diff options
| author | Joseph Myers <joseph@codesourcery.com> | 2015-11-05 21:56:31 +0000 |
|---|---|---|
| committer | Joseph Myers <joseph@codesourcery.com> | 2015-11-05 21:56:31 +0000 |
| commit | 01189b083b8ce54fc568fd2ab0dd433d869ede0e (patch) | |
| tree | 15620545ab980f008bdec41540a43f1efea79ed0 /sysdeps/x86_64 | |
| parent | 939fdd1c36dde8884543bc9c3b8506e76c9ff0e4 (diff) | |
| download | glibc-01189b083b8ce54fc568fd2ab0dd433d869ede0e.zip glibc-01189b083b8ce54fc568fd2ab0dd433d869ede0e.tar.gz glibc-01189b083b8ce54fc568fd2ab0dd433d869ede0e.tar.bz2 | |
Fix i386/x86_64 log* (1) zero sign for -ffinite-math-only (bug 19213).
For the -ffinite-math-only versions of various x86_64 and x86 log*
functions, a zero result from log* (1) is returned with incorrect sign
in round-downward mode. This patch fixes this in a similar way to the
previous fixes for the non-*_finite versions of the functions.
Tested for x86_64 and x86 (including an i586 build), together with a
patch that will be applied separately to enable the main libm-test.inc
tests for the finite-math-only functions.
[BZ #19213]
* sysdeps/i386/fpu/e_log.S (__log_finite): Ensure +0 is always
returned for argument 1.
* sysdeps/i386/fpu/e_logf.S (__logf_finite): Likewise.
* sysdeps/i386/fpu/e_logl.S (__logl_finite): Likewise.
* sysdeps/i386/i686/fpu/e_logl.S (__logl_finite): Likewise.
* sysdeps/x86_64/fpu/e_log10l.S (__log10l_finite): Likewise.
* sysdeps/x86_64/fpu/e_log2l.S (__log2l_finite): Likewise.
* sysdeps/x86_64/fpu/e_logl.S (__logl_finite): Likewise.
Diffstat (limited to 'sysdeps/x86_64')
| -rw-r--r-- | sysdeps/x86_64/fpu/e_log10l.S | 8 | ||||
| -rw-r--r-- | sysdeps/x86_64/fpu/e_log2l.S | 8 | ||||
| -rw-r--r-- | sysdeps/x86_64/fpu/e_logl.S | 8 |
3 files changed, 21 insertions, 3 deletions
diff --git a/sysdeps/x86_64/fpu/e_log10l.S b/sysdeps/x86_64/fpu/e_log10l.S index 2607ad1..8fa6164 100644 --- a/sysdeps/x86_64/fpu/e_log10l.S +++ b/sysdeps/x86_64/fpu/e_log10l.S @@ -79,7 +79,13 @@ ENTRY(__log10l_finite) fnstsw // x-1 : x : log10(2) andb $0x45, %ah jz 2b - fstp %st(1) // x-1 : log10(2) + fxam + fnstsw + andb $0x45, %ah + cmpb $0x40, %ah + jne 6f + fabs // log10(1) is +0 in all rounding modes. +6: fstp %st(1) // x-1 : log10(2) fyl2xp1 // log10(x) ret END(__log10l_finite) diff --git a/sysdeps/x86_64/fpu/e_log2l.S b/sysdeps/x86_64/fpu/e_log2l.S index c12906d..a063255 100644 --- a/sysdeps/x86_64/fpu/e_log2l.S +++ b/sysdeps/x86_64/fpu/e_log2l.S @@ -78,7 +78,13 @@ ENTRY(__log2l_finite) fnstsw // x-1 : x : 1 andb $0x45, %ah jz 2b - fstp %st(1) // x-1 : 1 + fxam + fnstsw + andb $0x45, %ah + cmpb $0x40, %ah + jne 6f + fabs // log2(1) is +0 in all rounding modes. +6: fstp %st(1) // x-1 : 1 fyl2xp1 // log(x) ret END (__log2l_finite) diff --git a/sysdeps/x86_64/fpu/e_logl.S b/sysdeps/x86_64/fpu/e_logl.S index 047b8db..dbe6fd5 100644 --- a/sysdeps/x86_64/fpu/e_logl.S +++ b/sysdeps/x86_64/fpu/e_logl.S @@ -81,7 +81,13 @@ ENTRY(__logl_finite) fnstsw // x-1 : x : log(2) andb $0x45, %ah jz 2b - fstp %st(1) // x-1 : log(2) + fxam + fnstsw + andb $0x45, %ah + cmpb $0x40, %ah + jne 7f + fabs // log(1) is +0 in all rounding modes. +7: fstp %st(1) // x-1 : log(2) fyl2xp1 // log(x) ret END (__logl_finite) |