Files
gcc/libgfortran/ieee/issignaling_fallback.h
T
Jakub Jelinek 133d0d422e fortran, libgfortran: Avoid using libquadmath for glibc 2.26+
As mentioned by Joseph in PR105101, glibc 2.26 or later has on x86
(both -m32/-m64), powerpc64le, ia64 and mips support for
*f128 math/complex APIs plus strtof128 and strfromf128, and these APIs allow
us to avoid libquadmath for Fortran purposes on these architectures,
replace *q math/complex APIs, strtof128 instead of strtoflt128 and,
while strfromf128 unfortunately isn't a perfect replacement to
quadmath_snprintf, it can be made to work.

The advantage of this is that when configured against such glibcs
(2.26 is now almost 5 years old), we can avoid linking against an extra shared
library and the math support in glibc is maintained better than libquadmath.

We need both a compiler change (so that for glibc 2.26+ it uses *f128 APIs
instead of *q) and library change.

The above mentioned problem with strfromf128 is that the strfrom* functions
are severely restricted versions of snprintf.  In libgfortran, we handle
!isfinite differently and just use snprintf/quadmath_snprintf for
%+-#.*{L,Q}{f,e} printing.
strfrom* doesn't allow +, -, # modifiers and it only supports .34 or
similar precision, not .* .  The L/Q etc. letters are omitted.
The + is there to force + sign at the start if it is positive.
Workaround in the patch is to add the + at the start manually for
!signbit (val).
The - (left alignment instead of right) I don't understand why we need it,
when minimum field width isn't specified (for strfrom* can't be specified),
no padding is ever added anywhere I believe.
The # is to force adding . - workaround is to search for first . or e or '\0'
character, if it is '\0', just append ., if it is e, insert . before e and
memmove the rest (which is just a few bytes, e, +/- and at most a few digits)
one byte later.
The .* case is handled by creating the format string for strfrom* by
snprintf into a temporary buffer.

As requested, this patch also switches from using __float128 type in
libgfortran to _Float128 which is equivalent on all arches that support
__float128.

The change is done in a backwards compatible change, when GCC is configured
against glibc 2.26 or newer, libgfortran.so.5 itself doesn't link against
-lquadmath nor uses any libquadmath APIs, libgfortran.a doesn't use any
libquadmath APIs either.  User programs and libraries when being linked
by gfortran driver are linked against -lgfortran and -lquadmath, but
the latter only in the --as-needed linker mode, which means it needs
to be around during linking and will be linked in if there are any
calls to math/complex functions with real(kind=16) or complex(kind=16)
in compilation units compiled by older versions of gcc, but if either
user code doesn't call those math/complex functions for the largest
supported kind, or the code is recompiled by gcc with this change in,
libquadmath won't be linked in.

