firefox 82 requires USE="wayland" to be built with vaapi support. Please add a virtual vaapi flag requiring use_wayland in order to enable vaapi. Reproducible: Always Steps to Reproduce: 1.compile firefox 2.enable flags in about:config 3. run firefox 4. try to play 4k video in youtube using vp9 codec Actual Results: no vaapi is being used Expected Results: vaapi takes over
Created attachment 671608 [details] emerge --info
Created attachment 671611 [details, diff] proposed patch to enable vaapi flag
imo vaapi? ( media-video/ffmpeg[vaapi] ) is much better
(In reply to Perfect Gentleman from comment #3) > imo > vaapi? ( media-video/ffmpeg[vaapi] ) > is much better yes, probably, are you sure firefox uses ffmpeg libs and does not reply on libva directly?
(In reply to Janpieter Sollie from comment #4) > (In reply to Perfect Gentleman from comment #3) > > imo > > vaapi? ( media-video/ffmpeg[vaapi] ) > > is much better > > yes, probably, are you sure firefox uses ffmpeg libs and does not reply on > libva directly? yes, I'm sure.
We don't have virtual flags.
I am not feeling comfortable with this hack. The whole thing is still new and upstream is only supporting VAAPI for Wayland at this point and considers X11 as legacy (https://bugzilla.mozilla.org/show_bug.cgi?id=1662496). Patches are still welcome but exposing such a hack to our users is too risky for now. Maybe with Firefox 86 when another batch of patches have landed...
> We don't have virtual flags. I meant here: a virtual flag is something like 'it doesn't really change anything when compiling, it just points to the user that, when a feature is required (vaapi), it must be compiled with another flag (wayland)'. >I am not feeling comfortable with this hack. neither am I, but is it a good idea to hide a feature when you can read everywhere 'ff 82 supports vaapi'? > The whole thing is still new and upstream is only supporting VAAPI for Wayland > at this point and considers X11 as legacy > (https://bugzilla.mozilla.org/show_bug.cgi?id=1662496). > Patches are still welcome but exposing such a hack to our users is too risky for > now. Maybe with Firefox 86 when another batch of patches have landed. And if we hard-mask the flag?
