From: Austin Zhang Subject: [PATCH] Using Intel CRC32 instruction to accelerate CRC32c algorithm by new crypto API. Date: Mon, 04 Aug 2008 05:35:07 -0400 Message-ID: <1217842507.20845.18.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: linux-kernel@vger.kernel.org, linux-crypto@vger.kernel.org To: herbert@gondor.apana.org.au, davem@davemloft.net Return-path: Received: from mga10.intel.com ([192.55.52.92]:56305 "EHLO fmsmga102.fm.intel.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752816AbYHDJgr (ORCPT ); Mon, 4 Aug 2008 05:36:47 -0400 Sender: linux-crypto-owner@vger.kernel.org List-ID: >From Nehalem processor onward, Intel processors can support hardware accelerated CRC32c algorithm with the new CRC32 instruction in SSE 4.2 instruction set. The patch detects the availability of the feature, and chooses the most proper way to calculate CRC32c checksum. Byte code instructions are used for compiler compatibility. No MMX / XMM registers is involved in the implementation. Signed-off-by: Austin Zhang Signed-off-by: Kent Liu --- arch/x86/crypto/Makefile | 2 arch/x86/crypto/crc32c-intel.c | 192 +++++++++++++++++++++++++++++++++++++++++ crypto/Kconfig | 11 ++ include/asm-x86/cpufeature.h | 2 4 files changed, 207 insertions(+) diff -Naurp linux-2.6/arch/x86/crypto/crc32c-intel.c linux-2.6-patch/arch/x86/crypto/crc32c-intel.c --- linux-2.6/arch/x86/crypto/crc32c-intel.c 1969-12-31 19:00:00.000000000 -0500 +++ linux-2.6-patch/arch/x86/crypto/crc32c-intel.c 2008-08-04 01:59:00.000000000 -0400 @@ -0,0 +1,192 @@ +/* + * Using hardware provided CRC32 instruction to accelerate the CRC32 disposal. + * CRC32C polynomial:0x1EDC6F41(BE)/0x82F63B78(LE) + * CRC32 is a new instruction in Intel SSE4.2, the reference can be found at: + * http://www.intel.com/products/processor/manuals/ + * Intel(R) 64 and IA-32 Architectures Software Developer's Manual + * Volume 2A: Instruction Set Reference, A-M + */ +#include +#include +#include +#include +#include + +#include + +#define CHKSUM_BLOCK_SIZE 1 +#define CHKSUM_DIGEST_SIZE 4 + +#ifdef CONFIG_X86_64 +#define REX_PRE "0x48, " +#define SCALE_F 8 +#else +#define REX_PRE +#define SCALE_F 4 +#endif + +u32 crc32c_intel_le_hw_byte(u32 crc, unsigned char const *data, size_t length) +{ + while (length--) { + __asm__ __volatile__( + ".byte 0xf2, 0xf, 0x38, 0xf0, 0xf1" + :"=S"(crc) + :"0"(crc), "c"(*data) + ); + data++; + } + + return crc; +} + +u32 __pure crc32c_intel_le_hw(u32 crc, unsigned char const *p, size_t len) +{ + unsigned int iquotient = len / SCALE_F; + unsigned int iremainder = len % SCALE_F; +#ifdef CONFIG_X86_64 + u64 *ptmp = (u64 *)p; +#else + u32 *ptmp = (u32 *)p; +#endif + + while (iquotient--) { + __asm__ __volatile__( + ".byte 0xf2, " REX_PRE "0xf, 0x38, 0xf1, 0xf1;" + :"=S"(crc) + :"0"(crc), "c"(*ptmp) + ); + ptmp++; + } + + if (iremainder) + crc = crc32c_intel_le_hw_byte(crc, (unsigned char *)ptmp, + iremainder); + + return crc; +} + +/* + * Setting the seed allows arbitrary accumulators and flexible XOR policy + * If your algorithm starts with ~0, then XOR with ~0 before you set + * the seed. + */ +static int crc32c_intel_setkey(struct crypto_ahash *hash, const u8 *key, + unsigned int keylen) +{ + u32 *mctx = crypto_ahash_ctx(hash); + + if (keylen != sizeof(u32)) { + crypto_ahash_set_flags(hash, CRYPTO_TFM_RES_BAD_KEY_LEN); + return -EINVAL; + } + *mctx = le32_to_cpup((__le32 *)key); + return 0; +} + +static int crc32c_intel_init(struct ahash_request *req) +{ + u32 *mctx = crypto_ahash_ctx(crypto_ahash_reqtfm(req)); + u32 *crcp = ahash_request_ctx(req); + + *crcp = *mctx; + + return 0; +} + +static int crc32c_intel_update(struct ahash_request *req) +{ + struct crypto_hash_walk walk; + u32 *crcp = ahash_request_ctx(req); + u32 crc = *crcp; + int nbytes; + + for (nbytes = crypto_hash_walk_first(req, &walk); nbytes; + nbytes = crypto_hash_walk_done(&walk, 0)) + crc = crc32c_intel_le_hw(crc, walk.data, nbytes); + + *crcp = crc; + return 0; +} + +static int crc32c_intel_final(struct ahash_request *req) +{ + u32 *crcp = ahash_request_ctx(req); + + *(__le32 *)req->result = ~cpu_to_le32p(crcp); + return 0; +} + +static int crc32c_intel_digest(struct ahash_request *req) +{ + struct crypto_hash_walk walk; + u32 *mctx = crypto_ahash_ctx(crypto_ahash_reqtfm(req)); + u32 crc = *mctx; + int nbytes; + + for (nbytes = crypto_hash_walk_first(req, &walk); nbytes; + nbytes = crypto_hash_walk_done(&walk, 0)) + crc = crc32c_intel_le_hw(crc, walk.data, nbytes); + + *(__le32 *)req->result = ~cpu_to_le32(crc); + return 0; +} + +static int crc32c_intel_cra_init(struct crypto_tfm *tfm) +{ + u32 *key = crypto_tfm_ctx(tfm); + + *key = ~0; + + tfm->crt_ahash.reqsize = sizeof(u32); + + if (cpu_has_xmm4_2) + return 0; + else + return -1; +} + +static struct crypto_alg alg = { + .cra_name = "crc32c", + .cra_driver_name = "crc32c-intel", + .cra_priority = 200, + .cra_flags = CRYPTO_ALG_TYPE_AHASH, + .cra_blocksize = CHKSUM_BLOCK_SIZE, + .cra_alignmask = 3, + .cra_ctxsize = sizeof(u32), + .cra_module = THIS_MODULE, + .cra_list = LIST_HEAD_INIT(alg.cra_list), + .cra_init = crc32c_intel_cra_init, + .cra_type = &crypto_ahash_type, + .cra_u = { + .ahash = { + .digestsize = CHKSUM_DIGEST_SIZE, + .setkey = crc32c_intel_setkey, + .init = crc32c_intel_init, + .update = crc32c_intel_update, + .final = crc32c_intel_final, + .digest = crc32c_intel_digest, + } + } +}; + + +static int __init crc32c_intel_mod_init(void) +{ + return crypto_register_alg(&alg); +} + +static void __exit crc32c_intel_mod_fini(void) +{ + crypto_unregister_alg(&alg); +} + +module_init(crc32c_intel_mod_init); +module_exit(crc32c_intel_mod_fini); + +MODULE_AUTHOR("Austin Zhang , Kent Liu "); +MODULE_DESCRIPTION("CRC32c (Castagnoli) optimization using Intel Hardware."); +MODULE_LICENSE("GPL"); + +MODULE_ALIAS("crc32c"); +MODULE_ALIAS("crc32c-intel"); + diff -Naurp linux-2.6/arch/x86/crypto/Makefile linux-2.6-patch/arch/x86/crypto/Makefile --- linux-2.6/arch/x86/crypto/Makefile 2008-08-04 01:08:00.000000000 -0400 +++ linux-2.6-patch/arch/x86/crypto/Makefile 2008-08-04 01:59:00.000000000 -0400 @@ -10,6 +10,8 @@ obj-$(CONFIG_CRYPTO_AES_X86_64) += aes-x obj-$(CONFIG_CRYPTO_TWOFISH_X86_64) += twofish-x86_64.o obj-$(CONFIG_CRYPTO_SALSA20_X86_64) += salsa20-x86_64.o +obj-$(CONFIG_CRYPTO_CRC32C_INTEL) += crc32c-intel.o + aes-i586-y := aes-i586-asm_32.o aes_glue.o twofish-i586-y := twofish-i586-asm_32.o twofish_glue.o salsa20-i586-y := salsa20-i586-asm_32.o salsa20_glue.o diff -Naurp linux-2.6/crypto/Kconfig linux-2.6-patch/crypto/Kconfig --- linux-2.6/crypto/Kconfig 2008-08-04 01:08:00.000000000 -0400 +++ linux-2.6-patch/crypto/Kconfig 2008-08-04 01:59:00.000000000 -0400 @@ -221,6 +221,17 @@ config CRYPTO_CRC32C See Castagnoli93. This implementation uses lib/libcrc32c. Module will be crc32c. +config CRYPTO_CRC32C_INTEL + tristate "CRC32c INTEL hardware acceleration" + select CRYPTO_HASH + help + In Intel processor with SSE4.2 supported, the processor will + support CRC32C implemetation using hardware accelerated CRC32 + instruction. This option will create 'crc32c-intel' module, + which will enable any routine to use the CRC32 instruction to + gain performance compared with software implementation. + Module will be crc32c-intel. + config CRYPTO_MD4 tristate "MD4 digest algorithm" select CRYPTO_ALGAPI diff -Naurp linux-2.6/include/asm-x86/cpufeature.h linux-2.6-patch/include/asm-x86/cpufeature.h --- linux-2.6/include/asm-x86/cpufeature.h 2008-08-04 01:08:08.000000000 -0400 +++ linux-2.6-patch/include/asm-x86/cpufeature.h 2008-08-04 01:59:00.000000000 -0400 @@ -91,6 +91,7 @@ #define X86_FEATURE_CX16 (4*32+13) /* CMPXCHG16B */ #define X86_FEATURE_XTPR (4*32+14) /* Send Task Priority Messages */ #define X86_FEATURE_DCA (4*32+18) /* Direct Cache Access */ +#define X86_FEATURE_XMM4_2 (4*32+20) /* Streaming SIMD Extensions-4.2 */ /* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */ #define X86_FEATURE_XSTORE (5*32+ 2) /* on-CPU RNG present (xstore insn) */ @@ -189,6 +190,7 @@ extern const char * const x86_power_flag #define cpu_has_gbpages boot_cpu_has(X86_FEATURE_GBPAGES) #define cpu_has_arch_perfmon boot_cpu_has(X86_FEATURE_ARCH_PERFMON) #define cpu_has_pat boot_cpu_has(X86_FEATURE_PAT) +#define cpu_has_xmm4_2 boot_cpu_has(X86_FEATURE_XMM4_2) #if defined(CONFIG_X86_INVLPG) || defined(CONFIG_X86_64) # define cpu_has_invlpg 1