EIO crash in CH was caused by CH trying to writev
before the interface has been assigned any addresses;
adding an address prior to passing the FD (via 80-vm-vt.network)
solves the issue. It is still unclear what causes the interrupt=None
in the vhost_user/passt branch