xref: /freebsd/contrib/bearssl/src/symcipher/aes_pwr8_cbcdec.c (revision 2aaf9152a852aba9eb2036b95f4948ee77988826)
1 /*
2  * Copyright (c) 2017 Thomas Pornin <pornin@bolet.org>
3  *
4  * Permission is hereby granted, free of charge, to any person obtaining
5  * a copy of this software and associated documentation files (the
6  * "Software"), to deal in the Software without restriction, including
7  * without limitation the rights to use, copy, modify, merge, publish,
8  * distribute, sublicense, and/or sell copies of the Software, and to
9  * permit persons to whom the Software is furnished to do so, subject to
10  * the following conditions:
11  *
12  * The above copyright notice and this permission notice shall be
13  * included in all copies or substantial portions of the Software.
14  *
15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
16  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
17  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
18  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
19  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
20  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
21  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 
25 #define BR_POWER_ASM_MACROS   1
26 #include "inner.h"
27 
28 #if BR_POWER8
29 
30 /* see bearssl_block.h */
31 void
br_aes_pwr8_cbcdec_init(br_aes_pwr8_cbcdec_keys * ctx,const void * key,size_t len)32 br_aes_pwr8_cbcdec_init(br_aes_pwr8_cbcdec_keys *ctx,
33 	const void *key, size_t len)
34 {
35 	ctx->vtable = &br_aes_pwr8_cbcdec_vtable;
36 	ctx->num_rounds = br_aes_pwr8_keysched(ctx->skey.skni, key, len);
37 }
38 
39 static void
cbcdec_128(const unsigned char * sk,const unsigned char * iv,unsigned char * buf,size_t num_blocks)40 cbcdec_128(const unsigned char *sk,
41 	const unsigned char *iv, unsigned char *buf, size_t num_blocks)
42 {
43 	long cc0, cc1, cc2, cc3;
44 
45 #if BR_POWER8_LE
46 	static const uint32_t idx2be[] = {
47 		0x03020100, 0x07060504, 0x0B0A0908, 0x0F0E0D0C
48 	};
49 #endif
50 
51 	cc0 = 0;
52 	cc1 = 16;
53 	cc2 = 32;
54 	cc3 = 48;
55 	asm volatile (
56 
57 		/*
58 		 * Load subkeys into v0..v10
59 		 */
60 		lxvw4x(32, %[cc0], %[sk])
61 		addi(%[cc0], %[cc0], 16)
62 		lxvw4x(33, %[cc0], %[sk])
63 		addi(%[cc0], %[cc0], 16)
64 		lxvw4x(34, %[cc0], %[sk])
65 		addi(%[cc0], %[cc0], 16)
66 		lxvw4x(35, %[cc0], %[sk])
67 		addi(%[cc0], %[cc0], 16)
68 		lxvw4x(36, %[cc0], %[sk])
69 		addi(%[cc0], %[cc0], 16)
70 		lxvw4x(37, %[cc0], %[sk])
71 		addi(%[cc0], %[cc0], 16)
72 		lxvw4x(38, %[cc0], %[sk])
73 		addi(%[cc0], %[cc0], 16)
74 		lxvw4x(39, %[cc0], %[sk])
75 		addi(%[cc0], %[cc0], 16)
76 		lxvw4x(40, %[cc0], %[sk])
77 		addi(%[cc0], %[cc0], 16)
78 		lxvw4x(41, %[cc0], %[sk])
79 		addi(%[cc0], %[cc0], 16)
80 		lxvw4x(42, %[cc0], %[sk])
81 		li(%[cc0], 0)
82 
83 #if BR_POWER8_LE
84 		/*
85 		 * v15 = constant for byteswapping words
86 		 */
87 		lxvw4x(47, 0, %[idx2be])
88 #endif
89 		/*
90 		 * Load IV into v24.
