I upgraded to 6.9.0 then to 6.9.1    and I have been having some issues.   My win10 vm keeps crashing.  Looks like video card or something.  THis was working perfect before upgrade.   Now I see this in my log.  Just keeps repeating.  

 

I have two graphics cards in server.  Both AMD rx580 and rx570.    I am not even using them at the moment.  Just ran the win10 vm on vnc to try and get it to work.  Just keeps crashing and corrupting.  Tried different ssd's so it is not a drive issue.  

 

Apr 3 12:16:57 Tower kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400000000).
Apr 3 12:16:57 Tower kernel: [drm] UVD and UVD ENC initialized successfully.
Apr 3 12:16:57 Tower kernel: [drm] VCE initialized successfully.
Apr 3 12:16:57 Tower kernel: amdgpu 0000:0a:00.0: [drm] Cannot find any crtc or sizes
Apr 3 12:17:08 Tower kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400000000).
Apr 3 12:17:08 Tower kernel: [drm] UVD and UVD ENC initialized successfully.
Apr 3 12:17:08 Tower kernel: [drm] VCE initialized successfully.
Apr 3 12:17:08 Tower kernel: amdgpu 0000:0a:00.0: [drm] Cannot find any crtc or sizes
Apr 3 12:17:19 Tower kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400000000).
Apr 3 12:17:19 Tower kernel: [drm] UVD and UVD ENC initialized successfully.
Apr 3 12:17:19 Tower kernel: [drm] VCE initialized successfully.
A

tower-diagnostics-20210403-1231.zip

You have two Radeon cards, both bound to the amdgpu driver. You can't pass a bound card through to a VM. Choose the one you want to pass through and bind it to the VFIO stub driver instead by going to Tools -> System Devices then ticking the appropriate boxes, then clicking "Bind selected to VFIO at boot" and re-booting.

 

Incidentally, your diagnostics are for 6.9.0, not 6.9.1.

 

Yeah.  I tried to downgrade thinking that was some sort of issue.   I will go back to 6.9.1   Thanks!  I will try the vfio thing