Andrew McCreight [:mccr8] Reporter
	Description • 9 years ago

Various tests leak 1 AudioChannelService object. This shows up with XPCOM leak checking, but does not trigger it because we are currently using a threshold-based system with some slack. In bug 1215148, I am adding a new leak checking method that has specific counts of leaked objects.

I suspect this does not show up in LSan because the service object is held alive in a global variable. I think it is a singleton leak so it is mostly a matter of maintaining leak checking hygiene.

This leaks in a number of tests:
  https://treeherder.mozilla.org/#/jobs?repo=try&revision=1c04c85d7acd

One easy way I found to reproduce it is this:
  ./mach mochitest -f plain --e10s toolkit/content/tests/widgets/

(I thought there was a bug on file for this already, but I can't find one.)

	Andrew McCreight [:mccr8] Reporter
	Comment 1 • 9 years ago

How hard would this be to fix, Andrea? Is there some reason that the observer shutdown stuff is only in the parent process?

	Andrew McCreight [:mccr8] Reporter
	Comment 3 • 9 years ago

Comment on attachment 8676071 [details] [diff] [review]
acs.patch

Review of attachment 8676071 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks!

	Andrew McCreight [:mccr8] Reporter
	Comment 4 • 9 years ago

Comment on attachment 8676071 [details] [diff] [review]
acs.patch

Review of attachment 8676071 [details] [diff] [review]:
-----------------------------------------------------------------

I just noticed that the observers that are added and removed in a non-parent process don't match up.

Also, I'm having trouble applying this patch to trunk, though I'm not sure why.

	Andrew McCreight [:mccr8] Reporter
	Comment 5 • 9 years ago

This makes it so that ipc:content-shutdown observation stuff only happens in the parent.

	Andrew McCreight [:mccr8] Reporter
	Comment 6 • 9 years ago

Comment on attachment 8676391 [details] [diff] [review]
Shut down AudioChannelService in the child process.

Review of attachment 8676391 [details] [diff] [review]:
-----------------------------------------------------------------

I guess you should technically review this. :) I'll do a try push, too.

	Andrew McCreight [:mccr8] Reporter
	Comment 7 • 9 years ago

try run: https://treeherder.mozilla.org/#/jobs?repo=try&revision=32c8f7f86d13

	Pulsebot
	Comment 8 • 9 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/afd0786c65f5

	Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)
	Comment 9 • 9 years ago

https://hg.mozilla.org/mozilla-central/rev/afd0786c65f5