As reported in bug 32045, it's incorrect for strtod/nan functions to
set errno based on overflowing payload (strtod should only set errno
for overflow / underflow of its actual result, and potentially if
nothing in the string can be parsed as a number at all; nan should be
a pure function that never sets it). Save and restore errno around
the internal strtoull call and add associated test coverage.
Tested for x86_64.
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the exp2m1 and exp10m1 functions (exp2(x)-1 and
exp10(x)-1, like expm1).
As with other such functions, these use type-generic templates that
could be replaced with faster and more accurate type-specific
implementations in future. Test inputs are copied from those for
expm1, plus some additions close to the overflow threshold (copied
from exp2 and exp10) and also some near the underflow threshold.
exp2m1 has the unusual property of having an input (M_MAX_EXP) where
whether the function overflows (under IEEE semantics) depends on the
rounding mode. Although these could reasonably be XFAILed in the
testsuite (as we do in some cases for arguments very close to a
function's overflow threshold when an error of a few ulps in the
implementation can result in the implementation not agreeing with an
ideal one on whether overflow takes place - the testsuite isn't smart
enough to handle this automatically), since these functions aren't
required to be correctly rounding, I made the implementation check for
and handle this case specially.
The Makefile ordering expected by lint-makefiles for the new functions
is a bit peculiar, but I implemented it in this patch so that the test
passes; I don't know why log2 also needed moving in one Makefile
variable setting when it didn't in my previous patches, but the
failure showed a different place was expected for that function as
well.
The powerpc64le IFUNC setup seems not to be as self-contained as one
might hope; it shouldn't be necessary to add IFUNCs for new functions
such as these simply to get them building, but without setting up
IFUNCs for the new functions, there were undefined references to
__GI___expm1f128 (that IFUNC machinery results in no such function
being defined, but doesn't stop include/math.h from doing the
redirection resulting in the exp2m1f128 and exp10m1f128
implementations expecting to call it).
Tested for x86_64 and x86, and with build-many-glibcs.py.
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the log10p1 functions (log10(1+x): like log1p, but for
base-10 logarithms).
This is directly analogous to the log2p1 implementation (except that
whereas log2p1 has a smaller underflow range than log1p, log10p1 has a
larger underflow range). The test inputs are copied from those for
log1p and log2p1, plus a few more inputs in that wider underflow
range.
Tested for x86_64 and x86, and with build-many-glibcs.py.
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the logp1 functions (aliases for log1p functions - the
name is intended to be more consistent with the new log2p1 and
log10p1, where clearly it would have been very confusing to name those
functions log21p and log101p). As aliases rather than new functions,
the content of this patch is somewhat different from those actually
adding new functions.
Tests are shared with log1p, so this patch *does* mechanically update
all affected libm-test-ulps files to expect the same errors for both
functions.
The vector versions of log1p on aarch64 and x86_64 are *not* updated
to have logp1 aliases (and thus there are no corresponding header,
tests, abilist or ulps changes for vector functions either). It would
be reasonable for such vector aliases and corresponding changes to
other files to be made separately. For now, the log1p tests instead
avoid testing logp1 in the vector case (a Makefile change is needed to
avoid problems with grep, used in generating the .c files for vector
function tests, matching more than one ALL_RM_TEST line in a file
testing multiple functions with the same inputs, when it assumes that
the .inc file only has a single such line).
Tested for x86_64 and x86, and with build-many-glibcs.py.
It basically copy the already in place rules for dynamic tests for
auto-generated math functions for all support types. To avoid the
need to duplicate .inc files, a .SECONDEXPANSION rules is adeed for
the gen-libm-test.py generation.
New tests are added on the new rules 'libm-test-funcs-auto-static',
'libm-test-funcs-noauto-static', and 'libm-test-funcs-narrow-static';
similar to the non-static counterparts.
To avoid add extra build and disk requirement, the new math static
tests are only enable with a new define 'build-math-static-tests'.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the log2p1 functions (log2(1+x): like log1p, but for
base-2 logarithms).
This illustrates the intended structure of implementations of all
these function families: define them initially with a type-generic
template implementation. If someone wishes to add type-specific
implementations, it is likely such implementations can be both faster
and more accurate than the type-generic one and can then override it
for types for which they are implemented (adding benchmarks would be
desirable in such cases to demonstrate that a new implementation is
indeed faster).
The test inputs are copied from those for log1p. Note that these
changes make gen-auto-libm-tests depend on MPFR 4.2 (or later).
The bulk of the changes are fairly generic for any such new function.
(sysdeps/powerpc/nofpu/Makefile only needs changing for those
type-generic templates that use fabs.)
Tested for x86_64 and x86, and with build-many-glibcs.py.
Starting from glibc 2.1, crt1.o contains _IO_stdin_used which is checked
by _IO_check_libio to provide binary compatibility for glibc 2.0. Add
crt1-2.0.o for tests against glibc 2.0. Define tests-2.0 for glibc 2.0
compatibility tests. Add and update glibc 2.0 compatibility tests for
stderr, matherr and pthread_kill.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
The implementations of trunc functions using x87 floating point (i386 and
x86_64 long double only) traps when FE_INEXACT is enabled. Although
this is a GNU extension outside the scope of the C standard, other
architectures that also support traps do not show this behavior.
The fix moves the implementation to a common one that holds any
exceptions with a 'fnclex' (libc_feholdexcept_setround_387).
Checked on x86_64-linux-gnu and i686-linux-gnu.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
The implementations of floor functions using x87 floating point (i386 and
86_64 long double only) traps when FE_INEXACT is enabled. Although
this is a GNU extension outside the scope of the C standard, other
architectures that also support traps do not show this behavior.
The fix moves the implementation to a common one that holds any
exceptions with a 'fnclex' (libc_feholdexcept_setround_387).
Checked on x86_64-linux-gnu and i686-linux-gnu.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
The implementations of ceil functions using x87 floating point (i386 and
x86_64 long double only) traps when FE_INEXACT is enabled. Although
this is a GNU extension outside the scope of the C standard, other
architectures that also support traps do not show this behavior.
