Kenji Arai
/
mbed-os_TYBLE16
Important changes to repositories hosted on mbed.com
Mbed hosted mercurial repositories are deprecated and are due to be permanently deleted in July 2026.
To keep a copy of this software download the repository Zip archive or clone locally using Mercurial.
It is also possible to export all your personal repositories from the account settings page.
Dependents: TYBLE16_simple_data_logger TYBLE16_MP3_Air
Embed:
(wiki syntax)
Show/hide line numbers
bn_mul.h
Go to the documentation of this file.
00001 /** 00002 * \file bn_mul.h 00003 * 00004 * \brief Multi-precision integer library 00005 */ 00006 /* 00007 * Copyright (C) 2006-2015, ARM Limited, All Rights Reserved 00008 * SPDX-License-Identifier: Apache-2.0 00009 * 00010 * Licensed under the Apache License, Version 2.0 (the "License"); you may 00011 * not use this file except in compliance with the License. 00012 * You may obtain a copy of the License at 00013 * 00014 * http://www.apache.org/licenses/LICENSE-2.0 00015 * 00016 * Unless required by applicable law or agreed to in writing, software 00017 * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT 00018 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 00019 * See the License for the specific language governing permissions and 00020 * limitations under the License. 00021 * 00022 * This file is part of mbed TLS (https://tls.mbed.org) 00023 */ 00024 /* 00025 * Multiply source vector [s] with b, add result 00026 * to destination vector [d] and set carry c. 00027 * 00028 * Currently supports: 00029 * 00030 * . IA-32 (386+) . AMD64 / EM64T 00031 * . IA-32 (SSE2) . Motorola 68000 00032 * . PowerPC, 32-bit . MicroBlaze 00033 * . PowerPC, 64-bit . TriCore 00034 * . SPARC v8 . ARM v3+ 00035 * . Alpha . MIPS32 00036 * . C, longlong . C, generic 00037 */ 00038 #ifndef MBEDTLS_BN_MUL_H 00039 #define MBEDTLS_BN_MUL_H 00040 00041 #if !defined(MBEDTLS_CONFIG_FILE) 00042 #include "mbedtls/config.h" 00043 #else 00044 #include MBEDTLS_CONFIG_FILE 00045 #endif 00046 00047 #include "mbedtls/bignum.h" 00048 00049 #if defined(MBEDTLS_HAVE_ASM) 00050 00051 #ifndef asm 00052 #define asm __asm 00053 #endif 00054 00055 /* armcc5 --gnu defines __GNUC__ but doesn't support GNU's extended asm */ 00056 #if defined(__GNUC__) && \ 00057 ( !defined(__ARMCC_VERSION) || __ARMCC_VERSION >= 6000000 ) 00058 00059 /* 00060 * Disable use of the i386 assembly code below if option -O0, to disable all 00061 * compiler optimisations, is passed, detected with __OPTIMIZE__ 00062 * This is done as the number of registers used in the assembly code doesn't 00063 * work with the -O0 option. 00064 */ 00065 #if defined(__i386__) && defined(__OPTIMIZE__) 00066 00067 #define MULADDC_INIT \ 00068 asm( \ 00069 "movl %%ebx, %0 \n\t" \ 00070 "movl %5, %%esi \n\t" \ 00071 "movl %6, %%edi \n\t" \ 00072 "movl %7, %%ecx \n\t" \ 00073 "movl %8, %%ebx \n\t" 00074 00075 #define MULADDC_CORE \ 00076 "lodsl \n\t" \ 00077 "mull %%ebx \n\t" \ 00078 "addl %%ecx, %%eax \n\t" \ 00079 "adcl $0, %%edx \n\t" \ 00080 "addl (%%edi), %%eax \n\t" \ 00081 "adcl $0, %%edx \n\t" \ 00082 "movl %%edx, %%ecx \n\t" \ 00083 "stosl \n\t" 00084 00085 #if defined(MBEDTLS_HAVE_SSE2) 00086 00087 #define MULADDC_HUIT \ 00088 "movd %%ecx, %%mm1 \n\t" \ 00089 "movd %%ebx, %%mm0 \n\t" \ 00090 "movd (%%edi), %%mm3 \n\t" \ 00091 "paddq %%mm3, %%mm1 \n\t" \ 00092 "movd (%%esi), %%mm2 \n\t" \ 00093 "pmuludq %%mm0, %%mm2 \n\t" \ 00094 "movd 4(%%esi), %%mm4 \n\t" \ 00095 "pmuludq %%mm0, %%mm4 \n\t" \ 00096 "movd 8(%%esi), %%mm6 \n\t" \ 00097 "pmuludq %%mm0, %%mm6 \n\t" \ 00098 "movd 