Re: module's use of security_ops

From: jmjonesat_private
Date: Fri Jun 08 2001 - 11:26:36 PDT

  • Next message: sarnoldat_private: "Re: module's use of security_ops"

    On Fri, 8 Jun 2001 sarnoldat_private wrote:
    
    > On Fri, Jun 08, 2001 at 10:57:48AM -0400, jmjonesat_private wrote:
    > >    a global variable KERNELSECURITY in the "manner" of errno without
    > >    passing anything?
    > 
    > Absolutely not. :)
    > 
    > Think "SMP" and I think you will agree that this global value is
    > probably not a wise move. :) (The problem continues to exist even on
    > monoproc machines, but not as obviously broken.)
    
    Acknowleged.  I can't get SMP into my brain for some reason, although, 
    there are "workarounds".
    
    > 
    > > 5)  For GODESSES'S SAKE, protect security_ops SOMEHOW, or this API is only 
    > >     e pluribus unum and vulnerable to "total replacement" from ANYWHERE
    > >     in kernel space. Provide a "non-functional copy", provide a module
    > >     check for revelation, SOMETHING.  The idea of a global export just 
    > >     seems to introduce too many new vulnerabilities.
    > 
    > I understand your concern. However, recall that this is the kernel
    > we are talking about; any code loaded into the kernel is completely
    > trusted. The assumption is made that kernel code is infallible and has
    > no malicious purpose.
    > 
    
    Since modules of any kind can be "insmod'd" into the kernel by the admin,
    "total trust" is a pretty liberal policy.
    
    
    > Not exporting the security_ops structure is only an obscurity defense
    > against code that is already assumed to do no harm. If there is kernel
    > code you don't trust, either #ifdef it out, or refuse to load the
    > module (if it comes in that form).
    > 
    
    Okay (sigh).  I think that assumption is unsupportable, but your
    conclusion follows if the assumption is true.
    
    How do I refuse a module loading without knowing if it needs
    security_ops, in a general sense?  Even minimally, exporting does 
    not leave an audit trail...
    
    > Or, run a microkernel you do trust. :)
    > 
    
    You throwing me out of Linux altogether? (^_^)
    
    I like microkernels, but there's no one (yet) that works as well as 
    Linux.  Period.
    
    > _______________________________________________
    > linux-security-module mailing list
    > linux-security-moduleat_private
    > http://mail.wirex.com/mailman/listinfo/linux-security-module
    > 
    
    
    _______________________________________________
    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 : Fri Jun 08 2001 - 11:27:25 PDT