Control: reassign -1 src:linux
Control: forcemerge 1102914 -1
On Sun, Apr 13, 2025 at 08:53:13AM +0200, Bernd Rinn wrote:
> Package: linux-image-amd64
> Version: 6.1.133-1
>
> Description:
> I noticed a regression with the most recent kernel 6.1.0-33 (6.1.133) of
> Debian Bookworm:
> libvirtd VMs don't start up anymore when they are being passed through a PCI
> device (in my case the PCI device is a network card). The error message is:
>
> """
> Apr 13 07:35:08 noc libvirtd\[4984\]: libvirt version: 9.0.0, package:
> 9.0.0-4+deb12u2 (Debian)
> Apr 13 07:35:08 xyz libvirtd\[4984\]: hostname: xyz
> Apr 13 07:35:08 xyz libvirtd\[4984\]: Unable to read from monitor: Connection
> reset by peer
> Apr 13 07:35:08 xyz libvirtd\[4984\]: internal error: qemu unexpectedly
> closed the monitor: 2025-04-13T05:35:08.436709Z qemu-system-x86_64: -device
> {"driver":"vfio-pci","host":"0000:05:00.0","id":"hostdev0","bus":"
> pci.8","addr":"0x0"}: VFIO_MAP_DMA failed: Cannot allocate memory
> 2025-04-13T05:35:08.447462Z qemu-system-x86_64: -device
> {"driver":"vfio-pci","host":"0000:05:00.0","id":"hostdev0","bus":"pci.8","addr":"0x0"}:
> vfio 0000:05:00.0: failed to set
> """
>
> The definition of the PCI device pass through looks like this:
>
> """
> <hostdev mode="subsystem" type="pci" managed="yes">
> <driver name="vfio"/>
> <source>
> <address domain="0x0000" bus="0x05" slot="0x00" function="0x0"/>
> </source>
> <alias name="hostdev0"/>
> <address type="pci" domain="0x0000" bus="0x08" slot="0x00" function="0x0"/>
> </hostdev>
> """
>
> When reverting the kernel back to previous version 6.1.0-32 (6.1.129), the
> VM boots up OK, i.e. the error is gone.
Given in the above PCI passthrough is involved I suspect this has the
same root cause as #1102914 .
Can you bisect the changes between 6.1.129 and 6.1.133 to identify the
offending commit? Can you test 6.1.134 from upstream?
As mentioned in the other bugreport I will look at it ASAP.
Regards,
Salvatore
