summaryrefslogtreecommitdiff
path: root/source4/ntvfs/common/brlock.c
blob: e01f458e7444abf76ab4f33e8ea8a7eb26690774 (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
/* 
   Unix SMB/CIFS implementation.

   generic byte range locking code

   Copyright (C) Andrew Tridgell 1992-2004
   Copyright (C) Jeremy Allison 1992-2000
   
   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 2 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, write to the Free Software
   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/

/* This module implements a tdb based byte range locking service,
   replacing the fcntl() based byte range locking previously
   used. This allows us to provide the same semantics as NT */

#include "includes.h"
#include "messages.h"

/*
  in this module a "DATA_BLOB *file_key" is a blob that uniquely identifies
  a file. For a local posix filesystem this will usually be a combination
  of the device and inode numbers of the file, but it can be anything 
  that uniquely idetifies a file for locking purposes, as long
  as it is applied consistently.
*/

/*
  the lock context contains the elements that define whether one
  lock is the same as another lock
*/
struct lock_context {
	servid_t server;
	uint16_t smbpid;
	uint16_t tid;
};

/* The data in brlock records is an unsorted linear array of these
   records.  It is unnecessary to store the count as tdb provides the
   size of the record */
struct lock_struct {
	struct lock_context context;
	uint64_t start;
	uint64_t size;
	uint16_t fnum;
	enum brl_type lock_type;
	void *notify_ptr;
};

struct brl_context {
	struct tdb_wrap *w;
	servid_t server;
	uint16_t tid;
	struct messaging_context *messaging_ctx;
	struct lock_struct last_lock;
};


/*
  Open up the brlock.tdb database. Close it down using
  talloc_free(). We need the messaging_ctx to allow for
  pending lock notifications.
*/
struct brl_context *brl_init(TALLOC_CTX *mem_ctx, servid_t server, uint16_t tid, 
			     struct messaging_context *messaging_ctx)
{
	char *path;
	struct brl_context *brl;

	brl = talloc_p(mem_ctx, struct brl_context);
	if (brl == NULL) {
		return NULL;
	}

	path = smbd_tmp_path(brl, "brlock.tdb");
	brl->w = tdb_wrap_open(brl, path, 0,  
			       TDB_DEFAULT, O_RDWR|O_CREAT, 0600);
	talloc_free(path);
	if (brl->w == NULL) {
		talloc_free(brl);
		return NULL;
	}

	brl->server = server;
	brl->tid = tid;
	brl->messaging_ctx = messaging_ctx;
	ZERO_STRUCT(brl->last_lock);

	return brl;
}


/*
  see if two locking contexts are equal
*/
static BOOL brl_same_context(struct lock_context *ctx1, struct lock_context *ctx2)
{
	return (ctx1->server == ctx2->server &&
		ctx1->smbpid == ctx2->smbpid &&
		ctx1->tid == ctx2->tid);
}

/*
  see if lck1 and lck2 overlap
*/
static BOOL brl_overlap(struct lock_struct *lck1, 
			struct lock_struct *lck2)
{
	/* this extra check is not redundent - it copes with locks
	   that go beyond the end of 64 bit file space */
	if (lck1->size != 0 &&
	    lck1->start == lck2->start &&
	    lck1->size == lck2->size) {
		return True;
	}
	    
	if (lck1->start >= (lck2->start+lck2->size) ||
	    lck2->start >= (lck1->start+lck1->size)) {
		return False;
	}
	return True;
} 

