bpf: test_tc_tunnel.sh needs reverse path filtering disabled
authorAlan Maguire <alan.maguire@oracle.com>
Mon, 25 Mar 2019 09:36:37 +0000 (09:36 +0000)
committerDaniel Borkmann <daniel@iogearbox.net>
Mon, 25 Mar 2019 14:01:54 +0000 (15:01 +0100)
commit0c4ea7f87abbdb56df616678bc23f10e51a0b4f8
treed79acfdfbe4348eaf73d8e5bb283bceeaaa3769b
parent62b31b42cff924c7d1e9a095b68ff3bbfc49b15b
bpf: test_tc_tunnel.sh needs reverse path filtering disabled

test_tc_tunnel.sh sets up a pair of namespaces connected by a
veth pair to verify encap/decap using bpf_skb_adjust_room.  In
testing this, it uses tunnel links as the peer of the bpf-based
encap/decap.  However because the same IP header is used for inner
and outer IP, when packets arrive at the tunnel interface they will
be dropped by reverse path filtering as those packets are expected
on the veth interface (where the destination IP of the decapped
packet is configured).

To avoid this, ensure reverse path filtering is disabled for the
namespace using tunneling.

Fixes: 98cdabcd0798 ("selftests/bpf: bpf tunnel encap test")
Signed-off-by: Alan Maguire <alan.maguire@oracle.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
tools/testing/selftests/bpf/test_tc_tunnel.sh