12(%%esi), %%mm7 \n\t" \ 00099 "pmuludq %%mm0, %%mm7 \n\t" \ 00100 "paddq %%mm2, %%mm1 \n\t" \ 00101 "movd 4(%%edi), %%mm3 \n\t" \ 00102 "paddq %%mm4, %%mm3 \n\t" \ 00103 "movd 8(%%edi), %%mm5 \n\t" \ 00104 "paddq %%mm6, %%mm5 \n\t" \ 00105 "movd 12(%%edi), %%mm4 \n\t" \ 00106 "paddq %%mm4, %%mm7 \n\t" \ 00107 "movd %%mm1, (%%edi) \n\t" \ 00108 "movd 16(%%esi), %%mm2 \n\t" \ 00109 "pmuludq %%mm0, %%mm2 \n\t" \ 00110 "psrlq $32, %%mm1 \n\t" \ 00111 "movd 20(%%esi), %%mm4 \n\t" \ 00112 "pmuludq %%mm0, %%mm4 \n\t" \ 00113 "paddq %%mm3, %%mm1 \n\t" \ 00114 "movd 24(%%esi), %%mm6 \n\t" \ 00115 "pmuludq %%mm0, %%mm6 \n\t" \ 00116 "movd %%mm1, 4(%%edi) \n\t" \ 00117 "psrlq $32, %%mm1 \n\t" \ 00118 "movd 28(%%esi), %%mm3 \n\t" \ 00119 "pmuludq %%mm0, %%mm3 \n\t" \ 00120 "paddq %%mm5, %%mm1 \n\t" \ 00121 "movd 16(%%edi), %%mm5 \n\t" \ 00122 "paddq %%mm5, %%mm2 \n\t" \ 00123 "movd %%mm1, 8(%%edi) \n\t" \ 00124 "psrlq $32, %%mm1 \n\t" \ 00125 "paddq %%mm7, %%mm1 \n\t" \ 00126 "movd 20(%%edi), %%mm5 \n\t" \ 00127 "paddq %%mm5, %%mm4 \n\t" \ 00128 "movd %%mm1, 12(%%edi) \n\t" \ 00129 "psrlq $32, %%mm1 \n\t" \ 00130 "paddq %%mm2, %%mm1 \n\t" \ 00131 "movd 24(%%edi), %%mm5 \n\t" \ 00132 "paddq %%mm5, %%mm6 \n\t" \ 00133 "movd %%mm1, 16(%%edi) \n\t" \ 00134 "psrlq $32, %%mm1 \n\t" \ 00135 "paddq %%mm4, %%mm1 \n\t" \ 00136 "movd 28(%%edi), %%mm5 \n\t" \ 00137 "paddq %%mm5, %%mm3 \n\t" \ 00138 "movd %%mm1, 20(%%edi) \n\t" \ 00139 "psrlq $32, %%mm1 \n\t" \ 00140 "paddq %%mm6, %%mm1 \n\t" \ 00141 "movd %%mm1, 24(%%edi) \n\t" \ 00142 "psrlq $32, %%mm1 \n\t" \ 00143 "paddq %%mm3, %%mm1 \n\t" \ 00144 "movd %%mm1, 28(%%edi) \n\t" \ 00145 "addl $32, %%edi \n\t" \ 00146 "addl $32, %%esi \n\t" \ 00147 "psrlq $32, %%mm1 \n\t" \ 00148 "movd %%mm1, %%ecx \n\t" 00149 00150 #define MULADDC_STOP \ 00151 "emms \n\t" \ 00152 "movl %4, %%ebx \n\t" \ 00153 "movl %%ecx, %1 \n\t" \ 00154 "movl %%edi, %2 \n\t" \ 00155 "movl %%esi, %3 \n\t" \ 00156 : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \ 00157 : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \ 00158 : "eax", "ebx", "ecx", "edx", "esi", "edi" \ 00159 ); 00160 00161 #else 00162 00163 #define MULADDC_STOP \ 00164 "movl %4, %%ebx \n\t" \ 00165 "movl %%ecx, %1 \n\t" \ 00166 "movl %%edi, %2 \n\t" \ 00167 "movl %%esi, %3 \n\t" \ 00168 : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \ 00169 : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \ 00170 : "eax", "ebx", "ecx", "edx", "esi", "edi" \ 00171 ); 00172 #endif /* SSE2 */ 00173 #endif /* i386 */ 00174 00175 #if defined(__amd64__) || defined (__x86_64__) 00176 00177 #define MULADDC_INIT \ 00178 asm( \ 00179 "xorq %%r8, %%r8\n" 00180 00181 #define MULADDC_CORE \ 00182 "movq (%%rsi), %%rax\n" \ 00183 "mulq %%rbx\n" \ 00184 "addq $8, %%rsi\n" \ 00185 "addq %%rcx, %%rax\n" \ 00186 "movq %%r8, %%rcx\n" \ 00187 "adcq $0, %%rdx\n" \ 00188 "nop \n" \ 00189 "addq %%rax, (%%rdi)\n" \ 00190 "adcq %%rdx, %%rcx\n" \ 00191 "addq $8, %%rdi\n" 00192 00193 #define MULADDC_STOP \ 00194 : "+c" (c), "+D" (d), "+S" (s) \ 00195 : "b" (b) \ 00196 : "rax", "rdx", "r8" \ 00197 ); 00198 00199 #endif /* AMD64 */ 00200 00201 #if defined(__aarch64__) 00202 00203 #define MULADDC_INIT \ 00204 asm( 00205 00206 #define MULADDC_CORE \ 00207 "ldr x4, [%2], #8 \n\t" \ 00208 "ldr x5, [%1] \n\t" \ 00209 "mul x6, x4, %3 \n\t" \ 00210 "umulh x7, x4, %3 \n\t" \ 00211 "adds x5, x5, x6 \n\t" \ 00212 "adc x7, x7, xzr \n\t" \ 00213 "adds x5, x5, %0 \n\t" \ 00214 "adc %0, x7, xzr \n\t" \ 00215 "str x5, [%1], #8 \n\t" 00216 00217 #define MULADDC_STOP \ 00218 : "+r" (c), "+r" (d), "+r" (s) \ 