2022-06-28  Jakub Jelinek  <jakub@redhat.com>

gcc/fortran/
	* gfortran.h (gfc_real_info): Add use_iec_60559 bitfield.
	* trans-types.h (gfc_real16_use_iec_60559): Declare.
	* trans-types.cc (gfc_real16_use_iec_60559): Define.
	(gfc_init_kinds): When building powerpc64le-linux libgfortran
	on glibc 2.26 to 2.31, set gfc_real16_use_iec_60559 and
	use_iec_60559.
	(gfc_build_real_type): Set gfc_real16_use_iec_60559 and use_iec_60559
	on glibc 2.26 or later.
	* trans-intrinsic.cc (gfc_build_intrinsic_lib_fndecls): Adjust
	comment.  Handle gfc_real16_use_iec_60559.
	(gfc_get_intrinsic_lib_fndecl): Handle use_iec_60559.
libgfortran/
	* configure.ac: Check for strtof128 and strfromf128.
	Check for math and complex *f128 functions.  Set
	have_iec_60559_libc_support to yes if *f128 support is around, for
	--enable-libquadmath-support default to "default" rather than yes if
	have_iec_60559_libc_support is yes.
	* acinclude.m4 (LIBGFOR_CHECK_FLOAT128): Test
	_Float128/_Complex _Float128 rather than __float128 and
	_Complex float __attribute__((mode(TC))).  If libquadmath support
	is defaulted and have_iec_60559_libc_support is yes, define and subst
	USE_IEC_60559.  Remove unused LIBGFOR_BUILD_QUAD conditional.
	* Makefile.am (kinds.h): Pass @USE_IEC_60559@ as an extra
	mk-kinds-h.sh argument.
	* mk-kinds-h.sh: Accept 4th use_iec_60559 argument.  Use
	_Float128/_Complex _Float128 types instead of __float128 and
	_Complex float __attribute__((mode(TC))), and if use_iec_60559 is yes,
	use f128 suffix instead of q and define GFC_REAL_16_USE_IEC_60559.
	* kinds-override.h: Use _Float128/_Complex _Float128 types instead of
	__float128 and _Complex float __attribute__((mode(TC))), if
	USE_IEC_60559 is defined, use f128 suffixes instead of q and
	define GFC_REAL_17_USE_IEC_60559.
	* libgfortran.h: Don't include quadmath_weak.h if USE_IEC_60559 is
	defined.
	(GFC_REAL_16_INFINITY, GFC_REAL_16_QUIET_NAN): Define
	for GFC_REAL_16_USE_IEC_60559 differently.
	* caf/single.c (convert_type): Use _Float128/_Complex _Float128
	instead of __float128 and _Complex float __attribute__((mode(TC))).
	For HAVE_GFC_REAL_10 when HAVE_GFC_REAL_16 isn't defined use
	_Complex long double instead of long double.
	* ieee/issignaling_fallback.h (ieee854_float128_shape_type): Use
	_Float128 instead of __float128.
	(__issignalingf128): Change argument type to _Float128.
	(issignaling): Use _Float128 instead of __float128 in _Generic.
	* intrinsics/cshift0.c (cshift0): Use _Float128 instead of __float128
	in a comment.  Fix a comment typo, logn double -> long double.
	* intrinsics/erfc_scaled.c (_THRESH, _M_2_SQRTPI, _INF, _ERFC, _EXP):
	Use different definitions if GFC_REAL_16_USE_IEC_60559.
	(_THRESH, _M_2_SQRTPI): Use GFC_REAL_17_LITERAL macro.
	(_ERFC, _EXP): Use different definitions if GFC_REAL_17_USE_IEC_60559.
	* intrinsics/spread_generic.c (spread, spread_scalar): Use _Float128
	instead of __float128 in a comment.  Fix a comment typo,
	logn double -> long double.
	* intrinsics/trigd.c (ENABLE_SIND, ENABLE_COSD, ENABLE_TAND): Handle
	GFC_REAL_16_USE_IEC_60559.
	* intrinsics/pack_generic.c (pack): Use _Float128 instead of
	__float128 in a comment.  Fix a comment typo, logn double ->
	long double.
	* intrinsics/unpack_generic.c (unpack1, unpack0): Likewise.
	* runtime/in_pack_generic.c (internal_pack): Likewise.
	* runtime/in_unpack_generic.c (internal_unpack): Likewise.
	* io/read.c (convert_real, convert_infnan): Handle
	GFC_REAL_16_USE_IEC_60559 and GFC_REAL_17_USE_IEC_60559.
	* io/transfer128.c (tmp1, tmp2): Don't define if libquadmath
	isn't needed.
	* io/write_float.def (gfor_strfromf128): New function.
	(DTOA2Q, FDTOA2Q): Define differently if
	GFC_REAL_16_USE_IEC_60559 or GFC_REAL_17_USE_IEC_60559.
	* m4/mtype.m4: Use different suffix if GFC_REAL_16_USE_IEC_60559
	or GFC_REAL_17_USE_IEC_60559.
	* config.h.in: Regenerated.
	* configure: Regenerated.
	* Makefile.in: Regenerated.
	* generated/bessel_r16.c: Regenerated.
	* generated/bessel_r17.c: Regenerated.
	* generated/norm2_r16.c: Regenerated.
	* generated/norm2_r17.c: Regenerated.
2022-06-28 13:14:45 +02:00

