962871 - Enable content/media mochitests on debug emulators

Status: RESOLVED FIXED

(Core :: Audio/Video, defect)

Description

[Jason Smith [:jsmith]]

bug 916135 successfully got the media mochitests running on non-debug emulators. We should work on getting those tests enabled on debug emulators.

Comment 1

[JW Wang [:jwwang] [:jw_wang]]

https://tbpl.mozilla.org/?tree=Try&rev=ae5539916705
Media mochitests are most green (withing 30 runs) on B2G debug with a few patches. It looks like the status is good enough to enable them on B2G debug again.

Comment 2

[JW Wang [:jwwang] [:jw_wang]]

Attachment: 962871_part1_remove_GC-v1.patch

Disable GC per token for it is sometimes very slow (see bug 1012516 comment 0) and causes exceeding 2hr limitation per chunk. It also helps reduce test duration.

Comment 3

[JW Wang [:jwwang] [:jw_wang]]

Attachment: 962871_part2_fix_testcases-v1.patch

Following patch part 1, we need to release decoders manually and aggressively to avoid timeout in waiting for decoding resources. It also set preload to "metadata" as appropriate for preload=auto will behave differently depending on pref settings.

Comment 4

[JW Wang [:jwwang] [:jw_wang]]

Attachment: 962871_part3_enable_tests-v1.patch

Enable mochitests on B2G debug. Also disable failed tests that will be fixed on other bugs.

Comment 5

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/8b181c601afb
https://hg.mozilla.org/integration/mozilla-inbound/rev/83d364b2f226
https://hg.mozilla.org/integration/mozilla-inbound/rev/5e4a2f99574a

Comment 6

[Andrew McCreight [:mccr8]]

Somewhat weirdly, this seems to have fixed a problem in B2G M9 that was stopping us from getting a leak log.  So thanks for fixing that, even if it was accidental.

Comment 7

[Andrew McCreight [:mccr8]]

Though unfortunately this exposed an existing leak, and khuey thinks it is better to not see M9 leaks than increase the leak threshold by 11kb for every test.

Comment 8

[Ryan VanderMeulen [:RyanVM]]

Backed out at mccr8's request. Sorry :(
https://hg.mozilla.org/integration/mozilla-inbound/rev/7704043c4fe5

Comment 9

[JW Wang [:jwwang] [:jw_wang]]

(In reply to Andrew McCreight [:mccr8] from comment #7)
> Though unfortunately this exposed an existing leak, and khuey thinks it is
> better to not see M9 leaks than increase the leak threshold by 11kb for
> every test.

I guess some leaked tests are squeezed from M8 to M9 and cause M9 to exceed the threshold.

Comment 10

[JW Wang [:jwwang] [:jw_wang]]

(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #8)
> Backed out at mccr8's request. Sorry :(
> https://hg.mozilla.org/integration/mozilla-inbound/rev/7704043c4fe5

Can we back out part 3 only (which enables the tests)? I think part 1 and 2 are good to go and we will need to check in part 3 only when the leaks are fixed. Having less code to check in also reduce the effort of rebase.

Comment 11

[Ryan VanderMeulen [:RyanVM]]

Yeah, that's a fair request. Sorry for not doing that in the first place.

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9ae65ea08644
https://hg.mozilla.org/integration/mozilla-inbound/rev/c585db172b7b

Comment 13

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/9ae65ea08644
https://hg.mozilla.org/mozilla-central/rev/c585db172b7b

Comment 14

[Ryan VanderMeulen [:RyanVM]]

Bug 1068268 is fixed, but it seems to me that we should probably wait for bug 1068276 to be fixed as well before attempting to re-land this. Do you agree, mccr8?

Comment 15

[Andrew McCreight [:mccr8]]

(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #14)
> Bug 1068268 is fixed, but it seems to me that we should probably wait for
> bug 1068276 to be fixed as well before attempting to re-land this. Do you
> agree, mccr8?

No, bug 962871 has to land before bug 1068276, or the tree will go orange.  Right now, B2G M9 is timing out and not producing a leak log.

Comment 16

[Ryan VanderMeulen [:RyanVM]]

Try says this is green.
https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=e9178d8a53ae

Comment 17

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/b2g-inbound/rev/da4de6b9cdab

Comment 18

[Andrew McCreight [:mccr8]]

Oh well, judging by that try push this just pushes the timeout from Mochitest-9 to Mochitest-10, which I guess makes sense.  I'll file a separate bug for that...

Comment 19

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/da4de6b9cdab