Merge branch 'bpf-fib-mtu-check'
authorDaniel Borkmann <daniel@iogearbox.net>
Tue, 22 May 2018 08:51:10 +0000 (10:51 +0200)
committerDaniel Borkmann <daniel@iogearbox.net>
Tue, 22 May 2018 08:51:11 +0000 (10:51 +0200)
commit3fb48d881dbe2ef5290d30c5e0a601e98c0be2b9
tree70615666ccc5c415f02efde857a8bb800a11791f
parentfd0bfa8d6e046237cb591bd3b51c87c56d4772b2
parent4f74fede40df8dbdb5261d50682491126675aac3
Merge branch 'bpf-fib-mtu-check'

David Ahern says:

====================
Packets that exceed the egress MTU can not be forwarded in the fast path.
Add IPv4 and IPv6 MTU helpers that take a FIB lookup result (versus the
typical dst path) and add the calls to bpf_ipv{4,6}_fib_lookup.

v2
- add ip6_mtu_from_fib6 to ipv6_stub
- only call the new MTU helpers for fib lookups in XDP path; skb
  path uses is_skb_forwardable to determine if the packet can be
  sent via the egress device from the FIB lookup
====================

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>