Win 10 VM with Passthrough powers on, but won't boot

[harshl]

I have a VM I use as my gaming machine. It has been working fine for months or maybe even over a year through reboots, unraid upgrades, etc. Out of no where I cam down to it sitting at a tianocore screen with the first assigned CPU pegged and all other idle. Nothing terribly interesting in the logs in my eyes, but that probably doesn't mean much.

 

I am assuming it rebooted after some Windows updates or something and never came back up. Nothing else would have rebooted it that I can see. unraid hadn't rebooted in months. I took the opportunity to update to 6.8.3 hoping it would magically fix itself, but it hasn't.

I also tried deleting and recreating the VM, but that didn't seem to change anything either.

If I remove all passthrough devices (USB controller and RTX2060) it boots up just fine and everything looks as it should within VNC. So the disks are fine and this thing has worked just fine for a long time. I don't know what has changed. It was fine when I used it on Thursday and when I came into the room on Monday it was stuck at the tianocore screen.

 

Anyway, any thoughts on how to get this thing going again? Diagnostics attached. Appreciate any guidance.

 

-Landon

thecube-diagnostics-20200615-1241.zip

[jonp]

Hi Landon and so sorry to hear you are running into trouble!  Let's see if we can sort this out.  So just to make sure we're on the same page, your first indication of the issue was when you had the VM running and left it running for a while (and you were in Windows and everything was fine), then you leave it for a while and when you came back, the monitor itself was displaying the TianoCore logo and would never proceed to boot after that, correct?  Then you tried updating the OS, which required a reboot of the physical hardware, and after rebooting, do you

 

A) get the exact same behavior with a TianoCore logo on the monitor or

B) does the monitor just stay black now?

 

Next question would be what happens if you remove the USB controller but not the GPU?  What about removing the GPU but not the USB controller?  Would like to narrow this down to a specific device causing the problem if possible.

 

Lastly, I see your system has an integrated graphics controller on the CPU.  Do you have a monitor attached to that?  If not, can you?  Also check the BIOS settings again as some motherboard will randomly flip which graphics device is the default for use with the VM.  Another thing you can try is to recreate the VM using SeaBIOS instead of OVMF and see if that works.

 

Let us know and I'll keep an eye on this thread to support you.

[harshl]

Jon! Thank you so much for the response! Appreciate all you guys do with UNRAID, what a great distro.

Quote

So just to make sure we're on the same page, your first indication of the issue was when you had the VM running and left it running for a while (and you were in Windows and everything was fine), then you leave it for a while and when you came back, the monitor itself was displaying the TianoCore logo and would never proceed to boot after that, correct?

That is correct. I also noticed once I removed all passthrough devices and booted it, that it was completing Windows updates, so 99% certain that Windows update is what rebooted it.

Quote

Then you tried updating the OS, which required a reboot of the physical hardware, and after rebooting, do you

 

A) get the exact same behavior with a TianoCore logo on the monitor or

B) does the monitor just stay black now?

A) exact same behavior

Quote

Next question would be what happens if you remove the USB controller but not the GPU?  What about removing the GPU but not the USB controller?

With the GPU removed and the USB controller present, it hangs the same way, with the exception that the VNC session does not show the TianoCore screen, but rather says the guest has not initialized the display yet. Still powered on and has the first assigned core pegged.

With the USB controller removed and the GPU present, it boots properly!

I actually tried this earlier, but I received a PCI error and it wouldn't power on. Due to the limited time I had, I gave up at that point.  This time I realized I only get that on the first power on attempt after changing passthrough devices. Subsequent power on attempts actually work properly.

 

After you got me pointed in the right direction of the USB controller, I disconnected all USB devices, reassigned the controller to the VM and booted again. I then hot plugged all of the USB devices and they all worked fine. When I was doing that, I realized that I had a USB CDROM attached that normally isn't. I disconnected only that device and shutdown, then powered the VM back on and it booted correctly. To confirm, I shutdown, plugged the USB CDROM drive back in and powered back on... stuck at the TianCore screen. So the USB CDROM was the culprit in my case and I had forgotten I connected it weeks ago.  I feel like a dummy for not remembering that change, but I am getting old... 😆

 

Sure do appreciate your guidance in helping me work through the appropriate steps.

 

In case it is of any use, attached is the PCI error I get each time I remove all passthrough devices, save the VM config, then add back in any passthrough device (video card or USB or both). The first power on will fail with this error or something very similar, device IDs might change, subsequent power on attempts will work perfectly fine.

 

Again, thank you so much for the help and the great product you all have created in UNRAID!

-Landon