The new generic expf and exp2f code don't need wrappers any more, they
set errno inline, so only use the wrappers on targets that need it.
(If the wrapper is needed, then the top level wrapper code is included,
otherwise empty w_exp*f.c is used to suppress the wrapper.)
A powerpc64 expf implementation includes the expf c code directly which
needed some changes.
* sysdeps/ieee754/flt-32/e_exp2f.c (__exp2f): Define without wrapper.
* sysdeps/ieee754/flt-32/e_expf.c (__expf): Likewise
* sysdeps/ieee754/flt-32/w_exp2f.c: New file.
* sysdeps/ieee754/flt-32/w_expf.c: New file.
* sysdeps/powerpc/powerpc64/fpu/multiarch/e_expf-ppc64.c: Update for
the new expf code.
* sysdeps/powerpc/powerpc64/fpu/multiarch/w_expf.c: New file.
* sysdeps/powerpc/powerpc64/power8/fpu/w_expf.c: New file.
* sysdeps/m68k/m680x0/fpu/w_exp2f.c: New file.
* sysdeps/m68k/m680x0/fpu/w_expf.c: New file.
* sysdeps/i386/fpu/w_exp2f.c: New file.
* sysdeps/i386/fpu/w_expf.c: New file.
* sysdeps/i386/i686/fpu/multiarch/w_expf.c: New file.
* sysdeps/x86_64/fpu/w_expf.c: New file.
Based on new expf and exp2f code from
https://github.com/ARM-software/optimized-routines/
with wrapper on aarch64:
expf reciprocal-throughput: 2.3x faster
expf latency: 1.7x faster
without wrapper on aarch64:
expf reciprocal-throughput: 3.3x faster
expf latency: 1.7x faster
without wrapper on aarch64:
exp2f reciprocal-throughput: 2.8x faster
exp2f latency: 1.3x faster
libm.so size on aarch64:
.text size: -152 bytes
.rodata size: -1740 bytes
expf/exp2f worst case nearest rounding error: 0.502 ulp
worst case non-nearest rounding error: 1 ulp
Error checks are inline and errno setting is in separate tail called
functions, but the wrappers are kept in this patch to handle the
_LIB_VERSION==_SVID_ case. (So e.g. errno is set twice for expf calls
and once for __expf_finite calls on targets where the new code is used.)
Double precision arithmetics is used which is expected to be faster on
most targets (including soft-float) than using single precision and it
is easier to get good precision result with it.
Const data is kept in a separate translation unit which complicates
maintenance a bit, but is expected to give good code for literal loads
on most targets and allows sharing data across expf, exp2f and powf.
(This data is disabled on i386, m68k and ia64 which have their own
expf, exp2f and powf code.)
Some details may need target specific tweaks:
- best convert and round to int operation in the arg reduction may be
different across targets.
- code was optimized on fma target, optimal polynomial eval may be
different without fma.
- gcc does not always generate good code for fp bit representation
access via unions or it may be inherently slow on some targets.
The libm-test-ulps will need adjustment because..
- The argument reduction ideally uses nearest rounded rint, but that is
not efficient on most targets, so the polynomial can get evaluated on a
wider interval in non-nearest rounding mode making 1 ulp errors common
in that case.
- The polynomial is evaluated such that it may have 1 ulp error on
negative tiny inputs with upward rounding.
* math/Makefile (type-float-routines): Add math_errf and e_exp2f_data.
* sysdeps/aarch64/fpu/math_private.h (TOINT_INTRINSICS): Define.
(roundtoint, converttoint): Likewise.
* sysdeps/ieee754/flt-32/e_expf.c: New implementation.
* sysdeps/ieee754/flt-32/e_exp2f.c: New implementation.
* sysdeps/ieee754/flt-32/e_exp2f_data.c: New file.
* sysdeps/ieee754/flt-32/math_config.h: New file.
* sysdeps/ieee754/flt-32/math_errf.c: New file.
* sysdeps/ieee754/flt-32/t_exp2f.h: Remove.
* sysdeps/i386/fpu/e_exp2f_data.c: New file.
* sysdeps/i386/fpu/math_errf.c: New file.
* sysdeps/ia64/fpu/e_exp2f_data.c: New file.
* sysdeps/ia64/fpu/math_errf.c: New file.
* sysdeps/m68k/m680x0/fpu/e_exp2f_data.c: New file.
