Digi XBee(R) ANSI C Host Library
bn_mul.h
Go to the documentation of this file.
1 
6 /*
7  * Copyright (C) 2006-2015, ARM Limited, All Rights Reserved
8  * SPDX-License-Identifier: Apache-2.0
9  *
10  * Licensed under the Apache License, Version 2.0 (the "License"); you may
11  * not use this file except in compliance with the License.
12  * You may obtain a copy of the License at
13  *
14  * http://www.apache.org/licenses/LICENSE-2.0
15  *
16  * Unless required by applicable law or agreed to in writing, software
17  * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
18  * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
19  * See the License for the specific language governing permissions and
20  * limitations under the License.
21  *
22  * This file is part of mbed TLS (https://tls.mbed.org)
23  */
24 /*
25  * Multiply source vector [s] with b, add result
26  * to destination vector [d] and set carry c.
27  *
28  * Currently supports:
29  *
30  * . IA-32 (386+) . AMD64 / EM64T
31  * . IA-32 (SSE2) . Motorola 68000
32  * . PowerPC, 32-bit . MicroBlaze
33  * . PowerPC, 64-bit . TriCore
34  * . SPARC v8 . ARM v3+
35  * . Alpha . MIPS32
36  * . C, longlong . C, generic
37  */
38 #ifndef MBEDTLS_BN_MUL_H
39 #define MBEDTLS_BN_MUL_H
40 
41 #if !defined(MBEDTLS_CONFIG_FILE)
42 #include "mbedtls/config.h"
43 #else
44 #include MBEDTLS_CONFIG_FILE
45 #endif
46 
47 #include "mbedtls/bignum.h"
48 
49 #if defined(MBEDTLS_HAVE_ASM)
50 
51 #ifndef asm
52 #define asm __asm
53 #endif
54 
55 /* armcc5 --gnu defines __GNUC__ but doesn't support GNU's extended asm */
56 #if defined(__GNUC__) && \
57  ( !defined(__ARMCC_VERSION) || __ARMCC_VERSION >= 6000000 )
58 
59 /*
60  * Disable use of the i386 assembly code below if option -O0, to disable all
61  * compiler optimisations, is passed, detected with __OPTIMIZE__
62  * This is done as the number of registers used in the assembly code doesn't
63  * work with the -O0 option.
64  */
65 #if defined(__i386__) && defined(__OPTIMIZE__)
66 
67 #define MULADDC_INIT \
68  asm( \
69  "movl %%ebx, %0 \n\t" \
70  "movl %5, %%esi \n\t" \
71  "movl %6, %%edi \n\t" \
72  "movl %7, %%ecx \n\t" \
73  "movl %8, %%ebx \n\t"
74 
75 #define MULADDC_CORE \
76  "lodsl \n\t" \
77  "mull %%ebx \n\t" \
78  "addl %%ecx, %%eax \n\t" \
79  "adcl $0, %%edx \n\t" \
80  "addl (%%edi), %%eax \n\t" \
81  "adcl $0, %%edx \n\t" \
82  "movl %%edx, %%ecx \n\t" \
83  "stosl \n\t"
84 
85 #if defined(MBEDTLS_HAVE_SSE2)
86 
87 #define MULADDC_HUIT \
88  "movd %%ecx, %%mm1 \n\t" \
89  "movd %%ebx, %%mm0 \n\t" \
90  "movd (%%edi), %%mm3 \n\t" \
91  "paddq %%mm3, %%mm1 \n\t" \
92  "movd (%%esi), %%mm2 \n\t" \
93  "pmuludq %%mm0, %%mm2 \n\t" \
94  "movd 4(%%esi), %%mm4 \n\t" \
95  "pmuludq %%mm0, %%mm4 \n\t" \
96  "movd 8(%%esi), %%mm6 \n\t" \
97  "pmuludq %%mm0, %%mm6 \n\t" \
98  "movd 12(%%esi), %%mm7 \n\t" \
99  "pmuludq %%mm0, %%mm7 \n\t" \
100  "paddq %%mm2, %%mm1 \n\t" \
101  "movd 4(%%edi), %%mm3 \n\t" \
102  "paddq %%mm4, %%mm3 \n\t" \
103  "movd 8(%%edi), %%mm5 \n\t" \
104  "paddq %%mm6, %%mm5 \n\t" \
105  "movd 12(%%edi), %%mm4 \n\t" \
106  "paddq %%mm4, %%mm7 \n\t" \
