https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95663
--- Comment #21 from Jeffrey A. Law <law at gcc dot gnu.org> --- I don't think there's anything inherently wrong with treating 0 + small offset similarly to 0 when it comes to -fdelete-null-pointer-checks. I suspect it'll break some poorly written low level code, though we fixed up some of that a year or so ago in response to some of Martin's diagnostic work which was flagging 0 + offset as problematical.