91 		 */
92 		lxvw4x(56, 0, %[iv])
93 #if BR_POWER8_LE
94 		vperm(24, 24, 24, 15)
95 #endif
96 
97 		mtctr(%[num_blocks])
98 	label(loop)
99 		/*
100 		 * Load next ciphertext words in v16..v19. Also save them
101 		 * in v20..v23.
102 		 */
103 		lxvw4x(48, %[cc0], %[buf])
104 		lxvw4x(49, %[cc1], %[buf])
105 		lxvw4x(50, %[cc2], %[buf])
106 		lxvw4x(51, %[cc3], %[buf])
107 #if BR_POWER8_LE
108 		vperm(16, 16, 16, 15)
109 		vperm(17, 17, 17, 15)
110 		vperm(18, 18, 18, 15)
111 		vperm(19, 19, 19, 15)
112 #endif
113 		vand(20, 16, 16)
114 		vand(21, 17, 17)
115 		vand(22, 18, 18)
116 		vand(23, 19, 19)
117 
118 		/*
119 		 * Decrypt the blocks.
120 		 */
121 		vxor(16, 16, 10)
122 		vxor(17, 17, 10)
123 		vxor(18, 18, 10)
124 		vxor(19, 19, 10)
125 		vncipher(16, 16, 9)
126 		vncipher(17, 17, 9)
127 		vncipher(18, 18, 9)
128 		vncipher(19, 19, 9)
129 		vncipher(16, 16, 8)
130 		vncipher(17, 17, 8)
131 		vncipher(18, 18, 8)
132 		vncipher(19, 19, 8)
133 		vncipher(16, 16, 7)
134 		vncipher(17, 17, 7)
135 		vncipher(18, 18, 7)
136 		vncipher(19, 19, 7)
137 		vncipher(16, 16, 6)
138 		vncipher(17, 17, 6)
139 		vncipher(18, 18, 6)
140 		vncipher(19, 19, 6)
141 		vncipher(16, 16, 5)
142 		vncipher(17, 17, 5)
143 		vncipher(18, 18, 5)
144 		vncipher(19, 19, 5)
145 		vncipher(16, 16, 4)
146 		vncipher(17, 17, 4)
147 		vncipher(18, 18, 4)
148 		vncipher(19, 19, 4)
149 		vncipher(16, 16, 3)
150 		vncipher(17, 17, 3)
151 		vncipher(18, 18, 3)
152 		vncipher(19, 19, 3)
153 		vncipher(16, 16, 2)
154 		vncipher(17, 17, 2)
155 		vncipher(18, 18, 2)
156 		vncipher(19, 19, 2)
157 		vncipher(16, 16, 1)
158 		vncipher(17, 17, 1)
159 		vncipher(18, 18, 1)
160 		vncipher(19, 19, 1)
161 		vncipherlast(16, 16, 0)
162 		vncipherlast(17, 17, 0)
163 		vncipherlast(18, 18, 0)
164 		vncipherlast(19, 19, 0)
165 
166 		/*
167 		 * XOR decrypted blocks with IV / previous block.
168 		 */
169 		vxor(16, 16, 24)
170 		vxor(17, 17, 20)
171 		vxor(18, 18, 21)
172 		vxor(19, 19, 22)
173 
174 		/*
175 		 * Store back result (with byteswap)
176 		 */
177 #if BR_POWER8_LE
178 		vperm(16, 16, 16, 15)
179 		vperm(17, 17, 17, 15)
180 		vperm(18, 18, 18, 15)
181 		vperm(19, 19, 19, 15)
182 #endif
183 		stxvw4x(48, %[cc0], %[buf])
184 		stxvw4x(49, %[cc1], %[buf])
185 		stxvw4x(50, %[cc2], %[buf])
186 		stxvw4x(51, %[cc3], %[buf])
187 
188 		/*
189 		 * Fourth encrypted block is IV for next run.
