Here are the patches I worked on today. It does seem to fix KIND=16 to use
Float128, but by not considering long double for KIND processing, it breaks the
tests that want to do ISO C binding to long double.
Feel free to completely ignore the patches and go off in a different
direction. But I thought it would be useful to share what I've done.
> From 443773ac040383311384577b48ecc0bd957ff328 Mon Sep 17 00:00:00 2001
> From: Michael Meissner <meiss...@linux.ibm.com>
> Date: Thu, 28 Oct 2021 23:23:53 -0400
> Subject: [PATCH] Initial patch for PowerPC Fortran KIND=16
This is a work in progress patch. It attempts to make Fortran KIND=16 to
always mean Float128 on PowerPC VSX systems. Unfortunately, in changing
KIND=16 to Float128, it breaks all of the ISO C bindings for long double
support in Fortran.
gcc/
2021-10-28 Michael Meissner <meiss...@the-meissners.org>
* config/rs6000/rs6000.h (FORTRAN_USE_FLOAT128): New macro.
(FORTRAN_USE_LONG_DOUBLE): New macro.
* tree.h (complex_float128_type_node): Define.
* doc/tm.texi.in (FORTRAN_USE_FLOAT128): Add documentation.
(FORTRAN_USE_LONG_DOUBLE): Likewise.
* doc/tm.texi: Regenerate.
gcc/fortran/
2021-10-28 Michael Meissner <meiss...@the-meissners.org>
* f95-lang.c (gfc_init_builtin_functions): Flesh out more Float128
support.
* gfortran.h (FORTRAN_USE_LONG_DOUBLE): Provide default
definition.
(FORTRAN_USE_FLOAT128): Likewise.
* trans-types.c (gfc_init_kinds): Add support for
FORTRAN_USE_LONG_DOUBLE and FORTRAN_USE_FLOAT128.
(gfc_build_real_type): Likewise.
(gfc_build_complex_type): Add support for Float128 complex.
---
gcc/config/rs6000/rs6000.h | 10 ++++++++++
gcc/doc/tm.texi | 13 +++++++++++++
gcc/doc/tm.texi.in | 13 +++++++++++++
gcc/fortran/f95-lang.c | 28 ++++++++++++++++++++++++++++
gcc/fortran/gfortran.h | 13 +++++++++++++
gcc/fortran/trans-types.c | 24 +++++++++++++++++-------
gcc/tree.h | 2 ++
7 files changed, 96 insertions(+), 7 deletions(-)
diff --git a/gcc/config/rs6000/rs6000.h b/gcc/config/rs6000/rs6000.h
index 3eba1c072cf..4e016e548db 100644
--- a/gcc/config/rs6000/rs6000.h
+++ b/gcc/config/rs6000/rs6000.h
@@ -2691,3 +2691,13 @@ while (0)
rs6000_asm_output_opcode (STREAM); \
} \
while (0)
+
+/* Whether Fortran should use long double or __float128 for KIND=16. If we
+ support IEEE 128-bit and long double is not IEEE 128-bit, then use the
+ _Float128 type for KIND=16. Otherwise use long double. */
+#undef FORTRAN_USE_FLOAT128
+#define FORTRAN_USE_FLOAT128 (TARGET_FLOAT128_TYPE && !TARGET_IEEEQUAD)
+
+#undef FORTRAN_USE_LONG_DOUBLE
+#define FORTRAN_USE_LONG_DOUBLE (!FORTRAN_USE_FLOAT128)
+
diff --git a/gcc/doc/tm.texi b/gcc/doc/tm.texi
index 902402d7503..13ecca2605c 100644
--- a/gcc/doc/tm.texi
+++ b/gcc/doc/tm.texi
@@ -12612,3 +12612,16 @@ counters are incremented using atomic operations.
Targets not supporting
64-bit atomic operations may override the default value and request a 32-bit
type.
@end deftypefn
+
+@defmac FORTRAN_USE_LONG_DOUBLE
+Define this macro to return true if Fortran should enable @code{long
+double} support for @code{KIND} processing. If you do not define this
+macro, Fortran always uses the @code{long double} type.
+@end defmac
+
+@defmac FORTRAN_USE_LONG_DOUBLE
+Define this macro to return true if Fortran should enable
+@code{_Float128} support for @code{KIND} processing. If you do not
+define this macro, Fortran will enable @code{_Float128} support if the
+quadmath library is built, and the mode @code{TFmode} is enabled.
+@end defmac
diff --git a/gcc/doc/tm.texi.in b/gcc/doc/tm.texi.in
index 86352dc9bd2..012ef1ecc98 100644
--- a/gcc/doc/tm.texi.in
+++ b/gcc/doc/tm.texi.in
@@ -8187,3 +8187,16 @@ maintainer is familiar with.
