Re: Kernel Security Extensions USENIX BOF Summary

From: jmjonesat_private
Date: Tue Jul 03 2001 - 13:41:22 PDT

  • Next message: Greg KH: "Re: attach_pathlabel"

    A bit long, but I'm trying...
    
    On Tue, 3 Jul 2001, Stephen Smalley wrote:
    
    > The basic idea was to provide a module and utility that could be used
    > to verify that for each system operation, the right set of hooks was
    > called and the kernel responded properly when those hooks returned
    > errors.  Otherwise, it would be very easy for a hook call to become
    > "lost" or bypassed when a set of updates were made to the kernel.
    > I think the idea was proposed by Steve Kramer of HP.
    
    Ouch!  Big Job.  Mighty hard diggin'!
    
    The idea of "responded properly", it seems to me, would be module 
    specific... appropriate responses would depend on the policy in 
    force.
    
    What I CAN do that might be generally useful is design a module that
    has an API that an application which exercises the "desirable" functions
    of the kernel can use to capture the results.  Such an API would be 
    called before an exercise to register a "key", then the application would
    make calls to a specific syscall, and the responses would be recorded
    and keyed on the "registered" object.  After the exercise, the app would
    read the results and process them however is important to the module 
    under test.
    
    The module under test would be subordinated to the test module.
    
    I might also be able to build a "tester application" that is somewhat
    generally useful, even with a simple scripting language to "tune" it, but,
    again, the specific results must be evaluated against the policy/module 
    in force.  I may be too feeble-minded to see a totally general solution,
    but I don't think "proper response" is something that can be defined at
    this (or maybe any) point.
    
    Would this be useful?  If I get several replies in the affirmative, I'll
    put it on the agenda.  It would be useful to our project, as
    well, I think.  Otherwise, I request modifications to my basic plan
    be discussed so we can synthesize a better response, or arguments why it's 
    not feasible at all.
    
    Testing LSMEXAMPLE v0.4 against lsm-2001_06_20... now, will be online 
    tonight.
    
    Sincerely,
    J. Melvin Jones
    
    |>------------------------------------------------------
    ||  J. MELVIN JONES            jmjonesat_private 
    |>------------------------------------------------------
    ||  Microcomputer Systems Consultant  
    ||  Software Developer
    ||  Web Site Design, Hosting, and Administration
    ||  Network and Systems Administration
    |>------------------------------------------------------
    ||  http://www.jmjones.com/
    |>------------------------------------------------------
    
    
    
    
    _______________________________________________
    linux-security-module mailing list
    linux-security-moduleat_private
    http://mail.wirex.com/mailman/listinfo/linux-security-module
    



    This archive was generated by hypermail 2b30 : Tue Jul 03 2001 - 13:42:37 PDT