summaryrefslogtreecommitdiff
path: root/source4/ntvfs/common/opendb_tdb.c
blob: a037e7e47eaf53ea9d6f05789b2f2ebd3ae8dd6b (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
/* 
   Unix SMB/CIFS implementation.

   Copyright (C) Andrew Tridgell 2004
   
   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/>.
*/

/*
  this is the open files database, tdb backend. It implements shared
  storage of what files are open between server instances, and
  implements the rules of shared access to files.

  The caller needs to provide a file_key, which specifies what file
  they are talking about. This needs to be a unique key across all
  filesystems, and is usually implemented in terms of a device/inode
  pair.

  Before any operations can be performed the caller needs to establish
  a lock on the record associated with file_key. That is done by
  calling odb_lock(). The caller releases this lock by calling
  talloc_free() on the returned handle.

  All other operations on a record are done by passing the odb_lock()
  handle back to this module. The handle contains internal
  information about what file_key is being operated on.
*/

#include "includes.h"
#include "system/filesys.h"
#include "lib/tdb/include/tdb.h"
#include "messaging/messaging.h"
#include "db_wrap.h"
#include "lib/messaging/irpc.h"
#include "librpc/gen_ndr/ndr_opendb.h"
#include "ntvfs/ntvfs.h"
#include "ntvfs/common/ntvfs_common.h"
#include "cluster/cluster.h"
#include "param/param.h"

struct odb_context {
	struct tdb_wrap *w;
	struct ntvfs_context *ntvfs_ctx;
	bool oplocks;
};

/*
  an odb lock handle. You must obtain one of these using odb_lock() before doing
  any other operations. 
*/
struct odb_lock {
	struct odb_context *odb;
	TDB_DATA key;
};

/*
  Open up the openfiles.tdb database. Close it down using
  talloc_free(). We need the messaging_ctx to allow for pending open
  notifications.
*/
static struct odb_context *odb_tdb_init(TALLOC_CTX *mem_ctx, 
					struct ntvfs_context *ntvfs_ctx)
{
	struct odb_context *odb;

	odb = talloc(mem_ctx, struct odb_context);
	if (odb == NULL) {
		return NULL;
	}

	odb->w = cluster_tdb_tmp_open(odb, "openfiles.tdb", TDB_DEFAULT);
	if (odb->w == NULL) {
		talloc_free(odb);
		return NULL;
	}

	odb->ntvfs_ctx = ntvfs_ctx;

	/* leave oplocks disabled by default until the code is working */
	odb->oplocks = lp_parm_bool(global_loadparm, NULL, "opendb", "oplocks", false);

	return odb;
}

/*
  destroy a lock on the database
*/
static int odb_lock_destructor(struct odb_lock *lck)
{
	tdb_chainunlock(lck->odb->w->tdb, lck->key);
	return 0;
}

/*
  get a lock on a entry in the odb. This call returns a lock handle,
  which the caller should unlock using talloc_free().
*/
static struct odb_lock *odb_tdb_lock(TALLOC_CTX *mem_ctx,
				     struct odb_context *odb, DATA_BLOB *file_key)
{
	struct odb_lock *lck;

	lck = talloc(mem_ctx, struct odb_lock);
	if (lck == NULL) {
		return NULL;
	}

	lck->odb = talloc_reference(lck, odb);
	lck->key.dptr = talloc_memdup(lck, file_key->data, file_key->length);
	lck->key.dsize = file_key->length;
	if (lck->key.dptr == NULL) {
		talloc_free(lck);
		return NULL;
	}

	if (tdb_chainlock(odb->w->tdb, lck->key) != 0) {
		talloc_free(lck);
		return NULL;
	}

	talloc_set_destructor(lck, odb_lock_destructor);
	
	return lck;
}

/*
  determine if two odb_entry structures conflict

  return NT_STATUS_OK on no conflict
*/
static NTSTATUS share_conflict(struct opendb_entry *e1, struct opendb_entry *e2)
{
	/* if either open involves no read.write or delete access then
	   it can't conflict */
	if (!(e1->access_mask & (SEC_FILE_WRITE_DATA |
				 SEC_FILE_APPEND_DATA |
				 SEC_FILE_READ_DATA |
				 SEC_FILE_EXECUTE |
				 SEC_STD_DELETE))) {
		return NT_STATUS_OK;
	}
	if (!(e2->access_mask & (SEC_FILE_WRITE_DATA |
				 SEC_FILE_APPEND_DATA |
				 SEC_FILE_READ_DATA |
				 SEC_FILE_EXECUTE |
				 SEC_STD_DELETE))) {
		return NT_STATUS_OK;
	}

