d03ab969 |
1 | /* -*-c-*- |
2 | * |
1a05a8ef |
3 | * $Id: mpx.c,v 1.10 2000/10/08 12:06:12 mdw Exp $ |
d03ab969 |
4 | * |
5 | * Low-level multiprecision arithmetic |
6 | * |
7 | * (c) 1999 Straylight/Edgeware |
8 | */ |
9 | |
10 | /*----- Licensing notice --------------------------------------------------* |
11 | * |
12 | * This file is part of Catacomb. |
13 | * |
14 | * Catacomb is free software; you can redistribute it and/or modify |
15 | * it under the terms of the GNU Library General Public License as |
16 | * published by the Free Software Foundation; either version 2 of the |
17 | * License, or (at your option) any later version. |
18 | * |
19 | * Catacomb is distributed in the hope that it will be useful, |
20 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
21 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
22 | * GNU Library General Public License for more details. |
23 | * |
24 | * You should have received a copy of the GNU Library General Public |
25 | * License along with Catacomb; if not, write to the Free |
26 | * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, |
27 | * MA 02111-1307, USA. |
28 | */ |
29 | |
30 | /*----- Revision history --------------------------------------------------* |
31 | * |
32 | * $Log: mpx.c,v $ |
1a05a8ef |
33 | * Revision 1.10 2000/10/08 12:06:12 mdw |
34 | * Provide @mpx_ueq@ for rapidly testing equality of two integers. |
35 | * |
99b30c23 |
36 | * Revision 1.9 2000/06/26 07:52:50 mdw |
37 | * Portability fix for the bug fix. |
38 | * |
ce76ff16 |
39 | * Revision 1.8 2000/06/25 12:59:02 mdw |
40 | * (mpx_udiv): Fix bug in quotient digit estimation. |
41 | * |
698bd937 |
42 | * Revision 1.7 1999/12/22 15:49:07 mdw |
43 | * New function for division by a small integer. |
44 | * |
42684bdb |
45 | * Revision 1.6 1999/11/20 22:43:44 mdw |
46 | * Integrate testing for MPX routines. |
47 | * |
dd517851 |
48 | * Revision 1.5 1999/11/20 22:23:27 mdw |
49 | * Add function versions of some low-level macros with wider use. |
50 | * |
f45a00c6 |
51 | * Revision 1.4 1999/11/17 18:04:09 mdw |
52 | * Add two's-complement functionality. Improve mpx_udiv a little by |
53 | * performing the multiplication of the divisor by q with the subtraction |
54 | * from r. |
55 | * |
501da53c |
56 | * Revision 1.3 1999/11/13 01:57:31 mdw |
57 | * Remove stray debugging code. |
58 | * |
c8a2f9ef |
59 | * Revision 1.2 1999/11/13 01:50:59 mdw |
60 | * Multiprecision routines finished and tested. |
61 | * |
d03ab969 |
62 | * Revision 1.1 1999/09/03 08:41:12 mdw |
63 | * Initial import. |
64 | * |
65 | */ |
66 | |
67 | /*----- Header files ------------------------------------------------------*/ |
68 | |
c8a2f9ef |
69 | #include <assert.h> |
d03ab969 |
70 | #include <stdio.h> |
71 | #include <stdlib.h> |
72 | #include <string.h> |
73 | |
74 | #include <mLib/bits.h> |
75 | |
76 | #include "mptypes.h" |
77 | #include "mpx.h" |
78 | |
79 | /*----- Loading and storing -----------------------------------------------*/ |
80 | |
81 | /* --- @mpx_storel@ --- * |
82 | * |
83 | * Arguments: @const mpw *v, *vl@ = base and limit of source vector |
c8a2f9ef |
84 | * @void *pp@ = pointer to octet array |
d03ab969 |
85 | * @size_t sz@ = size of octet array |
86 | * |
87 | * Returns: --- |
88 | * |
89 | * Use: Stores an MP in an octet array, least significant octet |
90 | * first. High-end octets are silently discarded if there |
91 | * isn't enough space for them. |
92 | */ |
93 | |
c8a2f9ef |
94 | void mpx_storel(const mpw *v, const mpw *vl, void *pp, size_t sz) |
d03ab969 |
95 | { |
96 | mpw n, w = 0; |
c8a2f9ef |
97 | octet *p = pp, *q = p + sz; |
d03ab969 |
98 | unsigned bits = 0; |
99 | |
100 | while (p < q) { |
101 | if (bits < 8) { |
102 | if (v >= vl) { |
103 | *p++ = U8(w); |
104 | break; |
105 | } |
106 | n = *v++; |
107 | *p++ = U8(w | n << bits); |
108 | w = n >> (8 - bits); |
109 | bits += MPW_BITS - 8; |
110 | } else { |
111 | *p++ = U8(w); |
112 | w >>= 8; |
113 | bits -= 8; |
114 | } |
115 | } |
116 | memset(p, 0, q - p); |
117 | } |
118 | |
119 | /* --- @mpx_loadl@ --- * |
120 | * |
121 | * Arguments: @mpw *v, *vl@ = base and limit of destination vector |
c8a2f9ef |
122 | * @const void *pp@ = pointer to octet array |
d03ab969 |
123 | * @size_t sz@ = size of octet array |
124 | * |
125 | * Returns: --- |
126 | * |
127 | * Use: Loads an MP in an octet array, least significant octet |
128 | * first. High-end octets are ignored if there isn't enough |
129 | * space for them. |
130 | */ |
131 | |
c8a2f9ef |
132 | void mpx_loadl(mpw *v, mpw *vl, const void *pp, size_t sz) |
d03ab969 |
133 | { |
134 | unsigned n; |
c8a2f9ef |
135 | mpw w = 0; |
136 | const octet *p = pp, *q = p + sz; |
d03ab969 |
