summaryrefslogtreecommitdiff
path: root/source3/lib/g_lock.c
blob: 8a44b22e11fb78e5b37d93fbbf7ffb71d8e86274 (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
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
/*
   Unix SMB/CIFS implementation.
   global locks based on dbwrap and messaging
   Copyright (C) 2009 by Volker Lendecke

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

   This program 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 General Public License for more details.

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

#include "includes.h"
#include "g_lock.h"

static NTSTATUS g_lock_force_unlock(struct g_lock_ctx *ctx, const char *name,
				    struct server_id pid);

struct g_lock_ctx {
	struct db_context *db;
	struct messaging_context *msg;
};

/*
 * The "g_lock.tdb" file contains records, indexed by the 0-terminated
 * lockname. The record contains an array of "struct g_lock_rec"
 * structures. Waiters have the lock_type with G_LOCK_PENDING or'ed.
 */

struct g_lock_rec {
	enum g_lock_type lock_type;
	struct server_id pid;
};

struct g_lock_ctx *g_lock_ctx_init(TALLOC_CTX *mem_ctx,
				   struct messaging_context *msg)
{
	struct g_lock_ctx *result;

	result = talloc(mem_ctx, struct g_lock_ctx);
	if (result == NULL) {
		return NULL;
	}
	result->msg = msg;

	result->db = db_open(result, lock_path("g_lock.tdb"), 0,
			     TDB_CLEAR_IF_FIRST, O_RDWR|O_CREAT, 0700);
	if (result->db == NULL) {
		DEBUG(1, ("g_lock_init: Could not open g_lock.tdb"));
		TALLOC_FREE(result);
		return NULL;
	}
	return result;
}

static bool g_lock_conflicts(enum g_lock_type lock_type,
			     const struct g_lock_rec *rec)
{
	enum g_lock_type rec_lock = rec->lock_type;

	if ((rec_lock & G_LOCK_PENDING) != 0) {
		return false;
	}

	/*
	 * Only tested write locks so far. Very likely this routine
	 * needs to be fixed for read locks....
	 */
	if ((lock_type == G_LOCK_READ) && (rec_lock == G_LOCK_READ)) {
		return false;
	}
	return true;
}

static bool g_lock_parse(TALLOC_CTX *mem_ctx, TDB_DATA data,
			 int *pnum_locks, struct g_lock_rec **plocks)
{
	int i, num_locks;
	struct g_lock_rec *locks;

	if ((data.dsize % sizeof(struct g_lock_rec)) != 0) {
		DEBUG(1, ("invalid lock record length %d\n", (int)data.dsize));
		return false;
	}

	num_locks = data.dsize / sizeof(struct g_lock_rec);
	locks = talloc_array(mem_ctx, struct g_lock_rec, num_locks);
	if (locks == NULL) {
		DEBUG(1, ("talloc failed\n"));
		return false;
	}

	memcpy(locks, data.dptr, data.dsize);

	DEBUG(10, ("locks:\n"));
	for (i=0; i<num_locks; i++) {
		DEBUGADD(10, ("%s: %s %s\n",
			      procid_str(talloc_tos(), &locks[i].pid),
			      ((locks[i].lock_type & 1) == G_LOCK_READ) ?
			      "read" : "write",
			      (locks[i].lock_type & G_LOCK_PENDING) ?
			      "(pending)" : "(owner)"));

		if (process_exists(locks[i].pid)) {
			continue;
		}
		DEBUGADD(10, ("%s does not exist -- discarding\n",
			      procid_str(talloc_tos(), &locks[i].pid)));

		if (i < (num_locks-1)) {
			locks[i] = locks[num_locks-1];
		}
		num_locks -= 1;
	}

	*plocks = locks;
	*pnum_locks = num_locks;
	return true;
}

static struct g_lock_rec *g_lock_addrec(TALLOC_CTX *mem_ctx,
					struct g_lock_rec *locks,
					int num_locks,
					const struct server_id pid,
					enum g_lock_type lock_type)
{
	struct g_lock_rec *result;

	result = talloc_realloc(mem_ctx, locks, struct g_lock_rec,
				num_locks+1);
	if (result == NULL) {
		return NULL;
	}

	result[num_locks].pid = pid;
	result[num_locks].lock_type = lock_type;
	return result;
}

static void g_lock_got_retry(struct messaging_context *msg,
			     void *private_data,
			     uint32_t msg_type,
			     struct server_id server_id,
			     DATA_BLOB *data);
static void g_lock_timedout(struct tevent_context *ev,
			    struct tevent_timer *te,
			    struct timeval current_time,
			    void *private_data);

static NTSTATUS g_lock_trylock(struct g_lock_ctx *ctx, const char *name,
			       enum g_lock_type lock_type)
{
	struct db_record *rec = NULL;
	struct g_lock_rec *locks = NULL;
	int i, num_locks;
	struct server_id self;
	int our_index;
	TDB_DATA data;
	NTSTATUS status = NT_STATUS_OK;
	NTSTATUS store_status;

again:
	rec = ctx->db->fetch_locked(ctx->db, talloc_tos(),
				    string_term_tdb_data(name));
	if (rec == NULL) {
		DEBUG(10, ("fetch_locked(\"%s\") failed\n", name));
		status = NT_STATUS_LOCK_NOT_GRANTED;
		goto done;
	}

	if (!g_lock_parse(talloc_tos(), rec->value, &num_locks, &locks)) {
		DEBUG(10, ("g_lock_parse for %s failed\n", name));
		status = NT_STATUS_INTERNAL_ERROR;
		goto done;
	}

	self = procid_self();
	our_index = -1;

	for (i=0; i<num_locks; i++) {
		if (procid_equal(&self, &locks[i].pid)) {
			if (our_index != -1) {
				DEBUG(1, ("g_lock_trylock: Added ourself "
					  "twice!\n"));
				status = NT_STATUS_INTERNAL_ERROR;
				goto done;
			}
			if ((locks[i].lock_type & G_LOCK_PENDING) == 0) {
				DEBUG(1, ("g_lock_trylock: Found ourself not "
					  "pending!\n"));
				status = NT_STATUS_INTERNAL_ERROR;
				goto done;
			}

			our_index = i;

			/* never conflict with ourself */
			continue;
		}
		if (g_lock_conflicts(lock_type, &locks[i])) {
			struct server_id pid = locks[i].pid;

			if (!process_exists(pid)) {
				TALLOC_FREE(locks);
				TALLOC_FREE(rec);
				status = g_lock_force_unlock(ctx, name, pid);
				if (!NT_STATUS_IS_OK(status)) {
					DEBUG(1, ("Could not unlock dead lock "
						  "holder!\n"));
					goto done;
				}
				goto again;
			}
			lock_type |= G_LOCK_PENDING;
		}
	}

	if (our_index == -1) {
		/* First round, add ourself */

		locks = g_lock_addrec(talloc_tos(), locks, num_locks,
				      self, lock_type);
		if (locks == NULL) {
			DEBUG(10, ("g_lock_addrec failed\n"));
			status = NT_STATUS_NO_MEMORY;
			goto done;
		}
	} else {
		/*
		 * Retry. We were pending last time. Overwrite the
		 * stored lock_type with what we calculated, we might
		 * have acquired the lock this time.
		 */
		locks[our_index].lock_type = lock_type;
	}

	data = make_tdb_data((uint8_t *)locks, talloc_get_size(locks));
	store_status = rec->store(rec, data, 0);
	if (!NT_STATUS_IS_OK(store_status)) {
		DEBUG(1, ("rec->store failed: %s\n",
			  nt_errstr(store_status)));
		status = store_status;
	}

done:
	TALLOC_FREE(locks);
	TALLOC_FREE(rec);

	if (NT_STATUS_IS_OK(status) && (lock_type & G_LOCK_PENDING) != 0) {
		return STATUS_PENDING;
	}

	return NT_STATUS_OK;
}

NTSTATUS g_lock_lock(struct g_lock_ctx *ctx, const char *name,
		     enum g_lock_type lock_type, struct timeval timeout)
{
	struct tevent_timer *te = NULL;
	NTSTATUS status;
	bool retry = false;
	struct timeval timeout_end;
	struct timeval timeout_remaining;
	struct timeval time_now;

