From: Marek Behún Date: Mon, 29 Apr 2019 20:40:43 +0000 (+0200) Subject: lib: Add xxhash support X-Git-Url: http://git.lede-project.org./?a=commitdiff_plain;h=83a486b6fdb9cc0d009de75d304fb51156425215;p=project%2Fbcm63xx%2Fu-boot.git lib: Add xxhash support This adds the xxhash support from Linux. Files are almost identical to those added to Linux in commit 5d240522 ("lib: Add xxhash module") (they haven't been touched since in Linux). The only difference is to add some includes to be compatible with U-Boot. Also SPDX lincese tags were added. Signed-off-by: Marek Behún --- diff --git a/include/linux/xxhash.h b/include/linux/xxhash.h new file mode 100644 index 0000000000..85feb67fbd --- /dev/null +++ b/include/linux/xxhash.h @@ -0,0 +1,229 @@ +/* SPDX-License-Identifier: (GPL-2.0 or BSD-2-Clause) */ +/* + * xxHash - Extremely Fast Hash algorithm + * Copyright (C) 2012-2016, Yann Collet. + * + * You can contact the author at: + * - xxHash homepage: http://cyan4973.github.io/xxHash/ + * - xxHash source repository: https://github.com/Cyan4973/xxHash + */ + +/* + * Notice extracted from xxHash homepage: + * + * xxHash is an extremely fast Hash algorithm, running at RAM speed limits. + * It also successfully passes all tests from the SMHasher suite. + * + * Comparison (single thread, Windows Seven 32 bits, using SMHasher on a Core 2 + * Duo @3GHz) + * + * Name Speed Q.Score Author + * xxHash 5.4 GB/s 10 + * CrapWow 3.2 GB/s 2 Andrew + * MumurHash 3a 2.7 GB/s 10 Austin Appleby + * SpookyHash 2.0 GB/s 10 Bob Jenkins + * SBox 1.4 GB/s 9 Bret Mulvey + * Lookup3 1.2 GB/s 9 Bob Jenkins + * SuperFastHash 1.2 GB/s 1 Paul Hsieh + * CityHash64 1.05 GB/s 10 Pike & Alakuijala + * FNV 0.55 GB/s 5 Fowler, Noll, Vo + * CRC32 0.43 GB/s 9 + * MD5-32 0.33 GB/s 10 Ronald L. Rivest + * SHA1-32 0.28 GB/s 10 + * + * Q.Score is a measure of quality of the hash function. + * It depends on successfully passing SMHasher test set. + * 10 is a perfect score. + * + * A 64-bits version, named xxh64 offers much better speed, + * but for 64-bits applications only. + * Name Speed on 64 bits Speed on 32 bits + * xxh64 13.8 GB/s 1.9 GB/s + * xxh32 6.8 GB/s 6.0 GB/s + */ + +#ifndef XXHASH_H +#define XXHASH_H + +#include + +/*-**************************** + * Simple Hash Functions + *****************************/ + +/** + * xxh32() - calculate the 32-bit hash of the input with a given seed. + * + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * Speed on Core 2 Duo @ 3 GHz (single thread, SMHasher benchmark) : 5.4 GB/s + * + * Return: The 32-bit hash of the data. + */ +uint32_t xxh32(const void *input, size_t length, uint32_t seed); + +/** + * xxh64() - calculate the 64-bit hash of the input with a given seed. + * + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * This function runs 2x faster on 64-bit systems, but slower on 32-bit systems. + * + * Return: The 64-bit hash of the data. + */ +uint64_t xxh64(const void *input, size_t length, uint64_t seed); + +/** + * xxhash() - calculate wordsize hash of the input with a given seed + * @input: The data to hash. + * @length: The length of the data to hash. + * @seed: The seed can be used to alter the result predictably. + * + * If the hash does not need to be comparable between machines with + * different word sizes, this function will call whichever of xxh32() + * or xxh64() is faster. + * + * Return: wordsize hash of the data. + */ + +static inline unsigned long xxhash(const void *input, size_t length, + uint64_t seed) +{ +#if BITS_PER_LONG == 64 + return xxh64(input, length, seed); +#else + return xxh32(input, length, seed); +#endif +} + +/*-**************************** + * Streaming Hash Functions + *****************************/ + +/* + * These definitions are only meant to allow allocation of XXH state + * statically, on stack, or in a struct