Commit 4b5c65b8 authored by Laurent Vivier's avatar Laurent Vivier
Browse files

target/m68k: implement flognp1 



Using a local m68k floatx80_lognp1()
[copied from previous:
Written by Andreas Grabher for Previous, NeXT Computer Emulator.]

Signed-off-by: default avatarLaurent Vivier <laurent@vivier.eu>
Message-Id: <20180305203910.10391-3-laurent@vivier.eu>
parent 9a069775

target/m68k/fpu_helper.c

+5 −0
Original line number Diff line number Diff line
@@ -557,3 +557,8 @@ void HELPER(fscale)(CPUM68KState *env, FPReg *res, FPReg *val0, FPReg *val1) 
{

    res->d = floatx80_scale(val1->d, val0->d, &env->fp_status);

}



void HELPER(flognp1)(CPUM68KState *env, FPReg *res, FPReg *val)

{

    res->d = floatx80_lognp1(val->d, &env->fp_status);

}

target/m68k/helper.h

+1 −0
Original line number Diff line number Diff line
@@ -68,6 +68,7 @@ DEF_HELPER_4(frem, void, env, fp, fp, fp) 
DEF_HELPER_3(fgetexp, void, env, fp, fp)

DEF_HELPER_3(fgetman, void, env, fp, fp)

DEF_HELPER_4(fscale, void, env, fp, fp, fp)

DEF_HELPER_3(flognp1, void, env, fp, fp)



DEF_HELPER_3(mac_move, void, env, i32, i32)

DEF_HELPER_3(macmulf, i64, env, i32, i32)

target/m68k/softfloat.c

+216 −0
Original line number Diff line number Diff line
@@ -21,6 +21,7 @@ 
#include "qemu/osdep.h"

#include "softfloat.h"

#include "fpu/softfloat-macros.h"

#include "softfloat_fpsp_tables.h"



static floatx80 propagateFloatx80NaNOneArg(floatx80 a, float_status *status)

{
@@ -274,3 +275,218 @@ floatx80 floatx80_move(floatx80 a, float_status *status) 
    return roundAndPackFloatx80(status->floatx80_rounding_precision, aSign,

                                aExp, aSig, 0, status);

}



/*----------------------------------------------------------------------------

| Algorithms for transcendental functions supported by MC68881 and MC68882

| mathematical coprocessors. The functions are derived from FPSP library.

*----------------------------------------------------------------------------*/



#define one_exp     0x3FFF

#define one_sig     LIT64(0x8000000000000000)



/*----------------------------------------------------------------------------

 | Function for compactifying extended double-precision floating point values.

 *----------------------------------------------------------------------------*/



static int32_t floatx80_make_compact(int32_t aExp, uint64_t aSig)

{

    return (aExp << 16) | (aSig >> 48);

}



/*----------------------------------------------------------------------------

 | Log base e of x plus 1

 *----------------------------------------------------------------------------*/



floatx80 floatx80_lognp1(floatx80 a, float_status *status)

{

    flag aSign;

    int32_t aExp;

    uint64_t aSig, fSig;



    int8_t user_rnd_mode, user_rnd_prec;



    int32_t compact, j, k;

    floatx80 fp0, fp1, fp2, fp3, f, logof2, klog2, saveu;



    aSig = extractFloatx80Frac(a);

    aExp = extractFloatx80Exp(a);

    aSign = extractFloatx80Sign(a);



    if (aExp == 0x7FFF) {

        if ((uint64_t) (aSig << 1)) {

            propagateFloatx80NaNOneArg(a, status);

        }

        if (aSign) {

            float_raise(float_flag_invalid, status);

            return floatx80_default_nan(status);

        }

        return packFloatx80(0, floatx80_infinity.high, floatx80_infinity.low);

    }



    if (aExp == 0 && aSig == 0) {

        return packFloatx80(aSign, 0, 0);

    }



    if (aSign && aExp >= one_exp) {

        if (aExp == one_exp && aSig == one_sig) {

            float_raise(float_flag_divbyzero, status);

            packFloatx80(aSign, floatx80_infinity.high, floatx80_infinity.low);

