Re: backward compat / access (was Re: Benchmarks)

From: Jesse Pollard (pollardat_private)
Date: Wed Apr 18 2001 - 10:25:15 PDT

  • Next message: Stephen Smalley: "Direction of the mailing list/effort"

    ---------  Received message begins Here  ---------
    
    > 
    > Seth Arnold wrote:
    > 
    > > * Luc Pardon <lucpat_private> [010418 01:02]:
    > > > An application developer may want to inform the user that (s)he
    > > > doesn't have sufficient rights to do something, without actually
    > > > attempt to do that "something" and set off all kinds of alarms.
    > >
    > > While I think I understand the reasoning involved, I know *I* would like
    > > to keep Linux as source-compatible with other Unix-like and Unix
    > > operating systems as possible. A mess of new syscalls (or one new
    > > syscall with a generic interface and many library wrappers) would only
    > > encourage non-standard code.
    > 
    > I see all of this as being part & parcel of module design, not
    > LSM interface design.  If you wanna write a module that provides this
    > functionality, go right ahead.  It will not be standard, and thus most
    > applications won't use that functionality.  But the freedom to do so is
    > exactly what LSM is about.
    > 
    > 
    > > Furthermore, I don't know if requiring the module to support such query
    > > interfaces is a good idea either -- the policy a module may desire to
    > > implement may wish to restrict this sort of information. Requiring the
    > > module to support query interfaces would leak this information, going
    > > against the module's design policy.
    > 
    > What LSM should be supporting is sufficient hooks to write such an
    > interface.  Even that is tentative:  I want to see an actual module that
    > needs this functionality before it goes into the LSM interface.
    
    Not a module, but a daemon like a user space NFS/Samba/other filesystem like
    daemon that has access privileges, but needs to have a requested access
    evaluated.
    
    It's only a POSSIBLE interface, not a required one. Personally, I don't
    think it would or should require a system call if there were a way to
    have an IOCTL to pass the request through. Much better than having to
    switch UIDs (or security contexts) to get an access evaluation done.
    
    -------------------------------------------------------------------------
    Jesse I Pollard, II
    Email: pollardat_private
    
    Any opinions expressed are solely my own.
    
    _______________________________________________
    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 : Wed Apr 18 2001 - 10:26:58 PDT