234f27b5ae
TestSuite has a static method LoadTestSuite that should properly configure the TestSuite instance (i.e. loaded status files and tests), however the method leaves some configuration logic to the caller. The leaky abstraction causes the caller to do a bunch of loading operations (see the removed methods in base_runner.py). This CL isolates the TestSuite loading logic to the static method only. This is a refactoring only change without any intended logical changes. R=machenbach@chromium.org CC=yangguo@chromium.org,sergiyb@chromium.org Bug: v8:8174 Change-Id: I105059c9c9e050f03bb584174e2bd7ceeae2b228 Reviewed-on: https://chromium-review.googlesource.com/c/1396417 Commit-Queue: Tamer Tas <tmrts@chromium.org> Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Sergiy Belozorov <sergiyb@chromium.org> Cr-Commit-Position: refs/heads/master@{#58744}
782 lines
26 KiB
Python
782 lines
26 KiB
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
|
|
import json
|
|
import multiprocessing
|
|
import optparse
|
|
import os
|
|
import shlex
|
|
import sys
|
|
|
|
|
|
# Add testrunner to the path.
|
|
sys.path.insert(
|
|
0,
|
|
os.path.dirname(
|
|
os.path.dirname(os.path.abspath(__file__))))
|
|
|
|
|
|
from testrunner.local import command
|
|
from testrunner.local import testsuite
|
|
from testrunner.local import utils
|
|
from testrunner.test_config import TestConfig
|
|
from testrunner.testproc import progress
|
|
from testrunner.testproc.rerun import RerunProc
|
|
from testrunner.testproc.shard import ShardProc
|
|
from testrunner.testproc.sigproc import SignalProc
|
|
from testrunner.testproc.timeout import TimeoutProc
|
|
|
|
|
|
BASE_DIR = (
|
|
os.path.dirname(
|
|
os.path.dirname(
|
|
os.path.dirname(
|
|
os.path.abspath(__file__)))))
|
|
|
|
DEFAULT_OUT_GN = 'out.gn'
