Subject: Re: bridge(4) and silent data corruption :-(
To: Michael Richardson <mcr@sandelman.ottawa.on.ca>
From: Sean Doran <smd@ab.use.net>
List: tech-net
Date: 05/03/2002 15:29:22
Michael Richardson <mcr@sandelman.ottawa.on.ca> writes:

>   There are lots of lines without "tcp sum ok" on them!
>   They all are from world.22 > state-a.

Right, I expect that from tcpdump.
You get [bad tcp cksum %x!] on inbound traffic that
doesn't sum right.

>   I also noticed that the tos^H^H^Hdscp byte varries a lot. 0x28, 0x20, 0x8, 0x0.
>   The ECN bits are never set (lower 2, right?). 

The varying TOS is expected behaviour from ssh/scp, which
twiddles things like IPTOS_THROUGHPUT, IPTOS_LOWDELAY and suchlike.

IOW, I don't see anything odd in the trace at all, and
that's at the root (or route) of the problem...

In CPH we tried to set up a bridge with the latest -alpha
snapshot but have run into userland woes (in particular,
no brconfig...).  I haven't looked into this personally
yet though.

     Sean.