[linux-block.git] / arch / x86 / crypto / camellia_aesni_avx2_glue.c

/*
 * Glue Code for x86_64/AVX2/AES-NI assembler optimized version of Camellia
 *
 * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/crypto.h>
#include <linux/err.h>
#include <crypto/ablk_helper.h>
#include <crypto/algapi.h>
#include <crypto/ctr.h>
#include <crypto/lrw.h>
#include <crypto/xts.h>
#include <asm/xcr.h>
#include <asm/xsave.h>
#include <asm/crypto/camellia.h>
#include <asm/crypto/glue_helper.h>

#define CAMELLIA_AESNI_PARALLEL_BLOCKS 16
#define CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS 32

/* 32-way AVX2/AES-NI parallel cipher functions */
asmlinkage void camellia_ecb_enc_32way(struct camellia_ctx *ctx, u8 *dst,
				       const u8 *src);
asmlinkage void camellia_ecb_dec_32way(struct camellia_ctx *ctx, u8 *dst,
				       const u8 *src);

asmlinkage void camellia_cbc_dec_32way(struct camellia_ctx *ctx, u8 *dst,
				       const u8 *src);
asmlinkage void camellia_ctr_32way(struct camellia_ctx *ctx, u8 *dst,
				   const u8 *src, le128 *iv);

asmlinkage void camellia_xts_enc_32way(struct camellia_ctx *ctx, u8 *dst,
				       const u8 *src, le128 *iv);
asmlinkage void camellia_xts_dec_32way(struct camellia_ctx *ctx, u8 *dst,
				       const u8 *src, le128 *iv);

static const struct common_glue_ctx camellia_enc = {
	.num_funcs = 4,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_16way) }
	}, {
		.num_blocks = 2,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk_2way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk) }
	} }
};

static const struct common_glue_ctx camellia_ctr = {
	.num_funcs = 4,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_16way) }
	}, {
		.num_blocks = 2,
		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr_2way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr) }
	} }
};

static const struct common_glue_ctx camellia_enc_xts = {
	.num_funcs = 3,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_16way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc) }
	} }
};

static const struct common_glue_ctx camellia_dec = {
	.num_funcs = 4,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_16way) }
	}, {
		.num_blocks = 2,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk_2way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk) }
	} }
};

static const struct common_glue_ctx camellia_dec_cbc = {
	.num_funcs = 4,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_16way) }
	}, {
		.num_blocks = 2,
		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_decrypt_cbc_2way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_dec_blk) }
	} }
};

static const struct common_glue_ctx camellia_dec_xts = {
	.num_funcs = 3,
	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,

	.funcs = { {
		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_32way) }
	}, {
		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_16way) }
	}, {
		.num_blocks = 1,
		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec) }
	} }
};

static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	return glue_ecb_crypt_128bit(&camellia_enc, desc, dst, src, nbytes);
}

static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	return glue_ecb_crypt_128bit(&camellia_dec, desc, dst, src, nbytes);
}

static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(camellia_enc_blk), desc,
				       dst, src, nbytes);
}

static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	return glue_cbc_decrypt_128bit(&camellia_dec_cbc, desc, dst, src,
				       nbytes);
}

static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		     struct scatterlist *src, unsigned int nbytes)
{
	return glue_ctr_crypt_128bit(&camellia_ctr, desc, dst, src, nbytes);
}

static inline bool camellia_fpu_begin(bool fpu_enabled, unsigned int nbytes)
{
	return glue_fpu_begin(CAMELLIA_BLOCK_SIZE,
			      CAMELLIA_AESNI_PARALLEL_BLOCKS, NULL, fpu_enabled,
			      nbytes);
}

static inline void camellia_fpu_end(bool fpu_enabled)
{
	glue_fpu_end(fpu_enabled);
}

static int camellia_setkey(struct crypto_tfm *tfm, const u8 *in_key,
			   unsigned int key_len)
{
	return __camellia_setkey(crypto_tfm_ctx(tfm), in_key, key_len,
				 &tfm->crt_flags);
}

struct crypt_priv {
	struct camellia_ctx *ctx;
	bool fpu_enabled;
};

static void encrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
{
	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
	struct crypt_priv *ctx = priv;
	int i;

