hashmap.c

   1 /*
   2  * Generic implementation of hash-based key value mappings.
   3  */
   4 #include "cache.h"
   5 #include "hashmap.h"
   6
   7 #define FNV32_BASE ((unsigned int) 0x811c9dc5)
   8 #define FNV32_PRIME ((unsigned int) 0x01000193)
   9
  10 unsigned int strhash(const char *str)
  11 {
  12         unsigned int c, hash = FNV32_BASE;
  13         while ((c = (unsigned char) *str++))
  14                 hash = (hash * FNV32_PRIME) ^ c;
  15         return hash;
  16 }
  17
  18 unsigned int strihash(const char *str)
  19 {
  20         unsigned int c, hash = FNV32_BASE;
  21         while ((c = (unsigned char) *str++)) {
  22                 if (c >= 'a' && c <= 'z')
  23                         c -= 'a' - 'A';
  24                 hash = (hash * FNV32_PRIME) ^ c;
  25         }
  26         return hash;
  27 }
  28
  29 unsigned int memhash(const void *buf, size_t len)
  30 {
  31         unsigned int hash = FNV32_BASE;
  32         unsigned char *ucbuf = (unsigned char *) buf;
  33         while (len--) {
  34                 unsigned int c = *ucbuf++;
  35                 hash = (hash * FNV32_PRIME) ^ c;
  36         }
  37         return hash;
  38 }
  39
  40 unsigned int memihash(const void *buf, size_t len)
  41 {
  42         unsigned int hash = FNV32_BASE;
  43         unsigned char *ucbuf = (unsigned char *) buf;
  44         while (len--) {
  45                 unsigned int c = *ucbuf++;
  46                 if (c >= 'a' && c <= 'z')
  47                         c -= 'a' - 'A';
  48                 hash = (hash * FNV32_PRIME) ^ c;
  49         }
  50         return hash;
  51 }
  52
  53 /*
  54  * Incoporate another chunk of data into a memihash
  55  * computation.
  56  */
  57 unsigned int memihash_cont(unsigned int hash_seed, const void *buf, size_t len)
  58 {
  59         unsigned int hash = hash_seed;
  60         unsigned char *ucbuf = (unsigned char *) buf;
  61         while (len--) {
  62                 unsigned int c = *ucbuf++;
  63                 if (c >= 'a' && c <= 'z')
  64                         c -= 'a' - 'A';
  65                 hash = (hash * FNV32_PRIME) ^ c;
  66         }
  67         return hash;
  68 }
  69
  70 #define HASHMAP_INITIAL_SIZE 64
  71 /* grow / shrink by 2^2 */
  72 #define HASHMAP_RESIZE_BITS 2
  73 /* load factor in percent */
  74 #define HASHMAP_LOAD_FACTOR 80
  75
  76 static void alloc_table(struct hashmap *map, unsigned int size)
  77 {
  78         map->tablesize = size;
  79         map->table = xcalloc(size, sizeof(struct hashmap_entry *));
  80
  81         /* calculate resize thresholds for new size */
  82         map->grow_at = (unsigned int) ((uint64_t) size * HASHMAP_LOAD_FACTOR / 100);
  83         if (size <= HASHMAP_INITIAL_SIZE)
  84                 map->shrink_at = 0;
  85         else
  86                 /*
  87                  * The shrink-threshold must be slightly smaller than
  88                  * (grow-threshold / resize-factor) to prevent erratic resizing,
  89                  * thus we divide by (resize-factor + 1).
  90                  */
  91                 map->shrink_at = map->grow_at / ((1 << HASHMAP_RESIZE_BITS) + 1);
  92 }
  93
  94 static inline int entry_equals(const struct hashmap *map,
  95                 const struct hashmap_entry *e1, const struct hashmap_entry *e2,
  96                 const void *keydata)
  97 {
  98         return (e1 == e2) ||
  99                (e1->hash == e2->hash &&
 100                 !map->cmpfn(map->cmpfn_data, e1, e2, keydata));
 101 }
 102
 103 static inline unsigned int bucket(const struct hashmap *map,
 104                 const struct hashmap_entry *key)
 105 {
 106         return key->hash & (map->tablesize - 1);
 107 }
 108
 109 int hashmap_bucket(const struct hashmap *map, unsigned int hash)
 110 {
 111         return hash & (map->tablesize - 1);
 112 }
 113
 114 static void rehash(struct hashmap *map, unsigned int newsize)
 115 {
 116         unsigned int i, oldsize = map->tablesize;
 117         struct hashmap_entry **oldtable = map->table;
 118
 119         alloc_table(map, newsize);
 120         for (i = 0; i < oldsize; i++) {
 121                 struct hashmap_entry *e = oldtable[i];
 122                 while (e) {
 123                         struct hashmap_entry *next = e->next;
 124                         unsigned int b = bucket(map, e);
 125                         e->next = map->table[b];
 126                         map->table[b] = e;
 127                         e = next;
 128                 }
 129         }
 130         free(oldtable);
 131 }
 132
 133 static inline struct hashmap_entry **find_entry_ptr(const struct hashmap *map,
 134                 const struct hashmap_entry *key, const void *keydata)
 135 {
 136         struct hashmap_entry **e = &map->table[bucket(map, key)];
 137         while (*e && !entry_equals(map, *e, key, keydata))
 138                 e = &(*e)->next;
 139         return e;
 140 }
 141
 142 static int always_equal(const void *unused_cmp_data,
 143                         const void *unused1,
 144                         const void *unused2,
 145                         const void *unused_keydata)
 146 {
 147         return 0;
 148 }
 149
 150 void hashmap_init(struct hashmap *map, hashmap_cmp_fn equals_function,
 151                 const void *cmpfn_data, size_t initial_size)
 152 {
 153         unsigned int size = HASHMAP_INITIAL_SIZE;
 154
 155         memset(map, 0, sizeof(*map));
 156
 157         map->cmpfn = equals_function ? equals_function : always_equal;
 158         map->cmpfn_data = cmpfn_data;
 159
 160         /* calculate initial table size and allocate the table */
 161         initial_size = (unsigned int) ((uint64_t) initial_size * 100
 162                         / HASHMAP_LOAD_FACTOR);
 163         while (initial_size > size)
 164                 size <<= HASHMAP_RESIZE_BITS;
 165         alloc_table(map, size);
 166
 167         /*
 168          * Keep track of the number of items in the map and
 169          * allow the map to automatically grow as necessary.
