nagai/kernel
3
0
Fork 0
forked from mirrors/linux
Code Pull requests Activity

tcp: align tcp_xmit_size_goal() on tcp_tso_autosize()

Browse source 
With some mss values, it is possible tcp_xmit_size_goal() puts
one segment more in TSO packet than tcp_tso_autosize().

We send then one TSO packet followed by one single MSS.

It is not a serious bug, but we can do slightly better, especially
for drivers using netif_set_gso_max_size() to lower gso_max_size.

Using same formula avoids these corner cases and makes
tcp_xmit_size_goal() a bit faster.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Fixes: 605ad7f184 ("tcp: refine TSO autosizing")
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Eric Dumazet 2015-03-05 08:03:06 -08:00 committed by David S. Miller
parent e9647d1e74
commit 6c09fa09d4
1 changed files with 3 additions and 7 deletions
repo.diff.show_diff_stats Download patch file Download diff file Expand all files Collapse all files

10
net/ipv4/tcp.c
View file

@ -835,17 +835,13 @@ static unsigned int tcp_xmit_size_goal(struct sock *sk, u32 mss_now,
int large_allowed) int large_allowed)
{ {
struct tcp_sock *tp = tcp_sk(sk); struct tcp_sock *tp = tcp_sk(sk);
u32 new_size_goal, size_goal, hlen; u32 new_size_goal, size_goal;
if (!large_allowed || !sk_can_gso(sk)) if (!large_allowed || !sk_can_gso(sk))
return mss_now; return mss_now;
/* Maybe we should/could use sk->sk_prot->max_header here ? */ /* Note : tcp_tso_autosize() will eventually split this later */
hlen = inet_csk(sk)->icsk_af_ops->net_header_len + new_size_goal = sk->sk_gso_max_size - 1 - MAX_TCP_HEADER;
inet_csk(sk)->icsk_ext_hdr_len +
tp->tcp_header_len;
new_size_goal = sk->sk_gso_max_size - 1 - hlen;
new_size_goal = tcp_bound_to_half_wnd(tp, new_size_goal); new_size_goal = tcp_bound_to_half_wnd(tp, new_size_goal);
/* We try hard to avoid divides here */ /* We try hard to avoid divides here */