On Tue, 2 Jul 2019 at 23:33, Jeffrey Walton <noloader@xxxxxxxxx> wrote: > > On Tue, Jul 2, 2019 at 6:31 PM Segher Boessenkool > <segher@xxxxxxxxxxxxxxxxxxx> wrote: > > > > On Tue, Jul 02, 2019 at 05:46:24PM -0400, Jeffrey Walton wrote: > > > We don't use <x86intrin.h> because it is missing on too many systems > > > we support (even those that use GCC). > > > > > > Is there a way to get GCC to provide the functions through a system > > > header as expected? > > > > #include <x86intrin.h> > > > > It *is* a system header. And it is provided by GCC, and it gives you > > _mm_roti_epi64, provided you have GCC 4.5 or later. > > > > If you don't want to use it, you're on your own. Good luck. > > I believe <x86intrin.h> is a GCC specific header. No it isn't: https://github.com/llvm/llvm-project/blob/4cf5743b7784b2c2bff8be82e1a5cc37f8c68a36/clang/lib/Headers/x86intrin.h > I believe the > expected header is <amminstrin.h>. > > Is there any reason GCC can't just follow conventions and stop making > the shit up as they go? As Segher said, these files were contributed by (and are largely maintained by) people from AMD and Intel. Maybe you should talk to them and stop complaining about GCC making shit up, since GCC didn't make it up. Using <x86intrin.h> is the convention.
