1 /* hash - hashing table processing.
3 Copyright (C) 1998-2004, 2006-2007, 2009-2010 Free Software Foundation, Inc.
5 Written by Jim Meyering, 1992.
7 This program is free software: you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
20 /* A generic hash table package. */
22 /* Define USE_OBSTACK to 1 if you want the allocator to use obstacks instead
23 of malloc. If you change USE_OBSTACK, you have to recompile! */
29 #include "bitrotate.h"
38 # ifndef obstack_chunk_alloc
39 # define obstack_chunk_alloc malloc
41 # ifndef obstack_chunk_free
42 # define obstack_chunk_free free
49 struct hash_entry *next;
54 /* The array of buckets starts at BUCKET and extends to BUCKET_LIMIT-1,
55 for a possibility of N_BUCKETS. Among those, N_BUCKETS_USED buckets
56 are not empty, there are N_ENTRIES active entries in the table. */
57 struct hash_entry *bucket;
58 struct hash_entry const *bucket_limit;
60 size_t n_buckets_used;
63 /* Tuning arguments, kept in a physically separate structure. */
64 const Hash_tuning *tuning;
66 /* Three functions are given to `hash_initialize', see the documentation
67 block for this function. In a word, HASHER randomizes a user entry
68 into a number up from 0 up to some maximum minus 1; COMPARATOR returns
69 true if two user entries compare equally; and DATA_FREER is the cleanup
70 function for a user entry. */
72 Hash_comparator comparator;
73 Hash_data_freer data_freer;
75 /* A linked list of freed struct hash_entry structs. */
76 struct hash_entry *free_entry_list;
79 /* Whenever obstacks are used, it is possible to allocate all overflowed
80 entries into a single stack, so they all can be freed in a single
81 operation. It is not clear if the speedup is worth the trouble. */
82 struct obstack entry_stack;
86 /* A hash table contains many internal entries, each holding a pointer to
87 some user-provided data (also called a user entry). An entry indistinctly
88 refers to both the internal entry and its associated user entry. A user
89 entry contents may be hashed by a randomization function (the hashing
90 function, or just `hasher' for short) into a number (or `slot') between 0
91 and the current table size. At each slot position in the hash table,
92 starts a linked chain of entries for which the user data all hash to this
93 slot. A bucket is the collection of all entries hashing to the same slot.
95 A good `hasher' function will distribute entries rather evenly in buckets.
96 In the ideal case, the length of each bucket is roughly the number of
97 entries divided by the table size. Finding the slot for a data is usually
98 done in constant time by the `hasher', and the later finding of a precise
99 entry is linear in time with the size of the bucket. Consequently, a
100 larger hash table size (that is, a larger number of buckets) is prone to
101 yielding shorter chains, *given* the `hasher' function behaves properly.