|
|
|
|
# Map of test name synonyms to lists of test suites. Should be ordered by
|
|
# expected runtimes (suites with slow test cases first). These groups are
|
|
# invoked in separate steps on the bots.
|
|
TEST_MAP = {
|
|
# This needs to stay in sync with test/bot_default.isolate.
|
|
"bot_default": [
|
|
"debugger",
|
|
"mjsunit",
|
|
"cctest",
|
|
"wasm-spec-tests",
|
|
"inspector",
|
|
"webkit",
|
|
"mkgrokdump",
|
|
"wasm-js",
|
|
"fuzzer",
|
|
"message",
|
|
"preparser",
|
|
"intl",
|
|
"unittests",
|
|
],
|
|
# This needs to stay in sync with test/default.isolate.
|
|
"default": [
|
|
"debugger",
|
|
"mjsunit",
|
|
"cctest",
|
|
"wasm-spec-tests",
|
|
"inspector",
|
|
"mkgrokdump",
|
|
"wasm-js",
|
|
"fuzzer",
|
|
"message",
|
|
"preparser",
|
|
"intl",
|
|
"unittests",
|
|
],
|
|
# This needs to stay in sync with test/d8_default.isolate.
|
|
"d8_default": [
|
|
"debugger",
|
|
"mjsunit",
|
|
"webkit",
|
|
"message",
|
|
"preparser",
|
|
"intl",
|
|
],
|
|
# This needs to stay in sync with test/optimize_for_size.isolate.
|
|
"optimize_for_size": [
|
|
"debugger",
|
|
"mjsunit",
|
|
"cctest",
|
|
"inspector",
|
|
"webkit",
|
|
"intl",
|
|
],
|
|
"unittests": [
|
|
"unittests",
|
|
],
|
|
}
|
|
|
|
# Double the timeout for these:
|
|
SLOW_ARCHS = ["arm",
|
|
"mips",
|
|
"mipsel",
|
|
"mips64",
|
|
"mips64el",
|
|
"s390",
|
|
"s390x",
|
|
"arm64"]
|
|
|
|
|
|
class ModeConfig(object):
|
|
def __init__(self, flags, timeout_scalefactor, status_mode, execution_mode):
|
|
self.flags = flags
|
|
self.timeout_scalefactor = timeout_scalefactor
|
|
self.status_mode = status_mode
|
|
self.execution_mode = execution_mode
|
|
|
|
|
|
DEBUG_FLAGS = ["--nohard-abort", "--enable-slow-asserts", "--verify-heap"]
|
|
RELEASE_FLAGS = ["--nohard-abort"]
|
|
MODES = {
|
|
"debug": ModeConfig(
|
|
flags=DEBUG_FLAGS,
|
|
timeout_scalefactor=4,
|
|
status_mode="debug",
|
|
execution_mode="debug",
|
|
),
|
|
"optdebug": ModeConfig(
|
|
flags=DEBUG_FLAGS,
|
|
timeout_scalefactor=4,
|
|
status_mode="debug",
|
|
execution_mode="debug",
|
|
),
|
|
"release": ModeConfig(
|
|
flags=RELEASE_FLAGS,
|
|
timeout_scalefactor=1,
|
|
status_mode="release",
|
|
execution_mode="release",
|
|
),
|
|
# Normal trybot release configuration. There, dchecks are always on which
|
|
# implies debug is set. Hence, the status file needs to assume debug-like
|
|
# behavior/timeouts.
|
|
"tryrelease": ModeConfig(
|
|
flags=RELEASE_FLAGS,
|
|
timeout_scalefactor=1,
|
|
status_mode="debug",
|
|
execution_mode="release",
|
|
),
|
|
# This mode requires v8 to be compiled with dchecks and slow dchecks.
|
|
"slowrelease": ModeConfig(
|
|
flags=RELEASE_FLAGS + ["--enable-slow-asserts"],
|
|
timeout_scalefactor=2,
|
|
status_mode="debug",
|
|
execution_mode="release",
|
|
),
|
|
}
|
|
|
|
PROGRESS_INDICATORS = {
|
|
'verbose': progress.VerboseProgressIndicator,
|
|
'dots': progress.DotsProgressIndicator,
|
|
'color': progress.ColorProgressIndicator,
|
|
'mono': progress.MonochromeProgressIndicator,
|
|
}
|
|
|
|
class TestRunnerError(Exception):
|
|
pass
|
|
|
|
|
|
class BuildConfig(object):
|
|
def __init__(self, build_config):
|
|
# In V8 land, GN's x86 is called ia32.
|
|
if build_config['v8_target_cpu'] == 'x86':
|
|
self.arch = 'ia32'
|
|
else:
|
|
self.arch = build_config['v8_target_cpu']
|
|
|
|
self.asan = build_config['is_asan']
|
|
self.cfi_vptr = build_config['is_cfi']
|
|
self.dcheck_always_on = build_config['dcheck_always_on']
|
|
self.gcov_coverage = build_config['is_gcov_coverage']
|
|
self.is_android = build_config['is_android']
|
|
self.is_debug = build_config['is_debug']
|
|
self.msan = build_config['is_msan']
|
|
self.no_i18n = not build_config['v8_enable_i18n_support']
|
|
self.no_snap = not build_config['v8_use_snapshot']
|
|
self.predictable = build_config['v8_enable_verify_predictable']
|
|
self.tsan = build_config['is_tsan']
|
|
self.ubsan_vptr = build_config['is_ubsan_vptr']
|
|
self.embedded_builtins = build_config['v8_enable_embedded_builtins']
|
|
self.verify_csa = build_config['v8_enable_verify_csa']
|
|
self.lite_mode = build_config['v8_enable_lite_mode']
|
|
self.pointer_compression = build_config['v8_enable_pointer_compression']
|
|
# Export only for MIPS target
|
|
if self.arch in ['mips', 'mipsel', 'mips64', 'mips64el']:
|
|
