This mini series contains a fix for a bug noticed when analyzing a reported
merge conflict between bpf-next and net tree [0].
Apart from fixing a corner-case that affects use of BPF sk_lookup in tandem
with UDP reuseport groups with connected sockets, it should make the
conflict resolution with net tree easier.
These changes don't replicate the improved UDP socket lookup behavior from
net tree, where commit efc6b6f6c311 ("udp: Improve load balancing for
SO_REUSEPORT.") is present.
Happy to do it as a follow up. For the moment I didn't want to make things
more confusing when it comes to what got fixed where and why.
Thanks,
-jkbs
Cc: Kuniyuki Iwashima <[email protected]>
Cc: Willem de Bruijn <[email protected]>
[0] https://lore.kernel.org/linux-next/[email protected]/
Jakub Sitnicki (2):
udp: Don't discard reuseport selection when group has connections
selftests/bpf: Test BPF socket lookup and reuseport with connections
net/ipv4/udp.c | 5 +-
net/ipv6/udp.c | 5 +-
.../selftests/bpf/prog_tests/sk_lookup.c | 54 ++++++++++++++++++-
3 files changed, 55 insertions(+), 9 deletions(-)
--
2.25.4
