Control: tags -1 + moreinfo Hi Mathias,
On Fri, Jan 10, 2025 at 01:17:37AM +0000, Mathias Gibbens wrote: > Package: src:linux > Version: 6.12.8-1 > Severity: normal > X-Debbugs-Cc: gib...@debian.org > > I have been running testing on this machine for the past four months, > and the AMD GPU has worked just fine with the 6.10 and 6.11 series > kernels. However, a couple of weeks ago when 6.12.5 migrated to > testing, I began to notice numerous kernel errors: > > > [ 2269.495240] amdgpu 0000:e3:00.0: PCIe Bus Error: severity=Correctable, > > type=Transaction Layer, (Receiver ID) > > [ 2269.495250] amdgpu 0000:e3:00.0: device [1002:744c] error > > status/mask=00002000/00000000 > > [ 2269.495256] amdgpu 0000:e3:00.0: [13] NonFatalErr > > [ 2269.495275] pcieport 0000:e0:01.1: AER: Correctable error message > > received from 0000:e3:00.0 > > While the GPU still works, I have also observed over the past couple > of weeks that sometimes when I turn on my monitor after a period of > inactivity (such as overnight) the GPU fails to "wake up" and nothing > is displayed. The system is still responsive (if I switch to a virtual > terminal I can CTRL+ALT+DEL to reboot the system), but there's no > graphical output. > > I did try downgrading firmware-amd-graphics to 20240909-2 just in > case there might have been something caused by the latest firmware > updates, but the issue persists. > > I am running the KDE Plasma 6 desktop with wayland, but have been > doing so for a couple of months now without issue. Assuming you still can reproduce the issue as well with the most recent iteration in unstable for 6.12.10, can you please report the issue to upstream at https://gitlab.freedesktop.org/drm/amd (and report back the the upstream issue here, or set the forwarded yourself). Thanks already, Regards, Salvatore
