Re: [PATCH 3/3] stddef: Define max_align_t

[Szabolcs Nagy <nsz@port70.net>]

* Khem Raj <raj.khem@gmail.com> [2014-04-27 22:51:34 -0700]:
> On Sun, Apr 27, 2014 at 7:03 PM, Rich Felker <dalias@libc.org> wrote:
> >> +typedef struct {
> >> +      long long __max_align_ll __attribute__((__aligned__(__alignof__(long long))));
> >> +        long double __max_align_ld __attribute__((__aligned__(__alignof__(long double))));
> >> +} max_align_t;
> >> +
> >
> > As far as I can tell, there's no reason to use the attribute here.
> > What's it there for? Also a union would probably be nicer than a
> > struct, but perhaps it doesn't matter.
> 
> union does not return correct alignment where as struct did. I just
> tried to match
> what clang also has
> 
> http://reviews.llvm.org/rL201729
> 

nice..

i think i386 abi is non-conforming to the c11 alignment requirements now:
long long has 8 byte alignment, but in a struct/union it has only 4
(this is why the attrs are needed above)

	long long x; // _Alignof(x) == 8
	struct {long long x;} y; // _Alignof(y.x) == 4

i think the standard requires that all (addressable) long long objects
should have the same alignment (or stricter) than _Alignof(x)

max_align_t is defined to be the "greatest alignment supported in all
contexts", i don't know why it is not just

	typedef char max_align_t __attribute__((aligned(__BIGGEST_ALIGNMENT__)));

which gives 16 byte alignment on i386 gcc, i thought it was supported
in all contexts

if gcc and clang went with the same definition we should follow, but
this makes the type less meaningful