	/* data IO access masks. This is skipped if the two open handles
	   are on different streams (as in that case the masks don't
	   interact) */
	if (e1->stream_id != e2->stream_id) {
		return NT_STATUS_OK;
	}

#define CHECK_MASK(am, right, sa, share) \
	if (((am) & (right)) && !((sa) & (share))) return NT_STATUS_SHARING_VIOLATION

	CHECK_MASK(e1->access_mask, SEC_FILE_WRITE_DATA | SEC_FILE_APPEND_DATA,
		   e2->share_access, NTCREATEX_SHARE_ACCESS_WRITE);
	CHECK_MASK(e2->access_mask, SEC_FILE_WRITE_DATA | SEC_FILE_APPEND_DATA,
		   e1->share_access, NTCREATEX_SHARE_ACCESS_WRITE);
	
	CHECK_MASK(e1->access_mask, SEC_FILE_READ_DATA | SEC_FILE_EXECUTE,
		   e2->share_access, NTCREATEX_SHARE_ACCESS_READ);
	CHECK_MASK(e2->access_mask, SEC_FILE_READ_DATA | SEC_FILE_EXECUTE,
		   e1->share_access, NTCREATEX_SHARE_ACCESS_READ);

	CHECK_MASK(e1->access_mask, SEC_STD_DELETE,
		   e2->share_access, NTCREATEX_SHARE_ACCESS_DELETE);
	CHECK_MASK(e2->access_mask, SEC_STD_DELETE,
		   e1->share_access, NTCREATEX_SHARE_ACCESS_DELETE);

	return NT_STATUS_OK;
}

/*
  pull a record, translating from the db format to the opendb_file structure defined
  in opendb.idl
*/
static NTSTATUS odb_pull_record(struct odb_lock *lck, struct opendb_file *file)
{
	struct odb_context *odb = lck->odb;
	TDB_DATA dbuf;
	DATA_BLOB blob;
	NTSTATUS status;
		
	dbuf = tdb_fetch(odb->w->tdb, lck->key);
	if (dbuf.dptr == NULL) {
		return NT_STATUS_OBJECT_NAME_NOT_FOUND;
	}

	blob.data = dbuf.dptr;
	blob.length = dbuf.dsize;

	status = ndr_pull_struct_blob(&blob, lck, file, (ndr_pull_flags_fn_t)ndr_pull_opendb_file);

	free(dbuf.dptr);

	return status;
}

/*
  push a record, translating from the opendb_file structure defined in opendb.idl
*/
static NTSTATUS odb_push_record(struct odb_lock *lck, struct opendb_file *file)
{
	struct odb_context *odb = lck->odb;
	TDB_DATA dbuf;
	DATA_BLOB blob;
	NTSTATUS status;
	int ret;

	if (file->num_entries == 0) {
		ret = tdb_delete(odb->w->tdb, lck->key);
		if (ret != 0) {
			return NT_STATUS_INTERNAL_DB_CORRUPTION;
		}
		return NT_STATUS_OK;
	}

	status = ndr_push_struct_blob(&blob, lck, file, (ndr_push_flags_fn_t)ndr_push_opendb_file);
	NT_STATUS_NOT_OK_RETURN(status);

	dbuf.dptr = blob.data;
	dbuf.dsize = blob.length;
		
	ret = tdb_store(odb->w->tdb, lck->key, dbuf, TDB_REPLACE);
	data_blob_free(&blob);
	if (ret != 0) {
		return NT_STATUS_INTERNAL_DB_CORRUPTION;
	}

	return NT_STATUS_OK;
}

/*
  send an oplock break to a client
*/
static NTSTATUS odb_oplock_break_send(struct odb_context *odb, struct opendb_entry *e)
{
	/* tell the server handling this open file about the need to send the client
	   a break */
	return messaging_send_ptr(odb->ntvfs_ctx->msg_ctx, e->server, 
				  MSG_NTVFS_OPLOCK_BREAK, e->file_handle);
}