	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);

	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
		camellia_ecb_enc_32way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
	}

	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
		camellia_ecb_enc_16way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
	}

	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
		camellia_enc_blk_2way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
	}

	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
		camellia_enc_blk(ctx->ctx, srcdst, srcdst);
}

static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
{
	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
	struct crypt_priv *ctx = priv;
	int i;

	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);

	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
		camellia_ecb_dec_32way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
	}

	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
		camellia_ecb_dec_16way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
	}

	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
		camellia_dec_blk_2way(ctx->ctx, srcdst, srcdst);
		srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
		nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
	}

	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
		camellia_dec_blk(ctx->ctx, srcdst, srcdst);
}

static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
	struct crypt_priv crypt_ctx = {
		.ctx = &ctx->camellia_ctx,
		.fpu_enabled = false,
	};
	struct lrw_crypt_req req = {
		.tbuf = buf,
		.tbuflen = sizeof(buf),

		.table_ctx = &ctx->lrw_table,
		.crypt_ctx = &crypt_ctx,
		.crypt_fn = encrypt_callback,
	};
	int ret;

	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
	ret = lrw_crypt(desc, dst, src, nbytes, &req);
	camellia_fpu_end(crypt_ctx.fpu_enabled);

	return ret;
}

static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
	struct crypt_priv crypt_ctx = {
		.ctx = &ctx->camellia_ctx,
		.fpu_enabled = false,
	};
	struct lrw_crypt_req req = {
		.tbuf = buf,
		.tbuflen = sizeof(buf),

		.table_ctx = &ctx->lrw_table,
		.crypt_ctx = &crypt_ctx,
		.crypt_fn = decrypt_callback,
	};
	int ret;

	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
	ret = lrw_crypt(desc, dst, src, nbytes, &req);
	camellia_fpu_end(crypt_ctx.fpu_enabled);

	return ret;
}

static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);

	return glue_xts_crypt_128bit(&camellia_enc_xts, desc, dst, src, nbytes,
				     XTS_TWEAK_CAST(camellia_enc_blk),
				     &ctx->tweak_ctx, &ctx->crypt_ctx);
}

static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
		       struct scatterlist *src, unsigned int nbytes)
{
	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);

	return glue_xts_crypt_128bit(&camellia_dec_xts, desc, dst, src, nbytes,
				     XTS_TWEAK_CAST(camellia_enc_blk),
				     &ctx->tweak_ctx, &ctx->crypt_ctx);
}

