glibc/math/Makefile
Joseph Myers 050f29c188 Fix lgamma (negative) inaccuracy (bug 2542, bug 2543, bug 2558).
The existing implementations of lgamma functions (except for the ia64
versions) use the reflection formula for negative arguments.  This
suffers large inaccuracy from cancellation near zeros of lgamma (near
where the gamma function is +/- 1).

This patch fixes this inaccuracy.  For arguments above -2, there are
no zeros and no large cancellation, while for sufficiently large
negative arguments the zeros are so close to integers that even for
integers +/- 1ulp the log(gamma(1-x)) term dominates and cancellation
is not significant.  Thus, it is only necessary to take special care
about cancellation for arguments around a limited number of zeros.

Accordingly, this patch uses precomputed tables of relevant zeros,
expressed as the sum of two floating-point values.  The log of the
ratio of two sines can be computed accurately using log1p in cases
where log would lose accuracy.  The log of the ratio of two gamma(1-x)
values can be computed using Stirling's approximation (the difference
between two values of that approximation to lgamma being computable
without computing the two values and then subtracting), with
appropriate adjustments (which don't reduce accuracy too much) in
cases where 1-x is too small to use Stirling's approximation directly.

In the interval from -3 to -2, using the ratios of sines and of
gamma(1-x) can still produce too much cancellation between those two
parts of the computation (and that interval is also the worst interval
for computing the ratio between gamma(1-x) values, which computation
becomes more accurate, while being less critical for the final result,
for larger 1-x).  Because this can result in errors slightly above
those accepted in glibc, this interval is instead dealt with by
polynomial approximations.  Separate polynomial approximations to
(|gamma(x)|-1)(x-n)/(x-x0) are used for each interval of length 1/8
from -3 to -2, where n (-3 or -2) is the nearest integer to the
1/8-interval and x0 is the zero of lgamma in the relevant half-integer
interval (-3 to -2.5 or -2.5 to -2).

Together, the two approaches are intended to give sufficient accuracy
for all negative arguments in the problem range.  Outside that range,
the previous implementation continues to be used.

Tested for x86_64, x86, mips64 and powerpc.  The mips64 and powerpc
testing shows up pre-existing problems for ldbl-128 and ldbl-128ibm
with large negative arguments giving spurious "invalid" exceptions
(exposed by newly added tests for cases this patch doesn't affect the
logic for); I'll address those problems separately.

	[BZ #2542]
	[BZ #2543]
	[BZ #2558]
	* sysdeps/ieee754/dbl-64/e_lgamma_r.c (__ieee754_lgamma_r): Call
	__lgamma_neg for arguments from -28.0 to -2.0.
	* sysdeps/ieee754/flt-32/e_lgammaf_r.c (__ieee754_lgammaf_r): Call
	__lgamma_negf for arguments from -15.0 to -2.0.
	* sysdeps/ieee754/ldbl-128/e_lgammal_r.c (__ieee754_lgammal_r):
	Call __lgamma_negl for arguments from -48.0 or -50.0 to -2.0.
	* sysdeps/ieee754/ldbl-96/e_lgammal_r.c (__ieee754_lgammal_r):
	Call __lgamma_negl for arguments from -33.0 to -2.0.
	* sysdeps/ieee754/dbl-64/lgamma_neg.c: New file.
	* sysdeps/ieee754/dbl-64/lgamma_product.c: Likewise.
	* sysdeps/ieee754/flt-32/lgamma_negf.c: Likewise.
	* sysdeps/ieee754/flt-32/lgamma_productf.c: Likewise.
	* sysdeps/ieee754/ldbl-128/lgamma_negl.c: Likewise.
	* sysdeps/ieee754/ldbl-128/lgamma_productl.c: Likewise.
	* sysdeps/ieee754/ldbl-128ibm/lgamma_negl.c: Likewise.
	* sysdeps/ieee754/ldbl-128ibm/lgamma_productl.c: Likewise.
	* sysdeps/ieee754/ldbl-96/lgamma_negl.c: Likewise.
	* sysdeps/ieee754/ldbl-96/lgamma_product.c: Likewise.
	* sysdeps/ieee754/ldbl-96/lgamma_productl.c: Likewise.
	* sysdeps/generic/math_private.h (__lgamma_negf): New prototype.
	(__lgamma_neg): Likewise.
	(__lgamma_negl): Likewise.
	(__lgamma_product): Likewise.
	(__lgamma_productl): Likewise.
	* math/Makefile (libm-calls): Add lgamma_neg and lgamma_product.
	* math/auto-libm-test-in: Add more tests of lgamma.
	* math/auto-libm-test-out: Regenerated.
	* sysdeps/i386/fpu/libm-test-ulps: Update.
	* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-09-10 22:27:58 +00:00