* sysdeps/m68k/m680x0/fpu/math_errf.c: New file.
sysdeps/ieee754/flt-32/e_exp2f.c declares two variable as "static
const volatile float". Maybe this use of "volatile" was originally
intended to inhibit optimization of underflowing / overflowing
operations such as TWOM100 * TWOM100; in any case, it's not currently
needed, as given -frounding-math constant folding of such expressions
is properly disabled when it would be unsafe. This patch removes the
unnecessary use of "volatile".
Tested for x86_64.
* sysdeps/ieee754/flt-32/e_exp2f.c (TWOM100): Remove volatile.
(TWO127): Likewise.
Various exp2 implementations in glibc can miss underflow exceptions
when the scaling down part of the calculation is exact (or, in the x86
case, when the conversion from extended precision to the target
precision is exact). This patch forces the exception in a similar way
to previous fixes.
The x86 exp2f changes may in fact not be needed for this purpose -
it's likely to be the case that no argument of type float has an exp2
result so close to an exact subnormal float value that it equals that
value when rounded to 64 bits (even taking account of variation
between different x86 implementations). However, they are included
for consistency with the changes to exp2 and so as to fix the exp2f
part of bug 18875 by ensuring that excess range and precision is
removed from underflowing return values.
Tested for x86_64, x86 and mips64.
[BZ #16521]
[BZ #18875]
* math/e_exp2l.c (__ieee754_exp2l): Force underflow exception for
small results.
* sysdeps/i386/fpu/e_exp2.S (dbl_min): New object.
(MO): New macro.
(__ieee754_exp2): For small results, force underflow exception and
remove excess range and precision from return value.
* sysdeps/i386/fpu/e_exp2f.S (flt_min): New object.
(MO): New macro.
(__ieee754_exp2f): For small results, force underflow exception
and remove excess range and precision from return value.
* sysdeps/i386/fpu/e_exp2l.S (ldbl_min): New object.
(MO): New macro.
(__ieee754_exp2l): Force underflow exception for small results.
* sysdeps/ieee754/dbl-64/e_exp2.c (__ieee754_exp2): Likewise.
* sysdeps/ieee754/flt-32/e_exp2f.c (__ieee754_exp2f): Likewise.
* sysdeps/x86_64/fpu/e_exp2l.S (ldbl_min): New object.
(MO): New macro.
(__ieee754_exp2l): Force underflow exception for small results.
* math/auto-libm-test-in: Add more tests or exp2.
* math/auto-libm-test-out: Regenerated.
The dbl-64 and flt-32 implementations of exp2 functions produce
spurious underflow exceptions. The underlying reason is the same in
both cases: the computation works as (2^a - 1)*2^b + 2^b for suitably
chosen a and b, where a has small magnitude so 2^a - 1 can be computed
with a low-degree polynomial approximation, and (2^a - 1)*2^b can
underflow even when the final result does not. This patch fixes this
by adjusting the threshold for when scaling is used to avoid
intermediate underflow so it works for any possible value of a where
the final result would not underflow.
Tested for x86_64 and x86.
[BZ #18219]
* sysdeps/ieee754/dbl-64/e_exp2.c (__ieee754_exp2): Reduce
threshold on absolute value of exponent for which scaling is used.
* sysdeps/ieee754/flt-32/e_exp2f.c (__ieee754_exp2f): Likewise.
* math/auto-libm-test-in: Add more tests of exp2.
* math/auto-libm-test-out: Regenerated.
This patch fixes the remaining part of bug 16560, spurious underflows
from exp2 of arguments close to 0 (when the result is close to 1, so
should not underflow), by just using 1+x instead of a more complicated
calculation when the argument is sufficiently small.
Tested for x86_64, x86 and mips64.
[BZ #16560]
* math/e_exp2l.c [LDBL_MANT_DIG == 106] (LDBL_EPSILON): Undefine
and redefine.
(__ieee754_exp2l): Do not multiply small fractional parts by
M_LN2l.
* sysdeps/i386/fpu/e_exp2l.S (__ieee754_exp2l): Just add 1 to
small argument.
* sysdeps/ieee754/dbl-64/e_exp2.c (__ieee754_exp2): Likewise.
* sysdeps/ieee754/flt-32/e_exp2f.c (__ieee754_exp2f): Likewise.
* sysdeps/x86_64/fpu/e_exp2l.S (__ieee754_exp2l): Likewise.
* math/auto-libm-test-in: Add more tests of exp2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/e_expl.c: Mark c0 and c1 as const.
