[fio.git] / lib / bloom.c

#include <stdlib.h>
#include <inttypes.h>

#include "bloom.h"
#include "../hash.h"
#include "../minmax.h"
#include "../crc/xxhash.h"
#include "../crc/murmur3.h"
#include "../crc/crc32c.h"
#include "../crc/fnv.h"

struct bloom {
	uint64_t nentries;

	uint32_t *map;
};

#define BITS_PER_INDEX	(sizeof(uint32_t) * 8)
#define BITS_INDEX_MASK	(BITS_PER_INDEX - 1)

struct bloom_hash {
	unsigned int seed;
	uint32_t (*fn)(const void *, uint32_t, uint32_t);
};

static uint32_t bloom_crc32c(const void *buf, uint32_t len, uint32_t seed)
{
	return fio_crc32c(buf, len);
}

static uint32_t bloom_fnv(const void *buf, uint32_t len, uint32_t seed)
{
	return fnv(buf, len, seed);
}

#define BLOOM_SEED	0x8989

static struct bloom_hash hashes[] = {
	{
		.seed = BLOOM_SEED,
		.fn = jhash,
	},
	{
		.seed = BLOOM_SEED,
		.fn = XXH32,
	},
	{
		.seed = BLOOM_SEED,
		.fn = murmurhash3,
	},
	{
		.seed = BLOOM_SEED,
		.fn = bloom_crc32c,
	},
	{
		.seed = BLOOM_SEED,
		.fn = bloom_fnv,
	},
};

#define N_HASHES	5

struct bloom *bloom_new(uint64_t entries)
{
	struct bloom *b;
	size_t no_uints;

	crc32c_intel_probe();

	b = malloc(sizeof(*b));
	b->nentries = entries;
	no_uints = (entries + BITS_PER_INDEX - 1) / BITS_PER_INDEX;
	b->map = calloc(no_uints, sizeof(uint32_t));
	if (!b->map) {
		free(b);
		return NULL;
	}

	return b;
}

void bloom_free(struct bloom *b)
{
	free(b->map);
	free(b);
}

static bool __bloom_check(struct bloom *b, const void *data, unsigned int len,
			  bool set)
{
	uint32_t hash[N_HASHES];
	int i, was_set;

	for (i = 0; i < N_HASHES; i++) {
		hash[i] = hashes[i].fn(data, len, hashes[i].seed);
		hash[i] = hash[i] % b->nentries;
	}

	was_set = 0;
	for (i = 0; i < N_HASHES; i++) {
		const unsigned int index = hash[i] / BITS_PER_INDEX;
		const unsigned int bit = hash[i] & BITS_INDEX_MASK;

		if (b->map[index] & (1U << bit))
			was_set++;
		if (set)
			b->map[index] |= 1U << bit;
	}

	return was_set == N_HASHES;
}

bool bloom_set(struct bloom *b, uint32_t *data, unsigned int nwords)
{
	return __bloom_check(b, data, nwords * sizeof(uint32_t), true);
}

bool bloom_string(struct bloom *b, const char *data, unsigned int len,
		  bool set)
{
	return __bloom_check(b, data, len, set);
}
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	7	#include "../crc/xxhash.h"
f83ffd02	8	#include "../crc/murmur3.h"
78583f91 JA	9	#include "../crc/crc32c.h"
78583f91 JA	10	#include "../crc/fnv.h"
652ae149 JA	11
	12	struct bloom {
	13	uint64_t nentries;
	14
	15	uint32_t *map;
	16	};
	17
	18	#define BITS_PER_INDEX (sizeof(uint32_t) * 8)
	19	#define BITS_INDEX_MASK (BITS_PER_INDEX - 1)
	20
899834b5 JA	21	struct bloom_hash {
	22	unsigned int seed;
	23	uint32_t (fn)(const void , uint32_t, uint32_t);
	24	};
	25
78583f91 JA	26	static uint32_t bloom_crc32c(const void *buf, uint32_t len, uint32_t seed)
	27	{
	28	return fio_crc32c(buf, len);
	29	}
	30
	31	static uint32_t bloom_fnv(const void *buf, uint32_t len, uint32_t seed)
	32	{
	33	return fnv(buf, len, seed);
	34	}
	35
	36	#define BLOOM_SEED 0x8989
	37
a89ba4b1	38	static struct bloom_hash hashes[] = {
899834b5	39	{
78583f91	40	.seed = BLOOM_SEED,
899834b5 JA	41	.fn = jhash,
	42	},
	43	{
78583f91	44	.seed = BLOOM_SEED,
899834b5 JA	45	.fn = XXH32,
	46	},
	47	{
78583f91	48	.seed = BLOOM_SEED,
9f0e365d	49	.fn = murmurhash3,
899834b5	50	},
78583f91 JA	51	{
	52	.seed = BLOOM_SEED,
	53	.fn = bloom_crc32c,
	54	},
	55	{
	56	.seed = BLOOM_SEED,
	57	.fn = bloom_fnv,
	58	},
899834b5 JA	59	};
899834b5 JA	60
78583f91	61	#define N_HASHES 5
652ae149 JA	62
	63	struct bloom *bloom_new(uint64_t entries)
	64	{
	65	struct bloom *b;
	66	size_t no_uints;
	67
78583f91 JA	68	crc32c_intel_probe();
78583f91 JA	69
652ae149 JA	70	b = malloc(sizeof(*b));
	71	b->nentries = entries;
	72	no_uints = (entries + BITS_PER_INDEX - 1) / BITS_PER_INDEX;
	73	b->map = calloc(no_uints, sizeof(uint32_t));
	74	if (!b->map) {
	75	free(b);
	76	return NULL;
	77	}
	78
	79	return b;
	80	}
	81
	82	void bloom_free(struct bloom *b)
	83	{
	84	free(b->map);
	85	free(b);
	86	}
	87
0a301e93	88	static bool __bloom_check(struct bloom b, const void data, unsigned int len,
c0d75983	89	bool set)
652ae149	90	{
899834b5	91	uint32_t hash[N_HASHES];
652ae149 JA	92	int i, was_set;
652ae149 JA	93
899834b5	94	for (i = 0; i < N_HASHES; i++) {
7790f697	95	hash[i] = hashes[i].fn(data, len, hashes[i].seed);
899834b5 JA	96	hash[i] = hash[i] % b->nentries;
899834b5 JA	97	}
652ae149 JA	98
	99	was_set = 0;
	100	for (i = 0; i < N_HASHES; i++) {
899834b5 JA	101	const unsigned int index = hash[i] / BITS_PER_INDEX;
899834b5 JA	102	const unsigned int bit = hash[i] & BITS_INDEX_MASK;
652ae149 JA	103
	104	if (b->map[index] & (1U << bit))
	105	was_set++;
	106	if (set)
	107	b->map[index] \|= 1U << bit;
	108	}
	109
	110	return was_set == N_HASHES;
	111	}
	112
c0d75983	113	bool bloom_set(struct bloom b, uint32_t data, unsigned int nwords)
652ae149	114	{
7790f697	115	return __bloom_check(b, data, nwords * sizeof(uint32_t), true);
652ae149	116	}
0a301e93	117
eb50727a JA	118	bool bloom_string(struct bloom b, const char data, unsigned int len,
eb50727a JA	119	bool set)
0a301e93	120	{
eb50727a	121	return __bloom_check(b, data, len, set);
0a301e93	122	}