bn_asm.c revision 59191
1/* crypto/bn/bn_asm.c */ 2/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com) 3 * All rights reserved. 4 * 5 * This package is an SSL implementation written 6 * by Eric Young (eay@cryptsoft.com). 7 * The implementation was written so as to conform with Netscapes SSL. 8 * 9 * This library is free for commercial and non-commercial use as long as 10 * the following conditions are aheared to. The following conditions 11 * apply to all code found in this distribution, be it the RC4, RSA, 12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation 13 * included with this distribution is covered by the same copyright terms 14 * except that the holder is Tim Hudson (tjh@cryptsoft.com). 15 * 16 * Copyright remains Eric Young's, and as such any Copyright notices in 17 * the code are not to be removed. 18 * If this package is used in a product, Eric Young should be given attribution 19 * as the author of the parts of the library used. 20 * This can be in the form of a textual message at program startup or 21 * in documentation (online or textual) provided with the package. 22 * 23 * Redistribution and use in source and binary forms, with or without 24 * modification, are permitted provided that the following conditions 25 * are met: 26 * 1. Redistributions of source code must retain the copyright 27 * notice, this list of conditions and the following disclaimer. 28 * 2. Redistributions in binary form must reproduce the above copyright 29 * notice, this list of conditions and the following disclaimer in the 30 * documentation and/or other materials provided with the distribution. 31 * 3. All advertising materials mentioning features or use of this software 32 * must display the following acknowledgement: 33 * "This product includes cryptographic software written by 34 * Eric Young (eay@cryptsoft.com)" 35 * The word 'cryptographic' can be left out if the rouines from the library 36 * being used are not cryptographic related :-). 37 * 4. If you include any Windows specific code (or a derivative thereof) from 38 * the apps directory (application code) you must include an acknowledgement: 39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" 40 * 41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND 42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 51 * SUCH DAMAGE. 52 * 53 * The licence and distribution terms for any publically available version or 54 * derivative of this code cannot be changed. i.e. this code cannot simply be 55 * copied and put under another distribution licence 56 * [including the GNU Public Licence.] 