v8/tools/testrunner/standard_runner.py
Michal Majewski 7f01c9303f [test] Implement variant generators as processors
Bug: v8:6917
Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng;luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: Ib5bfdf4d6fee6102f62c7334a1b22146f1a1fc5b
Reviewed-on: https://chromium-review.googlesource.com/857376
Commit-Queue: Michał Majewski <majeski@google.com>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#50525}
2018-01-12 10:39:20 +00:00

638 lines
25 KiB
Python
Executable File

#!/usr/bin/env python
#
# Copyright 2017 the V8 project authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.
from collections import OrderedDict
from os.path import join
import multiprocessing
import os
import random
import shlex
import subprocess
import sys
import time
# Adds testrunner to the path hence it has to be imported at the beggining.
import base_runner
from testrunner.local import execution
from testrunner.local import progress
from testrunner.local import testsuite
from testrunner.local import utils
from testrunner.local import verbose
from testrunner.local.variants import ALL_VARIANTS
from testrunner.objects import context
from testrunner.objects import predictable
from testrunner.testproc.execution import ExecutionProc
from testrunner.testproc.filter import StatusFileFilterProc
from testrunner.testproc.loader import LoadProc
from testrunner.testproc.progress import (VerboseProgressIndicator,
ResultsTracker)
from testrunner.testproc.rerun import RerunProc
from testrunner.testproc.variant import VariantProc
TIMEOUT_DEFAULT = 60
# Variants ordered by expected runtime (slowest first).
VARIANTS = ["default"]
MORE_VARIANTS = [
"stress",
"stress_incremental_marking",
"nooptimization",
"stress_background_compile",
"wasm_traps",
]
VARIANT_ALIASES = {
# The default for developer workstations.
"dev": VARIANTS,
# Additional variants, run on all bots.
"more": MORE_VARIANTS,
# Shortcut for the two above ("more" first - it has the longer running tests).
"exhaustive": MORE_VARIANTS + VARIANTS,
# Additional variants, run on a subset of bots.
"extra": ["future", "liftoff", "trusted"],
}
GC_STRESS_FLAGS = ["--gc-interval=500", "--stress-compaction",
"--concurrent-recompilation-queue-length=64",
"--concurrent-recompilation-delay=500",
"--concurrent-recompilation"]
# Double the timeout for these:
SLOW_ARCHS = ["arm",
"mips",
"mipsel",
"mips64",
"mips64el",
"s390",
"s390x",
"arm64"]
PREDICTABLE_WRAPPER = os.path.join(
base_runner.BASE_DIR, 'tools', 'predictable_wrapper.py')
class StandardTestRunner(base_runner.BaseTestRunner):
def __init__(self, *args, **kwargs):
super(StandardTestRunner, self).__init__(*args, **kwargs)
self.sancov_dir = None
def _do_execute(self, options, args):
if options.swarming:
# Swarming doesn't print how isolated commands are called. Lets make
# this less cryptic by printing it ourselves.
print ' '.join(sys.argv)
if utils.GuessOS() == "macos":
# TODO(machenbach): Temporary output for investigating hanging test
# driver on mac.
print "V8 related processes running on this host:"
try:
print subprocess.check_output(
"ps -e | egrep 'd8|cctest|unittests'", shell=True)
except Exception:
pass
suite_paths = utils.GetSuitePaths(join(self.basedir, "test"))
# Use default tests if no test configuration was provided at the cmd line.
if len(args) == 0:
args = ["default"]