@hook TARGET_MEMTAG_UNTAGGED_POINTER
@hook TARGET_GCOV_TYPE_SIZE
+
+@defmac FORTRAN_USE_LONG_DOUBLE
+Define this macro to return true if Fortran should enable @code{long
+double} support for @code{KIND} processing. If you do not define this
+macro, Fortran always uses the @code{long double} type.
+@end defmac
+
+@defmac FORTRAN_USE_LONG_DOUBLE
+Define this macro to return true if Fortran should enable
+@code{_Float128} support for @code{KIND} processing. If you do not
+define this macro, Fortran will enable @code{_Float128} support if the
+quadmath library is built, and the mode @code{TFmode} is enabled.
+@end defmac
diff --git a/gcc/fortran/f95-lang.c b/gcc/fortran/f95-lang.c
index 58dcaf01d75..b8117dc72b4 100644
--- a/gcc/fortran/f95-lang.c
+++ b/gcc/fortran/f95-lang.c
@@ -674,9 +674,11 @@ gfc_init_builtin_functions (void)
tree mfunc_float[6];
tree mfunc_double[6];
tree mfunc_longdouble[6];
+ tree mfunc_float128[6];
tree mfunc_cfloat[6];
tree mfunc_cdouble[6];
tree mfunc_clongdouble[6];
+ tree mfunc_cfloat128[6];
tree func_cfloat_float, func_float_cfloat;
tree func_cdouble_double, func_double_cdouble;
tree func_clongdouble_longdouble, func_longdouble_clongdouble;
@@ -691,9 +693,11 @@ gfc_init_builtin_functions (void)
build_builtin_fntypes (mfunc_float, float_type_node);
build_builtin_fntypes (mfunc_double, double_type_node);
build_builtin_fntypes (mfunc_longdouble, long_double_type_node);
+ build_builtin_fntypes (mfunc_float128, float128_type_node);
build_builtin_fntypes (mfunc_cfloat, complex_float_type_node);
build_builtin_fntypes (mfunc_cdouble, complex_double_type_node);
build_builtin_fntypes (mfunc_clongdouble, complex_long_double_type_node);
+ build_builtin_fntypes (mfunc_cfloat128, complex_float128_type_node);
func_cfloat_float = build_function_type_list (float_type_node,
complex_float_type_node,
@@ -736,6 +740,8 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_roundl", mfunc_longdouble[0],
BUILT_IN_ROUNDL, "roundl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ gfc_define_builtin ("__builtin_roundf128", mfunc_float128[0],
+ BUILT_IN_ROUNDF128, "roundf128",
ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_round", mfunc_double[0],
BUILT_IN_ROUND, "round", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_roundf", mfunc_float[0],
@@ -743,6 +749,8 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_truncl", mfunc_longdouble[0],
BUILT_IN_TRUNCL, "truncl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ gfc_define_builtin ("__builtin_truncf128", mfunc_float128[0],
+ BUILT_IN_TRUNCF128, "truncl",
ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_trunc", mfunc_double[0],
BUILT_IN_TRUNC, "trunc", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_truncf", mfunc_float[0],
@@ -750,6 +758,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_cabsl", func_clongdouble_longdouble,
BUILT_IN_CABSL, "cabsl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_cabsf128. */
gfc_define_builtin ("__builtin_cabs", func_cdouble_double,
BUILT_IN_CABS, "cabs", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_cabsf", func_cfloat_float,
@@ -758,6 +767,9 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_copysignl", mfunc_longdouble[1],
BUILT_IN_COPYSIGNL, "copysignl",
ATTR_CONST_NOTHROW_LEAF_LIST);
+ gfc_define_builtin ("__builtin_copysignf128", mfunc_longdouble[1],
+ BUILT_IN_COPYSIGNF128, "copysignf128",
+ ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_copysign", mfunc_double[1],
BUILT_IN_COPYSIGN, "copysign",
ATTR_CONST_NOTHROW_LEAF_LIST);
@@ -768,6 +780,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_nextafterl", mfunc_longdouble[1],
BUILT_IN_NEXTAFTERL, "nextafterl",
ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_nextafterf128. */
gfc_define_builtin ("__builtin_nextafter", mfunc_double[1],
BUILT_IN_NEXTAFTER, "nextafter",
ATTR_CONST_NOTHROW_LEAF_LIST);
@@ -781,6 +794,8 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_rintl", mfunc_longdouble[0],
BUILT_IN_RINTL, "rintl", attr);
+ gfc_define_builtin ("__builtin_rintf128", mfunc_float128[0],