57 */ 58 59#ifndef BN_DEBUG 60# undef NDEBUG /* avoid conflicting definitions */ 61# define NDEBUG 62#endif 63 64#include <stdio.h> 65#include <assert.h> 66#include "cryptlib.h" 67#include "bn_lcl.h" 68 69#if defined(BN_LLONG) || defined(BN_UMULT_HIGH) 70 71BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) 72 { 73 BN_ULONG c1=0; 74 75 assert(num >= 0); 76 if (num <= 0) return(c1); 77 78 while (num&~3) 79 { 80 mul_add(rp[0],ap[0],w,c1); 81 mul_add(rp[1],ap[1],w,c1); 82 mul_add(rp[2],ap[2],w,c1); 83 mul_add(rp[3],ap[3],w,c1); 84 ap+=4; rp+=4; num-=4; 85 } 86 if (num) 87 { 88 mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1; 89 mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1; 90 mul_add(rp[2],ap[2],w,c1); return c1; 91 } 92 93 return(c1); 94 } 95 96BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) 97 { 98 BN_ULONG c1=0; 99 100 assert(num >= 0); 101 if (num <= 0) return(c1); 102 103 while (num&~3) 104 { 105 mul(rp[0],ap[0],w,c1); 106 mul(rp[1],ap[1],w,c1); 107 mul(rp[2],ap[2],w,c1); 108 mul(rp[3],ap[3],w,c1); 109 ap+=4; rp+=4; num-=4; 110 } 111 if (num) 112 { 113 mul(rp[0],ap[0],w,c1); if (--num == 0) return c1; 114 mul(rp[1],ap[1],w,c1); if (--num == 0) return c1; 115 mul(rp[2],ap[2],w,c1); 116 } 117 return(c1); 118 } 119 120void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n) 121 { 122 assert(n >= 0); 123 if (n <= 0) return; 124 while (n&~3) 125 { 126 sqr(r[0],r[1],a[0]); 127 sqr(r[2],r[3],a[1]); 128 sqr(r[4],r[5],a[2]); 129 sqr(r[6],r[7],a[3]); 130 a+=4; r+=8; n-=4; 131 } 132 if (n) 133 { 134 sqr(r[0],r[1],a[0]); if (--n == 0) return; 135 sqr(r[2],r[3],a[1]); if (--n == 0) return; 136 sqr(r[4],r[5],a[2]); 137 } 138 } 139 140#else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ 141 142BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) 143 { 144 BN_ULONG c=0; 145 BN_ULONG bl,bh; 146 147 assert(num >= 0); 148 if (num <= 0) return((BN_ULONG)0); 149 150 bl=LBITS(w); 151 bh=HBITS(w); 152 153 for (;;) 154 { 155 mul_add(rp[0],ap[0],bl,bh,c); 156 if (--num == 0) break; 157 mul_add(rp[1],ap[1],bl,bh,c); 158 if (--num == 0) break; 159 mul_add(rp[2],ap[2],bl,bh,c); 160 if (--num == 0) break; 161 mul_add(rp[3],ap[3],bl,bh,c); 162 if (--num == 0) break; 163 ap+=4; 164 rp+=4; 165 } 166 return(c); 167 } 168 169BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) 170 { 171 BN_ULONG carry=0; 172 BN_ULONG bl,bh; 173 174 assert(num >= 0); 175 if (num <= 0) return((BN_ULONG)0); 176 177 bl=LBITS(w); 178 bh=HBITS(w); 179 180 for (;;) 181 { 182 mul(rp[0],ap[0],bl,bh,carry); 183 if (--num == 0) break; 184 mul(rp[1],ap[1],bl,bh,carry); 185 if (--num == 0) break; 186 mul(rp[2],ap[2],bl,bh,carry); 187 if (--num == 0) break; 188 mul(rp[3],ap[3],bl,bh,carry); 189 if (--num == 0) break; 190 ap+=4; 191 rp+=4; 192 } 193 return(carry); 194 } 195 196void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n) 197 { 198 assert(n >= 0); 199 if (n <= 0) return; 200 for (;;) 201 { 202 sqr64(r[0],r[1],a[0]); 203 if (--n == 0) break; 204 205 sqr64(r[2],r[3],a[1]); 206 if (--n == 0) break; 207 208 sqr64(r[4],r[5],a[2]); 209 if (--n == 0) break; 210 211 sqr64(r[6],r[7],a[3]); 212 if (--n == 0) break; 