Re: receiving ack in SYN_SENT?

Alan Cox (alan@lxorguk.ukuu.org.uk)
Thu, 29 Oct 1998 18:40:55 +0000 (GMT)

> 0.618439 me.3725 > q9.yahoo.com.http: S 2542078338:2542078338(0)
> win 32120 <mss 1460,sackOK,timestamp 119740339[|tcp]> (DF)
> 0.728439 q9.yahoo.com.http > me.3725: . ack 2542078339 win 17520 (DF)
> 0.728439 me.3725 > q9.yahoo.com.http: R 2542078339:2542078339(0) win 0
>
> v1 of TCP/IP Illustrated says basically to send RST whenever
> an inappopriate packet is received. here, we're expecting a SYN
> or SYN,ACK, but get a bare ACK. is this a known bug in some TCP
> stacks? or just a way of signalling busy? seems like resending
> the SYN would be a nicer way to handle this...

It indicates the other end is either broken, or that it thought there was
an established session. We do actually want to send an RST since if this
wasnt a bug (you should pester the yahoo folk about that trace I guess)
then it indicates somehow one side thinks we have a session going on this
port and the other does not. An RST helps tear that half dangling connection
down.

Its also not a known bug. Running an analyser at the site it tells me the
site doesnt talk valid HTTP/1.0 either as HEAD / HTTP/1.0 returns the whole
page.

Alan

-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.rutgers.edu