2 This file is part of GNUnet.
3 (C) 2001, 2002, 2003, 2004, 2006, 2008 Christian Grothoff (and other contributing authors)
5 GNUnet is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; either version 2, or (at your
8 option) any later version.
10 GNUnet is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with GNUnet; see the file COPYING. If not, write to the
17 Free Software Foundation, Inc., 59 Temple Place - Suite 330,
18 Boston, MA 02111-1307, USA.
21 * @file util/container_bloomfilter.c
22 * @brief data structure used to reduce disk accesses.
24 * The idea basically: Create a signature for each element in the
25 * database. Add those signatures to a bit array. When doing a lookup,
26 * check if the bit array matches the signature of the requested
27 * element. If yes, address the disk, otherwise return 'not found'.
29 * A property of the bloom filter is that sometimes we will have
30 * a match even if the element is not on the disk (then we do
31 * an unnecessary disk access), but what's most important is that
32 * we never get a single "false negative".
34 * To be able to delete entries from the bloom filter, we maintain
35 * a 4 bit counter in the file on the drive (we still use only one
38 * @author Igor Wronsky
39 * @author Christian Grothoff
43 #include "gnunet_common.h"
44 #include "gnunet_container_lib.h"
45 #include "gnunet_disk_lib.h"
47 #define LOG(kind,...) GNUNET_log_from (kind, "util", __VA_ARGS__)
49 #define LOG_STRERROR(kind,syscall) GNUNET_log_from_strerror (kind, "util", syscall)
51 #define LOG_STRERROR_FILE(kind,syscall,filename) GNUNET_log_from_strerror_file (kind, "util", syscall, filename)
53 struct GNUNET_CONTAINER_BloomFilter
57 * The actual bloomfilter bit array
62 * Filename of the filter
67 * The bit counter file on disk
69 struct GNUNET_DISK_FileHandle *fh;
72 * How many bits we set for each stored element
74 unsigned int addressesPerElement;
77 * Size of bitArray in bytes
86 * Get size of the bloom filter.
88 * @param bf the filter
89 * @return number of bytes used for the data of the bloom filter
92 GNUNET_CONTAINER_bloomfilter_get_size (const struct GNUNET_CONTAINER_BloomFilter
97 return bf->bitArraySize;
102 * Copy an existing memory. Any association with a file
103 * on-disk will be lost in the process.
104 * @param bf the filter to copy
105 * @return copy of the bf
107 struct GNUNET_CONTAINER_BloomFilter *
108 GNUNET_CONTAINER_bloomfilter_copy (const struct GNUNET_CONTAINER_BloomFilter
111 return GNUNET_CONTAINER_bloomfilter_init (bf->bitArray, bf->bitArraySize,
112 bf->addressesPerElement);
117 * Sets a bit active in the bitArray. Increment bit-specific
118 * usage counter on disk only if below 4bit max (==15).
120 * @param bitArray memory area to set the bit in
121 * @param bitIdx which bit to set
124 setBit (char *bitArray, unsigned int bitIdx)
127 unsigned int targetBit;
129 arraySlot = bitIdx / 8;
130 targetBit = (1L << (bitIdx % 8));
131 bitArray[arraySlot] |= targetBit;
135 * Clears a bit from bitArray. Bit is cleared from the array
136 * only if the respective usage counter on the disk hits/is zero.
138 * @param bitArray memory area to set the bit in
139 * @param bitIdx which bit to unset
142 clearBit (char *bitArray, unsigned int bitIdx)
145 unsigned int targetBit;
148 targetBit = (1L << (bitIdx % 8));
149 bitArray[slot] = bitArray[slot] & (~targetBit);
153 * Checks if a bit is active in the bitArray
155 * @param bitArray memory area to set the bit in
156 * @param bitIdx which bit to test
157 * @return GNUNET_YES if the bit is set, GNUNET_NO if not.
