diff options
author | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2014-05-20 16:07:49 -0500 |
---|---|---|
committer | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2014-05-20 16:07:49 -0500 |
commit | e13bccd3de53227dd4d6ece08873e9bcdd76b08f (patch) | |
tree | 49177b99cb7347c9625fbab5ce29cc0a640f9caf | |
parent | 4406c41c1d6088abf01c216e49700cd3f8f01fcc (diff) | |
download | glibc-e13bccd3de53227dd4d6ece08873e9bcdd76b08f.zip glibc-e13bccd3de53227dd4d6ece08873e9bcdd76b08f.tar.gz glibc-e13bccd3de53227dd4d6ece08873e9bcdd76b08f.tar.bz2 |
PowerPC: Fix copysignf optimization macro
This patch fixes the __copysignf optimized macro meant to internal libm
usage when used with constant value. Without the explicit cast to
float, if it is used with const double value (for instance, on
s_casinhf.c) double constants will be used and it may lead to precision
issues in some algorithms.
It fixes the following failures on PPC64/POWER7:
Failure: Test: Real part of: cacos_downward (inf + 0 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_downward (inf - 0 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_downward (inf + 0.5 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_downward (inf - 0.5 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_towardzero (inf + 0 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_towardzero (inf - 0 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_towardzero (inf + 0.5 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
Failure: Test: Real part of: cacos_towardzero (inf - 0.5 i)
Result:
is: 1.19209289550781250000e-07 0x1.00000000000000000000p-23
should be: 0.00000000000000000000e+00 0x0.00000000000000000000p+0
-rw-r--r-- | ChangeLog | 5 | ||||
-rw-r--r-- | sysdeps/powerpc/fpu/math_private.h | 4 |
2 files changed, 8 insertions, 1 deletions
@@ -1,3 +1,8 @@ +2014-05-20 Adhemerval Zanella <azanella@linux.vnet.ibm.com> + + * sysdeps/powerpc/fpu/math_private.h [__copysignf]: Fix copysign macro + optimization when used with float constants. + 2014-05-20 Aurelien Jarno <aurelien@aurel32.net> [BZ #16915] diff --git a/sysdeps/powerpc/fpu/math_private.h b/sysdeps/powerpc/fpu/math_private.h index dde153d..1ec4881 100644 --- a/sysdeps/powerpc/fpu/math_private.h +++ b/sysdeps/powerpc/fpu/math_private.h @@ -166,11 +166,13 @@ __ieee754_sqrtf (float __x) # ifndef __copysignf # define __copysignf(x, y) \ ({ float __z; \ + float __x = x; \ + float __y = y; \ __asm __volatile ( \ " fcpsgn %0,%1,%2\n" \ " frsp %0,%0\n" \ : "=f" (__z) \ - : "f" (y), "f" (x)); \ + : "f" (__y), "f" (__x));\ __z; }) # endif |