clarification regarding tcp_sequence() fn

[Madanlal Musuvathi <madan@xxxxxxxxxxxxxxx>]

Hi,
 In tcp_input.c the function tcp_sequence is defined as follows

static inline int tcp_sequence(struct tcp_opt *tp, u32 seq, u32 end_seq)
{
        return  !before(end_seq, tp->rcv_wup) &&
                !after(seq, tp->rcv_nxt + tcp_receive_window(tp));
}

I read this as (after inlining tcp_receive_window()): 
  A received tcp packet is in sequence if
     end_seq >= tp->rcv_wup &&
     seq <= tp->rcv_wup + tp->rcv_wnd  

However, this function accepts a packet that is just to the left of the
window (end_seq == tp->rcv_wup). Is this intentional?

If such a packet arrives, RFC requires that a duplicate acknowledgment be
sent.  The linux tcp (atleast 2.4.19 & the latest 2.5 version from
bitkeeper) would not acknowledge such packets in some cases. (Specifically
those cases where the conditions in __tcp_ack_snd_check() fail.)

madan