Re: patch: policy update by id

[jamal <hadi@xxxxxxxxxx>]

On Wed, 2005-27-04 at 19:43 -0700, David S. Miller wrote:
> On Thu, 28 Apr 2005 12:07:54 +1000
> Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote:
> 
> > You know what, I actually agree with you :) But you'll need to convince
> > Dave:
> > 
> > http://www.uwsg.iu.edu/hypermail/linux/net/0305.3/0018.html
> 
> I'm willing to reneg on that position if you can convince me
> that security minded folks won't be surprised by this pseudo-
> aliasing.  For example, do firewall systems tend to support
> such priority schemes?  If so, I guess we can do it.

Well, the tc classifiers are a good example. Priorities are used 
as ambiguity resolvers. 

After reading that URL though i think either way would be fine ..

rule1:
reject ipsrc A/32 ipdst B/32 with different priorities if entered more
than once; 
** but we allow the second rule ipsrc A/24 ipdst B/24 - only thing would
probably be useful to add is ensure a different priority is used. This
may be a little involved.

BTW, a weird ambiguity resolver is iptables - it just prepends rules.

cheers,
jamal