The fix moves the implementation to a common one that holds any
exceptions with a 'fnclex' (libc_feholdexcept_setround_387).
Checked on x86_64-linux-gnu and i686-linux-gnu.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Reflow all long lines adding comment terminators.
Sort all reflowed text using scripts/sort-makefile-lines.py.
No code generation changes observed in binary artifacts.
No regressions on x86_64 and i686.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
GCC 13 has added more _FloatN and _FloatNx versions of existing
<math.h> and <complex.h> built-in functions, for use in libstdc++-v3.
This breaks the glibc build because of how those functions are defined
as aliases to functions with the same ABI but different types. Add
appropriate -fno-builtin-* options for compiling relevant files, as
already done for the case of long double functions aliasing double
ones and based on the list of files used there.
I fixed some mistakes in that list of double files that I noticed
while implementing this fix, but there may well be more such
(harmless) cases, in this list or the new one (files that don't
actually exist or don't define the named functions as aliases so don't
need the options). I did try to exclude cases where glibc doesn't
define certain functions for _FloatN or _FloatNx types at all from the
new uses of -fno-builtin-* options. As with the options for double
files (see the commit message for commit
49348beafe, "Fix build with GCC 10 when
long double = double."), it's deliberate that the options are used
even if GCC currently doesn't have a built-in version of a given
functions, so providing some level of future-proofing against more
such built-in functions being added in future.
Tested with build-many-glibcs.py for aarch64-linux-gnu
powerpc-linux-gnu powerpc64le-linux-gnu x86_64-linux-gnu (compilers
and glibcs builds) with GCC mainline.
I used these shell commands:
../glibc/scripts/update-copyrights $PWD/../gnulib/build-aux/update-copyright
(cd ../glibc && git commit -am"[this commit message]")
and then ignored the output, which consisted lines saying "FOO: warning:
copyright statement not found" for each of 7061 files FOO.
I then removed trailing white space from math/tgmath.h,
support/tst-support-open-dev-null-range.c, and
sysdeps/x86_64/multiarch/strlen-vec.S, to work around the following
obscure pre-commit check failure diagnostics from Savannah. I don't
know why I run into these diagnostics whereas others evidently do not.
remote: *** 912-#endif
remote: *** 913:
remote: *** 914-
remote: *** error: lines with trailing whitespace found
...
remote: *** error: sysdeps/unix/sysv/linux/statx_cp.c: trailing lines
glibc has had exp10 functions since long before they were
standardized; now they are standardized in TS 18661-4 and C2X, they
are also specified there to have a corresponding type-generic macro.
Add one to <tgmath.h>, so fixing bug 26108.
glibc doesn't have other functions from TS 18661-4 yet, but when
added, it will be natural to add the type-generic macro for each
function family at the same time as the functions.
Tested for x86_64.
C2X adds new <math.h> functions for floating-point maximum and
minimum, corresponding to the new operations that were added in IEEE
754-2019 because of concerns about the old operations not being
associative in the presence of signaling NaNs. fmaximum and fminimum
handle NaNs like most <math.h> functions (any NaN argument means the
result is a quiet NaN). fmaximum_num and fminimum_num handle both
quiet and signaling NaNs the way fmax and fmin handle quiet NaNs (if
one argument is a number and the other is a NaN, return the number),
but still raise "invalid" for a signaling NaN argument, making them
exceptions to the normal rule that a function with a floating-point
result raising "invalid" also returns a quiet NaN. fmaximum_mag,
fminimum_mag, fmaximum_mag_num and fminimum_mag_num are corresponding
functions returning the argument with greatest or least absolute
value. All these functions also treat +0 as greater than -0. There
are also corresponding <tgmath.h> type-generic macros.
Add these functions to glibc. The implementations use type-generic
templates based on those for fmax, fmin, fmaxmag and fminmag, and test
inputs are based on those for those functions with appropriate
adjustments to the expected results. The RISC-V maintainers might
wish to add optimized versions of fmaximum_num and fminimum_num (for
float and double), since RISC-V (F extension version 2.2 and later)
provides instructions corresponding to those functions - though it
might be at least as useful to add architecture-independent built-in
functions to GCC and teach the RISC-V back end to expand those
functions inline, which is what you generally want for functions that
can be implemented with a single instruction.
Tested for x86_64 and x86, and with build-many-glibcs.py.
This patch adds the narrowing fused multiply-add functions from TS
18661-1 / TS 18661-3 / C2X to glibc's libm: ffma, ffmal, dfmal,
f32fmaf64, f32fmaf32x, f32xfmaf64 for all configurations; f32fmaf64x,
f32fmaf128, f64fmaf64x, f64fmaf128, f32xfmaf64x, f32xfmaf128,
f64xfmaf128 for configurations with _Float64x and _Float128;
__f32fmaieee128 and __f64fmaieee128 aliases in the powerpc64le case
(for calls to ffmal and dfmal when long double is IEEE binary128).
Corresponding tgmath.h macro support is also added.
The changes are mostly similar to those for the other narrowing
functions previously added, especially that for sqrt, so the
description of those generally applies to this patch as well. As with
sqrt, I reused the same test inputs in auto-libm-test-in as for
non-narrowing fma rather than adding extra or separate inputs for
narrowing fma. The tests in libm-test-narrow-fma.inc also follow
those for non-narrowing fma.
The non-narrowing fma has a known bug (bug 6801) that it does not set
errno on errors (overflow, underflow, Inf * 0, Inf - Inf). Rather
than fixing this or having narrowing fma check for errors when
non-narrowing does not (complicating the cases when narrowing fma can
otherwise be an alias for a non-narrowing function), this patch does
not attempt to check for errors from narrowing fma and set errno; the
CHECK_NARROW_FMA macro is still present, but as a placeholder that
does nothing, and this missing errno setting is considered to be
covered by the existing bug rather than needing a separate open bug.
missing-errno annotations are duly added to many of the
auto-libm-test-in test inputs for fma.
