Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 8 Sep 1997 00:28:39 -0700 (PDT)
From:      "Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com>
To:        nate@mt.sri.com (Nate Williams)
Cc:        brian@awfulhak.org, freebsd-stable@freebsd.org
Subject:   Re: Don Croyle: make world failing at ppp install (again)
Message-ID:  <199709080728.AAA16253@GndRsh.aac.dev.com>
In-Reply-To: <199709080556.XAA18293@rocky.mt.sri.com> from Nate Williams at "Sep 7, 97 11:56:23 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
You can wave your hands all around about ease of use vs doing it
right, but the bottom line is as ppp stands today it is a security
hole, and security holes are bad karma.

Okay the group network cuts down the exposure, no you only have to deal
with a fist full of users who can bring your router down.

I simply fix most of the problem by rm'ing the user land ppp files,
use the kernel version, make sure I don't have any tun drivers, etc.

> > Running ppp does _NOT_ *requires* write access to the routing table,
> > this is much much much better handled by properly configuring
> > a real routing daemon and running real routing protocols.
> 
> Bzzt, thanks for playing, but for 99.9999999% of the folks who run a PPP
> connection, a 'real routing daemon' is way overkill and will cause them
> no-end of headaches.  

And for those 99.9999% of the folks /sbin/routed -q will do just what
they need.  Now was that so hard.  I didn't say the only real routing
daemon was gated, but for server side ppp boxes it's a lot more guttsy
than /sbin/routed.  If you have VLSM run routed in ripv2 mode.

> 
> > Infact I have to go to great pains to _stop_ what ppp tries to do to
> > the routing tables, gated handles it MUCH better!
> 
> Gated handles nothing better unless you've got a spare 40 hours to
> dedicate to figuring out how it works.  Gated is only necessary if
> you've got multiple 'routes', and most (see above) folks have a single
> network connection which is their PPP link.
> 
> Engineering is finding the best solution for most folks, optimizing it
> for it while trying to not penalize the rest of the folks.  What ijppp
> does is take the engineering approach, and not find the 'best/most
> complicated/gated' solution.

And leaves a big security hole....


-- 
Rod Grimes                                      rgrimes@gndrsh.aac.dev.com
Accurate Automation, Inc.                   Reliable computers for FreeBSD



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199709080728.AAA16253>