Hi! > The only negative impact it would have is that > if someone has old QEMU installed at define time and later upgrades to > a QEMU with PCI support before starting the guest, they'll lack the > PCI controller. This won't actually happen, because together with the missing controller he/she will not have PCI devices. Attempt to add a device without a bus immediately produces "No PCI bus" report, and the XML gets rejected. And, right after the upgrade, as soon as he/she adds some PCI devices to use, the XML will be re-parsed, and PCI bus description will be added. Because reparsing happens after any change. > So from that POV, maybe we should just clean up the current merged > patch so that it is properly exercised in the test suite by providing > capabilities I will try to do it soon. Kind regards, Pavel Fedin Expert Engineer Samsung Electronics Research center Russia -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list