2 * Glue Code for the AVX assembler implemention of the Cast6 Cipher
4 * Copyright (C) 2012 Johannes Goetzfried
5 * <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
24 #include <linux/module.h>
25 #include <linux/hardirq.h>
26 #include <linux/types.h>
27 #include <linux/crypto.h>
28 #include <linux/err.h>
29 #include <crypto/algapi.h>
30 #include <crypto/cast6.h>
31 #include <crypto/cryptd.h>
32 #include <crypto/b128ops.h>
33 #include <crypto/ctr.h>
34 #include <crypto/lrw.h>
35 #include <crypto/xts.h>
37 #include <asm/xsave.h>
38 #include <asm/crypto/ablk_helper.h>
39 #include <asm/crypto/glue_helper.h>
41 #define CAST6_PARALLEL_BLOCKS 8
43 asmlinkage
void cast6_ecb_enc_8way(struct cast6_ctx
*ctx
, u8
*dst
,
45 asmlinkage
void cast6_ecb_dec_8way(struct cast6_ctx
*ctx
, u8
*dst
,
48 asmlinkage
void cast6_cbc_dec_8way(struct cast6_ctx
*ctx
, u8
*dst
,
50 asmlinkage
void cast6_ctr_8way(struct cast6_ctx
*ctx
, u8
*dst
, const u8
*src
,
53 static void cast6_crypt_ctr(void *ctx
, u128
*dst
, const u128
*src
, le128
*iv
)
57 le128_to_be128(&ctrblk
, iv
);
60 __cast6_encrypt(ctx
, (u8
*)&ctrblk
, (u8
*)&ctrblk
);
61 u128_xor(dst
, src
, (u128
*)&ctrblk
);
64 static const struct common_glue_ctx cast6_enc
= {
66 .fpu_blocks_limit
= CAST6_PARALLEL_BLOCKS
,
69 .num_blocks
= CAST6_PARALLEL_BLOCKS
,
70 .fn_u
= { .ecb
= GLUE_FUNC_CAST(cast6_ecb_enc_8way
) }
73 .fn_u
= { .ecb
= GLUE_FUNC_CAST(__cast6_encrypt
) }
77 static const struct common_glue_ctx cast6_ctr
= {
79 .fpu_blocks_limit
= CAST6_PARALLEL_BLOCKS
,
82 .num_blocks
= CAST6_PARALLEL_BLOCKS
,
83 .fn_u
= { .ctr
= GLUE_CTR_FUNC_CAST(cast6_ctr_8way
) }
86 .fn_u
= { .ctr
= GLUE_CTR_FUNC_CAST(cast6_crypt_ctr
) }
90 static const struct common_glue_ctx cast6_dec
= {
92 .fpu_blocks_limit
= CAST6_PARALLEL_BLOCKS
,
95 .num_blocks
= CAST6_PARALLEL_BLOCKS
,
96 .fn_u
= { .ecb
= GLUE_FUNC_CAST(cast6_ecb_dec_8way
) }
99 .fn_u
= { .ecb
= GLUE_FUNC_CAST(__cast6_decrypt
) }
103 static const struct common_glue_ctx cast6_dec_cbc
= {
105 .fpu_blocks_limit
= CAST6_PARALLEL_BLOCKS
,
108 .num_blocks
= CAST6_PARALLEL_BLOCKS
,
109 .fn_u
= { .cbc
= GLUE_CBC_FUNC_CAST(cast6_cbc_dec_8way
) }
112 .fn_u
= { .cbc
= GLUE_CBC_FUNC_CAST(__cast6_decrypt
) }
116 static int ecb_encrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
117 struct scatterlist
*src
, unsigned int nbytes
)
119 return glue_ecb_crypt_128bit(&cast6_enc
, desc
, dst
, src
, nbytes
);
122 static int ecb_decrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
123 struct scatterlist
*src
, unsigned int nbytes
)
125 return glue_ecb_crypt_128bit(&cast6_dec
, desc
, dst
, src
, nbytes
);
128 static int cbc_encrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
129 struct scatterlist
*src
, unsigned int nbytes
)
131 return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(__cast6_encrypt
