diff options
| author | Joseph Myers <joseph@codesourcery.com> | 2015-01-12 23:02:14 +0000 |
|---|---|---|
| committer | Joseph Myers <joseph@codesourcery.com> | 2015-01-12 23:02:14 +0000 |
| commit | 5a9e4c09a2601a8100ea9a1f7bc0360782cd1625 (patch) | |
| tree | f70efee88fe34c9f405165c4241978724ccf3282 /NEWS | |
| parent | 34e93d6c76b7a6eaf94697a172b958704fd33396 (diff) | |
| download | glibc-5a9e4c09a2601a8100ea9a1f7bc0360782cd1625.zip glibc-5a9e4c09a2601a8100ea9a1f7bc0360782cd1625.tar.gz glibc-5a9e4c09a2601a8100ea9a1f7bc0360782cd1625.tar.bz2 | |
Fix ldbl-96 scalblnl underflowing results (bug 17803).
The ldbl-96 implementation of scalblnl (used for x86_64 and ia64) uses
a condition k <= -63 to determine when a standard underflowing result
tiny*__copysignl(tiny,x) should be returned. However, that condition
corresponds to values with exponent -16446 or less, and in the case of
-16446, the correct result for round-to-nearest depends on whether the
value is exactly 0x1p-16446 (half the least subnormal) or more than
that. This patch fixes the bug by changing the condition to k <= -64
and accordingly adjusting the exponent by 64 not 63 when converting to
a normal value.
Tested for x86_64.
[BZ #17803]
* sysdeps/ieee754/ldbl-96/s_scalblnl.c (twom63): Rename to
twom64. Adjust value to 0x1p-64L.
(__scalblnl): Only return standard underflowing result for K <=
-64 not K <= -63; adjust exponent for underflowing result by 64
not 63.
* math/libm-test.inc (scalbn_test_data): Add more tests.
(scalbln_test_data): Likewise.
Diffstat (limited to 'NEWS')
| -rw-r--r-- | NEWS | 2 |
1 files changed, 1 insertions, 1 deletions
@@ -17,7 +17,7 @@ Version 2.21 17633, 17634, 17635, 17647, 17653, 17657, 17658, 17664, 17665, 17668, 17682, 17717, 17719, 17722, 17723, 17724, 17725, 17732, 17733, 17744, 17745, 17746, 17747, 17748, 17775, 17777, 17780, 17781, 17782, 17791, - 17793, 17796, 17797, 17806, 17834 + 17793, 17796, 17797, 17803, 17806, 17834 * Added support for TSX lock elision of pthread mutexes on powerpc32, powerpc64 and powerpc64le. This may improve lock scaling of existing programs on |