[deliverable/linux.git] / arch / x86 / crypto / chacha20_glue.c

/*
 * ChaCha20 256-bit cipher algorithm, RFC7539, SIMD glue code
 *
 * Copyright (C) 2015 Martin Willi
 *
 * 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 <crypto/algapi.h>
#include <crypto/chacha20.h>
#include <linux/crypto.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <asm/fpu/api.h>
#include <asm/simd.h>

#define CHACHA20_STATE_ALIGN 16

asmlinkage void chacha20_block_xor_ssse3(u32 *state, u8 *dst, const u8 *src);
asmlinkage void chacha20_4block_xor_ssse3(u32 *state, u8 *dst, const u8 *src);
#ifdef CONFIG_AS_AVX2
asmlinkage void chacha20_8block_xor_avx2(u32 *state, u8 *dst, const u8 *src);
static bool chacha20_use_avx2;
#endif

static void chacha20_dosimd(u32 *state, u8 *dst, const u8 *src,
			    unsigned int bytes)
{
	u8 buf[CHACHA20_BLOCK_SIZE];

#ifdef CONFIG_AS_AVX2
	if (chacha20_use_avx2) {
		while (bytes >= CHACHA20_BLOCK_SIZE * 8) {
			chacha20_8block_xor_avx2(state, dst, src);
			bytes -= CHACHA20_BLOCK_SIZE * 8;
			src += CHACHA20_BLOCK_SIZE * 8;
			dst += CHACHA20_BLOCK_SIZE * 8;
			state[12] += 8;
		}
	}
#endif
	while (bytes >= CHACHA20_BLOCK_SIZE * 4) {
		chacha20_4block_xor_ssse3(state, dst, src);
		bytes -= CHACHA20_BLOCK_SIZE * 4;
		src += CHACHA20_BLOCK_SIZE * 4;
		dst += CHACHA20_BLOCK_SIZE * 4;
		state[12] += 4;
	}
	while (bytes >= CHACHA20_BLOCK_SIZE) {
		chacha20_block_xor_ssse3(state, dst, src);
		bytes -= CHACHA20_BLOCK_SIZE;
		src += CHACHA20_BLOCK_SIZE;
		dst += CHACHA20_BLOCK_SIZE;
		state[12]++;
	}
	if (bytes) {
		memcpy(buf, src, bytes);
		chacha20_block_xor_ssse3(state, buf, buf);
		memcpy(dst, buf, bytes);
	}
}

static int chacha20_simd(struct blkcipher_desc *desc, struct scatterlist *dst,
			 struct scatterlist *src, unsigned int nbytes)
{
	u32 *state, state_buf[16 + (CHACHA20_STATE_ALIGN / sizeof(u32)) - 1];
	struct blkcipher_walk walk;
	int err;

	if (nbytes <= CHACHA20_BLOCK_SIZE || !may_use_simd())
		return crypto_chacha20_crypt(desc, dst, src, nbytes);

	state = (u32 *)roundup((uintptr_t)state_buf, CHACHA20_STATE_ALIGN);

	blkcipher_walk_init(&walk, dst, src, nbytes);
	err = blkcipher_walk_virt_block(desc, &walk, CHACHA20_BLOCK_SIZE);

	crypto_chacha20_init(state, crypto_blkcipher_ctx(desc->tfm), walk.iv);

	kernel_fpu_begin();

	while (walk.nbytes >= CHACHA20_BLOCK_SIZE) {
		chacha20_dosimd(state, walk.dst.virt.addr, walk.src.virt.addr,
				rounddown(walk.nbytes, CHACHA20_BLOCK_SIZE));
		err = blkcipher_walk_done(desc, &walk,
					  walk.nbytes % CHACHA20_BLOCK_SIZE);
	}

	if (walk.nbytes) {
		chacha20_dosimd(state, walk.dst.virt.addr, walk.src.virt.addr,
				walk.nbytes);
		err = blkcipher_walk_done(desc, &walk, 0);
	}

	kernel_fpu_end();