), desc
,
135 static int cbc_decrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
136 struct scatterlist
*src
, unsigned int nbytes
)
138 return glue_cbc_decrypt_128bit(&cast6_dec_cbc
, desc
, dst
, src
,
142 static int ctr_crypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
143 struct scatterlist
*src
, unsigned int nbytes
)
145 return glue_ctr_crypt_128bit(&cast6_ctr
, desc
, dst
, src
, nbytes
);
148 static inline bool cast6_fpu_begin(bool fpu_enabled
, unsigned int nbytes
)
150 return glue_fpu_begin(CAST6_BLOCK_SIZE
, CAST6_PARALLEL_BLOCKS
,
151 NULL
, fpu_enabled
, nbytes
);
154 static inline void cast6_fpu_end(bool fpu_enabled
)
156 glue_fpu_end(fpu_enabled
);
160 struct cast6_ctx
*ctx
;
164 static void encrypt_callback(void *priv
, u8
*srcdst
, unsigned int nbytes
)
166 const unsigned int bsize
= CAST6_BLOCK_SIZE
;
167 struct crypt_priv
*ctx
= priv
;
170 ctx
->fpu_enabled
= cast6_fpu_begin(ctx
->fpu_enabled
, nbytes
);
172 if (nbytes
== bsize
* CAST6_PARALLEL_BLOCKS
) {
173 cast6_ecb_enc_8way(ctx
->ctx
, srcdst
, srcdst
);
177 for (i
= 0; i
< nbytes
/ bsize
; i
++, srcdst
+= bsize
)
178 __cast6_encrypt(ctx
->ctx
, srcdst
, srcdst
);
181 static void decrypt_callback(void *priv
, u8
*srcdst
, unsigned int nbytes
)
183 const unsigned int bsize
= CAST6_BLOCK_SIZE
;
184 struct crypt_priv
*ctx
= priv
;
187 ctx
->fpu_enabled
= cast6_fpu_begin(ctx
->fpu_enabled
, nbytes
);
189 if (nbytes
== bsize
* CAST6_PARALLEL_BLOCKS
) {
190 cast6_ecb_dec_8way(ctx
->ctx
, srcdst
, srcdst
);
194 for (i
= 0; i
< nbytes
/ bsize
; i
++, srcdst
+= bsize
)
195 __cast6_decrypt(ctx
->ctx
, srcdst
, srcdst
);
198 struct cast6_lrw_ctx
{
199 struct lrw_table_ctx lrw_table
;
200 struct cast6_ctx cast6_ctx
;
203 static int lrw_cast6_setkey(struct crypto_tfm
*tfm
, const u8
*key
,
206 struct cast6_lrw_ctx
*ctx
= crypto_tfm_ctx(tfm
);
209 err
= __cast6_setkey(&ctx
->cast6_ctx
, key
, keylen
- CAST6_BLOCK_SIZE
,
214 return lrw_init_table(&ctx
->lrw_table
, key
+ keylen
- CAST6_BLOCK_SIZE
);
217 static int lrw_encrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
218 struct scatterlist
*src
, unsigned int nbytes
)
220 struct cast6_lrw_ctx
*ctx
= crypto_blkcipher_ctx(desc
->tfm
);
221 be128 buf
[CAST6_PARALLEL_BLOCKS
];
222 struct crypt_priv crypt_ctx
= {
223 .ctx
= &ctx
->cast6_ctx
,
224 .fpu_enabled
= false,
226 struct lrw_crypt_req req
= {
228 .tbuflen
= sizeof(buf
),
230 .table_ctx
= &ctx
->lrw_table
,
231 .crypt_ctx
= &crypt_ctx
,
232 .crypt_fn
= encrypt_callback
,
236 desc
->flags
&= ~CRYPTO_TFM_REQ_MAY_SLEEP
;
237 ret
= lrw_crypt(desc
, dst
, src
, nbytes
, &req
);
238 cast6_fpu_end(crypt_ctx
.fpu_enabled
);
243 static int lrw_decrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
244 struct scatterlist
*src
, unsigned int nbytes
)
246 struct cast6_lrw_ctx
*ctx
