[3054] in linux-net channel archive
Re: Linux TCP/IP wedges talking to DG AViiON
daemon@ATHENA.MIT.EDU (Alan Cox)
Wed May 29 14:59:53 1996
From: alan@lxorguk.ukuu.org.uk (Alan Cox)
To: ktk@ktk.bih.harvard.edu (Kris Karas)
Date: Tue, 28 May 1996 23:50:43 +0100 (BST)
Cc: alan@lxorguk.ukuu.org.uk, linux-net@vger.rutgers.edu
In-Reply-To: <199605281913.PAA01486@ktk.bih.harvard.edu> from "Kris Karas" at May 28, 96 03:13:41 pm
> 14:16:40.703796 AViiON.smtp > Linux.1105: . ack 7788 win 511
Ok so far.
> 14:16:40.903796 Linux.1105 > AViiON.smtp: P 7788:8299(511) ack 210 win 29696
We send the full window
> 14:16:40.913796 AViiON.smtp > Linux.1105: . ack 8299 win 963
It acks the block.
> 14:16:40.913796 Linux.1105 > AViiON.smtp: P 7788:8556(768) ack 210 win 29696
We send the full block now that the window has opened enough. Because of the
way Linux tcp works for performance reasons mainly we will in these odd window
cases send the full version of an existing frame we sent partially before. This
is 100% legal.
> 14:16:40.913796 AViiON.smtp > Linux.1105: . ack 8299 win 963
The aviion fails to ack it. Its thrown the bigger frame away which it should
not have done.
> 14:16:41.103796 Linux.1105 > AViiON.smtp: P 7788:8556(768) ack 210 win 29696
We send the block again
> 14:16:41.113796 AViiON.smtp > Linux.1105: . ack 8299 win 963
The aviion fails to ack it again. This looks very wrong behaviour.
> ..... ad infinitum.
>
> Any suggestions/help would be gratefully appreciated.
I think you need to refer that tcpdump to AViiON. What Linux does is a little
unusual but it is correct. If someone in AViiON wants a Linux side contact
give them my address.
Alan
