]> git.saurik.com Git - apple/security.git/blob - OSX/libsecurity_apple_csp/open_ssl/bn/bn_asm.c
Security-59754.41.1.tar.gz
[apple/security.git] / OSX / libsecurity_apple_csp / open_ssl / bn / bn_asm.c
1 /*
2 * Copyright (c) 2000-2001,2011,2014 Apple Inc. All Rights Reserved.
3 *
4 * The contents of this file constitute Original Code as defined in and are
5 * subject to the Apple Public Source License Version 1.2 (the 'License').
6 * You may not use this file except in compliance with the License. Please obtain
7 * a copy of the License at http://www.apple.com/publicsource and read it before
8 * using this file.
9 *
10 * This Original Code and all software distributed under the License are
11 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS
12 * OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, INCLUDING WITHOUT
13 * LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR
14 * PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. Please see the License for the
15 * specific language governing rights and limitations under the License.
16 */
17
18
19 /* crypto/bn/bn_asm.c */
20 /* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
21 * All rights reserved.
22 *
23 * This package is an SSL implementation written
24 * by Eric Young (eay@cryptsoft.com).
25 * The implementation was written so as to conform with Netscapes SSL.
26 *
27 * This library is free for commercial and non-commercial use as long as
28 * the following conditions are aheared to. The following conditions
29 * apply to all code found in this distribution, be it the RC4, RSA,
30 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
31 * included with this distribution is covered by the same copyright terms
32 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
33 *
34 * Copyright remains Eric Young's, and as such any Copyright notices in
35 * the code are not to be removed.
36 * If this package is used in a product, Eric Young should be given attribution
37 * as the author of the parts of the library used.
38 * This can be in the form of a textual message at program startup or
39 * in documentation (online or textual) provided with the package.
40 *
41 * Redistribution and use in source and binary forms, with or without
42 * modification, are permitted provided that the following conditions
43 * are met:
44 * 1. Redistributions of source code must retain the copyright
45 * notice, this list of conditions and the following disclaimer.
46 * 2. Redistributions in binary form must reproduce the above copyright
47 * notice, this list of conditions and the following disclaimer in the
48 * documentation and/or other materials provided with the distribution.
49 * 3. All advertising materials mentioning features or use of this software
50 * must display the following acknowledgement:
51 * "This product includes cryptographic software written by
52 * Eric Young (eay@cryptsoft.com)"
53 * The word 'cryptographic' can be left out if the rouines from the library
54 * being used are not cryptographic related :-).
