mirror of
https://github.com/verilator/verilator.git
synced 2025-04-05 12:12:39 +00:00
Tests: Improve driver performance with forkserver
This commit is contained in:
Wilson Snyder
parent
1543b202b2
commit
e990415ee0
@ -48,6 +48,8 @@ All_Scenarios = {
|
||||
test = None
|
||||
Arg_Tests = []
|
||||
Arg_Driver_Verilator_Flags = []
|
||||
Quitting = False
|
||||
Vltmt_Threads = 3
|
||||
|
||||
# So an 'import vltest_bootstrap' inside test files will do nothing
|
||||
sys.modules['vltest_bootstrap'] = {}
|
||||
@ -223,14 +225,16 @@ class Capabilities:
|
||||
|
||||
class Forker:
|
||||
|
||||
class Process:
|
||||
class Job:
|
||||
|
||||
def __init__(self, _id, name, scenario, rerun_skipping, run_pre_start, run_on_start,
|
||||
run_on_finish):
|
||||
def __init__(self, _id, name, scenario, args, quiet, rerun_skipping, run_pre_start,
|
||||
run_on_start, run_on_finish):
|
||||
self.fail_max_skip = False
|
||||
self.id = _id
|
||||
self.name = name
|
||||
self.scenario = scenario
|
||||
self.args = args
|
||||
self.quiet = quiet
|
||||
self.rerun_skipping = rerun_skipping
|
||||
self.run_pre_start = run_pre_start
|
||||
self.run_on_start = run_on_start
|
||||
@ -248,8 +252,8 @@ class Forker:
|
||||
def __init__(self, max_processes):
|
||||
self._max_processes = max_processes
|
||||
self._id_next = 0
|
||||
self._left = collections.deque() # deque of Process
|
||||
self._running = {} # key of pid, value of Process
|
||||
self._left = collections.deque() # deque of Job
|
||||
self._running = {} # key of pid, value of Job
|
||||
|
||||
def is_any_left(self) -> bool:
|
||||
return self.num_running() > 0 or (len(self._left) > 0 and not Quitting)
|
||||
@ -257,21 +261,27 @@ class Forker:
|
||||
def max_proc(self, n: int) -> None:
|
||||
self._max_processes = n
|
||||
|
||||
def poll(self) -> None:
|
||||
def poll(self) -> bool:
|
||||
"""Run threads, returning if more work to do (if False, sleep)"""
|
||||
# We don't use SIGCHLD as conflicted with other handler, instead just poll
|
||||
completed = [] # Need two passes to avoid changing the list we are iterating
|
||||
nrunning = 0
|
||||
more_now = False
|
||||
for process in self._running.values():
|
||||
if process.exitcode is not None:
|
||||
completed.append(process)
|
||||
more_now = True
|
||||
else:
|
||||
nrunning += 1
|
||||
for process in completed:
|
||||
self._finished(process)
|
||||
# Start new work now, so running in background while we then collect completions
|
||||
while len(self._left) and nrunning < self._max_processes and not Quitting:
|
||||
process = self._left.popleft()
|
||||
self._run(process)
|
||||
nrunning += 1
|
||||
more_now = True
|
||||
for process in completed:
|
||||
self._finished(process)
|
||||
return more_now
|
||||
|
||||
def running(self) -> list:
|
||||
return self._running.values()
|
||||
@ -279,15 +289,18 @@ class Forker:
|
||||
def num_running(self) -> int:
|
||||
return len(self._running)
|
||||
|
||||
def schedule(self, name, scenario, rerun_skipping, run_pre_start, run_on_start, run_on_finish):
|
||||
def schedule(self, name, scenario, args, quiet, rerun_skipping, run_pre_start, run_on_start,
|
||||
run_on_finish):
|
||||
# print("-Forker::schedule: [" + name + "]")
|
||||
process = Forker.Process(self._id_next,
|
||||
name=name,
|
||||
scenario=scenario,
|
||||
rerun_skipping=rerun_skipping,
|
||||
run_pre_start=run_pre_start,
|
||||
run_on_start=run_on_start,
|
||||
run_on_finish=run_on_finish)
|
||||
process = Forker.Job(self._id_next,
|
||||
name=name,
|
||||
scenario=scenario,
|
||||
args=args,
|
||||
quiet=quiet,
|
||||
rerun_skipping=rerun_skipping,
|
||||
run_pre_start=run_pre_start,
|
||||
run_on_start=run_on_start,
|
||||
run_on_finish=run_on_finish)
|
||||
self._id_next += 1
|
||||
self._left.append(process)
|
||||
|
||||
@ -300,8 +313,10 @@ class Forker:
|
||||
# print("-Forker: [" + process.name + "] run_pre_start")
|
||||
process.run_pre_start(process)
|
||||
|
||||
ctx = multiprocessing.get_context('fork')
|
||||
