1 /*
2 * Copyright (C) 2013 Andrea Mazzoleni
3 *
4 * This program is free software: you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation, either version 3 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program. If not, see <http://www.gnu.org/licenses/>.
16 */
17
18 /*
19 * Derivative work from SpookyV2.cpp/h
20 *
21 * WARNING!!!! Note that this implementation doesn't use the short hash optimization
22 * resulting in different hashes for any length shorter than 192 bytes
23 *
24 * SpookyHash
25 * http://burtleburtle.net/bob/hash/spooky.html
26 *
27 * Exact source used as reference:
28 * http://burtleburtle.net/bob/c/SpookyV2.h
29 * http://burtleburtle.net/bob/c/SpookyV2.cpp
30 */
31
32 // Spooky Hash
33 // A 128-bit noncryptographic hash, for checksums and table lookup
34 // By Bob Jenkins. Public domain.
35 // Oct 31 2010: published framework, disclaimer ShortHash isn't right
36 // Nov 7 2010: disabled ShortHash
37 // Oct 31 2011: replace End, ShortMix, ShortEnd, enable ShortHash again
38 // April 10 2012: buffer overflow on platforms without unaligned reads
39 // July 12 2012: was passing out variables in final to in/out in short
40 // July 30 2012: I reintroduced the buffer overflow
41 // August 5 2012: SpookyV2: d = should be d += in short hash, and remove extra mix from long hash
42 //
43 // Up to 3 bytes/cycle for long messages. Reasonably fast for short messages.
44 // All 1 or 2 bit deltas achieve avalanche within 1% bias per output bit.
45 //
46 // This was developed for and tested on 64-bit x86-compatible processors.
47 // It assumes the processor is little-endian. There is a macro
48 // controlling whether unaligned reads are allowed (by default they are).
49 // This should be an equally good hash on big-endian machines, but it will
50 // compute different results on them than on little-endian machines.
51 //
52 // Google's CityHash has similar specs to SpookyHash, and CityHash is faster
53 // on new Intel boxes. MD4 and MD5 also have similar specs, but they are orders
54 // of magnitude slower. CRCs are two or more times slower, but unlike
55 // SpookyHash, they have nice math for combining the CRCs of pieces to form
56 // the CRCs of wholes. There are also cryptographic hashes, but those are even
57 // slower than MD5.
58 //
59
60 #define Mix(data, s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11) \
61 s0 += data[0]; s2 ^= s10; s11 ^= s0; s0 = util_rotl64(s0, 11); s11 += s1; \
62 s1 += data[1]; s3 ^= s11; s0 ^= s1; s1 = util_rotl64(s1, 32); s0 += s2; \
63 s2 += data[2]; s4 ^= s0; s1 ^= s2; s2 = util_rotl64(s2, 43); s1 += s3; \
64 s3 += data[3]; s5 ^= s1; s2 ^= s3; s3 = util_rotl64(s3, 31); s2 += s4; \
65 s4 += data[4]; s6 ^= s2; s3 ^= s4; s4 = util_rotl64(s4, 17); s3 += s5; \
66 s5 += data[5]; s7 ^= s3; s4 ^= s5; s5 = util_rotl64(s5, 28); s4 += s6; \
67 s6 += data[6]; s8 ^= s4; s5 ^= s6; s6 = util_rotl64(s6, 39); s5 += s7; \
68 s7 += data[7]; s9 ^= s5; s6 ^= s7; s7 = util_rotl64(s7, 57); s6 += s8; \
69 s8 += data[8]; s10 ^= s6; s7 ^= s8; s8 = util_rotl64(s8, 55); s7 += s9; \
70 s9 += data[9]; s11 ^= s7; s8 ^= s9; s9 = util_rotl64(s9, 54); s8 += s10; \
71 s10 += data[10]; s0 ^= s8; s9 ^= s10; s10 = util_rotl64(s10, 22); s9 += s11; \
