2034722 - Regression between 149.0.2-1 and 150.0-1 cannot use v4l2loopback based webcam: USER_FORBIDDED_CAPTURE_VIDEO Error: Video loadedmetadata timeout 3 with canon EOS as webcam

Status: RESOLVED FIXED

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

Description

[Samuel Wein]

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:149.0) Gecko/20100101 Firefox/149.0

Steps to reproduce:

Start canon EOS webcam with: gphoto2 --stdout --set-config liveviewsize=0 --capture-movie | ffmpeg -i - -vcodec rawvideo -pix_fmt yuv420p -threads 0 -f v4l2 -s:v 1366x768 -r 25 /dev/video0
Loaded firefox snap on ubuntu 26.04 after update to 150.0-1, joined a zoom meeting in the web app. Tried to turn on my webcam.

Actual results:

I was unable to connect to camera. The console reported: USER_FORBIDDED_CAPTURE_VIDEO Error: Video loadedmetadata timeout 3 with canon EOS as webcam

Expected results:

Webcam should have started up, as it did in 149.0.2-1

Comment 1

[Samuel Wein]

I've confirmed that a) the camera device still works (I get a stream with ffplay), and b) the issue is not on the zoom webapp side. I can't get a preview in the firefox permission popup window, and can't get the camera to work on other webcam testing sites.

Comment 2

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

The Bugbug bot thinks this bug should belong to the 'Core::WebRTC: Audio/Video' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 3

[Andreas Pehrson [:pehrsons]]

Hi Samuel, thanks for the report. Does it reproduce (i.e. also not starting and/or showing frames) with the camera option on https://mozilla.github.io/webrtc-landing/gum_test.html? Would you be able to reproduce this with debug logging enabled?
If so, the instructions are:

• Startup the browser in a fresh profile, i.e. no extensions and no existing tabs.
• Go to about:logging, select the WebRTC preset and click Start Logging.
• Open a new tab where you reproduce the issue, ideally on gum_test.html if it reproduces there, otherwise Zoom web works.
• Once reproduced, go back to about:logging, click Stop Logging.
• A new tab will appear with the Firefox Profiler. In the top right corner click to upload the profile. Make sure to include hidden threads! Do as you wish with the other options, as long as the timeline contains the getUserMedia request and response.
• When uploaded, share the link here.

Thank you!

Comment 4

[Andreas Pehrson [:pehrsons]]

FWIW USER_FORBIDDED_CAPTURE_VIDEO seems to be a Zoom error. Unclear if it's specific to gUM rejecting with NotAllowedError or all gUM rejection. The profile will tell.

Comment 5

[Samuel Wein]

Yup, it is replicated on the gum_test. I've taken logs which are available here: https://share.firefox.dev/4e7ef2W
I also ran Mozregression:

 7:16.57 INFO: Narrowed integration regression window from [6de5bbf9, 81adc19b] (3 builds) to [7a02c406, 81adc19b] (2 builds) (~1 steps left)
 7:16.57 INFO: No more integration revisions, bisection finished.
 7:16.57 INFO: Last good revision: 7a02c40691b27da758d8070e3ae925e65da0a555
 7:16.57 INFO: First bad revision: 81adc19b420f055dd9e5b1ef30962829b403437d
 7:16.57 INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=7a02c40691b27da758d8070e3ae925e65da0a555&tochange=81adc19b420f055dd9e5b1ef30962829b403437d

Comment 6

[Samuel Wein]

Hi all,
any more information that I can provide on this?

Comment 7

[Andreas Pehrson [:pehrsons]]

Brilliant, thanks. I see this in the log, but not either of these, meaning we returned early here.

Likely regressor is https://hg-edge.mozilla.org/mozilla-central/rev/d35564be296062a87d5fbd94d9ba7df60fb81e5e.

I also see from the logs (in the parent process getUserMedia request for the prompt preview) that we are using hardcoded capabilities. That only happens when the device reports none:

Start	Duration	Name	Details
16.876s		LogMessages	(D/MediaManager) Hardcoded capability: 854 x 640 x 30 maxFps, unknown. Distance = 0
16.876s		LogMessages	(D/MediaManager) Hardcoded capability: 854 x 480 x 30 maxFps, unknown. Distance = 0
16.876s		LogMessages	(D/MediaManager) Capability: 854 x 640 x 30 maxFps, unknown. Distance = 250
16.876s		LogMessages	(D/MediaManager) Capability: 854 x 480 x 30 maxFps, unknown. Distance = 0
16.876s		LogMessages	(D/MediaManager) Chosen capability: 854 x 480 x 30 maxFps, unknown. Distance = 0
16.876s		LogMessages	(D/MediaManager) ChooseCapability(kFeasibility) for mCapability (Allocate) --
16.876s		LogMessages	(D/MediaManager) Video device 0 allocated
16.876s		LogMessages	(D/MediaManager) virtual void mozilla::MediaEngineRemoteVideoSource::SetTrack(const RefPtr<MediaTrack> &, const PrincipalHandle &)
16.876s		LogMessages	(D/MediaManager) virtual nsresult mozilla::MediaEngineRemoteVideoSource::Start()
16.876s		LogMessages	(D/MediaManager) nsresult mozilla::MediaEngineRemoteVideoSource::StartCapture(const NormalizedConstraints &, const dom::VideoResizeModeEnum &)
16.877s		LogMessages	(D/MediaManager) started videoinput device 7ef768d83100

