Fix possible crash at exit on iOS #11781
Conversation
marcelofg55
force-pushed
the
iphone_crashfix
branch
2 times, most recently
from
61c2194 to
7201a14
Compare
marcelofg55
changed the title
marcelofg55
force-pushed
the
iphone_crashfix
branch
from
7201a14 to
116e735
Compare
marcelofg55
changed the title
| @@ -726,7 +726,8 @@ void AudioServerSW::_thread_func(void *self) { | |||
|
|
|||
| while (!as->exit_update_thread) { | |||
| as->_update_streams(true); | |||
| OS::get_singleton()->delay_usec(5000); | |||
| if (OS::get_singleton()) | |||
| OS::get_singleton()->delay_usec(5000); | |||
There was a problem hiding this comment.
This looks a little suspect. We just sleep in hopes of avoiding a race?
There was a problem hiding this comment.
It sleeps a little to avoid calling _update_streams all the time and eat a lot of cpu.
There was a problem hiding this comment.
I could change this if (OS::get_singleton()) for an ERR_BREAK.
There was a problem hiding this comment.
If it isn't to avoid a race I guess this is OK, but how do we know this is the right amount of time to wait? Can't we have some kind of callback?
There was a problem hiding this comment.
We process the audio in its own thread, which I think its great we take advantage of multiple cores and also avoid taking process time from the main thread, so a callback would be a bad idea in my opinion. And the timer has been like this for a long time I think so I don't want to mess with it, and 5ms if fine in my opinion.
Fixes the crash described at #11623 which happens when the iOS app tries to quit and the render callback keeps getting called (because AudioDriverIphone::finish wasn't clearing the property).