[flang] Change complex type define in runtime for clang-cl

[rovka]

Summary

When compiling the runtime with a version of clang-cl newer than 12, we
define CMPLXF as __builtin_complex, which returns a float _Complex type.
This errors out in contexts where the result of CMPLXF is expected to be
a float_Complex_t. This is defined as _Fcomplex whenever _MSC_VER is
defined and as float _Complex otherwise.

This patch defines float_Complex_t & friends as _Fcomplex only when
we're using "true" MSVC, and not just clang-pretending-to-be-MSVC. This
should only affect clang-cl >= 12.

[Meinersbur]

pre-merge check fails because disk is full.

flang/runtime/complex-reduction.h

23

! has a higher precedence than >=

With "true" msvc, __clang_major__ will not be defined and evaluate to 0, so it turns out to work. Still consider checking for whether __clang_major__ is defined to make intention clear.

[rovka]

Thanks for the review, I've updated all relevant sites to check if clang_major is defined.

[Meinersbur]

LGTM

flang/runtime/complex-reduction.h

 #ifndef FORTRAN_RUNTIME_COMPLEX_REDUCTION_H_
 #define FORTRAN_RUNTIME_COMPLEX_REDUCTION_H_
 
 #include "flang/Runtime/entry-names.h"
 #include <complex.h>
 
 struct CppDescriptor; /* dummy type name for Fortran::runtime::Descriptor */
 
-#ifdef _MSC_VER
+#if defined _MSC_VER && ! __clang_major__ >= 12
 typedef _Fcomplex float_Complex_t;
 typedef _Dcomplex double_Complex_t;
 typedef _Lcomplex long_double_Complex_t;
 #else
 typedef float _Complex float_Complex_t;
 typedef double _Complex double_Complex_t;
 typedef long double long_double_Complex_t;
 #endif