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  * Incorporate 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 struct hashmap_entry *unused1,
 144                         const struct hashmap_entry *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, ssize_t entry_offset)
 175 {
 176         if (!map || !map->table)
 177                 return;
 178         if (entry_offset >= 0) { /* called by hashmap_free_entries */
 179                 struct hashmap_iter iter;
 180                 struct hashmap_entry *e;
 181
 182                 hashmap_iter_init(map, &iter);
 183                 while ((e = hashmap_iter_next(&iter)))
 184                         /*
 185                          * like container_of, but using caller-calculated
 186                          * offset (caller being hashmap_free_entries)
 187                          */
 188                         free((char *)e - entry_offset);
 189         }
 190         free(map->table);
 191         memset(map, 0, sizeof(*map));
 192 }
 193
 194 struct hashmap_entry *hashmap_get(const struct hashmap *map,
 195                                 const struct hashmap_entry *key,
 196                                 const void *keydata)
 197 {
 198         return *find_entry_ptr(map, key, keydata);
 199 }
 200
 201 struct hashmap_entry *hashmap_get_next(const struct hashmap *map,
 202                         const struct hashmap_entry *entry)
 203 {
 204         struct hashmap_entry *e = entry->next;
 205         for (; e; e = e->next)
 206                 if (entry_equals(map, entry, e, NULL))
 207                         return e;
 208         return NULL;
 209 }
 210
 211 void hashmap_add(struct hashmap *map, struct hashmap_entry *entry)
 212 {
 213         unsigned int b = bucket(map, entry);
 214
 215         /* add entry */
 216         entry->next = map->table[b];
 217         map->table[b] = entry;
 218
 219         /* fix size and rehash if appropriate */
 220         if (map->do_count_items) {
 221                 map->private_size++;
 222                 if (map->private_size > map->grow_at)
 223                         rehash(map, map->tablesize << HASHMAP_RESIZE_BITS);
 224         }
 225 }
 226
 227 struct hashmap_entry *hashmap_remove(struct hashmap *map,
 228                                         const struct hashmap_entry *key,
 229                                         const void *keydata)
 230 {
 231         struct hashmap_entry *old;
 232         struct hashmap_entry **e = find_entry_ptr(map, key, keydata);
 233         if (!*e)
 234                 return NULL;
 235
 236         /* remove existing entry */
 237         old = *e;
 238         *e = old->next;
 239         old->next = NULL;
 240
 241         /* fix size and rehash if appropriate */
 242         if (map->do_count_items) {
 243                 map->private_size--;
 244                 if (map->private_size < map->shrink_at)
 245                         rehash(map, map->tablesize >> HASHMAP_RESIZE_BITS);
 246         }
 247
 248         return old;
 249 }
 250
 251 struct hashmap_entry *hashmap_put(struct hashmap *map,
 252                                 struct hashmap_entry *entry)
 253 {
 254         struct hashmap_entry *old = hashmap_remove(map, entry, NULL);
 255         hashmap_add(map, entry);
 256         return old;
 257 }
 258
 259 void hashmap_iter_init(struct hashmap *map, struct hashmap_iter *iter)
 260 {
 261         iter->map = map;
 262         iter->tablepos = 0;
 263         iter->next = NULL;
 264 }
 265
 266 struct hashmap_entry *hashmap_iter_next(struct hashmap_iter *iter)
 267 {
 268         struct hashmap_entry *current = iter->next;
 269         for (;;) {
 270                 if (current) {
 271                         iter->next = current->next;
 272                         return current;
 273                 }
 274
 275                 if (iter->tablepos >= iter->map->tablesize)
 276                         return NULL;
 277
 278                 current = iter->map->table[iter->tablepos++];
 279         }
 280 }
 281
 282 struct pool_entry {
 283         struct hashmap_entry ent;
 284         size_t len;
 285         unsigned char data[FLEX_ARRAY];
 286 };
 287
 288 static int pool_entry_cmp(const void *unused_cmp_data,
 289                           const struct hashmap_entry *eptr,
 290                           const struct hashmap_entry *entry_or_key,
 291                           const void *keydata)
 292 {
 293         const struct pool_entry *e1, *e2;
 294
 295         e1 = container_of(eptr, const struct pool_entry, ent);
 296         e2 = container_of(entry_or_key, const struct pool_entry, ent);
 297
 298         return e1->data != keydata &&
 299                (e1->len != e2->len || memcmp(e1->data, keydata, e1->len));
 300 }
 301
 302 const void *memintern(const void *data, size_t len)
 303 {
 304         static struct hashmap map;
 305         struct pool_entry key, *e;
 306
 307         /* initialize string pool hashmap */
 308         if (!map.tablesize)
 309                 hashmap_init(&map, pool_entry_cmp, NULL, 0);
 310
 311         /* lookup interned string in pool */
 312         hashmap_entry_init(&key.ent, memhash(data, len));
 313         key.len = len;
 314         e = hashmap_get_entry(&map, &key, ent, data);
 315         if (!e) {
 316                 /* not found: create it */
 317                 FLEX_ALLOC_MEM(e, data, data, len);
 318                 hashmap_entry_init(&e->ent, key.ent.hash);
 319                 e->len = len;
 320                 hashmap_add(&map, &e->ent);
 321         }
 322         return e->data;
 323 }