summaryrefslogtreecommitdiff
path: root/lib/tdb2/tdb1_lock.c
blob: f21a22cacd10457a0191fcf7bd9c8b9065ca8ae2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
 /*
   Unix SMB/CIFS implementation.

   trivial database library

   Copyright (C) Andrew Tridgell              1999-2005
   Copyright (C) Paul `Rusty' Russell		   2000
   Copyright (C) Jeremy Allison			   2000-2003

     ** NOTE! The following LGPL license applies to the tdb
     ** library. This does NOT imply that all of Samba is released
     ** under the LGPL

   This library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
   License as published by the Free Software Foundation; either
   version 3 of the License, or (at your option) any later version.

   This library is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public
   License along with this library; if not, see <http://www.gnu.org/licenses/>.
*/

#include "tdb1_private.h"

/* list -1 is the alloc list, otherwise a hash chain. */
static tdb1_off_t lock_offset(int list)
{
	return TDB1_FREELIST_TOP + 4*list;
}

/* a byte range locking function - return 0 on success
   this functions locks/unlocks 1 byte at the specified offset.

   On error, errno is also set so that errors are passed back properly
   through tdb1_open().

   note that a len of zero means lock to end of file
*/
int tdb1_brlock(struct tdb_context *tdb,
	       int rw_type, tdb1_off_t offset, size_t len,
	       enum tdb_lock_flags flags)
{
	enum TDB_ERROR ecode = tdb_brlock(tdb, rw_type, offset, len, flags
					  | TDB_LOCK_NOCHECK);
	if (ecode == TDB_SUCCESS)
		return 0;
	tdb->last_error = ecode;
	return -1;
}

int tdb1_brunlock(struct tdb_context *tdb,
		 int rw_type, tdb1_off_t offset, size_t len)
{
	enum TDB_ERROR ecode = tdb_brunlock(tdb, rw_type, offset, len);
	if (ecode == TDB_SUCCESS)
		return 0;
	tdb->last_error = ecode;
	return -1;
}

int tdb1_allrecord_upgrade(struct tdb_context *tdb)
{
	enum TDB_ERROR ecode = tdb_allrecord_upgrade(tdb, TDB1_FREELIST_TOP);
	if (ecode == TDB_SUCCESS)
		return 0;
	tdb->last_error = ecode;
	return -1;
}

static struct tdb_lock *tdb1_find_nestlock(struct tdb_context *tdb,
					   tdb1_off_t offset)
{
	unsigned int i;

	for (i=0; i<tdb->file->num_lockrecs; i++) {
		if (tdb->file->lockrecs[i].off == offset) {
			return &tdb->file->lockrecs[i];
		}
	}
	return NULL;
}

/* lock an offset in the database. */
int tdb1_nest_lock(struct tdb_context *tdb, uint32_t offset, int ltype,
		  enum tdb_lock_flags flags)
{
	enum TDB_ERROR ecode;

	if (offset >= lock_offset(tdb->tdb1.header.hash_size)) {
		tdb->last_error = tdb_logerr(tdb, TDB_ERR_LOCK, TDB_LOG_ERROR,
					"tdb1_lock: invalid offset %u for"
					" ltype=%d",
					offset, ltype);
		return -1;
	}

	ecode = tdb_nest_lock(tdb, offset, ltype, flags | TDB_LOCK_NOCHECK);
	if (unlikely(ecode != TDB_SUCCESS)) {
		tdb->last_error = ecode;
		return -1;
	}
	return 0;
}

static int tdb1_lock_and_recover(struct tdb_context *tdb)
{
	int ret;

	/* We need to match locking order in transaction commit. */
	if (tdb1_brlock(tdb, F_WRLCK, TDB1_FREELIST_TOP, 0,
			TDB_LOCK_WAIT|TDB_LOCK_NOCHECK)) {
		return -1;
	}

	if (tdb1_brlock(tdb, F_WRLCK, TDB1_OPEN_LOCK, 1,
			TDB_LOCK_WAIT|TDB_LOCK_NOCHECK)) {
		tdb1_brunlock(tdb, F_WRLCK, TDB1_FREELIST_TOP, 0);
		return -1;
	}

	ret = tdb1_transaction_recover(tdb);

	tdb1_brunlock(tdb, F_WRLCK, TDB1_OPEN_LOCK, 1);
	tdb1_brunlock(tdb, F_WRLCK, TDB1_FREELIST_TOP, 0);

	return ret;
}

static bool have_data_locks(const struct tdb_context *tdb)
{
	unsigned int i;

	for (i = 0; i < tdb->file->num_lockrecs; i++) {
		if (tdb->file->lockrecs[i].off >= lock_offset(-1))
			return true;
	}
	return false;
}

static int tdb1_lock_list(struct tdb_context *tdb, int list, int ltype,
			 enum tdb_lock_flags waitflag)
{
	int ret;
	bool check = false;

