2 Unix SMB/Netbios implementation.
4 Samba database functions
5 Copyright (C) Andrew Tridgell 1999-2000
6 Copyright (C) Luke Kenneth Casson Leighton 2000
7 Copyright (C) Paul `Rusty' Russell 2000
8 Copyright (C) Jeremy Allison 2000
10 This program is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 2 of the License, or
13 (at your option) any later version.
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program; if not, write to the Free Software
22 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
44 #define TDB_MAGIC_FOOD "TDB file\n"
45 #define TDB_VERSION (0x26011967 + 6)
46 #define TDB_MAGIC (0x26011999U)
47 #define TDB_FREE_MAGIC (~TDB_MAGIC)
48 #define TDB_DEAD_MAGIC (0xFEE1DEAD)
49 #define TDB_ALIGNMENT 4
50 #define MIN_REC_SIZE (2*sizeof(struct list_struct) + TDB_ALIGNMENT)
51 #define DEFAULT_HASH_SIZE 131
52 #define TDB_PAGE_SIZE 0x2000
53 #define FREELIST_TOP (sizeof(struct tdb_header))
54 #define TDB_ALIGN(x,a) (((x) + (a)-1) & ~((a)-1))
55 #define TDB_BYTEREV(x) (((((x)&0xff)<<24)|((x)&0xFF00)<<8)|(((x)>>8)&0xFF00)|((x)>>24))
56 #define TDB_DEAD(r) ((r)->magic == TDB_DEAD_MAGIC)
57 #define TDB_BAD_MAGIC(r) ((r)->magic != TDB_MAGIC && !TDB_DEAD(r))
58 #define TDB_HASH_TOP(hash) (FREELIST_TOP + (BUCKET(hash)+1)*sizeof(tdb_off))
59 #define TDB_LOG(x) (tdb->log_fn?((tdb->log_fn x),0) : 0)
70 #define MAP_FAILED ((void *)-1)
73 #define BUCKET(hash) ((hash) % tdb->header.hash_size)
76 /* all contexts, to ensure no double-opens (fcntl locks don't nest!) */
77 static TDB_CONTEXT *tdbs = NULL;
79 static void tdb_munmap(TDB_CONTEXT *tdb)
81 if (tdb->flags & TDB_INTERNAL)
86 munmap(tdb->map_ptr, tdb->map_size);
91 static void tdb_mmap(TDB_CONTEXT *tdb)
93 if (tdb->flags & TDB_INTERNAL)
97 if (!(tdb->flags & TDB_NOMMAP)) {
98 tdb->map_ptr = mmap(NULL, tdb->map_size,
99 PROT_READ|(tdb->read_only? 0:PROT_WRITE),
100 MAP_SHARED|MAP_FILE, tdb->fd, 0);
103 * NB. When mmap fails it returns MAP_FAILED *NOT* NULL !!!!
106 if (tdb->map_ptr == MAP_FAILED) {
108 TDB_LOG((tdb, 2, "tdb_mmap failed for size %d (%s)\n",
109 tdb->map_size, strerror(errno)));
119 /* Endian conversion: we only ever deal with 4 byte quantities */
120 static void *convert(void *buf, u32 size)
123 for (i = 0; i < size / 4; i++)
124 p[i] = TDB_BYTEREV(p[i]);
127 #define DOCONV() (tdb->flags & TDB_CONVERT)
128 #define CONVERT(x) (DOCONV() ? convert(&x, sizeof(x)) : &x)
130 /* the body of the database is made of one list_struct for the free space
131 plus a separate data list for each hash value */
133 tdb_off next; /* offset of the next record in the list */
134 tdb_len rec_len; /* total byte length of record */
135 tdb_len key_len; /* byte length of key */
136 tdb_len data_len; /* byte length of data */
137 u32 full_hash; /* the full 32 bit hash of the key */
138 u32 magic; /* try to catch errors */
139 /* the following union is implied:
141 char record[rec_len];
146 u32 totalsize; (tailer)
151 /* a byte range locking function - return 0 on success
152 this functions locks/unlocks 1 byte at the specified offset */
153 static int tdb_brlock(TDB_CONTEXT *tdb, tdb_off offset,
154 int rw_type, int lck_type, int probe)
158 if (tdb->flags & TDB_NOLOCK)
164 fl.l_whence = SEEK_SET;
169 if (fcntl(tdb->fd,lck_type,&fl)) {
171 TDB_LOG((tdb, 5,"tdb_brlock failed at offset %d rw_type=%d lck_type=%d\n",
172 offset, rw_type, lck_type));
174 return TDB_ERRCODE(TDB_ERR_LOCK, -1);
179 /* lock a list in the database. list -1 is the alloc list */
180 static int tdb_lock(TDB_CONTEXT *tdb, int list, int ltype)
182 if (list < -1 || list >= (int)tdb->header.hash_size) {
183 TDB_LOG((tdb, 0,"tdb_lock: invalid list %d for ltype=%d\n",
187 if (tdb->flags & TDB_NOLOCK)
190 /* Since fcntl locks don't nest, we do a lock for the first one,
191 and simply bump the count for future ones */
192 if (tdb->locked[list+1].count == 0) {
193 if (tdb->header.rwlocks) {
194 if (tdb_spinlock(tdb, list, ltype)) {
195 TDB_LOG((tdb, 0, "tdb_lock spinlock on list ltype=%d\n",
199 } else if (tdb_brlock(tdb,FREELIST_TOP+4*list,ltype,F_SETLKW, 0)) {
200 TDB_LOG((tdb, 0,"tdb_lock failed on list %d ltype=%d (%s)\n",
201 list, ltype, strerror(errno)));
204 tdb->locked[list+1].ltype = ltype;
206 tdb->locked[list+1].count++;
210 /* unlock the database: returns void because it's too late for errors. */
211 static void tdb_unlock(TDB_CONTEXT *tdb, int list, int ltype)
213 if (tdb->flags & TDB_NOLOCK)
217 if (list < -1 || list >= (int)tdb->header.hash_size)
219 if (tdb->locked[list+1].count==0)
222 if (tdb->locked[list+1].count == 1) {
223 /* Down to last nested lock: unlock underneath */
224 if (tdb->header.rwlocks)
225 tdb_spinunlock(tdb, list, ltype);
227 tdb_brlock(tdb, FREELIST_TOP+4*list, F_UNLCK, F_SETLKW, 0);
229 tdb->locked[list+1].count--;
232 /* This is based on the hash agorithm from gdbm */
233 static u32 tdb_hash(TDB_DATA *key)
235 u32 value; /* Used to compute the hash value. */
236 u32 i; /* Used to cycle through random values. */
238 /* Set the initial value from the key size. */
239 for (value = 0x238F13AF * key->dsize, i=0; i < key->dsize; i++)
