both are for running turbine retransmit over AF_XDP. xdp writes raw frames and bypasses the kernel, so it also skips the vlan driver: if your egress route resolves through a tagged sub-interface (vlanN), the frames go out untagged and the fabric drops them. so yeah, tagged sub-interfaces like that are exactly the case it fixes.
the thing that actually pushed it was multicast retransmit. multicast never ARPs, so xdp couldn't resolve a dest mac and dropped the packets. so one patch derives the multicast mac (rfc 1112), the other writes the 802.1Q tag in userspace, and together they let multicast fanout work over a tagged vlan. it's auto-detected from netlink :D