Re: Runt packets ??

Donald Becker (becker@cesdis1.gsfc.nasa.gov)
Fri, 11 Sep 1998 00:33:38 -0400 (EDT)


On Thu, 10 Sep 1998, Kris Carlier wrote:

> > Sep 9 11:38:21 picard kernel: ICMP: runt packet
> > Sep 9 11:38:41 picard kernel: ICMP: runt packet
> > Sep 9 11:41:21 picard kernel: ICMP: runt packet
> > Sep 9 11:41:37 picard kernel: ICMP: runt packet
> >
> > We have been seeing a lot of "odd" reverse ARP requests on the segments
> > recently - this may indicate something ?
> >
>
> runts or giants might result from late collisions. Is your network within
> the standards as far as distances are concerned ?

True runts should be rejected by the length filter on the chip.
Corrupted packets caused by late collisions should fail the CRC.

The only device vulnerability is slightly oversized frames on some chips.
Many of my drivers provide 1536 (or 1536 - delta) byte buffers for receive
packets, expecting that the chip will enforce the 1514/1518 byte limit if it
wants to.

Unlike the minimum size, there is no physical limitation behind the Ethernet
maximum packet length. The reason behind the 1514 byte limit (1518 with
CRC, 1526 with pramble) was quick network access delay at 3Mbps, and that
1536 is an easy number for the hardware jabber detect.

Donald Becker becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center, Greenbelt, MD. 20771
301-286-0882 http://cesdis.gsfc.nasa.gov/people/becker/whoiam.html

-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.rutgers.edu