273 lines
11 KiB
Makefile

# Copyright (C) 1996-2015 Free Software Foundation, Inc.
# This file is part of the GNU C Library.
# The GNU C Library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
# The GNU C Library 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
# Lesser General Public License for more details.
# You should have received a copy of the GNU Lesser General Public
# License along with the GNU C Library; if not, see
# <http://www.gnu.org/licenses/>.
# Makefile for the math library.
subdir := math
include ../Makeconfig
# Installed header files.
headers := math.h bits/mathcalls.h bits/mathinline.h bits/huge_val.h \
bits/huge_valf.h bits/huge_vall.h bits/inf.h bits/nan.h \
fpu_control.h complex.h bits/cmathcalls.h fenv.h \
bits/fenv.h bits/fenvinline.h bits/mathdef.h tgmath.h \
bits/math-finite.h bits/math-vector.h \
bits/libm-simd-decl-stubs.h
# FPU support code.
aux := setfpucw fpu_control
# Build the -lm library.
extra-libs := libm
extra-libs-others = $(extra-libs)
libm-support = s_lib_version s_matherr s_signgam \
fclrexcpt fgetexcptflg fraiseexcpt fsetexcptflg \
ftestexcept fegetround fesetround fegetenv feholdexcpt \
fesetenv feupdateenv t_exp fedisblxcpt feenablxcpt \
fegetexcept
libm-calls = e_acos e_acosh e_asin e_atan2 e_atanh e_cosh e_exp e_fmod \
e_hypot e_j0 e_j1 e_jn e_lgamma_r e_log e_log10 e_pow \
e_rem_pio2 e_remainder e_scalb e_sinh e_sqrt e_gamma_r \
e_ilogb \
k_cos k_rem_pio2 k_sin k_tan s_asinh s_atan s_cbrt \
s_ceil s_cos s_erf s_expm1 s_fabs \
s_floor s_log1p w_log1p s_logb \
s_nextafter s_nexttoward s_rint s_scalbln w_scalbln \
s_significand s_sin s_tan s_tanh w_acos w_acosh w_asin \
w_atan2 w_atanh w_cosh w_drem w_exp w_exp2 w_exp10 w_fmod \
w_tgamma w_hypot w_j0 w_j1 w_jn w_lgamma w_lgamma_r \
w_log w_log10 w_pow w_remainder w_scalb w_sinh w_sqrt \
w_ilogb \
s_fpclassify s_fmax s_fmin s_fdim s_nan s_trunc \
s_remquo e_log2 e_exp2 s_round s_nearbyint s_sincos \
conj cimag creal cabs carg s_cexp s_csinh s_ccosh s_clog \
s_catan s_casin s_ccos s_csin s_ctan s_ctanh s_cacos \
s_casinh s_cacosh s_catanh s_csqrt s_cpow s_cproj s_clog10 \
s_fma s_lrint s_llrint s_lround s_llround e_exp10 w_log2 \
s_isinf_ns s_issignaling $(calls:s_%=m_%) x2y2m1 k_casinh \
gamma_product k_standard lgamma_neg lgamma_product
dbl-only-routines := branred doasin dosincos halfulp mpa mpatan2 \
mpatan mpexp mplog mpsqrt mptan sincos32 slowexp \
slowpow sincostab
libm-routines = $(strip $(libm-support) $(libm-calls) \
$(patsubst %_rf,%f_r,$(libm-calls:=f)) \
$(long-m-$(long-double-fcts))) \
$(dbl-only-routines)
long-m-routines = $(patsubst %_rl,%l_r,$(libm-calls:=l))
long-m-support = t_sincosl k_sincosl
long-m-yes = $(long-m-routines) $(long-m-support)
# These functions are in libc instead of libm because __printf_fp
# calls them, so any program using printf will need them linked in,
# and we don't want to have to link every program with -lm.
# In libm-calls (above), list m_foo in place of s_foo for any
# routine that should be compiled separately for its libc and libm versions.
calls = s_isinf s_isnan s_finite s_copysign s_modf s_scalbn s_frexp s_ldexp \
s_signbit
generated += $(foreach s,.c .S l.c l.S f.c f.S,$(calls:s_%=m_%$s))
routines = $(calls) $(calls:=f) $(long-c-$(long-double-fcts))
long-c-yes = $(calls:=l)
ifeq ($(build-mathvec),yes)
# We need to install libm.so as linker script
# for more comfortable use of vector math library.
install-lib-ldscripts := libm.so