107  "movd %%mm1, (%%edi) \n\t" \
108  "movd 16(%%esi), %%mm2 \n\t" \
109  "pmuludq %%mm0, %%mm2 \n\t" \
110  "psrlq $32, %%mm1 \n\t" \
111  "movd 20(%%esi), %%mm4 \n\t" \
112  "pmuludq %%mm0, %%mm4 \n\t" \
113  "paddq %%mm3, %%mm1 \n\t" \
114  "movd 24(%%esi), %%mm6 \n\t" \
115  "pmuludq %%mm0, %%mm6 \n\t" \
116  "movd %%mm1, 4(%%edi) \n\t" \
117  "psrlq $32, %%mm1 \n\t" \
118  "movd 28(%%esi), %%mm3 \n\t" \
119  "pmuludq %%mm0, %%mm3 \n\t" \
120  "paddq %%mm5, %%mm1 \n\t" \
121  "movd 16(%%edi), %%mm5 \n\t" \
122  "paddq %%mm5, %%mm2 \n\t" \
123  "movd %%mm1, 8(%%edi) \n\t" \
124  "psrlq $32, %%mm1 \n\t" \
125  "paddq %%mm7, %%mm1 \n\t" \
126  "movd 20(%%edi), %%mm5 \n\t" \
127  "paddq %%mm5, %%mm4 \n\t" \
128  "movd %%mm1, 12(%%edi) \n\t" \
129  "psrlq $32, %%mm1 \n\t" \
130  "paddq %%mm2, %%mm1 \n\t" \
131  "movd 24(%%edi), %%mm5 \n\t" \
132  "paddq %%mm5, %%mm6 \n\t" \
133  "movd %%mm1, 16(%%edi) \n\t" \
134  "psrlq $32, %%mm1 \n\t" \
135  "paddq %%mm4, %%mm1 \n\t" \
136  "movd 28(%%edi), %%mm5 \n\t" \
137  "paddq %%mm5, %%mm3 \n\t" \
138  "movd %%mm1, 20(%%edi) \n\t" \
139  "psrlq $32, %%mm1 \n\t" \
140  "paddq %%mm6, %%mm1 \n\t" \
141  "movd %%mm1, 24(%%edi) \n\t" \
142  "psrlq $32, %%mm1 \n\t" \
143  "paddq %%mm3, %%mm1 \n\t" \
144  "movd %%mm1, 28(%%edi) \n\t" \
145  "addl $32, %%edi \n\t" \
146  "addl $32, %%esi \n\t" \
147  "psrlq $32, %%mm1 \n\t" \
148  "movd %%mm1, %%ecx \n\t"
149 
150 #define MULADDC_STOP \
151  "emms \n\t" \
152  "movl %4, %%ebx \n\t" \
153  "movl %%ecx, %1 \n\t" \
154  "movl %%edi, %2 \n\t" \
155  "movl %%esi, %3 \n\t" \
156  : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \
157  : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \
158  : "eax", "ebx", "ecx", "edx", "esi", "edi" \
159  );
160 
161 #else
162 
163 #define MULADDC_STOP \
164  "movl %4, %%ebx \n\t" \
165  "movl %%ecx, %1 \n\t" \
166  "movl %%edi, %2 \n\t" \
167  "movl %%esi, %3 \n\t" \
168  : "=m" (t), "=m" (c), "=m" (d), "=m" (s) \
169  : "m" (t), "m" (s), "m" (d), "m" (c), "m" (b) \
170  : "eax", "ebx", "ecx", "edx", "esi", "edi" \
171  );
172 #endif /* SSE2 */
173 #endif /* i386 */
174 
175 #if defined(__amd64__) || defined (__x86_64__)
176 
177 #define MULADDC_INIT \
178  asm( \
179  "xorq %%r8, %%r8\n"
180 
181 #define MULADDC_CORE \
182  "movq (%%rsi), %%rax\n" \
183  "mulq %%rbx\n" \
184  "addq $8, %%rsi\n" \
185  "addq %%rcx, %%rax\n" \
186  "movq %%r8, %%rcx\n" \
187  "adcq $0, %%rdx\n" \
188  "nop \n" \
189  "addq %%rax, (%%rdi)\n" \
190  "adcq %%rdx, %%rcx\n" \
191  "addq $8, %%rdi\n"
192 
193 #define MULADDC_STOP \
194  : "+c" (c), "+D" (d), "+S" (s) \
195  : "b" (b) \
196  : "rax", "rdx", "r8" \
197  );
198 
199 #endif /* AMD64 */
200 
201 #if defined(__aarch64__)
202 
203 #define MULADDC_INIT \
204  asm(
205 
206 #define MULADDC_CORE \
207  "ldr x4, [%2], #8 \n\t" \
208  "ldr x5, [%1] \n\t" \
209  "mul x6, x4, %3 \n\t" \
210  "umulh x7, x4, %3 \n\t" \
211  "adds x5, x5, x6 \n\t" \
212  "adc x7, x7, xzr \n\t" \
213  "adds x5, x5, %0 \n\t" \
214  "adc %0, x7, xzr \n\t" \
215  "str x5, [%1], #8 \n\t"
216 
217 #define MULADDC_STOP \
218  : "+r" (c), "+r" (d), "+r" (s) \
219  : "r" (b) \
220  : "x4", "x5", "x6", "x7", "cc" \
221  );
222 
223 #endif /* Aarch64 */
224 
225 #if defined(__mc68020__) || defined(__mcpu32__)
226 
227 #define MULADDC_INIT \
228  asm( \
229  "movl %3, %%a2 \n\t" \
230  "movl %4, %%a3 \n\t" \
231  "movl %5, %%d3 \n\t" \
232  "movl %6, %%d2 \n\t" \
233  "moveq #0, %%d0 \n\t"
234 
235 #define MULADDC_CORE \
236  "movel %%a2@+, %%d1 \n\t" \
237  "mulul %%d2, %%d4:%%d1 \n\t" \
238  "addl %%d3, %%d1 \n\t" \
239  "addxl %%d0, %%d4 \n\t" \
240  "moveq #0, %%d3 \n\t" \
241  "addl %%d1, %%a3@+ \n\t" \
242  "addxl %%d4, %%d3 \n\t"
243 
244 #define MULADDC_STOP \
245  "movl %%d3, %0 \n\t" \
246  "movl %%a3, %1 \n\t" \
247  "movl %%a2, %2 \n\t" \
248  : "=m" (c), "=m" (d), "=m" (s) \
249  : "m" (s), "m" (d), "m" (c), "m" (b) \
250  : "d0", "d1", "d2", "d3", "d4", "a2", "a3" \
251  );
252 
253 #define MULADDC_HUIT \
254  "movel %%a2@+, %%d1 \n\t" \
255  "mulul %%d2, %%d4:%%d1 \n\t" \
256  "addxl %%d3, %%d1 \n\t" \
257  "addxl %%d0, %%d4 \n\t" \
258  "addl %%d1, %%a3@+ \n\t" \
259  "movel %%a2@+, %%d1 \n\t" \
260  "mulul %%d2, %%d3:%%d1 \n\t" \
261  "addxl %%d4, %%d1 \n\t" \
262  "addxl %%d0, %%d3 \n\t" \
263  "addl %%d1, %%a3@+ \n\t" \
264  "movel %%a2@+, %%d1 \n\t" \
265  "mulul %%d2, %%d4:%%d1 \n\t" \
266  "addxl %%d3, %%d1 \n\t" \
267  "addxl %%d0, %%d4 \n\t" \
268  "addl %%d1, %%a3@+ \n\t" \
269  "movel %%a2@+, %%d1 \n\t" \
270  "mulul %%d2, %%d3:%%d1 \n\t" \
271  "addxl %%d4, %%d1 \n\t" \
272  "addxl %%d0, %%d3 \n\t" \
273  "addl %%d1, %%a3@+ \n\t" \
274  "movel %%a2@+, %%d1 \n\t" \
275  "mulul %%d2, %%d4:%%d1 \n\t" \
276  "addxl %%d3, %%d1 \n\t" \
277  "addxl %%d0, %%d4 \n\t" \
278  "addl %%d1, %%a3@+ \n\t" \
279  "movel %%a2@+, %%d1 \n\t" \
280  "mulul %%d2, %%d3:%%d1 \n\t" \
281  "addxl %%d4, %%d1 \n\t" \
282  "addxl %%d0, %%d3 \n\t" \
283  "addl %%d1, %%a3@+ \n\t" \
284  "movel %%a2@+, %%d1 \n\t" \
285  "mulul %%d2, %%d4:%%d1 \n\t" \
286  "addxl %%d3, %%d1 \n\t" \
287  "addxl %%d0, %%d4 \n\t" \
288  "addl %%d1, %%a3@+ \n\t" \
289  "movel %%a2@+, %%d1 \n\t" \
290  "mulul %%d2, %%d3:%%d1 \n\t" \
291  "addxl %%d4, %%d1 \n\t" \
292  "addxl %%d0, %%d3 \n\t" \
293  "addl %%d1, %%a3@+ \n\t" \
294  "addxl %%d0, %%d3 \n\t"
295 
296 #endif /* MC68000 */
297 
298 #if defined(__powerpc64__) || defined(__ppc64__)
299 
300 #if defined(__MACH__) && defined(__APPLE__)
301 
302 #define MULADDC_INIT \
303  asm( \
304  "ld r3, %3 \n\t" \
305  "ld r4, %4 \n\t" \
306  "ld r5, %5 \n\t" \
307  "ld r6, %6 \n\t" \
308  "addi r3, r3, -8 \n\t" \
309  "addi r4, r4, -8 \n\t" \
310  "addic r5, r5, 0 \n\t"
311 
312 #define MULADDC_CORE \
313  "ldu r7, 8(r3) \n\t" \
314  "mulld r8, r7, r6 \n\t" \
315  "mulhdu r9, r7, r6 \n\t" \
316  "adde r8, r8, r5 \n\t" \
317  "ld r7, 8(r4) \n\t" \
318  "addze r5, r9 \n\t" \
319  "addc r8, r8, r7 \n\t" \
320  "stdu r8, 8(r4) \n\t"
321 
322 #define MULADDC_STOP \
323  "addze r5, r5 \n\t" \
324  "addi r4, r4, 8 \n\t" \
325  "addi r3, r3, 8 \n\t" \
326  "std r5, %0 \n\t" \
327  "std r4, %1 \n\t" \
328  "std r3, %2 \n\t" \
329  : "=m" (c), "=m" (d), "=m" (s) \
330  : "m" (s), "m" (d), "m" (c), "m" (b) \
331  : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
332  );
333 
334 
335 #else /* __MACH__ && __APPLE__ */
336 
337 #define MULADDC_INIT \
338  asm( \
339  "ld %%r3, %3 \n\t" \
340  "ld %%r4, %4 \n\t" \
341  "ld %%r5, %5 \n\t" \
342  "ld %%r6, %6 \n\t" \
343  "addi %%r3, %%r3, -8 \n\t" \
344  "addi %%r4, %%r4, -8 \n\t" \
345  "addic %%r5, %%r5, 0 \n\t"
346 
347 #define MULADDC_CORE \
348  "ldu %%r7, 8(%%r3) \n\t" \
349  "mulld %%r8, %%r7, %%r6 \n\t" \
350  "mulhdu %%r9, %%r7, %%r6 \n\t" \
351  "adde %%r8, %%r8, %%r5 \n\t" \
352  "ld %%r7, 8(%%r4) \n\t" \
353  "addze %%r5, %%r9 \n\t" \
354  "addc %%r8, %%r8, %%r7 \n\t" \
355  "stdu %%r8, 8(%%r4) \n\t"
356 
357 #define MULADDC_STOP \
358  "addze %%r5, %%r5 \n\t" \
359  "addi %%r4, %%r4, 8 \n\t" \
360  "addi %%r3, %%r3, 8 \n\t" \
361  "std %%r5, %0 \n\t" \
362  "std %%r4, %1 \n\t" \
363  "std %%r3, %2 \n\t" \
364  : "=m" (c), "=m" (d), "=m" (s) \
365  : "m" (s), "m" (d), "m" (c), "m" (b) \
366  : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
367  );
368 
369 #endif /* __MACH__ && __APPLE__ */
370 
371 #elif defined(__powerpc__) || defined(__ppc__) /* end PPC64/begin PPC32 */
372 
373 #if defined(__MACH__) && defined(__APPLE__)
374 
375 #define MULADDC_INIT \
376  asm( \
377  "lwz r3, %3 \n\t" \
378  "lwz r4, %4 \n\t" \
379  "lwz r5, %5 \n\t" \
380  "lwz r6, %6 \n\t" \
381  "addi r3, r3, -4 \n\t" \
382  "addi r4, r4, -4 \n\t" \
383  "addic r5, r5, 0 \n\t"
384 
385 #define MULADDC_CORE \
386  "lwzu r7, 4(r3) \n\t" \
387  "mullw r8, r7, r6 \n\t" \
388  "mulhwu r9, r7, r6 \n\t" \
389  "adde r8, r8, r5 \n\t" \
390  "lwz r7, 4(r4) \n\t" \
391  "addze r5, r9 \n\t" \
392  "addc r8, r8, r7 \n\t" \
393  "stwu r8, 4(r4) \n\t"
394 
395 #define MULADDC_STOP \
396  "addze r5, r5 \n\t" \
397  "addi r4, r4, 4 \n\t" \
398  "addi r3, r3, 4 \n\t" \
399  "stw r5, %0 \n\t" \
400  "stw r4, %1 \n\t" \
401  "stw r3, %2 \n\t" \
402  : "=m" (c), "=m" (d), "=m" (s) \
403  : "m" (s), "m" (d), "m" (c), "m" (b) \
404  : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
405  );
406 
407 #else /* __MACH__ && __APPLE__ */
408 
409 #define MULADDC_INIT \
410  asm( \
411  "lwz %%r3, %3 \n\t" \
412  "lwz %%r4, %4 \n\t" \
413  "lwz %%r5, %5 \n\t" \
414  "lwz %%r6, %6 \n\t" \
415  "addi %%r3, %%r3, -4 \n\t" \
416  "addi %%r4, %%r4, -4 \n\t" \
417  "addic %%r5, %%r5, 0 \n\t"
418 
419 #define MULADDC_CORE \
420  "lwzu %%r7, 4(%%r3) \n\t" \
421  "mullw %%r8, %%r7, %%r6 \n\t" \
422  "mulhwu %%r9, %%r7, %%r6 \n\t" \
423  "adde %%r8, %%r8, %%r5 \n\t" \
424  "lwz %%r7, 4(%%r4) \n\t" \
425  "addze %%r5, %%r9 \n\t" \
426  "addc %%r8, %%r8, %%r7 \n\t" \
427  "stwu %%r8, 4(%%r4) \n\t"
428 
429 #define MULADDC_STOP \
430  "addze %%r5, %%r5 \n\t" \
431  "addi %%r4, %%r4, 4 \n\t" \
432  "addi %%r3, %%r3, 4 \n\t" \
433  "stw %%r5, %0 \n\t" \
434  "stw %%r4, %1 \n\t" \
435  "stw %%r3, %2 \n\t" \
436  : "=m" (c), "=m" (d), "=m" (s) \
437  : "m" (s), "m" (d), "m" (c), "m" (b) \
438  : "r3", "r4", "r5", "r6", "r7", "r8", "r9" \
439  );
440 
441 #endif /* __MACH__ && __APPLE__ */
442 
443 #endif /* PPC32 */
444 
445 /*
446  * The Sparc(64) assembly is reported to be broken.
447  * Disable it for now, until we're able to fix it.
448  */
449 #if 0 && defined(__sparc__)
450 #if defined(__sparc64__)
451 
452 #define MULADDC_INIT \
453  asm( \
454  "ldx %3, %%o0 \n\t" \
455  "ldx %4, %%o1 \n\t" \
456  "ld %5, %%o2 \n\t" \
457  "ld %6, %%o3 \n\t"
458 
459 #define MULADDC_CORE \
460  "ld [%%o0], %%o4 \n\t" \
461  "inc 4, %%o0 \n\t" \
462  "ld [%%o1], %%o5 \n\t" \
463  "umul %%o3, %%o4, %%o4 \n\t" \
464  "addcc %%o4, %%o2, %%o4 \n\t" \
465  "rd %%y, %%g1 \n\t" \
466  "addx %%g1, 0, %%g1 \n\t" \
467  "addcc %%o4, %%o5, %%o4 \n\t" \
468  "st %%o4, [%%o1] \n\t" \
469  "addx %%g1, 0, %%o2 \n\t" \
470  "inc 4, %%o1 \n\t"
471 
472  #define MULADDC_STOP \
473  "st %%o2, %0 \n\t" \
474  "stx %%o1, %1 \n\t" \
475  "stx %%o0, %2 \n\t" \
476  : "=m" (c), "=m" (d), "=m" (s) \
477  : "m" (s), "m" (d), "m" (c), "m" (b) \
478  : "g1", "o0", "o1", "o2", "o3", "o4", \
479  "o5" \
480  );
481 
482 #else /* __sparc64__ */
483 
484 #define MULADDC_INIT \
485  asm( \
486  "ld %3, %%o0 \n\t" \
487  "ld %4, %%o1 \n\t" \
488  "ld %5, %%o2 \n\t" \
489  "ld %6, %%o3 \n\t"
490 
491 #define MULADDC_CORE \
492  "ld [%%o0], %%o4 \n\t" \
493  "inc 4, %%o0 \n\t" \
494  "ld [%%o1], %%o5 \n\t" \
495  "umul %%o3, %%o4, %%o4 \n\t" \
496  "addcc %%o4, %%o2, %%o4 \n\t" \
497  "rd %%y, %%g1 \n\t" \
498  "addx %%g1, 0, %%g1 \n\t" \
499  "addcc %%o4, %%o5, %%o4 \n\t" \
500  "st %%o4, [%%o1] \n\t" \
501  "addx %%g1, 0, %%o2 \n\t" \
502  "inc 4, %%o1 \n\t"
503 
504 #define MULADDC_STOP \
505  "st %%o2, %0 \n\t" \
506  "st %%o1, %1 \n\t" \
507  "st %%o0, %2 \n\t" \
508  : "=m" (c), "=m" (d), "=m" (s) \
509  : "m" (s), "m" (d), "m" (c), "m" (b) \
510  : "g1", "o0", "o1", "o2", "o3", "o4", \
511  "o5" \
512  );
513 
514 #endif /* __sparc64__ */
515 #endif /* __sparc__ */
516 
517 #if defined(__microblaze__) || defined(microblaze)
518 
519 #define MULADDC_INIT \
520  asm( \
521  "lwi r3, %3 \n\t" \
522  "lwi r4, %4 \n\t" \
523  "lwi r5, %5 \n\t" \
524  "lwi r6, %6 \n\t" \
525  "andi r7, r6, 0xffff \n\t" \
526  "bsrli r6, r6, 16 \n\t"
527 
528 #define MULADDC_CORE \
529  "lhui r8, r3, 0 \n\t" \
530  "addi r3, r3, 2 \n\t" \
531  "lhui r9, r3, 0 \n\t" \
532  "addi r3, r3, 2 \n\t" \
533  "mul r10, r9, r6 \n\t" \
534  "mul r11, r8, r7 \n\t" \
535  "mul r12, r9, r7 \n\t" \
536  "mul r13, r8, r6 \n\t" \
537  "bsrli r8, r10, 16 \n\t" \
538  "bsrli r9, r11, 16 \n\t" \
539  "add r13, r13, r8 \n\t" \
540  "add r13, r13, r9 \n\t" \
541  "bslli r10, r10, 16 \n\t" \
542  "bslli r11, r11, 16 \n\t" \
543  "add r12, r12, r10 \n\t" \
544  "addc r13, r13, r0 \n\t" \
545  "add r12, r12, r11 \n\t" \
546  "addc r13, r13, r0 \n\t" \
547  "lwi r10, r4, 0 \n\t" \
548  "add r12, r12, r10 \n\t" \
549  "addc r13, r13, r0 \n\t" \
550  "add r12, r12, r5 \n\t" \
551  "addc r5, r13, r0 \n\t" \
552  "swi r12, r4, 0 \n\t" \
553  "addi r4, r4, 4 \n\t"
554 
555 #define MULADDC_STOP \
556  "swi r5, %0 \n\t" \
557  "swi r4, %1 \n\t" \
558  "swi r3, %2 \n\t" \
559  : "=m" (c), "=m" (d), "=m" (s) \
560  : "m" (s), "m" (d), "m" (c), "m" (b) \
561  : "r3", "r4", "r5", "r6", "r7", "r8", \
562  "r9", "r10", "r11", "r12", "r13" \
563  );
564 
565 #endif /* MicroBlaze */
566 
567 #if defined(__tricore__)
568 
569 #define MULADDC_INIT \
570  asm( \
571  "ld.a %%a2, %3 \n\t" \
572  "ld.a %%a3, %4 \n\t" \
573  "ld.w %%d4, %5 \n\t" \
574  "ld.w %%d1, %6 \n\t" \
575  "xor %%d5, %%d5 \n\t"
576 
577 #define MULADDC_CORE \
578  "ld.w %%d0, [%%a2+] \n\t" \
579  "madd.u %%e2, %%e4, %%d0, %%d1 \n\t" \
580  "ld.w %%d0, [%%a3] \n\t" \
581  "addx %%d2, %%d2, %%d0 \n\t" \
582  "addc %%d3, %%d3, 0 \n\t" \
583  "mov %%d4, %%d3 \n\t" \
584  "st.w [%%a3+], %%d2 \n\t"
585 
586 #define MULADDC_STOP \
587  "st.w %0, %%d4 \n\t" \
588  "st.a %1, %%a3 \n\t" \
589  "st.a %2, %%a2 \n\t" \
590  : "=m" (c), "=m" (d), "=m" (s) \
591  : "m" (s), "m" (d), "m" (c), "m" (b) \
592  : "d0", "d1", "e2", "d4", "a2", "a3" \
593  );
594 
595 #endif /* TriCore */
596 
597 /*
598  * Note, gcc -O0 by default uses r7 for the frame pointer, so it complains about
599  * our use of r7 below, unless -fomit-frame-pointer is passed.
600  *
601  * On the other hand, -fomit-frame-pointer is implied by any -Ox options with
602  * x !=0, which we can detect using __OPTIMIZE__ (which is also defined by
603  * clang and armcc5 under the same conditions).
604  *
605  * So, only use the optimized assembly below for optimized build, which avoids
606  * the build error and is pretty reasonable anyway.
607  */
608 #if defined(__GNUC__) && !defined(__OPTIMIZE__)
609 #define MULADDC_CANNOT_USE_R7
610 #endif
611 
612 #if defined(__arm__) && !defined(MULADDC_CANNOT_USE_R7)
613 
614 #if defined(__thumb__) && !defined(__thumb2__)
615 
616 #define MULADDC_INIT \
617  asm( \
618  "ldr r0, %3 \n\t" \
619  "ldr r1, %4 \n\t" \
620  "ldr r2, %5 \n\t" \
621  "ldr r3, %6 \n\t" \
622  "lsr r7, r3, #16 \n\t" \
623  "mov r9, r7 \n\t" \
624  "lsl r7, r3, #16 \n\t" \
625  "lsr r7, r7, #16 \n\t" \
626  "mov r8, r7 \n\t"
627 
628 #define MULADDC_CORE \
629  "ldmia r0!, {r6} \n\t" \
630  "lsr r7, r6, #16 \n\t" \
631  "lsl r6, r6, #16 \n\t" \
632  "lsr r6, r6, #16 \n\t" \
633  "mov r4, r8 \n\t" \
634  "mul r4, r6 \n\t" \
635  "mov r3, r9 \n\t" \
636  "mul r6, r3 \n\t" \
637  "mov r5, r9 \n\t" \
638  "mul r5, r7 \n\t" \
639  "mov r3, r8 \n\t" \
640  "mul r7, r3 \n\t" \
641  "lsr r3, r6, #16 \n\t" \
642  "add r5, r5, r3 \n\t" \
643  "lsr r3, r7, #16 \n\t" \
644  "add r5, r5, r3 \n\t" \
645  "add r4, r4, r2 \n\t" \
646  "mov r2, #0 \n\t" \
647  "adc r5, r2 \n\t" \
648  "lsl r3, r6, #16 \n\t" \
649  "add r4, r4, r3 \n\t" \
650  "adc r5, r2 \n\t" \
651  "lsl r3, r7, #16 \n\t" \
652  "add r4, r4, r3 \n\t" \
653  "adc r5, r2 \n\t" \
654  "ldr r3, [r1] \n\t" \
655  "add r4, r4, r3 \n\t" \
656  "adc r2, r5 \n\t" \
657  "stmia r1!, {r4} \n\t"
658 
659 #define MULADDC_STOP \
660  "str r2, %0 \n\t" \
661  "str r1, %1 \n\t" \
662  "str r0, %2 \n\t" \
663  : "=m" (c), "=m" (d), "=m" (s) \
664  : "m" (s), "m" (d), "m" (c), "m" (b) \
665  : "r0", "r1", "r2", "r3", "r4", "r5", \
666  "r6", "r7", "r8", "r9", "cc" \
667  );
668 
669 #elif (__ARM_ARCH >= 6) && \
670  defined (__ARM_FEATURE_DSP) && (__ARM_FEATURE_DSP == 1)
671 
672 #define MULADDC_INIT \
673  asm(
674 
675 #define MULADDC_CORE \
676  "ldr r0, [%0], #4 \n\t" \
677  "ldr r1, [%1] \n\t" \
678  "umaal r1, %2, %3, r0 \n\t" \
679  "str r1, [%1], #4 \n\t"
680 
681 #define MULADDC_STOP \
682  : "=r" (s), "=r" (d), "=r" (c) \
683  : "r" (b), "0" (s), "1" (d), "2" (c) \
684  : "r0", "r1", "memory" \
685  );
686 
687 #else
688 
689 #define MULADDC_INIT \
690  asm( \
691  "ldr r0, %3 \n\t" \
692  "ldr r1, %4 \n\t" \
693  "ldr r2, %5 \n\t" \
694  "ldr r3, %6 \n\t"
695 
696 #define MULADDC_CORE \
697  "ldr r4, [r0], #4 \n\t" \
698  "mov r5, #0 \n\t" \
699  "ldr r6, [r1] \n\t" \
700  "umlal r2, r5, r3, r4 \n\t" \
701  "adds r7, r6, r2 \n\t" \
702  "adc r2, r5, #0 \n\t" \
703  "str r7, [r1], #4 \n\t"
704 
705 #define MULADDC_STOP \
706  "str r2, %0 \n\t" \
707  "str r1, %1 \n\t" \
708  "str r0, %2 \n\t" \
709  : "=m" (c), "=m" (d), "=m" (s) \
710  : "m" (s), "m" (d), "m" (c), "m" (b) \
711  : "r0", "r1", "r2", "r3", "r4", "r5", \
712  "r6", "r7", "cc" \
713  );
714 
715 #endif /* Thumb */
716 
717 #endif /* ARMv3 */
718 
719 #if defined(__alpha__)
720 
721 #define MULADDC_INIT \
722  asm( \
723  "ldq $1, %3 \n\t" \
724  "ldq $2, %4 \n\t" \
725  "ldq $3, %5 \n\t" \
726  "ldq $4, %6 \n\t"
727 
728 #define MULADDC_CORE \
729  "ldq $6, 0($1) \n\t" \
730  "addq $1, 8, $1 \n\t" \
731  "mulq $6, $4, $7 \n\t" \