	DEBUG(10, ("Trying to acquire lock %d for %s\n", (int)lock_type,
		   name));

	if (lock_type & ~1) {
		DEBUG(1, ("Got invalid lock type %d for %s\n",
			  (int)lock_type, name));
		return NT_STATUS_INVALID_PARAMETER;
	}

#ifdef CLUSTER_SUPPORT
	if (lp_clustering()) {
		status = ctdb_watch_us(messaging_ctdbd_connection());
		if (!NT_STATUS_IS_OK(status)) {
			DEBUG(10, ("could not register retry with ctdb: %s\n",
				   nt_errstr(status)));
			goto done;
		}
	}
#endif

	status = messaging_register(ctx->msg, &retry, MSG_DBWRAP_G_LOCK_RETRY,
				    g_lock_got_retry);
	if (!NT_STATUS_IS_OK(status)) {
		DEBUG(10, ("messaging_register failed: %s\n",
			   nt_errstr(status)));
		return status;
	}

	time_now = timeval_current();
	timeout_end = timeval_sum(&time_now, &timeout);

	while (true) {
		fd_set _r_fds;
		fd_set *r_fds = NULL;
		int max_fd = 0;
		int ret;

		status = g_lock_trylock(ctx, name, lock_type);
		if (NT_STATUS_IS_OK(status)) {
			DEBUG(10, ("Got lock %s\n", name));
			break;
		}
		if (!NT_STATUS_EQUAL(status, STATUS_PENDING)) {
			DEBUG(10, ("g_lock_trylock failed: %s\n",
				   nt_errstr(status)));
			break;
		}

		DEBUG(10, ("g_lock_trylock: Did not get lock, waiting...\n"));

		/* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
		 *             !!! HACK ALERT --- FIX ME !!!
		 * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
		 * What we really want to do here is to react to
		 * MSG_DBWRAP_G_LOCK_RETRY messages that are either sent
		 * by a client doing g_lock_unlock or by ourselves when
		 * we receive a CTDB_SRVID_SAMBA_NOTIFY or
		 * CTDB_SRVID_RECONFIGURE message from ctdbd, i.e. when
		 * either a client holding a lock or a complete node
		 * has died.
		 *
		 * Doing this properly involves calling tevent_loop_once(),
		 * but doing this here with the main ctdbd messaging context
		 * creates a nested event loop when g_lock_lock() is called
		 * from the main event loop, e.g. in a tcon_and_X where the
		 * share_info.tdb needs to be initialized and is locked by
		 * another process, or when the remore registry is accessed
		 * for writing and some other process already holds a lock
		 * on the registry.tdb.
		 *
		 * So as a quick fix, we act a little corasely here: we do
		 * a select on the ctdb connection fd and when it is readable
		 * or we get EINTR, then we retry without actually parsing
		 * any ctdb packages or dispatching messages. This means that
		 * we retry more often than intended by design, but this does
		 * not harm and it is unobtrusive. When we have finished,
		 * the main loop will pick up all the messages and ctdb
		 * packets. The only extra twist is that we cannot use timed
		 * events here but have to handcode a timeout.
		 */

#ifdef CLUSTER_SUPPORT
		if (lp_clustering()) {
			struct ctdbd_connection *conn = messaging_ctdbd_connection();

			r_fds = &_r_fds;
			FD_ZERO(r_fds);
			max_fd = ctdbd_conn_get_fd(conn);
			FD_SET(max_fd, r_fds);
		}
#endif

		time_now = timeval_current();
		timeout_remaining = timeval_until(&time_now, &timeout_end);

		ret = sys_select(max_fd + 1, r_fds, NULL, NULL,
				 &timeout_remaining);

