Add bloom filter

[fio.git] / lib / bloom.c

diff --git a/lib/bloom.c b/lib/bloom.c
new file mode 100644 (file)
index 0000000..fbae808
--- /dev/null
+++ b/lib/bloom.c
@@ -0,0 +1,73 @@
+#include <stdlib.h>
+#include <inttypes.h>
+
+#include "bloom.h"
+#include "../hash.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)
+
+static unsigned int jhash_init[] = { 0, 0x12db635, 0x2a4a53 };
+#define N_HASHES       3
+
+struct bloom *bloom_new(uint64_t entries)
+{
+       struct bloom *b;
+       size_t no_uints;
+
+       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 int __bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords,
+                        int set)
+{
+       uint32_t hashes[N_HASHES];
+       int i, was_set;
+
+       for (i = 0; i < N_HASHES; i++)
+               hashes[i] = jhash(data, nwords, jhash_init[i]) % b->nentries;
+
+       was_set = 0;
+       for (i = 0; i < N_HASHES; i++) {
+               const unsigned int index = hashes[i] / BITS_PER_INDEX;
+               const unsigned int bit = hashes[i] & BITS_INDEX_MASK;
+
+               if (b->map[index] & (1U << bit))
+                       was_set++;
+               if (set)
+                       b->map[index] |= 1U << bit;
+       }
+
+       return was_set == N_HASHES;
+}
+
+int bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords)
+{
+       return __bloom_check(b, data, nwords, 0);
+}
+
+int bloom_set(struct bloom *b, uint32_t *data, unsigned int nwords)
+{
+       return __bloom_check(b, data, nwords, 1);
+}