Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6

[profile/ivi/kernel-adaptation-intel-automotive.git] / net / ipv4 / tcp_output.c

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index cbfef8b..67f84f5 100644 (file)
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1035,6 +1035,13 @@ static void tcp_cwnd_validate(struct sock *sk)
  * introducing MSS oddities to segment boundaries. In rare cases where
  * mss_now != mss_cache, we will request caller to create a small skb
  * per input skb which could be mostly avoided here (if desired).
+ *
+ * We explicitly want to create a request for splitting write queue tail
+ * to a small skb for Nagle purposes while avoiding unnecessary modulos,
+ * thus all the complexity (cwnd_len is always MSS multiple which we
+ * return whenever allowed by the other factors). Basically we need the
+ * modulo only when the receiver window alone is the limiting factor or
+ * when we would be allowed to send the split-due-to-Nagle skb fully.
  */
 static unsigned int tcp_mss_split_point(struct sock *sk, struct sk_buff *skb,
                                        unsigned int mss_now, unsigned int cwnd)
@@ -1048,10 +1055,11 @@ static unsigned int tcp_mss_split_point(struct sock *sk, struct sk_buff *skb,
        if (likely(cwnd_len <= window && skb != tcp_write_queue_tail(sk)))
                return cwnd_len;
 
-       if (skb == tcp_write_queue_tail(sk) && cwnd_len <= skb->len)
+       needed = min(skb->len, window);
+
+       if (skb == tcp_write_queue_tail(sk) && cwnd_len <= needed)
                return cwnd_len;
 
-       needed = min(skb->len, window);
        return needed - needed % mss_now;
 }