Eric Dumazet 6646145be9 udp: fix busy polling
[ Upstream commit a54d51fb2dfb846aedf3751af501e9688db447f5 ]

Generic sk_busy_loop_end() only looks at sk->sk_receive_queue
for presence of packets.

Problem is that for UDP sockets after blamed commit, some packets
could be present in another queue: udp_sk(sk)->reader_queue

In some cases, a busy poller could spin until timeout expiration,
even if some packets are available in udp_sk(sk)->reader_queue.

v3: - make sk_busy_loop_end() nicer (Willem)

v2: - add a READ_ONCE(sk->sk_family) in sk_is_inet() to avoid KCSAN splats.
    - add a sk_is_inet() check in sk_is_udp() (Willem feedback)
    - add a sk_is_inet() check in sk_is_tcp().

Fixes: 2276f58ac589 ("udp: use a separate rx queue for packet reception")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2024-01-31 16:17:04 -08:00
..
2021-07-01 13:19:48 -07:00
2022-08-23 10:43:11 +02:00
2022-07-28 22:06:15 -07:00
2022-08-09 22:14:02 -07:00
2022-09-30 18:17:17 -07:00
2021-12-16 07:18:35 -08:00
2023-11-20 11:52:15 +01:00
2021-10-15 11:33:08 +01:00
2024-01-31 16:17:04 -08:00
2023-09-19 12:28:03 +02:00
2021-10-13 09:40:46 -07:00
2022-05-11 12:43:10 +01:00
2023-10-10 22:00:46 +02:00
2023-04-13 16:55:23 +02:00
2022-06-19 10:00:02 +01:00
2023-06-14 11:15:20 +02:00
2024-01-31 16:17:04 -08:00
2021-08-03 13:05:26 +01:00
2024-01-10 17:10:27 +01:00
2024-01-10 17:10:27 +01:00
2023-08-11 12:08:17 +02:00