--- /dev/null
+/*\r
+xxHash - Fast Hash algorithm\r
+Copyright (C) 2012-2013, Yann Collet.\r
+BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)\r
+\r
+Redistribution and use in source and binary forms, with or without\r
+modification, are permitted provided that the following conditions are\r
+met:\r
+\r
+* Redistributions of source code must retain the above copyright\r
+notice, this list of conditions and the following disclaimer.\r
+* Redistributions in binary form must reproduce the above\r
+copyright notice, this list of conditions and the following disclaimer\r
+in the documentation and/or other materials provided with the\r
+distribution.\r
+\r
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS\r
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT\r
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR\r
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT\r
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,\r
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT\r
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\r
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\r
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\r
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE\r
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\r
+\r
+You can contact the author at :\r
+- xxHash source repository : http://code.google.com/p/xxhash/\r
+*/\r
+\r
+\r
+\r
+//**************************************\r
+// Tuning parameters\r
+//**************************************\r
+// Unaligned memory access is automatically enabled for "common" CPU, such as x86.\r
+// For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected.\r
+// If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance.\r
+// You can also enable this parameter if you know your input data will always be aligned (boundaries of 4, for U32).\r
+#if defined(__ARM_FEATURE_UNALIGNED) || defined(__i386) || defined(_M_IX86) || defined(__x86_64__) || defined(_M_X64)\r
+# define XXH_USE_UNALIGNED_ACCESS 1\r
+#endif\r
+\r
+// XXH_ACCEPT_NULL_INPUT_POINTER :\r
+// If the input pointer is a null pointer, xxHash default behavior is to crash, since it is a bad input.\r
+// If this option is enabled, xxHash output for null input pointers will be the same as a null-length input.\r
+// This option has a very small performance cost (only measurable on small inputs).\r
+// By default, this option is disabled. To enable it, uncomment below define :\r
+//#define XXH_ACCEPT_NULL_INPUT_POINTER 1\r
+\r
+// XXH_FORCE_NATIVE_FORMAT :\r
+// By default, xxHash library provides endian-independant Hash values, based on little-endian convention.\r
+// Results are therefore identical for little-endian and big-endian CPU.\r
+// This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format.\r
+// Should endian-independance be of no importance for your application, you may uncomment the #define below.\r
+// It will improve speed for Big-endian CPU.\r
+// This option has no impact on Little_Endian CPU.\r
+//#define XXH_FORCE_NATIVE_FORMAT 1\r
+\r
+\r
+//**************************************\r
+// Compiler Options\r
+//**************************************\r
+#if defined(_MSC_VER) && !defined(__cplusplus) // Visual Studio\r
+# define inline __inline // Visual C is not C99, but supports some kind of inline\r
+#endif\r
+\r
+\r
+//**************************************\r
+// Includes & Memory related functions\r
+//**************************************\r
+#include "xxhash.h"\r
+// Modify the local functions below should you wish to use some other memory related routines\r
+// for malloc(), free()\r
+#include <stdlib.h>\r
+static inline void* XXH_malloc(size_t s) { return malloc(s); }\r
