1LIT(1) LLVM LIT(1)
2
6 lit - LLVM Integrated Tester
7
9 lit [options] [tests]
10
12 lit is a portable tool for executing LLVM and Clang style test suites,
13 summarizing their results, and providing indication of failures. lit
14 is designed to be a lightweight testing tool with as simple a user in‐
15 terface as possible.
16 lit should be run with one or more tests to run specified on the com‐
18 mand line. Tests can be either individual test files or directories to
19 search for tests (see TEST DISCOVERY).
20 Each specified test will be executed (potentially concurrently) and
22 once all tests have been run lit will print summary information on the
23 number of tests which passed or failed (see TEST STATUS RESULTS). The
24 lit program will execute with a non-zero exit code if any tests fail.
25 By default lit will use a succinct progress display and will only print
27 summary information for test failures. See OUTPUT OPTIONS for options
28 controlling the lit progress display and output.
29 lit also includes a number of options for controlling how tests are ex‐
31 ecuted (specific features may depend on the particular test format).
32 See EXECUTION OPTIONS for more information.
33 Finally, lit also supports additional options for only running a subset
35 of the options specified on the command line, see SELECTION OPTIONS for
36 more information.
37 lit parses options from the environment variable LIT_OPTS after parsing
39 options from the command line. LIT_OPTS is primarily useful for sup‐
40 plementing or overriding the command-line options supplied to lit by
41 check targets defined by a project's build system.
42 Users interested in the lit architecture or designing a lit testing im‐
44 plementation should see LIT INFRASTRUCTURE.
45
47 -h, --help
48 Show the lit help message.
49 -j N, --workers=N
51 Run N tests in parallel. By default, this is automatically cho‐
52 sen to match the number of detected available CPUs.
53 --config-prefix=NAME
55 Search for NAME.cfg and NAME.site.cfg when searching for test
56 suites, instead of lit.cfg and lit.site.cfg.
57 -D NAME[=VALUE], --param NAME[=VALUE]
59 Add a user defined parameter NAME with the given VALUE (or the
60 empty string if not given). The meaning and use of these param‐
61 eters is test suite dependent.
62
64 -q, --quiet
65 Suppress any output except for test failures.
66 -s, --succinct
68 Show less output, for example don't show information on tests
69 that pass. Also show a progress bar, unless --no-progress-bar
70 is specified.
71 -v, --verbose
73 Show more information on test failures, for example the entire
74 test output instead of just the test result.
75 -vv, --echo-all-commands
77 Echo all commands to stdout, as they are being executed. This
78 can be valuable for debugging test failures, as the last echoed
79 command will be the one which has failed. lit normally inserts
80 a no-op command (: in the case of bash) with argument 'RUN: at
81 line N' before each command pipeline, and this option also
82 causes those no-op commands to be echoed to stdout to help you
83 locate the source line of the failed command. This option im‐
84 plies --verbose.
85 -a, --show-all
87 Show more information about all tests, for example the entire
88 test commandline and output.
89 --no-progress-bar
91 Do not use curses based progress bar.
92 --show-unsupported
94 Show the names of unsupported tests.
95 --show-xfail
97 Show the names of tests that were expected to fail.
98
100 --path=PATH
101 Specify an additional PATH to use when searching for executables
102 in tests.
103 --vg Run individual tests under valgrind (using the memcheck tool).
105 The --error-exitcode argument for valgrind is used so that val‐
106 grind failures will cause the program to exit with a non-zero
107 status.
108 When this option is enabled, lit will also automatically provide
110 a "valgrind" feature that can be used to conditionally disable
111 (or expect failure in) certain tests.
112 --vg-arg=ARG
114 When --vg is used, specify an additional argument to pass to
115 valgrind itself.
116 --vg-leak
118 When --vg is used, enable memory leak checks. When this option
119 is enabled, lit will also automatically provide a "vg_leak" fea‐
120 ture that can be used to conditionally disable (or expect fail‐
121 ure in) certain tests.
122 --time-tests
124 Track the wall time individual tests take to execute and in‐
125 cludes the results in the summary output. This is useful for
126 determining which tests in a test suite take the most time to
127 execute.
128 --ignore-fail
130 Exit with status zero even if some tests fail.
131 --no-indirectly-run-check
133 Do not error if a test would not be run if the user had speci‐
134 fied the containing directory instead of naming the test di‐
135 rectly.
136
138 By default, lit will run failing tests first, then run tests in de‐
139 scending execution time order to optimize concurrency. The execution
140 order can be changed using the --order option.