190 		 */
191 		vand(24, 23, 23)
192 
193 		addi(%[buf], %[buf], 64)
194 
195 		bdnz(loop)
196 
197 : [cc0] "+b" (cc0), [cc1] "+b" (cc1), [cc2] "+b" (cc2), [cc3] "+b" (cc3),
198   [buf] "+b" (buf)
199 : [sk] "b" (sk), [iv] "b" (iv), [num_blocks] "b" (num_blocks >> 2)
200 #if BR_POWER8_LE
201 	, [idx2be] "b" (idx2be)
202 #endif
203 : "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8", "v9",
204   "v10", "v11", "v12", "v13", "v14", "v15", "v16", "v17", "v18", "v19",
205   "v20", "v21", "v22", "v23", "v24", "v25", "v26", "v27", "v28", "v29",
206   "ctr", "memory"
207 	);
208 }
209 
210 static void
cbcdec_192(const unsigned char * sk,const unsigned char * iv,unsigned char * buf,size_t num_blocks)211 cbcdec_192(const unsigned char *sk,
212 	const unsigned char *iv, unsigned char *buf, size_t num_blocks)
213 {
214 	long cc0, cc1, cc2, cc3;
215 
216 #if BR_POWER8_LE
217 	static const uint32_t idx2be[] = {
218 		0x03020100, 0x07060504, 0x0B0A0908, 0x0F0E0D0C
219 	};
220 #endif
221 
222 	cc0 = 0;
223 	cc1 = 16;
224 	cc2 = 32;
225 	cc3 = 48;
226 	asm volatile (
227 
228 		/*
229 		 * Load subkeys into v0..v12
230 		 */
231 		lxvw4x(32, %[cc0], %[sk])
232 		addi(%[cc0], %[cc0], 16)
233 		lxvw4x(33, %[cc0], %[sk])
234 		addi(%[cc0], %[cc0], 16)
235 		lxvw4x(34, %[cc0], %[sk])
236 		addi(%[cc0], %[cc0], 16)
237 		lxvw4x(35, %[cc0], %[sk])
238 		addi(%[cc0], %[cc0], 16)
239 		lxvw4x(36, %[cc0], %[sk])
240 		addi(%[cc0], %[cc0], 16)
241 		lxvw4x(37, %[cc0], %[sk])
242 		addi(%[cc0], %[cc0], 16)
243 		lxvw4x(38, %[cc0], %[sk])
244 		addi(%[cc0], %[cc0], 16)
245 		lxvw4x(39, %[cc0], %[sk])
246 		addi(%[cc0], %[cc0], 16)
247 		lxvw4x(40, %[cc0], %[sk])
248 		addi(%[cc0], %[cc0], 16)
249 		lxvw4x(41, %[cc0], %[sk])
250 		addi(%[cc0], %[cc0], 16)
251 		lxvw4x(42, %[cc0], %[sk])
252 		addi(%[cc0], %[cc0], 16)
253 		lxvw4x(43, %[cc0], %[sk])
254 		addi(%[cc0], %[cc0], 16)
255 		lxvw4x(44, %[cc0], %[sk])
256 		li(%[cc0], 0)
257 
258 #if BR_POWER8_LE
259 		/*
260 		 * v15 = constant for byteswapping words
261 		 */
262 		lxvw4x(47, 0, %[idx2be])
263 #endif
264 		/*
265 		 * Load IV into v24.
266 		 */
267 		lxvw4x(56, 0, %[iv])
268 #if BR_POWER8_LE
269 		vperm(24, 24, 24, 15)
270 #endif
271 
272 		mtctr(%[num_blocks])
273 	label(loop)
274 		/*
275 		 * Load next ciphertext words in v16..v19. Also save them
276 		 * in v20..v23.
277 		 */
278 		lxvw4x(48, %[cc0], %[buf])
279 		lxvw4x(49, %[cc1], %[buf])
280 		lxvw4x(50, %[cc2], %[buf])
281 		lxvw4x(51, %[cc3], %[buf])
282 #if BR_POWER8_LE
283 		vperm(16, 16, 16, 15)
284 		vperm(17, 17, 17, 15)
285 		vperm(18, 18, 18, 15)
286 		vperm(19, 19, 19, 15)
287 #endif
288 		vand(20, 16, 16)
289 		vand(21, 17, 17)
290 		vand(22, 18, 18)
291 		vand(23, 19, 19)
292 
293 		/*
294 		 * Decrypt the blocks.