+static inline void XXH_free (void* p) { free(p); }\r
+// for memcpy()\r
+#include <string.h>\r
+static inline void* XXH_memcpy(void* dest, const void* src, size_t size) { return memcpy(dest,src,size); }\r
+\r
+\r
+//**************************************\r
+// CPU Feature Detection\r
+//**************************************\r
+// Little Endian or Big Endian ?\r
+// You can overwrite the #define below if you know your architecture endianess\r
+#if defined(XXH_FORCE_NATIVE_FORMAT) && (XXH_FORCE_NATIVE_FORMAT==1)\r
+// Force native format. The result will be endian dependant.\r
+# define XXH_BIG_ENDIAN 0\r
+#elif defined (__GLIBC__)\r
+# include <endian.h>\r
+# if (__BYTE_ORDER == __BIG_ENDIAN)\r
+# define XXH_BIG_ENDIAN 1\r
+# endif\r
+#elif (defined(__BIG_ENDIAN__) || defined(__BIG_ENDIAN) || defined(_BIG_ENDIAN)) && !(defined(__LITTLE_ENDIAN__) || defined(__LITTLE_ENDIAN) || defined(_LITTLE_ENDIAN))\r
+# define XXH_BIG_ENDIAN 1\r
+#elif defined(__sparc) || defined(__sparc__) \\r
+ || defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) \\r
+ || defined(__hpux) || defined(__hppa) \\r
+ || defined(_MIPSEB) || defined(__s390__)\r
+# define XXH_BIG_ENDIAN 1\r
+#endif\r
+\r
+#if !defined(XXH_BIG_ENDIAN)\r
+// Little Endian assumed. PDP Endian and other very rare endian format are unsupported.\r
+# define XXH_BIG_ENDIAN 0\r
+#endif\r
+\r
+\r
+//**************************************\r
+// Basic Types\r
+//**************************************\r
+#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L // C99\r
+# include <stdint.h>\r
+ typedef uint8_t BYTE;\r
+ typedef uint16_t U16;\r
+ typedef uint32_t U32;\r
+ typedef int32_t S32;\r
+ typedef uint64_t U64;\r
+#else\r
+ typedef unsigned char BYTE;\r
+ typedef unsigned short U16;\r
+ typedef unsigned int U32;\r
+ typedef signed int S32;\r
+ typedef unsigned long long U64;\r
+#endif\r
+\r
+#if defined(__GNUC__) && !defined(XXH_USE_UNALIGNED_ACCESS)\r
+# define _PACKED __attribute__ ((packed))\r
+#else\r
+# define _PACKED\r
+#endif\r
+\r
+#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__)\r
+# pragma pack(push, 1)\r
+#endif\r
+\r
+typedef struct _U32_S { U32 v; } _PACKED U32_S;\r
+\r
+#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__)\r
+# pragma pack(pop)\r
+#endif\r
+\r
+#define A32(x) (((U32_S *)(x))->v)\r
+\r
+\r
+//***************************************\r
+// Compiler-specific Functions and Macros\r
+//***************************************\r
+#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)\r
+\r
+// Note : although _rotl exists for minGW (GCC under windows), performance seems poor\r
+#if defined(_MSC_VER)\r
+# define XXH_rotl32(x,r) _rotl(x,r)\r
+#else\r
+# define XXH_rotl32(x,r) ((x << r) | (x >> (32 - r)))\r
+#endif\r
+\r
+#if defined(_MSC_VER) // Visual Studio\r
+# define XXH_swap32 _byteswap_ulong\r
+#elif GCC_VERSION >= 403\r
+# define XXH_swap32 __builtin_bswap32\r
+#else\r
+static inline U32 XXH_swap32 (U32 x) {\r
+ return ((x << 24) & 0xff000000 ) |\r
+ ((x << 8) & 0x00ff0000 ) |\r
+ ((x >> 8) & 0x0000ff00 ) |\r
+ ((x >> 24) & 0x000000ff );}\r
+#endif\r
+\r
+\r
+//**************************************\r
+// Constants\r
+//**************************************\r
+#define PRIME32_1 2654435761U\r
+#define PRIME32_2 2246822519U\r
+#define PRIME32_3 3266489917U\r
+#define PRIME32_4 668265263U\r
+#define PRIME32_5 374761393U\r
+\r
+\r
+//**************************************\r
+// Macros\r
+//**************************************\r
+#define XXH_STATIC_ASSERT(c) { enum { XXH_static_assert = 1/(!!(c)) }; } // use only *after* variable declarations\r
+#define XXH_LE32(p) (XXH_BIG_ENDIAN ? XXH_swap32(A32(p)) : A32(p))\r
+#define XXH_alignedLE32(p) (XXH_BIG_ENDIAN ? XXH_swap32(*(U32*)(p)) : *(U32*)(p))\r
