mirror of
https://sourceware.org/git/glibc.git
synced 2024-11-27 15:30:07 +00:00
bb769ab68f
2001-04-06 Ulrich Drepper <drepper@redhat.com> * sysdeps/generic/e_scalb.c: Don't use FE_INVALID unless it's available. * sysdeps/generic/e_scalbf.c: Likewise. * sysdeps/generic/e_scalbl.c: Likewise. Patch by Mark Hatle <fray@mvista.com>. 2001-04-05 David S. Miller <davem@redhat.com> * sysdeps/sparc/abort-instr.h: New file. 2001-04-05 Matt Wilson <msw@redhat.com> * sysdeps/unix/sysv/linux/alpha/getsysstats.c (GET_NPROCS_PARSER): added code to parse new 2.4 format. (GET_NPROCS_CONF_PARSER): Likewise. 2001-04-05 David S. Miller <davem@redhat.com> * scripts/config.sub: Recognize sparcv9b like sparcv9. * configure.in: Add sparcv9b. * sysdeps/sparc/sparc32/sparcv9b/Implies: New file * sysdeps/sparc/sparc32/sparcv9b/memcpy.S: New file * sysdeps/sparc/sparc64/sparcv9b/memcpy.S: New file
69 lines
1.5 KiB
C
69 lines
1.5 KiB
C
/* e_scalbf.c -- float version of e_scalb.c.
|
|
* Conversion to float by Ian Lance Taylor, Cygnus Support, ian@cygnus.com.
|
|
*/
|
|
|
|
/*
|
|
* ====================================================
|
|
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
|
|
*
|
|
* Developed at SunPro, a Sun Microsystems, Inc. business.
|
|
* Permission to use, copy, modify, and distribute this
|
|
* software is freely granted, provided that this notice
|
|
* is preserved.
|
|
* ====================================================
|
|
*/
|
|
|
|
#if defined(LIBM_SCCS) && !defined(lint)
|
|
static char rcsid[] = "$NetBSD: e_scalbf.c,v 1.3 1995/05/10 20:46:12 jtc Exp $";
|
|
#endif
|
|
|
|
#include <fenv.h>
|
|
#include "math.h"
|
|
#include "math_private.h"
|
|
|
|
#ifdef _SCALB_INT
|
|
#ifdef __STDC__
|
|
float __ieee754_scalbf(float x, int fn)
|
|
#else
|
|
float __ieee754_scalbf(x,fn)
|
|
float x; int fn;
|
|
#endif
|
|
#else
|
|
#ifdef __STDC__
|
|
float __ieee754_scalbf(float x, float fn)
|
|
#else
|
|
float __ieee754_scalbf(x,fn)
|
|
float x, fn;
|
|
#endif
|
|
#endif
|
|
{
|
|
#ifdef _SCALB_INT
|
|
return __scalbnf(x,fn);
|
|
#else
|
|
if (__isnanf(x)||__isnanf(fn)) return x*fn;
|
|
if (!__finitef(fn)) {
|
|
if(fn>(float)0.0) return x*fn;
|
|
else if (x == 0)
|
|
return x;
|
|
else if (!__finitef (x))
|
|
{
|
|
# ifdef FE_INVALID
|
|
feraiseexcept (FE_INVALID);
|
|
# endif
|
|
return __nanf ("");
|
|
}
|
|
else return x/(-fn);
|
|
}
|
|
if (__rintf(fn)!=fn)
|
|
{
|
|
# ifdef FE_INVALID
|
|
feraiseexcept (FE_INVALID);
|
|
# endif
|
|
return __nanf ("");
|
|
}
|
|
if ( fn > (float)65000.0) return __scalbnf(x, 65000);
|
|
if (-fn > (float)65000.0) return __scalbnf(x,-65000);
|
|
return __scalbnf(x,(int)fn);
|
|
#endif
|
|
}
|