bloom: add string version

[fio.git] / lib / bloom.c

diff --git a/lib/bloom.c b/lib/bloom.c
index b469fdedac0d57cc133bbbdc4067032a4950e13f..c2e6c11f2e1444020ae18225f8277938827c37f9 100644 (file)
--- a/lib/bloom.c
+++ b/lib/bloom.c
@@ -5,7 +5,9 @@
 #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;
@@ -21,27 +23,42 @@ struct bloom_hash {
        uint32_t (*fn)(const void *, uint32_t, uint32_t);
 };
 
-static uint32_t b_crc32c(const void *buf, uint32_t len, uint32_t seed)
+static uint32_t bloom_crc32c(const void *buf, uint32_t len, uint32_t seed)
 {
        return fio_crc32c(buf, len);
 }
 
-struct bloom_hash hashes[] = {
+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 = 0x8989,
+               .seed = BLOOM_SEED,
                .fn = jhash,
        },
        {
-               .seed = 0x8989,
+               .seed = BLOOM_SEED,
                .fn = XXH32,
        },
        {
-               .seed = 0,
-               .fn = b_crc32c,
+               .seed = BLOOM_SEED,
+               .fn = murmurhash3,
+       },
+       {
+               .seed = BLOOM_SEED,
+               .fn = bloom_crc32c,
+       },
+       {
+               .seed = BLOOM_SEED,
+               .fn = bloom_fnv,
        },
 };
 
-#define N_HASHES       3
+#define N_HASHES       5
 
 #define MIN_ENTRIES    1073741824UL
 
@@ -71,14 +88,14 @@ void bloom_free(struct bloom *b)
        free(b);
 }
 
-static int __bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords,
-                        int set)
+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, nwords, hashes[i].seed);
+               hash[i] = hashes[i].fn(data, len, hashes[i].seed);
                hash[i] = hash[i] % b->nentries;
        }
 
@@ -96,12 +113,12 @@ static int __bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords,
        return was_set == N_HASHES;
 }
 
-int bloom_check(struct bloom *b, uint32_t *data, unsigned int nwords)
+bool bloom_set(struct bloom *b, uint32_t *data, unsigned int nwords)
 {
-       return __bloom_check(b, data, nwords, 0);
+       return __bloom_check(b, data, nwords * sizeof(uint32_t), true);
 }
 
-int bloom_set(struct bloom *b, uint32_t *data, unsigned int nwords)
+bool bloom_set_string(struct bloom *b, const char *data, unsigned int len)
 {
-       return __bloom_check(b, data, nwords, 1);
+       return __bloom_check(b, data, len, true);
 }