xe_exec_reset.gt-reset-stress fails on TGL
The below is the dmesg on TGL:
5570 [ 1297.488444] xe 0000:00:02.0: [drm] *ERROR* Timed out wait for G2H, fence 175, action 3003
5571 [ 1297.500499] xe 0000:00:02.0: [drm] *ERROR* GuC PC query task state failed: -ETIME
5572 [ 1297.675424] xe 0000:00:02.0: [drm] *ERROR* GT reset failed, err=-62
5573 [ 1298.031070] xe 0000:00:02.0: [drm] G2H fence (175) not found!
This isn't on tip BTW, but can double check on tip shortly.
In general I think we need to harden the reset path to never actually fail as currently if this happens the machine is junked which isn't good.
@rodrigovivi - What do you think?