103 Long buckets slow down the lookup algorithm. One might use big hash table
104 sizes in hope to reduce the average length of buckets, but this might
105 become inordinate, as unused slots in the hash table take some space. The
106 best bet is to make sure you are using a good `hasher' function (beware
107 that those are not that easy to write! :-), and to use a table size
108 larger than the actual number of entries. */
110 /* If an insertion makes the ratio of nonempty buckets to table size larger
111 than the growth threshold (a number between 0.0 and 1.0), then increase
112 the table size by multiplying by the growth factor (a number greater than
113 1.0). The growth threshold defaults to 0.8, and the growth factor
114 defaults to 1.414, meaning that the table will have doubled its size
115 every second time 80% of the buckets get used. */
116 #define DEFAULT_GROWTH_THRESHOLD 0.8
117 #define DEFAULT_GROWTH_FACTOR 1.414
119 /* If a deletion empties a bucket and causes the ratio of used buckets to
120 table size to become smaller than the shrink threshold (a number between
121 0.0 and 1.0), then shrink the table by multiplying by the shrink factor (a
122 number greater than the shrink threshold but smaller than 1.0). The shrink
123 threshold and factor default to 0.0 and 1.0, meaning that the table never
125 #define DEFAULT_SHRINK_THRESHOLD 0.0
126 #define DEFAULT_SHRINK_FACTOR 1.0
128 /* Use this to initialize or reset a TUNING structure to
129 some sensible values. */
130 static const Hash_tuning default_tuning =
132 DEFAULT_SHRINK_THRESHOLD,
133 DEFAULT_SHRINK_FACTOR,
134 DEFAULT_GROWTH_THRESHOLD,
135 DEFAULT_GROWTH_FACTOR,
139 /* Information and lookup. */
141 /* The following few functions provide information about the overall hash
142 table organization: the number of entries, number of buckets and maximum
143 length of buckets. */
145 /* Return the number of buckets in the hash table. The table size, the total
146 number of buckets (used plus unused), or the maximum number of slots, are
147 the same quantity. */
150 hash_get_n_buckets (const Hash_table *table)
152 return table->n_buckets;
155 /* Return the number of slots in use (non-empty buckets). */
158 hash_get_n_buckets_used (const Hash_table *table)
160 return table->n_buckets_used;
163 /* Return the number of active entries. */
166 hash_get_n_entries (const Hash_table *table)
168 return table->n_entries;
171 /* Return the length of the longest chain (bucket). */
174 hash_get_max_bucket_length (const Hash_table *table)
176 struct hash_entry const *bucket;
177 size_t max_bucket_length = 0;
179 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
183 struct hash_entry const *cursor = bucket;
184 size_t bucket_length = 1;
186 while (cursor = cursor->next, cursor)
189 if (bucket_length > max_bucket_length)
190 max_bucket_length = bucket_length;
194 return max_bucket_length;
197 /* Do a mild validation of a hash table, by traversing it and checking two
201 hash_table_ok (const Hash_table *table)
203 struct hash_entry const *bucket;
204 size_t n_buckets_used = 0;
205 size_t n_entries = 0;
207 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
211 struct hash_entry const *cursor = bucket;
213 /* Count bucket head. */
217 /* Count bucket overflow. */
218 while (cursor = cursor->next, cursor)
223 if (n_buckets_used == table->n_buckets_used && n_entries == table->n_entries)
230 hash_print_statistics (const Hash_table *table, FILE *stream)
232 size_t n_entries = hash_get_n_entries (table);
233 size_t n_buckets = hash_get_n_buckets (table);
234 size_t n_buckets_used = hash_get_n_buckets_used (table);
235 size_t max_bucket_length = hash_get_max_bucket_length (table);
237 fprintf (stream, "# entries: %lu\n", (unsigned long int) n_entries);
238 fprintf (stream, "# buckets: %lu\n", (unsigned long int) n_buckets);
239 fprintf (stream, "# buckets used: %lu (%.2f%%)\n",
240 (unsigned long int) n_buckets_used,
241 (100.0 * n_buckets_used) / n_buckets);
242 fprintf (stream, "max bucket length: %lu\n",
243 (unsigned long int) max_bucket_length);
246 /* If ENTRY matches an entry already in the hash table, return the
247 entry from the table. Otherwise, return NULL. */
250 hash_lookup (const Hash_table *table, const void *entry)
252 struct hash_entry const *bucket
253 = table->bucket + table->hasher (entry, table->n_buckets);
254 struct hash_entry const *cursor;
256 if (! (bucket < table->bucket_limit))
259 if (bucket->data == NULL)
262 for (cursor = bucket; cursor; cursor = cursor->next)
263 if (entry == cursor->data || table->comparator (entry, cursor->data))
271 /* The functions in this page traverse the hash table and process the
272 contained entries. For the traversal to work properly, the hash table
273 should not be resized nor modified while any particular entry is being
274 processed. In particular, entries should not be added, and an entry
275 may be removed only if there is no shrink threshold and the entry being
276 removed has already been passed to hash_get_next. */
278 /* Return the first data in the table, or NULL if the table is empty. */
281 hash_get_first (const Hash_table *table)
283 struct hash_entry const *bucket;
285 if (table->n_entries == 0)
288 for (bucket = table->bucket; ; bucket++)
289 if (! (bucket < table->bucket_limit))
291 else if (bucket->data)
295 /* Return the user data for the entry following ENTRY, where ENTRY has been
296 returned by a previous call to either `hash_get_first' or `hash_get_next'.
