Let me clarify my question, I am using unmodified stock qemu (1.4.0) and only the location of vhost file descriptor is different from default (I don't remove the default /dev/vhost-net directory)

The VHOST file descriptor is present in "/dev/vhost-0" directory instead of the default "/dev/vhost-net". When I try to give vhostfd in the XML I get a different error:

2014-07-24 10:42:22.507+0000: 3181: error : qemuMonitorOpenUnix:293 : failed to connect to monitor socket: No such process

2014-07-24 10:42:22.507+0000: 3181: error : qemuProcessWaitForMonitor:1806 : internal error process exited while connecting to monitor: qemu-system-x86_64: -netdev type=tap,id=net1,script=no,downscript=no,ifname=port4,vhost=on,vhostfd=/dev/vhost-0: Device 'tap' could not be initialized

Whereas if I don't give the vhostfd option in the XML, it uses the default /dev/vhost-net directory and throws the following error:

2014-07-24 10:41:35.970+0000: 3183: error : qemuMonitorOpenUnix:293 : failed to connect to monitor socket: No such process

2014-07-24 10:41:35.970+0000: 3183: error : qemuProcessWaitForMonitor:1806 : internal error process exited while connecting to monitor: qemu-system-x86_64: -netdev type=tap,id=net1,script=no,downscript=no,ifname=port4,vhost=on: vhost-net requested but could not be initialized

qemu-system-x86_64: -netdev type=tap,id=net1,script=no,downscript=no,ifname=port4,vhost=on: Device 'tap' could not be initialized

On 24 July 2014 13:47, Asadullah Hussain <asadxflow@gmail.com> wrote:

The qemu that I am using isn't modified at all. It's the VHOST drivers that are mounted elsewhere i.e., not on default /dev/vhost-net.

Here is the command that I use to launch my qemu VM:
qemu-system-x86_64 -cpu host -boot order=c -hda /root/Disks/ubuntu1.qcow2 -m 1024M -smp 2 --enable-kvm -name 'client 1' -nographic -vnc :2 -net none -no-reboot -mem-path /dev/hugepages -mem-prealloc -netdev type=tap,id=net1,script=no,downscript=no,ifname=port3,vhost=on -device virtio-net-pci,netdev=net1,mac=00:00:00:00:00:01,csum=off,gso=off,guest_tso4=off,guest_tso6=off,guest_ecn=off -netdev type=tap,id=net2,script=no,downscript=no,ifname=port4,vhost=on -device virtio-net-pci,netdev=net2,mac=00:00:00:00:00:02,csum=off,gso=off,guest_tso4=off,guest_tso6=off,guest_ecn=off

As you can see I have created tap interfaces port3 & port4 and I just want to connect the VM to these VMs.

On 24 July 2014 13:31, Daniel P. Berrange <berrange@redhat.com> wrote:

On Thu, Jul 24, 2014 at 08:42:47AM +0500, Asadullah Hussain wrote:
> Thanks Wangkai but libvirt infact pre-configures network tap devices and
> pre-opens /dev/vhost-net file and passes these to qemu as open, ready to
> use file descriptors.
>
> My issue is that my qemu requires that /dev/vhost-net directory is removed
> and it's kernel module be unloaded. Then it creates a vhost -net device
> using its own configuration.

Libvirt is only written to work against official upstream QEMU. If you
have modified your QEMU to work in a different way wrt net devices then
you'll have to modify your libvirt to cope or workaround the problems.

Regards,
Daniel
--
|: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org -o- http://virt-manager.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|

--
Cheers,

Asadullah Hussain

Cheers,

Asadullah Hussain