1/*
  2 * Using hardware provided CRC32 instruction to accelerate the CRC32 disposal.
  3 * CRC32C polynomial:0x1EDC6F41(BE)/0x82F63B78(LE)
  4 * CRC32 is a new instruction in Intel SSE4.2, the reference can be found at:
  5 * http://www.intel.com/products/processor/manuals/
  6 * Intel(R) 64 and IA-32 Architectures Software Developer's Manual
  7 * Volume 2A: Instruction Set Reference, A-M
  8 *
  9 * Copyright (C) 2008 Intel Corporation
 10 * Authors: Austin Zhang <austin_zhang@linux.intel.com>
 11 *          Kent Liu <kent.liu@intel.com>
 12 *
 13 * This program is free software; you can redistribute it and/or modify it
 14 * under the terms and conditions of the GNU General Public License,
 15 * version 2, as published by the Free Software Foundation.
 16 *
 17 * This program is distributed in the hope it will be useful, but WITHOUT
 18 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 19 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 20 * more details.
 21 *
 22 * You should have received a copy of the GNU General Public License along with
 23 * this program; if not, write to the Free Software Foundation, Inc.,
 24 * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 25 *
 26 */
 27#include <linux/init.h>
 28#include <linux/module.h>
 29#include <linux/string.h>
 30#include <linux/kernel.h>
 31#include <crypto/internal/hash.h>
 32
 33#include <asm/cpufeature.h>
 34#include <asm/cpu_device_id.h>
 35
 36#define CHKSUM_BLOCK_SIZE	1
 37#define CHKSUM_DIGEST_SIZE	4
 38
 39#define SCALE_F	sizeof(unsigned long)
 40
 41#ifdef CONFIG_X86_64
 42#define REX_PRE "0x48, "
 43#else
 44#define REX_PRE
 45#endif
 46
 47static u32 crc32c_intel_le_hw_byte(u32 crc, unsigned char const *data, size_t length)
 48{
 49	while (length--) {
 50		__asm__ __volatile__(
 51			".byte 0xf2, 0xf, 0x38, 0xf0, 0xf1"
 52			:"=S"(crc)
 53			:"0"(crc), "c"(*data)
 54		);
 55		data++;
 56	}
 57
 58	return crc;
 59}
 60
 61static u32 __pure crc32c_intel_le_hw(u32 crc, unsigned char const *p, size_t len)
 62{
 63	unsigned int iquotient = len / SCALE_F;
 64	unsigned int iremainder = len % SCALE_F;
 65	unsigned long *ptmp = (unsigned long *)p;
 66
 67	while (iquotient--) {
 68		__asm__ __volatile__(
 69			".byte 0xf2, " REX_PRE "0xf, 0x38, 0xf1, 0xf1;"
 70			:"=S"(crc)
 71			:"0"(crc), "c"(*ptmp)
 72		);
 73		ptmp++;
 74	}
 75
 76	if (iremainder)
 77		crc = crc32c_intel_le_hw_byte(crc, (unsigned char *)ptmp,
 78				 iremainder);
 79
 80	return crc;
 81}
 82
 83/*
 84 * Setting the seed allows arbitrary accumulators and flexible XOR policy
 85 * If your algorithm starts with ~0, then XOR with ~0 before you set
 86 * the seed.
 87 */
 88static int crc32c_intel_setkey(struct crypto_shash *hash, const u8 *key,
 89			unsigned int keylen)
 90{
 91	u32 *mctx = crypto_shash_ctx(hash);
 92
 93	if (keylen != sizeof(u32)) {
 94		crypto_shash_set_flags(hash, CRYPTO_TFM_RES_BAD_KEY_LEN);
 95		return -EINVAL;
 96	}
 97	*mctx = le32_to_cpup((__le32 *)key);
 98	return 0;
 99}
100
101static int crc32c_intel_init(struct shash_desc *desc)
102{
103	u32 *mctx = crypto_shash_ctx(desc->tfm);
104	u32 *crcp = shash_desc_ctx(desc);
105
106	*crcp = *mctx;
107
108	return 0;
109}
110
111static int crc32c_intel_update(struct shash_desc *desc, const u8 *data,
112			       unsigned int len)
113{
114	u32 *crcp = shash_desc_ctx(desc);
115
116	*crcp = crc32c_intel_le_hw(*crcp, data, len);
117	return 0;
118}
119
120static int __crc32c_intel_finup(u32 *crcp, const u8 *data, unsigned int len,
121				u8 *out)
122{
123	*(__le32 *)out = ~cpu_to_le32(crc32c_intel_le_hw(*crcp, data, len));
124	return 0;
125}
126
127static int crc32c_intel_finup(struct shash_desc *desc, const u8 *data,
128			      unsigned int len, u8 *out)
129{
130	return __crc32c_intel_finup(shash_desc_ctx(desc), data, len, out);
131}
132
133static int crc32c_intel_final(struct shash_desc *desc, u8 *out)
134{
135	u32 *crcp = shash_desc_ctx(desc);
136
137	*(__le32 *)out = ~cpu_to_le32p(crcp);
138	return 0;
139}
140
141static int crc32c_intel_digest(struct shash_desc *desc, const u8 *data,
142			       unsigned int len, u8 *out)
143{
144	return __crc32c_intel_finup(crypto_shash_ctx(desc->tfm), data, len,
145				    out);
146}
147
148static int crc32c_intel_cra_init(struct crypto_tfm *tfm)
149{
150	u32 *key = crypto_tfm_ctx(tfm);
151
152	*key = ~0;
153
154	return 0;
155}
156
157static struct shash_alg alg = {
158	.setkey			=	crc32c_intel_setkey,
159	.init			=	crc32c_intel_init,
160	.update			=	crc32c_intel_update,
161	.final			=	crc32c_intel_final,
162	.finup			=	crc32c_intel_finup,
163	.digest			=	crc32c_intel_digest,
164	.descsize		=	sizeof(u32),
165	.digestsize		=	CHKSUM_DIGEST_SIZE,
166	.base			=	{
167		.cra_name		=	"crc32c",
168		.cra_driver_name	=	"crc32c-intel",
169		.cra_priority		=	200,
170		.cra_blocksize		=	CHKSUM_BLOCK_SIZE,
171		.cra_ctxsize		=	sizeof(u32),
172		.cra_module		=	THIS_MODULE,
173		.cra_init		=	crc32c_intel_cra_init,
174	}
175};
176
177static const struct x86_cpu_id crc32c_cpu_id[] = {
178	X86_FEATURE_MATCH(X86_FEATURE_XMM4_2),
179	{}
180};
181MODULE_DEVICE_TABLE(x86cpu, crc32c_cpu_id);
182
183static int __init crc32c_intel_mod_init(void)
184{
185	if (!x86_match_cpu(crc32c_cpu_id))
186		return -ENODEV;
187	return crypto_register_shash(&alg);
188}
189
190static void __exit crc32c_intel_mod_fini(void)
191{
192	crypto_unregister_shash(&alg);
193}
194
195module_init(crc32c_intel_mod_init);
196module_exit(crc32c_intel_mod_fini);
197
198MODULE_AUTHOR("Austin Zhang <austin.zhang@intel.com>, Kent Liu <kent.liu@intel.com>");
199MODULE_DESCRIPTION("CRC32c (Castagnoli) optimization using Intel Hardware.");
200MODULE_LICENSE("GPL");
201
202MODULE_ALIAS("crc32c");
203MODULE_ALIAS("crc32c-intel");