This completes adding all the new functions from TS 18661-1 to glibc,
so will be followed by corresponding stdc-predef.h changes to define
__STDC_IEC_60559_BFP__ and __STDC_IEC_60559_COMPLEX__, as the support
for TS 18661-1 will be at a similar level to that for C standard
floating-point facilities up to C11 (pragmas not implemented, but
library functions done). (There are still further changes to be done
to implement changes to the types of fromfp functions from N2548.)
Tested as followed: natively with the full glibc testsuite for x86_64
(GCC 11, 7, 6) and x86 (GCC 11); with build-many-glibcs.py with GCC
11, 7 and 6; cross testing of math/ tests for powerpc64le, powerpc32
hard float, mips64 (all three ABIs, both hard and soft float). The
different GCC versions are to cover the different cases in tgmath.h
and tgmath.h tests properly (GCC 6 has _Float* only as typedefs in
glibc headers, GCC 7 has proper _Float* support, GCC 8 adds
__builtin_tgmath).
This patch adds the narrowing square root functions from TS 18661-1 /
TS 18661-3 / C2X to glibc's libm: fsqrt, fsqrtl, dsqrtl, f32sqrtf64,
f32sqrtf32x, f32xsqrtf64 for all configurations; f32sqrtf64x,
f32sqrtf128, f64sqrtf64x, f64sqrtf128, f32xsqrtf64x, f32xsqrtf128,
f64xsqrtf128 for configurations with _Float64x and _Float128;
__f32sqrtieee128 and __f64sqrtieee128 aliases in the powerpc64le case
(for calls to fsqrtl and dsqrtl when long double is IEEE binary128).
Corresponding tgmath.h macro support is also added.
The changes are mostly similar to those for the other narrowing
functions previously added, so the description of those generally
applies to this patch as well. However, the not-actually-narrowing
cases (where the two types involved in the function have the same
floating-point format) are aliased to sqrt, sqrtl or sqrtf128 rather
than needing a separately built not-actually-narrowing function such
as was needed for add / sub / mul / div. Thus, there is no
__nldbl_dsqrtl name for ldbl-opt because no such name was needed
(whereas the other functions needed such a name since the only other
name for that entry point was e.g. f32xaddf64, not reserved by TS
18661-1); the headers are made to arrange for sqrt to be called in
that case instead.
The DIAG_* calls in sysdeps/ieee754/soft-fp/s_dsqrtl.c are because
they were observed to be needed in GCC 7 testing of
riscv32-linux-gnu-rv32imac-ilp32. The other sysdeps/ieee754/soft-fp/
files added didn't need such DIAG_* in any configuration I tested with
build-many-glibcs.py, but if they do turn out to be needed in more
files with some other configuration / GCC version, they can always be
added there.
I reused the same test inputs in auto-libm-test-in as for
non-narrowing sqrt rather than adding extra or separate inputs for
narrowing sqrt. The tests in libm-test-narrow-sqrt.inc also follow
those for non-narrowing sqrt.
Tested as followed: natively with the full glibc testsuite for x86_64
(GCC 11, 7, 6) and x86 (GCC 11); with build-many-glibcs.py with GCC
11, 7 and 6; cross testing of math/ tests for powerpc64le, powerpc32
hard float, mips64 (all three ABIs, both hard and soft float). The
different GCC versions are to cover the different cases in tgmath.h
and tgmath.h tests properly (GCC 6 has _Float* only as typedefs in
glibc headers, GCC 7 has proper _Float* support, GCC 8 adds
__builtin_tgmath).
Some math functions (such as __isnan*) are built into both libm and
libc because they are needed in libc. The symbol gets exported from
libc.so and not libm.so, because of which dynamic linking works fine;
the symbols are always resolved from libc.so and libm.so uses its
internal copy of the same function if needed.
When linking statically though, the libm variants get used throughout
because the symbols are exported in both archives and libm.a is
searched first.
This patch removes these duplicate objects from the libm.a archive so
that programs always link to libc in both, the static and dynamic
case. The difference this will cause is that libm uses of these
functions will start using the libc versions in the !SHARED case.
This is harmless at the moment because the objects are identical
except for their names.
Some of these duplicates could be removed from libm.so too, but I
avoided that in the interest of retaining an internal reference if at
all those functions get used within libm in future.
Reviewed-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Finally remove all mpa related files, headers, declarations, probes, unused
tables and update makefiles.
Reviewed-By: Paul Zimmermann <Paul.Zimmermann@inria.fr>
compat_symbol_reference is now available without tests-internal.
Do not build the test at all on glibc versions that lack the symbols,
to avoid spurious UNSUPPORTED results.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
tests-internal is no longer needed because compat_symbol_reference
now works in regular tests.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
I used these shell commands:
../glibc/scripts/update-copyrights $PWD/../gnulib/build-aux/update-copyright
(cd ../glibc && git commit -am"[this commit message]")
and then ignored the output, which consisted lines saying "FOO: warning:
copyright statement not found" for each of 6694 files FOO.
I then removed trailing white space from benchtests/bench-pthread-locks.c
and iconvdata/tst-iconv-big5-hkscs-to-2ucs4.c, to work around this
diagnostic from Savannah:
remote: *** pre-commit check failed ...
remote: *** error: lines with trailing whitespace found
remote: error: hook declined to update refs/heads/master
Similar to string2.h (18b10de7ce) and string3.h (09a596cc2c) this
patch removes the fenvinline.h on all architectures. Currently
only powerpc implements some optimizations. This kind of optimization
is better implemented by the compiler (which handles the architecture
ISA transparently).
Also, for the specific optimized powerpc implementation the code is
becoming convoluted and these micro-optimization are hardly wildly
used, even more being a possible hotspot in realword cases
(non-default rounding are used only on specific cases and exception
handling are done most likely only on errors path). Only x86
implements similar optimization (on fenv.h) also indicates that
these should no be on libc.
The math/test-fenv already covers all math/test-fenvinline tests,
so it is safe to remove it.
