18c2ecf20Sopenharmony_ci// SPDX-License-Identifier: GPL-2.0-or-later
28c2ecf20Sopenharmony_ci/*
38c2ecf20Sopenharmony_ci * Glue Code for the AVX assembler implementation of the Cast6 Cipher
48c2ecf20Sopenharmony_ci *
58c2ecf20Sopenharmony_ci * Copyright (C) 2012 Johannes Goetzfried
68c2ecf20Sopenharmony_ci *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
78c2ecf20Sopenharmony_ci *
88c2ecf20Sopenharmony_ci * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
98c2ecf20Sopenharmony_ci */
108c2ecf20Sopenharmony_ci
118c2ecf20Sopenharmony_ci#include <linux/module.h>
128c2ecf20Sopenharmony_ci#include <linux/types.h>
138c2ecf20Sopenharmony_ci#include <linux/crypto.h>
148c2ecf20Sopenharmony_ci#include <linux/err.h>
158c2ecf20Sopenharmony_ci#include <crypto/algapi.h>
168c2ecf20Sopenharmony_ci#include <crypto/cast6.h>
178c2ecf20Sopenharmony_ci#include <crypto/internal/simd.h>
188c2ecf20Sopenharmony_ci#include <crypto/xts.h>
198c2ecf20Sopenharmony_ci#include <asm/crypto/glue_helper.h>
208c2ecf20Sopenharmony_ci
218c2ecf20Sopenharmony_ci#define CAST6_PARALLEL_BLOCKS 8
228c2ecf20Sopenharmony_ci
238c2ecf20Sopenharmony_ciasmlinkage void cast6_ecb_enc_8way(const void *ctx, u8 *dst, const u8 *src);
248c2ecf20Sopenharmony_ciasmlinkage void cast6_ecb_dec_8way(const void *ctx, u8 *dst, const u8 *src);
258c2ecf20Sopenharmony_ci
268c2ecf20Sopenharmony_ciasmlinkage void cast6_cbc_dec_8way(const void *ctx, u8 *dst, const u8 *src);
278c2ecf20Sopenharmony_ciasmlinkage void cast6_ctr_8way(const void *ctx, u8 *dst, const u8 *src,
288c2ecf20Sopenharmony_ci			       le128 *iv);
298c2ecf20Sopenharmony_ci
308c2ecf20Sopenharmony_ciasmlinkage void cast6_xts_enc_8way(const void *ctx, u8 *dst, const u8 *src,
318c2ecf20Sopenharmony_ci				   le128 *iv);
328c2ecf20Sopenharmony_ciasmlinkage void cast6_xts_dec_8way(const void *ctx, u8 *dst, const u8 *src,
338c2ecf20Sopenharmony_ci				   le128 *iv);
348c2ecf20Sopenharmony_ci
358c2ecf20Sopenharmony_cistatic int cast6_setkey_skcipher(struct crypto_skcipher *tfm,
368c2ecf20Sopenharmony_ci				 const u8 *key, unsigned int keylen)
378c2ecf20Sopenharmony_ci{
388c2ecf20Sopenharmony_ci	return cast6_setkey(&tfm->base, key, keylen);
398c2ecf20Sopenharmony_ci}
408c2ecf20Sopenharmony_ci
418c2ecf20Sopenharmony_cistatic void cast6_xts_enc(const void *ctx, u8 *dst, const u8 *src, le128 *iv)
428c2ecf20Sopenharmony_ci{
438c2ecf20Sopenharmony_ci	glue_xts_crypt_128bit_one(ctx, dst, src, iv, __cast6_encrypt);
448c2ecf20Sopenharmony_ci}
458c2ecf20Sopenharmony_ci
468c2ecf20Sopenharmony_cistatic void cast6_xts_dec(const void *ctx, u8 *dst, const u8 *src, le128 *iv)
