1178063 - twit.tv mp4 audio/video stutter

Status: RESOLVED WORKSFORME

(Core :: Audio/Video: Playback, defect)

Description

[Alice0775 White]

[Tracking Requested - why for this release]: This is regression since Firefox36.

This was originally reported by http://forums.mozillazine.org/viewtopic.php?p=14213915#p14213915 .

Reproducible: always

Steps To Reproduce:
1. Open https://twit.tv/shows/security-now/episodes/513?autostart=true
   (or direct link of mp4, http://www.podtrac.com/pts/redirect.mp4/twit.cachefly.net/video/sn/sn0512/sn0512_h264m_864x480_500.mp4 )
2. Wait for 10-20 sec (depended on network speed), so that the video will auto start 
3. Click Pause and Play button if necessary


Actual Results:
audio/video stutter

Expected Results:
No stutter

Regression window:

Beta pushlog:
https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=5d185a7d03b5&tochange=96d0d77a3462

Aurora pushlog:
https://hg.mozilla.org/releases/mozilla-aurora/pushloghtml?fromchange=ee5c1552ca19&tochange=ba32d4735526

m-i pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=a638073e104d&tochange=6ceb11cd7fc1

REGRESSED by:
6ceb11cd7fc1	Chris Pearce — Bug 1112445 - Ignore the audio stream when determining whether we should skip-t-o-next-keyframe for async readers. r=mattwoodrow

Comment 1

[Peja Stija]

All video's in that site have problems in Firefox 38 stable also, although the get out of the "buffer" phase, video gets out of sync pretty fast.

Comment 2

[Chris Pearce [:cpearce (Not reading bugmail)]]

Can't repro on my Win8.1 PC. Alice, are you testing in a VM or on a native Win7 machine?

Comment 3

[Alice0775 White]

native Win7

Comment 4

[Alice0775 White]

Attachment: media logging

set NSPR_LOG_MODULES=timestamp,MediaDecoder:5,MediaSource:5,MediaPromise:5,MP4Demuxer:5,nsMediaElement:5,nsMediaElementEvents:5

Comment 5

[Alice0775 White]

(In reply to Chris Pearce (:cpearce) from comment #2)
> Can't repro on my Win8.1 PC.

On Windows8 on VMWare.
It is easy to reproduce the stutter if set custom network speed to 8-10Mbps.
And I can also confirm the same regression range on the VM.

Comment 6

[Alice0775 White]

In aditthin to the comment 5.

I setup local http server. 

I cannot reproduce the problem on 100Mbps network speed between local http server and the Win8.1VM.
However, I can reproduce the problem on 10Mbps network speed between them.
And I can also confirm the same regression range.


So, this mean, Bug 1112445 seems to interfere media buffering and makes low performance.

Comment 7

[Peja Stija]

Momentarily, the video at https://twit.tv/shows/security-now/episodes/513?autostart=true starts & plays fine, no buffering issues, no desyncs.

Comment 8

[Alice0775 White]

Fixed range:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=d7e196b07cee&tochange=54bf2c8de576

So, Bug 1179110 eventually fixes this problem.

:bholley and :cpearce,

Could you up uplift Bug 1179110 to Aurora41.0a1, 40beta and 38ESR?

Comment 9

[Bobby Holley (:bholley)]

(In reply to Alice0775 White from comment #8)
> Fixed range:
> https://hg.mozilla.org/integration/mozilla-inbound/
> pushloghtml?fromchange=d7e196b07cee&tochange=54bf2c8de576
> 
> So, Bug 1179110 eventually fixes this problem.
> 
> :bholley and :cpearce,
> 
> Could you up uplift Bug 1179110 to Aurora41.0a1, 40beta and 38ESR?

This is being discussed in bug 1179110 comment 14 and onward.

Comment 10

[Kate Glazko]

In bug 1179110, it says that Firefox 40 is unaffected. Do we need tracking/uplift for 40, as it appears that a solution from bug 1179110 doesn't affect 40?

Comment 11

[Bobby Holley (:bholley)]

(In reply to Kate Glazko from comment #10)
> In bug 1179110, it says that Firefox 40 is unaffected. Do we need
> tracking/uplift for 40, as it appears that a solution from bug 1179110
> doesn't affect 40?

Part 2 of bug 1179110 could probably be uplifted in a way that it would fix this bug. The problem described in bug 1179110 as filed is 41+.

Comment 12

[Kate Glazko]

Thank you for clarifying! In that case, will track this bug for 40 as it is a regression. Adding bug 1179110 as a dependency as that bug could fix this issue. Please keep us posted on when/if an uplift to 40 would be happening.

Comment 13

[Kate Glazko]

Untracking 41, 42 because they are tracked in 1179110. Leaving 40 tracked here as 40 is not addressed in bug 1179110.

Comment 14

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Adding a tracking flag for ESR38 release as this seems to have a significant end-user impact.

Chris, can you comment on whether this ought to be fixed in the upcoming ESR38 release? Thanks!

Comment 15

[Chris Pearce [:cpearce (Not reading bugmail)]]

Was this fixed in Firefox 41 by the partial uplift of bug 1179110? If so you should follow up with the dev jya in that bug to see about getting it uplifted to 38ESR.

Comment 16

[Lawrence Mandel [:lmandel] (use needinfo)]

This is now wontfix for 40. Is 41 fixed?