xref: /freebsd/sys/contrib/ck/src/ck_ht_hash.h (revision 43a5ec4eb41567cc92586503212743d89686d78f)
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