55 * 4. If you include any Windows specific code (or a derivative thereof) from
56 * the apps directory (application code) you must include an acknowledgement:
57 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
58 *
59 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
60 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
63 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69 * SUCH DAMAGE.
70 *
71 * The licence and distribution terms for any publically available version or
72 * derivative of this code cannot be changed. i.e. this code cannot simply be
73 * copied and put under another distribution licence
74 * [including the GNU Public Licence.]
75 */
76
77 #ifndef BN_DEBUG
78 # undef NDEBUG /* avoid conflicting definitions */
79 # define NDEBUG
80 #endif
81
82 #include <stdio.h>
83 #include <security_utilities/simulatecrash_assert.h>
84 #include "cryptlib.h"
85 #include "bn_lcl.h"
86
87 #if defined(BN_LLONG) || defined(BN_UMULT_HIGH)
88
89 BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
90 {
91 BN_ULONG c1=0;
92
93 assert(num >= 0);
94 if (num <= 0) return(c1);
95
96 while (num&~3)
97 {
98 mul_add(rp[0],ap[0],w,c1);
99 mul_add(rp[1],ap[1],w,c1);
100 mul_add(rp[2],ap[2],w,c1);
101 mul_add(rp[3],ap[3],w,c1);
102 ap+=4; rp+=4; num-=4;
103 }
104 if (num)
105 {
106 mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1;
107 mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1;
108 mul_add(rp[2],ap[2],w,c1); return c1;
109 }
110
111 return(c1);
112 }
113
114 BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
115 {
116 BN_ULONG c1=0;
117
118 assert(num >= 0);
119 if (num <= 0) return(c1);
120
121 while (num&~3)
122 {
123 mul(rp[0],ap[0],w,c1);
124 mul(rp[1],ap[1],w,c1);
125 mul(rp[2],ap[2],w,c1);
126 mul(rp[3],ap[3],w,c1);
127 ap+=4; rp+=4; num-=4;
128 }
129 if (num)
130 {
131 mul(rp[0],ap[0],w,c1); if (--num == 0) return c1;
132 mul(rp[1],ap[1],w,c1); if (--num == 0) return c1;
133 mul(rp[2],ap[2],w,c1);
134 }
135 return(c1);
136 }
137
138 void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
139 {
140 assert(n >= 0);
141 if (n <= 0) return;
142 while (n&~3)
143 {
144 sqr(r[0],r[1],a[0]);
145 sqr(r[2],r[3],a[1]);
146 sqr(r[4],r[5],a[2]);
147 sqr(r[6],r[7],a[3]);
148 a+=4; r+=8; n-=4;
149 }
150 if (n)
151 {
152 sqr(r[0],r[1],a[0]); if (--n == 0) return;
153 sqr(r[2],r[3],a[1]); if (--n == 0) return;
154 sqr(r[4],r[5],a[2]);
155 }
156 }
157
158 #else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
159
160 BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
161 {
162 BN_ULONG c=0;
163 BN_ULONG bl,bh;
164
165 assert(num >= 0);
166 if (num <= 0) return((BN_ULONG)0);
167
168 bl=LBITS(w);
169 bh=HBITS(w);
170
171 for (;;)
172 {
173 mul_add(rp[0],ap[0],bl,bh,c);
174 if (--num == 0) break;
175 mul_add(rp[1],ap[1],bl,bh,c);
176 if (--num == 0) break;
177 mul_add(rp[2],ap[2],bl,bh,c);
178 if (--num == 0) break;
179 mul_add(rp[3],ap[3],bl,bh,c);
180 if (--num == 0) break;
181 ap+=4;
182 rp+=4;
183 }
184 return(c);
185 }
186
187 BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w)
188 {
189 BN_ULONG carry=0;
