On my AMD Renoir laptop, I’m experiencing machine freezes (which I notice when getting back to the scrensaver-locked session). dom0 logs show quite a number of:
xen-blkback: Scheduled work from previous purge is still busy, cannot purge list
… after an initial
May 28 20:35:13 dom0 kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_low timeout, signaled seq=9710147, emitted seq=9710150
May 28 20:35:13 dom0 kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process Xorg pid 5620 thread X:cs0 pid 6197
May 28 20:35:13 dom0 kernel: amdgpu 0000:07:00.0: amdgpu: GPU reset begin!
May 28 20:35:13 dom0 kernel: amdgpu 0000:07:00.0: amdgpu: SMU: I'm very busy for your command: index:8 param:0x00000000 message:DisallowGfxOff
May 28 20:35:13 dom0 kernel: amdgpu 0000:07:00.0: amdgpu: Failed to disable gfxoff!
after this several other problems can appear, e.g.
May 28 20:35:42 dom0 kernel: snd_hda_intel 0000:07:00.6: IRQ timing workaround is activated for card #1. Suggest a bigger bdl_pos_adj.
May 28 20:35:59 dom0 kernel: snd_hda_intel 0000:07:00.6: Refused to change power state from D0 to D3hot
May 27 19:11:48 dom0 kernel: [drm:amdgpu_acpi_power_shift_control [amdgpu]] *ERROR* ATCS PSC update failed
May 27 19:11:48 dom0 kernel: [drm] smart shift update failed
May 27 19:11:54 dom0 kernel: xen-blkback: backend/vbd/23/51728: using 2 queues, protocol 1 (x86_64-abi) persistent grants
May 27 19:11:55 dom0 qmemman.systemstate[2790]: dom '0' still hold more memory than have assigned (4150550528 > 4092113018)
Latest dnf transaction upgraded only microcode-ctl containing only intel microcodes, and qubes-dist-upgrade, and seems hard to blame; but OTOH the previous one ~2w ago was only qubes-* packages, and the system did have longer-lived and crash-less session since then. Previous transaction did update the kernel and xen, but here again there were long sessions with no such issue since then.
Does that ring any bell?