lsquic_xxhash.c revision 7d09751d
1/* Copyright (c) 2017 - 2020 LiteSpeed Technologies Inc. See LICENSE. */ 2/* 3xxHash - Fast Hash algorithm 4Copyright (C) 2012-2014, Yann Collet. 5BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) 6 7Redistribution and use in source and binary forms, with or without 8modification, are permitted provided that the following conditions are 9met: 10 11* Redistributions of source code must retain the above copyright 12notice, this list of conditions and the following disclaimer. 13* Redistributions in binary form must reproduce the above 14copyright notice, this list of conditions and the following disclaimer 15in the documentation and/or other materials provided with the 16distribution. 17 18THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 19"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 20LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 21A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 22OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 23SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 24LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 25DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 26THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 27(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 28OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 29 30You can contact the author at : 31- xxHash source repository : http://code.google.com/p/xxhash/ 32- public discussion board : https://groups.google.com/forum/#!forum/lz4c 33*/ 34 35 36//************************************** 37// Tuning parameters 38//************************************** 39// Unaligned memory access is automatically enabled for "common" CPU, such as x86. 40// For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected. 41// If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance. 42// You can also enable this parameter if you know your input data will always be aligned (boundaries of 4, for U32). 43#if defined(__ARM_FEATURE_UNALIGNED) || defined(__i386) || defined(_M_IX86) || defined(__x86_64__) || defined(_M_X64) 44# define XXH_USE_UNALIGNED_ACCESS 1 45#endif 46 47// XXH_ACCEPT_NULL_INPUT_POINTER : 48// If the input pointer is a null pointer, xxHash default behavior is to trigger a memory access error, since it is a bad pointer. 49// When this option is enabled, xxHash output for null input pointers will be the same as a null-length input. 50// This option has a very small performance cost (only measurable on small inputs). 51// By default, this option is disabled. To enable it, uncomment below define : 52// #define XXH_ACCEPT_NULL_INPUT_POINTER 1 53 54// XXH_FORCE_NATIVE_FORMAT : 55// By default, xxHash library provides endian-independant Hash values, based on little-endian convention. 56// Results are therefore identical for little-endian and big-endian CPU. 57// This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format. 58// Should endian-independance be of no importance for your application, you may set the #define below to 1. 