summaryrefslogtreecommitdiffstats
path: root/source3/smbd/oplock_onefs.c
blob: 9c6287ba342651d2e3138b96a33ecc70489c8ea2 (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
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
/*
 * Unix SMB/CIFS implementation.
 * Support for OneFS kernel oplocks
 *
 * Copyright (C) Volker Lendecke 2007
 * Copyright (C) Tim Prouty, 2009
 *
 * 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/>.
 */

#define DBGC_CLASS DBGC_LOCKING

#include "includes.h"

#if HAVE_ONEFS
#include "oplock_onefs.h"
#include "smbd/globals.h"

#include <ifs/ifs_syscalls.h>
#include <isi_ecs/isi_ecs_oplocks.h>
#include <sys/proc.h>

struct onefs_oplocks_context {
	struct kernel_oplocks *ctx;
	const struct oplocks_event_ops *onefs_ops;
	int onefs_event_fd;
	struct fd_event *read_fde;
};

enum onefs_callback_state {
	ONEFS_OPEN_FILE,
	ONEFS_WAITING_FOR_OPLOCK
};

struct onefs_callback_record {
	struct onefs_callback_record *prev, *next;
	uint64_t id;
	enum onefs_callback_state state;
	union {
		files_struct *fsp;	/* ONEFS_OPEN_FILE */
		uint64_t mid;		/* ONEFS_WAITING_FOR_OPLOCK */
	} data;
};

/**
 * Internal list of files (along with additional state) that have outstanding
 * oplocks or requests for oplocks.
 */
struct onefs_callback_record *callback_recs;

/**
 * Convert a onefs_callback_record to a debug string using the dbg_ctx().
 */
const char *onefs_cb_record_str_dbg(const struct onefs_callback_record *r)
{
	char *result;

	if (r == NULL) {
		result = talloc_strdup(talloc_tos(), "NULL callback record");
		return result;
	}

	switch (r->state) {
	case ONEFS_OPEN_FILE:
		result = talloc_asprintf(talloc_tos(), "cb record %llu for "
					 "file %s", r->id,
					 fsp_str_dbg(r->data.fsp));
	case ONEFS_WAITING_FOR_OPLOCK:
		result = talloc_asprintf(talloc_tos(), "cb record %llu for "
					 "pending mid %llu", r->id,
					 (unsigned long long)r->data.mid);
		break;
	default:
		result = talloc_asprintf(talloc_tos(), "cb record %llu unknown "
					 "state %d", r->id, r->state);
		break;
	}

	return result;
}

/**
 * Traverse the list of onefs_callback_records and print all entries.
 */
static void debug_cb_records(const char *fn)
{
	struct onefs_callback_record *rec;

	if (DEBUGLEVEL < 10)
		return;

	DEBUG(10, ("cb records (%s):\n", fn));

	for (rec = callback_recs; rec; rec = rec->next) {
		DEBUGADD(10, ("%s\n", onefs_cb_record_str_dbg(rec)));
	}
}

/**
 * Find a callback record in the list of outstanding oplock operations.
 *
 * Once n ifs_createfile requests an oplock on a file, the kernel communicates
 * with samba via the oplock event channel by sending events that reference an
 * id.  This function maps that id to the onefs_callback_record that was
 * created for it during the initial setup on open (onefs_oplock_wait_record).
 * When a matching id is found in the onefs_callback_record list, the
 * callback_type is checked to make sure the record is in in the correct
 * state.
 */
static struct onefs_callback_record *onefs_find_cb(uint64_t id,
    enum onefs_callback_state expected_state)
{
	struct onefs_callback_record *rec;

	debug_cb_records("onefs_find_cb");

	for (rec = callback_recs; rec; rec = rec->next) {
		if (rec->id == id) {
			DEBUG(10, ("found %s\n",
				   onefs_cb_record_str_dbg(rec)));
			break;
		}
	}

	if (rec == NULL) {
		DEBUG(5, ("Could not find callback record for id %llu\n", id));
		return NULL;
	}

	if (rec->state != expected_state) {
		DEBUG(0, ("Expected cb type %d, got %s", expected_state,
			  onefs_cb_record_str_dbg(rec)));
		SMB_ASSERT(0);
		return NULL;
	}

	return rec;
}

/**
 * Remove and free a callback record from the callback record list.
 */
void destroy_onefs_callback_record(uint64_t id)
{
	struct onefs_callback_record *rec;

	debug_cb_records("destroy_onefs_callback_record");

	if (id == 0) {
		DEBUG(10, ("destroy_onefs_callback_record: Nothing to "
			   "destroy\n"));
		return;
	}

	for (rec = callback_recs; rec; rec = rec->next) {
		if (rec->id == id) {
			DLIST_REMOVE(callback_recs, rec);
			SAFE_FREE(rec);
			DEBUG(10, ("removed cb rec %llu\n", id));
			return;
		}
	}