The powerpc fegetround optimization is moved to internal
fenv_libc.h.
The BZ#94193 [1] the corresponding GCC bug for adding replacements
for these on powerpc.
Checked on x86_64-linux-gnu and powerpc64le-linux-gnu.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94193
With mathinline removal there is no need to keep building and testing
inline math tests.
The gen-libm-tests.py support to generate ULP_I_* is removed and all
libm-test-ulps files are updated to longer have the
i{float,double,ldouble} entries. The support for no-test-inline is
also removed from both gen-auto-libm-tests and the
auto-libm-test-out-* were regenerated.
Checked on x86_64-linux-gnu and i686-linux-gnu.
On platforms where long double has the same ABI as double, glibc
defines long double functions as aliases for the corresponding double
functions. The declarations of those functions in <math.h> are
disabled to avoid problems with aliases having incompatible types, but
GCC 10 now gives errors for incompatible types when the long double
function is known to GCC as a built-in function, not just when there
is an incompatible header declaration.
This patch fixes those errors by using appropriate
-fno-builtin-<function> options to compile the double functions. The
list of CFLAGS-* settings is an appropriately adapted version of that
in sysdeps/ieee754/ldbl-opt/Makefile used there for building nldbl-*.c
files; in particular, the options are used even if GCC does not
currently have a built-in function of a given function, so that adding
such a built-in function in future will not break the glibc build.
Thus, various of the CFLAGS-* settings are only for future-proofing
and may not currently be needed (and it's possible some could be
irrelevant for other reasons).
Tested with build-many-glibcs.py for arm-linux-gnueabi (compilers and
glibcs builds), where it fixes the build that previously failed.
This patch creates test-ibm128* tests from the long double function tests.
In order to explicitly test IBM long double functions -mabi=ibmlongdouble is
added to CFLAGS.
Likewise, update the test headers to correct choose ULPs when redirects
are enabled.
Co-authored-by: Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
Co-authored-by: Paul E. Murphy <murphyp@linux.vnet.ibm.com>
This is a preparatory patch to enable building a _Float128
variant to ease reuse when building a _Float128 variant to
alias this long double only symbol.
Notably, stubs are added where missing to the native _Float128
sysdep dir to prevent building these newly templated variants
created inside the build directories.
Also noteworthy are the changes around LIBM_SVID_COMPAT. These
changes are not intuitive. The templated version is only
enabled when !LIBM_SVID_COMPAT, and the compat version is
predicated entirely on LIBM_SVID_COMPAT. Thus, exactly one is
stubbed out entirely when building. The nldbl scalb compat
files are updated to account for this.
Likewise, fixup the reuse of m68k's e_scalb{f,l}.c to include
it's override of e_scalb.c. Otherwise, the search path finds
the templated copy in the build directory. This could be
futher simplified by providing an overridden template, but I
lack the hardware to verify.
Remove _finite tests and references from x86_64. Rather than calling
__exp_finite, use exp directly (since it's the same entry point).
x86_64 builds and passes testsuite.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Remove the finite-math tests from the testsuite - these are no longer
useful after removing math-finite.h header.
Passes buildmanyglibc, build&test on x86_64 and AArch64.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
When adding some of the TS 18661 narrowing functions for glibc 2.28, I
deferred adding corresponding <tgmath.h> support because of unresolved
questions about the specification for those type-generic macros,
especially in relation to _FloatN and _FloatNx types.
Those issues are now clarified in the response to Clarification
Request 13 to TS 18661-3, and this patch adds the deferred tgmath.h
support. As with other tgmath.h macros, there are fairly
straightforward implementations based on __builtin_tgmath for GCC 8
and later, which result in exactly the right function being called in
each case, and more complicated implementations for GCC 7 and earlier,
which generally result in a function being called whose arguments have
the right format (i.e. an alias for the right function), but which
might not be exactly the function name specified by TS 18661.
In one case with older compilers (f32x* macros, where the type
_Float64x exists and all the arguments have type _Float32 or
_Float32x), there is a further relaxation and the function called may
have arguments narrower than the one specified by the TS, but still
wide enough to represent the arguments exactly, so the result of the
call is unchanged (as this does not affect any case where rounding of
integer arguments might be involved). With GCC 6 or before this is
inherently unavoidable (but still harmless and not detectable by how
the compiled program behaves, unless it redefines the functions in
question like the testcases do) because _Float32x and _Float64 are
both typedefs for double in that case but the specified semantics
result in different functions, with different argument formats, being
called for those two argument types.
Tests for the new macros are handled through gen-tgmath-tests.py,
which deals with the special-case handling for older GCC.
Tested as follows: with the full glibc testsuite on x86_64 and x86
(with GCC 6, 7 and 8); with the math/ tests on aarch64 and arm (with
GCC 6, 7 and 8); with build-many-glibcs.py (with GCC 6, 7 and 9).
* math/tgmath.h [__HAVE_FLOAT128X]: Give error.
[(__HAVE_FLOAT64X && !__HAVE_FLOAT128)
|| (__HAVE_FLOAT128 && !__HAVE_FLOAT64X)]: Likewise.
(__TGMATH_2_NARROW_F): Likewise.
(__TGMATH_2_NARROW_D): New macro.
(__TGMATH_2_NARROW_F16): Likewise.
(__TGMATH_2_NARROW_F32): Likewise.
(__TGMATH_2_NARROW_F64): Likewise.
(__TGMATH_2_NARROW_F32X): Likewise.
(__TGMATH_2_NARROW_F64X): Likewise.
[__HAVE_BUILTIN_TGMATH] (__TGMATH_NARROW_FUNCS_F): Likewise.
[__HAVE_BUILTIN_TGMATH] (__TGMATH_NARROW_FUNCS_F16): Likewise.
[__HAVE_BUILTIN_TGMATH] (__TGMATH_NARROW_FUNCS_F32): Likewise.
[__HAVE_BUILTIN_TGMATH] (__TGMATH_NARROW_FUNCS_F64): Likewise.