295 		 */
296 		vxor(16, 16, 12)
297 		vxor(17, 17, 12)
298 		vxor(18, 18, 12)
299 		vxor(19, 19, 12)
300 		vncipher(16, 16, 11)
301 		vncipher(17, 17, 11)
302 		vncipher(18, 18, 11)
303 		vncipher(19, 19, 11)
304 		vncipher(16, 16, 10)
305 		vncipher(17, 17, 10)
306 		vncipher(18, 18, 10)
307 		vncipher(19, 19, 10)
308 		vncipher(16, 16, 9)
309 		vncipher(17, 17, 9)
310 		vncipher(18, 18, 9)
311 		vncipher(19, 19, 9)
312 		vncipher(16, 16, 8)
313 		vncipher(17, 17, 8)
314 		vncipher(18, 18, 8)
315 		vncipher(19, 19, 8)
316 		vncipher(16, 16, 7)
317 		vncipher(17, 17, 7)
318 		vncipher(18, 18, 7)
319 		vncipher(19, 19, 7)
320 		vncipher(16, 16, 6)
321 		vncipher(17, 17, 6)
322 		vncipher(18, 18, 6)
323 		vncipher(19, 19, 6)
324 		vncipher(16, 16, 5)
325 		vncipher(17, 17, 5)
326 		vncipher(18, 18, 5)
327 		vncipher(19, 19, 5)
328 		vncipher(16, 16, 4)
329 		vncipher(17, 17, 4)
330 		vncipher(18, 18, 4)
331 		vncipher(19, 19, 4)
332 		vncipher(16, 16, 3)
333 		vncipher(17, 17, 3)
334 		vncipher(18, 18, 3)
335 		vncipher(19, 19, 3)
336 		vncipher(16, 16, 2)
337 		vncipher(17, 17, 2)
338 		vncipher(18, 18, 2)
339 		vncipher(19, 19, 2)
340 		vncipher(16, 16, 1)
341 		vncipher(17, 17, 1)
342 		vncipher(18, 18, 1)
343 		vncipher(19, 19, 1)
344 		vncipherlast(16, 16, 0)
345 		vncipherlast(17, 17, 0)
346 		vncipherlast(18, 18, 0)
347 		vncipherlast(19, 19, 0)
348 
349 		/*
350 		 * XOR decrypted blocks with IV / previous block.
351 		 */
352 		vxor(16, 16, 24)
353 		vxor(17, 17, 20)
354 		vxor(18, 18, 21)
355 		vxor(19, 19, 22)
356 
357 		/*
358 		 * Store back result (with byteswap)
359 		 */
360 #if BR_POWER8_LE
361 		vperm(16, 16, 16, 15)
362 		vperm(17, 17, 17, 15)
363 		vperm(18, 18, 18, 15)
364 		vperm(19, 19, 19, 15)
365 #endif
366 		stxvw4x(48, %[cc0], %[buf])
367 		stxvw4x(49, %[cc1], %[buf])
368 		stxvw4x(50, %[cc2], %[buf])
369 		stxvw4x(51, %[cc3], %[buf])
370 
371 		/*
372 		 * Fourth encrypted block is IV for next run.