213 214 a+=4; 215 r+=8; 216 } 217 } 218 219#endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ 220 221#if defined(BN_LLONG) && defined(BN_DIV2W) 222 223BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) 224 { 225 return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d)); 226 } 227 228#else 229 230/* Divide h-l by d and return the result. */ 231/* I need to test this some more :-( */ 232BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) 233 { 234 BN_ULONG dh,dl,q,ret=0,th,tl,t; 235 int i,count=2; 236 237 if (d == 0) return(BN_MASK2); 238 239 i=BN_num_bits_word(d); 240 if ((i != BN_BITS2) && (h > (BN_ULONG)1<<i)) 241 { 242#if !defined(NO_STDIO) && !defined(WIN16) 243 fprintf(stderr,"Division would overflow (%d)\n",i); 244#endif 245 abort(); 246 } 247 i=BN_BITS2-i; 248 if (h >= d) h-=d; 249 250 if (i) 251 { 252 d<<=i; 253 h=(h<<i)|(l>>(BN_BITS2-i)); 254 l<<=i; 255 } 256 dh=(d&BN_MASK2h)>>BN_BITS4; 257 dl=(d&BN_MASK2l); 258 for (;;) 259 { 260 if ((h>>BN_BITS4) == dh) 261 q=BN_MASK2l; 262 else 263 q=h/dh; 264 265 th=q*dh; 266 tl=dl*q; 267 for (;;) 268 { 269 t=h-th; 270 if ((t&BN_MASK2h) || 271 ((tl) <= ( 272 (t<<BN_BITS4)| 273 ((l&BN_MASK2h)>>BN_BITS4)))) 274 break; 275 q--; 276 th-=dh; 277 tl-=dl; 278 } 279 t=(tl>>BN_BITS4); 280 tl=(tl<<BN_BITS4)&BN_MASK2h; 281 th+=t; 282 283 if (l < tl) th++; 284 l-=tl; 285 if (h < th) 286 { 287 h+=d; 288 q--; 289 } 290 h-=th; 291 292 if (--count == 0) break; 293 294 ret=q<<BN_BITS4; 295 h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2; 296 l=(l&BN_MASK2l)<<BN_BITS4; 297 } 298 ret|=q; 299 return(ret); 300 } 301#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */ 302 303#ifdef BN_LLONG 304BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) 305 { 306 BN_ULLONG ll=0; 307 308 assert(n >= 0); 309 if (n <= 0) return((BN_ULONG)0); 310 311 for (;;) 312 { 313 ll+=(BN_ULLONG)a[0]+b[0]; 314 r[0]=(BN_ULONG)ll&BN_MASK2; 315 ll>>=BN_BITS2; 316 if (--n <= 0) break; 317 318 ll+=(BN_ULLONG)a[1]+b[1]; 319 r[1]=(BN_ULONG)ll&BN_MASK2; 320 ll>>=BN_BITS2; 321 if (--n <= 0) break; 322 323 ll+=(BN_ULLONG)a[2]+b[2]; 324 r[2]=(BN_ULONG)ll&BN_MASK2; 325 ll>>=BN_BITS2; 326 if (--n <= 0) break; 327 328 ll+=(BN_ULLONG)a[3]+b[3]; 329 r[3]=(BN_ULONG)ll&BN_MASK2; 330 ll>>=BN_BITS2; 331 if (--n <= 0) break; 332 333 a+=4; 334 b+=4; 335 r+=4; 336 } 337 return((BN_ULONG)ll); 338 } 339#else /* !BN_LLONG */ 340BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) 341 { 342 BN_ULONG c,l,t; 343 344 assert(n >= 0); 345 if (n <= 0) return((BN_ULONG)0); 346 347 c=0; 348 for (;;) 349 { 350 t=a[0]; 351 t=(t+c)&BN_MASK2; 352 c=(t < c); 353 l=(t+b[0])&BN_MASK2; 354 c+=(l < t); 355 r[0]=l; 356 if (--n <= 0) break; 357 358 t=a[1]; 359 t=(t+c)&BN_MASK2; 360 c=(t < c); 361 l=(t+b[1])&BN_MASK2; 362 c+=(l < t); 363 r[1]=l; 364 if (--n <= 0) break; 365 366 t=a[2]; 367 t=(t+c)&BN_MASK2; 368 c=(t < c); 369 