160 testBit (char *bitArray, unsigned int bitIdx)
163 unsigned int targetBit;
166 targetBit = (1L << (bitIdx % 8));
167 if (bitArray[slot] & targetBit)
174 * Sets a bit active in the bitArray and increments
175 * bit-specific usage counter on disk (but only if
176 * the counter was below 4 bit max (==15)).
178 * @param bitArray memory area to set the bit in
179 * @param bitIdx which bit to test
180 * @param fh A file to keep the 4 bit address usage counters in
183 incrementBit (char *bitArray, unsigned int bitIdx,
184 const struct GNUNET_DISK_FileHandle *fh)
190 unsigned int targetLoc;
192 setBit (bitArray, bitIdx);
193 if (GNUNET_DISK_handle_invalid (fh))
195 /* Update the counter file on disk */
196 fileSlot = bitIdx / 2;
197 targetLoc = bitIdx % 2;
199 GNUNET_assert (fileSlot ==
200 GNUNET_DISK_file_seek (fh, fileSlot, GNUNET_DISK_SEEK_SET));
201 if (1 != GNUNET_DISK_file_read (fh, &value, 1))
204 high = (value & (~0xF)) >> 4;
216 value = ((high << 4) | low);
217 GNUNET_assert (fileSlot ==
218 GNUNET_DISK_file_seek (fh, fileSlot, GNUNET_DISK_SEEK_SET));
219 GNUNET_assert (1 == GNUNET_DISK_file_write (fh, &value, 1));
223 * Clears a bit from bitArray if the respective usage
224 * counter on the disk hits/is zero.
226 * @param bitArray memory area to set the bit in
227 * @param bitIdx which bit to test
228 * @param fh A file to keep the 4bit address usage counters in
231 decrementBit (char *bitArray, unsigned int bitIdx,
232 const struct GNUNET_DISK_FileHandle *fh)
238 unsigned int targetLoc;
240 if (GNUNET_DISK_handle_invalid (fh))
241 return; /* cannot decrement! */
242 /* Each char slot in the counter file holds two 4 bit counters */
243 fileSlot = bitIdx / 2;
244 targetLoc = bitIdx % 2;
245 GNUNET_DISK_file_seek (fh, fileSlot, GNUNET_DISK_SEEK_SET);
246 if (1 != GNUNET_DISK_file_read (fh, &value, 1))
249 high = (value & 0xF0) >> 4;
251 /* decrement, but once we have reached the max, never go back! */
254 if ((low > 0) && (low < 0xF))
258 clearBit (bitArray, bitIdx);
263 if ((high > 0) && (high < 0xF))
267 clearBit (bitArray, bitIdx);
270 value = ((high << 4) | low);
271 GNUNET_DISK_file_seek (fh, fileSlot, GNUNET_DISK_SEEK_SET);
272 GNUNET_assert (1 == GNUNET_DISK_file_write (fh, &value, 1));
275 #define BUFFSIZE 65536
278 * Creates a file filled with zeroes
280 * @param fh the file handle
281 * @param size the size of the file
282 * @return GNUNET_OK if created ok, GNUNET_SYSERR otherwise
285 makeEmptyFile (const struct GNUNET_DISK_FileHandle *fh, size_t size)
288 size_t bytesleft = size;
291 if (GNUNET_DISK_handle_invalid (fh))
292 return GNUNET_SYSERR;
293 buffer = GNUNET_malloc (BUFFSIZE);
294 memset (buffer, 0, BUFFSIZE);
295 GNUNET_DISK_file_seek (fh, 0, GNUNET_DISK_SEEK_SET);
297 while (bytesleft > 0)
299 if (bytesleft > BUFFSIZE)
301 res = GNUNET_DISK_file_write (fh, buffer, BUFFSIZE);
302 bytesleft -= BUFFSIZE;
306 res = GNUNET_DISK_file_write (fh, buffer, bytesleft);
309 GNUNET_assert (res != GNUNET_SYSERR);
311 GNUNET_free (buffer);
315 /* ************** GNUNET_CONTAINER_BloomFilter iterator ********* */
318 * Iterator (callback) method to be called by the
319 * bloomfilter iterator on each bit that is to be
320 * set or tested for the key.
