2 Unix SMB/CIFS implementation.
4 trivial database library
6 Copyright (C) Andrew Tridgell 2005
8 ** NOTE! The following LGPL license applies to the tdb
9 ** library. This does NOT imply that all of Samba is released
12 This library is free software; you can redistribute it and/or
13 modify it under the terms of the GNU Lesser General Public
14 License as published by the Free Software Foundation; either
15 version 3 of the License, or (at your option) any later version.
17 This library is distributed in the hope that it will be useful,
18 but WITHOUT ANY WARRANTY; without even the implied warranty of
19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 Lesser General Public License for more details.
22 You should have received a copy of the GNU Lesser General Public
23 License along with this library; if not, see <http://www.gnu.org/licenses/>.
26 #include "tdb_private.h"
31 - only allow a single transaction at a time per database. This makes
32 using the transaction API simpler, as otherwise the caller would
33 have to cope with temporary failures in transactions that conflict
34 with other current transactions
36 - keep the transaction recovery information in the same file as the
37 database, using a special 'transaction recovery' record pointed at
38 by the header. This removes the need for extra journal files as
39 used by some other databases
41 - dynamically allocated the transaction recover record, re-using it
42 for subsequent transactions. If a larger record is needed then
43 tdb_free() the old record to place it on the normal tdb freelist
44 before allocating the new record
46 - during transactions, keep a linked list of writes all that have
47 been performed by intercepting all tdb_write() calls. The hooked
48 transaction versions of tdb_read() and tdb_write() check this
49 linked list and try to use the elements of the list in preference
52 - don't allow any locks to be held when a transaction starts,
53 otherwise we can end up with deadlock (plus lack of lock nesting
54 in posix locks would mean the lock is lost)
56 - if the caller gains a lock during the transaction but doesn't
57 release it then fail the commit
59 - allow for nested calls to tdb_transaction_start(), re-using the
60 existing transaction record. If the inner transaction is cancelled
61 then a subsequent commit will fail
63 - keep a mirrored copy of the tdb hash chain heads to allow for the
64 fast hash heads scan on traverse, updating the mirrored copy in
65 the transaction version of tdb_write
67 - allow callers to mix transaction and non-transaction use of tdb,
68 although once a transaction is started then an exclusive lock is
69 gained until the transaction is committed or cancelled
71 - the commit stategy involves first saving away all modified data
72 into a linearised buffer in the transaction recovery area, then
73 marking the transaction recovery area with a magic value to
74 indicate a valid recovery record. In total 4 fsync/msync calls are
75 needed per commit to prevent race conditions. It might be possible
76 to reduce this to 3 or even 2 with some more work.
78 - check for a valid recovery record on open of the tdb, while the
79 global lock is held. Automatically recover from the transaction
80 recovery area if needed, then continue with the open as
81 usual. This allows for smooth crash recovery with no administrator
84 - if TDB_NOSYNC is passed to flags in tdb_open then transactions are
85 still available, but no transaction recovery area is used and no
86 fsync/msync calls are made.
88 - if TDB_NO_NESTING is passed to flags in tdb open then transaction
89 nesting is disabled. tdb_transaction_start() will then implicitely
90 cancel any pending transactions and always start a new transaction
91 context instead of nesting.
97 hold the context of any current transaction
99 struct tdb_transaction {
100 /* we keep a mirrored copy of the tdb hash heads here so
101 tdb_next_hash_chain() can operate efficiently */
102 uint32_t *hash_heads;
104 /* the original io methods - used to do IOs to the real db */
105 const struct tdb_methods *io_methods;
107 /* the list of transaction blocks. When a block is first
108 written to, it gets created in this list */
111 uint32_t block_size; /* bytes in each block */
