Subversion Repositories HelenOS-historic

/*

 * Copyright (C) 2005 Josef Cejka

 * All rights reserved.

 *

 * Redistribution and use in source and binary forms, with or without

 * modification, are permitted provided that the following conditions

 * are met:

 *

 * - Redistributions of source code must retain the above copyright

 *   notice, this list of conditions and the following disclaimer.

 * - Redistributions in binary form must reproduce the above copyright

 *   notice, this list of conditions and the following disclaimer in the

 *   documentation and/or other materials provided with the distribution.

 * - The name of the author may not be used to endorse or promote products

 *   derived from this software without specific prior written permission.

 *

 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR

 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES

 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.

 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,

 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT

 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,

 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY

 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT

 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF

 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 */

#include<sftypes.h>

#include<mul.h>

#include<comparison.h>

/** Multiply two 32 bit float numbers

 *

 */

float32 mulFloat32(float32 a, float32 b)

{

    float32 result;

    __u64 mant1, mant2;

    __s32 exp;

    result.parts.sign = a.parts.sign ^ b.parts.sign;

    if (isFloat32NaN(a) || isFloat32NaN(b) ) {

        /* TODO: fix SigNaNs */

        if (isFloat32SigNaN(a)) {

            result.parts.mantisa = a.parts.mantisa;

            result.parts.exp = a.parts.exp;

            return result;

        };

        if (isFloat32SigNaN(b)) { /* TODO: fix SigNaN */

            result.parts.mantisa = b.parts.mantisa;

            result.parts.exp = b.parts.exp;

            return result;

        };

        /* set NaN as result */

        result.binary = FLOAT32_NAN;

        return result;

    };

    if (isFloat32Infinity(a)) {

        if (isFloat32Zero(b)) {

            /* FIXME: zero * infinity */

            result.binary = FLOAT32_NAN;

            return result;

        }

        result.parts.mantisa = a.parts.mantisa;

        result.parts.exp = a.parts.exp;

        return result;

    }

    if (isFloat32Infinity(b)) {

        if (isFloat32Zero(a)) {

            /* FIXME: zero * infinity */

            result.binary = FLOAT32_NAN;

            return result;

        }

        result.parts.mantisa = b.parts.mantisa;

        result.parts.exp = b.parts.exp;

        return result;

    }

    /* exp is signed so we can easy detect underflow */

    exp = a.parts.exp + b.parts.exp;

    exp -= FLOAT32_BIAS;

    if (exp >= FLOAT32_MAX_EXPONENT) {

        /* FIXME: overflow */

        /* set infinity as result */

        result.binary = FLOAT32_INF;

        result.parts.sign = a.parts.sign ^ b.parts.sign;

        return result;

    };

    if (exp < 0) {

        /* FIXME: underflow */

        /* return signed zero */

        result.parts.mantisa = 0x0;

        result.parts.exp = 0x0;

        return result;

    };

    mant1 = a.parts.mantisa;

    if (a.parts.exp > 0) {

        mant1 |= FLOAT32_HIDDEN_BIT_MASK;

    } else {

        ++exp;

    };

    mant2 = b.parts.mantisa;

    if (b.parts.exp > 0) {

        mant2 |= FLOAT32_HIDDEN_BIT_MASK;

    } else {

        ++exp;

    };

    mant1 <<= 1; /* one bit space for rounding */

    mant1 = mant1 * mant2;

/* round and return */

    while ((exp < FLOAT32_MAX_EXPONENT) && (mant1 >= ( 1 << (FLOAT32_MANTISA_SIZE + 2)))) {

        /* 23 bits of mantisa + one more for hidden bit (all shifted 1 bit left)*/

        ++exp;

        mant1 >>= 1;

    };

    /* rounding */

    //++mant1; /* FIXME: not works - without it is ok */

    mant1 >>= 1; /* shift off rounding space */

    if ((exp < FLOAT32_MAX_EXPONENT) && (mant1 >= (1 << (FLOAT32_MANTISA_SIZE + 1)))) {

        ++exp;

        mant1 >>= 1;

    };

    if (exp >= FLOAT32_MAX_EXPONENT ) {

        /* TODO: fix overflow */

        /* return infinity*/

        result.parts.exp = FLOAT32_MAX_EXPONENT;

        result.parts.mantisa = 0x0;

        return result;

    }

    exp -= FLOAT32_MANTISA_SIZE;

    if (exp <= FLOAT32_MANTISA_SIZE) {

        /* denormalized number */

        mant1 >>= 1; /* denormalize */

        while ((mant1 > 0) && (exp < 0)) {

            mant1 >>= 1;

            ++exp;

        };

        if (mant1 == 0) {

            /* FIXME : underflow */

        result.parts.exp = 0;

        result.parts.mantisa = 0;

        return result;

        };

    };

    result.parts.exp = exp;

    result.parts.mantisa = mant1 & ( (1 << FLOAT32_MANTISA_SIZE) - 1);

    return result; 

}

/** Multiply two 64 bit float numbers

 *

 */

float64 mulFloat64(float64 a, float64 b)

{

    float64 result;