137 | unsigned bits = 0; |
138 | |
139 | if (v >= vl) |
140 | return; |
141 | while (p < q) { |
142 | n = U8(*p++); |
143 | w |= n << bits; |
144 | bits += 8; |
145 | if (bits >= MPW_BITS) { |
146 | *v++ = MPW(w); |
147 | w = n >> (MPW_BITS - bits + 8); |
148 | bits -= MPW_BITS; |
149 | if (v >= vl) |
150 | return; |
151 | } |
152 | } |
153 | *v++ = w; |
154 | MPX_ZERO(v, vl); |
155 | } |
156 | |
157 | /* --- @mpx_storeb@ --- * |
158 | * |
159 | * Arguments: @const mpw *v, *vl@ = base and limit of source vector |
c8a2f9ef |
160 | * @void *pp@ = pointer to octet array |
d03ab969 |
161 | * @size_t sz@ = size of octet array |
162 | * |
163 | * Returns: --- |
164 | * |
165 | * Use: Stores an MP in an octet array, most significant octet |
166 | * first. High-end octets are silently discarded if there |
167 | * isn't enough space for them. |
168 | */ |
169 | |
c8a2f9ef |
170 | void mpx_storeb(const mpw *v, const mpw *vl, void *pp, size_t sz) |
d03ab969 |
171 | { |
172 | mpw n, w = 0; |
c8a2f9ef |
173 | octet *p = pp, *q = p + sz; |
d03ab969 |
174 | unsigned bits = 0; |
175 | |
176 | while (q > p) { |
177 | if (bits < 8) { |
178 | if (v >= vl) { |
179 | *--q = U8(w); |
180 | break; |
181 | } |
182 | n = *v++; |
183 | *--q = U8(w | n << bits); |
184 | w = n >> (8 - bits); |
185 | bits += MPW_BITS - 8; |
186 | } else { |
187 | *--q = U8(w); |
188 | w >>= 8; |
189 | bits -= 8; |
190 | } |
191 | } |
192 | memset(p, 0, q - p); |
193 | } |
194 | |
195 | /* --- @mpx_loadb@ --- * |
196 | * |
197 | * Arguments: @mpw *v, *vl@ = base and limit of destination vector |
c8a2f9ef |
198 | * @const void *pp@ = pointer to octet array |
d03ab969 |
199 | * @size_t sz@ = size of octet array |
200 | * |
201 | * Returns: --- |
202 | * |
203 | * Use: Loads an MP in an octet array, most significant octet |
204 | * first. High-end octets are ignored if there isn't enough |
205 | * space for them. |
206 | */ |
207 | |
c8a2f9ef |
208 | void mpx_loadb(mpw *v, mpw *vl, const void *pp, size_t sz) |
d03ab969 |
209 | { |
210 | unsigned n; |
c8a2f9ef |
211 | mpw w = 0; |
212 | const octet *p = pp, *q = p + sz; |
d03ab969 |
213 | unsigned bits = 0; |
214 | |
215 | if (v >= vl) |
216 | return; |
217 | while (q > p) { |
218 | n = U8(*--q); |
219 | w |= n << bits; |
220 | bits += 8; |
221 | if (bits >= MPW_BITS) { |
222 | *v++ = MPW(w); |
223 | w = n >> (MPW_BITS - bits + 8); |
224 | bits -= MPW_BITS; |
225 | if (v >= vl) |
226 | return; |
227 | } |
228 | } |
229 | *v++ = w; |
230 | MPX_ZERO(v, vl); |
231 | } |
232 | |
233 | /*----- Logical shifting --------------------------------------------------*/ |
234 | |
235 | /* --- @mpx_lsl@ --- * |
236 | * |
237 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
238 | * @const mpw *av, *avl@ = source vector base and limit |
239 | * @size_t n@ = number of bit positions to shift by |
240 | * |
241 | * Returns: --- |
242 | * |
243 | * Use: Performs a logical shift left operation on an integer. |
244 | */ |
245 | |
246 | void mpx_lsl(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, size_t n) |
247 | { |
248 | size_t nw; |
249 | unsigned nb; |
250 | |
251 | /* --- Trivial special case --- */ |
252 | |
253 | if (n == 0) |
254 | MPX_COPY(dv, dvl, av, avl); |
255 | |
256 | /* --- Single bit shifting --- */ |
257 | |
258 | else if (n == 1) { |
259 | mpw w = 0; |
260 | while (av < avl) { |
261 | mpw t; |
262 | if (dv >= dvl) |
263 | goto done; |
264 | t = *av++; |
265 | *dv++ = MPW((t << 1) | w); |
266 | w = t >> (MPW_BITS - 1); |
267 | } |
268 | if (dv >= dvl) |
269 | goto done; |
270 | *dv++ = MPW(w); |
271 | MPX_ZERO(dv, dvl); |
c8a2f9ef |
272 | goto done; |
d03ab969 |
273 | } |
274 | |
275 | /* --- Break out word and bit shifts for more sophisticated work --- */ |
276 | |
277 | nw = n / MPW_BITS; |
278 | nb = n % MPW_BITS; |
279 | |
280 | /* --- Handle a shift by a multiple of the word size --- */ |
281 | |
282 | if (nb == 0) { |
283 | MPX_COPY(dv + nw, dvl, av, avl); |
284 | memset(dv, 0, MPWS(nw)); |
285 | } |
286 | |
c8a2f9ef |
287 | /* --- And finally the difficult case --- * |
288 | * |
289 | * This is a little convoluted, because I have to start from the end and |
290 | * work backwards to avoid overwriting the source, if they're both the same |
291 | * block of memory. |
292 | */ |
d03ab969 |
293 | |
294 | else { |
295 | mpw w; |
296 | size_t nr = MPW_BITS - nb; |
c8a2f9ef |
297 | size_t dvn = dvl - dv; |
298 | size_t avn = avl - av; |
d03ab969 |
299 | |
c8a2f9ef |
300 | if (dvn <= nw) { |
d03ab969 |
301 | MPX_ZERO(dv, dvl); |
302 | goto done; |
303 | } |
d03ab969 |
304 | |
c8a2f9ef |
305 | if (dvn > avn + nw) { |
306 | size_t off = avn + nw + 1; |
307 | MPX_ZERO(dv + off, dvl); |
308 | dvl = dv + off; |
