1202012 - Video decode error where Firefox thinks valid MP4 is corrupt

Status: RESOLVED FIXED

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

Description

[Wade]

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:40.0) Gecko/20100101 Firefox/40.0
Build ID: 20150826023504

Steps to reproduce:

Go here:
http://www.concertvault.com/the-who/video/young-man-blues_1293014538.html

Video tries to auto-start with the 480P bitrate, which is this file:
http://pmd.wolfgangsvault.com/20120509/269/1_sr9eafcy_1_dtme0cbf_1.mp4


Actual results:

In the Concert Vault URL sometimes I see an error about a problem with playback, and others I just see the loading spinner.

The pmd.wolfgangsvault.com URL explicitly says that the file is corrupt, despite it working in other browsers and players on the same computer as well as other browsers and players on non-windows computers.

From a technician at Kaltura:
"The exact same file plays properly on Chrome (Windows and OSX), IE11, Firefox (OSX), Safari (OSX and iOS).
It also plays correctly on standalone players such as VLC, Windows Media Player, and MPC-HC."


Expected results:

The video should play on the Concert Vault URL just like the other bitrates of the same file.

Comment 1

[Alice0775 White]

[Tracking Requested - why for this release]: regression since Firefox35

Comment 2

[Alice0775 White]

[Tracking Requested - why for this release]: regression since Firefox40

Comment 3

[Alice0775 White]

Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=e17f5910da75&tochange=65492fe83362

Regressed by: Bug 1153049

Comment 4

[Liz Henry (:lizzard) (relman/hg->git project)]

Tracking for 41+

Comment 5

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

Matthew, can you please help investigate or find an owner? It is unclear to me the extent of this issue and whether it should be blocking 41. 

It is getting too late to fix this in FF41 given that we don't have a patch ready. Let's try to get this fixed in 42.

Comment 6

[Matthew Gregan [:kinetik]]

Redirecting based on the regression range in comment 3.

Comment 7

[Jean-Yves Avenard [:jya]]

File http://pmd.wolfgangsvault.com/20120509/269/1_sr9eafcy_1_dtme0cbf_1.mp4 plays here in 43 and 42 (OS X) just fine.

If bug 1153049 were to have caused that regression, it would have been on all platforms.

Comment 8

[Jean-Yves Avenard [:jya]]

Attachment: Use QTv2 header when available.

We really only played this file prior bug 1153049 by chance. The default sampling rate in the old constructor was set at 44100Hz and if the sampling rate provided was nonsensical is was ignored. In this MP4 metadata the sampling rate has a value of 1 and the channel count is 3. The WMF audio decoder errors on these values.
This is a file in "QTv2" format ; an Apple extension : we are supposed to ignore the normal values..

Comment 9

[Jean-Yves Avenard [:jya]]

Attachment: P2. Make stagefright works on big-endian systems.

The comment about hton64 not working on big endian caught my eyes. That renders the entire stagefright unusable on big-endian system. So quick fix. This with ffmpeg will allow media playback on big-endian machine.

Comment 10

[Ralph Giles (:rillian)]

Comment on attachment 8659980 [details] [diff] [review]
P2. Make stagefright works on big-endian systems.

Review of attachment 8659980 [details] [diff] [review]:
-----------------------------------------------------------------

What about tests for the new header parsing?

::: media/libstagefright/frameworks/av/media/libstagefright/Utils.cpp
@@ +72,3 @@
>  }
>  
> +// XXX warning: this won't work on big-endian host.

Why not fix both of them while you're here?

Comment 11

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/cbea1d09d413
https://hg.mozilla.org/integration/mozilla-inbound/rev/d087ef333c9a

Comment 12

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/cbea1d09d413
https://hg.mozilla.org/mozilla-central/rev/d087ef333c9a

Comment 13

[Jean-Yves Avenard [:jya]]

Comment on attachment 8659978 [details] [diff] [review]
Use QTv2 header when available.

Approval Request Comment
[Feature/regressing bug #]: 1153049. While a regression, it used to play thanks to sheer luck
[User impact if declined]: Some content will not play on windows.
[Describe test coverage new/current, TreeHerder]: Local test, in central
[Risks and why]: Low ; this is reading a custom property that we just used to skip over.
[String/UUID change made/needed]: None

could uplift this to beta, but I'm guessing it's too late

Comment 14

[Sylvestre Ledru [:Sylvestre]]

Comment on attachment 8659978 [details] [diff] [review]
Use QTv2 header when available.

Indeed too late for beta but still plenty of time for aurora.

Comment 15

[Carsten Book [:Tomcat]]

jya: is this approval-aurora request also for p2 ? (like should part 1 and 2 checked in into aurora ?

Comment 16

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/8c8a7797bf0e
https://hg.mozilla.org/releases/mozilla-aurora/rev/8fefcb8ca48c