[BZ #16824] Fix failing y1 due to too large ulps in downward/upward rounding mode.

This commit is contained in:
Stefan Liebler 2014-04-16 12:52:52 +02:00 committed by Andreas Krebbel
parent 9860b04502
commit 8ea587db2b
3 changed files with 8 additions and 1 deletions

View File

@ -1,3 +1,9 @@
2014-04-16 Stefan Liebler <stli@linux.vnet.ibm.com>
[BZ #16824]
* sysdeps/ieee754/ldbl-128/e_j1l.c (__ieee754_y1l):
Set round-to-nearest internally to reduce error accumulation.
2014-04-16 Alan Modra <amodra@gmail.com>
[BZ #16740]

2
NEWS
View File

@ -14,7 +14,7 @@ Version 2.20
16610, 16611, 16613, 16619, 16623, 16632, 16634, 16639, 16642, 16648,
16649, 16670, 16674, 16677, 16680, 16683, 16689, 16695, 16701, 16706,
16707, 16712, 16713, 16714, 16731, 16739, 16740, 16743, 16758, 16759,
16760, 16770, 16786, 16789, 16799, 16800, 16815, 16831, 16838.
16760, 16770, 16786, 16789, 16799, 16800, 16815, 16824, 16831, 16838.
* Running the testsuite no longer terminates as soon as a test fails.
Instead, a file tests.sum (xtests.sum from "make xcheck") is generated,

View File

@ -856,6 +856,7 @@ __ieee754_y1l (long double x)
if (xx <= 2.0L)
{
/* 0 <= x <= 2 */
SET_RESTORE_ROUNDL (FE_TONEAREST);
z = xx * xx;
p = xx * neval (z, Y0_2N, NY0_2N) / deval (z, Y0_2D, NY0_2D);
p = -TWOOPI / xx + p;