* sysdeps/ieee754/flt-32/s_expm1f.c: Make sure huge and tiny end
up in .rodata.
* sysdeps/ieee754/flt-32/e_exp2f.c: Make sure TWO127 and TWOM100
end up in .rodata.
* sysdeps/ieee754/flt-32/e_expf.c: Likewise.
* sysdeps/ieee754/dbl-64/e_exp2.c: Make sure TWO1023 and TWOM1000
end up in .rodata.
2001-07-06 Paul Eggert <eggert@twinsun.com>
* manual/argp.texi: Remove ignored LGPL copyright notice; it's
not appropriate for documentation anyway.
* manual/libc-texinfo.sh: "Library General Public License" ->
"Lesser General Public License".
2001-07-06 Andreas Jaeger <aj@suse.de>
* All files under GPL/LGPL version 2: Place under LGPL version
2.1.
2001-06-04 Bruno Haible <haible@clisp.cons.org>
* iconv/loop.c (UNICODE_TAG_HANDLER): New macro.
* iconv/gconv_simple.c (__gconv_transform_internal_ascii): Invoke
UNICODE_TAG_HANDLER.
(__gconv_transform_internal_ucs2): Likewise.
(__gconv_transform_internal_ucs2reverse): Likewise.
* iconvdata/8bit-gap.c (BODY for TO_LOOP): Invoke UNICODE_TAG_HANDLER.
* iconvdata/8bit-generic.c (BODY for TO_LOOP): Likewise.
* iconvdata/ansi_x3.110.c (BODY for TO_LOOP): Likewise.
* iconvdata/big5.c (BODY for TO_LOOP): Likewise.
* iconvdata/big5hkscs.c (BODY for TO_LOOP): Likewise.
* iconvdata/cp1255.c (BODY for TO_LOOP): Likewise.
* iconvdata/cp1258.c (BODY for TO_LOOP): Likewise.
* iconvdata/euc-cn.c (BODY for TO_LOOP): Likewise.
* iconvdata/euc-jp.c (BODY for TO_LOOP): Likewise.
* iconvdata/euc-kr.c (BODY for TO_LOOP): Likewise.
* iconvdata/euc-tw.c (BODY for TO_LOOP): Likewise.
* iconvdata/gbk.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm930.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm932.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm933.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm935.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm937.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm939.c (BODY for TO_LOOP): Likewise.
* iconvdata/ibm943.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso646.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso8859-1.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso_6937.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso_6937-2.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso-2022-cn.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso-2022-cn-ext.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso-2022-kr.c (BODY for TO_LOOP): Likewise.
* iconvdata/johab.c (BODY for TO_LOOP): Likewise.
* iconvdata/sjis.c (BODY for TO_LOOP): Likewise.
* iconvdata/t.61.c (BODY for TO_LOOP): Likewise.
* iconvdata/uhc.c (BODY for TO_LOOP): Likewise.
* iconvdata/unicode.c (BODY for TO_LOOP): Likewise.
* iconvdata/iso-2022-jp.c (TAG_none, TAG_language, TAG_language_j,
TAG_language_ja, TAG_language_k, TAG_language_ko, TAG_language_z,
TAG_language_zh, CURRENT_TAG_MASK): New enum values.
(EMIT_SHIFT_TO_INIT): Don't emit an escape sequence if ASCII_set
is already selected but set2 or tag are set.
(conversion): New enum type.
(cvlist_t): New type.
(CVLIST, CVLIST_FIRST, CVLIST_REST): New macros.
(conversion_lists): New array.
(BODY for TO_LOOP): Keep track of Unicode 3.1 language tag. If "ja",
prefer conversion to Japanese character sets. If "zh", prefer
conversion to GB2312. If "ko", prefer conversion to KSC5601. Small
optimizations.
(INIT_PARAMS): Add tag.
(UPDATE_PARAMS): Add tag.
2001-06-04 Bruno Haible <haible@clisp.cons.org>
* locale/programs/locfile.c (write_locale_data): Before creat(),
unlink the file, to avoid crashing the processes that mmap it. Change
a double slash to a single slash. Free fname in case of error return.
2001-06-02 Jakub Jelinek <jakub@redhat.com>
* sysdeps/i386/fpu/s_frexpl.S (__frexpl): Mostly revert 2000-12-03
changes, do the special handling for denormal numbers, not for
normalized numbers (patch by <trevin@xmission.com>).
