Hello! > 00:04:0 is VFIO passthrough, and i put everything to bus#0 by hands, for > MSI-X to work. I > could also leave devices at bus#2, just in > this case i don't get MSI-X, and this, for example, makes vhost-net unable to > use irqfds (or > it cannot initialize at all, again, i > already don't remember), because this requires per-event irqs.
Just tested Q35 machine on x86-64, and retested the same on ARM64. Sorry, but please ignore what i said before about MSI-X. Everything works fine, MSI-X is OK on bus#2 after the bridge, and this was either some old bug in old qemu, or i was simply doing something crappy and wrong. So, i don't care anymore about this bridge complex on PCIe bus, it does not make my living any worse. If libvirt strictly distinguishes between PCI and PCIe, let it just be there. And, to complete the testing, i've just migrated the Q35 machine and the bridge also breaks down. 'lspci -v' starts reporting "Unknown header type: 7f" for the bridge and for all devices behind it (and all devices fail). But it just means it should be fixed in qemu. Kind regards, Pavel Fedin Expert Engineer Samsung Electronics Research center Russia -- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
