https://bugzilla.suse.com/show_bug.cgi?id=1219986 https://bugzilla.suse.com/show_bug.cgi?id=1219986#c7 --- Comment #7 from Tiziano Müller <tm@dev-zero.ch> --- (In reply to James Fehlig from comment #6)

Heh, yeah, I forgot about the udev interface backend. It, and the interface driver as a whole, do not get any love upstream. In fact, the idea of deprecating it has been floated

https://lists.libvirt.org/archives/list/devel@lists.libvirt.org/thread/ REKSP5LHZHC7TQ3FRDNUKRVQHJOD7GKL/#MWEFFP6W5IK4XPBKTGUB2TKTUGW3ELAK

Makes sense, also netcf does not seem to get much attention anymore. In fact, I am pretty sure that error code path for netconfig (in the opensuse fork of libvirt) has never been tested, otherwise that hang would have been found much earlier as it should also occur as root if netconfig could not initialise.

Do you use the virInterface* APIs (aka, 'virsh iface-*')? AFAICT, the udev backend only supports read operations

https://gitlab.com/libvirt/libvirt/-/blob/master/src/interface/ interface_backend_udev.c?ref_type=heads#L1241

I only testd with `virsh iface-list` and `virsh iface-dumpxml`, to be honest. My use case is vagrant-libvirt, which does an enumeration of the devices (via the fog libvirt ruby library), and kept hanging when using qemu:///session for the same reason `virsh iface-list` was hanging. So, read-only is good enough. With the proposed patch (and some minor modifications to vagrant-libvirt) I was finally able to get a VM up and running via qemu:///session. -- You are receiving this mail because: You are on the CC list for the bug.