 170          */
 171         map->do_count_items = 1;
 172 }
 173
 174 void hashmap_free(struct hashmap *map, int free_entries)
 175 {
 176         if (!map || !map->table)
 177                 return;
 178         if (free_entries) {
 179                 struct hashmap_iter iter;
 180                 struct hashmap_entry *e;
 181                 hashmap_iter_init(map, &iter);
 182                 while ((e = hashmap_iter_next(&iter)))
 183                         free(e);
 184         }
 185         free(map->table);
 186         memset(map, 0, sizeof(*map));
 187 }
 188
 189 void *hashmap_get(const struct hashmap *map, const struct hashmap_entry *key,
 190                 const void *keydata)
 191 {
 192         return *find_entry_ptr(map, key, keydata);
 193 }
 194
 195 void *hashmap_get_next(const struct hashmap *map,
 196                         const struct hashmap_entry *entry)
 197 {
 198         struct hashmap_entry *e = entry->next;
 199         for (; e; e = e->next)
 200                 if (entry_equals(map, entry, e, NULL))
 201                         return e;
 202         return NULL;
 203 }
 204
 205 void hashmap_add(struct hashmap *map, struct hashmap_entry *entry)
 206 {
 207         unsigned int b = bucket(map, entry);
 208
 209         /* add entry */
 210         entry->next = map->table[b];
 211         map->table[b] = entry;
 212
 213         /* fix size and rehash if appropriate */
 214         if (map->do_count_items) {
 215                 map->private_size++;
 216                 if (map->private_size > map->grow_at)
 217                         rehash(map, map->tablesize << HASHMAP_RESIZE_BITS);
 218         }
 219 }
 220
 221 void *hashmap_remove(struct hashmap *map, const struct hashmap_entry *key,
 222                 const void *keydata)
 223 {
 224         struct hashmap_entry *old;
 225         struct hashmap_entry **e = find_entry_ptr(map, key, keydata);
 226         if (!*e)
 227                 return NULL;
 228
 229         /* remove existing entry */
 230         old = *e;
 231         *e = old->next;
 232         old->next = NULL;
 233
 234         /* fix size and rehash if appropriate */
 235         if (map->do_count_items) {
 236                 map->private_size--;
 237                 if (map->private_size < map->shrink_at)
 238                         rehash(map, map->tablesize >> HASHMAP_RESIZE_BITS);
 239         }
 240
 241         return old;
 242 }
 243
 244 void *hashmap_put(struct hashmap *map, void *entry)
 245 {
 246         struct hashmap_entry *old = hashmap_remove(map, entry, NULL);
 247         hashmap_add(map, entry);
 248         return old;
 249 }
 250
 251 void hashmap_iter_init(struct hashmap *map, struct hashmap_iter *iter)
 252 {
 253         iter->map = map;
 254         iter->tablepos = 0;
 255         iter->next = NULL;
 256 }
 257
 258 void *hashmap_iter_next(struct hashmap_iter *iter)
 259 {
 260         struct hashmap_entry *current = iter->next;
 261         for (;;) {
 262                 if (current) {
 263                         iter->next = current->next;
 264                         return current;
 265                 }
 266
 267                 if (iter->tablepos >= iter->map->tablesize)
 268                         return NULL;
 269
 270                 current = iter->map->table[iter->tablepos++];
 271         }
 272 }
 273
 274 struct pool_entry {
 275         struct hashmap_entry ent;
 276         size_t len;
 277         unsigned char data[FLEX_ARRAY];
 278 };
 279
 280 static int pool_entry_cmp(const void *unused_cmp_data,
 281                           const struct pool_entry *e1,
 282                           const struct pool_entry *e2,
 283                           const unsigned char *keydata)
 284 {
 285         return e1->data != keydata &&
 286                (e1->len != e2->len || memcmp(e1->data, keydata, e1->len));
 287 }
 288
 289 const void *memintern(const void *data, size_t len)
 290 {
 291         static struct hashmap map;
 292         struct pool_entry key, *e;
 293
 294         /* initialize string pool hashmap */
 295         if (!map.tablesize)
 296                 hashmap_init(&map, (hashmap_cmp_fn) pool_entry_cmp, NULL, 0);
 297
 298         /* lookup interned string in pool */
 299         hashmap_entry_init(&key.ent, memhash(data, len));
 300         key.len = len;
 301         e = hashmap_get(&map, &key.ent, data);
 302         if (!e) {
 303                 /* not found: create it */
 304                 FLEX_ALLOC_MEM(e, data, data, len);
 305                 hashmap_entry_init(&e->ent, key.ent.hash);
 306                 e->len = len;
 307                 hashmap_add(&map, &e->ent);
 308         }
 309         return e->data;
 310 }