59// It will improve speed for Big-endian CPU. 60// This option has no impact on Little_Endian CPU. 61#define XXH_FORCE_NATIVE_FORMAT 0 62 63//************************************** 64// Compiler Specific Options 65//************************************** 66// Disable some Visual warning messages 67#ifdef _MSC_VER // Visual Studio 68# pragma warning(disable : 4127) // disable: C4127: conditional expression is constant 69#endif 70 71#ifdef _MSC_VER // Visual Studio 72# define FORCE_INLINE static __forceinline 73#else 74# ifdef __GNUC__ 75# define FORCE_INLINE static inline __attribute__((always_inline)) 76# else 77# define FORCE_INLINE static inline 78# endif 79#endif 80 81//************************************** 82// Includes & Memory related functions 83//************************************** 84#include "lsquic_xxhash.h" 85// Modify the local functions below should you wish to use some other memory routines 86// for malloc(), free() 87#include <stdlib.h> 88static void *XXH_malloc(size_t s) { return malloc(s); } 89static void XXH_free(void *p) { free(p); } 90// for memcpy() 91#include <string.h> 92static void *XXH_memcpy(void *dest, const void *src, size_t size) 93{ 94 return memcpy(dest, src, size); 95} 96 97 98//************************************** 99// Basic Types 100//************************************** 101#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99 102# include <stdint.h> 103typedef uint8_t BYTE; 104typedef uint16_t U16; 105typedef uint32_t U32; 106typedef int32_t S32; 107typedef uint64_t U64; 108#else 109typedef unsigned char BYTE; 110typedef unsigned short U16; 111typedef unsigned int U32; 112typedef signed int S32; 113typedef unsigned long long U64; 114#endif 115 116#if defined(__GNUC__) && !defined(XXH_USE_UNALIGNED_ACCESS) 117# define _PACKED __attribute__ ((packed)) 118#else 119# define _PACKED 120#endif 121 122#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) 123# ifdef __IBMC__ 124# pragma pack(1) 125# else 126# pragma pack(push, 1) 127# endif 128#endif 129 130typedef struct _U32_S 131{ 132 U32 v; 133} _PACKED U32_S; 134typedef struct _U64_S 135{ 136 U64 v; 137} _PACKED U64_S; 138 139#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) 140# pragma pack(pop) 141#endif 142 143#define A32(x) (((U32_S *)(x))->v) 144#define A64(x) (((U64_S *)(x))->v) 145 146 147//*************************************** 148// Compiler-specific Functions and Macros 149//*************************************** 150#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) 151 152// Note : although _rotl exists for minGW (GCC under windows), performance seems poor 153#if defined(_MSC_VER) 154# define XXH_rotl32(x,r) _rotl(x,r) 155# define XXH_rotl64(x,r) _rotl64(x,r) 156#else 157# define XXH_rotl32(x,r) ((x << r) | (x >> (32 - r))) 158# define XXH_rotl64(x,r) ((x << r) | (x >> (64 - r))) 159#endif 160 161#if defined(_MSC_VER) // Visual Studio 162# define XXH_swap32 _byteswap_ulong 163# define XXH_swap64 _byteswap_uint64 164#elif GCC_VERSION >= 403 165# define XXH_swap32 __builtin_bswap32 166# define XXH_swap64 __builtin_bswap64 167#else 168static inline U32 XXH_swap32(U32 x) 169{ 170 return ((x << 24) & 0xff000000) | 171 ((x << 8) & 0x00ff0000) | 172 ((x >> 8) & 0x0000ff00) | 173 ((x >> 24) & 0x000000ff); 174} 175static inline U64 XXH_swap64(U64 x) 176{ 177 return ((x << 56) & 0xff00000000000000ULL) | 178 ((x << 40) & 0x00ff000000000000ULL) | 179 ((x << 24) & 