Subversion Repositories HelenOS-historic

Rev

Rev 804 | Rev 829 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
731 cejka 1
/*
2
 * Copyright (C) 2005 Josef Cejka
3
 * All rights reserved.
4
 *
5
 * Redistribution and use in source and binary forms, with or without
6
 * modification, are permitted provided that the following conditions
7
 * are met:
8
 *
9
 * - Redistributions of source code must retain the above copyright
10
 *   notice, this list of conditions and the following disclaimer.
11
 * - Redistributions in binary form must reproduce the above copyright
12
 *   notice, this list of conditions and the following disclaimer in the
13
 *   documentation and/or other materials provided with the distribution.
14
 * - The name of the author may not be used to endorse or promote products
15
 *   derived from this software without specific prior written permission.
16
 *
17
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27
 */
28
 
29
#include<sftypes.h>
30
#include<add.h>
828 cejka 31
#include<div.h>
731 cejka 32
#include<comparison.h>
828 cejka 33
#include<mul.h>
731 cejka 34
 
35
float32 divFloat32(float32 a, float32 b)
36
{
804 cejka 37
    float32 result;
38
    __s32 aexp, bexp, cexp;
39
    __u64 afrac, bfrac, cfrac;
731 cejka 40
 
804 cejka 41
    result.parts.sign = a.parts.sign ^ b.parts.sign;
42
 
43
    if (isFloat32NaN(a)) {
44
        if (isFloat32SigNaN(a)) {
45
            /*FIXME: SigNaN*/
46
        }
47
        /*NaN*/
48
        return a;
49
    }
50
 
51
    if (isFloat32NaN(b)) {
52
        if (isFloat32SigNaN(b)) {
53
            /*FIXME: SigNaN*/
54
        }
55
        /*NaN*/
56
        return b;
57
    }
58
 
59
    if (isFloat32Infinity(a)) {
60
        if (isFloat32Infinity(b)) {
61
            /*FIXME: inf / inf */
62
            result.binary = FLOAT32_NAN;
63
            return result;
64
        }
65
        /* inf / num */
66
        result.parts.exp = a.parts.exp;
67
        result.parts.fraction = a.parts.fraction;
68
        return result;
69
    }
70
 
71
    if (isFloat32Infinity(b)) {
72
        if (isFloat32Zero(a)) {
73
            /* FIXME 0 / inf */
74
            result.parts.exp = 0;
75
            result.parts.fraction = 0;
76
            return result;
77
        }
78
        /* FIXME: num / inf*/
79
        result.parts.exp = 0;
80
        result.parts.fraction = 0;
81
        return result;
82
    }
83
 
84
    if (isFloat32Zero(b)) {
85
        if (isFloat32Zero(a)) {
86
            /*FIXME: 0 / 0*/
87
            result.binary = FLOAT32_NAN;
88
            return result;
89
        }
90
        /* FIXME: division by zero */
91
        result.parts.exp = 0;
92
        result.parts.fraction = 0;
93
        return result;
94
    }
95
 
96
 
97
    afrac = a.parts.fraction;
98
    aexp = a.parts.exp;
99
    bfrac = b.parts.fraction;
100
    bexp = b.parts.exp;
101
 
102
    /* denormalized numbers */
103
    if (aexp == 0) {
104
        if (afrac == 0) {
105
        result.parts.exp = 0;
106
        result.parts.fraction = 0;
107
        return result;
108
        }
109
        /* normalize it*/
110
 
111
        afrac <<= 1;
112
            /* afrac is nonzero => it must stop */ 
113
        while (! (afrac & FLOAT32_HIDDEN_BIT_MASK) ) {
114
            afrac <<= 1;
115
            aexp--;
116
        }
117
    }
118
 
119
    if (bexp == 0) {
120
        bfrac <<= 1;
121
            /* bfrac is nonzero => it must stop */ 
122
        while (! (bfrac & FLOAT32_HIDDEN_BIT_MASK) ) {
123
            bfrac <<= 1;
124
            bexp--;
125
        }
126
    }
127
 
128
    afrac = (afrac | FLOAT32_HIDDEN_BIT_MASK ) << (32 - FLOAT32_FRACTION_SIZE - 1 );
129
    bfrac = (bfrac | FLOAT32_HIDDEN_BIT_MASK ) << (32 - FLOAT32_FRACTION_SIZE );
130
 
131
    if ( bfrac <= (afrac << 1) ) {
132
        afrac >>= 1;
133
        aexp++;
134
    }
135
 
