 342With captured variables, ``CHECK-DAG:`` is able to match valid topological 343orderings of a DAG with edges from the definition of a variable to its use. 344It's useful, e.g., when your test cases need to match different output 345sequences from the instruction scheduler. For example, 346 347.. code-block:: llvm 348 349 ; CHECK-DAG: add [[REG1:r[0-9]+]], r1, r2 350 ; CHECK-DAG: add [[REG2:r[0-9]+]], r3, r4 351 ; CHECK: mul r5, [[REG1]], [[REG2]] 352 353In this case, any order of that two ``add`` instructions will be allowed. 354 355If you are defining `and` using variables in the same ``CHECK-DAG:`` block, 356be aware that the definition rule can match `after` its use. 357 358So, for instance, the code below will pass: 359 360.. code-block:: text 361 362 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 363 ; CHECK-DAG: vmov.32 [[REG2]][1] 364 vmov.32 d0[1] 365 vmov.32 d0[0] 366 367While this other code, will not: 368 369.. code-block:: text 370 371 ; CHECK-DAG: vmov.32 [[REG2:d[0-9]+]][0] 372 ; CHECK-DAG: vmov.32 [[REG2]][1] 373 vmov.32 d1[1] 374 vmov.32 d0[0] 375 376While this can be very useful, it's also dangerous, because in the case of 377register sequence, you must have a strong order (read before write, copy before 378use, etc). If the definition your test is looking for doesn't match (because 379of a bug in the compiler), it may match further away from the use, and mask 380real bugs away. 381 382In those cases, to enforce the order, use a non-DAG directive between DAG-blocks. 383 384A ``CHECK-DAG:`` directive skips matches that overlap the matches of any 385preceding ``CHECK-DAG:`` directives in the same ``CHECK-DAG:`` block. Not only 386is this non-overlapping behavior consistent with other directives, but it's 387also necessary to handle sets of non-unique strings or patterns. For example, 388the following directives look for unordered log entries for two tasks in a 389parallel program, such as the OpenMP runtime: 390 391.. code-block:: text 392 393 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin 394 // CHECK-DAG: [[THREAD_ID]]: task_end 395 // 396 // CHECK-DAG: [[THREAD_ID:[0-9]+]]: task_begin 397 // CHECK-DAG: [[THREAD_ID]]: task_end 398 399The second pair of directives is guaranteed not to match the same log entries 400as the first pair even though the patterns are identical and even if the text 401of the log entries is identical because the thread ID manages to be reused. 402 403The "CHECK-LABEL:" directive 404~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 405 406Sometimes in a file containing multiple tests divided into logical blocks, one 407or more ``CHECK:`` directives may inadvertently succeed by matching lines in a 408later block. While an error will usually eventually be generated, the check 409flagged as causing the error may not actually bear any relationship to the 410actual source of the problem. 411 412In order to produce better error messages in these cases, the "``CHECK-LABEL:``" 413directive can be used. It is treated identically to a normal ``CHECK`` 414directive except that FileCheck makes an additional assumption that a line 415matched by the directive cannot also be matched by any other check present in 416``match-filename``; this is intended to be used for lines containing labels or 417other unique identifiers. Conceptually, the presence of ``CHECK-LABEL`` divides 418the input stream into separate blocks, each of which is processed independently, 419preventing a ``CHECK:`` directive in one block matching a line in another block. 420If ``--enable-var-scope`` is in effect, all local variables are cleared at the 421beginning of the block. 422 423For example, 424 425.. code-block:: llvm 426 427 define %struct.C* @C_ctor_base(%struct.C* %this, i32 %x) { 428 entry: 429 ; CHECK-LABEL: C_ctor_base: 430 ; CHECK: mov [[SAVETHIS:r[0-9]+]], r0 431 ; CHECK: bl A_ctor_base 432 ; CHECK: mov r0, [[SAVETHIS]] 433 %0 = bitcast %struct.C* %this to %struct.A* 434 %call = tail call %struct.A* @A_ctor_base(%struct.A* %0) 435 %1 = bitcast %struct.C* %this to %struct.B* 436 %call2 = tail call %struct.B* @B_ctor_base(%struct.B* %1, i32 %x) 437 ret %struct.C* %this 438 } 439 440 define %struct.D* @D_ctor_base(%struct.D* %this, i32 %x) { 441 entry: 442 ; CHECK-LABEL: D_ctor_base: 443 444The use of ``CHECK-LABEL:`` directives in this case ensures that the three 445``CHECK:`` directives only accept lines corresponding to the body of the 446``@C_ctor_base`` function, even if the patterns match lines found later in 447the file. Furthermore, if one of these three ``CHECK:`` directives fail, 448FileCheck will recover by continuing to the next block, allowing multiple test 449failures to be detected in a single invocation. 450 451There is no requirement that ``CHECK-LABEL:`` directives contain strings that 452correspond to actual syntactic labels in a source or output language: they must 453simply uniquely match a single line in the file being verified. 