# Expand arguments with grouped tests. The args should reflect the list
# of suites as otherwise filters would break.
def ExpandTestGroups(name):
if name in base_runner.TEST_MAP:
return [suite for suite in base_runner.TEST_MAP[name]]
else:
return [name]
args = reduce(lambda x, y: x + y,
[ExpandTestGroups(arg) for arg in args],
[])
args_suites = OrderedDict() # Used as set
for arg in args:
args_suites[arg.split('/')[0]] = True
suite_paths = [ s for s in args_suites if s in suite_paths ]
suites = []
for root in suite_paths:
if options.verbose:
print '>>> Loading test suite: %s' % root
suite = testsuite.TestSuite.LoadTestSuite(
os.path.join(self.basedir, "test", root))
if suite:
suites.append(suite)
try:
return self._execute(args, options, suites)
except KeyboardInterrupt:
return 2
def _add_parser_options(self, parser):
parser.add_option("--sancov-dir",
help="Directory where to collect coverage data")
parser.add_option("--cfi-vptr",
help="Run tests with UBSAN cfi_vptr option.",
default=False, action="store_true")
parser.add_option("--novfp3",
help="Indicates that V8 was compiled without VFP3"
" support",
default=False, action="store_true")
parser.add_option("--cat", help="Print the source of the tests",
default=False, action="store_true")
parser.add_option("--slow-tests",
help="Regard slow tests (run|skip|dontcare)",
default="dontcare")
parser.add_option("--pass-fail-tests",
help="Regard pass|fail tests (run|skip|dontcare)",
default="dontcare")
parser.add_option("--gc-stress",
help="Switch on GC stress mode",
default=False, action="store_true")
parser.add_option("--command-prefix",
help="Prepended to each shell command used to run a"
" test",
default="")
parser.add_option("--extra-flags",
help="Additional flags to pass to each test command",
action="append", default=[])
parser.add_option("--infra-staging", help="Use new test runner features",
default=False, action="store_true")
parser.add_option("--isolates", help="Whether to test isolates",
default=False, action="store_true")
parser.add_option("-j", help="The number of parallel tasks to run",
default=0, type="int")
parser.add_option("--no-harness", "--noharness",
help="Run without test harness of a given suite",
default=False, action="store_true")
parser.add_option("--no-presubmit", "--nopresubmit",
help='Skip presubmit checks (deprecated)',
default=False, dest="no_presubmit", action="store_true")
parser.add_option("--no-sorting", "--nosorting",
help="Don't sort tests according to duration of last"
" run.",
default=False, dest="no_sorting", action="store_true")
parser.add_option("--no-variants", "--novariants",
help="Deprecated. "
"Equivalent to passing --variants=default",
default=False, dest="no_variants", action="store_true")
parser.add_option("--variants",
help="Comma-separated list of testing variants;"
" default: \"%s\"" % ",".join(VARIANTS))
parser.add_option("--exhaustive-variants",
default=False, action="store_true",
help="Deprecated. "
"Equivalent to passing --variants=exhaustive")
parser.add_option("-p", "--progress",
help=("The style of progress indicator"
" (verbose, dots, color, mono)"),
choices=progress.PROGRESS_INDICATORS.keys(),
default="mono")
parser.add_option("--quickcheck", default=False, action="store_true",
help=("Quick check mode (skip slow tests)"))
parser.add_option("--report", help="Print a summary of the tests to be"
" run",
default=False, action="store_true")
parser.add_option("--json-test-results",
help="Path to a file for storing json results.")
parser.add_option("--flakiness-results",
help="Path to a file for storing flakiness json.")
parser.add_option("--rerun-failures-count",
help=("Number of times to rerun each failing test case."
" Very slow tests will be rerun only once."),
default=0, type="int")
parser.add_option("--rerun-failures-max",
help="Maximum number of failing test cases to rerun.",
default=100, type="int")
parser.add_option("--shard-count",
help="Split testsuites into this number of shards",
default=1, type="int")
parser.add_option("--shard-run",
help="Run this shard from the split up tests.",
default=1, type="int")
parser.add_option("--dont-skip-slow-simulator-tests",
help="Don't skip more slow tests when using a"
" simulator.",
default=False, action="store_true",
dest="dont_skip_simulator_slow_tests")
parser.add_option("--swarming",
help="Indicates running test driver on swarming.",
default=False, action="store_true")
parser.add_option("--time", help="Print timing information after running",
default=False, action="store_true")
parser.add_option("-t", "--timeout", help="Timeout in seconds",
default=TIMEOUT_DEFAULT, type="int")
parser.add_option("--warn-unused", help="Report unused rules",
default=False, action="store_true")
parser.add_option("--junitout", help="File name of the JUnit output")
parser.add_option("--junittestsuite",
help="The testsuite name in the JUnit output file",
default="v8tests")
parser.add_option("--random-seed", default=0, dest="random_seed",
help="Default seed for initializing random generator",
type=int)
parser.add_option("--random-seed-stress-count", default=1, type="int",
dest="random_seed_stress_count",
help="Number of runs with different random seeds")
def _process_options(self, options):
global VARIANTS
if options.sancov_dir:
self.sancov_dir = options.sancov_dir
if not os.path.exists(self.sancov_dir):
print("sancov-dir %s doesn't exist" % self.sancov_dir)
raise base_runner.TestRunnerError()
options.command_prefix = shlex.split(options.command_prefix)
options.extra_flags = sum(map(shlex.split, options.extra_flags), [])
if options.gc_stress:
options.extra_flags += GC_STRESS_FLAGS
if self.build_config.asan:
options.extra_flags.append("--invoke-weak-callbacks")
options.extra_flags.append("--omit-quit")
if options.novfp3:
options.extra_flags.append("--noenable-vfp3")
if options.no_variants: # pragma: no cover
print ("Option --no-variants is deprecated. "
"Pass --variants=default instead.")
