1326026 - When output video frame to surface, send dummy samples.

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: Audio/Video, defect, P1)

Description

[John Lin [:jhlin][:jolin]]

Since RemoteVideoDecoder [1] does not use the data in output buffers anyway, use dummy samples as outputs saves some shared memory allocation and IPC percel serialization efforts.

Also, logcat dumps in bug 1298300 attachment 8822140 [details] shows some devices (MTK chipset) returns null output buffers after format changes sometimes. Use dummy samples instead of shared memory also works around that.

Comment 1

[John Lin [:jhlin][:jolin]]

Attachment: Bug 1326026 - use bufferless samples for outputs when rendering to surface.

Review commit: https://reviewboard.mozilla.org/r/101290/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/101290/

Comment 2

[John Lin [:jhlin][:jolin]]

Comment on attachment 8822345 [details]
Bug 1326026 - use bufferless samples for outputs when rendering to surface.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/101290/diff/1-2/

Comment 3

[John Lin [:jhlin][:jolin]]

Attachment: Bug 1326026 - use bufferless samples for outputs when rendering to surface.

Review commit: https://reviewboard.mozilla.org/r/101308/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/101308/

Comment 4

[James Willcox (:snorp) ([e-mail]) (he/him)]

Comment on attachment 8822359 [details]
Bug 1326026 - use bufferless samples for outputs when rendering to surface.

https://reviewboard.mozilla.org/r/101308/#review103076

Comment 5

[Rachelle Yang [:ryang][[e-mail]]]

Hi John,

May I know when this patch will be landed ? Will it also be merged in to Firefox53?
Since I suppose the feature (pref-on Bug 1311960) of moving Android audio/video decoder out of application process will be enabled and landed in Firefox 53?

Can you please help with this ?? Thank you very much !

Comment 6

[John Lin [:jhlin][:jolin]]

Comment on attachment 8822359 [details]
Bug 1326026 - use bufferless samples for outputs when rendering to surface.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/101308/diff/1-2/

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/4b3a24d9a501
use bufferless samples for outputs when rendering to surface. r=snorp

Comment 8

[John Lin [:jhlin][:jolin]]

(In reply to Rachelle Yang [:ryang][[e-mail]] from comment #5)
> Hi John,
> 
> May I know when this patch will be landed ? Will it also be merged in to
> Firefox53?
> Since I suppose the feature (pref-on Bug 1311960) of moving Android
> audio/video decoder out of application process will be enabled and landed in
> Firefox 53?
> 
> Can you please help with this ?? Thank you very much !

 Thanks for the reminder and sorry for not noticing the autoland failure. Just landed it in inbound manually.

Comment 9

[Rachelle Yang [:ryang][[e-mail]]]

Thank you, John!

Comment 10

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/4b3a24d9a501

Comment 11

[Oana Horvath [:ohorvath]]

I see this bug should fix the problem reported on Bug 1298300 & Bug 1331391, but I've ran again across the issue of videos not working on youtube or other video pages (mp4, webm or html5). 
Out of the devices that I've checked on (including other devices running on Android 4), the issue was isolated to the Asus Transformers Pad (Android 4.2.1): the browser doesn't crash, but the video keeps buffering without any video or audio output.
It is reproducible also on Nightly 54. Please let me know if you think I should file a new bug.

Comment 12

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

Per comment 11, ni john.

Comment 13

[John Lin [:jhlin][:jolin]]

(In reply to Oana Horvath from comment #11)
> I see this bug should fix the problem reported on Bug 1298300 & Bug 1331391,
> but I've ran again across the issue of videos not working on youtube or
> other video pages (mp4, webm or html5). 
> Out of the devices that I've checked on (including other devices running on
> Android 4), the issue was isolated to the Asus Transformers Pad (Android
> 4.2.1): the browser doesn't crash, but the video keeps buffering without any
> video or audio output.
> It is reproducible also on Nightly 54. Please let me know if you think I
> should file a new bug.

 Oana, could you please help capture logcat output? A new bug should be filed if the remote process crash stack is different.

Comment 14

[Oana Horvath [:ohorvath]]

Attachment: logcat output - bug 1326026.txt

I attached the logcat file captured today (2017-2-2) on the Asus Transformers Pad (Android 4) device.

Comment 15

[John Lin [:jhlin][:jolin]]

(In reply to Oana Horvath from comment #14)
> Created attachment 8832849 [details]
> logcat output - bug 1326026.txt
> 
> I attached the logcat file captured today (2017-2-2) on the Asus
> Transformers Pad (Android 4) device.

 Oana, thanks a lot for the log. It doesn't show similar error as this bug, so I guess a new bug is needed after all. Also, do you mind if I share a private build with you to collect more debug info?

Comment 16

[Oana Horvath [:ohorvath]]

(In reply to John Lin [:jolin][:jhlin] from comment #15)
> (In reply to Oana Horvath from comment #14)
> > Created attachment 8832849 [details]
> > logcat output - bug 1326026.txt
> > 
> > I attached the logcat file captured today (2017-2-2) on the Asus
> > Transformers Pad (Android 4) device.
> 
>  Oana, thanks a lot for the log. It doesn't show similar error as this bug,
> so I guess a new bug is needed after all. Also, do you mind if I share a
> private build with you to collect more debug info?

Sure, happy to help.