l=(t+b[2])&BN_MASK2; 370 c+=(l < t); 371 r[2]=l; 372 if (--n <= 0) break; 373 374 t=a[3]; 375 t=(t+c)&BN_MASK2; 376 c=(t < c); 377 l=(t+b[3])&BN_MASK2; 378 c+=(l < t); 379 r[3]=l; 380 if (--n <= 0) break; 381 382 a+=4; 383 b+=4; 384 r+=4; 385 } 386 return((BN_ULONG)c); 387 } 388#endif /* !BN_LLONG */ 389 390BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) 391 { 392 BN_ULONG t1,t2; 393 int c=0; 394 395 assert(n >= 0); 396 if (n <= 0) return((BN_ULONG)0); 397 398 for (;;) 399 { 400 t1=a[0]; t2=b[0]; 401 r[0]=(t1-t2-c)&BN_MASK2; 402 if (t1 != t2) c=(t1 < t2); 403 if (--n <= 0) break; 404 405 t1=a[1]; t2=b[1]; 406 r[1]=(t1-t2-c)&BN_MASK2; 407 if (t1 != t2) c=(t1 < t2); 408 if (--n <= 0) break; 409 410 t1=a[2]; t2=b[2]; 411 r[2]=(t1-t2-c)&BN_MASK2; 412 if (t1 != t2) c=(t1 < t2); 413 if (--n <= 0) break; 414 415 t1=a[3]; t2=b[3]; 416 r[3]=(t1-t2-c)&BN_MASK2; 417 if (t1 != t2) c=(t1 < t2); 418 if (--n <= 0) break; 419 420 a+=4; 421 b+=4; 422 r+=4; 423 } 424 return(c); 425 } 426 427#ifdef BN_MUL_COMBA 428 429#undef bn_mul_comba8 430#undef bn_mul_comba4 431#undef bn_sqr_comba8 432#undef bn_sqr_comba4 433 434/* mul_add_c(a,b,c0,c1,c2) -- c+=a*b for three word number c=(c2,c1,c0) */ 435/* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */ 436/* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */ 437/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */ 438 439#ifdef BN_LLONG 440#define mul_add_c(a,b,c0,c1,c2) \ 441 t=(BN_ULLONG)a*b; \ 442 t1=(BN_ULONG)Lw(t); \ 443 t2=(BN_ULONG)Hw(t); \ 444 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ 445 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 446 447#define mul_add_c2(a,b,c0,c1,c2) \ 448 t=(BN_ULLONG)a*b; \ 449 tt=(t+t)&BN_MASK; \ 450 if (tt < t) c2++; \ 451 t1=(BN_ULONG)Lw(tt); \ 452 t2=(BN_ULONG)Hw(tt); \ 453 c0=(c0+t1)&BN_MASK2; \ 454 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \ 455 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 456 457#define sqr_add_c(a,i,c0,c1,c2) \ 458 t=(BN_ULLONG)a[i]*a[i]; \ 459 t1=(BN_ULONG)Lw(t); \ 460 t2=(BN_ULONG)Hw(t); \ 461 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ 462 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 463 464#define sqr_add_c2(a,i,j,c0,c1,c2) \ 465 mul_add_c2((a)[i],(a)[j],c0,c1,c2) 466 467#elif defined(BN_UMULT_HIGH) 468 469#define mul_add_c(a,b,c0,c1,c2) { \ 470 BN_ULONG ta=(a),tb=(b); \ 471 t1 = ta * tb; \ 472 t2 = BN_UMULT_HIGH(ta,tb); \ 473 c0 += t1; t2 += (c0<t1)?1:0; \ 474 c1 += t2; c2 += (c1<t2)?1:0; \ 475 } 476 477#define mul_add_c2(a,b,c0,c1,c2) { \ 478 BN_ULONG ta=(a),tb=(b),t0; \ 479 t1 = BN_UMULT_HIGH(ta,tb); \ 480 t0 = ta * tb; \ 481 t2 = t1+t1; c2 += (t2<t1)?1:0; \ 482 t1 = t0+t0; t2 += (t1<t0)?1:0; \ 483 c0 += t1; t2 += (c0<t1)?1:0; \ 484 c1 += t2; c2 += (c1<t2)?1:0; \ 485 } 486 487#define sqr_add_c(a,i,c0,c1,c2) { \ 488 BN_ULONG ta=(a)[i]; \ 489 t1 = ta * ta; \ 490 t2 = BN_UMULT_HIGH(ta,ta); \ 491 c0 += t1; t2 += (c0<t1)?1:0; \ 492 c1 += t2; c2 += (c1<t2)?1:0; \ 493 } 494 495#define sqr_add_c2(a,i,j,c0,c1,c2) \ 496 mul_add_c2((a)[i],(a)[j],c0,c1,c2) 497 498#else /* !BN_LLONG */ 499#define mul_add_c(a,b,c0,c1,c2) \ 500 t1=LBITS(a); t2=HBITS(a); \ 501 bl=LBITS(b); bh=HBITS(b); \ 502 mul64(t1,t2,bl,bh); \ 503 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ 504 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 505 506#define mul_add_c2(a,b,c0,c1,c2) \ 507 t1=LBITS(a); t2=HBITS(a); \ 508 bl=LBITS(b); bh=HBITS(b); \ 509 mul64(t1,t2,bl,bh); \ 510 if (t2 & BN_TBIT) c2++; \ 511 t2=(t2+t2)&BN_MASK2; \ 512 if (t1 & BN_TBIT) t2++; \ 513 t1=(t1+t1)&BN_MASK2; \ 514 c0=(c0+t1)&BN_MASK2; \ 515 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \ 516 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 517 518#define sqr_add_c(a,i,c0,c1,c2) \ 519 sqr64(t1,t2,(a)[i]); \ 520 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ 521 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; 522 523#define sqr_add_c2(a,i,j,c0,c1,c2) \ 524 mul_add_c2((a)[i],(a)[j],c0,c1,c2) 525#endif /* !BN_LLONG */ 526 527void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 528 { 529#ifdef BN_LLONG 530 BN_ULLONG t; 531#else 532 BN_ULONG bl,bh; 533#endif 534 BN_ULONG t1,t2; 535 BN_ULONG c1,c2,c3; 536 537 c1=0; 538 c2=0; 539 c3=0; 540 mul_add_c(a[0],b[0],c1,c2,c3); 541 r[0]=c1; 542 c1=0; 543 mul_add_c(a[0],b[1],c2,c3,c1); 544 mul_add_c(a[1],b[0],c2,c3,c1); 545 r[1]=c2; 546 c2=0; 547 mul_add_c(a[2],b[0],c3,c1,c2); 548 mul_add_c(a[1],b[1],c3,c1,c2); 549 mul_add_c(a[0],b[2],c3,c1,c2); 550 r[2]=c3; 551 c3=0; 552 mul_add_c(a[0],b[3],c1,c2,c3); 553 mul_add_c(a[1],b[2],c1,c2,c3); 554 mul_add_c(a[2],b[1],c1,c2,c3); 555 mul_add_c(a[3],b[0],c1,c2,c3); 556 r[3]=c1; 557 c1=0; 558 mul_add_c(a[4],b[0],c2,c3,c1); 559 mul_add_c(a[3],b[1],c2,c3,c1); 560 mul_add_c(a[2],b[2],c2,c3,c1); 561 mul_add_c(a[1],b[3],c2,c3,c1); 562 mul_add_c(a[0],b[4],c2,c3,c1); 563 r[4]=c2; 564 c2=0; 565 mul_add_c(a[0],b[5],c3,c1,c2); 566 mul_add_c(a[1],b[4],c3,c1,c2); 567 mul_add_c(a[2],b[3],c3,c1,c2); 568 mul_add_c(a[3],b[2],c3,c1,c2); 569 mul_add_c(a[4],b[1],c3,c1,c2); 570 mul_add_c(a[5],b[0],c3,c1,c2); 571 r[5]=c3; 572 c3=0; 573 mul_add_c(a[6],b[0],c1,c2,c3); 574 mul_add_c(a[5],b[1],c1,c2,c3); 575 mul_add_c(a[4],b[2],c1,c2,c3); 576 mul_add_c(a[3],b[3],c1,c2,c3); 577 mul_add_c(a[2],b[4],c1,c2,c3); 578 mul_add_c(a[1],b[5],c1,c2,c3); 579 mul_add_c(a[0],b[6],c1,c2,c3); 580 r[6]=c1; 581 c1=0; 582 mul_add_c(a[0],b[7],c2,c3,c1); 583 mul_add_c(a[1],b[6],c2,c3,c1); 584 mul_add_c(a[2],b[5],c2,c3,c1); 585 mul_add_c(a[3],b[4],c2,c3,c1); 586 mul_add_c(a[4],b[3],c2,c3,c1); 587 mul_add_c(a[5],b[2],c2,c3,c1); 588 mul_add_c(a[6],b[1],c2,c3,c1); 589 mul_add_c(a[7],b[0],c2,c3,c1); 590 r[7]=c2; 591 c2=0; 592 mul_add_c(a[7],b[1],c3,c1,c2); 593 mul_add_c(a[6],b[2],c3,c1,c2); 594 