240 value = (value + (key->dptr[i] << (i*5 % 24)));
242 return (1103515243 * value + 12345);
245 /* check for an out of bounds access - if it is out of bounds then
246 see if the database has been expanded by someone else and expand
248 note that "len" is the minimum length needed for the db
250 static int tdb_oob(TDB_CONTEXT *tdb, tdb_off len, int probe)
253 if (len <= tdb->map_size)
255 if (tdb->flags & TDB_INTERNAL)
258 if (fstat(tdb->fd, &st) == -1)
259 return TDB_ERRCODE(TDB_ERR_IO, -1);
261 if (st.st_size < (size_t)len) {
263 TDB_LOG((tdb, 0,"tdb_oob len %d beyond eof at %d\n",
264 (int)len, (int)st.st_size));
266 return TDB_ERRCODE(TDB_ERR_IO, -1);
269 /* Unmap, update size, remap */
271 tdb->map_size = st.st_size;
276 /* write a lump of data at a specified offset */
277 static int tdb_write(TDB_CONTEXT *tdb, tdb_off off, void *buf, tdb_len len)
279 if (tdb_oob(tdb, off + len, 0) != 0)
283 memcpy(off + (char *)tdb->map_ptr, buf, len);
284 else if (lseek(tdb->fd, off, SEEK_SET) != off
285 || write(tdb->fd, buf, len) != (ssize_t)len) {
286 TDB_LOG((tdb, 0,"tdb_write failed at %d len=%d (%s)\n",
287 off, len, strerror(errno)));
288 return TDB_ERRCODE(TDB_ERR_IO, -1);
293 /* read a lump of data at a specified offset, maybe convert */
294 static int tdb_read(TDB_CONTEXT *tdb,tdb_off off,void *buf,tdb_len len,int cv)
296 if (tdb_oob(tdb, off + len, 0) != 0)
300 memcpy(buf, off + (char *)tdb->map_ptr, len);
301 else if (lseek(tdb->fd, off, SEEK_SET) != off
302 || read(tdb->fd, buf, len) != (ssize_t)len) {
303 TDB_LOG((tdb, 0,"tdb_read failed at %d len=%d (%s)\n",
304 off, len, strerror(errno)));
305 return TDB_ERRCODE(TDB_ERR_IO, -1);
312 /* read a lump of data, allocating the space for it */
313 static char *tdb_alloc_read(TDB_CONTEXT *tdb, tdb_off offset, tdb_len len)
317 if (!(buf = malloc(len))) {
318 TDB_LOG((tdb, 0,"tdb_alloc_read malloc failed len=%d (%s)\n",
319 len, strerror(errno)));
320 return TDB_ERRCODE(TDB_ERR_OOM, buf);
322 if (tdb_read(tdb, offset, buf, len, 0) == -1) {
329 /* read/write a tdb_off */
330 static int ofs_read(TDB_CONTEXT *tdb, tdb_off offset, tdb_off *d)
332 return tdb_read(tdb, offset, (char*)d, sizeof(*d), DOCONV());
334 static int ofs_write(TDB_CONTEXT *tdb, tdb_off offset, tdb_off *d)
337 return tdb_write(tdb, offset, CONVERT(off), sizeof(*d));
340 /* read/write a record */
341 static int rec_read(TDB_CONTEXT *tdb, tdb_off offset, struct list_struct *rec)
343 if (tdb_read(tdb, offset, rec, sizeof(*rec),DOCONV()) == -1)
345 if (TDB_BAD_MAGIC(rec)) {
346 TDB_LOG((tdb, 0,"rec_read bad magic 0x%x at offset=%d\n", rec->magic, offset));
347 return TDB_ERRCODE(TDB_ERR_CORRUPT, -1);
349 return tdb_oob(tdb, rec->next+sizeof(*rec), 0);
351 static int rec_write(TDB_CONTEXT *tdb, tdb_off offset, struct list_struct *rec)
353 struct list_struct r = *rec;
354 return tdb_write(tdb, offset, CONVERT(r), sizeof(r));
357 /* read a freelist record and check for simple errors */
358 static int rec_free_read(TDB_CONTEXT *tdb, tdb_off off, struct list_struct *rec)
360 if (tdb_read(tdb, off, rec, sizeof(*rec),DOCONV()) == -1)
362 if (rec->magic != TDB_FREE_MAGIC) {
363 TDB_LOG((tdb, 0,"rec_free_read bad magic 0x%x at offset=%d\n",
365 return TDB_ERRCODE(TDB_ERR_CORRUPT, -1);
367 if (tdb_oob(tdb, rec->next+sizeof(*rec), 0) != 0)
372 /* update a record tailer (must hold allocation lock) */
373 static int update_tailer(TDB_CONTEXT *tdb, tdb_off offset,
374 const struct list_struct *rec)
378 /* Offset of tailer from record header */
379 totalsize = sizeof(*rec) + rec->rec_len;
380 return ofs_write(tdb, offset + totalsize - sizeof(tdb_off),
385 static tdb_off tdb_dump_record(TDB_CONTEXT *tdb, tdb_off offset)
387 struct list_struct rec;
388 tdb_off tailer_ofs, tailer;
390 if (tdb_read(tdb, offset, (char *)&rec, sizeof(rec), DOCONV()) == -1) {
391 printf("ERROR: failed to read record at %u\n", offset);
395 printf(" rec: offset=%u next=%d rec_len=%d key_len=%d data_len=%d full_hash=0x%x magic=0x%x\n",
396 offset, rec.next, rec.rec_len, rec.key_len, rec.data_len, rec.full_hash, rec.magic);
398 tailer_ofs = offset + sizeof(rec) + rec.rec_len - sizeof(tdb_off);
399 if (ofs_read(tdb, tailer_ofs, &tailer) == -1) {
400 printf("ERROR: failed to read tailer at %u\n", tailer_ofs);
404 if (tailer != rec.rec_len + sizeof(rec)) {
405 printf("ERROR: tailer does not match record! tailer=%u totalsize=%u\n", tailer, rec.rec_len + sizeof(rec));
410 static void tdb_dump_chain(TDB_CONTEXT *tdb, int i)
412 tdb_off rec_ptr, top;
414 top = TDB_HASH_TOP(i);
416 tdb_lock(tdb, i, F_WRLCK);
418 if (ofs_read(tdb, top, &rec_ptr) == -1) {
419 tdb_unlock(tdb, i, F_WRLCK);
424 printf("hash=%d\n", i);
427 rec_ptr = tdb_dump_record(tdb, rec_ptr);
429 tdb_unlock(tdb, i, F_WRLCK);
432 void tdb_dump_all(TDB_CONTEXT *tdb)
436 for (i=0;i<tdb->header.hash_size;i++) {
437 tdb_dump_chain(tdb, i);
439 printf("freelist:\n");
440 tdb_dump_chain(tdb, -1);
443 void tdb_printfreelist(TDB_CONTEXT *tdb)
446 tdb_off offset, rec_ptr, last_ptr;
447 struct list_struct rec, lastrec, newrec;
449 tdb_lock(tdb, -1, F_WRLCK);
452 offset = FREELIST_TOP;