/*
 See if lock2 can be added when lock1 is in place.
*/
static BOOL brl_conflict(struct lock_struct *lck1, 
			 struct lock_struct *lck2)
{
	/* pending locks don't conflict with anything */
	if (lck1->lock_type >= PENDING_READ_LOCK ||
	    lck2->lock_type >= PENDING_READ_LOCK) {
		return False;
	}

	if (lck1->lock_type == READ_LOCK && lck2->lock_type == READ_LOCK) {
		return False;
	}

	if (brl_same_context(&lck1->context, &lck2->context) &&
	    lck2->lock_type == READ_LOCK && lck1->fnum == lck2->fnum) {
		return False;
	}

	return brl_overlap(lck1, lck2);
} 


/*
 Check to see if this lock conflicts, but ignore our own locks on the
 same fnum only.
*/
static BOOL brl_conflict_other(struct lock_struct *lck1, struct lock_struct *lck2)
{
	/* pending locks don't conflict with anything */
	if (lck1->lock_type >= PENDING_READ_LOCK ||
	    lck2->lock_type >= PENDING_READ_LOCK) {
		return False;
	}

	if (lck1->lock_type == READ_LOCK && lck2->lock_type == READ_LOCK) 
		return False;

	/*
	 * note that incoming write calls conflict with existing READ
	 * locks even if the context is the same. JRA. See LOCKTEST7
	 * in smbtorture.
	 */
	if (brl_same_context(&lck1->context, &lck2->context) &&
	    lck1->fnum == lck2->fnum &&
	    (lck2->lock_type == READ_LOCK || lck1->lock_type == WRITE_LOCK)) {
		return False;
	}

	return brl_overlap(lck1, lck2);
} 


/*
  amazingly enough, w2k3 "remembers" whether the last lock failure
  is the same as this one and changes its error code. I wonder if any
  app depends on this?
*/
static NTSTATUS brl_lock_failed(struct brl_context *brl, struct lock_struct *lock)
{
	if (lock->context.server == brl->last_lock.context.server &&
	    lock->context.tid == brl->last_lock.context.tid &&
	    lock->fnum == brl->last_lock.fnum &&
	    lock->start == brl->last_lock.start &&
	    lock->size == brl->last_lock.size) {
		return NT_STATUS_FILE_LOCK_CONFLICT;
	}
	brl->last_lock = *lock;
	if (lock->start >= 0xEF000000 && 
	    (lock->start >> 63) == 0) {
		/* amazing the little things you learn with a test
		   suite. Locks beyond this offset (as a 64 bit
		   number!) always generate the conflict error code,
		   unless the top bit is set */
		return NT_STATUS_FILE_LOCK_CONFLICT;
	}
	return NT_STATUS_LOCK_NOT_GRANTED;
}

/*
  Lock a range of bytes.  The lock_type can be a PENDING_*_LOCK, in
  which case a real lock is first tried, and if that fails then a
  pending lock is created. When the pending lock is triggered (by
  someone else closing an overlapping lock range) a messaging
  notification is sent, identified by the notify_ptr
*/
NTSTATUS brl_lock(struct brl_context *brl,
		  DATA_BLOB *file_key, 
		  uint16_t smbpid,
		  uint16_t fnum, 
		  uint64_t start, uint64_t size, 
		  enum brl_type lock_type,
		  void *notify_ptr)
{
	TDB_DATA kbuf, dbuf;
	int count, i;
	struct lock_struct lock, *locks;
	char *tp;
	NTSTATUS status;

	kbuf.dptr = (char *)file_key->data;
	kbuf.dsize = file_key->length;

	if (tdb_chainlock(brl->w->tdb, kbuf) != 0) {
		return NT_STATUS_INTERNAL_DB_CORRUPTION;
	}

	/* if this is a pending lock, then with the chainlock held we
	   try to get the real lock. If we succeed then we don't need
	   to make it pending. This prevents a possible race condition
	   where the pending lock gets created after the lock that is
	   preventing the real lock gets removed */
	if (lock_type >= PENDING_READ_LOCK) {
		enum brl_type rw = (lock_type==PENDING_READ_LOCK? READ_LOCK : WRITE_LOCK);
		status = brl_lock(brl, file_key, smbpid, fnum, start, size, rw, NULL);
		if (NT_STATUS_IS_OK(status)) {
			tdb_chainunlock(brl->w->tdb, kbuf);
			return NT_STATUS_OK;
		}
	}

	dbuf = tdb_fetch(brl->w->tdb, kbuf);

