d03ab969 |
1 | /* -*-c-*- |
2 | * |
0f32e0f8 |
3 | * $Id: mpx.c,v 1.11 2001/04/03 19:36:05 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 $ |
0f32e0f8 |
33 | * Revision 1.11 2001/04/03 19:36:05 mdw |
34 | * Add some simple bitwise operations so that Perl can use them. |
35 | * |
1a05a8ef |
36 | * Revision 1.10 2000/10/08 12:06:12 mdw |
37 | * Provide @mpx_ueq@ for rapidly testing equality of two integers. |
38 | * |
99b30c23 |
39 | * Revision 1.9 2000/06/26 07:52:50 mdw |
40 | * Portability fix for the bug fix. |
41 | * |
ce76ff16 |
42 | * Revision 1.8 2000/06/25 12:59:02 mdw |
43 | * (mpx_udiv): Fix bug in quotient digit estimation. |
44 | * |
698bd937 |
45 | * Revision 1.7 1999/12/22 15:49:07 mdw |
46 | * New function for division by a small integer. |
47 | * |
42684bdb |
48 | * Revision 1.6 1999/11/20 22:43:44 mdw |
49 | * Integrate testing for MPX routines. |
50 | * |
dd517851 |
51 | * Revision 1.5 1999/11/20 22:23:27 mdw |
52 | * Add function versions of some low-level macros with wider use. |
53 | * |
f45a00c6 |
54 | * Revision 1.4 1999/11/17 18:04:09 mdw |
55 | * Add two's-complement functionality. Improve mpx_udiv a little by |
56 | * performing the multiplication of the divisor by q with the subtraction |
57 | * from r. |
58 | * |
501da53c |
59 | * Revision 1.3 1999/11/13 01:57:31 mdw |
60 | * Remove stray debugging code. |
61 | * |
c8a2f9ef |
62 | * Revision 1.2 1999/11/13 01:50:59 mdw |
63 | * Multiprecision routines finished and tested. |
64 | * |
d03ab969 |
65 | * Revision 1.1 1999/09/03 08:41:12 mdw |
66 | * Initial import. |
67 | * |
68 | */ |
69 | |
70 | /*----- Header files ------------------------------------------------------*/ |
71 | |
c8a2f9ef |
72 | #include <assert.h> |
d03ab969 |
73 | #include <stdio.h> |
74 | #include <stdlib.h> |
75 | #include <string.h> |
76 | |
77 | #include <mLib/bits.h> |
78 | |
79 | #include "mptypes.h" |
80 | #include "mpx.h" |
81 | |
82 | /*----- Loading and storing -----------------------------------------------*/ |
83 | |
84 | /* --- @mpx_storel@ --- * |
85 | * |
86 | * Arguments: @const mpw *v, *vl@ = base and limit of source vector |
c8a2f9ef |
87 | * @void *pp@ = pointer to octet array |
d03ab969 |
88 | * @size_t sz@ = size of octet array |
89 | * |
90 | * Returns: --- |
91 | * |
92 | * Use: Stores an MP in an octet array, least significant octet |
93 | * first. High-end octets are silently discarded if there |
94 | * isn't enough space for them. |
95 | */ |
96 | |
c8a2f9ef |
97 | void mpx_storel(const mpw *v, const mpw *vl, void *pp, size_t sz) |
d03ab969 |
98 | { |
99 | mpw n, w = 0; |
c8a2f9ef |
100 | octet *p = pp, *q = p + sz; |
d03ab969 |
101 | unsigned bits = 0; |
102 | |
103 | while (p < q) { |
104 | if (bits < 8) { |
105 | if (v >= vl) { |
106 | *p++ = U8(w); |
107 | break; |
108 | } |
109 | n = *v++; |
110 | *p++ = U8(w | n << bits); |
111 | w = n >> (8 - bits); |
112 | bits += MPW_BITS - 8; |
113 | } else { |
114 | *p++ = U8(w); |
115 | w >>= 8; |
116 | bits -= 8; |
117 | } |
118 | } |
119 | memset(p, 0, q - p); |
120 | } |
121 | |
122 | /* --- @mpx_loadl@ --- * |
123 | * |
124 | * Arguments: @mpw *v, *vl@ = base and limit of destination vector |
c8a2f9ef |
125 | * @const void *pp@ = pointer to octet array |
d03ab969 |
126 | * @size_t sz@ = size of octet array |
127 | * |
128 | * Returns: --- |
129 | * |
130 | * Use: Loads an MP in an octet array, least significant octet |
131 | * first. High-end octets are ignored if there isn't enough |
132 | * space for them. |
133 | */ |
134 | |
c8a2f9ef |
135 | void mpx_loadl(mpw *v, mpw *vl, const void *pp, size_t sz) |
d03ab969 |
136 | { |
137 | unsigned n; |
c8a2f9ef |
138 | mpw w = 0; |
139 | const octet *p = pp, *q = p + sz; |
d03ab969 |
140 | unsigned bits = 0; |
141 | |
142 | if (v >= vl) |
143 | return; |
144 | while (p < q) { |
145 | n = U8(*p++); |
146 | w |= n << bits; |
147 | bits += 8; |
148 | if (bits >= MPW_BITS) { |
149 | *v++ = MPW(w); |
150 | w = n >> (MPW_BITS - bits + 8); |
151 | bits -= MPW_BITS; |
152 | if (v >= vl) |
153 | return; |
154 | } |
155 | } |
156 | *v++ = w; |
157 | MPX_ZERO(v, vl); |
158 | } |
159 | |
160 | /* --- @mpx_storeb@ --- * |
161 | * |
162 | * Arguments: @const mpw *v, *vl@ = base and limit of source vector |
c8a2f9ef |
163 | * @void *pp@ = pointer to octet array |
d03ab969 |
164 | * @size_t sz@ = size of octet array |
165 | * |
166 | * Returns: --- |
167 | * |
168 | * Use: Stores an MP in an octet array, most significant octet |
169 | * first. High-end octets are silently discarded if there |
170 | * isn't enough space for them. |
171 | */ |
172 | |
c8a2f9ef |
173 | void mpx_storeb(const mpw *v, const mpw *vl, void *pp, size_t sz) |
d03ab969 |
174 | { |
175 | mpw n, w = 0; |
c8a2f9ef |
176 | octet *p = pp, *q = p + sz; |
