On Sun, Jun 28, 2009 at 05:46:51PM +0200, Mirko Raasch wrote: > Hi, > > i want to use more then one pci card in my debian guest. > > But when i start the guest, the log says: > > libvirtd: 15:44:55.459: warning : pciTrySecondaryBusReset:483 : Other > devices on bus with 0000:05:01.0, not doing bus reset > IIUC, what libvirt is complaining about is that there are other devices in the PCI bus which are not associated with this guest, and thus there is no way to safely reset the device you are trying to assign, without endangering the host OS or other guest OS > With "/usr/bin/qemu-system-x86_64 -m 512 -name test -hda > /media/vm/test.img -boot c -vnc :2 -k de -pcidevice host=05:01.0 > -pcidevice host=05:02.0 -pcidevice host=05:03.0" there seems to be no > problems. > > How can i start my guest with three pci devices passed through and libvirt? > The difference is that when you launch QEMU manually there is no checking for whether the PCI devices are in use by other guests, or by the host OS. So while it may launch QEMU, it is not running safely, and eg, if your guest OS does a PCI bus reset it could kill/harm your host OS. Regards, Daniel

Hi Mirko, On Sun, 2009-06-28 at 17:46 +0200, Mirko Raasch wrote: > Hi, > > i want to use more then one pci card in my debian guest. > > But when i start the guest, the log says: > > libvirtd: 15:44:55.459: warning : pciTrySecondaryBusReset:483 : Other > devices on bus with 0000:05:01.0, not doing bus reset > > With "/usr/bin/qemu-system-x86_64 -m 512 -name test -hda > /media/vm/test.img -boot c -vnc :2 -k de -pcidevice host=05:01.0 > -pcidevice host=05:02.0 -pcidevice host=05:03.0" there seems to be no > problems. > I'm afraid this is purely a limitation with the current code in qemu: /* For now, we just refuse to do a secondary bus reset * if there are other devices/functions behind the bus. * In future, we could allow it so long as those devices * are not in use by the host or other guests. */ if (pciBusContainsOtherDevices(conn, dev)) { VIR_WARN("Other devices on bus with %s, not doing bus reset", dev->name); return -1; } We also have a bug tracking this issue: https://bugzilla.redhat.com/499678 Fixes welcome! :-) Cheers, Mark.