454 /* read in the freelist top */
455 if (ofs_read(tdb, offset, &rec_ptr) == -1) {
459 printf("freelist top=[0x%08x]\n", rec_ptr );
461 if (tdb_read(tdb, rec_ptr, (char *)&rec, sizeof(rec), DOCONV()) == -1) {
465 if (rec.magic != TDB_FREE_MAGIC) {
466 printf("bad magic 0x%08x in free list\n", rec.magic);
470 printf("entry offset=[0x%08x], rec.rec_len = [0x%08x (%d)]\n", rec.next, rec.rec_len, rec.rec_len );
471 total_free += rec.rec_len;
473 /* move to the next record */
476 printf("total rec_len = [0x%08x (%d)]\n", total_free, total_free );
478 tdb_unlock(tdb, -1, F_WRLCK);
482 /* Remove an element from the freelist. Must have alloc lock. */
483 static int remove_from_freelist(TDB_CONTEXT *tdb, tdb_off off, tdb_off next)
487 /* read in the freelist top */
488 last_ptr = FREELIST_TOP;
489 while (ofs_read(tdb, last_ptr, &i) != -1 && i != 0) {
491 /* We've found it! */
492 return ofs_write(tdb, last_ptr, &next);
494 /* Follow chain (next offset is at start of record) */
497 TDB_LOG((tdb, 0,"remove_from_freelist: not on list at off=%d\n", off));
498 return TDB_ERRCODE(TDB_ERR_CORRUPT, -1);
501 /* Add an element into the freelist. Merge adjacent records if
503 static int tdb_free(TDB_CONTEXT *tdb, tdb_off offset, struct list_struct *rec)
507 /* Allocation and tailer lock */
508 if (tdb_lock(tdb, -1, F_WRLCK) != 0)
511 /* set an initial tailer, so if we fail we don't leave a bogus record */
512 update_tailer(tdb, offset, rec);
514 /* Look right first (I'm an Australian, dammit) */
515 right = offset + sizeof(*rec) + rec->rec_len;
516 if (right + sizeof(*rec) <= tdb->map_size) {
517 struct list_struct r;
519 if (tdb_read(tdb, right, &r, sizeof(r), DOCONV()) == -1) {
520 TDB_LOG((tdb, 0, "tdb_free: right read failed at %u\n", right));
524 /* If it's free, expand to include it. */
525 if (r.magic == TDB_FREE_MAGIC) {
526 if (remove_from_freelist(tdb, right, r.next) == -1) {
527 TDB_LOG((tdb, 0, "tdb_free: right free failed at %u\n", right));
530 rec->rec_len += sizeof(r) + r.rec_len;
536 left = offset - sizeof(tdb_off);
537 if (left > TDB_HASH_TOP(tdb->header.hash_size-1)) {
538 struct list_struct l;
541 /* Read in tailer and jump back to header */
542 if (ofs_read(tdb, left, &leftsize) == -1) {
543 TDB_LOG((tdb, 0, "tdb_free: left offset read failed at %u\n", left));
546 left = offset - leftsize;
548 /* Now read in record */
549 if (tdb_read(tdb, left, &l, sizeof(l), DOCONV()) == -1) {
550 TDB_LOG((tdb, 0, "tdb_free: left read failed at %u (%u)\n", left, leftsize));
554 /* If it's free, expand to include it. */
555 if (l.magic == TDB_FREE_MAGIC) {
556 if (remove_from_freelist(tdb, left, l.next) == -1) {
557 TDB_LOG((tdb, 0, "tdb_free: left free failed at %u\n", left));
561 rec->rec_len += leftsize;
567 if (update_tailer(tdb, offset, rec) == -1) {
568 TDB_LOG((tdb, 0, "tdb_free: update_tailer failed at %u\n", offset));
572 /* Now, prepend to free list */
573 rec->magic = TDB_FREE_MAGIC;
575 if (ofs_read(tdb, FREELIST_TOP, &rec->next) == -1 ||
576 rec_write(tdb, offset, rec) == -1 ||
577 ofs_write(tdb, FREELIST_TOP, &offset) == -1) {
578 TDB_LOG((tdb, 0, "tdb_free record write failed at offset=%d\n", offset));
582 /* And we're done. */
583 tdb_unlock(tdb, -1, F_WRLCK);
587 tdb_unlock(tdb, -1, F_WRLCK);
592 /* expand a file. we prefer to use ftruncate, as that is what posix
593 says to use for mmap expansion */
594 static int expand_file(TDB_CONTEXT *tdb, tdb_off size, tdb_off addition)
597 #if HAVE_FTRUNCATE_EXTEND
598 if (ftruncate(tdb->fd, size+addition) != 0) {
599 TDB_LOG((tdb, 0, "expand_file ftruncate to %d failed (%s)\n",
600 size+addition, strerror(errno)));
605 if (lseek(tdb->fd, (size+addition) - 1, SEEK_SET) != (size+addition) - 1 ||
606 write(tdb->fd, &b, 1) != 1) {
607 TDB_LOG((tdb, 0, "expand_file to %d failed (%s)\n",
608 size+addition, strerror(errno)));
612 /* now fill the file with something. This ensures that the file isn't sparse, which would be
613 very bad if we ran out of disk. This must be done with write, not via mmap */
614 memset(buf, 0x42, sizeof(buf));
615 if (lseek(tdb->fd, size, SEEK_SET) != size)
618 int n = addition>sizeof(buf)?sizeof(buf):addition;
619 int ret = write(tdb->fd, buf, n);
621 TDB_LOG((tdb, 0, "expand_file write of %d failed (%s)\n",
622 n, strerror(errno)));
631 /* expand the database at least size bytes by expanding the underlying
632 file and doing the mmap again if necessary */
633 static int tdb_expand(TDB_CONTEXT *tdb, tdb_off size)
635 struct list_struct rec;
638 if (tdb_lock(tdb, -1, F_WRLCK) == -1) {
639 TDB_LOG((tdb, 0, "lock failed in tdb_expand\n"));
643 /* must know about any previous expansions by another process */
644 tdb_oob(tdb, tdb->map_size + 1, 1);
646 /* always make room for at least 10 more records, and round
647 the database up to a multiple of TDB_PAGE_SIZE */
648 size = TDB_ALIGN(tdb->map_size + size*10, TDB_PAGE_SIZE) - tdb->map_size;
650 if (!(tdb->flags & TDB_INTERNAL))
654 * We must ensure the file is unmapped before doing this
655 * to ensure consistency with systems like OpenBSD where
656 * writes and mmaps are not consistent.
