So, i came home after a few days away, and saw an error on my TV that a program had stopped working, and windows was completely frozen.  It does not respond to any input from the mouse, nor the keyboard.

 

I tried shutting it down with the unRAID GUI, but that didn't work either.  So, I issued a xl destroy windows8 command from Putty, and the GUI shows it's now shutdown.

 

Ok, I restarted the VM from the GUI, and it shows it's started, but the same frozen screen I saw when I got home is still on the screen.  I have since shutdown the VM with xl destroy again, and the GUI once again shows it's shutdown, but I still see the same frozen screen, even though the VM isn't even running.

 

I got this error when I issued the xl destroy command...

 

root@media:~# xl destroy windows8
libxl: error: libxl_pci.c:990:libxl__device_pci_reset: The kernel doesn't support reset from sysfs for PCI device 0000:00:14.0

 

That refers to this device...

 

00:14.0 USB controller Intel Corporation Lynx Point USB xHCI Host Controller (rev 04)

 

not the GPU, as I would have expected.

 

So, how is the screen showing anything when no VM is running, and how do I clear it, without rebooting unRAID?

 

I haven't found a way without rebooting unRAID. It seems that the GPU needs a power-cycle to reset itself when it gets into this state.

 

I wonder if using Xen hot plug syntax would work from unRAID cl;

 

xl pci-detach <domain-id> <pci device> <guest virtual slot number>
xl pci-attach <domain-id> <pci device> <guest virtual slot number>

 

it might also be necessary to unbind and re-bind the device;

 

echo 0000:08:00.0 > /sys/bus/pci/devices/0000:08:00.0/driver/unbind
echo 0000:08:00.0 > /sys/bus/pci/drivers/pciback/new_slot
echo 0000:08:00.0 > /sys/bus/pci/drivers/pciback/bind

 

Must try this next time I get a lock up.

 

Peter

 

ah, that sounds familiar.  I think IronicBadger has something on his blog about setting this up.  I vaguely recall seeing something, bu didn't know I needed it, so I haven't looked into it yet.  I'll have to look into that now.  Plus, I'll save this thread to keep your solution handy also.

 

Thanks for the reply.

Well, the startup / shutdown scripts that Irinic provides are necessary but won't help in this instance.

 

The purpose of those scripts is to disable and re-enable the GPU on shutdown and statrup respectively. This works great as long as you correctly shutdown your VM.

 

However, in cases where the VM OS crashes and you need to issue an xl destroy (equivalent of yanking the power cable), the GPU won't be correctly ejected and may enter this locked state.

 

By all means, you need those scripts set up but you may still encounter locked GPUs following VM OS crash. (not too frequent, thankfully).

 

Peter

Good to know.  I'll work on getting that all setup later today.  I'm leaving the house for a while, so it'll have to wait.

 

If/when it hangs again, I'll try your suggested resolution, and report back.