xref: /qemu/include/qemu/xxhash.h (revision 2e8f72ac)
1 /*
2  * xxHash - Fast Hash algorithm
3  * Copyright (C) 2012-2016, Yann Collet
4  *
5  * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions are
9  * met:
10  *
11  * + Redistributions of source code must retain the above copyright
12  * notice, this list of conditions and the following disclaimer.
13  * + Redistributions in binary form must reproduce the above
14  * copyright notice, this list of conditions and the following disclaimer
15  * in the documentation and/or other materials provided with the
16  * distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29  *
30  * You can contact the author at :
31  * - xxHash source repository : https://github.com/Cyan4973/xxHash
32  */
33 
34 #ifndef QEMU_XXHASH_H
35 #define QEMU_XXHASH_H
36 
37 #include "qemu/bitops.h"
38 
39 #define PRIME32_1   2654435761U
40 #define PRIME32_2   2246822519U
41 #define PRIME32_3   3266489917U
42 #define PRIME32_4    668265263U
43 #define PRIME32_5    374761393U
44 
45 #define QEMU_XXHASH_SEED 1
46 
47 /*
48  * xxhash32, customized for input variables that are not guaranteed to be
49  * contiguous in memory.
50  */
51 static inline uint32_t
52 qemu_xxhash7(uint64_t ab, uint64_t cd, uint32_t e, uint32_t f, uint32_t g)
53 {
54     uint32_t v1 = QEMU_XXHASH_SEED + PRIME32_1 + PRIME32_2;
55     uint32_t v2 = QEMU_XXHASH_SEED + PRIME32_2;
56     uint32_t v3 = QEMU_XXHASH_SEED + 0;
57     uint32_t v4 = QEMU_XXHASH_SEED - PRIME32_1;
58     uint32_t a = ab;
59     uint32_t b = ab >> 32;
60     uint32_t c = cd;
61     uint32_t d = cd >> 32;
62     uint32_t h32;
63 
64     v1 += a * PRIME32_2;
65     v1 = rol32(v1, 13);
66     v1 *= PRIME32_1;
67 
68     v2 += b * PRIME32_2;
69     v2 = rol32(v2, 13);
70     v2 *= PRIME32_1;
71 
72     v3 += c * PRIME32_2;
73     v3 = rol32(v3, 13);
74     v3 *= PRIME32_1;
75 
76     v4 += d * PRIME32_2;
77     v4 = rol32(v4, 13);
78     v4 *= PRIME32_1;
79 
80     h32 = rol32(v1, 1) + rol32(v2, 7) + rol32(v3, 12) + rol32(v4, 18);
81     h32 += 28;
82 
83     h32 += e * PRIME32_3;
84     h32  = rol32(h32, 17) * PRIME32_4;
85 
86     h32 += f * PRIME32_3;
87     h32  = rol32(h32, 17) * PRIME32_4;
88 
89     h32 += g * PRIME32_3;
90     h32  = rol32(h32, 17) * PRIME32_4;
91 
92     h32 ^= h32 >> 15;
93     h32 *= PRIME32_2;
94     h32 ^= h32 >> 13;
95     h32 *= PRIME32_3;
96     h32 ^= h32 >> 16;
97 
98     return h32;
99 }
100 
101 static inline uint32_t qemu_xxhash2(uint64_t ab)
102 {
103     return qemu_xxhash7(ab, 0, 0, 0, 0);
104 }
105 
106 static inline uint32_t qemu_xxhash4(uint64_t ab, uint64_t cd)
107 {
108     return qemu_xxhash7(ab, cd, 0, 0, 0);
109 }
110 
111 static inline uint32_t qemu_xxhash5(uint64_t ab, uint64_t cd, uint32_t e)
112 {
113     return qemu_xxhash7(ab, cd, e, 0, 0);
114 }
115 
116 static inline uint32_t qemu_xxhash6(uint64_t ab, uint64_t cd, uint32_t e,
117                                     uint32_t f)
118 {
119     return qemu_xxhash7(ab, cd, e, f, 0);
120 }
121 
122 /*
123  * Component parts of the XXH64 algorithm from
124  * https://github.com/Cyan4973/xxHash/blob/v0.8.0/xxhash.h
125  *
126  * The complete algorithm looks like
127  *
128  *  i = 0;