process.mprocess = ctx.Process(target=lambda: self.child_start(process))
|
||||
ctx = multiprocessing.get_context('forkserver')
|
||||
process.mprocess = ctx.Process( #
|
||||
target=forker.child_start, # pylint: disable=used-before-assignment
|
||||
args=(process, ))
|
||||
process.mprocess.start()
|
||||
|
||||
# print("-Forker: [" + process.name + "] RUNNING pid=" + str(process.pid))
|
||||
@ -355,6 +370,8 @@ class Runner:
|
||||
self.left_cnt += 1
|
||||
forker.schedule(name=py_filename,
|
||||
scenario=scenario,
|
||||
args=Args,
|
||||
quiet=self.quiet,
|
||||
rerun_skipping=rerun_skipping,
|
||||
run_pre_start=self._run_pre_start_static,
|
||||
run_on_start=self._run_on_start_static,
|
||||
@ -380,15 +397,15 @@ class Runner:
|
||||
|
||||
@staticmethod
|
||||
def _run_on_start_static(process) -> None:
|
||||
Runner.runner._run_on_start(process) # pylint: disable=protected-access
|
||||
|
||||
def _run_on_start(self, process) -> None:
|
||||
# Running in context of child, so can't pass data to parent directly
|
||||
if self.quiet:
|
||||
if process.quiet:
|
||||
sys.stdout = open(os.devnull, 'w') # pylint: disable=R1732,unspecified-encoding
|
||||
sys.stderr = open(os.devnull, 'w') # pylint: disable=R1732,unspecified-encoding
|
||||
|
||||
print("=" * 70)
|
||||
global Args
|
||||
Args = process.args
|
||||
|
||||
global test
|
||||
test = VlTest(py_filename=process.name,
|
||||
scenario=process.scenario,
|
||||
@ -396,9 +413,9 @@ class Runner:
|
||||
test.oprint("=" * 50)
|
||||
test._prep()
|
||||
if process.rerun_skipping:
|
||||
print(" ---------- Earlier logfiles below; test was rerunnable = 0\n")
|
||||
os.system("cat $test->{obj_dir}/*.log")
|
||||
print(" ---------- Earlier logfiles above; test was rerunnable = 0\n")
|
||||
print(" ---------- Earlier logfiles below; test was rerunnable = False\n")
|
||||
os.system("cat " + test.obj_dir + "/*.log")
|
||||
print(" ---------- Earlier logfiles above; test was rerunnable = False\n")
|
||||
elif process.fail_max_skip:
|
||||
test.skip("Too many test failures; exceeded --fail-max")
|
||||
else:
|
||||
@ -457,10 +474,11 @@ class Runner:
|
||||
self.print_summary(force=True)
|
||||
# Wait for all children to finish
|
||||
while forker.is_any_left():
|
||||
forker.poll()
|
||||
if not interactive_debugger:
|
||||
more_now = forker.poll()
|
||||
if not Args.interactive_debugger:
|
||||
self.print_summary(force=False)
|
||||
time.sleep(0.1)
|
||||
if not more_now:
|
||||
time.sleep(0.1)
|
||||
self.report(None)
|
||||
self.report(self.driver_log_filename)
|
||||
|
||||
@ -505,7 +523,7 @@ class Runner:
|
||||
@staticmethod
|
||||
def _py_filename_adjust(py_filename: str,
|
||||
tdir_def: str) -> list: # Return (py_filename, t_dir)
|
||||
for tdir in Test_Dirs: # pylint: disable=redefined-outer-name
|
||||
for tdir in Args.test_dirs: # pylint: disable=redefined-outer-name
|
||||
# t_dir used both absolutely and under obj_dir
|
||||
try_py_filename = tdir + "/" + os.path.basename(py_filename)
|
||||
if os.path.exists(try_py_filename):
|
||||
@ -517,7 +535,7 @@ class Runner:
|
||||
return (py_filename, os.path.abspath(tdir_def))
|
||||
|
||||
def sprint_summary(self) -> str:
|
||||
delta = time.time() - Start
|
||||
delta = time.time() - Start # pylint: disable=used-before-assignment
|
||||
# Fudge of 120% works out about right so ETA correctly predicts completion time
|
||||
eta = 1.2 * ((self.all_cnt * (delta / ((self.all_cnt - self.left_cnt) + 0.001))) - delta)
|
||||
if delta < 10:
|
||||
@ -542,7 +560,7 @@ class Runner:
|
||||
def _manual_args(self) -> str:
|
||||
# Return command line with scenarios stripped
|
||||
out = []
|
||||
for oarg in Orig_ARGV_Sw:
|
||||
for oarg in Args.orig_argv_sw:
|
||||
showarg = True
|
||||
for val in All_Scenarios.values():
|
||||
for allscarg in val:
|
||||
@ -623,7 +641,7 @@ class VlTest:
|
||||
self.vlt_all = self.vlt or self.vltmt # Any Verilator scenario
|
||||
|
||||
(self.py_filename, self.t_dir) = Runner._py_filename_adjust(self.py_filename, ".")