478c2ecf20Sopenharmony_ci{
488c2ecf20Sopenharmony_ci	glue_xts_crypt_128bit_one(ctx, dst, src, iv, __cast6_decrypt);
498c2ecf20Sopenharmony_ci}
508c2ecf20Sopenharmony_ci
518c2ecf20Sopenharmony_cistatic void cast6_crypt_ctr(const void *ctx, u8 *d, const u8 *s, le128 *iv)
528c2ecf20Sopenharmony_ci{
538c2ecf20Sopenharmony_ci	be128 ctrblk;
548c2ecf20Sopenharmony_ci	u128 *dst = (u128 *)d;
558c2ecf20Sopenharmony_ci	const u128 *src = (const u128 *)s;
568c2ecf20Sopenharmony_ci
578c2ecf20Sopenharmony_ci	le128_to_be128(&ctrblk, iv);
588c2ecf20Sopenharmony_ci	le128_inc(iv);
598c2ecf20Sopenharmony_ci
608c2ecf20Sopenharmony_ci	__cast6_encrypt(ctx, (u8 *)&ctrblk, (u8 *)&ctrblk);
618c2ecf20Sopenharmony_ci	u128_xor(dst, src, (u128 *)&ctrblk);
628c2ecf20Sopenharmony_ci}
638c2ecf20Sopenharmony_ci
648c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_enc = {
658c2ecf20Sopenharmony_ci	.num_funcs = 2,
668c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
678c2ecf20Sopenharmony_ci
688c2ecf20Sopenharmony_ci	.funcs = { {
698c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
708c2ecf20Sopenharmony_ci		.fn_u = { .ecb = cast6_ecb_enc_8way }
718c2ecf20Sopenharmony_ci	}, {
728c2ecf20Sopenharmony_ci		.num_blocks = 1,
738c2ecf20Sopenharmony_ci		.fn_u = { .ecb = __cast6_encrypt }
748c2ecf20Sopenharmony_ci	} }
758c2ecf20Sopenharmony_ci};
768c2ecf20Sopenharmony_ci
778c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_ctr = {
788c2ecf20Sopenharmony_ci	.num_funcs = 2,
798c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
808c2ecf20Sopenharmony_ci
818c2ecf20Sopenharmony_ci	.funcs = { {
828c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
838c2ecf20Sopenharmony_ci		.fn_u = { .ctr = cast6_ctr_8way }
848c2ecf20Sopenharmony_ci	}, {
858c2ecf20Sopenharmony_ci		.num_blocks = 1,
868c2ecf20Sopenharmony_ci		.fn_u = { .ctr = cast6_crypt_ctr }
878c2ecf20Sopenharmony_ci	} }
888c2ecf20Sopenharmony_ci};
898c2ecf20Sopenharmony_ci
908c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_enc_xts = {
918c2ecf20Sopenharmony_ci	.num_funcs = 2,
928c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
938c2ecf20Sopenharmony_ci
948c2ecf20Sopenharmony_ci	.funcs = { {
958c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
968c2ecf20Sopenharmony_ci		.fn_u = { .xts = cast6_xts_enc_8way }
978c2ecf20Sopenharmony_ci	}, {
988c2ecf20Sopenharmony_ci		.num_blocks = 1,
998c2ecf20Sopenharmony_ci		.fn_u = { .xts = cast6_xts_enc }
1008c2ecf20Sopenharmony_ci	} }
1018c2ecf20Sopenharmony_ci};
1028c2ecf20Sopenharmony_ci