mul_add_c(a[5],b[3],c3,c1,c2); 595 mul_add_c(a[4],b[4],c3,c1,c2); 596 mul_add_c(a[3],b[5],c3,c1,c2); 597 mul_add_c(a[2],b[6],c3,c1,c2); 598 mul_add_c(a[1],b[7],c3,c1,c2); 599 r[8]=c3; 600 c3=0; 601 mul_add_c(a[2],b[7],c1,c2,c3); 602 mul_add_c(a[3],b[6],c1,c2,c3); 603 mul_add_c(a[4],b[5],c1,c2,c3); 604 mul_add_c(a[5],b[4],c1,c2,c3); 605 mul_add_c(a[6],b[3],c1,c2,c3); 606 mul_add_c(a[7],b[2],c1,c2,c3); 607 r[9]=c1; 608 c1=0; 609 mul_add_c(a[7],b[3],c2,c3,c1); 610 mul_add_c(a[6],b[4],c2,c3,c1); 611 mul_add_c(a[5],b[5],c2,c3,c1); 612 mul_add_c(a[4],b[6],c2,c3,c1); 613 mul_add_c(a[3],b[7],c2,c3,c1); 614 r[10]=c2; 615 c2=0; 616 mul_add_c(a[4],b[7],c3,c1,c2); 617 mul_add_c(a[5],b[6],c3,c1,c2); 618 mul_add_c(a[6],b[5],c3,c1,c2); 619 mul_add_c(a[7],b[4],c3,c1,c2); 620 r[11]=c3; 621 c3=0; 622 mul_add_c(a[7],b[5],c1,c2,c3); 623 mul_add_c(a[6],b[6],c1,c2,c3); 624 mul_add_c(a[5],b[7],c1,c2,c3); 625 r[12]=c1; 626 c1=0; 627 mul_add_c(a[6],b[7],c2,c3,c1); 628 mul_add_c(a[7],b[6],c2,c3,c1); 629 r[13]=c2; 630 c2=0; 631 mul_add_c(a[7],b[7],c3,c1,c2); 632 r[14]=c3; 633 r[15]=c1; 634 } 635 636void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 637 { 638#ifdef BN_LLONG 639 BN_ULLONG t; 640#else 641 BN_ULONG bl,bh; 642#endif 643 BN_ULONG t1,t2; 644 BN_ULONG c1,c2,c3; 645 646 c1=0; 647 c2=0; 648 c3=0; 649 mul_add_c(a[0],b[0],c1,c2,c3); 650 r[0]=c1; 651 c1=0; 652 mul_add_c(a[0],b[1],c2,c3,c1); 653 mul_add_c(a[1],b[0],c2,c3,c1); 654 r[1]=c2; 655 c2=0; 656 mul_add_c(a[2],b[0],c3,c1,c2); 657 mul_add_c(a[1],b[1],c3,c1,c2); 658 mul_add_c(a[0],b[2],c3,c1,c2); 659 r[2]=c3; 660 c3=0; 661 mul_add_c(a[0],b[3],c1,c2,c3); 662 mul_add_c(a[1],b[2],c1,c2,c3); 663 mul_add_c(a[2],b[1],c1,c2,c3); 664 mul_add_c(a[3],b[0],c1,c2,c3); 665 r[3]=c1; 666 c1=0; 667 mul_add_c(a[3],b[1],c2,c3,c1); 668 mul_add_c(a[2],b[2],c2,c3,c1); 669 mul_add_c(a[1],b[3],c2,c3,c1); 670 r[4]=c2; 671 c2=0; 672 mul_add_c(a[2],b[3],c3,c1,c2); 673 mul_add_c(a[3],b[2],c3,c1,c2); 674 r[5]=c3; 675 c3=0; 676 mul_add_c(a[3],b[3],c1,c2,c3); 677 r[6]=c1; 678 r[7]=c2; 679 } 680 681void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a) 682 { 683#ifdef BN_LLONG 684 BN_ULLONG t,tt; 685#else 686 BN_ULONG bl,bh; 687#endif 688 BN_ULONG t1,t2; 689 BN_ULONG c1,c2,c3; 690 691 c1=0; 692 c2=0; 693 c3=0; 694 sqr_add_c(a,0,c1,c2,c3); 695 r[0]=c1; 696 c1=0; 697 sqr_add_c2(a,1,0,c2,c3,c1); 698 r[1]=c2; 699 c2=0; 700 sqr_add_c(a,1,c3,c1,c2); 701 sqr_add_c2(a,2,0,c3,c1,c2); 702 r[2]=c3; 703 c3=0; 704 sqr_add_c2(a,3,0,c1,c2,c3); 705 sqr_add_c2(a,2,1,c1,c2,c3); 706 r[3]=c1; 707 c1=0; 708 sqr_add_c(a,2,c2,c3,c1); 709 sqr_add_c2(a,3,1,c2,c3,c1); 710 sqr_add_c2(a,4,0,c2,c3,c1); 711 r[4]=c2; 712 c2=0; 713 sqr_add_c2(a,5,0,c3,c1,c2); 714 sqr_add_c2(a,4,1,c3,c1,c2); 715 sqr_add_c2(a,3,2,c3,c1,c2); 716 r[5]=c3; 717 c3=0; 718 sqr_add_c(a,3,c1,c2,c3); 719 sqr_add_c2(a,4,2,c1,c2,c3); 720 sqr_add_c2(a,5,1,c1,c2,c3); 721 