|
||||
for tdir in Test_Dirs: # pylint: disable=redefined-outer-name
|
||||
for tdir in Args.test_dirs: # pylint: disable=redefined-outer-name
|
||||
# t_dir used both absolutely and under obj_dir
|
||||
self.t_dir = None
|
||||
if os.path.exists(tdir + "/" + self.name + ".py"):
|
||||
@ -1682,7 +1700,7 @@ class VlTest:
|
||||
data = rawbuf[0:got]
|
||||
if tee:
|
||||
sys.stdout.write(data.decode('latin-1'))
|
||||
if interactive_debugger:
|
||||
if Args.interactive_debugger:
|
||||
sys.stdout.flush()
|
||||
if logfh:
|
||||
logfh.write(data)
|
||||
@ -1728,8 +1746,8 @@ class VlTest:
|
||||
|
||||
# Read the log file a couple of times to allow for NFS delays
|
||||
if check_finished:
|
||||
for tryn in range(self.tries() - 1, -1, -1):
|
||||
if tryn != self.tries() - 1:
|
||||
for tryn in range(Args.log_retries - 1, -1, -1):
|
||||
if tryn != Args.log_retries - 1:
|
||||
time.sleep(1)
|
||||
moretry = tryn != 0
|
||||
if not self._run_log_try(logfile, check_finished, moretry):
|
||||
@ -2122,8 +2140,8 @@ class VlTest:
|
||||
|
||||
def files_identical(self, fn1: str, fn2: str, is_logfile=False) -> None:
|
||||
"""Test if two files have identical contents"""
|
||||
for tryn in range(self.tries() - 1, -1, -1):
|
||||
if tryn != self.tries() - 1:
|
||||
for tryn in range(Args.log_retries, -1, -1):
|
||||
if tryn != Args.log_retries - 1:
|
||||
time.sleep(1)
|
||||
moretry = tryn != 0
|
||||
if not self._files_identical_try(
|
||||
@ -2139,7 +2157,7 @@ class VlTest:
|
||||
f1 = None
|
||||
if not moretry:
|
||||
self.error("Files_identical file does not exist: " + fn1)
|
||||
return True
|
||||
return True # Retry
|
||||
try:
|
||||
f2 = open( # pylint: disable=consider-using-with
|
||||
fn2, 'r', encoding='latin-1', newline='\n')
|
||||
@ -2148,7 +2166,7 @@ class VlTest:
|
||||
if not moretry:
|
||||
self.copy_if_golden(fn1, fn2)
|
||||
self.error("Files_identical file does not exist: " + fn2)
|
||||
return True
|
||||
return True # Retry
|
||||
again = self._files_identical_reader(f1,
|
||||
f2,
|
||||
fn1=fn1,
|
||||
@ -2213,7 +2231,7 @@ class VlTest:
|
||||
if l1 != l2:
|
||||
# print(" clnGOT="+pformat(l1s)+"\n clnEXP="+pformat(l2s))
|
||||
if moretry:
|
||||
return True
|
||||
return True # Retry
|
||||
self.error_keep_going("Line " + str(lineno_m1) + " miscompares; " + fn1 + " != " +
|
||||
fn2)
|
||||
for c in range(min(len(l1), len(l2))):
|
||||
@ -2231,9 +2249,9 @@ class VlTest:
|
||||
else:
|
||||
print("To update reference: HARNESS_UPDATE_GOLDEN=1 {command} or --golden",
|
||||
file=sys.stderr)
|
||||
return False
|
||||
return False # No retry - bad
|
||||
|
||||
return False # Good
|
||||
return False # No retry - good
|
||||
|
||||
def files_identical_sorted(self, fn1: str, fn2: str, is_logfile=False) -> None:
|
||||
"""Test if two files, after sorting both, have identical contents"""
|
||||
@ -2379,16 +2397,6 @@ class VlTest:
|
||||
VlTest._cached_cfg_with_ccache = bool(re.match(r'OBJCACHE \?= ccache', mkf))
|
||||
return VlTest._cached_cfg_with_ccache
|
||||
|
||||
@staticmethod
|