659 /* expand the file itself */
660 if (!(tdb->flags & TDB_INTERNAL)) {
661 if (expand_file(tdb, tdb->map_size, size) != 0)
665 tdb->map_size += size;
667 if (tdb->flags & TDB_INTERNAL)
668 tdb->map_ptr = realloc(tdb->map_ptr, tdb->map_size);
671 * We must ensure the file is remapped before adding the space
672 * to ensure consistency with systems like OpenBSD where
673 * writes and mmaps are not consistent.
676 /* We're ok if the mmap fails as we'll fallback to read/write */
680 /* form a new freelist record */
681 memset(&rec,'\0',sizeof(rec));
682 rec.rec_len = size - sizeof(rec);
684 /* link it into the free list */
685 offset = tdb->map_size - size;
686 if (tdb_free(tdb, offset, &rec) == -1)
689 tdb_unlock(tdb, -1, F_WRLCK);
692 tdb_unlock(tdb, -1, F_WRLCK);
696 /* allocate some space from the free list. The offset returned points
697 to a unconnected list_struct within the database with room for at
698 least length bytes of total data
700 0 is returned if the space could not be allocated
702 static tdb_off tdb_allocate(TDB_CONTEXT *tdb, tdb_len length,
703 struct list_struct *rec)
705 tdb_off rec_ptr, last_ptr, newrec_ptr;
706 struct list_struct newrec;
708 if (tdb_lock(tdb, -1, F_WRLCK) == -1)
711 /* Extra bytes required for tailer */
712 length += sizeof(tdb_off);
715 last_ptr = FREELIST_TOP;
717 /* read in the freelist top */
718 if (ofs_read(tdb, FREELIST_TOP, &rec_ptr) == -1)
721 /* keep looking until we find a freelist record big enough */
723 if (rec_free_read(tdb, rec_ptr, rec) == -1)
726 if (rec->rec_len >= length) {
727 /* found it - now possibly split it up */
728 if (rec->rec_len > length + MIN_REC_SIZE) {
729 /* Length of left piece */
730 length = TDB_ALIGN(length, TDB_ALIGNMENT);
732 /* Right piece to go on free list */
733 newrec.rec_len = rec->rec_len
734 - (sizeof(*rec) + length);
735 newrec_ptr = rec_ptr + sizeof(*rec) + length;
737 /* And left record is shortened */
738 rec->rec_len = length;
742 /* Remove allocated record from the free list */
743 if (ofs_write(tdb, last_ptr, &rec->next) == -1)
746 /* Update header: do this before we drop alloc
747 lock, otherwise tdb_free() might try to
748 merge with us, thinking we're free.
