tcp: use DSACKs that arrive when packets_out is 0
authorNeal Cardwell <ncardwell@google.com>
Wed, 16 Nov 2011 08:58:02 +0000 (08:58 +0000)
committerDavid S. Miller <davem@davemloft.net>
Sun, 27 Nov 2011 23:54:09 +0000 (18:54 -0500)
The bug: Senders ignored DSACKs after recovery when there were no
outstanding packets (a common scenario for HTTP servers).

The change: when there are no outstanding packets (the "no_queue" goto
label), call tcp_fastretrans_alert() in order to use DSACKs to undo
congestion window reductions.

Other patches in this series will provide other changes that are
necessary to fully fix this problem.

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_input.c

index f772aaa4319ea5c716f3e35cf83cf01622f0c2bc..b49e41864037b6b30220c776a8ff2027461d66b8 100644 (file)
@@ -3788,6 +3788,10 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
        return 1;
 
 no_queue:
+       /* If data was DSACKed, see if we can undo a cwnd reduction. */
+       if (flag & FLAG_DSACKING_ACK)
+               tcp_fastretrans_alert(sk, pkts_acked, newly_acked_sacked,
+                                     is_dupack, flag);
        /* If this ack opens up a zero window, clear backoff.  It was
         * being used to time the probes, and is probably far higher than
         * it needs to be for normal retransmission.