* math/test-misc.c (main): Test frexpl with denormal arguments.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* math/libm-test.inc (llround_test): Add two new llround tests.
* sysdeps/ieee754/ldbl-96/s_llroundl.c (__llroundl): Don't allow
overflow when rounding away from zero.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* math/Makefile (libm-calls): Add e_log2, w_log2, remove s_log2.
* math/math_private.h (__ieee754_log2, __ieee754_log2f,
__ieee754_log2l): New prototypes.
* sysdeps/generic/w_log2.c: New file.
* sysdeps/generic/w_log2f.c: New file.
* sysdeps/generic/w_log2l.c: New file.
* sysdeps/generic/s_log2l.c: Move...
* sysdeps/generic/e_log2l.c: ...to here. Rename to __ieee754_log2l.
* sysdeps/ieee754/k_standard.c (__kernel_standard): Handle log2(0)
and log2(x < 0).
* sysdeps/i386/fpu/s_log2.S: Move...
* sysdeps/i386/fpu/e_log2.S: ...to here. Rename to __ieee754_log2.
* sysdeps/i386/fpu/s_log2f.S: Move...
* sysdeps/i386/fpu/e_log2f.S: ...to here. Rename to __ieee754_log2f.
* sysdeps/i386/fpu/s_log2l.S: Move...
* sysdeps/i386/fpu/e_log2l.S: ...to here. Rename to __ieee754_log2l.
* sysdeps/m68k/fpu/s_log2.S: Move...
* sysdeps/m68k/fpu/e_log2.S: ...to here. Rename to __ieee754_log2.
* sysdeps/m68k/fpu/s_log2f.S: Move...
* sysdeps/m68k/fpu/e_log2f.S: ...to here. Rename to __ieee754_log2f.
* sysdeps/m68k/fpu/s_log2l.S: Move...
* sysdeps/m68k/fpu/e_log2l.S: ...to here. Rename to __ieee754_log2l.
* sysdeps/ieee754/dbl-64/s_log2.c: Move...
* sysdeps/ieee754/dbl-64/e_log2.c: ...to here. Rename to
__ieee754_log2.
* sysdeps/ieee754/flt-32/s_log2f.c: Move...
* sysdeps/ieee754/flt-32/e_log2f.c: ...to here. Rename to
__ieee754_log2f.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* sysdeps/generic/w_exp2.c (u_threshold): Lower threshold so that
even arguments which result in denormalized exp2 are accepted.
(__exp2): Arguments equal to u_threshold already result into
underflow.
* sysdeps/generic/w_exp2f.c (u_threshold, __exp2f): Likewise.
* sysdeps/generic/w_exp2l.c (u_threshold, __exp2l): Likewise.
* sysdeps/ieee754/dbl-64/e_exp2.c (__ieee754_exp2): Lomark was too
low, with corrected lowmark use greaterequal, not greater.
* sysdeps/ieee754/flt-32/e_exp2f.c (__ieee754_exp2f): Likewise.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* math/libm-test.inc (ilogb_test): Test that ilogb(+-Inf) == INT_MAX.
* sysdeps/i386/fpu/s_ilogb.S (__ilogb): Return INT_MAX for +-Inf.
* sysdeps/i386/fpu/s_ilogbf.S (__ilogbf): Likewise.
* sysdeps/i386/fpu/s_ilogbl.S (__ilogbl): Likewise.
* sysdeps/ieee754/dbl-64/s_ilogb.c (__ilogb): Likewise.
* sysdeps/ieee754/flt-32/s_ilogbf.c (__ilogbf): Likewise.
* sysdeps/ieee754/ldbl-128/s_ilogbl.c (__ilogbl): Likewise.
* sysdeps/ieee754/ldbl-96/s_ilogbl.c (__ilogbl): Likewise.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* sysdeps/generic/w_coshl.c (__coshl): Test if finite argument
gave non-finite result instead of using constant in generic
version.
* sysdeps/generic/w_coshf.c (__coshf): Likewise.
* sysdeps/generic/w_cosh.c (__cosh): Likewise.
* sysdeps/generic/w_exp10.c (o_threshold, u_threshold): Remove.
(__exp10): Test if finite argument gave non-finite result.
* sysdeps/generic/w_exp10f.c (o_threshold, u_threshold, __exp10f):
Likewise.
* sysdeps/generic/w_exp10l.c (o_threshold, u_threshold, __exp10l):
Likewise.