        }

        float_raise(float_flag_invalid, status);

        return floatx80_default_nan(status);

    }



    if (aExp < 0x3f99 || (aExp == 0x3f99 && aSig == one_sig)) {

        /* <= min threshold */

        float_raise(float_flag_inexact, status);

        return floatx80_move(a, status);

    }



    user_rnd_mode = status->float_rounding_mode;

    user_rnd_prec = status->floatx80_rounding_precision;

    status->float_rounding_mode = float_round_nearest_even;

    status->floatx80_rounding_precision = 80;



    compact = floatx80_make_compact(aExp, aSig);



    fp0 = a; /* Z */

    fp1 = a;



    fp0 = floatx80_add(fp0, float32_to_floatx80(make_float32(0x3F800000),

                       status), status); /* X = (1+Z) */



    aExp = extractFloatx80Exp(fp0);

    aSig = extractFloatx80Frac(fp0);



    compact = floatx80_make_compact(aExp, aSig);



    if (compact < 0x3FFE8000 || compact > 0x3FFFC000) {

        /* |X| < 1/2 or |X| > 3/2 */

        k = aExp - 0x3FFF;

        fp1 = int32_to_floatx80(k, status);



        fSig = (aSig & LIT64(0xFE00000000000000)) | LIT64(0x0100000000000000);

        j = (fSig >> 56) & 0x7E; /* DISPLACEMENT FOR 1/F */



        f = packFloatx80(0, 0x3FFF, fSig); /* F */

        fp0 = packFloatx80(0, 0x3FFF, aSig); /* Y */



        fp0 = floatx80_sub(fp0, f, status); /* Y-F */



    lp1cont1:

        /* LP1CONT1 */

        fp0 = floatx80_mul(fp0, log_tbl[j], status); /* FP0 IS U = (Y-F)/F */

        logof2 = packFloatx80(0, 0x3FFE, LIT64(0xB17217F7D1CF79AC));

        klog2 = floatx80_mul(fp1, logof2, status); /* FP1 IS K*LOG2 */

        fp2 = floatx80_mul(fp0, fp0, status); /* FP2 IS V=U*U */



        fp3 = fp2;

        fp1 = fp2;



        fp1 = floatx80_mul(fp1, float64_to_floatx80(

                           make_float64(0x3FC2499AB5E4040B), status),

                           status); /* V*A6 */

        fp2 = floatx80_mul(fp2, float64_to_floatx80(

                           make_float64(0xBFC555B5848CB7DB), status),

                           status); /* V*A5 */

        fp1 = floatx80_add(fp1, float64_to_floatx80(

                           make_float64(0x3FC99999987D8730), status),

                           status); /* A4+V*A6 */

        fp2 = floatx80_add(fp2, float64_to_floatx80(

                           make_float64(0xBFCFFFFFFF6F7E97), status),

                           status); /* A3+V*A5 */

        fp1 = floatx80_mul(fp1, fp3, status); /* V*(A4+V*A6) */

        fp2 = floatx80_mul(fp2, fp3, status); /* V*(A3+V*A5) */

        fp1 = floatx80_add(fp1, float64_to_floatx80(

                           make_float64(0x3FD55555555555A4), status),

                           status); /* A2+V*(A4+V*A6) */

        fp2 = floatx80_add(fp2, float64_to_floatx80(

                           make_float64(0xBFE0000000000008), status),

                           status); /* A1+V*(A3+V*A5) */

        fp1 = floatx80_mul(fp1, fp3, status); /* V*(A2+V*(A4+V*A6)) */

        fp2 = floatx80_mul(fp2, fp3, status); /* V*(A1+V*(A3+V*A5)) */

        fp1 = floatx80_mul(fp1, fp0, status); /* U*V*(A2+V*(A4+V*A6)) */

        fp0 = floatx80_add(fp0, fp2, status); /* U+V*(A1+V*(A3+V*A5)) */



        fp1 = floatx80_add(fp1, log_tbl[j + 1],

                           status); /* LOG(F)+U*V*(A2+V*(A4+V*A6)) */

        fp0 = floatx80_add(fp0, fp1, status); /* FP0 IS LOG(F) + LOG(1+U) */



        status->float_rounding_mode = user_rnd_mode;

        status->floatx80_rounding_precision = user_rnd_prec;



        a = floatx80_add(fp0, klog2, status);



        float_raise(float_flag_inexact, status);



        return a;

