In case of underflow, return -0.0 or +0.0, depending on the sign of
the argument. This makes sparc64 pass regress/lib/libc/ldexp.
This commit is contained in:
parent
dd05ba5c10
commit
47923e6bf5
|
@ -1,4 +1,4 @@
|
|||
/* $NetBSD: ldexp_ieee754.c,v 1.3 2003/10/27 03:09:43 mrg Exp $ */
|
||||
/* $NetBSD: ldexp_ieee754.c,v 1.4 2003/12/30 23:19:07 martin Exp $ */
|
||||
|
||||
/*-
|
||||
* Copyright (c) 1999 The NetBSD Foundation, Inc.
|
||||
|
@ -38,7 +38,7 @@
|
|||
|
||||
#include <sys/cdefs.h>
|
||||
#if defined(LIBC_SCCS) && !defined(lint)
|
||||
__RCSID("$NetBSD: ldexp_ieee754.c,v 1.3 2003/10/27 03:09:43 mrg Exp $");
|
||||
__RCSID("$NetBSD: ldexp_ieee754.c,v 1.4 2003/12/30 23:19:07 martin Exp $");
|
||||
#endif /* LIBC_SCCS and not lint */
|
||||
|
||||
#include <sys/types.h>
|
||||
|
@ -76,14 +76,14 @@ ldexp(double val, int expon)
|
|||
*/
|
||||
if (expon <= -DBL_FRACBITS) {
|
||||
errno = ERANGE;
|
||||
return (0.0);
|
||||
return (val < 0.0 ? -0.0 : 0.0);
|
||||
}
|
||||
mul.dblu_d = 0.0;
|
||||
mul.dblu_dbl.dbl_exp = expon + DBL_EXP_BIAS;
|
||||
u.dblu_d *= mul.dblu_d;
|
||||
if (u.dblu_d == 0.0) {
|
||||
errno = ERANGE;
|
||||
return (0.0);
|
||||
return (val < 0.0 ? -0.0 : 0.0);
|
||||
}
|
||||
return (u.dblu_d);
|
||||
} else {
|
||||
|
@ -113,7 +113,7 @@ ldexp(double val, int expon)
|
|||
*/
|
||||
if (newexp <= -DBL_FRACBITS) {
|
||||
errno = ERANGE;
|
||||
return (0.0);
|
||||
return (val < 0.0 ? -0.0 : 0.0);
|
||||
}
|
||||
/*
|
||||
* Denormalize the result. We do this with a multiply. If
|
||||
|
|
Loading…
Reference in New Issue