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 * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
26 #include <linux/module.h>
27 #include <linux/hardirq.h>
28 #include <linux/types.h>
29 #include <linux/crypto.h>
30 #include <linux/err.h>
31 #include <crypto/algapi.h>
32 #include <crypto/cast6.h>
33 #include <crypto/cryptd.h>
34 #include <crypto/b128ops.h>
35 #include <crypto/ctr.h>
36 #include <crypto/lrw.h>
37 #include <crypto/xts.h>
39 #include <asm/xsave.h>
40 #include <asm/crypto/ablk_helper.h>
41 #include <asm/crypto/glue_helper.h>
43 #define CAST6_PARALLEL_BLOCKS 8
45 asmlinkage void cast6_ecb_enc_8way(struct cast6_ctx *ctx, u8 *dst,
47 asmlinkage void cast6_ecb_dec_8way(struct cast6_ctx *ctx, u8 *dst,
50 asmlinkage void cast6_cbc_dec_8way(struct cast6_ctx *ctx, u8 *dst,
52 asmlinkage void cast6_ctr_8way(struct cast6_ctx *ctx, u8 *dst, const u8 *src,
55 asmlinkage void cast6_xts_enc_8way(struct cast6_ctx *ctx, u8 *dst,
56 const u8 *src, le128 *iv);
57 asmlinkage void cast6_xts_dec_8way(struct cast6_ctx *ctx, u8 *dst,
58 const u8 *src, le128 *iv);
60 static void cast6_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
62 glue_xts_crypt_128bit_one(ctx, dst, src, iv,
63 GLUE_FUNC_CAST(__cast6_encrypt));
66 static void cast6_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
68 glue_xts_crypt_128bit_one(ctx, dst, src, iv,
69 GLUE_FUNC_CAST(__cast6_decrypt));
72 static void cast6_crypt_ctr(void *ctx, u128 *dst, const u128 *src, le128 *iv)
76 le128_to_be128(&ctrblk, iv);
79 __cast6_encrypt(ctx, (u8 *)&ctrblk, (u8 *)&ctrblk);
80 u128_xor(dst, src, (u128 *)&ctrblk);
83 static const struct common_glue_ctx cast6_enc = {
85 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
88 .num_blocks = CAST6_PARALLEL_BLOCKS,
89 .fn_u = { .ecb = GLUE_FUNC_CAST(cast6_ecb_enc_8way) }
92 .fn_u = { .ecb = GLUE_FUNC_CAST(__cast6_encrypt) }
96 static const struct common_glue_ctx cast6_ctr = {
98 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
101 .num_blocks = CAST6_PARALLEL_BLOCKS,
102 .fn_u = { .ctr = GLUE_CTR_FUNC_CAST(cast6_ctr_8way) }
105 .fn_u = { .ctr = GLUE_CTR_FUNC_CAST(cast6_crypt_ctr) }
109 static const struct common_glue_ctx cast6_enc_xts = {
111 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
114 .num_blocks = CAST6_PARALLEL_BLOCKS,
115 .fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_enc_8way) }
118 .fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_enc) }
122 static const struct common_glue_ctx cast6_dec = {
124 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
127 .num_blocks = CAST6_PARALLEL_BLOCKS,
128 .fn_u = { .ecb = GLUE_FUNC_CAST(cast6_ecb_dec_8way) }
131 .fn_u = { .ecb = GLUE_FUNC_CAST(__cast6_decrypt) }
135 static const struct common_glue_ctx cast6_dec_cbc = {
137 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
140 .num_blocks = CAST6_PARALLEL_BLOCKS,
141 .fn_u = { .cbc = GLUE_CBC_FUNC_CAST(cast6_cbc_dec_8way) }
144 .fn_u = { .cbc = GLUE_CBC_FUNC_CAST(__cast6_decrypt) }
148 static const struct common_glue_ctx cast6_dec_xts = {
150 .fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
153 .num_blocks = CAST6_PARALLEL_BLOCKS,
154 .fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_dec_8way) }
157 .fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_dec) }
161 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
162 struct scatterlist *src, unsigned int nbytes)
164 return glue_ecb_crypt_128bit(&cast6_enc, desc, dst, src, nbytes);
167 static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
