Why hooks in sys_iopl and sys_ioperm?

From: Emily Ratliff (ratliffat_private)
Date: Tue Jul 23 2002 - 16:19:46 PDT

  • Next message: James Morris: "Re: Got suggestions to reduce the locks in stacker.c?"

    Looking at the arch specific stuff, there is a hook in sys_iopl and 
    sys_ioperm. Paul Mackerras pointed out to me that all I/O on the ppc is 
    memory mapped.  After hearing that, I was concerned that this might lead 
    to alternate platform security holes if the module writer puts different 
    checks in the sys_iopl and sys_ioperm hooks than in the memory mapping 
    hooks. But none of the current modules, selinux, dte, lids, owlsm 
    capabilities, dummy actually use the hooks. SELinux has a comment to say 
    that they use the CAP_RAWIO capability to handle the issue. Are these 
    hooks really desirable given that they might lead to easily overlooked 
    security bugs on non-Intel platforms? If they are really desirable, should 
    we warn the module writer that their module might not work as expected on 
    non-Intel platforms in the comment for them in security.h?
    
    Emily
    
    -- 
    Emily Ratliff
    IBM Linux Technology Center, Security
    
    _______________________________________________
    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 Jul 23 2002 - 16:21:01 PDT