1038c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_dec = {
1048c2ecf20Sopenharmony_ci	.num_funcs = 2,
1058c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
1068c2ecf20Sopenharmony_ci
1078c2ecf20Sopenharmony_ci	.funcs = { {
1088c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
1098c2ecf20Sopenharmony_ci		.fn_u = { .ecb = cast6_ecb_dec_8way }
1108c2ecf20Sopenharmony_ci	}, {
1118c2ecf20Sopenharmony_ci		.num_blocks = 1,
1128c2ecf20Sopenharmony_ci		.fn_u = { .ecb = __cast6_decrypt }
1138c2ecf20Sopenharmony_ci	} }
1148c2ecf20Sopenharmony_ci};
1158c2ecf20Sopenharmony_ci
1168c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_dec_cbc = {
1178c2ecf20Sopenharmony_ci	.num_funcs = 2,
1188c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
1198c2ecf20Sopenharmony_ci
1208c2ecf20Sopenharmony_ci	.funcs = { {
1218c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
1228c2ecf20Sopenharmony_ci		.fn_u = { .cbc = cast6_cbc_dec_8way }
1238c2ecf20Sopenharmony_ci	}, {
1248c2ecf20Sopenharmony_ci		.num_blocks = 1,
1258c2ecf20Sopenharmony_ci		.fn_u = { .cbc = __cast6_decrypt }
1268c2ecf20Sopenharmony_ci	} }
1278c2ecf20Sopenharmony_ci};
1288c2ecf20Sopenharmony_ci
1298c2ecf20Sopenharmony_cistatic const struct common_glue_ctx cast6_dec_xts = {
1308c2ecf20Sopenharmony_ci	.num_funcs = 2,
1318c2ecf20Sopenharmony_ci	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
1328c2ecf20Sopenharmony_ci
1338c2ecf20Sopenharmony_ci	.funcs = { {
1348c2ecf20Sopenharmony_ci		.num_blocks = CAST6_PARALLEL_BLOCKS,
1358c2ecf20Sopenharmony_ci		.fn_u = { .xts = cast6_xts_dec_8way }
1368c2ecf20Sopenharmony_ci	}, {
1378c2ecf20Sopenharmony_ci		.num_blocks = 1,
1388c2ecf20Sopenharmony_ci		.fn_u = { .xts = cast6_xts_dec }
1398c2ecf20Sopenharmony_ci	} }
1408c2ecf20Sopenharmony_ci};
1418c2ecf20Sopenharmony_ci
1428c2ecf20Sopenharmony_cistatic int ecb_encrypt(struct skcipher_request *req)
1438c2ecf20Sopenharmony_ci{
1448c2ecf20Sopenharmony_ci	return glue_ecb_req_128bit(&cast6_enc, req);
1458c2ecf20Sopenharmony_ci}
1468c2ecf20Sopenharmony_ci
1478c2ecf20Sopenharmony_cistatic int ecb_decrypt(struct skcipher_request *req)
1488c2ecf20Sopenharmony_ci{
1498c2ecf20Sopenharmony_ci	return glue_ecb_req_128bit(&cast6_dec, req);
1508c2ecf20Sopenharmony_ci}
1518c2ecf20Sopenharmony_ci
1528c2ecf20Sopenharmony_cistatic int cbc_encrypt(struct skcipher_request *req)
1538c2ecf20Sopenharmony_ci{
1548c2ecf20Sopenharmony_ci	return glue_cbc_encrypt_req_128bit(__cast6_encrypt, req);
1558c2ecf20Sopenharmony_ci}
1568c2ecf20Sopenharmony_ci
1578c2ecf20Sopenharmony_cistatic int cbc_decrypt(struct skcipher_request *req)
1588c2ecf20Sopenharmony_ci{
1598c2ecf20Sopenharmony_ci	return glue_cbc_decrypt_req_128bit(&cast6_dec_cbc, req);
1608c2ecf20Sopenharmony_ci}
1618c2ecf20Sopenharmony_ci
1628c2ecf20Sopenharmony_cistatic int ctr_crypt(struct skcipher_request *req)
1638c2ecf20Sopenharmony_ci{
1648c2ecf20Sopenharmony_ci	return glue_ctr_req_128bit(&cast6_ctr, req);
1658c2ecf20Sopenharmony_ci}
1668c2ecf20Sopenharmony_ci
1678c2ecf20Sopenharmony_cistruct cast6_xts_ctx {
1688c2ecf20Sopenharmony_ci	struct cast6_ctx tweak_ctx;
1698c2ecf20Sopenharmony_ci	struct cast6_ctx crypt_ctx;
1708c2ecf20Sopenharmony_ci};
1718c2ecf20Sopenharmony_ci
1728c2ecf20Sopenharmony_cistatic int xts_cast6_setkey(struct crypto_skcipher *tfm, const u8 *key,
1738c2ecf20Sopenharmony_ci			    unsigned int keylen)
1748c2ecf20Sopenharmony_ci{
1758c2ecf20Sopenharmony_ci	struct cast6_xts_ctx *ctx = crypto_skcipher_ctx(tfm);
1768c2ecf20Sopenharmony_ci	int err;
1778c2ecf20Sopenharmony_ci
1788c2ecf20Sopenharmony_ci	err = xts_verify_key(tfm, key, keylen);
1798c2ecf20Sopenharmony_ci	if (err)
1808c2ecf20Sopenharmony_ci		return err;
1818c2ecf20Sopenharmony_ci
1828c2ecf20Sopenharmony_ci	/* first half of xts-key is for crypt */
1838c2ecf20Sopenharmony_ci	err = __cast6_setkey(&ctx->crypt_ctx, key, keylen / 2);
1848c2ecf20Sopenharmony_ci	if (err)
1858c2ecf20Sopenharmony_ci		return err;
1868c2ecf20Sopenharmony_ci
1878c2ecf20Sopenharmony_ci	/* second half of xts-key is for tweak */
1888c2ecf20Sopenharmony_ci	return __cast6_setkey(&ctx->tweak_ctx, key + keylen / 2, keylen / 2);
1898c2ecf20Sopenharmony_ci}
1908c2ecf20Sopenharmony_ci
1918c2ecf20Sopenharmony_cistatic int xts_encrypt(struct skcipher_request *req)
1928c2ecf20Sopenharmony_ci{
1938c2ecf20Sopenharmony_ci	struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
1948c2ecf20Sopenharmony_ci	struct cast6_xts_ctx *ctx = crypto_skcipher_ctx(tfm);
1958c2ecf20Sopenharmony_ci
1968c2ecf20Sopenharmony_ci	return glue_xts_req_128bit(&cast6_enc_xts, req, __cast6_encrypt,
1978c2ecf20Sopenharmony_ci				   &ctx->tweak_ctx, &ctx->crypt_ctx, false);
1988c2ecf20Sopenharmony_ci}
1998c2ecf20Sopenharmony_ci
2008c2ecf20Sopenharmony_cistatic int xts_decrypt(struct skcipher_request *req)
2018c2ecf20Sopenharmony_ci{
2028c2ecf20Sopenharmony_ci	struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
2038c2ecf20Sopenharmony_ci	struct cast6_xts_ctx *ctx = crypto_skcipher_ctx(tfm);
2048c2ecf20Sopenharmony_ci
2058c2ecf20Sopenharmony_ci	return glue_xts_req_128bit(&cast6_dec_xts, req, __cast6_encrypt,
