<div dir="ltr"><div dir="ltr"><div>Thanks Laine,</div><div><br></div><div>Mea Culpa, <br></div><div><br></div><div>Couldn't reproduce, or found the multiple duplicate entries in the guest pci domain/bus/slot/function space which got resolved easily.<br></div><div><br></div><div><br></div><div><br></div><div><br></div><div>[root@grit tmp]# virsh list<br> Id Name State<br>--------------------<br><br>[root@grit tmp]# virsh list --all<br> Id Name State<br>------------------------------<br> - openbsd6.7 shut off<br> - openindiana shut off<br> - proxmox6.2 shut off<br> - ubuntu20.04 shut off<br> - win2k19 shut off<br><br>[root@grit tmp]# virsh version<br>Compiled against library: libvirt 6.1.0<br>Using library: libvirt 6.1.0<br>Using API: QEMU 6.1.0<br>Running hypervisor: QEMU 4.2.1<br><br>[root@grit tmp]# virsh dumpxml ubuntu20.04 | gzip -c > ubunt2004.xml.gz <br>[root@grit tmp]# chown dan: ubunt2004.xml.gz<br>[root@grit tmp]# virsh edit win2k19 <br></div><div><br></div><div>(paste the 4 <hostdev mode='subsystem' type='pci' managed='yes'> devices from ubuntu2004 to the devices section)</div><div>save and exit.<br></div><div><br></div><div>error: XML error: Attempted double use of PCI Address 0000:08:00.0<br>Failed. Try again? [y,n,i,f,?]: <br><br></div><div>[root@grit tmp]# virsh dumpxml win2k19 | gzip -c > win2k19.xml.gz</div><div><br></div><br></div><div>Looking closer it seemed the 0000:08:00.0 was the guest address that already had a device defined on it.</div><div><br></div><div><br></div><div>Removing those in virt-manager and readding them generated a warning that they were in another definition, however</div><div>it did accept them.</div><div><br></div><div>This, for 4 pci devices, confused libvrtd in the meantime however it was still functional.</div><div><br></div><div>Aug 18 10:31:27 grit libvirtd[106082]: internal error: failed to get number of host interfaces: unspecified error - errors in loading some config files<br>Aug 18 10:31:55 grit libvirtd[106082]: internal error: failed to get number of host interfaces: unspecified error - errors in loading some config files<br>Aug 18 10:32:17 grit libvirtd[106082]: internal error: failed to get number of host interfaces: unspecified error - errors in loading some config files<br>Aug 18 10:32:32 grit libvirtd[106082]: internal error: failed to get number of host interfaces: unspecified error - errors in loading some config files<br></div><div><br></div><div dir="ltr"><div><br></div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Aug 18, 2020 at 1:00 AM Laine Stump <<a href="mailto:laine@redhat.com">laine@redhat.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On 8/8/20 11:53 PM, Daniel Black wrote:<br> > <br> > In attempting to isolate vfio-pci problems between two different guest <br> > instances, the creation of a second guest (with existing guest shutdown) <br> > resulted in:.<br> > <br> > Aug 09 12:43:23 grit libvirtd[6716]: internal error: Device 0000:01:00.3 <br> > is already in use<br> > Aug 09 12:43:23 grit libvirtd[6716]: internal error: Device 0000:01:00.3 <br> > is already in use<br> > Aug 09 12:43:23 grit libvirtd[6716]: Failed to allocate PCI device list: <br> > internal error: Device 0000:01:00.3 is already in use<br> <br> Hmm. Normally the error that would be logged if a device is already in <br> use would say something like this:<br> <br> error: Failed to start domain Win10-GPU<br> error: Requested operation is not valid: PCI device 0000:05:00.0 is in<br> use by driver QEMU, domain F30<br> <br> So you're encountering this in an unexpected place.<br> <br> > <br> > Compiled against library: libvirt 6.1.0<br> > Using library: libvirt 6.1.0<br> > Using API: QEMU 6.1.0<br> > Running hypervisor: QEMU 4.2.1<br> > (fc32 default install)<br> > <br> > The upstream code seems also to test definitions rather than active <br> > uses of the PCI device.<br> <br> <br> That isn't the case. You're misunderstanding what devices are on the <br> list. (see below for details)<br> <br> > <br> > My potentially naive patch to correct this (but not the failing test <br> > cases) would be:<br> > <br> > diff --git a/src/util/virpci.c b/src/util/virpci.c<br> > index 47c671daa0..a00c5e6f44 100644<br> > --- a/src/util/virpci.c<br> > +++ b/src/util/virpci.c<br> > @@ -1597,7 +1597,7 @@ int<br> > virPCIDeviceListAdd(virPCIDeviceListPtr list,<br> > virPCIDevicePtr dev)<br> > {<br> > - if (virPCIDeviceListFind(list, dev)) {<br> > + if (virPCIDeviceBusContainsActiveDevices(dev, list)) {<br> > virReportError(VIR_ERR_INTERNAL_ERROR,<br> > _("Device %s is already in use"), dev->name);<br> > return -1;<br> > <br> > Is this too simplistic or undesirable a feature request/implementation?<br> <br> Only devices that are currently in use by a guest (activePCIHostdevs), <br> or that libvirt is in the process of detaching from the guest + vfio and <br> rebinding to the device's host driver (inactivePCIHostdevs) are on <br> either list of PCI devices maintained by libvirt. Once a device is <br> completely detached from the guest and (if "managed='yes'" was set in <br> the XML config) re-binded to the natural host driver for the device, it <br> is removed from the list and can be used elsewhere.<br> <br> I just tested this with an assigned GPU + soundcard on two guests to <br> verify that it works properly. (I'm running the latest upstream master <br> though, so it's not an exact replication of your test)<br> <br> <br> > <br> > I'd be more than grateful if someone carries this through as I'm unsure <br> > when I may get time for this.<br> <br> <br> Can you provide the XML for your <hostdev> in the two guests, and the <br> exact sequence of commands that lead to this error? There is definitely <br> either a bug in the code, or a bug in what you're doing. By seeing the <br> sequence of events, we can either attempt to replicate it, or let you <br> know what change you need to make to your workflow to eliminate the error.<br> <br> </blockquote></div></div>