190 BN_ULONG bl,bh;
191
192 assert(num >= 0);
193 if (num <= 0) return((BN_ULONG)0);
194
195 bl=LBITS(w);
196 bh=HBITS(w);
197
198 for (;;)
199 {
200 mul(rp[0],ap[0],bl,bh,carry);
201 if (--num == 0) break;
202 mul(rp[1],ap[1],bl,bh,carry);
203 if (--num == 0) break;
204 mul(rp[2],ap[2],bl,bh,carry);
205 if (--num == 0) break;
206 mul(rp[3],ap[3],bl,bh,carry);
207 if (--num == 0) break;
208 ap+=4;
209 rp+=4;
210 }
211 return(carry);
212 }
213
214 void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n)
215 {
216 assert(n >= 0);
217 if (n <= 0) return;
218 for (;;)
219 {
220 sqr64(r[0],r[1],a[0]);
221 if (--n == 0) break;
222
223 sqr64(r[2],r[3],a[1]);
224 if (--n == 0) break;
225
226 sqr64(r[4],r[5],a[2]);
227 if (--n == 0) break;
228
229 sqr64(r[6],r[7],a[3]);
230 if (--n == 0) break;
231
232 a+=4;
233 r+=8;
234 }
235 }
236
237 #endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
238
239 #if defined(BN_LLONG) && defined(BN_DIV2W)
240
241 BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
242 {
243 return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d));
244 }
245
246 #else
247
248 /* Divide h-l by d and return the result. */
249 /* I need to test this some more :-( */
250 BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
251 {
252 BN_ULONG dh,dl,q,ret=0,th,tl,t;
253 int i,count=2;
254
255 if (d == 0) return(BN_MASK2);
256
257 i=BN_num_bits_word(d);
258 if ((i != BN_BITS2) && (h > (BN_ULONG)1<<i))
259 {
260 #if !defined(NO_STDIO) && !defined(WIN16)
261 fprintf(stderr,"Division would overflow (%d)\n",i);
262 #endif
263 abort();
264 }
265 i=BN_BITS2-i;
266 if (h >= d) h-=d;
267
268 if (i)
269 {
270 d<<=i;
271 h=(h<<i)|(l>>(BN_BITS2-i));
272 l<<=i;
273 }
274 dh=(d&BN_MASK2h)>>BN_BITS4;
275 dl=(d&BN_MASK2l);
276 for (;;)
277 {
278 if ((h>>BN_BITS4) == dh)
279 q=BN_MASK2l;
280 else
281 q=h/dh;
282
283 th=q*dh;
284 tl=dl*q;
285 for (;;)
286 {
287 t=h-th;
288 if ((t&BN_MASK2h) ||
289 ((tl) <= (
290 (t<<BN_BITS4)|
291 ((l&BN_MASK2h)>>BN_BITS4))))
292 break;
293 q--;
294 th-=dh;
295 tl-=dl;
296 }
297 t=(tl>>BN_BITS4);
298 tl=(tl<<BN_BITS4)&BN_MASK2h;
299 th+=t;
300
301 if (l < tl) th++;
302 l-=tl;
303 if (h < th)
304 {
305 h+=d;
306 q--;
307 }
308 h-=th;
309
310 if (--count == 0) break;
311
312 ret=q<<BN_BITS4;
313 h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2;
314 l=(l&BN_MASK2l)<<BN_BITS4;
315 }
316 ret|=q;
317 return(ret);
318 }
319 #endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */
320
321 #ifdef BN_LLONG
322 BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
323 {
324 BN_ULLONG ll=0;
325
326 assert(n >= 0);
327 if (n <= 0) return((BN_ULONG)0);
328
329 for (;;)
330 {
331 ll+=(BN_ULLONG)a[0]+b[0];
332 r[0]=(BN_ULONG)ll&BN_MASK2;
333 ll>>=BN_BITS2;
334 if (--n <= 0) break;
335
336 ll+=(BN_ULLONG)a[1]+b[1];
337 r[1]=(BN_ULONG)ll&BN_MASK2;
338 ll>>=BN_BITS2;
339 if (--n <= 0) break;
340
341 ll+=(BN_ULLONG)a[2]+b[2];
342 r[2]=(BN_ULONG)ll&BN_MASK2;
343 ll>>=BN_BITS2;
344 if (--n <= 0) break;
345
346 ll+=(BN_ULLONG)a[3]+b[3];
347 r[3]=(BN_ULONG)ll&BN_MASK2;
