Re: [PATCH net v3 3/4] ipmr: set up GSO state for tunnel vifs

[Julius Bairaktaris]

> What happens here if the GSO skb already has skb->encapsulation set?
[ ... ]
> Would it make sense to segment in software, or reject, GSO skbs that
> already have skb->encapsulation set before calling
> iptunnel_handle_offloads()?

Agreed. v4 drops such a packet in ip_encap() before touching its GSO
state. It was dropped in ipip_gso_segment() before this series as well,
so a GRE or VXLAN aggregate sent to a tunnel vif behaves as it did, and
the inner headers of its own tunnel stay intact for the other vifs.

> Should this use iph_set_totlen(iph, skb->len) instead?

Agreed, fixed in v4.