Re: How to clear compile error when using rdrand64_step due to typedef'ing a 64-bit type?

[Marc Glisse <marc dot glisse at inria dot fr>]

On Fri, 19 Aug 2016, Jonathan Wakely wrote:

> On 19 August 2016 at 09:53, Jonathan Wakely wrote:
> > On 19 August 2016 at 09:07, Jeffrey Walton wrote:
> > > > You've basically gone chasing after some unrelated definition (that of
> > > > uint64_t) which has nothing to do with _rdrand64(), looking at
> > > > unrelated implementation details of that different type, and the
> > > > simple solution is to just use the type in the function declaration.
> > >
> > > That's how Intel documents it.
> >
> > They document it as unsigned __int64 but that's not a standard type.
> >
> > That's not a standard type. You found some definition of __int64 in a
> > different set of GCC headers and chased that, but there's no guarantee
> > that the header you looked in defined it consistently with how it's
> > used in the Intel docs.
> >
> > > A long time ago I was told to use Intel's documentation because GCC
> > > does not provide it (cf.,
> > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=68081).
> >
> > GCC might not document the function, but the header unambiguously
> > defines the function signature, and it takes unsigned long long*.
>
> Arguably the GCC intrinsic should use uint64_t* but it doesn't, so the
> answer to your original question is still to simply use the right
> type. And the right type was shown in the error messages you quoted in
> your first mail.
>
> If a function is declared to take X* and you get an error when passing
> Y* the solution is to pass X* instead.

We could overload the function in C++ in 64-bit mode, but that may break 
code that tries to use a function pointer.

--
Marc Glisse