348 ll>>=BN_BITS2;
349 if (--n <= 0) break;
350
351 a+=4;
352 b+=4;
353 r+=4;
354 }
355 return((BN_ULONG)ll);
356 }
357 #else /* !BN_LLONG */
358 BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
359 {
360 BN_ULONG c,l,t;
361
362 assert(n >= 0);
363 if (n <= 0) return((BN_ULONG)0);
364
365 c=0;
366 for (;;)
367 {
368 t=a[0];
369 t=(t+c)&BN_MASK2;
370 c=(t < c);
371 l=(t+b[0])&BN_MASK2;
372 c+=(l < t);
373 r[0]=l;
374 if (--n <= 0) break;
375
376 t=a[1];
377 t=(t+c)&BN_MASK2;
378 c=(t < c);
379 l=(t+b[1])&BN_MASK2;
380 c+=(l < t);
381 r[1]=l;
382 if (--n <= 0) break;
383
384 t=a[2];
385 t=(t+c)&BN_MASK2;
386 c=(t < c);
387 l=(t+b[2])&BN_MASK2;
388 c+=(l < t);
389 r[2]=l;
390 if (--n <= 0) break;
391
392 t=a[3];
393 t=(t+c)&BN_MASK2;
394 c=(t < c);
395 l=(t+b[3])&BN_MASK2;
396 c+=(l < t);
397 r[3]=l;
398 if (--n <= 0) break;
399
400 a+=4;
401 b+=4;
402 r+=4;
403 }
404 return((BN_ULONG)c);
405 }
406 #endif /* !BN_LLONG */
407
408 BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
409 {
410 BN_ULONG t1,t2;
411 int c=0;
412
413 assert(n >= 0);
414 if (n <= 0) return((BN_ULONG)0);
415
416 for (;;)
417 {
418 t1=a[0]; t2=b[0];
419 r[0]=(t1-t2-c)&BN_MASK2;
420 if (t1 != t2) c=(t1 < t2);
421 if (--n <= 0) break;
422
423 t1=a[1]; t2=b[1];
424 r[1]=(t1-t2-c)&BN_MASK2;
425 if (t1 != t2) c=(t1 < t2);
426 if (--n <= 0) break;
427
428 t1=a[2]; t2=b[2];
429 r[2]=(t1-t2-c)&BN_MASK2;
430 if (t1 != t2) c=(t1 < t2);
431 if (--n <= 0) break;
432
433 t1=a[3]; t2=b[3];
434 r[3]=(t1-t2-c)&BN_MASK2;
435 if (t1 != t2) c=(t1 < t2);
436 if (--n <= 0) break;
437
438 a+=4;
439 b+=4;
440 r+=4;
441 }
442 return(c);
443 }
444
445 #ifdef BN_MUL_COMBA
446
447 #undef bn_mul_comba8
448 #undef bn_mul_comba4
449 #undef bn_sqr_comba8
450 #undef bn_sqr_comba4
451
452 /* mul_add_c(a,b,c0,c1,c2) -- c+=a*b for three word number c=(c2,c1,c0) */
453 /* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */
454 /* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
455 /* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */
456
457 #ifdef BN_LLONG
458 #define mul_add_c(a,b,c0,c1,c2) \
459 t=(BN_ULLONG)a*b; \
460 t1=(BN_ULONG)Lw(t); \
461 t2=(BN_ULONG)Hw(t); \
462 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
463 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
464
465 #define mul_add_c2(a,b,c0,c1,c2) \
466 t=(BN_ULLONG)a*b; \
467 tt=(t+t)&BN_MASK; \
468 if (tt < t) c2++; \
469 t1=(BN_ULONG)Lw(tt); \
470 t2=(BN_ULONG)Hw(tt); \
471 c0=(c0+t1)&BN_MASK2; \
472 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
473 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
474
475 #define sqr_add_c(a,i,c0,c1,c2) \
476 t=(BN_ULLONG)a[i]*a[i]; \
477 t1=(BN_ULONG)Lw(t); \
478 t2=(BN_ULONG)Hw(t); \
479 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
480 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
481
482 #define sqr_add_c2(a,i,j,c0,c1,c2) \
483 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
484
485 #elif defined(BN_UMULT_HIGH)
486
487 #define mul_add_c(a,b,c0,c1,c2) { \
488 BN_ULONG ta=(a),tb=(b); \
489 t1 = ta * tb; \