assert not options.variants
options.variants = "default"
if options.exhaustive_variants: # pragma: no cover
# TODO(machenbach): Switch infra to --variants=exhaustive after M65.
print ("Option --exhaustive-variants is deprecated. "
"Pass --variants=exhaustive instead.")
# This is used on many bots. It includes a larger set of default
# variants.
# Other options for manipulating variants still apply afterwards.
assert not options.variants
options.variants = "exhaustive"
if options.quickcheck:
assert not options.variants
options.variants = "stress,default"
options.slow_tests = "skip"
options.pass_fail_tests = "skip"
if self.build_config.predictable:
options.variants = "default"
options.extra_flags.append("--predictable")
options.extra_flags.append("--verify_predictable")
options.extra_flags.append("--no-inline-new")
# Add predictable wrapper to command prefix.
options.command_prefix = (
[sys.executable, PREDICTABLE_WRAPPER] + options.command_prefix)
# TODO(machenbach): Figure out how to test a bigger subset of variants on
# msan.
if self.build_config.msan:
options.variants = "default"
if options.j == 0:
options.j = multiprocessing.cpu_count()
if options.random_seed_stress_count <= 1 and options.random_seed == 0:
options.random_seed = self._random_seed()
# Use developer defaults if no variant was specified.
options.variants = options.variants or "dev"
if options.variants == "infra_staging":
options.variants = "exhaustive"
options.infra_staging = True
# Resolve variant aliases and dedupe.
# TODO(machenbach): Don't mutate global variable. Rather pass mutated
# version as local variable.
VARIANTS = list(set(reduce(
list.__add__,
(VARIANT_ALIASES.get(v, [v]) for v in options.variants.split(",")),
[],
)))
if not set(VARIANTS).issubset(ALL_VARIANTS):
print "All variants must be in %s" % str(ALL_VARIANTS)
raise base_runner.TestRunnerError()
def CheckTestMode(name, option): # pragma: no cover
if not option in ["run", "skip", "dontcare"]:
print "Unknown %s mode %s" % (name, option)
raise base_runner.TestRunnerError()
CheckTestMode("slow test", options.slow_tests)
CheckTestMode("pass|fail test", options.pass_fail_tests)
if self.build_config.no_i18n:
base_runner.TEST_MAP["bot_default"].remove("intl")
base_runner.TEST_MAP["default"].remove("intl")
def _setup_env(self):
super(StandardTestRunner, self)._setup_env()
symbolizer_option = self._get_external_symbolizer_option()
if self.sancov_dir:
os.environ['ASAN_OPTIONS'] = ":".join([
'coverage=1',
'coverage_dir=%s' % self.sancov_dir,
symbolizer_option,
"allow_user_segv_handler=1",
])
def _random_seed(self):
seed = 0
while not seed:
seed = random.SystemRandom().randint(-2147483648, 2147483647)
return seed
def _execute(self, args, options, suites):
print(">>> Running tests for %s.%s" % (self.build_config.arch,
self.mode_name))