112 uint32_t last_block_size; /* number of valid bytes in the last block */
114 /* non-zero when an internal transaction error has
115 occurred. All write operations will then fail until the
116 transaction is ended */
117 int transaction_error;
119 /* when inside a transaction we need to keep track of any
120 nested tdb_transaction_start() calls, as these are allowed,
121 but don't create a new transaction */
124 /* set when a prepare has already occurred */
126 tdb_off_t magic_offset;
128 /* old file size before transaction */
129 tdb_len_t old_map_size;
131 /* we should re-pack on commit */
137 read while in a transaction. We need to check first if the data is in our list
138 of transaction elements, then if not do a real read
140 static int transaction_read(struct tdb_context *tdb, tdb_off_t off, void *buf,
141 tdb_len_t len, int cv)
145 /* Only a commit is allowed on a prepared transaction */
146 if (tdb->transaction->prepared) {
147 tdb->ecode = TDB_ERR_EINVAL;
148 TDB_LOG((tdb, TDB_DEBUG_FATAL, "transaction_read: transaction already prepared, read not allowed\n"));
149 tdb->transaction->transaction_error = 1;
153 /* break it down into block sized ops */
154 while (len + (off % tdb->transaction->block_size) > tdb->transaction->block_size) {
155 tdb_len_t len2 = tdb->transaction->block_size - (off % tdb->transaction->block_size);
156 if (transaction_read(tdb, off, buf, len2, cv) != 0) {
161 buf = (void *)(len2 + (char *)buf);
168 blk = off / tdb->transaction->block_size;
170 /* see if we have it in the block list */
171 if (tdb->transaction->num_blocks <= blk ||
172 tdb->transaction->blocks[blk] == NULL) {
173 /* nope, do a real read */
174 if (tdb->transaction->io_methods->tdb_read(tdb, off, buf, len, cv) != 0) {
180 /* it is in the block list. Now check for the last block */
181 if (blk == tdb->transaction->num_blocks-1) {
182 if (len > tdb->transaction->last_block_size) {
187 /* now copy it out of this block */
188 memcpy(buf, tdb->transaction->blocks[blk] + (off % tdb->transaction->block_size), len);
190 tdb_convert(buf, len);
195 TDB_LOG((tdb, TDB_DEBUG_FATAL, "transaction_read: failed at off=%d len=%d\n", off, len));
196 tdb->ecode = TDB_ERR_IO;
197 tdb->transaction->transaction_error = 1;
203 write while in a transaction
205 static int transaction_write(struct tdb_context *tdb, tdb_off_t off,
206 const void *buf, tdb_len_t len)
210 /* Only a commit is allowed on a prepared transaction */
211 if (tdb->transaction->prepared) {
212 tdb->ecode = TDB_ERR_EINVAL;
213 TDB_LOG((tdb, TDB_DEBUG_FATAL, "transaction_write: transaction already prepared, write not allowed\n"));
214 tdb->transaction->transaction_error = 1;
218 /* if the write is to a hash head, then update the transaction
220 if (len == sizeof(tdb_off_t) && off >= FREELIST_TOP &&
221 off < FREELIST_TOP+TDB_HASHTABLE_SIZE(tdb)) {
222 uint32_t chain = (off-FREELIST_TOP) / sizeof(tdb_off_t);
223 memcpy(&tdb->transaction->hash_heads[chain], buf, len);
226 /* break it up into block sized chunks */
227 while (len + (off % tdb->transaction->block_size) > tdb->transaction->block_size) {
228 tdb_len_t len2 = tdb->transaction->block_size - (off % tdb->transaction->block_size);
229 if (transaction_write(tdb, off, buf, len2) != 0) {
235 buf = (const void *)(len2 + (const char *)buf);
243 blk = off / tdb->transaction->block_size;
244 off = off % tdb->transaction->block_size;
246 if (tdb->transaction->num_blocks <= blk) {
247 uint8_t **new_blocks;
248 /* expand the blocks array */
249 if (tdb->transaction->blocks == NULL) {
250 new_blocks = (uint8_t **)malloc(
251 (blk+1)*sizeof(uint8_t *));
253 new_blocks = (uint8_t **)realloc(
254 tdb->transaction->blocks,
255 (blk+1)*sizeof(uint8_t *));
257 if (new_blocks == NULL) {
258 tdb->ecode = TDB_ERR_OOM;
261 memset(&new_blocks[tdb->transaction->num_blocks], 0,
262 (1+(blk - tdb->transaction->num_blocks))*sizeof(uint8_t *));
263 tdb->transaction->blocks = new_blocks;
264 tdb->transaction->num_blocks = blk+1;
265 tdb->transaction->last_block_size = 0;
268 /* allocate and fill a block? */
269 if (tdb->transaction->blocks[blk] == NULL) {
270 tdb->transaction->blocks[blk] = (uint8_t *)calloc(tdb->transaction->block_size, 1);
271 if (tdb->transaction->blocks[blk] == NULL) {
272 tdb->ecode = TDB_ERR_OOM;