373 		 */
374 		vand(24, 23, 23)
375 
376 		addi(%[buf], %[buf], 64)
377 
378 		bdnz(loop)
379 
380 : [cc0] "+b" (cc0), [cc1] "+b" (cc1), [cc2] "+b" (cc2), [cc3] "+b" (cc3),
381   [buf] "+b" (buf)
382 : [sk] "b" (sk), [iv] "b" (iv), [num_blocks] "b" (num_blocks >> 2)
383 #if BR_POWER8_LE
384 	, [idx2be] "b" (idx2be)
385 #endif
386 : "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8", "v9",
387   "v10", "v11", "v12", "v13", "v14", "v15", "v16", "v17", "v18", "v19",
388   "v20", "v21", "v22", "v23", "v24", "v25", "v26", "v27", "v28", "v29",
389   "ctr", "memory"
390 	);
391 }
392 
393 static void
cbcdec_256(const unsigned char * sk,const unsigned char * iv,unsigned char * buf,size_t num_blocks)394 cbcdec_256(const unsigned char *sk,
395 	const unsigned char *iv, unsigned char *buf, size_t num_blocks)
396 {
397 	long cc0, cc1, cc2, cc3;
398 
399 #if BR_POWER8_LE
400 	static const uint32_t idx2be[] = {
401 		0x03020100, 0x07060504, 0x0B0A0908, 0x0F0E0D0C
402 	};
403 #endif
404 
405 	cc0 = 0;
406 	cc1 = 16;
407 	cc2 = 32;
408 	cc3 = 48;
409 	asm volatile (
410 
411 		/*
412 		 * Load subkeys into v0..v14
413 		 */
414 		lxvw4x(32, %[cc0], %[sk])
415 		addi(%[cc0], %[cc0], 16)
416 		lxvw4x(33, %[cc0], %[sk])
417 		addi(%[cc0], %[cc0], 16)
418 		lxvw4x(34, %[cc0], %[sk])
419 		addi(%[cc0], %[cc0], 16)
420 		lxvw4x(35, %[cc0], %[sk])
421 		addi(%[cc0], %[cc0], 16)
422 		lxvw4x(36, %[cc0], %[sk])
423 		addi(%[cc0], %[cc0], 16)
424 		lxvw4x(37, %[cc0], %[sk])
425 		addi(%[cc0], %[cc0], 16)
426 		lxvw4x(38, %[cc0], %[sk])
427 		addi(%[cc0], %[cc0], 16)
428 		lxvw4x(39, %[cc0], %[sk])
429 		addi(%[cc0], %[cc0], 16)
430 		lxvw4x(40, %[cc0], %[sk])
431 		addi(%[cc0], %[cc0], 16)
432 		lxvw4x(41, %[cc0], %[sk])
433 		addi(%[cc0], %[cc0], 16)
434 		lxvw4x(42, %[cc0], %[sk])
435 		addi(%[cc0], %[cc0], 16)
436 		lxvw4x(43, %[cc0], %[sk])
437 		addi(%[cc0], %[cc0], 16)
438 		lxvw4x(44, %[cc0], %[sk])
439 		addi(%[cc0], %[cc0], 16)
440 		lxvw4x(45, %[cc0], %[sk])
441 		addi(%[cc0], %[cc0], 16)
442 		lxvw4x(46, %[cc0], %[sk])
443 		li(%[cc0], 0)
444 
445 #if BR_POWER8_LE
446 		/*
447 		 * v15 = constant for byteswapping words
448 		 */
449 		lxvw4x(47, 0, %[idx2be])
450 #endif
451 		/*
452 		 * Load IV into v24.
453 		 */
454 		lxvw4x(56, 0, %[iv])
455 #if BR_POWER8_LE
456 		vperm(24, 24, 24, 15)
457 #endif
458 
459 		mtctr(%[num_blocks])
460 	label(loop)
461 		/*
462 		 * Load next ciphertext words in v16..v19. Also save them
463 		 * in v20..v23.
464 		 */
465 		lxvw4x(48, %[cc0], %[buf])
466 		lxvw4x(49, %[cc1], %[buf])
467 		lxvw4x(50, %[cc2], %[buf])
468 		lxvw4x(51, %[cc3], %[buf])
469 #if BR_POWER8_LE
470 		vperm(16, 16, 16, 15)
471 		vperm(17, 17, 17, 15)
472 		vperm(18, 18, 18, 15)
473 		vperm(19, 19, 19, 15)
474 #endif
475 		vand(20, 16, 16)
476 		vand(21, 17, 17)
477 		vand(22, 18, 18)
478 		vand(23, 19, 19)
479 
480 		/*
481 		 * Decrypt the blocks.
