https://bugs.kde.org/show_bug.cgi?id=508085

--- Comment #13 from Harald Sitter <[email protected]> ---
A somewhat crude way to trigger the crash is 

void KFileWidget::slotOk()
{
    QThread::sleep(2);

and then smashing the ok button. eventually a race condition should happen and
crash.

With this trick we've managed to confirm that this is indeed a problem with
nested eventloops and more specifically the fact that they process user input
events, even though kjob tells them not to
https://invent.kde.org/frameworks/kcoreaddons/-/blob/master/src/lib/jobs/kjob.cpp#L224
which turns out is probably a bug in Qt not implementing this correctly for
wayland.

fortunately this seems to be fixed by
https://invent.kde.org/qt/qt/qtbase/-/commit/6ff771b86f7b961a5ff63f148cf69ecb3415dd95
(needs someone to confirm it actually fixes the problem though). And maybe we
are lucky and can get that commit into Qt 6.10.

That leaves us with a temporary problem that we could possibly patch with
David's proposed MR. A more reasonable long term solution may be to not have
blocking stat calls. We do not know if the backing storage will respond in a
timely manner.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to