+ BUILT_IN_RINTF128, "rintf128", attr);
gfc_define_builtin ("__builtin_rint", mfunc_double[0],
BUILT_IN_RINT, "rint", attr);
gfc_define_builtin ("__builtin_rintf", mfunc_float[0],
@@ -788,6 +803,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_remainderl", mfunc_longdouble[1],
BUILT_IN_REMAINDERL, "remainderl", attr);
+ /* no __builtin_remainderf128. */
gfc_define_builtin ("__builtin_remainder", mfunc_double[1],
BUILT_IN_REMAINDER, "remainder", attr);
gfc_define_builtin ("__builtin_remainderf", mfunc_float[1],
@@ -795,6 +811,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_logbl", mfunc_longdouble[0],
BUILT_IN_LOGBL, "logbl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_logbf128. */
gfc_define_builtin ("__builtin_logb", mfunc_double[0],
BUILT_IN_LOGB, "logb", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_logbf", mfunc_float[0],
@@ -803,6 +820,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_frexpl", mfunc_longdouble[4],
BUILT_IN_FREXPL, "frexpl", ATTR_NOTHROW_LEAF_LIST);
+ /* no __builtin_frexpf128. */
gfc_define_builtin ("__builtin_frexp", mfunc_double[4],
BUILT_IN_FREXP, "frexp", ATTR_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_frexpf", mfunc_float[4],
@@ -810,6 +828,8 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_fabsl", mfunc_longdouble[0],
BUILT_IN_FABSL, "fabsl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ gfc_define_builtin ("__builtin_fabsf128", mfunc_float128[0],
+ BUILT_IN_FABSF128, "fabsf128",
ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_fabs", mfunc_double[0],
BUILT_IN_FABS, "fabs", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_fabsf", mfunc_float[0],
@@ -817,6 +837,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_scalbnl", mfunc_longdouble[2],
BUILT_IN_SCALBNL, "scalbnl",
ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_scalbnf128. */
gfc_define_builtin ("__builtin_scalbn", mfunc_double[2],
BUILT_IN_SCALBN, "scalbn", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_scalbnf", mfunc_float[2],
@@ -824,6 +845,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_fmodl", mfunc_longdouble[1],
BUILT_IN_FMODL, "fmodl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_fmodf128. */
gfc_define_builtin ("__builtin_fmod", mfunc_double[1],
BUILT_IN_FMOD, "fmod", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_fmodf", mfunc_float[1],
@@ -872,18 +894,21 @@ gfc_init_builtin_functions (void)
/* These are used to implement the ** operator. */
gfc_define_builtin ("__builtin_powl", mfunc_longdouble[1],
BUILT_IN_POWL, "powl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_powf128. */
gfc_define_builtin ("__builtin_pow", mfunc_double[1],
BUILT_IN_POW, "pow", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_powf", mfunc_float[1],
BUILT_IN_POWF, "powf", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_cpowl", mfunc_clongdouble[1],
BUILT_IN_CPOWL, "cpowl", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_cpowf128. */
gfc_define_builtin ("__builtin_cpow", mfunc_cdouble[1],
BUILT_IN_CPOW, "cpow", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_cpowf", mfunc_cfloat[1],
BUILT_IN_CPOWF, "cpowf", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_powil", mfunc_longdouble[2],
BUILT_IN_POWIL, "powil", ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_powif128. */
gfc_define_builtin ("__builtin_powi", mfunc_double[2],
BUILT_IN_POWI, "powi", ATTR_CONST_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_powif", mfunc_float[2],
@@ -895,6 +920,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_cbrtl", mfunc_longdouble[0],
BUILT_IN_CBRTL, "cbrtl",
ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_cbrtf128. */
gfc_define_builtin ("__builtin_cbrt", mfunc_double[0],
BUILT_IN_CBRT, "cbrt",
ATTR_CONST_NOTHROW_LEAF_LIST);
@@ -904,6 +930,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_cexpil", func_longdouble_clongdouble,
BUILT_IN_CEXPIL, "cexpil",
ATTR_CONST_NOTHROW_LEAF_LIST);
+ /* no __builtin_cexpif128. */
gfc_define_builtin ("__builtin_cexpi", func_double_cdouble,