141 The timing data is stored in the test_exec_root in a file named
143 .lit_test_times.txt. If this file does not exist, then lit checks the
144 test_source_root for the file to optionally accelerate clean builds.
145 --shuffle
147 Run the tests in a random order, not failing/slowest first. Dep‐
148 recated, use --order instead.
149 --max-failures N
151 Stop execution after the given number N of failures. An integer
152 argument should be passed on the command line prior to execu‐
153 tion.
154 --max-tests=N
156 Run at most N tests and then terminate.
157 --max-time=N
159 Spend at most N seconds (approximately) running tests and then
160 terminate. Note that this is not an alias for --timeout; the
161 two are different kinds of maximums.
162 --num-shards=M
164 Divide the set of selected tests into M equal-sized subsets or
165 "shards", and run only one of them. Must be used with the
166 --run-shard=N option, which selects the shard to run. The envi‐
167 ronment variable LIT_NUM_SHARDS can also be used in place of
168 this option. These two options provide a coarse mechanism for
169 partitioning large testsuites, for parallel execution on sepa‐
170 rate machines (say in a large testing farm).
171 --order={lexical,random,smart}
173 Define the order in which tests are run. The supported values
174 are:
175 • lexical - tests will be run in lexical order according to the
177 test file path. This option is useful when predictable test
178 order is desired.
179 • random - tests will be run in random order.
181 • smart - tests that failed previously will be run first, then
183 the remaining tests, all in descending execution time order.
184 This is the default as it optimizes concurrency.
185 --run-shard=N
187 Select which shard to run, assuming the --num-shards=M option
188 was provided. The two options must be used together, and the
189 value of N must be in the range 1..M. The environment variable
190 LIT_RUN_SHARD can also be used in place of this option.
191 --timeout=N
193 Spend at most N seconds (approximately) running each individual
194 test. 0 means no time limit, and 0 is the default. Note that
195 this is not an alias for --max-time; the two are different kinds
196 of maximums.
197 --filter=REGEXP
199 Run only those tests whose name matches the regular expression
200 specified in REGEXP. The environment variable LIT_FILTER can be
201 also used in place of this option, which is especially useful in
202 environments where the call to lit is issued indirectly.
203 --filter-out=REGEXP
205 Filter out those tests whose name matches the regular expression
206 specified in REGEXP. The environment variable LIT_FILTER_OUT can
207 be also used in place of this option, which is especially useful
208 in environments where the call to lit is issued indirectly.
209 --xfail=LIST
211 Treat those tests whose name is in the semicolon separated list
212 LIST as XFAIL. This can be helpful when one does not want to
213 modify the test suite. The environment variable LIT_XFAIL can be
214 also used in place of this option, which is especially useful in
215 environments where the call to lit is issued indirectly.
216 A test name can specified as a file name relative to the test
218 suite directory. For example:
219 LIT_XFAIL="affinity/kmp-hw-subset.c;offloading/memory_manager.cpp"
221 In this case, all of the following tests are treated as XFAIL:
223 libomp :: affinity/kmp-hw-subset.c
225 libomptarget :: nvptx64-nvidia-cuda :: offloading/memory_manager.cpp
226 libomptarget :: x86_64-pc-linux-gnu :: offloading/memory_manager.cpp
227 Alternatively, a test name can be specified as the full test
229 name reported in LIT output. For example, we can adjust the
230 previous example not to treat the nvptx64-nvidia-cuda version of
231 offloading/memory_manager.cpp as XFAIL:
232 LIT_XFAIL="affinity/kmp-hw-subset.c;libomptarget :: x86_64-pc-linux-gnu :: offloading/memory_manager.cpp"
234 --xfail-not=LIST
236 Do not treat the specified tests as XFAIL. The environment
237 variable LIT_XFAIL_NOT can also be used in place of this option.
238 The syntax is the same as for --xfail and LIT_XFAIL.
239 --xfail-not and LIT_XFAIL_NOT always override all other XFAIL
240 specifications, including an --xfail appearing later on the com‐
241 mand line. The primary purpose is to suppress an XPASS result
242 without modifying a test case that uses the XFAIL directive.
243
245 --debug
246 Run lit in debug mode, for debugging configuration issues and
247 lit itself.
248 --show-suites
250 List the discovered test suites and exit.
251 --show-tests
253 List all of the discovered tests and exit.
254
256 lit will exit with an exit code of 1 if there are any FAIL or XPASS re‐
257 sults. Otherwise, it will exit with the status 0. Other exit codes
258 are used for non-test related failures (for example a user error or an
259 internal program error).
