udp: increments sk_drops in __udp_queue_rcv_skb()
authorEric Dumazet <dada1@cosmosbay.com>
Mon, 2 Feb 2009 21:41:57 +0000 (13:41 -0800)
committerDavid S. Miller <davem@davemloft.net>
Mon, 2 Feb 2009 21:41:57 +0000 (13:41 -0800)
Commit 93821778def10ec1e69aa3ac10adee975dad4ff3 (udp: Fix rcv socket
locking) accidentally removed sk_drops increments for UDP IPV4
sockets.

This field can be used to detect incorrect sizing of socket receive
buffers.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/udp.c

index b7faffe5c02984cccae813b3d32090fc0b66ff23..1ab180bad72a11195f0997ff6d15d77ce9fd5713 100644 (file)
@@ -1015,9 +1015,11 @@ static int __udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 
        if ((rc = sock_queue_rcv_skb(sk, skb)) < 0) {
                /* Note that an ENOMEM error is charged twice */
-               if (rc == -ENOMEM)
+               if (rc == -ENOMEM) {
                        UDP_INC_STATS_BH(sock_net(sk), UDP_MIB_RCVBUFERRORS,
                                         is_udplite);
+                       atomic_inc(&sk->sk_drops);
+               }
                goto drop;
        }