||||
def tries() -> int:
|
||||
# Number of retries when reading logfiles, generally only need many
|
||||
# retries when system is busy running a lot of tests
|
||||
if not forker.running:
|
||||
return 2
|
||||
if len(Arg_Tests) > 3:
|
||||
return 7
|
||||
return 2
|
||||
|
||||
def glob_some(self, pattern: str) -> list:
|
||||
"""Return list of filenames matching a glob, with at least one match required."""
|
||||
files = glob.glob(pattern)
|
||||
@ -2653,156 +2661,162 @@ def run_them() -> None:
|
||||
|
||||
######################################################################
|
||||
######################################################################
|
||||
# Globals
|
||||
# Main
|
||||
|
||||
os.environ['PYTHONUNBUFFERED'] = "1"
|
||||
if __name__ == '__main__':
|
||||
os.environ['PYTHONUNBUFFERED'] = "1"
|
||||
|
||||
if ('VERILATOR_ROOT' not in os.environ) and os.path.isfile('../bin/verilator'):
|
||||
os.environ['VERILATOR_ROOT'] = os.getcwd() + "/.."
|
||||
if 'MAKE' not in os.environ:
|
||||
os.environ['MAKE'] = "make"
|
||||
if 'CXX' not in os.environ:
|
||||
os.environ['CXX'] = "c++"
|
||||
if 'TEST_REGRESS' in os.environ:
|
||||
sys.exit("%Error: TEST_REGRESS environment variable is already set")
|
||||
os.environ['TEST_REGRESS'] = os.getcwd()
|
||||
if ('VERILATOR_ROOT' not in os.environ) and os.path.isfile('../bin/verilator'):
|
||||
os.environ['VERILATOR_ROOT'] = os.getcwd() + "/.."
|
||||
if 'MAKE' not in os.environ:
|
||||
os.environ['MAKE'] = "make"
|
||||
if 'CXX' not in os.environ:
|
||||
os.environ['CXX'] = "c++"
|
||||
if 'TEST_REGRESS' in os.environ:
|
||||
sys.exit("%Error: TEST_REGRESS environment variable is already set")
|
||||
os.environ['TEST_REGRESS'] = os.getcwd()
|
||||
|
||||
forker = None
|
||||
Start = time.time()
|
||||
Vltmt_Threads = 3
|
||||
_Parameter_Next_Level = None
|
||||
Quitting = False
|
||||
forker = None
|
||||
Start = time.time()
|
||||
_Parameter_Next_Level = None
|
||||
|
||||
def sig_int(signum, env) -> None: # pylint: disable=unused-argument
|
||||
global Quitting
|
||||
if Quitting:
|
||||
sys.exit("\nQuitting (immediately)...")
|
||||
print("\nQuitting... (send another interrupt signal for immediate quit)")
|
||||
Quitting = True
|
||||
if forker:
|
||||
forker.kill_tree_all()
|
||||
|
||||
def sig_int(signum, env) -> None: # pylint: disable=unused-argument
|
||||
global Quitting
|
||||
if Quitting:
|
||||
sys.exit("\nQuitting (immediately)...")
|
||||
print("\nQuitting... (send another interrupt signal for immediate quit)")
|
||||
Quitting = True
|
||||
if forker:
|
||||
forker.kill_tree_all()
|
||||
signal.signal(signal.SIGINT, sig_int)
|
||||
|
||||
#---------------------------------------------------------------------
|
||||
|
||||
signal.signal(signal.SIGINT, sig_int)
|
||||
parser = argparse.ArgumentParser(
|
||||
allow_abbrev=False,
|
||||
formatter_class=argparse.RawDescriptionHelpFormatter,
|
||||
description="""Run Verilator regression tests""",
|
||||
epilog="""driver.py invokes Verilator or another simulator on each test file.
