https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95908
--- Comment #1 from Richard Earnshaw <rearnsha at gcc dot gnu.org> --- I'm sure the code we generate doesn't match your expectations. What's more, I don't think it's really practical to meet them. To do so would require emitting a mask operation after practically every arithmetic operation done on pointers to ensure the value is kept within the valid range of the address space, since we have no idea how close an object might be to the end of that space. So my inclination is to resolve this a WONTFIX; but I'll leave it open for now so that others can comment.