482 		 */
483 		vxor(16, 16, 14)
484 		vxor(17, 17, 14)
485 		vxor(18, 18, 14)
486 		vxor(19, 19, 14)
487 		vncipher(16, 16, 13)
488 		vncipher(17, 17, 13)
489 		vncipher(18, 18, 13)
490 		vncipher(19, 19, 13)
491 		vncipher(16, 16, 12)
492 		vncipher(17, 17, 12)
493 		vncipher(18, 18, 12)
494 		vncipher(19, 19, 12)
495 		vncipher(16, 16, 11)
496 		vncipher(17, 17, 11)
497 		vncipher(18, 18, 11)
498 		vncipher(19, 19, 11)
499 		vncipher(16, 16, 10)
500 		vncipher(17, 17, 10)
501 		vncipher(18, 18, 10)
502 		vncipher(19, 19, 10)
503 		vncipher(16, 16, 9)
504 		vncipher(17, 17, 9)
505 		vncipher(18, 18, 9)
506 		vncipher(19, 19, 9)
507 		vncipher(16, 16, 8)
508 		vncipher(17, 17, 8)
509 		vncipher(18, 18, 8)
510 		vncipher(19, 19, 8)
511 		vncipher(16, 16, 7)
512 		vncipher(17, 17, 7)
513 		vncipher(18, 18, 7)
514 		vncipher(19, 19, 7)
515 		vncipher(16, 16, 6)
516 		vncipher(17, 17, 6)
517 		vncipher(18, 18, 6)
518 		vncipher(19, 19, 6)
519 		vncipher(16, 16, 5)
520 		vncipher(17, 17, 5)
521 		vncipher(18, 18, 5)
522 		vncipher(19, 19, 5)
523 		vncipher(16, 16, 4)
524 		vncipher(17, 17, 4)
525 		vncipher(18, 18, 4)
526 		vncipher(19, 19, 4)
527 		vncipher(16, 16, 3)
528 		vncipher(17, 17, 3)
529 		vncipher(18, 18, 3)
530 		vncipher(19, 19, 3)
531 		vncipher(16, 16, 2)
532 		vncipher(17, 17, 2)
533 		vncipher(18, 18, 2)
534 		vncipher(19, 19, 2)
535 		vncipher(16, 16, 1)
536 		vncipher(17, 17, 1)
537 		vncipher(18, 18, 1)
538 		vncipher(19, 19, 1)
539 		vncipherlast(16, 16, 0)
540 		vncipherlast(17, 17, 0)
541 		vncipherlast(18, 18, 0)
542 		vncipherlast(19, 19, 0)
543 
544 		/*
545 		 * XOR decrypted blocks with IV / previous block.
546 		 */
547 		vxor(16, 16, 24)
548 		vxor(17, 17, 20)
549 		vxor(18, 18, 21)
550 		vxor(19, 19, 22)
551 
552 		/*
553 		 * Store back result (with byteswap)
554 		 */
555 #if BR_POWER8_LE
556 		vperm(16, 16, 16, 15)
557 		vperm(17, 17, 17, 15)
558 		vperm(18, 18, 18, 15)
559 		vperm(19, 19, 19, 15)
560 #endif
561 		stxvw4x(48, %[cc0], %[buf])
562 		stxvw4x(49, %[cc1], %[buf])
563 		stxvw4x(50, %[cc2], %[buf])
564 		stxvw4x(51, %[cc3], %[buf])
565 
566 		/*
567 		 * Fourth encrypted block is IV for next run.
