On Fri, Jul 22, 2016 at 04:55:16PM +0200, Pavel Hrdina wrote: > Hi, > > So there was made a decision that default ROM bars for network devices > shouldn't > be packaged for ARM architecture. Based on this thorough explanation [1] it
What do you mean by 'shouldn't be packaged for ARM architecture' ? It sounds like you are referring to the QEMU RPM packaging, which would sound like a bogus decision, since you can run non-arm emulator binaries on the arm host arch. > makes sense that it's useless. However it leads to a small issue with > starting > a guest on ARM hosts with PCI network device, if there are no ROM files > included > it will fail with error message: > > -device > virtio-net-pci,netdev=hostnet0,id=net0,mac=52:54:00:10:e9:80,bus=pci.2,addr=0x1: > failed to find romfile "efi-virtio.rom" > > there is also a bug [2]. > > It was originally created against virt-manager, but I've moved it to libvirt > temporally because I don't think that this should be a job for virt-manager to > workaround this ugly behavior of QEMU, because all other layers using libvirt > would have to do the same workaround. However I also thing that it isn't a > job > for libvirt too, because if you don't specify anything about using or not > using > a ROM bar (you just don't care about it) the QEMU process should start without > any error. This means that the default behavior should be based on the fact > whether there is or isn't a required ROM file to load. > > So I would like to start a discussion where would be the best place to fix > this. Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|
