Hi Stephen,
> 2. Need more error checks. macpair won't work with TUN device. > and what happens if mac address is specified? Following Ferruh's feedback, and since the use-case is a bit different, I was thinking perhaps it would make more sense to add this functionality as a new device type rather than piggyback on the tap device (like --vdev=net_veth). This would allow for better isolation, both in code and in the configuration options, and would also be more accurate as the dpdk device is completely virtual while the tap device is the peer one. > 3. Should the mac address devarg, take two args? If we go with net_veth as above, then we can either drop the mac address option for simplicity, and let the user set them manually (the dpdk interface via dpdk API, and the kernel interface via ip / ifconfig), or provide an option to set either or both mac addresses as preferred. If it sounds good, I'll rework the patch first to implement it as a new net_veth device, post it, and then try to work out some basic tests (the latter might take some time tbh). Thanks!