260
262 The inputs passed to lit can be either individual tests, or entire di‐
263 rectories or hierarchies of tests to run. When lit starts up, the
264 first thing it does is convert the inputs into a complete list of tests
265 to run as part of test discovery.
266 In the lit model, every test must exist inside some test suite. lit
268 resolves the inputs specified on the command line to test suites by
269 searching upwards from the input path until it finds a lit.cfg or
270 lit.site.cfg file. These files serve as both a marker of test suites
271 and as configuration files which lit loads in order to understand how
272 to find and run the tests inside the test suite.
273 Once lit has mapped the inputs into test suites it traverses the list
275 of inputs adding tests for individual files and recursively searching
276 for tests in directories.
277 This behavior makes it easy to specify a subset of tests to run, while
279 still allowing the test suite configuration to control exactly how
280 tests are interpreted. In addition, lit always identifies tests by the
281 test suite they are in, and their relative path inside the test suite.
282 For appropriately configured projects, this allows lit to provide con‐
283 venient and flexible support for out-of-tree builds.
284
286 Each test ultimately produces one of the following eight results:
287 PASS
289 The test succeeded.
290 FLAKYPASS
292 The test succeeded after being re-run more than once. This only ap‐
293 plies to tests containing an ALLOW_RETRIES: annotation.
294 XFAIL
296 The test failed, but that is expected. This is used for test for‐
297 mats which allow specifying that a test does not currently work, but
298 wish to leave it in the test suite.
299 XPASS
301 The test succeeded, but it was expected to fail. This is used for
302 tests which were specified as expected to fail, but are now succeed‐
303 ing (generally because the feature they test was broken and has been
304 fixed).
305 FAIL
307 The test failed.
308 UNRESOLVED
310 The test result could not be determined. For example, this occurs
311 when the test could not be run, the test itself is invalid, or the
312 test was interrupted.
313 UNSUPPORTED
315 The test is not supported in this environment. This is used by test
316 formats which can report unsupported tests.
317 TIMEOUT
319 The test was run, but it timed out before it was able to complete.
320 This is considered a failure.
321 Depending on the test format tests may produce additional information
323 about their status (generally only for failures). See the OUTPUT OP‐
324 TIONS section for more information.
325
327 This section describes the lit testing architecture for users inter‐
328 ested in creating a new lit testing implementation, or extending an ex‐
329 isting one.
330 lit proper is primarily an infrastructure for discovering and running
332 arbitrary tests, and to expose a single convenient interface to these
333 tests. lit itself doesn't know how to run tests, rather this logic is
334 defined by test suites.
335 TEST SUITES
337 As described in TEST DISCOVERY, tests are always located inside a test
338 suite. Test suites serve to define the format of the tests they con‐
339 tain, the logic for finding those tests, and any additional information
340 to run the tests.
341 lit identifies test suites as directories containing lit.cfg or
343 lit.site.cfg files (see also --config-prefix). Test suites are ini‐
344 tially discovered by recursively searching up the directory hierarchy
345 for all the input files passed on the command line. You can use
346 --show-suites to display the discovered test suites at startup.
347 Once a test suite is discovered, its config file is loaded. Config
349 files themselves are Python modules which will be executed. When the
350 config file is executed, two important global variables are predefined:
351 lit_config
353 The global lit configuration object (a LitConfig instance), which
354 defines the builtin test formats, global configuration parameters,
355 and other helper routines for implementing test configurations.
356 config
358 This is the config object (a TestingConfig instance) for the test
359 suite, which the config file is expected to populate. The following
360 variables are also available on the config object, some of which
361 must be set by the config and others are optional or predefined:
362 name [required] The name of the test suite, for use in reports and
364 diagnostics.
365 test_format [required] The test format object which will be used to
367 discover and run tests in the test suite. Generally this will be a
368 builtin test format available from the lit.formats module.
369 test_source_root The filesystem path to the test suite root. For
371 out-of-dir builds this is the directory that will be scanned for
372 tests.
373 test_exec_root For out-of-dir builds, the path to the test suite
375 root inside the object directory. This is where tests will be run
376 and temporary output files placed.
377 environment A dictionary representing the environment to use when
379 executing tests in the suite.
380 standalone_tests When true, mark a directory with tests expected to
382 be run standalone. Test discovery is disabled for that directory and
383 --no-indirectly-run-check is in effect. lit.suffixes and lit.ex‐
384 cludes must be empty when this variable is true.