0x0000ff0000000000ULL) | 180 ((x << 8) & 0x000000ff00000000ULL) | 181 ((x >> 8) & 0x00000000ff000000ULL) | 182 ((x >> 24) & 0x0000000000ff0000ULL) | 183 ((x >> 40) & 0x000000000000ff00ULL) | 184 ((x >> 56) & 0x00000000000000ffULL); 185} 186#endif 187 188 189//************************************** 190// Constants 191//************************************** 192#define PRIME32_1 2654435761U 193#define PRIME32_2 2246822519U 194#define PRIME32_3 3266489917U 195#define PRIME32_4 668265263U 196#define PRIME32_5 374761393U 197 198#define PRIME64_1 11400714785074694791ULL 199#define PRIME64_2 14029467366897019727ULL 200#define PRIME64_3 1609587929392839161ULL 201#define PRIME64_4 9650029242287828579ULL 202#define PRIME64_5 2870177450012600261ULL 203 204//************************************** 205// Architecture Macros 206//************************************** 207typedef enum { XXH_bigEndian = 0, XXH_littleEndian = 1 } XXH_endianess; 208#ifndef XXH_CPU_LITTLE_ENDIAN // It is possible to define XXH_CPU_LITTLE_ENDIAN externally, for example using a compiler switch 209static const int one = 1; 210# define XXH_CPU_LITTLE_ENDIAN (*(char*)(&one)) 211#endif 212 213 214//************************************** 215// Macros 216//************************************** 217#define XXH_STATIC_ASSERT(c) { enum { XXH_static_assert = 1/(!!(c)) }; } // use only *after* variable declarations 218 219 220//**************************** 221// Memory reads 222//**************************** 223typedef enum { XXH_aligned, XXH_unaligned } XXH_alignment; 224 225FORCE_INLINE U32 XXH_readLE32_align(const void *ptr, XXH_endianess endian, 226 XXH_alignment align) 227{ 228 if (align == XXH_unaligned) 229 return endian == XXH_littleEndian ? A32(ptr) : XXH_swap32(A32(ptr)); 230 else 231 return endian == XXH_littleEndian ? *(U32 *)ptr : XXH_swap32(*(U32 *)ptr); 232} 233 234FORCE_INLINE U32 XXH_readLE32(const void *ptr, XXH_endianess endian) 235{ 236 return XXH_readLE32_align(ptr, endian, XXH_unaligned); 237} 238 239FORCE_INLINE U64 XXH_readLE64_align(const void *ptr, XXH_endianess endian, 240 XXH_alignment align) 241{ 242 if (align == XXH_unaligned) 243 return endian == XXH_littleEndian ? A64(ptr) : XXH_swap64(A64(ptr)); 244 else 245 return endian == XXH_littleEndian ? *(U64 *)ptr : XXH_swap64(*(U64 *)ptr); 246} 247 248FORCE_INLINE U64 XXH_readLE64(const void *ptr, XXH_endianess endian) 249{ 250 return XXH_readLE64_align(ptr, endian, XXH_unaligned); 251} 252 253 254//**************************** 255// Simple Hash Functions 256//**************************** 257FORCE_INLINE U32 XXH32_endian_align(const void *input, size_t len, 258 U32 seed, XXH_endianess endian, XXH_alignment align) 259{ 260 const BYTE *p = (const BYTE *)input; 261 const BYTE *bEnd = p + len; 262 U32 h32; 263#define XXH_get32bits(p) XXH_readLE32_align(p, endian, align) 264 265#ifdef XXH_ACCEPT_NULL_INPUT_POINTER 266 if (p == NULL) 267 { 268 len = 0; 269 bEnd = p = (const BYTE *)(size_t)16; 270 } 271#endif 272 273 if (len >= 16) 274 { 275 const BYTE *const limit = bEnd - 16; 276 U32 v1 = seed + PRIME32_1 + PRIME32_2; 277 U32 v2 = seed + PRIME32_2; 278 U32 v3 = seed + 0; 279 U32 v4 = seed - PRIME32_1; 280 281 do 282 { 283 v1 += XXH_get32bits(p) * PRIME32_2; 284 v1 = XXH_rotl32(v1, 13); 285 v1 *= PRIME32_1; 286 p += 4; 287 v2 += XXH_get32bits(p) * PRIME32_2; 288 v2 = XXH_rotl32(v2, 13); 289 v2 *= PRIME32_1; 290 p += 4; 291 v3 += XXH_get32bits(p) * PRIME32_2; 292 v3 = XXH_rotl32(v3, 13); 293 v3 *= PRIME32_1; 294 p += 4; 295 v4 += XXH_get32bits(p) * PRIME32_2; 296 v4 = XXH_rotl32(v4, 13); 297 v4 *= PRIME32_1; 298 p += 4; 299 } 300 while (p <= limit); 301 302 h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 303 12) + XXH_rotl32(v4, 18); 304 } 305 else 306 h32 = seed + PRIME32_5; 307 308 h32 += (U32) len; 309 310 while (p + 4 <= bEnd) 311 { 312 h32 += XXH_get32bits(p) * PRIME32_3; 313 h32 = XXH_rotl32(h32, 17) * PRIME32_4 ; 314 p += 4; 315 } 316 317 while (p < bEnd) 318 { 319 h32 += (*p) * PRIME32_5; 320 h32 = XXH_rotl32(h32, 11) * PRIME32_1 ; 321 p++; 322 } 323 324 h32 ^= h32 >> 15; 325 h32 *= PRIME32_2; 326 h32 ^= h32 >> 13; 327 h32 *= PRIME32_3; 328 h32 ^= h32 >> 16; 329 330 return h32; 331} 332 333 334unsigned int XXH32(const void *input, size_t len, unsigned seed) 335{ 336#if 0 337 // Simple version, good for code maintenance, but unfortunately slow for small inputs 338 XXH32_state_t state; 339 XXH32_reset(&state, seed); 340 XXH32_update(&state, input, len); 341 return XXH32_digest(&state); 342#else 343 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 344 345# if !defined(XXH_USE_UNALIGNED_ACCESS) 346 if ((((size_t)input) & 3) == 347 0) // Input is aligned, let's leverage the speed advantage 348 { 349 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 350 return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); 351 else 352 return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned); 353 } 354# endif 355 356 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 357 return XXH32_endian_align(input, len, seed, XXH_littleEndian, 358 XXH_unaligned); 359 else 360 return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned); 361#endif 362} 363 364FORCE_INLINE U64 XXH64_endian_align(const void *input, size_t len, 365 U64 seed, XXH_endianess endian, XXH_alignment align) 366{ 367 const BYTE *p = (const BYTE *)input; 368 const BYTE *bEnd = p + len; 369 U64 h64; 370#define XXH_get64bits(p) XXH_readLE64_align(p, endian, align) 371 372#ifdef XXH_ACCEPT_NULL_INPUT_POINTER 373 if (p == NULL) 374 { 375 len = 0; 376 bEnd = p = (const BYTE *)(size_t)32; 377 } 378#endif 379 380 if (len >= 32) 381 { 382 const BYTE *const limit = bEnd - 32; 383 U64 v1 = seed + PRIME64_1 + PRIME64_2; 384 U64 v2 = seed + PRIME64_2; 385 U64 v3 = seed + 0; 386 U64 v4 = seed - PRIME64_1; 387 388 do 389 { 390 v1 += XXH_get64bits(p) * PRIME64_2; 391 p += 8; 392 v1 = XXH_rotl64(v1, 31); 393 v1 *= PRIME64_1; 394 v2 += XXH_get64bits(p) * PRIME64_2; 395 p += 8; 396 v2 = XXH_rotl64(v2, 31); 397 v2 *= PRIME64_1; 398 v3 += XXH_get64bits(p) * PRIME64_2; 399 p += 8; 400 v3 = XXH_rotl64(v3, 31); 401 v3 *= PRIME64_1; 402 v4 += XXH_get64bits(p) * PRIME64_2; 403 p += 8; 404 v4 = XXH_rotl64(v4, 31); 405 v4 *= PRIME64_1; 406 } 407 while (p <= limit); 408 409 h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 410 12) + XXH_rotl64(v4, 18); 411 412 v1 *= PRIME64_2; 413 v1 = XXH_rotl64(v1, 31); 414 v1 *= PRIME64_1; 415 h64 ^= v1; 416 h64 = h64 * PRIME64_1 + PRIME64_4; 417 418 v2 *= PRIME64_2; 419 v2 = XXH_rotl64(v2, 31); 420 v2 *= PRIME64_1; 421 h64 ^= v2; 422 h64 = h64 * PRIME64_1 + PRIME64_4; 423 424 v3 *= PRIME64_2; 425 v3 = XXH_rotl64(v3, 31); 426 v3 *= PRIME64_1; 427 h64 ^= v3; 428 h64 = h64 * PRIME64_1 + PRIME64_4; 429 430 v4 *= PRIME64_2; 431 v4 = XXH_rotl64(v4, 31); 432 v4 *= PRIME64_1; 433 h64 ^= v4; 434 h64 = h64 * PRIME64_1 + PRIME64_4; 435 } 436 else 437 h64 = seed + PRIME64_5; 438 439 h64 += (U64) len; 440 441 while (p + 8 <= bEnd) 442 { 443 U64 k1 = XXH_get64bits(p); 444 k1 *= PRIME64_2; 445 k1 = XXH_rotl64(k1, 31); 446 k1 *= PRIME64_1; 447 h64 ^= k1; 448 h64 = XXH_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; 449 p += 8; 450 } 451 452 if (p + 4 <= bEnd) 453 { 454 h64 ^= (U64)(XXH_get32bits(p)) * PRIME64_1; 455 h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; 456 p += 4; 457 } 458 459 while (p < bEnd) 460 { 461 h64 ^= (*p) * PRIME64_5; 462 h64 = XXH_rotl64(h64, 11) * PRIME64_1; 463 p++; 464 } 465 466 h64 ^= h64 >> 33; 467 h64 *= PRIME64_2; 468 h64 ^= h64 >> 29; 469 h64 *= PRIME64_3; 470 h64 ^= h64 >> 32; 471 472 return h64; 473} 474 475 476unsigned long long XXH64(const void *input, size_t len, 477 unsigned long long seed) 478{ 479#if 0 480 // Simple version, good for code maintenance, but unfortunately slow for small inputs 481 XXH64_state_t state; 482 XXH64_reset(&state, seed); 483 XXH64_update(&state, input, len); 484 return XXH64_digest(&state); 485#else 486 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 487 488# if !defined(XXH_USE_UNALIGNED_ACCESS) 489 if ((((size_t)input) & 7) == 490 0) // Input is aligned, let's leverage the speed advantage 491 { 492 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 493 return XXH64_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); 494 else 495 return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned); 496 } 497# endif 498 499 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 500 return XXH64_endian_align(input, len, seed, XXH_littleEndian, 501 XXH_unaligned); 502 else 503 return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned); 504#endif 505} 506 507/**************************************************** 508 * Advanced Hash Functions 509****************************************************/ 510 511/*** Allocation ***/ 512typedef struct 513{ 514 U64 total_len; 515 U32 seed; 516 U32 v1; 517 U32 v2; 518 U32 v3; 519 U32 v4; 520 U32 mem32[4]; /* defined as U32 for alignment */ 521 U32 memsize; 522} XXH_istate32_t; 523 524typedef struct 525{ 526 U64 total_len; 527 U64 seed; 528 U64 v1; 529 U64 v2; 530 U64 v3; 531 U64 v4; 532 U64 mem64[4]; /* defined as U64 for alignment */ 533 U32 memsize; 534} XXH_istate64_t; 535 536 537XXH32_state_t *XXH32_createState(void) 538{ 539 XXH_STATIC_ASSERT(sizeof(XXH32_state_t) >= sizeof( 540 XXH_istate32_t)); // A compilation error here means XXH32_state_t is not large enough 541 return (XXH32_state_t *)XXH_malloc(sizeof(XXH32_state_t)); 542} 543XXH_errorcode XXH32_freeState(XXH32_state_t *statePtr) 544{ 545 XXH_free(statePtr); 546 return XXH_OK; 547}; 548 549XXH64_state_t *XXH64_createState(void) 550{ 551 XXH_STATIC_ASSERT(sizeof(XXH64_state_t) >= sizeof( 552 XXH_istate64_t)); // A compilation error here means XXH64_state_t is not large enough 553 return (XXH64_state_t *)XXH_malloc(sizeof(XXH64_state_t)); 554} 555XXH_errorcode XXH64_freeState(XXH64_state_t *statePtr) 556{ 557 XXH_free(statePtr); 558 return XXH_OK; 559}; 560 561 562/*** Hash feed ***/ 563 564XXH_errorcode XXH32_reset(XXH32_state_t *state_in, U32 seed) 565{ 566 XXH_istate32_t *state = (XXH_istate32_t *) state_in; 567 state->seed = seed; 568 state->v1 = seed + PRIME32_1 + PRIME32_2; 569 state->v2 = seed + PRIME32_2; 570 state->v3 = seed + 0; 571 state->v4 = seed - PRIME32_1; 572 state->total_len = 0; 573 state->memsize = 0; 574 return XXH_OK; 575} 576 577XXH_errorcode XXH64_reset(XXH64_state_t *state_in, unsigned long long seed) 578{ 579 XXH_istate64_t *state = (XXH_istate64_t *) state_in; 580 state->seed = seed; 581 state->v1 = seed + PRIME64_1 + PRIME64_2; 582 state->v2 = seed + PRIME64_2; 583 state->v3 = seed + 0; 584 state->v4 = seed - PRIME64_1; 585 state->total_len = 0; 586 state->memsize = 0; 587 return XXH_OK; 588} 589 590 591FORCE_INLINE XXH_errorcode XXH32_update_endian(XXH32_state_t *state_in, 592 const void *input, size_t len, XXH_endianess endian) 593{ 594 XXH_istate32_t *state = (XXH_istate32_t *) state_in; 595 const BYTE *p = (const BYTE *)input; 596 const BYTE *const bEnd = p + len; 597 598#ifdef XXH_ACCEPT_NULL_INPUT_POINTER 599 if (input == NULL) return XXH_ERROR; 600#endif 601 602 state->total_len += len; 603 604 if (state->memsize + len < 16) // fill in tmp buffer 605 { 606 XXH_memcpy((BYTE *)(state->mem32) + state->memsize, input, len); 607 state->memsize += (U32)len; 608 return XXH_OK; 609 } 610 611 if (state->memsize) // some data left from previous update 612 { 613 XXH_memcpy((BYTE *)(state->mem32) + state->memsize, input, 614 16 - state->memsize); 615 { 616 const U32 *p32 = state->mem32; 617 state->v1 += XXH_readLE32(p32, endian) * PRIME32_2; 618 state->v1 = XXH_rotl32(state->v1, 13); 619 state->v1 *= PRIME32_1; 620 p32++; 621 state->v2 += XXH_readLE32(p32, endian) * PRIME32_2; 622 state->v2 = XXH_rotl32(state->v2, 13); 623 state->v2 *= PRIME32_1; 624 p32++; 625 state->v3 += XXH_readLE32(p32, endian) * PRIME32_2; 626 state->v3 = XXH_rotl32(state->v3, 13); 627 state->v3 *= PRIME32_1; 628 p32++; 629 state->v4 += XXH_readLE32(p32, endian) * PRIME32_2; 630 state->v4 = XXH_rotl32(state->v4, 13); 631 state->v4 *= PRIME32_1; 632 p32++; 633 } 634 p += 16 - state->memsize; 635 state->memsize = 0; 636 } 637 638 if (p <= bEnd - 16) 639 { 640 const BYTE *const limit = bEnd - 16; 641 U32 v1 = state->v1; 642 U32 v2 = state->v2; 643 U32 v3 = state->v3; 644 U32 v4 = state->v4; 645 646 do 647 { 648 v1 += XXH_readLE32(p, endian) * PRIME32_2; 649 v1 = XXH_rotl32(v1, 13); 650 v1 *= PRIME32_1; 651 p += 4; 652 v2 += XXH_readLE32(p, endian) * PRIME32_2; 653 v2 = XXH_rotl32(v2, 13); 654 v2 *= PRIME32_1; 655 p += 4; 656 v3 += XXH_readLE32(p, endian) * PRIME32_2; 657 v3 = XXH_rotl32(v3, 13); 658 v3 *= PRIME32_1; 659 p += 4; 660 v4 += XXH_readLE32(p, endian) * PRIME32_2; 661 v4 = XXH_rotl32(v4, 13); 662 v4 *= PRIME32_1; 663 p += 4; 664 } 665 while (p <= limit); 666 667 state->v1 = v1; 668 state->v2 = v2; 669 state->v3 = v3; 670 state->v4 = v4; 671 } 672 673 if (p < bEnd) 674 { 675 XXH_memcpy(state->mem32, p, bEnd - p); 676 state->memsize = (int)(bEnd - p); 