[__HAVE_BUILTIN_TGMATH] (__TGMATH_NARROW_FUNCS_F32X): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (fadd): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (dadd): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (fdiv): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (ddiv): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (fmul): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (dmul): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (fsub): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT_C2X)] (dsub): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT16] (f16add):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT16] (f16div):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT16] (f16mul):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT16] (f16sub):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32] (f32add):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32] (f32div):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32] (f32mul):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32] (f32sub):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64
&& (__HAVE_FLOAT64X || __HAVE_FLOAT128)] (f64add): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64
&& (__HAVE_FLOAT64X || __HAVE_FLOAT128)] (f64div): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64
&& (__HAVE_FLOAT64X || __HAVE_FLOAT128)] (f64mul): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64
&& (__HAVE_FLOAT64X || __HAVE_FLOAT128)] (f64sub): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32X] (f32xadd):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32X] (f32xdiv):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32X] (f32xmul):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT32X] (f32xsub):
Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64X
&& (__HAVE_FLOAT128X || __HAVE_FLOAT128)] (f64xadd): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64X
&& (__HAVE_FLOAT128X || __HAVE_FLOAT128)] (f64xdiv): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64X
&& (__HAVE_FLOAT128X || __HAVE_FLOAT128)] (f64xmul): Likewise.
[__GLIBC_USE (IEC_60559_TYPES_EXT) && __HAVE_FLOAT64X
&& (__HAVE_FLOAT128X || __HAVE_FLOAT128)] (f64xsub): Likewise.
* math/gen-tgmath-tests.py (Type): Add members
non_standard_real_argument_types_list, long_double_type,
complex_float64_type and float32x_ext_type.
(Type.__init__): Set the new members.
(Type.floating_type): Add new argument floatn.
(Type.real_floating_type): Likewise.
(Type.can_combine_types): Likewise.
(Type.combine_types): Likewise.
(Type.init_types): Create internal Float32x_ext type.
(Tests.__init__): Define Float32x_ext in generated C code.
(Tests.add_tests): Handle narrowing functions.
(Tests.add_all_tests): Likewise.
(Tests.tests_text): Allow variation in mant_dig for narrowing
functions with compilers before GCC 8.
* math/Makefile (tgmath3-narrow-types): New variable.
(tgmath3-narrow-macros): Likewise.
(tgmath3-macros): Add $(tgmath3-narrow-macros).
The resolution of C floating-point Clarification Request 25
<http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2397.htm#dr_25> is
that the totalorder and totalordermag functions should take pointer
arguments, and this has been adopted in C2X (with const added; note
that the integration of this change into C2X is present in the C
standard git repository but postdates the most recent public PDF
draft).
This patch updates glibc accordingly. As a defect resolution, the API
is changed unconditionally rather than supporting any sort of TS
18661-1 mode for compilation with the old version of the API. There
are compat symbols for existing binaries that pass floating-point
arguments directly. As a consequence of changing to pointer
arguments, there are no longer type-generic macros in tgmath.h for
these functions.
Because of the fairly complicated logic for creating libm function
aliases and determining the set of aliases to create in a given glibc
configuration, rather than duplicating all that in individual source
files to create the versioned and compat symbols, the source files for
the various versions of totalorder functions are set up to redefine
weak_alias before using libm_alias_* macros to create the symbols
required. In turn, this requires creating a separate alias for each
symbol version pointing to the same implementation (see binutils bug
<https://sourceware.org/bugzilla/show_bug.cgi?id=23840>), which is
done automatically using __COUNTER__. (As I noted in
<https://sourceware.org/ml/libc-alpha/2018-10/msg00631.html>, it might
well make sense for glibc's symbol versioning macros to do that alias
creation with __COUNTER__ themselves, which would somewhat simplify
the logic in the totalorder source files.)
It is of course desirable to test the compat symbols. I did this with
the generic libm-test machinery, but didn't wish to duplicate the
actual tables of test inputs and outputs, and thought it risky to
attempt to have a single object file refer to both default and compat
versions of the same function in order to test them together. Thus, I
created libm-test-compat_totalorder.inc and
libm-test-compat_totalordermag.inc which include the generated .c
files (with the processed version of those tables of inputs) from the
non-compat tests, and added appropriate dependencies. I think this
provides sufficient test coverage for the compat symbols without also
needing to make the special ldbl-96 and ldbl-128ibm tests (of
peculiarities relating to the representations of those formats that
can't be covered in the generic tests) run for the compat symbols.
Tests of compat symbols need to be internal tests, meaning _ISOMAC is
not defined. Making some libm-test tests into internal tests showed
up two other issues. GCC diagnoses duplicate macro definitions of
__STDC_* macros, including __STDC_WANT_IEC_60559_TYPES_EXT__; I added
an appropriate conditional and filed
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91451> for this issue.
On ia64, include/setjmp.h ends up getting included indirectly from
libm-symbols.h, resulting in conflicting definitions of the STR macro
(also defined in libm-test-driver.c); I renamed the macros in
include/setjmp.h. (It's arguable that we should have common internal
headers used everywhere for stringizing and concatenation macros.)
Tested for x86_64 and x86, and with build-many-glibcs.py.
* math/bits/mathcalls.h
[__GLIBC_USE (IEC_60559_BFP_EXT) || __MATH_DECLARING_FLOATN]
(totalorder): Take pointer arguments.
[__GLIBC_USE (IEC_60559_BFP_EXT) || __MATH_DECLARING_FLOATN]
(totalordermag): Likewise.
* manual/arith.texi (totalorder): Likewise.
(totalorderf): Likewise.
(totalorderl): Likewise.
(totalorderfN): Likewise.
(totalorderfNx): Likewise.
(totalordermag): Likewise.
(totalordermagf): Likewise.
(totalordermagl): Likewise.
(totalordermagfN): Likewise.
(totalordermagfNx): Likewise.
* math/tgmath.h (__TGMATH_BINARY_REAL_RET_ONLY): Remove macro.
[__GLIBC_USE (IEC_60559_BFP_EXT)] (totalorder): Likewise.
