tcp: fix tcp_rearm_rto()
authorEric Dumazet <edumazet@google.com>
Thu, 18 May 2017 16:15:58 +0000 (09:15 -0700)
committerDavid S. Miller <davem@davemloft.net>
Thu, 18 May 2017 17:20:31 +0000 (13:20 -0400)
skbs in (re)transmit queue no longer have a copy of jiffies
at the time of the transmit : skb->skb_mstamp is now in usec unit,
with no correlation to tcp_jiffies32.

We have to convert rto from jiffies to usec, compute a time difference
in usec, then convert the delta to HZ units.

Fixes: 9a568de4818d ("tcp: switch TCP TS option (RFC 7323) to 1ms clock")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_input.c

index 9a5a9e8eda899666501cca06b37948ab64ae79b2..aa1eef150dc470fe70816dd5c248a925d01df20e 100644 (file)
@@ -3002,14 +3002,14 @@ void tcp_rearm_rto(struct sock *sk)
                if (icsk->icsk_pending == ICSK_TIME_REO_TIMEOUT ||
                    icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) {
                        struct sk_buff *skb = tcp_write_queue_head(sk);
-                       const u32 rto_time_stamp =
-                               tcp_skb_timestamp(skb) + rto;
-                       s32 delta = (s32)(rto_time_stamp - tcp_jiffies32);
-                       /* delta may not be positive if the socket is locked
+                       u64 rto_time_stamp = skb->skb_mstamp +
+                                            jiffies_to_usecs(rto);
+                       s64 delta_us = rto_time_stamp - tp->tcp_mstamp;
+                       /* delta_us may not be positive if the socket is locked
                         * when the retrans timer fires and is rescheduled.
                         */
-                       if (delta > 0)
-                               rto = delta;
+                       if (delta_us > 0)
+                               rto = usecs_to_jiffies(delta_us);
                }
                inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, rto,
                                          TCP_RTO_MAX);