Re: Some New Hooks

From: Chris Wright (chrisat_private)
Date: Tue Jun 12 2001 - 09:27:23 PDT

  • Next message: Greg KH: "Re: New LSM patch for consideration"

    * Chris Vance (cvanceat_private) wrote:
    > 
    > I have added hooks in the following locations:
    
    thanks again for the patch.  i finally had the time last night to review
    the hooks.  some comments below...
    
    > fcntl/fcntl64 (sys_fcntl/sys_fcntl64): Added hooks to authorize these
    > 	operations.  Additionally, a security field was added to the
    > 	fown_struct so that the attributes of the owning process could
    > 	be maintained for later use in send_sigio_to_task.
    >
    > fcntl (send_sigio_to_task): Added hook to verify signal permissions.
    
    i'm not real familiar with this code path, so please correct my
    misunderstandings.
    
    the fown_ops->alloc_security in F_SETOWN is a perfect example of needing
    to separate allocation from setting the security blob.   i think this
    should really be set (not alloc), as the space should have already been
    allocated, (perhaps in get_empty_filp).  and i don't see any free (granted
    this may be a simple 32 bit security id superimposed on the void *, but
    we should get the chance to free in case it is actual allocated space).
    should the blob be set when a lease is established, or directory notification
    turned on?  since both of these actions maninpulate the f_owner.pid, and
    these can generate io signals to be sent.  i want to make sure that the
    send_sigiotask() check will always have a properly filled in f_owner.security
    blob.
    
    -chris
    
    _______________________________________________
    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 Jun 12 2001 - 09:35:29 PDT