	/* a allrecord lock allows us to avoid per chain locks */
	if (tdb->file->allrecord_lock.count &&
	    (ltype == tdb->file->allrecord_lock.ltype || ltype == F_RDLCK)) {
		return 0;
	}

	if (tdb->file->allrecord_lock.count) {
		tdb->last_error = TDB_ERR_LOCK;
		ret = -1;
	} else {
		/* Only check when we grab first data lock. */
		check = !have_data_locks(tdb);
		ret = tdb1_nest_lock(tdb, lock_offset(list), ltype, waitflag);

		if (ret == 0 && check && tdb1_needs_recovery(tdb)) {
			tdb1_nest_unlock(tdb, lock_offset(list), ltype);

			if (tdb1_lock_and_recover(tdb) == -1) {
				return -1;
			}
			return tdb1_lock_list(tdb, list, ltype, waitflag);
		}
	}
	return ret;
}

/* lock a list in the database. list -1 is the alloc list */
int tdb1_lock(struct tdb_context *tdb, int list, int ltype)
{
	int ret;

	ret = tdb1_lock_list(tdb, list, ltype, TDB_LOCK_WAIT);
	/* Don't log for EAGAIN and EINTR: they could have overridden lock fns */
	if (ret && errno != EAGAIN && errno != EINTR) {
		tdb_logerr(tdb, tdb->last_error, TDB_LOG_ERROR,
			   "tdb1_lock failed on list %d "
			   "ltype=%d (%s)",  list, ltype, strerror(errno));
	}
	return ret;
}

int tdb1_nest_unlock(struct tdb_context *tdb, uint32_t offset, int ltype)
{
	enum TDB_ERROR ecode;

	/* Sanity checks */
	if (offset >= lock_offset(tdb->tdb1.header.hash_size)) {
		tdb->last_error = tdb_logerr(tdb, TDB_ERR_LOCK, TDB_LOG_ERROR,
					"tdb1_unlock: offset %u invalid (%d)",
					offset, tdb->tdb1.header.hash_size);
		return -1;
	}

	ecode = tdb_nest_unlock(tdb, offset, ltype);
	if (unlikely(ecode != TDB_SUCCESS)) {
		tdb->last_error = ecode;
		return -1;
	}
	return 0;
}

int tdb1_unlock(struct tdb_context *tdb, int list, int ltype)
{
	/* a global lock allows us to avoid per chain locks */
	if (tdb->file->allrecord_lock.count &&
	    (ltype == tdb->file->allrecord_lock.ltype || ltype == F_RDLCK)) {
		return 0;
	}

	if (tdb->file->allrecord_lock.count) {
		tdb->last_error = TDB_ERR_LOCK;
		return -1;
	}

	return tdb1_nest_unlock(tdb, lock_offset(list), ltype);
}

/*
  get the transaction lock
 */
int tdb1_transaction_lock(struct tdb_context *tdb, int ltype,
			 enum tdb_lock_flags lockflags)
{
	return tdb1_nest_lock(tdb, TDB1_TRANSACTION_LOCK, ltype, lockflags);
}

/*
  release the transaction lock
 */
int tdb1_transaction_unlock(struct tdb_context *tdb, int ltype)
{
	return tdb1_nest_unlock(tdb, TDB1_TRANSACTION_LOCK, ltype);
}

/* lock/unlock entire database.  It can only be upgradable if you have some
 * other way of guaranteeing exclusivity (ie. transaction write lock).
 * We do the locking gradually to avoid being starved by smaller locks. */
int tdb1_allrecord_lock(struct tdb_context *tdb, int ltype,
		       enum tdb_lock_flags flags, bool upgradable)
{
	enum TDB_ERROR ecode;