129  *  if (len >= 32) {
130  *      v1 = seed + XXH_PRIME64_1 + XXH_PRIME64_2;
131  *      v2 = seed + XXH_PRIME64_2;
132  *      v3 = seed + 0;
133  *      v4 = seed - XXH_PRIME64_1;
134  *      do {
135  *          v1 = XXH64_round(v1, get64bits(input + i));
136  *          v2 = XXH64_round(v2, get64bits(input + i + 8));
137  *          v3 = XXH64_round(v3, get64bits(input + i + 16));
138  *          v4 = XXH64_round(v4, get64bits(input + i + 24));
139  *      } while ((i += 32) <= len);
140  *      h64 = XXH64_mergerounds(v1, v2, v3, v4);
141  *  } else {
142  *      h64 = seed + XXH_PRIME64_5;
143  *  }
144  *  h64 += len;
145  *
146  *  for (; i + 8 <= len; i += 8) {
147  *      h64 ^= XXH64_round(0, get64bits(input + i));
148  *      h64 = rol64(h64, 27) * XXH_PRIME64_1 + XXH_PRIME64_4;
149  *  }
150  *  for (; i + 4 <= len; i += 4) {
151  *      h64 ^= get32bits(input + i) * PRIME64_1;
152  *      h64 = rol64(h64, 23) * XXH_PRIME64_2 + XXH_PRIME64_3;
153  *  }
154  *  for (; i < len; i += 1) {
155  *      h64 ^= get8bits(input + i) * XXH_PRIME64_5;
156  *      h64 = rol64(h64, 11) * XXH_PRIME64_1;
157  *  }
158  *
159  *  return XXH64_avalanche(h64)
160  *
161  * Exposing the pieces instead allows for simplified usage when
162  * the length is a known constant and the inputs are in registers.
163  */
164 #define XXH_PRIME64_1   0x9E3779B185EBCA87ULL
165 #define XXH_PRIME64_2   0xC2B2AE3D27D4EB4FULL
166 #define XXH_PRIME64_3   0x165667B19E3779F9ULL
167 #define XXH_PRIME64_4   0x85EBCA77C2B2AE63ULL
168 #define XXH_PRIME64_5   0x27D4EB2F165667C5ULL
169 
170 static inline uint64_t XXH64_round(uint64_t acc, uint64_t input)
171 {
172     return rol64(acc + input * XXH_PRIME64_2, 31) * XXH_PRIME64_1;
173 }
174 
175 static inline uint64_t XXH64_mergeround(uint64_t acc, uint64_t val)
176 {
177     return (acc ^ XXH64_round(0, val)) * XXH_PRIME64_1 + XXH_PRIME64_4;
178 }
179 
180 static inline uint64_t XXH64_mergerounds(uint64_t v1, uint64_t v2,
181                                          uint64_t v3, uint64_t v4)
182 {
183     uint64_t h64;
184 
185     h64 = rol64(v1, 1) + rol64(v2, 7) + rol64(v3, 12) + rol64(v4, 18);
186     h64 = XXH64_mergeround(h64, v1);
187     h64 = XXH64_mergeround(h64, v2);
188     h64 = XXH64_mergeround(h64, v3);
189     h64 = XXH64_mergeround(h64, v4);
190 
191     return h64;
192 }
193 
194 static inline uint64_t XXH64_avalanche(uint64_t h64)
195 {
196     h64 ^= h64 >> 33;
197     h64 *= XXH_PRIME64_2;
198     h64 ^= h64 >> 29;
199     h64 *= XXH_PRIME64_3;
200     h64 ^= h64 >> 32;
201     return h64;
202 }
203 
204 static inline uint64_t qemu_xxhash64_4(uint64_t a, uint64_t b,
205                                        uint64_t c, uint64_t d)
206 {
207     uint64_t v1 = QEMU_XXHASH_SEED + XXH_PRIME64_1 + XXH_PRIME64_2;
208     uint64_t v2 = QEMU_XXHASH_SEED + XXH_PRIME64_2;
209     uint64_t v3 = QEMU_XXHASH_SEED + 0;
210     uint64_t v4 = QEMU_XXHASH_SEED - XXH_PRIME64_1;
211 
212     v1 = XXH64_round(v1, a);
213     v2 = XXH64_round(v2, b);
214     v3 = XXH64_round(v3, c);
215     v4 = XXH64_round(v4, d);
216 
217     return XXH64_avalanche(XXH64_mergerounds(v1, v2, v3, v4));
218 }
219 
220 #endif /* QEMU_XXHASH_H */
221