install_subdir: $(inst_libdir)/libm.so
$(inst_libdir)/libm.so: $(common-objpfx)format.lds \
$(libm) \
$(libmvec) \
$(+force)
(echo '/* GNU ld script'; echo '*/';\
cat $<; \
echo 'GROUP ( $(slibdir)/libm.so$(libm.so-version) ' \
'AS_NEEDED ( $(slibdir)/libmvec.so$(libmvec.so-version) ) )' \
) > $@
endif
# Rules for the test suite.
tests = test-matherr test-fenv atest-exp atest-sincos atest-exp2 basic-test \
test-misc test-fpucw test-fpucw-ieee tst-definitions test-tgmath \
test-tgmath-ret bug-nextafter bug-nexttoward bug-tgmath1 \
test-tgmath-int test-tgmath2 test-powl tst-CMPLX tst-CMPLX2 test-snan \
test-fenv-tls test-fenv-preserve test-fenv-return test-fenvinline \
$(tests-static)
tests-static = test-fpucw-static test-fpucw-ieee-static
# We do the `long double' tests only if this data type is available and
# distinct from `double'.
test-longdouble-yes = test-ldouble test-ildoubl
ifneq (no,$(PERL))
libm-vec-tests = $(addprefix test-,$(libmvec-tests))
libm-tests = test-float test-double $(test-longdouble-$(long-double-fcts)) \
test-ifloat test-idouble $(libm-vec-tests)
libm-tests.o = $(addsuffix .o,$(libm-tests))
tests += $(libm-tests)
libm-tests-generated = libm-test-ulps.h libm-have-vector-test.h libm-test.c
generated += $(libm-tests-generated) libm-test.stmp
# This is needed for dependencies
before-compile += $(objpfx)libm-test.c
ulps-file = $(firstword $(wildcard $(sysdirs:%=%/libm-test-ulps)))
$(addprefix $(objpfx), $(libm-tests-generated)): $(objpfx)libm-test.stmp
$(objpfx)libm-test.stmp: $(ulps-file) libm-test.inc gen-libm-test.pl \
gen-libm-have-vector-test.sh auto-libm-test-out
$(make-target-directory)
$(PERL) gen-libm-test.pl -u $< -o "$(objpfx)"
$(SHELL) gen-libm-have-vector-test.sh > $(objpfx)libm-have-vector-test.h
@echo > $@
$(objpfx)test-float.o: $(objpfx)libm-test.stmp
$(objpfx)test-ifloat.o: $(objpfx)libm-test.stmp
$(objpfx)test-double.o: $(objpfx)libm-test.stmp
$(objpfx)test-idouble.o: $(objpfx)libm-test.stmp
$(objpfx)test-ldouble.o: $(objpfx)libm-test.stmp
$(objpfx)test-ildoubl.o: $(objpfx)libm-test.stmp
endif
libm-test-fast-math-cflags = -fno-builtin -D__FAST_MATH__ -DTEST_FAST_MATH
libm-test-vec-cflags = $(libm-test-fast-math-cflags) -fno-inline \
-ffloat-store -D_OPENMP=201307 -Wno-unknown-pragmas
CFLAGS-test-double-vlen2.c = $(libm-test-vec-cflags)
CFLAGS-test-double-vlen4.c = $(libm-test-vec-cflags)
CFLAGS-test-double-vlen4-wrappers.c = $(double-vlen4-arch-ext-cflags)
CFLAGS-test-double-vlen8.c = $(libm-test-vec-cflags)
CFLAGS-test-double-vlen8-wrappers.c = $(double-vlen8-arch-ext-cflags)
CFLAGS-test-float-vlen4.c = $(libm-test-vec-cflags)
CFLAGS-test-float-vlen8.c = $(libm-test-vec-cflags)
CFLAGS-test-float-vlen8-wrappers.c = $(float-vlen8-arch-ext-cflags)
CFLAGS-test-float-vlen16.c = $(libm-test-vec-cflags)
CFLAGS-test-float-vlen16-wrappers.c = $(float-vlen16-arch-ext-cflags)
CFLAGS-test-float.c = -fno-inline -ffloat-store -fno-builtin
CFLAGS-test-double.c = -fno-inline -ffloat-store -fno-builtin
CFLAGS-test-ldouble.c = -fno-inline -ffloat-store -fno-builtin
CFLAGS-test-tgmath.c = -fno-builtin
# The following testcase uses very long lines (>3 million), so it may take a
# while to compile it. See: http://llvm.org/bugs/show_bug.cgi?id=14106 and
# http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54402
CFLAGS-test-tgmath2.c = -fno-builtin
CFLAGS-test-tgmath-ret.c = -fno-builtin
CFLAGS-test-powl.c = -fno-builtin