749 (Thanks Jeremy Allison). */
750 rec->magic = TDB_MAGIC;
751 if (rec_write(tdb, rec_ptr, rec) == -1)
754 /* Did we create new block? */
756 /* Update allocated record tailer (we
758 if (update_tailer(tdb, rec_ptr, rec) == -1)
761 /* Free new record */
762 if (tdb_free(tdb, newrec_ptr, &newrec) == -1)
766 /* all done - return the new record offset */
767 tdb_unlock(tdb, -1, F_WRLCK);
770 /* move to the next record */
774 /* we didn't find enough space. See if we can expand the
775 database and if we can then try again */
776 if (tdb_expand(tdb, length + sizeof(*rec)) == 0)
779 tdb_unlock(tdb, -1, F_WRLCK);
783 /* initialise a new database with a specified hash size */
784 static int tdb_new_database(TDB_CONTEXT *tdb, int hash_size)
786 struct tdb_header *newdb;
789 /* We make it up in memory, then write it out if not internal */
790 size = sizeof(struct tdb_header) + (hash_size+1)*sizeof(tdb_off);
791 if (!(newdb = calloc(size, 1)))
792 return TDB_ERRCODE(TDB_ERR_OOM, -1);
794 /* Fill in the header */
795 newdb->version = TDB_VERSION;
796 newdb->hash_size = hash_size;
798 newdb->rwlocks = size;
800 if (tdb->flags & TDB_INTERNAL) {
801 tdb->map_size = size;
802 tdb->map_ptr = (char *)newdb;
803 memcpy(&tdb->header, newdb, sizeof(tdb->header));
804 /* Convert the `ondisk' version if asked. */
808 if (lseek(tdb->fd, 0, SEEK_SET) == -1)
811 if (ftruncate(tdb->fd, 0) == -1)
814 /* This creates an endian-converted header, as if read from disk */
816 memcpy(&tdb->header, newdb, sizeof(tdb->header));
817 /* Don't endian-convert the magic food! */
818 memcpy(newdb->magic_food, TDB_MAGIC_FOOD, strlen(TDB_MAGIC_FOOD)+1);
819 if (write(tdb->fd, newdb, size) != size)
822 ret = tdb_create_rwlocks(tdb->fd, hash_size);
829 /* Returns 0 on fail. On success, return offset of record, and fills
831 static tdb_off tdb_find(TDB_CONTEXT *tdb, TDB_DATA key, u32 hash,
832 struct list_struct *r)
836 /* read in the hash top */
837 if (ofs_read(tdb, TDB_HASH_TOP(hash), &rec_ptr) == -1)
840 /* keep looking until we find the right record */
842 if (rec_read(tdb, rec_ptr, r) == -1)
845 if (!TDB_DEAD(r) && hash==r->full_hash && key.dsize==r->key_len) {
847 /* a very likely hit - read the key */
848 k = tdb_alloc_read(tdb, rec_ptr + sizeof(*r),
853 if (memcmp(key.dptr, k, key.dsize) == 0) {
861 return TDB_ERRCODE(TDB_ERR_NOEXIST, 0);
864 /* If they do lockkeys, check that this hash is one they locked */
865 static int tdb_keylocked(TDB_CONTEXT *tdb, u32 hash)
868 if (!tdb->lockedkeys)
870 for (i = 0; i < tdb->lockedkeys[0]; i++)
871 if (tdb->lockedkeys[i+1] == hash)
873 return TDB_ERRCODE(TDB_ERR_NOLOCK, 0);
876 /* As tdb_find, but if you succeed, keep the lock */
877 static tdb_off tdb_find_lock(TDB_CONTEXT *tdb, TDB_DATA key, int locktype,
878 struct list_struct *rec)
882 hash = tdb_hash(&key);
883 if (!tdb_keylocked(tdb, hash))
885 if (tdb_lock(tdb, BUCKET(hash), locktype) == -1)
887 if (!(rec_ptr = tdb_find(tdb, key, hash, rec)))
888 tdb_unlock(tdb, BUCKET(hash), locktype);
892 enum TDB_ERROR tdb_error(TDB_CONTEXT *tdb)
897 static struct tdb_errname {
898 enum TDB_ERROR ecode; const char *estring;
899 } emap[] = { {TDB_SUCCESS, "Success"},
900 {TDB_ERR_CORRUPT, "Corrupt database"},
901 {TDB_ERR_IO, "IO Error"},
902 {TDB_ERR_LOCK, "Locking error"},
903 {TDB_ERR_OOM, "Out of memory"},
904 {TDB_ERR_EXISTS, "Record exists"},
905 {TDB_ERR_NOLOCK, "Lock exists on other keys"},
906 {TDB_ERR_NOEXIST, "Record does not exist"} };
908 /* Error string for the last tdb error */
909 const char *tdb_errorstr(TDB_CONTEXT *tdb)
912 for (i = 0; i < sizeof(emap) / sizeof(struct tdb_errname); i++)
913 if (tdb->ecode == emap[i].ecode)
914 return emap[i].estring;
915 return "Invalid error code";
918 /* update an entry in place - this only works if the new data size
919 is <= the old data size and the key exists.
922 static int tdb_update(TDB_CONTEXT *tdb, TDB_DATA key, TDB_DATA dbuf)
924 struct list_struct rec;
929 if (!(rec_ptr = tdb_find_lock(tdb, key, F_WRLCK, &rec)))
932 /* must be long enough key, data and tailer */
933 if (rec.rec_len < key.dsize + dbuf.dsize + sizeof(tdb_off)) {
934 tdb->ecode = TDB_SUCCESS; /* Not really an error */
938 if (tdb_write(tdb, rec_ptr + sizeof(rec) + rec.key_len,
939 dbuf.dptr, dbuf.dsize) == -1)
942 if (dbuf.dsize != rec.data_len) {
944 rec.data_len = dbuf.dsize;
945 ret = rec_write(tdb, rec_ptr, &rec);
949 tdb_unlock(tdb, BUCKET(rec.full_hash), F_WRLCK);
953 /* find an entry in the database given a key */
954 TDB_DATA tdb_fetch(TDB_CONTEXT *tdb, TDB_DATA key)
957 struct list_struct rec;
960 /* find which hash bucket it is in */
961 if (!(rec_ptr = tdb_find_lock(tdb,key,F_RDLCK,&rec)))
964 ret.dptr = tdb_alloc_read(tdb, rec_ptr + sizeof(rec) + rec.key_len,
966 ret.dsize = rec.data_len;