273 tdb->transaction->transaction_error = 1;
276 if (tdb->transaction->old_map_size > blk * tdb->transaction->block_size) {
277 tdb_len_t len2 = tdb->transaction->block_size;
278 if (len2 + (blk * tdb->transaction->block_size) > tdb->transaction->old_map_size) {
279 len2 = tdb->transaction->old_map_size - (blk * tdb->transaction->block_size);
281 if (tdb->transaction->io_methods->tdb_read(tdb, blk * tdb->transaction->block_size,
282 tdb->transaction->blocks[blk],
284 SAFE_FREE(tdb->transaction->blocks[blk]);
285 tdb->ecode = TDB_ERR_IO;
288 if (blk == tdb->transaction->num_blocks-1) {
289 tdb->transaction->last_block_size = len2;
294 /* overwrite part of an existing block */
296 memset(tdb->transaction->blocks[blk] + off, 0, len);
298 memcpy(tdb->transaction->blocks[blk] + off, buf, len);
300 if (blk == tdb->transaction->num_blocks-1) {
301 if (len + off > tdb->transaction->last_block_size) {
302 tdb->transaction->last_block_size = len + off;
309 TDB_LOG((tdb, TDB_DEBUG_FATAL, "transaction_write: failed at off=%d len=%d\n",
310 (blk*tdb->transaction->block_size) + off, len));
311 tdb->transaction->transaction_error = 1;
317 write while in a transaction - this varient never expands the transaction blocks, it only
318 updates existing blocks. This means it cannot change the recovery size
320 static int transaction_write_existing(struct tdb_context *tdb, tdb_off_t off,
321 const void *buf, tdb_len_t len)
325 /* break it up into block sized chunks */
326 while (len + (off % tdb->transaction->block_size) > tdb->transaction->block_size) {
327 tdb_len_t len2 = tdb->transaction->block_size - (off % tdb->transaction->block_size);
328 if (transaction_write_existing(tdb, off, buf, len2) != 0) {
334 buf = (const void *)(len2 + (const char *)buf);
342 blk = off / tdb->transaction->block_size;
343 off = off % tdb->transaction->block_size;
345 if (tdb->transaction->num_blocks <= blk ||
346 tdb->transaction->blocks[blk] == NULL) {
350 if (blk == tdb->transaction->num_blocks-1 &&
351 off + len > tdb->transaction->last_block_size) {
352 if (off >= tdb->transaction->last_block_size) {
355 len = tdb->transaction->last_block_size - off;
358 /* overwrite part of an existing block */
359 memcpy(tdb->transaction->blocks[blk] + off, buf, len);
366 accelerated hash chain head search, using the cached hash heads
368 static void transaction_next_hash_chain(struct tdb_context *tdb, uint32_t *chain)
371 for (;h < tdb->header.hash_size;h++) {
372 /* the +1 takes account of the freelist */
373 if (0 != tdb->transaction->hash_heads[h+1]) {
381 out of bounds check during a transaction
383 static int transaction_oob(struct tdb_context *tdb, tdb_off_t len, int probe)
385 if (len <= tdb->map_size) {
388 tdb->ecode = TDB_ERR_IO;
393 transaction version of tdb_expand().
395 static int transaction_expand_file(struct tdb_context *tdb, tdb_off_t size,
398 /* add a write to the transaction elements, so subsequent
399 reads see the zero data */
400 if (transaction_write(tdb, size, NULL, addition) != 0) {
404 tdb->transaction->need_repack = true;
410 brlock during a transaction - ignore them
412 static int transaction_brlock(struct tdb_context *tdb, tdb_off_t offset,
413 int rw_type, int lck_type, int probe, size_t len)
418 static const struct tdb_methods transaction_methods = {
421 transaction_next_hash_chain,
423 transaction_expand_file,
430 static int transaction_sync(struct tdb_context *tdb, tdb_off_t offset, tdb_len_t length)
432 if (tdb->flags & TDB_NOSYNC) {
436 if (fsync(tdb->fd) != 0) {
437 tdb->ecode = TDB_ERR_IO;
438 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction: fsync failed\n"));
443 tdb_off_t moffset = offset & ~(tdb->page_size-1);
444 if (msync(moffset + (char *)tdb->map_ptr,
445 length + (offset - moffset), MS_SYNC) != 0) {
446 tdb->ecode = TDB_ERR_IO;
447 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction: msync failed - %s\n",
456 int tdb_transaction_cancel_internal(struct tdb_context *tdb)
460 if (tdb->transaction == NULL) {
461 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_cancel: no transaction\n"));
465 if (tdb->transaction->nesting != 0) {
466 tdb->transaction->transaction_error = 1;
467 tdb->transaction->nesting--;
471 tdb->map_size = tdb->transaction->old_map_size;
473 /* free all the transaction blocks */