309 | w = 0; |
310 | } else { |
311 | avl = av + dvn - nw; |
312 | w = *--avl << nb; |
d03ab969 |
313 | } |
314 | |
c8a2f9ef |
315 | while (avl > av) { |
316 | mpw t = *--avl; |
317 | *--dvl = (t >> nr) | w; |
318 | w = t << nb; |
d03ab969 |
319 | } |
c8a2f9ef |
320 | |
321 | *--dvl = w; |
322 | MPX_ZERO(dv, dvl); |
d03ab969 |
323 | } |
324 | |
325 | done:; |
326 | } |
327 | |
328 | /* --- @mpx_lsr@ --- * |
329 | * |
330 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
331 | * @const mpw *av, *avl@ = source vector base and limit |
332 | * @size_t n@ = number of bit positions to shift by |
333 | * |
334 | * Returns: --- |
335 | * |
336 | * Use: Performs a logical shift right operation on an integer. |
337 | */ |
338 | |
339 | void mpx_lsr(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, size_t n) |
340 | { |
341 | size_t nw; |
342 | unsigned nb; |
343 | |
344 | /* --- Trivial special case --- */ |
345 | |
346 | if (n == 0) |
347 | MPX_COPY(dv, dvl, av, avl); |
348 | |
349 | /* --- Single bit shifting --- */ |
350 | |
351 | else if (n == 1) { |
352 | mpw w = *av++ >> 1; |
353 | while (av < avl) { |
354 | mpw t; |
355 | if (dv >= dvl) |
356 | goto done; |
357 | t = *av++; |
358 | *dv++ = MPW((t << (MPW_BITS - 1)) | w); |
359 | w = t >> 1; |
360 | } |
361 | if (dv >= dvl) |
362 | goto done; |
363 | *dv++ = MPW(w); |
364 | MPX_ZERO(dv, dvl); |
c8a2f9ef |
365 | goto done; |
d03ab969 |
366 | } |
367 | |
368 | /* --- Break out word and bit shifts for more sophisticated work --- */ |
369 | |
370 | nw = n / MPW_BITS; |
371 | nb = n % MPW_BITS; |
372 | |
373 | /* --- Handle a shift by a multiple of the word size --- */ |
374 | |
375 | if (nb == 0) |
376 | MPX_COPY(dv, dvl, av + nw, avl); |
377 | |
378 | /* --- And finally the difficult case --- */ |
379 | |
380 | else { |
381 | mpw w; |
382 | size_t nr = MPW_BITS - nb; |
383 | |
384 | av += nw; |
385 | w = *av++; |
386 | while (av < avl) { |
387 | mpw t; |
388 | if (dv >= dvl) |
389 | goto done; |
390 | t = *av++; |
391 | *dv++ = MPW((w >> nb) | (t << nr)); |
392 | w = t; |
393 | } |
394 | if (dv < dvl) { |
395 | *dv++ = MPW(w >> nb); |
396 | MPX_ZERO(dv, dvl); |
397 | } |
398 | } |
399 | |
400 | done:; |
401 | } |
402 | |
403 | /*----- Unsigned arithmetic -----------------------------------------------*/ |
404 | |
f45a00c6 |
405 | /* --- @mpx_2c@ --- * |
406 | * |
407 | * Arguments: @mpw *dv, *dvl@ = destination vector |
408 | * @const mpw *v, *vl@ = source vector |
409 | * |
410 | * Returns: --- |
411 | * |
412 | * Use: Calculates the two's complement of @v@. |
413 | */ |
414 | |
415 | void mpx_2c(mpw *dv, mpw *dvl, const mpw *v, const mpw *vl) |
416 | { |
417 | mpw c = 0; |
418 | while (dv < dvl && v < vl) |
419 | *dv++ = c = MPW(~*v++); |
420 | if (dv < dvl) { |
421 | if (c > MPW_MAX / 2) |
422 | c = MPW(~0); |
423 | while (dv < dvl) |
424 | *dv++ = c; |
425 | } |
426 | MPX_UADDN(dv, dvl, 1); |
427 | } |
428 | |
1a05a8ef |
429 | /* --- @mpx_ueq@ --- * |
430 | * |
431 | * Arguments: @const mpw *av, *avl@ = first argument vector base and limit |
432 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
433 | * |
434 | * Returns: Nonzero if the two vectors are equal. |
435 | * |
436 | * Use: Performs an unsigned integer test for equality. |
437 | */ |
438 | |
439 | int mpx_ueq(const mpw *av, const mpw *avl, const mpw *bv, const mpw *bvl) |
440 | { |
441 | MPX_SHRINK(av, avl); |
442 | MPX_SHRINK(bv, bvl); |
443 | if (avl - av != bvl - bv) |
444 | return (0); |
445 | while (av < avl) { |
446 | if (*av++ != *bv++) |
447 | return (0); |
448 | } |
449 | return (1); |
450 | } |
451 | |
d03ab969 |
452 | /* --- @mpx_ucmp@ --- * |
453 | * |
454 | * Arguments: @const mpw *av, *avl@ = first argument vector base and limit |
455 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
456 | * |
457 | * Returns: Less than, equal to, or greater than zero depending on |
458 | * whether @a@ is less than, equal to or greater than @b@, |
459 | * respectively. |
460 | * |
461 | * Use: Performs an unsigned integer comparison. |
462 | */ |
463 | |
464 | int mpx_ucmp(const mpw *av, const mpw *avl, const mpw *bv, const mpw *bvl) |
465 | { |
466 | MPX_SHRINK(av, avl); |
467 | MPX_SHRINK(bv, bvl); |
468 | |
469 | if (avl - av > bvl - bv) |
470 | return (+1); |
471 | else if (avl - av < bvl - bv) |
472 | return (-1); |
473 | else while (avl > av) { |
474 | mpw a = *--avl, b = *--bvl; |
475 | if (a > b) |
476 | return (+1); |
477 | else if (a < b) |
478 | return (-1); |
479 | } |
480 | return (0); |
481 | } |
1a05a8ef |
482 | |
d03ab969 |
483 | /* --- @mpx_uadd@ --- * |
484 | * |
485 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
486 | * @const mpw *av, *avl@ = first addend vector base and limit |
