new | ask | show | jobs
yjftsjthsd-h 2 days ago  [ - ]

I trust hardware to mostly be reasonable when starting from off, but we're discussing the case where it's on and stays on but gets handed from one kernel to another and I don't trust it nearly as well in that case. I think the comparison is kexec rather than hibernate, and while it often works, kexec can result in misbehaving hardware.

yencabulator 7 hours ago  [ - ]

Many peripherals have a mechanism to reset the device, to get it back to a known good state. Generally device drivers will do this when they receive a message they don't understand from the device, or a command sent to the device times out without response.

Here's my graphics chip getting reset:

  [drm:mes_v11_0_submit_pkt_and_poll_completion.constprop.0 [amdgpu]] *ERROR* MES failed to response msg=3
  [drm:amdgpu_mes_unmap_legacy_queue [amdgpu]] *ERROR* failed to unmap legacy queue
  [drm:gfx_v11_0_hw_fini [amdgpu]] *ERROR* failed to halt cp gfx
  amdgpu 0000:c6:00.0: amdgpu: MODE2 reset
  amdgpu 0000:c6:00.0: amdgpu: GPU reset succeeded, trying to resume