d03ab969 |
177 | unsigned bits = 0; |
178 | |
179 | while (q > p) { |
180 | if (bits < 8) { |
181 | if (v >= vl) { |
182 | *--q = U8(w); |
183 | break; |
184 | } |
185 | n = *v++; |
186 | *--q = U8(w | n << bits); |
187 | w = n >> (8 - bits); |
188 | bits += MPW_BITS - 8; |
189 | } else { |
190 | *--q = U8(w); |
191 | w >>= 8; |
192 | bits -= 8; |
193 | } |
194 | } |
195 | memset(p, 0, q - p); |
196 | } |
197 | |
198 | /* --- @mpx_loadb@ --- * |
199 | * |
200 | * Arguments: @mpw *v, *vl@ = base and limit of destination vector |
c8a2f9ef |
201 | * @const void *pp@ = pointer to octet array |
d03ab969 |
202 | * @size_t sz@ = size of octet array |
203 | * |
204 | * Returns: --- |
205 | * |
206 | * Use: Loads an MP in an octet array, most significant octet |
207 | * first. High-end octets are ignored if there isn't enough |
208 | * space for them. |
209 | */ |
210 | |
c8a2f9ef |
211 | void mpx_loadb(mpw *v, mpw *vl, const void *pp, size_t sz) |
d03ab969 |
212 | { |
213 | unsigned n; |
c8a2f9ef |
214 | mpw w = 0; |
215 | const octet *p = pp, *q = p + sz; |
d03ab969 |
216 | unsigned bits = 0; |
217 | |
218 | if (v >= vl) |
219 | return; |
220 | while (q > p) { |
221 | n = U8(*--q); |
222 | w |= n << bits; |
223 | bits += 8; |
224 | if (bits >= MPW_BITS) { |
225 | *v++ = MPW(w); |
226 | w = n >> (MPW_BITS - bits + 8); |
227 | bits -= MPW_BITS; |
228 | if (v >= vl) |
229 | return; |
230 | } |
231 | } |
232 | *v++ = w; |
233 | MPX_ZERO(v, vl); |
234 | } |
235 | |
236 | /*----- Logical shifting --------------------------------------------------*/ |
237 | |
238 | /* --- @mpx_lsl@ --- * |
239 | * |
240 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
241 | * @const mpw *av, *avl@ = source vector base and limit |
242 | * @size_t n@ = number of bit positions to shift by |
243 | * |
244 | * Returns: --- |
245 | * |
246 | * Use: Performs a logical shift left operation on an integer. |
247 | */ |
248 | |
249 | void mpx_lsl(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, size_t n) |
250 | { |
251 | size_t nw; |
252 | unsigned nb; |
253 | |
254 | /* --- Trivial special case --- */ |
255 | |
256 | if (n == 0) |
257 | MPX_COPY(dv, dvl, av, avl); |
258 | |
259 | /* --- Single bit shifting --- */ |
260 | |
261 | else if (n == 1) { |
262 | mpw w = 0; |
263 | while (av < avl) { |
264 | mpw t; |
265 | if (dv >= dvl) |
266 | goto done; |
267 | t = *av++; |
268 | *dv++ = MPW((t << 1) | w); |
269 | w = t >> (MPW_BITS - 1); |
270 | } |
271 | if (dv >= dvl) |
272 | goto done; |
273 | *dv++ = MPW(w); |
274 | MPX_ZERO(dv, dvl); |
c8a2f9ef |
275 | goto done; |
d03ab969 |
276 | } |
277 | |
278 | /* --- Break out word and bit shifts for more sophisticated work --- */ |
279 | |
280 | nw = n / MPW_BITS; |
281 | nb = n % MPW_BITS; |
282 | |
283 | /* --- Handle a shift by a multiple of the word size --- */ |
284 | |
285 | if (nb == 0) { |
286 | MPX_COPY(dv + nw, dvl, av, avl); |
287 | memset(dv, 0, MPWS(nw)); |
288 | } |
289 | |
c8a2f9ef |
290 | /* --- And finally the difficult case --- * |
291 | * |
292 | * This is a little convoluted, because I have to start from the end and |
293 | * work backwards to avoid overwriting the source, if they're both the same |
294 | * block of memory. |
295 | */ |
d03ab969 |
296 | |
297 | else { |
298 | mpw w; |
299 | size_t nr = MPW_BITS - nb; |
c8a2f9ef |
300 | size_t dvn = dvl - dv; |
301 | size_t avn = avl - av; |
d03ab969 |
302 | |
c8a2f9ef |
303 | if (dvn <= nw) { |
d03ab969 |
304 | MPX_ZERO(dv, dvl); |
305 | goto done; |
306 | } |
d03ab969 |
307 | |
c8a2f9ef |
308 | if (dvn > avn + nw) { |
309 | size_t off = avn + nw + 1; |
310 | MPX_ZERO(dv + off, dvl); |
311 | dvl = dv + off; |
312 | w = 0; |
313 | } else { |
314 | avl = av + dvn - nw; |
315 | w = *--avl << nb; |
d03ab969 |
316 | } |
317 | |
c8a2f9ef |
318 | while (avl > av) { |
319 | mpw t = *--avl; |
320 | *--dvl = (t >> nr) | w; |
321 | w = t << nb; |
d03ab969 |
322 | } |
c8a2f9ef |
323 | |
324 | *--dvl = w; |
325 | MPX_ZERO(dv, dvl); |
d03ab969 |
326 | } |
327 | |
328 | done:; |
329 | } |
330 | |
331 | /* --- @mpx_lsr@ --- * |
332 | * |
333 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
334 | * @const mpw *av, *avl@ = source vector base and limit |
335 | * @size_t n@ = number of bit positions to shift by |
336 | * |
337 | * Returns: --- |
338 | * |
339 | * Use: Performs a logical shift right operation on an integer. |
340 | */ |
341 | |
342 | void mpx_lsr(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, size_t n) |
343 | { |
344 | size_t nw; |
345 | unsigned nb; |
346 | |
347 | /* --- Trivial special case --- */ |
348 | |
349 | if (n == 0) |
350 | MPX_COPY(dv, dvl, av, avl); |
351 | |
352 | /* --- Single bit shifting --- */ |
353 | |
354 | else if (n == 1) { |
355 | mpw w = *av++ >> 1; |
356 | while (av < avl) { |
357 | mpw t; |
358 | if (dv >= dvl) |
359 | goto done; |
360 | t = *av++; |
361 | *dv++ = MPW((t << (MPW_BITS - 1)) | w); |
362 | w = t >> 1; |
363 | } |
364 | if (dv >= dvl) |
365 | goto done; |
366 | *dv++ = MPW(w); |
367 | MPX_ZERO(dv, dvl); |
c8a2f9ef |
368 | goto done; |
d03ab969 |
369 | } |
370 | |
371 | /* --- Break out word and bit shifts for more sophisticated work --- */ |
372 | |
373 | nw = n / MPW_BITS; |
374 | nb = n % MPW_BITS; |
375 | |
376 | /* --- Handle a shift by a multiple of the word size --- */ |
377 | |
378 | if (nb == 0) |
379 | MPX_COPY(dv, dvl, av + nw, avl); |
380 | |
381 | /* --- And finally the difficult case --- */ |
382 | |
383 | else { |
384 | mpw w; |
385 | size_t nr = MPW_BITS - nb; |
386 | |
387 | av += nw; |
388 | w = *av++; |
389 | while (av < avl) { |
390 | mpw t; |
391 | if (dv >= dvl) |
392 | goto done; |
393 | t = *av++; |
394 | *dv++ = MPW((w >> nb) | (t << nr)); |
395 | w = t; |
396 | } |
397 | if (dv < dvl) { |
398 | *dv++ = MPW(w >> nb); |
399 | MPX_ZERO(dv, dvl); |
400 | } |
401 | } |
402 | |
403 | done:; |
404 | } |
405 | |
0f32e0f8 |
406 | /*----- Bitwise operations ------------------------------------------------*/ |
407 | |
408 | /* --- @mpx_and@, @mpx_or@, @mpx_xor@, @mpx_not@ --- * |
409 | * |
410 | * Arguments: @mpw *dv, *dvl@ = destination vector |
411 | * @const mpw *av, *avl@ = first source vector |
412 | * @const mpw *bv, *bvl@ = second source vector |
413 | * |
414 | * Returns: --- |
415 | * |
416 | * Use; Does the obvious bitwise operations. |
417 | */ |
418 | |
419 | #define MPX_BITBINOP(name, op) \ |
420 | \ |
421 | void mpx_##name(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, \ |
422 | const mpw *bv, const mpw *bvl) \ |
423 | { \ |
424 | MPX_SHRINK(av, avl); \ |
425 | MPX_SHRINK(bv, bvl); \ |
426 | \ |
427 | while (dv < dvl) { \ |
428 | mpw a, b; \ |
429 | a = (av < avl) ? *av++ : 0; \ |
430 | b = (bv < bvl) ? *bv++ : 0; \ |
431 | *dv++ = a op b; \ |
432 | } \ |
433 | } |
434 | |
435 | MPX_BITBINOP(and, &) |
436 | MPX_BITBINOP(or, |) |
437 | MPX_BITBINOP(xor, ^) |
438 | |
439 | void mpx_not(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl) |
440 | { |
441 | MPX_SHRINK(av, avl); |
442 | |
443 | while (dv < dvl) { |
444 | mpw a; |
445 | a = (av < avl) ? *av++ : 0; |
446 | *dv++ = ~a; |
447 | } |
448 | } |
449 | |
d03ab969 |
450 | /*----- Unsigned arithmetic -----------------------------------------------*/ |
451 | |
f45a00c6 |
452 | /* --- @mpx_2c@ --- * |
453 | * |
454 | * Arguments: @mpw *dv, *dvl@ = destination vector |
455 | * @const mpw *v, *vl@ = source vector |
456 | * |
457 | * Returns: --- |
458 | * |
459 | * Use: Calculates the two's complement of @v@. |
460 | */ |
461 | |
462 | void mpx_2c(mpw *dv, mpw *dvl, const mpw *v, const mpw *vl) |
463 | { |
464 | mpw c = 0; |
465 | while (dv < dvl && v < vl) |
466 | *dv++ = c = MPW(~*v++); |
467 | if (dv < dvl) { |
468 | if (c > MPW_MAX / 2) |
469 | c = MPW(~0); |
470 | while (dv < dvl) |
471 | *dv++ = c; |
472 | } |
473 | MPX_UADDN(dv, dvl, 1); |
474 | } |
475 | |
1a05a8ef |
476 | /* --- @mpx_ueq@ --- * |
477 | * |
478 | * Arguments: @const mpw *av, *avl@ = first argument vector base and limit |
479 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
480 | * |
481 | * Returns: Nonzero if the two vectors are equal. |
482 | * |
483 | * Use: Performs an unsigned integer test for equality. |
484 | */ |
485 | |
486 | int mpx_ueq(const mpw *av, const mpw *avl, const mpw *bv, const mpw *bvl) |
487 | { |
488 | MPX_SHRINK(av, avl); |
489 | MPX_SHRINK(bv, bvl); |
490 | if (avl - av != bvl - bv) |
491 | return (0); |
492 | while (av < avl) { |
493 | if (*av++ != *bv++) |
494 | return (0); |
495 | } |
496 | return (1); |
497 | } |
498 | |
d03ab969 |
499 | /* --- @mpx_ucmp@ --- * |
500 | * |
501 | * Arguments: @const mpw *av, *avl@ = first argument vector base and limit |
502 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
503 | * |
504 | * Returns: Less than, equal to, or greater than zero depending on |
505 | * whether @a@ is less than, equal to or greater than @b@, |
506 | * respectively. |
507 | * |
508 | * Use: Performs an unsigned integer comparison. |
509 | */ |
510 | |
511 | int mpx_ucmp(const mpw *av, const mpw *avl, const mpw *bv, const mpw *bvl) |
512 | { |
513 | MPX_SHRINK(av, avl); |
514 | MPX_SHRINK(bv, bvl); |
515 | |
516 | if (avl - av > bvl - bv) |
517 | return (+1); |
518 | else if (avl - av < bvl - bv) |
519 | return (-1); |
520 | else while (avl > av) { |
521 | mpw a = *--avl, b = *--bvl; |
522 | if (a > b) |
523 | return (+1); |
524 | else if (a < b) |
525 | return (-1); |
526 | } |
527 | return (0); |
528 | } |
1a05a8ef |
529 | |
d03ab969 |
530 | /* --- @mpx_uadd@ --- * |
