All skb in socket write queue should be properly timestamped.
In case of FastOpen, we special case the SYN+DATA 'message' as we
queue in socket wrote queue the two fallback skbs:
1) SYN message by itself.
2) DATA segment by itself.
We should make sure these skbs have proper timestamps.
Add a WARN_ON_ONCE() to eventually catch future violations.
Fixes: 740b0f1841f6 ("tcp: switch rtt estimations to usec resolution")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
                        flag |= FLAG_RETRANS_DATA_ACKED;
                } else {
                        last_ackt = skb->skb_mstamp;
+                       WARN_ON_ONCE(last_ackt.v64 == 0);
                        if (!first_ackt.v64)
                                first_ackt = last_ackt;
 
 
        tcp_connect_queue_skb(sk, data);
        fo->copied = data->len;
 
+       /* syn_data is about to be sent, we need to take current time stamps
+        * for the packets that are in write queue : SYN packet and DATA
+        */
+       skb_mstamp_get(&syn->skb_mstamp);
+       data->skb_mstamp = syn->skb_mstamp;
+
        if (tcp_transmit_skb(sk, syn_data, 0, sk->sk_allocation) == 0) {
                tp->syn_data = (fo->copied > 0);
                NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPORIGDATASENT);