| 1 | /** |
| 2 | * \file bn_mul.h |
| 3 | * |
| 4 | * Copyright (C) 2006-2010, Brainspark B.V. |
| 5 | * |
| 6 | * This file is part of PolarSSL (http://www.polarssl.org) |
| 7 | * Lead Maintainer: Paul Bakker <polarssl_maintainer at polarssl.org> |
| 8 | * |
| 9 | * All rights reserved. |
| 10 | * |
| 11 | * This program is free software; you can redistribute it and/or modify |
| 12 | * it under the terms of the GNU General Public License as published by |
| 13 | * the Free Software Foundation; either version 2 of the License, or |
| 14 | * (at your option) any later version. |
| 15 | * |
| 16 | * This program is distributed in the hope that it will be useful, |
| 17 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 18 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 19 | * GNU General Public License for more details. |
| 20 | * |
| 21 | * You should have received a copy of the GNU General Public License along |
| 22 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 23 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
| 24 | */ |
| 25 | /* |
| 26 | * Multiply source vector [s] with b, add result |
| 27 | * to destination vector [d] and set carry c. |
| 28 | * |
| 29 | * Currently supports: |
| 30 | * |
| 31 | * . IA-32 (386+) . AMD64 / EM64T |
| 32 | * . IA-32 (SSE2) . Motorola 68000 |
| 33 | * . PowerPC, 32-bit . MicroBlaze |
| 34 | * . PowerPC, 64-bit . TriCore |
| 35 | * . SPARC v8 . ARM v3+ |
| 36 | * . Alpha . MIPS32 |
| 37 | * . C, longlong . C, generic |
| 38 | */ |
| 39 | |
| 40 | #ifndef POLARSSL_BN_MUL_H |
| 41 | #define POLARSSL_BN_MUL_H |
| 42 | |
| 43 | #if defined(POLARSSL_HAVE_ASM) |
| 44 | |
| 45 | #if defined(__GNUC__) |
| 46 | #if defined(__i386__) |
| 47 | |
| 48 | #define MULADDC_INIT \ |
| 49 | asm( " \ |
| 50 | movl %%ebx, %0; \ |
| 51 | movl %5, %%esi; \ |
| 52 | movl %6, %%edi; \ |
| 53 | movl %7, %%ecx; \ |
| 54 | movl %8, %%ebx; \ |
| 55 | " |
| 56 | |
| 57 | #define MULADDC_CORE \ |
| 58 | " \ |
| 59 | lodsl; \ |
| 60 | mull %%ebx; \ |
| 61 | addl %%ecx, %%eax; \ |
| 62 | adcl $0, %%edx; \ |
| 63 | addl (%%edi), %%eax; \ |
| 64 | adcl $0, %%edx; \ |
| 65 | movl %%edx, %%ecx; \ |
| 66 | stosl; \ |
| 67 | " |
| 68 | |
| 69 | #if defined(POLARSSL_HAVE_SSE2) |
| 70 | |
| 71 | #define MULADDC_HUIT \ |
| 72 | " \ |
| 73 | movd %%ecx, %%mm1; \ |
| 74 | movd %%ebx, %%mm0; \ |
| 75 | movd (%%edi), %%mm3; \ |
| 76 | paddq %%mm3, %%mm1; \ |
| 77 | movd (%%esi), %%mm2; \ |
| 78 | pmuludq %%mm0, %%mm2; \ |
| 79 | movd 4(%%esi), %%mm4; \ |
| 80 | pmuludq %%mm0, %%mm4; \ |
| 81 | movd 8(%%esi), %%mm6; \ |
| 82 | pmuludq %%mm0, %%mm6; \ |
| 83 | movd 12(%%esi), %%mm7; \ |
| 84 | pmuludq %%mm0, %%mm7; \ |
| 85 | paddq %%mm2, %%mm1; \ |
| 86 | movd 4(%%edi), %%mm3; \ |
| 87 | paddq %%mm4, %%mm3; \ |
| 88 | movd 8(%%edi), %%mm5; \ |
| 89 | paddq %%mm6, %%mm5; \ |
| 90 | movd 12(%%edi), %%mm4; \ |
| 91 | paddq %%mm4, %%mm7; \ |
| 92 | movd %%mm1, (%%edi); \ |
| 93 | movd 16(%%esi), %%mm2; \ |
| 94 | pmuludq %%mm0, %%mm2; \ |
| 95 | psrlq $32, %%mm1; \ |
| 96 | movd 20(%%esi), %%mm4; \ |
| 97 | pmuludq %%mm0, %%mm4; \ |
| 98 | paddq %%mm3, %%mm1; \ |
| 99 | movd 24(%%esi), %%mm6; \ |
| 100 | pmuludq %%mm0, %%mm6; \ |
| 101 | movd %%mm1, 4(%%edi); \ |
