1 /* 2 * Copyright 2012-2015 Samy Al Bahra 3 * Copyright 2011-2014 AppNexus, Inc. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 */ 26 27 #ifndef CK_HT_HASH_H 28 #define CK_HT_HASH_H 29 30 /* 31 * This is the Murmur hash written by Austin Appleby. 32 */ 33 34 #include <ck_stdint.h> 35 #include <ck_string.h> 36 37 //----------------------------------------------------------------------------- 38 // MurmurHash3 was written by Austin Appleby, and is placed in the public 39 // domain. The author hereby disclaims copyright to this source code. 40 41 // Note - The x86 and x64 versions do _not_ produce the same results, as the 42 // algorithms are optimized for their respective platforms. You can still 43 // compile and run any of them on any platform, but your performance with the 44 // non-native version will be less than optimal. 45 46 //----------------------------------------------------------------------------- 47 // Platform-specific functions and macros 48 49 // Microsoft Visual Studio 50 51 #if defined(_MSC_VER) 52 53 #define FORCE_INLINE __forceinline 54 55 #include <stdlib.h> 56 57 #define ROTL32(x,y) _rotl(x,y) 58 #define ROTL64(x,y) _rotl64(x,y) 59 60 #define BIG_CONSTANT(x) (x) 61 62 // Other compilers 63 64 #else // defined(_MSC_VER) 65 66 #define FORCE_INLINE inline __attribute__((always_inline)) 67 68 static inline uint32_t rotl32 ( uint32_t x, int8_t r ) 69 { 70 return (x << r) | (x >> (32 - r)); 71 } 72 73 static inline uint64_t rotl64 ( uint64_t x, int8_t r ) 74 { 75 return (x << r) | (x >> (64 - r)); 76 } 77 78 #define ROTL32(x,y) rotl32(x,y) 79 #define ROTL64(x,y) rotl64(x,y) 80 81 #define BIG_CONSTANT(x) (x##LLU) 82 83 #endif // !defined(_MSC_VER) 84 85 //----------------------------------------------------------------------------- 86 // Block read - if your platform needs to do endian-swapping or can only 87 // handle aligned reads, do the conversion here 88 89 FORCE_INLINE static uint32_t getblock ( const uint32_t * p, int i ) 90 { 91 #ifdef __s390x__ 92 uint32_t res; 93 94 __asm__ (" lrv %0,%1\n" 95 : "=r" (res) : "Q" (p[i]) : "cc", "mem"); 96 return res; 97 #else 98 return p[i]; 99 #endif /* !__s390x__ */ 100 } 101 102 //----------------------------------------------------------------------------- 103 // Finalization mix - force all bits of a hash block to avalanche 104 105 FORCE_INLINE static uint32_t fmix ( uint32_t h ) 106 { 107 h ^= h >> 16; 108 h *= 0x85ebca6b; 109 h ^= h >> 13; 110 h *= 0xc2b2ae35; 111 h ^= h >> 16; 112 113 return h; 114 } 115 116 //----------------------------------------------------------------------------- 117 118 static inline void MurmurHash3_x86_32 ( const void * key, int len, 119 uint32_t seed, uint32_t * out ) 120 { 121 const uint8_t * data = (const uint8_t*)key; 122 const int nblocks = len / 4; 123 int i; 124 125 uint32_t h1 = seed; 126 127 uint32_t c1 = 0xcc9e2d51; 128 uint32_t c2 = 0x1b873593; 129 130 //---------- 131 // body 132 133 const uint32_t * blocks = (const uint32_t *)(const void *)(data + nblocks*4); 134 135 for(i = -nblocks; i; i++) 