967 tdb_unlock(tdb, BUCKET(rec.full_hash), F_RDLCK);
971 /* check if an entry in the database exists
973 note that 1 is returned if the key is found and 0 is returned if not found
974 this doesn't match the conventions in the rest of this module, but is
977 int tdb_exists(TDB_CONTEXT *tdb, TDB_DATA key)
979 struct list_struct rec;
981 if (tdb_find_lock(tdb, key, F_RDLCK, &rec) == 0)
983 tdb_unlock(tdb, BUCKET(rec.full_hash), F_RDLCK);
987 /* record lock stops delete underneath */
988 static int lock_record(TDB_CONTEXT *tdb, tdb_off off)
990 return off ? tdb_brlock(tdb, off, F_RDLCK, F_SETLKW, 0) : 0;
992 /* write locks override our own fcntl readlocks, so check it here */
993 static int write_lock_record(TDB_CONTEXT *tdb, tdb_off off)
995 struct tdb_traverse_lock *i;
996 for (i = &tdb->travlocks; i; i = i->next)
999 return tdb_brlock(tdb, off, F_WRLCK, F_SETLKW, 1);
1001 static int write_unlock_record(TDB_CONTEXT *tdb, tdb_off off)
1003 return tdb_brlock(tdb, off, F_UNLCK, F_SETLKW, 0);
1005 /* fcntl locks don't stack: avoid unlocking someone else's */
1006 static int unlock_record(TDB_CONTEXT *tdb, tdb_off off)
1008 struct tdb_traverse_lock *i;
1013 for (i = &tdb->travlocks; i; i = i->next)
1016 return (count == 1 ? tdb_brlock(tdb, off, F_UNLCK, F_SETLKW, 0) : 0);
1019 /* actually delete an entry in the database given the offset */
1020 static int do_delete(TDB_CONTEXT *tdb, tdb_off rec_ptr, struct list_struct*rec)
1022 tdb_off last_ptr, i;
1023 struct list_struct lastrec;
1025 if (tdb->read_only) return -1;
1027 if (write_lock_record(tdb, rec_ptr) == -1) {
1028 /* Someone traversing here: mark it as dead */
1029 rec->magic = TDB_DEAD_MAGIC;
1030 return rec_write(tdb, rec_ptr, rec);
1032 write_unlock_record(tdb, rec_ptr);
1034 /* find previous record in hash chain */
1035 if (ofs_read(tdb, TDB_HASH_TOP(rec->full_hash), &i) == -1)
1037 for (last_ptr = 0; i != rec_ptr; last_ptr = i, i = lastrec.next)
1038 if (rec_read(tdb, i, &lastrec) == -1)
1041 /* unlink it: next ptr is at start of record. */
1043 last_ptr = TDB_HASH_TOP(rec->full_hash);
1044 if (ofs_write(tdb, last_ptr, &rec->next) == -1)
1047 /* recover the space */
1048 if (tdb_free(tdb, rec_ptr, rec) == -1)
1053 /* Uses traverse lock: 0 = finish, -1 = error, other = record offset */
1054 static int tdb_next_lock(TDB_CONTEXT *tdb, struct tdb_traverse_lock *tlock,
1055 struct list_struct *rec)
1057 int want_next = (tlock->off != 0);
1059 /* No traversal allows if you've called tdb_lockkeys() */
1060 if (tdb->lockedkeys)
1061 return TDB_ERRCODE(TDB_ERR_NOLOCK, -1);
1063 /* Lock each chain from the start one. */
1064 for (; tlock->hash < tdb->header.hash_size; tlock->hash++) {
1065 if (tdb_lock(tdb, tlock->hash, F_WRLCK) == -1)
1068 /* No previous record? Start at top of chain. */
1070 if (ofs_read(tdb, TDB_HASH_TOP(tlock->hash),
1074 /* Otherwise unlock the previous record. */
1075 unlock_record(tdb, tlock->off);
1079 /* We have offset of old record: grab next */
1080 if (rec_read(tdb, tlock->off, rec) == -1)
1082 tlock->off = rec->next;
1085 /* Iterate through chain */
1086 while( tlock->off) {
1088 if (rec_read(tdb, tlock->off, rec) == -1)
1090 if (!TDB_DEAD(rec)) {
1091 /* Woohoo: we found one! */
1092 lock_record(tdb, tlock->off);
1095 /* Try to clean dead ones from old traverses */
1096 current = tlock->off;
1097 tlock->off = rec->next;
1098 do_delete(tdb, current, rec);
1100 tdb_unlock(tdb, tlock->hash, F_WRLCK);
1103 /* We finished iteration without finding anything */
1104 return TDB_ERRCODE(TDB_SUCCESS, 0);
1108 tdb_unlock(tdb, tlock->hash, F_WRLCK);
1112 /* traverse the entire database - calling fn(tdb, key, data) on each element.
1113 return -1 on error or the record count traversed
1114 if fn is NULL then it is not called
1115 a non-zero return value from fn() indicates that the traversal should stop
1117 int tdb_traverse(TDB_CONTEXT *tdb, tdb_traverse_func fn, void *state)
1120 struct list_struct rec;
1121 struct tdb_traverse_lock tl = { NULL, 0, 0 };
1124 /* This was in the initializaton, above, but the IRIX compiler
1125 * did not like it. crh
1127 tl.next = tdb->travlocks.next;
1129 /* fcntl locks don't stack: beware traverse inside traverse */
1130 tdb->travlocks.next = &tl;
1132 /* tdb_next_lock places locks on the record returned, and its chain */
1133 while ((ret = tdb_next_lock(tdb, &tl, &rec)) > 0) {
1135 /* now read the full record */
1136 key.dptr = tdb_alloc_read(tdb, tl.off + sizeof(rec),
1137 rec.key_len + rec.data_len);
1139 tdb_unlock(tdb, tl.hash, F_WRLCK);
1140 unlock_record(tdb, tl.off);
1141 tdb->travlocks.next = tl.next;
1144 key.dsize = rec.key_len;
1145 dbuf.dptr = key.dptr + rec.key_len;
1146 dbuf.dsize = rec.data_len;
1148 /* Drop chain lock, call out */
1149 tdb_unlock(tdb, tl.hash, F_WRLCK);
1150 if (fn && fn(tdb, key, dbuf, state)) {