self.mips_arch_variant = build_config['mips_arch_variant']
|
|
self.mips_use_msa = build_config['mips_use_msa']
|
|
|
|
def __str__(self):
|
|
detected_options = []
|
|
|
|
if self.asan:
|
|
detected_options.append('asan')
|
|
if self.cfi_vptr:
|
|
detected_options.append('cfi_vptr')
|
|
if self.dcheck_always_on:
|
|
detected_options.append('dcheck_always_on')
|
|
if self.gcov_coverage:
|
|
detected_options.append('gcov_coverage')
|
|
if self.msan:
|
|
detected_options.append('msan')
|
|
if self.no_i18n:
|
|
detected_options.append('no_i18n')
|
|
if self.no_snap:
|
|
detected_options.append('no_snap')
|
|
if self.predictable:
|
|
detected_options.append('predictable')
|
|
if self.tsan:
|
|
detected_options.append('tsan')
|
|
if self.ubsan_vptr:
|
|
detected_options.append('ubsan_vptr')
|
|
if self.embedded_builtins:
|
|
detected_options.append('embedded_builtins')
|
|
if self.verify_csa:
|
|
detected_options.append('verify_csa')
|
|
if self.lite_mode:
|
|
detected_options.append('lite_mode')
|
|
if self.pointer_compression:
|
|
detected_options.append('pointer_compression')
|
|
|
|
return '\n'.join(detected_options)
|
|
|
|
|
|
class BaseTestRunner(object):
|
|
def __init__(self, basedir=None):
|
|
self.basedir = basedir or BASE_DIR
|
|
self.outdir = None
|
|
self.build_config = None
|
|
self.mode_name = None
|
|
self.mode_options = None
|
|
self.target_os = None
|
|
|
|
def execute(self, sys_args=None):
|
|
if sys_args is None: # pragma: no cover
|
|
sys_args = sys.argv[1:]
|
|
try:
|
|
parser = self._create_parser()
|
|
options, args = self._parse_args(parser, sys_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)
|
|
|
|
self._load_build_config(options)
|
|
command.setup(self.target_os)
|
|
|
|
try:
|
|
self._process_default_options(options)
|
|
self._process_options(options)
|
|
except TestRunnerError:
|
|
parser.print_help()
|
|
raise
|
|
|
|
args = self._parse_test_args(args)
|
|
suites = self._load_testsuites(args, options)
|
|
self._setup_env()
|
|
print(">>> Running tests for %s.%s" % (self.build_config.arch,
|
|
self.mode_name))
|
|
tests = [t for s in suites for t in s.tests]
|
|
return self._do_execute(tests, args, options)
|
|
except TestRunnerError:
|
|
return utils.EXIT_CODE_INTERNAL_ERROR
|
|
except KeyboardInterrupt:
|
|
return utils.EXIT_CODE_INTERRUPTED
|
|
finally:
|
|
command.tear_down()
|
|
|
|
def _create_parser(self):
|
|
parser = optparse.OptionParser()
|
|
parser.usage = '%prog [options] [tests]'
|
|
parser.description = """TESTS: %s""" % (TEST_MAP["default"])
|
|
self._add_parser_default_options(parser)
|
|
self._add_parser_options(parser)
|
|
return parser
|
|
|
|
def _add_parser_default_options(self, parser):
|
|
parser.add_option("--gn", help="Scan out.gn for the last built"
|
|
" configuration",
|
|
default=False, action="store_true")
|
|
parser.add_option("--outdir", help="Base directory with compile output",
|
|
default="out")
|
|
parser.add_option("--buildbot", help="DEPRECATED!",
|
|
default=False, action="store_true")
|
|
parser.add_option("--arch",
|
|
help="The architecture to run tests for")
|
|
parser.add_option("-m", "--mode",
|
|
help="The test mode in which to run (uppercase for ninja"
|
|
" and buildbot builds): %s" % MODES.keys())
|
|
parser.add_option("--shell-dir", help="DEPRECATED! Executables from build "
|
|
"directory will be used")
|
|
parser.add_option("--test-root", help="Root directory of the test suites",
|
|
default=os.path.join(self.basedir, 'test'))
|
|
parser.add_option("--total-timeout-sec", default=0, type="int",
|
|
help="How long should fuzzer run")
|
|
parser.add_option("--swarming", default=False, action="store_true",
|
|
help="Indicates running test driver on swarming.")
|
|
|
|
parser.add_option("-j", help="The number of parallel tasks to run",
|
|
default=0, type=int)
|
|
|
|
# Shard
|
|
parser.add_option("--shard-count", default=1, type=int,
|
|
help="Split tests into this number of shards")
|
|
parser.add_option("--shard-run", default=1, type=int,
|
|
help="Run this shard from the split up tests.")
|
|
|
|
# Progress
|
|
parser.add_option("-p", "--progress",
|
|
choices=PROGRESS_INDICATORS.keys(), default="mono",
|
|
help="The style of progress indicator (verbose, dots, "
|
|
"color, mono)")
|
|
parser.add_option("--json-test-results",
|
|
help="Path to a file for storing json results.")