    } else if (compact < 0x3FFEF07D || compact > 0x3FFF8841) {

        /* |X| < 1/16 or |X| > -1/16 */

        /* LP1CARE */

        fSig = (aSig & LIT64(0xFE00000000000000)) | LIT64(0x0100000000000000);

        f = packFloatx80(0, 0x3FFF, fSig); /* F */

        j = (fSig >> 56) & 0x7E; /* DISPLACEMENT FOR 1/F */



        if (compact >= 0x3FFF8000) { /* 1+Z >= 1 */

            /* KISZERO */

            fp0 = floatx80_sub(float32_to_floatx80(make_float32(0x3F800000),

                               status), f, status); /* 1-F */

            fp0 = floatx80_add(fp0, fp1, status); /* FP0 IS Y-F = (1-F)+Z */

            fp1 = packFloatx80(0, 0, 0); /* K = 0 */

        } else {

            /* KISNEG */

            fp0 = floatx80_sub(float32_to_floatx80(make_float32(0x40000000),

                               status), f, status); /* 2-F */

            fp1 = floatx80_add(fp1, fp1, status); /* 2Z */

            fp0 = floatx80_add(fp0, fp1, status); /* FP0 IS Y-F = (2-F)+2Z */

            fp1 = packFloatx80(1, one_exp, one_sig); /* K = -1 */

        }

        goto lp1cont1;

    } else {

        /* LP1ONE16 */

        fp1 = floatx80_add(fp1, fp1, status); /* FP1 IS 2Z */

        fp0 = floatx80_add(fp0, float32_to_floatx80(make_float32(0x3F800000),

                           status), status); /* FP0 IS 1+X */



        /* LP1CONT2 */

        fp1 = floatx80_div(fp1, fp0, status); /* U */

        saveu = fp1;

        fp0 = floatx80_mul(fp1, fp1, status); /* FP0 IS V = U*U */

        fp1 = floatx80_mul(fp0, fp0, status); /* FP1 IS W = V*V */



        fp3 = float64_to_floatx80(make_float64(0x3F175496ADD7DAD6),

                                  status); /* B5 */

        fp2 = float64_to_floatx80(make_float64(0x3F3C71C2FE80C7E0),

                                  status); /* B4 */

        fp3 = floatx80_mul(fp3, fp1, status); /* W*B5 */

        fp2 = floatx80_mul(fp2, fp1, status); /* W*B4 */

        fp3 = floatx80_add(fp3, float64_to_floatx80(

                           make_float64(0x3F624924928BCCFF), status),

                           status); /* B3+W*B5 */

        fp2 = floatx80_add(fp2, float64_to_floatx80(

                           make_float64(0x3F899999999995EC), status),

                           status); /* B2+W*B4 */

        fp1 = floatx80_mul(fp1, fp3, status); /* W*(B3+W*B5) */

        fp2 = floatx80_mul(fp2, fp0, status); /* V*(B2+W*B4) */

        fp1 = floatx80_add(fp1, float64_to_floatx80(

                           make_float64(0x3FB5555555555555), status),

                           status); /* B1+W*(B3+W*B5) */



        fp0 = floatx80_mul(fp0, saveu, status); /* FP0 IS U*V */

        fp1 = floatx80_add(fp1, fp2,

                           status); /* B1+W*(B3+W*B5) + V*(B2+W*B4) */

        fp0 = floatx80_mul(fp0, fp1,

                           status); /* U*V*([B1+W*(B3+W*B5)] + [V*(B2+W*B4)]) */



        status->float_rounding_mode = user_rnd_mode;

        status->floatx80_rounding_precision = user_rnd_prec;



        a = floatx80_add(fp0, saveu, status);



        /*if (!floatx80_is_zero(a)) { */

            float_raise(float_flag_inexact, status);

        /*} */



        return a;

