On Wed, Jan 22, 2003 at 03:39:49PM -0800, Adam Lofstedt wrote:
> I've tried to block all the non-routable addresses from my external
> interface in my ipf.rules.  My internal network is using 10.0.0.0
> scheme.  Since rdr rewrites the addresses before the rule set is
> checked, I can't block packets from 10.0.0.0 on my external interface.
> 
> Does anyone know if there is a security implication because of this?
> Right now I can block in quick 10.0.0.0 and block in quick 10.0.0.255 to
> avoid smurf attacks.  Anyone know if not being able to block other
> 10.0.0.0 addresses poses a threat?  Just curious...

Keep in mind that the traffic you'd want to block is in the opposite
direction.

Unauthorized inbound traffic would have a -source- of 10.x.x.x coming
in, whereas valid traffic would have a -destination- of 10.x.x.x.

Likewise, unauthorized outbound traffic would have a -destination- of
10.x.x.x while valid traffic would have a -source- of 10.x.x.x.

So, these rules should work:

block in quick on <external> from 10.0.0.0/8 to any
block out quick on <external> from any to 10.0.0.0/8

pass in quick on <external> from any to 10.0.0.0/8
pass out quick on <external> from 10.0.0.0/8 to any

Obviously you'll have more complicated rules, but point being the two
block rules above shouldn't interfere with valid traffic. The only
case where this might be a problem is using a bridging firewall on
OpenBSD. If I recall, you can't filter by direction in that case.

-c

Reply via email to