2001-06-04 Jakub Jelinek <jakub@redhat.com>
* sysdeps/ieee754/ldbl-96/e_coshl.c (__ieee754_coshl): Fix
overflow threshold constant (log(LDBL_MAX)+M_LN2l).
2001-05-29 Bruno Haible <haible@clisp.cons.org>
* locale/programs/ld-ctype.c (idx_table): New struct type.
(idx_table_init, idx_table_get, idx_table_add): New functions.
(MAX_CHARNAMES_IDX): Remove macro.
(locale_ctype_t): Change type of charnames_idx field.
(ctype_startup): Change initialization of charnames_idx field.
(find_idx): Use idx_table_get and idx_table_add for speed.
* locale/programs/charmap.c (charmap_new_char): Fix ucs4 value
computation of characters in a range.
2001-05-29 Bruno Haible <haible@clisp.cons.org>
* iconvdata/gb18030.c (__fourbyte_to_ucs1): Add mappings for <U03F4>,
<U03F5>.
(__ucs_to_gb18030_tab1): Likewise.
(BODY for FROM_LOOP): Add mapping for <U00010000>..<U0010FFFF>.
(BODY for TO_LOOP): Likewise.
* iconvdata/tst-table-charmap.sh: Update for charmaps containing
<U00xxxxxx> syntax.
* iconvdata/tst-table-from.c (bmp_only): New variable.
(utf8_decode): If bmp_only, don't return characters outside Unicode
plane 0.
(main): When testing UTF-8 or GB18030, set bmp_only to 1. Don't print
a conversion line if utf8_decode returns NULL.
* iconvdata/tst-table-to.c (main): When testing encodings other than
UTF-8 and GB18030, loop upto U+30000 instead of U+10000. Use UTF-8
instead of UCS-2 as input.
* iconvdata/tst-table.sh: For GB18030, use only the part < 0x10000
of the charmap.
2001-05-29 Bruno Haible <haible@clisp.cons.org>
* iconvdata/cns11643l1.c: Update to Unicode 3.1.
(__cns11643l1_to_ucs4_tab): Regenerated.
(__cns11643l1_from_ucs4_tab12): Regenerated.
* iconvdata/cns11643.c: Update to Unicode 3.1.
(__cns11643l14_to_ucs4_tab): Remove array.
(__cns11643l3_to_ucs4_tab, __cns11643l4_to_ucs4_tab,
__cns11643l5_to_ucs4_tab, __cns11643l6_to_ucs4_tab,
__cns11643l7_to_ucs4_tab, __cns11643l15_to_ucs4_tab): New arrays.
(__cns11643_from_ucs4p0_tab): Renamed from __cns11643_from_ucs4_tab.
(__cns11643_from_ucs4p2_tab): New array.
* iconvdata/cns11643.h (__cns11643l14_to_ucs4_tab): Remove declaration.
(__cns11643l3_to_ucs4_tab, __cns11643l4_to_ucs4_tab,
__cns11643l5_to_ucs4_tab, __cns11643l6_to_ucs4_tab,
__cns11643l7_to_ucs4_tab, __cns11643l15_to_ucs4_tab): New declarations.
(cns11643_to_ucs4): Treat planes 3, 4, 5, 6, 7, 15 instead of 14.
(__cns11643_from_ucs4_tab): Remove declaration.
(__cns11643_from_ucs4p0_tab, __cns11643_from_ucs4p2_tab): New
declarations.
(ucs4_to_cns11643): Update for new arrays. Treat U+3400..U+4DFF and
U+20000..U+2A6D6.
* iconvdata/cns11643l2.h (__cns11643_from_ucs4_tab): Remove
declaration.
(__cns11643_from_ucs4p0_tab): New declaration.
(ucs4_to_cns11643l2): Update for new arrays.
* iconvdata/iso-2022-cn-ext.c (BODY for FROM_LOOP): Handle planes
3 to 7.
(BODY for TO_LOOP): Handle planes 3 to 7, instead of plane 14.
* iconvdata/EUC-TW.irreversible: New file.
* iconvdata/tst-table.sh: Use it.
* iconvdata/Makefile (distribute): Add CP1255.irreversible,
CP1258.irreversible, EUC-TW.irreversible.
2001-05-29 Bruno Haible <haible@clisp.cons.org>
* locale/C-translit.h.in: Add transliterations for new Unicode 3.1
mathematical symbols.
