[cap-talk] SELinux vs. capabilities

[Rob Meijer]

On Sun, July 29, 2007 17:49, Jonathan S. Shapiro wrote:

>> I took Jonathan to have meant that one could re-create SELinux (or a
>> system with equivalent security properties) out of capability
>> privmities, which wouldn't necessarily be available to anything
>> running under SELinux.  In other words, the SELinux implementation
>> wouldn't expose or pass any capabilities to the software running on it
>> that just sees an SELinux execution environment.
>
> Yes. This is what I had in mind. I think that the contained subsystems
> can be given restricted capabilities, but yes, I was thinking to build
> "on top".

Would it not be better than to first look at the possible added value of
combining the two paradigms, rather than looking if implementing one in
terms of the other, only for the sake of showing that it can be done, what
to me seems like just an other academic exercise without any real
practical use.
Only if by doing so, we would arrive at a point where features of both
paradigms can also be combined in ways useful to the user would the
exercise  be worth the trouble IMHO.

Rob