diff options
| author | Linus Torvalds <torvalds@g5.osdl.org> | 2006-10-20 19:27:38 +0200 |
|---|---|---|
| committer | Linus Torvalds <torvalds@g5.osdl.org> | 2006-10-20 19:27:38 +0200 |
| commit | c1448791648d44ecbf8cadb192704e3e6eb3bb0e (patch) | |
| tree | 606cb5289bc8770e925761c6e1617eb44a6dafef /net/ipv4/tcp_output.c | |
| parent | [PATCH] one more ARM IRQ fix (diff) | |
| parent | [Bluetooth] Fix HID disconnect NULL pointer dereference (diff) | |
| download | linux-c1448791648d44ecbf8cadb192704e3e6eb3bb0e.tar.xz linux-c1448791648d44ecbf8cadb192704e3e6eb3bb0e.zip | |
Merge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (36 commits)
[Bluetooth] Fix HID disconnect NULL pointer dereference
[Bluetooth] Add missing entry for Nokia DTL-4 PCMCIA card
[Bluetooth] Add support for newer ANYCOM USB dongles
[NET]: Can use __get_cpu_var() instead of per_cpu() in loopback driver.
[IPV4] inet_peer: Group together avl_left, avl_right, v4daddr to speedup lookups on some CPUS
[TCP]: One NET_INC_STATS() could be NET_INC_STATS_BH in tcp_v4_err()
[NETFILTER]: Missing check for CAP_NET_ADMIN in iptables compat layer
[NETPOLL]: initialize skb for UDP
[IPV6]: Fix route.c warnings when multiple tables are disabled.
[TG3]: Bump driver version and release date.
[TG3]: Add lower bound checks for tx ring size.
[TG3]: Fix set ring params tx ring size implementation
[NET]: reduce per cpu ram used for loopback stats
[IPv6] route: Fix prohibit and blackhole routing decision
[DECNET]: Fix input routing bug
[TCP]: Bound TSO defer time
[IPv4] fib: Remove unused fib_config members
[IPV6]: Always copy rt->u.dst.error when copying a rt6_info.
[IPV6]: Make IPV6_SUBTREES depend on IPV6_MULTIPLE_TABLES.
[IPV6]: Clean up BACKTRACK().
...
Diffstat (limited to 'net/ipv4/tcp_output.c')
| -rw-r--r-- | net/ipv4/tcp_output.c | 20 |
1 files changed, 15 insertions, 5 deletions
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index f22536e32cb1..ca406157724c 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -1096,10 +1096,14 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_ u32 send_win, cong_win, limit, in_flight; if (TCP_SKB_CB(skb)->flags & TCPCB_FLAG_FIN) - return 0; + goto send_now; if (icsk->icsk_ca_state != TCP_CA_Open) - return 0; + goto send_now; + + /* Defer for less than two clock ticks. */ + if (!tp->tso_deferred && ((jiffies<<1)>>1) - (tp->tso_deferred>>1) > 1) + goto send_now; in_flight = tcp_packets_in_flight(tp); @@ -1115,7 +1119,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_ /* If a full-sized TSO skb can be sent, do it. */ if (limit >= 65536) - return 0; + goto send_now; if (sysctl_tcp_tso_win_divisor) { u32 chunk = min(tp->snd_wnd, tp->snd_cwnd * tp->mss_cache); @@ -1125,7 +1129,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_ */ chunk /= sysctl_tcp_tso_win_divisor; if (limit >= chunk) - return 0; + goto send_now; } else { /* Different approach, try not to defer past a single * ACK. Receiver should ACK every other full sized @@ -1133,11 +1137,17 @@ static int tcp_tso_should_defer(struct sock *sk, struct tcp_sock *tp, struct sk_ * then send now. */ if (limit > tcp_max_burst(tp) * tp->mss_cache) - return 0; + goto send_now; } /* Ok, it looks like it is advisable to defer. */ + tp->tso_deferred = 1 | (jiffies<<1); + return 1; + +send_now: + tp->tso_deferred = 0; + return 0; } /* Create a new MTU probe if we are ready. |