    }

}

target/m68k/softfloat.h

+1 −0
Original line number Diff line number Diff line
@@ -27,4 +27,5 @@ floatx80 floatx80_getman(floatx80 a, float_status *status); 
floatx80 floatx80_getexp(floatx80 a, float_status *status);

floatx80 floatx80_scale(floatx80 a, floatx80 b, float_status *status);

floatx80 floatx80_move(floatx80 a, float_status *status);

floatx80 floatx80_lognp1(floatx80 a, float_status *status);

#endif

target/m68k/softfloat_fpsp_tables.h

0 → 100644
+154 −0
Original line number Diff line number Diff line 
/*

 * Ported from a work by Andreas Grabher for Previous, NeXT Computer Emulator,

 * derived from NetBSD M68040 FPSP functions,

 * derived from release 2a of the SoftFloat IEC/IEEE Floating-point Arithmetic

 * Package. Those parts of the code (and some later contributions) are

 * provided under that license, as detailed below.

 * It has subsequently been modified by contributors to the QEMU Project,

 * so some portions are provided under:

 *  the SoftFloat-2a license

 *  the BSD license

 *  GPL-v2-or-later

 *

 * Any future contributions to this file will be taken to be licensed under

 * the Softfloat-2a license unless specifically indicated otherwise.

 */



/* Portions of this work are licensed under the terms of the GNU GPL,

 * version 2 or later. See the COPYING file in the top-level directory.

