At least with both Firefox and Chrome using the test detailed below, it seems that my display is only receiving every other frame (with perfect cadence) that is rendered by the browsers. I've tested with both OpenGL2.1 and 3.0 compositors and can reproduce on both. I cannot reproduce this on Gnome shell. kwin 5.16 had the same issue a few months ago so this is not a new regression. STEPS TO REPRODUCE 1. Visit https://www.testufo.com/blackframes#count=2&bonusufo=1&equalizer=1&background=000000 in either Chrome or FF OBSERVED RESULT 1. The middle UFO is either completely missing (black frames only being shown) or twice the expected brightness (black frames not being shown). 2. The topmost and bottommost UFOs are operating at the same speed (visibly, both show proper framerates according to the browsers - this is a presentation issue not rendering). EXPECTED RESULT 1. The middle UFO should look substantially sharper but at the same brightness and speed as the top UFO. This one is using black-frame-insertion. 2. The bottommost UFO should be showing at twice the framerate (visibly smoother, usually looks slightly 'ahead' as well). SOFTWARE/OS VERSIONS Linux/KDE Plasma: Arch with Plasma 5.18.0 KDE Plasma Version: 5.18.0 KDE Frameworks Version: 5.67.0 Qt Version: 5.14.1 ADDITIONAL INFORMATION This is using the Nvidia binary driver on X11. I have had other people via Discord/IRC duplicate the issue, though it may only impact Nvidia users as no AMD users reported back.
I can reproduce this on 5700XT using mesa-git 20.1.0_devel.122121.089e1fb287e-1 SOFTWARE/OS VERSIONS Linux/KDE Plasma: Arch KDE Plasma Version: 5.18.4 KDE Frameworks Version: 5.68.0 Qt Version: 5.14.2
Update on this, I have just switched to kwinft and can confirm this issue is resolved by avoiding the use of kwin. Also bumping the version on this bug to 5.20.2 as I've confirmed it still exists with the latest kwin upstream release.
Fixed with the compositing scheduling rewrite.