*ERROR* GT0: load failed: status = 0x400000A0

[ 4945.952369] xe 0000:06:00.0: [drm] *ERROR* GT0: load failed: status = 0x400000A0, time = 4ms, freq = 2100MHz (req 2400MHz), done = -1
[ 4945.952386] xe 0000:06:00.0: [drm] *ERROR* GT0: load failed: status: Reset = 0, BootROM = 0x50, UKernel = 0x00, MIA = 0x00, Auth = 0x01
[ 4945.952391] xe 0000:06:00.0: [drm] *ERROR* GT0: firmware signature verification failed
[ 4945.952397] xe 0000:06:00.0: [drm] *ERROR* CRITICAL: Xe has declared device 0000:06:00.0 as wedged.
               IOCTLs and executions are blocked. Only a rebind may clear the failure
               Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/xe/kernel/issues/new
[ 4946.012416] xe 0000:06:00.0: [drm] *ERROR* GT0: GuC mmio request 0x508: no reply 0x508
[ 4946.012425] xe 0000:06:00.0: [drm] *ERROR* GT0: Failed to enable GuC CT (-ETIMEDOUT)
[ 4946.012451] xe 0000:06:00.0: [drm] *ERROR* GT0: resume failed (-ETIMEDOUT)

This happens fairly reliably for me when I leave the system unattended for a while (and the screens turn off due to swaylock). Interestingly, I can not reproduce the issue manually (with swaymsg), so presumably it only happens when the screens are turned off for several minutes at least. From the user perspective, the screens simply refuse to turn back on but switching to a tty works.

FWIW I used to have a similar problem with amdgpu which I was only able to resolve by switching to the Arc A750 card (it's working fine with i915). It seems there's something about my combination of monitors (Samsung LC49G95T and LG 27MU67) that's stressing GPU drivers.

I can also provide logs from sway if needed.