| 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 "bn_lcl.h" |
| 67 | |
| 68 | #if defined(BN_LLONG) || defined(BN_UMULT_HIGH) |
| 69 | |
| 70 | BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) |
| 71 | { |
| 72 | BN_ULONG c1=0; |
| 73 | |
| 74 | assert(num >= 0); |
| 75 | if (num <= 0) return(c1); |
| 76 | |
| 77 | while (num&~3) |
| 78 | { |
| 79 | mul_add(rp[0],ap[0],w,c1); |
| 80 | mul_add(rp[1],ap[1],w,c1); |
| 81 | mul_add(rp[2],ap[2],w,c1); |
| 82 | mul_add(rp[3],ap[3],w,c1); |
| 83 | ap+=4; rp+=4; num-=4; |
| 84 | } |
| 85 | if (num) |
| 86 | { |
| 87 | mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1; |
| 88 | mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1; |
| 89 | mul_add(rp[2],ap[2],w,c1); return c1; |
| 90 | } |
| 91 | |
| 92 | return(c1); |
| 93 | } |
| 94 | |
| 95 | BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) |
| 96 | { |
| 97 | BN_ULONG c1=0; |
| 98 | |
| 99 | assert(num >= 0); |
| 100 | if (num <= 0) return(c1); |
| 101 | |
| 102 | while (num&~3) |
| 103 | { |
| 104 | mul(rp[0],ap[0],w,c1); |
| 105 | mul(rp[1],ap[1],w,c1); |
| 106 | mul(rp[2],ap[2],w,c1); |
| 107 | mul(rp[3],ap[3],w,c1); |
| 108 | ap+=4; rp+=4; num-=4; |
| 109 | } |
| 110 | if (num) |
| 111 | { |
| 112 | mul(rp[0],ap[0],w,c1); if (--num == 0) return c1; |
| 113 | mul(rp[1],ap[1],w,c1); if (--num == 0) return c1; |
| 114 | mul(rp[2],ap[2],w,c1); |
| 115 | } |
| 116 | return(c1); |
| 117 | } |
| 118 | |
| 119 | void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n) |
| 120 | { |
| 121 | assert(n >= 0); |
| 122 | if (n <= 0) return; |
| 123 | while (n&~3) |
| 124 | { |
| 125 | sqr(r[0],r[1],a[0]); |
| 126 | sqr(r[2],r[3],a[1]); |
| 127 | sqr(r[4],r[5],a[2]); |
| 128 | sqr(r[6],r[7],a[3]); |
| 129 | a+=4; r+=8; n-=4; |
| 130 | } |
| 131 | if (n) |
| 132 | { |
| 133 | sqr(r[0],r[1],a[0]); if (--n == 0) return; |
| 134 | sqr(r[2],r[3],a[1]); if (--n == 0) return; |
| 135 | sqr(r[4],r[5],a[2]); |
| 136 | } |
| 137 | } |
| 138 | |
| 139 | #else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ |
| 140 | |
| 141 | BN_ULONG bn_mul_add_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) |
| 142 | { |
| 143 | BN_ULONG c=0; |
| 144 | BN_ULONG bl,bh; |
| 145 | |
| 146 | assert(num >= 0); |
| 147 | if (num <= 0) return((BN_ULONG)0); |
| 148 | |
| 149 | bl=LBITS(w); |
| 150 | bh=HBITS(w); |
| 151 | |
| 152 | for (;;) |
| 153 | { |
| 154 | mul_add(rp[0],ap[0],bl,bh,c); |
| 155 | if (--num == 0) break; |
| 156 | mul_add(rp[1],ap[1],bl,bh,c); |
| 157 | if (--num == 0) break; |
| 158 | mul_add(rp[2],ap[2],bl,bh,c); |
| 159 | if (--num == 0) break; |
| 160 | mul_add(rp[3],ap[3],bl,bh,c); |
| 161 | if (--num == 0) break; |