474 for (i=0;i<tdb->transaction->num_blocks;i++) {
475 if (tdb->transaction->blocks[i] != NULL) {
476 free(tdb->transaction->blocks[i]);
479 SAFE_FREE(tdb->transaction->blocks);
481 if (tdb->transaction->magic_offset) {
482 const struct tdb_methods *methods = tdb->transaction->io_methods;
485 /* remove the recovery marker */
486 if (methods->tdb_write(tdb, tdb->transaction->magic_offset, &zero, 4) == -1 ||
487 transaction_sync(tdb, tdb->transaction->magic_offset, 4) == -1) {
488 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_cancel: failed to remove recovery magic\n"));
493 /* remove any global lock created during the transaction */
494 if (tdb->global_lock.count != 0) {
495 tdb_brlock(tdb, FREELIST_TOP, F_UNLCK, F_SETLKW, 0, 4*tdb->header.hash_size);
496 tdb->global_lock.count = 0;
499 /* remove any locks created during the transaction */
500 if (tdb->num_locks != 0) {
501 for (i=0;i<tdb->num_lockrecs;i++) {
502 tdb_brlock(tdb,FREELIST_TOP+4*tdb->lockrecs[i].list,
503 F_UNLCK,F_SETLKW, 0, 1);
506 tdb->num_lockrecs = 0;
507 SAFE_FREE(tdb->lockrecs);
510 /* restore the normal io methods */
511 tdb->methods = tdb->transaction->io_methods;
513 tdb_brlock(tdb, FREELIST_TOP, F_UNLCK, F_SETLKW, 0, 0);
514 tdb_transaction_unlock(tdb);
515 SAFE_FREE(tdb->transaction->hash_heads);
516 SAFE_FREE(tdb->transaction);
522 start a tdb transaction. No token is returned, as only a single
523 transaction is allowed to be pending per tdb_context
525 int tdb_transaction_start(struct tdb_context *tdb)
527 /* some sanity checks */
528 if (tdb->read_only || (tdb->flags & TDB_INTERNAL) || tdb->traverse_read) {
529 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_start: cannot start a transaction on a read-only or internal db\n"));
530 tdb->ecode = TDB_ERR_EINVAL;
534 /* cope with nested tdb_transaction_start() calls */
535 if (tdb->transaction != NULL) {
536 tdb_trace(tdb, "tdb_transaction_start");
537 if (!tdb->flags & TDB_NO_NESTING) {
538 tdb->transaction->nesting++;
539 TDB_LOG((tdb, TDB_DEBUG_TRACE, "tdb_transaction_start: nesting %d\n",
540 tdb->transaction->nesting));
543 tdb_transaction_cancel_internal(tdb);
544 TDB_LOG((tdb, TDB_DEBUG_TRACE, "tdb_transaction_start: cancelling previous transaction\n"));
548 if (tdb->num_locks != 0 || tdb->global_lock.count) {
549 /* the caller must not have any locks when starting a
550 transaction as otherwise we'll be screwed by lack
551 of nested locks in posix */
552 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_start: cannot start a transaction with locks held\n"));
553 tdb->ecode = TDB_ERR_LOCK;
557 if (tdb->travlocks.next != NULL) {
558 /* you cannot use transactions inside a traverse (although you can use
559 traverse inside a transaction) as otherwise you can end up with
561 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_start: cannot start a transaction within a traverse\n"));
562 tdb->ecode = TDB_ERR_LOCK;
566 tdb->transaction = (struct tdb_transaction *)
567 calloc(sizeof(struct tdb_transaction), 1);
568 if (tdb->transaction == NULL) {
569 tdb->ecode = TDB_ERR_OOM;
573 /* a page at a time seems like a reasonable compromise between compactness and efficiency */
574 tdb->transaction->block_size = tdb->page_size;
576 /* get the transaction write lock. This is a blocking lock. As
577 discussed with Volker, there are a number of ways we could
578 make this async, which we will probably do in the future */
579 if (tdb_transaction_lock(tdb, F_WRLCK) == -1) {
580 SAFE_FREE(tdb->transaction->blocks);
581 SAFE_FREE(tdb->transaction);
585 /* get a read lock from the freelist to the end of file. This
586 is upgraded to a write lock during the commit */
587 if (tdb_brlock(tdb, FREELIST_TOP, F_RDLCK, F_SETLKW, 0, 0) == -1) {
588 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_start: failed to get hash locks\n"));
589 tdb->ecode = TDB_ERR_LOCK;
593 /* setup a copy of the hash table heads so the hash scan in
594 traverse can be fast */
595 tdb->transaction->hash_heads = (uint32_t *)
596 calloc(tdb->header.hash_size+1, sizeof(uint32_t));
597 if (tdb->transaction->hash_heads == NULL) {
598 tdb->ecode = TDB_ERR_OOM;
601 if (tdb->methods->tdb_read(tdb, FREELIST_TOP, tdb->transaction->hash_heads,
602 TDB_HASHTABLE_SIZE(tdb), 0) != 0) {
