|
| 1 | + |
| 2 | +/*============================================================================ |
| 3 | +
|
| 4 | +This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic |
| 5 | +Package, Release 3d, by John R. Hauser. |
| 6 | +
|
| 7 | +Copyright 2011, 2012, 2013, 2014 The Regents of the University of California. |
| 8 | +All rights reserved. |
| 9 | +
|
| 10 | +Redistribution and use in source and binary forms, with or without |
| 11 | +modification, are permitted provided that the following conditions are met: |
| 12 | +
|
| 13 | + 1. Redistributions of source code must retain the above copyright notice, |
| 14 | + this list of conditions, and the following disclaimer. |
| 15 | +
|
| 16 | + 2. Redistributions in binary form must reproduce the above copyright notice, |
| 17 | + this list of conditions, and the following disclaimer in the documentation |
| 18 | + and/or other materials provided with the distribution. |
| 19 | +
|
| 20 | + 3. Neither the name of the University nor the names of its contributors may |
| 21 | + be used to endorse or promote products derived from this software without |
| 22 | + specific prior written permission. |
| 23 | +
|
| 24 | +THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS "AS IS", AND ANY |
| 25 | +EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED |
| 26 | +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, ARE |
| 27 | +DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY |
| 28 | +DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
| 29 | +(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
| 30 | +LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
| 31 | +ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 32 | +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS |
| 33 | +SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 34 | +
|
| 35 | +=============================================================================*/ |
| 36 | + |
| 37 | +#include <stdbool.h> |
| 38 | +#include <stdint.h> |
| 39 | +#include "platform.h" |
| 40 | +#include "internals.h" |
| 41 | +#include "specialize.h" |
| 42 | +#include "softfloat.h" |
| 43 | + |
| 44 | +float128_t f128_div( float128_t a, float128_t b ) |
| 45 | +{ |
| 46 | + union ui128_f128 uA; |
| 47 | + uint_fast64_t uiA64, uiA0; |
| 48 | + bool signA; |
| 49 | + int_fast32_t expA; |
| 50 | + struct uint128 sigA; |
| 51 | + union ui128_f128 uB; |
| 52 | + uint_fast64_t uiB64, uiB0; |
| 53 | + bool signB; |
| 54 | + int_fast32_t expB; |
| 55 | + struct uint128 sigB; |
| 56 | + bool signZ; |
| 57 | + struct exp32_sig128 normExpSig; |
| 58 | + int_fast32_t expZ; |
| 59 | + struct uint128 rem; |
| 60 | + uint_fast32_t recip32; |
| 61 | + int ix; |
| 62 | + uint_fast64_t q64; |
| 63 | + uint_fast32_t q; |
| 64 | + struct uint128 term; |
| 65 | + uint_fast32_t qs[3]; |
| 66 | + uint_fast64_t sigZExtra; |
| 67 | + struct uint128 sigZ, uiZ; |
| 68 | + union ui128_f128 uZ; |
| 69 | + |
| 70 | + /*------------------------------------------------------------------------ |
| 71 | + *------------------------------------------------------------------------*/ |
| 72 | + uA.f = a; |
| 73 | + uiA64 = uA.ui.v64; |
| 74 | + uiA0 = uA.ui.v0; |
| 75 | + signA = signF128UI64( uiA64 ); |
| 76 | + expA = expF128UI64( uiA64 ); |
| 77 | + sigA.v64 = fracF128UI64( uiA64 ); |
| 78 | + sigA.v0 = uiA0; |
| 79 | + uB.f = b; |
| 80 | + uiB64 = uB.ui.v64; |
| 81 | + uiB0 = uB.ui.v0; |
| 82 | + signB = signF128UI64( uiB64 ); |
| 83 | + expB = expF128UI64( uiB64 ); |
| 84 | + sigB.v64 = fracF128UI64( uiB64 ); |
| 85 | + sigB.v0 = uiB0; |
| 86 | + signZ = signA ^ signB; |
| 87 | + /*------------------------------------------------------------------------ |
| 88 | + *------------------------------------------------------------------------*/ |
| 89 | + if ( expA == 0x7FFF ) { |
| 90 | + if ( sigA.v64 | sigA.v0 ) goto propagateNaN; |
| 91 | + if ( expB == 0x7FFF ) { |
| 92 | + if ( sigB.v64 | sigB.v0 ) goto propagateNaN; |
| 93 | + goto invalid; |
| 94 | + } |
| 95 | + goto infinity; |
| 96 | + } |
| 97 | + if ( expB == 0x7FFF ) { |
| 98 | + if ( sigB.v64 | sigB.v0 ) goto propagateNaN; |
| 99 | + goto zero; |
| 100 | + } |
| 101 | + /*------------------------------------------------------------------------ |
| 102 | + *------------------------------------------------------------------------*/ |
| 103 | + if ( ! expB ) { |
| 104 | + if ( ! (sigB.v64 | sigB.v0) ) { |
| 105 | + if ( ! (expA | sigA.v64 | sigA.v0) ) goto invalid; |