|
|
parser.add_option("--junitout", help="File name of the JUnit output")
|
|
parser.add_option("--junittestsuite", default="v8tests",
|
|
help="The testsuite name in the JUnit output file")
|
|
parser.add_option("--exit-after-n-failures", type="int", default=100,
|
|
help="Exit after the first N failures instead of "
|
|
"running all tests. Pass 0 to disable this feature.")
|
|
|
|
# Rerun
|
|
parser.add_option("--rerun-failures-count", default=0, type=int,
|
|
help="Number of times to rerun each failing test case. "
|
|
"Very slow tests will be rerun only once.")
|
|
parser.add_option("--rerun-failures-max", default=100, type=int,
|
|
help="Maximum number of failing test cases to rerun")
|
|
|
|
# Test config
|
|
parser.add_option("--command-prefix", default="",
|
|
help="Prepended to each shell command used to run a test")
|
|
parser.add_option("--extra-flags", action="append", default=[],
|
|
help="Additional flags to pass to each test command")
|
|
parser.add_option("--isolates", action="store_true", default=False,
|
|
help="Whether to test isolates")
|
|
parser.add_option("--no-harness", "--noharness",
|
|
default=False, action="store_true",
|
|
help="Run without test harness of a given suite")
|
|
parser.add_option("--random-seed", default=0, type=int,
|
|
help="Default seed for initializing random generator")
|
|
parser.add_option("--run-skipped", help="Also run skipped tests.",
|
|
default=False, action="store_true")
|
|
parser.add_option("-t", "--timeout", default=60, type=int,
|
|
help="Timeout for single test in seconds")
|
|
parser.add_option("-v", "--verbose", default=False, action="store_true",
|
|
help="Verbose output")
|
|
|
|
# TODO(machenbach): Temporary options for rolling out new test runner
|
|
# features.
|
|
parser.add_option("--mastername", default='',
|
|
help="Mastername property from infrastructure. Not "
|
|
"setting this option indicates manual usage.")
|
|
parser.add_option("--buildername", default='',
|
|
help="Buildername property from infrastructure. Not "
|
|
"setting this option indicates manual usage.")
|
|
|
|
def _add_parser_options(self, parser):
|
|
pass
|
|
|
|
def _parse_args(self, parser, sys_args):
|
|
options, args = parser.parse_args(sys_args)
|
|
|
|
if any(map(lambda v: v and ',' in v,
|
|
[options.arch, options.mode])): # pragma: no cover
|
|
print 'Multiple arch/mode are deprecated'
|
|
raise TestRunnerError()
|
|
|
|
return options, args
|
|
|
|
def _load_build_config(self, options):
|
|
for outdir in self._possible_outdirs(options):
|
|
try:
|
|
self.build_config = self._do_load_build_config(outdir, options.verbose)
|
|
except TestRunnerError:
|
|
pass
|
|
|
|
if not self.build_config: # pragma: no cover
|
|
print 'Failed to load build config'
|
|
raise TestRunnerError
|
|
|
|
print 'Build found: %s' % self.outdir
|
|
if str(self.build_config):
|
|
print '>>> Autodetected:'
|
|
print self.build_config
|
|
|
|
# Represents the OS where tests are run on. Same as host OS except for
|
|
# Android, which is determined by build output.
|
|
if self.build_config.is_android:
|
|
self.target_os = 'android'
|
|
else:
|
|
self.target_os = utils.GuessOS()