487 | * @const mpw *bv, *bvl@ = second addend vector base and limit |
488 | * |
489 | * Returns: --- |
490 | * |
491 | * Use: Performs unsigned integer addition. If the result overflows |
492 | * the destination vector, high-order bits are discarded. This |
493 | * means that two's complement addition happens more or less for |
494 | * free, although that's more a side-effect than anything else. |
495 | * The result vector may be equal to either or both source |
496 | * vectors, but may not otherwise overlap them. |
497 | */ |
498 | |
499 | void mpx_uadd(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
500 | const mpw *bv, const mpw *bvl) |
501 | { |
502 | mpw c = 0; |
503 | |
504 | while (av < avl || bv < bvl) { |
505 | mpw a, b; |
506 | mpd x; |
507 | if (dv >= dvl) |
508 | return; |
509 | a = (av < avl) ? *av++ : 0; |
510 | b = (bv < bvl) ? *bv++ : 0; |
511 | x = (mpd)a + (mpd)b + c; |
512 | *dv++ = MPW(x); |
513 | c = x >> MPW_BITS; |
514 | } |
515 | if (dv < dvl) { |
516 | *dv++ = c; |
517 | MPX_ZERO(dv, dvl); |
518 | } |
519 | } |
520 | |
dd517851 |
521 | /* --- @mpx_uaddn@ --- * |
522 | * |
523 | * Arguments: @mpw *dv, *dvl@ = source and destination base and limit |
524 | * @mpw n@ = other addend |
525 | * |
526 | * Returns: --- |
527 | * |
528 | * Use: Adds a small integer to a multiprecision number. |
529 | */ |
530 | |
531 | void mpx_uaddn(mpw *dv, mpw *dvl, mpw n) { MPX_UADDN(dv, dvl, n); } |
532 | |
d03ab969 |
533 | /* --- @mpx_usub@ --- * |
534 | * |
535 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
536 | * @const mpw *av, *avl@ = first argument vector base and limit |
537 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
538 | * |
539 | * Returns: --- |
540 | * |
541 | * Use: Performs unsigned integer subtraction. If the result |
542 | * overflows the destination vector, high-order bits are |
543 | * discarded. This means that two's complement subtraction |
544 | * happens more or less for free, althuogh that's more a side- |
545 | * effect than anything else. The result vector may be equal to |
546 | * either or both source vectors, but may not otherwise overlap |
547 | * them. |
548 | */ |
549 | |
550 | void mpx_usub(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
551 | const mpw *bv, const mpw *bvl) |
552 | { |
553 | mpw c = 0; |
554 | |
555 | while (av < avl || bv < bvl) { |
556 | mpw a, b; |
557 | mpd x; |
558 | if (dv >= dvl) |
559 | return; |
560 | a = (av < avl) ? *av++ : 0; |
561 | b = (bv < bvl) ? *bv++ : 0; |
c8a2f9ef |
562 | x = (mpd)a - (mpd)b - c; |
d03ab969 |
563 | *dv++ = MPW(x); |
c8a2f9ef |
564 | if (x >> MPW_BITS) |
565 | c = 1; |
566 | else |
567 | c = 0; |
d03ab969 |
568 | } |
c8a2f9ef |
569 | if (c) |
570 | c = MPW_MAX; |
d03ab969 |
571 | while (dv < dvl) |
c8a2f9ef |
572 | *dv++ = c; |
d03ab969 |
573 | } |
574 | |
dd517851 |
575 | /* --- @mpx_usubn@ --- * |
576 | * |
577 | * Arguments: @mpw *dv, *dvl@ = source and destination base and limit |
578 | * @n@ = subtrahend |
579 | * |
580 | * Returns: --- |
581 | * |
582 | * Use: Subtracts a small integer from a multiprecision number. |
583 | */ |
584 | |
585 | void mpx_usubn(mpw *dv, mpw *dvl, mpw n) { MPX_USUBN(dv, dvl, n); } |
586 | |
d03ab969 |
587 | /* --- @mpx_umul@ --- * |
588 | * |
589 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
590 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
591 | * @const mpw *bv, *bvl@ = multiplier vector base and limit |
592 | * |
593 | * Returns: --- |
594 | * |
595 | * Use: Performs unsigned integer multiplication. If the result |
596 | * overflows the desination vector, high-order bits are |
597 | * discarded. The result vector may not overlap the argument |
598 | * vectors in any way. |
599 | */ |
600 | |
601 | void mpx_umul(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
602 | const mpw *bv, const mpw *bvl) |
603 | { |
604 | /* --- This is probably worthwhile on a multiply --- */ |
605 | |
606 | MPX_SHRINK(av, avl); |
607 | MPX_SHRINK(bv, bvl); |
608 | |
609 | /* --- Deal with a multiply by zero --- */ |
610 | |
611 | if (bv == bvl) { |
c8a2f9ef |
612 | MPX_ZERO(dv, dvl); |
d03ab969 |
613 | return; |
614 | } |
615 | |
616 | /* --- Do the initial multiply and initialize the accumulator --- */ |
617 | |
618 | MPX_UMULN(dv, dvl, av, avl, *bv++); |
619 | |
620 | /* --- Do the remaining multiply/accumulates --- */ |
621 | |
c8a2f9ef |
622 | while (dv < dvl && bv < bvl) { |
d03ab969 |
623 | mpw m = *bv++; |
c8a2f9ef |
624 | mpw c = 0; |
d03ab969 |
625 | const mpw *avv = av; |
626 | mpw *dvv = ++dv; |
627 | |
628 | while (avv < avl) { |
629 | mpd x; |
630 | if (dvv >= dvl) |
631 | goto next; |
c8a2f9ef |
632 | x = (mpd)*dvv + (mpd)m * (mpd)*avv++ + c; |