603 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_start: failed to read hash heads\n"));
604 tdb->ecode = TDB_ERR_IO;
608 /* make sure we know about any file expansions already done by
610 tdb->methods->tdb_oob(tdb, tdb->map_size + 1, 1);
611 tdb->transaction->old_map_size = tdb->map_size;
613 /* finally hook the io methods, replacing them with
614 transaction specific methods */
615 tdb->transaction->io_methods = tdb->methods;
616 tdb->methods = &transaction_methods;
618 /* Trace at the end, so we get sequence number correct. */
619 tdb_trace(tdb, "tdb_transaction_start");
623 tdb_brlock(tdb, FREELIST_TOP, F_UNLCK, F_SETLKW, 0, 0);
624 tdb_transaction_unlock(tdb);
625 SAFE_FREE(tdb->transaction->blocks);
626 SAFE_FREE(tdb->transaction->hash_heads);
627 SAFE_FREE(tdb->transaction);
633 cancel the current transaction
635 int tdb_transaction_cancel(struct tdb_context *tdb)
637 tdb_trace(tdb, "tdb_transaction_cancel");
638 return tdb_transaction_cancel_internal(tdb);
642 work out how much space the linearised recovery data will consume
644 static tdb_len_t tdb_recovery_size(struct tdb_context *tdb)
646 tdb_len_t recovery_size = 0;
649 recovery_size = sizeof(uint32_t);
650 for (i=0;i<tdb->transaction->num_blocks;i++) {
651 if (i * tdb->transaction->block_size >= tdb->transaction->old_map_size) {
654 if (tdb->transaction->blocks[i] == NULL) {
657 recovery_size += 2*sizeof(tdb_off_t);
658 if (i == tdb->transaction->num_blocks-1) {
659 recovery_size += tdb->transaction->last_block_size;
661 recovery_size += tdb->transaction->block_size;
665 return recovery_size;
669 allocate the recovery area, or use an existing recovery area if it is
672 static int tdb_recovery_allocate(struct tdb_context *tdb,
673 tdb_len_t *recovery_size,
674 tdb_off_t *recovery_offset,
675 tdb_len_t *recovery_max_size)
677 struct list_struct rec;
678 const struct tdb_methods *methods = tdb->transaction->io_methods;
679 tdb_off_t recovery_head;
681 if (tdb_ofs_read(tdb, TDB_RECOVERY_HEAD, &recovery_head) == -1) {
682 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to read recovery head\n"));
688 if (recovery_head != 0 &&
689 methods->tdb_read(tdb, recovery_head, &rec, sizeof(rec), DOCONV()) == -1) {
690 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to read recovery record\n"));
694 *recovery_size = tdb_recovery_size(tdb);
696 if (recovery_head != 0 && *recovery_size <= rec.rec_len) {
697 /* it fits in the existing area */
698 *recovery_max_size = rec.rec_len;
699 *recovery_offset = recovery_head;
703 /* we need to free up the old recovery area, then allocate a
704 new one at the end of the file. Note that we cannot use
705 tdb_allocate() to allocate the new one as that might return
706 us an area that is being currently used (as of the start of
708 if (recovery_head != 0) {
709 if (tdb_free(tdb, recovery_head, &rec) == -1) {
710 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to free previous recovery area\n"));
715 /* the tdb_free() call might have increased the recovery size */
716 *recovery_size = tdb_recovery_size(tdb);
718 /* round up to a multiple of page size */
719 *recovery_max_size = TDB_ALIGN(sizeof(rec) + *recovery_size, tdb->page_size) - sizeof(rec);
720 *recovery_offset = tdb->map_size;
721 recovery_head = *recovery_offset;
723 if (methods->tdb_expand_file(tdb, tdb->transaction->old_map_size,
724 (tdb->map_size - tdb->transaction->old_map_size) +
725 sizeof(rec) + *recovery_max_size) == -1) {
726 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to create recovery area\n"));
730 /* remap the file (if using mmap) */
731 methods->tdb_oob(tdb, tdb->map_size + 1, 1);
733 /* we have to reset the old map size so that we don't try to expand the file
734 again in the transaction commit, which would destroy the recovery area */
735 tdb->transaction->old_map_size = tdb->map_size;
737 /* write the recovery header offset and sync - we can sync without a race here
738 as the magic ptr in the recovery record has not been set */
739 CONVERT(recovery_head);
740 if (methods->tdb_write(tdb, TDB_RECOVERY_HEAD,
741 &recovery_head, sizeof(tdb_off_t)) == -1) {
742 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to write recovery head\n"));
745 if (transaction_write_existing(tdb, TDB_RECOVERY_HEAD, &recovery_head, sizeof(tdb_off_t)) == -1) {