| 102 | psrlq $32, %%mm1; \ |
| 103 | movd 28(%%esi), %%mm3; \ |
| 104 | pmuludq %%mm0, %%mm3; \ |
| 105 | paddq %%mm5, %%mm1; \ |
| 106 | movd 16(%%edi), %%mm5; \ |
| 107 | paddq %%mm5, %%mm2; \ |
| 108 | movd %%mm1, 8(%%edi); \ |
| 109 | psrlq $32, %%mm1; \ |
| 110 | paddq %%mm7, %%mm1; \ |
| 111 | movd 20(%%edi), %%mm5; \ |
| 112 | paddq %%mm5, %%mm4; \ |
| 113 | movd %%mm1, 12(%%edi); \ |
| 114 | psrlq $32, %%mm1; \ |
| 115 | paddq %%mm2, %%mm1; \ |
| 116 | movd 24(%%edi), %%mm5; \ |
| 117 | paddq %%mm5, %%mm6; \ |
| 118 | movd %%mm1, 16(%%edi); \ |
| 119 | psrlq $32, %%mm1; \ |
| 120 | paddq %%mm4, %%mm1; \ |
| 121 | movd 28(%%edi), %%mm5; \ |
| 122 | paddq %%mm5, %%mm3; \ |
| 123 | movd %%mm1, 20(%%edi); \ |
| 124 | psrlq $32, %%mm1; \ |
| 125 | paddq %%mm6, %%mm1; \ |
| 126 | movd %%mm1, 24(%%edi); \ |
| 127 | psrlq $32, %%mm1; \ |
| 128 | paddq %%mm3, %%mm1; \ |
| 129 | movd %%mm1, 28(%%edi); \ |
| 130 | addl $32, %%edi; \ |
| 131 | addl $32, %%esi; \ |
| 132 | psrlq $32, %%mm1; \ |
| 133 | movd %%mm1, %%ecx; \ |
| 134 | " |
| 135 | |
| 136 | #define MULADDC_STOP \ |
| 137 | " \ |
| 138 | emms; \ |
| 139 | movl %4, %%ebx; \ |
| 140 | movl %%ecx, %1; \ |
| 141 | movl %%edi, %2; \ |
| 142 | movl %%esi, %3; \ |
| 143 | " \ |
| 144 | : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \ |
| 145 | : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \ |
| 146 | : "eax", "ecx", "edx", "esi", "edi" \ |
| 147 | ); |
| 148 | |
| 149 | #else |
| 150 | |
| 151 | #define MULADDC_STOP \ |
| 152 | " \ |
| 153 | movl %4, %%ebx; \ |
| 154 | movl %%ecx, %1; \ |
| 155 | movl %%edi, %2; \ |
| 156 | movl %%esi, %3; \ |
| 157 | " \ |
| 158 | : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \ |
| 159 | : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \ |
| 160 | : "eax", "ecx", "edx", "esi", "edi" \ |
| 161 | ); |
| 162 | #endif /* SSE2 */ |
| 163 | #endif /* i386 */ |
| 164 | |
| 165 | #if defined(__amd64__) || defined (__x86_64__) |
| 166 | |
| 167 | #define MULADDC_INIT \ |
| 168 | asm( "movq %0, %%rsi " :: "m" (s)); \ |
| 169 | asm( "movq %0, %%rdi " :: "m" (d)); \ |
| 170 | asm( "movq %0, %%rcx " :: "m" (c)); \ |
| 171 | asm( "movq %0, %%rbx " :: "m" (b)); \ |
| 172 | asm( "xorq %r8, %r8 " ); |
| 173 | |
| 174 | #define MULADDC_CORE \ |
| 175 | asm( "movq (%rsi),%rax " ); \ |
| 176 | asm( "mulq %rbx " ); \ |
| 177 | asm( "addq $8, %rsi " ); \ |
| 178 | asm( "addq %rcx, %rax " ); \ |
| 179 | asm( "movq %r8, %rcx " ); \ |
| 180 | asm( "adcq $0, %rdx " ); \ |
| 181 | asm( "nop " ); \ |
| 182 | asm( "addq %rax, (%rdi) " ); \ |
| 183 | asm( "adcq %rdx, %rcx " ); \ |
| 184 | asm( "addq $8, %rdi " ); |
| 185 | |
| 186 | #define MULADDC_STOP \ |
| 187 | asm( "movq %%rcx, %0 " : "=m" (c)); \ |
| 188 | asm( "movq %%rdi, %0 " : "=m" (d)); \ |
| 189 | asm( "movq %%rsi, %0 " : "=m" (s) :: \ |
| 190 | "rax", "rcx", "rdx", "rbx", "rsi", "rdi", "r8" ); |
| 191 | |
| 192 | #endif /* AMD64 */ |
| 193 | |
| 194 | #if defined(__mc68020__) || defined(__mcpu32__) |
| 195 | |
| 196 | #define MULADDC_INIT \ |
| 197 | asm( "movl %0, %%a2 " :: "m" (s)); \ |
| 198 | asm( "movl %0, %%a3 " :: "m" (d)); \ |
| 199 | asm( "movl %0, %%d3 " :: "m" (c)); \ |
| 200 | asm( "movl %0, %%d2 " :: "m" (b)); \ |
