Windows 10 VM completely freezing after 6.11.1 upgrade

[snolly]

13 minutes ago, SimonF said:

And you still are running on 6.10.x? 6.11.1 has updated QEMU and Libvirt to 6.10.x

Used to when I first posted in here. Now on 6.11.1. Freezes occur in both versions. 

Edited October 22, 2022 by snolly

[SimonF]

27 minutes ago, snolly said:

Used to when I first posted in here. Now on 6.11.1. Freezes occur in both versions. 

Not related to this issue but do you pass thru a vbios for your 3060.

[snolly]

49 minutes ago, SimonF said:

Not related to this issue but do you pass thru a vbios for your 3060.

Nope. Never did, not even when this VM was Win10. With Win10 no freezes whatsoever.

[projektdotnet]

This appears to also affect Linux guests as well. I use a 6 core VM with EndeavourOS as my primary desktop with a passthrough AMD RX 6600 gfx card as the primary display/audio device. When I am actively using the VM I haven't had it lock up (though I've not been using it for usually more than an hour) but if I leave it idle for a few hours it would lock up with no clear indication in the logs as to why. The attached XML is after I've removed the shared memory from it as well as the shared directory, those are the only changes I've made to it before upload. I'll report back if it freezes again after those changes.

 

Host Hardware
Supermicro X11SCH-LN4F
128GiB ecc
Intel(R) Xeon(R) E-2146G CPU @ 3.50GHz
Radeon RX 6600
Navi 21/23 HDMI/DP Audio Controller

 

 

jaynestown-diagnostics-20221022-0548.zip serenity.xml

[johnsanc]

I tried Q35 and i440fx snd the results are exactly the same. If memoryBacking is included a freeze will occur within a few hours.

[NLS]

My VM Win 11 Pro uses Q35 (latest).
I booted 5 minutes ago... I'll let it be and see if it freezes...

 

[projektdotnet]

23 hours ago, projektdotnet said:

This appears to also affect Linux guests as well. I use a 6 core VM with EndeavourOS as my primary desktop with a passthrough AMD RX 6600 gfx card as the primary display/audio device. When I am actively using the VM I haven't had it lock up (though I've not been using it for usually more than an hour) but if I leave it idle for a few hours it would lock up with no clear indication in the logs as to why. The attached XML is after I've removed the shared memory from it as well as the shared directory, those are the only changes I've made to it before upload. I'll report back if it freezes again after those changes.

 

Host Hardware
Supermicro X11SCH-LN4F
128GiB ecc
Intel(R) Xeon(R) E-2146G CPU @ 3.50GHz
Radeon RX 6600
Navi 21/23 HDMI/DP Audio Controller

 

 

jaynestown-diagnostics-20221022-0548.zip 184.61 kB · 0 downloads serenity.xml 7.54 kB · 0 downloads

 

As a follow up to my previous comment, after removing the memoryBacking I haven't had any issues pop back up and it's been running since that comment without a single lock-up.

[NLS]

4 hours ago, projektdotnet said:

 

As a follow up to my previous comment, after removing the memoryBacking I haven't had any issues pop back up and it's been running since that comment without a single lock-up.

 

? Only if you don't use virtiofs you can remove that.

If you use it, it needs it.

Anyway, my report is that my VM has NOT crashed since yesterday that I booted it to let it idle and test.

[projektdotnet]

19 hours ago, NLS said:

 

? Only if you don't use virtiofs you can remove that.

If you use it, it needs it.

 

Right, I wasn't using that, I'd planned to so I'd had the tag setup but never actually mounted it. After removing the tag and the  backing I stopped having lockup issues, I was just reporting back that it fixed the issue for Linux guests (I assume it will for Windows guests as well)

[snolly]

I am having lockups without having memoryBacking in my XML

[NLS]

There is a 225.2 driver ISO... anybody tried it?

 

[NLS]

12 minutes ago, NLS said:

There is a 225.2 driver ISO... anybody tried it?

 

 

...and I did. It works. Fixes the install issue and after re-creating virtiofs service, it works as before (and hopefully better).

 

[ghost82]

Looking at the github commits, the only thing that changed from 225-1 to 225-2 is a patch for the viostor driver, for changing the discard sector alignement if not explicitly set, so I don't think your installation issues were fixed by this new version.

[NLS]

1 hour ago, ghost82 said:

Looking at the github commits, the only thing that changed from 225-1 to 225-2 is a patch for the viostor driver, for changing the discard sector alignement if not explicitly set, so I don't think your installation issues were fixed by this new version.