297 Return NULL if there are no more entries. */
300 hash_get_next (const Hash_table *table, const void *entry)
302 struct hash_entry const *bucket
303 = table->bucket + table->hasher (entry, table->n_buckets);
304 struct hash_entry const *cursor;
306 if (! (bucket < table->bucket_limit))
309 /* Find next entry in the same bucket. */
310 for (cursor = bucket; cursor; cursor = cursor->next)
311 if (cursor->data == entry && cursor->next)
312 return cursor->next->data;
314 /* Find first entry in any subsequent bucket. */
315 while (++bucket < table->bucket_limit)
323 /* Fill BUFFER with pointers to active user entries in the hash table, then
324 return the number of pointers copied. Do not copy more than BUFFER_SIZE
328 hash_get_entries (const Hash_table *table, void **buffer,
332 struct hash_entry const *bucket;
333 struct hash_entry const *cursor;
335 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
339 for (cursor = bucket; cursor; cursor = cursor->next)
341 if (counter >= buffer_size)
343 buffer[counter++] = cursor->data;
351 /* Call a PROCESSOR function for each entry of a hash table, and return the
352 number of entries for which the processor function returned success. A
353 pointer to some PROCESSOR_DATA which will be made available to each call to
354 the processor function. The PROCESSOR accepts two arguments: the first is
355 the user entry being walked into, the second is the value of PROCESSOR_DATA
356 as received. The walking continue for as long as the PROCESSOR function
357 returns nonzero. When it returns zero, the walking is interrupted. */
360 hash_do_for_each (const Hash_table *table, Hash_processor processor,
361 void *processor_data)
364 struct hash_entry const *bucket;
365 struct hash_entry const *cursor;
367 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
371 for (cursor = bucket; cursor; cursor = cursor->next)
373 if (! processor (cursor->data, processor_data))
383 /* Allocation and clean-up. */
385 /* Return a hash index for a NUL-terminated STRING between 0 and N_BUCKETS-1.
386 This is a convenience routine for constructing other hashing functions. */
390 /* About hashings, Paul Eggert writes to me (FP), on 1994-01-01: "Please see
391 B. J. McKenzie, R. Harries & T. Bell, Selecting a hashing algorithm,
392 Software--practice & experience 20, 2 (Feb 1990), 209-224. Good hash
393 algorithms tend to be domain-specific, so what's good for [diffutils'] io.c
394 may not be good for your application." */
397 hash_string (const char *string, size_t n_buckets)
399 # define HASH_ONE_CHAR(Value, Byte) \
400 ((Byte) + rotl_sz (Value, 7))
405 for (; (ch = *string); string++)
406 value = HASH_ONE_CHAR (value, ch);
407 return value % n_buckets;
409 # undef HASH_ONE_CHAR
412 #else /* not USE_DIFF_HASH */
414 /* This one comes from `recode', and performs a bit better than the above as
415 per a few experiments. It is inspired from a hashing routine found in the
416 very old Cyber `snoop', itself written in typical Greg Mansfield style.
