diff options
| author | Joseph Myers <joseph@codesourcery.com> | 2014-05-14 12:35:40 +0000 |
|---|---|---|
| committer | Joseph Myers <joseph@codesourcery.com> | 2014-05-14 12:35:40 +0000 |
| commit | 913d03c864ea2547e97f8d2d30fc71a008d4d103 (patch) | |
| tree | f5b230d8c416181ef9998a5cf497d7cd89091b72 /sysdeps/i386 | |
| parent | 0bf061d3e37e0f72c2378cbded67c58df4f58a4b (diff) | |
| download | glibc-913d03c864ea2547e97f8d2d30fc71a008d4d103.tar.gz glibc-913d03c864ea2547e97f8d2d30fc71a008d4d103.tar.xz glibc-913d03c864ea2547e97f8d2d30fc71a008d4d103.zip | |
Fix acosh (1) in round-downward mode (bug 16927).
According to C99 and C11 Annex F, acosh (1) should be +0 in all rounding modes. However, some implementations in glibc wrongly return -0 in round-downward mode (which is what you get if you end up computing log1p (-0), via 1 - 1 being -0 in round-downward mode). This patch fixes the problem implementations, by correcting the test for an exact 1 value in the ldbl-96 implementation to allow for the explicit high bit of the mantissa, and by inserting fabs instructions in the i386 implementations; tests of acosh are duly converted to ALL_RM_TEST. I believe all the other sysdeps/ieee754 implementations are already OK (I haven't checked the ia64 versions, but if buggy then that will be obvious from the results of test runs after this patch is in). Tested x86_64 and x86 and ulps updated accordingly. [BZ #16927] * sysdeps/i386/fpu/e_acosh.S (__ieee754_acosh): Use fabs on x-1 value. * sysdeps/i386/fpu/e_acoshf.S (__ieee754_acoshf): Likewise. * sysdeps/i386/fpu/e_acoshl.S (__ieee754_acoshl): Likewise. * sysdeps/ieee754/ldbl-96/e_acoshl.c (__ieee754_acoshl): Correct for explicit high bit of mantissa when testing for argument equal to 1. * math/libm-test.inc (acosh_test): Use ALL_RM_TEST. * sysdeps/i386/fpu/libm-test-ulps: Update. * sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
Diffstat (limited to 'sysdeps/i386')
| -rw-r--r-- | sysdeps/i386/fpu/e_acosh.S | 1 | ||||
| -rw-r--r-- | sysdeps/i386/fpu/e_acoshf.S | 1 | ||||
| -rw-r--r-- | sysdeps/i386/fpu/e_acoshl.S | 1 | ||||
| -rw-r--r-- | sysdeps/i386/fpu/libm-test-ulps | 12 |
4 files changed, 15 insertions, 0 deletions
diff --git a/sysdeps/i386/fpu/e_acosh.S b/sysdeps/i386/fpu/e_acosh.S index 98a329179b..c66781c985 100644 --- a/sysdeps/i386/fpu/e_acosh.S +++ b/sysdeps/i386/fpu/e_acosh.S @@ -52,6 +52,7 @@ ENTRY(__ieee754_acosh) // 1 <= x <= 2 => y = log1p(x-1+sqrt(2*(x-1)+(x-1)^2)) fsubl MO(one) // x-1 : log(2) + fabs // acosh(1) is +0 in all rounding modes fld %st // x-1 : x-1 : log(2) fmul %st(1) // (x-1)^2 : x-1 : log(2) fadd %st(1) // x-1+(x-1)^2 : x-1 : log(2) diff --git a/sysdeps/i386/fpu/e_acoshf.S b/sysdeps/i386/fpu/e_acoshf.S index db9cf337b2..fa35d50fd0 100644 --- a/sysdeps/i386/fpu/e_acoshf.S +++ b/sysdeps/i386/fpu/e_acoshf.S @@ -52,6 +52,7 @@ ENTRY(__ieee754_acoshf) // 1 <= x <= 2 => y = log1p(x-1+sqrt(2*(x-1)+(x-1)^2)) fsubl MO(one) // x-1 : log(2) + fabs // acosh(1) is +0 in all rounding modes fld %st // x-1 : x-1 : log(2) fmul %st(1) // (x-1)^2 : x-1 : log(2) fadd %st(1) // x-1+(x-1)^2 : x-1 : log(2) diff --git a/sysdeps/i386/fpu/e_acoshl.S b/sysdeps/i386/fpu/e_acoshl.S index a832155d17..38d8110550 100644 --- a/sysdeps/i386/fpu/e_acoshl.S +++ b/sysdeps/i386/fpu/e_acoshl.S @@ -59,6 +59,7 @@ ENTRY(__ieee754_acoshl) // 1 <= x <= 2 => y = log1p(x-1+sqrt(2*(x-1)+(x-1)^2)) fsubl MO(one) // x-1 : log(2) + fabs // acosh(1) is +0 in all rounding modes fld %st // x-1 : x-1 : log(2) fmul %st(1) // (x-1)^2 : x-1 : log(2) fadd %st(1) // x-1+(x-1)^2 : x-1 : log(2) diff --git a/sysdeps/i386/fpu/libm-test-ulps b/sysdeps/i386/fpu/libm-test-ulps index aea6c51b23..ccef44ac63 100644 --- a/sysdeps/i386/fpu/libm-test-ulps +++ b/sysdeps/i386/fpu/libm-test-ulps @@ -21,6 +21,18 @@ Function: "acos_upward": ildouble: 1 ldouble: 1 +Function: "acosh_downward": +ildouble: 2 +ldouble: 2 + +Function: "acosh_towardzero": +ildouble: 2 +ldouble: 2 + +Function: "acosh_upward": +ildouble: 1 +ldouble: 1 + Function: "asin_downward": double: 1 float: 1 |