= crypto_blkcipher_ctx(desc
->tfm
);
247 be128 buf
[CAST6_PARALLEL_BLOCKS
];
248 struct crypt_priv crypt_ctx
= {
249 .ctx
= &ctx
->cast6_ctx
,
250 .fpu_enabled
= false,
252 struct lrw_crypt_req req
= {
254 .tbuflen
= sizeof(buf
),
256 .table_ctx
= &ctx
->lrw_table
,
257 .crypt_ctx
= &crypt_ctx
,
258 .crypt_fn
= decrypt_callback
,
262 desc
->flags
&= ~CRYPTO_TFM_REQ_MAY_SLEEP
;
263 ret
= lrw_crypt(desc
, dst
, src
, nbytes
, &req
);
264 cast6_fpu_end(crypt_ctx
.fpu_enabled
);
269 static void lrw_exit_tfm(struct crypto_tfm
*tfm
)
271 struct cast6_lrw_ctx
*ctx
= crypto_tfm_ctx(tfm
);
273 lrw_free_table(&ctx
->lrw_table
);
276 struct cast6_xts_ctx
{
277 struct cast6_ctx tweak_ctx
;
278 struct cast6_ctx crypt_ctx
;
281 static int xts_cast6_setkey(struct crypto_tfm
*tfm
, const u8
*key
,
284 struct cast6_xts_ctx
*ctx
= crypto_tfm_ctx(tfm
);
285 u32
*flags
= &tfm
->crt_flags
;
288 /* key consists of keys of equal size concatenated, therefore
289 * the length must be even
292 *flags
|= CRYPTO_TFM_RES_BAD_KEY_LEN
;
296 /* first half of xts-key is for crypt */
297 err
= __cast6_setkey(&ctx
->crypt_ctx
, key
, keylen
/ 2, flags
);
301 /* second half of xts-key is for tweak */
302 return __cast6_setkey(&ctx
->tweak_ctx
, key
+ keylen
/ 2, keylen
/ 2,
306 static int xts_encrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
307 struct scatterlist
*src
, unsigned int nbytes
)
309 struct cast6_xts_ctx
*ctx
= crypto_blkcipher_ctx(desc
->tfm
);
310 be128 buf
[CAST6_PARALLEL_BLOCKS
];
311 struct crypt_priv crypt_ctx
= {
312 .ctx
= &ctx
->crypt_ctx
,
313 .fpu_enabled
= false,
315 struct xts_crypt_req req
= {
317 .tbuflen
= sizeof(buf
),
319 .tweak_ctx
= &ctx
->tweak_ctx
,
320 .tweak_fn
= XTS_TWEAK_CAST(__cast6_encrypt
),
321 .crypt_ctx
= &crypt_ctx
,
322 .crypt_fn
= encrypt_callback
,
326 desc
->flags
&= ~CRYPTO_TFM_REQ_MAY_SLEEP
;
327 ret
= xts_crypt(desc
, dst
, src
, nbytes
, &req
);
328 cast6_fpu_end(crypt_ctx
.fpu_enabled
);
333 static int xts_decrypt(struct blkcipher_desc
*desc
, struct scatterlist
*dst
,
334 struct scatterlist
*src
, unsigned int nbytes
)
336 struct cast6_xts_ctx
*ctx
= crypto_blkcipher_ctx(desc
->tfm
);
337 be128 buf
[CAST6_PARALLEL_BLOCKS
];
338 struct crypt_priv crypt_ctx
= {
339 .ctx
= &ctx
->crypt_ctx
,
340 .fpu_enabled
= false,
342 struct xts_crypt_req req
= {
344 .tbuflen
= sizeof(buf
),
346 .tweak_ctx
= &ctx
->tweak_ctx
,
347 .tweak_fn
= XTS_TWEAK_CAST(__cast6_encrypt
),
348 .crypt_ctx
= &crypt_ctx
,
349 .crypt_fn
= decrypt_callback
,
353 desc
->flags
&= ~CRYPTO_TFM_REQ_MAY_SLEEP
;
354 ret
= xts_crypt(desc
, dst
, src
, nbytes
, &req
);
355 cast6_fpu_end(crypt_ctx
.fpu_enabled
);
360 static struct crypto_alg cast6_algs
[10] = { {
361 .cra_name
= "__ecb-cast6-avx",
362 .cra_driver_name
= "__driver-ecb-cast6-avx",
364 .cra_flags
= CRYPTO_ALG_TYPE_BLKCIPHER
,