677 } 678 679 return XXH_OK; 680} 681 682XXH_errorcode XXH32_update(XXH32_state_t *state_in, const void *input, 683 size_t len) 684{ 685 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 686 687 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 688 return XXH32_update_endian(state_in, input, len, XXH_littleEndian); 689 else 690 return XXH32_update_endian(state_in, input, len, XXH_bigEndian); 691} 692 693 694 695FORCE_INLINE U32 XXH32_digest_endian(const XXH32_state_t *state_in, 696 XXH_endianess endian) 697{ 698 XXH_istate32_t *state = (XXH_istate32_t *) state_in; 699 const BYTE *p = (const BYTE *)state->mem32; 700 BYTE *bEnd = (BYTE *)(state->mem32) + state->memsize; 701 U32 h32; 702 703 if (state->total_len >= 16) 704 h32 = XXH_rotl32(state->v1, 1) + XXH_rotl32(state->v2, 705 7) + XXH_rotl32(state->v3, 12) + XXH_rotl32(state->v4, 18); 706 else 707 h32 = state->seed + PRIME32_5; 708 709 h32 += (U32) state->total_len; 710 711 while (p + 4 <= bEnd) 712 { 713 h32 += XXH_readLE32(p, endian) * PRIME32_3; 714 h32 = XXH_rotl32(h32, 17) * PRIME32_4; 715 p += 4; 716 } 717 718 while (p < bEnd) 719 { 720 h32 += (*p) * PRIME32_5; 721 h32 = XXH_rotl32(h32, 11) * PRIME32_1; 722 p++; 723 } 724 725 h32 ^= h32 >> 15; 726 h32 *= PRIME32_2; 727 h32 ^= h32 >> 13; 728 h32 *= PRIME32_3; 729 h32 ^= h32 >> 16; 730 731 return h32; 732} 733 734 735U32 XXH32_digest(const XXH32_state_t *state_in) 736{ 737 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 738 739 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 740 return XXH32_digest_endian(state_in, XXH_littleEndian); 741 else 742 return XXH32_digest_endian(state_in, XXH_bigEndian); 743} 744 745 746FORCE_INLINE XXH_errorcode XXH64_update_endian(XXH64_state_t *state_in, 747 const void *input, size_t len, XXH_endianess endian) 748{ 749 XXH_istate64_t *state = (XXH_istate64_t *) state_in; 750 const BYTE *p = (const BYTE *)input; 751 const BYTE *const bEnd = p + len; 752 753#ifdef XXH_ACCEPT_NULL_INPUT_POINTER 754 if (input == NULL) return XXH_ERROR; 755#endif 756 757 state->total_len += len; 758 759 if (state->memsize + len < 32) // fill in tmp buffer 760 { 761 XXH_memcpy(((BYTE *)state->mem64) + state->memsize, input, len); 762 state->memsize += (U32)len; 763 return XXH_OK; 764 } 765 766 if (state->memsize) // some data left from previous update 767 { 768 XXH_memcpy(((BYTE *)state->mem64) + state->memsize, input, 769 32 - state->memsize); 770 { 771 const U64 *p64 = state->mem64; 772 state->v1 += XXH_readLE64(p64, endian) * PRIME64_2; 773 state->v1 = XXH_rotl64(state->v1, 31); 774 state->v1 *= PRIME64_1; 775 p64++; 776 state->v2 += XXH_readLE64(p64, endian) * PRIME64_2; 777 state->v2 = XXH_rotl64(state->v2, 31); 778 state->v2 *= PRIME64_1; 779 p64++; 780 state->v3 += XXH_readLE64(p64, endian) * PRIME64_2; 781 state->v3 = XXH_rotl64(state->v3, 31); 782 state->v3 *= PRIME64_1; 783 p64++; 784 state->v4 += XXH_readLE64(p64, endian) * PRIME64_2; 785 state->v4 = XXH_rotl64(state->v4, 31); 786 state->v4 *= PRIME64_1; 787 p64++; 788 } 789 p += 32 - state->memsize; 790 state->memsize = 0; 791 } 792 793 if (p + 32 <= bEnd) 794 { 795 const BYTE *const limit = bEnd - 32; 796 U64 v1 = state->v1; 797 U64 v2 = state->v2; 798 U64 v3 = state->v3; 799 U64 v4 = state->v4; 