+\r
+\r
+\r
+//****************************\r
+// Simple Hash Functions\r
+//****************************\r
+\r
+#if !defined(XXH_USE_UNALIGNED_ACCESS)\r
+// Specific version, for aligned 32-bits input. Useless for CPU supporting unaligned access.\r
+static U32 XXH32_alignedInput(const void* input, int len, U32 seed)\r
+{\r
+ const BYTE* p = (const BYTE*)input;\r
+ const BYTE* const bEnd = p + len;\r
+ U32 h32;\r
+\r
+ if (len>=16)\r
+ {\r
+ const BYTE* const limit = bEnd - 16;\r
+ U32 v1 = seed + PRIME32_1 + PRIME32_2;\r
+ U32 v2 = seed + PRIME32_2;\r
+ U32 v3 = seed + 0;\r
+ U32 v4 = seed - PRIME32_1;\r
+ do\r
+ {\r
+ v1 += XXH_alignedLE32(p) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4;\r
+ v2 += XXH_alignedLE32(p) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4;\r
+ v3 += XXH_alignedLE32(p) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4;\r
+ v4 += XXH_alignedLE32(p) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4;\r
+ } while (p<=limit);\r
+ h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18);\r
+ }\r
+ else { h32 = seed + PRIME32_5; }\r
+ h32 += (U32) len;\r
+ while (p<=bEnd-4)\r
+ {\r
+ h32 += XXH_alignedLE32(p) * PRIME32_3;\r
+ h32 = XXH_rotl32(h32, 17) * PRIME32_4 ;\r
+ p+=4;\r
+ }\r
+ while (p<bEnd)\r
+ {\r
+ h32 += (*p) * PRIME32_5;\r
+ h32 = XXH_rotl32(h32, 11) * PRIME32_1 ;\r
+ p++;\r
+ }\r
+ h32 ^= h32 >> 15;\r
+ h32 *= PRIME32_2;\r
+ h32 ^= h32 >> 13;\r
+ h32 *= PRIME32_3;\r
+ h32 ^= h32 >> 16;\r
+ return h32;\r
+}\r
+#endif\r
+\r
+U32 XXH32(const void* input, int len, U32 seed)\r
+{\r
+#if 0\r
+ // Simple version, good for code maintenance, but unfortunately slow for small inputs\r
+ void* state = XXH32_init(seed);\r
+ XXH32_update(state, input, len);\r
+ return XXH32_digest(state);\r
+#else\r
+\r
+ const BYTE* p = (const BYTE*)input;\r
+ const BYTE* const bEnd = p + len;\r
+ U32 h32;\r
+\r
+#ifdef XXH_ACCEPT_NULL_INPUT_POINTER\r
+ if (p==NULL) { len=0; p=(const BYTE*)16; }\r
+#endif\r
+\r
+#if !defined(XXH_USE_UNALIGNED_ACCESS)\r
+ if ((((U32)p) & 3) == 0) return XXH32_alignedInput(input, len, seed); // Input is aligned, let's leverage the speed advantage\r
+#endif\r
+\r
+ if (len>=16)\r
+ {\r
+ const BYTE* const limit = bEnd - 16;\r
+ U32 v1 = seed + PRIME32_1 + PRIME32_2;\r
+ U32 v2 = seed + PRIME32_2;\r
+ U32 v3 = seed + 0;\r
+ U32 v4 = seed - PRIME32_1;\r
+\r
+ do\r
+ {\r
+ v1 += XXH_LE32(p) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4;\r
+ v2 += XXH_LE32(p) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4;\r
+ v3 += XXH_LE32(p) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4;\r
+ v4 += XXH_LE32(p) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4;\r
+ } while (p<=limit);\r
+\r
+ h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18);\r
+ }\r
+ else\r
+ {\r
+ h32 = seed + PRIME32_5;\r
+ }\r
+\r
+ h32 += (U32) len;\r
+\r
+ while (p<=bEnd-4)\r
+ {\r
+ h32 += XXH_LE32(p) * PRIME32_3;\r
+ h32 = XXH_rotl32(h32, 17) * PRIME32_4 ;\r
+ p+=4;\r
+ }\r
+\r
+ while (p<bEnd)\r
+ {\r
+ h32 += (*p) * PRIME32_5;\r
+ h32 = XXH_rotl32(h32, 11) * PRIME32_1 ;\r
+ p++;\r
+ }\r
+\r
+ h32 ^= h32 >> 15;\r
+ h32 *= PRIME32_2;\r
+ h32 ^= h32 >> 13;\r
+ h32 *= PRIME32_3;\r
+ h32 ^= h32 >> 16;\r
+\r
+ return h32;\r
+\r
+#endif\r
+}\r
+\r
+\r
+//****************************\r
+// Advanced Hash Functions\r
+//****************************\r
+\r
+struct XXH_state32_t\r
+{\r
+ U64 total_len;\r
+ U32 seed;\r
+ U32 v1;\r
+ U32 v2;\r
+ U32 v3;\r
+ U32 v4;\r
+ int memsize;\r
+ char memory[16];\r
+};\r
+\r
+\r
+int XXH32_sizeofState() \r
+{\r
+ XXH_STATIC_ASSERT(XXH32_SIZEOFSTATE >= sizeof(struct XXH_state32_t)); // A compilation error here means XXH32_SIZEOFSTATE is not large enough\r