417 (By the way, what happened to this excellent man? Is he still alive?) */
420 hash_string (const char *string, size_t n_buckets)
425 for (; (ch = *string); string++)
426 value = (value * 31 + ch) % n_buckets;
430 #endif /* not USE_DIFF_HASH */
432 /* Return true if CANDIDATE is a prime number. CANDIDATE should be an odd
433 number at least equal to 11. */
436 is_prime (size_t candidate)
439 size_t square = divisor * divisor;
441 while (square < candidate && (candidate % divisor))
444 square += 4 * divisor;
448 return (candidate % divisor ? true : false);
451 /* Round a given CANDIDATE number up to the nearest prime, and return that
452 prime. Primes lower than 10 are merely skipped. */
455 next_prime (size_t candidate)
457 /* Skip small primes. */
461 /* Make it definitely odd. */
464 while (SIZE_MAX != candidate && !is_prime (candidate))
471 hash_reset_tuning (Hash_tuning *tuning)
473 *tuning = default_tuning;
476 /* If the user passes a NULL hasher, we hash the raw pointer. */
478 raw_hasher (const void *data, size_t n)
480 /* When hashing unique pointers, it is often the case that they were
481 generated by malloc and thus have the property that the low-order
482 bits are 0. As this tends to give poorer performance with small
483 tables, we rotate the pointer value before performing division,
484 in an attempt to improve hash quality. */
485 size_t val = rotr_sz ((size_t) data, 3);
489 /* If the user passes a NULL comparator, we use pointer comparison. */
491 raw_comparator (const void *a, const void *b)
497 /* For the given hash TABLE, check the user supplied tuning structure for
498 reasonable values, and return true if there is no gross error with it.
499 Otherwise, definitively reset the TUNING field to some acceptable default
500 in the hash table (that is, the user loses the right of further modifying
501 tuning arguments), and return false. */
504 check_tuning (Hash_table *table)
506 const Hash_tuning *tuning = table->tuning;
508 if (tuning == &default_tuning)
511 /* Be a bit stricter than mathematics would require, so that
512 rounding errors in size calculations do not cause allocations to
513 fail to grow or shrink as they should. The smallest allocation
514 is 11 (due to next_prime's algorithm), so an epsilon of 0.1
515 should be good enough. */
518 if (epsilon < tuning->growth_threshold
519 && tuning->growth_threshold < 1 - epsilon
520 && 1 + epsilon < tuning->growth_factor
521 && 0 <= tuning->shrink_threshold
522 && tuning->shrink_threshold + epsilon < tuning->shrink_factor
523 && tuning->shrink_factor <= 1
524 && tuning->shrink_threshold + epsilon < tuning->growth_threshold)
527 table->tuning = &default_tuning;
531 /* Compute the size of the bucket array for the given CANDIDATE and
532 TUNING, or return 0 if there is no possible way to allocate that
536 compute_bucket_size (size_t candidate, const Hash_tuning *tuning)
538 if (!tuning->is_n_buckets)
540 float new_candidate = candidate / tuning->growth_threshold;
541 if (SIZE_MAX <= new_candidate)
543 candidate = new_candidate;
545 candidate = next_prime (candidate);
546 if (xalloc_oversized (candidate, sizeof (struct hash_entry *)))
551 /* Allocate and return a new hash table, or NULL upon failure. The initial
552 number of buckets is automatically selected so as to _guarantee_ that you
553 may insert at least CANDIDATE different user entries before any growth of
554 the hash table size occurs. So, if have a reasonably tight a-priori upper
555 bound on the number of entries you intend to insert in the hash table, you
556 may save some table memory and insertion time, by specifying it here. If
557 the IS_N_BUCKETS field of the TUNING structure is true, the CANDIDATE
558 argument has its meaning changed to the wanted number of buckets.
560 TUNING points to a structure of user-supplied values, in case some fine
561 tuning is wanted over the default behavior of the hasher. If TUNING is
562 NULL, the default tuning parameters are used instead. If TUNING is
563 provided but the values requested are out of bounds or might cause
564 rounding errors, return NULL.
566 The user-supplied HASHER function, when not NULL, accepts two
567 arguments ENTRY and TABLE_SIZE. It computes, by hashing ENTRY contents, a
568 slot number for that entry which should be in the range 0..TABLE_SIZE-1.
