[rsbac] Re: How does RSBAC relate to the Linux Security Module framework (LSM)?

[Peter Busser]

Hello,

> could you perhaps also give me your personal view on how things will 
> develop in the future, e.g. probably several of the security enhancement 
> projects for linux will merge or at least use a common infrastructure 
> and try to find some common standards so that these enhancements can go 
> into the mainstream linux distributions like debian, redhat, suse, ... 
> and perhaps even the linux standard base project could integrate such a 
> common infrastructure?

I doubt this will ever happen. Because there has already been a lot of
discussion before LSM went into the kernel. The problem is that for the Linux
kernel people, speed is more important than security.

If kernel related security projects would like to standardise on one common
infrastructure, then using the RSBAC framework would be a good starting point.
Many people look at RSBAC as a set of models, but what makes RSBAC really
unique is in fact its framework.

You could use the same RSBAC framework to implement SELinux. Or the gr-security
ACLs if you want. I would say that, from a security point of view, the RSBAC
framework is what LSM should have been like.

The RSBAC framework could be ported to other *NIX operating systems (like e.g.
BSD) and that would also make it possible to share the security modules between
the different *NIX platforms. Because the RSBAC framework already provides all
the groundwork and plumbing, it also helps developers to create new security
solutions more easilly.

> i am also a bit surprised that a project like SELinux, which seems to 
> have a similar objective as RSBAC, started after RSBAC was already a 
> stable system running on different platforms? only from what i've read 
> so far SELinux seems not to add any significant advantages over RSBAC, 
> or does it in some areas?

There is in fact parallel evolution going on here. The framework provided by
SELinux was first implemented on Flask, some kind of microkernel. And then it
was converted to Linux.

I don't think the objectives are the same. You can compare SELinux to a games
console and RSBAC to a PC. SELinux is designed to provide only one security
solution. RSBAC is designed to support MANY security solutions. And you can
even activate them at the same time.

> at the moment i am just a bit irritated by the numberous projects having 
> similar or overlapping objectives. is there perhaps somewhere a good 
> overview that sets all the relevant projects into relation, e.g. how 
> they could be combined or how they complement each other (perhaps even 
> with a bit of historic insight)?

Well, this diversity is in fact a good and healthy thing. It means people have
a choice and developers have to do their best to provide better solutions. It
would help if the different projects standardised on one security
infrastructure, so users could simply compile a kernel with the framework and
then load the security solution of their choice as kernel module. If they don't
like it, they can combine it with other solutions. Or simply throw it out.

Groetjes,
Peter Busser
-- 
The Adamantix Project
Taking trustworthy software out of the labs, and into the real world
http://www.adamantix.org/