f6d34da58acf5f9dd39f8f791d2fff81c79ca806
[openwrt/openwrt.git] / package / utils / px5g-standalone / src / polarssl / bn_mul.h
1 /**
2 * \file bn_mul.h
3 *
4 * Based on XySSL: Copyright (C) 2006-2008 Christophe Devine
5 *
6 * Copyright (C) 2009 Paul Bakker <polarssl_maintainer at polarssl dot org>
7 *
8 * All rights reserved.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 *
14 * * Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * * Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * * Neither the names of PolarSSL or XySSL nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
24 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
25 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
26 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
27 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
28 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
29 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
30 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
31 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
32 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
33 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 */
35 /*
36 * Multiply source vector [s] with b, add result
37 * to destination vector [d] and set carry c.
38 *
39 * Currently supports:
40 *
41 * . IA-32 (386+) . AMD64 / EM64T
42 * . IA-32 (SSE2) . Motorola 68000
43 * . PowerPC, 32-bit . MicroBlaze
44 * . PowerPC, 64-bit . TriCore
45 * . SPARC v8 . ARM v3+
46 * . Alpha . MIPS32
47 * . C, longlong . C, generic
48 */
49 #ifndef POLARSSL_BN_MUL_H
50 #define POLARSSL_BN_MUL_H
51
52 #include "polarssl/config.h"
53
54 #if defined(POLARSSL_HAVE_ASM)
55
56 #if defined(__GNUC__)
57 #if defined(__i386__)
58
59 #define MULADDC_INIT \
60 asm( "movl %%ebx, %0 " : "=m" (t)); \
61 asm( "movl %0, %%esi " :: "m" (s)); \
62 asm( "movl %0, %%edi " :: "m" (d)); \
63 asm( "movl %0, %%ecx " :: "m" (c)); \
64 asm( "movl %0, %%ebx " :: "m" (b));
65
66 #define MULADDC_CORE \
67 asm( "lodsl " ); \
68 asm( "mull %ebx " ); \
69 asm( "addl %ecx, %eax " ); \
70 asm( "adcl $0, %edx " ); \
71 asm( "addl (%edi), %eax " ); \
72 asm( "adcl $0, %edx " ); \
73 asm( "movl %edx, %ecx " ); \
74 asm( "stosl " );
75
76 #if defined(POLARSSL_HAVE_SSE2)
77
78 #define MULADDC_HUIT \
79 asm( "movd %ecx, %mm1 " ); \
80 asm( "movd %ebx, %mm0 " ); \
81 asm( "movd (%edi), %mm3 " ); \
82 asm( "paddq %mm3, %mm1 " ); \
83 asm( "movd (%esi), %mm2 " ); \
84 asm( "pmuludq %mm0, %mm2 " ); \
85 asm( "movd 4(%esi), %mm4 " ); \
86 asm( "pmuludq %mm0, %mm4 " ); \
87 asm( "movd 8(%esi), %mm6 " ); \
88 asm( "pmuludq %mm0, %mm6 " ); \
89 asm( "movd 12(%esi), %mm7 " ); \
90 asm( "pmuludq %mm0, %mm7 " ); \
91 asm( "paddq %mm2, %mm1 " ); \
92 asm( "movd 4(%edi), %mm3 " ); \
93 asm( "paddq %mm4, %mm3 " ); \
94 asm( "movd 8(%edi), %mm5 " ); \
95 asm( "paddq %mm6, %mm5 " ); \
96 asm( "movd 12(%edi), %mm4 " ); \
97 asm( "paddq %mm4, %mm7 " ); \
98 asm( "movd %mm1, (%edi) " ); \
99 asm( "movd 16(%esi), %mm2 " ); \
100 asm( "pmuludq %mm0, %mm2 " ); \
101 asm( "psrlq $32, %mm1 " ); \
102 asm( "movd 20(%esi), %mm4 " ); \
103 asm( "pmuludq %mm0, %mm4 " ); \
104 asm( "paddq %mm3, %mm1 " ); \
105 asm( "movd 24(%esi), %mm6 " ); \
106 asm( "pmuludq %mm0, %mm6 " ); \
107 asm( "movd %mm1, 4(%edi) " ); \
108 asm( "psrlq $32, %mm1 " ); \
109 asm( "movd 28(%esi), %mm3 " ); \
110 asm( "pmuludq %mm0, %mm3 " ); \
111 asm( "paddq %mm5, %mm1 " ); \
112 asm( "movd 16(%edi), %mm5 " ); \
113 asm( "paddq %mm5, %mm2 " ); \
114 asm( "movd %mm1, 8(%edi) " ); \
115 asm( "psrlq $32, %mm1 " ); \
116 asm( "paddq %mm7, %mm1 " ); \
117 asm( "movd 20(%edi), %mm5 " ); \
118 asm( "paddq %mm5, %mm4 " ); \
119 asm( "movd %mm1, 12(%edi) " ); \
120 asm( "psrlq $32, %mm1 " ); \
121 asm( "paddq %mm2, %mm1 " ); \
122 asm( "movd 24(%edi), %mm5 " ); \
123 asm( "paddq %mm5, %mm6 " ); \
124 asm( "movd %mm1, 16(%edi) " ); \
125 asm( "psrlq $32, %mm1 " ); \
126 asm( "paddq %mm4, %mm1 " ); \
127 asm( "movd 28(%edi), %mm5 " ); \
128 asm( "paddq %mm5, %mm3 " ); \
129 asm( "movd %mm1, 20(%edi) " ); \
130 asm( "psrlq $32, %mm1 " ); \
131 asm( "paddq %mm6, %mm1 " ); \
132 asm( "movd %mm1, 24(%edi) " ); \
133 asm( "psrlq $32, %mm1 " ); \
134 asm( "paddq %mm3, %mm1 " ); \
135 asm( "movd %mm1, 28(%edi) " ); \
136 asm( "addl $32, %edi " ); \
137 asm( "addl $32, %esi " ); \
138 asm( "psrlq $32, %mm1 " ); \
139 asm( "movd %mm1, %ecx " );
140
141 #define MULADDC_STOP \
142 asm( "emms " ); \
143 asm( "movl %0, %%ebx " :: "m" (t)); \
144 asm( "movl %%ecx, %0 " : "=m" (c)); \
145 asm( "movl %%edi, %0 " : "=m" (d)); \
146 asm( "movl %%esi, %0 " : "=m" (s) :: \
147 "eax", "ecx", "edx", "esi", "edi" );
148
149 #else
150
151 #define MULADDC_STOP \
152 asm( "movl %0, %%ebx " :: "m" (t)); \
153 asm( "movl %%ecx, %0 " : "=m" (c)); \
154 asm( "movl %%edi, %0 " : "=m" (d)); \
155 asm( "movl %%esi, %0 " : "=m" (s) :: \
156 "eax", "ecx", "edx", "esi", "edi" );
157
158 #endif /* SSE2 */
159 #endif /* i386 */
160
161 #if defined(__amd64__) || defined (__x86_64__)
162
163 #define MULADDC_INIT \
164 asm( "movq %0, %%rsi " :: "m" (s)); \
165 asm( "movq %0, %%rdi " :: "m" (d)); \
166 asm( "movq %0, %%rcx " :: "m" (c)); \
167 asm( "movq %0, %%rbx " :: "m" (b)); \
168 asm( "xorq %r8, %r8 " );
169
170 #define MULADDC_CORE \
171 asm( "movq (%rsi),%rax " ); \
172 asm( "mulq %rbx " ); \
173 asm( "addq $8, %rsi " ); \
174 asm( "addq %rcx, %rax " ); \
175 asm( "movq %r8, %rcx " ); \
176 asm( "adcq $0, %rdx " ); \
177 asm( "nop " ); \
178 asm( "addq %rax, (%rdi) " ); \
179 asm( "adcq %rdx, %rcx " ); \
180 asm( "addq $8, %rdi " );
181
182 #define MULADDC_STOP \
183 asm( "movq %%rcx, %0 " : "=m" (c)); \
184 asm( "movq %%rdi, %0 " : "=m" (d)); \
185 asm( "movq %%rsi, %0 " : "=m" (s) :: \
186 "rax", "rcx", "rdx", "rbx", "rsi", "rdi", "r8" );
187
188 #endif /* AMD64 */
189
190 #if defined(__mc68020__) || defined(__mcpu32__)
191
192 #define MULADDC_INIT \
193 asm( "movl %0, %%a2 " :: "m" (s)); \
194 asm( "movl %0, %%a3 " :: "m" (d)); \
195 asm( "movl %0, %%d3 " :: "m" (c)); \
196 asm( "movl %0, %%d2 " :: "m" (b)); \
197 asm( "moveq #0, %d0 " );
198
199 #define MULADDC_CORE \
200 asm( "movel %a2@+, %d1 " ); \
201 asm( "mulul %d2, %d4:%d1 " ); \
202 asm( "addl %d3, %d1 " ); \
203 asm( "addxl %d0, %d4 " ); \
204 asm( "moveq #0, %d3 " ); \
205 asm( "addl %d1, %a3@+ " ); \
206 asm( "addxl %d4, %d3 " );
207
208 #define MULADDC_STOP \
209 asm( "movl %%d3, %0 " : "=m" (c)); \
210 asm( "movl %%a3, %0 " : "=m" (d)); \
211 asm( "movl %%a2, %0 " : "=m" (s) :: \
212 "d0", "d1", "d2", "d3", "d4", "a2", "a3" );
213
214 #define MULADDC_HUIT \
215 asm( "movel %a2@+, %d1 " ); \
216 asm( "mulul %d2, %d4:%d1 " ); \
217 asm( "addxl %d3, %d1 " ); \
218 asm( "addxl %d0, %d4 " ); \
219 asm( "addl %d1, %a3@+ " ); \
220 asm( "movel %a2@+, %d1 " ); \
221 asm( "mulul %d2, %d3:%d1 " ); \
222 asm( "addxl %d4, %d1 " ); \
223 asm( "addxl %d0, %d3 " ); \
224 asm( "addl %d1, %a3@+ " ); \
225 asm( "movel %a2@+, %d1 " ); \
226 asm( "mulul %d2, %d4:%d1 " ); \
227 asm( "addxl %d3, %d1 " ); \
228 asm( "addxl %d0, %d4 " ); \
229 asm( "addl %d1, %a3@+ " ); \
230 asm( "movel %a2@+, %d1 " ); \
231 asm( "mulul %d2, %d3:%d1 " ); \
232 asm( "addxl %d4, %d1 " ); \
233 asm( "addxl %d0, %d3 " ); \
234 asm( "addl %d1, %a3@+ " ); \
235 asm( "movel %a2@+, %d1 " ); \
236 asm( "mulul %d2, %d4:%d1 " ); \
237 asm( "addxl %d3, %d1 " ); \
238 asm( "addxl %d0, %d4 " ); \
239 asm( "addl %d1, %a3@+ " ); \
240 asm( "movel %a2@+, %d1 " ); \
241 asm( "mulul %d2, %d3:%d1 " ); \
242 asm( "addxl %d4, %d1 " ); \
243 asm( "addxl %d0, %d3 " ); \
244 asm( "addl %d1, %a3@+ " ); \
245 asm( "movel %a2@+, %d1 " ); \
246 asm( "mulul %d2, %d4:%d1 " ); \
247 asm( "addxl %d3, %d1 " ); \
248 asm( "addxl %d0, %d4 " ); \
249 asm( "addl %d1, %a3@+ " ); \
250 asm( "movel %a2@+, %d1 " ); \
251 asm( "mulul %d2, %d3:%d1 " ); \
252 asm( "addxl %d4, %d1 " ); \
253 asm( "addxl %d0, %d3 " ); \
254 asm( "addl %d1, %a3@+ " ); \
255 asm( "addxl %d0, %d3 " );
256
257 #endif /* MC68000 */
258
259 #if defined(__powerpc__) || defined(__ppc__)
260 #if defined(__powerpc64__) || defined(__ppc64__)
261
262 #if defined(__MACH__) && defined(__APPLE__)
263
264 #define MULADDC_INIT \
265 asm( "ld r3, %0 " :: "m" (s)); \
266 asm( "ld r4, %0 " :: "m" (d)); \
267 asm( "ld r5, %0 " :: "m" (c)); \
268 asm( "ld r6, %0 " :: "m" (b)); \
269 asm( "addi r3, r3, -8 " ); \
270 asm( "addi r4, r4, -8 " ); \
271 asm( "addic r5, r5, 0 " );
272
273 #define MULADDC_CORE \
274 asm( "ldu r7, 8(r3) " ); \
275 asm( "mulld r8, r7, r6 " ); \
276 asm( "mulhdu r9, r7, r6 " ); \
277 asm( "adde r8, r8, r5 " ); \
278 asm( "ld r7, 8(r4) " ); \
279 asm( "addze r5, r9 " ); \
280 asm( "addc r8, r8, r7 " ); \
281 asm( "stdu r8, 8(r4) " );
282
283 #define MULADDC_STOP \
284 asm( "addze r5, r5 " ); \
285 asm( "addi r4, r4, 8 " ); \
286 asm( "addi r3, r3, 8 " ); \
287 asm( "std r5, %0 " : "=m" (c)); \
288 asm( "std r4, %0 " : "=m" (d)); \
289 asm( "std r3, %0 " : "=m" (s) :: \
290 "r3", "r4", "r5", "r6", "r7", "r8", "r9" );
291
292 #else
293
294 #define MULADDC_INIT \
295 asm( "ld %%r3, %0 " :: "m" (s)); \
296 asm( "ld %%r4, %0 " :: "m" (d)); \
297 asm( "ld %%r5, %0 " :: "m" (c)); \
298 asm( "ld %%r6, %0 " :: "m" (b)); \
299 asm( "addi %r3, %r3, -8 " ); \
300 asm( "addi %r4, %r4, -8 " ); \
301 asm( "addic %r5, %r5, 0 " );
302
303 #define MULADDC_CORE \
304 asm( "ldu %r7, 8(%r3) " ); \
305 asm( "mulld %r8, %r7, %r6 " ); \
306 asm( "mulhdu %r9, %r7, %r6 " ); \
307 asm( "adde %r8, %r8, %r5 " ); \
308 asm( "ld %r7, 8(%r4) " ); \
309 asm( "addze %r5, %r9 " ); \
310 asm( "addc %r8, %r8, %r7 " ); \
311 asm( "stdu %r8, 8(%r4) " );
312
313 #define MULADDC_STOP \
314 asm( "addze %r5, %r5 " ); \
315 asm( "addi %r4, %r4, 8 " ); \
316 asm( "addi %r3, %r3, 8 " ); \
317 asm( "std %%r5, %0 " : "=m" (c)); \
318 asm( "std %%r4, %0 " : "=m" (d)); \
319 asm( "std %%r3, %0 " : "=m" (s) :: \
320 "r3", "r4", "r5", "r6", "r7", "r8", "r9" );
321
322 #endif
323
324 #else /* PPC32 */
325
326 #if defined(__MACH__) && defined(__APPLE__)
327
328 #define MULADDC_INIT \
329 asm( "lwz r3, %0 " :: "m" (s)); \
330 asm( "lwz r4, %0 " :: "m" (d)); \
331 asm( "lwz r5, %0 " :: "m" (c)); \
332 asm( "lwz r6, %0 " :: "m" (b)); \
333 asm( "addi r3, r3, -4 " ); \
334 asm( "addi r4, r4, -4 " ); \
335 asm( "addic r5, r5, 0 " );
336
337 #define MULADDC_CORE \
338 asm( "lwzu r7, 4(r3) " ); \
339 asm( "mullw r8, r7, r6 " ); \
340 asm( "mulhwu r9, r7, r6 " ); \
341 asm( "adde r8, r8, r5 " ); \
342 asm( "lwz r7, 4(r4) " ); \
343 asm( "addze r5, r9 " ); \
344 asm( "addc r8, r8, r7 " ); \
345 asm( "stwu r8, 4(r4) " );
346
347 #define MULADDC_STOP \
348 asm( "addze r5, r5 " ); \
349 asm( "addi r4, r4, 4 " ); \
350 asm( "addi r3, r3, 4 " ); \
351 asm( "stw r5, %0 " : "=m" (c)); \
352 asm( "stw r4, %0 " : "=m" (d)); \
353 asm( "stw r3, %0 " : "=m" (s) :: \
354 "r3", "r4", "r5", "r6", "r7", "r8", "r9" );
355
356 #else
357
358 #define MULADDC_INIT \
359 asm( "lwz %%r3, %0 " :: "m" (s)); \
360 asm( "lwz %%r4, %0 " :: "m" (d)); \
361 asm( "lwz %%r5, %0 " :: "m" (c)); \
362 asm( "lwz %%r6, %0 " :: "m" (b)); \
363 asm( "addi %r3, %r3, -4 " ); \
364 asm( "addi %r4, %r4, -4 " ); \
365 asm( "addic %r5, %r5, 0 " );
366
367 #define MULADDC_CORE \
368 asm( "lwzu %r7, 4(%r3) " ); \
369 asm( "mullw %r8, %r7, %r6 " ); \
370 asm( "mulhwu %r9, %r7, %r6 " ); \
371 asm( "adde %r8, %r8, %r5 " ); \
372 asm( "lwz %r7, 4(%r4) " ); \
373 asm( "addze %r5, %r9 " ); \
374 asm( "addc %r8, %r8, %r7 " ); \
375 asm( "stwu %r8, 4(%r4) " );
376
377 #define MULADDC_STOP \
378 asm( "addze %r5, %r5 " ); \
379 asm( "addi %r4, %r4, 4 " ); \
380 asm( "addi %r3, %r3, 4 " ); \
381 asm( "stw %%r5, %0 " : "=m" (c)); \
382 asm( "stw %%r4, %0 " : "=m" (d)); \
383 asm( "stw %%r3, %0 " : "=m" (s) :: \
384 "r3", "r4", "r5", "r6", "r7", "r8", "r9" );
385
386 #endif
387
388 #endif /* PPC32 */
389 #endif /* PPC64 */
390
391 #if defined(__sparc__)
392
393 #define MULADDC_INIT \
394 asm( "ld %0, %%o0 " :: "m" (s)); \
395 asm( "ld %0, %%o1 " :: "m" (d)); \
396 asm( "ld %0, %%o2 " :: "m" (c)); \
397 asm( "ld %0, %%o3 " :: "m" (b));
398
399 #define MULADDC_CORE \
400 asm( "ld [%o0], %o4 " ); \
401 asm( "inc 4, %o0 " ); \
402 asm( "ld [%o1], %o5 " ); \
403 asm( "umul %o3, %o4, %o4 " ); \
404 asm( "addcc %o4, %o2, %o4 " ); \
405 asm( "rd %y, %g1 " ); \
406 asm( "addx %g1, 0, %g1 " ); \
407 asm( "addcc %o4, %o5, %o4 " ); \
408 asm( "st %o4, [%o1] " ); \
409 asm( "addx %g1, 0, %o2 " ); \
410 asm( "inc 4, %o1 " );
411
412 #define MULADDC_STOP \
413 asm( "st %%o2, %0 " : "=m" (c)); \
414 asm( "st %%o1, %0 " : "=m" (d)); \
415 asm( "st %%o0, %0 " : "=m" (s) :: \
416 "g1", "o0", "o1", "o2", "o3", "o4", "o5" );
417
418 #endif /* SPARCv8 */
419
420 #if defined(__microblaze__) || defined(microblaze)
421
422 #define MULADDC_INIT \
423 asm( "lwi r3, %0 " :: "m" (s)); \
424 asm( "lwi r4, %0 " :: "m" (d)); \
425 asm( "lwi r5, %0 " :: "m" (c)); \
426 asm( "lwi r6, %0 " :: "m" (b)); \
427 asm( "andi r7, r6, 0xffff" ); \
428 asm( "bsrli r6, r6, 16 " );
429
430 #define MULADDC_CORE \
431 asm( "lhui r8, r3, 0 " ); \
432 asm( "addi r3, r3, 2 " ); \
433 asm( "lhui r9, r3, 0 " ); \
434 asm( "addi r3, r3, 2 " ); \
435 asm( "mul r10, r9, r6 " ); \
436 asm( "mul r11, r8, r7 " ); \
437 asm( "mul r12, r9, r7 " ); \
438 asm( "mul r13, r8, r6 " ); \
439 asm( "bsrli r8, r10, 16 " ); \
440 asm( "bsrli r9, r11, 16 " ); \
441 asm( "add r13, r13, r8 " ); \
442 asm( "add r13, r13, r9 " ); \
443 asm( "bslli r10, r10, 16 " ); \
444 asm( "bslli r11, r11, 16 " ); \
445 asm( "add r12, r12, r10 " ); \
446 asm( "addc r13, r13, r0 " ); \
447 asm( "add r12, r12, r11 " ); \
448 asm( "addc r13, r13, r0 " ); \
449 asm( "lwi r10, r4, 0 " ); \
450 asm( "add r12, r12, r10 " ); \
451 asm( "addc r13, r13, r0 " ); \
452 asm( "add r12, r12, r5 " ); \
453 asm( "addc r5, r13, r0 " ); \
454 asm( "swi r12, r4, 0 " ); \
455 asm( "addi r4, r4, 4 " );
456
457 #define MULADDC_STOP \
458 asm( "swi r5, %0 " : "=m" (c)); \
459 asm( "swi r4, %0 " : "=m" (d)); \
460 asm( "swi r3, %0 " : "=m" (s) :: \
461 "r3", "r4" , "r5" , "r6" , "r7" , "r8" , \
462 "r9", "r10", "r11", "r12", "r13" );
463
464 #endif /* MicroBlaze */
465
466 #if defined(__tricore__)
467
468 #define MULADDC_INIT \
469 asm( "ld.a %%a2, %0 " :: "m" (s)); \
470 asm( "ld.a %%a3, %0 " :: "m" (d)); \
471 asm( "ld.w %%d4, %0 " :: "m" (c)); \
472 asm( "ld.w %%d1, %0 " :: "m" (b)); \
473 asm( "xor %d5, %d5 " );
474
475 #define MULADDC_CORE \
476 asm( "ld.w %d0, [%a2+] " ); \
477 asm( "madd.u %e2, %e4, %d0, %d1 " ); \
478 asm( "ld.w %d0, [%a3] " ); \
479 asm( "addx %d2, %d2, %d0 " ); \
480 asm( "addc %d3, %d3, 0 " ); \
481 asm( "mov %d4, %d3 " ); \
482 asm( "st.w [%a3+], %d2 " );
483
484 #define MULADDC_STOP \
485 asm( "st.w %0, %%d4 " : "=m" (c)); \
486 asm( "st.a %0, %%a3 " : "=m" (d)); \
487 asm( "st.a %0, %%a2 " : "=m" (s) :: \
488 "d0", "d1", "e2", "d4", "a2", "a3" );
489
490 #endif /* TriCore */
491
492 #if defined(__arm__)
493
494 #define MULADDC_INIT \
495 asm( "ldr r0, %0 " :: "m" (s)); \
496 asm( "ldr r1, %0 " :: "m" (d)); \
497 asm( "ldr r2, %0 " :: "m" (c)); \
498 asm( "ldr r3, %0 " :: "m" (b));
499
500 #define MULADDC_CORE \
501 asm( "ldr r4, [r0], #4 " ); \
502 asm( "mov r5, #0 " ); \
503 asm( "ldr r6, [r1] " ); \
504 asm( "umlal r2, r5, r3, r4 " ); \
505 asm( "adds r7, r6, r2 " ); \
506 asm( "adc r2, r5, #0 " ); \
507 asm( "str r7, [r1], #4 " );
508
509 #define MULADDC_STOP \
510 asm( "str r2, %0 " : "=m" (c)); \
511 asm( "str r1, %0 " : "=m" (d)); \
512 asm( "str r0, %0 " : "=m" (s) :: \
513 "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7" );
514
515 #endif /* ARMv3 */
516
517 #if defined(__alpha__)
518
519 #define MULADDC_INIT \
520 asm( "ldq $1, %0 " :: "m" (s)); \
521 asm( "ldq $2, %0 " :: "m" (d)); \
522 asm( "ldq $3, %0 " :: "m" (c)); \
523 asm( "ldq $4, %0 " :: "m" (b));
524
525 #define MULADDC_CORE \
526 asm( "ldq $6, 0($1) " ); \
527 asm( "addq $1, 8, $1 " ); \
528 asm( "mulq $6, $4, $7 " ); \
529 asm( "umulh $6, $4, $6 " ); \
530 asm( "addq $7, $3, $7 " ); \
531 asm( "cmpult $7, $3, $3 " ); \
532 asm( "ldq $5, 0($2) " ); \
533 asm( "addq $7, $5, $7 " ); \
534 asm( "cmpult $7, $5, $5 " ); \
535 asm( "stq $7, 0($2) " ); \
536 asm( "addq $2, 8, $2 " ); \
537 asm( "addq $6, $3, $3 " ); \
538 asm( "addq $5, $3, $3 " );
539
540 #define MULADDC_STOP \
541 asm( "stq $3, %0 " : "=m" (c)); \
542 asm( "stq $2, %0 " : "=m" (d)); \
543 asm( "stq $1, %0 " : "=m" (s) :: \
544 "$1", "$2", "$3", "$4", "$5", "$6", "$7" );
545
546 #endif /* Alpha */
547
548 #if defined(__mips__)
549
550 #define MULADDC_INIT \
551 asm( "lw $10, %0 " :: "m" (s)); \
552 asm( "lw $11, %0 " :: "m" (d)); \
553 asm( "lw $12, %0 " :: "m" (c)); \
554 asm( "lw $13, %0 " :: "m" (b));
555
556 #define MULADDC_CORE \
557 asm( "lw $14, 0($10) " ); \
558 asm( "multu $13, $14 " ); \
559 asm( "addi $10, $10, 4 " ); \
560 asm( "mflo $14 " ); \
561 asm( "mfhi $9 " ); \
562 asm( "addu $14, $12, $14 " ); \
563 asm( "lw $15, 0($11) " ); \
564 asm( "sltu $12, $14, $12 " ); \
565 asm( "addu $15, $14, $15 " ); \
566 asm( "sltu $14, $15, $14 " ); \
567 asm( "addu $12, $12, $9 " ); \
568 asm( "sw $15, 0($11) " ); \
569 asm( "addu $12, $12, $14 " ); \
570 asm( "addi $11, $11, 4 " );
571
572 #define MULADDC_STOP \
573 asm( "sw $12, %0 " : "=m" (c)); \
574 asm( "sw $11, %0 " : "=m" (d)); \
575 asm( "sw $10, %0 " : "=m" (s) :: \
576 "$9", "$10", "$11", "$12", "$13", "$14", "$15" );
577
578 #endif /* MIPS */
579 #endif /* GNUC */
580
581 #if (defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__)
582
583 #define MULADDC_INIT \
584 __asm mov esi, s \
585 __asm mov edi, d \
586 __asm mov ecx, c \
587 __asm mov ebx, b
588
589 #define MULADDC_CORE \
590 __asm lodsd \
591 __asm mul ebx \
592 __asm add eax, ecx \
593 __asm adc edx, 0 \
594 __asm add eax, [edi] \
595 __asm adc edx, 0 \
596 __asm mov ecx, edx \
597 __asm stosd
598
599 #if defined(POLARSSL_HAVE_SSE2)
600
601 #define EMIT __asm _emit
602
603 #define MULADDC_HUIT \
604 EMIT 0x0F EMIT 0x6E EMIT 0xC9 \
605 EMIT 0x0F EMIT 0x6E EMIT 0xC3 \
606 EMIT 0x0F EMIT 0x6E EMIT 0x1F \
607 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
608 EMIT 0x0F EMIT 0x6E EMIT 0x16 \
609 EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
610 EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x04 \
611 EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
612 EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x08 \
613 EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
614 EMIT 0x0F EMIT 0x6E EMIT 0x7E EMIT 0x0C \
615 EMIT 0x0F EMIT 0xF4 EMIT 0xF8 \
616 EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
617 EMIT 0x0F EMIT 0x6E EMIT 0x5F EMIT 0x04 \
618 EMIT 0x0F EMIT 0xD4 EMIT 0xDC \
619 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x08 \
620 EMIT 0x0F EMIT 0xD4 EMIT 0xEE \
621 EMIT 0x0F EMIT 0x6E EMIT 0x67 EMIT 0x0C \
622 EMIT 0x0F EMIT 0xD4 EMIT 0xFC \
623 EMIT 0x0F EMIT 0x7E EMIT 0x0F \
624 EMIT 0x0F EMIT 0x6E EMIT 0x56 EMIT 0x10 \
625 EMIT 0x0F EMIT 0xF4 EMIT 0xD0 \
626 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
627 EMIT 0x0F EMIT 0x6E EMIT 0x66 EMIT 0x14 \
628 EMIT 0x0F EMIT 0xF4 EMIT 0xE0 \
629 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
630 EMIT 0x0F EMIT 0x6E EMIT 0x76 EMIT 0x18 \
631 EMIT 0x0F EMIT 0xF4 EMIT 0xF0 \
632 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x04 \
633 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
634 EMIT 0x0F EMIT 0x6E EMIT 0x5E EMIT 0x1C \
635 EMIT 0x0F EMIT 0xF4 EMIT 0xD8 \
