OpenSolaris_b135/lib/libnsl/rpc/xdr_float.c

Compare this file to the similar file:
Show the results in this format:

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */

/*
 * Copyright 2006 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

/* Copyright (c) 1983, 1984, 1985, 1986, 1987, 1988, 1989 AT&T */
/* All Rights Reserved */
/*
 * Portions of this source code were derived from Berkeley
 * 4.3 BSD under license from the Regents of the University of
 * California.
 */

#pragma ident	"%Z%%M%	%I%	%E% SMI"

/*
 * Generic XDR routines impelmentation.
 *
 * These are the "floating point" xdr routines used to (de)serialize
 * most common data items.  See xdr.h for more info on the interface to
 * xdr.
 */

#include "mt.h"
#include <sys/types.h>
#include <stdio.h>
#include <rpc/types.h>
#include <rpc/xdr.h>

/*
 * This routine works on Suns, 3b2, 68000s, 386 and Vaxen in a manner
 * which is very efficient as bit twiddling is all that is needed.  All
 * other machines can use this code but the code is inefficient as
 * various mathematical operations are used to generate the ieee format.
 * In addition rounding errors may occur due to the calculations involved.
 * To be most efficient, new machines should have their own ifdefs.
 * The encoding routines will fail if the machines try to encode a
 * float/double whose value can not be represented by the ieee format,
 * e.g. the exponent is too big/small.
 *	ieee largest  float  = (2 ^ 128)  * 0x1.fffff
 *	ieee smallest float  = (2 ^ -127) * 0x1.00000
 *	ieee largest  double = (2 ^ 1024)  * 0x1.fffff
 *	ieee smallest double = (2 ^ -1023) * 0x1.00000
 * The decoding routines assumes that the receiving machine can handle
 * floats/doubles as large/small as the values stated above.  If you
 * use a machine which can not represent these values, you will need
 * to put ifdefs in the decode sections to identify areas of failure.
 */

#if defined(vax)

/*
 * What IEEE single precision floating point looks like this on a
 * vax.
 */

struct	ieee_single {
	unsigned int	mantissa: 23;
	unsigned int	exp	: 8;
	unsigned int	sign    : 1;
};

#define	IEEE_SNG_BIAS	0x7f
#define	VAX_SNG_BIAS    0x81


/* Vax single precision floating point */
struct	vax_single {
	unsigned int	mantissa1 : 7;
	unsigned int	exp	: 8;
	unsigned int	sign	: 1;
	unsigned int	mantissa2 : 16;
};

#define	VAX_SNG_BIAS	0x81

static struct sgl_limits {
	struct vax_single s;
	struct ieee_single ieee;
} sgl_limits[2] = {
	{{ 0x7f, 0xff, 0x0, 0xffff },	/* Max Vax */
	{ 0x0, 0xff, 0x0 }},		/* Max IEEE */
	{{ 0x0, 0x0, 0x0, 0x0 },	/* Min Vax */
	{ 0x0, 0x0, 0x0 }}		/* Min IEEE */
};
#endif /* vax */

bool_t
xdr_float(XDR *xdrs, float *fp)
{
#if defined(vax)
	struct ieee_single is;
	struct vax_single vs, *vsp;
	struct sgl_limits *lim;
	size_t i;
#endif

	switch (xdrs->x_op) {

	case XDR_ENCODE:
#if defined(mc68000) || defined(sparc) || defined(u3b2) || \
	defined(u3b15) || defined(i386) || defined(amd64)
		return (XDR_PUTINT32(xdrs, (int *)fp));
#else
#if defined(vax)
		vs = *((struct vax_single *)fp);
		if ((vs.exp == 1) || (vs.exp == 2)) {
			/* map these to subnormals */
			is.exp = 0;
			is.mantissa = (vs.mantissa1 << 16) | vs.mantissa2;
			/* lose some precision */
			is.mantissa >>= 3 - vs.exp;
			is.mantissa += (1 << (20 + vs.exp));
			goto shipit;
		}
		for (i = 0, lim = sgl_limits;
			i < (int)(sizeof (sgl_limits) /
					sizeof (struct sgl_limits));
			i++, lim++) {
			if ((vs.mantissa2 == lim->s.mantissa2) &&
				(vs.exp == lim->s.exp) &&
				(vs.mantissa1 == lim->s.mantissa1)) {
				is = lim->ieee;
				goto shipit;
			}
		}
		is.exp = vs.exp - VAX_SNG_BIAS + IEEE_SNG_BIAS;
		is.mantissa = (vs.mantissa1 << 16) | vs.mantissa2;
	shipit:
		is.sign = vs.sign;
		return (XDR_PUTINT32(xdrs, (int32_t *)&is));
#else
		{
		/*
		 * Every machine can do this, its just not very efficient.
		 * In addtion, some rounding errors may occur do to the
		 * calculations involved.
		 */
		float f;
		int neg = 0;
		int exp = 0;
		int32_t val;