568 		 */
569 		vand(24, 23, 23)
570 
571 		addi(%[buf], %[buf], 64)
572 
573 		bdnz(loop)
574 
575 : [cc0] "+b" (cc0), [cc1] "+b" (cc1), [cc2] "+b" (cc2), [cc3] "+b" (cc3),
576   [buf] "+b" (buf)
577 : [sk] "b" (sk), [iv] "b" (iv), [num_blocks] "b" (num_blocks >> 2)
578 #if BR_POWER8_LE
579 	, [idx2be] "b" (idx2be)
580 #endif
581 : "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8", "v9",
582   "v10", "v11", "v12", "v13", "v14", "v15", "v16", "v17", "v18", "v19",
583   "v20", "v21", "v22", "v23", "v24", "v25", "v26", "v27", "v28", "v29",
584   "ctr", "memory"
585 	);
586 }
587 
588 /* see bearssl_block.h */
589 void
br_aes_pwr8_cbcdec_run(const br_aes_pwr8_cbcdec_keys * ctx,void * iv,void * data,size_t len)590 br_aes_pwr8_cbcdec_run(const br_aes_pwr8_cbcdec_keys *ctx,
591 	void *iv, void *data, size_t len)
592 {
593 	unsigned char nextiv[16];
594 	unsigned char *buf;
595 
596 	if (len == 0) {
597 		return;
598 	}
599 	buf = data;
600 	memcpy(nextiv, buf + len - 16, 16);
601 	if (len >= 64) {
602 		size_t num_blocks;
603 		unsigned char tmp[16];
604 
605 		num_blocks = (len >> 4) & ~(size_t)3;
606 		memcpy(tmp, buf + (num_blocks << 4) - 16, 16);
607 		switch (ctx->num_rounds) {
608 		case 10:
609 			cbcdec_128(ctx->skey.skni, iv, buf, num_blocks);
610 			break;
611 		case 12:
612 			cbcdec_192(ctx->skey.skni, iv, buf, num_blocks);
613 			break;
614 		default:
615 			cbcdec_256(ctx->skey.skni, iv, buf, num_blocks);
616 			break;
617 		}
618 		buf += num_blocks << 4;
619 		len &= 63;
620 		memcpy(iv, tmp, 16);
621 	}
622 	if (len > 0) {
623 		unsigned char tmp[64];
624 
625 		memcpy(tmp, buf, len);
626 		memset(tmp + len, 0, (sizeof tmp) - len);
627 		switch (ctx->num_rounds) {
628 		case 10:
629 			cbcdec_128(ctx->skey.skni, iv, tmp, 4);
630 			break;
631 		case 12:
632 			cbcdec_192(ctx->skey.skni, iv, tmp, 4);
633 			break;
634 		default:
635 			cbcdec_256(ctx->skey.skni, iv, tmp, 4);
636 			break;
637 		}
638 		memcpy(buf, tmp, len);
639 	}
640 	memcpy(iv, nextiv, 16);
641 }
642 
643 /* see bearssl_block.h */
644 const br_block_cbcdec_class br_aes_pwr8_cbcdec_vtable = {
645 	sizeof(br_aes_pwr8_cbcdec_keys),
646 	16,
647 	4,
648 	(void (*)(const br_block_cbcdec_class **, const void *, size_t))
649 		&br_aes_pwr8_cbcdec_init,
650 	(void (*)(const br_block_cbcdec_class *const *, void *, void *, size_t))
651 		&br_aes_pwr8_cbcdec_run
652 };
653 
654 /* see bearssl_block.h */
655 const br_block_cbcdec_class *
br_aes_pwr8_cbcdec_get_vtable(void)656 br_aes_pwr8_cbcdec_get_vtable(void)
657 {
658 	return br_aes_pwr8_supported() ? &br_aes_pwr8_cbcdec_vtable : NULL;
659 }
660 
661 #else
662 
663 /* see bearssl_block.h */
664 const br_block_cbcdec_class *
br_aes_pwr8_cbcdec_get_vtable(void)665 br_aes_pwr8_cbcdec_get_vtable(void)
666 {
667 	return NULL;
668 }
669 
670 #endif
671