[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
A place to start
Hi,
A simple proposal : Total Bastard Linux (TBL)
TBL is designed to provide a higher barrier to programs that run on top
of linux as well as more of the kernel itself. It is a new set of
kernel patches based upon including some of the well defined patches
(openwall, stackguard) and the rest of them, along with some other
tightening fixes.
The goal is to make a version of the Linux kernel that is intolerant of
bad programs and bad modules. It will of course run a subset of the
programs available for linux.
An example of what I'm referring to is the development of libsafe.
Many of the distributions include but do not make libsafe part of the
default installation because it will stop some programs from working.
However TBL, is designed more to say tough, fix it.
Years ago, I had to port a large program from SunOS 4.1.1 on a sparc to
SCO unix running on an intel box. Unfortunately we had taken advantage
of a variety of nice features in the way SunOS did memory management
like padding to word boundaries allowing off by 1 errors in a variety of
sections. We eventually ended up writing a malloc replacement to
replicate the behavior on SunOS to raise the bar on the SunOS box to the
same level as the SCO box.
TBL is designed to set a higher bar. If a programmer can get his/her
module to run with TBL it is more than likely going to run on other
versions of Linux and will be more secure since it has had to comply
with a tighter environment. TBL may also have advantages to application
writers how can take advantage of a stricter environment.
TBL is not designed to replace the hand audit process that the kernel
audit project is going to pursue. It is intended to help find those
areas that should be audited first. It will also help provide a tool to
help make sure that our changes do not introduce new unexpected side
effects.
I think TBL could be a good first project for the audit project and
would provide us with the possibility of gathering some more information
on the Linux kernel base. If there is enough interest we could probably
put together a reasonable set of patches against the standard kernel.
What do you think?
Aaron Grothe
==========
"The Journey is the Reward" - Old Zen Buddhist Saying
Kernel-audit: discussion list for security and the linux kernel
Archive: http://mail.nl.linux.org/kernel-audit/