Hi, I went on and installed the respective image from
https://snapshot.debian.org/package/linux/5.10.46-5/#linux-image-5.10.0-8-amd64-unsigned_5.10.46-5
and had no luck, no change whatsoever.
Later next week I'll try running Fedora or Arch on a different drive to
see if it has been corrected on the firmware/kernel side, like the git said.
El 10/06/22 a las 10:57, Diederik de Haas escribió:
Could you try *downgrading* your kernel to 5.10.46-5, to be found here:
https://snapshot.debian.org/package/linux/5.10.46-5/#linux-image-5.10.0-8-amd64-unsigned_5.10.46-5
The reason for this is https://bugs.debian.org/990312 with a possible
similar
cause and therefor solution, which was part of 5.10.46-3.
The fix was reverting 2 commits wrt 'CP_MEC_DOORBELL_RANGE' and in
kernel 5.10.61 there was another commit related to that, and I'd like
to know
if that caused a regression. I don't expect it, but it's better to verify.
The symptom of bug 990312 were easy to detect with:
cat /sys/class/drm/card0/device/gpu_busy_percent