323 * @param bf the filter to manipulate
324 * @param bit the current bit
325 * @return GNUNET_YES to continue, GNUNET_NO to stop early
327 typedef int (*BitIterator) (void *cls,
328 const struct GNUNET_CONTAINER_BloomFilter * bf,
333 * Call an iterator for each bit that the bloomfilter
334 * must test or set for this element.
336 * @param bf the filter
337 * @param callback the method to call
338 * @param arg extra argument to callback
339 * @param key the key for which we iterate over the BF bits
342 iterateBits (const struct GNUNET_CONTAINER_BloomFilter *bf,
343 BitIterator callback, void *arg, const GNUNET_HashCode * key)
345 GNUNET_HashCode tmp[2];
348 unsigned int slot = 0;
350 bitCount = bf->addressesPerElement;
355 while (slot < (sizeof (GNUNET_HashCode) / sizeof (uint32_t)))
359 (((uint32_t *) & tmp[round & 1])[slot]) &
360 ((bf->bitArraySize * 8) - 1)))
369 GNUNET_CRYPTO_hash (&tmp[round & 1], sizeof (GNUNET_HashCode),
370 &tmp[(round + 1) & 1]);
379 * Callback: increment bit
381 * @param cls pointer to writeable form of bf
382 * @param bf the filter to manipulate
383 * @param bit the bit to increment
387 incrementBitCallback (void *cls, const struct GNUNET_CONTAINER_BloomFilter *bf,
390 struct GNUNET_CONTAINER_BloomFilter *b = cls;
392 incrementBit (b->bitArray, bit, bf->fh);
398 * Callback: decrement bit
400 * @param cls pointer to writeable form of bf
401 * @param bf the filter to manipulate
402 * @param bit the bit to decrement
406 decrementBitCallback (void *cls, const struct GNUNET_CONTAINER_BloomFilter *bf,
409 struct GNUNET_CONTAINER_BloomFilter *b = cls;
411 decrementBit (b->bitArray, bit, bf->fh);
417 * Callback: test if all bits are set
419 * @param cls pointer set to GNUNET_NO if bit is not set
420 * @param bf the filter
421 * @param bit the bit to test
422 * @return YES if the bit is set, NO if not
425 testBitCallback (void *cls, const struct GNUNET_CONTAINER_BloomFilter *bf,
430 if (GNUNET_NO == testBit (bf->bitArray, bit))
438 /* *********************** INTERFACE **************** */
441 * Load a bloom-filter from a file.
