>Thats what concerned me: clearly the changes to sys_init were not
>tested on a v4-only host. That suggests we're not in the habit of
>testing all IPv6-feature changes in an IPv4-only kernel.
>I think that needs to change.

	in most cases (like telnetd or telnet) we implement address family
	independence for IPv4/v6 support.  All the binaries SHOULD work for
	both cases.  pppd was an exception.  I agree I should have tested
	it carefully.

itojun