Re: compartment || the other way of compartment

From: karin@root66.org
Date: Thu Jan 06 2000 - 12:49:37 EST


On Thu, Jan 06, 2000 at 09:53:38AM -0600, Jesse Pollard wrote:
>
> You are actually talking about using compartmented mode operation and
> mandatory access controls. I do believe that more complete capability
> list support will be usefull. Tracking and configuration of capability lists
> has been brought up several times without final resolution, since all
> implementations seem to need modifications to the filesystem for support.
> Some suggestions that came up:
> 1. modify the binary header to contain the capabilities of the binary.
> problem: anyone that can create a binary would be able to set the
> capability list of that binary. How is this prevented.
> 2. only inherit capabilities from a parent process.
> problem: how are the capabilities initially set? how can a non-
> privilege process (say a user shell) run a privileged
> application (say passwd)?

You could also just have a config file like /etc/auditopen containing the capability list of the binaries.
What is wrong with that?

> Other topics that came up:
>
> 1. what is a suitable list of capabilities?

You could say that the max list of capabilities is the complete list of sys_calls.

> 2. What granularity is supported by the capabilities?

What do you mean?

> 3. Can site defined capabilities be defined?
> 4. (related to 3) Can application specific capabilities be defined?

You could think of extending the configfile from a binary specific to a binary and process specific. But as far as I can see binary specific will to enough. Could you give me an example where binary specific configfiles are not enough?

>
> See the RSBAC project at http://www.rsbac.de/rsbac/
> for an implementation of MAC controls, including compartmented operation.
> -------------------------------------------------------------------------
> Jesse I Pollard, II
> Email: pollard@navo.hpc.mil
>
> Any opinions expressed are solely my own.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/

        Frank van Vliet
                karin@root66.org
                        RooT66 - http://root66.org
                        ShellOracle - http://www.shelloracle.cjb.com

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Fri Jan 07 2000 - 21:00:06 EST