|
||||
See docs/internals.rst in the distribution for more information.
|
||||
|
||||
Orig_ARGV_Sw = []
|
||||
for arg in sys.argv:
|
||||
if re.match(r'^-', arg) and not re.match(r'^-j', arg):
|
||||
Orig_ARGV_Sw.append(arg)
|
||||
Copyright 2024-2024 by Wilson Snyder. This program is free software; you
|
||||
can redistribute it and/or modify it under the terms of either the GNU
|
||||
Lesser General Public License Version 3 or the Perl Artistic License
|
||||
Version 2.0.
|
||||
|
||||
#---------------------------------------------------------------------
|
||||
SPDX-License-Identifier: LGPL-3.0-only OR Artistic-2.0""")
|
||||
|
||||
parser = argparse.ArgumentParser(
|
||||
allow_abbrev=False,
|
||||
formatter_class=argparse.RawDescriptionHelpFormatter,
|
||||
description="""Run Verilator regression tests""",
|
||||
epilog="""driver.py invokes Verilator or another simulator on each test file.
|
||||
See docs/internals.rst in the distribution for more information.
|
||||
parser.add_argument('--benchmark', action='store', help='enable benchmarking')
|
||||
parser.add_argument('--debug', action='store_const', const=9, help='enable debug')
|
||||
# --debugi: see _parameter()
|
||||
parser.add_argument('--fail-max',
|
||||
action='store',
|
||||
default=None,
|
||||
help='run Verilator executable with gdb')
|
||||
parser.add_argument('--gdb', action='store_true', help='run Verilator executable with gdb')
|
||||
parser.add_argument('--gdbbt',
|
||||
action='store_true',
|
||||
help='run Verilated executable with gdb and backtrace')
|
||||
parser.add_argument('--gdbsim', action='store_true', help='run Verilated executable with gdb')
|
||||
parser.add_argument('--golden', '--gold', action='store_true', help='update golden .out files')
|
||||
parser.add_argument('--hashset', action='store', help='split tests based on <set>/<numsets>')
|
||||
parser.add_argument('--jobs',
|
||||
'-j',
|
||||
action='store',
|
||||
default=0,
|
||||
type=int,
|
||||
help='parallel job count (0=cpu count)')
|
||||
parser.add_argument('--quiet',
|
||||
action='store_true',
|
||||
help='suppress output except failures and progress')
|
||||
parser.add_argument('--rerun', action='store_true', help='rerun all tests that fail')
|
||||
parser.add_argument('--rr', action='store_true', help='run Verilator executable with rr')
|
||||
parser.add_argument('--rrsim', action='store_true', help='run Verilated executable with rr')
|
||||
parser.add_argument('--sanitize', action='store_true', help='run address sanitizer')
|
||||
parser.add_argument('--site',
|
||||
action='store_true',
|
||||
help='include VERILATOR_TEST_SITE test list')
|
||||
parser.add_argument('--stop', action='store_true', help='stop on the first error')
|
||||
parser.add_argument('--trace', action='store_true', help='enable simulator waveform tracing')
|
||||
parser.add_argument('--verbose',
|
||||
action='store_true',
|
||||
help='compile and run test in verbose mode')
|
||||
parser.add_argument(
|
||||
'--verilation', # -no-verilation undocumented debugging
|
||||
action='store_true',
|
||||
default=True,
|
||||
help="don't run verilator compile() phase")
|
||||
parser.add_argument('--verilated-debug',
|
||||
action='store_true',
|
||||
help='enable Verilated executable debug')
|
||||
## Scenarios
|
||||
for scen, v in All_Scenarios.items():
|
||||
parser.add_argument('--' + scen,
|
||||
dest='scenarios',
|
||||
action='append_const',
|
||||
const=scen,
|
||||
help='scenario-enable ' + scen)
|
||||
|
||||
Copyright 2024-2024 by Wilson Snyder. This program is free software; you
|
||||
can redistribute it and/or modify it under the terms of either the GNU
|
||||
Lesser General Public License Version 3 or the Perl Artistic License
|
||||
Version 2.0.