 */



#ifndef TARGET_M68K_SOFTFLOAT_FPSP_TABLES_H

#define TARGET_M68K_SOFTFLOAT_FPSP_TABLES_H



static const floatx80 log_tbl[128] = {

    make_floatx80_init(0x3FFE, 0xFE03F80FE03F80FE),

    make_floatx80_init(0x3FF7, 0xFF015358833C47E2),

    make_floatx80_init(0x3FFE, 0xFA232CF252138AC0),

    make_floatx80_init(0x3FF9, 0xBDC8D83EAD88D549),

    make_floatx80_init(0x3FFE, 0xF6603D980F6603DA),

    make_floatx80_init(0x3FFA, 0x9CF43DCFF5EAFD48),

    make_floatx80_init(0x3FFE, 0xF2B9D6480F2B9D65),

    make_floatx80_init(0x3FFA, 0xDA16EB88CB8DF614),

    make_floatx80_init(0x3FFE, 0xEF2EB71FC4345238),

    make_floatx80_init(0x3FFB, 0x8B29B7751BD70743),

    make_floatx80_init(0x3FFE, 0xEBBDB2A5C1619C8C),

    make_floatx80_init(0x3FFB, 0xA8D839F830C1FB49),

    make_floatx80_init(0x3FFE, 0xE865AC7B7603A197),

    make_floatx80_init(0x3FFB, 0xC61A2EB18CD907AD),

    make_floatx80_init(0x3FFE, 0xE525982AF70C880E),

    make_floatx80_init(0x3FFB, 0xE2F2A47ADE3A18AF),

    make_floatx80_init(0x3FFE, 0xE1FC780E1FC780E2),

    make_floatx80_init(0x3FFB, 0xFF64898EDF55D551),

    make_floatx80_init(0x3FFE, 0xDEE95C4CA037BA57),

    make_floatx80_init(0x3FFC, 0x8DB956A97B3D0148),

    make_floatx80_init(0x3FFE, 0xDBEB61EED19C5958),

    make_floatx80_init(0x3FFC, 0x9B8FE100F47BA1DE),

    make_floatx80_init(0x3FFE, 0xD901B2036406C80E),

    make_floatx80_init(0x3FFC, 0xA9372F1D0DA1BD17),

    make_floatx80_init(0x3FFE, 0xD62B80D62B80D62C),

    make_floatx80_init(0x3FFC, 0xB6B07F38CE90E46B),

    make_floatx80_init(0x3FFE, 0xD3680D3680D3680D),

    make_floatx80_init(0x3FFC, 0xC3FD032906488481),

    make_floatx80_init(0x3FFE, 0xD0B69FCBD2580D0B),

    make_floatx80_init(0x3FFC, 0xD11DE0FF15AB18CA),

    make_floatx80_init(0x3FFE, 0xCE168A7725080CE1),

    make_floatx80_init(0x3FFC, 0xDE1433A16C66B150),

    make_floatx80_init(0x3FFE, 0xCB8727C065C393E0),

    make_floatx80_init(0x3FFC, 0xEAE10B5A7DDC8ADD),

    make_floatx80_init(0x3FFE, 0xC907DA4E871146AD),

    make_floatx80_init(0x3FFC, 0xF7856E5EE2C9B291),

    make_floatx80_init(0x3FFE, 0xC6980C6980C6980C),

    make_floatx80_init(0x3FFD, 0x82012CA5A68206D7),

    make_floatx80_init(0x3FFE, 0xC4372F855D824CA6),

    make_floatx80_init(0x3FFD, 0x882C5FCD7256A8C5),

    make_floatx80_init(0x3FFE, 0xC1E4BBD595F6E947),

    make_floatx80_init(0x3FFD, 0x8E44C60B4CCFD7DE),

    make_floatx80_init(0x3FFE, 0xBFA02FE80BFA02FF),

    make_floatx80_init(0x3FFD, 0x944AD09EF4351AF6),

    make_floatx80_init(0x3FFE, 0xBD69104707661AA3),

    make_floatx80_init(0x3FFD, 0x9A3EECD4C3EAA6B2),

    make_floatx80_init(0x3FFE, 0xBB3EE721A54D880C),

    make_floatx80_init(0x3FFD, 0xA0218434353F1DE8),

    make_floatx80_init(0x3FFE, 0xB92143FA36F5E02E),

    make_floatx80_init(0x3FFD, 0xA5F2FCABBBC506DA),

    make_floatx80_init(0x3FFE, 0xB70FBB5A19BE3659),

    make_floatx80_init(0x3FFD, 0xABB3B8BA2AD362A5),

    make_floatx80_init(0x3FFE, 0xB509E68A9B94821F),

    make_floatx80_init(0x3FFD, 0xB1641795CE3CA97B),

    make_floatx80_init(0x3FFE, 0xB30F63528917C80B),

    make_floatx80_init(0x3FFD, 0xB70475515D0F1C61),

    make_floatx80_init(0x3FFE, 0xB11FD3B80B11FD3C),

    make_floatx80_init(0x3FFD, 0xBC952AFEEA3D13E1),

    make_floatx80_init(0x3FFE, 0xAF3ADDC680AF3ADE),

    make_floatx80_init(0x3FFD, 0xC2168ED0F458BA4A),

    make_floatx80_init(0x3FFE, 0xAD602B580AD602B6),

    make_floatx80_init(0x3FFD, 0xC788F439B3163BF1),

    make_floatx80_init(0x3FFE, 0xAB8F69E28359CD11),