252 lines
5.5 KiB
C

/* Fallback implementation of issignaling macro.
Copyright (C) 2022 Free Software Foundation, Inc.
Contributed by Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
This file is part of the GNU Fortran runtime library (libgfortran).
Libgfortran is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public
License as published by the Free Software Foundation; either
version 3 of the License, or (at your option) any later version.
Libgfortran is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
Under Section 7 of GPL version 3, you are granted additional
permissions described in the GCC Runtime Library Exception, version
3.1, as published by the Free Software Foundation.
You should have received a copy of the GNU General Public License and
a copy of the GCC Runtime Library Exception along with this program;
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
<http://www.gnu.org/licenses/>. */
#include "libgfortran.h"
/* This header provides an implementation of the type-generic issignaling macro.
Some points of note:
- This header is only included if the issignaling macro is not defined.
- All targets for which Fortran IEEE modules are supported currently have
the high-order bit of the NaN mantissa clear for signaling (and set
for quiet), as recommended by IEEE.
- We use the __*_IS_IEC_60559__ macros to make sure we only deal with formats
we know. For other floating-point formats, we consider all NaNs as quiet.
*/
typedef union
{
float value;
uint32_t word;
} ieee_float_shape_type;
static inline int
__issignalingf (float x)
{
#if __FLT_IS_IEC_60559__
uint32_t xi;
ieee_float_shape_type u;
u.value = x;
xi = u.word;
xi ^= 0x00400000;
return (xi & 0x7fffffff) > 0x7fc00000;
#else
return 0;
#endif
}
typedef union
{
double value;
uint64_t word;
} ieee_double_shape_type;
static inline int
__issignaling (double x)
{
#if __DBL_IS_IEC_60559__
ieee_double_shape_type u;
uint64_t xi;
u.value = x;
xi = u.word;
xi ^= UINT64_C (0x0008000000000000);
return (xi & UINT64_C (0x7fffffffffffffff)) > UINT64_C (0x7ff8000000000000);
#else
return 0;
#endif
}
#if __LDBL_DIG__ == __DBL_DIG__
/* Long double is the same as double. */
static inline int
__issignalingl (long double x)
{
return __issignaling (x);
}
#elif (__LDBL_DIG__ == 18) && __LDBL_IS_IEC_60559__
/* Long double is x86 extended type. */
typedef union
{
long double value;
struct
{
#if __FLOAT_WORD_ORDER__ == __ORDER_BIG_ENDIAN__
int sign_exponent:16;
unsigned int empty:16;
uint32_t msw;
uint32_t lsw;
#elif __FLOAT_WORD_ORDER__ == __ORDER_LITTLE_ENDIAN__
uint32_t lsw;
uint32_t msw;
int sign_exponent:16;
unsigned int empty:16;
#endif
} parts;
} ieee_long_double_shape_type;
static inline int
__issignalingl (long double x)
{
int ret;
uint32_t exi, hxi, lxi;
ieee_long_double_shape_type u;
u.value = x;
exi = u.parts.sign_exponent;
hxi = u.parts.msw;
lxi = u.parts.lsw;
/* Pseudo numbers on x86 are always signaling. */
ret = (exi & 0x7fff) && ((hxi & 0x80000000) == 0);
hxi ^= 0x40000000;
hxi |= (lxi | -lxi) >> 31;
return ret || (((exi & 0x7fff) == 0x7fff) && (hxi > 0xc0000000));
}
#elif (__LDBL_DIG__ == 31)
/* Long double is 128-bit IBM extended type. */
static inline int
__issignalingl (long double x)
{
union { long double value; double parts[2]; } u;
u.value = x;
return __issignaling (u.parts[0]);
}
#elif (__LDBL_DIG__ == 33) && __LDBL_IS_IEC_60559__
/* Long double is 128-bit type. */
typedef union
{
long double value;
struct
{
#if __FLOAT_WORD_ORDER__ == __ORDER_BIG_ENDIAN__
uint64_t msw;
uint64_t lsw;
#elif __FLOAT_WORD_ORDER__ == __ORDER_LITTLE_ENDIAN__
uint64_t lsw;
uint64_t msw;
#endif
} parts64;
} ieee854_long_double_shape_type;
static inline int
__issignalingl (long double x)
{
uint64_t hxi, lxi;
ieee854_long_double_shape_type u;
u.value = x;
hxi = u.parts64.msw;
lxi = u.parts64.lsw;
hxi ^= UINT64_C (0x0000800000000000);
hxi |= (lxi | -lxi) >> 63;
return (hxi & UINT64_C (0x7fffffffffffffff)) > UINT64_C (0x7fff800000000000);
}
#else
static inline int
__issignalingl (long double x)
{
return 0;
}
#endif
#if defined(GFC_REAL_16_IS_FLOAT128)
/* We have a _Float128 type. */
typedef union
{
_Float128 value;
struct
{
#if __FLOAT_WORD_ORDER__ == __ORDER_BIG_ENDIAN__
uint64_t msw;
uint64_t lsw;
#elif __FLOAT_WORD_ORDER__ == __ORDER_LITTLE_ENDIAN__
uint64_t lsw;
uint64_t msw;
#endif
} parts64;
} ieee854_float128_shape_type;
static inline int
__issignalingf128 (_Float128 x)
{
uint64_t hxi, lxi;
ieee854_float128_shape_type u;
u.value = x;
hxi = u.parts64.msw;
lxi = u.parts64.lsw;
hxi ^= UINT64_C (0x0000800000000000);
hxi |= (lxi | -lxi) >> 63;
return (hxi & UINT64_C (0x7fffffffffffffff)) > UINT64_C (0x7fff800000000000);
}
#endif
/* Define the type-generic macro based on the functions above. */
#if defined(GFC_REAL_16_IS_FLOAT128)
# define issignaling(X) \
_Generic ((X), \
_Float128: __issignalingf128, \
float: __issignalingf, \
double: __issignaling, \
long double: __issignalingl)(X)
#else
# define issignaling(X) \
_Generic ((X), \
float: __issignalingf, \
double: __issignaling, \
long double: __issignalingl)(X)
#endif