Re: permissive vs. restrictive issue

From: jmjonesat_private
Date: Mon Jun 04 2001 - 06:27:52 PDT

  • Next message: Stephen Smalley: "Re: permissive vs. restrictive issue and solutions..."

    Okay, I'm beginning to be overwhelmed with the options
    again, so let me summarize where we are (for my own 
    benefit) and see if I've got it right.
    
    A) There is clearly a desire to support restrictive-only,
       restrictive-permissive, and permissive-only (possibly) 
       modules with the LSM interface.
    
    B) Nearly everyone agrees that, in *some* context, the 
       pre-LSM kernel security logic must be preserved exactly as 
       is... somewhere.
    
    C) The main "bone of contention" seems to be where "somewhere" 
       is, and if it is to be "required" or "optional".  Some wish 
       to keep it in the kernel and add the module 
       results to the logic in the same manner we have it now.
       The problem with this approach is that you create a logical
       conundrum by having to decide if it's AND or OR, NAND or NOR
       when inserting the module results into the logic.  Because
       much of the logic supports the "dually" permissive/restrictive 
       capability_plug, this muddies the waters.
    
       Some think that dividing the hooks into two sets of hooks,
       one permissive, one restrictive, all in the kernel, is the 
       answer.  This complicates the interface, but, as Mr. Wagner
       points out, it doesn't have to make too much of a mess if 
       the implementation is clever.
    
       Some would rather move the existing logic out of the "kernel 
       proper" to a shared space.  The options here seem to be 
       a shared source code library, a stackable module, or 
       something-inbetween (flip-flop kernel logic?).
    
       I think the "leave it ALL up to each individual module" has 
       somewhat died in committee.
    
    D) Considerations relevant to the "where issue" are verification,
       acceptability and accessability to the kernel developers, and 
       points of philosophy.
    
    Chris Wright gave us a clear list of 3 possible solutions.  Can 
    somebody refine it, now, to a list of fewer-than-five options 
    that we can, perhaps, use to form a consensus and start moving 
    forward again?  Have we actually eliminated any of the original
    three, or just merged them back into a continuum of "some of one,
    some of another, a little of factor X"?
    
    I do better on multiple choice tests than essay questions.
    
    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 : Mon Jun 04 2001 - 06:28:32 PDT