/*
  register an open file in the open files database. This implements the share_access
  rules

  Note that the path is only used by the delete on close logic, not
  for comparing with other filenames
*/
static NTSTATUS odb_tdb_open_file(struct odb_lock *lck, void *file_handle,
				  uint32_t stream_id, uint32_t share_access, 
				  uint32_t access_mask, bool delete_on_close,
				  const char *path, 
				  uint32_t oplock_level, uint32_t *oplock_granted)
{
	struct odb_context *odb = lck->odb;
	struct opendb_entry e;
	int i;
	struct opendb_file file;
	NTSTATUS status;

	if (odb->oplocks == false) {
		oplock_level = OPLOCK_NONE;
	}

	status = odb_pull_record(lck, &file);
	if (NT_STATUS_EQUAL(status, NT_STATUS_OBJECT_NAME_NOT_FOUND)) {
		/* initialise a blank structure */
		ZERO_STRUCT(file);
		file.path = path;
	} else {
		NT_STATUS_NOT_OK_RETURN(status);
	}

	/* see if it conflicts */
	e.server          = odb->ntvfs_ctx->server_id;
	e.file_handle     = file_handle;
	e.stream_id       = stream_id;
	e.share_access    = share_access;
	e.access_mask     = access_mask;
	e.delete_on_close = delete_on_close;
	e.oplock_level    = OPLOCK_NONE;
		
	/* see if anyone has an oplock, which we need to break */
	for (i=0;i<file.num_entries;i++) {
		if (file.entries[i].oplock_level == OPLOCK_BATCH) {
			/* a batch oplock caches close calls, which
			   means the client application might have
			   already closed the file. We have to allow
			   this close to propogate by sending a oplock
			   break request and suspending this call
			   until the break is acknowledged or the file
			   is closed */
			odb_oplock_break_send(odb, &file.entries[i]);
			return NT_STATUS_OPLOCK_NOT_GRANTED;
		}
	}

	if (file.delete_on_close || 
	    (file.num_entries != 0 && delete_on_close)) {
		/* while delete on close is set, no new opens are allowed */
		return NT_STATUS_DELETE_PENDING;
	}

	/* check for sharing violations */
	for (i=0;i<file.num_entries;i++) {
		status = share_conflict(&file.entries[i], &e);
		NT_STATUS_NOT_OK_RETURN(status);
	}

	/* we now know the open could succeed, but we need to check
	   for any exclusive oplocks. We can't grant a second open
	   till these are broken. Note that we check for batch oplocks
	   before checking for sharing violations, and check for
	   exclusive oplocks afterwards. */
	for (i=0;i<file.num_entries;i++) {
		if (file.entries[i].oplock_level == OPLOCK_EXCLUSIVE) {
			odb_oplock_break_send(odb, &file.entries[i]);
			return NT_STATUS_OPLOCK_NOT_GRANTED;
		}
	}

	/*
	  possibly grant an exclusive or batch oplock if this is the only client
	  with the file open. We don't yet grant levelII oplocks.
	*/
	if (oplock_granted != NULL) {
		if ((oplock_level == OPLOCK_BATCH ||
		     oplock_level == OPLOCK_EXCLUSIVE) &&
		    file.num_entries == 0) {
			(*oplock_granted) = oplock_level;
		} else {
			(*oplock_granted) = OPLOCK_NONE;
		}
		e.oplock_level = (*oplock_granted);
	}