| 201 | asm( "moveq #0, %d0 " ); |
| 202 | |
| 203 | #define MULADDC_CORE \ |
| 204 | asm( "movel %a2@+, %d1 " ); \ |
| 205 | asm( "mulul %d2, %d4:%d1 " ); \ |
| 206 | asm( "addl %d3, %d1 " ); \ |
| 207 | asm( "addxl %d0, %d4 " ); \ |
| 208 | asm( "moveq #0, %d3 " ); \ |
| 209 | asm( "addl %d1, %a3@+ " ); \ |
| 210 | asm( "addxl %d4, %d3 " ); |
| 211 | |
| 212 | #define MULADDC_STOP \ |
| 213 | asm( "movl %%d3, %0 " : "=m" (c)); \ |
| 214 | asm( "movl %%a3, %0 " : "=m" (d)); \ |
| 215 | asm( "movl %%a2, %0 " : "=m" (s) :: \ |
| 216 | "d0", "d1", "d2", "d3", "d4", "a2", "a3" ); |
| 217 | |
| 218 | #define MULADDC_HUIT \ |
| 219 | asm( "movel %a2@+, %d1 " ); \ |
| 220 | asm( "mulul %d2, %d4:%d1 " ); \ |
| 221 | asm( "addxl %d3, %d1 " ); \ |
| 222 | asm( "addxl %d0, %d4 " ); \ |
| 223 | asm( "addl %d1, %a3@+ " ); \ |
| 224 | asm( "movel %a2@+, %d1 " ); \ |
| 225 | asm( "mulul %d2, %d3:%d1 " ); \ |
| 226 | asm( "addxl %d4, %d1 " ); \ |
| 227 | asm( "addxl %d0, %d3 " ); \ |
| 228 | asm( "addl %d1, %a3@+ " ); \ |
| 229 | asm( "movel %a2@+, %d1 " ); \ |
| 230 | asm( "mulul %d2, %d4:%d1 " ); \ |
| 231 | asm( "addxl %d3, %d1 " ); \ |
| 232 | asm( "addxl %d0, %d4 " ); \ |
| 233 | asm( "addl %d1, %a3@+ " ); \ |
| 234 | asm( "movel %a2@+, %d1 " ); \ |
| 235 | asm( "mulul %d2, %d3:%d1 " ); \ |
| 236 | asm( "addxl %d4, %d1 " ); \ |
| 237 | asm( "addxl %d0, %d3 " ); \ |
| 238 | asm( "addl %d1, %a3@+ " ); \ |
| 239 | asm( "movel %a2@+, %d1 " ); \ |
| 240 | asm( "mulul %d2, %d4:%d1 " ); \ |
| 241 | asm( "addxl %d3, %d1 " ); \ |
| 242 | asm( "addxl %d0, %d4 " ); \ |
| 243 | asm( "addl %d1, %a3@+ " ); \ |
| 244 | asm( "movel %a2@+, %d1 " ); \ |
| 245 | asm( "mulul %d2, %d3:%d1 " ); \ |
| 246 | asm( "addxl %d4, %d1 " ); \ |
| 247 | asm( "addxl %d0, %d3 " ); \ |
| 248 | asm( "addl %d1, %a3@+ " ); \ |
| 249 | asm( "movel %a2@+, %d1 " ); \ |
| 250 | asm( "mulul %d2, %d4:%d1 " ); \ |
| 251 | asm( "addxl %d3, %d1 " ); \ |
| 252 | asm( "addxl %d0, %d4 " ); \ |
| 253 | asm( "addl %d1, %a3@+ " ); \ |
| 254 | asm( "movel %a2@+, %d1 " ); \ |
| 255 | asm( "mulul %d2, %d3:%d1 " ); \ |
| 256 | asm( "addxl %d4, %d1 " ); \ |
| 257 | asm( "addxl %d0, %d3 " ); \ |
| 258 | asm( "addl %d1, %a3@+ " ); \ |
| 259 | asm( "addxl %d0, %d3 " ); |
| 260 | |
| 261 | #endif /* MC68000 */ |
| 262 | |
| 263 | #if defined(__powerpc__) || defined(__ppc__) |
| 264 | #if defined(__powerpc64__) || defined(__ppc64__) |
| 265 | |
| 266 | #if defined(__MACH__) && defined(__APPLE__) |
| 267 | |
| 268 | #define MULADDC_INIT \ |
| 269 | asm( "ld r3, %0 " :: "m" (s)); \ |
| 270 | asm( "ld r4, %0 " :: "m" (d)); \ |
| 271 | asm( "ld r5, %0 " :: "m" (c)); \ |
| 272 | asm( "ld r6, %0 " :: "m" (b)); \ |
| 273 | asm( "addi r3, r3, -8 " ); \ |
| 274 | asm( "addi r4, r4, -8 " ); \ |
| 275 | asm( "addic r5, r5, 0 " ); |
| 276 | |
| 277 | #define MULADDC_CORE \ |
| 278 | asm( "ldu r7, 8(r3) " ); \ |
| 279 | asm( "mulld r8, r7, r6 " ); \ |
| 280 | asm( "mulhdu r9, r7, r6 " ); \ |
| 281 | asm( "adde r8, r8, r5 " ); \ |
| 282 | asm( "ld r7, 8(r4) " ); \ |
| 283 | asm( "addze r5, r9 " ); \ |
| 284 | asm( "addc r8, r8, r7 " ); \ |
| 285 | asm( "stdu r8, 8(r4) " ); |
| 286 | |
| 287 | #define MULADDC_STOP \ |
| 288 | asm( "addze r5, r5 " ); \ |
| 289 | asm( "addi r4, r4, 8 " ); \ |