	return err;
}

static struct crypto_alg alg = {
	.cra_name		= "chacha20",
	.cra_driver_name	= "chacha20-simd",
	.cra_priority		= 300,
	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
	.cra_blocksize		= 1,
	.cra_type		= &crypto_blkcipher_type,
	.cra_ctxsize		= sizeof(struct chacha20_ctx),
	.cra_alignmask		= sizeof(u32) - 1,
	.cra_module		= THIS_MODULE,
	.cra_u			= {
		.blkcipher = {
			.min_keysize	= CHACHA20_KEY_SIZE,
			.max_keysize	= CHACHA20_KEY_SIZE,
			.ivsize		= CHACHA20_IV_SIZE,
			.geniv		= "seqiv",
			.setkey		= crypto_chacha20_setkey,
			.encrypt	= chacha20_simd,
			.decrypt	= chacha20_simd,
		},
	},
};

static int __init chacha20_simd_mod_init(void)
{
	if (!boot_cpu_has(X86_FEATURE_SSSE3))
		return -ENODEV;

#ifdef CONFIG_AS_AVX2
	chacha20_use_avx2 = boot_cpu_has(X86_FEATURE_AVX) &&
			    boot_cpu_has(X86_FEATURE_AVX2) &&
			    cpu_has_xfeatures(XFEATURE_MASK_SSE | XFEATURE_MASK_YMM, NULL);
#endif
	return crypto_register_alg(&alg);
}

static void __exit chacha20_simd_mod_fini(void)
{
	crypto_unregister_alg(&alg);
}