732  "umulh $6, $4, $6 \n\t" \
733  "addq $7, $3, $7 \n\t" \
734  "cmpult $7, $3, $3 \n\t" \
735  "ldq $5, 0($2) \n\t" \
736  "addq $7, $5, $7 \n\t" \
737  "cmpult $7, $5, $5 \n\t" \
738  "stq $7, 0($2) \n\t" \
739  "addq $2, 8, $2 \n\t" \
740  "addq $6, $3, $3 \n\t" \
741  "addq $5, $3, $3 \n\t"
742 
743 #define MULADDC_STOP \
744  "stq $3, %0 \n\t" \
745  "stq $2, %1 \n\t" \
746  "stq $1, %2 \n\t" \
747  : "=m" (c), "=m" (d), "=m" (s) \
748  : "m" (s), "m" (d), "m" (c), "m" (b) \
749  : "$1", "$2", "$3", "$4", "$5", "$6", "$7" \
750  );
751 #endif /* Alpha */
752 
753 #if defined(__mips__) && !defined(__mips64)
754 
755 #define MULADDC_INIT \
756  asm( \
757  "lw $10, %3 \n\t" \
758  "lw $11, %4 \n\t" \
759  "lw $12, %5 \n\t" \
760  "lw $13, %6 \n\t"
761 
762 #define MULADDC_CORE \
763  "lw $14, 0($10) \n\t" \
764  "multu $13, $14 \n\t" \
765  "addi $10, $10, 4 \n\t" \
766  "mflo $14 \n\t" \
767  "mfhi $9 \n\t" \
768  "addu $14, $12, $14 \n\t" \
769  "lw $15, 0($11) \n\t" \
770  "sltu $12, $14, $12 \n\t" \
771  "addu $15, $14, $15 \n\t" \
772  "sltu $14, $15, $14 \n\t" \
773  "addu $12, $12, $9 \n\t" \
774  "sw $15, 0($11) \n\t" \
775  "addu $12, $12, $14 \n\t" \
776  "addi $11, $11, 4 \n\t"
777 
778 #define MULADDC_STOP \
779  "sw $12, %0 \n\t" \
780  "sw $11, %1 \n\t" \
781  "sw $10, %2 \n\t" \
782  : "=m" (c), "=m" (d), "=m" (s) \
783  : "m" (s), "m" (d), "m" (c), "m" (b) \
784  : "$9", "$10", "$11", "$12", "$13", "$14", "$15", "lo", "hi" \
785  );
786 
787 #endif /* MIPS */
788 #endif /* GNUC */
789 
790 #if (defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__)
791 
792 #define MULADDC_INIT \
793  __asm mov esi, s \
794  __asm mov edi, d \
795  __asm mov ecx, c \
796  __asm mov ebx, b
797 
798 #define MULADDC_CORE \
799  __asm lodsd \
800  __asm mul ebx \
801  __asm add eax, ecx \
802  __asm adc edx, 0 \
803  __asm add eax, [edi] \
804  __asm adc edx, 0 \
805  __asm mov ecx, edx \
806  __asm stosd
807 
808 #if defined(MBEDTLS_HAVE_SSE2)
809 
810 #define EMIT __asm _emit
811 
812 #define MULADDC_HUIT \
813  EMIT 0x0F EMIT 0x6E EMIT 0xC9 \
814  EMIT 0x0F EMIT 0x6E EMIT 0xC3 \
815  EMIT 0x0F EMIT 0x6E EMIT 0x1F \
816  EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
817  EMIT 0x0F EMIT 0x6E EMIT 0x16 \
818  EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
819  EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x04 \
820  EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
821  EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x08 \
822  EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
823  EMIT 0x0F EMIT 0x6E EMIT 0x7E EMIT 0x0C \
824  EMIT 0x0F EMIT 0xF4 EMIT 0xF8 \
825  EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
826  EMIT 0x0F EMIT 0x6E EMIT 0x5F EMIT 0x04 \
827  EMIT 0x0F EMIT 0xD4 EMIT 0xDC \
828  EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x08 \
829  EMIT 0x0F EMIT 0xD4 EMIT 0xEE \
830  EMIT 0x0F EMIT 0x6E EMIT 0x67 EMIT 0x0C \
831  EMIT 0x0F EMIT 0xD4 EMIT 0xFC \
832  EMIT 0x0F EMIT 0x7E EMIT 0x0F \
833  EMIT 0x0F EMIT 0x6E EMIT 0x56 EMIT 0x10 \
834  EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
835  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
836  EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x14 \
837  EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
838  EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
839  EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x18 \