static struct crypto_alg cmll_algs[10] = { {
	.cra_name		= "__ecb-camellia-aesni-avx2",
	.cra_driver_name	= "__driver-ecb-camellia-aesni-avx2",
	.cra_priority		= 0,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct camellia_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_blkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_u = {
		.blkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.setkey		= camellia_setkey,
			.encrypt	= ecb_encrypt,
			.decrypt	= ecb_decrypt,
		},
	},
}, {
	.cra_name		= "__cbc-camellia-aesni-avx2",
	.cra_driver_name	= "__driver-cbc-camellia-aesni-avx2",
	.cra_priority		= 0,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct camellia_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_blkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_u = {
		.blkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.setkey		= camellia_setkey,
			.encrypt	= cbc_encrypt,
			.decrypt	= cbc_decrypt,
		},
	},
}, {
	.cra_name		= "__ctr-camellia-aesni-avx2",
	.cra_driver_name	= "__driver-ctr-camellia-aesni-avx2",
	.cra_priority		= 0,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= 1,
	.cra_ctxsize		= sizeof(struct camellia_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_blkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_u = {
		.blkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= camellia_setkey,
			.encrypt	= ctr_crypt,
			.decrypt	= ctr_crypt,
		},
	},
}, {
	.cra_name		= "__lrw-camellia-aesni-avx2",
	.cra_driver_name	= "__driver-lrw-camellia-aesni-avx2",
	.cra_priority		= 0,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct camellia_lrw_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_blkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_exit		= lrw_camellia_exit_tfm,
	.cra_u = {
		.blkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE +
					  CAMELLIA_BLOCK_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE +
					  CAMELLIA_BLOCK_SIZE,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= lrw_camellia_setkey,
			.encrypt	= lrw_encrypt,
			.decrypt	= lrw_decrypt,
		},
	},
}, {
	.cra_name		= "__xts-camellia-aesni-avx2",
	.cra_driver_name	= "__driver-xts-camellia-aesni-avx2",
	.cra_priority		= 0,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct camellia_xts_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_blkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_u = {
		.blkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE * 2,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE * 2,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= xts_camellia_setkey,
			.encrypt	= xts_encrypt,
			.decrypt	= xts_decrypt,
		},
	},
}, {
	.cra_name		= "ecb(camellia)",
	.cra_driver_name	= "ecb-camellia-aesni-avx2",
	.cra_priority		= 500,
	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct async_helper_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_ablkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_init		= ablk_init,
	.cra_exit		= ablk_exit,
	.cra_u = {
		.ablkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.setkey		= ablk_set_key,
			.encrypt	= ablk_encrypt,
			.decrypt	= ablk_decrypt,
		},
	},
}, {
	.cra_name		= "cbc(camellia)",
	.cra_driver_name	= "cbc-camellia-aesni-avx2",
	.cra_priority		= 500,
	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct async_helper_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_ablkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_init		= ablk_init,
	.cra_exit		= ablk_exit,
	.cra_u = {
		.ablkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= ablk_set_key,
			.encrypt	= __ablk_encrypt,
			.decrypt	= ablk_decrypt,
		},
	},
}, {
	.cra_name		= "ctr(camellia)",
	.cra_driver_name	= "ctr-camellia-aesni-avx2",
	.cra_priority		= 500,
	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
	.cra_blocksize		= 1,
	.cra_ctxsize		= sizeof(struct async_helper_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_ablkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_init		= ablk_init,
	.cra_exit		= ablk_exit,
	.cra_u = {
		.ablkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= ablk_set_key,
			.encrypt	= ablk_encrypt,
			.decrypt	= ablk_encrypt,
			.geniv		= "chainiv",
		},
	},
}, {
	.cra_name		= "lrw(camellia)",
	.cra_driver_name	= "lrw-camellia-aesni-avx2",
	.cra_priority		= 500,
	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct async_helper_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_ablkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_init		= ablk_init,
	.cra_exit		= ablk_exit,
	.cra_u = {
		.ablkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE +
					  CAMELLIA_BLOCK_SIZE,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE +
					  CAMELLIA_BLOCK_SIZE,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= ablk_set_key,
			.encrypt	= ablk_encrypt,
			.decrypt	= ablk_decrypt,
		},
	},
}, {
	.cra_name		= "xts(camellia)",
	.cra_driver_name	= "xts-camellia-aesni-avx2",
	.cra_priority		= 500,
	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
	.cra_ctxsize		= sizeof(struct async_helper_ctx),
	.cra_alignmask		= 0,
	.cra_type		= &crypto_ablkcipher_type,
	.cra_module		= THIS_MODULE,
	.cra_init		= ablk_init,
	.cra_exit		= ablk_exit,
	.cra_u = {
		.ablkcipher = {
			.min_keysize	= CAMELLIA_MIN_KEY_SIZE * 2,
			.max_keysize	= CAMELLIA_MAX_KEY_SIZE * 2,
			.ivsize		= CAMELLIA_BLOCK_SIZE,
			.setkey		= ablk_set_key,
			.encrypt	= ablk_encrypt,
			.decrypt	= ablk_decrypt,
		},
	},
} };

static int __init camellia_aesni_init(void)
{
	u64 xcr0;

	if (!cpu_has_avx2 || !cpu_has_avx || !cpu_has_aes || !cpu_has_osxsave) {
		pr_info("AVX2 or AES-NI instructions are not detected.\n");
		return -ENODEV;
	}

	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
		pr_info("AVX2 detected but unusable.\n");
		return -ENODEV;
	}

	return crypto_register_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
}

static void __exit camellia_aesni_fini(void)
{
	crypto_unregister_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
}

module_init(camellia_aesni_init);
module_exit(camellia_aesni_fini);

MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("Camellia Cipher Algorithm, AES-NI/AVX2 optimized");
MODULE_ALIAS("camellia");
MODULE_ALIAS("camellia-asm");
Commit	Line	Data
f3f935a7 JK	1	/*
	2	* Glue Code for x86_64/AVX2/AES-NI assembler optimized version of Camellia
	3	*
	4	* Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
	5	*
	6	* This program is free software; you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation; either version 2 of the License, or
	9	* (at your option) any later version.
	10	*
	11	*/
	12
	13	#include <linux/module.h>
	14	#include <linux/types.h>
	15	#include <linux/crypto.h>
	16	#include <linux/err.h>
801201aa	17	#include <crypto/ablk_helper.h>
f3f935a7 JK	18	#include <crypto/algapi.h>
	19	#include <crypto/ctr.h>
	20	#include <crypto/lrw.h>
	21	#include <crypto/xts.h>
	22	#include <asm/xcr.h>
	23	#include <asm/xsave.h>
	24	#include <asm/crypto/camellia.h>
f3f935a7 JK	25	#include <asm/crypto/glue_helper.h>
	26
	27	#define CAMELLIA_AESNI_PARALLEL_BLOCKS 16
	28	#define CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS 32
	29
	30	/* 32-way AVX2/AES-NI parallel cipher functions */
	31	asmlinkage void camellia_ecb_enc_32way(struct camellia_ctx ctx, u8 dst,
	32	const u8 *src);
	33	asmlinkage void camellia_ecb_dec_32way(struct camellia_ctx ctx, u8 dst,
	34	const u8 *src);
	35
	36	asmlinkage void camellia_cbc_dec_32way(struct camellia_ctx ctx, u8 dst,
	37	const u8 *src);
	38	asmlinkage void camellia_ctr_32way(struct camellia_ctx ctx, u8 dst,
	39	const u8 src, le128 iv);
	40
	41	asmlinkage void camellia_xts_enc_32way(struct camellia_ctx ctx, u8 dst,
	42	const u8 src, le128 iv);
	43	asmlinkage void camellia_xts_dec_32way(struct camellia_ctx ctx, u8 dst,
	44	const u8 src, le128 iv);
	45
	46	static const struct common_glue_ctx camellia_enc = {
	47	.num_funcs = 4,
	48	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
	49
	50	.funcs = { {
	51	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
	52	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_32way) }
	53	}, {
	54	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
	55	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_16way) }
	56	}, {
	57	.num_blocks = 2,
	58	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk_2way) }
	59	}, {
	60	.num_blocks = 1,
	61	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk) }
	62	} }
	63	};
	64
	65	static const struct common_glue_ctx camellia_ctr = {
	66	.num_funcs = 4,
	67	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
	68
	69	.funcs = { {
	70	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
	71	.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_32way) }
	72	}, {
	73	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
	74	.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_16way) }
	75	}, {
	76	.num_blocks = 2,
	77	.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr_2way) }
	78	}, {
	79	.num_blocks = 1,
	80	.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr) }
	81	} }
	82	};
	83
	84	static const struct common_glue_ctx camellia_enc_xts = {
	85	.num_funcs = 3,
	86	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
	87
	88	.funcs = { {
89	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
90	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_32way) }
91	}, {
92	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
93	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_16way) }
94	}, {
95	.num_blocks = 1,
96	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc) }
97	} }
98	};
99
100	static const struct common_glue_ctx camellia_dec = {
101	.num_funcs = 4,
102	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
103
104	.funcs = { {
105	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
106	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_32way) }
107	}, {
108	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