531 | * |
532 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
533 | * @const mpw *av, *avl@ = first addend vector base and limit |
534 | * @const mpw *bv, *bvl@ = second addend vector base and limit |
535 | * |
536 | * Returns: --- |
537 | * |
538 | * Use: Performs unsigned integer addition. If the result overflows |
539 | * the destination vector, high-order bits are discarded. This |
540 | * means that two's complement addition happens more or less for |
541 | * free, although that's more a side-effect than anything else. |
542 | * The result vector may be equal to either or both source |
543 | * vectors, but may not otherwise overlap them. |
544 | */ |
545 | |
546 | void mpx_uadd(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
547 | const mpw *bv, const mpw *bvl) |
548 | { |
549 | mpw c = 0; |
550 | |
551 | while (av < avl || bv < bvl) { |
552 | mpw a, b; |
553 | mpd x; |
554 | if (dv >= dvl) |
555 | return; |
556 | a = (av < avl) ? *av++ : 0; |
557 | b = (bv < bvl) ? *bv++ : 0; |
558 | x = (mpd)a + (mpd)b + c; |
559 | *dv++ = MPW(x); |
560 | c = x >> MPW_BITS; |
561 | } |
562 | if (dv < dvl) { |
563 | *dv++ = c; |
564 | MPX_ZERO(dv, dvl); |
565 | } |
566 | } |
567 | |
dd517851 |
568 | /* --- @mpx_uaddn@ --- * |
569 | * |
570 | * Arguments: @mpw *dv, *dvl@ = source and destination base and limit |
571 | * @mpw n@ = other addend |
572 | * |
573 | * Returns: --- |
574 | * |
575 | * Use: Adds a small integer to a multiprecision number. |
576 | */ |
577 | |
578 | void mpx_uaddn(mpw *dv, mpw *dvl, mpw n) { MPX_UADDN(dv, dvl, n); } |
579 | |
d03ab969 |
580 | /* --- @mpx_usub@ --- * |
581 | * |
582 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
583 | * @const mpw *av, *avl@ = first argument vector base and limit |
584 | * @const mpw *bv, *bvl@ = second argument vector base and limit |
585 | * |
586 | * Returns: --- |
587 | * |
588 | * Use: Performs unsigned integer subtraction. If the result |
589 | * overflows the destination vector, high-order bits are |
590 | * discarded. This means that two's complement subtraction |
591 | * happens more or less for free, althuogh that's more a side- |
592 | * effect than anything else. The result vector may be equal to |
593 | * either or both source vectors, but may not otherwise overlap |
594 | * them. |
595 | */ |
596 | |
597 | void mpx_usub(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
598 | const mpw *bv, const mpw *bvl) |
599 | { |
600 | mpw c = 0; |
601 | |
602 | while (av < avl || bv < bvl) { |
603 | mpw a, b; |
604 | mpd x; |
605 | if (dv >= dvl) |
606 | return; |
607 | a = (av < avl) ? *av++ : 0; |
608 | b = (bv < bvl) ? *bv++ : 0; |
c8a2f9ef |
609 | x = (mpd)a - (mpd)b - c; |
d03ab969 |
610 | *dv++ = MPW(x); |
c8a2f9ef |
611 | if (x >> MPW_BITS) |
612 | c = 1; |
613 | else |
614 | c = 0; |
d03ab969 |
615 | } |
c8a2f9ef |
616 | if (c) |
617 | c = MPW_MAX; |
d03ab969 |
618 | while (dv < dvl) |
c8a2f9ef |
619 | *dv++ = c; |
d03ab969 |
620 | } |
621 | |
dd517851 |
622 | /* --- @mpx_usubn@ --- * |
623 | * |
624 | * Arguments: @mpw *dv, *dvl@ = source and destination base and limit |
625 | * @n@ = subtrahend |
626 | * |
627 | * Returns: --- |
628 | * |
629 | * Use: Subtracts a small integer from a multiprecision number. |
630 | */ |
631 | |
632 | void mpx_usubn(mpw *dv, mpw *dvl, mpw n) { MPX_USUBN(dv, dvl, n); } |
633 | |
d03ab969 |
634 | /* --- @mpx_umul@ --- * |
635 | * |
636 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
637 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
638 | * @const mpw *bv, *bvl@ = multiplier vector base and limit |
639 | * |
640 | * Returns: --- |
641 | * |
642 | * Use: Performs unsigned integer multiplication. If the result |
643 | * overflows the desination vector, high-order bits are |
644 | * discarded. The result vector may not overlap the argument |
645 | * vectors in any way. |
646 | */ |
647 | |
648 | void mpx_umul(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, |
649 | const mpw *bv, const mpw *bvl) |
650 | { |
651 | /* --- This is probably worthwhile on a multiply --- */ |
652 | |
653 | MPX_SHRINK(av, avl); |
654 | MPX_SHRINK(bv, bvl); |
655 | |
656 | /* --- Deal with a multiply by zero --- */ |
657 | |
658 | if (bv == bvl) { |
c8a2f9ef |
659 | MPX_ZERO(dv, dvl); |
d03ab969 |
660 | return; |
661 | } |
662 | |
663 | /* --- Do the initial multiply and initialize the accumulator --- */ |
664 | |
665 | MPX_UMULN(dv, dvl, av, avl, *bv++); |
666 | |
667 | /* --- Do the remaining multiply/accumulates --- */ |
668 | |
c8a2f9ef |
669 | while (dv < dvl && bv < bvl) { |
d03ab969 |
670 | mpw m = *bv++; |
c8a2f9ef |
671 | mpw c = 0; |
d03ab969 |
672 | const mpw *avv = av; |
673 | mpw *dvv = ++dv; |
674 | |
675 | while (avv < avl) { |