800 801 do 802 { 803 v1 += XXH_readLE64(p, endian) * PRIME64_2; 804 v1 = XXH_rotl64(v1, 31); 805 v1 *= PRIME64_1; 806 p += 8; 807 v2 += XXH_readLE64(p, endian) * PRIME64_2; 808 v2 = XXH_rotl64(v2, 31); 809 v2 *= PRIME64_1; 810 p += 8; 811 v3 += XXH_readLE64(p, endian) * PRIME64_2; 812 v3 = XXH_rotl64(v3, 31); 813 v3 *= PRIME64_1; 814 p += 8; 815 v4 += XXH_readLE64(p, endian) * PRIME64_2; 816 v4 = XXH_rotl64(v4, 31); 817 v4 *= PRIME64_1; 818 p += 8; 819 } 820 while (p <= limit); 821 822 state->v1 = v1; 823 state->v2 = v2; 824 state->v3 = v3; 825 state->v4 = v4; 826 } 827 828 if (p < bEnd) 829 { 830 XXH_memcpy(state->mem64, p, bEnd - p); 831 state->memsize = (int)(bEnd - p); 832 } 833 834 return XXH_OK; 835} 836 837XXH_errorcode XXH64_update(XXH64_state_t *state_in, const void *input, 838 size_t len) 839{ 840 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 841 842 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 843 return XXH64_update_endian(state_in, input, len, XXH_littleEndian); 844 else 845 return XXH64_update_endian(state_in, input, len, XXH_bigEndian); 846} 847 848 849 850FORCE_INLINE U64 XXH64_digest_endian(const XXH64_state_t *state_in, 851 XXH_endianess endian) 852{ 853 XXH_istate64_t *state = (XXH_istate64_t *) state_in; 854 const BYTE *p = (const BYTE *)state->mem64; 855 BYTE *bEnd = (BYTE *)state->mem64 + state->memsize; 856 U64 h64; 857 858 if (state->total_len >= 32) 859 { 860 U64 v1 = state->v1; 861 U64 v2 = state->v2; 862 U64 v3 = state->v3; 863 U64 v4 = state->v4; 864 865 h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 866 12) + XXH_rotl64(v4, 18); 867 868 v1 *= PRIME64_2; 869 v1 = XXH_rotl64(v1, 31); 870 v1 *= PRIME64_1; 871 h64 ^= v1; 872 h64 = h64 * PRIME64_1 + PRIME64_4; 873 874 v2 *= PRIME64_2; 875 v2 = XXH_rotl64(v2, 31); 876 v2 *= PRIME64_1; 877 h64 ^= v2; 878 h64 = h64 * PRIME64_1 + PRIME64_4; 879 880 v3 *= PRIME64_2; 881 v3 = XXH_rotl64(v3, 31); 882 v3 *= PRIME64_1; 883 h64 ^= v3; 884 h64 = h64 * PRIME64_1 + PRIME64_4; 885 886 v4 *= PRIME64_2; 887 v4 = XXH_rotl64(v4, 31); 888 v4 *= PRIME64_1; 889 h64 ^= v4; 890 h64 = h64 * PRIME64_1 + PRIME64_4; 891 } 892 else 893 h64 = state->seed + PRIME64_5; 894 895 h64 += (U64) state->total_len; 896 897 while (p + 8 <= bEnd) 898 { 899 U64 k1 = XXH_readLE64(p, endian); 900 k1 *= PRIME64_2; 901 k1 = XXH_rotl64(k1, 31); 902 k1 *= PRIME64_1; 903 h64 ^= k1; 904 h64 = XXH_rotl64(h64, 27) * PRIME64_1 + PRIME64_4; 905 p += 8; 906 } 907 908 if (p + 4 <= bEnd) 909 { 910 h64 ^= (U64)(XXH_readLE32(p, endian)) * PRIME64_1; 911 h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3; 912 p += 4; 913 } 914 915 while (p < bEnd) 916 { 917 h64 ^= (*p) * PRIME64_5; 918 h64 = XXH_rotl64(h64, 11) * PRIME64_1; 919 p++; 920 } 921 922 h64 ^= h64 >> 33; 923 h64 *= PRIME64_2; 924 h64 ^= h64 >> 29; 925 h64 *= PRIME64_3; 926 h64 ^= h64 >> 32; 927 928 return h64; 929} 930 931 932unsigned long long XXH64_digest(const XXH64_state_t *state_in) 933{ 934 XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; 935 936 if ((endian_detected == XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) 937 return XXH64_digest_endian(state_in, XXH_littleEndian); 938 else 939 return XXH64_digest_endian(state_in, XXH_bigEndian); 940} 941 942 943