Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 17 Sep 1999 14:04:10 -0600
From:      Warner Losh <imp@village.org>
To:        Brett Glass <brett@lariat.org>
Cc:        Liam Slusser <liam@tiora.net>, Kenny Drobnack <kdrobnac@mission.mvnc.edu>, "Harry M. Leitzell" <Harry_M_Leitzell@cmu.edu>, security@FreeBSD.ORG
Subject:   Re: BPF on in 3.3-RC GENERIC kernel 
Message-ID:  <199909172004.OAA04763@harmony.village.org>
In-Reply-To: Your message of "Thu, 16 Sep 1999 18:54:24 MDT." <4.2.0.58.19990916185341.00aaf100@localhost> 
References:  <4.2.0.58.19990916185341.00aaf100@localhost>  <Pine.GSO.3.96.990916150427.5757E-100000@mission.mvnc.edu> 

next in thread | previous in thread | raw e-mail | index | archive | help
In message <4.2.0.58.19990916185341.00aaf100@localhost> Brett Glass writes:
: At 04:14 PM 9/16/99 -0700, Liam Slusser wrote:
: 
: >Right...but if the system was hacked what would stop the hacker from
: >building BPF in a kernel?
: 
: securelevel=2 or securelevel=3.
: 
: Or Tripwire.

Dreamer.  This is covered in the archives.  Unless every file that
root could run before secure level is raised, including any
dynamically linked in libraries is set immutable, you loose.  However,
there is a hole the size of a truck right now in that anything that is
listed in the rc.d directories gets run before the secure level is
changed.  Did you remember to list them all in the list of files that
have schg tagged on them?  Also, some programs can be run based on the
presense or absense of files in /etc, which cannot be marked schg or
users will be unable to change their passwords, finger information,
etc.  If even one of these items is missed, I can force a reboot after
putting my trojan into that file, load the module and then put the
file back to how it was before in an environment that will give me a
good chance of restoring the damage before Tripwire can run.  Also,
most of the commands in /etc/rc do not include complete paths, so path
games are possible.  Did I forget anything?  Likely.  This was a 5
minute glance at the source.  What would a couple of hours have
revealed?

If you can fix all of these problems trivially, then you might have an
arguement.  As it is, it takes a hell of a lot of work to keep root
from running completely arbitrary commands on boot.  The pain of
recompiling the kernel is minor in comparison.  The pain of not having
DHCP is much worse...

Warner


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-security" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909172004.OAA04763>