net: speedup udp receive path

Since commit 95766fff ([UDP]: Add memory accounting.), 
each received packet needs one extra sock_lock()/sock_release() pair.

This added latency because of possible backlog handling. Then later,
ticket spinlocks added yet another latency source in case of DDOS.

This patch introduces lock_sock_bh() and unlock_sock_bh()
synchronization primitives, avoiding one atomic operation and backlog
processing.

skb_free_datagram_locked() uses them instead of full blown
lock_sock()/release_sock(). skb is orphaned inside locked section for
proper socket memory reclaim, and finally freed outside of it.

UDP receive path now take the socket spinlock only once.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 2 insertions, 2 deletions

diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 3ead20ad9d07..91c60f0090a4 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -424,7 +424,7 @@ out:
         return err;
 
 csum_copy_err:
-        lock_sock(sk);
+        lock_sock_bh(sk);
         if (!skb_kill_datagram(sk, skb, flags)) {
                 if (is_udp4)
                         UDP_INC_STATS_USER(sock_net(sk),
@@ -433,7 +433,7 @@ csum_copy_err:
                         UDP6_INC_STATS_USER(sock_net(sk),
                                         UDP_MIB_INERRORS, is_udplite);
         }
-        release_sock(sk);
+        unlock_sock_bh(sk);
 
         if (flags & MSG_DONTWAIT)
                 return -EAGAIN;