385 suffixes For lit test formats which scan directories for tests, this
387 variable is a list of suffixes to identify test files. Used by:
388 ShTest.
389 substitutions For lit test formats which substitute variables into a
391 test script, the list of substitutions to perform. Used by: ShTest.
392 unsupported Mark an unsupported directory, all tests within it will
394 be reported as unsupported. Used by: ShTest.
395 parent The parent configuration, this is the config object for the
397 directory containing the test suite, or None.
398 root The root configuration. This is the top-most lit configuration
400 in the project.
401 pipefail Normally a test using a shell pipe fails if any of the com‐
403 mands on the pipe fail. If this is not desired, setting this vari‐
404 able to false makes the test fail only if the last command in the
405 pipe fails.
406 available_features A set of features that can be used in XFAIL, RE‐
408 QUIRES, and UNSUPPORTED directives.
409 TEST DISCOVERY
411 Once test suites are located, lit recursively traverses the source di‐
412 rectory (following test_source_root) looking for tests. When lit en‐
413 ters a sub-directory, it first checks to see if a nested test suite is
414 defined in that directory. If so, it loads that test suite recur‐
415 sively, otherwise it instantiates a local test config for the directory
416 (see LOCAL CONFIGURATION FILES).
417 Tests are identified by the test suite they are contained within, and
419 the relative path inside that suite. Note that the relative path may
420 not refer to an actual file on disk; some test formats (such as
421 GoogleTest) define "virtual tests" which have a path that contains both
422 the path to the actual test file and a subpath to identify the virtual
423 test.
424 LOCAL CONFIGURATION FILES
426 When lit loads a subdirectory in a test suite, it instantiates a local
427 test configuration by cloning the configuration for the parent direc‐
428 tory --- the root of this configuration chain will always be a test
429 suite. Once the test configuration is cloned lit checks for a lit.lo‐
430 cal.cfg file in the subdirectory. If present, this file will be loaded
431 and can be used to specialize the configuration for each individual di‐
432 rectory. This facility can be used to define subdirectories of op‐
433 tional tests, or to change other configuration parameters --- for exam‐
434 ple, to change the test format, or the suffixes which identify test
435 files.
436 SUBSTITUTIONS
438 lit allows patterns to be substituted inside RUN commands. It also pro‐
439 vides the following base set of substitutions, which are defined in
440 TestRunner.py:
441 ┌────────────────────────┬────────────────────────────┐
443 │Macro │ Substitution │
444 ├────────────────────────┼────────────────────────────┤
445 │%s │ source path (path to the │
446 │ │ file currently being run) │
447 ├────────────────────────┼────────────────────────────┤
448 │%S │ source dir (directory of │
449 │ │ the file currently being │
450 │ │ run) │
451 ├────────────────────────┼────────────────────────────┤
452 │%p │ same as %S │
453 ├────────────────────────┼────────────────────────────┤
454 │%{pathsep} │ path separator │
455 ├────────────────────────┼────────────────────────────┤
456 │%t │ temporary file name unique │
457 │ │ to the test │
458 ├────────────────────────┼────────────────────────────┤
459 │%basename_t │ The last path component of │
460 │ │ %t but without the .tmp │
461 │ │ extension │
462 ├────────────────────────┼────────────────────────────┤
463 │%T │ parent directory of %t │
464 │ │ (not unique, deprecated, │
465 │ │ do not use) │
466 ├────────────────────────┼────────────────────────────┤
467 │%% │ % │
468 ├────────────────────────┼────────────────────────────┤
469 │%/s │ %s but \ is replaced by / │
470 ├────────────────────────┼────────────────────────────┤
471 │%/S │ %S but \ is replaced by / │
472 └────────────────────────┴────────────────────────────┘
473 │%/p │ %p but \ is replaced by / │
476 ├────────────────────────┼────────────────────────────┤
477 │%/t │ %t but \ is replaced by / │
478 ├────────────────────────┼────────────────────────────┤
479 │%/T │ %T but \ is replaced by / │
480 ├────────────────────────┼────────────────────────────┤
481 │%{/s:regex_replacement} │ %/s but escaped for use in │
482 │ │ the replacement of a s@@@ │
483 │ │ command in sed │
484 ├────────────────────────┼────────────────────────────┤
485 │%{/S:regex_replacement} │ %/S but escaped for use in │
486 │ │ the replacement of a s@@@ │
487 │ │ command in sed │