 

Well who knows. I know what happened. Maybe the change was in the installer itself, so it wouldn't be in the commits.

[ghost82]

10 minutes ago, NLS said:

 

Well who knows. I know what happened. Maybe the change was in the installer itself, so it wouldn't be in the commits.

 

Take into account that the virtio repository doesn't have only the virtio drivers, but all related to virtio; here you can see all the repositories, included that of the virtio installer:

https://github.com/orgs/virtio-win/repositories

 

0.1.225-1 was released on october 20th, 0.1.225-2 was released on october 24th, so if you look for commits in every repository between these 2 dates, you will find what changed.

https://fedorapeople.org/groups/virt/virtio-win/CHANGELOG

 

Anyway, this was only to give some more additional information, and very happy that in some way, as you said, who knows..., a new installation fixed your issues.

Edited October 26, 2022 by ghost82

[NLS]

2 minutes ago, ghost82 said:

 

...

 

I know, I already saw those.

 

 

[Gracey400]

Hoping I can use this thread to get some help. Neither of my VMs (W10 & W11) will boot after upgrading to 6.11.1.

 

They firstly crash, but I can get them into the Windows recovery screen by attempting the boot again.

 

Thanks in advance

[NLS]

If everything else fails, just recreate the VM using the same disks.
This will create a fresh XML.

 

When you manage to boot, use 225.2, 225.1 seems to be buggy somehow.

Also, why not use the latest chipsets emulated by KVM when running Win10 and 11?

W10 and W11 are surprisingly resilient to hardware (and virtual hardware) changes.

 

[snolly]

FYI. Using 225.2 seems to have fixed freezing issues. Win11 VM is up and running for 24h straight.

[NLS]

1 hour ago, snolly said:

FYI. Using 225.2 seems to have fixed freezing issues. Win11 VM is up and running for 24h straight.

 

Well this is weird because obvious differences seem more than what is shown in the commits.
IF that is the actual source that makes the build (because I am starting to have my doubts).

 

[johnsanc]

snolly's freezing issue was clearly unrelated the the freeze issue most of the people in this thread have. Again the freeze that shows no errors, blue screens, or other indication that anything is wrong is related to the presence of memoryBacking in the config XML. It likely has nothing to do with virtio drivers.

 

Hopefully we hear some kind of a response about this on where to go from here, but for now virtio-fs is not feasible for a lot of people simply because it requires memoryBacking.

[UNOPARATOR]

Any workarounds or solutions for this?

[blink515]

I'm having the same issue with my Win 10 VM freezing after upgrading unRAID to 6.11.5 and using virtio-fs.

 

I also noticed that all of the CPU cores I have pinned for this VM and not shared with anything else will show 100% usage until I force stop the VM. After I start the VM again I sometimes see the same image that was last shown during the freeze and have to force stop a second time to get the machine back to a normal state. I have a RTX3080 passed through and using Q35-7.1 for the machine type.

 

I'm using virtio-win-0.1.229 at the moment. Should I roll back to a different driver version? I was previously using an older version that didn't support virtio-fs since I was on unRAID 6.9.0.

 

Not sure if this is helpful, but I was trying to reinstall the virtio drivers and instantly get a freeze every time I try. See below lines from unRAID logs when this occurs.

 

Jan 18 01:15:12 Tower kernel: vfio-pci 0000:0e:00.0: No more image in the PCI ROM
Jan 18 01:15:12 Tower kernel: vfio-pci 0000:0e:00.0: No more image in the PCI ROM
Jan 18 01:15:14 Tower kernel: vfio-pci 0000:08:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0028 address=0xfffffffdf8000000 flags=0x0008]

 

Happy to start a new thread if preferred.

Edited January 18, 2023 by blink515
Added partial log info

[dja]

Seeing the same lockups with Win11.  Trying 16 Gb memory idea. Uuuugh.

[RhinoMedia]

On 10/14/2022 at 2:13 AM, SimonF said:

Do you have or are you able to create a Linux VM on your system to see if the issue is Linux or Windows guest specific?

I can confirm this happens in Linux. I have 5 shares mapped in fstab in and Xubuntu VM and in an KDE Neon VM. Connect to Xubuntu via Windows RDP and to KDE via NoMachine. This happens when I am away from the computer. Both VMs lose the remote connect and the allocated cpus are pinned at 100%. The only way to resolve it is to force stop the VM and restart it. If I remove the VirtioFS drives from fstab and use 9P or SMB it is fine.