168 struct scatterlist *src, unsigned int nbytes)
170 return glue_ecb_crypt_128bit(&cast6_dec, desc, dst, src, nbytes);
173 static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
174 struct scatterlist *src, unsigned int nbytes)
176 return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(__cast6_encrypt), desc,
180 static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
181 struct scatterlist *src, unsigned int nbytes)
183 return glue_cbc_decrypt_128bit(&cast6_dec_cbc, desc, dst, src,
187 static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
188 struct scatterlist *src, unsigned int nbytes)
190 return glue_ctr_crypt_128bit(&cast6_ctr, desc, dst, src, nbytes);
193 static inline bool cast6_fpu_begin(bool fpu_enabled, unsigned int nbytes)
195 return glue_fpu_begin(CAST6_BLOCK_SIZE, CAST6_PARALLEL_BLOCKS,
196 NULL, fpu_enabled, nbytes);
199 static inline void cast6_fpu_end(bool fpu_enabled)
201 glue_fpu_end(fpu_enabled);
205 struct cast6_ctx *ctx;
209 static void encrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
211 const unsigned int bsize = CAST6_BLOCK_SIZE;
212 struct crypt_priv *ctx = priv;
215 ctx->fpu_enabled = cast6_fpu_begin(ctx->fpu_enabled, nbytes);
217 if (nbytes == bsize * CAST6_PARALLEL_BLOCKS) {
218 cast6_ecb_enc_8way(ctx->ctx, srcdst, srcdst);
222 for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
223 __cast6_encrypt(ctx->ctx, srcdst, srcdst);
226 static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
228 const unsigned int bsize = CAST6_BLOCK_SIZE;
229 struct crypt_priv *ctx = priv;
232 ctx->fpu_enabled = cast6_fpu_begin(ctx->fpu_enabled, nbytes);
234 if (nbytes == bsize * CAST6_PARALLEL_BLOCKS) {
235 cast6_ecb_dec_8way(ctx->ctx, srcdst, srcdst);
239 for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
240 __cast6_decrypt(ctx->ctx, srcdst, srcdst);
243 struct cast6_lrw_ctx {
244 struct lrw_table_ctx lrw_table;
245 struct cast6_ctx cast6_ctx;
248 static int lrw_cast6_setkey(struct crypto_tfm *tfm, const u8 *key,
251 struct cast6_lrw_ctx *ctx = crypto_tfm_ctx(tfm);
254 err = __cast6_setkey(&ctx->cast6_ctx, key, keylen - CAST6_BLOCK_SIZE,
259 return lrw_init_table(&ctx->lrw_table, key + keylen - CAST6_BLOCK_SIZE);
262 static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
263 struct scatterlist *src, unsigned int nbytes)
265 struct cast6_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
266 be128 buf[CAST6_PARALLEL_BLOCKS];
267 struct crypt_priv crypt_ctx = {
268 .ctx = &ctx->cast6_ctx,
269 .fpu_enabled = false,
271 struct lrw_crypt_req req = {
273 .tbuflen = sizeof(buf),
275 .table_ctx = &ctx->lrw_table,
276 .crypt_ctx = &crypt_ctx,
277 .crypt_fn = encrypt_callback,
281 desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
282 ret = lrw_crypt(desc, dst, src, nbytes, &req);
283 cast6_fpu_end(crypt_ctx.fpu_enabled);
288 static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
289 struct scatterlist *src, unsigned int nbytes)
291 struct cast6_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
292 be128 buf[CAST6_PARALLEL_BLOCKS];
293 struct crypt_priv crypt_ctx = {
294 .ctx = &ctx->cast6_ctx,
295 .fpu_enabled = false,
297 struct lrw_crypt_req req = {
299 .tbuflen = sizeof(buf),
301 .table_ctx = &ctx->lrw_table,
302 .crypt_ctx = &crypt_ctx,
303 .crypt_fn = decrypt_callback,
307 desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
308 ret = lrw_crypt(desc, dst, src, nbytes, &req);
309 cast6_fpu_end(crypt_ctx.fpu_enabled);
314 static void lrw_exit_tfm(struct crypto_tfm *tfm)
316 struct cast6_lrw_ctx *ctx = crypto_tfm_ctx(tfm);