| 290 | asm( "addi r3, r3, 8 " ); \ |
| 291 | asm( "std r5, %0 " : "=m" (c)); \ |
| 292 | asm( "std r4, %0 " : "=m" (d)); \ |
| 293 | asm( "std r3, %0 " : "=m" (s) :: \ |
| 294 | "r3", "r4", "r5", "r6", "r7", "r8", "r9" ); |
| 295 | |
| 296 | #else |
| 297 | |
| 298 | #define MULADDC_INIT \ |
| 299 | asm( "ld %%r3, %0 " :: "m" (s)); \ |
| 300 | asm( "ld %%r4, %0 " :: "m" (d)); \ |
| 301 | asm( "ld %%r5, %0 " :: "m" (c)); \ |
| 302 | asm( "ld %%r6, %0 " :: "m" (b)); \ |
| 303 | asm( "addi %r3, %r3, -8 " ); \ |
| 304 | asm( "addi %r4, %r4, -8 " ); \ |
| 305 | asm( "addic %r5, %r5, 0 " ); |
| 306 | |
| 307 | #define MULADDC_CORE \ |
| 308 | asm( "ldu %r7, 8(%r3) " ); \ |
| 309 | asm( "mulld %r8, %r7, %r6 " ); \ |
| 310 | asm( "mulhdu %r9, %r7, %r6 " ); \ |
| 311 | asm( "adde %r8, %r8, %r5 " ); \ |
| 312 | asm( "ld %r7, 8(%r4) " ); \ |
| 313 | asm( "addze %r5, %r9 " ); \ |
| 314 | asm( "addc %r8, %r8, %r7 " ); \ |
| 315 | asm( "stdu %r8, 8(%r4) " ); |
| 316 | |
| 317 | #define MULADDC_STOP \ |
| 318 | asm( "addze %r5, %r5 " ); \ |
| 319 | asm( "addi %r4, %r4, 8 " ); \ |
| 320 | asm( "addi %r3, %r3, 8 " ); \ |
| 321 | asm( "std %%r5, %0 " : "=m" (c)); \ |
| 322 | asm( "std %%r4, %0 " : "=m" (d)); \ |
| 323 | asm( "std %%r3, %0 " : "=m" (s) :: \ |
| 324 | "r3", "r4", "r5", "r6", "r7", "r8", "r9" ); |
| 325 | |
| 326 | #endif |
| 327 | |
| 328 | #else /* PPC32 */ |
| 329 | |
| 330 | #if defined(__MACH__) && defined(__APPLE__) |
| 331 | |
| 332 | #define MULADDC_INIT \ |
| 333 | asm( "lwz r3, %0 " :: "m" (s)); \ |
| 334 | asm( "lwz r4, %0 " :: "m" (d)); \ |
| 335 | asm( "lwz r5, %0 " :: "m" (c)); \ |
| 336 | asm( "lwz r6, %0 " :: "m" (b)); \ |
| 337 | asm( "addi r3, r3, -4 " ); \ |
| 338 | asm( "addi r4, r4, -4 " ); \ |
| 339 | asm( "addic r5, r5, 0 " ); |
| 340 | |
| 341 | #define MULADDC_CORE \ |
| 342 | asm( "lwzu r7, 4(r3) " ); \ |
| 343 | asm( "mullw r8, r7, r6 " ); \ |
| 344 | asm( "mulhwu r9, r7, r6 " ); \ |
| 345 | asm( "adde r8, r8, r5 " ); \ |
| 346 | asm( "lwz r7, 4(r4) " ); \ |
| 347 | asm( "addze r5, r9 " ); \ |
| 348 | asm( "addc r8, r8, r7 " ); \ |
| 349 | asm( "stwu r8, 4(r4) " ); |
| 350 | |
| 351 | #define MULADDC_STOP \ |
| 352 | asm( "addze r5, r5 " ); \ |
| 353 | asm( "addi r4, r4, 4 " ); \ |
| 354 | asm( "addi r3, r3, 4 " ); \ |
| 355 | asm( "stw r5, %0 " : "=m" (c)); \ |
| 356 | asm( "stw r4, %0 " : "=m" (d)); \ |
| 357 | asm( "stw r3, %0 " : "=m" (s) :: \ |
| 358 | "r3", "r4", "r5", "r6", "r7", "r8", "r9" ); |
| 359 | |
| 360 | #else |
| 361 | |
| 362 | #define MULADDC_INIT \ |
| 363 | asm( "lwz %%r3, %0 " :: "m" (s)); \ |
| 364 | asm( "lwz %%r4, %0 " :: "m" (d)); \ |
| 365 | asm( "lwz %%r5, %0 " :: "m" (c)); \ |
| 366 | asm( "lwz %%r6, %0 " :: "m" (b)); \ |
| 367 | asm( "addi %r3, %r3, -4 " ); \ |
| 368 | asm( "addi %r4, %r4, -4 " ); \ |
| 369 | asm( "addic %r5, %r5, 0 " ); |
| 370 | |
| 371 | #define MULADDC_CORE \ |
| 372 | asm( "lwzu %r7, 4(%r3) " ); \ |
| 373 | asm( "mullw %r8, %r7, %r6 " ); \ |
| 374 | asm( "mulhwu %r9, %r7, %r6 " ); \ |
| 375 | asm( "adde %r8, %r8, %r5 " ); \ |
| 376 | asm( "lwz %r7, 4(%r4) " ); \ |
| 377 | asm( "addze %r5, %r9 " ); \ |
| 378 | asm( "addc %r8, %r8, %r7 " ); \ |
| 379 | asm( "stwu %r8, 4(%r4) " ); |
| 380 | |
