Pedro Roque Marques <roque@di.fc.ul.pt> writes:
> Eric> Looking over the code I expect this is over zealous checking
> Eric> of potential bug conditions on my part. You can get the
> Eric> above message if TCP is doing a zero window probe when it
> Eric> gets the FIN from the far end. I've revised the bug check to
> Eric> actually look for the error condition I'm worried about
> Eric> rather than the short cut check I was using.
>
>That would be mostly unusual.
>I believe that what is happening is that the ack is being processed
>before tcp_fin in tcp_ack and removes the fin segment out of the
>queue. Pretty normal situation.
I don't think this is what is happening. If the ack and the fin
arrive in the same packet and we are in FIN_WAIT1 then it seems to me
that the code in tcp_ack should move us into FIN_WAIT2 before we
get to the code in tcp_ack.
Also, given how common this case is, I would have expected to see
a flood of such messages about pre2.0.14, and I have not.
Cheers,
-- eric