[__GLIBC_USE (IEC_60559_BFP_EXT)] (totalordermag): Likewise.
* math/Versions (GLIBC_2.31): Add totalorder, totalorderf,
totalorderl, totalordermag, totalordermagf, totalordermagl,
totalorderf32, totalorderf64, totalorderf32x, totalordermagf32,
totalordermagf64, totalordermagf32x, totalorderf64x,
totalordermagf64x, totalorderf128 and totalordermagf128.
* math/Makefile (libm-test-funcs-noauto): Add compat_totalorder
and compat_totalordermag.
(libm-test-funcs-compat): New variable.
(libm-tests-compat): Likewise.
(tests): Do not include compat tests.
(tests-internal): Add compat tests.
($(foreach t,$(libm-tests-base),
$(objpfx)$(t)-compat_totalorder.o)): Depend
on $(objpfx)libm-test-totalorder.c.
($(foreach t,$(libm-tests-base),
$(objpfx)$(t)-compat_totalordermag.o): Depend on
$(objpfx)libm-test-totalordermag.c.
(tgmath3-macros): Remove totalorder and totalordermag.
* math/libm-test-compat_totalorder.inc: New file.
* math/libm-test-compat_totalordermag.inc: Likewise.
* math/libm-test-driver.c (struct test_ff_i_data): Update comment.
(RUN_TEST_fpfp_b): New macro.
(RUN_TEST_LOOP_fpfp_b): Likewise.
* math/libm-test-totalorder.inc (totalorder_test_data): Use
TEST_fpfp_b.
(totalorder_test): Condition on [!COMPAT_TEST].
(do_test): Likewise.
* math/libm-test-totalordermag.inc (totalordermag_test_data): Use
TEST_fpfp_b.
(totalordermag_test): Condition on [!COMPAT_TEST].
(do_test): Likewise.
* math/gen-tgmath-tests.py (Tests.add_all_tests): Remove
totalorder and totalordermag.
* math/test-tgmath.c (NCALLS): Change to 132.
(F(compile_test)): Do not call totalorder or totalordermag.
(F(totalorder)): Remove.
(F(totalordermag)): Likewise.
* include/float.h (__STDC_WANT_IEC_60559_TYPES_EXT__): Do not
define if [__STDC_WANT_IEC_60559_TYPES_EXT__].
* include/setjmp.h [!_ISOMAC] (STR_HELPER): Rename to
SJSTR_HELPER.
[!_ISOMAC] (STR): Rename to SJSTR. Update call to STR_HELPER.
[!_ISOMAC] (TEST_SIZE): Update call to STR.
[!_ISOMAC] (TEST_ALIGN): Likewise.
[!_ISOMAC] (TEST_OFFSET): Likewise.
* sysdeps/ieee754/dbl-64/s_totalorder.c: Include <shlib-compat.h>
and <first-versions.h>.
(__totalorder): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/dbl-64/s_totalordermag.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalordermag): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/dbl-64/wordsize-64/s_totalorder.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalorder): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/dbl-64/wordsize-64/s_totalordermag.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalordermag): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/float128/float128_private.h
(__totalorder_compatl): New macro.
(__totalordermag_compatl): Likewise.
* sysdeps/ieee754/flt-32/s_totalorderf.c: Include <shlib-compat.h>
and <first-versions.h>.
(__totalorderf): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/flt-32/s_totalordermagf.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalordermagf): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/ldbl-128/s_totalorderl.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalorderl): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/ldbl-128/s_totalordermagl.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalordermagl): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/ldbl-128ibm/s_totalorderl.c: Include
<shlib-compat.h>.
(__totalorderl): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/ldbl-128ibm/s_totalordermagl.c: Include
<shlib-compat.h>.
(__totalordermagl): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/ldbl-96/s_totalorderl.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalorderl): Take pointer arguments. Add symbol versions and
compat symbols.
* sysdeps/ieee754/ldbl-96/s_totalordermagl.c: Include
<shlib-compat.h> and <first-versions.h>.
(__totalordermagl): Take pointer arguments. Add symbol versions
and compat symbols.
* sysdeps/ieee754/ldbl-opt/nldbl-totalorder.c (totalorderl): Take
pointer arguments.
* sysdeps/ieee754/ldbl-opt/nldbl-totalordermag.c (totalordermagl):
Likewise.
* sysdeps/ieee754/ldbl-128ibm/test-totalorderl-ldbl-128ibm.c
(do_test): Update calls to totalorderl and totalordermagl.
* sysdeps/ieee754/ldbl-96/test-totalorderl-ldbl-96.c (do_test):
Update calls to totalorderl and totalordermagl.
* sysdeps/mach/hurd/i386/libm.abilist: Update.
* sysdeps/unix/sysv/linux/aarch64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/alpha/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/arm/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/csky/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/hppa/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/i386/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/ia64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/m68k/coldfire/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/m68k/m680x0/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/microblaze/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/mips/mips64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/nios2/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist:
Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist:
Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/be/libm.abilist:
Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libm.abilist:
Likewise.
* sysdeps/unix/sysv/linux/riscv/rv64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/sh/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc32/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/x86_64/64/libm.abilist: Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist: Likewise.
This patch makes Python 3.4 or later a required tool for building
glibc, so allowing changes of awk, perl etc. code used in the build
and test to Python code without any such changes needing makefile
conditionals or to handle older Python versions.
This patch makes the configure test for Python check the version and
give an error if Python is missing or too old, and removes makefile
conditionals that are no longer needed. It does not itself convert
any code from another language to Python, and does not remove any
compatibility with older Python versions from existing scripts.
Tested for x86_64.
* configure.ac (PYTHON_PROG): Use AC_CHECK_PROG_VER. Set
critic_missing for versions before 3.4.
* configure: Regenerated.
* manual/install.texi (Tools for Compilation): Document
requirement for Python to build glibc.
* INSTALL: Regenerated.
* Rules [PYTHON]: Make code unconditional.
* benchtests/Makefile [PYTHON]: Likewise.