490 t2 = BN_UMULT_HIGH(ta,tb); \
491 c0 += t1; t2 += (c0<t1)?1:0; \
492 c1 += t2; c2 += (c1<t2)?1:0; \
493 }
494
495 #define mul_add_c2(a,b,c0,c1,c2) { \
496 BN_ULONG ta=(a),tb=(b),t0; \
497 t1 = BN_UMULT_HIGH(ta,tb); \
498 t0 = ta * tb; \
499 t2 = t1+t1; c2 += (t2<t1)?1:0; \
500 t1 = t0+t0; t2 += (t1<t0)?1:0; \
501 c0 += t1; t2 += (c0<t1)?1:0; \
502 c1 += t2; c2 += (c1<t2)?1:0; \
503 }
504
505 #define sqr_add_c(a,i,c0,c1,c2) { \
506 BN_ULONG ta=(a)[i]; \
507 t1 = ta * ta; \
508 t2 = BN_UMULT_HIGH(ta,ta); \
509 c0 += t1; t2 += (c0<t1)?1:0; \
510 c1 += t2; c2 += (c1<t2)?1:0; \
511 }
512
513 #define sqr_add_c2(a,i,j,c0,c1,c2) \
514 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
515
516 #else /* !BN_LLONG */
517 #define mul_add_c(a,b,c0,c1,c2) \
518 t1=LBITS(a); t2=HBITS(a); \
519 bl=LBITS(b); bh=HBITS(b); \
520 mul64(t1,t2,bl,bh); \
521 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
522 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
523
524 #define mul_add_c2(a,b,c0,c1,c2) \
525 t1=LBITS(a); t2=HBITS(a); \
526 bl=LBITS(b); bh=HBITS(b); \
527 mul64(t1,t2,bl,bh); \
528 if (t2 & BN_TBIT) c2++; \
529 t2=(t2+t2)&BN_MASK2; \
530 if (t1 & BN_TBIT) t2++; \
531 t1=(t1+t1)&BN_MASK2; \
532 c0=(c0+t1)&BN_MASK2; \
533 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
534 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
535
536 #define sqr_add_c(a,i,c0,c1,c2) \
537 sqr64(t1,t2,(a)[i]); \
538 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
539 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
540
541 #define sqr_add_c2(a,i,j,c0,c1,c2) \
542 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
543 #endif /* !BN_LLONG */
544
545 void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
546 {
547 #ifdef BN_LLONG
548 BN_ULLONG t;
549 #else
550 BN_ULONG bl,bh;
551 #endif
552 BN_ULONG t1,t2;
553 BN_ULONG c1,c2,c3;
554
555 c1=0;
556 c2=0;
557 c3=0;
558 mul_add_c(a[0],b[0],c1,c2,c3);
559 r[0]=c1;
560 c1=0;
561 mul_add_c(a[0],b[1],c2,c3,c1);
562 mul_add_c(a[1],b[0],c2,c3,c1);
563 r[1]=c2;
564 c2=0;
565 mul_add_c(a[2],b[0],c3,c1,c2);
566 mul_add_c(a[1],b[1],c3,c1,c2);
567 mul_add_c(a[0],b[2],c3,c1,c2);
568 r[2]=c3;
569 c3=0;
570 mul_add_c(a[0],b[3],c1,c2,c3);
571 mul_add_c(a[1],b[2],c1,c2,c3);
572 mul_add_c(a[2],b[1],c1,c2,c3);
573 mul_add_c(a[3],b[0],c1,c2,c3);
574 r[3]=c1;
575 c1=0;
576 mul_add_c(a[4],b[0],c2,c3,c1);
577 mul_add_c(a[3],b[1],c2,c3,c1);
578 mul_add_c(a[2],b[2],c2,c3,c1);
579 mul_add_c(a[1],b[3],c2,c3,c1);
580 mul_add_c(a[0],b[4],c2,c3,c1);
581 r[4]=c2;
582 c2=0;
583 mul_add_c(a[0],b[5],c3,c1,c2);
584 mul_add_c(a[1],b[4],c3,c1,c2);
585 mul_add_c(a[2],b[3],c3,c1,c2);
586 mul_add_c(a[3],b[2],c3,c1,c2);
587 mul_add_c(a[4],b[1],c3,c1,c2);
588 mul_add_c(a[5],b[0],c3,c1,c2);
589 r[5]=c3;
590 c3=0;
591 mul_add_c(a[6],b[0],c1,c2,c3);
592 mul_add_c(a[5],b[1],c1,c2,c3);
593 mul_add_c(a[4],b[2],c1,c2,c3);
594 mul_add_c(a[3],b[3],c1,c2,c3);
595 mul_add_c(a[2],b[4],c1,c2,c3);
596 mul_add_c(a[1],b[5],c1,c2,c3);
597 mul_add_c(a[0],b[6],c1,c2,c3);
598 r[6]=c1;
599 c1=0;
600 mul_add_c(a[0],b[7],c2,c3,c1);