sqr_add_c2(a,6,0,c1,c2,c3); 722 r[6]=c1; 723 c1=0; 724 sqr_add_c2(a,7,0,c2,c3,c1); 725 sqr_add_c2(a,6,1,c2,c3,c1); 726 sqr_add_c2(a,5,2,c2,c3,c1); 727 sqr_add_c2(a,4,3,c2,c3,c1); 728 r[7]=c2; 729 c2=0; 730 sqr_add_c(a,4,c3,c1,c2); 731 sqr_add_c2(a,5,3,c3,c1,c2); 732 sqr_add_c2(a,6,2,c3,c1,c2); 733 sqr_add_c2(a,7,1,c3,c1,c2); 734 r[8]=c3; 735 c3=0; 736 sqr_add_c2(a,7,2,c1,c2,c3); 737 sqr_add_c2(a,6,3,c1,c2,c3); 738 sqr_add_c2(a,5,4,c1,c2,c3); 739 r[9]=c1; 740 c1=0; 741 sqr_add_c(a,5,c2,c3,c1); 742 sqr_add_c2(a,6,4,c2,c3,c1); 743 sqr_add_c2(a,7,3,c2,c3,c1); 744 r[10]=c2; 745 c2=0; 746 sqr_add_c2(a,7,4,c3,c1,c2); 747 sqr_add_c2(a,6,5,c3,c1,c2); 748 r[11]=c3; 749 c3=0; 750 sqr_add_c(a,6,c1,c2,c3); 751 sqr_add_c2(a,7,5,c1,c2,c3); 752 r[12]=c1; 753 c1=0; 754 sqr_add_c2(a,7,6,c2,c3,c1); 755 r[13]=c2; 756 c2=0; 757 sqr_add_c(a,7,c3,c1,c2); 758 r[14]=c3; 759 r[15]=c1; 760 } 761 762void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a) 763 { 764#ifdef BN_LLONG 765 BN_ULLONG t,tt; 766#else 767 BN_ULONG bl,bh; 768#endif 769 BN_ULONG t1,t2; 770 BN_ULONG c1,c2,c3; 771 772 c1=0; 773 c2=0; 774 c3=0; 775 sqr_add_c(a,0,c1,c2,c3); 776 r[0]=c1; 777 c1=0; 778 sqr_add_c2(a,1,0,c2,c3,c1); 779 r[1]=c2; 780 c2=0; 781 sqr_add_c(a,1,c3,c1,c2); 782 sqr_add_c2(a,2,0,c3,c1,c2); 783 r[2]=c3; 784 c3=0; 785 sqr_add_c2(a,3,0,c1,c2,c3); 786 sqr_add_c2(a,2,1,c1,c2,c3); 787 r[3]=c1; 788 c1=0; 789 sqr_add_c(a,2,c2,c3,c1); 790 sqr_add_c2(a,3,1,c2,c3,c1); 791 r[4]=c2; 792 c2=0; 793 sqr_add_c2(a,3,2,c3,c1,c2); 794 r[5]=c3; 795 c3=0; 796 sqr_add_c(a,3,c1,c2,c3); 797 r[6]=c1; 798 r[7]=c2; 799 } 800#else /* !BN_MUL_COMBA */ 801 802/* hmm... is it faster just to do a multiply? */ 803#undef bn_sqr_comba4 804void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a) 805 { 806 BN_ULONG t[8]; 807 bn_sqr_normal(r,a,4,t); 808 } 809 810#undef bn_sqr_comba8 811void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a) 812 { 813 BN_ULONG t[16]; 814 bn_sqr_normal(r,a,8,t); 815 } 816 817void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 818 { 819 r[4]=bn_mul_words( &(r[0]),a,4,b[0]); 820 r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]); 821 r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]); 822 r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]); 823 } 824 825void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) 826 { 827 r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]); 828 r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]); 829 r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]); 830 r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]); 831 r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]); 832 r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]); 833 r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]); 834 r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]); 835 } 836 837#endif /* !BN_MUL_COMBA */ 838