636 EMIT 0x0F EMIT 0xD4 EMIT 0xCD \
637 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x10 \
638 EMIT 0x0F EMIT 0xD4 EMIT 0xD5 \
639 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x08 \
640 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
641 EMIT 0x0F EMIT 0xD4 EMIT 0xCF \
642 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x14 \
643 EMIT 0x0F EMIT 0xD4 EMIT 0xE5 \
644 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x0C \
645 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
646 EMIT 0x0F EMIT 0xD4 EMIT 0xCA \
647 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x18 \
648 EMIT 0x0F EMIT 0xD4 EMIT 0xF5 \
649 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x10 \
650 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
651 EMIT 0x0F EMIT 0xD4 EMIT 0xCC \
652 EMIT 0x0F EMIT 0x6E EMIT 0x6F EMIT 0x1C \
653 EMIT 0x0F EMIT 0xD4 EMIT 0xDD \
654 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x14 \
655 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
656 EMIT 0x0F EMIT 0xD4 EMIT 0xCE \
657 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x18 \
658 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
659 EMIT 0x0F EMIT 0xD4 EMIT 0xCB \
660 EMIT 0x0F EMIT 0x7E EMIT 0x4F EMIT 0x1C \
661 EMIT 0x83 EMIT 0xC7 EMIT 0x20 \
662 EMIT 0x83 EMIT 0xC6 EMIT 0x20 \
663 EMIT 0x0F EMIT 0x73 EMIT 0xD1 EMIT 0x20 \
664 EMIT 0x0F EMIT 0x7E EMIT 0xC9
665
666 #define MULADDC_STOP \
667 EMIT 0x0F EMIT 0x77 \
668 __asm mov c, ecx \
669 __asm mov d, edi \
670 __asm mov s, esi \
671
672 #else
673
674 #define MULADDC_STOP \
675 __asm mov c, ecx \
676 __asm mov d, edi \
677 __asm mov s, esi \
678
679 #endif /* SSE2 */
680 #endif /* MSVC */
681
682 #endif /* POLARSSL_HAVE_ASM */
683
684 #if !defined(MULADDC_CORE)
685 #if defined(POLARSSL_HAVE_LONGLONG)
686
687 #define MULADDC_INIT \
688 { \
689 t_dbl r; \
690 t_int r0, r1;
691
692 #define MULADDC_CORE \
693 r = *(s++) * (t_dbl) b; \
694 r0 = r; \
695 r1 = r >> biL; \
696 r0 += c; r1 += (r0 < c); \
697 r0 += *d; r1 += (r0 < *d); \
698 c = r1; *(d++) = r0;
699
700 #define MULADDC_STOP \
701 }
702
703 #else
704 #define MULADDC_INIT \
705 { \
706 t_int s0, s1, b0, b1; \
707 t_int r0, r1, rx, ry; \
708 b0 = ( b << biH ) >> biH; \
709 b1 = ( b >> biH );
710
711 #define MULADDC_CORE \
712 s0 = ( *s << biH ) >> biH; \
713 s1 = ( *s >> biH ); s++; \
714 rx = s0 * b1; r0 = s0 * b0; \
715 ry = s1 * b0; r1 = s1 * b1; \
716 r1 += ( rx >> biH ); \
717 r1 += ( ry >> biH ); \
718 rx <<= biH; ry <<= biH; \
719 r0 += rx; r1 += (r0 < rx); \
720 r0 += ry; r1 += (r0 < ry); \
721 r0 += c; r1 += (r0 < c); \
722 r0 += *d; r1 += (r0 < *d); \
723 c = r1; *(d++) = r0;
724
725 #define MULADDC_STOP \
726 }
727
728 #endif /* C (generic) */
729 #endif /* C (longlong) */
730
731 #endif /* bn_mul.h */