		if (ret == -1) {
			if (errno != EINTR) {
				DEBUG(1, ("error calling select: %s\n",
					  strerror(errno)));
				status = NT_STATUS_INTERNAL_ERROR;
				break;
			}
			/*
			 * errno == EINTR:
			 * This means a signal was received.
			 * It might have been a MSG_DBWRAP_G_LOCK_RETRY message.
			 * ==> retry
			 */
		} else if (ret == 0) {
			if (timeval_expired(&timeout_end)) {
				DEBUG(10, ("g_lock_lock timed out\n"));
				status = NT_STATUS_LOCK_NOT_GRANTED;
				break;
			} else {
				DEBUG(10, ("select returned 0 but timeout not "
					   "not expired: strange - retrying\n"));
			}
		} else if (ret != 1) {
			DEBUG(1, ("invalid return code of select: %d\n", ret));
			status = NT_STATUS_INTERNAL_ERROR;
			break;
		}
		/*
		 * ret == 1:
		 * This means ctdbd has sent us some data.
		 * Might be a CTDB_SRVID_RECONFIGURE or a
		 * CTDB_SRVID_SAMBA_NOTIFY message.
		 * ==> retry
		 */
	}

done:

	if (!NT_STATUS_IS_OK(status)) {
		NTSTATUS unlock_status;

		unlock_status = g_lock_unlock(ctx, name);

		if (!NT_STATUS_IS_OK(unlock_status)) {
			DEBUG(1, ("Could not remove ourself from the locking "
				  "db: %s\n", nt_errstr(status)));
		}
	}

	messaging_deregister(ctx->msg, MSG_DBWRAP_G_LOCK_RETRY, &retry);
	TALLOC_FREE(te);

	return status;
}

static void g_lock_got_retry(struct messaging_context *msg,
			     void *private_data,
			     uint32_t msg_type,
			     struct server_id server_id,
			     DATA_BLOB *data)
{
	bool *pretry = (bool *)private_data;

	DEBUG(10, ("Got retry message from pid %s\n",
		   procid_str(talloc_tos(), &server_id)));

	*pretry = true;
}

static void g_lock_timedout(struct tevent_context *ev,
			    struct tevent_timer *te,
			    struct timeval current_time,
			    void *private_data)
{
	bool *ptimedout = (bool *)private_data;
	*ptimedout = true;
	TALLOC_FREE(te);
}

static NTSTATUS g_lock_force_unlock(struct g_lock_ctx *ctx, const char *name,
				    struct server_id pid)
{
	struct db_record *rec = NULL;
	struct g_lock_rec *locks = NULL;
	int i, num_locks;
	enum g_lock_type lock_type;
	NTSTATUS status;

	rec = ctx->db->fetch_locked(ctx->db, talloc_tos(),
				    string_term_tdb_data(name));
	if (rec == NULL) {
		DEBUG(10, ("fetch_locked(\"%s\") failed\n", name));
		status = NT_STATUS_INTERNAL_ERROR;
		goto done;
	}

	if (!g_lock_parse(talloc_tos(), rec->value, &num_locks, &locks)) {
		DEBUG(10, ("g_lock_parse for %s failed\n", name));
		status = NT_STATUS_INTERNAL_ERROR;
		goto done;
	}

	for (i=0; i<num_locks; i++) {
		if (procid_equal(&pid, &locks[i].pid)) {
			break;
		}
	}

	if (i == num_locks) {
		DEBUG(10, ("g_lock_force_unlock: Lock not found\n"));
		status = NT_STATUS_INTERNAL_ERROR;
		goto done;
	}

	lock_type = locks[i].lock_type;

	if (i < (num_locks-1)) {
		locks[i] = locks[num_locks-1];
	}
	num_locks -= 1;

	if (num_locks == 0) {
		status = rec->delete_rec(rec);
	} else {
		TDB_DATA data;
		data = make_tdb_data((uint8_t *)locks,
				     sizeof(struct g_lock_rec) * num_locks);
		status = rec->store(rec, data, 0);
	}

	if (!NT_STATUS_IS_OK(status)) {
		DEBUG(1, ("g_lock_force_unlock: Could not store record: %s\n",
			  nt_errstr(status)));
		goto done;
	}

	if ((lock_type & G_LOCK_PENDING) == 0) {
		/*
		 * We've been the lock holder. Tell all others to retry.
		 */
		for (i=0; i<num_locks; i++) {
			if ((locks[i].lock_type & G_LOCK_PENDING) == 0) {
				continue;
			}