* conform/Makefile [PYTHON]: Likewise.
* manual/Makefile [PYTHON]: Likewise.
* math/Makefile [PYTHON]: Likewise.
The algorithm is exp(y * log(x)), where log(x) is computed with about
1.3*2^-68 relative error (1.5*2^-68 without fma), returning the result
in two doubles, and the exp part uses the same algorithm (and lookup
tables) as exp, but takes the input as two doubles and a sign (to handle
negative bases with odd integer exponent). The __exp1 internal symbol
is no longer necessary.
There is separate code path when fma is not available but the worst case
error is about 0.54 ULP in both cases. The lookup table and consts for
log are 4168 bytes. The .rodata+.text is decreased by 37908 bytes on
aarch64. The non-nearest rounding error is less than 1 ULP.
Improvements on Cortex-A72 compared to current glibc master:
pow thruput: 2.40x in [0.01 11.1]x[0.01 11.1]
pow latency: 1.84x in [0.01 11.1]x[0.01 11.1]
Tested on
aarch64-linux-gnu (defined __FP_FAST_FMA, TOINT_INTRINSICS) and
arm-linux-gnueabihf (!defined __FP_FAST_FMA, !TOINT_INTRINSICS) and
x86_64-linux-gnu (!defined __FP_FAST_FMA, !TOINT_INTRINSICS) and
powerpc64le-linux-gnu (defined __FP_FAST_FMA, !TOINT_INTRINSICS) targets.
* NEWS: Mention pow improvements.
* math/Makefile (type-double-routines): Add e_pow_log_data.
* sysdeps/generic/math_private.h (__exp1): Remove.
* sysdeps/i386/fpu/e_pow_log_data.c: New file.
* sysdeps/ia64/fpu/e_pow_log_data.c: New file.
* sysdeps/ieee754/dbl-64/Makefile (CFLAGS-e_pow.c): Allow fma
contraction.
* sysdeps/ieee754/dbl-64/e_exp.c (__exp1): Remove.
(exp_inline): Remove.
(__ieee754_exp): Only single double input is handled.
* sysdeps/ieee754/dbl-64/e_pow.c: Rewrite.
* sysdeps/ieee754/dbl-64/e_pow_log_data.c: New file.
* sysdeps/ieee754/dbl-64/math_config.h (issignaling_inline): Define.
(__pow_log_data): Define.
* sysdeps/ieee754/dbl-64/upow.h: Remove.
* sysdeps/ieee754/dbl-64/upow.tbl: Remove.
* sysdeps/m68k/m680x0/fpu/e_pow_log_data.c: New file.
* sysdeps/x86_64/fpu/multiarch/Makefile (CFLAGS-e_pow-fma.c): Allow fma
contraction.
(CFLAGS-e_pow-fma4.c): Likewise.
Similar algorithm is used as in log: log2(2^k x) = k + log2(c) + log2(x/c)
where the last term is approximated by a polynomial of x/c - 1, the first
order coefficient is about 1/ln2 in this case.
There is separate code path when fma instruction is not available for
computing x/c - 1 precisely, for which the table size is doubled.
The worst case error is 0.547 ULP (0.55 without fma), the read only
global data size is 1168 bytes (2192 without fma) on aarch64. The
non-nearest rounding error is less than 1 ULP.
Improvements on Cortex-A72 compared to current glibc master:
log2 thruput: 2.00x in [0.01 11.1]
log2 latency: 2.04x in [0.01 11.1]
log2 thruput: 2.17x in [0.999 1.001]
log2 latency: 2.88x in [0.999 1.001]
Tested on
aarch64-linux-gnu (defined __FP_FAST_FMA)
arm-linux-gnueabihf (!defined __FP_FAST_FMA)
x86_64-linux-gnu (!defined __FP_FAST_FMA)
powerpc64le-linxu-gnu (defined __FP_FAST_FMA)
targets.
* NEWS: Mention log2 improvements.
* math/Makefile (type-double-routines): Add e_log2_data.
* sysdeps/i386/fpu/e_log2_data.c: New file.
* sysdeps/ia64/fpu/e_log2_data.c: New file.
* sysdeps/ieee754/dbl-64/e_log2.c: Rewrite.
* sysdeps/ieee754/dbl-64/e_log2_data.c: New file.
* sysdeps/ieee754/dbl-64/math_config.h (__log2_data): Add.
* sysdeps/ieee754/dbl-64/wordsize-64/e_log2.c: Remove.
* sysdeps/m68k/m680x0/fpu/e_log2_data.c: New file.
Optimized log using carefully generated lookup table with 1/c and log(c)
values for small intervalls around 1. The log(c) is very near a double
precision value, it has about 62 bits precision. The algorithm is
log(2^k x) = k log(2) + log(c) + log(x/c), where the last term is
approximated by a polynomial of x/c - 1. Near 1 a single polynomial of
x - 1 is used.
There is separate code path when fma instruction is not available for
computing x/c - 1 precisely, in which case the table size is doubled.
The code uses __builtin_fma under __FP_FAST_FMA to ensure it is inlined
as an instruction.
With the default configuration settings the worst case error is 0.519 ULP
(and 0.520 without fma), the rodata size is 2192 bytes (4240 without fma).
The non-nearest rounding error is less than 1 ULP.
Improvements on Cortex-A72 compared to current glibc master:
log thruput: 3.28x in [0.01 11.1]
log latency: 2.23x in [0.01 11.1]
log thruput: 1.56x in [0.999 1.001]
log latency: 1.57x in [0.999 1.001]
Tested on
aarch64-linux-gnu (defined __FP_FAST_FMA)
arm-linux-gnueabihf (!defined __FP_FAST_FMA)
x86_64-linux-gnu (!defined __FP_FAST_FMA)
powerpc64le-linux-gnu (defined __FP_FAST_FMA)
targets.
* NEWS: Mention log improvement.
* math/Makefile (type-double-routines): Add e_log_data.
* sysdeps/i386/fpu/e_log_data.c: New file.