|
||||
(Args, rest) = parser.parse_known_intermixed_args()
|
||||
|
||||
SPDX-License-Identifier: LGPL-3.0-only OR Artistic-2.0""")
|
||||
for arg in rest:
|
||||
_parameter(arg)
|
||||
|
||||
parser.add_argument('--benchmark', action='store', help='enable benchmarking')
|
||||
parser.add_argument('--debug', action='store_const', const=9, help='enable debug')
|
||||
# --debugi: see _parameter()
|
||||
parser.add_argument('--fail-max',
|
||||
action='store',
|
||||
default=None,
|
||||
help='run Verilator executable with gdb')
|
||||
parser.add_argument('--gdb', action='store_true', help='run Verilator executable with gdb')
|
||||
parser.add_argument('--gdbbt',
|
||||
action='store_true',
|
||||
help='run Verilated executable with gdb and backtrace')
|
||||
parser.add_argument('--gdbsim', action='store_true', help='run Verilated executable with gdb')
|
||||
parser.add_argument('--golden', '--gold', action='store_true', help='update golden .out files')
|
||||
parser.add_argument('--hashset', action='store', help='split tests based on <set>/<numsets>')
|
||||
parser.add_argument('--jobs',
|
||||
'-j',
|
||||
action='store',
|
||||
default=0,
|
||||
type=int,
|
||||
help='parallel job count (0=cpu count)')
|
||||
parser.add_argument('--quiet',
|
||||
action='store_true',
|
||||
help='suppress output except failures and progress')
|
||||
parser.add_argument('--rerun', action='store_true', help='rerun all tests that fail')
|
||||
parser.add_argument('--rr', action='store_true', help='run Verilator executable with rr')
|
||||
parser.add_argument('--rrsim', action='store_true', help='run Verilated executable with rr')
|
||||
parser.add_argument('--sanitize', action='store_true', help='run address sanitizer')
|
||||
parser.add_argument('--site', action='store_true', help='include VERILATOR_TEST_SITE test list')
|
||||
parser.add_argument('--stop', action='store_true', help='stop on the first error')
|
||||
parser.add_argument('--trace', action='store_true', help='enable simulator waveform tracing')
|
||||
parser.add_argument('--verbose', action='store_true', help='compile and run test in verbose mode')
|
||||
parser.add_argument(
|
||||
'--verilation', # -no-verilation undocumented debugging
|
||||
action='store_true',
|
||||
default=True,
|
||||
help="don't run verilator compile() phase")
|
||||
parser.add_argument('--verilated-debug',
|
||||
action='store_true',
|
||||
help='enable Verilated executable debug')
|
||||
## Scenarios
|
||||
for scen, v in All_Scenarios.items():
|
||||
parser.add_argument('--' + scen,
|
||||
dest='scenarios',
|
||||
action='append_const',
|
||||
const=scen,
|
||||
help='scenario-enable ' + scen)
|
||||
if Args.debug:
|
||||
Arg_Driver_Verilator_Flags.append("--debug --no-skip-identical")
|
||||
logging.basicConfig(level=logging.DEBUG)
|
||||
logging.info("In driver.py, ARGV=" + ' '.join(sys.argv))
|
||||
|
||||
(Args, rest) = parser.parse_known_intermixed_args()
|
||||
# Use Args for some global information, so gets passed to forked process
|
||||
Args.interactive_debugger = Args.gdb or Args.gdbsim or Args.rr or Args.rrsim
|
||||
if Args.jobs > 1 and Args.interactive_debugger:
|
||||
sys.exit("%Error: Unable to use -j > 1 with --gdb* and --rr* options")
|
||||
|
||||
for arg in rest:
|
||||
_parameter(arg)
|
||||
if Args.golden:
|
||||
os.environ['HARNESS_UPDATE_GOLDEN'] = '1'
|
||||
if Args.jobs == 0:
|
||||
Args.jobs = 1 if Args.interactive_debugger else calc_jobs()
|
||||
if not Args.scenarios:
|
||||
Args.scenarios = []
