C++ Embedded 'time.sleep()' is not working on Windows host due to 'Py_InitializeEx(0)'

[hafizbilal100]

BPO	41686
Nosy	@pfmoore, @vstinner, @tjguk, @zware, @eryksun, @zooba
PRs	bpo-41686: Always create the SIGINT event on Windows #23344 bpo-41686: Refactor signal_exec() #23346 [3.9] bpo-41686: Always create the SIGINT event on Windows (GH-23344) #23347 [3.8] bpo-41686: Always create the SIGINT event on Windows (GH-23344) (GH-23347) #23349 bpo-41686: Move _Py_RestoreSignals() to signalmodule.c #23353

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2020-11-17.21:26:11.457>
created_at = <Date 2020-09-01.12:32:23.361>
labels = ['type-bug', '3.8', '3.9', '3.10', 'expert-C-API', 'OS-windows']
title = "C++ Embedded 'time.sleep()' is not working on Windows host due to 'Py_InitializeEx(0)'"
updated_at = <Date 2020-11-17.21:56:51.779>
user = 'https://bugs.python.org/hafizbilal100'

bugs.python.org fields:

activity = <Date 2020-11-17.21:56:51.779>
actor = 'vstinner'
assignee = 'none'
closed = True
closed_date = <Date 2020-11-17.21:26:11.457>
closer = 'vstinner'
components = ['Windows', 'C API']
creation = <Date 2020-09-01.12:32:23.361>
creator = 'hafizbilal100'
dependencies = []
files = []
hgrepos = []
issue_num = 41686
keywords = ['patch']
message_count = 13.0
messages = ['376189', '376196', '376219', '376220', '376232', '376265', '381257', '381265', '381266', '381286', '381288', '381292', '381294']
nosy_count = 7.0
nosy_names = ['paul.moore', 'vstinner', 'tim.golden', 'zach.ware', 'eryksun', 'steve.dower', 'hafizbilal100']
pr_nums = ['23344', '23346', '23347', '23349', '23353']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue41686'
versions = ['Python 3.8', 'Python 3.9', 'Python 3.10']

[hafizbilal100]

Hi,

time.sleep() function is not working on Windows host if I use C API Py_InitializeEx(0) function.

Issue could be reproduce with following simple example.

int main()
{
   Py_InitializeEx(0);
   PyRun_SimpleString("import time \n"
		      "start = time.time() \n"
		      "time.sleep(4) \n"
		      "print('elapsed: %d' % (time.time() - start))");
}

output:

elapsed: 0

Note, issue is not reproduce on Linux host.

[eryksun]

The SIGINT event gets created when the _signal extension module is imported. Until then, _PyOS_SigintEvent() returns NULL. But currently all code that calls _PyOS_SigintEvent() assumes it returns a valid handle. This has to be fixed to support Py_InitializeEx(0).

For example, pysleep could call WinAPI Sleep on the main thread if the interrupt event isn't configured:

        ul_millis = (unsigned long)millisecs;
        hInterruptEvent = _PyOS_SigintEvent();

        if (ul_millis == 0 || !hInterruptEvent || !_PyOS_IsMainThread()) {
            Py_BEGIN_ALLOW_THREADS
            Sleep(ul_millis);
            Py_END_ALLOW_THREADS
            break;
        }

[vstinner]

Py_InitializeEx() should be modified to always initialize sigint_event, even if install_sigs is equal to zero.

Maybe the variable should be moved somewhere else. Maybe in the time module?

[vstinner]

By the way, it would be nice to add error handling on:

    sigint_event = CreateEvent(NULL, TRUE, FALSE, FALSE);

See also PR bpo-22049.

[eryksun]

Maybe in the time module?

The SIGINT event is also needed by PyOS_Readline, _io (builtin), _winapi (builtin), and _multiprocessing (pyd). Is the time module guaranteed to be imported in 3.x? It appears to get imported incidentally via _PyImportZip_Init (zipimport).

Could sigint_event be relocated to a platform-dependent extension of _PyRuntimeState that gets initialized and finalized independent of any module?

[zooba]

Could sigint_event be relocated to a platform-dependent extension of _PyRuntimeState that gets initialized and finalized independent of any module?

This, or leaving it where it is and making sure it gets initialised (it really is about emulating certain signals). Either sounds fine to me.

[vstinner]

New changeset 0ae323b by Victor Stinner in branch 'master':
bpo-41686: Always create the SIGINT event on Windows (GH-23344)
0ae323b

[vstinner]

New changeset cda23be by Victor Stinner in branch 'master':
bpo-41686: Refactor signal_exec() (GH-23346)
cda23be

[vstinner]

New changeset 05a5d69 by Victor Stinner in branch '3.9':
bpo-41686: Always create the SIGINT event on Windows (GH-23344) (GH-23347)
05a5d69

[vstinner]

New changeset a702bd4 by Victor Stinner in branch '3.8':
bpo-41686: Always create the SIGINT event on Windows (GH-23344) (GH-23347) (GH-23349)
a702bd4

[vstinner]

Ok, the issue should now be fixed in 3.8, 3.9 and master branches. Thanks for the bug report hafiz bilal.

[vstinner]

New changeset 29aa624 by Victor Stinner in branch 'master':
bpo-41686: Move _Py_RestoreSignals() to signalmodule.c (GH-23353)
29aa624

[vstinner]

bpo-41686: Move _Py_RestoreSignals() to signalmodule.c (GH-23353)

Sorry, this change was for bpo-41713.