569 This slot number is then returned.
571 The user-supplied COMPARATOR function, when not NULL, accepts two
572 arguments pointing to user data, it then returns true for a pair of entries
573 that compare equal, or false otherwise. This function is internally called
574 on entries which are already known to hash to the same bucket index,
575 but which are distinct pointers.
577 The user-supplied DATA_FREER function, when not NULL, may be later called
578 with the user data as an argument, just before the entry containing the
579 data gets freed. This happens from within `hash_free' or `hash_clear'.
580 You should specify this function only if you want these functions to free
581 all of your `data' data. This is typically the case when your data is
582 simply an auxiliary struct that you have malloc'd to aggregate several
586 hash_initialize (size_t candidate, const Hash_tuning *tuning,
587 Hash_hasher hasher, Hash_comparator comparator,
588 Hash_data_freer data_freer)
594 if (comparator == NULL)
595 comparator = raw_comparator;
597 table = malloc (sizeof *table);
602 tuning = &default_tuning;
603 table->tuning = tuning;
604 if (!check_tuning (table))
606 /* Fail if the tuning options are invalid. This is the only occasion
607 when the user gets some feedback about it. Once the table is created,
608 if the user provides invalid tuning options, we silently revert to
609 using the defaults, and ignore further request to change the tuning
614 table->n_buckets = compute_bucket_size (candidate, tuning);
615 if (!table->n_buckets)
618 table->bucket = calloc (table->n_buckets, sizeof *table->bucket);
619 if (table->bucket == NULL)
621 table->bucket_limit = table->bucket + table->n_buckets;
622 table->n_buckets_used = 0;
623 table->n_entries = 0;
625 table->hasher = hasher;
626 table->comparator = comparator;
627 table->data_freer = data_freer;
629 table->free_entry_list = NULL;
631 obstack_init (&table->entry_stack);
640 /* Make all buckets empty, placing any chained entries on the free list.
641 Apply the user-specified function data_freer (if any) to the datas of any
645 hash_clear (Hash_table *table)
647 struct hash_entry *bucket;
649 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
653 struct hash_entry *cursor;
654 struct hash_entry *next;
656 /* Free the bucket overflow. */
657 for (cursor = bucket->next; cursor; cursor = next)
659 if (table->data_freer)
660 table->data_freer (cursor->data);
664 /* Relinking is done one entry at a time, as it is to be expected
665 that overflows are either rare or short. */
666 cursor->next = table->free_entry_list;
667 table->free_entry_list = cursor;
670 /* Free the bucket head. */
671 if (table->data_freer)
672 table->data_freer (bucket->data);
678 table->n_buckets_used = 0;
679 table->n_entries = 0;
682 /* Reclaim all storage associated with a hash table. If a data_freer
683 function has been supplied by the user when the hash table was created,
684 this function applies it to the data of each entry before freeing that
688 hash_free (Hash_table *table)
690 struct hash_entry *bucket;
691 struct hash_entry *cursor;
692 struct hash_entry *next;
694 /* Call the user data_freer function. */
695 if (table->data_freer && table->n_entries)
697 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
701 for (cursor = bucket; cursor; cursor = cursor->next)
702 table->data_freer (cursor->data);
709 obstack_free (&table->entry_stack, NULL);
713 /* Free all bucket overflowed entries. */
714 for (bucket = table->bucket; bucket < table->bucket_limit; bucket++)
716 for (cursor = bucket->next; cursor; cursor = next)
723 /* Also reclaim the internal list of previously freed entries. */
724 for (cursor = table->free_entry_list; cursor; cursor = next)
732 /* Free the remainder of the hash table structure. */
733 free (table->bucket);
737 /* Insertion and deletion. */
739 /* Get a new hash entry for a bucket overflow, possibly by recycling a
740 previously freed one. If this is not possible, allocate a new one. */
742 static struct hash_entry *
743 allocate_entry (Hash_table *table)
745 struct hash_entry *new;
747 if (table->free_entry_list)
749 new = table->free_entry_list;
750 table->free_entry_list = new->next;
755 new = obstack_alloc (&table->entry_stack, sizeof *new);
757 new = malloc (sizeof *new);
764 /* Free a hash entry which was part of some bucket overflow,
765 saving it for later recycling. */
768 free_entry (Hash_table *table, struct hash_entry *entry)
771 entry->next = table->free_entry_list;
772 table->free_entry_list = entry;
775 /* This private function is used to help with insertion and deletion. When
776 ENTRY matches an entry in the table, return a pointer to the corresponding
777 user data and set *BUCKET_HEAD to the head of the selected bucket.