for example. + * Do not use members directly. + */ + +/** + * struct xxh32_state - private xxh32 state, do not use members directly + */ +struct xxh32_state { + uint32_t total_len_32; + uint32_t large_len; + uint32_t v1; + uint32_t v2; + uint32_t v3; + uint32_t v4; + uint32_t mem32[4]; + uint32_t memsize; +}; + +/** + * struct xxh32_state - private xxh64 state, do not use members directly + */ +struct xxh64_state { + uint64_t total_len; + uint64_t v1; + uint64_t v2; + uint64_t v3; + uint64_t v4; + uint64_t mem64[4]; + uint32_t memsize; +}; + +/** + * xxh32_reset() - reset the xxh32 state to start a new hashing operation + * + * @state: The xxh32 state to reset. + * @seed: Initialize the hash state with this seed. + * + * Call this function on any xxh32_state to prepare for a new hashing operation. + */ +void xxh32_reset(struct xxh32_state *state, uint32_t seed); + +/** + * xxh32_update() - hash the data given and update the xxh32 state + * + * @state: The xxh32 state to update. + * @input: The data to hash. + * @length: The length of the data to hash. + * + * After calling xxh32_reset() call xxh32_update() as many times as necessary. + * + * Return: Zero on success, otherwise an error code. + */ +int xxh32_update(struct xxh32_state *state, const void *input, size_t length); + +/** + * xxh32_digest() - produce the current xxh32 hash + * + * @state: Produce the current xxh32 hash of this state. + * + * A hash value can be produced at any time. It is still possible to continue + * inserting input into the hash state after a call to xxh32_digest(), and + * generate new hashes later on, by calling xxh32_digest() again. + * + * Return: The xxh32 hash stored in the state. + */ +uint32_t xxh32_digest(const struct xxh32_state *state); + +/** + * xxh64_reset() - reset the xxh64 state to start a new hashing operation + * + * @state: The xxh64 state to reset. + * @seed: Initialize the hash state with this seed. + */ +void xxh64_reset(struct xxh64_state *state, uint64_t seed); + +/** + * xxh64_update() - hash the data given and update the xxh64 state + * @state: The xxh64 state to update. + * @input: The data to hash. + * @length: The length of the data to hash. + * + * After calling xxh64_reset() call xxh64_update() as many times as necessary. + * + * Return: Zero on success, otherwise an error code. + */ +int xxh64_update(struct xxh64_state *state, const void *input, size_t length); + +/** + * xxh64_digest() - produce the current xxh64 hash + * + * @state: Produce the current xxh64 hash of this state. + * + * A hash value can be produced at any time. It is still possible to continue + * inserting input into the hash state after a call to xxh64_digest(), and + * generate new hashes later on, by calling xxh64_digest() again. + * + * Return: The xxh64 hash stored in the state. + */ +uint64_t xxh64_digest(const struct xxh64_state *state); + +/*-************************** + * Utils + ***************************/ + +/** + * xxh32_copy_state() - copy the source state into the destination state + * + * @src: The source xxh32 state. + * @dst: The destination xxh32 state. + */ +void xxh32_copy_state(struct xxh32_state *dst, const struct xxh32_state *src); + +/** + * xxh64_copy_state() - copy the source state into the destination state + * + * @src: The source xxh64 state. + * @dst: The destination xxh64 state. + */ +void xxh64_copy_state(struct xxh64_state *dst, const struct xxh64_state *src); + +#endif /* XXHASH_H */ diff --git a/lib/Kconfig b/lib/Kconfig index 38012506d2..1ed69fafc7 100644 --- a/lib/Kconfig +++ b/lib/Kconfig @@ -327,6 +327,9 @@ config MD5 config CRC32C bool +config XXHASH + bool + endmenu menu "Compression Support" diff --git a/lib/Makefile b/lib/Makefile index 47829bfed5..6a52543530 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -37,6 +37,7 @@ obj-$(CONFIG_GENERATE_SMBIOS_TABLE) += smbios.o obj-$(CONFIG_IMAGE_SPARSE) += image-sparse.o obj-y += ldiv.o obj-$(CONFIG_MD5) += md5.o +obj-$(CONFIG_XXHASH) += xxhash.o obj-y += net_utils.o obj-$(CONFIG_PHYSMEM) += physmem.o obj-y += rc4.o diff --git a/lib/xxhash.c b/lib/xxhash.c new file mode 100644 index 0000000000..2fb4dc6d02 --- /dev/null +++ b/lib/xxhash.c @@ -0,0 +1,467 @@ +// SPDX-License-Identifier: (GPL-2.0 or BSD-2-Clause) +/* + * xxHash - Extremely Fast Hash algorithm + * Copyright (C) 2012-2016, Yann Collet. + * + * You can contact the author at: + * - xxHash homepage: http://cyan4973.github.io/xxHash/ + * - xxHash source repository: https://github.com/Cyan4973/xxHash + */ + +#include +#include +#include +#include +#include +#include +#include + +/*-************************************* + * Macros + **************************************/ +#define xxh_rotl32(x, r) ((x << r) | (x >> (32 - r))) +#define xxh_rotl64(x, r) ((x << r) | (x >> (64 - r))) + +#ifdef __LITTLE_ENDIAN +# define XXH_CPU_LITTLE_ENDIAN 1 +#else +# define XXH_CPU_LITTLE_ENDIAN 0 +#endif + +/*-************************************* + * Constants + **************************************/ +static const uint32_t PRIME32_1 = 2654435761U; +static const uint32_t PRIME32_2 = 2246822519U; +static const uint32_t PRIME32_3 = 3266489917U; +static const uint32_t PRIME32_4 = 668265263U; +static const uint32_t PRIME32_5 = 374761393U; + +static const uint64_t PRIME64_1 = 11400714785074694791ULL; +static const uint64_t PRIME64_2 = 14029467366897019727ULL; +static const uint64_t PRIME64_3 = 1609587929392839161ULL; +static const uint64_t PRIME64_4 = 9650029242287828579ULL; +static const uint64_t PRIME64_5 = 2870177450012600261ULL; + +/*-************************** + * Utils + ***************************/ +void xxh32_copy_state(struct xxh32_state *dst, const struct xxh32_state *src) +{ + memcpy(dst, src, sizeof(*dst)); +} +EXPORT_SYMBOL(xxh32_copy_state); + +void xxh64_copy_state(struct xxh64_state *dst, const struct xxh64_state *src) +{ + memcpy(dst, src, sizeof(*dst)); +} +EXPORT_SYMBOL(xxh64_copy_state); + +/*-*************************** + * Simple Hash Functions + ****************************/ +static uint32_t xxh32_round(uint32_t seed, const uint32_t input) +{ + seed += input * PRIME32_2; + seed = xxh_rotl32(seed, 13); + seed *= PRIME32_1; + return seed; +} + +uint32_t xxh32(const void *input, const size_t len, const uint32_t seed) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *b_end = p + len; + uint32_t h32; + + if (len >= 16) { + const uint8_t *const limit = b_end - 16; + uint32_t v1 = seed + PRIME32_1 + PRIME32_2; + uint32_t v2 = seed + PRIME32_2; + uint32_t v3 = seed + 0; + uint32_t v4 = seed - PRIME32_1; + + do { + v1 = xxh32_round(v1, get_unaligned_le32(p)); + p += 4; + v2 = xxh32_round(v2, get_unaligned_le32(p)); + p += 4; + v3 = xxh32_round(v3, get_unaligned_le32(p)); + p += 4; + v4 = xxh32_round(v4, get_unaligned_le32(p)); + p += 4; + } while (p <= limit); + + h32 = xxh_rotl32(v1, 1) + xxh_rotl32(v2, 7) + + xxh_rotl32(v3, 12) + xxh_rotl32(v4, 18); + } else { + h32 = seed + PRIME32_5; + } + + h32 += (uint32_t)len; + + while (p + 4 <= b_end) { + h32 += get_unaligned_le32(p) * PRIME32_3; + h32 = xxh_rotl32(h32, 17) * PRIME32_4; + p += 4; + } + + while (p < b_end) { + h32 += (*p) * PRIME32_5; + h32 = xxh_rotl32(h32, 11) * PRIME32_1; + p++; + } + + h32 ^= h32 >> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} +EXPORT_SYMBOL(xxh32); + +static uint64_t xxh64_round(uint64_t acc, const uint64_t input) +{ + acc += input * PRIME64_2; + acc = xxh_rotl64(acc, 31); + acc *= PRIME64_1; + return acc; +} + +static uint64_t xxh64_merge_round(uint64_t acc, uint64_t val) +{ + val = xxh64_round(0, val); + acc ^= val; + acc = acc * PRIME64_1 + PRIME64_4; + return acc; +} + +uint64_t xxh64(const void *input, const