LSM (Was: Re: [rsbac] Version 1.2.2 uploaded)

[Peter Busser]

Hi!

> |> I will try to use and translate LSM hooks,
> That was where I was going next.  I think hooking into LSM is a good idea;
> you're using them for what they were meant for, and it's a good way of
> keeping things more grounded in a formalized method of adding a security
> infrastructure.  If the LSM folks keep things straight, then its a win-win
> situation for everyone (at least, in the long term).

Well, it remains to be seen wether it is really a ``win-win situation''. It may
be the case that RSBAC needs more than LSM provides, which means you still have
to patch the kernel.

The last time I downloaded the 2.4 LSM patch, it was more than 2 (more like
2.5) times bigger than RSBAC's kernel patch. It patched more than twice as
many source files in the kernel tree. And it provides only hooks, nothing else,
no functionality whatsoever, whereas the RSBAC kernel patch does some useful
things in a few places.

> |> where suitable, and only patch in 
> |> my own hooks where necessary. If 2.6.0-test takes as long as 2.4.0-test, I 
> |> will probably be ready before 2.6.0 comes out.
> 
> I would guess that it won't take as long as 2.4, simply because of the LSM
> base.  So, unless LSM is poorly documented, it should be smooth sailing.

LSM is not the end-all to security stuff, the objects provided by the Linux
kernel through LSM still have to be translated to the abstractions needed by
the RSBAC framework.

It also remains to be seen how portable LSM will be between kernel versions,
since it depends on internal kernel structures which can change between
versions.

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