676 | mpd x; |
677 | if (dvv >= dvl) |
678 | goto next; |
c8a2f9ef |
679 | x = (mpd)*dvv + (mpd)m * (mpd)*avv++ + c; |
680 | *dvv++ = MPW(x); |
d03ab969 |
681 | c = x >> MPW_BITS; |
682 | } |
c8a2f9ef |
683 | MPX_UADDN(dvv, dvl, c); |
d03ab969 |
684 | next:; |
685 | } |
686 | } |
687 | |
dd517851 |
688 | /* --- @mpx_umuln@ --- * |
689 | * |
690 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
691 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
692 | * @mpw m@ = multiplier |
693 | * |
694 | * Returns: --- |
695 | * |
696 | * Use: Multiplies a multiprecision integer by a single-word value. |
697 | * The destination and source may be equal. The destination |
698 | * is completely cleared after use. |
699 | */ |
700 | |
701 | void mpx_umuln(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, mpw m) |
702 | { |
703 | MPX_UMULN(dv, dvl, av, avl, m); |
704 | } |
705 | |
706 | /* --- @mpx_umlan@ --- * |
707 | * |
708 | * Arguments: @mpw *dv, *dvl@ = destination/accumulator base and limit |
709 | * @const mpw *av, *avl@ = multiplicand vector base and limit |
710 | * @mpw m@ = multiplier |
711 | * |
712 | * Returns: --- |
713 | * |
714 | * Use: Multiplies a multiprecision integer by a single-word value |
715 | * and adds the result to an accumulator. |
716 | */ |
717 | |
718 | void mpx_umlan(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl, mpw m) |
719 | { |
720 | MPX_UMLAN(dv, dvl, av, avl, m); |
721 | } |
722 | |
c8a2f9ef |
723 | /* --- @mpx_usqr@ --- * |
724 | * |
725 | * Arguments: @mpw *dv, *dvl@ = destination vector base and limit |
726 | * @const mpw *av, *av@ = source vector base and limit |
727 | * |
728 | * Returns: --- |
729 | * |
730 | * Use: Performs unsigned integer squaring. The result vector must |
731 | * not overlap the source vector in any way. |
732 | */ |
733 | |
734 | void mpx_usqr(mpw *dv, mpw *dvl, const mpw *av, const mpw *avl) |
735 | { |
736 | MPX_ZERO(dv, dvl); |
737 | |
738 | /* --- Main loop --- */ |
739 | |
740 | while (av < avl) { |
741 | const mpw *avv = av; |
742 | mpw *dvv = dv; |
743 | mpw a = *av; |
744 | mpd c; |
745 | |
746 | /* --- Stop if I've run out of destination --- */ |
747 | |
748 | if (dvv >= dvl) |
749 | break; |
750 | |
751 | /* --- Work out the square at this point in the proceedings --- */ |
752 | |
753 | { |
c8a2f9ef |
754 | mpd x = (mpd)a * (mpd)a + *dvv; |
755 | *dvv++ = MPW(x); |
756 | c = MPW(x >> MPW_BITS); |
757 | } |
758 | |
759 | /* --- Now fix up the rest of the vector upwards --- */ |
760 | |
761 | avv++; |
762 | while (dvv < dvl && avv < avl) { |
c8a2f9ef |
763 | mpd x = (mpd)a * (mpd)*avv++; |
764 | mpd y = ((x << 1) & MPW_MAX) + c + *dvv; |
765 | c = (x >> (MPW_BITS - 1)) + (y >> MPW_BITS); |
766 | *dvv++ = MPW(y); |
767 | } |
768 | while (dvv < dvl && c) { |
769 | mpd x = c + *dvv; |
770 | *dvv++ = MPW(x); |
771 | c = x >> MPW_BITS; |
772 | } |
773 | |
774 | /* --- Get ready for the next round --- */ |
775 | |
776 | av++; |
777 | dv += 2; |
778 | } |
779 | } |
780 | |
d03ab969 |
781 | /* --- @mpx_udiv@ --- * |
782 | * |
783 | * Arguments: @mpw *qv, *qvl@ = quotient vector base and limit |
784 | * @mpw *rv, *rvl@ = dividend/remainder vector base and limit |
785 | * @const mpw *dv, *dvl@ = divisor vector base and limit |
c8a2f9ef |
786 | * @mpw *sv, *svl@ = scratch workspace |
d03ab969 |
787 | * |
788 | * Returns: --- |
789 | * |
790 | * Use: Performs unsigned integer division. If the result overflows |
791 | * the quotient vector, high-order bits are discarded. (Clearly |
792 | * the remainder vector can't overflow.) The various vectors |
793 | * may not overlap in any way. Yes, I know it's a bit odd |
794 | * requiring the dividend to be in the result position but it |
795 | * does make some sense really. The remainder must have |
c8a2f9ef |
796 | * headroom for at least two extra words. The scratch space |
f45a00c6 |
797 | * must be at least one word larger than the divisor. |
d03ab969 |
798 | */ |
799 | |
800 | void mpx_udiv(mpw *qv, mpw *qvl, mpw *rv, mpw *rvl, |
c8a2f9ef |
801 | const mpw *dv, const mpw *dvl, |
802 | mpw *sv, mpw *svl) |
d03ab969 |
803 | { |
d03ab969 |
804 | unsigned norm = 0; |
805 | size_t scale; |
806 | mpw d, dd; |
807 | |
808 | /* --- Initialize the quotient --- */ |
809 | |
810 | MPX_ZERO(qv, qvl); |
811 | |
c8a2f9ef |
812 | /* --- Perform some sanity checks --- */ |
813 | |
814 | MPX_SHRINK(dv, dvl); |
815 | assert(((void)"division by zero in mpx_udiv", dv < dvl)); |
816 | |
d03ab969 |
817 | /* --- Normalize the divisor --- * |
818 | * |
819 | * The algorithm requires that the divisor be at least two digits long. |
820 | * This is easy to fix. |
821 | */ |
822 | |
c8a2f9ef |
823 | { |
824 | unsigned b; |
d03ab969 |
825 | |
c8a2f9ef |
826 | d = dvl[-1]; |