| 381 | #define MULADDC_STOP \ |
| 382 | asm( "addze %r5, %r5 " ); \ |
| 383 | asm( "addi %r4, %r4, 4 " ); \ |
| 384 | asm( "addi %r3, %r3, 4 " ); \ |
| 385 | asm( "stw %%r5, %0 " : "=m" (c)); \ |
| 386 | asm( "stw %%r4, %0 " : "=m" (d)); \ |
| 387 | asm( "stw %%r3, %0 " : "=m" (s) :: \ |
| 388 | "r3", "r4", "r5", "r6", "r7", "r8", "r9" ); |
| 389 | |
| 390 | #endif |
| 391 | |
| 392 | #endif /* PPC32 */ |
| 393 | #endif /* PPC64 */ |
| 394 | |
| 395 | #if defined(__sparc__) |
| 396 | |
| 397 | #define MULADDC_INIT \ |
| 398 | asm( "ld %0, %%o0 " :: "m" (s)); \ |
| 399 | asm( "ld %0, %%o1 " :: "m" (d)); \ |
| 400 | asm( "ld %0, %%o2 " :: "m" (c)); \ |
| 401 | asm( "ld %0, %%o3 " :: "m" (b)); |
| 402 | |
| 403 | #define MULADDC_CORE \ |
| 404 | asm( "ld [%o0], %o4 " ); \ |
| 405 | asm( "inc 4, %o0 " ); \ |
| 406 | asm( "ld [%o1], %o5 " ); \ |
| 407 | asm( "umul %o3, %o4, %o4 " ); \ |
| 408 | asm( "addcc %o4, %o2, %o4 " ); \ |
| 409 | asm( "rd %y, %g1 " ); \ |
| 410 | asm( "addx %g1, 0, %g1 " ); \ |
| 411 | asm( "addcc %o4, %o5, %o4 " ); \ |
| 412 | asm( "st %o4, [%o1] " ); \ |
| 413 | asm( "addx %g1, 0, %o2 " ); \ |
| 414 | asm( "inc 4, %o1 " ); |
| 415 | |
| 416 | #define MULADDC_STOP \ |
| 417 | asm( "st %%o2, %0 " : "=m" (c)); \ |
| 418 | asm( "st %%o1, %0 " : "=m" (d)); \ |
| 419 | asm( "st %%o0, %0 " : "=m" (s) :: \ |
| 420 | "g1", "o0", "o1", "o2", "o3", "o4", "o5" ); |
| 421 | |
| 422 | #endif /* SPARCv8 */ |
| 423 | |
| 424 | #if defined(__microblaze__) || defined(microblaze) |
| 425 | |
| 426 | #define MULADDC_INIT \ |
| 427 | asm( "lwi r3, %0 " :: "m" (s)); \ |
| 428 | asm( "lwi r4, %0 " :: "m" (d)); \ |
| 429 | asm( "lwi r5, %0 " :: "m" (c)); \ |
| 430 | asm( "lwi r6, %0 " :: "m" (b)); \ |
| 431 | asm( "andi r7, r6, 0xffff" ); \ |
| 432 | asm( "bsrli r6, r6, 16 " ); |
| 433 | |
| 434 | #define MULADDC_CORE \ |
| 435 | asm( "lhui r8, r3, 0 " ); \ |
| 436 | asm( "addi r3, r3, 2 " ); \ |
| 437 | asm( "lhui r9, r3, 0 " ); \ |
| 438 | asm( "addi r3, r3, 2 " ); \ |
| 439 | asm( "mul r10, r9, r6 " ); \ |
| 440 | asm( "mul r11, r8, r7 " ); \ |
| 441 | asm( "mul r12, r9, r7 " ); \ |
| 442 | asm( "mul r13, r8, r6 " ); \ |
| 443 | asm( "bsrli r8, r10, 16 " ); \ |
| 444 | asm( "bsrli r9, r11, 16 " ); \ |
| 445 | asm( "add r13, r13, r8 " ); \ |
| 446 | asm( "add r13, r13, r9 " ); \ |
| 447 | asm( "bslli r10, r10, 16 " ); \ |
| 448 | asm( "bslli r11, r11, 16 " ); \ |
| 449 | asm( "add r12, r12, r10 " ); \ |
| 450 | asm( "addc r13, r13, r0 " ); \ |
| 451 | asm( "add r12, r12, r11 " ); \ |
| 452 | asm( "addc r13, r13, r0 " ); \ |
| 453 | asm( "lwi r10, r4, 0 " ); \ |
| 454 | asm( "add r12, r12, r10 " ); \ |
| 455 | asm( "addc r13, r13, r0 " ); \ |
| 456 | asm( "add r12, r12, r5 " ); \ |
| 457 | asm( "addc r5, r13, r0 " ); \ |
| 458 | asm( "swi r12, r4, 0 " ); \ |
| 459 | asm( "addi r4, r4, 4 " ); |
| 460 | |
| 461 | #define MULADDC_STOP \ |
| 462 | asm( "swi r5, %0 " : "=m" (c)); \ |
| 463 | asm( "swi r4, %0 " : "=m" (d)); \ |
| 464 | asm( "swi r3, %0 " : "=m" (s) :: \ |
| 465 | "r3", "r4" , "r5" , "r6" , "r7" , "r8" , \ |
| 466 | "r9", "r10", "r11", "r12", "r13" ); |
| 467 | |
| 468 | #endif /* MicroBlaze */ |
| 469 | |
| 470 | #if defined(__tricore__) |
| 471 | |
| 472 | #define MULADDC_INIT \ |
