Commit | Line | Data |
---|---|---|
652ae149 JA |
1 | #include <stdlib.h> |
2 | #include <inttypes.h> | |
3 | ||
4 | #include "bloom.h" | |
5 | #include "../hash.h" | |
265c0032 | 6 | #include "../minmax.h" |
899834b5 JA |
7 | #include "../crc/xxhash.h" |
8 | #include "../crc/crc32c.h" | |
652ae149 JA |
9 | |
10 | struct bloom { | |
11 | uint64_t nentries; | |
12 | ||
13 | uint32_t *map; | |
14 | }; | |
15 | ||
16 | #define BITS_PER_INDEX (sizeof(uint32_t) * 8) | |
17 | #define BITS_INDEX_MASK (BITS_PER_INDEX - 1) | |
18 | ||
899834b5 JA |
19 | struct bloom_hash { |
20 | unsigned int seed; | |
21 | uint32_t (*fn)(const void *, uint32_t, uint32_t); | |
22 | }; | |
23 | ||
24 | static uint32_t b_crc32c(const void *buf, uint32_t len, uint32_t seed) | |
25 | { | |
26 | return fio_crc32c(buf, len); | |
27 | } | |
28 | ||
29 | struct bloom_hash hashes[] = { | |
30 | { | |
31 | .seed = 0x8989, | |
32 | .fn = jhash, | |
33 | }, | |
34 | { | |
35 | .seed = 0x8989, | |
36 | .fn = XXH32, | |
37 | }, | |
38 | { | |
39 | .seed = 0, | |
40 | .fn = b_crc32c, | |
41 | }, | |
42 | }; | |
43 | ||
652ae149 JA |
44 | #define N_HASHES 3 |
45 | ||
265c0032 JA |
46 | #define MIN_ENTRIES 1073741824UL |
47 | ||
652ae149 JA |
48 | struct bloom *bloom_new(uint64_t entries) |
49 | { | |
50 | struct bloom *b; | |
51 | size_t no_uints; | |
52 | ||
899834b5 JA |
53 | crc32c_intel_probe(); |
54 | ||
652ae149 JA |
55 | b = malloc(sizeof(*b)); |
56 | b->nentries = entries; | |
57 | no_uints = (entries + BITS_PER_INDEX - 1) / BITS_PER_INDEX; | |
265c0032 | 58 | no_uints = max((unsigned long) no_uints, MIN_ENTRIES); |
652ae149 JA |
59 | b->map = calloc(no_uints, sizeof(uint32_t)); |
60 | if (!b->map) { | |
61 | free(b); | |
62 | return NULL; | |
63 | } | |
64 | ||
65 | return b; | |
66 | } | |
67 | ||
68 | void bloom_free(struct bloom *b) | |
69 | { | |
70 | free(b->map); | |
71 | free(b); | |
72 | } | |
73 | ||
74 | static int __bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords, | |
75 | int set) | |
76 | { | |
899834b5 | 77 | uint32_t hash[N_HASHES]; |
652ae149 JA |
78 | int i, was_set; |
79 | ||
899834b5 JA |
80 | for (i = 0; i < N_HASHES; i++) { |
81 | hash[i] = hashes[i].fn(data, nwords, hashes[i].seed); | |
82 | hash[i] = hash[i] % b->nentries; | |
83 | } | |
652ae149 JA |
84 | |
85 | was_set = 0; | |
86 | for (i = 0; i < N_HASHES; i++) { | |
899834b5 JA |
87 | const unsigned int index = hash[i] / BITS_PER_INDEX; |
88 | const unsigned int bit = hash[i] & BITS_INDEX_MASK; | |
652ae149 JA |
89 | |
90 | if (b->map[index] & (1U << bit)) | |
91 | was_set++; | |
92 | if (set) | |
93 | b->map[index] |= 1U << bit; | |
94 | } | |
95 | ||
96 | return was_set == N_HASHES; | |
97 | } | |
98 | ||
99 | int bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords) | |
100 | { | |
101 | return __bloom_check(b, data, nwords, 0); | |
102 | } | |
103 | ||
104 | int bloom_set(struct bloom *b, uint32_t *data, unsigned int nwords) | |
105 | { | |
106 | return __bloom_check(b, data, nwords, 1); | |
107 | } |