    make_floatx80_init(0x3FFD, 0xCCECAC08BF04565D),

    make_floatx80_init(0x3FFE, 0xA9C84A47A07F5638),

    make_floatx80_init(0x3FFD, 0xD24204872DD85160),

    make_floatx80_init(0x3FFE, 0xA80A80A80A80A80B),

    make_floatx80_init(0x3FFD, 0xD78949923BC3588A),

    make_floatx80_init(0x3FFE, 0xA655C4392D7B73A8),

    make_floatx80_init(0x3FFD, 0xDCC2C4B49887DACC),

    make_floatx80_init(0x3FFE, 0xA4A9CF1D96833751),

    make_floatx80_init(0x3FFD, 0xE1EEBD3E6D6A6B9E),

    make_floatx80_init(0x3FFE, 0xA3065E3FAE7CD0E0),

    make_floatx80_init(0x3FFD, 0xE70D785C2F9F5BDC),

    make_floatx80_init(0x3FFE, 0xA16B312EA8FC377D),

    make_floatx80_init(0x3FFD, 0xEC1F392C5179F283),

    make_floatx80_init(0x3FFE, 0x9FD809FD809FD80A),

    make_floatx80_init(0x3FFD, 0xF12440D3E36130E6),

    make_floatx80_init(0x3FFE, 0x9E4CAD23DD5F3A20),

    make_floatx80_init(0x3FFD, 0xF61CCE92346600BB),

    make_floatx80_init(0x3FFE, 0x9CC8E160C3FB19B9),

    make_floatx80_init(0x3FFD, 0xFB091FD38145630A),

    make_floatx80_init(0x3FFE, 0x9B4C6F9EF03A3CAA),

    make_floatx80_init(0x3FFD, 0xFFE97042BFA4C2AD),

    make_floatx80_init(0x3FFE, 0x99D722DABDE58F06),

    make_floatx80_init(0x3FFE, 0x825EFCED49369330),

    make_floatx80_init(0x3FFE, 0x9868C809868C8098),

    make_floatx80_init(0x3FFE, 0x84C37A7AB9A905C9),

    make_floatx80_init(0x3FFE, 0x97012E025C04B809),

    make_floatx80_init(0x3FFE, 0x87224C2E8E645FB7),

    make_floatx80_init(0x3FFE, 0x95A02568095A0257),

    make_floatx80_init(0x3FFE, 0x897B8CAC9F7DE298),

    make_floatx80_init(0x3FFE, 0x9445809445809446),

    make_floatx80_init(0x3FFE, 0x8BCF55DEC4CD05FE),

    make_floatx80_init(0x3FFE, 0x92F113840497889C),

    make_floatx80_init(0x3FFE, 0x8E1DC0FB89E125E5),

    make_floatx80_init(0x3FFE, 0x91A2B3C4D5E6F809),

    make_floatx80_init(0x3FFE, 0x9066E68C955B6C9B),

    make_floatx80_init(0x3FFE, 0x905A38633E06C43B),

    make_floatx80_init(0x3FFE, 0x92AADE74C7BE59E0),

    make_floatx80_init(0x3FFE, 0x8F1779D9FDC3A219),

    make_floatx80_init(0x3FFE, 0x94E9BFF615845643),

    make_floatx80_init(0x3FFE, 0x8DDA520237694809),

    make_floatx80_init(0x3FFE, 0x9723A1B720134203),

    make_floatx80_init(0x3FFE, 0x8CA29C046514E023),

    make_floatx80_init(0x3FFE, 0x995899C890EB8990),

    make_floatx80_init(0x3FFE, 0x8B70344A139BC75A),

    make_floatx80_init(0x3FFE, 0x9B88BDAA3A3DAE2F),

    make_floatx80_init(0x3FFE, 0x8A42F8705669DB46),

    make_floatx80_init(0x3FFE, 0x9DB4224FFFE1157C),

    make_floatx80_init(0x3FFE, 0x891AC73AE9819B50),

    make_floatx80_init(0x3FFE, 0x9FDADC268B7A12DA),

    make_floatx80_init(0x3FFE, 0x87F78087F78087F8),

    make_floatx80_init(0x3FFE, 0xA1FCFF17CE733BD4),

    make_floatx80_init(0x3FFE, 0x86D905447A34ACC6),

    make_floatx80_init(0x3FFE, 0xA41A9E8F5446FB9F),

    make_floatx80_init(0x3FFE, 0x85BF37612CEE3C9B),

    make_floatx80_init(0x3FFE, 0xA633CD7E6771CD8B),

    make_floatx80_init(0x3FFE, 0x84A9F9C8084A9F9D),

    make_floatx80_init(0x3FFE, 0xA8489E600B435A5E),

    make_floatx80_init(0x3FFE, 0x839930523FBE3368),

    make_floatx80_init(0x3FFE, 0xAA59233CCCA4BD49),

    make_floatx80_init(0x3FFE, 0x828CBFBEB9A020A3),

    make_floatx80_init(0x3FFE, 0xAC656DAE6BCC4985),

    make_floatx80_init(0x3FFE, 0x81848DA8FAF0D277),

    make_floatx80_init(0x3FFE, 0xAE6D8EE360BB2468),

    make_floatx80_init(0x3FFE, 0x8080808080808081),

    make_floatx80_init(0x3FFE, 0xB07197A23C46C654)

};

#endif