633 | *dvv++ = MPW(x); |
d03ab969 |
634 | c = x >> MPW_BITS; |
635 | } |
c8a2f9ef |
636 | MPX_UADDN(dvv, dvl, c); |
d03ab969 |
637 | next:; |
638 | } |
639 | } |
640 | |
dd517851 |
641 | /* --- @mpx_umuln@ --- * |
642 | * |
643 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
644 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
645 | * @mpw m@ = multiplier |
646 | * |
647 | * Returns: --- |
648 | * |
649 | * Use: Multiplies a multiprecision integer by a single-word value. |
650 | * The destination and source may be equal. The destination |
651 | * is completely cleared after use. |
652 | */ |
653 | |
654 | void mpx_umuln(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, mpw m) |
655 | { |
656 | MPX_UMULN(dv, dvl, av, avl, m); |
657 | } |
658 | |
659 | /* --- @mpx_umlan@ --- * |
660 | * |
661 | * Arguments: @mpw *dv, *dvl@ = destination/accumulator base and limit |
662 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
663 | * @mpw m@ = multiplier |
664 | * |
665 | * Returns: --- |
666 | * |
667 | * Use: Multiplies a multiprecision integer by a single-word value |
668 | * and adds the result to an accumulator. |
669 | */ |
670 | |
671 | void mpx_umlan(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, mpw m) |
672 | { |
673 | MPX_UMLAN(dv, dvl, av, avl, m); |
674 | } |
675 | |
c8a2f9ef |
676 | /* --- @mpx_usqr@ --- * |
677 | * |
678 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
679 | * @const mpw *av, *av@ = source vector base and limit |
680 | * |
681 | * Returns: --- |
682 | * |
683 | * Use: Performs unsigned integer squaring. The result vector must |
684 | * not overlap the source vector in any way. |
685 | */ |
686 | |
687 | void mpx_usqr(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl) |
688 | { |
689 | MPX_ZERO(dv, dvl); |
690 | |
691 | /* --- Main loop --- */ |
692 | |
693 | while (av < avl) { |
694 | const mpw *avv = av; |
695 | mpw *dvv = dv; |
696 | mpw a = *av; |
697 | mpd c; |
698 | |
699 | /* --- Stop if I've run out of destination --- */ |
700 | |
701 | if (dvv >= dvl) |
702 | break; |
703 | |
704 | /* --- Work out the square at this point in the proceedings --- */ |
705 | |
706 | { |
c8a2f9ef |
707 | mpd x = (mpd)a * (mpd)a + *dvv; |
708 | *dvv++ = MPW(x); |
709 | c = MPW(x >> MPW_BITS); |
710 | } |
711 | |
712 | /* --- Now fix up the rest of the vector upwards --- */ |
713 | |
714 | avv++; |
715 | while (dvv < dvl && avv < avl) { |
c8a2f9ef |
716 | mpd x = (mpd)a * (mpd)*avv++; |
717 | mpd y = ((x << 1) & MPW_MAX) + c + *dvv; |
718 | c = (x >> (MPW_BITS - 1)) + (y >> MPW_BITS); |
719 | *dvv++ = MPW(y); |
720 | } |
721 | while (dvv < dvl && c) { |
722 | mpd x = c + *dvv; |
723 | *dvv++ = MPW(x); |
724 | c = x >> MPW_BITS; |
725 | } |
726 | |
727 | /* --- Get ready for the next round --- */ |
728 | |
729 | av++; |
730 | dv += 2; |
731 | } |
732 | } |
733 | |
d03ab969 |
734 | /* --- @mpx_udiv@ --- * |
735 | * |
736 | * Arguments: @mpw *qv, *qvl@ = quotient vector base and limit |
737 | * @mpw *rv, *rvl@ = dividend/remainder vector base and limit |
738 | * @const mpw *dv, *dvl@ = divisor vector base and limit |
c8a2f9ef |
739 | * @mpw *sv, *svl@ = scratch workspace |
d03ab969 |
740 | * |
741 | * Returns: --- |
742 | * |
743 | * Use: Performs unsigned integer division. If the result overflows |
744 | * the quotient vector, high-order bits are discarded. (Clearly |
745 | * the remainder vector can't overflow.) The various vectors |
746 | * may not overlap in any way. Yes, I know it's a bit odd |
747 | * requiring the dividend to be in the result position but it |
748 | * does make some sense really. The remainder must have |
c8a2f9ef |
749 | * headroom for at least two extra words. The scratch space |
f45a00c6 |
750 | * must be at least one word larger than the divisor. |
d03ab969 |
751 | */ |
752 | |
753 | void mpx_udiv(mpw *qv, mpw *qvl, mpw *rv, mpw *rvl, |
c8a2f9ef |
754 | const mpw *dv, const mpw *dvl, |
755 | mpw *sv, mpw *svl) |
d03ab969 |
756 | { |
d03ab969 |
757 | unsigned norm = 0; |
758 | size_t scale; |
759 | mpw d, dd; |
760 | |
761 | /* --- Initialize the quotient --- */ |
762 | |
763 | MPX_ZERO(qv, qvl); |
764 | |
c8a2f9ef |
765 | /* --- Perform some sanity checks --- */ |
766 | |
767 | MPX_SHRINK(dv, dvl); |
768 | assert(((void)"division by zero in mpx_udiv", dv < dvl)); |
769 | |
d03ab969 |
770 | /* --- Normalize the divisor --- * |
771 | * |
772 | * The algorithm requires that the divisor be at least two digits long. |
773 | * This is easy to fix. |
774 | */ |
775 | |
c8a2f9ef |
776 | { |
777 | unsigned b; |
d03ab969 |
778 | |
c8a2f9ef |
779 | d = dvl[-1]; |
780 | for (b = MPW_BITS / 2; b; b >>= 1) { |
781 | if (d < (MPW_MAX >> b)) { |
782 | d <<= b; |
