rtl8139 problem

[Santiago Garcia Mantinan]

  This message is in MIME format.  The first part should be readable text,
  while the remaining parts are likely unreadable without MIME-aware tools.
  Send mail to mime@docserver.cac.washington.edu for more info.

---1463811839-1771354263-949763780=:435
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Content-ID: <Pine.LNX.3.96.1000213131055.318C@mantianito.arroutada.es>

Hi!

It is long time since I found this problem, wich I believe I can now
replicate. I have written about this to Donald but I think he must be
too busy to take a look at it, so I'd like to see if anyone on the list
can confirmate this.

The problem was found on a Linux bridge with rtl8139 based cards, where it
causes that windows clients conected to the bridge loose packages.

Now I have found what I think is the problem, and it is not related to any
bridge or anything, so it is easy to see on any network.

The thing is that you do, for example a ping on a network and try to see
its length with for example a tcpdump -e, well, the hosts with rtl8129
cards will see a legth 4 bytes greater than the hosts with other cards, on
the ping example, the correct length is 98, while the tcpdump on the
rtl8129 shows 102. This happens with all the packages, and wether the
rtl8139 card is on promiscuous mode or not, and wether the package was for
it, or from it or not.

This is what is causing the problems on the bridge when packages are near
the mtu limit, but it can cause many other problems, so let's see if any
of you can confirmate this so it can be fixed.

Thanks in advance.

Regards...

Manty/BestiaTester -> http://www.mantianito.ddns.org

---1463811839-1771354263-949763780=:435--
 | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
 |  body of the mail, include only the text:
 |   unsubscribe this-list-name youraddress@wherever.org
 | You will be unsubscribed as speedily as possible.