8808-type Packets after Unrelated Crash

Pedro Ribeiro pribeiro@isel.pt
Tue Feb 2 16:48:22 1999

Ooops as far as a can remember from a document read some days ago that is
something related to new frame format 802.1p/802.1q for VLANS ...

  Pedro Ribeiro
  Online: http://www.isel.pt/~pribeiro/
  IRC(PTnet) Nick: PAntMaR
  e-Mail: Personal:  pribeiro@isel.pt
          Admin:     admin@isel.pt

-----Original Message-----
From: Stephen Wynne <stevemw@mindspring.com>
To: linux-eepro100@beowulf.gsfc.nasa.gov
Date: Tue Feb  2 16:48:22 1999
Subject: 8808-type Packets after Unrelated Crash

>First, I'd like to thank the developers for this driver and say that
>it normally does an excellent job of keeping my system on the network.
>Although it initially wasn't the eepro100 driver to blame, I recently
>took down our entire network at work (involving a Cisco router) when
>my system crashed and began spewing type ethernet type "8808" packets.
>Whatever the initial rate, it settles down to about one per second,
>but it's enough to keep the router very confused. I don't know much
>more than that since I don't manage the network gear where I work.
>The crash was caused (and the details are unclear to me) when part of
>an unrelated kernel module we're developing dereferenced a null
>function pointer. I have no idea what happened right before that
>or afterward. I can reproduce the problem, however.
>Does anyone have any tips for learning more about how to debug this
>sort of thing? I was reduced to running some cheap DOS-based software
>for sniffing Ethernet packets as I worked to try to isolate the
>problem, but it was still ineffectual in helping to determine what the
>kernel was actually doing (since my whole system was locked up).
>I know where to go in the kernel to see what happened approximately up
>until the crash occurred, but I don't know how to actually debug right
>on through it. Moreover, the "oops" we managed to capture wasn't very
>helpful (lots of stuff was missing, including the EIP).
>Any tips would be appreciated, although this is mostly an academic
>exercise at this point. Again, I'm looking for answers to these
>1. How to better debug these kinds of Linux kernel faults?
>2. What could cause an eepro100 on a 10mbit network to start spewing
>   '8808' type packets?
>3. Are the eepro100 developers interested in adding some resilience
>   in this kind of situation? (If it's even possible.)
>Because I feel there's so much for me to learn in this problem, I'd
>be happy for the opportunity to work with you on this, if it seems
>at all productive.
>Thank you,