1114928 - [FFOS][MSE] GonkMediaDataDecoder Throws a false alarm

Status: RESOLVED FIXED

(Core :: Audio/Video, defect)

Description

[Blake Wu [:bwu][:blakewu]]

Play the link[1] on Youtube via dash and it just lasts 4 second and stops. 
Refreshing browser doesn't re-trigger anything. It looks like it hang or gets stuck.  

If I play [1] with MAC's nightly version, 37.0a1 (2014-12-22), it can play well.

[1]https://www.youtube.com/watch?v=wmgzfn1VtQ4

Comment 1

[Blake Wu [:bwu][:blakewu]]

It looks there is a false alarm when getting EOS frame [1][2] and it makes playback stop. 

[1]http://dxr.mozilla.org/mozilla-central/source/dom/media/fmp4/gonk/GonkVideoDecoderManager.cpp?from=GonkVideoDecoderManager.cpp&case=true#404
[2]http://dxr.mozilla.org/mozilla-central/source/dom/media/fmp4/gonk/GonkAudioDecoderManager.cpp?from=GonkAudioDecoderManager.cpp#180

Comment 2

[Blake Wu [:bwu][:blakewu]]

Attachment: Bug-1114928-Have-an-initial-value-for-Ouput-result.patch

Have an correct initial value for "rv" to avoid checking a garbage value when mDrainComplete is false.

Comment 3

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

Comment on attachment 8545038 [details] [diff] [review]
Bug-1114928-Have-an-initial-value-for-Ouput-result.patch

Review of attachment 8545038 [details] [diff] [review]:
-----------------------------------------------------------------

You should configure hg (or git, or whatever) to give you 8 lines of context instead of the default 4.

::: dom/media/fmp4/gonk/GonkMediaDataDecoder.cpp
@@ +89,4 @@
>  GonkMediaDataDecoder::ProcessOutput()
>  {
>    nsRefPtr<MediaData> output;
> +  nsresult rv = NS_ERROR_ABORT;

I don't object to it, but I don't see how it helps.

Comment 4

[Blake Wu [:bwu][:blakewu]]

(In reply to Anthony Jones (:kentuckyfriedtakahe, :k17e) from comment #3)
> Comment on attachment 8545038 [details] [diff] [review]
> Bug-1114928-Have-an-initial-value-for-Ouput-result.patch
> 
> Review of attachment 8545038 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> You should configure hg (or git, or whatever) to give you 8 lines of context
> instead of the default 4.
> 
> ::: dom/media/fmp4/gonk/GonkMediaDataDecoder.cpp
> @@ +89,4 @@
> >  GonkMediaDataDecoder::ProcessOutput()
> >  {
> >    nsRefPtr<MediaData> output;
> > +  nsresult rv = NS_ERROR_ABORT;
> 
> I don't object to it, but I don't see how it helps.
"rv" will be only assigned value @[1]. If the while loop is not entered, rv will carry a garbage value which will be checked incorrectly in the codes behind the while loop and throws a false alarm.

[1]http://dxr.mozilla.org/mozilla-central/source/dom/media/fmp4/gonk/GonkMediaDataDecoder.cpp?from=GonkMediaDataDecoder.cpp&case=true#94

Comment 5

[Blake Wu [:bwu][:blakewu]]

Attachment: Final-Bug-1114928-Have-an-initial-value-for-Ouput-result.patch

Carry r+ from ajones.

Comment 6

[Blake Wu [:bwu][:blakewu]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=fbebc7fe3e93
Building results on try server are good enough. No need to run test cases due to pref off.

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/adbabd8de78c

Comment 8

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/adbabd8de78c