tcp: fix ABC in tcp_slow_start()
authorEric Dumazet <edumazet@google.com>
Fri, 20 Jul 2012 05:02:33 +0000 (05:02 +0000)
committerDavid S. Miller <davem@davemloft.net>
Fri, 20 Jul 2012 17:59:41 +0000 (10:59 -0700)
When/if sysctl_tcp_abc > 1, we expect to increase cwnd by 2 if the
received ACK acknowledges more than 2*MSS bytes, in tcp_slow_start()

Problem is this RFC 3465 statement is not correctly coded, as
the while () loop increases snd_cwnd one by one.

Add a new variable to avoid this off-by one error.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Nandita Dukkipati <nanditad@google.com>
Cc: John Heffner <johnwheffner@gmail.com>
Cc: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_cong.c

index 04dbd7a..4d4db16 100644 (file)
@@ -307,6 +307,7 @@ EXPORT_SYMBOL_GPL(tcp_is_cwnd_limited);
 void tcp_slow_start(struct tcp_sock *tp)
 {
        int cnt; /* increase in packets */
+       unsigned int delta = 0;
 
        /* RFC3465: ABC Slow start
         * Increase only after a full MSS of bytes is acked
@@ -333,9 +334,9 @@ void tcp_slow_start(struct tcp_sock *tp)
        tp->snd_cwnd_cnt += cnt;
        while (tp->snd_cwnd_cnt >= tp->snd_cwnd) {
                tp->snd_cwnd_cnt -= tp->snd_cwnd;
-               if (tp->snd_cwnd < tp->snd_cwnd_clamp)
-                       tp->snd_cwnd++;
+               delta++;
        }
+       tp->snd_cwnd = min(tp->snd_cwnd + delta, tp->snd_cwnd_clamp);
 }
 EXPORT_SYMBOL_GPL(tcp_slow_start);