[comp.protocols.tcp-ip] Loose and Strict source routing

cpw%sneezy@LANL.GOV (C. Philip Wood) (01/31/89)

RFC 1009 states that it is important for gateways to implement both the
Loose and Strict Source Route IP options (Page 13).  Why?

Hacker's in the know have indicated that, with those tools, they can
bypass IP address checks which various hosts have set up to attempt a
source host access check.

Is it worth considering not honoring these options in a production
environment such as MILNET and ARPANET?   Or, could Hosts be required
to have an operating system option which allows an administrator to
signal that these packets be dropped?  What other ways are there to
defend against this kind of internet spoofing?

Are there other ways to spoof a source address in an Internet
environment besides LSRR and SSRR?

Phil Wood, cpw@lanl.gov

CERF@A.ISI.EDU (02/01/89)

Phil,

Unless I'm badly mistaken, there isn't any guarantee that
a non-source-routed internet packet has a valid source
address. Of course, responses to such a spoofed packet may
not make it back to the origin unless a cooperating gateway
helps out, or the source is on an Ethernet and is operating
in promiscuous mode. I suggest that, if source authentication
is an issue, you will need stronger tools/mechanisms than
avoiding the use of source routing of either type.

The general problem of authentication in the Internet is
very important, applies to many areas including, for example,
various control methods (e.g. network management subsystems)
and will probably require some form of cryptographic protection
to solve. The cryptography need not be used to conceal
information - merely to provide an unforgeable authentication
of the source.

Vint Cerf