Consider to increase the default value for `media.eme.max-throughput-ms`
Categories
(Core :: Audio/Video: Playback, defect, P2)
Tracking
()
People
(Reporter: alwu, Assigned: bryce)
References
(Blocks 1 open bug)
Details
Attachments
(1 file)
The pref media.eme.max-throughput-ms
was introduced by bug 1718223, and we also saw the same issue reported in bug 1721847.
Some reddit threads also show that this problem happening a lot among different sites. We should consider to increase the default value if this situation happens a lot.
[1]
https://www.reddit.com/r/Udemy/comments/mvpqts/drm_on_udemy_videos_since_march_breaks_legitimate/
https://www.reddit.com/r/firefox/comments/n1t0iy/crackling_while_while_playing_back_udemy_videos/
Reporter | ||
Comment 3•3 years ago
|
||
Update, our EME expert is now away, and I will ping him to update his thought about this when he's back on Feb. Currently adjusting this pref to a larger value should be enough for avoiding audio stuttering on websites using EME.
Comment 5•3 years ago
|
||
[Tracking Requested - why for this release]:
Seems that it is breaking some important websites like udemy
Comment 6•3 years ago
|
||
Alastor, can we get a patch up today? If it is safe to change this pref value, we may still have time on Monday to uplift it in our 97 release candidate. Thanks
Comment 7•3 years ago
|
||
What are the downsides to setting a higher default value? Would be good to understand the risk we'd be taking by changing it in the RC build.
Reporter | ||
Comment 8•3 years ago
|
||
As I say in comment3, I'm still waiting our EME expert back (Feb 1st), and he knows more on this topic than me. I would let him make the call on this.
Assignee | ||
Comment 9•3 years ago
|
||
(In reply to Ryan VanderMeulen [:RyanVM] from comment #7)
What are the downsides to setting a higher default value? Would be good to understand the risk we'd be taking by changing it in the RC build.
Theoretically, increasing the number in current Firefoxes is safe and the worst that will happen is that Widevine will internally rate limit. I.e. if we feed too fast, we may not get data back that fast, but it won't fall over.
I say theoretically because Widevine is a black box and we have code that indicates much more serious issues if Gecko doesn't rate limit (e.g. fatal playback issues).
Due to lack of test coverage for Widevine, I propose a small bump for this. We've seen that improve behaviour, and it has less risk than a large bump or outright removal of the Gecko rate limiting. I'll get a patch done up.
Assignee | ||
Comment 10•3 years ago
|
||
This allows for faster playback rates of EME media. This should allow up to ~5
times speed. Users in the wild have reported this value working without bustage,
so we have some evidence this is a safe value.
Updated•3 years ago
|
Comment 11•3 years ago
|
||
Assignee | ||
Comment 12•3 years ago
|
||
To expand on comment 7, while I'm relatively confident this will be fine, I think we should keep an eye on 2 things
- Fatal playback issues -- playback stops.
- Discontinuities in audio -- choppy audio, popping.
If we get any reports of these following landing in EME media, please NI me.
Comment 13•3 years ago
|
||
bugherder |
Comment 14•3 years ago
|
||
Been going back and forth on this one, but I think it's safer to let this fix ride the 98 train and give it a full Beta cycle for any fallout to appear rather than taking this in 97.0rc2 to ship in less than a week.
Description
•