1841709 - WebM VP8/VP9 not supported in FF Nightly, fails in Decode metadata (builds/worker/checkouts/gecko/dom/media/MediaDecoderStateMachine.cpp:372 and worker/checkouts/gecko/dom/media/MediaDecoderStateMachineBase.cpp:166)

Status: RESOLVED INVALID

(Core :: Audio/Video, defect)

Description

[Frank de Lange]

FF Nightly (tested with 116 and 117) on Linux does not support WebM VP8/VP9 video, when attempting to play such a video the user is presented with No video with supported format and MIME type found on screen and the following on stderr:

[Child 244810, MediaDecoderStateMachine #1] WARNING: Decoder=7fd727e9aa00 state=DECODING_METADATA Decode metadata failed, shutting down decoder: file /builds/worker/checkouts/gecko/dom/media/MediaDecoderStateMachine.cpp:372
[Child 244810, MediaDecoderStateMachine #1] WARNING: Decoder=7fd727e9aa00 Decode error: NS_ERROR_DOM_MEDIA_METADATA_ERR (0x806e0006): file /builds/worker/checkouts/gecko/dom/media/MediaDecoderStateMachineBase.cpp:166

How to reproduce: using FF Nightly 116/117 on Linux, open a WebM video and watch it fail: https://base-n.de/webm/out9.webm

Another way to show the missing WebM support is to open the HTML5 test site (https://html5test.com/) and notice the failing video support (WebM Vorbis and Opus audio work, only VP8/VP9 video are failing).

Bug was discovered while testing Invidious (https://github.com/iv-org/invidious/issues/3966)

Comment 1

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

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

Comment 2

[Jim Mathies [:jimm]]

Appears Linux specific, the video in question plays on Windows.

Comment 3

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

The severity field is not set for this bug.
:jimm, could you have a look please?

For more information, please visit BugBot documentation.

Comment 4

[Jim Mathies [:jimm]]

Alastor, do you have a linux setup you can test this on?

Comment 5

[Paul Adenot (:padenot)]

This works here on linux, today's nightly.

Comment 6

[Paul Adenot (:padenot)]

Please try with our builds (https://www.mozilla.org/en-US/firefox/channel/desktop/#nightly), and otherwise provide the raw data from about:support.

Comment 7

[Frank de Lange]

Attachment: about:support (raw)

It has been a while and Nightly is now at 125.0a1 but the results are still the same. Here's the raw data for about:support

Comment 8

[Frank de Lange]

Attachment: about:config (text)

...and here's the textual version for completeness' sake

Comment 9

[Frank de Lange]

Attachment: about:config (text)

(duplicate of above, ignore)

Comment 10

[Frank de Lange]

(empty, ignore)

Comment 11

[Paul Adenot (:padenot)]

Go to about:config, search for media.rdd-process.enabled, flip that back to true, restart the browser.

Comment 12

[Frank de Lange]

Flipping media.rdd-process.enabled and playing a video leads to an immediate hard kernel crash (which is why that option was disabled to begin with) which can only be resolved by power cycling the machine. Kernel crashes are not Firefox problems but most likely related to the video driver so in that respect this bug can be seen as 'resolved', alas it is not for me.

Comment 13

[Paul Adenot (:padenot)]

Yes, disable acceleration as well, that's the other pref you have flipped that aren't the default: media.ffmpeg.vaapi.enabled and media.hardware-video-decoding.force-enabled. Generally, all this is expected to work well and is tested. If it doesn't work, try a new profile -- if that makes it work, it's something on your side that was modified. about:support has a list of modification that have been done on the Firefox profile.

Comment 14

[Frank de Lange]

That seems to work for now, no crash yet and the video I linked above plays without problems. Hardware acceleration seems to trigger a driver bug but as said that is not under Firefox' purview. This bug can remain closed.

Comment 15

[Paul Adenot (:padenot)]

You seem to have a GPU that's a bit old unfortunately, so I'm not sure that we can make it work. It's even using a software implementation for the graphics rendering, so that doesn't bode well (and doing video decoding acceleration without hardware accelerated graphics rendering isn't particularly useful). This GPU might even be in our blocklist because of knows driver bugs.

Sorry about that.