| 473 | asm( "ld.a %%a2, %0 " :: "m" (s)); \ |
| 474 | asm( "ld.a %%a3, %0 " :: "m" (d)); \ |
| 475 | asm( "ld.w %%d4, %0 " :: "m" (c)); \ |
| 476 | asm( "ld.w %%d1, %0 " :: "m" (b)); \ |
| 477 | asm( "xor %d5, %d5 " ); |
| 478 | |
| 479 | #define MULADDC_CORE \ |
| 480 | asm( "ld.w %d0, [%a2+] " ); \ |
| 481 | asm( "madd.u %e2, %e4, %d0, %d1 " ); \ |
| 482 | asm( "ld.w %d0, [%a3] " ); \ |
| 483 | asm( "addx %d2, %d2, %d0 " ); \ |
| 484 | asm( "addc %d3, %d3, 0 " ); \ |
| 485 | asm( "mov %d4, %d3 " ); \ |
| 486 | asm( "st.w [%a3+], %d2 " ); |
| 487 | |
| 488 | #define MULADDC_STOP \ |
| 489 | asm( "st.w %0, %%d4 " : "=m" (c)); \ |
| 490 | asm( "st.a %0, %%a3 " : "=m" (d)); \ |
| 491 | asm( "st.a %0, %%a2 " : "=m" (s) :: \ |
| 492 | "d0", "d1", "e2", "d4", "a2", "a3" ); |
| 493 | |
| 494 | #endif /* TriCore */ |
| 495 | |
| 496 | #if defined(__arm__) |
| 497 | |
| 498 | #define MULADDC_INIT \ |
| 499 | asm( "ldr r0, %0 " :: "m" (s)); \ |
| 500 | asm( "ldr r1, %0 " :: "m" (d)); \ |
| 501 | asm( "ldr r2, %0 " :: "m" (c)); \ |
| 502 | asm( "ldr r3, %0 " :: "m" (b)); |
| 503 | |
| 504 | #define MULADDC_CORE \ |
| 505 | asm( "ldr r4, [r0], #4 " ); \ |
| 506 | asm( "mov r5, #0 " ); \ |
| 507 | asm( "ldr r6, [r1] " ); \ |
| 508 | asm( "umlal r2, r5, r3, r4 " ); \ |
| 509 | asm( "adds r7, r6, r2 " ); \ |
| 510 | asm( "adc r2, r5, #0 " ); \ |
| 511 | asm( "str r7, [r1], #4 " ); |
| 512 | |
| 513 | #define MULADDC_STOP \ |
| 514 | asm( "str r2, %0 " : "=m" (c)); \ |
| 515 | asm( "str r1, %0 " : "=m" (d)); \ |
| 516 | asm( "str r0, %0 " : "=m" (s) :: \ |
| 517 | "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7" ); |
| 518 | |
| 519 | #endif /* ARMv3 */ |
| 520 | |
| 521 | #if defined(__alpha__) |
| 522 | |
| 523 | #define MULADDC_INIT \ |
| 524 | asm( "ldq $1, %0 " :: "m" (s)); \ |
| 525 | asm( "ldq $2, %0 " :: "m" (d)); \ |
| 526 | asm( "ldq $3, %0 " :: "m" (c)); \ |
| 527 | asm( "ldq $4, %0 " :: "m" (b)); |
| 528 | |
| 529 | #define MULADDC_CORE \ |
| 530 | asm( "ldq $6, 0($1) " ); \ |
| 531 | asm( "addq $1, 8, $1 " ); \ |
| 532 | asm( "mulq $6, $4, $7 " ); \ |
| 533 | asm( "umulh $6, $4, $6 " ); \ |
| 534 | asm( "addq $7, $3, $7 " ); \ |
| 535 | asm( "cmpult $7, $3, $3 " ); \ |
| 536 | asm( "ldq $5, 0($2) " ); \ |
| 537 | asm( "addq $7, $5, $7 " ); \ |
| 538 | asm( "cmpult $7, $5, $5 " ); \ |
| 539 | asm( "stq $7, 0($2) " ); \ |
| 540 | asm( "addq $2, 8, $2 " ); \ |
| 541 | asm( "addq $6, $3, $3 " ); \ |
| 542 | asm( "addq $5, $3, $3 " ); |
| 543 | |
| 544 | #define MULADDC_STOP \ |
| 545 | asm( "stq $3, %0 " : "=m" (c)); \ |
| 546 | asm( "stq $2, %0 " : "=m" (d)); \ |
| 547 | asm( "stq $1, %0 " : "=m" (s) :: \ |
| 548 | "$1", "$2", "$3", "$4", "$5", "$6", "$7" ); |
| 549 | |
| 550 | #endif /* Alpha */ |
| 551 | |
| 552 | #if defined(__mips__) |
| 553 | |
| 554 | #define MULADDC_INIT \ |
| 555 | asm( "lw $10, %0 " :: "m" (s)); \ |
| 556 | asm( "lw $11, %0 " :: "m" (d)); \ |
| 557 | asm( "lw $12, %0 " :: "m" (c)); \ |
| 558 | asm( "lw $13, %0 " :: "m" (b)); |
| 559 | |
| 560 | #define MULADDC_CORE \ |
| 561 | asm( "lw $14, 0($10) " ); \ |
| 562 | asm( "multu $13, $14 " ); \ |
| 563 | asm( "addi $10, $10, 4 " ); \ |
| 564 | asm( "mflo $14 " ); \ |
| 565 | asm( "mfhi $9 " ); \ |
