On Sat, 20 Jul 2002, Lars Eggert wrote: > Jun-ichiro itojun Hagino wrote: > > I looked through RFC826 and it seems that the operation performed by > > Lars was a Bad Thing. RFC826 input processing explicitly suggests us > > to update ARP cache entry without checking arp operation type. > > > > therefore, it is unsafe to transmit ARP_REQUEST with spoofed IP > > source address - it will overwrite ARP entries of neighbors. > > I re-read it, too, and you are of course right. > > It's sad though how easy a DoS attack can be - as easy as mistyping the > IP address of your machine. > > It might be worthwhile to investigate if 826 should be updated. Someone > at IETF mentioned that Linux explicitly violates 826 and does not update > the local cache based on the contents of spoofed packets, and thus seems > more resilient than the BSDs (against this particular bug). How does one tell, in principle, that the source IP address (ar$spa) in an ARP packet is in fact spoofed? //cmh