This is ieee-float.c in view mode; [Download] [Up]
/* IEEE floating point support routines, for GDB, the GNU Debugger.
Copyright (C) 1991 Free Software Foundation, Inc.
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
#include "defs.h"
#include "ieee-float.h"
#include <math.h> /* ldexp */
/* Convert an IEEE extended float to a double.
FROM is the address of the extended float.
Store the double in *TO. */
void
ieee_extended_to_double (ext_format, from, to)
const struct ext_format *ext_format;
char *from;
double *to;
{
unsigned char *ufrom = (unsigned char *)from;
double dto;
unsigned long mant0, mant1, exponent;
memcpy (&mant0, &from[MANBYTE_H], 4);
memcpy (&mant1, &from[MANBYTE_L], 4);
exponent = ((ufrom[EXPBYTE_H] & (unsigned char)~SIGNMASK) << 8) | ufrom[EXPBYTE_L];
#if 0
/* We can't do anything useful with a NaN anyway, so ignore its
difference. It will end up as Infinity or something close. */
if (exponent == EXT_EXP_NAN) {
/* We have a NaN source. */
dto = 0.123456789; /* Not much else useful to do -- we don't know if
the host system even *has* NaNs, nor how to
generate an innocuous one if it does. */
} else
#endif
if (exponent == 0 && mant0 == 0 && mant1 == 0) {
dto = 0;
} else {
/* Build the result algebraically. Might go infinite, underflow, etc;
who cares. */
mant0 |= 0x80000000;
dto = ldexp ((double)mant0, exponent - EXT_EXP_BIAS - 31);
dto += ldexp ((double)mant1, exponent - EXT_EXP_BIAS - 31 - 32);
if (ufrom[EXPBYTE_H] & SIGNMASK) /* If negative... */
dto = -dto; /* ...negate. */
}
*to = dto;
}
/* The converse: convert the double *FROM to an extended float
and store where TO points. */
void
double_to_ieee_extended (ext_format, from, to)
const struct ext_format *ext_format;
double *from;
char *to;
{
double dfrom = *from;
unsigned long twolongs[2];
unsigned long mant0, mant1, exponent;
unsigned char tobytes[8];
memset (to, 0, TOTALSIZE);
if (dfrom == 0)
return; /* Result is zero */
if (dfrom != dfrom) {
/* From is NaN */
to[EXPBYTE_H] = (unsigned char)(EXT_EXP_NAN >> 8);
to[EXPBYTE_L] = (unsigned char)EXT_EXP_NAN;
to[MANBYTE_H] = 1; /* Be sure it's not infinity, but NaN value is irrel */
return; /* Result is NaN */
}
if (dfrom < 0)
to[SIGNBYTE] |= SIGNMASK; /* Set negative sign */
/* How to tell an infinity from an ordinary number? FIXME-someday */
/* The following code assumes that the host has IEEE doubles. FIXME-someday.
It also assumes longs are 32 bits! FIXME-someday. */
memcpy (twolongs, from, 8);
memcpy (tobytes, from, 8);
#if HOST_BYTE_ORDER == BIG_ENDIAN
exponent = ((tobytes[1] & 0xF0) >> 4) | (tobytes[0] & 0x7F) << 4;
mant0 = (twolongs[0] << 11) | twolongs[1] >> 21;
mant1 = (twolongs[1] << 11);
#else
exponent = ((tobytes[6] & 0xF0) >> 4) | (tobytes[7] & 0x7F) << 4;
mant0 = (twolongs[1] << 11) | twolongs[0] >> 21;
mant1 = (twolongs[0] << 11);
#endif
/* Fiddle with leading 1-bit, implied in double, explicit in extended. */
if (exponent == 0)
mant0 &= 0x7FFFFFFF;
else
mant0 |= 0x80000000;
exponent -= DBL_EXP_BIAS; /* Get integer exp */
exponent += EXT_EXP_BIAS; /* Offset for extended */
/* OK, now store it in extended format. */
to[EXPBYTE_H] |= (unsigned char)(exponent >> 8); /* Retain sign */
to[EXPBYTE_L] = (unsigned char) exponent;
memcpy (&to[MANBYTE_H], &mant0, 4);
memcpy (&to[MANBYTE_L], &mant1, 4);
}
#ifdef IEEE_DEBUG
/* Test some numbers to see that extended/double conversion works for them. */
ieee_test (n)
int n;
{
union { double d; int i[2]; } di;
double result;
int i;
char exten[16];
extern struct ext_format ext_format_68881;
for (i = 0; i < n; i++) {
di.i[0] = (random() << 16) | (random() & 0xffff);
di.i[1] = (random() << 16) | (random() & 0xffff);
double_to_ieee_extended (&ext_format_68881, &di.d, exten);
ieee_extended_to_double (&ext_format_68881, exten, &result);
if (di.d != result)
printf ("Differ: %x %x %g => %x %x %g\n", di.d, di.d, result, result);
}
}
#endif
These are the contents of the former NiCE NeXT User Group NeXTSTEP/OpenStep software archive, currently hosted by Netfuture.ch.