		f = *fp;
		if (f == 0) {
			val = 0;
			return (XDR_PUTINT32(xdrs, &val));
		}
		if (f < 0) {
			f = 0 - f;
			neg = 1;
		}
		while (f < 1) {
			f = f * 2;
			--exp;
		}
		while (f >= 2) {
			f = f/2;
			++exp;
		}
		if ((exp > 128) || (exp < -127)) {
			/* over or under flowing ieee exponent */
			return (FALSE);
		}
		val = neg;
		val = val << 8;		/* for the exponent */
		val += 127 + exp;	/* 127 is the bias */
		val = val << 23;	/* for the mantissa */
		val += (int32_t)((f - 1) * 8388608);	/* 2 ^ 23 */
		return (XDR_PUTINT32(xdrs, &val));
		}
#endif
#endif

	case XDR_DECODE:
#if defined(mc68000) || defined(sparc) || defined(u3b2) || \
	defined(u3b15) || defined(i386) || defined(amd64)
		return (XDR_GETINT32(xdrs, (int *)fp));
#else
#if defined(vax)
		vsp = (struct vax_single *)fp;
		if (!XDR_GETINT32(xdrs, (int32_t *)&is))
			return (FALSE);

		for (i = 0, lim = sgl_limits;
			i < (int)(sizeof (sgl_limits) /
					sizeof (struct sgl_limits));
			i++, lim++) {
			if ((is.exp == lim->ieee.exp) &&
				(is.mantissa == lim->ieee.mantissa)) {
				*vsp = lim->s;
				goto doneit;
			} else if ((is.exp == 0) && (lim->ieee.exp == 0)) {
			    /* Special Case */
			    unsigned tmp = is.mantissa >> 20;
			    if (tmp >= 4) {
			    vsp->exp = 2;
			    } else if (tmp >= 2) {
			    vsp->exp = 1;
			    } else {
				*vsp = min.s;
				break;
			    }	/* else */
			    tmp = is.mantissa - (1 << (20 + vsp->exp));
			    tmp <<= 3 - vsp->exp;
			    vsp->mantissa2 = tmp;
			    vsp->mantissa1 = (tmp >> 16);
			    goto doneit;
		    }
		vsp->exp = is.exp - IEEE_SNG_BIAS + VAX_SNG_BIAS;
		vsp->mantissa2 = is.mantissa;
		vsp->mantissa1 = (is.mantissa >> 16);
	doneit:
		vsp->sign = is.sign;
		return (TRUE);
#else
		{
		/*
		 * Every machine can do this, its just not very
		 * efficient.  It assumes that the decoding machine's
		 * float can represent any value in the range of
		 *	ieee largest  float  = (2 ^ 128)  * 0x1.fffff
		 *	to
		 *	ieee smallest float  = (2 ^ -127) * 0x1.00000
		 * In addtion, some rounding errors may occur do to the
		 * calculations involved.
		 */
		float f;
		int neg = 0;
		int exp = 0;
		int32_t val;

		if (!XDR_GETINT32(xdrs, (int32_t *)&val))
			return (FALSE);
		neg = val & 0x80000000;
		exp = (val & 0x7f800000) >> 23;
		exp -= 127;		/* subtract exponent base */
		f = (val & 0x007fffff) * 0.00000011920928955078125;
		/* 2 ^ -23 */
		f++;
		while (exp != 0) {
			if (exp < 0) {
				f = f/2.0;
				++exp;
			} else {
				f = f * 2.0;
				--exp;
			}
		}
		if (neg)
			f = 0 - f;
		*fp = f;
		}
		return (TRUE);
#endif
#endif

	case XDR_FREE:
		return (TRUE);
	}
	return (FALSE);
}

/*
 * This routine works on Suns (Sky / 68000's) and Vaxen.
 */

#if defined(vax)
/* What IEEE double precision floating point looks like on a Vax */
struct	ieee_double {
	unsigned int	mantissa1 : 20;
	unsigned int	exp	  : 11;
	unsigned int	sign	  : 1;
	unsigned int	mantissa2 : 32;
};