746 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_recovery_allocate: failed to write recovery head\n"));
755 setup the recovery data that will be used on a crash during commit
757 static int transaction_setup_recovery(struct tdb_context *tdb,
758 tdb_off_t *magic_offset)
760 tdb_len_t recovery_size;
761 unsigned char *data, *p;
762 const struct tdb_methods *methods = tdb->transaction->io_methods;
763 struct list_struct *rec;
764 tdb_off_t recovery_offset, recovery_max_size;
765 tdb_off_t old_map_size = tdb->transaction->old_map_size;
766 uint32_t magic, tailer;
770 check that the recovery area has enough space
772 if (tdb_recovery_allocate(tdb, &recovery_size,
773 &recovery_offset, &recovery_max_size) == -1) {
777 data = (unsigned char *)malloc(recovery_size + sizeof(*rec));
779 tdb->ecode = TDB_ERR_OOM;
783 rec = (struct list_struct *)data;
784 memset(rec, 0, sizeof(*rec));
787 rec->data_len = recovery_size;
788 rec->rec_len = recovery_max_size;
789 rec->key_len = old_map_size;
792 /* build the recovery data into a single blob to allow us to do a single
793 large write, which should be more efficient */
794 p = data + sizeof(*rec);
795 for (i=0;i<tdb->transaction->num_blocks;i++) {
799 if (tdb->transaction->blocks[i] == NULL) {
803 offset = i * tdb->transaction->block_size;
804 length = tdb->transaction->block_size;
805 if (i == tdb->transaction->num_blocks-1) {
806 length = tdb->transaction->last_block_size;
809 if (offset >= old_map_size) {
812 if (offset + length > tdb->transaction->old_map_size) {
813 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_setup_recovery: transaction data over new region boundary\n"));
815 tdb->ecode = TDB_ERR_CORRUPT;
818 memcpy(p, &offset, 4);
819 memcpy(p+4, &length, 4);
823 /* the recovery area contains the old data, not the
824 new data, so we have to call the original tdb_read
826 if (methods->tdb_read(tdb, offset, p + 8, length, 0) != 0) {
828 tdb->ecode = TDB_ERR_IO;
835 tailer = sizeof(*rec) + recovery_max_size;
836 memcpy(p, &tailer, 4);
839 /* write the recovery data to the recovery area */
840 if (methods->tdb_write(tdb, recovery_offset, data, sizeof(*rec) + recovery_size) == -1) {
841 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_setup_recovery: failed to write recovery data\n"));
843 tdb->ecode = TDB_ERR_IO;
846 if (transaction_write_existing(tdb, recovery_offset, data, sizeof(*rec) + recovery_size) == -1) {
847 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_setup_recovery: failed to write secondary recovery data\n"));
849 tdb->ecode = TDB_ERR_IO;
853 /* as we don't have ordered writes, we have to sync the recovery
854 data before we update the magic to indicate that the recovery
856 if (transaction_sync(tdb, recovery_offset, sizeof(*rec) + recovery_size) == -1) {
863 magic = TDB_RECOVERY_MAGIC;
866 *magic_offset = recovery_offset + offsetof(struct list_struct, magic);
868 if (methods->tdb_write(tdb, *magic_offset, &magic, sizeof(magic)) == -1) {
869 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_setup_recovery: failed to write recovery magic\n"));
870 tdb->ecode = TDB_ERR_IO;
873 if (transaction_write_existing(tdb, *magic_offset, &magic, sizeof(magic)) == -1) {
874 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_setup_recovery: failed to write secondary recovery magic\n"));
875 tdb->ecode = TDB_ERR_IO;
879 /* ensure the recovery magic marker is on disk */
880 if (transaction_sync(tdb, *magic_offset, sizeof(magic)) == -1) {
887 static int tdb_transaction_prepare_commit_internal(struct tdb_context *tdb)
889 const struct tdb_methods *methods;
891 if (tdb->transaction == NULL) {
892 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: no transaction\n"));
896 if (tdb->transaction->prepared) {
897 tdb->ecode = TDB_ERR_EINVAL;
898 tdb_transaction_cancel(tdb);
899 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: transaction already prepared\n"));
903 if (tdb->transaction->transaction_error) {
904 tdb->ecode = TDB_ERR_IO;
905 tdb_transaction_cancel_internal(tdb);
906 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: transaction error pending\n"));
911 if (tdb->transaction->nesting != 0) {
912 tdb->transaction->nesting--;
917 /* store seqnum now, before reading becomes illegal. */