2068c2ecf20Sopenharmony_ci				   &ctx->tweak_ctx, &ctx->crypt_ctx, true);
2078c2ecf20Sopenharmony_ci}
2088c2ecf20Sopenharmony_ci
2098c2ecf20Sopenharmony_cistatic struct skcipher_alg cast6_algs[] = {
2108c2ecf20Sopenharmony_ci	{
2118c2ecf20Sopenharmony_ci		.base.cra_name		= "__ecb(cast6)",
2128c2ecf20Sopenharmony_ci		.base.cra_driver_name	= "__ecb-cast6-avx",
2138c2ecf20Sopenharmony_ci		.base.cra_priority	= 200,
2148c2ecf20Sopenharmony_ci		.base.cra_flags		= CRYPTO_ALG_INTERNAL,
2158c2ecf20Sopenharmony_ci		.base.cra_blocksize	= CAST6_BLOCK_SIZE,
2168c2ecf20Sopenharmony_ci		.base.cra_ctxsize	= sizeof(struct cast6_ctx),
2178c2ecf20Sopenharmony_ci		.base.cra_module	= THIS_MODULE,
2188c2ecf20Sopenharmony_ci		.min_keysize		= CAST6_MIN_KEY_SIZE,
2198c2ecf20Sopenharmony_ci		.max_keysize		= CAST6_MAX_KEY_SIZE,
2208c2ecf20Sopenharmony_ci		.setkey			= cast6_setkey_skcipher,
2218c2ecf20Sopenharmony_ci		.encrypt		= ecb_encrypt,
2228c2ecf20Sopenharmony_ci		.decrypt		= ecb_decrypt,
2238c2ecf20Sopenharmony_ci	}, {
2248c2ecf20Sopenharmony_ci		.base.cra_name		= "__cbc(cast6)",
2258c2ecf20Sopenharmony_ci		.base.cra_driver_name	= "__cbc-cast6-avx",
2268c2ecf20Sopenharmony_ci		.base.cra_priority	= 200,
2278c2ecf20Sopenharmony_ci		.base.cra_flags		= CRYPTO_ALG_INTERNAL,
2288c2ecf20Sopenharmony_ci		.base.cra_blocksize	= CAST6_BLOCK_SIZE,
2298c2ecf20Sopenharmony_ci		.base.cra_ctxsize	= sizeof(struct cast6_ctx),
2308c2ecf20Sopenharmony_ci		.base.cra_module	= THIS_MODULE,
2318c2ecf20Sopenharmony_ci		.min_keysize		= CAST6_MIN_KEY_SIZE,
2328c2ecf20Sopenharmony_ci		.max_keysize		= CAST6_MAX_KEY_SIZE,
2338c2ecf20Sopenharmony_ci		.ivsize			= CAST6_BLOCK_SIZE,
2348c2ecf20Sopenharmony_ci		.setkey			= cast6_setkey_skcipher,
2358c2ecf20Sopenharmony_ci		.encrypt		= cbc_encrypt,
2368c2ecf20Sopenharmony_ci		.decrypt		= cbc_decrypt,
2378c2ecf20Sopenharmony_ci	}, {
2388c2ecf20Sopenharmony_ci		.base.cra_name		= "__ctr(cast6)",
2398c2ecf20Sopenharmony_ci		.base.cra_driver_name	= "__ctr-cast6-avx",
2408c2ecf20Sopenharmony_ci		.base.cra_priority	= 200,
2418c2ecf20Sopenharmony_ci		.base.cra_flags		= CRYPTO_ALG_INTERNAL,
2428c2ecf20Sopenharmony_ci		.base.cra_blocksize	= 1,
2438c2ecf20Sopenharmony_ci		.base.cra_ctxsize	= sizeof(struct cast6_ctx),
2448c2ecf20Sopenharmony_ci		.base.cra_module	= THIS_MODULE,
2458c2ecf20Sopenharmony_ci		.min_keysize		= CAST6_MIN_KEY_SIZE,
2468c2ecf20Sopenharmony_ci		.max_keysize		= CAST6_MAX_KEY_SIZE,
2478c2ecf20Sopenharmony_ci		.ivsize			= CAST6_BLOCK_SIZE,
2488c2ecf20Sopenharmony_ci		.chunksize		= CAST6_BLOCK_SIZE,
