On Fri, Nov 05, 2021 at 03:22:27AM +0100, Emiel Kollof wrote: > On Tue, 2 Nov 2021 21:17:43 +1300 > Avon Robertson <[email protected]> wrote: > > > Please read the preceding post before this. > > > > The information below is the second post made by me pertaining to > > thread 'drmfreeze'. It is the first? post made to determine when the > > amdgpu firmware introduced this bug. > > I have reinstalled my system with 7.0 (release with latest syspatches). > I still have the same problem. Here's my dmesg when it corrupts. > > I can still get into the machine with ssh, but the screen is all > corrupted, and the xenocara session is frozen.
I believe amdgpu ring timeouts have been seen with all of the ports of amdgpu and can be triggered with a concurrent piglit run. As discussed in the archives it isn't any one particular test which triggers it. 6.9 drm is ported from linux 5.7, 7.0 drm from linux 5.10. I had hoped the 5.10 drm might help some problems such as corruption seen on renoir without compositing https://gitlab.freedesktop.org/mesa/mesa/-/issues/3699 but problems remain.
