2 * Glue Code for the AVX assembler implemention of the Cast5 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/ablk_helper.h>
30 #include <crypto/algapi.h>
31 #include <crypto/cast5.h>
32 #include <crypto/cryptd.h>
33 #include <crypto/ctr.h>
35 #include <asm/xsave.h>
36 #include <asm/crypto/glue_helper.h>
38 #define CAST5_PARALLEL_BLOCKS 16
40 asmlinkage void cast5_ecb_enc_16way(struct cast5_ctx *ctx, u8 *dst,
42 asmlinkage void cast5_ecb_dec_16way(struct cast5_ctx *ctx, u8 *dst,
44 asmlinkage void cast5_cbc_dec_16way(struct cast5_ctx *ctx, u8 *dst,
46 asmlinkage void cast5_ctr_16way(struct cast5_ctx *ctx, u8 *dst, const u8 *src,
49 static inline bool cast5_fpu_begin(bool fpu_enabled, unsigned int nbytes)
51 return glue_fpu_begin(CAST5_BLOCK_SIZE, CAST5_PARALLEL_BLOCKS,
52 NULL, fpu_enabled, nbytes);
55 static inline void cast5_fpu_end(bool fpu_enabled)
57 return glue_fpu_end(fpu_enabled);
60 static int ecb_crypt(struct blkcipher_desc *desc, struct blkcipher_walk *walk,
64 struct cast5_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
65 const unsigned int bsize = CAST5_BLOCK_SIZE;
67 void (*fn)(struct cast5_ctx *ctx, u8 *dst, const u8 *src);
70 fn = (enc) ? cast5_ecb_enc_16way : cast5_ecb_dec_16way;
72 err = blkcipher_walk_virt(desc, walk);
73 desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
75 while ((nbytes = walk->nbytes)) {
76 u8 *wsrc = walk->src.virt.addr;
77 u8 *wdst = walk->dst.virt.addr;
79 fpu_enabled = cast5_fpu_begin(false, nbytes);
81 /* Process multi-block batch */
82 if (nbytes >= bsize * CAST5_PARALLEL_BLOCKS) {
86 wsrc += bsize * CAST5_PARALLEL_BLOCKS;
87 wdst += bsize * CAST5_PARALLEL_BLOCKS;
88 nbytes -= bsize * CAST5_PARALLEL_BLOCKS;
89 } while (nbytes >= bsize * CAST5_PARALLEL_BLOCKS);
95 fn = (enc) ? __cast5_encrypt : __cast5_decrypt;
97 /* Handle leftovers */
104 } while (nbytes >= bsize);
107 cast5_fpu_end(fpu_enabled);
108 err = blkcipher_walk_done(desc, walk, nbytes);
113 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
114 struct scatterlist *src, unsigned int nbytes)
116 struct blkcipher_walk walk;
118 blkcipher_walk_init(&walk, dst, src, nbytes);
119 return ecb_crypt(desc, &walk, true);
122 static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
123 struct scatterlist *src, unsigned int nbytes)
125 struct blkcipher_walk walk;
127 blkcipher_walk_init(&walk, dst, src, nbytes);
128 return ecb_crypt(desc, &walk, false);
131 static unsigned int __cbc_encrypt(struct blkcipher_desc *desc,
132 struct blkcipher_walk *walk)
134 struct cast5_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
135 const unsigned int bsize = CAST5_BLOCK_SIZE;
136 unsigned int nbytes = walk->nbytes;
137 u64 *src = (u64 *)walk->src.virt.addr;
138 u64 *dst = (u64 *)walk->dst.virt.addr;
139 u64 *iv = (u64 *)walk->iv;
143 __cast5_encrypt(ctx, (u8 *)dst, (u8 *)dst);
149 } while (nbytes >= bsize);
151 *(u64 *)walk->iv = *iv;
155 static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
156 struct scatterlist *src, unsigned int nbytes)
158 struct blkcipher_walk walk;
161 blkcipher_walk_init(&walk, dst, src, nbytes);
162 err = blkcipher_walk_virt(desc, &walk);
164 while ((nbytes = walk.nbytes)) {
165 nbytes = __cbc_encrypt(desc, &walk);
166 err = blkcipher_walk_done(desc, &walk, nbytes);
172 static unsigned int __cbc_decrypt(struct blkcipher_desc *desc,
173 struct blkcipher_walk *walk)
175 struct cast5_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