109	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_16way) }
110	}, {
111	.num_blocks = 2,
112	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk_2way) }
113	}, {
114	.num_blocks = 1,
115	.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk) }
116	} }
117	};
118
119	static const struct common_glue_ctx camellia_dec_cbc = {
120	.num_funcs = 4,
121	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
122
123	.funcs = { {
124	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
125	.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_32way) }
126	}, {
127	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
128	.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_16way) }
129	}, {
130	.num_blocks = 2,
131	.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_decrypt_cbc_2way) }
132	}, {
133	.num_blocks = 1,
134	.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_dec_blk) }
135	} }
136	};
137
138	static const struct common_glue_ctx camellia_dec_xts = {
139	.num_funcs = 3,
140	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
141
142	.funcs = { {
143	.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
144	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_32way) }
145	}, {
146	.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
147	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_16way) }
148	}, {
149	.num_blocks = 1,
150	.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec) }
151	} }
152	};
153
154	static int ecb_encrypt(struct blkcipher_desc desc, struct scatterlist dst,
155	struct scatterlist *src, unsigned int nbytes)
156	{
157	return glue_ecb_crypt_128bit(&camellia_enc, desc, dst, src, nbytes);
158	}
159
160	static int ecb_decrypt(struct blkcipher_desc desc, struct scatterlist dst,
161	struct scatterlist *src, unsigned int nbytes)
162	{
163	return glue_ecb_crypt_128bit(&camellia_dec, desc, dst, src, nbytes);
164	}
165
166	static int cbc_encrypt(struct blkcipher_desc desc, struct scatterlist dst,
167	struct scatterlist *src, unsigned int nbytes)
168	{
169	return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(camellia_enc_blk), desc,
170	dst, src, nbytes);
171	}
172
173	static int cbc_decrypt(struct blkcipher_desc desc, struct scatterlist dst,
174	struct scatterlist *src, unsigned int nbytes)
175	{
176	return glue_cbc_decrypt_128bit(&camellia_dec_cbc, desc, dst, src,
177	nbytes);
178	}
179
180	static int ctr_crypt(struct blkcipher_desc desc, struct scatterlist dst,
181	struct scatterlist *src, unsigned int nbytes)
182	{
183	return glue_ctr_crypt_128bit(&camellia_ctr, desc, dst, src, nbytes);
184	}
185
186	static inline bool camellia_fpu_begin(bool fpu_enabled, unsigned int nbytes)
187	{
188	return glue_fpu_begin(CAMELLIA_BLOCK_SIZE,
189	CAMELLIA_AESNI_PARALLEL_BLOCKS, NULL, fpu_enabled,
190	nbytes);
191	}
192
193	static inline void camellia_fpu_end(bool fpu_enabled)
194	{
195	glue_fpu_end(fpu_enabled);
196	}
197
198	static int camellia_setkey(struct crypto_tfm tfm, const u8 in_key,
199	unsigned int key_len)
200	{
201	return __camellia_setkey(crypto_tfm_ctx(tfm), in_key, key_len,
202	&tfm->crt_flags);
203	}
204
205	struct crypt_priv {
206	struct camellia_ctx *ctx;
207	bool fpu_enabled;
208	};
209
210	static void encrypt_callback(void priv, u8 srcdst, unsigned int nbytes)
211	{
212	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
213	struct crypt_priv *ctx = priv;
214	int i;
215
216	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);
217
218	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
219	camellia_ecb_enc_32way(ctx->ctx, srcdst, srcdst);
220	srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
221	nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
222	}
223
224	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
225	camellia_ecb_enc_16way(ctx->ctx, srcdst, srcdst);
226	srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
