Video issue when accessing https://mozilla.github.io/webrtc-landing/pc_test.html
Categories
(Core :: WebRTC: Audio/Video, defect, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox72 | --- | affected |
People
(Reporter: sflorean, Unassigned)
References
Details
Attachments
(1 file)
25.84 KB,
text/plain
|
Details |
Environment:
Build: Firefox Nightly Preview 72.0a1 - #13030620
Devices:
- One Plus 5T (Android 9.0)
- Motorola Moto G6 (Android 8.0)
Steps to reproduce:
- Open Fenix and go to https://mozilla.github.io/webrtc-landing/pc_test.html;
- Check H.264 and tap on the "Start" button.
- Allow permissions.
Expected result: The video call starts.
Actual results: the video doesn't start and on the bottom of the page the following message is displayed: "Failure callback: Starting video failed". Nightly crashes if the page is reloaded.
Notes:
- the issue is reproducible even if the prefs are true or false in about:config:
"media.webrtc.platformencoder" and "media.webrtc.hw.h264.enabled"
Comment 1•6 years ago
•
|
||
I can reproduce this issue also with the HTC Desire 820 (Android 6.0.1).
Additionally from those behaviors, sometimes the page doesn't require the permissions. After restarting the page or reloading the page, Nightly crashes.
Notes:
- This issue is not 100% reproducible.
Comment 2•6 years ago
|
||
This appears to be failing at the point of obtaining the local camera, so this presumably happens regardless of the H.264 option—since that option only takes effect later—and presumably means there's not self-view as well.
I'm unable to repro on Samsung s8 (Android 9) FWIW.
After restarting the page or reloading the page, Nightly crashes.
Crash stats would be helpful, either here or in a separate bug if we're not sure these are related?
Additionally from those behaviors, sometimes the page doesn't require the permissions.
Not requiring permissions sounds concerning and Fenix-specific, probably deserving of its own bug.
Comment 4•6 years ago
|
||
Comment 5•6 years ago
|
||
Hello, I can't sent you guys a crash report because is not registered any crash report in the system.
Anyway, there is a logcat in with the errors from the app in those moments, I hope it helps.
If you guys need additional testing or help, let me know, thanks.
Updated•6 years ago
|
Comment 6•6 years ago
|
||
jib: See comment 4 for a log attachment (lots of errors, but I'm not sure what I'm looking for).
Comment 7•6 years ago
|
||
I'm not sure either. Who do we have looking at the android camera code these days?
Comment 8•6 years ago
|
||
John, you have experience reading logcats, if not camera stuff. Does anything in the logs in comment 4 jump out at you?
bumping over to :jolin
Comment 10•6 years ago
|
||
It is working on both of the phones I've tested and I see a lot of the same "errors" in the logcat. What jumps out at me are:
2019-11-05 09:07:28.940 957-1361/? E/mm-camera: <IMGLIB><ERROR> 303: AllocateBuffers: Invalid dimensions 0x0
...
2019-11-05 09:07:28.943 957-19072/? E/mm-camera: <IFACE ><ERROR> 2791: iface_util_modify_plane_info_for_native_buf: iface_util_modify_plane_info_for_native_buf: Invalid cam_format 2088533116
There is nothing like that in my logcat. Maybe we're picking a bad format on the affected phones? My understanding is that we are using old/deprecated camera APIs on Android. The webrtc.org code we're using was removed from upstream years ago.
Comment 11•6 years ago
|
||
Dan, I assume we're going to move over to maintained code at some stage, so I don't think it's worth it to investigate this? I haven't seen a bug for the webrtc.org update however, so can't dupe it.
Comment 12•6 years ago
|
||
I'm looking at updating the Android camera code from upstream in Bug 1578073. That might fix things here.
Updated•6 years ago
|
Updated•3 years ago
|
![]() |
||
Updated•1 year ago
|
Description
•