601 mul_add_c(a[1],b[6],c2,c3,c1);
602 mul_add_c(a[2],b[5],c2,c3,c1);
603 mul_add_c(a[3],b[4],c2,c3,c1);
604 mul_add_c(a[4],b[3],c2,c3,c1);
605 mul_add_c(a[5],b[2],c2,c3,c1);
606 mul_add_c(a[6],b[1],c2,c3,c1);
607 mul_add_c(a[7],b[0],c2,c3,c1);
608 r[7]=c2;
609 c2=0;
610 mul_add_c(a[7],b[1],c3,c1,c2);
611 mul_add_c(a[6],b[2],c3,c1,c2);
612 mul_add_c(a[5],b[3],c3,c1,c2);
613 mul_add_c(a[4],b[4],c3,c1,c2);
614 mul_add_c(a[3],b[5],c3,c1,c2);
615 mul_add_c(a[2],b[6],c3,c1,c2);
616 mul_add_c(a[1],b[7],c3,c1,c2);
617 r[8]=c3;
618 c3=0;
619 mul_add_c(a[2],b[7],c1,c2,c3);
620 mul_add_c(a[3],b[6],c1,c2,c3);
621 mul_add_c(a[4],b[5],c1,c2,c3);
622 mul_add_c(a[5],b[4],c1,c2,c3);
623 mul_add_c(a[6],b[3],c1,c2,c3);
624 mul_add_c(a[7],b[2],c1,c2,c3);
625 r[9]=c1;
626 c1=0;
627 mul_add_c(a[7],b[3],c2,c3,c1);
628 mul_add_c(a[6],b[4],c2,c3,c1);
629 mul_add_c(a[5],b[5],c2,c3,c1);
630 mul_add_c(a[4],b[6],c2,c3,c1);
631 mul_add_c(a[3],b[7],c2,c3,c1);
632 r[10]=c2;
633 c2=0;
634 mul_add_c(a[4],b[7],c3,c1,c2);
635 mul_add_c(a[5],b[6],c3,c1,c2);
636 mul_add_c(a[6],b[5],c3,c1,c2);
637 mul_add_c(a[7],b[4],c3,c1,c2);
638 r[11]=c3;
639 c3=0;
640 mul_add_c(a[7],b[5],c1,c2,c3);
641 mul_add_c(a[6],b[6],c1,c2,c3);
642 mul_add_c(a[5],b[7],c1,c2,c3);
643 r[12]=c1;
644 c1=0;
645 mul_add_c(a[6],b[7],c2,c3,c1);
646 mul_add_c(a[7],b[6],c2,c3,c1);
647 r[13]=c2;
648 c2=0;
649 mul_add_c(a[7],b[7],c3,c1,c2);
650 r[14]=c3;
651 r[15]=c1;
652 }
653
654 void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
655 {
656 #ifdef BN_LLONG
657 BN_ULLONG t;
658 #else
659 BN_ULONG bl,bh;
660 #endif
661 BN_ULONG t1,t2;
662 BN_ULONG c1,c2,c3;
663
664 c1=0;
665 c2=0;
666 c3=0;
667 mul_add_c(a[0],b[0],c1,c2,c3);
668 r[0]=c1;
669 c1=0;
670 mul_add_c(a[0],b[1],c2,c3,c1);
671 mul_add_c(a[1],b[0],c2,c3,c1);
672 r[1]=c2;
673 c2=0;
674 mul_add_c(a[2],b[0],c3,c1,c2);
675 mul_add_c(a[1],b[1],c3,c1,c2);
676 mul_add_c(a[0],b[2],c3,c1,c2);
677 r[2]=c3;
678 c3=0;
679 mul_add_c(a[0],b[3],c1,c2,c3);
680 mul_add_c(a[1],b[2],c1,c2,c3);
681 mul_add_c(a[2],b[1],c1,c2,c3);
682 mul_add_c(a[3],b[0],c1,c2,c3);
683 r[3]=c1;
684 c1=0;
685 mul_add_c(a[3],b[1],c2,c3,c1);
686 mul_add_c(a[2],b[2],c2,c3,c1);
687 mul_add_c(a[1],b[3],c2,c3,c1);
688 r[4]=c2;
689 c2=0;
690 mul_add_c(a[2],b[3],c3,c1,c2);
691 mul_add_c(a[3],b[2],c3,c1,c2);
692 r[5]=c3;
693 c3=0;
694 mul_add_c(a[3],b[3],c1,c2,c3);
695 r[6]=c1;
696 r[7]=c2;
697 }
698
699 void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
700 {
701 #ifdef BN_LLONG
702 BN_ULLONG t,tt;
703 #else
704 BN_ULONG bl,bh;
705 #endif
706 BN_ULONG t1,t2;
707 BN_ULONG c1,c2,c3;
708
709 c1=0;
710 c2=0;
711 c3=0;
712 sqr_add_c(a,0,c1,c2,c3);
713 r[0]=c1;
714 c1=0;
715 sqr_add_c2(a,1,0,c2,c3,c1);
716 r[1]=c2;
717 c2=0;
718 sqr_add_c(a,1,c3,c1,c2);
719 sqr_add_c2(a,2,0,c3,c1,c2);
720 r[2]=c3;
721 c3=0;
722 sqr_add_c2(a,3,0,c1,c2,c3);
723 sqr_add_c2(a,2,1,c1,c2,c3);
724 r[3]=c1;
725 c1=0;
726 sqr_add_c(a,2,c2,c3,c1);
727 sqr_add_c2(a,3,1,c2,c3,c1);
728 sqr_add_c2(a,4,0,c2,c3,c1);
729 r[4]=c2;
730 c2=0;
731 sqr_add_c2(a,5,0,c3,c1,c2);