/* Vax double precision floating point */
struct  vax_double {
	unsigned int	mantissa1 : 7;
	unsigned int	exp	  : 8;
	unsigned int	sign	  : 1;
	unsigned int	mantissa2 : 16;
	unsigned int	mantissa3 : 16;
	unsigned int	mantissa4 : 16;
};

#define	VAX_DBL_BIAS	0x81
#define	IEEE_DBL_BIAS	0x3ff
#define	MASK(nbits)	((1 << nbits) - 1)

static struct dbl_limits {
	struct	vax_double d;
	struct	ieee_double ieee;
} dbl_limits[2] = {
	{{ 0x7f, 0xff, 0x0, 0xffff, 0xffff, 0xffff },	/* Max Vax */
	{ 0x0, 0x7ff, 0x0, 0x0 }},			/* Max IEEE */
	{{ 0x0, 0x0, 0x0, 0x0, 0x0, 0x0},		/* Min Vax */
	{ 0x0, 0x0, 0x0, 0x0 }}				/* Min IEEE */
};

#endif /* vax */


bool_t
xdr_double(XDR *xdrs, double *dp)
{
	int *lp;
#if defined(vax)
	struct	ieee_double id;
	struct	vax_double vd;
	struct dbl_limits *lim;
	size_t i;
#endif

	switch (xdrs->x_op) {

	case XDR_ENCODE:
#if defined(mc68000) || defined(u3b2) || defined(u3b15) || \
	defined(_LONG_LONG_HTOL)
		lp = (int *)dp;
		return (XDR_PUTINT32(xdrs, lp++) && XDR_PUTINT32(xdrs, lp));
#else
#if defined(_LONG_LONG_LTOH)
		lp = (int *)dp;
		lp++;
		return (XDR_PUTINT32(xdrs, lp--) && XDR_PUTINT32(xdrs, lp));
#else
#if defined(vax)
		vd = *((struct vax_double *)dp);
		for (i = 0, lim = dbl_limits;
			i < (int)(sizeof (dbl_limits) /
					sizeof (struct dbl_limits));
			i++, lim++) {
			if ((vd.mantissa4 == lim->d.mantissa4) &&
				(vd.mantissa3 == lim->d.mantissa3) &&
				(vd.mantissa2 == lim->d.mantissa2) &&
				(vd.mantissa1 == lim->d.mantissa1) &&
				(vd.exp == lim->d.exp)) {
				id = lim->ieee;
				goto shipit;
			}
		}
		id.exp = vd.exp - VAX_DBL_BIAS + IEEE_DBL_BIAS;
		id.mantissa1 = (vd.mantissa1 << 13) | (vd.mantissa2 >> 3);
		id.mantissa2 = ((vd.mantissa2 & MASK(3)) << 29) |
				(vd.mantissa3 << 13) |
				((vd.mantissa4 >> 3) & MASK(13));
	shipit:
		id.sign = vd.sign;
		lp = (int32_t *)&id;
#else
		{
		/*
		 * Every machine can do this, its just not very efficient.
		 * In addtion, some rounding errors may occur do to the
		 * calculations involved.
		 */
		double d;
		int neg = 0;
		int exp = 0;
		int32_t val[2];

		d = *dp;
		if (d == 0) {
			val[0] = 0;
			val[1] = 0;
			lp = val;
			return (XDR_PUTINT32(xdrs, lp++) &&
				XDR_PUTINT32(xdrs, lp));
		}
		if (d < 0) {
			d = 0 - d;
			neg = 1;
		}
		while (d < 1) {
			d = d * 2;
			--exp;
		}
		while (d >= 2) {
			d = d/2;
			++exp;
		}
		if ((exp > 1024) || (exp < -1023)) {
			/* over or under flowing ieee exponent */
			return (FALSE);
		}
		val[0] = neg;
		val[0] = val[0] << 11;	/* for the exponent */
		val[0] += 1023 + exp;	/* 1023 is the bias */
		val[0] = val[0] << 20;	/* for the mantissa */
		val[0] += (int32_t)((d - 1) * 1048576);	/* 2 ^ 20 */
		val[1] += (int32_t)((((d - 1) * 1048576) - val[0])
							* 4294967296);
		/* 2 ^ 32 */
		lp = val;
		}
#endif
		return (XDR_PUTINT32(xdrs, lp++) && XDR_PUTINT32(xdrs, lp));
#endif
#endif