840  EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
841  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x04 \
842  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
843  EMIT 0x0F EMIT 0x6E EMIT 0x5E EMIT 0x1C \
844  EMIT 0x0F EMIT 0xF4 EMIT 0xD8 \
845  EMIT 0x0F EMIT 0xD4 EMIT 0xCD \
846  EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x10 \
847  EMIT 0x0F EMIT 0xD4 EMIT 0xD5 \
848  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x08 \
849  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
850  EMIT 0x0F EMIT 0xD4 EMIT 0xCF \
851  EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x14 \
852  EMIT 0x0F EMIT 0xD4 EMIT 0xE5 \
853  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x0C \
854  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
855  EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
856  EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x18 \
857  EMIT 0x0F EMIT 0xD4 EMIT 0xF5 \
858  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x10 \
859  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
860  EMIT 0x0F EMIT 0xD4 EMIT 0xCC \
861  EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x1C \
862  EMIT 0x0F EMIT 0xD4 EMIT 0xDD \
863  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x14 \
864  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
865  EMIT 0x0F EMIT 0xD4 EMIT 0xCE \
866  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x18 \
867  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
868  EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
869  EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x1C \
870  EMIT 0x83 EMIT 0xC7 EMIT 0x20 \
871  EMIT 0x83 EMIT 0xC6 EMIT 0x20 \
872  EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
873  EMIT 0x0F EMIT 0x7E EMIT 0xC9
874 
875 #define MULADDC_STOP \
876  EMIT 0x0F EMIT 0x77 \
877  __asm mov c, ecx \
878  __asm mov d, edi \
879  __asm mov s, esi \
880 
881 #else
882 
883 #define MULADDC_STOP \
884  __asm mov c, ecx \
885  __asm mov d, edi \
886  __asm mov s, esi \
887 
888 #endif /* SSE2 */
889 #endif /* MSVC */
890 
891 #endif /* MBEDTLS_HAVE_ASM */
892 
893 #if !defined(MULADDC_CORE)
894 #if defined(MBEDTLS_HAVE_UDBL)
895 
896 #define MULADDC_INIT \
897 { \
898  mbedtls_t_udbl r; \
899  mbedtls_mpi_uint r0, r1;
900 
901 #define MULADDC_CORE \
902  r = *(s++) * (mbedtls_t_udbl) b; \
903  r0 = (mbedtls_mpi_uint) r; \
904  r1 = (mbedtls_mpi_uint)( r >> biL ); \
905  r0 += c; r1 += (r0 < c); \
906  r0 += *d; r1 += (r0 < *d); \
907  c = r1; *(d++) = r0;
908 
909 #define MULADDC_STOP \
910 }
911 
912 #else
913 #define MULADDC_INIT \
914 { \
915  mbedtls_mpi_uint s0, s1, b0, b1; \
916  mbedtls_mpi_uint r0, r1, rx, ry; \
917  b0 = ( b << biH ) >> biH; \
918  b1 = ( b >> biH );
919 
920 #define MULADDC_CORE \
921  s0 = ( *s << biH ) >> biH; \
922  s1 = ( *s >> biH ); s++; \
923  rx = s0 * b1; r0 = s0 * b0; \
924  ry = s1 * b0; r1 = s1 * b1; \
925  r1 += ( rx >> biH ); \
926  r1 += ( ry >> biH ); \
927  rx <<= biH; ry <<= biH; \
928  r0 += rx; r1 += (r0 < rx); \
929  r0 += ry; r1 += (r0 < ry); \
930  r0 += c; r1 += (r0 < c); \
931  r0 += *d; r1 += (r0 < *d); \
932  c = r1; *(d++) = r0;
933 
934 #define MULADDC_STOP \
935 }
936 
937 #endif /* C (generic) */
938 #endif /* C (longlong) */
939 
940 #endif /* bn_mul.h */
bignum.h
Multi-precision integer library.
Generated on Sat Nov 21 2020 22:08:39 for Digi XBee(R) ANSI C Host Library by   doxygen 1.8.11