Handle cythonize(..., nthreads=1) for "spawn" start method #3262
Description
Configuration
- Cython 0.29.14
- Systems with
multiprocessing.get_start_method() == 'spawn':- generally affects Windows,
- since Python 3.8 also macOS with
multiprocessing's new default behavior (ref: bpo-33725: multiprocessing uses spawn by default on macOS python/cpython#13603).
Observed behavior
When
Cython.Build.cythonize(..., nthreads=1)is called- from a typical
setup.pyscript - that does not guard its execution with the
if __name__ == '__main__'idiom - and
multiprocessing's start method is set to"spawn",
then
- starting a new process will run the initial code again up until the next
cythonize/multiprocessing.Poolcall, - which will then repeatedly fail with a
RuntimeError: An attempt has been made to start a new process before the current process has finished its bootstrapping phase. This probably means that you are not using fork to start your child processes and you have forgotten to use the proper idiom in the main module: if __name__ == '__main__': freeze_support() ... The "freeze_support()" line can be omitted if the program is not going to be frozen to produce an executable. - practically ad infinitum due to
result.get(99999).
Small equivalent reproducer
Details
$ cat test.py
import sys
import multiprocessing
if sys.argv[1] != 'guard' or __name__ == '__main__':
multiprocessing.set_start_method(sys.argv[2], force=True)
with multiprocessing.Pool(1) as pool:
pool.map_async(id, [0], chunksize=1).get(float(sys.argv[3]))
$ python test.py guard fork 1
$ python test.py no-guard fork 1
$ python test.py guard spawn 1
$ python test.py no-guard spawn 1
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 105, in spawn_main
exitcode = _main(fd)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 114, in _main
prepare(preparation_data)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 225, in prepare
_fixup_main_from_path(data['init_main_from_path'])
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 277, in _fixup_main_from_path
run_name="__mp_main__")
File "/opt/conda/lib/python3.7/runpy.py", line 263, in run_path
pkg_name=pkg_name, script_name=fname)
File "/opt/conda/lib/python3.7/runpy.py", line 96, in _run_module_code
mod_name, mod_spec, pkg_name, script_name)
File "/opt/conda/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/conda/z/test.py", line 5, in <module>
with multiprocessing.Pool(1) as pool:
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 119, in Pool
context=self.get_context())
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 176, in __init__
self._repopulate_pool()
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 241, in _repopulate_pool
w.start()
File "/opt/conda/lib/python3.7/multiprocessing/process.py", line 112, in start
self._popen = self._Popen(self)
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 284, in _Popen
return Popen(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 32, in __init__
super().__init__(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_fork.py", line 20, in __init__
self._launch(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 42, in _launch
prep_data = spawn.get_preparation_data(process_obj._name)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 143, in get_preparation_data
_check_not_importing_main()
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 136, in _check_not_importing_main
is not going to be frozen to produce an executable.''')
RuntimeError:
An attempt has been made to start a new process before the
current process has finished its bootstrapping phase.
This probably means that you are not using fork to start your
child processes and you have forgotten to use the proper idiom
in the main module:
if __name__ == '__main__':
freeze_support()
...
The "freeze_support()" line can be omitted if the program
is not going to be frozen to produce an executable.
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 105, in spawn_main
exitcode = _main(fd)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 114, in _main
prepare(preparation_data)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 225, in prepare
_fixup_main_from_path(data['init_main_from_path'])
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 277, in _fixup_main_from_path
run_name="__mp_main__")
File "/opt/conda/lib/python3.7/runpy.py", line 263, in run_path
pkg_name=pkg_name, script_name=fname)
File "/opt/conda/lib/python3.7/runpy.py", line 96, in _run_module_code
mod_name, mod_spec, pkg_name, script_name)
File "/opt/conda/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/conda/z/test.py", line 5, in <module>
with multiprocessing.Pool(1) as pool:
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 119, in Pool
context=self.get_context())
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 176, in __init__
self._repopulate_pool()
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 241, in _repopulate_pool
w.start()
File "/opt/conda/lib/python3.7/multiprocessing/process.py", line 112, in start
self._popen = self._Popen(self)
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 284, in _Popen
return Popen(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 32, in __init__
super().__init__(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_fork.py", line 20, in __init__
self._launch(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 42, in _launch
prep_data = spawn.get_preparation_data(process_obj._name)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 143, in get_preparation_data
_check_not_importing_main()
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 136, in _check_not_importing_main
is not going to be frozen to produce an executable.''')
