diff options
| author | Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> | 2008-09-20 21:25:15 -0700 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2008-09-20 21:25:15 -0700 |
| commit | ef9da47c7cc64d69526331f315e76b5680d4048f (patch) | |
| tree | 97407b83c7c339dbe1fd0948c30f35e1680562ec /net/ipv4/tcp_output.c | |
| parent | f0ceb0ed86b4792a4ed9d3438f5f7572e48f9803 (diff) | |
| download | linux-ef9da47c7cc64d69526331f315e76b5680d4048f.tar.bz2 | |
tcp: don't clear retransmit_skb_hint when not necessary
Most importantly avoid doing it with cumulative ACK. Not clearing
means that we no longer need n^2 processing in resolution of each
fast recovery.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/tcp_output.c')
| -rw-r--r-- | net/ipv4/tcp_output.c | 8 |
1 files changed, 5 insertions, 3 deletions
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index f900fae8b87e..239cea7b6c0e 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -750,7 +750,7 @@ int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len, BUG_ON(len > skb->len); - tcp_clear_all_retrans_hints(tp); + tcp_clear_retrans_hints_partial(tp); nsize = skb_headlen(skb) - len; if (nsize < 0) nsize = 0; @@ -1823,7 +1823,9 @@ static void tcp_retrans_try_collapse(struct sock *sk, struct sk_buff *skb, tp->packets_out -= tcp_skb_pcount(next_skb); /* changed transmit queue under us so clear hints */ - tcp_clear_all_retrans_hints(tp); + tcp_clear_retrans_hints_partial(tp); + if (next_skb == tp->retransmit_skb_hint) + tp->retransmit_skb_hint = skb; sk_wmem_free_skb(sk, next_skb); } @@ -1853,7 +1855,7 @@ void tcp_simple_retransmit(struct sock *sk) } } - tcp_clear_all_retrans_hints(tp); + tcp_clear_retrans_hints_partial(tp); if (prior_lost == tp->lost_out) return; |