selftests/bpf: test that GSO works in lwt_ip_encap
authorPeter Oskolkov <posk@google.com>
Tue, 5 Mar 2019 00:27:09 +0000 (16:27 -0800)
committerDaniel Borkmann <daniel@iogearbox.net>
Thu, 7 Mar 2019 09:41:40 +0000 (10:41 +0100)
commit17a90a78847324b1a34b0de833492cbd2366361d
treefaaf5c0521063b03c7d8040a992c5e0b60dc20a5
parentea0371f7879987cff70e21d808e3e9fea624c051
selftests/bpf: test that GSO works in lwt_ip_encap

Add a test on egress that a large TCP packet successfully goes through
the lwt+bpf encap tunnel.

Although there is no direct evidence that GSO worked, as opposed to
e.g. TCP segmentation or IP fragmentation (maybe a kernel stats counter
should be added to track the number of failed GSO attempts?), without
the previous patch in the patchset this test fails, and printk-debugging
showed that software-based GSO succeeded here (veth is not compatible with
SKB_GSO_DODGY, so GSO happens in the software stack).

Also removed an unnecessary nodad and added a missed failed flag.

Signed-off-by: Peter Oskolkov <posk@google.com>
Acked-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
tools/testing/selftests/bpf/test_lwt_ip_encap.sh