CPPFLAGS-test-ifloat.c = -U__LIBC_INTERNAL_MATH_INLINES \
$(libm-test-fast-math-cflags)
CPPFLAGS-test-idouble.c = -U__LIBC_INTERNAL_MATH_INLINES \
$(libm-test-fast-math-cflags)
CPPFLAGS-test-ildoubl.c = -U__LIBC_INTERNAL_MATH_INLINES \
$(libm-test-fast-math-cflags)
# The -lieee module sets the _LIB_VERSION_ switch to IEEE mode
# for error handling in the -lm functions.
install-lib += libieee.a
non-lib.a += libieee.a
extra-objs += libieee.a ieee-math.o
include ../Rules
ifneq (no,$(PERL))
# This must come after the inclusion of sysdeps Makefiles via Rules.
$(addprefix $(objpfx), $(addsuffix .o, $(libm-vec-tests))): $(objpfx)libm-test.stmp
# Run the math programs to automatically generate ULPs files.
.PHONY: regen-ulps
run-regen-ulps = $(test-wrapper-env) \
$(run-program-env) \
$($*-ENV) $(rtld-prefix) $${run}
regen-ulps: $(addprefix $(objpfx),$(libm-tests))
rm -f $(objpfx)ULPs; rm -f $(objpfx)NewUlps; \
cp $(ulps-file) $(objpfx)libm-test-ulps; \
for run in $^; do \
echo "Regenerating ULPs for $${run}"; \
$(run-regen-ulps) -u -o $(objpfx); \
cat $(objpfx)ULPs >> $(objpfx)libm-test-ulps; \
rm $(objpfx)ULPs; \
done; \
$(PERL) gen-libm-test.pl -o $(objpfx) -n -u $(objpfx)libm-test-ulps; \
echo "Automatic regeneration of ULPs complete."; \
echo "Difference between the current baseline and the new baseline is:";\
diff -urN $(ulps-file) $(objpfx)NewUlps; \
echo "Copy $(objpfx)NewUlps to $(ulps-file) (relative to source)."
else
regen-ulps:
@echo "Automatic regeneration of ULPs requires perl."; \
exit 1;
endif
# The generated sysd-rules file defines rules like this for sources
# coming from sysdeps/ directories. These rules find the generic sources.
define o-iterator-doit
$(objpfx)m_%$o: s_%.c $(before-compile); $$(compile-command.c)
endef
object-suffixes-left := $(all-object-suffixes)
include $(o-iterator)
# This file defines the default _LIB_VERSION variable that controls
# the error return conventions for the math functions.
CPPFLAGS-s_lib_version.c := -D_POSIX_MODE
# We don't want the fdlibm code to use the inline math functions,
# only the fdlibm code.
math-CPPFLAGS += -D__NO_MATH_INLINES -D__LIBC_INTERNAL_MATH_INLINES
ifneq ($(long-double-fcts),yes)
# The `double' and `long double' types are the same on this machine.
# We won't compile the `long double' code at all. Tell the `double' code
# to define aliases for the `FUNCl' names. To avoid type conflicts in
# defining those aliases, tell <math.h> to declare the `FUNCl' names with
# `double' instead of `long double'.
math-CPPFLAGS += -DNO_LONG_DOUBLE -D_Mlong_double_=double
endif
# The -lieee library is actually an object file.
# The module just defines the _LIB_VERSION_ variable.
# It's not a library to make sure it is linked in instead of s_lib_version.o.
$(objpfx)libieee.a: $(objpfx)ieee-math.o
rm -f $@
$(patsubst %/,cd % &&,$(objpfx)) \
$(LN_S) $(<F) $(@F)
$(addprefix $(objpfx),$(filter-out $(tests-static) $(libm-vec-tests),$(tests))): $(libm)
$(addprefix $(objpfx),$(tests-static)): $(objpfx)libm.a
$(addprefix $(objpfx), $(libm-vec-tests)): $(objpfx)%: $(libm) $(libmvec) \
$(objpfx)%-wrappers.o
gmp-objs = $(patsubst %,$(common-objpfx)stdlib/%.o,\
add_n sub_n cmp addmul_1 mul_1 mul_n divmod_1 \
lshift rshift mp_clz_tab udiv_qrnnd inlines \
$(gmp-sysdep_routines))
$(objpfx)atest-exp: $(gmp-objs)
$(objpfx)atest-sincos: $(gmp-objs)
$(objpfx)atest-exp2: $(gmp-objs)
$(objpfx)test-fenv-tls: $(shared-thread-library)