Re: [libvirt] CPU topology 'sockets' handling guest vs host