783 | norm += b; |
784 | } |
785 | } |
786 | if (dv + 1 == dvl) |
787 | norm += MPW_BITS; |
d03ab969 |
788 | } |
d03ab969 |
789 | |
790 | /* --- Normalize the dividend/remainder to match --- */ |
791 | |
c8a2f9ef |
792 | if (norm) { |
c8a2f9ef |
793 | mpx_lsl(rv, rvl, rv, rvl, norm); |
f45a00c6 |
794 | mpx_lsl(sv, svl, dv, dvl, norm); |
c8a2f9ef |
795 | dv = sv; |
f45a00c6 |
796 | dvl = svl; |
c8a2f9ef |
797 | MPX_SHRINK(dv, dvl); |
798 | } |
799 | |
d03ab969 |
800 | MPX_SHRINK(rv, rvl); |
c8a2f9ef |
801 | d = dvl[-1]; |
802 | dd = dvl[-2]; |
d03ab969 |
803 | |
804 | /* --- Work out the relative scales --- */ |
805 | |
806 | { |
807 | size_t rvn = rvl - rv; |
c8a2f9ef |
808 | size_t dvn = dvl - dv; |
d03ab969 |
809 | |
810 | /* --- If the divisor is clearly larger, notice this --- */ |
811 | |
812 | if (dvn > rvn) { |
813 | mpx_lsr(rv, rvl, rv, rvl, norm); |
814 | return; |
815 | } |
816 | |
817 | scale = rvn - dvn; |
818 | } |
819 | |
820 | /* --- Calculate the most significant quotient digit --- * |
821 | * |
822 | * Because the divisor has its top bit set, this can only happen once. The |
823 | * pointer arithmetic is a little contorted, to make sure that the |
824 | * behaviour is defined. |
825 | */ |
826 | |
827 | if (MPX_UCMP(rv + scale, rvl, >=, dv, dvl)) { |
828 | mpx_usub(rv + scale, rvl, rv + scale, rvl, dv, dvl); |
829 | if (qvl - qv > scale) |
830 | qv[scale] = 1; |
831 | } |
832 | |
833 | /* --- Now for the main loop --- */ |
834 | |
835 | { |
c8a2f9ef |
836 | mpw *rvv = rvl - 2; |
d03ab969 |
837 | |
838 | while (scale) { |
c8a2f9ef |
839 | mpw q; |
840 | mpd rh; |
d03ab969 |
841 | |
842 | /* --- Get an estimate for the next quotient digit --- */ |
843 | |
c8a2f9ef |
844 | mpw r = rvv[1]; |
845 | mpw rr = rvv[0]; |
846 | mpw rrr = *--rvv; |
847 | |
848 | scale--; |
849 | rh = ((mpd)r << MPW_BITS) | rr; |
d03ab969 |
850 | if (r == d) |
851 | q = MPW_MAX; |
c8a2f9ef |
852 | else |
853 | q = MPW(rh / d); |
d03ab969 |
854 | |
855 | /* --- Refine the estimate --- */ |
856 | |
857 | { |
858 | mpd yh = (mpd)d * q; |
ce76ff16 |
859 | mpd yy = (mpd)dd * q; |
860 | mpw yl; |
c8a2f9ef |
861 | |
ce76ff16 |
862 | if (yy > MPW_MAX) |
863 | yh += yy >> MPW_BITS; |
864 | yl = MPW(yy); |
c8a2f9ef |
865 | |
866 | while (yh > rh || (yh == rh && yl > rrr)) { |
867 | q--; |
868 | yh -= d; |
ce76ff16 |
869 | if (yl < dd) |
870 | yh--; |
99b30c23 |
871 | yl = MPW(yl - dd); |
c8a2f9ef |
872 | } |
873 | } |
874 | |
875 | /* --- Remove a chunk from the dividend --- */ |
876 | |
877 | { |
878 | mpw *svv; |
879 | const mpw *dvv; |
f45a00c6 |
880 | mpw mc = 0, sc = 0; |
c8a2f9ef |
881 | |
f45a00c6 |
882 | /* --- Calculate the size of the chunk --- * |
883 | * |
884 | * This does the whole job of calculating @r >> scale - qd@. |
885 | */ |
c8a2f9ef |
886 | |
f45a00c6 |
887 | for (svv = rv + scale, dvv = dv; |
888 | dvv < dvl && svv < rvl; |
889 | svv++, dvv++) { |
890 | mpd x = (mpd)*dvv * (mpd)q + mc; |
891 | mc = x >> MPW_BITS; |
892 | x = (mpd)*svv - MPW(x) - sc; |
c8a2f9ef |
893 | *svv = MPW(x); |
f45a00c6 |
894 | if (x >> MPW_BITS) |
895 | sc = 1; |
896 | else |
897 | sc = 0; |
898 | } |
899 | |
900 | if (svv < rvl) { |
901 | mpd x = (mpd)*svv - mc - sc; |
902 | *svv++ = MPW(x); |
903 | if (x >> MPW_BITS) |
904 | sc = MPW_MAX; |
905 | else |
906 | sc = 0; |
907 | while (svv < rvl) |
908 | *svv++ = sc; |
c8a2f9ef |
909 | } |
c8a2f9ef |
910 | |
f45a00c6 |
911 | /* --- Fix if the quotient was too large --- * |
c8a2f9ef |
912 | * |
f45a00c6 |
913 | * This doesn't seem to happen very often. |
c8a2f9ef |
914 | */ |
915 | |
c8a2f9ef |
916 | if (rvl[-1] > MPW_MAX / 2) { |
917 | mpx_uadd(rv + scale, rvl, rv + scale, rvl, dv, dvl); |
918 | q--; |
919 | } |
920 | } |
921 | |
922 | /* --- Done for another iteration --- */ |
923 | |
924 | if (qvl - qv > scale) |
925 | qv[scale] = q; |
926 | r = rr; |
927 | rr = rrr; |
928 | } |
929 | } |
930 | |
931 | /* --- Now fiddle with unnormalizing and things --- */ |
932 | |
933 | mpx_lsr(rv, rvl, rv, rvl, norm); |
d03ab969 |
934 | } |
935 | |
698bd937 |
936 | /* --- @mpx_udivn@ --- * |
937 | * |
938 | * Arguments: @mpw *qv, *qvl@ = storage for the quotient (may overlap |
939 | * dividend) |
940 | * @const mpw *rv, *rvl@ = dividend |
941 | * @mpw d@ = single-precision divisor |
942 | * |
943 | * Returns: Remainder after divison. |
944 | * |
945 | * Use: Performs a single-precision division operation. |
946 | */ |
947 | |
948 | mpw mpx_udivn(mpw *qv, mpw *qvl, const mpw *rv, const mpw *rvl, mpw d) |
949 | { |
950 | size_t i; |
951 | size_t ql = qvl - qv; |
952 | mpd r = 0; |
953 | |
954 | i = rvl - rv; |
955 | while (i > 0) { |
956 | i--; |
957 | r = (r << MPW_BITS) | rv[i]; |
958 | if (i < ql) |
