Re: quotactl hook

From: jmjonesat_private
Date: Sat Sep 01 2001 - 16:25:03 PDT

  • Next message: Greg KH: "Re: quotactl hook"

    On Sat, 1 Sep 2001, Greg KH wrote:
    
    > On Sat, Sep 01, 2001 at 06:46:36PM -0400, jmjonesat_private wrote:
    > > I don't have any specific response to this assertion, but, respectfully,
    > > ask for someone (even Greg ;)) to clarify "the direction LSM is heading",
    > > hopefully with regard to:
    > 
    > My opinions:
    > 
    > > 1) authoritative hooks: YES, NO, CONDITIONAL (how?)
    > 
    > No.  I've already talked about why I feel this way.  Please see the
    > archives.
    
    
    I had not thought this was the "policy", at this point, but rather had
    believed that SGI was reviewing "some authoritative" and had responded
    positively to a limitted application.
    
    If it's a complete "NO", I will procede forward based on that.
    
    > 
    > > 2) DAC bypass (as an option), YES, NO, CONDITIONAL (how?)
    > 
    > Do you mean the MAC before DAC discussion?
    > Personally I don't care.  But I like the current DAC before MAC
    > implementation that LSM seems to follow.
    
    I mean DAC (or, more accurately, IN-KERNEL checks) should execute and then
    provide the result to the module, which can return a code that is
    authoritative, if desired. This allows, ummmmm, not bypass, but
    override.
    
    While this is not optimum for my project, it is acceptable.  My module can
    replicate the DAC logic and add it's own "spin", and make a decision based
    upon that code... and log the discrepancy against the in-kernel logic
    accordingly.  
    
    I don't know how LSM can support this without SOME authoritative hooks,
    but, then again, a set of "DAC decision" hooks that simply record the 
    result but don't allow override might be between the two positions.
    
    If the current "decision/consensus" is restrictive_only, that answers my
    question.  I don't think this is the case, unless the "consensus" is
    weighted according to individual.
    
    > 
    > > 3) Support for loadable modules NOT compiled into the kernel (I've 
    > >    seen some "not an issue because we're suggesting compiling in" 
    > >    discussions that have short-circuited (perhaps not intentionally)
    > >    issues that may be relevant to allowing a module to slide into 
    > >    a system that has run for a while before the module is loaded.
    > > 
    > >    YES, NO, CONDITIONAL (how?)
    > 
    > Yes.  This support is there right now.  You just have to get the logic
    > correct in your module, being able to handle the tasks and other objects
    > that were created before your module was loaded.
    
    Yes, and No.  Tasks may run before the module loads and some of the
    "rebuttals" toward void blobs have been aimed at "it doesn't matter" with
    the "compiled in" consideration.
    
    I'd think that a mechanism to init blobs reliably without conflict would
    be a consideration if this support was to be applied.  I admit I haven't
    looked "very closely" at the patch with this regard... our prototypes
    assume that a NULL is an un-allocated blob and don't deal with locking.
    
    We DID see locking as a consdir-able issue and hoped it would get more
    discussion.
    
    Of course, if others have progressed farther and determined it to not be
    an issue, that's the power of open development, and we are grateful.
    
    > 
    > And if you want to recommend that your module be compiled into the
    > kernel (like SELinux does) that's your option.
    
    Yes, that's true, but is a little "smelly" of one-module-think.  I am one,
    and I think there are others, thinking along lines of stackable/loadable
    modules that can rise and fall.
    
    If LSM is really SSM (standard security module) and we're developing a
    standard, let's name it such... it might be easier to sell.  Sorry, a
    little overboard for this list.
    
    > 
    > > I'm dealing with developers in my project that insist that it may be
    > > necessary for us to "branch", and create a patch that removes LSM and
    > > reapplies a specific patch to the kernel to address our functionality.
    > > I'd rather not go that direction, but a few things that may be necessary
    > > are probably going to need a "plus-patch", and some other things that are
    > > admittedly possible, but require significant "manipulation" with the
    > > current patch may be better done with "plus-patches."
    > 
    > A branch of a patch, and I thought I had heard of everything :)
    > Fine, all the lsm work is opensource, and if it doesn't meet your needs
    > in certain ways, feel free to change it for your own usages.  Just
    > respect the current license and everyone will be happy.
    
    The LSM project has explored and implemented an excellent solution to a 
    variety of modules/policies.  There have been several mentions of policies
    and modules that may be "going left."  I can see that the "big names"
    working here wouldn't need some of the "leftward" options, but I readily
    admit that WE are working on one, and don't believe that we're the only
    ones.
    
    Perfect Solution: LSM addresses our needs.
    
    Moderate Solution: LSM addresses most of our needs, and we apply a very 
                       small patch to extend it to our needs, which,
                       hopefully, will be considered in Stage 2 or Stage 3
                       (or whatever) by LSM.
    
    Extreme Solution:  remove LSM and apply our own patch.  I don't think this
                       should be "evil" to the big-interests here... since
                       they all have had to apply proprietary patches to date.
    
    > 
    > Did this help? 
    > 
    
    Yes, very much, but I still hope for other responses... if for no other
    reason than to coordinate and agree-by-consensus some of these points for
    LSM
    
    > thanks,
    > 
    > greg k-h
    > 
    
    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 : Sat Sep 01 2001 - 16:27:28 PDT