    __u64 mant1, mant2;

    __s32 exp;

    result.parts.sign = a.parts.sign ^ b.parts.sign;

    if (isFloat64NaN(a) || isFloat64NaN(b) ) {

        /* TODO: fix SigNaNs */

        if (isFloat64SigNaN(a)) {

            result.parts.mantisa = a.parts.mantisa;

            result.parts.exp = a.parts.exp;

            return result;

        };

        if (isFloat64SigNaN(b)) { /* TODO: fix SigNaN */

            result.parts.mantisa = b.parts.mantisa;

            result.parts.exp = b.parts.exp;

            return result;

        };

        /* set NaN as result */

        result.binary = FLOAT64_NAN;

        return result;

    };

    if (isFloat64Infinity(a)) {

        if (isFloat64Zero(b)) {

            /* FIXME: zero * infinity */

            result.binary = FLOAT64_NAN;

            return result;

        }

        result.parts.mantisa = a.parts.mantisa;

        result.parts.exp = a.parts.exp;

        return result;

    }

    if (isFloat64Infinity(b)) {

        if (isFloat64Zero(a)) {

            /* FIXME: zero * infinity */

            result.binary = FLOAT64_NAN;

            return result;

        }

        result.parts.mantisa = b.parts.mantisa;

        result.parts.exp = b.parts.exp;

        return result;

    }

    /* exp is signed so we can easy detect underflow */

    exp = a.parts.exp + b.parts.exp;

    exp -= FLOAT64_BIAS;

    if (exp >= FLOAT64_MAX_EXPONENT) {

        /* FIXME: overflow */

        /* set infinity as result */

        result.binary = FLOAT64_INF;

        result.parts.sign = a.parts.sign ^ b.parts.sign;

        return result;

    };

    if (exp < 0) {

        /* FIXME: underflow */

        /* return signed zero */

        result.parts.mantisa = 0x0;

        result.parts.exp = 0x0;

        return result;

    };

    mant1 = a.parts.mantisa;

    if (a.parts.exp > 0) {

        mant1 |= FLOAT64_HIDDEN_BIT_MASK;

    } else {

        ++exp;

    };

    mant2 = b.parts.mantisa;

    if (b.parts.exp > 0) {

        mant2 |= FLOAT64_HIDDEN_BIT_MASK;

    } else {

        ++exp;

    };

    mant1 <<= 1; /* one bit space for rounding */

    mul64integers(mant1, mant2, &mant1, &mant2);

/* round and return */

    /* FIXME: ugly soulution is to shift whole mant2 >> as in 32bit version

     * Here is is more slower because we have to shift two numbers with carry

     * Better is find first nonzero bit and make only one shift

     * Third version is to shift both numbers a bit to right and result will be then

     * placed in higher part of result. Then lower part will be good only for rounding.

     */

    while ((exp < FLOAT64_MAX_EXPONENT) && (mant2 > 0 )) {

        mant1 >>= 1;

        mant1 &= ((mant2 & 0x1) << 63);

        mant2 >>= 1;

        ++exp;

    }

    while ((exp < FLOAT64_MAX_EXPONENT) && (mant1 >= ( (__u64)1 << (FLOAT64_MANTISA_SIZE + 2)))) {

        ++exp;

        mant1 >>= 1;

    };

    /* rounding */

    //++mant1; /* FIXME: not works - without it is ok */

    mant1 >>= 1; /* shift off rounding space */

    if ((exp < FLOAT64_MAX_EXPONENT) && (mant1 >= ((__u64)1 << (FLOAT64_MANTISA_SIZE + 1)))) {

        ++exp;

        mant1 >>= 1;

    };

    if (exp >= FLOAT64_MAX_EXPONENT ) {

        /* TODO: fix overflow */

        /* return infinity*/

        result.parts.exp = FLOAT64_MAX_EXPONENT;

        result.parts.mantisa = 0x0;

        return result;

    }

    exp -= FLOAT64_MANTISA_SIZE;

    if (exp <= FLOAT64_MANTISA_SIZE) {

        /* denormalized number */

        mant1 >>= 1; /* denormalize */

        while ((mant1 > 0) && (exp < 0)) {

            mant1 >>= 1;

            ++exp;

        };

        if (mant1 == 0) {

            /* FIXME : underflow */

        result.parts.exp = 0;

        result.parts.mantisa = 0;

        return result;

        };

    };

    result.parts.exp = exp;

    result.parts.mantisa = mant1 & ( ((__u64)1 << FLOAT64_MANTISA_SIZE) - 1);

    return result; 

}

/** Multiply two 64 bit numbers and return result in two parts

 * @param a first operand

 * @param b second operand

 * @param lo lower part from result

 * @param hi higher part of result

 */

void mul64integers(__u64 a,__u64 b, __u64 *lo, __u64 *hi)

{

    __u64 low, high, middle1, middle2;

    __u32 alow, blow;

    alow = a & 0xFFFFFFFF;

    blow = b & 0xFFFFFFFF;

    a <<= 32;

    b <<= 32;

    low = (__u64)alow * blow;

    middle1 = a * blow;

    middle2 = alow * b;

    high = a * b;

    middle1 += middle2;

    high += ((__u64)(middle1 < middle2) << 32) + middle1>>32;

    middle1 << 32;

    low += middle1;

    high += (low < middle1);

    *lo = low;

    *hi = high;

    return;

}