1151 /* They want us to terminate traversal */
1152 unlock_record(tdb, tl.off);
1153 tdb->travlocks.next = tl.next;
1159 tdb->travlocks.next = tl.next;
1166 /* find the first entry in the database and return its key */
1167 TDB_DATA tdb_firstkey(TDB_CONTEXT *tdb)
1170 struct list_struct rec;
1172 /* release any old lock */
1173 unlock_record(tdb, tdb->travlocks.off);
1174 tdb->travlocks.off = tdb->travlocks.hash = 0;
1176 if (tdb_next_lock(tdb, &tdb->travlocks, &rec) <= 0)
1178 /* now read the key */
1179 key.dsize = rec.key_len;
1180 key.dptr =tdb_alloc_read(tdb,tdb->travlocks.off+sizeof(rec),key.dsize);
1181 tdb_unlock(tdb, BUCKET(tdb->travlocks.hash), F_WRLCK);
1185 /* find the next entry in the database, returning its key */
1186 TDB_DATA tdb_nextkey(TDB_CONTEXT *tdb, TDB_DATA oldkey)
1189 TDB_DATA key = tdb_null;
1190 struct list_struct rec;
1193 /* Is locked key the old key? If so, traverse will be reliable. */
1194 if (tdb->travlocks.off) {
1195 if (tdb_lock(tdb,tdb->travlocks.hash,F_WRLCK))
1197 if (rec_read(tdb, tdb->travlocks.off, &rec) == -1
1198 || !(k = tdb_alloc_read(tdb,tdb->travlocks.off+sizeof(rec),
1200 || memcmp(k, oldkey.dptr, oldkey.dsize) != 0) {
1201 /* No, it wasn't: unlock it and start from scratch */
1202 unlock_record(tdb, tdb->travlocks.off);
1203 tdb_unlock(tdb, tdb->travlocks.hash, F_WRLCK);
1204 tdb->travlocks.off = 0;
1211 if (!tdb->travlocks.off) {
1212 /* No previous element: do normal find, and lock record */
1213 tdb->travlocks.off = tdb_find_lock(tdb, oldkey, F_WRLCK, &rec);
1214 if (!tdb->travlocks.off)
1216 tdb->travlocks.hash = BUCKET(rec.full_hash);
1217 lock_record(tdb, tdb->travlocks.off);
1219 oldhash = tdb->travlocks.hash;
1221 /* Grab next record: locks chain and returned record,
1222 unlocks old record */
1223 if (tdb_next_lock(tdb, &tdb->travlocks, &rec) > 0) {
1224 key.dsize = rec.key_len;
1225 key.dptr = tdb_alloc_read(tdb, tdb->travlocks.off+sizeof(rec),
1227 /* Unlock the chain of this new record */
1228 tdb_unlock(tdb, tdb->travlocks.hash, F_WRLCK);
1230 /* Unlock the chain of old record */
1231 tdb_unlock(tdb, BUCKET(oldhash), F_WRLCK);
1235 /* delete an entry in the database given a key */
1236 int tdb_delete(TDB_CONTEXT *tdb, TDB_DATA key)
1239 struct list_struct rec;
1242 if (!(rec_ptr = tdb_find_lock(tdb, key, F_WRLCK, &rec)))
1244 ret = do_delete(tdb, rec_ptr, &rec);
1245 tdb_unlock(tdb, BUCKET(rec.full_hash), F_WRLCK);
1249 /* store an element in the database, replacing any existing element
1252 return 0 on success, -1 on failure
1254 int tdb_store(TDB_CONTEXT *tdb, TDB_DATA key, TDB_DATA dbuf, int flag)
1256 struct list_struct rec;
1262 /* find which hash bucket it is in */
1263 hash = tdb_hash(&key);
1264 if (!tdb_keylocked(tdb, hash))
1266 if (tdb_lock(tdb, BUCKET(hash), F_WRLCK) == -1)
1269 /* check for it existing, on insert. */
1270 if (flag == TDB_INSERT) {
1271 if (tdb_exists(tdb, key)) {
1272 tdb->ecode = TDB_ERR_EXISTS;
1276 /* first try in-place update, on modify or replace. */
1277 if (tdb_update(tdb, key, dbuf) == 0)
1279 if (flag == TDB_MODIFY && tdb->ecode == TDB_ERR_NOEXIST)
1282 /* reset the error code potentially set by the tdb_update() */
1283 tdb->ecode = TDB_SUCCESS;
1285 /* delete any existing record - if it doesn't exist we don't
1286 care. Doing this first reduces fragmentation, and avoids
1287 coalescing with `allocated' block before it's updated. */
1288 if (flag != TDB_INSERT)
1289 tdb_delete(tdb, key);
1291 /* Copy key+value *before* allocating free space in case malloc
1292 fails and we are left with a dead spot in the tdb. */
1294 if (!(p = (char *)malloc(key.dsize + dbuf.dsize))) {
1295 tdb->ecode = TDB_ERR_OOM;
1299 memcpy(p, key.dptr, key.dsize);
1300 memcpy(p+key.dsize, dbuf.dptr, dbuf.dsize);
1302 /* now we're into insert / modify / replace of a record which
1303 * we know could not be optimised by an in-place store (for
1304 * various reasons). */
1305 if (!(rec_ptr = tdb_allocate(tdb, key.dsize + dbuf.dsize, &rec)))
1308 /* Read hash top into next ptr */
1309 if (ofs_read(tdb, TDB_HASH_TOP(hash), &rec.next) == -1)
1312 rec.key_len = key.dsize;
1313 rec.data_len = dbuf.dsize;
1314 rec.full_hash = hash;
1315 rec.magic = TDB_MAGIC;
1317 /* write out and point the top of the hash chain at it */
1318 if (rec_write(tdb, rec_ptr, &rec) == -1
1319 || tdb_write(tdb, rec_ptr+sizeof(rec), p, key.dsize+dbuf.dsize)==-1
1320 || ofs_write(tdb, TDB_HASH_TOP(hash), &rec_ptr) == -1) {
1322 /* Need to tdb_unallocate() here */
1328 tdb_unlock(tdb, BUCKET(hash), F_WRLCK);
1332 /* open the database, creating it if necessary
1334 The open_flags and mode are passed straight to the open call on the
1335 database file. A flags value of O_WRONLY is invalid. The hash size
1336 is advisory, use zero for a default value.