| 566 | asm( "addu $14, $12, $14 " ); \ |
| 567 | asm( "lw $15, 0($11) " ); \ |
| 568 | asm( "sltu $12, $14, $12 " ); \ |
| 569 | asm( "addu $15, $14, $15 " ); \ |
| 570 | asm( "sltu $14, $15, $14 " ); \ |
| 571 | asm( "addu $12, $12, $9 " ); \ |
| 572 | asm( "sw $15, 0($11) " ); \ |
| 573 | asm( "addu $12, $12, $14 " ); \ |
| 574 | asm( "addi $11, $11, 4 " ); |
| 575 | |
| 576 | #define MULADDC_STOP \ |
| 577 | asm( "sw $12, %0 " : "=m" (c)); \ |
| 578 | asm( "sw $11, %0 " : "=m" (d)); \ |
| 579 | asm( "sw $10, %0 " : "=m" (s) :: \ |
| 580 | "$9", "$10", "$11", "$12", "$13", "$14", "$15" ); |
| 581 | |
| 582 | #endif /* MIPS */ |
| 583 | #endif /* GNUC */ |
| 584 | |
| 585 | #if (defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__) |
| 586 | |
| 587 | #define MULADDC_INIT \ |
| 588 | __asm mov esi, s \ |
| 589 | __asm mov edi, d \ |
| 590 | __asm mov ecx, c \ |
| 591 | __asm mov ebx, b |
| 592 | |
| 593 | #define MULADDC_CORE \ |
| 594 | __asm lodsd \ |
| 595 | __asm mul ebx \ |
| 596 | __asm add eax, ecx \ |
| 597 | __asm adc edx, 0 \ |
| 598 | __asm add eax, [edi] \ |
| 599 | __asm adc edx, 0 \ |
| 600 | __asm mov ecx, edx \ |
| 601 | __asm stosd |
| 602 | |
| 603 | #if defined(POLARSSL_HAVE_SSE2) |
| 604 | |
| 605 | #define EMIT __asm _emit |
| 606 | |
| 607 | #define MULADDC_HUIT \ |
| 608 | EMIT 0x0F EMIT 0x6E EMIT 0xC9 \ |
| 609 | EMIT 0x0F EMIT 0x6E EMIT 0xC3 \ |
| 610 | EMIT 0x0F EMIT 0x6E EMIT 0x1F \ |
| 611 | EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ |
| 612 | EMIT 0x0F EMIT 0x6E EMIT 0x16 \ |
| 613 | EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \ |
| 614 | EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x04 \ |
| 615 | EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \ |
| 616 | EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x08 \ |
| 617 | EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \ |
| 618 | EMIT 0x0F EMIT 0x6E EMIT 0x7E EMIT 0x0C \ |
| 619 | EMIT 0x0F EMIT 0xF4 EMIT 0xF8 \ |
| 620 | EMIT 0x0F EMIT 0xD4 EMIT 0xCA \ |
| 621 | EMIT 0x0F EMIT 0x6E EMIT 0x5F EMIT 0x04 \ |
| 622 | EMIT 0x0F EMIT 0xD4 EMIT 0xDC \ |
| 623 | EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x08 \ |
| 624 | EMIT 0x0F EMIT 0xD4 EMIT 0xEE \ |
| 625 | EMIT 0x0F EMIT 0x6E EMIT 0x67 EMIT 0x0C \ |
| 626 | EMIT 0x0F EMIT 0xD4 EMIT 0xFC \ |
| 627 | EMIT 0x0F EMIT 0x7E EMIT 0x0F \ |
| 628 | EMIT 0x0F EMIT 0x6E EMIT 0x56 EMIT 0x10 \ |
| 629 | EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \ |
| 630 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 631 | EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x14 \ |
| 632 | EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \ |
| 633 | EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ |
| 634 | EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x18 \ |
| 635 | EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \ |
| 636 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x04 \ |
| 637 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 638 | EMIT 0x0F EMIT 0x6E EMIT 0x5E EMIT 0x1C \ |
| 639 | EMIT 0x0F EMIT 0xF4 EMIT 0xD8 \ |
| 640 | EMIT 0x0F EMIT 0xD4 EMIT 0xCD \ |
| 641 | EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x10 \ |
| 642 | EMIT 0x0F EMIT 0xD4 EMIT 0xD5 \ |