136
    cexp = aexp - bexp + FLOAT32_BIAS - 2;
137
 
138
    cfrac = (afrac << 32) / bfrac;
139
    if ((  cfrac & 0x3F ) == 0) {
140
        cfrac |= ( bfrac * cfrac != afrac << 32 );
141
    }
142
 
143
    /* pack and round */
144
 
828 cejka 145
    /* find first nonzero digit and shift result and detect possibly underflow */
804 cejka 146
    while ((cexp > 0) && (cfrac) && (!(cfrac & (FLOAT32_HIDDEN_BIT_MASK << 7 )))) {
147
        cexp--;
148
        cfrac <<= 1;
149
            /* TODO: fix underflow */
150
    };
151
 
152
    cfrac += (0x1 << 6); /* FIXME: 7 is not sure*/
153
 
154
    if (cfrac & (FLOAT32_HIDDEN_BIT_MASK << 7)) {
155
        ++cexp;
156
        cfrac >>= 1;
157
        }  
158
 
159
    /* check overflow */
160
    if (cexp >= FLOAT32_MAX_EXPONENT ) {
161
        /* FIXME: overflow, return infinity */
162
        result.parts.exp = FLOAT32_MAX_EXPONENT;
163
        result.parts.fraction = 0;
164
        return result;
165
    }
166
 
167
    if (cexp < 0) {
168
        /* FIXME: underflow */
169
        result.parts.exp = 0;
170
        if ((cexp + FLOAT32_FRACTION_SIZE) < 0) {
171
            result.parts.fraction = 0;
172
            return result;
173
        }
174
        cfrac >>= 1;
175
        while (cexp < 0) {
176
            cexp ++;
177
            cfrac >>= 1;
178
        }
179
 
180
    } else {
181
        result.parts.exp = (__u32)cexp;
182
    }
183
 
184
    result.parts.fraction = ((cfrac >> 6) & (~FLOAT32_HIDDEN_BIT_MASK));
185
 
186
    return result; 
731 cejka 187
}
188
 
828 cejka 189
float64 divFloat64(float64 a, float64 b)
190
{
191
    float64 result;
192
    __s32 aexp, bexp, cexp;
193
    __u64 afrac, bfrac, cfrac;
194
    __u64 remlo, remhi;
195
 
196
    result.parts.sign = a.parts.sign ^ b.parts.sign;
197
 
198
    if (isFloat64NaN(a)) {
199
        if (isFloat64SigNaN(a)) {
200
            /*FIXME: SigNaN*/
201
        }
202
        /*NaN*/
203
        return a;
204
    }
205
 
206
    if (isFloat64NaN(b)) {
207
        if (isFloat64SigNaN(b)) {
208
            /*FIXME: SigNaN*/
209
        }
210
        /*NaN*/
211
        return b;
212
    }
213
 
214
    if (isFloat64Infinity(a)) {
215
        if (isFloat64Infinity(b)) {
216
            /*FIXME: inf / inf */
217
            result.binary = FLOAT64_NAN;
218
            return result;
219
        }
220
        /* inf / num */
221
        result.parts.exp = a.parts.exp;
222
        result.parts.fraction = a.parts.fraction;
223
        return result;
224
    }
225
 
226
    if (isFloat64Infinity(b)) {
227
        if (isFloat64Zero(a)) {
228
            /* FIXME 0 / inf */
229
            result.parts.exp = 0;
230
            result.parts.fraction = 0;
231
            return result;
232
        }
233
        /* FIXME: num / inf*/
234
        result.parts.exp = 0;
235
        result.parts.fraction = 0;
236
        return result;
237
    }
238
 
239
    if (isFloat64Zero(b)) {
240
        if (isFloat64Zero(a)) {
241
            /*FIXME: 0 / 0*/
242
            result.binary = FLOAT64_NAN;
243
            return result;
244
        }
245
        /* FIXME: division by zero */
246
        result.parts.exp = 0;
247
        result.parts.fraction = 0;
248
        return result;
249
    }
250
 
251
 
252
    afrac = a.parts.fraction;
253
    aexp = a.parts.exp;
254
    bfrac = b.parts.fraction;
255
    bexp = b.parts.exp;
256
 
257
    /* denormalized numbers */
258
    if (aexp == 0) {
259
        if (afrac == 0) {
260
        result.parts.exp = 0;
261
        result.parts.fraction = 0;
262
        return result;
263
        }
264
        /* normalize it*/
265
 
266
        afrac <<= 1;
267
            /* afrac is nonzero => it must stop */ 
268
        while (! (afrac & FLOAT64_HIDDEN_BIT_MASK) ) {
269
            afrac <<= 1;
270
            aexp--;
271
        }
272
    }
273
 