959 | qv[i] = r / d; |
960 | r %= d; |
961 | } |
962 | return (MPW(r)); |
963 | } |
964 | |
42684bdb |
965 | /*----- Test rig ----------------------------------------------------------*/ |
966 | |
967 | #ifdef TEST_RIG |
968 | |
969 | #include <mLib/alloc.h> |
970 | #include <mLib/dstr.h> |
971 | #include <mLib/quis.h> |
972 | #include <mLib/testrig.h> |
973 | |
974 | #include "mpscan.h" |
975 | |
976 | #define ALLOC(v, vl, sz) do { \ |
977 | size_t _sz = (sz); \ |
978 | mpw *_vv = xmalloc(MPWS(_sz)); \ |
979 | mpw *_vvl = _vv + _sz; \ |
980 | (v) = _vv; \ |
981 | (vl) = _vvl; \ |
982 | } while (0) |
983 | |
984 | #define LOAD(v, vl, d) do { \ |
985 | const dstr *_d = (d); \ |
986 | mpw *_v, *_vl; \ |
987 | ALLOC(_v, _vl, MPW_RQ(_d->len)); \ |
988 | mpx_loadb(_v, _vl, _d->buf, _d->len); \ |
989 | (v) = _v; \ |
990 | (vl) = _vl; \ |
991 | } while (0) |
992 | |
993 | #define MAX(x, y) ((x) > (y) ? (x) : (y)) |
994 | |
995 | static void dumpbits(const char *msg, const void *pp, size_t sz) |
996 | { |
997 | const octet *p = pp; |
998 | fputs(msg, stderr); |
999 | for (; sz; sz--) |
1000 | fprintf(stderr, " %02x", *p++); |
1001 | fputc('\n', stderr); |
1002 | } |
1003 | |
1004 | static void dumpmp(const char *msg, const mpw *v, const mpw *vl) |
1005 | { |
1006 | fputs(msg, stderr); |
1007 | MPX_SHRINK(v, vl); |
1008 | while (v < vl) |
1009 | fprintf(stderr, " %08lx", (unsigned long)*--vl); |
1010 | fputc('\n', stderr); |
1011 | } |
1012 | |
1013 | static int chkscan(const mpw *v, const mpw *vl, |
1014 | const void *pp, size_t sz, int step) |
1015 | { |
1016 | mpscan mps; |
1017 | const octet *p = pp; |
1018 | unsigned bit = 0; |
1019 | int ok = 1; |
1020 | |
1021 | mpscan_initx(&mps, v, vl); |
1022 | while (sz) { |
1023 | unsigned x = *p; |
1024 | int i; |
1025 | p += step; |
1026 | for (i = 0; i < 8 && MPSCAN_STEP(&mps); i++) { |
1027 | if (MPSCAN_BIT(&mps) != (x & 1)) { |
1028 | fprintf(stderr, |
1029 | "\n*** error, step %i, bit %u, expected %u, found %u\n", |
1030 | step, bit, x & 1, MPSCAN_BIT(&mps)); |
1031 | ok = 0; |
1032 | } |
1033 | x >>= 1; |
1034 | bit++; |
1035 | } |
1036 | sz--; |
1037 | } |
1038 | |
1039 | return (ok); |
1040 | } |
1041 | |
1042 | static int loadstore(dstr *v) |
1043 | { |
1044 | dstr d = DSTR_INIT; |
1045 | size_t sz = MPW_RQ(v->len) * 2, diff; |
1046 | mpw *m, *ml; |
1047 | int ok = 1; |
1048 | |
1049 | dstr_ensure(&d, v->len); |
1050 | m = xmalloc(MPWS(sz)); |
1051 | |
1052 | for (diff = 0; diff < sz; diff += 5) { |
1053 | size_t oct; |
1054 | |
1055 | ml = m + sz - diff; |
1056 | |
1057 | mpx_loadl(m, ml, v->buf, v->len); |
1058 | if (!chkscan(m, ml, v->buf, v->len, +1)) |
1059 | ok = 0; |
1060 | MPX_OCTETS(oct, m, ml); |
1061 | mpx_storel(m, ml, d.buf, d.sz); |
1062 | if (memcmp(d.buf, v->buf, oct) != 0) { |
1063 | dumpbits("\n*** storel failed", d.buf, d.sz); |
1064 | ok = 0; |
1065 | } |
1066 | |
1067 | mpx_loadb(m, ml, v->buf, v->len); |
1068 | if (!chkscan(m, ml, v->buf + v->len - 1, v->len, -1)) |
1069 | ok = 0; |
1070 | MPX_OCTETS(oct, m, ml); |
1071 | mpx_storeb(m, ml, d.buf, d.sz); |
1072 | if (memcmp(d.buf + d.sz - oct, v->buf + v->len - oct, oct) != 0) { |
1073 | dumpbits("\n*** storeb failed", d.buf, d.sz); |
1074 | ok = 0; |
1075 | } |
1076 | } |
1077 | |
1078 | if (!ok) |
1079 | dumpbits("input data", v->buf, v->len); |
1080 | |
1081 | free(m); |
1082 | dstr_destroy(&d); |
1083 | return (ok); |
1084 | } |
1085 | |
1086 | static int lsl(dstr *v) |
1087 | { |
1088 | mpw *a, *al; |
1089 | int n = *(int *)v[1].buf; |
1090 | mpw *c, *cl; |
1091 | mpw *d, *dl; |
1092 | int ok = 1; |
1093 | |
1094 | LOAD(a, al, &v[0]); |
1095 | LOAD(c, cl, &v[2]); |
1096 | ALLOC(d, dl, al - a + (n + MPW_BITS - 1) / MPW_BITS); |
1097 | |
1098 | mpx_lsl(d, dl, a, al, n); |
1a05a8ef |
1099 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1100 | fprintf(stderr, "\n*** lsl(%i) failed\n", n); |
1101 | dumpmp(" a", a, al); |
1102 | dumpmp("expected", c, cl); |
1103 | dumpmp(" result", d, dl); |
1104 | ok = 0; |
1105 | } |
1106 | |
1107 | free(a); free(c); free(d); |
1108 | return (ok); |
1109 | } |
1110 | |
1111 | static int lsr(dstr *v) |
1112 | { |
1113 | mpw *a, *al; |
1114 | int n = *(int *)v[1].buf; |
1115 | mpw *c, *cl; |
1116 | mpw *d, *dl; |
1117 | int ok = 1; |
1118 | |
1119 | LOAD(a, al, &v[0]); |
1120 | LOAD(c, cl, &v[2]); |
1121 | ALLOC(d, dl, al - a + (n + MPW_BITS - 1) / MPW_BITS + 1); |
1122 | |
1123 | mpx_lsr(d, dl, a, al, n); |
1a05a8ef |
1124 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1125 | fprintf(stderr, "\n*** lsr(%i) failed\n", n); |
1126 | dumpmp(" a", a, al); |
1127 | dumpmp("expected", c, cl); |
1128 | dumpmp(" result", d, dl); |
1129 | ok = 0; |
1130 | } |