+ return sizeof(struct XXH_state32_t); \r
+}\r
+\r
+\r
+XXH_errorcode XXH32_resetState(void* state_in, U32 seed)\r
+{ \r
+ struct XXH_state32_t * state = (struct XXH_state32_t *) state_in;\r
+ state->seed = seed;\r
+ state->v1 = seed + PRIME32_1 + PRIME32_2;\r
+ state->v2 = seed + PRIME32_2;\r
+ state->v3 = seed + 0;\r
+ state->v4 = seed - PRIME32_1;\r
+ state->total_len = 0;\r
+ state->memsize = 0;\r
+ return XXH_OK;\r
+}\r
+\r
+\r
+void* XXH32_init (U32 seed)\r
+{\r
+ void* state = XXH_malloc (sizeof(struct XXH_state32_t));\r
+ XXH32_resetState(state, seed);\r
+ return state;\r
+}\r
+\r
+\r
+XXH_errorcode XXH32_update (void* state_in, const void* input, int len)\r
+{\r
+ struct XXH_state32_t * state = (struct XXH_state32_t *) state_in;\r
+ const BYTE* p = (const BYTE*)input;\r
+ const BYTE* const bEnd = p + len;\r
+\r
+#ifdef XXH_ACCEPT_NULL_INPUT_POINTER\r
+ if (input==NULL) return XXH_ERROR;\r
+#endif\r
+\r
+ state->total_len += len;\r
+\r
+ if (state->memsize + len < 16) // fill in tmp buffer\r
+ {\r
+ XXH_memcpy(state->memory + state->memsize, input, len);\r
+ state->memsize += len;\r
+ return XXH_OK;\r
+ }\r
+\r
+ if (state->memsize) // some data left from previous update\r
+ {\r
+ XXH_memcpy(state->memory + state->memsize, input, 16-state->memsize);\r
+ {\r
+ const U32* p32 = (const U32*)state->memory;\r
+ state->v1 += XXH_LE32(p32) * PRIME32_2; state->v1 = XXH_rotl32(state->v1, 13); state->v1 *= PRIME32_1; p32++;\r
+ state->v2 += XXH_LE32(p32) * PRIME32_2; state->v2 = XXH_rotl32(state->v2, 13); state->v2 *= PRIME32_1; p32++; \r
+ state->v3 += XXH_LE32(p32) * PRIME32_2; state->v3 = XXH_rotl32(state->v3, 13); state->v3 *= PRIME32_1; p32++;\r
+ state->v4 += XXH_LE32(p32) * PRIME32_2; state->v4 = XXH_rotl32(state->v4, 13); state->v4 *= PRIME32_1; p32++;\r
+ }\r
+ p += 16-state->memsize;\r
+ state->memsize = 0;\r
+ }\r
+\r
+ if (p <= bEnd-16)\r
+ {\r
+ const BYTE* const limit = bEnd - 16;\r
+ U32 v1 = state->v1;\r
+ U32 v2 = state->v2;\r
+ U32 v3 = state->v3;\r
+ U32 v4 = state->v4;\r
+\r
+ do\r
+ {\r
+ v1 += XXH_LE32(p) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4;\r
+ v2 += XXH_LE32(p) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4;\r
+ v3 += XXH_LE32(p) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4;\r
+ v4 += XXH_LE32(p) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4;\r
+ } while (p<=limit);\r
+\r
+ state->v1 = v1;\r
+ state->v2 = v2;\r
+ state->v3 = v3;\r
+ state->v4 = v4;\r
+ }\r
+\r
+ if (p < bEnd)\r
+ {\r
+ XXH_memcpy(state->memory, p, bEnd-p);\r
+ state->memsize = (int)(bEnd-p);\r
+ }\r
+\r
+ return XXH_OK;\r
+}\r
+\r
+\r
+U32 XXH32_intermediateDigest (void* state_in)\r
+{\r
+ struct XXH_state32_t * state = (struct XXH_state32_t *) state_in;\r
+ BYTE * p = (BYTE*)state->memory;\r
+ BYTE* bEnd = (BYTE*)state->memory + state->memsize;\r
+ U32 h32;\r
+\r
+ if (state->total_len >= 16)\r
+ {\r
+ h32 = XXH_rotl32(state->v1, 1) + XXH_rotl32(state->v2, 7) + XXH_rotl32(state->v3, 12) + XXH_rotl32(state->v4, 18);\r
+ }\r
+ else\r
+ {\r
+ h32 = state->seed + PRIME32_5;\r
+ }\r
+\r
+ h32 += (U32) state->total_len;\r
+\r
+ while (p<=bEnd-4)\r
+ {\r
+ h32 += XXH_LE32(p) * PRIME32_3;\r
+ h32 = XXH_rotl32(h32, 17) * PRIME32_4;\r
+ p+=4;\r
+ }\r
+\r
+ while (p<bEnd)\r
+ {\r
+ h32 += (*p) * PRIME32_5;\r
+ h32 = XXH_rotl32(h32, 11) * PRIME32_1;\r
+ p++;\r
+ }\r
+\r
+ h32 ^= h32 >> 15;\r
+ h32 *= PRIME32_2;\r
+ h32 ^= h32 >> 13;\r
+ h32 *= PRIME32_3;\r
+ h32 ^= h32 >> 16;\r
+\r
+ return h32;\r
+}\r
+\r
+\r
+U32 XXH32_digest (void* state_in)\r
+{\r
+ U32 h32 = XXH32_intermediateDigest(state_in);\r
+\r
+ XXH_free(state_in);\r
+\r
+ return h32;\r
+}\r
projects / c11concurrency-benchmarks.git / blobdiff