827 | for (b = MPW_BITS / 2; b; b >>= 1) { |
828 | if (d < (MPW_MAX >> b)) { |
829 | d <<= b; |
830 | norm += b; |
831 | } |
832 | } |
833 | if (dv + 1 == dvl) |
834 | norm += MPW_BITS; |
d03ab969 |
835 | } |
d03ab969 |
836 | |
837 | /* --- Normalize the dividend/remainder to match --- */ |
838 | |
c8a2f9ef |
839 | if (norm) { |
c8a2f9ef |
840 | mpx_lsl(rv, rvl, rv, rvl, norm); |
f45a00c6 |
841 | mpx_lsl(sv, svl, dv, dvl, norm); |
c8a2f9ef |
842 | dv = sv; |
f45a00c6 |
843 | dvl = svl; |
c8a2f9ef |
844 | MPX_SHRINK(dv, dvl); |
845 | } |
846 | |
d03ab969 |
847 | MPX_SHRINK(rv, rvl); |
c8a2f9ef |
848 | d = dvl[-1]; |
849 | dd = dvl[-2]; |
d03ab969 |
850 | |
851 | /* --- Work out the relative scales --- */ |
852 | |
853 | { |
854 | size_t rvn = rvl - rv; |
c8a2f9ef |
855 | size_t dvn = dvl - dv; |
d03ab969 |
856 | |
857 | /* --- If the divisor is clearly larger, notice this --- */ |
858 | |
859 | if (dvn > rvn) { |
860 | mpx_lsr(rv, rvl, rv, rvl, norm); |
861 | return; |
862 | } |
863 | |
864 | scale = rvn - dvn; |
865 | } |
866 | |
867 | /* --- Calculate the most significant quotient digit --- * |
868 | * |
869 | * Because the divisor has its top bit set, this can only happen once. The |
870 | * pointer arithmetic is a little contorted, to make sure that the |
871 | * behaviour is defined. |
872 | */ |
873 | |
874 | if (MPX_UCMP(rv + scale, rvl, >=, dv, dvl)) { |
875 | mpx_usub(rv + scale, rvl, rv + scale, rvl, dv, dvl); |
876 | if (qvl - qv > scale) |
877 | qv[scale] = 1; |
878 | } |
879 | |
880 | /* --- Now for the main loop --- */ |
881 | |
882 | { |
c8a2f9ef |
883 | mpw *rvv = rvl - 2; |
d03ab969 |
884 | |
885 | while (scale) { |
c8a2f9ef |
886 | mpw q; |
887 | mpd rh; |
d03ab969 |
888 | |
889 | /* --- Get an estimate for the next quotient digit --- */ |
890 | |
c8a2f9ef |
891 | mpw r = rvv[1]; |
892 | mpw rr = rvv[0]; |
893 | mpw rrr = *--rvv; |
894 | |
895 | scale--; |
896 | rh = ((mpd)r << MPW_BITS) | rr; |
d03ab969 |
897 | if (r == d) |
898 | q = MPW_MAX; |
c8a2f9ef |
899 | else |
900 | q = MPW(rh / d); |
d03ab969 |
901 | |
902 | /* --- Refine the estimate --- */ |
903 | |
904 | { |
905 | mpd yh = (mpd)d * q; |
ce76ff16 |
906 | mpd yy = (mpd)dd * q; |
907 | mpw yl; |
c8a2f9ef |
908 | |
ce76ff16 |
909 | if (yy > MPW_MAX) |
910 | yh += yy >> MPW_BITS; |
911 | yl = MPW(yy); |
c8a2f9ef |
912 | |
913 | while (yh > rh || (yh == rh && yl > rrr)) { |
914 | q--; |
915 | yh -= d; |
ce76ff16 |
916 | if (yl < dd) |
917 | yh--; |
99b30c23 |
918 | yl = MPW(yl - dd); |
c8a2f9ef |
919 | } |
920 | } |
921 | |
922 | /* --- Remove a chunk from the dividend --- */ |
923 | |
924 | { |
925 | mpw *svv; |
926 | const mpw *dvv; |
f45a00c6 |
927 | mpw mc = 0, sc = 0; |
c8a2f9ef |
928 | |
f45a00c6 |
929 | /* --- Calculate the size of the chunk --- * |
930 | * |
931 | * This does the whole job of calculating @r >> scale - qd@. |
932 | */ |
c8a2f9ef |
933 | |
f45a00c6 |
934 | for (svv = rv + scale, dvv = dv; |
935 | dvv < dvl && svv < rvl; |
936 | svv++, dvv++) { |
937 | mpd x = (mpd)*dvv * (mpd)q + mc; |
938 | mc = x >> MPW_BITS; |
939 | x = (mpd)*svv - MPW(x) - sc; |
c8a2f9ef |
940 | *svv = MPW(x); |
f45a00c6 |
941 | if (x >> MPW_BITS) |
942 | sc = 1; |
943 | else |
944 | sc = 0; |
945 | } |
946 | |
947 | if (svv < rvl) { |
948 | mpd x = (mpd)*svv - mc - sc; |
949 | *svv++ = MPW(x); |
950 | if (x >> MPW_BITS) |
951 | sc = MPW_MAX; |
952 | else |
953 | sc = 0; |
954 | while (svv < rvl) |
955 | *svv++ = sc; |
c8a2f9ef |
956 | } |
c8a2f9ef |
957 | |
f45a00c6 |
958 | /* --- Fix if the quotient was too large --- * |
c8a2f9ef |
959 | * |
f45a00c6 |
960 | * This doesn't seem to happen very often. |
c8a2f9ef |
961 | */ |
962 | |
c8a2f9ef |
963 | if (rvl[-1] > MPW_MAX / 2) { |
964 | mpx_uadd(rv + scale, rvl, rv + scale, rvl, dv, dvl); |
965 | q--; |
966 | } |
967 | } |
968 | |
969 | /* --- Done for another iteration --- */ |
970 | |
971 | if (qvl - qv > scale) |
972 | qv[scale] = q; |
973 | r = rr; |
974 | rr = rrr; |
975 | } |
976 | } |
977 | |
978 | /* --- Now fiddle with unnormalizing and things --- */ |
979 | |
980 | mpx_lsr(rv, rvl, rv, rvl, norm); |
d03ab969 |
981 | } |
982 | |
698bd937 |
983 | /* --- @mpx_udivn@ --- * |
984 | * |
985 | * Arguments: @mpw *qv, *qvl@ = storage for the quotient (may overlap |
986 | * dividend) |
987 | * @const mpw *rv, *rvl@ = dividend |
988 | * @mpw d@ = single-precision divisor |
989 | * |
990 | * Returns: Remainder after divison. |
991 | * |
992 | * Use: Performs a single-precision division operation. |
993 | */ |
994 | |
995 | mpw mpx_udivn(mpw *qv, mpw *qvl, const mpw *rv, const mpw *rvl, mpw d) |
996 | { |
997 | size_t i; |
998 | size_t ql = qvl - qv; |
999 | mpd r = 0; |
1000 | |
1001 | i = rvl - rv; |
1002 | while (i > 0) { |
1003 | i--; |
