Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!tuvie!nestroy!usenet
From: mah@dec1.wu-wien.ac.at (Michael Haberler)
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: NE1000 PD problem found, WATCH FOR BAD CARD ADRESSES!
Message-ID: <1991Mar07.180439.5382@nestroy.wu-wien.ac.at>
Date: 7 Mar 91 18:04:39 GMT
References: <1991Mar06.004805.5386@nestroy.wu-wien.ac.at> <1991Mar06.144438.21803@nestroy.wu-wien.ac.at>
Sender: usenet@nestroy.wu-wien.ac.at (The Usenet User)
Reply-To: mah@nestroy.wu-wien.ac.at
Organization: Wirtschaftsuniversitaet Wien
Lines: 32

I found the solution to the problem with the NE1000 packet driver.

Actually its not a problem with the software *at all*.

This trace shows it. Well, I didnt see it right away, either:

Wed Mar 06 12:49:15 1991  - ec0 recv:
Ether: len 60 08:00:2b:14:18:f6->47:54:43:00:05:9b type IP
IP: len 32 137.208.1.5->137.208.1.128 ihl 20 ttl 255 prot ICMP
ICMP: type Echo Reply id 65535 seq 0


If you look at the ethernet destination address, you'll find 
47:54:43 as the manufacturer code, which is hex for the string
'GTC'. Unfortunately, the first bit of 'G' is 1, so to KA9Q looks
like a multicast or broadcast adress (can somebody enlighten me on
this bit's meaning? I just inferred that from a list of ethernet
type fields, multicast adresses and manufacturer number culled
from the net).

The card `manual' reads 'Local Area Network GT-1000 Ethernet Card,
Users Manual'; no further hints about the company are given, 
so watch out.

Those turkeys dont have a registered manufacturer number, and 
dont know about its meaning. I could imagine that several
other  software packages break with such a card.

I think for me that means an option to the attach command to set
the hardware adress, or mask it to a sane value :-(

- michael