[cfe-dev] RFC: __atomic_* support for gcc 4.7 compatibility

[Jeffrey Yasskin]

On Tue, Apr 10, 2012 at 3:20 PM, Howard Hinnant <hhinnant at apple.com> wrote:
> On Apr 10, 2012, at 6:11 PM, David Chisnall wrote:
>
>> On 10 Apr 2012, at 22:57, Howard Hinnant wrote:
>>
>>> David, could you revert the initialization statements to use either plain C++, or the __atomic_init intrinsic?
>>
>> Sorry, I thought I'd done that already when I fixed initialisation of _Atomic types in clang.  I'm just about to fall asleep, but I'll do it in the morning.
>
> Of course.  I'd much rather you be awake when you do this. :-)
>
>>
>> __atomic_init() is still required for atomic_init() (in both C11 and C++11) - these are currently using __atomic_store, in libc++, so I'll fix them at the same time.
>
> template <class _Tp>
> inline _LIBCPP_INLINE_VISIBILITY
> void
> atomic_init(atomic<_Tp>* __o, _Tp __d)
> {
>    __o->__a_ = __d;
> }
>
> Not right?

If __o->__a_ is an _Atomic(_Tp), then C11 says "Loads and stores of
objects with atomic types are done with memory_order_seq_cst
semantics." (6.2.6.1), and the simple assignment gives you the
equivalent of atomic_store(&__o->__a_, __d, memory_order_seq_cst).

You could get the effect of atomic_init with a memcpy, but only by
knowing the _Atomic(_Tp)'s layout on the target platform.