diff options
| author | Roland McGrath <roland@gnu.org> | 1996-03-21 15:20:37 +0000 |
|---|---|---|
| committer | Roland McGrath <roland@gnu.org> | 1996-03-21 15:20:37 +0000 |
| commit | ce563359aff66775c71c3e84e7d06a69c199eac0 (patch) | |
| tree | 5d131b8826fba365414fafd25ee7d61eaba7ef0b /sysdeps/ieee754 | |
| parent | 38bb44bc19539b14a22a2e931c001c58b6cb1b95 (diff) | |
Wed Mar 20 11:28:49 1996 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* stdio-common/printf.h, stdio-common/printf_fp.c,
stdio-common/vfprintf.c: Place const in parameter list at
correct place.
Wed Mar 20 23:58.12 1996 Ulrich Drepper <drepper@gnu.ai.mit.edu>
* sysdeps/i386/ldbl2mpn.c: Copy of former version of
sysdeps/ieee754/ldbl2mpn.c.
* sysdeps/ieee754/ldbl2mpn.c: Remove i386 specific unification.
Reported by Andreas Schwab.
Wed Mar 20 19:58:43 1996 Roland McGrath <roland@charlie-brown.gnu.ai.mit.edu>
* version.c: Include version.h to define RELEASE and VERSION macros.
(__libc_release, __libc_version): Use them.
* locale/SYS_libc.c (_libc_intl_domainname): Include ../version.h and
use VERSION to define domainname as `SYS_GNU_libc-VERSION'.
* Make-dist (rel+vers): Snarf the macro values from version.h.
* MakeTAGS (extract): Pass $(XGETTEXTFLAGS-$(@F)).
(XGETTEXTFLAGS-siglist.pot, XGETTEXTFLAGS-errlist.pot): New variables;
pass -a for these files.
* Makerules (po/%.pot): Depend on FORCE target so recursive make is
always run.
Diffstat (limited to 'sysdeps/ieee754')
| -rw-r--r-- | sysdeps/ieee754/ldbl2mpn.c | 7 |
1 files changed, 0 insertions, 7 deletions
diff --git a/sysdeps/ieee754/ldbl2mpn.c b/sysdeps/ieee754/ldbl2mpn.c index 8db73cd27b..b72af2ef47 100644 --- a/sysdeps/ieee754/ldbl2mpn.c +++ b/sysdeps/ieee754/ldbl2mpn.c @@ -65,13 +65,6 @@ __mpn_extract_long_double (mp_ptr res_ptr, mp_size_t size, one bit, and its exponent is in fact the format minimum. */ int cnt; - /* One problem with Intel's 80-bit format is that the explicit - leading one in the normalized representation has to be zero - for denormalized number. If it is one, the number is according - to Intel's specification an invalid number. We make the - representation unique by explicitly clearing this bit. */ - res_ptr[N - 1] &= ~(1L << ((LDBL_MANT_DIG - 1) % BITS_PER_MP_LIMB)); - if (res_ptr[N - 1] != 0) { count_leading_zeros (cnt, res_ptr[N - 1]); |