274
    if (bexp == 0) {
275
        bfrac <<= 1;
276
            /* bfrac is nonzero => it must stop */ 
277
        while (! (bfrac & FLOAT64_HIDDEN_BIT_MASK) ) {
278
            bfrac <<= 1;
279
            bexp--;
280
        }
281
    }
282
 
283
    afrac = (afrac | FLOAT64_HIDDEN_BIT_MASK ) << (64 - FLOAT64_FRACTION_SIZE - 2 );
284
    bfrac = (bfrac | FLOAT64_HIDDEN_BIT_MASK ) << (64 - FLOAT64_FRACTION_SIZE - 1);
285
 
286
    if ( bfrac <= (afrac << 1) ) {
287
        afrac >>= 1;
288
        aexp++;
289
    }
290
 
291
    cexp = aexp - bexp + FLOAT64_BIAS - 2;
292
 
293
    cfrac = divFloat64estim(afrac, bfrac);
294
 
295
    if ((  cfrac & 0x1FF ) <= 2) { /*FIXME:?? */
296
        mul64integers( bfrac, cfrac, &remlo, &remhi);
297
        /* (__u128)afrac << 64 - ( ((__u128)remhi<<64) + (__u128)remlo )*/ 
298
        remhi = afrac - remhi - ( remlo > 0);
299
        remlo = - remlo;
300
 
301
        while ((__s64) remhi < 0) {
302
            cfrac--;
303
            remlo += bfrac;
304
            remhi += ( remlo < bfrac );
305
        }
306
        cfrac |= ( remlo != 0 );
307
    }
308
 
309
    /* pack and round */
310
 
311
    /* find first nonzero digit and shift result and detect possibly underflow */
312
    while ((cexp > 0) && (cfrac) && (!(cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1 ) )))) {
313
        cexp--;
314
        cfrac <<= 1;
315
            /* TODO: fix underflow */
316
    };
317
 
318
 
319
    cfrac >>= 1;
320
    ++cexp;
321
    cfrac += (0x1 << (64 - FLOAT64_FRACTION_SIZE - 3));
322
 
323
    if (cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1 ))) {
324
        ++cexp;
325
        cfrac >>= 1;
326
        }  
327
 
328
    /* check overflow */
329
    if (cexp >= FLOAT64_MAX_EXPONENT ) {
330
        /* FIXME: overflow, return infinity */
331
        result.parts.exp = FLOAT64_MAX_EXPONENT;
332
        result.parts.fraction = 0;
333
        return result;
334
    }
335
 
336
    if (cexp < 0) {
337
        /* FIXME: underflow */
338
        result.parts.exp = 0;
339
        if ((cexp + FLOAT64_FRACTION_SIZE) < 0) {
340
            result.parts.fraction = 0;
341
            return result;
342
        }
343
        cfrac >>= 1;
344
        while (cexp < 0) {
345
            cexp ++;
346
            cfrac >>= 1;
347
        }
348
        return result;
349
 
350
    } else {
351
        cexp ++; /*normalized*/
352
        result.parts.exp = (__u32)cexp;
353
    }
354
 
355
    result.parts.fraction = ((cfrac >>(64 - FLOAT64_FRACTION_SIZE - 2 ) ) & (~FLOAT64_HIDDEN_BIT_MASK));
356
 
357
    return result; 
358
}
359
 
360
__u64 divFloat64estim(__u64 a, __u64 b)
361
{
362
    __u64 bhi;
363
    __u64 remhi, remlo;
364
    __u64 result;
365
 
366
    if ( b <= a ) {
367
        return 0xFFFFFFFFFFFFFFFFull;
368
    }
369
 
370
    bhi = b >> 32;
371
    result = ((bhi << 32) <= a) ?( 0xFFFFFFFFull << 32) : ( a / bhi) << 32;
372
    mul64integers(b, result, &remlo, &remhi);
373
 
374
    remhi = a - remhi - (remlo > 0);
375
    remlo = - remlo;
376
 
377
    b <<= 32;
378
    while ( (__s64) remhi < 0 ) {
379
            result -= 0x1ll << 32; 
380
            remlo += b;
381
            remhi += bhi + ( remlo < b );
382
        }
383
    remhi = (remhi << 32) | (remlo >> 32);
384
    if (( bhi << 32) <= remhi) {
385
        result |= 0xFFFFFFFF;
386
    } else {
387
        result |= remhi / bhi;
388
    }
389
 
390
 
391
    return result;
392
}
393