optiluca

Members
  • Posts

    8
  • Joined

  • Last visited

optiluca's Achievements

Noob

Noob (1/14)

1

Reputation

  1. The offending device was a USB controller that I was passing through to the VM. It's still unclear why this started happening in 6.12.5 - but I've worked around it by not passing through the entire USB controller, but only the individual devices.
  2. I tried 6.12.5, just to be able to confirm when the issue appeared, and can confirm that 6.12.5 also presents this error. I've downgraded back to 6.12.4 for now.
  3. They seem to be unchanged. Going back to my original issue, the error is: Verify all devices in group 20 are bound to vfio- or pci-stub and not already in use And IOMMU group 20 only contains 1 device, and it is bound to vfio-...
  4. In my case the PCI passthrough is for the graphics card (I'm passing through the integrated graphics on a Ryzen 5600G). I have not found a fix, I'm stuck on Unraid 6.12.4.
  5. It's attached in the original post, unless you're referring to something different?
  6. A VM passthrough that had been working up to 6.12.4 inclusive (relevant config files/screenshots attached)W10.xml no longer works with 6.12.6 (I skipped .5, but from the release notes I suspect the issue was introduced there). Launching the VM now throws an Execution Error: internal error: qemu unexpectedly closed the monitor: 2023-12-08T18:46:07.612555Z qemu-system-x86_64: -device {"driver":"vfio-pci","host":"0000:06:00.4","id":"hostdev3","bus":"pci.4","addr":"0x0.0x3"}: vfio 0000:06:00.4: error getting device from group 20: Invalid argument Verify all devices in group 20 are bound to vfio- or pci-stub and not already in use IOMMU group 20 only contains 1 device, and it is bound to vfio- (as per the attached screenshot - the page and log output look exactly the same between 6.12.4 and 6.12.6. Rolling back to 6.12.4 restores VM functionality. deathstar-diagnostics-20231208-1950.zip
  7. Just wanted to thank Tanne for the guide. By following it I managed to get my 5600G iGPU passed through successfully. Additionally to all the points listed here, I had to change a few things (which took the best part of a day to figure out, so here they are for everyone else!): In your BIOS Make sure that the iGPU is enabled as the default (my ASRock mobo had external GPU as default. But I have no external GPU!). I disabled CSM as well - no idea if it was needed or not, but didn't seem to hurt. I think the notation used in step 6.3 might be obsolete. I was getting error messages about "hostdev0" not being a valid device. I had to change that whole section to: <qemu:override> <qemu:device alias='hostdev0'> <qemu:frontend> <qemu:property name='x-vga' type='bool' value='true'/> </qemu:frontend> </qemu:device> </qemu:override> I was seeing the same warning message as @ven0m above. Unraid was using the GPU for its terminal output, I had to modify the Syslinux configuration in Unraid and add video=efifb:off to the kernel argument list. After a reboot, the warning was gone (and indeed Unraid stopped writing output to my monitor during the boot sequence!) In general, I found myself "fighting" against whatever xml validation mechanism is in place for the VM configuration in Unraid. Often I'd make changes, save them, but re-opening the config would reveal that my configuration had been removed. I guess you guys already know this, but I found it quite confusing at first... On my Mobo I had to set PCIe ACS override=both and VFIO allow unsafe interrupts=yes, as my VGA and HDMI audio controllers shared a group with a tonne of other devices. I tested the system for an hour or so yesterday, and everything seemed OK, apart from some slight audio stutters every now and then. I'll be keeping it monitored. Hopefully someone will find these additions helpful. Luca