Re: Future: RSBAC and LSM

From: Amon Ott <ao@rsbac.org>
Subject: Re: Future: RSBAC and LSM
Date: Thu, 30 Aug 2001 09:45:41 +0200

Next Article (by Date): Re: Future: RSBAC and Unix sockets Amon Ott
Previous Article (by Date): Future: RSBAC and Unix sockets Stanislav Ievlev
Top of Thread: Future: RSBAC and LSM Stanislav Ievlev
Next in Thread: Re: Future: RSBAC and LSM James Morris
Articles sorted by: [Date] [Author] [Subject]

On Don, 30 Aug 2001 Stanislav Ievlev wrote:
> Hello All!
> I've just seen pre-patches for LSM. As I understand, RSBAC 1.1.2 will be 
> ported to LSM.
> 
> One question:
> RSBAC sometimes uses two ADF calls in syscalls: one for decision and one 
> for notification (e.g. in sys_unlink)
> But LSM already use only one LSM call.
> How to solve it?

It will only work, if after the single call the syscall will always succeed.

I am still not sure about LSM - it is very Linux specific, and anything without
further abstraction will be hard to port. This means that I would have to make
an abstraction module, which translates LSM hooks into RSBAC decision requests.

The problem is that LSM will probably be there in 2.6, and not using it can be
a disadvantage. Still, I would have preferred something more like a real
framework, not only a collection of hooks. Linus made clear that he would not
accept anything that might slow down the system more than a few cycles per
system call, so the current LSM design is probably the best which could be got
for the first version. Later, there might be extensions like notification calls
- which many model implementations rely on anyway.

Amon.
-
To unsubscribe from the rsbac list, send a mail to
majordomo@rsbac.org with
unsubscribe rsbac
as single line in the body.

Go to Compuniverse LWGate Home Page.