136 { 137 uint32_t k1 = getblock(blocks,i); 138 139 k1 *= c1; 140 k1 = ROTL32(k1,15); 141 k1 *= c2; 142 143 h1 ^= k1; 144 h1 = ROTL32(h1,13); 145 h1 = h1*5+0xe6546b64; 146 } 147 148 //---------- 149 // tail 150 151 const uint8_t * tail = (const uint8_t*)(data + nblocks*4); 152 153 uint32_t k1 = 0; 154 155 switch(len & 3) 156 { 157 case 3: k1 ^= tail[2] << 16; 158 /* fall through */ 159 case 2: k1 ^= tail[1] << 8; 160 /* fall through */ 161 case 1: k1 ^= tail[0]; 162 k1 *= c1; k1 = ROTL32(k1,15); k1 *= c2; h1 ^= k1; 163 }; 164 165 //---------- 166 // finalization 167 168 h1 ^= len; 169 170 h1 = fmix(h1); 171 172 *(uint32_t *)out = h1; 173 } 174 175 static inline uint64_t MurmurHash64A ( const void * key, int len, uint64_t seed ) 176 { 177 const uint64_t m = BIG_CONSTANT(0xc6a4a7935bd1e995); 178 const int r = 47; 179 180 uint64_t h = seed ^ (len * m); 181 182 const uint64_t * data = (const uint64_t *)key; 183 const uint64_t * end = data + (len/8); 184 185 while(data != end) 186 { 187 uint64_t k; 188 189 if (!((uintptr_t)data & 0x7)) 190 k = *data++; 191 else { 192 memcpy(&k, data, sizeof(k)); 193 data++; 194 } 195 196 k *= m; 197 k ^= k >> r; 198 k *= m; 199 200 h ^= k; 201 h *= m; 202 } 203 204 const unsigned char * data2 = (const unsigned char*)data; 205 206 switch(len & 7) 207 { 208 case 7: h ^= (uint64_t)(data2[6]) << 48; 209 /* fall through */ 210 case 6: h ^= (uint64_t)(data2[5]) << 40; 211 /* fall through */ 212 case 5: h ^= (uint64_t)(data2[4]) << 32; 213 /* fall through */ 214 case 4: h ^= (uint64_t)(data2[3]) << 24; 215 /* fall through */ 216 case 3: h ^= (uint64_t)(data2[2]) << 16; 217 /* fall through */ 218 case 2: h ^= (uint64_t)(data2[1]) << 8; 219 /* fall through */ 220 case 1: h ^= (uint64_t)(data2[0]); 221 h *= m; 222 }; 223 224 h ^= h >> r; 225 h *= m; 226 h ^= h >> r; 227 228 return h; 229 } 230 231 232 // 64-bit hash for 32-bit platforms 233 234 static inline uint64_t MurmurHash64B ( const void * key, int len, uint64_t seed ) 235 { 236 const uint32_t m = 0x5bd1e995; 237 const int r = 24; 238 239 uint32_t h1 = (uint32_t)(seed) ^ len; 240 uint32_t h2 = (uint32_t)(seed >> 32); 241 242 const uint32_t * data = (const uint32_t *)key; 243 244 while(len >= 8) 245 { 246 uint32_t k1 = *data++; 247 k1 *= m; k1 ^= k1 >> r; k1 *= m; 248 h1 *= m; h1 ^= k1; 249 len -= 4; 250 251 uint32_t k2 = *data++; 252 k2 *= m; k2 ^= k2 >> r; k2 *= m; 253 h2 *= m; h2 ^= k2; 254 len -= 4; 255 } 256 257 if(len >= 4) 258 { 259 uint32_t k1 = *data++; 260 k1 *= m; k1 ^= k1 >> r; k1 *= m; 261 h1 *= m; h1 ^= k1; 262 len -= 4; 263 } 264 265 switch(len) 266 { 267 case 3: h2 ^= ((const unsigned char*)data)[2] << 16; 268 /* fall through */ 269 case 2: h2 ^= ((const unsigned char*)data)[1] << 8; 270 /* fall through */ 271 case 1: h2 ^= ((const unsigned char*)data)[0]; 272 h2 *= m; 273 }; 274 275 h1 ^= h2 >> 18; h1 *= m; 276 h2 ^= h1 >> 22; h2 *= m; 277 h1 ^= h2 >> 17; h1 *= m; 278 h2 ^= h1 >> 19; h2 *= m; 279 280 uint64_t h = h1; 281 282 h = (h << 32) | h2; 283 284 return h; 285 } 286 287 #endif /* CK_HT_HASH_H */ 288