UnRaid 6.9 VM Issue after removing ACS override and Syslinux config

[smick]

Hi,

 

I recently updated to 6.9 and I'm unable to successfully start my Windows 10 VM after attempting to clean up overrides I had set with ACS override set to "both" and an exclusion in the syslinux configuration. I removed the acs override and restored the Unraid OS section of syslinux config to:

 

kernel /bzimage
append initrd=/bzroot

 

After rebooting and reserving the HW in System Devices and rebooting again. My WIndows 10 will not boot but gives an error that starts with "internal error: qemu unexpectedly closed the monitor: 20210...". Diagnostics and error image are attached.

 

Thanks!

 

Steve

tower-diagnostics-20210331-2257.zip

[bastl]

The ACS override changes the groupings of your PCI devices. By removing ACS you might have regrouped devices you wanna passthrough. Make sure the specific device is still in it's own group. Check under Tools >>> System Devices

 

"28:00.0 Intel Corporation 82576 Gigabit" is grouped with lot of other devices like "26:00.0 USB controller Renesas Technology Corp. uPD720201 USB 3.0 Host Controller" and "21:00.0 Network controller [0280]: Intel Corporation Wireless-AC 9260". Passthrough only works if a device you wanna passthrough is separated in it's own group. Look at all these devices together in group 14. 28:00.0 from your error log is a part of this group.

 

/sys/kernel/iommu_groups/13/devices/0000:01:00.0
/sys/kernel/iommu_groups/14/devices/0000:03:00.0
/sys/kernel/iommu_groups/14/devices/0000:03:00.1
/sys/kernel/iommu_groups/14/devices/0000:03:00.2
/sys/kernel/iommu_groups/14/devices/0000:20:00.0
/sys/kernel/iommu_groups/14/devices/0000:20:01.0
/sys/kernel/iommu_groups/14/devices/0000:20:04.0
/sys/kernel/iommu_groups/14/devices/0000:20:05.0
/sys/kernel/iommu_groups/14/devices/0000:20:06.0
/sys/kernel/iommu_groups/14/devices/0000:20:07.0
/sys/kernel/iommu_groups/14/devices/0000:21:00.0
/sys/kernel/iommu_groups/14/devices/0000:22:00.0
/sys/kernel/iommu_groups/14/devices/0000:26:00.0
/sys/kernel/iommu_groups/14/devices/0000:28:00.0
/sys/kernel/iommu_groups/14/devices/0000:28:00.1
/sys/kernel/iommu_groups/15/devices/0000:29:00.0

 

Best way to fix it is to reenable ACS Override. Not sure why you removed it in the first place if it worked with it enabled. An Unraid Update won't change the underlying reporting from the motherboards bios. The PCI groupings only changes if in some cases a BIOS update from your motherboard vendor reorganizes it.

[smick]

Thanks for the reply. I tried reenabling ACS override and it didn't work. I tried that along with using flash creator to go back to my 6.9.1 backup that worked before I started this exercise. I also tried going back to a 6.8.3 backup I had from three weeks ago and neither of them worked. When I start the VM, I see my monitor flash but no Windows logo or login. It seems its running in the background because I hear audio switching sounds like its trying to do something with the audio. If I try stopping the VM is does not stop itself. I have to do force stop.

 

I attached the diagnostics from 6.8.3 and the backup 6.9.1 I made just before trying this. 

 

A couple reasons why I tried this:

 

1) I was hoping to get rid of the Fix Common Problems message about ASC override being on and I shouldn't need to use the old pass through methods. I was hoping to clean up my setup now that 6.9 simplified the pass through process.

2) I have never been able to pass through my GT 1030 video card to Mac OS. I bought an AMD RX-560 I hoped to use to get both Windows and Mac OS running with pass through.

 

I thought my "out" would be the backup I made prior to trying this new pass through method. That backup does not boot Windows properly either. Shouldn't the restored backup always work the same? I have made no hardware changes.

 

I'm really stuck here. Nothing I try seems to work. 

 

Steve

Unraid Backup v6.9.1 tower-diagnostics-20210401-0938.zip UnRaid Backup v6.8.3 tower-diagnostics-20210401-0920.zip

[turnipisum]

Have you tried recreating the vm's? 

[smick]

No I haven't. Do you mean create a new Windows 10 VM and manually make my primary vdisk location the same as the offending VM? Should I be worried that it would corrupt the image? 

 

[smick]

I tried to recreate another Windows 10 test VM I had. I verified that I could get to the login screen using VNC with no pass throughs. Then I deleted the VM and recreated it with the pass throughs. No luck. Nothing from the video card seen on the monitor.

[smick]

I was able get the Windows 10 VM to boot. I left all the previous overrides in place and isolated the video card and associated audio in the device manager. I then passed through everything BUT the video card audio (passed through the MB audio controller) and to my amazement it booted successfully. 

 

I saved a backup of the flash drive and I admit I'm a little gun shy on trying anything further.

 

Question - Assuming I make no hardware changes, shouldn't reverting to a saved working backup with the USB creator always get me back to a working config? For some reason, that did not work for me.

 

Thanks for the help.

[ljm42]

2 hours ago, smick said:

1) I was hoping to get rid of the Fix Common Problems message about ASC override being on and I shouldn't need to use the old pass through methods. I was hoping to clean up my setup now that 6.9 simplified the pass through process.

 

I'm guessing the FCP warning was for "Legacy PCI Stubbing found". It would have said:

Quote

vfio-pci.ids or xen-pciback.hide found within syslinux.cfg. For best results on Unraid 6.9+, it is recommended to remove those methods of isolating devices for use within a VM and instead utilize the options within Tools - System Devices. See HERE for more details

 

It said this because your syslinux contains "vfio-pci.ids=1022:145f,8086:10c9":

BOOT_IMAGE=/bzimage vfio-pci.ids=1022:145f,8086:10c9 pcie_acs_override=downstream,multifunction initrd=/bzroot

"vfio-pci.ids=1022:145f,8086:10c9" can be removed from that line and you can use point and click on the System Devices page to pick which devices to bind to vfio-pci. Explanation and instructions are here:

  

 

 

14 minutes ago, smick said:

Question - Assuming I make no hardware changes, shouldn't reverting to a saved working backup with the USB creator always get me back to a working config? For some reason, that did not work for me.

 

The syslinux is stored on the flash drive, so restoring to a previous flash backup would have restored that. But VM configs are not stored on the flash drive so your VM changes would not have been restored.

[smick]

Yes "Legacy PCI Stubbing found" was the message. As I stated in my first message, I started by both removing that section and removing ACS override. This first time I did not bind the devices in System Devices. From there I realized my realized my network cards were in a very busy group 14 and bastl from above pointed out I still needed ASC override. Long story but after that I tried many combinations but could never get the video output until the last attempt. The only thing I know for sure is the last time I did not attempt to pass though the video card audio. 

 

From the error link instructions, it sounds like I should be able to remove "vfio-pci.ids=1022:145f,8086:10c9" and check the box in System Devices next to the corresponding "[1022:145f] 2a:00.3 USB controller: Advanced Micro Devices, Inc. [AMD] Zeppelin USB 3.0 Host controller", reboot and all will be well.