488 ├────────────────────────┼────────────────────────────┤
489 │%{/p:regex_replacement} │ %/p but escaped for use in │
490 │ │ the replacement of a s@@@ │
491 │ │ command in sed │
492 ├────────────────────────┼────────────────────────────┤
493 │%{/t:regex_replacement} │ %/t but escaped for use in │
494 │ │ the replacement of a s@@@ │
495 │ │ command in sed │
496 ├────────────────────────┼────────────────────────────┤
497 │%{/T:regex_replacement} │ %/T but escaped for use in │
498 │ │ the replacement of a s@@@ │
499 │ │ command in sed │
500 ├────────────────────────┼────────────────────────────┤
501 │%:s │ On Windows, %/s but a : is │
502 │ │ removed if its the second │
503 │ │ character. Otherwise, %s │
504 │ │ but with a single leading │
505 │ │ / removed. │
506 ├────────────────────────┼────────────────────────────┤
507 │%:S │ On Windows, %/S but a : is │
508 │ │ removed if its the second │
509 │ │ character. Otherwise, %S │
510 │ │ but with a single leading │
511 │ │ / removed. │
512 ├────────────────────────┼────────────────────────────┤
513 │%:p │ On Windows, %/p but a : is │
514 │ │ removed if its the second │
515 │ │ character. Otherwise, %p │
516 │ │ but with a single leading │
517 │ │ / removed. │
518 ├────────────────────────┼────────────────────────────┤
519 │%:t │ On Windows, %/t but a : is │
520 │ │ removed if its the second │
521 │ │ character. Otherwise, %t │
522 │ │ but with a single leading │
523 │ │ / removed. │
524 ├────────────────────────┼────────────────────────────┤
525 │%:T │ On Windows, %/T but a : is │
526 │ │ removed if its the second │
527 │ │ character. Otherwise, %T │
528 │ │ but with a single leading │
529 │ │ / removed. │
530 └────────────────────────┴────────────────────────────┘
531 Other substitutions are provided that are variations on this base set
533 and further substitution patterns can be defined by each test module.
534 See the modules LOCAL CONFIGURATION FILES.
535 By default, substitutions are expanded exactly once, so that if e.g. a
537 substitution %build is defined in top of another substitution %cxx,
538 %build will expand to %cxx textually, not to what %cxx expands to.
539 However, if the recursiveExpansionLimit property of the TestingConfig
540 is set to a non-negative integer, substitutions will be expanded recur‐
541 sively until that limit is reached. It is an error if the limit is
542 reached and expanding substitutions again would yield a different re‐
543 sult.
544 More detailed information on substitutions can be found in the LLVM
546 Testing Infrastructure Guide.
547 TEST RUN OUTPUT FORMAT
549 The lit output for a test run conforms to the following schema, in both
550 short and verbose modes (although in short mode no PASS lines will be
551 shown). This schema has been chosen to be relatively easy to reliably
552 parse by a machine (for example in buildbot log scraping), and for
553 other tools to generate.
554 Each test result is expected to appear on a line that matches:
556 <result code>: <test name> (<progress info>)
558 where <result-code> is a standard test result such as PASS, FAIL,
560 XFAIL, XPASS, UNRESOLVED, or UNSUPPORTED. The performance result codes
561 of IMPROVED and REGRESSED are also allowed.
562 The <test name> field can consist of an arbitrary string containing no
564 newline.
565 The <progress info> field can be used to report progress information
567 such as (1/300) or can be empty, but even when empty the parentheses
568 are required.
569 Each test result may include additional (multiline) log information in
571 the following format:
572 <log delineator> TEST '(<test name>)' <trailing delineator>
574 ... log message ...
575 <log delineator>
576 where <test name> should be the name of a preceding reported test, <log
578 delineator> is a string of "*" characters at least four characters long
579 (the recommended length is 20), and <trailing delineator> is an arbi‐
580 trary (unparsed) string.
581 The following is an example of a test run output which consists of four
583 tests A, B, C, and D, and a log message for the failing test C:
584 PASS: A (1 of 4)
586 PASS: B (2 of 4)
587 FAIL: C (3 of 4)
588 ******************** TEST 'C' FAILED ********************
589 Test 'C' failed as a result of exit code 1.
590 ********************
591 PASS: D (4 of 4)
592 LIT EXAMPLE TESTS
594 The lit distribution contains several example implementations of test
595 suites in the ExampleTests directory.
596
598 valgrind(1)
599
601 Maintained by the LLVM Team (https://llvm.org/).
602
604 2003-2023, LLVM Project
60514 2023-07-20 LIT(1)