00219 : "r" (b) \ 00220 : "x4", "x5", "x6", "x7", "cc" \ 00221 ); 00222 00223 #endif /* Aarch64 */ 00224 00225 #if defined(__mc68020__) || defined(__mcpu32__) 00226 00227 #define MULADDC_INIT \ 00228 asm( \ 00229 "movl %3, %%a2 \n\t" \ 00230 "movl %4, %%a3 \n\t" \ 00231 "movl %5, %%d3 \n\t" \ 00232 "movl %6, %%d2 \n\t" \ 00233 "moveq #0, %%d0 \n\t" 00234 00235 #define MULADDC_CORE \ 00236 "movel %%a2@+, %%d1 \n\t" \ 00237 "mulul %%d2, %%d4:%%d1 \n\t" \ 00238 "addl %%d3, %%d1 \n\t" \ 00239 "addxl %%d0, %%d4 \n\t" \ 00240 "moveq #0, %%d3 \n\t" \ 00241 "addl %%d1, %%a3@+ \n\t" \ 00242 "addxl %%d4, %%d3 \n\t" 00243 00244 #define MULADDC_STOP \ 00245 "movl %%d3, %0 \n\t" \ 00246 "movl %%a3, %1 \n\t" \ 00247 "movl %%a2, %2 \n\t" \ 00248 : "=m" (c), "=m" (d), "=m" (s) \ 00249 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00250 : "d0", "d1", "d2", "d3", "d4", "a2", "a3" \ 00251 ); 00252 00253 #define MULADDC_HUIT \ 00254 "movel %%a2@+, %%d1 \n\t" \ 00255 "mulul %%d2, %%d4:%%d1 \n\t" \ 00256 "addxl %%d3, %%d1 \n\t" \ 00257 "addxl %%d0, %%d4 \n\t" \ 00258 "addl %%d1, %%a3@+ \n\t" \ 00259 "movel %%a2@+, %%d1 \n\t" \ 00260 "mulul %%d2, %%d3:%%d1 \n\t" \ 00261 "addxl %%d4, %%d1 \n\t" \ 00262 "addxl %%d0, %%d3 \n\t" \ 00263 "addl %%d1, %%a3@+ \n\t" \ 00264 "movel %%a2@+, %%d1 \n\t" \ 00265 "mulul %%d2, %%d4:%%d1 \n\t" \ 00266 "addxl %%d3, %%d1 \n\t" \ 00267 "addxl %%d0, %%d4 \n\t" \ 00268 "addl %%d1, %%a3@+ \n\t" \ 00269 "movel %%a2@+, %%d1 \n\t" \ 00270 "mulul %%d2, %%d3:%%d1 \n\t" \ 00271 "addxl %%d4, %%d1 \n\t" \ 00272 "addxl %%d0, %%d3 \n\t" \ 00273 "addl %%d1, %%a3@+ \n\t" \ 00274 "movel %%a2@+, %%d1 \n\t" \ 00275 "mulul %%d2, %%d4:%%d1 \n\t" \ 00276 "addxl %%d3, %%d1 \n\t" \ 00277 "addxl %%d0, %%d4 \n\t" \ 00278 "addl %%d1, %%a3@+ \n\t" \ 00279 "movel %%a2@+, %%d1 \n\t" \ 00280 "mulul %%d2, %%d3:%%d1 \n\t" \ 00281 "addxl %%d4, %%d1 \n\t" \ 00282 "addxl %%d0, %%d3 \n\t" \ 00283 "addl %%d1, %%a3@+ \n\t" \ 00284 "movel %%a2@+, %%d1 \n\t" \ 00285 "mulul %%d2, %%d4:%%d1 \n\t" \ 00286 "addxl %%d3, %%d1 \n\t" \ 00287 "addxl %%d0, %%d4 \n\t" \ 00288 "addl %%d1, %%a3@+ \n\t" \ 00289 "movel %%a2@+, %%d1 \n\t" \ 00290 "mulul %%d2, %%d3:%%d1 \n\t" \ 00291 "addxl %%d4, %%d1 \n\t" \ 00292 "addxl %%d0, %%d3 \n\t" \ 00293 "addl %%d1, %%a3@+ \n\t" \ 00294 "addxl %%d0, %%d3 \n\t" 00295 00296 #endif /* MC68000 */ 00297 00298 #if defined(__powerpc64__) || defined(__ppc64__) 00299 00300 #if defined(__MACH__) && defined(__APPLE__) 00301 00302 #define MULADDC_INIT \ 00303 asm( \ 00304 "ld r3, %3 \n\t" \ 00305 "ld r4, %4 \n\t" \ 00306 "ld r5, %5 \n\t" \ 00307 "ld r6, %6 \n\t" \ 00308 "addi r3, r3, -8 \n\t" \ 00309 "addi r4, r4, -8 \n\t" \ 00310 "addic r5, r5, 0 \n\t" 00311 00312 #define MULADDC_CORE \ 00313 "ldu r7, 8(r3) \n\t" \ 00314 "mulld r8, r7, r6 \n\t" \ 00315 "mulhdu r9, r7, r6 \n\t" \ 00316 "adde r8, r8, r5 \n\t" \ 00317 "ld r7, 8(r4) \n\t" \ 00318 "addze r5, r9 \n\t" \ 00319 "addc r8, r8, r7 \n\t" \ 00320 "stdu r8, 8(r4) \n\t" 00321 00322 #define MULADDC_STOP \ 00323 "addze r5, r5 \n\t" \ 00324 "addi r4, r4, 8 \n\t" \ 00325 "addi r3, r3, 8 \n\t" \ 00326 "std r5, %0 \n\t" \ 00327 "std r4, %1 \n\t" \ 00328 "std r3, %2 \n\t" \ 00329 : "=m" (c), "=m" (d), "=m" (s) \ 00330 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00331 : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \ 00332 ); 00333 00334 00335 #else /* __MACH__ && __APPLE__ */ 00336 00337 #define MULADDC_INIT \ 00338 asm( \ 00339 "ld %%r3, %3 \n\t" \ 00340 "ld %%r4, %4 \n\t" \ 00341 "ld %%r5, %5 \n\t" \ 00342 "ld %%r6, %6 \n\t" \ 00343 "addi %%r3, %%r3, -8 \n\t" \ 00344 "addi %%r4, %%r4, -8 \n\t" \ 00345 "addic %%r5, %%r5, 0 \n\t" 00346 00347 #define MULADDC_CORE \ 00348 "ldu %%r7, 8(%%r3) \n\t" \ 00349 "mulld %%r8, %%r7, %%r6 \n\t" \ 00350 "mulhdu %%r9, %%r7, %%r6 \n\t" \ 00351 "adde %%r8, %%r8, %%r5 \n\t" \ 00352 "ld %%r7, 8(%%r4) \n\t" \ 00353 "addze %%r5, %%r9 \n\t" \ 00354 "addc %%r8, %%r8, %%r7 \n\t" \ 00355 "stdu %%r8, 8(%%r4) \n\t" 00356 00357 #define MULADDC_STOP \ 00358 "addze %%r5, %%r5 \n\t" \ 00359 "addi %%r4, %%r4, 8 \n\t" \ 00360 "addi %%r3, %%r3, 8 \n\t" \ 00361 "std %%r5, %0 \n\t" \ 00362 "std %%r4, %1 \n\t" \ 00363 "std %%r3, %2 \n\t" \ 00364 : "=m" (c), "=m" (d), "=m" (s) \ 00365 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00366 : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \ 00367 ); 00368 00369 #endif /* __MACH__ && __APPLE__ */ 00370 00371 #elif defined(__powerpc__) || defined(__ppc__) /* end PPC64/begin PPC32 */ 00372 00373 #if defined(__MACH__) && defined(__APPLE__) 00374 00375 #define MULADDC_INIT \ 00376 asm( \ 00377 "lwz r3, %3 \n\t" \ 00378 "lwz r4, %4 \n\t" \ 00379 "lwz r5, %5 \n\t" \ 00380 "lwz r6, %6 \n\t" \ 00381 "addi r3, r3, -4 \n\t" \ 00382 "addi r4, r4, -4 \n\t" \ 00383 "addic r5, r5, 0 \n\t" 00384 00385 #define MULADDC_CORE \ 00386 "lwzu r7, 4(r3) \n\t" \ 00387 "mullw r8, r7, r6 \n\t" \ 00388 "mulhwu r9, r7, r6 \n\t" \ 00389 "adde r8, r8, r5 \n\t" \ 00390 "lwz r7, 4(r4) \n\t" \ 00391 "addze r5, r9 \n\t" \ 00392 "addc r8, r8, r7 \n\t" \ 00393 "stwu r8, 4(r4) \n\t" 00394 00395 #define MULADDC_STOP \ 00396 "addze r5, r5 \n\t" \ 00397 "addi r4, r4, 4 \n\t" \ 00398 "addi r3, r3, 4 \n\t" \ 00399 "stw r5, %0 \n\t" \ 00400 "stw r4, %1 \n\t" \ 00401 "stw r3, %2 \n\t" \ 00402 : "=m" (c), "=m" (d), "=m" (s) \ 00403 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00404 : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \ 00405 ); 00406 00407 #else /* __MACH__ && __APPLE__ */ 00408 00409 #define MULADDC_INIT \ 00410 asm( \ 00411 "lwz %%r3, %3 \n\t" \ 00412 "lwz %%r4, %4 \n\t" \ 00413 "lwz %%r5, %5 \n\t" \ 00414 "lwz %%r6, %6 \n\t" \ 00415 "addi %%r3, %%r3, -4 \n\t" \ 00416 "addi %%r4, %%r4, -4 \n\t" \ 00417 "addic %%r5, %%r5, 0 \n\t" 00418 00419 #define MULADDC_CORE \ 00420 "lwzu %%r7, 4(%%r3) \n\t" \ 00421 "mullw %%r8, %%r7, %%r6 \n\t" \ 00422 "mulhwu %%r9, %%r7, %%r6 \n\t" \ 00423 "adde %%r8, %%r8, %%r5 \n\t" \ 00424 "lwz %%r7, 4(%%r4) \n\t" \ 00425 "addze %%r5, %%r9 \n\t" \ 00426 "addc %%r8, %%r8, %%r7 \n\t" \ 00427 "stwu %%r8, 4(%%r4) \n\t" 00428 00429 #define MULADDC_STOP \ 00430 "addze %%r5, %%r5 \n\t" \ 00431 "addi %%r4, %%r4, 4 \n\t" \ 00432 "addi %%r3, %%r3, 4 \n\t" \ 00433 "stw %%r5, %0 \n\t" \ 00434 "stw %%r4, %1 \n\t" \ 00435 "stw %%r3, %2 \n\t" \ 00436 : "=m" (c), "=m" (d), "=m" (s) \ 00437 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00438 : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \ 00439 ); 00440 00441 #endif /* __MACH__ && __APPLE__ */ 00442 00443 #endif /* PPC32 */ 00444 00445 /* 00446 * The Sparc(64) assembly is reported to be broken. 00447 * Disable it for now, until we're able to fix it. 00448 */ 00449 #if 0 && defined(__sparc__) 00450 #if defined(__sparc64__) 00451 00452 #define MULADDC_INIT \ 00453 asm( \ 00454 "ldx %3, %%o0 \n\t" \ 00455 "ldx %4, %%o1 \n\t" \ 00456 "ld %5, %%o2 \n\t" \ 00457 "ld %6, %%o3 \n\t" 00458 00459 #define MULADDC_CORE \ 00460 "ld [%%o0], %%o4 \n\t" \ 00461 "inc 4, %%o0 \n\t" \ 00462 "ld [%%o1], %%o5 \n\t" \ 00463 "umul %%o3, %%o4, %%o4 \n\t" \ 00464 "addcc %%o4, %%o2, %%o4 \n\t" \ 00465 "rd %%y, %%g1 \n\t" \ 00466 "addx %%g1, 0, %%g1 \n\t" \ 00467 "addcc %%o4, %%o5, %%o4 \n\t" \ 00468 "st %%o4, [%%o1] \n\t" \ 00469 "addx %%g1, 0, %%o2 \n\t" \ 00470 "inc 4, %%o1 \n\t" 00471 00472 #define MULADDC_STOP \ 00473 "st %%o2, %0 \n\t" \ 00474 "stx %%o1, %1 \n\t" \ 00475 "stx %%o0, %2 \n\t" \ 00476 : "=m" (c), "=m" (d), "=m" (s) \ 00477 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00478 : "g1", "o0", "o1", "o2", "o3", "o4", \ 00479 "o5" \ 00480 ); 00481 00482 #else /* __sparc64__ */ 00483 00484 #define MULADDC_INIT \ 00485 asm( \ 00486 "ld %3, %%o0 \n\t" \ 00487 "ld %4, %%o1 \n\t" \ 00488 "ld %5, %%o2 \n\t" \ 00489 "ld %6, %%o3 \n\t" 00490 00491 #define MULADDC_CORE \ 00492 "ld [%%o0], %%o4 \n\t" \ 00493 "inc 4, %%o0 \n\t" \ 00494 "ld [%%o1], %%o5 \n\t" \ 00495 "umul %%o3, %%o4, %%o4 \n\t" \ 00496 "addcc %%o4, %%o2, %%o4 \n\t" \ 00497 "rd %%y, %%g1 \n\t" \ 00498 "addx %%g1, 0, %%g1 \n\t" \ 00499 "addcc %%o4, %%o5, %%o4 \n\t" \ 00500 "st %%o4, [%%o1] \n\t" \ 00501 "addx %%g1, 0, %%o2 \n\t" \ 00502 "inc 4, %%o1 \n\t" 00503 00504 #define MULADDC_STOP \ 00505 "st %%o2, %0 \n\t" \ 00506 "st %%o1, %1 \n\t" \ 00507 "st %%o0, %2 \n\t" \ 00508 : "=m" (c), "=m" (d), "=m" (s) \ 00509 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00510 : "g1", "o0", "o1", "o2", "o3", "o4", \ 00511 "o5" \ 00512 ); 00513 00514 #endif /* __sparc64__ */ 00515 #endif /* __sparc__ */ 00516 00517 #if defined(__microblaze__) || defined(microblaze) 00518 00519 #define MULADDC_INIT \ 00520 asm( \ 00521 "lwi r3, %3 \n\t" \ 00522 "lwi r4, %4 \n\t" \ 00523 "lwi r5, %5 \n\t" \ 00524 "lwi r6, %6 \n\t" \ 00525 "andi r7, r6, 0xffff \n\t" \ 00526 "bsrli r6, r6, 16 \n\t" 00527 00528 #define MULADDC_CORE \ 00529 "lhui r8, r3, 0 \n\t" \ 00530 "addi r3, r3, 2 \n\t" \ 00531 "lhui r9, r3, 0 \n\t" \ 00532 "addi r3, r3, 2 \n\t" \ 00533 "mul r10, r9, r6 \n\t" \ 00534 "mul r11, r8, r7 \n\t" \ 00535 "mul r12, r9, r7 \n\t" \ 00536 "mul r13, r8, r6 \n\t" \ 00537 "bsrli r8, r10, 16 \n\t" \ 00538 "bsrli r9, r11, 16 \n\t" \ 00539 "add r13, r13, r8 \n\t" \ 00540 "add r13, r13, r9 \n\t" \ 00541 "bslli r10, r10, 16 \n\t" \ 00542 "bslli r11, r11, 16 \n\t" \ 00543 "add r12, r12, r10 \n\t" \ 00544 "addc r13, r13, r0 \n\t" \ 00545 "add r12, r12, r11 \n\t" \ 00546 "addc r13, r13, r0 \n\t" \ 00547 "lwi r10, r4, 0 \n\t" \ 00548 "add r12, r12, r10 \n\t" \ 00549 "addc r13, r13, r0 \n\t" \ 00550 "add r12, r12, r5 \n\t" \ 00551 "addc r5, r13, r0 \n\t" \ 00552 "swi r12, r4, 0 \n\t" \ 00553 "addi r4, r4, 4 \n\t" 00554 00555 #define MULADDC_STOP \ 00556 "swi r5, %0 \n\t" \ 00557 "swi r4, %1 \n\t" \ 00558 "swi r3, %2 \n\t" \ 00559 : "=m" (c), "=m" (d), "=m" (s) \ 00560 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00561 : "r3", "r4", "r5", "r6", "r7", "r8", \ 00562 "r9", "r10", "r11", "r12", "r13" \ 00563 ); 00564 00565 #endif /* MicroBlaze */ 00566 00567 #if defined(__tricore__) 00568 00569 #define MULADDC_INIT \ 00570 asm( \ 00571 "ld.a %%a2, %3 \n\t" \ 00572 "ld.a %%a3, %4 \n\t" \ 00573 "ld.w %%d4, %5 \n\t" \ 00574 "ld.w %%d1, %6 \n\t" \ 00575 "xor %%d5, %%d5 \n\t" 00576 00577 #define MULADDC_CORE \ 00578 "ld.w %%d0, [%%a2+] \n\t" \ 00579 "madd.u %%e2, %%e4, %%d0, %%d1 \n\t" \ 00580 "ld.w %%d0, [%%a3] \n\t" \ 00581 "addx %%d2, %%d2, %%d0 \n\t" \ 00582 "addc %%d3, %%d3, 0 \n\t" \ 00583 "mov %%d4, %%d3 \n\t" \ 00584 "st.w [%%a3+], %%d2 \n\t" 00585 00586 #define MULADDC_STOP \ 00587 "st.w %0, %%d4 \n\t" \ 00588 "st.a %1, %%a3 \n\t" \ 00589 "st.a %2, %%a2 \n\t" \ 00590 : "=m" (c), "=m" (d), "=m" (s) \ 00591 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00592 : "d0", "d1", "e2", "d4", "a2", "a3" \ 00593 ); 00594 00595 #endif /* TriCore */ 00596 00597 /* 00598 * Note, gcc -O0 by default uses r7 for the frame pointer, so it complains about 00599 * our use of r7 below, unless -fomit-frame-pointer is passed. 