> From: Paul Kranenburg <pk@cs.few.eur.nl>
> Date: Tue, 9 Sep 1997 22:44:11 +0200 (MET DST)

> Since a couple of days the kernel logs messages when 0.0.0.0 is detected
> in a IP packet's source IP field:
> 
> 	arp: zero IP addr from link address ...
> 
> Since this is quite normal behaviour for BOOTP/DHCP clients when they
> first broadcast their discovery packets, I consider this logging (to the
> console, per the default syslog.conf) as rather useless noise.
> 
> I suggest that at least the check be conjuncted with a test for a broadcast
> destination address.
> 
> -pk

Oh yeah, I meant to ask about this.  I never had a chance to analyze
the traffic that was causing these messages.  Are there indeed valid
ARP requests that can have zero for either of the protocol addresses?
Can anyone explain why we get such packets?  Are they valid?

I also noticed that the RARP input routine will forward some kinds of
RARP packets to the ARP input routine, so perhaps these might explain
the "arp: zero IP addr" messages.

Gordon