227	nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
228	}
229
230	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
231	camellia_enc_blk_2way(ctx->ctx, srcdst, srcdst);
232	srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
233	nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
234	}
235
236	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
237	camellia_enc_blk(ctx->ctx, srcdst, srcdst);
238	}
239
240	static void decrypt_callback(void priv, u8 srcdst, unsigned int nbytes)
241	{
242	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
243	struct crypt_priv *ctx = priv;
244	int i;
245
246	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);
247
248	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
249	camellia_ecb_dec_32way(ctx->ctx, srcdst, srcdst);
250	srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
251	nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
252	}
253
254	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
255	camellia_ecb_dec_16way(ctx->ctx, srcdst, srcdst);
256	srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
257	nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
258	}
259
260	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
261	camellia_dec_blk_2way(ctx->ctx, srcdst, srcdst);
262	srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
263	nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
264	}
265
266	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
267	camellia_dec_blk(ctx->ctx, srcdst, srcdst);
268	}
269
270	static int lrw_encrypt(struct blkcipher_desc desc, struct scatterlist dst,
271	struct scatterlist *src, unsigned int nbytes)
272	{
273	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
274	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
275	struct crypt_priv crypt_ctx = {
276	.ctx = &ctx->camellia_ctx,
277	.fpu_enabled = false,
278	};
279	struct lrw_crypt_req req = {
280	.tbuf = buf,
281	.tbuflen = sizeof(buf),
282
283	.table_ctx = &ctx->lrw_table,
284	.crypt_ctx = &crypt_ctx,
285	.crypt_fn = encrypt_callback,
286	};
287	int ret;
288
289	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
290	ret = lrw_crypt(desc, dst, src, nbytes, &req);
291	camellia_fpu_end(crypt_ctx.fpu_enabled);
292
293	return ret;
294	}
295
296	static int lrw_decrypt(struct blkcipher_desc desc, struct scatterlist dst,
297	struct scatterlist *src, unsigned int nbytes)
298	{
299	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
300	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
301	struct crypt_priv crypt_ctx = {
302	.ctx = &ctx->camellia_ctx,
303	.fpu_enabled = false,
304	};
305	struct lrw_crypt_req req = {
306	.tbuf = buf,
307	.tbuflen = sizeof(buf),
308
309	.table_ctx = &ctx->lrw_table,
310	.crypt_ctx = &crypt_ctx,
311	.crypt_fn = decrypt_callback,
312	};
313	int ret;
314
315	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
316	ret = lrw_crypt(desc, dst, src, nbytes, &req);
317	camellia_fpu_end(crypt_ctx.fpu_enabled);
318
319	return ret;
320	}
321
322	static int xts_encrypt(struct blkcipher_desc desc, struct scatterlist dst,
323	struct scatterlist *src, unsigned int nbytes)
324	{
325	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
326
327	return glue_xts_crypt_128bit(&camellia_enc_xts, desc, dst, src, nbytes,
328	XTS_TWEAK_CAST(camellia_enc_blk),
329	&ctx->tweak_ctx, &ctx->crypt_ctx);
330	}
331
332	static int xts_decrypt(struct blkcipher_desc desc, struct scatterlist dst,
333	struct scatterlist *src, unsigned int nbytes)
334	{
335	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
336
337	return glue_xts_crypt_128bit(&camellia_dec_xts, desc, dst, src, nbytes,
338	XTS_TWEAK_CAST(camellia_enc_blk),
339	&ctx->tweak_ctx, &ctx->crypt_ctx);
340	}
341
342	static struct crypto_alg cmll_algs[10] = { {
343	.cra_name = "__ecb-camellia-aesni-avx2",
344	.cra_driver_name = "__driver-ecb-camellia-aesni-avx2",
345	.cra_priority = 0,