00600 * 00601 * On the other hand, -fomit-frame-pointer is implied by any -Ox options with 00602 * x !=0, which we can detect using __OPTIMIZE__ (which is also defined by 00603 * clang and armcc5 under the same conditions). 00604 * 00605 * So, only use the optimized assembly below for optimized build, which avoids 00606 * the build error and is pretty reasonable anyway. 00607 */ 00608 #if defined(__GNUC__) && !defined(__OPTIMIZE__) 00609 #define MULADDC_CANNOT_USE_R7 00610 #endif 00611 00612 #if defined(__arm__) && !defined(MULADDC_CANNOT_USE_R7) 00613 00614 #if defined(__thumb__) && !defined(__thumb2__) 00615 00616 #define MULADDC_INIT \ 00617 asm( \ 00618 "ldr r0, %3 \n\t" \ 00619 "ldr r1, %4 \n\t" \ 00620 "ldr r2, %5 \n\t" \ 00621 "ldr r3, %6 \n\t" \ 00622 "lsr r7, r3, #16 \n\t" \ 00623 "mov r9, r7 \n\t" \ 00624 "lsl r7, r3, #16 \n\t" \ 00625 "lsr r7, r7, #16 \n\t" \ 00626 "mov r8, r7 \n\t" 00627 00628 #define MULADDC_CORE \ 00629 "ldmia r0!, {r6} \n\t" \ 00630 "lsr r7, r6, #16 \n\t" \ 00631 "lsl r6, r6, #16 \n\t" \ 00632 "lsr r6, r6, #16 \n\t" \ 00633 "mov r4, r8 \n\t" \ 00634 "mul r4, r6 \n\t" \ 00635 "mov r3, r9 \n\t" \ 00636 "mul r6, r3 \n\t" \ 00637 "mov r5, r9 \n\t" \ 00638 "mul r5, r7 \n\t" \ 00639 "mov r3, r8 \n\t" \ 00640 "mul r7, r3 \n\t" \ 00641 "lsr r3, r6, #16 \n\t" \ 00642 "add r5, r5, r3 \n\t" \ 00643 "lsr r3, r7, #16 \n\t" \ 00644 "add r5, r5, r3 \n\t" \ 00645 "add r4, r4, r2 \n\t" \ 00646 "mov r2, #0 \n\t" \ 00647 "adc r5, r2 \n\t" \ 00648 "lsl r3, r6, #16 \n\t" \ 00649 "add r4, r4, r3 \n\t" \ 00650 "adc r5, r2 \n\t" \ 00651 "lsl r3, r7, #16 \n\t" \ 00652 "add r4, r4, r3 \n\t" \ 00653 "adc r5, r2 \n\t" \ 00654 "ldr r3, [r1] \n\t" \ 00655 "add r4, r4, r3 \n\t" \ 00656 "adc r2, r5 \n\t" \ 00657 "stmia r1!, {r4} \n\t" 00658 00659 #define MULADDC_STOP \ 00660 "str r2, %0 \n\t" \ 00661 "str r1, %1 \n\t" \ 00662 "str r0, %2 \n\t" \ 00663 : "=m" (c), "=m" (d), "=m" (s) \ 00664 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00665 : "r0", "r1", "r2", "r3", "r4", "r5", \ 00666 "r6", "r7", "r8", "r9", "cc" \ 00667 ); 00668 00669 #elif (__ARM_ARCH >= 6) && \ 00670 defined (__ARM_FEATURE_DSP) && (__ARM_FEATURE_DSP == 1) 00671 00672 #define MULADDC_INIT \ 00673 asm( 00674 00675 #define MULADDC_CORE \ 00676 "ldr r0, [%0], #4 \n\t" \ 00677 "ldr r1, [%1] \n\t" \ 00678 "umaal r1, %2, %3, r0 \n\t" \ 00679 "str r1, [%1], #4 \n\t" 00680 00681 #define MULADDC_STOP \ 00682 : "=r" (s), "=r" (d), "=r" (c) \ 00683 : "r" (b), "0" (s), "1" (d), "2" (c) \ 00684 : "r0", "r1", "memory" \ 00685 ); 00686 00687 #else 00688 00689 #define MULADDC_INIT \ 00690 asm( \ 00691 "ldr r0, %3 \n\t" \ 00692 "ldr r1, %4 \n\t" \ 00693 "ldr r2, %5 \n\t" \ 00694 "ldr r3, %6 \n\t" 00695 00696 #define MULADDC_CORE \ 00697 "ldr r4, [r0], #4 \n\t" \ 00698 "mov r5, #0 \n\t" \ 00699 "ldr r6, [r1] \n\t" \ 00700 "umlal r2, r5, r3, r4 \n\t" \ 00701 "adds r7, r6, r2 \n\t" \ 00702 "adc r2, r5, #0 \n\t" \ 00703 "str r7, [r1], #4 \n\t" 00704 00705 #define MULADDC_STOP \ 00706 "str r2, %0 \n\t" \ 00707 "str r1, %1 \n\t" \ 00708 "str r0, %2 \n\t" \ 00709 : "=m" (c), "=m" (d), "=m" (s) \ 00710 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00711 : "r0", "r1", "r2", "r3", "r4", "r5", \ 00712 "r6", "r7", "cc" \ 00713 ); 00714 00715 #endif /* Thumb */ 00716 00717 #endif /* ARMv3 */ 00718 00719 #if defined(__alpha__) 00720 00721 #define MULADDC_INIT \ 00722 asm( \ 00723 "ldq $1, %3 \n\t" \ 00724 "ldq $2, %4 \n\t" \ 00725 "ldq $3, %5 \n\t" \ 00726 "ldq $4, %6 \n\t" 00727 00728 #define MULADDC_CORE \ 00729 "ldq $6, 0($1) \n\t" \ 00730 "addq $1, 8, $1 \n\t" \ 00731 "mulq $6, $4, $7 \n\t" \ 00732 "umulh $6, $4, $6 \n\t" \ 00733 "addq $7, $3, $7 \n\t" \ 00734 "cmpult $7, $3, $3 \n\t" \ 00735 "ldq $5, 0($2) \n\t" \ 00736 "addq $7, $5, $7 \n\t" \ 00737 "cmpult $7, $5, $5 \n\t" \ 00738 "stq $7, 0($2) \n\t" \ 00739 "addq $2, 8, $2 \n\t" \ 00740 "addq $6, $3, $3 \n\t" \ 00741 "addq $5, $3, $3 \n\t" 00742 00743 #define MULADDC_STOP \ 00744 "stq $3, %0 \n\t" \ 00745 "stq $2, %1 \n\t" \ 00746 "stq $1, %2 \n\t" \ 00747 : "=m" (c), "=m" (d), "=m" (s) \ 00748 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00749 : "$1", "$2", "$3", "$4", "$5", "$6", "$7" \ 00750 ); 00751 #endif /* Alpha */ 00752 00753 #if defined(__mips__) && !defined(__mips64) 00754 00755 #define MULADDC_INIT \ 00756 asm( \ 00757 "lw $10, %3 \n\t" \ 00758 "lw $11, %4 \n\t" \ 00759 "lw $12, %5 \n\t" \ 00760 "lw $13, %6 \n\t" 00761 00762 #define MULADDC_CORE \ 00763 "lw $14, 0($10) \n\t" \ 00764 "multu $13, $14 \n\t" \ 00765 "addi $10, $10, 4 \n\t" \ 00766 "mflo $14 \n\t" \ 00767 "mfhi $9 \n\t" \ 00768 "addu $14, $12, $14 \n\t" \ 00769 "lw $15, 0($11) \n\t" \ 00770 "sltu $12, $14, $12 \n\t" \ 00771 "addu $15, $14, $15 \n\t" \ 00772 "sltu $14, $15, $14 \n\t" \ 00773 "addu $12, $12, $9 \n\t" \ 00774 "sw $15, 0($11) \n\t" \ 00775 "addu $12, $12, $14 \n\t" \ 00776 "addi $11, $11, 4 \n\t" 00777 00778 #define MULADDC_STOP \ 00779 "sw $12, %0 \n\t" \ 00780 "sw $11, %1 \n\t" \ 00781 "sw $10, %2 \n\t" \ 00782 : "=m" (c), "=m" (d), "=m" (s) \ 00783 : "m" (s), "m" (d), "m" (c), "m" (b) \ 00784 : "$9", "$10", "$11", "$12", "$13", "$14", "$15", "lo", "hi" \ 00785 ); 00786 00787 #endif /* MIPS */ 00788 #endif /* GNUC */ 00789 00790 #if (defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__) 00791 00792 #define MULADDC_INIT \ 00793 __asm mov esi, s \ 00794 __asm mov edi, d \ 00795 __asm mov ecx, c \ 00796 __asm mov ebx, b 00797 00798 #define MULADDC_CORE \ 00799 __asm lodsd \ 00800 __asm mul ebx \ 00801 __asm add eax, ecx \ 00802 __asm adc edx, 0 \ 00803 __asm add eax, [edi] \ 00804 __asm adc edx, 0 \ 00805 __asm mov ecx, edx \ 00806 __asm stosd 00807 00808 #if defined(MBEDTLS_HAVE_SSE2) 00809 00810 #define EMIT __asm _emit 00811 00812 #define MULADDC_HUIT \ 00813 EMIT 0x0F EMIT 0x6E EMIT 0xC9 \ 00814 EMIT 0x0F EMIT 0x6E EMIT 0xC3 \ 00815 EMIT 0x0F EMIT 0x6E EMIT 0x1F \ 00816 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ 00817 EMIT 0x0F EMIT 0x6E EMIT 0x16 \ 00818 EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \ 00819 EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x04 \ 00820 EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \ 00821 EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x08 \ 00822 EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \ 00823 EMIT 0x0F EMIT 0x6E EMIT 0x7E EMIT 0x0C \ 00824 EMIT 0x0F EMIT 0xF4 EMIT 0xF8 \ 00825 EMIT 0x0F EMIT 0xD4 EMIT 0xCA \ 00826 EMIT 0x0F EMIT 0x6E EMIT 0x5F EMIT 0x04 \ 00827 EMIT 0x0F EMIT 0xD4 EMIT 0xDC \ 00828 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x08 \ 00829 EMIT 0x0F EMIT 0xD4 EMIT 0xEE \ 00830 EMIT 0x0F EMIT 0x6E EMIT 0x67 EMIT 0x0C \ 00831 EMIT 0x0F EMIT 0xD4 EMIT 0xFC \ 00832 EMIT 0x0F EMIT 0x7E EMIT 0x0F \ 00833 EMIT 0x0F EMIT 0x6E EMIT 0x56 EMIT 0x10 \ 00834 EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \ 00835 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00836 EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x14 \ 00837 EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \ 00838 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ 00839 EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x18 \ 00840 EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \ 00841 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x04 \ 00842 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00843 EMIT 0x0F EMIT 0x6E EMIT 0x5E EMIT 0x1C \ 00844 EMIT 0x0F EMIT 0xF4 EMIT 0xD8 \ 00845 EMIT 0x0F EMIT 0xD4 EMIT 0xCD \ 00846 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x10 \ 00847 EMIT 0x0F EMIT 0xD4 EMIT 0xD5 \ 00848 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x08 \ 00849 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00850 EMIT 0x0F EMIT 0xD4 EMIT 0xCF \ 00851 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x14 \ 00852 EMIT 0x0F EMIT 0xD4 EMIT 0xE5 \ 00853 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x0C \ 00854 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00855 EMIT 0x0F EMIT 0xD4 EMIT 0xCA \ 00856 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x18 \ 00857 EMIT 0x0F EMIT 0xD4 EMIT 0xF5 \ 00858 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x10 \ 00859 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00860 EMIT 0x0F EMIT 0xD4 EMIT 0xCC \ 00861 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x1C \ 00862 EMIT 0x0F EMIT 0xD4 EMIT 0xDD \ 00863 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x14 \ 00864 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00865 EMIT 0x0F EMIT 0xD4 EMIT 0xCE \ 00866 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x18 \ 00867 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00868 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \ 00869 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x1C \ 00870 EMIT 0x83 EMIT 0xC7 EMIT 0x20 \ 00871 EMIT 0x83 EMIT 0xC6 EMIT 0x20 \ 00872 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \ 00873 EMIT 0x0F EMIT 0x7E EMIT 0xC9 00874 00875 #define MULADDC_STOP \ 00876 EMIT 0x0F EMIT 0x77 \ 00877 __asm mov c, ecx \ 00878 __asm mov d, edi \ 00879 __asm mov s, esi \ 00880 00881 #else 00882 00883 #define MULADDC_STOP \ 00884 __asm mov c, ecx \ 00885 __asm mov d, edi \ 00886 __asm mov s, esi \ 00887 00888 #endif /* SSE2 */ 00889 #endif /* MSVC */ 00890 00891 #endif /* MBEDTLS_HAVE_ASM */ 00892 00893 #if !defined(MULADDC_CORE) 00894 #if defined(MBEDTLS_HAVE_UDBL) 00895 00896 #define MULADDC_INIT \ 00897 { \ 00898 mbedtls_t_udbl r; \ 00899 mbedtls_mpi_uint r0, r1; 00900 00901 #define MULADDC_CORE \ 00902 r = *(s++) * (mbedtls_t_udbl) b; \ 00903 r0 = (mbedtls_mpi_uint) r; \ 00904 r1 = (mbedtls_mpi_uint)( r >> biL ); \ 00905 r0 += c; r1 += (r0 < c); \ 00906 r0 += *d; r1 += (r0 < *d); \ 00907 c = r1; *(d++) = r0; 00908 00909 #define MULADDC_STOP \ 00910 } 00911 00912 #else 00913 #define MULADDC_INIT \ 00914 { \ 00915 mbedtls_mpi_uint s0, s1, b0, b1; \ 00916 mbedtls_mpi_uint r0, r1, rx, ry; \ 00917 b0 = ( b << biH ) >> biH; \ 00918 b1 = ( b >> biH ); 00919 00920 #define MULADDC_CORE \ 00921 s0 = ( *s << biH ) >> biH; \ 00922 s1 = ( *s >> biH ); s++; \ 00923 rx = s0 * b1; r0 = s0 * b0; \ 00924 ry = s1 * b0; r1 = s1 * b1; \ 00925 r1 += ( rx >> biH ); \ 00926 r1 += ( ry >> biH ); \ 00927 rx <<= biH; ry <<= biH; \ 00928 r0 += rx; r1 += (r0 < rx); \ 00929 r0 += ry; r1 += (r0 < ry); \ 00930 r0 += c; r1 += (r0 < c); \ 00931 r0 += *d; r1 += (r0 < *d); \ 00932 c = r1; *(d++) = r0; 00933 00934 #define MULADDC_STOP \ 00935 } 00936 00937 #endif /* C (generic) */ 00938 #endif /* C (longlong) */ 00939 00940 #endif /* bn_mul.h */
Generated on Tue Jul 12 2022 13:54:03 by
1.7.2