732 sqr_add_c2(a,4,1,c3,c1,c2);
733 sqr_add_c2(a,3,2,c3,c1,c2);
734 r[5]=c3;
735 c3=0;
736 sqr_add_c(a,3,c1,c2,c3);
737 sqr_add_c2(a,4,2,c1,c2,c3);
738 sqr_add_c2(a,5,1,c1,c2,c3);
739 sqr_add_c2(a,6,0,c1,c2,c3);
740 r[6]=c1;
741 c1=0;
742 sqr_add_c2(a,7,0,c2,c3,c1);
743 sqr_add_c2(a,6,1,c2,c3,c1);
744 sqr_add_c2(a,5,2,c2,c3,c1);
745 sqr_add_c2(a,4,3,c2,c3,c1);
746 r[7]=c2;
747 c2=0;
748 sqr_add_c(a,4,c3,c1,c2);
749 sqr_add_c2(a,5,3,c3,c1,c2);
750 sqr_add_c2(a,6,2,c3,c1,c2);
751 sqr_add_c2(a,7,1,c3,c1,c2);
752 r[8]=c3;
753 c3=0;
754 sqr_add_c2(a,7,2,c1,c2,c3);
755 sqr_add_c2(a,6,3,c1,c2,c3);
756 sqr_add_c2(a,5,4,c1,c2,c3);
757 r[9]=c1;
758 c1=0;
759 sqr_add_c(a,5,c2,c3,c1);
760 sqr_add_c2(a,6,4,c2,c3,c1);
761 sqr_add_c2(a,7,3,c2,c3,c1);
762 r[10]=c2;
763 c2=0;
764 sqr_add_c2(a,7,4,c3,c1,c2);
765 sqr_add_c2(a,6,5,c3,c1,c2);
766 r[11]=c3;
767 c3=0;
768 sqr_add_c(a,6,c1,c2,c3);
769 sqr_add_c2(a,7,5,c1,c2,c3);
770 r[12]=c1;
771 c1=0;
772 sqr_add_c2(a,7,6,c2,c3,c1);
773 r[13]=c2;
774 c2=0;
775 sqr_add_c(a,7,c3,c1,c2);
776 r[14]=c3;
777 r[15]=c1;
778 }
779
780 void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
781 {
782 #ifdef BN_LLONG
783 BN_ULLONG t,tt;
784 #else
785 BN_ULONG bl,bh;
786 #endif
787 BN_ULONG t1,t2;
788 BN_ULONG c1,c2,c3;
789
790 c1=0;
791 c2=0;
792 c3=0;
793 sqr_add_c(a,0,c1,c2,c3);
794 r[0]=c1;
795 c1=0;
796 sqr_add_c2(a,1,0,c2,c3,c1);
797 r[1]=c2;
798 c2=0;
799 sqr_add_c(a,1,c3,c1,c2);
800 sqr_add_c2(a,2,0,c3,c1,c2);
801 r[2]=c3;
802 c3=0;
803 sqr_add_c2(a,3,0,c1,c2,c3);
804 sqr_add_c2(a,2,1,c1,c2,c3);
805 r[3]=c1;
806 c1=0;
807 sqr_add_c(a,2,c2,c3,c1);
808 sqr_add_c2(a,3,1,c2,c3,c1);
809 r[4]=c2;
810 c2=0;
811 sqr_add_c2(a,3,2,c3,c1,c2);
812 r[5]=c3;
813 c3=0;
814 sqr_add_c(a,3,c1,c2,c3);
815 r[6]=c1;
816 r[7]=c2;
817 }
818 #else /* !BN_MUL_COMBA */
819
820 /* hmm... is it faster just to do a multiply? */
821 #undef bn_sqr_comba4
822 void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
823 {
824 BN_ULONG t[8];
825 bn_sqr_normal(r,a,4,t);
826 }
827
828 #undef bn_sqr_comba8
829 void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
830 {
831 BN_ULONG t[16];
832 bn_sqr_normal(r,a,8,t);
833 }
834
835 void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
836 {
837 r[4]=bn_mul_words( &(r[0]),a,4,b[0]);
838 r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]);
839 r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]);
840 r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]);
841 }
842
843 void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
844 {
845 r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]);
846 r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]);
847 r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]);
848 r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]);
849 r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]);
850 r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]);
851 r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]);
852 r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]);
853 }
854
855 #endif /* !BN_MUL_COMBA */