443 * @param filename the name of the file (or the prefix)
444 * @param size the size of the bloom-filter (number of
445 * bytes of storage space to use)
446 * @param k the number of GNUNET_CRYPTO_hash-functions to apply per
447 * element (number of bits set per element in the set)
448 * @return the bloomfilter
450 struct GNUNET_CONTAINER_BloomFilter *
451 GNUNET_CONTAINER_bloomfilter_load (const char *filename, size_t size,
454 struct GNUNET_CONTAINER_BloomFilter *bf;
460 GNUNET_assert (NULL != filename);
461 if ((k == 0) || (size == 0))
466 while ( (ui < size) &&
469 size = ui; /* make sure it's a power of 2 */
471 bf = GNUNET_malloc (sizeof (struct GNUNET_CONTAINER_BloomFilter));
472 /* Try to open a bloomfilter file */
474 GNUNET_DISK_file_open (filename,
475 GNUNET_DISK_OPEN_READWRITE |
476 GNUNET_DISK_OPEN_CREATE,
477 GNUNET_DISK_PERM_USER_READ |
478 GNUNET_DISK_PERM_USER_WRITE);
484 bf->filename = GNUNET_strdup (filename);
486 bf->bitArray = GNUNET_malloc_large (size);
487 if (bf->bitArray == NULL)
490 GNUNET_DISK_file_close (bf->fh);
491 GNUNET_free (bf->filename);
495 bf->bitArraySize = size;
496 bf->addressesPerElement = k;
497 memset (bf->bitArray, 0, bf->bitArraySize);
499 /* Read from the file what bits we can */
500 rbuff = GNUNET_malloc (BUFFSIZE);
502 while (pos < size * 8)
506 res = GNUNET_DISK_file_read (bf->fh, rbuff, BUFFSIZE);
509 LOG_STRERROR_FILE (GNUNET_ERROR_TYPE_WARNING, "read", bf->filename);
512 break; /* is ok! we just did not use that many bits yet */
513 for (i = 0; i < res; i++)
515 if ((rbuff[i] & 0x0F) != 0)
516 setBit (bf->bitArray, pos + i * 2);
517 if ((rbuff[i] & 0xF0) != 0)
518 setBit (bf->bitArray, pos + i * 2 + 1);
522 pos += BUFFSIZE * 2; /* 2 bits per byte in the buffer */
530 * Create a bloom filter from raw bits.
532 * @param data the raw bits in memory (maybe NULL,
533 * in which case all bits should be considered
535 * @param size the size of the bloom-filter (number of
536 * bytes of storage space to use); also size of data
537 * -- unless data is NULL
538 * @param k the number of GNUNET_CRYPTO_hash-functions to apply per
539 * element (number of bits set per element in the set)
540 * @return the bloomfilter
542 struct GNUNET_CONTAINER_BloomFilter *
543 GNUNET_CONTAINER_bloomfilter_init (const char *data, size_t size,
546 struct GNUNET_CONTAINER_BloomFilter *bf;
549 if ((k == 0) || (size == 0))
559 bf = GNUNET_malloc (sizeof (struct GNUNET_CONTAINER_BloomFilter));
562 bf->bitArray = GNUNET_malloc_large (size);
563 if (bf->bitArray == NULL)
568 bf->bitArraySize = size;
569 bf->addressesPerElement = k;
571 memcpy (bf->bitArray, data, size);
573 memset (bf->bitArray, 0, bf->bitArraySize);
579 * Copy the raw data of this bloomfilter into
580 * the given data array.
582 * @param bf bloomfilter to take the raw data from
583 * @param data where to write the data
584 * @param size the size of the given data array
585 * @return GNUNET_SYSERR if the data array is not big enough
588 GNUNET_CONTAINER_bloomfilter_get_raw_data (const struct
589 GNUNET_CONTAINER_BloomFilter *bf,
590 char *data, size_t size)
593 return GNUNET_SYSERR;
594 if (bf->bitArraySize != size)
595 return GNUNET_SYSERR;
596 memcpy (data, bf->bitArray, size);
602 * Free the space associated with a filter
603 * in memory, flush to drive if needed (do not
604 * free the space on the drive)
606 * @param bf the filter
609 GNUNET_CONTAINER_bloomfilter_free (struct GNUNET_CONTAINER_BloomFilter *bf)
614 GNUNET_DISK_file_close (bf->fh);
615 GNUNET_free_non_null (bf->filename);
616 GNUNET_free (bf->bitArray);
622 * Reset a bloom filter to empty. Clears the file on disk.
624 * @param bf the filter
627 GNUNET_CONTAINER_bloomfilter_clear (struct GNUNET_CONTAINER_BloomFilter *bf)
632 memset (bf->bitArray, 0, bf->bitArraySize);
633 if (bf->filename != NULL)
634 makeEmptyFile (bf->fh, bf->bitArraySize * 4);
639 * Test if an element is in the filter.