918 tdb_ofs_read(tdb, TDB_SEQNUM_OFS, &tdb->transaction_prepare_seqnum);
921 /* check for a null transaction */
922 if (tdb->transaction->blocks == NULL) {
926 methods = tdb->transaction->io_methods;
928 /* if there are any locks pending then the caller has not
929 nested their locks properly, so fail the transaction */
930 if (tdb->num_locks || tdb->global_lock.count) {
931 tdb->ecode = TDB_ERR_LOCK;
932 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: locks pending on commit\n"));
933 tdb_transaction_cancel_internal(tdb);
937 /* upgrade the main transaction lock region to a write lock */
938 if (tdb_brlock_upgrade(tdb, FREELIST_TOP, 0) == -1) {
939 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: failed to upgrade hash locks\n"));
940 tdb->ecode = TDB_ERR_LOCK;
941 tdb_transaction_cancel_internal(tdb);
945 /* get the global lock - this prevents new users attaching to the database
947 if (tdb_brlock(tdb, GLOBAL_LOCK, F_WRLCK, F_SETLKW, 0, 1) == -1) {
948 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_prepare_commit: failed to get global lock\n"));
949 tdb->ecode = TDB_ERR_LOCK;
950 tdb_transaction_cancel_internal(tdb);
954 if (!(tdb->flags & TDB_NOSYNC)) {
955 /* write the recovery data to the end of the file */
956 if (transaction_setup_recovery(tdb, &tdb->transaction->magic_offset) == -1) {
957 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_prepare_commit: failed to setup recovery data\n"));
958 tdb_brlock(tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0, 1);
959 tdb_transaction_cancel_internal(tdb);
964 tdb->transaction->prepared = true;
966 /* expand the file to the new size if needed */
967 if (tdb->map_size != tdb->transaction->old_map_size) {
968 if (methods->tdb_expand_file(tdb, tdb->transaction->old_map_size,
970 tdb->transaction->old_map_size) == -1) {
971 tdb->ecode = TDB_ERR_IO;
972 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_prepare_commit: expansion failed\n"));
973 tdb_brlock(tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0, 1);
974 tdb_transaction_cancel_internal(tdb);
977 tdb->map_size = tdb->transaction->old_map_size;
978 methods->tdb_oob(tdb, tdb->map_size + 1, 1);
981 /* Keep the global lock until the actual commit */
987 prepare to commit the current transaction
989 int tdb_transaction_prepare_commit(struct tdb_context *tdb)
991 tdb_trace(tdb, "tdb_transaction_prepare_commit");
992 return tdb_transaction_prepare_commit_internal(tdb);
996 commit the current transaction
998 int tdb_transaction_commit(struct tdb_context *tdb)
1000 const struct tdb_methods *methods;
1004 if (tdb->transaction == NULL) {
1005 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_commit: no transaction\n"));
1009 /* If we've prepared, can't read seqnum. */
1010 if (tdb->transaction->prepared) {
1011 tdb_trace_seqnum(tdb, tdb->transaction_prepare_seqnum,
1012 "tdb_transaction_commit");
1014 tdb_trace(tdb, "tdb_transaction_commit");
1017 if (tdb->transaction->transaction_error) {
1018 tdb->ecode = TDB_ERR_IO;
1019 tdb_transaction_cancel(tdb);
1020 TDB_LOG((tdb, TDB_DEBUG_ERROR, "tdb_transaction_commit: transaction error pending\n"));
1025 if (tdb->transaction->nesting != 0) {
1026 tdb->transaction->nesting--;
1030 /* check for a null transaction */
1031 if (tdb->transaction->blocks == NULL) {
1032 tdb_transaction_cancel_internal(tdb);
1036 if (!tdb->transaction->prepared) {
1037 int ret = tdb_transaction_prepare_commit_internal(tdb);
1042 methods = tdb->transaction->io_methods;
1044 /* perform all the writes */
1045 for (i=0;i<tdb->transaction->num_blocks;i++) {
1049 if (tdb->transaction->blocks[i] == NULL) {
1053 offset = i * tdb->transaction->block_size;
1054 length = tdb->transaction->block_size;
1055 if (i == tdb->transaction->num_blocks-1) {
1056 length = tdb->transaction->last_block_size;
1059 if (methods->tdb_write(tdb, offset, tdb->transaction->blocks[i], length) == -1) {
1060 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_commit: write failed during commit\n"));
1062 /* we've overwritten part of the data and
1063 possibly expanded the file, so we need to
1064 run the crash recovery code */
1065 tdb->methods = methods;
1066 tdb_transaction_recover(tdb);
1068 tdb_transaction_cancel_internal(tdb);