	case XDR_DECODE:
#if defined(mc68000) || defined(u3b2) || defined(u3b15) || \
	defined(_LONG_LONG_HTOL)
		lp = (int *)dp;
		return (XDR_GETINT32(xdrs, lp++) && XDR_GETINT32(xdrs, lp));
#else
#if defined(_LONG_LONG_LTOH)
		lp = (int *)dp;
		lp++;
		return (XDR_GETINT32(xdrs, lp--) && XDR_GETINT32(xdrs, lp));
#else
#if defined(vax)
		lp = (int32_t *)&id;
		if (!XDR_GETINT32(xdrs, lp++) || !XDR_GETINT32(xdrs, lp))
			return (FALSE);
		for (i = 0, lim = dbl_limits;
			i < sizeof (dbl_limits)/sizeof (struct dbl_limits);
			i++, lim++) {
			if ((id.mantissa2 == lim->ieee.mantissa2) &&
				(id.mantissa1 == lim->ieee.mantissa1) &&
				(id.exp == lim->ieee.exp)) {
				vd = lim->d;
				goto doneit;
			}
		}
		vd.exp = id.exp - IEEE_DBL_BIAS + VAX_DBL_BIAS;
		vd.mantissa1 = (id.mantissa1 >> 13);
		vd.mantissa2 = ((id.mantissa1 & MASK(13)) << 3) |
				(id.mantissa2 >> 29);
		vd.mantissa3 = (id.mantissa2 >> 13);
		vd.mantissa4 = (id.mantissa2 << 3);
	doneit:
		vd.sign = id.sign;
		*dp = *((double *)&vd);
		return (TRUE);
#else
		{
		/*
		 * Every machine can do this, its just not very
		 * efficient.  It assumes that the decoding machine's
		 * double can represent any value in the range of
		 *	ieee largest  double  = (2 ^ 1024)  * 0x1.fffffffffffff
		 *	to
		 *	ieee smallest double  = (2 ^ -1023) * 0x1.0000000000000
		 * In addtion, some rounding errors may occur do to the
		 * calculations involved.
		 */
		double d;
		int neg = 0;
		int exp = 0;
		int32_t val[2];

		lp = val;
		if (!XDR_GETINT32(xdrs, lp++) || !XDR_GETINT32(xdrs, lp))
			return (FALSE);
		neg = val[0] & 0x80000000;
		exp = (val[0] & 0x7ff00000) >> 20;
		exp -= 1023;		/* subtract exponent base */
		d = (val[0] & 0x000fffff) * 0.00000095367431640625;
		/* 2 ^ -20 */
		d += (val[1] * 0.0000000000000002220446049250313);
		/* 2 ^ -52 */
		d++;
		while (exp != 0) {
			if (exp < 0) {
				d = d/2.0;
				++exp;
			} else {
				d = d * 2.0;
				--exp;
			}
		}
		if (neg)
			d = 0 - d;
		*dp = d;
		}
#endif
#endif
#endif

	case XDR_FREE:
		return (TRUE);
	}
	return (FALSE);
}

/* ARGSUSED */
bool_t
xdr_quadruple(XDR *xdrs, long double *fp)
{
/*
 * The Sparc uses IEEE FP encoding, so just do a byte copy
 */

#if !defined(sparc)
	return (FALSE);
#else
	switch (xdrs->x_op) {
	case XDR_ENCODE:
		return (XDR_PUTBYTES(xdrs, (char *)fp, sizeof (long double)));
	case XDR_DECODE:
		return (XDR_GETBYTES(xdrs, (char *)fp, sizeof (long double)));
	case XDR_FREE:
		return (TRUE);
	}
	return (FALSE);
#endif
}