Re: [RFC] LSM fix for stupid "empty" functions

[Greg KH]

On Mon, Dec 02, 2002 at 01:00:27PM +1100, James Morris wrote:
> On Sun, 1 Dec 2002, Greg KH wrote:
> 
> > > I think we still want to make sure that the module author has explicitly
> > > accounted for all of the hooks, in case new hooks are added.
> > 
> > But with this patch, if the module author hasn't specified a hook, they
> > get the "dummy" ones.  So the structure should always be full of
> > pointers, making the VERIFY_STRUCT macro pointless.
> 
> Yes, but defaulting unspecified hooks to dummy operations could be
> dangerous.  A module might appear to compile and run perfectly well, but 
> be missing some important new hook.

One could argue that a "important new hook" would provide a sane dummy
operation, or that if the module doesn't need it, why would it want to
provide it?  :)

Anyway, there's no way to resolve both this percieved problem, and the
"smaller and easier" patch that I proposed, right?  Unless we want to
export all dummy operation functions for all modules to use?  I could do
that, but it's pretty messy...

thanks,

greg k-h