Yes, level 2 should always work now. However, level 2 is already incompatible with immediate release.
There are three levels of optimization that can be performed. Only the highest level requires the strict alignment.
Also needs documentation; what it's for, why 256
AMD hardware has no programmable stride and the stride must always be width * bpp rounded up to 256. This is hardcoded everywhere in mesa.
a...
This is slow - in most cases slower than memcpy:ing the buffer.
The compositor doesn't have access to a faster memcpy implementation than the client.
I've looked into this in the past but...
Logs are under ~/.local/share/jay/log/jay. Based on the vibes, you're not using systemd.
There are two commits.
I don't know how to verify we're hitting the zero-copy path
You can run the current master of https://github.com/mahkoh/jay. I'm not aware of any other compositor that tries to do this at the...
It is nowadays possible for the compositor to import the memfd created by foot as a udmabuf. This allows the compositor to perform the copy from the foot shm to the GPU texture on the GPU without...
I've gone ahead and written a replacement for my usecases: https://docs.rs/linearize
I also took the opportunity to fix some of the remaining issues with enum-map:
- The traits of the map,...
wm_capabilities.minimize