318 lrw_free_table(&ctx->lrw_table);
321 struct cast6_xts_ctx {
322 struct cast6_ctx tweak_ctx;
323 struct cast6_ctx crypt_ctx;
326 static int xts_cast6_setkey(struct crypto_tfm *tfm, const u8 *key,
329 struct cast6_xts_ctx *ctx = crypto_tfm_ctx(tfm);
330 u32 *flags = &tfm->crt_flags;
333 /* key consists of keys of equal size concatenated, therefore
334 * the length must be even
337 *flags |= CRYPTO_TFM_RES_BAD_KEY_LEN;
341 /* first half of xts-key is for crypt */
342 err = __cast6_setkey(&ctx->crypt_ctx, key, keylen / 2, flags);
346 /* second half of xts-key is for tweak */
347 return __cast6_setkey(&ctx->tweak_ctx, key + keylen / 2, keylen / 2,
351 static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
352 struct scatterlist *src, unsigned int nbytes)
354 struct cast6_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
356 return glue_xts_crypt_128bit(&cast6_enc_xts, desc, dst, src, nbytes,
357 XTS_TWEAK_CAST(__cast6_encrypt),
358 &ctx->tweak_ctx, &ctx->crypt_ctx);
361 static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
362 struct scatterlist *src, unsigned int nbytes)
364 struct cast6_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
366 return glue_xts_crypt_128bit(&cast6_dec_xts, desc, dst, src, nbytes,
367 XTS_TWEAK_CAST(__cast6_encrypt),
368 &ctx->tweak_ctx, &ctx->crypt_ctx);
371 static struct crypto_alg cast6_algs[10] = { {
372 .cra_name = "__ecb-cast6-avx",
373 .cra_driver_name = "__driver-ecb-cast6-avx",
375 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
376 .cra_blocksize = CAST6_BLOCK_SIZE,
377 .cra_ctxsize = sizeof(struct cast6_ctx),
379 .cra_type = &crypto_blkcipher_type,
380 .cra_module = THIS_MODULE,
383 .min_keysize = CAST6_MIN_KEY_SIZE,
384 .max_keysize = CAST6_MAX_KEY_SIZE,
385 .setkey = cast6_setkey,
386 .encrypt = ecb_encrypt,
387 .decrypt = ecb_decrypt,
391 .cra_name = "__cbc-cast6-avx",
392 .cra_driver_name = "__driver-cbc-cast6-avx",
394 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
395 .cra_blocksize = CAST6_BLOCK_SIZE,
396 .cra_ctxsize = sizeof(struct cast6_ctx),
398 .cra_type = &crypto_blkcipher_type,
399 .cra_module = THIS_MODULE,
402 .min_keysize = CAST6_MIN_KEY_SIZE,
403 .max_keysize = CAST6_MAX_KEY_SIZE,
404 .setkey = cast6_setkey,
405 .encrypt = cbc_encrypt,
406 .decrypt = cbc_decrypt,
410 .cra_name = "__ctr-cast6-avx",
411 .cra_driver_name = "__driver-ctr-cast6-avx",
413 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
415 .cra_ctxsize = sizeof(struct cast6_ctx),
417 .cra_type = &crypto_blkcipher_type,
418 .cra_module = THIS_MODULE,
421 .min_keysize = CAST6_MIN_KEY_SIZE,
422 .max_keysize = CAST6_MAX_KEY_SIZE,
423 .ivsize = CAST6_BLOCK_SIZE,
424 .setkey = cast6_setkey,
425 .encrypt = ctr_crypt,
426 .decrypt = ctr_crypt,
430 .cra_name = "__lrw-cast6-avx",
431 .cra_driver_name = "__driver-lrw-cast6-avx",
433 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
434 .cra_blocksize = CAST6_BLOCK_SIZE,
435 .cra_ctxsize = sizeof(struct cast6_lrw_ctx),
437 .cra_type = &crypto_blkcipher_type,
438 .cra_module = THIS_MODULE,
439 .cra_exit = lrw_exit_tfm,
442 .min_keysize = CAST6_MIN_KEY_SIZE +
444 .max_keysize = CAST6_MAX_KEY_SIZE +
446 .ivsize = CAST6_BLOCK_SIZE,
447 .setkey = lrw_cast6_setkey,
448 .encrypt = lrw_encrypt,
449 .decrypt = lrw_decrypt,
453 .cra_name = "__xts-cast6-avx",
454 .cra_driver_name = "__driver-xts-cast6-avx",
456 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
457 .cra_blocksize = CAST6_BLOCK_SIZE,
458 .cra_ctxsize = sizeof(struct cast6_xts_ctx),
460 .cra_type = &crypto_blkcipher_type,