	lock.context.smbpid = smbpid;
	lock.context.server = brl->server;
	lock.context.tid = brl->tid;
	lock.start = start;
	lock.size = size;
	lock.fnum = fnum;
	lock.lock_type = lock_type;
	lock.notify_ptr = notify_ptr;

	if (dbuf.dptr) {
		/* there are existing locks - make sure they don't conflict */
		locks = (struct lock_struct *)dbuf.dptr;
		count = dbuf.dsize / sizeof(*locks);
		for (i=0; i<count; i++) {
			if (brl_conflict(&locks[i], &lock)) {
				status = brl_lock_failed(brl, &lock);
				goto fail;
			}
		}
	}

	/* no conflicts - add it to the list of locks */
	tp = Realloc(dbuf.dptr, dbuf.dsize + sizeof(*locks));
	if (!tp) {
		status = NT_STATUS_NO_MEMORY;
		goto fail;
	} else {
		dbuf.dptr = tp;
	}
	memcpy(dbuf.dptr + dbuf.dsize, &lock, sizeof(lock));
	dbuf.dsize += sizeof(lock);

	if (tdb_store(brl->w->tdb, kbuf, dbuf, TDB_REPLACE) != 0) {
		status = NT_STATUS_INTERNAL_DB_CORRUPTION;
		goto fail;
	}

	free(dbuf.dptr);
	tdb_chainunlock(brl->w->tdb, kbuf);

	/* the caller needs to know if the real lock was granted. If
	   we have reached here then it must be a pending lock that
	   was granted, so tell them the lock failed */
	if (lock_type >= PENDING_READ_LOCK) {
		return brl_lock_failed(brl, &lock);
	}

	return NT_STATUS_OK;

 fail:

	free(dbuf.dptr);
	tdb_chainunlock(brl->w->tdb, kbuf);
	return status;
}


/*
  we are removing a lock that might be holding up a pending lock. Scan for pending
  locks that cover this range and if we find any then notify the server that it should
  retry the lock
*/
static void brl_notify_unlock(struct brl_context *brl,
			      struct lock_struct *locks, int count, 
			      struct lock_struct *removed_lock)
{
	int i, last_notice;

	/* the last_notice logic is to prevent stampeding on a lock
	   range. It prevents us sending hundreds of notifies on the
	   same range of bytes. It doesn't prevent all possible
	   stampedes, but it does prevent the most common problem */
	last_notice = -1;

	for (i=0;i<count;i++) {
		if (locks[i].lock_type >= PENDING_READ_LOCK &&
		    brl_overlap(&locks[i], removed_lock)) {
			if (last_notice != -1 && brl_overlap(&locks[i], &locks[last_notice])) {
				continue;
			}
			if (locks[i].lock_type == PENDING_WRITE_LOCK) {
				last_notice = i;
			}
			messaging_send_ptr(brl->messaging_ctx, locks[i].context.server, 
					   MSG_BRL_RETRY, locks[i].notify_ptr);
		}
	}
}


/*
  send notifications for all pending locks - the file is being closed by this
  user
*/
static void brl_notify_all(struct brl_context *brl,
			   struct lock_struct *locks, int count)
{
	int i;
	for (i=0;i<count;i++) {
		if (locks->lock_type >= PENDING_READ_LOCK) {
			brl_notify_unlock(brl, locks, count, &locks[i]);
		}
	}
}



/*
 Unlock a range of bytes.
*/
NTSTATUS brl_unlock(struct brl_context *brl,
		    DATA_BLOB *file_key, 
		    uint16_t smbpid,
		    uint16_t fnum,
		    uint64_t start, uint64_t size)
{
	TDB_DATA kbuf, dbuf;
	int count, i;
	struct lock_struct *locks;
	struct lock_context context;
	NTSTATUS status;

	kbuf.dptr = (char *)file_key->data;
	kbuf.dsize = file_key->length;

	if (tdb_chainlock(brl->w->tdb, kbuf) != 0) {
		return NT_STATUS_INTERNAL_DB_CORRUPTION;
	}

	dbuf = tdb_fetch(brl->w->tdb, kbuf);
	if (!dbuf.dptr) {
		tdb_chainunlock(brl->w->tdb, kbuf);
		return NT_STATUS_RANGE_NOT_LOCKED;
	}

	context.smbpid = smbpid;
	context.server = brl->server;
	context.tid = brl->tid;