size_t len, const uint64_t seed) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + uint64_t h64; + + if (len >= 32) { + const uint8_t *const limit = b_end - 32; + uint64_t v1 = seed + PRIME64_1 + PRIME64_2; + uint64_t v2 = seed + PRIME64_2; + uint64_t v3 = seed + 0; + uint64_t v4 = seed - PRIME64_1; + + do { + v1 = xxh64_round(v1, get_unaligned_le64(p)); + p += 8; + v2 = xxh64_round(v2, get_unaligned_le64(p)); + p += 8; + v3 = xxh64_round(v3, get_unaligned_le64(p)); + p += 8; + v4 = xxh64_round(v4, get_unaligned_le64(p)); + p += 8; + } while (p <= limit); + + h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) + + xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18); + h64 = xxh64_merge_round(h64, v1); + h64 = xxh64_merge_round(h64, v2); + h64 = xxh64_merge_round(h64, v3); + h64 = xxh64_merge_round(h64, v4); + + } else { + h64 = seed + PRIME64_5; + } + + h64 += (uint64_t)len; + + while (p + 8 <= b_end) { + const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p)); + + h64 ^= k1; + h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; + p += 8; + } + + if (p + 4 <= b_end) { + h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1; + h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p += 4; + } + + while (p < b_end) { + h64 ^= (*p) * PRIME64_5; + h64 = xxh_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; + + return h64; +} +EXPORT_SYMBOL(xxh64); + +/*-************************************************** + * Advanced Hash Functions + ***************************************************/ +void xxh32_reset(struct xxh32_state *statePtr, const uint32_t seed) +{ + /* use a local state for memcpy() to avoid strict-aliasing warnings */ + struct xxh32_state state; + + memset(&state, 0, sizeof(state)); + state.v1 = seed + PRIME32_1 + PRIME32_2; + state.v2 = seed + PRIME32_2; + state.v3 = seed + 0; + state.v4 = seed - PRIME32_1; + memcpy(statePtr, &state, sizeof(state)); +} +EXPORT_SYMBOL(xxh32_reset); + +void xxh64_reset(struct xxh64_state *statePtr, const uint64_t seed) +{ + /* use a local state for memcpy() to avoid strict-aliasing warnings */ + struct xxh64_state state; + + memset(&state, 0, sizeof(state)); + state.v1 = seed + PRIME64_1 + PRIME64_2; + state.v2 = seed + PRIME64_2; + state.v3 = seed + 0; + state.v4 = seed - PRIME64_1; + memcpy(statePtr, &state, sizeof(state)); +} +EXPORT_SYMBOL(xxh64_reset); + +int xxh32_update(struct xxh32_state *state, const void *input, const size_t len) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + + if (input == NULL) + return -EINVAL; + + state->total_len_32 += (uint32_t)len; + state->large_len |= (len >= 16) | (state->total_len_32 >= 16); + + if (state->memsize + len < 16) { /* fill in tmp buffer */ + memcpy((uint8_t *)(state->mem32) + state->memsize, input, len); + state->memsize += (uint32_t)len; + return 0; + } + + if (state->memsize) { /* some data left from previous update */ + const uint32_t *p32 = state->mem32; + + memcpy((uint8_t *)(state->mem32) + state->memsize, input, + 16 - state->memsize); + + state->v1 = xxh32_round(state->v1, get_unaligned_le32(p32)); + p32++; + state->v2 = xxh32_round(state->v2, get_unaligned_le32(p32)); + p32++; + state->v3 = xxh32_round(state->v3, get_unaligned_le32(p32)); + p32++; + state->v4 = xxh32_round(state->v4, get_unaligned_le32(p32)); + p32++; + + p += 16-state->memsize; + state->memsize = 0; + } + + if (p <= b_end - 16) { + const uint8_t *const limit = b_end - 16; + uint32_t v1 = state->v1; + uint32_t v2 = state->v2; + uint32_t v3 = state->v3; + uint32_t v4 = state->v4; + + do { + v1 = xxh32_round(v1, get_unaligned_le32(p)); + p += 4; + v2 = xxh32_round(v2, get_unaligned_le32(p)); + p += 4; + v3 = xxh32_round(v3, get_unaligned_le32(p)); + p += 4; + v4 = xxh32_round(v4, get_unaligned_le32(p)); + p += 4; + } while (p <= limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < b_end) { + memcpy(state->mem32, p, (size_t)(b_end-p)); + state->memsize = (uint32_t)(b_end-p); + } + + return 0; +} +EXPORT_SYMBOL(xxh32_update); + +uint32_t xxh32_digest(const struct xxh32_state *state) +{ + const uint8_t *p = (const uint8_t *)state->mem32; + const uint8_t *const b_end = (const uint8_t *)(state->mem32) + + state->memsize; + uint32_t h32; + + if (state->large_len) { + h32 = xxh_rotl32(state->v1, 1) + xxh_rotl32(state->v2, 7) + + xxh_rotl32(state->v3, 12) + xxh_rotl32(state->v4, 18); + } else { + h32 = state->v3 /* == seed */ + PRIME32_5; + } + + h32 += state->total_len_32; + + while (p + 4 <= b_end) { + h32 += get_unaligned_le32(p) * PRIME32_3; + h32 = xxh_rotl32(h32, 17) * PRIME32_4; + p += 4; + } + + while (p < b_end) { + h32 += (*p) * PRIME32_5; + h32 = xxh_rotl32(h32, 11) * PRIME32_1; + p++; + } + + h32 ^= h32 >> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} +EXPORT_SYMBOL(xxh32_digest); + +int xxh64_update(struct xxh64_state *state, const void *input, const size_t len) +{ + const uint8_t *p = (const uint8_t *)input; + const uint8_t *const b_end = p + len; + + if (input == NULL) + return -EINVAL; + + state->total_len += len; + + if (state->memsize + len < 32) { /* fill in tmp buffer */ + memcpy(((uint8_t *)state->mem64) + state->memsize, input, len); + state->memsize += (uint32_t)len; + return 0; + } + + if (state->memsize) { /* tmp buffer is full */ + uint64_t *p64 = state->mem64; + + memcpy(((uint8_t *)p64) + state->memsize, input, + 32 - state->memsize); + + state->v1 = xxh64_round(state->v1, get_unaligned_le64(p64)); + p64++; + state->v2 = xxh64_round(state->v2, get_unaligned_le64(p64)); + p64++; + state->v3 = xxh64_round(state->v3, get_unaligned_le64(p64)); + p64++; + state->v4 = xxh64_round(state->v4, get_unaligned_le64(p64)); + + p += 32 - state->memsize; + state->memsize = 0; + } + + if (p + 32 <= b_end) { + const uint8_t *const limit = b_end - 32; + uint64_t v1 = state->v1; + uint64_t v2 = state->v2; + uint64_t v3 = state->v3; + uint64_t v4 = state->v4; + + do { + v1 = xxh64_round(v1, get_unaligned_le64(p)); + p += 8; + v2 = xxh64_round(v2, get_unaligned_le64(p)); + p += 8; + v3 = xxh64_round(v3, get_unaligned_le64(p)); + p += 8; + v4 = xxh64_round(v4, get_unaligned_le64(p)); + p += 8; + } while (p <= limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < b_end) { + memcpy(state->mem64, p, (size_t)(b_end-p)); + state->memsize = (uint32_t)(b_end - p); + } + + return 0; +} +EXPORT_SYMBOL(xxh64_update); + +uint64_t xxh64_digest(const struct xxh64_state *state) +{ + const uint8_t *p = (const uint8_t *)state->mem64; + const uint8_t *const b_end = (const uint8_t *)state->mem64 + + state->memsize; + uint64_t h64; + + if (state->total_len >= 32) { + const uint64_t v1 = state->v1; + const uint64_t v2 = state->v2; + const uint64_t v3 = state->v3; + const uint64_t v4 = state->v4; + + h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) + + xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18); + h64 = xxh64_merge_round(h64, v1); + h64 = xxh64_merge_round(h64, v2); + h64 = xxh64_merge_round(h64, v3); + h64 = xxh64_merge_round(h64, v4); + } else { + h64 = state->v3 + PRIME64_5; + } + + h64 += (uint64_t)state->total_len; + + while (p + 8 <= b_end) { + const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p)); + + h64 ^= k1; + h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; + p += 8; + } + + if (p + 4 <= b_end) { + h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1; + h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; + p += 4; + } + + while (p < b_end) { + h64 ^= (*p) * PRIME64_5; + h64 = xxh_rotl64(h64, 11) * PRIME64_1; + p++; + } + + h64 ^= h64 >> 33; + h64 *= PRIME64_2; + h64 ^= h64 >> 29; + h64 *= PRIME64_3; + h64 ^= h64 >> 32; + + return h64; +} +EXPORT_SYMBOL(xxh64_digest);