# Populate context object.
# Simulators are slow, therefore allow a longer timeout.
if self.build_config.arch in SLOW_ARCHS:
options.timeout *= 2
options.timeout *= self.mode_options.timeout_scalefactor
if self.build_config.predictable:
# Predictable mode is slower.
options.timeout *= 2
ctx = context.Context(self.build_config.arch,
self.mode_options.execution_mode,
self.outdir,
self.mode_options.flags,
options.verbose,
options.timeout,
options.isolates,
options.command_prefix,
options.extra_flags,
self.build_config.no_i18n,
options.random_seed,
options.no_sorting,
options.rerun_failures_count,
options.rerun_failures_max,
options.no_harness,
use_perf_data=not options.swarming,
sancov_dir=self.sancov_dir,
infra_staging=options.infra_staging)
# TODO(all): Combine "simulator" and "simulator_run".
# TODO(machenbach): In GN we can derive simulator run from
# target_arch != v8_target_arch in the dumped build config.
simulator_run = (
not options.dont_skip_simulator_slow_tests and
self.build_config.arch in [
'arm64', 'arm', 'mipsel', 'mips', 'mips64', 'mips64el', 'ppc',
'ppc64', 's390', 's390x'] and
bool(base_runner.ARCH_GUESS) and
self.build_config.arch != base_runner.ARCH_GUESS)
# Find available test suites and read test cases from them.
variables = {
"arch": self.build_config.arch,
"asan": self.build_config.asan,
"byteorder": sys.byteorder,
"dcheck_always_on": self.build_config.dcheck_always_on,
"deopt_fuzzer": False,
"gc_fuzzer": False,
"gc_stress": options.gc_stress,
"gcov_coverage": self.build_config.gcov_coverage,
"isolates": options.isolates,
"mode": self.mode_options.status_mode,
"msan": self.build_config.msan,
"no_harness": options.no_harness,
"no_i18n": self.build_config.no_i18n,
"no_snap": self.build_config.no_snap,
"novfp3": options.novfp3,
"predictable": self.build_config.predictable,
"simulator": utils.UseSimulator(self.build_config.arch),
"simulator_run": simulator_run,
"system": utils.GuessOS(),
"tsan": self.build_config.tsan,
"ubsan_vptr": self.build_config.ubsan_vptr,
}
all_tests = []
num_tests = 0
for s in suites:
s.ReadStatusFile(variables)
s.ReadTestCases(ctx)
if len(args) > 0:
s.FilterTestCasesByArgs(args)
all_tests += s.tests
# First filtering by status applying the generic rules (tests without
# variants)
if options.warn_unused:
tests = [(t.name, t.variant) for t in s.tests]
s.statusfile.warn_unused_rules(tests, check_variant_rules=False)
s.FilterTestCasesByStatus(options.slow_tests, options.pass_fail_tests)
if options.cat:
verbose.PrintTestSource(s.tests)
continue
if not options.infra_staging:
variant_gen = s.CreateLegacyVariantsGenerator(VARIANTS)
variant_tests = [ t.create_variant(v, flags)
for t in s.tests
for v in variant_gen.FilterVariantsByTest(t)
for flags in variant_gen.GetFlagSets(t, v) ]
else:
# Variants will be created in the test processors pipeline
variant_tests = s.tests
if options.random_seed_stress_count > 1:
# Duplicate test for random seed stress mode.
def iter_seed_flags():
for _ in range(0, options.random_seed_stress_count):
# Use given random seed for all runs (set by default in
# execution.py) or a new random seed if none is specified.
if options.random_seed:
yield []
else:
yield ["--random-seed=%d" % self._random_seed()]
s.tests = [
t.create_variant(t.variant, flags, 'seed-stress-%d' % n)
for t in variant_tests
for n, flags in enumerate(iter_seed_flags())
]
else:
s.tests = variant_tests
# Second filtering by status applying also the variant-dependent rules.
if options.warn_unused:
tests = [(t.name, t.variant) for t in s.tests]
s.statusfile.warn_unused_rules(tests, check_variant_rules=True)
if not options.infra_staging:
s.FilterTestCasesByStatus(options.slow_tests, options.pass_fail_tests)
s.tests = self._shard_tests(s.tests, options)
for t in s.tests:
t.cmd = t.get_command(ctx)
num_tests += len(s.tests)
if options.cat:
return 0 # We're done here.