72 s11 += data[11]; s1 ^= s9; s10 ^= s11; s11 = util_rotl64(s11, 46); s10 += s0;
73
74 #define EndPartial(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11) \
75 h11 += h1; h2 ^= h11; h1 = util_rotl64(h1, 44); \
76 h0 += h2; h3 ^= h0; h2 = util_rotl64(h2, 15); \
77 h1 += h3; h4 ^= h1; h3 = util_rotl64(h3, 34); \
78 h2 += h4; h5 ^= h2; h4 = util_rotl64(h4, 21); \
79 h3 += h5; h6 ^= h3; h5 = util_rotl64(h5, 38); \
80 h4 += h6; h7 ^= h4; h6 = util_rotl64(h6, 33); \
81 h5 += h7; h8 ^= h5; h7 = util_rotl64(h7, 10); \
82 h6 += h8; h9 ^= h6; h8 = util_rotl64(h8, 13); \
83 h7 += h9; h10 ^= h7; h9 = util_rotl64(h9, 38); \
84 h8 += h10; h11 ^= h8; h10 = util_rotl64(h10, 53); \
85 h9 += h11; h0 ^= h9; h11 = util_rotl64(h11, 42); \
86 h10 += h0; h1 ^= h10; h0 = util_rotl64(h0, 54);
87
88 #define End(data, h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11) \
89 h0 += data[0]; h1 += data[1]; h2 += data[2]; h3 += data[3]; \
90 h4 += data[4]; h5 += data[5]; h6 += data[6]; h7 += data[7]; \
91 h8 += data[8]; h9 += data[9]; h10 += data[10]; h11 += data[11]; \
92 EndPartial(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11); \
93 EndPartial(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11); \
94 EndPartial(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11);
95
96 // number of uint64_t's in internal state
97 #define sc_numVars 12
98
99 // size of the internal state
100 #define sc_blockSize (sc_numVars * 8)
101
102 //
103 // sc_const: a constant which:
104 // * is not zero
105 // * is odd
106 // * is a not-very-regular mix of 1's and 0's
107 // * does not need any other special mathematical properties
108 //
109 #define sc_const 0xdeadbeefdeadbeefLL
110
SpookyHash128(const void * data,size_t size,const uint8_t * seed,uint8_t * digest)111 void SpookyHash128(const void* data, size_t size, const uint8_t* seed, uint8_t* digest)
112 {
113 uint64_t h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11;
114 uint64_t buf[sc_numVars];
115 size_t nblocks;
116 const uint64_t* blocks;
117 const uint64_t* end;
118 size_t size_remainder;
119 #if WORDS_BIGENDIAN
120 unsigned i;
121 #endif
122
123 h9 = util_read64(seed + 0);
124 h10 = util_read64(seed + 8);
125
126 h0 = h3 = h6 = h9;
127 h1 = h4 = h7 = h10;
128 h2 = h5 = h8 = h11 = sc_const;
129
130 nblocks = size / sc_blockSize;
131 blocks = data;
132 end = blocks + nblocks * sc_numVars;
133
134 /* body */
135 while (blocks < end) {
136 #if WORDS_BIGENDIAN
137 for (i = 0; i < sc_numVars; ++i)
138 buf[i] = util_swap64(blocks[i]);
139 Mix(buf, h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11);
140 #else
141 Mix(blocks, h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11);
142 #endif
143 blocks += sc_numVars;
144 }
145
146 /* tail */
147 size_remainder = (size - ((const uint8_t*)end - (const uint8_t*)data));
148 memcpy(buf, end, size_remainder);
149 memset(((uint8_t*)buf) + size_remainder, 0, sc_blockSize - size_remainder);
150 ((uint8_t*)buf)[sc_blockSize - 1] = size_remainder;
151
152 /* finalization */
153 #if WORDS_BIGENDIAN
154 for (i = 0; i < sc_numVars; ++i)
155 buf[i] = util_swap64(buf[i]);
156 #endif
157 End(buf, h0, h1, h2, h3, h4, h5, h6, h7, h8, h9, h10, h11);
158
159 util_write64(digest + 0, h0);
160 util_write64(digest + 8, h1);
161 }
162
163