346	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
347	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
348	.cra_ctxsize = sizeof(struct camellia_ctx),
349	.cra_alignmask = 0,
350	.cra_type = &crypto_blkcipher_type,
351	.cra_module = THIS_MODULE,
352	.cra_u = {
353	.blkcipher = {
354	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
355	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
356	.setkey = camellia_setkey,
357	.encrypt = ecb_encrypt,
358	.decrypt = ecb_decrypt,
359	},
360	},
361	}, {
362	.cra_name = "__cbc-camellia-aesni-avx2",
363	.cra_driver_name = "__driver-cbc-camellia-aesni-avx2",
364	.cra_priority = 0,
365	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
366	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
367	.cra_ctxsize = sizeof(struct camellia_ctx),
368	.cra_alignmask = 0,
369	.cra_type = &crypto_blkcipher_type,
370	.cra_module = THIS_MODULE,
371	.cra_u = {
372	.blkcipher = {
373	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
374	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
375	.setkey = camellia_setkey,
376	.encrypt = cbc_encrypt,
377	.decrypt = cbc_decrypt,
378	},
379	},
380	}, {
381	.cra_name = "__ctr-camellia-aesni-avx2",
382	.cra_driver_name = "__driver-ctr-camellia-aesni-avx2",
383	.cra_priority = 0,
384	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
385	.cra_blocksize = 1,
386	.cra_ctxsize = sizeof(struct camellia_ctx),
387	.cra_alignmask = 0,
388	.cra_type = &crypto_blkcipher_type,
389	.cra_module = THIS_MODULE,
390	.cra_u = {
391	.blkcipher = {
392	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
393	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
394	.ivsize = CAMELLIA_BLOCK_SIZE,
395	.setkey = camellia_setkey,
396	.encrypt = ctr_crypt,
397	.decrypt = ctr_crypt,
398	},
399	},
400	}, {
401	.cra_name = "__lrw-camellia-aesni-avx2",
402	.cra_driver_name = "__driver-lrw-camellia-aesni-avx2",
403	.cra_priority = 0,
404	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
405	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
406	.cra_ctxsize = sizeof(struct camellia_lrw_ctx),
407	.cra_alignmask = 0,
408	.cra_type = &crypto_blkcipher_type,
409	.cra_module = THIS_MODULE,
410	.cra_exit = lrw_camellia_exit_tfm,
411	.cra_u = {
412	.blkcipher = {
413	.min_keysize = CAMELLIA_MIN_KEY_SIZE +
414	CAMELLIA_BLOCK_SIZE,
415	.max_keysize = CAMELLIA_MAX_KEY_SIZE +
416	CAMELLIA_BLOCK_SIZE,
417	.ivsize = CAMELLIA_BLOCK_SIZE,
418	.setkey = lrw_camellia_setkey,
419	.encrypt = lrw_encrypt,
420	.decrypt = lrw_decrypt,
421	},
422	},
423	}, {
424	.cra_name = "__xts-camellia-aesni-avx2",
425	.cra_driver_name = "__driver-xts-camellia-aesni-avx2",
426	.cra_priority = 0,
427	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
428	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
429	.cra_ctxsize = sizeof(struct camellia_xts_ctx),
430	.cra_alignmask = 0,
431	.cra_type = &crypto_blkcipher_type,
432	.cra_module = THIS_MODULE,
433	.cra_u = {
434	.blkcipher = {
435	.min_keysize = CAMELLIA_MIN_KEY_SIZE * 2,
436	.max_keysize = CAMELLIA_MAX_KEY_SIZE * 2,
437	.ivsize = CAMELLIA_BLOCK_SIZE,
438	.setkey = xts_camellia_setkey,
439	.encrypt = xts_encrypt,
440	.decrypt = xts_decrypt,
441	},
442	},
443	}, {
444	.cra_name = "ecb(camellia)",
445	.cra_driver_name = "ecb-camellia-aesni-avx2",
446	.cra_priority = 500,
447	.cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER \| CRYPTO_ALG_ASYNC,
448	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
449	.cra_ctxsize = sizeof(struct async_helper_ctx),
450	.cra_alignmask = 0,
451	.cra_type = &crypto_ablkcipher_type,
452	.cra_module = THIS_MODULE,
453	.cra_init = ablk_init,
454	.cra_exit = ablk_exit,
455	.cra_u = {
456	.ablkcipher = {
457	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
458	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
459	.setkey = ablk_set_key,
460	.encrypt = ablk_encrypt,
461	.decrypt = ablk_decrypt,