1338 return is NULL on error */
1339 TDB_CONTEXT *tdb_open(char *name, int hash_size, int tdb_flags,
1340 int open_flags, mode_t mode)
1342 TDB_CONTEXT tdb, *ret, *i;
1344 int rev = 0, locked;
1346 memset(&tdb, 0, sizeof(tdb));
1350 tdb.lockedkeys = NULL;
1351 tdb.flags = tdb_flags;
1353 if ((open_flags & O_ACCMODE) == O_WRONLY)
1356 hash_size = DEFAULT_HASH_SIZE;
1357 if ((open_flags & O_ACCMODE) == O_RDONLY) {
1359 /* read only databases don't do locking */
1360 tdb.flags |= TDB_NOLOCK;
1363 /* internal databases don't mmap or lock, and start off cleared */
1364 if (tdb.flags & TDB_INTERNAL) {
1365 tdb.flags |= (TDB_NOLOCK | TDB_NOMMAP);
1366 tdb.flags &= ~TDB_CLEAR_IF_FIRST;
1367 tdb_new_database(&tdb, hash_size);
1371 if ((tdb.fd = open(name, open_flags, mode)) == -1)
1374 /* ensure there is only one process initialising at once */
1375 if (tdb_brlock(&tdb, GLOBAL_LOCK, F_WRLCK, F_SETLKW, 0) == -1)
1378 /* we need to zero database if we are the only one with it open */
1379 if ((locked = (tdb_brlock(&tdb, ACTIVE_LOCK, F_WRLCK, F_SETLK, 0) == 0))
1380 && (tdb_flags & TDB_CLEAR_IF_FIRST)) {
1381 open_flags |= O_CREAT;
1382 if (ftruncate(tdb.fd, 0) == -1)
1386 if (read(tdb.fd, &tdb.header, sizeof(tdb.header)) != sizeof(tdb.header)
1387 || strcmp(tdb.header.magic_food, TDB_MAGIC_FOOD) != 0
1388 || (tdb.header.version != TDB_VERSION
1389 && !(rev = (tdb.header.version==TDB_BYTEREV(TDB_VERSION))))) {
1390 /* its not a valid database - possibly initialise it */
1391 if (!(open_flags & O_CREAT) || tdb_new_database(&tdb, hash_size) == -1)
1393 rev = (tdb.flags & TDB_CONVERT);
1396 tdb.flags &= ~TDB_CONVERT;
1398 tdb.flags |= TDB_CONVERT;
1399 convert(&tdb.header, sizeof(tdb.header));
1401 if (fstat(tdb.fd, &st) == -1)
1404 /* Is it already in the open list? If so, fail. */
1405 for (i = tdbs; i; i = i->next) {
1406 if (i->device == st.st_dev && i->inode == st.st_ino) {
1413 /* map the database and fill in the return structure */
1414 tdb.name = (char *)strdup(name);
1417 tdb.map_size = st.st_size;
1418 tdb.device = st.st_dev;
1419 tdb.inode = st.st_ino;
1420 tdb.locked = calloc(tdb.header.hash_size+1, sizeof(tdb.locked[0]));
1425 tdb_clear_spinlocks(&tdb);
1426 if (tdb_brlock(&tdb, ACTIVE_LOCK, F_UNLCK, F_SETLK, 0) == -1)
1429 /* leave this lock in place to indicate it's in use */
1430 if (tdb_brlock(&tdb, ACTIVE_LOCK, F_RDLCK, F_SETLKW, 0) == -1)
1434 if (!(ret = malloc(sizeof(tdb))))
1437 if (tdb_brlock(&tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0) == -1)
1445 if (tdb.flags & TDB_INTERNAL)
1459 /* close a database */
1460 int tdb_close(TDB_CONTEXT *tdb)
1466 if (tdb->flags & TDB_INTERNAL)
1474 ret = close(tdb->fd);
1477 if (tdb->lockedkeys)
1478 free(tdb->lockedkeys);
1480 /* Remove from contexts list */
1481 for (i = &tdbs; *i; i = &(*i)->next) {
1488 memset(tdb, 0, sizeof(*tdb));
1494 /* lock/unlock entire database */
1495 int tdb_lockall(TDB_CONTEXT *tdb)
1499 /* There are no locks on read-only dbs */
1501 return TDB_ERRCODE(TDB_ERR_LOCK, -1);
1502 if (tdb->lockedkeys)
1503 return TDB_ERRCODE(TDB_ERR_NOLOCK, -1);
1504 for (i = 0; i < tdb->header.hash_size; i++)
1505 if (tdb_lock(tdb, i, F_WRLCK))
1508 /* If error, release locks we have... */
1509 if (i < tdb->header.hash_size) {
1512 for ( j = 0; j < i; j++)
1513 tdb_unlock(tdb, j, F_WRLCK);
1514 return TDB_ERRCODE(TDB_ERR_NOLOCK, -1);
1519 void tdb_unlockall(TDB_CONTEXT *tdb)
1522 for (i=0; i < tdb->header.hash_size; i++)
1523 tdb_unlock(tdb, i, F_WRLCK);
1526 int tdb_lockkeys(TDB_CONTEXT *tdb, u32 number, TDB_DATA keys[])
1530 /* Can't lock more keys if already locked */
1531 if (tdb->lockedkeys)
1532 return TDB_ERRCODE(TDB_ERR_NOLOCK, -1);
1533 if (!(tdb->lockedkeys = malloc(sizeof(u32) * (number+1))))
1534 return TDB_ERRCODE(TDB_ERR_OOM, -1);
1535 /* First number in array is # keys */
1536 tdb->lockedkeys[0] = number;
1538 /* Insertion sort by bucket */
1539 for (i = 0; i < number; i++) {
1540 hash = tdb_hash(&keys[i]);
1541 for (j = 0; j < i && BUCKET(tdb->lockedkeys[j+1]) < BUCKET(hash); j++);
1542 memmove(&tdb->lockedkeys[j+2], &tdb->lockedkeys[j+1], sizeof(u32) * (i-j));
1543 tdb->lockedkeys[j+1] = hash;
1545 /* Finally, lock in order */
1546 for (i = 0; i < number; i++)
1547 if (tdb_lock(tdb, i, F_WRLCK))
1550 /* If error, release locks we have... */
1552 for ( j = 0; j < i; j++)
1553 tdb_unlock(tdb, j, F_WRLCK);
1554 free(tdb->lockedkeys);
1555 tdb->lockedkeys = NULL;
1556 return TDB_ERRCODE(TDB_ERR_NOLOCK, -1);
1561 /* Unlock the keys previously locked by tdb_lockkeys() */
1562 void tdb_unlockkeys(TDB_CONTEXT *tdb)
1565 for (i = 0; i < tdb->lockedkeys[0]; i++)
1566 tdb_unlock(tdb, tdb->lockedkeys[i+1], F_WRLCK);
1567 free(tdb->lockedkeys);
1568 tdb->lockedkeys = NULL;
1571 /* lock/unlock one hash chain. This is meant to be used to reduce
1572 contention - it cannot guarantee how many records will be locked */
1573 int tdb_chainlock(TDB_CONTEXT *tdb, TDB_DATA key)
1575 return tdb_lock(tdb, BUCKET(tdb_hash(&key)), F_WRLCK);
1577 void tdb_chainunlock(TDB_CONTEXT *tdb, TDB_DATA key)
1579 tdb_unlock(tdb, BUCKET(tdb_hash(&key)), F_WRLCK);
1583 /* register a loging function */
1584 void tdb_logging_function(TDB_CONTEXT *tdb, void (*fn)(TDB_CONTEXT *, int , const char *, ...))