461 .cra_module = THIS_MODULE,
464 .min_keysize = CAST6_MIN_KEY_SIZE * 2,
465 .max_keysize = CAST6_MAX_KEY_SIZE * 2,
466 .ivsize = CAST6_BLOCK_SIZE,
467 .setkey = xts_cast6_setkey,
468 .encrypt = xts_encrypt,
469 .decrypt = xts_decrypt,
473 .cra_name = "ecb(cast6)",
474 .cra_driver_name = "ecb-cast6-avx",
476 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
477 .cra_blocksize = CAST6_BLOCK_SIZE,
478 .cra_ctxsize = sizeof(struct async_helper_ctx),
480 .cra_type = &crypto_ablkcipher_type,
481 .cra_module = THIS_MODULE,
482 .cra_init = ablk_init,
483 .cra_exit = ablk_exit,
486 .min_keysize = CAST6_MIN_KEY_SIZE,
487 .max_keysize = CAST6_MAX_KEY_SIZE,
488 .setkey = ablk_set_key,
489 .encrypt = ablk_encrypt,
490 .decrypt = ablk_decrypt,
494 .cra_name = "cbc(cast6)",
495 .cra_driver_name = "cbc-cast6-avx",
497 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
498 .cra_blocksize = CAST6_BLOCK_SIZE,
499 .cra_ctxsize = sizeof(struct async_helper_ctx),
501 .cra_type = &crypto_ablkcipher_type,
502 .cra_module = THIS_MODULE,
503 .cra_init = ablk_init,
504 .cra_exit = ablk_exit,
507 .min_keysize = CAST6_MIN_KEY_SIZE,
508 .max_keysize = CAST6_MAX_KEY_SIZE,
509 .ivsize = CAST6_BLOCK_SIZE,
510 .setkey = ablk_set_key,
511 .encrypt = __ablk_encrypt,
512 .decrypt = ablk_decrypt,
516 .cra_name = "ctr(cast6)",
517 .cra_driver_name = "ctr-cast6-avx",
519 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
521 .cra_ctxsize = sizeof(struct async_helper_ctx),
523 .cra_type = &crypto_ablkcipher_type,
524 .cra_module = THIS_MODULE,
525 .cra_init = ablk_init,
526 .cra_exit = ablk_exit,
529 .min_keysize = CAST6_MIN_KEY_SIZE,
530 .max_keysize = CAST6_MAX_KEY_SIZE,
531 .ivsize = CAST6_BLOCK_SIZE,
532 .setkey = ablk_set_key,
533 .encrypt = ablk_encrypt,
534 .decrypt = ablk_encrypt,
539 .cra_name = "lrw(cast6)",
540 .cra_driver_name = "lrw-cast6-avx",
542 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
543 .cra_blocksize = CAST6_BLOCK_SIZE,
544 .cra_ctxsize = sizeof(struct async_helper_ctx),
546 .cra_type = &crypto_ablkcipher_type,
547 .cra_module = THIS_MODULE,
548 .cra_init = ablk_init,
549 .cra_exit = ablk_exit,
552 .min_keysize = CAST6_MIN_KEY_SIZE +
554 .max_keysize = CAST6_MAX_KEY_SIZE +
556 .ivsize = CAST6_BLOCK_SIZE,
557 .setkey = ablk_set_key,
558 .encrypt = ablk_encrypt,
559 .decrypt = ablk_decrypt,
563 .cra_name = "xts(cast6)",
564 .cra_driver_name = "xts-cast6-avx",
566 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
567 .cra_blocksize = CAST6_BLOCK_SIZE,
568 .cra_ctxsize = sizeof(struct async_helper_ctx),
570 .cra_type = &crypto_ablkcipher_type,
571 .cra_module = THIS_MODULE,
572 .cra_init = ablk_init,
573 .cra_exit = ablk_exit,
576 .min_keysize = CAST6_MIN_KEY_SIZE * 2,
577 .max_keysize = CAST6_MAX_KEY_SIZE * 2,
578 .ivsize = CAST6_BLOCK_SIZE,
579 .setkey = ablk_set_key,
580 .encrypt = ablk_encrypt,
581 .decrypt = ablk_decrypt,
586 static int __init cast6_init(void)
590 if (!cpu_has_avx || !cpu_has_osxsave) {
591 pr_info("AVX instructions are not detected.\n");
595 xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
596 if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
597 pr_info("AVX detected but unusable.\n");
601 return crypto_register_algs(cast6_algs, ARRAY_SIZE(cast6_algs));
604 static void __exit cast6_exit(void)
606 crypto_unregister_algs(cast6_algs, ARRAY_SIZE(cast6_algs));
609 module_init(cast6_init);
610 module_exit(cast6_exit);
612 MODULE_DESCRIPTION("Cast6 Cipher Algorithm, AVX optimized");
613 MODULE_LICENSE("GPL");
614 MODULE_ALIAS("cast6");