> The standard guarantees that function pointers are unique. Forcing stat > et.al to be local would violate this, in case a shared library returns a > pointer to one of them. Then what would violate the standards would be returning such an address from a libc function. If this is not done, it is not a problem.