module_init(chacha20_simd_mod_init);
module_exit(chacha20_simd_mod_fini);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Martin Willi <martin@strongswan.org>");
MODULE_DESCRIPTION("chacha20 cipher algorithm, SIMD accelerated");
MODULE_ALIAS_CRYPTO("chacha20");
MODULE_ALIAS_CRYPTO("chacha20-simd");
Commit	Line	Data
c9320b6d MW	1	/*
	2	* ChaCha20 256-bit cipher algorithm, RFC7539, SIMD glue code
	3	*
	4	* Copyright (C) 2015 Martin Willi
	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	#include <crypto/algapi.h>
	13	#include <crypto/chacha20.h>
	14	#include <linux/crypto.h>
	15	#include <linux/kernel.h>
	16	#include <linux/module.h>
	17	#include <asm/fpu/api.h>
	18	#include <asm/simd.h>
	19
	20	#define CHACHA20_STATE_ALIGN 16
	21
	22	asmlinkage void chacha20_block_xor_ssse3(u32 state, u8 dst, const u8 *src);
274f938e	23	asmlinkage void chacha20_4block_xor_ssse3(u32 state, u8 dst, const u8 *src);
3d1e93cd MW	24	#ifdef CONFIG_AS_AVX2
	25	asmlinkage void chacha20_8block_xor_avx2(u32 state, u8 dst, const u8 *src);
	26	static bool chacha20_use_avx2;
	27	#endif
c9320b6d MW	28
	29	static void chacha20_dosimd(u32 state, u8 dst, const u8 *src,
	30	unsigned int bytes)
	31	{
	32	u8 buf[CHACHA20_BLOCK_SIZE];
	33
3d1e93cd MW	34	#ifdef CONFIG_AS_AVX2
	35	if (chacha20_use_avx2) {
	36	while (bytes >= CHACHA20_BLOCK_SIZE * 8) {
	37	chacha20_8block_xor_avx2(state, dst, src);
	38	bytes -= CHACHA20_BLOCK_SIZE * 8;
	39	src += CHACHA20_BLOCK_SIZE * 8;
	40	dst += CHACHA20_BLOCK_SIZE * 8;
	41	state[12] += 8;
	42	}
	43	}
	44	#endif
274f938e MW	45	while (bytes >= CHACHA20_BLOCK_SIZE * 4) {
	46	chacha20_4block_xor_ssse3(state, dst, src);
	47	bytes -= CHACHA20_BLOCK_SIZE * 4;
	48	src += CHACHA20_BLOCK_SIZE * 4;
	49	dst += CHACHA20_BLOCK_SIZE * 4;
	50	state[12] += 4;
	51	}
c9320b6d MW	52	while (bytes >= CHACHA20_BLOCK_SIZE) {
	53	chacha20_block_xor_ssse3(state, dst, src);
	54	bytes -= CHACHA20_BLOCK_SIZE;
	55	src += CHACHA20_BLOCK_SIZE;
	56	dst += CHACHA20_BLOCK_SIZE;
	57	state[12]++;
	58	}
	59	if (bytes) {
	60	memcpy(buf, src, bytes);
	61	chacha20_block_xor_ssse3(state, buf, buf);
	62	memcpy(dst, buf, bytes);
	63	}
	64	}
	65
	66	static int chacha20_simd(struct blkcipher_desc desc, struct scatterlist dst,
	67	struct scatterlist *src, unsigned int nbytes)
	68	{
	69	u32 *state, state_buf[16 + (CHACHA20_STATE_ALIGN / sizeof(u32)) - 1];
	70	struct blkcipher_walk walk;
	71	int err;
	72
7ea0da1d	73	if (nbytes <= CHACHA20_BLOCK_SIZE \|\| !may_use_simd())
c9320b6d MW	74	return crypto_chacha20_crypt(desc, dst, src, nbytes);
	75
	76	state = (u32 *)roundup((uintptr_t)state_buf, CHACHA20_STATE_ALIGN);
	77
	78	blkcipher_walk_init(&walk, dst, src, nbytes);
	79	err = blkcipher_walk_virt_block(desc, &walk, CHACHA20_BLOCK_SIZE);
	80
	81	crypto_chacha20_init(state, crypto_blkcipher_ctx(desc->tfm), walk.iv);
	82
	83	kernel_fpu_begin();
	84
	85	while (walk.nbytes >= CHACHA20_BLOCK_SIZE) {
	86	chacha20_dosimd(state, walk.dst.virt.addr, walk.src.virt.addr,
	87	rounddown(walk.nbytes, CHACHA20_BLOCK_SIZE));
	88	err = blkcipher_walk_done(desc, &walk,
	89	walk.nbytes % CHACHA20_BLOCK_SIZE);
	90	}
	91
	92	if (walk.nbytes) {
	93	chacha20_dosimd(state, walk.dst.virt.addr, walk.src.virt.addr,
	94	walk.nbytes);
	95	err = blkcipher_walk_done(desc, &walk, 0);
	96	}
	97
	98	kernel_fpu_end();
	99
	100	return err;
	101	}
	102
	103	static struct crypto_alg alg = {
	104	.cra_name = "chacha20",
	105	.cra_driver_name = "chacha20-simd",
	106	.cra_priority = 300,
	107	.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER,
	108	.cra_blocksize = 1,
	109	.cra_type = &crypto_blkcipher_type,
	110	.cra_ctxsize = sizeof(struct chacha20_ctx),
	111	.cra_alignmask = sizeof(u32) - 1,
	112	.cra_module = THIS_MODULE,
	113	.cra_u = {
	114	.blkcipher = {
	115	.min_keysize = CHACHA20_KEY_SIZE,
	116	.max_keysize = CHACHA20_KEY_SIZE,
	117	.ivsize = CHACHA20_IV_SIZE,
	118	.geniv = "seqiv",
	119	.setkey = crypto_chacha20_setkey,
	120	.encrypt = chacha20_simd,
	121	.decrypt = chacha20_simd,
	122	},
	123	},
	124	};
	125
	126	static int __init chacha20_simd_mod_init(void)
	127	{
362f924b	128	if (!boot_cpu_has(X86_FEATURE_SSSE3))
c9320b6d MW	129	return -ENODEV;
c9320b6d MW	130
3d1e93cd	131	#ifdef CONFIG_AS_AVX2
da154e82 BP	132	chacha20_use_avx2 = boot_cpu_has(X86_FEATURE_AVX) &&
da154e82 BP	133	boot_cpu_has(X86_FEATURE_AVX2) &&
d91cab78	134	cpu_has_xfeatures(XFEATURE_MASK_SSE \| XFEATURE_MASK_YMM, NULL);
3d1e93cd	135	#endif
c9320b6d MW	136	return crypto_register_alg(&alg);
	137	}
	138
	139	static void __exit chacha20_simd_mod_fini(void)
	140	{
	141	crypto_unregister_alg(&alg);
	142	}
	143
	144	module_init(chacha20_simd_mod_init);
	145	module_exit(chacha20_simd_mod_fini);
	146
	147	MODULE_LICENSE("GPL");
	148	MODULE_AUTHOR("Martin Willi <martin@strongswan.org>");
	149	MODULE_DESCRIPTION("chacha20 cipher algorithm, SIMD accelerated");
	150	MODULE_ALIAS_CRYPTO("chacha20");
	151	MODULE_ALIAS_CRYPTO("chacha20-simd");