1004 | r = (r << MPW_BITS) | rv[i]; |
1005 | if (i < ql) |
1006 | qv[i] = r / d; |
1007 | r %= d; |
1008 | } |
1009 | return (MPW(r)); |
1010 | } |
1011 | |
42684bdb |
1012 | /*----- Test rig ----------------------------------------------------------*/ |
1013 | |
1014 | #ifdef TEST_RIG |
1015 | |
1016 | #include <mLib/alloc.h> |
1017 | #include <mLib/dstr.h> |
1018 | #include <mLib/quis.h> |
1019 | #include <mLib/testrig.h> |
1020 | |
1021 | #include "mpscan.h" |
1022 | |
1023 | #define ALLOC(v, vl, sz) do { \ |
1024 | size_t _sz = (sz); \ |
1025 | mpw *_vv = xmalloc(MPWS(_sz)); \ |
1026 | mpw *_vvl = _vv + _sz; \ |
1027 | (v) = _vv; \ |
1028 | (vl) = _vvl; \ |
1029 | } while (0) |
1030 | |
1031 | #define LOAD(v, vl, d) do { \ |
1032 | const dstr *_d = (d); \ |
1033 | mpw *_v, *_vl; \ |
1034 | ALLOC(_v, _vl, MPW_RQ(_d->len)); \ |
1035 | mpx_loadb(_v, _vl, _d->buf, _d->len); \ |
1036 | (v) = _v; \ |
1037 | (vl) = _vl; \ |
1038 | } while (0) |
1039 | |
1040 | #define MAX(x, y) ((x) > (y) ? (x) : (y)) |
1041 | |
1042 | static void dumpbits(const char *msg, const void *pp, size_t sz) |
1043 | { |
1044 | const octet *p = pp; |
1045 | fputs(msg, stderr); |
1046 | for (; sz; sz--) |
1047 | fprintf(stderr, " %02x", *p++); |
1048 | fputc('\n', stderr); |
1049 | } |
1050 | |
1051 | static void dumpmp(const char *msg, const mpw *v, const mpw *vl) |
1052 | { |
1053 | fputs(msg, stderr); |
1054 | MPX_SHRINK(v, vl); |
1055 | while (v < vl) |
1056 | fprintf(stderr, " %08lx", (unsigned long)*--vl); |
1057 | fputc('\n', stderr); |
1058 | } |
1059 | |
1060 | static int chkscan(const mpw *v, const mpw *vl, |
1061 | const void *pp, size_t sz, int step) |
1062 | { |
1063 | mpscan mps; |
1064 | const octet *p = pp; |
1065 | unsigned bit = 0; |
1066 | int ok = 1; |
1067 | |
1068 | mpscan_initx(&mps, v, vl); |
1069 | while (sz) { |
1070 | unsigned x = *p; |
1071 | int i; |
1072 | p += step; |
1073 | for (i = 0; i < 8 && MPSCAN_STEP(&mps); i++) { |
1074 | if (MPSCAN_BIT(&mps) != (x & 1)) { |
1075 | fprintf(stderr, |
1076 | "\n*** error, step %i, bit %u, expected %u, found %u\n", |
1077 | step, bit, x & 1, MPSCAN_BIT(&mps)); |
1078 | ok = 0; |
1079 | } |
1080 | x >>= 1; |
1081 | bit++; |
1082 | } |
1083 | sz--; |
1084 | } |
1085 | |
1086 | return (ok); |
1087 | } |
1088 | |
1089 | static int loadstore(dstr *v) |
1090 | { |
1091 | dstr d = DSTR_INIT; |
1092 | size_t sz = MPW_RQ(v->len) * 2, diff; |
1093 | mpw *m, *ml; |
1094 | int ok = 1; |
1095 | |
1096 | dstr_ensure(&d, v->len); |
1097 | m = xmalloc(MPWS(sz)); |
1098 | |
1099 | for (diff = 0; diff < sz; diff += 5) { |
1100 | size_t oct; |
1101 | |
1102 | ml = m + sz - diff; |
1103 | |
1104 | mpx_loadl(m, ml, v->buf, v->len); |
1105 | if (!chkscan(m, ml, v->buf, v->len, +1)) |
1106 | ok = 0; |
1107 | MPX_OCTETS(oct, m, ml); |
1108 | mpx_storel(m, ml, d.buf, d.sz); |
1109 | if (memcmp(d.buf, v->buf, oct) != 0) { |
1110 | dumpbits("\n*** storel failed", d.buf, d.sz); |
1111 | ok = 0; |
1112 | } |
1113 | |
1114 | mpx_loadb(m, ml, v->buf, v->len); |
1115 | if (!chkscan(m, ml, v->buf + v->len - 1, v->len, -1)) |
1116 | ok = 0; |
1117 | MPX_OCTETS(oct, m, ml); |
1118 | mpx_storeb(m, ml, d.buf, d.sz); |
1119 | if (memcmp(d.buf + d.sz - oct, v->buf + v->len - oct, oct) != 0) { |
1120 | dumpbits("\n*** storeb failed", d.buf, d.sz); |
1121 | ok = 0; |
1122 | } |
1123 | } |
1124 | |
1125 | if (!ok) |
1126 | dumpbits("input data", v->buf, v->len); |
1127 | |
1128 | free(m); |
1129 | dstr_destroy(&d); |
1130 | return (ok); |
1131 | } |
1132 | |
1133 | static int lsl(dstr *v) |
1134 | { |
1135 | mpw *a, *al; |
1136 | int n = *(int *)v[1].buf; |
1137 | mpw *c, *cl; |
1138 | mpw *d, *dl; |
1139 | int ok = 1; |
1140 | |
1141 | LOAD(a, al, &v[0]); |
1142 | LOAD(c, cl, &v[2]); |
1143 | ALLOC(d, dl, al - a + (n + MPW_BITS - 1) / MPW_BITS); |
1144 | |
1145 | mpx_lsl(d, dl, a, al, n); |
1a05a8ef |
1146 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1147 | fprintf(stderr, "\n*** lsl(%i) failed\n", n); |
1148 | dumpmp(" a", a, al); |
1149 | dumpmp("expected", c, cl); |
1150 | dumpmp(" result", d, dl); |
1151 | ok = 0; |
1152 | } |
1153 | |
1154 | free(a); free(c); free(d); |
1155 | return (ok); |
1156 | } |
1157 | |
1158 | static int lsr(dstr *v) |
1159 | { |
1160 | mpw *a, *al; |
1161 | int n = *(int *)v[1].buf; |
1162 | mpw *c, *cl; |
1163 | mpw *d, *dl; |
1164 | int ok = 1; |
1165 | |
1166 | LOAD(a, al, &v[0]); |
1167 | LOAD(c, cl, &v[2]); |
1168 | ALLOC(d, dl, al - a + (n + MPW_BITS - 1) / MPW_BITS + 1); |
1169 | |
1170 | mpx_lsr(d, dl, a, al, n); |
1a05a8ef |
1171 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1172 | fprintf(stderr, "\n*** lsr(%i) failed\n", n); |
1173 | dumpmp(" a", a, al); |
1174 | dumpmp("expected", c, cl); |