* sysdeps/ia64/fpu/e_log_data.c: New file.
* sysdeps/ieee754/dbl-64/e_log.c: Rewrite.
* sysdeps/ieee754/dbl-64/e_log_data.c: New file.
* sysdeps/ieee754/dbl-64/math_config.h (__log_data): Add.
* sysdeps/ieee754/dbl-64/ulog.h: Remove.
* sysdeps/ieee754/dbl-64/ulog.tbl: Remove.
* sysdeps/m68k/m680x0/fpu/e_log_data.c: New file.
Optimized exp and exp2 implementations using a lookup table for
fractional powers of 2. There are several variants, see e_exp_data.c,
they can be selected by modifying math_config.h allowing different
tradeoffs.
The default selection should be acceptable as generic libm code.
Worst case error is 0.509 ULP for exp and 0.507 ULP for exp2, on
aarch64 the rodata size is 2160 bytes, shared between exp and exp2.
On aarch64 .text + .rodata size decreased by 24912 bytes.
The non-nearest rounding error is less than 1 ULP even on targets
without efficient round implementation (although the error rate is
higher in that case). Targets with single instruction, rounding mode
independent, to nearest integer rounding and conversion can use them
by setting TOINT_INTRINSICS and adding the necessary code to their
math_private.h.
The __exp1 code uses the same algorithm, so the error bound of pow
increased a bit.
New double precision error handling code was added following the
style of the single precision error handling code.
Improvements on Cortex-A72 compared to current glibc master:
exp thruput: 1.61x in [-9.9 9.9]
exp latency: 1.53x in [-9.9 9.9]
exp thruput: 1.13x in [0.5 1]
exp latency: 1.30x in [0.5 1]
exp2 thruput: 2.03x in [-9.9 9.9]
exp2 latency: 1.64x in [-9.9 9.9]
For small (< 1) inputs the current exp code uses a separate algorithm
so the speed up there is less.
Was tested on
aarch64-linux-gnu (TOINT_INTRINSICS, fma contraction) and
arm-linux-gnueabihf (!TOINT_INTRINSICS, no fma contraction) and
x86_64-linux-gnu (!TOINT_INTRINSICS, no fma contraction) and
powerpc64le-linux-gnu (!TOINT_INTRINSICS, fma contraction) targets,
only non-nearest rounding ulp errors increase and they are within
acceptable bounds (ulp updates are in separate patches).
* NEWS: Mention exp and exp2 improvements.
* math/Makefile (libm-support): Remove t_exp.
(type-double-routines): Add math_err and e_exp_data.
* sysdeps/aarch64/libm-test-ulps: Update.
* sysdeps/arm/libm-test-ulps: Update.
* sysdeps/i386/fpu/e_exp_data.c: New file.
* sysdeps/i386/fpu/math_err.c: New file.
* sysdeps/i386/fpu/t_exp.c: Remove.
* sysdeps/ia64/fpu/e_exp_data.c: New file.
* sysdeps/ia64/fpu/math_err.c: New file.
* sysdeps/ia64/fpu/t_exp.c: Remove.
* sysdeps/ieee754/dbl-64/e_exp.c: Rewrite.
* sysdeps/ieee754/dbl-64/e_exp2.c: Rewrite.
* sysdeps/ieee754/dbl-64/e_exp_data.c: New file.
* sysdeps/ieee754/dbl-64/e_pow.c (__ieee754_pow): Update error bound.
* sysdeps/ieee754/dbl-64/eexp.tbl: Remove.
* sysdeps/ieee754/dbl-64/math_config.h: New file.
* sysdeps/ieee754/dbl-64/math_err.c: New file.
* sysdeps/ieee754/dbl-64/t_exp.c: Remove.
* sysdeps/ieee754/dbl-64/t_exp2.h: Remove.
* sysdeps/ieee754/dbl-64/uexp.h: Remove.
* sysdeps/ieee754/dbl-64/uexp.tbl: Remove.
* sysdeps/m68k/m680x0/fpu/e_exp_data.c: New file.
* sysdeps/m68k/m680x0/fpu/math_err.c: New file.
* sysdeps/m68k/m680x0/fpu/t_exp.c: Remove.
* sysdeps/powerpc/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Update.
This patch is a complete rewrite of sincosf. The new version is
significantly faster, as well as simple and accurate.
The worst-case ULP is 0.5607, maximum relative error is 0.5303 * 2^-23 over
all 4 billion inputs. In non-nearest rounding modes the error is 1ULP.
The algorithm uses 3 main cases: small inputs which don't need argument
reduction, small inputs which need a simple range reduction and large inputs
requiring complex range reduction. The code uses approximate integer
comparisons to quickly decide between these cases.
The small range reducer uses a single reduction step to handle values up to
120.0. It is fastest on targets which support inlined round instructions.
The large range reducer uses integer arithmetic for simplicity. It does a
32x96 bit multiply to compute a 64-bit modulo result. This is more than
accurate enough to handle the worst-case cancellation for values close to
an integer multiple of PI/4. It could be further optimized, however it is
already much faster than necessary.
sincosf throughput gains on Cortex-A72:
* |x| < 0x1p-12 : 1.6x
* |x| < M_PI_4 : 1.7x
* |x| < 2 * M_PI: 1.5x
* |x| < 120.0 : 1.8x
* |x| < Inf : 2.3x
* math/Makefile: Add s_sincosf_data.c.
* sysdeps/ia64/fpu/s_sincosf_data.c: New file.
* sysdeps/ieee754/flt-32/s_sincosf.h (abstop12): Add new function.
(sincosf_poly): Likewise.
(reduce_small): Likewise.
(reduce_large): Likewise.
* sysdeps/ieee754/flt-32/s_sincosf.c (sincosf): Rewrite.
* sysdeps/ieee754/flt-32/s_sincosf_data.c: New file with sincosf data.
* sysdeps/m68k/m680x0/fpu/s_sincosf_data.c: New file.
* sysdeps/x86_64/fpu/s_sincosf_data.c: New file.
