Qt Ctrl-C broken on windows

[tacaswell]

@anntzer @tacaswell
Hi! I have just tested the master branch on Windows and found that there is indeed a problem with the QSocketNotifier (it appeared here: #13306 (comment)). Previously, on that Windows machine, I was running my old code with QAbstractSocket which was fine, and did not test the newest version with QSocketNotifier there (my fault, sorry). I have tested on Linux, and there is no such problem there.

The code to reproduce on Windows (in Jupyter):

%matplotlib qt
from matplotlib import pyplot as plt
plt.plot([1,2])
plt.pause(.1)
plt.pause(.1)
plt.pause(.1)

The problem seems to be that if socketpair generates a socket with the same fileno value as it generated in the previous pause() call, the QSocketNotifier unexpectedly fires and hangs here:

matplotlib/lib/matplotlib/backends/qt_compat.py

Line 232 in b600026

sn.activated.connect(lambda *args: rsock.recv(1))

because nothing was actually written to the wakeup_fd. I don't understand how that can be.

Workaround fix:

rsock.set_blocking(False)

@sn.activated.connect
def on_signal(*args):
    try:
        signal = rsock.recv(1)  # clear the socket to re-arm the notifier
    except BlockingIOError:
        pass # false-trigger, nothing on rsock

What should we do with this? Is there a better solution?

Originally posted by @vdrhtc in #13306 (comment)

[QuLogic]

Is it the current QSocketNotifier that fires, or the previous one? Are we just missing some cleanup?

[vdrhtc]

@QuLogic

Here how it looks like, I don't think it is the same QSocketNotifier, it is a different object every time. But false-trigger only happens if the fileno is re-used (1092 in this case, for 1028 nothing happens). So maybe it is necessary to clear the rsock somehow.

[QuLogic]

Does something like this work?

import select
import socket


for i in range(10):
    wsock, rsock = socket.socketpair()
    wsock.setblocking(False)
    rfd = rsock.fileno()
    wfd = wsock.fileno()
    print('rsock:', rfd, 'wsock:', wfd)
    ready = select.select([rfd], [], [], 2)
    print('ready:', ready)
    wsock.close()
    rsock.close()

It should print out that no sockets are ready. Windows support was only added in 3.5, but maybe there's a bug?

Maybe we should call socket.shutdown on the sockets to force them to be disabled, even though we do close them:

Shut down one or both halves of the connection. If how is SHUT_RD, further receives are disallowed. If how is SHUT_WR, further sends are disallowed. If how is SHUT_RDWR, further sends and receives are disallowed.

[QuLogic]

Also, does reusing the existing socket pair work?

wsock, rsock = socket.socketpair()
wsock.setblocking(False)
sn = QtCore.QSocketNotifier(
    rsock.fileno(), _enum('QtCore.QSocketNotifier.Type').Read
)
sn.activated.connect(lambda *args: print('activated 1'))
del sn
sn = QtCore.QSocketNotifier(
    rsock.fileno(), _enum('QtCore.QSocketNotifier.Type').Read
)
sn.activated.connect(lambda *args: print('activated 2'))

[vdrhtc]

Nothing unexpected in the first test despite occasional re-uses:

Second test also works:

If I remove the del sn line, then there is a warning about multiple notifiers for the same socket in the terminal, and no activation happens on both notifiers.

Calling

wsock.shutdown(socket.SHUT_RDWR)
rsock.shutdown(socket.SHUT_RDWR)
wsock.close()
rsock.close()

does not help.

[QuLogic]

Actually, for that second test, maybe we should run the event loop in between?

wsock, rsock = socket.socketpair()
wsock.setblocking(False)

sn = QtCore.QSocketNotifier(
    rsock.fileno(), _enum('QtCore.QSocketNotifier.Type').Read
)
sn.activated.connect(lambda *args: print('activated 1'))
# 5 seconds of event processing should be enough?
event_loop = QtCore.QEventLoop()
timer = QtCore.QTimer.singleShot(5000, event_loop.quit)
qt_compat._exec(event_loop)

del sn

sn = QtCore.QSocketNotifier(
    rsock.fileno(), _enum('QtCore.QSocketNotifier.Type').Read
)
sn.activated.connect(lambda *args: print('activated 2'))
# 5 seconds of event processing should be enough?
event_loop = QtCore.QEventLoop()
timer = QtCore.QTimer.singleShot(5000, event_loop.quit)
qt_compat._exec(event_loop)

[vdrhtc]

@QuLogic For some reason, the event loop does not exit after 5s and then the program hangs...

I've obtained the previous results by running %matplotlib qt first (it probably starts the event loop correctly by itself)

[timhoffm]

Is this fixed by #20907?

[QuLogic]

It should be.