if options.report:
verbose.PrintReport(all_tests)
# Run the tests.
start_time = time.time()
progress_indicator = progress.IndicatorNotifier()
progress_indicator.Register(
progress.PROGRESS_INDICATORS[options.progress]())
if options.junitout: # pragma: no cover
progress_indicator.Register(progress.JUnitTestProgressIndicator(
options.junitout, options.junittestsuite))
if options.json_test_results:
progress_indicator.Register(progress.JsonTestProgressIndicator(
options.json_test_results,
self.build_config.arch,
self.mode_options.execution_mode,
ctx.random_seed))
if options.flakiness_results: # pragma: no cover
progress_indicator.Register(progress.FlakinessTestProgressIndicator(
options.flakiness_results))
if self.build_config.predictable:
outproc_factory = predictable.get_outproc
else:
outproc_factory = None
if options.infra_staging:
exit_code = self._run_test_procs(suites, options, progress_indicator,
ctx, outproc_factory)
else:
runner = execution.Runner(suites, progress_indicator, ctx,
outproc_factory)
exit_code = runner.Run(options.j)
overall_duration = time.time() - start_time
if options.time:
verbose.PrintTestDurations(suites, runner.outputs, overall_duration)
if num_tests == 0:
print("Warning: no tests were run!")
if exit_code == 1 and options.json_test_results:
print("Force exit code 0 after failures. Json test results file "
"generated with failure information.")
exit_code = 0
if self.sancov_dir:
# If tests ran with sanitizer coverage, merge coverage files in the end.
try:
print "Merging sancov files."
subprocess.check_call([
sys.executable,
join(self.basedir, "tools", "sanitizers", "sancov_merger.py"),
"--coverage-dir=%s" % self.sancov_dir])
except:
print >> sys.stderr, "Error: Merging sancov files failed."
exit_code = 1
return exit_code
def _shard_tests(self, tests, options):
# Read gtest shard configuration from environment (e.g. set by swarming).
# If none is present, use values passed on the command line.
shard_count = int(
os.environ.get('GTEST_TOTAL_SHARDS', options.shard_count))
shard_run = os.environ.get('GTEST_SHARD_INDEX')
if shard_run is not None:
# The v8 shard_run starts at 1, while GTEST_SHARD_INDEX starts at 0.
shard_run = int(shard_run) + 1
else:
shard_run = options.shard_run
if options.shard_count > 1:
# Log if a value was passed on the cmd line and it differs from the
# environment variables.
if options.shard_count != shard_count: # pragma: no cover
print("shard_count from cmd line differs from environment variable "
"GTEST_TOTAL_SHARDS")
if (options.shard_run > 1 and
options.shard_run != shard_run): # pragma: no cover
print("shard_run from cmd line differs from environment variable "
"GTEST_SHARD_INDEX")
if shard_count < 2:
return tests
if shard_run < 1 or shard_run > shard_count:
# TODO(machenbach): Turn this into an assert. If that's wrong on the
# bots, printing will be quite useless. Or refactor this code to make
# sure we get a return code != 0 after testing if we got here.
print "shard-run not a valid number, should be in [1:shard-count]"
print "defaulting back to running all tests"
return tests
count = 0
shard = []
for test in tests:
if count % shard_count == shard_run - 1:
shard.append(test)
count += 1
return shard
def _run_test_procs(self, suites, options, progress_indicator, context,
outproc_factory):
jobs = options.j
print '>>> Running with test processors'
loader = LoadProc()
results = ResultsTracker(count_subtests=False)
indicators = progress_indicator.ToProgressIndicatorProcs()
procs = [
loader,
VariantProc(VARIANTS),
StatusFileFilterProc(options.slow_tests, options.pass_fail_tests),
results,
] + indicators
if context.rerun_failures_count:
procs.append(RerunProc(
context.rerun_failures_count,
context.rerun_failures_max
))
execproc = ExecutionProc(jobs, context)
procs.append(execproc)
for i in xrange(0, len(procs) - 1):
procs[i].connect_to(procs[i + 1])
tests = [t for s in suites for t in s.tests]
tests.sort(key=lambda t: t.is_slow, reverse=True)
loader.load_tests(tests)
for indicator in indicators:
indicator.starting()
execproc.start()
for indicator in indicators:
indicator.finished()
if results.failed:
return 1
if results.remaining:
return 2
return 0
if __name__ == '__main__':
sys.exit(StandardTestRunner().execute())