	/* there are existing locks - find a match */
	locks = (struct lock_struct *)dbuf.dptr;
	count = dbuf.dsize / sizeof(*locks);

	for (i=0; i<count; i++) {
		struct lock_struct *lock = &locks[i];
		
		if (brl_same_context(&lock->context, &context) &&
		    lock->fnum == fnum &&
		    lock->start == start &&
		    lock->size == size &&
		    lock->notify_ptr == NULL) {
			/* found it - delete it */
			if (count == 1) {
				if (tdb_delete(brl->w->tdb, kbuf) != 0) {
					status = NT_STATUS_INTERNAL_DB_CORRUPTION;
					goto fail;
				}
			} else {
				struct lock_struct removed_lock = *lock;
				if (i < count-1) {
					memmove(&locks[i], &locks[i+1], 
						sizeof(*locks)*((count-1) - i));
				}
				count--;

				/* send notifications for any relevant pending locks */
				brl_notify_unlock(brl, locks, count, &removed_lock);

				dbuf.dsize = count * sizeof(*locks);

				if (tdb_store(brl->w->tdb, kbuf, dbuf, TDB_REPLACE) != 0) {
					status = NT_STATUS_INTERNAL_DB_CORRUPTION;
					goto fail;
				}
			}
			
			free(dbuf.dptr);
			tdb_chainunlock(brl->w->tdb, kbuf);
			return NT_STATUS_OK;
		}
	}
	
	/* we didn't find it */
	status = NT_STATUS_RANGE_NOT_LOCKED;

 fail:
	free(dbuf.dptr);
	tdb_chainunlock(brl->w->tdb, kbuf);
	return status;
}


/*
  remove a pending lock. This is called when the caller has either
  given up trying to establish a lock or when they have succeeded in
  getting it. In either case they no longer need to be notified.
*/
NTSTATUS brl_remove_pending(struct brl_context *brl,
			    DATA_BLOB *file_key, 
			    void *notify_ptr)
{
	TDB_DATA kbuf, dbuf;
	int count, i;
	struct lock_struct *locks;
	NTSTATUS status;

	kbuf.dptr = (char *)file_key->data;
	kbuf.dsize = file_key->length;

	if (tdb_chainlock(brl->w->tdb, kbuf) != 0) {
		return NT_STATUS_INTERNAL_DB_CORRUPTION;
	}

	dbuf = tdb_fetch(brl->w->tdb, kbuf);
	if (!dbuf.dptr) {
		tdb_chainunlock(brl->w->tdb, kbuf);
		return NT_STATUS_RANGE_NOT_LOCKED;
	}

	/* there are existing locks - find a match */
	locks = (struct lock_struct *)dbuf.dptr;
	count = dbuf.dsize / sizeof(*locks);

	for (i=0; i<count; i++) {
		struct lock_struct *lock = &locks[i];
		
		if (lock->notify_ptr == notify_ptr &&
		    lock->context.server == brl->server) {
			/* found it - delete it */
			if (count == 1) {
				if (tdb_delete(brl->w->tdb, kbuf) != 0) {
					status = NT_STATUS_INTERNAL_DB_CORRUPTION;
					goto fail;
				}
			} else {
				if (i < count-1) {
					memmove(&locks[i], &locks[i+1], 
						sizeof(*locks)*((count-1) - i));
				}
				count--;
				dbuf.dsize = count * sizeof(*locks);
				if (tdb_store(brl->w->tdb, kbuf, dbuf, TDB_REPLACE) != 0) {
					status = NT_STATUS_INTERNAL_DB_CORRUPTION;
					goto fail;
				}
			}
			
			free(dbuf.dptr);
			tdb_chainunlock(brl->w->tdb, kbuf);
			return NT_STATUS_OK;
		}
	}
	
