[vfio-users] GPU not available for vfio-pci, in use by virtio-blk-pci

Bradley Davis bradleydavisjr at gmail.com
Wed Sep 23 03:53:53 UTC 2015


I figured it out after stepping away for a bit... I hadn't thought I had
made any changes in the last month but in reality I had added a couple raw
disk images to the VM through virt-manager. When I did this, it created 2
conflicts with how my xml was setup.

1) The first line for the qemu domain type gets over written with what is
supposed to be there stock. I am supposed to have this: <domain type='kvm'
xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'>
This allows my custom qemu:commandline to validate and run.
2) Also by using virt-manager, it didn't check to make sure I wasn't
already using a domain slot elsewhere. When I added the disks, they took
over the slot which my GPU was attached to which was slot 15 or
addr=0x0f...my gpu was being called though my qemu:commandline arguments.

The one drawback about not constantly messing with VFIO and PCIe
pass-through is that you forget what you learned and are unable to spot
seemingly obvious issues. I apologize for my poor terminology, i am out of
practice :-|

I hope my messup helps others.


On Tue, Sep 22, 2015 at 9:52 PM, Bradley Davis <bradleydavisjr at gmail.com>
wrote:

> So this error only crops up randomly. I'm not sure what exactly I do, but
> it only happens every couple of weeks when I reboot. Currently running
> Windows 10 and decided to reboot the VM to apply patches and to get over
> into the Inside Preview for the latest build - i was running GM stable. On
> reboot, I get strange graphical errors and artifacts. I destroy the VM and
> tell it to start again and I get this:
>
> [root at kvmhost1 ~]# virsh start GamingMachine
> error: Failed to start domain GamingMachine
> error: internal error: early end of file from monitor: possible problem:
> 2015-09-23T02:19:00.062246Z qemu-system-x86_64: -device
> vfio-pci,host=04:00.0,addr=0x0f,addr=0x0f,bus=pci.0,x-vid=0x10DE,x-did=0x11BA,x-ss-vid=0x10DE,x-ss-did=0x0965,x-vga=on:
> PCI: slot 15 function 0 not available for vfio-pci, in use by virtio-blk-pci
> 2015-09-23T02:19:00.062556Z qemu-system-x86_64: -device
> vfio-pci,host=04:00.0,addr=0x0f,addr=0x0f,bus=pci.0,x-vid=0x10DE,x-did=0x11BA,x-ss-vid=0x10DE,x-ss-did=0x0965,x-vga=on:
> Device 'vfio-pci' could not be initialized
>
> This error doesn't occur with every reboot. Normally I get a good amount
> of reboots with my VM for updates and software installs.
>
> The only remedy that I know of currently is to reboot... I'm reading some
> other vfio-users threads talking about binding issues and will look more
> into them.
>
> Current kernel is: 4.1.2-200.fc22.x86_64
>
> What are your thoughts?
>
> Thanks,
> Brad
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/vfio-users/attachments/20150922/9aee8e44/attachment.htm>


More information about the vfio-users mailing list