365 .cra_blocksize
= CAST6_BLOCK_SIZE
,
366 .cra_ctxsize
= sizeof(struct cast6_ctx
),
368 .cra_type
= &crypto_blkcipher_type
,
369 .cra_module
= THIS_MODULE
,
372 .min_keysize
= CAST6_MIN_KEY_SIZE
,
373 .max_keysize
= CAST6_MAX_KEY_SIZE
,
374 .setkey
= cast6_setkey
,
375 .encrypt
= ecb_encrypt
,
376 .decrypt
= ecb_decrypt
,
380 .cra_name
= "__cbc-cast6-avx",
381 .cra_driver_name
= "__driver-cbc-cast6-avx",
383 .cra_flags
= CRYPTO_ALG_TYPE_BLKCIPHER
,
384 .cra_blocksize
= CAST6_BLOCK_SIZE
,
385 .cra_ctxsize
= sizeof(struct cast6_ctx
),
387 .cra_type
= &crypto_blkcipher_type
,
388 .cra_module
= THIS_MODULE
,
391 .min_keysize
= CAST6_MIN_KEY_SIZE
,
392 .max_keysize
= CAST6_MAX_KEY_SIZE
,
393 .setkey
= cast6_setkey
,
394 .encrypt
= cbc_encrypt
,
395 .decrypt
= cbc_decrypt
,
399 .cra_name
= "__ctr-cast6-avx",
400 .cra_driver_name
= "__driver-ctr-cast6-avx",
402 .cra_flags
= CRYPTO_ALG_TYPE_BLKCIPHER
,
404 .cra_ctxsize
= sizeof(struct cast6_ctx
),
406 .cra_type
= &crypto_blkcipher_type
,
407 .cra_module
= THIS_MODULE
,
410 .min_keysize
= CAST6_MIN_KEY_SIZE
,
411 .max_keysize
= CAST6_MAX_KEY_SIZE
,
412 .ivsize
= CAST6_BLOCK_SIZE
,
413 .setkey
= cast6_setkey
,
414 .encrypt
= ctr_crypt
,
415 .decrypt
= ctr_crypt
,
419 .cra_name
= "__lrw-cast6-avx",
420 .cra_driver_name
= "__driver-lrw-cast6-avx",
422 .cra_flags
= CRYPTO_ALG_TYPE_BLKCIPHER
,
423 .cra_blocksize
= CAST6_BLOCK_SIZE
,
424 .cra_ctxsize
= sizeof(struct cast6_lrw_ctx
),
426 .cra_type
= &crypto_blkcipher_type
,
427 .cra_module
= THIS_MODULE
,
428 .cra_exit
= lrw_exit_tfm
,
431 .min_keysize
= CAST6_MIN_KEY_SIZE
+
433 .max_keysize
= CAST6_MAX_KEY_SIZE
+
435 .ivsize
= CAST6_BLOCK_SIZE
,
436 .setkey
= lrw_cast6_setkey
,
437 .encrypt
= lrw_encrypt
,
438 .decrypt
= lrw_decrypt
,
442 .cra_name
= "__xts-cast6-avx",
443 .cra_driver_name
= "__driver-xts-cast6-avx",
445 .cra_flags
= CRYPTO_ALG_TYPE_BLKCIPHER
,
446 .cra_blocksize
= CAST6_BLOCK_SIZE
,
447 .cra_ctxsize
= sizeof(struct cast6_xts_ctx
),
449 .cra_type
= &crypto_blkcipher_type
,
450 .cra_module
= THIS_MODULE
,
453 .min_keysize
= CAST6_MIN_KEY_SIZE
* 2,
454 .max_keysize
= CAST6_MAX_KEY_SIZE
* 2,
455 .ivsize
= CAST6_BLOCK_SIZE
,
456 .setkey
= xts_cast6_setkey
,
457 .encrypt
= xts_encrypt
,
458 .decrypt
= xts_decrypt
,
462 .cra_name
= "ecb(cast6)",
463 .cra_driver_name
= "ecb-cast6-avx",
465 .cra_flags
= CRYPTO_ALG_TYPE_ABLKCIPHER
| CRYPTO_ALG_ASYNC
,
466 .cra_blocksize
= CAST6_BLOCK_SIZE
,
467 .cra_ctxsize
= sizeof(struct async_helper_ctx
),
469 .cra_type
= &crypto_ablkcipher_type
,
470 .cra_module
= THIS_MODULE
,
471 .cra_init
= ablk_init
,
472 .cra_exit
= ablk_exit
,
475 .min_keysize
= CAST6_MIN_KEY_SIZE
,
476 .max_keysize
= CAST6_MAX_KEY_SIZE
,
477 .setkey
= ablk_set_key
,
478 .encrypt
= ablk_encrypt
,
479 .decrypt
= ablk_decrypt
,
483 .cra_name
= "cbc(cast6)",
484 .cra_driver_name
