This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: PATCH: PR math/12775: Typo in sysdeps/x86_64/fpu/e_powl.S
On Wed, May 18, 2011 at 09:30:03AM -0700, H.J. Lu wrote:
> @@ -90,7 +90,8 @@ distribute += $(filter-out $(generated),$(long-m-yes:=.c) $(long-c-yes:=.c))
> # Rules for the test suite.
> tests = test-matherr test-fenv atest-exp atest-sincos atest-exp2 basic-test \
> test-misc test-fpucw tst-definitions test-tgmath test-tgmath-ret \
> - bug-nextafter bug-nexttoward bug-tgmath1 test-tgmath-int test-tgmath2
> + bug-nextafter bug-nexttoward bug-tgmath1 test-tgmath-int \
> + test-tgmath2 test-powl
> # We do the `long double' tests only if this data type is available and
> # distinct from `double'.
> test-longdouble-yes = test-ldouble test-ildoubl
> @@ -131,12 +132,14 @@ CFLAGS-test-ldouble.c = -fno-inline -ffloat-store -fno-builtin
> CFLAGS-test-tgmath.c = -fno-builtin
> CFLAGS-test-tgmath2.c = -fno-builtin
> CFLAGS-test-tgmath-ret.c = -fno-builtin
> +CFLAGS-test-powl.c = -O0
Why not -fno-builtin instead of -O0?
> +#ifndef NO_LONG_DOUBLE
> + {
> + long double x = 1e-20;
> + union ieee854_long_double u;
> + u.ieee.mantissa0 = 1;
> + u.ieee.mantissa1 = 1;
> + u.ieee.exponent = 0;
> + u.ieee.negative = 0;
This won't work correctly for IEEE quad long double or IBM double double.
So, either you should add the test for extended long double only, or
add guards here for extended long double too (e.g. LDBL_MANT_DIG == 64).
Jakub