1069 tdb_brlock(tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0, 1);
1071 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_commit: write failed\n"));
1074 SAFE_FREE(tdb->transaction->blocks[i]);
1077 SAFE_FREE(tdb->transaction->blocks);
1078 tdb->transaction->num_blocks = 0;
1080 /* ensure the new data is on disk */
1081 if (transaction_sync(tdb, 0, tdb->map_size) == -1) {
1085 tdb_brlock(tdb, GLOBAL_LOCK, F_UNLCK, F_SETLKW, 0, 1);
1088 TODO: maybe write to some dummy hdr field, or write to magic
1089 offset without mmap, before the last sync, instead of the
1093 /* on some systems (like Linux 2.6.x) changes via mmap/msync
1094 don't change the mtime of the file, this means the file may
1095 not be backed up (as tdb rounding to block sizes means that
1096 file size changes are quite rare too). The following forces
1097 mtime changes when a transaction completes */
1099 utime(tdb->name, NULL);
1102 need_repack = tdb->transaction->need_repack;
1104 /* use a transaction cancel to free memory and remove the
1105 transaction locks */
1106 tdb_transaction_cancel_internal(tdb);
1109 return tdb_repack(tdb);
1117 recover from an aborted transaction. Must be called with exclusive
1118 database write access already established (including the global
1119 lock to prevent new processes attaching)
1121 int tdb_transaction_recover(struct tdb_context *tdb)
1123 tdb_off_t recovery_head, recovery_eof;
1124 unsigned char *data, *p;
1126 struct list_struct rec;
1128 /* find the recovery area */
1129 if (tdb_ofs_read(tdb, TDB_RECOVERY_HEAD, &recovery_head) == -1) {
1130 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to read recovery head\n"));
1131 tdb->ecode = TDB_ERR_IO;
1135 if (recovery_head == 0) {
1136 /* we have never allocated a recovery record */
1140 /* read the recovery record */
1141 if (tdb->methods->tdb_read(tdb, recovery_head, &rec,
1142 sizeof(rec), DOCONV()) == -1) {
1143 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to read recovery record\n"));
1144 tdb->ecode = TDB_ERR_IO;
1148 if (rec.magic != TDB_RECOVERY_MAGIC) {
1149 /* there is no valid recovery data */
1153 if (tdb->read_only) {
1154 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: attempt to recover read only database\n"));
1155 tdb->ecode = TDB_ERR_CORRUPT;
1159 recovery_eof = rec.key_len;
1161 data = (unsigned char *)malloc(rec.data_len);
1163 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to allocate recovery data\n"));
1164 tdb->ecode = TDB_ERR_OOM;
1168 /* read the full recovery data */
1169 if (tdb->methods->tdb_read(tdb, recovery_head + sizeof(rec), data,
1170 rec.data_len, 0) == -1) {
1171 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to read recovery data\n"));
1172 tdb->ecode = TDB_ERR_IO;
1176 /* recover the file data */
1178 while (p+8 < data + rec.data_len) {
1184 memcpy(&len, p+4, 4);
1186 if (tdb->methods->tdb_write(tdb, ofs, p+8, len) == -1) {
1188 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to recover %d bytes at offset %d\n", len, ofs));
1189 tdb->ecode = TDB_ERR_IO;
1197 if (transaction_sync(tdb, 0, tdb->map_size) == -1) {
1198 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to sync recovery\n"));
1199 tdb->ecode = TDB_ERR_IO;
1203 /* if the recovery area is after the recovered eof then remove it */
1204 if (recovery_eof <= recovery_head) {
1205 if (tdb_ofs_write(tdb, TDB_RECOVERY_HEAD, &zero) == -1) {
1206 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to remove recovery head\n"));
1207 tdb->ecode = TDB_ERR_IO;
1212 /* remove the recovery magic */
1213 if (tdb_ofs_write(tdb, recovery_head + offsetof(struct list_struct, magic),
1215 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to remove recovery magic\n"));
1216 tdb->ecode = TDB_ERR_IO;
1220 /* reduce the file size to the old size */
1222 if (ftruncate(tdb->fd, recovery_eof) != 0) {
1223 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to reduce to recovery size\n"));
1224 tdb->ecode = TDB_ERR_IO;
1227 tdb->map_size = recovery_eof;
1230 if (transaction_sync(tdb, 0, recovery_eof) == -1) {
1231 TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_transaction_recover: failed to sync2 recovery\n"));
1232 tdb->ecode = TDB_ERR_IO;
1236 TDB_LOG((tdb, TDB_DEBUG_TRACE, "tdb_transaction_recover: recovered %d byte database\n",