778 Otherwise, return NULL. When DELETE is true and ENTRY matches an entry in
779 the table, unlink the matching entry. */
782 hash_find_entry (Hash_table *table, const void *entry,
783 struct hash_entry **bucket_head, bool delete)
785 struct hash_entry *bucket
786 = table->bucket + table->hasher (entry, table->n_buckets);
787 struct hash_entry *cursor;
789 if (! (bucket < table->bucket_limit))
792 *bucket_head = bucket;
794 /* Test for empty bucket. */
795 if (bucket->data == NULL)
798 /* See if the entry is the first in the bucket. */
799 if (entry == bucket->data || table->comparator (entry, bucket->data))
801 void *data = bucket->data;
807 struct hash_entry *next = bucket->next;
809 /* Bump the first overflow entry into the bucket head, then save
810 the previous first overflow entry for later recycling. */
812 free_entry (table, next);
823 /* Scan the bucket overflow. */
824 for (cursor = bucket; cursor->next; cursor = cursor->next)
826 if (entry == cursor->next->data
827 || table->comparator (entry, cursor->next->data))
829 void *data = cursor->next->data;
833 struct hash_entry *next = cursor->next;
835 /* Unlink the entry to delete, then save the freed entry for later
837 cursor->next = next->next;
838 free_entry (table, next);
845 /* No entry found. */
849 /* Internal helper, to move entries from SRC to DST. Both tables must
850 share the same free entry list. If SAFE, only move overflow
851 entries, saving bucket heads for later, so that no allocations will
852 occur. Return false if the free entry list is exhausted and an
856 transfer_entries (Hash_table *dst, Hash_table *src, bool safe)
858 struct hash_entry *bucket;
859 struct hash_entry *cursor;
860 struct hash_entry *next;
861 for (bucket = src->bucket; bucket < src->bucket_limit; bucket++)
865 struct hash_entry *new_bucket;
867 /* Within each bucket, transfer overflow entries first and
868 then the bucket head, to minimize memory pressure. After
869 all, the only time we might allocate is when moving the
870 bucket head, but moving overflow entries first may create
871 free entries that can be recycled by the time we finally
872 get to the bucket head. */
873 for (cursor = bucket->next; cursor; cursor = next)
876 new_bucket = (dst->bucket + dst->hasher (data, dst->n_buckets));
878 if (! (new_bucket < dst->bucket_limit))
883 if (new_bucket->data)
885 /* Merely relink an existing entry, when moving from a
886 bucket overflow into a bucket overflow. */
887 cursor->next = new_bucket->next;
888 new_bucket->next = cursor;
892 /* Free an existing entry, when moving from a bucket
893 overflow into a bucket header. */
894 new_bucket->data = data;
895 dst->n_buckets_used++;
896 free_entry (dst, cursor);
899 /* Now move the bucket head. Be sure that if we fail due to
900 allocation failure that the src table is in a consistent
906 new_bucket = (dst->bucket + dst->hasher (data, dst->n_buckets));
908 if (! (new_bucket < dst->bucket_limit))
911 if (new_bucket->data)
913 /* Allocate or recycle an entry, when moving from a bucket
914 header into a bucket overflow. */
915 struct hash_entry *new_entry = allocate_entry (dst);
917 if (new_entry == NULL)
920 new_entry->data = data;
921 new_entry->next = new_bucket->next;
922 new_bucket->next = new_entry;
926 /* Move from one bucket header to another. */
927 new_bucket->data = data;
928 dst->n_buckets_used++;
931 src->n_buckets_used--;
936 /* For an already existing hash table, change the number of buckets through
937 specifying CANDIDATE. The contents of the hash table are preserved. The
938 new number of buckets is automatically selected so as to _guarantee_ that