	/* tdb_lock_gradual() doesn't know about tdb->tdb1.traverse_read. */
	if (tdb->tdb1.traverse_read && !(tdb->flags & TDB_NOLOCK)) {
		tdb->last_error = tdb_logerr(tdb, TDB_ERR_LOCK,
					     TDB_LOG_USE_ERROR,
					     "tdb1_allrecord_lock during"
					     " tdb1_read_traverse");
		return -1;
	}

	if (tdb->file->allrecord_lock.count
	    && tdb->file->allrecord_lock.ltype == ltype) {
		tdb->file->allrecord_lock.count++;
		return 0;
	}

	if (tdb1_have_extra_locks(tdb)) {
		/* can't combine global and chain locks */
		tdb->last_error = tdb_logerr(tdb, TDB_ERR_LOCK,
					     TDB_LOG_USE_ERROR,
					     "tdb1_allrecord_lock holding"
					     " other locks");
		return -1;
	}

	if (upgradable && ltype != F_RDLCK) {
		/* tdb error: you can't upgrade a write lock! */
		tdb->last_error = tdb_logerr(tdb, TDB_ERR_LOCK,
					     TDB_LOG_ERROR,
					     "tdb1_allrecord_lock cannot"
					     " have upgradable write lock");
		return -1;
	}

	/* We cover two kinds of locks:
	 * 1) Normal chain locks.  Taken for almost all operations.
	 * 3) Individual records locks.  Taken after normal or free
	 *    chain locks.
	 *
	 * It is (1) which cause the starvation problem, so we're only
	 * gradual for that. */
	ecode = tdb_lock_gradual(tdb, ltype, flags | TDB_LOCK_NOCHECK,
				 TDB1_FREELIST_TOP, tdb->tdb1.header.hash_size * 4);
	if (ecode != TDB_SUCCESS) {
		tdb->last_error = ecode;
		return -1;
	}

	/* Grab individual record locks. */
	if (tdb1_brlock(tdb, ltype, lock_offset(tdb->tdb1.header.hash_size), 0,
		       flags) == -1) {
		tdb1_brunlock(tdb, ltype, TDB1_FREELIST_TOP,
			     tdb->tdb1.header.hash_size * 4);
		return -1;
	}

	/* FIXME: Temporary cast. */
	tdb->file->allrecord_lock.owner = (void *)(struct tdb_context *)tdb;
	tdb->file->allrecord_lock.count = 1;
	/* If it's upgradable, it's actually exclusive so we can treat
	 * it as a write lock. */
	tdb->file->allrecord_lock.ltype = upgradable ? F_WRLCK : ltype;
	tdb->file->allrecord_lock.off = upgradable;

	if (tdb1_needs_recovery(tdb)) {
		tdb1_allrecord_unlock(tdb, ltype);
		if (tdb1_lock_and_recover(tdb) == -1) {
			return -1;
		}
		return tdb1_allrecord_lock(tdb, ltype, flags, upgradable);
	}

	return 0;
}



/* unlock entire db */
int tdb1_allrecord_unlock(struct tdb_context *tdb, int ltype)
{
	/* Don't try this during r/o traversal! */
	if (tdb->tdb1.traverse_read) {
		tdb->last_error = TDB_ERR_LOCK;
		return -1;
	}

	if (tdb->file->allrecord_lock.count == 0) {
		tdb->last_error = TDB_ERR_LOCK;
		return -1;
	}

	/* Upgradable locks are marked as write locks. */
	if (tdb->file->allrecord_lock.ltype != ltype
	    && (!tdb->file->allrecord_lock.off || ltype != F_RDLCK)) {
		tdb->last_error = TDB_ERR_LOCK;
		return -1;
	}

	if (tdb->file->allrecord_lock.count > 1) {
		tdb->file->allrecord_lock.count--;
		return 0;
	}

	if (tdb1_brunlock(tdb, ltype, TDB1_FREELIST_TOP, 0)) {
		tdb_logerr(tdb, tdb->last_error, TDB_LOG_ERROR,
			   "tdb1_unlockall failed (%s)", strerror(errno));
		return -1;
	}

	tdb->file->allrecord_lock.count = 0;
	tdb->file->allrecord_lock.ltype = 0;

	return 0;
}

/* lock/unlock one hash chain. This is meant to be used to reduce
   contention - it cannot guarantee how many records will be locked */
int tdb1_chainlock(struct tdb_context *tdb, TDB_DATA key)
{
	int ret = tdb1_lock(tdb,
			    TDB1_BUCKET(tdb_hash(tdb, key.dptr, key.dsize)),
			    F_WRLCK);
	return ret;
}