| 162 | ap+=4; |
| 163 | rp+=4; |
| 164 | } |
| 165 | return(c); |
| 166 | } |
| 167 | |
| 168 | BN_ULONG bn_mul_words(BN_ULONG *rp, BN_ULONG *ap, int num, BN_ULONG w) |
| 169 | { |
| 170 | BN_ULONG carry=0; |
| 171 | BN_ULONG bl,bh; |
| 172 | |
| 173 | assert(num >= 0); |
| 174 | if (num <= 0) return((BN_ULONG)0); |
| 175 | |
| 176 | bl=LBITS(w); |
| 177 | bh=HBITS(w); |
| 178 | |
| 179 | for (;;) |
| 180 | { |
| 181 | mul(rp[0],ap[0],bl,bh,carry); |
| 182 | if (--num == 0) break; |
| 183 | mul(rp[1],ap[1],bl,bh,carry); |
| 184 | if (--num == 0) break; |
| 185 | mul(rp[2],ap[2],bl,bh,carry); |
| 186 | if (--num == 0) break; |
| 187 | mul(rp[3],ap[3],bl,bh,carry); |
| 188 | if (--num == 0) break; |
| 189 | ap+=4; |
| 190 | rp+=4; |
| 191 | } |
| 192 | return(carry); |
| 193 | } |
| 194 | |
| 195 | void bn_sqr_words(BN_ULONG *r, BN_ULONG *a, int n) |
| 196 | { |
| 197 | assert(n >= 0); |
| 198 | if (n <= 0) return; |
| 199 | for (;;) |
| 200 | { |
| 201 | sqr64(r[0],r[1],a[0]); |
| 202 | if (--n == 0) break; |
| 203 | |
| 204 | sqr64(r[2],r[3],a[1]); |
| 205 | if (--n == 0) break; |
| 206 | |
| 207 | sqr64(r[4],r[5],a[2]); |
| 208 | if (--n == 0) break; |
| 209 | |
| 210 | sqr64(r[6],r[7],a[3]); |
| 211 | if (--n == 0) break; |
| 212 | |
| 213 | a+=4; |
| 214 | r+=8; |
| 215 | } |
| 216 | } |
| 217 | |
| 218 | #endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ |
| 219 | |
| 220 | #if defined(BN_LLONG) && defined(BN_DIV2W) |
| 221 | |
| 222 | BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) |
| 223 | { |
| 224 | return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d)); |
| 225 | } |
| 226 | |
| 227 | #else |
| 228 | |
| 229 | /* Divide h,l by d and return the result. */ |
| 230 | /* I need to test this some more :-( */ |
| 231 | BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) |
| 232 | { |
| 233 | BN_ULONG dh,dl,q,ret=0,th,tl,t; |
| 234 | int i,count=2; |
| 235 | |
| 236 | if (d == 0) return(BN_MASK2); |
| 237 | |
| 238 | i=BN_num_bits_word(d); |
| 239 | assert((i == BN_BITS2) || (h > (BN_ULONG)1<<i)); |
| 240 | |
| 241 | i=BN_BITS2-i; |
| 242 | if (h >= d) h-=d; |
| 243 | |
| 244 | if (i) |
| 245 | { |
| 246 | d<<=i; |
| 247 | h=(h<<i)|(l>>(BN_BITS2-i)); |
| 248 | l<<=i; |
| 249 | } |
| 250 | dh=(d&BN_MASK2h)>>BN_BITS4; |
| 251 | dl=(d&BN_MASK2l); |
| 252 | for (;;) |
| 253 | { |
| 254 | if ((h>>BN_BITS4) == dh) |
| 255 | q=BN_MASK2l; |
| 256 | else |
| 257 | q=h/dh; |
| 258 | |
| 259 | th=q*dh; |
| 260 | tl=dl*q; |
| 261 | for (;;) |
| 262 | { |
| 263 | t=h-th; |
| 264 | if ((t&BN_MASK2h) || |
| 265 | ((tl) <= ( |
| 266 | (t<<BN_BITS4)| |
| 267 | ((l&BN_MASK2h)>>BN_BITS4)))) |
| 268 | break; |
| 269 | q--; |