1131 | |
1132 | free(a); free(c); free(d); |
1133 | return (ok); |
1134 | } |
1135 | |
1136 | static int uadd(dstr *v) |
1137 | { |
1138 | mpw *a, *al; |
1139 | mpw *b, *bl; |
1140 | mpw *c, *cl; |
1141 | mpw *d, *dl; |
1142 | int ok = 1; |
1143 | |
1144 | LOAD(a, al, &v[0]); |
1145 | LOAD(b, bl, &v[1]); |
1146 | LOAD(c, cl, &v[2]); |
1147 | ALLOC(d, dl, MAX(al - a, bl - b) + 1); |
1148 | |
1149 | mpx_uadd(d, dl, a, al, b, bl); |
1a05a8ef |
1150 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1151 | fprintf(stderr, "\n*** uadd failed\n"); |
1152 | dumpmp(" a", a, al); |
1153 | dumpmp(" b", b, bl); |
1154 | dumpmp("expected", c, cl); |
1155 | dumpmp(" result", d, dl); |
1156 | ok = 0; |
1157 | } |
1158 | |
1159 | free(a); free(b); free(c); free(d); |
1160 | return (ok); |
1161 | } |
1162 | |
1163 | static int usub(dstr *v) |
1164 | { |
1165 | mpw *a, *al; |
1166 | mpw *b, *bl; |
1167 | mpw *c, *cl; |
1168 | mpw *d, *dl; |
1169 | int ok = 1; |
1170 | |
1171 | LOAD(a, al, &v[0]); |
1172 | LOAD(b, bl, &v[1]); |
1173 | LOAD(c, cl, &v[2]); |
1174 | ALLOC(d, dl, al - a); |
1175 | |
1176 | mpx_usub(d, dl, a, al, b, bl); |
1a05a8ef |
1177 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1178 | fprintf(stderr, "\n*** usub failed\n"); |
1179 | dumpmp(" a", a, al); |
1180 | dumpmp(" b", b, bl); |
1181 | dumpmp("expected", c, cl); |
1182 | dumpmp(" result", d, dl); |
1183 | ok = 0; |
1184 | } |
1185 | |
1186 | free(a); free(b); free(c); free(d); |
1187 | return (ok); |
1188 | } |
1189 | |
1190 | static int umul(dstr *v) |
1191 | { |
1192 | mpw *a, *al; |
1193 | mpw *b, *bl; |
1194 | mpw *c, *cl; |
1195 | mpw *d, *dl; |
1196 | int ok = 1; |
1197 | |
1198 | LOAD(a, al, &v[0]); |
1199 | LOAD(b, bl, &v[1]); |
1200 | LOAD(c, cl, &v[2]); |
1201 | ALLOC(d, dl, (al - a) + (bl - b)); |
1202 | |
1203 | mpx_umul(d, dl, a, al, b, bl); |
1a05a8ef |
1204 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1205 | fprintf(stderr, "\n*** umul failed\n"); |
1206 | dumpmp(" a", a, al); |
1207 | dumpmp(" b", b, bl); |
1208 | dumpmp("expected", c, cl); |
1209 | dumpmp(" result", d, dl); |
1210 | ok = 0; |
1211 | } |
1212 | |
1213 | free(a); free(b); free(c); free(d); |
1214 | return (ok); |
1215 | } |
1216 | |
1217 | static int usqr(dstr *v) |
1218 | { |
1219 | mpw *a, *al; |
1220 | mpw *c, *cl; |
1221 | mpw *d, *dl; |
1222 | int ok = 1; |
1223 | |
1224 | LOAD(a, al, &v[0]); |
1225 | LOAD(c, cl, &v[1]); |
1226 | ALLOC(d, dl, 2 * (al - a)); |
1227 | |
1228 | mpx_usqr(d, dl, a, al); |
1a05a8ef |
1229 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1230 | fprintf(stderr, "\n*** usqr failed\n"); |
1231 | dumpmp(" a", a, al); |
1232 | dumpmp("expected", c, cl); |
1233 | dumpmp(" result", d, dl); |
1234 | ok = 0; |
1235 | } |
1236 | |
1237 | free(a); free(c); free(d); |
1238 | return (ok); |
1239 | } |
1240 | |
1241 | static int udiv(dstr *v) |
1242 | { |
1243 | mpw *a, *al; |
1244 | mpw *b, *bl; |
1245 | mpw *q, *ql; |
1246 | mpw *r, *rl; |
1247 | mpw *qq, *qql; |
1248 | mpw *s, *sl; |
1249 | int ok = 1; |
1250 | |
1251 | ALLOC(a, al, MPW_RQ(v[0].len) + 2); mpx_loadb(a, al, v[0].buf, v[0].len); |
1252 | LOAD(b, bl, &v[1]); |
1253 | LOAD(q, ql, &v[2]); |
1254 | LOAD(r, rl, &v[3]); |
1255 | ALLOC(qq, qql, al - a); |
1256 | ALLOC(s, sl, (bl - b) + 1); |
1257 | |
1258 | mpx_udiv(qq, qql, a, al, b, bl, s, sl); |
1a05a8ef |
1259 | if (!mpx_ueq(qq, qql, q, ql) || |
1260 | !mpx_ueq(a, al, r, rl)) { |
42684bdb |
1261 | fprintf(stderr, "\n*** udiv failed\n"); |
1262 | dumpmp(" divisor", b, bl); |
1263 | dumpmp("expect r", r, rl); |
1264 | dumpmp("result r", a, al); |
1265 | dumpmp("expect q", q, ql); |
1266 | dumpmp("result q", qq, qql); |
1267 | ok = 0; |
1268 | } |
1269 | |
1270 | free(a); free(b); free(r); free(q); free(s); free(qq); |
1271 | return (ok); |
1272 | } |
1273 | |
1274 | static test_chunk defs[] = { |
1275 | { "load-store", loadstore, { &type_hex, 0 } }, |
1276 | { "lsl", lsl, { &type_hex, &type_int, &type_hex, 0 } }, |
1277 | { "lsr", lsr, { &type_hex, &type_int, &type_hex, 0 } }, |
1278 | { "uadd", uadd, { &type_hex, &type_hex, &type_hex, 0 } }, |
1279 | { "usub", usub, { &type_hex, &type_hex, &type_hex, 0 } }, |
1280 | { "umul", umul, { &type_hex, &type_hex, &type_hex, 0 } }, |
1281 | { "usqr", usqr, { &type_hex, &type_hex, 0 } }, |
1282 | { "udiv", udiv, { &type_hex, &type_hex, &type_hex, &type_hex, 0 } }, |
1283 | { 0, 0, { 0 } } |
1284 | }; |
1285 | |
1286 | int main(int argc, char *argv[]) |
1287 | { |
1288 | test_run(argc, argv, defs, SRCDIR"/tests/mpx"); |
1289 | return (0); |
1290 | } |
1291 | |
42684bdb |
1292 | #endif |
1293 | |
d03ab969 |
1294 | /*----- That's all, folks -------------------------------------------------*/ |