RuntimeError:
An attempt has been made to start a new process before the
current process has finished its bootstrapping phase.
This probably means that you are not using fork to start your
child processes and you have forgotten to use the proper idiom
in the main module:
if __name__ == '__main__':
freeze_support()
...
The "freeze_support()" line can be omitted if the program
is not going to be frozen to produce an executable.
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 105, in spawn_main
exitcode = _main(fd)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 114, in _main
prepare(preparation_data)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 225, in prepare
_fixup_main_from_path(data['init_main_from_path'])
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 277, in _fixup_main_from_path
run_name="__mp_main__")
File "/opt/conda/lib/python3.7/runpy.py", line 263, in run_path
pkg_name=pkg_name, script_name=fname)
File "/opt/conda/lib/python3.7/runpy.py", line 96, in _run_module_code
mod_name, mod_spec, pkg_name, script_name)
File "/opt/conda/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/conda/z/test.py", line 5, in <module>
with multiprocessing.Pool(1) as pool:
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 119, in Pool
context=self.get_context())
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 176, in __init__
self._repopulate_pool()
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 241, in _repopulate_pool
w.start()
File "/opt/conda/lib/python3.7/multiprocessing/process.py", line 112, in start
self._popen = self._Popen(self)
File "/opt/conda/lib/python3.7/multiprocessing/context.py", line 284, in _Popen
return Popen(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 32, in __init__
super().__init__(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_fork.py", line 20, in __init__
self._launch(process_obj)
File "/opt/conda/lib/python3.7/multiprocessing/popen_spawn_posix.py", line 42, in _launch
prep_data = spawn.get_preparation_data(process_obj._name)
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 143, in get_preparation_data
_check_not_importing_main()
File "/opt/conda/lib/python3.7/multiprocessing/spawn.py", line 136, in _check_not_importing_main
is not going to be frozen to produce an executable.''')
RuntimeError:
An attempt has been made to start a new process before the
current process has finished its bootstrapping phase.
This probably means that you are not using fork to start your
child processes and you have forgotten to use the proper idiom
in the main module:
if __name__ == '__main__':
freeze_support()
...
The "freeze_support()" line can be omitted if the program
is not going to be frozen to produce an executable.
Traceback (most recent call last):
File "test.py", line 6, in <module>
pool.map_async(id, [0], chunksize=1).get(float(sys.argv[3]))
File "/opt/conda/lib/python3.7/multiprocessing/pool.py", line 653, in get
raise TimeoutError
multiprocessing.context.TimeoutError
Possible solutions/workarounds:
- Disable parallel processing if
multiprocessing.get_start_method() == 'spawn'and display a warning. - Raise an error if
nthreads and multiprocessing.get_start_method() == 'spawn'. - Try launching a process and only if that not works, disable parallel processing, e.g.:
if nthreads and multiprocessing.get_start_method() == 'spawn': p = multiprocessing.Process(target=sys.exit, args=(0,)) p.start() p.join(1) if p.exitcode != 0: print('Some warning', file=sys.stderr) nthreads = 0 - Forcefully set
multiprocessing.set_start_method('fork', force=True).
Caveats:
re. 1.: If a script/module properly guards execution with if __name__ == '__main__':, there wouldn't be a need to disable multiprocessing. Many (/most?) setup.py don't, though.
re. 2.: Cython already only warns and continues with serial processing if nthreads is non-zero but multiprocessing isn't available. So raising a error seems inconsequential.
re. 3.: If a script, e.g., setup.py, already does some persistent changes (e.g., adding release information to files), those would be executed again, which is most likely undesired behavior.
re. 4.: Won't work on Windows. And it might be unexpected/undesired by the user to change this.
I favor approach 1 as it is the "least breaking" change. "Least breaking" in the sense that scipts with nthreads that work with macOS and Python up 3.7 would continue to work, albeit serially instead of parallely processed.