641 * @param e the element
642 * @param bf the filter
643 * @return GNUNET_YES if the element is in the filter, GNUNET_NO if not
646 GNUNET_CONTAINER_bloomfilter_test (const struct GNUNET_CONTAINER_BloomFilter
647 *bf, const GNUNET_HashCode * e)
654 iterateBits (bf, &testBitCallback, &res, e);
660 * Add an element to the filter
662 * @param bf the filter
663 * @param e the element
666 GNUNET_CONTAINER_bloomfilter_add (struct GNUNET_CONTAINER_BloomFilter *bf,
667 const GNUNET_HashCode * e)
671 iterateBits (bf, &incrementBitCallback, bf, e);
676 * Or the entries of the given raw data array with the
677 * data of the given bloom filter. Assumes that
678 * the size of the data array and the current filter
681 * @param bf the filter
682 * @param data the data to or-in
683 * @param size number of bytes in data
686 GNUNET_CONTAINER_bloomfilter_or (struct GNUNET_CONTAINER_BloomFilter *bf,
687 const char *data, size_t size)
691 unsigned long long *fc;
692 const unsigned long long *dc;
696 if (bf->bitArraySize != size)
697 return GNUNET_SYSERR;
698 fc = (unsigned long long *) bf->bitArray;
699 dc = (const unsigned long long *) data;
700 n = size / sizeof (unsigned long long);
702 for (i = 0; i < n; i++)
704 for (i = n * sizeof (unsigned long long); i < size; i++)
705 bf->bitArray[i] |= data[i];
710 * Or the entries of the given raw data array with the
711 * data of the given bloom filter. Assumes that
712 * the size of the data array and the current filter
715 * @param bf the filter
716 * @param to_or the bloomfilter to or-in
717 * @param size number of bytes in data
720 GNUNET_CONTAINER_bloomfilter_or2 (struct GNUNET_CONTAINER_BloomFilter *bf,
721 const struct GNUNET_CONTAINER_BloomFilter
726 unsigned long long *fc;
727 const unsigned long long *dc;
731 if (bf->bitArraySize != size)
732 return GNUNET_SYSERR;
733 fc = (unsigned long long *) bf->bitArray;
734 dc = (const unsigned long long *) to_or->bitArray;
735 n = size / sizeof (unsigned long long);
737 for (i = 0; i < n; i++)
739 for (i = n * sizeof (unsigned long long); i < size; i++)
740 bf->bitArray[i] |= to_or->bitArray[i];
745 * Remove an element from the filter.
747 * @param bf the filter
748 * @param e the element to remove
751 GNUNET_CONTAINER_bloomfilter_remove (struct GNUNET_CONTAINER_BloomFilter *bf,
752 const GNUNET_HashCode * e)
756 if (bf->filename == NULL)
758 iterateBits (bf, &decrementBitCallback, bf, e);
762 * Resize a bloom filter. Note that this operation
763 * is pretty costly. Essentially, the bloom filter
764 * needs to be completely re-build.
766 * @param bf the filter
767 * @param iterator an iterator over all elements stored in the BF
768 * @param iterator_cls argument to the iterator function
769 * @param size the new size for the filter
770 * @param k the new number of GNUNET_CRYPTO_hash-function to apply per element
773 GNUNET_CONTAINER_bloomfilter_resize (struct GNUNET_CONTAINER_BloomFilter *bf,
774 GNUNET_HashCodeIterator iterator,
775 void *iterator_cls, size_t size,
781 GNUNET_free (bf->bitArray);
785 size = i; /* make sure it's a power of 2 */
787 bf->bitArraySize = size;
788 bf->bitArray = GNUNET_malloc (size);
789 memset (bf->bitArray, 0, bf->bitArraySize);
790 if (bf->filename != NULL)
791 makeEmptyFile (bf->fh, bf->bitArraySize * 4);
792 while (GNUNET_YES == iterator (iterator_cls, &hc))
793 GNUNET_CONTAINER_bloomfilter_add (bf, &hc);
796 /* end of container_bloomfilter.c */