2498c2ecf20Sopenharmony_ci		.setkey			= cast6_setkey_skcipher,
2508c2ecf20Sopenharmony_ci		.encrypt		= ctr_crypt,
2518c2ecf20Sopenharmony_ci		.decrypt		= ctr_crypt,
2528c2ecf20Sopenharmony_ci	}, {
2538c2ecf20Sopenharmony_ci		.base.cra_name		= "__xts(cast6)",
2548c2ecf20Sopenharmony_ci		.base.cra_driver_name	= "__xts-cast6-avx",
2558c2ecf20Sopenharmony_ci		.base.cra_priority	= 200,
2568c2ecf20Sopenharmony_ci		.base.cra_flags		= CRYPTO_ALG_INTERNAL,
2578c2ecf20Sopenharmony_ci		.base.cra_blocksize	= CAST6_BLOCK_SIZE,
2588c2ecf20Sopenharmony_ci		.base.cra_ctxsize	= sizeof(struct cast6_xts_ctx),
2598c2ecf20Sopenharmony_ci		.base.cra_module	= THIS_MODULE,
2608c2ecf20Sopenharmony_ci		.min_keysize		= 2 * CAST6_MIN_KEY_SIZE,
2618c2ecf20Sopenharmony_ci		.max_keysize		= 2 * CAST6_MAX_KEY_SIZE,
2628c2ecf20Sopenharmony_ci		.ivsize			= CAST6_BLOCK_SIZE,
2638c2ecf20Sopenharmony_ci		.setkey			= xts_cast6_setkey,
2648c2ecf20Sopenharmony_ci		.encrypt		= xts_encrypt,
2658c2ecf20Sopenharmony_ci		.decrypt		= xts_decrypt,
2668c2ecf20Sopenharmony_ci	},
2678c2ecf20Sopenharmony_ci};
2688c2ecf20Sopenharmony_ci
2698c2ecf20Sopenharmony_cistatic struct simd_skcipher_alg *cast6_simd_algs[ARRAY_SIZE(cast6_algs)];
2708c2ecf20Sopenharmony_ci
2718c2ecf20Sopenharmony_cistatic int __init cast6_init(void)
2728c2ecf20Sopenharmony_ci{
2738c2ecf20Sopenharmony_ci	const char *feature_name;
2748c2ecf20Sopenharmony_ci
2758c2ecf20Sopenharmony_ci	if (!cpu_has_xfeatures(XFEATURE_MASK_SSE | XFEATURE_MASK_YMM,
2768c2ecf20Sopenharmony_ci				&feature_name)) {
2778c2ecf20Sopenharmony_ci		pr_info("CPU feature '%s' is not supported.\n", feature_name);
2788c2ecf20Sopenharmony_ci		return -ENODEV;
2798c2ecf20Sopenharmony_ci	}
2808c2ecf20Sopenharmony_ci
2818c2ecf20Sopenharmony_ci	return simd_register_skciphers_compat(cast6_algs,
2828c2ecf20Sopenharmony_ci					      ARRAY_SIZE(cast6_algs),
2838c2ecf20Sopenharmony_ci					      cast6_simd_algs);
2848c2ecf20Sopenharmony_ci}
2858c2ecf20Sopenharmony_ci
2868c2ecf20Sopenharmony_cistatic void __exit cast6_exit(void)
2878c2ecf20Sopenharmony_ci{
2888c2ecf20Sopenharmony_ci	simd_unregister_skciphers(cast6_algs, ARRAY_SIZE(cast6_algs),
2898c2ecf20Sopenharmony_ci				  cast6_simd_algs);
2908c2ecf20Sopenharmony_ci}
2918c2ecf20Sopenharmony_ci
2928c2ecf20Sopenharmony_cimodule_init(cast6_init);
2938c2ecf20Sopenharmony_cimodule_exit(cast6_exit);
2948c2ecf20Sopenharmony_ci
2958c2ecf20Sopenharmony_ciMODULE_DESCRIPTION("Cast6 Cipher Algorithm, AVX optimized");
2968c2ecf20Sopenharmony_ciMODULE_LICENSE("GPL");
2978c2ecf20Sopenharmony_ciMODULE_ALIAS_CRYPTO("cast6");
298