[TCP]: Don't allow FRTO to take place while MTU is being probed
authorIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Tue, 8 Apr 2008 05:33:57 +0000 (22:33 -0700)
committerDavid S. Miller <davem@davemloft.net>
Tue, 8 Apr 2008 05:33:57 +0000 (22:33 -0700)
MTU probe can cause some remedies for FRTO because the normal
packet ordering may be violated allowing FRTO to make a wrong
decision (it might not be that serious threat for anything
though). Thus it's safer to not run FRTO while MTU probe is
underway.

It seems that the basic FRTO variant should also look for an
skb at probe_seq.start to check if that's retransmitted one
but I didn't implement it now (plain seqno in window check
isn't robust against wraparounds).

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_input.c

index b4812c3cbbcff574e8c001345b83f69b83ecf7d7..5119856017ab6274257331dcb458f0c65d6d9df0 100644 (file)
@@ -1691,11 +1691,16 @@ static inline void tcp_reset_reno_sack(struct tcp_sock *tp)
 int tcp_use_frto(struct sock *sk)
 {
        const struct tcp_sock *tp = tcp_sk(sk);
+       const struct inet_connection_sock *icsk = inet_csk(sk);
        struct sk_buff *skb;
 
        if (!sysctl_tcp_frto)
                return 0;
 
+       /* MTU probe and F-RTO won't really play nicely along currently */
+       if (icsk->icsk_mtup.probe_size)
+               return 0;
+
        if (IsSackFrto())
                return 1;