1175 | dumpmp(" result", d, dl); |
1176 | ok = 0; |
1177 | } |
1178 | |
1179 | free(a); free(c); free(d); |
1180 | return (ok); |
1181 | } |
1182 | |
1183 | static int uadd(dstr *v) |
1184 | { |
1185 | mpw *a, *al; |
1186 | mpw *b, *bl; |
1187 | mpw *c, *cl; |
1188 | mpw *d, *dl; |
1189 | int ok = 1; |
1190 | |
1191 | LOAD(a, al, &v[0]); |
1192 | LOAD(b, bl, &v[1]); |
1193 | LOAD(c, cl, &v[2]); |
1194 | ALLOC(d, dl, MAX(al - a, bl - b) + 1); |
1195 | |
1196 | mpx_uadd(d, dl, a, al, b, bl); |
1a05a8ef |
1197 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1198 | fprintf(stderr, "\n*** uadd failed\n"); |
1199 | dumpmp(" a", a, al); |
1200 | dumpmp(" b", b, bl); |
1201 | dumpmp("expected", c, cl); |
1202 | dumpmp(" result", d, dl); |
1203 | ok = 0; |
1204 | } |
1205 | |
1206 | free(a); free(b); free(c); free(d); |
1207 | return (ok); |
1208 | } |
1209 | |
1210 | static int usub(dstr *v) |
1211 | { |
1212 | mpw *a, *al; |
1213 | mpw *b, *bl; |
1214 | mpw *c, *cl; |
1215 | mpw *d, *dl; |
1216 | int ok = 1; |
1217 | |
1218 | LOAD(a, al, &v[0]); |
1219 | LOAD(b, bl, &v[1]); |
1220 | LOAD(c, cl, &v[2]); |
1221 | ALLOC(d, dl, al - a); |
1222 | |
1223 | mpx_usub(d, dl, a, al, b, bl); |
1a05a8ef |
1224 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1225 | fprintf(stderr, "\n*** usub failed\n"); |
1226 | dumpmp(" a", a, al); |
1227 | dumpmp(" b", b, bl); |
1228 | dumpmp("expected", c, cl); |
1229 | dumpmp(" result", d, dl); |
1230 | ok = 0; |
1231 | } |
1232 | |
1233 | free(a); free(b); free(c); free(d); |
1234 | return (ok); |
1235 | } |
1236 | |
1237 | static int umul(dstr *v) |
1238 | { |
1239 | mpw *a, *al; |
1240 | mpw *b, *bl; |
1241 | mpw *c, *cl; |
1242 | mpw *d, *dl; |
1243 | int ok = 1; |
1244 | |
1245 | LOAD(a, al, &v[0]); |
1246 | LOAD(b, bl, &v[1]); |
1247 | LOAD(c, cl, &v[2]); |
1248 | ALLOC(d, dl, (al - a) + (bl - b)); |
1249 | |
1250 | mpx_umul(d, dl, a, al, b, bl); |
1a05a8ef |
1251 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1252 | fprintf(stderr, "\n*** umul failed\n"); |
1253 | dumpmp(" a", a, al); |
1254 | dumpmp(" b", b, bl); |
1255 | dumpmp("expected", c, cl); |
1256 | dumpmp(" result", d, dl); |
1257 | ok = 0; |
1258 | } |
1259 | |
1260 | free(a); free(b); free(c); free(d); |
1261 | return (ok); |
1262 | } |
1263 | |
1264 | static int usqr(dstr *v) |
1265 | { |
1266 | mpw *a, *al; |
1267 | mpw *c, *cl; |
1268 | mpw *d, *dl; |
1269 | int ok = 1; |
1270 | |
1271 | LOAD(a, al, &v[0]); |
1272 | LOAD(c, cl, &v[1]); |
1273 | ALLOC(d, dl, 2 * (al - a)); |
1274 | |
1275 | mpx_usqr(d, dl, a, al); |
1a05a8ef |
1276 | if (!mpx_ueq(d, dl, c, cl)) { |
42684bdb |
1277 | fprintf(stderr, "\n*** usqr failed\n"); |
1278 | dumpmp(" a", a, al); |
1279 | dumpmp("expected", c, cl); |
1280 | dumpmp(" result", d, dl); |
1281 | ok = 0; |
1282 | } |
1283 | |
1284 | free(a); free(c); free(d); |
1285 | return (ok); |
1286 | } |
1287 | |
1288 | static int udiv(dstr *v) |
1289 | { |
1290 | mpw *a, *al; |
1291 | mpw *b, *bl; |
1292 | mpw *q, *ql; |
1293 | mpw *r, *rl; |
1294 | mpw *qq, *qql; |
1295 | mpw *s, *sl; |
1296 | int ok = 1; |
1297 | |
1298 | ALLOC(a, al, MPW_RQ(v[0].len) + 2); mpx_loadb(a, al, v[0].buf, v[0].len); |
1299 | LOAD(b, bl, &v[1]); |
1300 | LOAD(q, ql, &v[2]); |
1301 | LOAD(r, rl, &v[3]); |
1302 | ALLOC(qq, qql, al - a); |
1303 | ALLOC(s, sl, (bl - b) + 1); |
1304 | |
1305 | mpx_udiv(qq, qql, a, al, b, bl, s, sl); |
1a05a8ef |
1306 | if (!mpx_ueq(qq, qql, q, ql) || |
1307 | !mpx_ueq(a, al, r, rl)) { |
42684bdb |
1308 | fprintf(stderr, "\n*** udiv failed\n"); |
1309 | dumpmp(" divisor", b, bl); |
1310 | dumpmp("expect r", r, rl); |
1311 | dumpmp("result r", a, al); |
1312 | dumpmp("expect q", q, ql); |
1313 | dumpmp("result q", qq, qql); |
1314 | ok = 0; |
1315 | } |
1316 | |
1317 | free(a); free(b); free(r); free(q); free(s); free(qq); |
1318 | return (ok); |
1319 | } |
1320 | |
1321 | static test_chunk defs[] = { |
1322 | { "load-store", loadstore, { &type_hex, 0 } }, |
1323 | { "lsl", lsl, { &type_hex, &type_int, &type_hex, 0 } }, |
1324 | { "lsr", lsr, { &type_hex, &type_int, &type_hex, 0 } }, |
1325 | { "uadd", uadd, { &type_hex, &type_hex, &type_hex, 0 } }, |
1326 | { "usub", usub, { &type_hex, &type_hex, &type_hex, 0 } }, |
1327 | { "umul", umul, { &type_hex, &type_hex, &type_hex, 0 } }, |
1328 | { "usqr", usqr, { &type_hex, &type_hex, 0 } }, |
1329 | { "udiv", udiv, { &type_hex, &type_hex, &type_hex, &type_hex, 0 } }, |
1330 | { 0, 0, { 0 } } |
1331 | }; |
1332 | |
1333 | int main(int argc, char *argv[]) |
1334 | { |
1335 | test_run(argc, argv, defs, SRCDIR"/tests/mpx"); |
1336 | return (0); |
1337 | } |
1338 | |
42684bdb |
1339 | #endif |
1340 | |
d03ab969 |
1341 | /*----- That's all, folks -------------------------------------------------*/ |