454 455``CHECK-LABEL:`` directives cannot contain variable definitions or uses. 456 457FileCheck Pattern Matching Syntax 458~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 459 460All FileCheck directives take a pattern to match. 461For most uses of FileCheck, fixed string matching is perfectly sufficient. For 462some things, a more flexible form of matching is desired. To support this, 463FileCheck allows you to specify regular expressions in matching strings, 464surrounded by double braces: ``{{yourregex}}``. FileCheck implements a POSIX 465regular expression matcher; it supports Extended POSIX regular expressions 466(ERE). Because we want to use fixed string matching for a majority of what we 467do, FileCheck has been designed to support mixing and matching fixed string 468matching with regular expressions. This allows you to write things like this: 469 470.. code-block:: llvm 471 472 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}} 473 474In this case, any offset from the ESP register will be allowed, and any xmm 475register will be allowed. 476 477Because regular expressions are enclosed with double braces, they are 478visually distinct, and you don't need to use escape characters within the double 479braces like you would in C. In the rare case that you want to match double 480braces explicitly from the input, you can use something ugly like 481``{{[{][{]}}`` as your pattern. 482 483FileCheck Variables 484~~~~~~~~~~~~~~~~~~~ 485 486It is often useful to match a pattern and then verify that it occurs again 487later in the file. For codegen tests, this can be useful to allow any register, 488but verify that that register is used consistently later. To do this, 489:program:`FileCheck` allows named variables to be defined and substituted into 490patterns. Here is a simple example: 491 492.. code-block:: llvm 493 494 ; CHECK: test5: 495 ; CHECK: notw [[REGISTER:%[a-z]+]] 496 ; CHECK: andw {{.*}}[[REGISTER]] 497 498The first check line matches a regex ``%[a-z]+`` and captures it into the 499variable ``REGISTER``. The second line verifies that whatever is in 500``REGISTER`` occurs later in the file after an "``andw``". :program:`FileCheck` 501variable references are always contained in ``[[ ]]`` pairs, and their names can 502be formed with the regex ``[a-zA-Z_][a-zA-Z0-9_]*``. If a colon follows the name, 503then it is a definition of the variable; otherwise, it is a use. 504 505:program:`FileCheck` variables can be defined multiple times, and uses always 506get the latest value. Variables can also be used later on the same line they 507were defined on. For example: 508 509.. code-block:: llvm 510 511 ; CHECK: op [[REG:r[0-9]+]], [[REG]] 512 513Can be useful if you want the operands of ``op`` to be the same register, 514and don't care exactly which register it is. 515 516If ``--enable-var-scope`` is in effect, variables with names that 517start with ``$`` are considered to be global. All others variables are 518local. All local variables get undefined at the beginning of each 519CHECK-LABEL block. Global variables are not affected by CHECK-LABEL. 520This makes it easier to ensure that individual tests are not affected 521by variables set in preceding tests. 522 523FileCheck Expressions 524~~~~~~~~~~~~~~~~~~~~~ 525 526Sometimes there's a need to verify output which refers line numbers of the 527match file, e.g. when testing compiler diagnostics. This introduces a certain 528fragility of the match file structure, as "``CHECK:``" lines contain absolute 529line numbers in the same file, which have to be updated whenever line numbers 530change due to text addition or deletion. 531 532To support this case, FileCheck allows using ``[[@LINE]]``, 533``[[@LINE+<offset>]]``, ``[[@LINE-<offset>]]`` expressions in patterns. These 534expressions expand to a number of the line where a pattern is located (with an 535optional integer offset). 536 537This way match patterns can be put near the relevant test lines and include 538relative line number references, for example: 539 540.. code-block:: c++ 541 542 // CHECK: test.cpp:[[@LINE+4]]:6: error: expected ';' after top level declarator 543 // CHECK-NEXT: {{^int a}} 544 // CHECK-NEXT: {{^ \^}} 545 // CHECK-NEXT: {{^ ;}} 546 int a 547 548Matching Newline Characters 549~~~~~~~~~~~~~~~~~~~~~~~~~~~ 550 551To match newline characters in regular expressions the character class 552``[[:space:]]`` can be used. For example, the following pattern: 553 554.. code-block:: c++ 555 556 // CHECK: DW_AT_location [DW_FORM_sec_offset] ([[DLOC:0x[0-9a-f]+]]){{[[:space:]].*}}"intd" 557 558matches output of the form (from llvm-dwarfdump): 559 560.. code-block:: text 561 562 DW_AT_location [DW_FORM_sec_offset] (0x00000233) 563 DW_AT_name [DW_FORM_strp] ( .debug_str[0x000000c9] = "intd") 564 565letting us set the :program:`FileCheck` variable ``DLOC`` to the desired value 566``0x00000233``, extracted from the line immediately preceding "``intd``". 567