int tdb1_chainunlock(struct tdb_context *tdb, TDB_DATA key)
{
	return tdb1_unlock(tdb, TDB1_BUCKET(tdb_hash(tdb, key.dptr, key.dsize)),
			   F_WRLCK);
}

int tdb1_chainlock_read(struct tdb_context *tdb, TDB_DATA key)
{
	int ret;
	ret = tdb1_lock(tdb, TDB1_BUCKET(tdb_hash(tdb, key.dptr, key.dsize)),
			F_RDLCK);
	return ret;
}

int tdb1_chainunlock_read(struct tdb_context *tdb, TDB_DATA key)
{
	return tdb1_unlock(tdb, TDB1_BUCKET(tdb_hash(tdb, key.dptr, key.dsize)),
			   F_RDLCK);
}

/* record lock stops delete underneath */
int tdb1_lock_record(struct tdb_context *tdb, tdb1_off_t off)
{
	if (tdb->file->allrecord_lock.count) {
		return 0;
	}
	return off ? tdb1_brlock(tdb, F_RDLCK, off, 1, TDB_LOCK_WAIT) : 0;
}

/*
  Write locks override our own fcntl readlocks, so check it here.
  Note this is meant to be F_SETLK, *not* F_SETLKW, as it's not
  an error to fail to get the lock here.
*/
int tdb1_write_lock_record(struct tdb_context *tdb, tdb1_off_t off)
{
	struct tdb1_traverse_lock *i;
	for (i = &tdb->tdb1.travlocks; i; i = i->next)
		if (i->off == off)
			return -1;
	if (tdb->file->allrecord_lock.count) {
		if (tdb->file->allrecord_lock.ltype == F_WRLCK) {
			return 0;
		}
		return -1;
	}
	return tdb1_brlock(tdb, F_WRLCK, off, 1, TDB_LOCK_NOWAIT|TDB_LOCK_PROBE);
}

int tdb1_write_unlock_record(struct tdb_context *tdb, tdb1_off_t off)
{
	if (tdb->file->allrecord_lock.count) {
		return 0;
	}
	return tdb1_brunlock(tdb, F_WRLCK, off, 1);
}

/* fcntl locks don't stack: avoid unlocking someone else's */
int tdb1_unlock_record(struct tdb_context *tdb, tdb1_off_t off)
{
	struct tdb1_traverse_lock *i;
	uint32_t count = 0;

	if (tdb->file->allrecord_lock.count) {
		return 0;
	}

	if (off == 0)
		return 0;
	for (i = &tdb->tdb1.travlocks; i; i = i->next)
		if (i->off == off)
			count++;
	return (count == 1 ? tdb1_brunlock(tdb, F_RDLCK, off, 1) : 0);
}

bool tdb1_have_extra_locks(struct tdb_context *tdb)
{
	unsigned int extra = tdb->file->num_lockrecs;

	/* A transaction holds the lock for all records. */
	if (!tdb->tdb1.transaction && tdb->file->allrecord_lock.count) {
		return true;
	}

	/* We always hold the active lock if CLEAR_IF_FIRST. */
	if (tdb1_find_nestlock(tdb, TDB1_ACTIVE_LOCK)) {
		extra--;
	}

	/* In a transaction, we expect to hold the transaction lock */
	if (tdb->tdb1.transaction
	    && tdb1_find_nestlock(tdb, TDB1_TRANSACTION_LOCK)) {
		extra--;
	}

	return extra;
}

/* The transaction code uses this to remove all locks. */
void tdb1_release_transaction_locks(struct tdb_context *tdb)
{
	unsigned int i, active = 0;

	if (tdb->file->allrecord_lock.count != 0) {
		tdb1_brunlock(tdb, tdb->file->allrecord_lock.ltype, TDB1_FREELIST_TOP, 0);
		tdb->file->allrecord_lock.count = 0;
	}

	for (i=0;i<tdb->file->num_lockrecs;i++) {
		struct tdb_lock *lck = &tdb->file->lockrecs[i];

		/* Don't release the active lock!  Copy it to first entry. */
		if (lck->off == TDB1_ACTIVE_LOCK) {
			tdb->file->lockrecs[active++] = *lck;
		} else {
			tdb1_brunlock(tdb, lck->ltype, lck->off, 1);
		}
	}
	tdb->file->num_lockrecs = active;
	if (tdb->file->num_lockrecs == 0) {
		SAFE_FREE(tdb->file->lockrecs);
	}
}