939 the table may receive at least CANDIDATE different user entries, including
940 those already in the table, before any other growth of the hash table size
941 occurs. If TUNING->IS_N_BUCKETS is true, then CANDIDATE specifies the
942 exact number of buckets desired. Return true iff the rehash succeeded. */
945 hash_rehash (Hash_table *table, size_t candidate)
948 Hash_table *new_table;
949 size_t new_size = compute_bucket_size (candidate, table->tuning);
953 if (new_size == table->n_buckets)
955 new_table = &storage;
956 new_table->bucket = calloc (new_size, sizeof *new_table->bucket);
957 if (new_table->bucket == NULL)
959 new_table->n_buckets = new_size;
960 new_table->bucket_limit = new_table->bucket + new_size;
961 new_table->n_buckets_used = 0;
962 new_table->n_entries = 0;
963 new_table->tuning = table->tuning;
964 new_table->hasher = table->hasher;
965 new_table->comparator = table->comparator;
966 new_table->data_freer = table->data_freer;
968 /* In order for the transfer to successfully complete, we need
969 additional overflow entries when distinct buckets in the old
970 table collide into a common bucket in the new table. The worst
971 case possible is a hasher that gives a good spread with the old
972 size, but returns a constant with the new size; if we were to
973 guarantee table->n_buckets_used-1 free entries in advance, then
974 the transfer would be guaranteed to not allocate memory.
975 However, for large tables, a guarantee of no further allocation
976 introduces a lot of extra memory pressure, all for an unlikely
977 corner case (most rehashes reduce, rather than increase, the
978 number of overflow entries needed). So, we instead ensure that
979 the transfer process can be reversed if we hit a memory
980 allocation failure mid-transfer. */
982 /* Merely reuse the extra old space into the new table. */
984 new_table->entry_stack = table->entry_stack;
986 new_table->free_entry_list = table->free_entry_list;
988 if (transfer_entries (new_table, table, false))
990 /* Entries transferred successfully; tie up the loose ends. */
991 free (table->bucket);
992 table->bucket = new_table->bucket;
993 table->bucket_limit = new_table->bucket_limit;
994 table->n_buckets = new_table->n_buckets;
995 table->n_buckets_used = new_table->n_buckets_used;
996 table->free_entry_list = new_table->free_entry_list;
997 /* table->n_entries and table->entry_stack already hold their value. */
1001 /* We've allocated new_table->bucket (and possibly some entries),
1002 exhausted the free list, and moved some but not all entries into
1003 new_table. We must undo the partial move before returning
1004 failure. The only way to get into this situation is if new_table
1005 uses fewer buckets than the old table, so we will reclaim some
1006 free entries as overflows in the new table are put back into
1007 distinct buckets in the old table.
1009 There are some pathological cases where a single pass through the
1010 table requires more intermediate overflow entries than using two
1011 passes. Two passes give worse cache performance and takes
1012 longer, but at this point, we're already out of memory, so slow
1013 and safe is better than failure. */
1014 table->free_entry_list = new_table->free_entry_list;
1015 if (! (transfer_entries (table, new_table, true)
1016 && transfer_entries (table, new_table, false)))
1018 /* table->n_entries already holds its value. */
1019 free (new_table->bucket);
1023 /* If ENTRY matches an entry already in the hash table, return the pointer
1024 to the entry from the table. Otherwise, insert ENTRY and return ENTRY.