	/* it doesn't conflict, so add it to the end */
	file.entries = talloc_realloc(lck, file.entries, struct opendb_entry, 
				      file.num_entries+1);
	NT_STATUS_HAVE_NO_MEMORY(file.entries);

	file.entries[file.num_entries] = e;
	file.num_entries++;

	return odb_push_record(lck, &file);
}


/*
  register a pending open file in the open files database
*/
static NTSTATUS odb_tdb_open_file_pending(struct odb_lock *lck, void *private)
{
	struct odb_context *odb = lck->odb;
	struct opendb_file file;
	NTSTATUS status;
		
	status = odb_pull_record(lck, &file);
	NT_STATUS_NOT_OK_RETURN(status);

	file.pending = talloc_realloc(lck, file.pending, struct opendb_pending, 
				      file.num_pending+1);
	NT_STATUS_HAVE_NO_MEMORY(file.pending);

	file.pending[file.num_pending].server = odb->ntvfs_ctx->server_id;
	file.pending[file.num_pending].notify_ptr = private;

	file.num_pending++;

	return odb_push_record(lck, &file);
}


/*
  remove a opendb entry
*/
static NTSTATUS odb_tdb_close_file(struct odb_lock *lck, void *file_handle)
{
	struct odb_context *odb = lck->odb;
	struct opendb_file file;
	int i;
	NTSTATUS status;

	status = odb_pull_record(lck, &file);
	NT_STATUS_NOT_OK_RETURN(status);

	/* find the entry, and delete it */
	for (i=0;i<file.num_entries;i++) {
		if (file_handle == file.entries[i].file_handle &&
		    cluster_id_equal(&odb->ntvfs_ctx->server_id, &file.entries[i].server)) {
			if (file.entries[i].delete_on_close) {
				file.delete_on_close = true;
			}
			if (i < file.num_entries-1) {
				memmove(file.entries+i, file.entries+i+1, 
					(file.num_entries - (i+1)) * 
					sizeof(struct opendb_entry));
			}
			break;
		}
	}

	if (i == file.num_entries) {
		return NT_STATUS_UNSUCCESSFUL;
	}

	/* send any pending notifications, removing them once sent */
	for (i=0;i<file.num_pending;i++) {
		messaging_send_ptr(odb->ntvfs_ctx->msg_ctx, file.pending[i].server, 
				   MSG_PVFS_RETRY_OPEN, 
				   file.pending[i].notify_ptr);
	}
	file.num_pending = 0;

	file.num_entries--;
	
	return odb_push_record(lck, &file);
}


/*
  remove a pending opendb entry
*/
static NTSTATUS odb_tdb_remove_pending(struct odb_lock *lck, void *private)
{
	struct odb_context *odb = lck->odb;
	int i;
	NTSTATUS status;
	struct opendb_file file;

	status = odb_pull_record(lck, &file);
	NT_STATUS_NOT_OK_RETURN(status);

	/* find the entry, and delete it */
	for (i=0;i<file.num_pending;i++) {
		if (private == file.pending[i].notify_ptr &&
		    cluster_id_equal(&odb->ntvfs_ctx->server_id, &file.pending[i].server)) {
			if (i < file.num_pending-1) {
				memmove(file.pending+i, file.pending+i+1, 
					(file.num_pending - (i+1)) * 
					sizeof(struct opendb_pending));
			}
			break;
		}
	}

	if (i == file.num_pending) {
		return NT_STATUS_UNSUCCESSFUL;
	}

	file.num_pending--;
	
	return odb_push_record(lck, &file);
}


/*
  rename the path in a open file
*/
static NTSTATUS odb_tdb_rename(struct odb_lock *lck, const char *path)
{
	struct opendb_file file;
	NTSTATUS status;

	status = odb_pull_record(lck, &file);
	if (NT_STATUS_EQUAL(NT_STATUS_OBJECT_NAME_NOT_FOUND, status)) {
		/* not having the record at all is OK */
		return NT_STATUS_OK;
	}
	NT_STATUS_NOT_OK_RETURN(status);