462	},
463	},
464	}, {
465	.cra_name = "cbc(camellia)",
466	.cra_driver_name = "cbc-camellia-aesni-avx2",
467	.cra_priority = 500,
468	.cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER \| CRYPTO_ALG_ASYNC,
469	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
470	.cra_ctxsize = sizeof(struct async_helper_ctx),
471	.cra_alignmask = 0,
472	.cra_type = &crypto_ablkcipher_type,
473	.cra_module = THIS_MODULE,
474	.cra_init = ablk_init,
475	.cra_exit = ablk_exit,
476	.cra_u = {
477	.ablkcipher = {
478	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
479	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
480	.ivsize = CAMELLIA_BLOCK_SIZE,
481	.setkey = ablk_set_key,
482	.encrypt = __ablk_encrypt,
483	.decrypt = ablk_decrypt,
484	},
485	},
486	}, {
487	.cra_name = "ctr(camellia)",
488	.cra_driver_name = "ctr-camellia-aesni-avx2",
489	.cra_priority = 500,
490	.cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER \| CRYPTO_ALG_ASYNC,
491	.cra_blocksize = 1,
492	.cra_ctxsize = sizeof(struct async_helper_ctx),
493	.cra_alignmask = 0,
494	.cra_type = &crypto_ablkcipher_type,
495	.cra_module = THIS_MODULE,
496	.cra_init = ablk_init,
497	.cra_exit = ablk_exit,
498	.cra_u = {
499	.ablkcipher = {
500	.min_keysize = CAMELLIA_MIN_KEY_SIZE,
501	.max_keysize = CAMELLIA_MAX_KEY_SIZE,
502	.ivsize = CAMELLIA_BLOCK_SIZE,
503	.setkey = ablk_set_key,
504	.encrypt = ablk_encrypt,
505	.decrypt = ablk_encrypt,
506	.geniv = "chainiv",
507	},
508	},
509	}, {
510	.cra_name = "lrw(camellia)",
511	.cra_driver_name = "lrw-camellia-aesni-avx2",
512	.cra_priority = 500,
513	.cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER \| CRYPTO_ALG_ASYNC,
514	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
515	.cra_ctxsize = sizeof(struct async_helper_ctx),
516	.cra_alignmask = 0,
517	.cra_type = &crypto_ablkcipher_type,
518	.cra_module = THIS_MODULE,
519	.cra_init = ablk_init,
520	.cra_exit = ablk_exit,
521	.cra_u = {
522	.ablkcipher = {
523	.min_keysize = CAMELLIA_MIN_KEY_SIZE +
524	CAMELLIA_BLOCK_SIZE,
525	.max_keysize = CAMELLIA_MAX_KEY_SIZE +
526	CAMELLIA_BLOCK_SIZE,
527	.ivsize = CAMELLIA_BLOCK_SIZE,
528	.setkey = ablk_set_key,
529	.encrypt = ablk_encrypt,
530	.decrypt = ablk_decrypt,
531	},
532	},
533	}, {
534	.cra_name = "xts(camellia)",
535	.cra_driver_name = "xts-camellia-aesni-avx2",
536	.cra_priority = 500,
537	.cra_flags = CRYPTO_ALG_TYPE_ABLKCIPHER \| CRYPTO_ALG_ASYNC,
538	.cra_blocksize = CAMELLIA_BLOCK_SIZE,
539	.cra_ctxsize = sizeof(struct async_helper_ctx),
540	.cra_alignmask = 0,
541	.cra_type = &crypto_ablkcipher_type,
542	.cra_module = THIS_MODULE,
543	.cra_init = ablk_init,
544	.cra_exit = ablk_exit,
545	.cra_u = {
546	.ablkcipher = {
547	.min_keysize = CAMELLIA_MIN_KEY_SIZE * 2,
548	.max_keysize = CAMELLIA_MAX_KEY_SIZE * 2,
549	.ivsize = CAMELLIA_BLOCK_SIZE,
550	.setkey = ablk_set_key,
551	.encrypt = ablk_encrypt,
552	.decrypt = ablk_decrypt,
553	},
554	},
555	} };
556
557	static int __init camellia_aesni_init(void)
558	{
559	u64 xcr0;
560
561	if (!cpu_has_avx2 \|\| !cpu_has_avx \|\| !cpu_has_aes \|\| !cpu_has_osxsave) {
562	pr_info("AVX2 or AES-NI instructions are not detected.\n");
563	return -ENODEV;
564	}
565
566	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
567	if ((xcr0 & (XSTATE_SSE \| XSTATE_YMM)) != (XSTATE_SSE \| XSTATE_YMM)) {
568	pr_info("AVX2 detected but unusable.\n");
569	return -ENODEV;
570	}
571
572	return crypto_register_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
573	}
574
575	static void __exit camellia_aesni_fini(void)
576	{
577	crypto_unregister_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
578	}
579
580	module_init(camellia_aesni_init);
581	module_exit(camellia_aesni_fini);
582
583	MODULE_LICENSE("GPL");
584	MODULE_DESCRIPTION("Camellia Cipher Algorithm, AES-NI/AVX2 optimized");
585	MODULE_ALIAS("camellia");
586	MODULE_ALIAS("camellia-asm");