]> git.ipfire.org Git - thirdparty/gcc.git/blobdiff - libgfortran/kinds-override.h
Fortran: allow __float128 on targets where long double is not REAL(KIND=10)
[thirdparty/gcc.git] / libgfortran / kinds-override.h
index c9e874a3f383979149154c6c12b57c1b72f53dc7..5f7840b0c8982e2bbf30ee93e86b33c80b3a8202 100644 (file)
@@ -23,24 +23,9 @@ see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
 <http://www.gnu.org/licenses/>.  */
 
 
-/* What are the C types corresponding to the real(kind=10) and
-   real(kind=16) types? We currently rely on the following assumptions:
-     -- if real(kind=10) exists, i.e. if HAVE_GFC_REAL_10 is defined,
-        then it is necessarily the "long double" type
-     -- if real(kind=16) exists, then:
-         * if HAVE_GFC_REAL_10, real(kind=16) is "__float128"
-        * otherwise, real(kind=16) is "long double"
-   To allow to change this in the future, we create the
-   GFC_REAL_16_IS_FLOAT128 macro that is used throughout libgfortran.  */
-
-#if defined(HAVE_GFC_REAL_16)
-# if defined(HAVE_GFC_REAL_10)
-#  define GFC_REAL_16_IS_FLOAT128
-#  if !defined(HAVE_FLOAT128)
-#   error "Where has __float128 gone?"
-#  endif
-# else
-#  define GFC_REAL_16_IS_LONG_DOUBLE
-# endif
+/* Ensure that TFmode is available under.  */
+
+#if defined(GFC_REAL_16_IS_FLOAT128) && !defined(HAVE_FLOAT128)
+# error "Where has __float128 gone?"
 #endif