Merge pull request #14 from paulborile/wyhash-seed-fix2

fixed wyhash seeding

libfh/CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## 1.0.1 Sept Jan 2024
+
+- fixed seeding (was not working properly). Tests now show taht seeding works
+- breaking change on the hashfun signature. Additional void * used to pass fh_t where seed/secret is kept
+
+
 ## 1.0.0 Sept 2023
 
 - introduced buckets for collisions (size 8) heavily inspired by go map source

libfh/fh.c

@@ -22,21 +22,23 @@
 #include  "wyhash.h"
 
 // version
-static char version[] = "1.0.0";
+static char version[] = "1.0.1";
 
 static void wyhash_hash_init(fh_t *fh)
 {
-    make_secret(time(NULL), fh->seed);
+    fh->seed = time(NULL);
+    make_secret(fh->seed, fh->secret);
 }
 
-uint64_t fh_default_hash(char *key)
+uint64_t fh_default_hash(void *data, char *key)
 {
+    fh_t *fh = (fh_t *)data;
     if (key[0] == '\0' )
     {
         // return fixed value on empty string
         return 1;
     }
-    uint64_t h = wyhash(key, strlen(key), 0, _wyp);
+    uint64_t h = wyhash(key, strlen(key), fh->seed, fh->secret);
     return h;
 }
 
@@ -372,7 +374,7 @@ int fh_insert(fh_t *fh, char *key, void *block)
     // uint8_t mini_hash;
     // MAKE_MINIHASH(i, mini_hash);
 
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
     i = h & (fh->h_dim -1);
@@ -528,7 +530,7 @@ int fh_del(fh_t *fh, char *key)
     FH_KEY_CHECK(key);
 
     // find bucket
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
     bckt_num = h & (fh->h_dim -1);
@@ -553,7 +555,7 @@ int fh_dellocked(fh_t *fh, char *key, int locked_bucket)
     FH_CHECK(fh);
     FH_KEY_CHECK(key);
 
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
 
@@ -700,7 +702,7 @@ int fh_search(fh_t *fh, char *key, void *block, int block_size)
     }
 
     // find bucket
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
     i = h & (fh->h_dim -1);
@@ -763,7 +765,7 @@ void *fh_get(fh_t *fh, char *key, int *error)
     void *opaque = NULL;
 
     // find bucket
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
     i = h & (fh->h_dim -1);
@@ -810,7 +812,7 @@ void *fh_searchlock(fh_t *fh, char *key, int *slot, int *error)
     }
 
     // find bucket
-    uint64_t h = fh->hash_function(key);
+    uint64_t h = fh->hash_function(fh, key);
     uint8_t mini_hash;
     MAKE_MINIHASH(h, mini_hash); // get the 8 most sign bits
     i = h & (fh->h_dim -1);

libfh/fh.h

@@ -93,7 +93,7 @@ struct _f_hash {
     fh_bucket *h_bucket;
 };
 typedef struct _f_hash f_hash;
-typedef uint64_t (*fh_hash_fun)(char *key);
+typedef uint64_t (*fh_hash_fun)(void *data, char *key);
 
 // fh object
 struct _fh_t {
@@ -110,7 +110,8 @@ struct _fh_t {
     pthread_mutex_t *h_lock;
     int n_lock;
     f_hash *hash_table;
-    uint64_t seed[4]; // wyhash seed
+    uint64_t seed; // wyhash seed
+    uint64_t secret[4]; // wyhash secret
 };
 typedef struct _fh_t fh_t;
 
@@ -139,8 +140,7 @@ void *fh_searchlock(fh_t *fh, char *key, int *slot, int *error);
 int fh_dellocked(fh_t *fh, char *key, int locked_slot);
 int fh_releaselock(fh_t *fh, int slot);
 
-uint64_t fh_default_hash(char *key);
-// compute the hash size given initial dimension
+uint64_t fh_default_hash(void *data, char *key);// compute the hash size given initial dimension
 unsigned int fh_hash_size(unsigned int s);
 
 // elements in enumeration

libfh/test/BenchFH.cpp

@@ -11,11 +11,11 @@
 #include "fh.h"
 
 static void generate_random_str(int seed, char *str, int min_len, int max_len);
-uint64_t    murmur64a_hash(char *key);
+uint64_t    murmur64a_hash(void *data, char *key);
 /*
  * oat hash (one at a time hash), Bob Jenkins, used by cfu hash and perl
  */
-unsigned int fh_default_hash(char *key, int dim);
+uint64_t fh_default_hash(void *data, char *key);
 
 
 PICOBENCH_SUITE("BenchFH");
@@ -84,14 +84,15 @@ static void HashFunctionsDefaultX100(picobench::state &s)
     unsigned int hash;
     char key[65];
     int i = 0;
+    fh_t fh;
 
     generate_random_str(i, key, 32, 64);
 
     for (auto _ : s)
     {
         for (i = 0; i<100; i++)
         {
-            hash += fh_default_hash(key) & 127;
+            hash += fh_default_hash(&fh, key) & 127;
         }
     }
     // just to avoid optimizer removing all code..
@@ -108,14 +109,15 @@ static void HashFunctionsMurmurX100(picobench::state &s)
     unsigned int hash;
     char key[65];
     int i = 0;
+    fh_t fh;
 
     generate_random_str(i, key, 32, 64);
 
     for (auto _ : s)
     {
         for (i = 0; i<100; i++)
         {
-            hash += murmur64a_hash(key) & 127;
+            hash += murmur64a_hash(&fh, key) & 127;
         }
     }
     // just to avoid optimizer removing all code..
@@ -148,7 +150,7 @@ PICOBENCH(GenerateRandomString).label("GenerateRandomString").samples(10);
 
 #define BIG_CONSTANT(x) (x ## LLU)
 
-uint64_t    murmur64a_hash(char *key)
+uint64_t    murmur64a_hash(void *unused, char *key)
 {
     const uint64_t m = BIG_CONSTANT(0xc6a4a7935bd1e995);
     const int r = 47;

libfh/test/TestFH.cpp

@@ -84,7 +84,7 @@ void dofree(void *data)
 
 // worst hash function of the history
 // just to check if it works, all inserts will result in collisions
-uint64_t    custom_hash(char *key)
+uint64_t    custom_hash(void *data, char *key)
 {
     return 42;
 }
@@ -480,7 +480,7 @@ TEST(FH, test_attr_methods)
     attribute = 0;
     result = fh_getattr(fhash, FH_ATTR_COLLISION, &attribute);
     EXPECT_EQ(result, 1);
-    EXPECT_EQ(attribute, 3);
+    EXPECT_GE(attribute, 0);
 
     // Destroy hash table
     fh_destroy(fhash);