	file.path = path;
	return odb_push_record(lck, &file);
}

/*
  update delete on close flag on an open file
*/
static NTSTATUS odb_tdb_set_delete_on_close(struct odb_lock *lck, bool del_on_close)
{
	NTSTATUS status;
	struct opendb_file file;

	status = odb_pull_record(lck, &file);
	NT_STATUS_NOT_OK_RETURN(status);

	file.delete_on_close = del_on_close;

	return odb_push_record(lck, &file);
}

/*
  return the current value of the delete_on_close bit, and how many
  people still have the file open
*/
static NTSTATUS odb_tdb_get_delete_on_close(struct odb_context *odb, 
					    DATA_BLOB *key, bool *del_on_close, 
					    int *open_count, char **path)
{
	NTSTATUS status;
	struct opendb_file file;
	struct odb_lock *lck;

	lck = odb_lock(odb, odb, key);
	NT_STATUS_HAVE_NO_MEMORY(lck);

	status = odb_pull_record(lck, &file);
	if (NT_STATUS_EQUAL(NT_STATUS_OBJECT_NAME_NOT_FOUND, status)) {
		talloc_free(lck);
		(*del_on_close) = false;
		return NT_STATUS_OK;
	}
	if (!NT_STATUS_IS_OK(status)) {
		talloc_free(lck);
		return status;
	}

	(*del_on_close) = file.delete_on_close;
	if (open_count != NULL) {
		(*open_count) = file.num_entries;
	}
	if (path != NULL) {
		*path = talloc_strdup(odb, file.path);
		NT_STATUS_HAVE_NO_MEMORY(*path);
		if (file.num_entries == 1 && file.entries[0].delete_on_close) {
			(*del_on_close) = true;
		}
	}

	talloc_free(lck);

	return NT_STATUS_OK;
}


/*
  determine if a file can be opened with the given share_access,
  create_options and access_mask
*/
static NTSTATUS odb_tdb_can_open(struct odb_lock *lck,
				 uint32_t share_access, uint32_t create_options, 
				 uint32_t access_mask)
{
	struct odb_context *odb = lck->odb;
	NTSTATUS status;
	struct opendb_file file;
	struct opendb_entry e;
	int i;

	status = odb_pull_record(lck, &file);
	if (NT_STATUS_EQUAL(status, NT_STATUS_OBJECT_NAME_NOT_FOUND)) {
		return NT_STATUS_OK;
	}
	NT_STATUS_NOT_OK_RETURN(status);

	if ((create_options & NTCREATEX_OPTIONS_DELETE_ON_CLOSE) && 
	    file.num_entries != 0) {
		return NT_STATUS_SHARING_VIOLATION;
	}

	if (file.delete_on_close) {
		return NT_STATUS_DELETE_PENDING;
	}

	e.server       = odb->ntvfs_ctx->server_id;
	e.file_handle  = NULL;
	e.stream_id    = 0;
	e.share_access = share_access;
	e.access_mask  = access_mask;
		
	for (i=0;i<file.num_entries;i++) {
		status = share_conflict(&file.entries[i], &e);
		if (!NT_STATUS_IS_OK(status)) {
			/* note that we discard the error code
			   here. We do this as unless we are actually
			   doing an open (which comes via a different
			   function), we need to return a sharing
			   violation */
			return NT_STATUS_SHARING_VIOLATION;
		}
	}

	return NT_STATUS_OK;
}


static const struct opendb_ops opendb_tdb_ops = {
	.odb_init                = odb_tdb_init,
	.odb_lock                = odb_tdb_lock,
	.odb_open_file           = odb_tdb_open_file,
	.odb_open_file_pending   = odb_tdb_open_file_pending,
	.odb_close_file          = odb_tdb_close_file,
	.odb_remove_pending      = odb_tdb_remove_pending,
	.odb_rename              = odb_tdb_rename,
	.odb_set_delete_on_close = odb_tdb_set_delete_on_close,
	.odb_get_delete_on_close = odb_tdb_get_delete_on_close,
	.odb_can_open            = odb_tdb_can_open
};


void odb_tdb_init_ops(void)
{
	odb_set_ops(&opendb_tdb_ops);
}