On Fri, Dec 14, 2018 at 12:42:34PM +0100, Florian Weimer wrote: > * Rich Felker: > > > On Thu, Dec 13, 2018 at 05:04:59PM +0100, Florian Weimer wrote: > >> * Rich Felker: > >> > >> >> If the compiler can handle the zeroing, that would be great, though not > >> >> sure how (some __attribute__((zero)) which generates a type constructor > >> >> for such structure; it kind of departs from what the C language offers). > >> > > >> > The compiler fundamentally can't. At the very least it would require > >> > effective type tracking, which requires shadow memory and is even more > >> > controversial than -fstrict-aliasing (because in a sense it's a > >> > stronger version thereof). > >> > >> It's possible to do it with the right types. See _Bool on 32-bit Darwin > >> PowerPC for an example, which is four bytes instead of the usual one. > >> > >> Similarly, we could have integer types with trap representations. > >> Whether it is a good idea is a different matter, but the amount of > >> compiler magic required is actually limited. > > > > If you do this you just have LP64 with value range restricted to > > 32-bit. > > You have to a type different from long int for the relevant struct > fields. This type would have zero padding. Just upthread (Message-ID: <20181212165237.GT23599@xxxxxxxxxxxxxxxxxxxxx>) I explained why this does not work: >>> If on the other hand you tried to make just some pointers "wide >>> pointers", you'd also be completely breaking the specified API >>> contracts of standard interfaces. For example in struct iovec's >>> iov_base, &foo->iov_base is no longer a valid pointer to an object of >>> type void* that you can pass to interfaces expecting void**. Sloppy >>> misunderstandings like what you're making now are exactly why x32 is >>> already broken and buggy (&foo->tv_nsec already has wrong type for >>> struct timespec foo). Rich
