6by9

Added a patch to increase the timeout to the workflow "install toolchain" step as it appears to be failing regularly. Hopefully that'll get it through.

Upstream appear to be having some debate over this as to how it should be implemented, and referencing to previous attempts that caused discussion over 16bit precision or float and...

> CM5, Bookworm, kernel 6.12.47+rpt-rpi-v8 has this error never. CM5 has no hardware H264 decoder, so that's just using software decode. Are you running with a GUI running, or direct...

For debugging purposes please add `start_debug=1` to /boot/firmware/config.txt and reboot. When it goes wrong, grab the output from `sudo vclog -m` and `sudo vclog -a`. I have tried this on...

No smoking gun there. I was hoping for a VCOS_ABORT line or other error logging, but there isn't any. If I've read it right, `h264_parse_sei_ol line 2089` is only a...

> Would it be possible to make the bcm2835 hardware decode path more robust in this situation? Warnings are perfectly fine for me, but the hard stall (timeouts, "driver bug"...

And I'm surprised that kmsprint is giving `FB 722 1920x1080 XR24`. That would imply that it's doing GPU composition of the display, when it should be possible to pass the...

Branch rebased to keep it relevant. I've just noted https://lore.kernel.org/dri-devel/[email protected]/ So Xe won't work on the 16kB kernel at present. I don't see it having been merged, so at some...

Branch rebased.

@pigong I haven't really the time to dig into these drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c:639 amdgpu_irq_put would appear to be that the interrupt hasn't been enabled, so there are an unbalanced number of calls...