| 643 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x08 \ |
| 644 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 645 | EMIT 0x0F EMIT 0xD4 EMIT 0xCF \ |
| 646 | EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x14 \ |
| 647 | EMIT 0x0F EMIT 0xD4 EMIT 0xE5 \ |
| 648 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x0C \ |
| 649 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 650 | EMIT 0x0F EMIT 0xD4 EMIT 0xCA \ |
| 651 | EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x18 \ |
| 652 | EMIT 0x0F EMIT 0xD4 EMIT 0xF5 \ |
| 653 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x10 \ |
| 654 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 655 | EMIT 0x0F EMIT 0xD4 EMIT 0xCC \ |
| 656 | EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x1C \ |
| 657 | EMIT 0x0F EMIT 0xD4 EMIT 0xDD \ |
| 658 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x14 \ |
| 659 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 660 | EMIT 0x0F EMIT 0xD4 EMIT 0xCE \ |
| 661 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x18 \ |
| 662 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 663 | EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ |
| 664 | EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x1C \ |
| 665 | EMIT 0x83 EMIT 0xC7 EMIT 0x20 \ |
| 666 | EMIT 0x83 EMIT 0xC6 EMIT 0x20 \ |
| 667 | EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ |
| 668 | EMIT 0x0F EMIT 0x7E EMIT 0xC9 |
| 669 | |
| 670 | #define MULADDC_STOP \ |
| 671 | EMIT 0x0F EMIT 0x77 \ |
| 672 | __asm mov c, ecx \ |
| 673 | __asm mov d, edi \ |
| 674 | __asm mov s, esi \ |
| 675 | |
| 676 | #else |
| 677 | |
| 678 | #define MULADDC_STOP \ |
| 679 | __asm mov c, ecx \ |
| 680 | __asm mov d, edi \ |
| 681 | __asm mov s, esi \ |
| 682 | |
| 683 | #endif /* SSE2 */ |
| 684 | #endif /* MSVC */ |
| 685 | |
| 686 | #endif /* POLARSSL_HAVE_ASM */ |
| 687 | |
| 688 | #if !defined(MULADDC_CORE) |
| 689 | #if defined(POLARSSL_HAVE_LONGLONG) |
| 690 | |
| 691 | #define MULADDC_INIT \ |
| 692 | { \ |
| 693 | t_dbl r; \ |
| 694 | t_int r0, r1; |
| 695 | |
| 696 | #define MULADDC_CORE \ |
| 697 | r = *(s++) * (t_dbl) b; \ |
| 698 | r0 = r; \ |
| 699 | r1 = r >> biL; \ |
| 700 | r0 += c; r1 += (r0 < c); \ |
| 701 | r0 += *d; r1 += (r0 < *d); \ |
| 702 | c = r1; *(d++) = r0; |
| 703 | |
| 704 | #define MULADDC_STOP \ |
| 705 | } |
| 706 | |
| 707 | #else |
| 708 | #define MULADDC_INIT \ |
| 709 | { \ |
| 710 | t_int s0, s1, b0, b1; \ |
| 711 | t_int r0, r1, rx, ry; \ |
| 712 | b0 = ( b << biH ) >> biH; \ |
| 713 | b1 = ( b >> biH ); |
| 714 | |
| 715 | #define MULADDC_CORE \ |
| 716 | s0 = ( *s << biH ) >> biH; \ |
| 717 | s1 = ( *s >> biH ); s++; \ |
| 718 | rx = s0 * b1; r0 = s0 * b0; \ |
| 719 | ry = s1 * b0; r1 = s1 * b1; \ |
| 720 | r1 += ( rx >> biH ); \ |
| 721 | r1 += ( ry >> biH ); \ |
| 722 | rx <<= biH; ry <<= biH; \ |
| 723 | r0 += rx; r1 += (r0 < rx); \ |
| 724 | r0 += ry; r1 += (r0 < ry); \ |
| 725 | r0 += c; r1 += (r0 < c); \ |
| 726 | r0 += *d; r1 += (r0 < *d); \ |
| 727 | c = r1; *(d++) = r0; |
| 728 | |
| 729 | #define MULADDC_STOP \ |
| 730 | } |
| 731 | |
| 732 | #endif /* C (generic) */ |
| 733 | #endif /* C (longlong) */ |
| 734 | |
| 735 | #endif /* bn_mul.h */ |