| 106 | + softfloat_raiseFlags( softfloat_flag_infinite ); |
| 107 | + goto infinity; |
| 108 | + } |
| 109 | + normExpSig = softfloat_normSubnormalF128Sig( sigB.v64, sigB.v0 ); |
| 110 | + expB = normExpSig.exp; |
| 111 | + sigB = normExpSig.sig; |
| 112 | + } |
| 113 | + if ( ! expA ) { |
| 114 | + if ( ! (sigA.v64 | sigA.v0) ) goto zero; |
| 115 | + normExpSig = softfloat_normSubnormalF128Sig( sigA.v64, sigA.v0 ); |
| 116 | + expA = normExpSig.exp; |
| 117 | + sigA = normExpSig.sig; |
| 118 | + } |
| 119 | + /*------------------------------------------------------------------------ |
| 120 | + *------------------------------------------------------------------------*/ |
| 121 | + expZ = expA - expB + 0x3FFE; |
| 122 | + sigA.v64 |= UINT64_C( 0x0001000000000000 ); |
| 123 | + sigB.v64 |= UINT64_C( 0x0001000000000000 ); |
| 124 | + rem = sigA; |
| 125 | + if ( softfloat_lt128( sigA.v64, sigA.v0, sigB.v64, sigB.v0 ) ) { |
| 126 | + --expZ; |
| 127 | + rem = softfloat_add128( sigA.v64, sigA.v0, sigA.v64, sigA.v0 ); |
| 128 | + } |
| 129 | + recip32 = softfloat_approxRecip32_1( sigB.v64>>17 ); |
| 130 | + ix = 3; |
| 131 | + for (;;) { |
| 132 | + q64 = (uint_fast64_t) (uint32_t) (rem.v64>>19) * recip32; |
| 133 | + q = (q64 + 0x80000000)>>32; |
| 134 | + --ix; |
| 135 | + if ( ix < 0 ) break; |
| 136 | + rem = softfloat_shortShiftLeft128( rem.v64, rem.v0, 29 ); |
| 137 | + term = softfloat_mul128By32( sigB.v64, sigB.v0, q ); |
| 138 | + rem = softfloat_sub128( rem.v64, rem.v0, term.v64, term.v0 ); |
| 139 | + if ( rem.v64 & UINT64_C( 0x8000000000000000 ) ) { |
| 140 | + --q; |
| 141 | + rem = softfloat_add128( rem.v64, rem.v0, sigB.v64, sigB.v0 ); |
| 142 | + } |
| 143 | + qs[ix] = q; |
| 144 | + } |
| 145 | + /*------------------------------------------------------------------------ |
| 146 | + *------------------------------------------------------------------------*/ |
| 147 | + if ( ((q + 1) & 7) < 2 ) { |
| 148 | + rem = softfloat_shortShiftLeft128( rem.v64, rem.v0, 29 ); |
| 149 | + term = softfloat_mul128By32( sigB.v64, sigB.v0, q ); |
| 150 | + rem = softfloat_sub128( rem.v64, rem.v0, term.v64, term.v0 ); |
| 151 | + if ( rem.v64 & UINT64_C( 0x8000000000000000 ) ) { |
| 152 | + --q; |
| 153 | + rem = softfloat_add128( rem.v64, rem.v0, sigB.v64, sigB.v0 ); |
| 154 | + } else if ( softfloat_le128( sigB.v64, sigB.v0, rem.v64, rem.v0 ) ) { |
| 155 | + ++q; |
| 156 | + rem = softfloat_sub128( rem.v64, rem.v0, sigB.v64, sigB.v0 ); |
| 157 | + } |
| 158 | + if ( rem.v64 | rem.v0 ) q |= 1; |
| 159 | + } |
| 160 | + /*------------------------------------------------------------------------ |
| 161 | + *------------------------------------------------------------------------*/ |
| 162 | + sigZExtra = (uint64_t) ((uint_fast64_t) q<<60); |
| 163 | + term = softfloat_shortShiftLeft128( 0, qs[1], 54 ); |
| 164 | + sigZ = |
| 165 | + softfloat_add128( |
| 166 | + (uint_fast64_t) qs[2]<<19, ((uint_fast64_t) qs[0]<<25) + (q>>4), |
| 167 | + term.v64, term.v0 |
| 168 | + ); |
| 169 | + return |
| 170 | + softfloat_roundPackToF128( signZ, expZ, sigZ.v64, sigZ.v0, sigZExtra ); |
| 171 | + /*------------------------------------------------------------------------ |
| 172 | + *------------------------------------------------------------------------*/ |
| 173 | + propagateNaN: |
| 174 | + uiZ = softfloat_propagateNaNF128UI( uiA64, uiA0, uiB64, uiB0 ); |
| 175 | + goto uiZ; |
| 176 | + /*------------------------------------------------------------------------ |
| 177 | + *------------------------------------------------------------------------*/ |
| 178 | + invalid: |
| 179 | + softfloat_raiseFlags( softfloat_flag_invalid ); |
| 180 | + uiZ.v64 = defaultNaNF128UI64; |
| 181 | + uiZ.v0 = defaultNaNF128UI0; |
| 182 | + goto uiZ; |
| 183 | + /*------------------------------------------------------------------------ |
| 184 | + *------------------------------------------------------------------------*/ |
| 185 | + infinity: |
| 186 | + uiZ.v64 = packToF128UI64( signZ, 0x7FFF, 0 ); |
| 187 | + goto uiZ0; |
| 188 | + /*------------------------------------------------------------------------ |
| 189 | + *------------------------------------------------------------------------*/ |
| 190 | + zero: |
| 191 | + uiZ.v64 = packToF128UI64( signZ, 0, 0 ); |
| 192 | + uiZ0: |
| 193 | + uiZ.v0 = 0; |
| 194 | + uiZ: |
| 195 | + uZ.ui = uiZ; |
| 196 | + return uZ.f; |
| 197 | + |
| 198 | +} |
| 199 | + |
0 commit comments