https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92634
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> --- No those are still officially considered a referencing. In fact all three cases: &p->field does not dereference p, just as &*p and &p[i] do not. Should be considered a deference even though the deference does not happen. There are other similar bug reports that say similar things to the null pointer case.
