Avoid potential integer overflows and improve consistency. This should

fix bug #13918 "Segfault with precision = MPFR_PREC_MAX on 32-bit".
Note: this problem appeared in MPFR 3.0.0 when the precision type (now
mpfr_prec_t) was changed to a signed integer.

git-svn-id: svn://scm.gforge.inria.fr/svn/mpfr/trunk@8025 280ebfd0-de03-0410-8827-d642c229c3f4

1 files changed, 1 insertions, 1 deletions

diff --git a/src/exp.c b/src/exp.c
index 5fea7c19d..4a0cd3ec1 100644
--- a/src/exp.c
+++ b/src/exp.c
@@ -133,7 +133,7 @@ mpfr_exp (mpfr_ptr y, mpfr_srcptr x, mpfr_rnd_t rnd_mode)
               mp_size_t yn;
               int sh;
 
-              yn = 1 + (MPFR_PREC(y) - 1) / GMP_NUMB_BITS;
+              yn = MPFR_LIMB_SIZE (y);
               sh = (mpfr_prec_t) yn * GMP_NUMB_BITS - MPFR_PREC(y);
               MPFR_MANT(y)[0] += MPFR_LIMB_ONE << sh;
               inexact = 1;