|
|
|
|
# Returns possible build paths in order:
|
|
# gn
|
|
# outdir
|
|
# outdir/arch.mode
|
|
# Each path is provided in two versions: <path> and <path>/mode for buildbot.
|
|
def _possible_outdirs(self, options):
|
|
def outdirs():
|
|
if options.gn:
|
|
yield self._get_gn_outdir()
|
|
return
|
|
|
|
yield options.outdir
|
|
if options.arch and options.mode:
|
|
yield os.path.join(options.outdir,
|
|
'%s.%s' % (options.arch, options.mode))
|
|
|
|
for outdir in outdirs():
|
|
yield os.path.join(self.basedir, outdir)
|
|
|
|
# buildbot option
|
|
if options.mode:
|
|
yield os.path.join(self.basedir, outdir, options.mode)
|
|
|
|
def _get_gn_outdir(self):
|
|
gn_out_dir = os.path.join(self.basedir, DEFAULT_OUT_GN)
|
|
latest_timestamp = -1
|
|
latest_config = None
|
|
for gn_config in os.listdir(gn_out_dir):
|
|
gn_config_dir = os.path.join(gn_out_dir, gn_config)
|
|
if not os.path.isdir(gn_config_dir):
|
|
continue
|
|
if os.path.getmtime(gn_config_dir) > latest_timestamp:
|
|
latest_timestamp = os.path.getmtime(gn_config_dir)
|
|
latest_config = gn_config
|
|
if latest_config:
|
|
print(">>> Latest GN build found: %s" % latest_config)
|
|
return os.path.join(DEFAULT_OUT_GN, latest_config)
|
|
|
|
def _do_load_build_config(self, outdir, verbose=False):
|
|
build_config_path = os.path.join(outdir, "v8_build_config.json")
|
|
if not os.path.exists(build_config_path):
|
|
if verbose:
|
|
print("Didn't find build config: %s" % build_config_path)
|
|
raise TestRunnerError()
|
|
|
|
with open(build_config_path) as f:
|
|
try:
|
|
build_config_json = json.load(f)
|
|
except Exception: # pragma: no cover
|
|
print("%s exists but contains invalid json. Is your build up-to-date?"
|
|
% build_config_path)
|
|
raise TestRunnerError()
|
|
|
|
# In auto-detect mode the outdir is always where we found the build config.
|
|
# This ensures that we'll also take the build products from there.
|
|
self.outdir = os.path.dirname(build_config_path)
|
|
|
|
return BuildConfig(build_config_json)
|
|
|
|
def _process_default_options(self, options):
|
|
# We don't use the mode for more path-magic.
|
|
# Therefore transform the buildbot mode here to fix build_config value.
|
|
if options.mode:
|
|
options.mode = self._buildbot_to_v8_mode(options.mode)
|
|
|
|
build_config_mode = 'debug' if self.build_config.is_debug else 'release'
|
|
if options.mode:
|
|
if options.mode not in MODES: # pragma: no cover
|
|
print '%s mode is invalid' % options.mode
|
|
raise TestRunnerError()
|
|
if MODES[options.mode].execution_mode != build_config_mode:
|
|
print ('execution mode (%s) for %s is inconsistent with build config '
|
|
'(%s)' % (
|
|
MODES[options.mode].execution_mode,
|
|
options.mode,
|
|
build_config_mode))
|
|
raise TestRunnerError()
|
|
|
|
self.mode_name = options.mode
|
|
else:
|
|
self.mode_name = build_config_mode
|
|
|
|
self.mode_options = MODES[self.mode_name]
|
|
|
|
if options.arch and options.arch != self.build_config.arch:
|
|
print('--arch value (%s) inconsistent with build config (%s).' % (
|
|
options.arch, self.build_config.arch))
|
|
raise TestRunnerError()
|
|
|
|
if options.shell_dir: # pragma: no cover
|
|
print('Warning: --shell-dir is deprecated. Searching for executables in '
|
|
'build directory (%s) instead.' % self.outdir)
|
|
|
|
if options.j == 0:
|
|
if self.build_config.is_android:
|
|
# Adb isn't happy about multi-processed file pushing.
|
|
options.j = 1
|
|
else:
|
|
options.j = multiprocessing.cpu_count()
|
|
|
|
options.command_prefix = shlex.split(options.command_prefix)
|
|
options.extra_flags = sum(map(shlex.split, options.extra_flags), [])
|
|
|
|
def _buildbot_to_v8_mode(self, config):
|
|
"""Convert buildbot build configs to configs understood by the v8 runner.
|
|
|
|
V8 configs are always lower case and without the additional _x64 suffix
|
|
for 64 bit builds on windows with ninja.
|
|
"""
|
|
mode = config[:-4] if config.endswith('_x64') else config
|
|
return mode.lower()
|
|
|
|
def _process_options(self, options):
|
|
pass
|
|
|
|
def _setup_env(self):
|
|
# Use the v8 root as cwd as some test cases use "load" with relative paths.
|
|
os.chdir(self.basedir)
|
|
|
|
# Many tests assume an English interface.
|
|
os.environ['LANG'] = 'en_US.UTF-8'
|
|
|
|
symbolizer_option = self._get_external_symbolizer_option()
|
|
|
|
if self.build_config.asan:
|
|
asan_options = [
|
|
symbolizer_option,
|
|
'allow_user_segv_handler=1',
|
|
'allocator_may_return_null=1',
|
|
]
|
|
if not utils.GuessOS() in ['macos', 'windows']:
|
|
# LSAN is not available on mac and windows.
|
|
asan_options.append('detect_leaks=1')
|
|
else:
|
|
asan_options.append('detect_leaks=0')
|
|
os.environ['ASAN_OPTIONS'] = ":".join(asan_options)
|
|
|
|
if self.build_config.cfi_vptr:
|
|
os.environ['UBSAN_OPTIONS'] = ":".join([
|
|
'print_stacktrace=1',
|
|
'print_summary=1',
|
|
'symbolize=1',
|
|
symbolizer_option,
|
|
])
|
|
|
|
if self.build_config.ubsan_vptr:
|
|
os.environ['UBSAN_OPTIONS'] = ":".join([
|
|
'print_stacktrace=1',
|
|
symbolizer_option,
|
|
])
|
|
|
|
if self.build_config.msan:
|
|
os.environ['MSAN_OPTIONS'] = symbolizer_option
|
|
|
|
if self.build_config.tsan:
|
|
suppressions_file = os.path.join(
|
|
self.basedir,
|
|
'tools',
|
|
'sanitizers',
|
|
'tsan_suppressions.txt')
|
|
os.environ['TSAN_OPTIONS'] = " ".join([
|
|
symbolizer_option,
|
|
'suppressions=%s' % suppressions_file,
|
|
'exit_code=0',
|
|
'report_thread_leaks=0',
|
|
'history_size=7',
|
|
'report_destroy_locked=0',
|
|
])
|
|
|
|
def _get_external_symbolizer_option(self):
|
|
external_symbolizer_path = os.path.join(
|
|
self.basedir,
|
|
'third_party',
|
|
'llvm-build',
|
|
'Release+Asserts',
|
|
'bin',
|
|
'llvm-symbolizer',
|
|
)
|
|
|
|
if utils.IsWindows():
|
|
# Quote, because sanitizers might confuse colon as option separator.
|
|
external_symbolizer_path = '"%s.exe"' % external_symbolizer_path
|
|
|
|
return 'external_symbolizer_path=%s' % external_symbolizer_path
|
|
|
|
def _parse_test_args(self, args):
|
|
if not args:
|
|
args = self._get_default_suite_names()
|
|
|
|
# Expand arguments with grouped tests. The args should reflect the list
|
|
# of suites as otherwise filters would break.
|
|
def expand_test_group(name):
|
|
return TEST_MAP.get(name, [name])
|
|
|
|
return reduce(list.__add__, map(expand_test_group, args), [])
|
|
|
|
def _args_to_suite_names(self, args, test_root):
|
|
# Use default tests if no test configuration was provided at the cmd line.
|
|
all_names = set(utils.GetSuitePaths(test_root))
|
|
args_names = OrderedDict([(arg.split('/')[0], None) for arg in args]) # set
|
|
return [name for name in args_names if name in all_names]
|
|
|
|
def _get_default_suite_names(self):
|
|
return []
|
|
|
|
def _load_testsuites(self, args, options):
|
|
names = self._args_to_suite_names(args, options.test_root)
|
|
test_config = self._create_test_config(options)
|
|
variables = self._get_statusfile_variables(options)
|
|
suites = []
|
|
for name in names:
|
|
if options.verbose:
|
|
print '>>> Loading test suite: %s' % name
|
|
suite = testsuite.TestSuite.Load(
|
|
os.path.join(options.test_root, name), test_config)
|
|
|
|
if self._is_testsuite_supported(suite, options):
|
|
suite.load_tests_from_disk(variables)
|
|
suites.append(suite)
|
|
|
|
return suites
|
|
|
|
def _is_testsuite_supported(self, suite, options):
|
|
"""A predicate that can be overridden to filter out unsupported TestSuite
|
|
instances (see NumFuzzer for usage)."""