S3 because capability-less devices are not particularly common.

Comment 8

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

Set release status flags based on info from the regressing bug 2012483

Comment 9

[Andreas Pehrson [:pehrsons]]

Attachment: Bug 2034722 - Add DeviceInfo backend without capabilities, for testing. r?#webrtc-reviewers

Comment 10

[Andreas Pehrson [:pehrsons]]

Attachment: Bug 2034722 - Add gtest for StartStream with capability-less DeviceInfo. r?#webrtc-reviewers

Comment 11

[Andreas Pehrson [:pehrsons]]

Attachment: Bug 2034722 - Return a combined capability even when device reports no capabilities. r?#webrtc-reviewers

Some legit devices report no capabilities. In such cases we want to
start them with the requested capability and wait to see what we get.

Comment 12

[Samuel Wein]

Perfect. I can confirm that applying D296943 to the FIREFOX_150_0_1_BUILD1 tag fixes the problem for me.

Comment 13

[Andreas Pehrson [:pehrsons]]

Great! Thanks for checking. Let's uplift it as far as we can, and with the gtest we should avoid similar regressions in the future.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/a92521f7d65c
https://hg.mozilla.org/integration/autoland/rev/52139df4e29c
Add DeviceInfo backend without capabilities, for testing. r=webrtc-reviewers,mjf
https://github.com/mozilla-firefox/firefox/commit/3e1f32dc4097
https://hg.mozilla.org/integration/autoland/rev/29ad72dea30d
Add gtest for StartStream with capability-less DeviceInfo. r=webrtc-reviewers,mjf
https://github.com/mozilla-firefox/firefox/commit/6868ade819ba
https://hg.mozilla.org/integration/autoland/rev/b506d048711b
Return a combined capability even when device reports no capabilities. r=webrtc-reviewers,mjf

Comment 15

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/52139df4e29c
https://hg.mozilla.org/mozilla-central/rev/29ad72dea30d
https://hg.mozilla.org/mozilla-central/rev/b506d048711b

Comment 16

[Phabricator Automation]

firefox-beta Uplift Approval Request

• User impact if declined/Reason for urgency: Camera capture broken since 150 for a tiny subset of devices
• Code covered by automated testing?: yes
• Fix verified in Nightly?: yes
• Needs manual QE testing?: no
• Steps to reproduce for manual QE testing:
• Risk associated with taking this patch: low
• Explanation of risk level: Trivial change
• String changes made/needed?: None
• Is Android affected?: unknown

Comment 17

[Andreas Pehrson [:pehrsons]]

Attachment: Bug 2034722 - Return a combined capability even when device reports no capabilities.

Some legit devices report no capabilities. In such cases we want to
start them with the requested capability and wait to see what we get.

Original Revision: https://phabricator.services.mozilla.com/D296943

Comment 18

[Phabricator Automation]

firefox-release Uplift Approval Request

• User impact if declined/Reason for urgency: Camera capture broken since 150 for a tiny subset of devices
• Code covered by automated testing?: yes
• Fix verified in Nightly?: yes
• Needs manual QE testing?: no
• Steps to reproduce for manual QE testing:
• Risk associated with taking this patch: low
• Explanation of risk level: Trivial change
• String changes made/needed?: None
• Is Android affected?: unknown

Comment 19

[Andreas Pehrson [:pehrsons]]

Attachment: Bug 2034722 - Return a combined capability even when device reports no capabilities.

Some legit devices report no capabilities. In such cases we want to
start them with the requested capability and wait to see what we get.

Original Revision: https://phabricator.services.mozilla.com/D296943

Comment 20

[Pulsebot]

https://github.com/mozilla-firefox/firefox/commit/e24810dc315b
https://hg.mozilla.org/releases/mozilla-beta/rev/e4e071abee0f

Comment 21

[Pulsebot]

https://github.com/mozilla-firefox/firefox/commit/eac4d7aa240b
https://hg.mozilla.org/releases/mozilla-release/rev/9831e6bb60ec