06:52 MrCooper: mort_: sounds like something on the client side doesn't properly use the Wayland dma-buf protocol feedback (assuming the KMS primary plane's IN_FORMATS property doesn't list AFBC)
06:53 MrCooper: hmm, or maybe mutter considers the KMS device a "secondary GPU", in which case direct scanout can't work yet
06:55 MrCooper: I doubt it's the latter, mutter prefers a device with KMS connectors for the primary GPU
08:54 mort_: I know direct scanout *can* work, regular Chromium ends up using a linear framebuffer which gets scanout when it's fullscreen