	/* we didn't find it */
	status = NT_STATUS_RANGE_NOT_LOCKED;

 fail:
	free(dbuf.dptr);
	tdb_chainunlock(brl->w->tdb, kbuf);
	return status;
}


/*
  Test if we are allowed to perform IO on a region of an open file
*/
NTSTATUS brl_locktest(struct brl_context *brl,
		      DATA_BLOB *file_key, 
		      uint16_t fnum,
		      uint16 smbpid, 
		      uint64_t start, uint64_t size, 
		      enum brl_type lock_type)
{
	TDB_DATA kbuf, dbuf;
	int count, i;
	struct lock_struct lock, *locks;

	kbuf.dptr = (char *)file_key->data;
	kbuf.dsize = file_key->length;

	dbuf = tdb_fetch(brl->w->tdb, kbuf);
	if (dbuf.dptr == NULL) {
		return NT_STATUS_OK;
	}

	lock.context.smbpid = smbpid;
	lock.context.server = brl->server;
	lock.context.tid = brl->tid;
	lock.start = start;
	lock.size = size;
	lock.fnum = fnum;
	lock.lock_type = lock_type;

	/* there are existing locks - make sure they don't conflict */
	locks = (struct lock_struct *)dbuf.dptr;
	count = dbuf.dsize / sizeof(*locks);

	for (i=0; i<count; i++) {
		if (brl_conflict_other(&locks[i], &lock)) {
			free(dbuf.dptr);
			return NT_STATUS_FILE_LOCK_CONFLICT;
		}
	}

	free(dbuf.dptr);
	return NT_STATUS_OK;
}


/*
 Remove any locks associated with a open file.
*/
NTSTATUS brl_close(struct brl_context *brl,
		   DATA_BLOB *file_key, int fnum)
{
	TDB_DATA kbuf, dbuf;
	int count, i, dcount=0;
	struct lock_struct *locks;
	NTSTATUS status;

	kbuf.dptr = (char *)file_key->data;
	kbuf.dsize = file_key->length;

	if (tdb_chainlock(brl->w->tdb, kbuf) != 0) {
		return NT_STATUS_INTERNAL_DB_CORRUPTION;
	}

	dbuf = tdb_fetch(brl->w->tdb, kbuf);
	if (!dbuf.dptr) {
		tdb_chainunlock(brl->w->tdb, kbuf);
		return NT_STATUS_OK;
	}

	/* there are existing locks - remove any for this fnum */
	locks = (struct lock_struct *)dbuf.dptr;
	count = dbuf.dsize / sizeof(*locks);

	for (i=0; i<count; i++) {
		struct lock_struct *lock = &locks[i];

		if (lock->context.tid == brl->tid &&
		    lock->context.server == brl->server &&
		    lock->fnum == fnum) {
			/* found it - delete it */
			if (count > 1 && i < count-1) {
				memmove(&locks[i], &locks[i+1], 
					sizeof(*locks)*((count-1) - i));
			}
			count--;
			i--;
			dcount++;
		}
	}

	status = NT_STATUS_OK;

	if (count == 0) {
		if (tdb_delete(brl->w->tdb, kbuf) != 0) {
			status = NT_STATUS_INTERNAL_DB_CORRUPTION;
		}
	} else if (dcount != 0) {
		/* tell all pending lock holders for this file that
		   they have a chance now. This is a bit indiscriminant,
		   but works OK */
		brl_notify_all(brl, locks, count);

		dbuf.dsize = count * sizeof(*locks);

		if (tdb_store(brl->w->tdb, kbuf, dbuf, TDB_REPLACE) != 0) {
			status = NT_STATUS_INTERNAL_DB_CORRUPTION;
		}
	}

	free(dbuf.dptr);
	tdb_chainunlock(brl->w->tdb, kbuf);

	return status;
}