| 270 | th-=dh; |
| 271 | tl-=dl; |
| 272 | } |
| 273 | t=(tl>>BN_BITS4); |
| 274 | tl=(tl<<BN_BITS4)&BN_MASK2h; |
| 275 | th+=t; |
| 276 | |
| 277 | if (l < tl) th++; |
| 278 | l-=tl; |
| 279 | if (h < th) |
| 280 | { |
| 281 | h+=d; |
| 282 | q--; |
| 283 | } |
| 284 | h-=th; |
| 285 | |
| 286 | if (--count == 0) break; |
| 287 | |
| 288 | ret=q<<BN_BITS4; |
| 289 | h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2; |
| 290 | l=(l&BN_MASK2l)<<BN_BITS4; |
| 291 | } |
| 292 | ret|=q; |
| 293 | return(ret); |
| 294 | } |
| 295 | #endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */ |
| 296 | |
| 297 | #ifdef BN_LLONG |
| 298 | BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) |
| 299 | { |
| 300 | BN_ULLONG ll=0; |
| 301 | |
| 302 | assert(n >= 0); |
| 303 | if (n <= 0) return((BN_ULONG)0); |
| 304 | |
| 305 | for (;;) |
| 306 | { |
| 307 | ll+=(BN_ULLONG)a[0]+b[0]; |
| 308 | r[0]=(BN_ULONG)ll&BN_MASK2; |
| 309 | ll>>=BN_BITS2; |
| 310 | if (--n <= 0) break; |
| 311 | |
| 312 | ll+=(BN_ULLONG)a[1]+b[1]; |
| 313 | r[1]=(BN_ULONG)ll&BN_MASK2; |
| 314 | ll>>=BN_BITS2; |
| 315 | if (--n <= 0) break; |
| 316 | |
| 317 | ll+=(BN_ULLONG)a[2]+b[2]; |
| 318 | r[2]=(BN_ULONG)ll&BN_MASK2; |
| 319 | ll>>=BN_BITS2; |
| 320 | if (--n <= 0) break; |
| 321 | |
| 322 | ll+=(BN_ULLONG)a[3]+b[3]; |
| 323 | r[3]=(BN_ULONG)ll&BN_MASK2; |
| 324 | ll>>=BN_BITS2; |
| 325 | if (--n <= 0) break; |
| 326 | |
| 327 | a+=4; |
| 328 | b+=4; |
| 329 | r+=4; |
| 330 | } |
| 331 | return((BN_ULONG)ll); |
| 332 | } |
| 333 | #else /* !BN_LLONG */ |
| 334 | BN_ULONG bn_add_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) |
| 335 | { |
| 336 | BN_ULONG c,l,t; |
| 337 | |
| 338 | assert(n >= 0); |
| 339 | if (n <= 0) return((BN_ULONG)0); |
| 340 | |
| 341 | c=0; |
| 342 | for (;;) |
| 343 | { |
| 344 | t=a[0]; |
| 345 | t=(t+c)&BN_MASK2; |
| 346 | c=(t < c); |
| 347 | l=(t+b[0])&BN_MASK2; |
| 348 | c+=(l < t); |
| 349 | r[0]=l; |
| 350 | if (--n <= 0) break; |
| 351 | |
| 352 | t=a[1]; |
| 353 | t=(t+c)&BN_MASK2; |
| 354 | c=(t < c); |
| 355 | l=(t+b[1])&BN_MASK2; |
| 356 | c+=(l < t); |
| 357 | r[1]=l; |
| 358 | if (--n <= 0) break; |
| 359 | |
| 360 | t=a[2]; |
| 361 | t=(t+c)&BN_MASK2; |
| 362 | c=(t < c); |
| 363 | l=(t+b[2])&BN_MASK2; |
| 364 | c+=(l < t); |
| 365 | r[2]=l; |
| 366 | if (--n <= 0) break; |
| 367 | |
| 368 | t=a[3]; |
| 369 | t=(t+c)&BN_MASK2; |
| 370 | c=(t < c); |
| 371 | l=(t+b[3])&BN_MASK2; |
| 372 | c+=(l < t); |
| 373 | r[3]=l; |
| 374 | if (--n <= 0) break; |
| 375 | |
| 376 | a+=4; |
| 377 | b+=4; |
| 378 | r+=4; |
| 379 | } |
| 380 | return((BN_ULONG)c); |
| 381 | } |
| 382 | #endif /* !BN_LLONG */ |
| 383 | |
