That reminds me: there's a tiny buglet in net/ipv4/icmp.c of 1.2.13
which seems to have been fixed in 1.3.69 by commenting it out. If it
ever gets uncommented rather than deleted, it may as well be fixed:
Lines 290-292 in 1.2.13 (unindented) are
case ICMP_PROT_UNREACH:
printk(KERN_INFO "ICMP: %s:%d: protocol unreachable.\n",
in_ntoa(iph->daddr), ntohs(iph->protocol));
(1.3.69 has the same at lines 359-361 but with the printk line commented
out). iph->protocol is an unsigned char not a short so the ntohs() makes
the printk log a bogus number.
--Malcolm
-- Malcolm Beattie <mbeattie@sable.ox.ac.uk> Unix Systems Programmer Oxford University Computing Services