1025 Return NULL if the storage required for insertion cannot be allocated.
1026 This implementation does not support duplicate entries or insertion of
1030 hash_insert (Hash_table *table, const void *entry)
1033 struct hash_entry *bucket;
1035 /* The caller cannot insert a NULL entry. */
1039 /* If there's a matching entry already in the table, return that. */
1040 if ((data = hash_find_entry (table, entry, &bucket, false)) != NULL)
1043 /* If the growth threshold of the buckets in use has been reached, increase
1044 the table size and rehash. There's no point in checking the number of
1045 entries: if the hashing function is ill-conditioned, rehashing is not
1046 likely to improve it. */
1048 if (table->n_buckets_used
1049 > table->tuning->growth_threshold * table->n_buckets)
1051 /* Check more fully, before starting real work. If tuning arguments
1052 became invalid, the second check will rely on proper defaults. */
1053 check_tuning (table);
1054 if (table->n_buckets_used
1055 > table->tuning->growth_threshold * table->n_buckets)
1057 const Hash_tuning *tuning = table->tuning;
1059 (tuning->is_n_buckets
1060 ? (table->n_buckets * tuning->growth_factor)
1061 : (table->n_buckets * tuning->growth_factor
1062 * tuning->growth_threshold));
1064 if (SIZE_MAX <= candidate)
1067 /* If the rehash fails, arrange to return NULL. */
1068 if (!hash_rehash (table, candidate))
1071 /* Update the bucket we are interested in. */
1072 if (hash_find_entry (table, entry, &bucket, false) != NULL)
1077 /* ENTRY is not matched, it should be inserted. */
1081 struct hash_entry *new_entry = allocate_entry (table);
1083 if (new_entry == NULL)
1086 /* Add ENTRY in the overflow of the bucket. */
1088 new_entry->data = (void *) entry;
1089 new_entry->next = bucket->next;
1090 bucket->next = new_entry;
1092 return (void *) entry;
1095 /* Add ENTRY right in the bucket head. */
1097 bucket->data = (void *) entry;
1099 table->n_buckets_used++;
1101 return (void *) entry;
1104 /* If ENTRY is already in the table, remove it and return the just-deleted
1105 data (the user may want to deallocate its storage). If ENTRY is not in the
1106 table, don't modify the table and return NULL. */
1109 hash_delete (Hash_table *table, const void *entry)
1112 struct hash_entry *bucket;
1114 data = hash_find_entry (table, entry, &bucket, true);
1121 table->n_buckets_used--;
1123 /* If the shrink threshold of the buckets in use has been reached,
1124 rehash into a smaller table. */
1126 if (table->n_buckets_used
1127 < table->tuning->shrink_threshold * table->n_buckets)
1129 /* Check more fully, before starting real work. If tuning arguments
1130 became invalid, the second check will rely on proper defaults. */
1131 check_tuning (table);
1132 if (table->n_buckets_used
1133 < table->tuning->shrink_threshold * table->n_buckets)
1135 const Hash_tuning *tuning = table->tuning;
1137 (tuning->is_n_buckets
1138 ? table->n_buckets * tuning->shrink_factor
1139 : (table->n_buckets * tuning->shrink_factor
1140 * tuning->growth_threshold));
1142 if (!hash_rehash (table, candidate))
1144 /* Failure to allocate memory in an attempt to
1145 shrink the table is not fatal. But since memory
1146 is low, we can at least be kind and free any
1147 spare entries, rather than keeping them tied up
1148 in the free entry list. */
1150 struct hash_entry *cursor = table->free_entry_list;
1151 struct hash_entry *next;
1154 next = cursor->next;
1158 table->free_entry_list = NULL;
1173 hash_print (const Hash_table *table)
1175 struct hash_entry *bucket = (struct hash_entry *) table->bucket;
1177 for ( ; bucket < table->bucket_limit; bucket++)
1179 struct hash_entry *cursor;
1182 printf ("%lu:\n", (unsigned long int) (bucket - table->bucket));
1184 for (cursor = bucket; cursor; cursor = cursor->next)
1186 char const *s = cursor->data;
1189 printf (" %s\n", s);
1194 #endif /* TESTING */