			/*
			 * Ping all waiters to retry
			 */
			status = messaging_send(ctx->msg, locks[i].pid,
						MSG_DBWRAP_G_LOCK_RETRY,
						&data_blob_null);
			if (!NT_STATUS_IS_OK(status)) {
				DEBUG(1, ("sending retry to %s failed: %s\n",
					  procid_str(talloc_tos(),
						     &locks[i].pid),
					  nt_errstr(status)));
			}
		}
	}
done:

	TALLOC_FREE(locks);
	TALLOC_FREE(rec);
	return status;
}

NTSTATUS g_lock_unlock(struct g_lock_ctx *ctx, const char *name)
{
	NTSTATUS status;

	status = g_lock_force_unlock(ctx, name, procid_self());

#ifdef CLUSTER_SUPPORT
	if (lp_clustering()) {
		ctdb_unwatch(messaging_ctdbd_connection());
	}
#endif
	return status;
}

struct g_lock_locks_state {
	int (*fn)(const char *name, void *private_data);
	void *private_data;
};

static int g_lock_locks_fn(struct db_record *rec, void *priv)
{
	struct g_lock_locks_state *state = (struct g_lock_locks_state *)priv;

	if ((rec->key.dsize == 0) || (rec->key.dptr[rec->key.dsize-1] != 0)) {
		DEBUG(1, ("invalid key in g_lock.tdb, ignoring\n"));
		return 0;
	}
	return state->fn((char *)rec->key.dptr, state->private_data);
}

int g_lock_locks(struct g_lock_ctx *ctx,
		 int (*fn)(const char *name, void *private_data),
		 void *private_data)
{
	struct g_lock_locks_state state;

	state.fn = fn;
	state.private_data = private_data;

	return ctx->db->traverse_read(ctx->db, g_lock_locks_fn, &state);
}

NTSTATUS g_lock_dump(struct g_lock_ctx *ctx, const char *name,
		     int (*fn)(struct server_id pid,
			       enum g_lock_type lock_type,
			       void *private_data),
		     void *private_data)
{
	TDB_DATA data;
	int i, num_locks;
	struct g_lock_rec *locks = NULL;
	bool ret;

	if (ctx->db->fetch(ctx->db, talloc_tos(), string_term_tdb_data(name),
			   &data) != 0) {
		return NT_STATUS_NOT_FOUND;
	}

	if ((data.dsize == 0) || (data.dptr == NULL)) {
		return NT_STATUS_OK;
	}

	ret = g_lock_parse(talloc_tos(), data, &num_locks, &locks);

	TALLOC_FREE(data.dptr);

	if (!ret) {
		DEBUG(10, ("g_lock_parse for %s failed\n", name));
		return NT_STATUS_INTERNAL_ERROR;
	}

	for (i=0; i<num_locks; i++) {
		if (fn(locks[i].pid, locks[i].lock_type, private_data) != 0) {
			break;
		}
	}
	TALLOC_FREE(locks);
	return NT_STATUS_OK;
}

struct g_lock_get_state {
	bool found;
	struct server_id *pid;
};

static int g_lock_get_fn(struct server_id pid, enum g_lock_type lock_type,
			 void *priv)
{
	struct g_lock_get_state *state = (struct g_lock_get_state *)priv;

	if ((lock_type & G_LOCK_PENDING) != 0) {
		return 0;
	}

	state->found = true;
	*state->pid = pid;
	return 1;
}

NTSTATUS g_lock_get(struct g_lock_ctx *ctx, const char *name,
		    struct server_id *pid)
{
	struct g_lock_get_state state;
	NTSTATUS status;

	state.found = false;
	state.pid = pid;

	status = g_lock_dump(ctx, name, g_lock_get_fn, &state);
	if (!NT_STATUS_IS_OK(status)) {
		return status;
	}
	if (!state.found) {
		return NT_STATUS_NOT_FOUND;
	}
	return NT_STATUS_OK;
}