= "cbc-cast6-avx",
486 .cra_flags
= CRYPTO_ALG_TYPE_ABLKCIPHER
| CRYPTO_ALG_ASYNC
,
487 .cra_blocksize
= CAST6_BLOCK_SIZE
,
488 .cra_ctxsize
= sizeof(struct async_helper_ctx
),
490 .cra_type
= &crypto_ablkcipher_type
,
491 .cra_module
= THIS_MODULE
,
492 .cra_init
= ablk_init
,
493 .cra_exit
= ablk_exit
,
496 .min_keysize
= CAST6_MIN_KEY_SIZE
,
497 .max_keysize
= CAST6_MAX_KEY_SIZE
,
498 .ivsize
= CAST6_BLOCK_SIZE
,
499 .setkey
= ablk_set_key
,
500 .encrypt
= __ablk_encrypt
,
501 .decrypt
= ablk_decrypt
,
505 .cra_name
= "ctr(cast6)",
506 .cra_driver_name
= "ctr-cast6-avx",
508 .cra_flags
= CRYPTO_ALG_TYPE_ABLKCIPHER
| CRYPTO_ALG_ASYNC
,
510 .cra_ctxsize
= sizeof(struct async_helper_ctx
),
512 .cra_type
= &crypto_ablkcipher_type
,
513 .cra_module
= THIS_MODULE
,
514 .cra_init
= ablk_init
,
515 .cra_exit
= ablk_exit
,
518 .min_keysize
= CAST6_MIN_KEY_SIZE
,
519 .max_keysize
= CAST6_MAX_KEY_SIZE
,
520 .ivsize
= CAST6_BLOCK_SIZE
,
521 .setkey
= ablk_set_key
,
522 .encrypt
= ablk_encrypt
,
523 .decrypt
= ablk_encrypt
,
528 .cra_name
= "lrw(cast6)",
529 .cra_driver_name
= "lrw-cast6-avx",
531 .cra_flags
= CRYPTO_ALG_TYPE_ABLKCIPHER
| CRYPTO_ALG_ASYNC
,
532 .cra_blocksize
= CAST6_BLOCK_SIZE
,
533 .cra_ctxsize
= sizeof(struct async_helper_ctx
),
535 .cra_type
= &crypto_ablkcipher_type
,
536 .cra_module
= THIS_MODULE
,
537 .cra_init
= ablk_init
,
538 .cra_exit
= ablk_exit
,
541 .min_keysize
= CAST6_MIN_KEY_SIZE
+
543 .max_keysize
= CAST6_MAX_KEY_SIZE
+
545 .ivsize
= CAST6_BLOCK_SIZE
,
546 .setkey
= ablk_set_key
,
547 .encrypt
= ablk_encrypt
,
548 .decrypt
= ablk_decrypt
,
552 .cra_name
= "xts(cast6)",
553 .cra_driver_name
= "xts-cast6-avx",
555 .cra_flags
= CRYPTO_ALG_TYPE_ABLKCIPHER
| CRYPTO_ALG_ASYNC
,
556 .cra_blocksize
= CAST6_BLOCK_SIZE
,
557 .cra_ctxsize
= sizeof(struct async_helper_ctx
),
559 .cra_type
= &crypto_ablkcipher_type
,
560 .cra_module
= THIS_MODULE
,
561 .cra_init
= ablk_init
,
562 .cra_exit
= ablk_exit
,
565 .min_keysize
= CAST6_MIN_KEY_SIZE
* 2,
566 .max_keysize
= CAST6_MAX_KEY_SIZE
* 2,
567 .ivsize
= CAST6_BLOCK_SIZE
,
568 .setkey
= ablk_set_key
,
569 .encrypt
= ablk_encrypt
,
570 .decrypt
= ablk_decrypt
,
575 static int __init
cast6_init(void)
579 if (!cpu_has_avx
|| !cpu_has_osxsave
) {
580 pr_info("AVX instructions are not detected.\n");
584 xcr0
= xgetbv(XCR_XFEATURE_ENABLED_MASK
);
585 if ((xcr0
& (XSTATE_SSE
| XSTATE_YMM
)) != (XSTATE_SSE
| XSTATE_YMM
)) {
586 pr_info("AVX detected but unusable.\n");
590 return crypto_register_algs(cast6_algs
, ARRAY_SIZE(cast6_algs
));
593 static void __exit
cast6_exit(void)
595 crypto_unregister_algs(cast6_algs
, ARRAY_SIZE(cast6_algs
));
598 module_init(cast6_init
);
599 module_exit(cast6_exit
);
601 MODULE_DESCRIPTION("Cast6 Cipher Algorithm, AVX optimized");
602 MODULE_LICENSE("GPL");
603 MODULE_ALIAS("cast6");