BUILT_IN_CEXPI, "cexpi",
ATTR_CONST_NOTHROW_LEAF_LIST);
@@ -917,6 +944,7 @@ gfc_init_builtin_functions (void)
gfc_define_builtin ("__builtin_sincosl",
func_longdouble_longdoublep_longdoublep,
BUILT_IN_SINCOSL, "sincosl", ATTR_NOTHROW_LEAF_LIST);
+ /* no __builtin_sincosf128. */
gfc_define_builtin ("__builtin_sincos", func_double_doublep_doublep,
BUILT_IN_SINCOS, "sincos", ATTR_NOTHROW_LEAF_LIST);
gfc_define_builtin ("__builtin_sincosf", func_float_floatp_floatp,
diff --git a/gcc/fortran/gfortran.h b/gcc/fortran/gfortran.h
index c25d1cca3a8..15c940b0f31 100644
--- a/gcc/fortran/gfortran.h
+++ b/gcc/fortran/gfortran.h
@@ -3791,4 +3791,17 @@ bool gfc_is_reallocatable_lhs (gfc_expr *);
void finish_oacc_declare (gfc_namespace *, gfc_symbol *, bool);
void gfc_adjust_builtins (void);
+/* Whether to enable long double and/or float128 processing. */
+#ifndef FORTRAN_USE_LONG_DOUBLE
+#define FORTRAN_USE_LONG_DOUBLE 1
+#endif
+
+#ifndef FORTRAN_USE_FLOAT128
+#if defined(HAVE_TFmode) && defined(ENABLE_LIBQUADMATH_SUPPORT)
+#define FORTRAN_USE_FLOAT128 1
+#else
+#define FORTRAN_USE_FLOAT128 0
+#endif
+#endif
+
#endif /* GCC_GFORTRAN_H */
diff --git a/gcc/fortran/trans-types.c b/gcc/fortran/trans-types.c
index 1c78a906397..6803b832183 100644
--- a/gcc/fortran/trans-types.c
+++ b/gcc/fortran/trans-types.c
@@ -453,10 +453,10 @@ gfc_init_kinds (void)
continue;
if (mode != TYPE_MODE (float_type_node)
&& (mode != TYPE_MODE (double_type_node))
- && (mode != TYPE_MODE (long_double_type_node))
-#if defined(HAVE_TFmode) && defined(ENABLE_LIBQUADMATH_SUPPORT)
- && (mode != TFmode)
-#endif
+ && (!FORTRAN_USE_LONG_DOUBLE
+ || mode != TYPE_MODE (long_double_type_node))
+ && (!FORTRAN_USE_FLOAT128
+ || mode != TYPE_MODE (float128_type_node))
)
continue;
@@ -854,9 +854,14 @@ gfc_build_real_type (gfc_real_info *info)
info->c_float = 1;
if (mode_precision == DOUBLE_TYPE_SIZE)
info->c_double = 1;
- if (mode_precision == LONG_DOUBLE_TYPE_SIZE)
+ if (FORTRAN_USE_LONG_DOUBLE && mode_precision == LONG_DOUBLE_TYPE_SIZE)
info->c_long_double = 1;
- if (mode_precision != LONG_DOUBLE_TYPE_SIZE && mode_precision == 128)
+
+ /* Don't check for just mode_precision == 128 for Float128. On the PowerPC,
+ there are 3 different 128-bit floating point types (IEEE 128-bit, IBM
+ 128-bit, and the default long double, which is either of the other types).
+ The precision is used to differentiate between the types. */
+ if (FORTRAN_USE_FLOAT128 && IN_RANGE (mode_precision, 126, 128))
{
/* TODO: see PR101835. */
info->c_float128 = 1;
@@ -867,8 +872,11 @@ gfc_build_real_type (gfc_real_info *info)
return float_type_node;
if (TYPE_PRECISION (double_type_node) == mode_precision)
return double_type_node;
- if (TYPE_PRECISION (long_double_type_node) == mode_precision)
+ if (FORTRAN_USE_LONG_DOUBLE
+ && TYPE_PRECISION (long_double_type_node) == mode_precision)
return long_double_type_node;
+ if (FORTRAN_USE_FLOAT128 && IN_RANGE (mode_precision, 126, 128))
+ return float128_type_node;
new_type = make_node (REAL_TYPE);
TYPE_PRECISION (new_type) = mode_precision;
@@ -889,6 +897,8 @@ gfc_build_complex_type (tree scalar_type)
return complex_double_type_node;
if (scalar_type == long_double_type_node)
return complex_long_double_type_node;
+ if (scalar_type == float128_type_node)
+ return complex_float128_type_node;
new_type = make_node (COMPLEX_TYPE);
TREE_TYPE (new_type) = scalar_type;
diff --git a/gcc/tree.h b/gcc/tree.h
index 7542d97ce12..f3b47f81a09 100644
--- a/gcc/tree.h
+++ b/gcc/tree.h
@@ -4225,6 +4225,8 @@ tree_strip_any_location_wrapper (tree exp)
#define complex_double_type_node global_trees[TI_COMPLEX_DOUBLE_TYPE]
#define complex_long_double_type_node
global_trees[TI_COMPLEX_LONG_DOUBLE_TYPE]
+#define complex_float128_type_node global_trees[TI_COMPLEX_FLOAT128_TYPE]
+
#define COMPLEX_FLOATN_NX_TYPE_NODE(IDX)
global_trees[TI_COMPLEX_FLOATN_NX_TYPE_FIRST + (IDX)]
#define void_type_node global_trees[TI_VOID_TYPE]
--
2.31.1
--
Michael Meissner, IBM
PO Box 98, Ayer, Massachusetts, USA, 01432
email: meiss...@linux.ibm.com