|
||||
Args.scenarios.append('dist')
|
||||
Args.scenarios.append('vlt')
|
||||
|
||||
if Args.debug:
|
||||
Arg_Driver_Verilator_Flags.append("--debug --no-skip-identical")
|
||||
logging.basicConfig(level=logging.DEBUG)
|
||||
logging.info("In driver.py, ARGV=" + ' '.join(sys.argv))
|
||||
Args.orig_argv_sw = []
|
||||
for arg in sys.argv:
|
||||
if re.match(r'^-', arg) and not re.match(r'^-j', arg):
|
||||
Args.orig_argv_sw.append(arg)
|
||||
|
||||
interactive_debugger = Args.gdb or Args.gdbsim or Args.rr or Args.rrsim
|
||||
if Args.jobs > 1 and interactive_debugger:
|
||||
sys.exit("%Error: Unable to use -j > 1 with --gdb* and --rr* options")
|
||||
Args.test_dirs = ["t"]
|
||||
if 'VERILATOR_TESTS_SITE' in os.environ:
|
||||
if Args.site or len(Arg_Tests) >= 1:
|
||||
for tdir in os.environ['VERILATOR_TESTS_SITE'].split(':'):
|
||||
Args.test_dirs.append(tdir)
|
||||
|
||||
if Args.golden:
|
||||
os.environ['HARNESS_UPDATE_GOLDEN'] = '1'
|
||||
if Args.jobs == 0:
|
||||
Args.jobs = 1 if interactive_debugger else calc_jobs()
|
||||
if not Args.scenarios:
|
||||
Args.scenarios = []
|
||||
Args.scenarios.append('dist')
|
||||
Args.scenarios.append('vlt')
|
||||
if not Arg_Tests: # Run everything
|
||||
uniq = {}
|
||||
for tdir in Args.test_dirs:
|
||||
# Uniquify by inode, so different paths to same place get combined
|
||||
stats = os.stat(tdir)
|
||||
if stats.st_ino not in uniq:
|
||||
uniq[stats.st_ino] = 1
|
||||
Arg_Tests += sorted(glob.glob(tdir + "/t_*.py"))
|
||||
if Args.hashset:
|
||||
_calc_hashset()
|
||||
|
||||
forker = Forker(Args.jobs)
|
||||
# Number of retries when reading logfiles, generally only need many
|
||||
# retries when system is busy running a lot of tests
|
||||
Args.log_retries = 10 if (len(Arg_Tests) > 3) else 2
|
||||
|
||||
Test_Dirs = ["t"]
|
||||
if 'VERILATOR_TESTS_SITE' in os.environ:
|
||||
if Args.site or len(Arg_Tests) >= 1:
|
||||
for tdir in os.environ['VERILATOR_TESTS_SITE'].split(':'):
|
||||
Test_Dirs.append(tdir)
|
||||
forker = Forker(Args.jobs)
|
||||
|
||||
if not Arg_Tests: # Run everything
|
||||
uniq = {}
|
||||
for tdir in Test_Dirs:
|
||||
# Uniquify by inode, so different paths to same place get combined
|
||||
stats = os.stat(tdir)
|
||||
if stats.st_ino not in uniq:
|
||||
uniq[stats.st_ino] = 1
|
||||
Arg_Tests += sorted(glob.glob(tdir + "/t_*.py"))
|
||||
if Args.hashset:
|
||||
_calc_hashset()
|
||||
if len(Arg_Tests) >= 2 and Args.jobs >= 2:
|
||||
# Read supported into master process, so don't call every subprocess
|
||||
Capabilities.warmup_cache()
|
||||
# Without this tests such as t_debug_sigsegv_bt_bad.py will occasionally
|
||||
# block on input and cause a SIGSTOP, then a "fg" was needed to resume testing.
|
||||
print("== Many jobs; redirecting STDIN", file=sys.stderr)
|
||||
#
|
||||
sys.stdin = open("/dev/null", 'r', encoding="utf8") # pylint: disable=consider-using-with
|
||||
|
||||
if len(Arg_Tests) >= 2 and Args.jobs >= 2:
|
||||
# Read supported into master process, so don't call every subprocess
|
||||
Capabilities.warmup_cache()
|
||||
# Without this tests such as t_debug_sigsegv_bt_bad.py will occasionally
|
||||
# block on input and cause a SIGSTOP, then a "fg" was needed to resume testing.
|
||||
print("== Many jobs; redirecting STDIN", file=sys.stderr)
|
||||
#
|
||||
sys.stdin = open("/dev/null", 'r', encoding="utf8") # pylint: disable=consider-using-with
|
||||
|
||||
run_them()
|
||||
run_them()
|
||||
|
||||
Loading…
Reference in New Issue
Block a user