> Note that the *only* way to subvert this scheme is to use > inline assembly to force the symbol weak behind gcc's back. > All I can say here is "Don't Do That". OK, that seems reasonable to me, and I withdraw my suggestion that a change is required here :-)