On Mon, 14 Oct 2019 at 13:53, <hikaru...@gmail.com> wrote: > > From: Hikaru Nishida <hikaru...@gmail.com> > > An NSEvent is fired before applicationDidFinishLaunching on macOS > Catalina. This causes deadlock of iothread_lock by calling > bool_with_iothread_lock in handleEvent while its already locked. > This patch prevents to call bool_with_iothread_lock until the > app_started_sem is released to prevent this deadlock. > > Buglink: https://bugs.launchpad.net/qemu/+bug/1847906 > Signed-off-by: Hikaru Nishida <hikaru...@gmail.com> > --- > ui/cocoa.m | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/ui/cocoa.m b/ui/cocoa.m > index f12e21df6e..f16d341a0a 100644 > --- a/ui/cocoa.m > +++ b/ui/cocoa.m > @@ -134,6 +134,7 @@ > > static QemuSemaphore display_init_sem; > static QemuSemaphore app_started_sem; > +volatile sig_atomic_t allow_events; > > // Utility functions to run specified code block with iothread lock held > typedef void (^CodeBlock)(void); > @@ -729,6 +730,9 @@ - (void) handleMonitorInput:(NSEvent *)event > > - (bool) handleEvent:(NSEvent *)event > { > + if(!allow_events) { > + return false; > + } > return bool_with_iothread_lock(^{ > return [self handleEventLocked:event]; > }); > @@ -1897,6 +1901,7 @@ static void cocoa_display_init(DisplayState *ds, > DisplayOptions *opts) > /* Tell main thread to go ahead and create the app and enter the run > loop */ > qemu_sem_post(&display_init_sem); > qemu_sem_wait(&app_started_sem); > + allow_events = true; > COCOA_DEBUG("cocoa_display_init: app start completed\n");
If we do use a flag to fix this race, I think it would be better to set allow_events = true in the applicationDidFinishLaunching() function before posting the app_started_sem. Otherwise there's a window after applicationDidFinishLaunching() returns but before cocoa_display_init() resumes and sets the flag where we will unnecessarily drop events. Could you try that and check that it still fixes the hang? thanks -- PMM
