1853012 - On Firefox 119 Nightly and 118 beta 8 and 9 for Android, video playback occasionally stops while audio continues, then both resume from where video had stopped

Status: RESOLVED FIXED

(Core :: Audio/Video: cubeb, defect, P2)

Description

[Yannis Juglaret [:yannis]]

Device information: Fairphone 3, arm64, with Android 11, software version 8901.4.A.0025.0.

Steps to reproduce:

• Visit youtube.com on Firefox Nightly for Android.
• Watch any video of, say, at least 10 minutes.

Problem:

Every 60 to 120 seconds, the video playback stops and shows a spinning wheel. Meanwhile, audio continues. About 20 seconds later, video playback and audio both resume from the point where video playback had stopped.

Last known good build: Firefox Nightly 119.0.a1 2023-09-08-17-12-18 built from 3396ae36832c1e444d7373ab021be20072d57e89.
First known bad build: Firefox Nightly 119.0.a1 2023-09-09-04-03-11 built from eb062b89c03adc9b41355ee054d8582fda6b482d.
Regression range: here.

I'm tempted to say this could be regressed by bug 1848518 (which has just been uplifted to beta)?

I will attach extra information in the next comments.

Comment 1

[Alastor Wu [:alwu]]

Paul, kinetik,
This seems a cubeb regression, do y'all have any time to check this issue?
Thanks!

Comment 2

[Yannis Juglaret [:yannis]]

Attachment: logcat.txt

Here is a logcat of ~30 seconds, where the issue should be happening more or less during the last ~20 seconds.

Comment 3

[Yannis Juglaret [:yannis]]

Attachment: aboutsupport.json

Here is my about:support.

Comment 4

[Yannis Juglaret [:yannis]]

DailyMotion seems to behave a bit differently: video stops and shows continuously the last video frame without a spinning wheel, audio continues, video occasionally updates to a single more recent frame, then after ~1 minute audio and video resume without going back to the original point of desynchronisation.

Comment 5

[Donal Meehan [:dmeehan]]

Setting Bug 1848518 as the regressor, please correct if needed after the investigation in Comment 1

Comment 6

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1848518

Comment 7

[Yannis Juglaret [:yannis]]

I should probably mention why I'm pointing to bug 1848518 in particular. While I was using mozgression with GeckoViewExample, it turned out that "bad" builds of GVE had audio playback (together with the problem described here), whereas "good" builds of GVE had no audio playback (but didn't have the problem described here). That helped me reach the last known good build quicker ("do I have audio playback in GVE?"), then I confirmed in Firefox Nightly that this was indeed the last known good build for the problem described here.

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

The bug is marked as tracked for firefox118 (beta). We have limited time to fix this, the soft freeze is in 7 days. However, the bug still isn't assigned and has low severity.

:jimm, could you please find an assignee and increase the severity for this tracked bug? Given that it is a regression and we know the cause, we could also simply backout the regressor. If you disagree with the tracking decision, please talk with the release managers.

For more information, please visit BugBot documentation.

Comment 10

[jackyzy823]

FYI. Reverting this commit helps. https://github.com/mozilla/cubeb/commit/46f10c2eeaeef949a139f1e2b672989895cec98e

Sorry , wrongly tested.

The bug starts from this commit. https://github.com/mozilla/cubeb/commit/4dfcbd32fd854872040c7b9ff8a839b33f336f8f

Comment 11

[Yannis Juglaret [:yannis]]

(In reply to Yannis Juglaret [:yannis] from comment #0)

I'm tempted to say this could be regressed by bug 1848518 (which has just been uplifted to beta)?

I confirm that I can now reproduce the issue in Firefox 118 beta 8.

Comment 12

[jackyzy823]

After some testing:

I found that https://github.com/mozilla/cubeb/commit/4dfcbd32fd854872040c7b9ff8a839b33f336f8f?diff=unified#diff-481be8adcfca43a9a80fe319f0991d6deefd89501017a4e56c442b1a690affeaL1724-R1726 this line is the cause of bug.

change back from uin32_t to uint64_t fix the bug.

Some guess:
Uin32_MAX ( 4,294,967,295 ) / Sample_rate (48000) / 1000 -> 89.47s about 1min30s.

Comment 13

[Paul Adenot (:padenot)]

Yes, this is going through our CI currently, thanks for double-checking.

Comment 14

[Paul Adenot (:padenot)]

Attachment: Bug 1853012 - Update libcubeb to revision f9c118d. r?#cubeb-reviewers

Comment 15

[Paul Adenot (:padenot)]

Attachment: Bug 1853012 - Reapply patch on top of libcubeb. r?#cubeb-reviewers

Depends on D188348

Comment 16

[f38382014-buzilla]

Can uin64_t be used in armv7a 32bits

Comment 17

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/d5a0ac5140d8
Update libcubeb to revision f9c118d. r=cubeb-reviewers,chunmin
https://hg.mozilla.org/integration/autoland/rev/39b3e5649496
Reapply patch on top of libcubeb. r=cubeb-reviewers,chunmin

Comment 18

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/d5a0ac5140d8
https://hg.mozilla.org/mozilla-central/rev/39b3e5649496

Comment 19

[Ryan VanderMeulen [:RyanVM]]

Please nominate this for Beta approval.

Comment 20

[Yannis Juglaret [:yannis]]

The latest Nightly build works like a charm, thanks!

(In reply to f38382014-buzilla from comment #16)

Can uin64_t be used in armv7a 32bits

Yes. You can install Firefox Nightly from the Play Store right now to make sure that the issue is fixed for you too, or wait until the fix reaches Firefox Beta.

Comment 21

[Paul Adenot (:padenot)]

Comment on attachment 9353416 [details]
Bug 1853012 - Update libcubeb to revision f9c118d. r?#cubeb-reviewers

Beta/Release Uplift Approval Request

• User impact if declined: Video can can only play for 1:34 and then appears to be buffering (but audio continues), and then a/v sync is off when video starts again on Android version 10 and below.
• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: Yes
• Needs manual test from QE?: Yes
• If yes, steps to reproduce: Play a video that is longer than 1:34, it should work
• List of other uplifts needed: None
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): Patch found independently by two people, simple fix reverting to a previous version of a single line.
• String changes made/needed:
• Is Android affected?: Yes

Comment 22

[Paul Adenot (:padenot)]

1:24 actually but it doesn't matter.

Comment 23

[Pascal Chevrel:pascalc]

Comment on attachment 9353416 [details]
Bug 1853012 - Update libcubeb to revision f9c118d. r?#cubeb-reviewers

Approved for landing on mozilla-beta before the merge, will be in the 118.0 release candidate, thanks.

Comment 24

[Pulsebot]

https://hg.mozilla.org/releases/mozilla-beta/rev/55c9e3e2e368
https://hg.mozilla.org/releases/mozilla-beta/rev/899c187a45b8