176 const unsigned int bsize = CAST5_BLOCK_SIZE;
177 unsigned int nbytes = walk->nbytes;
178 u64 *src = (u64 *)walk->src.virt.addr;
179 u64 *dst = (u64 *)walk->dst.virt.addr;
182 /* Start of the last block. */
183 src += nbytes / bsize - 1;
184 dst += nbytes / bsize - 1;
188 /* Process multi-block batch */
189 if (nbytes >= bsize * CAST5_PARALLEL_BLOCKS) {
191 nbytes -= bsize * (CAST5_PARALLEL_BLOCKS - 1);
192 src -= CAST5_PARALLEL_BLOCKS - 1;
193 dst -= CAST5_PARALLEL_BLOCKS - 1;
195 cast5_cbc_dec_16way(ctx, (u8 *)dst, (u8 *)src);
204 } while (nbytes >= bsize * CAST5_PARALLEL_BLOCKS);
207 /* Handle leftovers */
209 __cast5_decrypt(ctx, (u8 *)dst, (u8 *)src);
221 *dst ^= *(u64 *)walk->iv;
222 *(u64 *)walk->iv = last_iv;
227 static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
228 struct scatterlist *src, unsigned int nbytes)
231 struct blkcipher_walk walk;
234 blkcipher_walk_init(&walk, dst, src, nbytes);
235 err = blkcipher_walk_virt(desc, &walk);
236 desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
238 while ((nbytes = walk.nbytes)) {
239 fpu_enabled = cast5_fpu_begin(false, nbytes);
240 nbytes = __cbc_decrypt(desc, &walk);
241 cast5_fpu_end(fpu_enabled);
242 err = blkcipher_walk_done(desc, &walk, nbytes);
247 static void ctr_crypt_final(struct blkcipher_desc *desc,
248 struct blkcipher_walk *walk)
250 struct cast5_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
251 u8 *ctrblk = walk->iv;
252 u8 keystream[CAST5_BLOCK_SIZE];
253 u8 *src = walk->src.virt.addr;
254 u8 *dst = walk->dst.virt.addr;
255 unsigned int nbytes = walk->nbytes;
257 __cast5_encrypt(ctx, keystream, ctrblk);
258 crypto_xor(keystream, src, nbytes);
259 memcpy(dst, keystream, nbytes);
261 crypto_inc(ctrblk, CAST5_BLOCK_SIZE);
264 static unsigned int __ctr_crypt(struct blkcipher_desc *desc,
265 struct blkcipher_walk *walk)
267 struct cast5_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
268 const unsigned int bsize = CAST5_BLOCK_SIZE;
269 unsigned int nbytes = walk->nbytes;
270 u64 *src = (u64 *)walk->src.virt.addr;
271 u64 *dst = (u64 *)walk->dst.virt.addr;
273 /* Process multi-block batch */
274 if (nbytes >= bsize * CAST5_PARALLEL_BLOCKS) {
276 cast5_ctr_16way(ctx, (u8 *)dst, (u8 *)src,
279 src += CAST5_PARALLEL_BLOCKS;
280 dst += CAST5_PARALLEL_BLOCKS;
281 nbytes -= bsize * CAST5_PARALLEL_BLOCKS;
282 } while (nbytes >= bsize * CAST5_PARALLEL_BLOCKS);
288 /* Handle leftovers */
295 ctrblk = *(u64 *)walk->iv;
296 be64_add_cpu((__be64 *)walk->iv, 1);
298 __cast5_encrypt(ctx, (u8 *)&ctrblk, (u8 *)&ctrblk);
304 } while (nbytes >= bsize);
310 static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
311 struct scatterlist *src, unsigned int nbytes)
314 struct blkcipher_walk walk;
317 blkcipher_walk_init(&walk, dst, src, nbytes);
318 err = blkcipher_walk_virt_block(desc, &walk, CAST5_BLOCK_SIZE);
319 desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
321 while ((nbytes = walk.nbytes) >= CAST5_BLOCK_SIZE) {
322 fpu_enabled = cast5_fpu_begin(false, nbytes);
323 nbytes = __ctr_crypt(desc, &walk);
324 cast5_fpu_end(fpu_enabled);
325 err = blkcipher_walk_done(desc, &walk, nbytes);
329 ctr_crypt_final(desc, &walk);
330 err = blkcipher_walk_done(desc, &walk, 0);
337 static struct crypto_alg cast5_algs[6] = { {
338 .cra_name = "__ecb-cast5-avx",
339 .cra_driver_name = "__driver-ecb-cast5-avx",
341 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER |
343 .cra_blocksize = CAST5_BLOCK_SIZE,
344 .cra_ctxsize = sizeof(struct cast5_ctx),
346 .cra_type = &crypto_blkcipher_type,
347 .cra_module = THIS_MODULE,
350 .min_keysize = CAST5_MIN_KEY_SIZE,
351 .max_keysize = CAST5_MAX_KEY_SIZE,
352 .setkey = cast5_setkey,
353 .encrypt = ecb_encrypt,
354 .decrypt = ecb_decrypt,
358 .cra_name = "__cbc-cast5-avx",
359 .cra_driver_name = "__driver-cbc-cast5-avx",
361 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER |
363 .cra_blocksize = CAST5_BLOCK_SIZE,
364 .cra_ctxsize = sizeof(struct cast5_ctx),
366 .cra_type = &crypto_blkcipher_type,
367 .cra_module = THIS_MODULE,
370 .min_keysize = CAST5_MIN_KEY_SIZE,
371 .max_keysize = CAST5_MAX_KEY_SIZE,
372 .setkey = cast5_setkey,
373 .encrypt = cbc_encrypt,
374 .decrypt = cbc_decrypt,
378 .cra_name = "__ctr-cast5-avx",
379 .cra_driver_name = "__driver-ctr-cast5-avx",
381 .cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER |
384 .cra_ctxsize = sizeof(struct cast5_ctx),
386 .cra_type = &crypto_blkcipher_type,
387 .cra_module = THIS_MODULE,
390 .min_keysize = CAST5_MIN_KEY_SIZE,
391 .max_keysize = CAST5_MAX_KEY_SIZE,
392 .ivsize = CAST5_BLOCK_SIZE,
393 .setkey = cast5_setkey,
394 .encrypt = ctr_crypt,
395 .decrypt = ctr_crypt,
399 .cra_name = "ecb(cast5)",
400 .cra_driver_name = "ecb-cast5-avx",
402 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
403 .cra_blocksize = CAST5_BLOCK_SIZE,
404 .cra_ctxsize = sizeof(struct async_helper_ctx),
406 .cra_type = &crypto_ablkcipher_type,
407 .cra_module = THIS_MODULE,
408 .cra_init = ablk_init,
409 .cra_exit = ablk_exit,
412 .min_keysize = CAST5_MIN_KEY_SIZE,
413 .max_keysize = CAST5_MAX_KEY_SIZE,
414 .setkey = ablk_set_key,
415 .encrypt = ablk_encrypt,
416 .decrypt = ablk_decrypt,
420 .cra_name = "cbc(cast5)",
421 .cra_driver_name = "cbc-cast5-avx",
423 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
424 .cra_blocksize = CAST5_BLOCK_SIZE,
425 .cra_ctxsize = sizeof(struct async_helper_ctx),
427 .cra_type = &crypto_ablkcipher_type,
428 .cra_module = THIS_MODULE,
429 .cra_init = ablk_init,
430 .cra_exit = ablk_exit,
433 .min_keysize = CAST5_MIN_KEY_SIZE,
434 .max_keysize = CAST5_MAX_KEY_SIZE,
435 .ivsize = CAST5_BLOCK_SIZE,
436 .setkey = ablk_set_key,
437 .encrypt = __ablk_encrypt,
438 .decrypt = ablk_decrypt,
442 .cra_name = "ctr(cast5)",
443 .cra_driver_name = "ctr-cast5-avx",
445 .cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
447 .cra_ctxsize = sizeof(struct async_helper_ctx),
449 .cra_type = &crypto_ablkcipher_type,
450 .cra_module = THIS_MODULE,
451 .cra_init = ablk_init,
452 .cra_exit = ablk_exit,
455 .min_keysize = CAST5_MIN_KEY_SIZE,
456 .max_keysize = CAST5_MAX_KEY_SIZE,
457 .ivsize = CAST5_BLOCK_SIZE,
458 .setkey = ablk_set_key,
459 .encrypt = ablk_encrypt,
460 .decrypt = ablk_encrypt,
466 static int __init cast5_init(void)
470 if (!cpu_has_avx || !cpu_has_osxsave) {
471 pr_info("AVX instructions are not detected.\n");
475 xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
476 if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
477 pr_info("AVX detected but unusable.\n");
481 return crypto_register_algs(cast5_algs, ARRAY_SIZE(cast5_algs));
484 static void __exit cast5_exit(void)
486 crypto_unregister_algs(cast5_algs, ARRAY_SIZE(cast5_algs));
489 module_init(cast5_init);
490 module_exit(cast5_exit);
492 MODULE_DESCRIPTION("Cast5 Cipher Algorithm, AVX optimized");
493 MODULE_LICENSE("GPL");
494 MODULE_ALIAS_CRYPTO("cast5");