|
|
return True
|
|
|
|
def _get_statusfile_variables(self, options):
|
|
simd_mips = (
|
|
self.build_config.arch in ['mipsel', 'mips', 'mips64', 'mips64el'] and
|
|
self.build_config.mips_arch_variant == "r6" and
|
|
self.build_config.mips_use_msa)
|
|
|
|
mips_arch_variant = (
|
|
self.build_config.arch in ['mipsel', 'mips', 'mips64', 'mips64el'] and
|
|
self.build_config.mips_arch_variant)
|
|
|
|
# TODO(machenbach): In GN we can derive simulator run from
|
|
# target_arch != v8_target_arch in the dumped build config.
|
|
return {
|
|
"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,
|
|
"endurance_fuzzer": False,
|
|
"gc_fuzzer": False,
|
|
"gc_stress": False,
|
|
"gcov_coverage": self.build_config.gcov_coverage,
|
|
"isolates": options.isolates,
|
|
"mips_arch_variant": mips_arch_variant,
|
|
"mode": self.mode_options.status_mode
|
|
if not self.build_config.dcheck_always_on
|
|
else "debug",
|
|
"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": False,
|
|
"optimize_for_size": "--optimize-for-size" in options.extra_flags,
|
|
"predictable": self.build_config.predictable,
|
|
"simd_mips": simd_mips,
|
|
"simulator_run": False,
|
|
"system": self.target_os,
|
|
"tsan": self.build_config.tsan,
|
|
"ubsan_vptr": self.build_config.ubsan_vptr,
|
|
"embedded_builtins": self.build_config.embedded_builtins,
|
|
"verify_csa": self.build_config.verify_csa,
|
|
"lite_mode": self.build_config.lite_mode,
|
|
"pointer_compression": self.build_config.pointer_compression,
|
|
}
|
|
|
|
def _create_test_config(self, options):
|
|
timeout = options.timeout * self._timeout_scalefactor(options)
|
|
return TestConfig(
|
|
command_prefix=options.command_prefix,
|
|
extra_flags=options.extra_flags,
|
|
isolates=options.isolates,
|
|
mode_flags=self.mode_options.flags,
|
|
no_harness=options.no_harness,
|
|
noi18n=self.build_config.no_i18n,
|
|
random_seed=options.random_seed,
|
|
run_skipped=options.run_skipped,
|
|
shell_dir=self.outdir,
|
|
timeout=timeout,
|
|
verbose=options.verbose,
|
|
)
|
|
|
|
def _timeout_scalefactor(self, options):
|
|
factor = self.mode_options.timeout_scalefactor
|
|
|
|
# Simulators are slow, therefore allow a longer timeout.
|
|
if self.build_config.arch in SLOW_ARCHS:
|
|
factor *= 2
|
|
|
|
# Predictable mode is slower.
|
|
if self.build_config.predictable:
|
|
factor *= 2
|
|
|
|
return factor
|
|
|
|
# TODO(majeski): remove options & args parameters
|
|
def _do_execute(self, suites, args, options):
|
|
raise NotImplementedError()
|
|
|
|
def _prepare_procs(self, procs):
|
|
procs = filter(None, procs)
|
|
for i in xrange(0, len(procs) - 1):
|
|
procs[i].connect_to(procs[i + 1])
|
|
procs[0].setup()
|
|
|
|
def _create_shard_proc(self, options):
|
|
myid, count = self._get_shard_info(options)
|
|
if count == 1:
|
|
return None
|
|
return ShardProc(myid - 1, count)
|
|
|
|
def _get_shard_info(self, options):
|
|
"""
|
|
Returns pair:
|
|
(id of the current shard [1; number of shards], number of shards)
|
|
"""
|
|
# 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_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 1, 1
|
|
|
|
return shard_run, shard_count
|
|
|
|
def _create_progress_indicators(self, options):
|
|
procs = [PROGRESS_INDICATORS[options.progress]()]
|
|
if options.junitout:
|
|
procs.append(progress.JUnitTestProgressIndicator(options.junitout,
|
|
options.junittestsuite))
|
|
if options.json_test_results:
|
|
procs.append(progress.JsonTestProgressIndicator(
|
|
options.json_test_results,
|
|
self.build_config.arch,
|
|
self.mode_options.execution_mode))
|
|
return procs
|
|
|
|
def _create_result_tracker(self, options):
|
|
return progress.ResultsTracker(options.exit_after_n_failures)
|
|
|
|
def _create_timeout_proc(self, options):
|
|
if not options.total_timeout_sec:
|
|
return None
|
|
return TimeoutProc(options.total_timeout_sec)
|
|
|
|
def _create_signal_proc(self):
|
|
return SignalProc()
|
|
|
|
def _create_rerun_proc(self, options):
|
|
if not options.rerun_failures_count:
|
|
return None
|
|
return RerunProc(options.rerun_failures_count,
|
|
options.rerun_failures_max)
|