Subject: Re: critical networking problem
To: Perry E. Metzger <perry@piermont.com>
From: David Brownlee <david@mono.org>
List: tech-net
Date: 03/29/1996 06:45:12
	Just tried this on
	NetBSD1.1/sparc (le0) connecting to IRIX

	(And as its probably pretty similar to NetBSD/1.1A)
	OpenBSD1.2/i386 (ep0) connecting to OpenBSD1.2/sparc

	Wil no problem. Also tried
	'dd if=/dev/zero of=kill2 count=5 bs=1024k'
	to make it use more data & no problems.
	Tried both with & without net.inet.tcp.rfc1323 (for light relief :)

                   David/abs             (david@mono.org)

+44 181 888 8949 System Manager, Southern Studios Ltd, Box 59, London N22 1AR
 >=- OpenBSD http://www.openbsd.org. Free un*x for {sparc,i386,m68k,...} -=<
       <<< Monochrome - Largest UK Internet BBS - telnet mono.org >>>


On Thu, 28 Mar 1996, Perry E. Metzger wrote:

>
> I have noticed that I can reproducably kill tcp connections by sending
> whole ethernet blocks of zeros down them. For example, if I try to
> upload via FTP the file created by the command
>
> dd if=/dev/zero of=kill2 count=5
>
> the connection dies, each and every time.
>
> This may have something to do with my ethernet driver (le0 on pci on
> i386) or it may be systemic -- I can't tell. I first noticed this bug
> a long time ago but it seemed to go away -- now its back. I've tried
> many, many different destination hosts and it makes no difference.
>
> Does this problem happen to anyone else? Could this be the cause of
> the dying sups people have from sup.netbsd.org?
>
> Perry
>