	DEBUG(0, ("Could not find cb rec %llu to delete", id));
	SMB_ASSERT(0);
}

/**
 * Initialize a callback record and add it to the list of outstanding callback
 * records.
 *
 * This is called in the open path before ifs_createfile so an id can be
 * passed in.  Each callback record can be in one of two states:
 *
 *   1. WAITING_FOR_OPLOCK: This is the initial state for all callback
 *   records.  If ifs_createfile can be completed syncronously without needing
 *   to break any level I oplocks, the state is transitioned to OPEN_FILE.
 *   Otherwise ifs_createfile will finish asynchronously and the open is
 *   deferred.  When the necessary level I opocks have been broken, and the
 *   open can be done, an event is sent by the kernel on the oplock event
 *   channel, which is handled by semlock_available_handler.  At this point
 *   the deferred open is retried.  Unless a level I oplock was acquired by
 *   another client, ifs_createfile will now complete synchronously.
 *
 *   2. OPEN_FILE: Once ifs_createfile completes, the callback record is
 *   transitioned to this state via onefs_set_oplock_callback.
 */
uint64_t onefs_oplock_wait_record(uint64_t mid)
{
	struct onefs_callback_record *result;
	static uint64_t id_generator = 0;

	if (!(result = SMB_MALLOC_P(struct onefs_callback_record))) {
		DEBUG(0, ("talloc failed\n"));
		return 0;
	}

	memset(result, '\0', sizeof(result));

	id_generator += 1;
	if (id_generator == 0) {
		/* Wow, that's a long-running smbd... */
		id_generator += 1;
	}

	result->id = id_generator;

	result->state = ONEFS_WAITING_FOR_OPLOCK;
	result->data.mid = mid;
	DLIST_ADD(callback_recs, result);

	DEBUG(10, ("New cb rec %llu created\n", result->id));

	return result->id;
}

/**
 * Transition the callback record state to OPEN_FILE.
 *
 * This is called after the file is opened and an fsp struct has been
 * allocated.  The mid is dropped in favor of storing the fsp.
 */
void onefs_set_oplock_callback(uint64_t id, files_struct *fsp)
{
	struct onefs_callback_record *cb;
	char *msg;

	DEBUG(10, ("onefs_set_oplock_callback called for cb rec %llu\n", id));

	if (!(cb = onefs_find_cb(id, ONEFS_WAITING_FOR_OPLOCK))) {
		if (asprintf(&msg, "Got invalid callback %lld\n", id) != -1) {
			smb_panic(msg);
		}
		smb_panic("Got invalid callback id\n");
	}

	/*
	 * Paranoia check
	 */
	if (open_was_deferred(cb->data.mid)) {
		if (asprintf(&msg, "Trying to upgrade callback for deferred "
			     "open mid=%llu\n", (unsigned long long)cb->data.mid) != -1) {
			smb_panic(msg);
		}
		smb_panic("Trying to upgrade callback for deferred open "
			  "mid\n");
	}

	cb->state = ONEFS_OPEN_FILE;
	cb->data.fsp = fsp;
}

/**
 * Using a callback record, initialize a share mode entry to pass to
 * share_mode_entry_to_message to send samba IPC messages.
 */
static void init_share_mode_entry(struct share_mode_entry *sme,
				  struct onefs_callback_record *cb,
				  int op_type)
{
	ZERO_STRUCT(*sme);

	sme->pid = procid_self();
	sme->op_type = op_type;
	sme->id = cb->data.fsp->file_id;
	sme->share_file_id = cb->data.fsp->fh->gen_id;
}

/**
 * Callback when a break-to-none event is received from the kernel.
 *
 * On OneFS level 1 oplocks are always broken to level 2 first, therefore an
 * async level 2 break message is always sent when breaking to none.  The
 * downside of this is that OneFS currently has no way to express breaking
 * directly from level 1 to none.
 */
static void oplock_break_to_none_handler(uint64_t id)
{
	struct onefs_callback_record *cb;
	struct share_mode_entry sme;
	char msg[MSG_SMB_SHARE_MODE_ENTRY_SIZE];

	DEBUG(10, ("oplock_break_to_none_handler called for id %llu\n", id));

	if (!(cb = onefs_find_cb(id, ONEFS_OPEN_FILE))) {
		DEBUG(3, ("oplock_break_to_none_handler: could not find "
			  "callback id %llu\n", id));
		return;
	}

	DEBUG(10, ("oplock_break_to_none_handler called for file %s\n",
		   fsp_str_dbg(cb->data.fsp)));

	init_share_mode_entry(&sme, cb, FORCE_OPLOCK_BREAK_TO_NONE);
	share_mode_entry_to_message(msg, &sme);
	messaging_send_buf(smbd_messaging_context(),
			   sme.pid,
			   MSG_SMB_ASYNC_LEVEL2_BREAK,
			   (uint8_t *)msg,
			   MSG_SMB_SHARE_MODE_ENTRY_SIZE);

	/*
	 * We could still receive an OPLOCK_REVOKED message, so keep the
	 * oplock_callback_id around.
	 */
}

/**
 * Callback when a break-to-level2 event is received from the kernel.
 *
 * Breaks from level 1 to level 2.
 */
static void oplock_break_to_level_two_handler(uint64_t id)
{
	struct onefs_callback_record *cb;
	struct share_mode_entry sme;
	char msg[MSG_SMB_SHARE_MODE_ENTRY_SIZE];

	DEBUG(10, ("oplock_break_to_level_two_handler called for id %llu\n",
		   id));

	if (!(cb = onefs_find_cb(id, ONEFS_OPEN_FILE))) {
		DEBUG(3, ("oplock_break_to_level_two_handler: could not find "
			  "callback id %llu\n", id));
		return;
	}

	DEBUG(10, ("oplock_break_to_level_two_handler called for file %s\n",
		   fsp_str_dbg(cb->data.fsp)));

	init_share_mode_entry(&sme, cb, LEVEL_II_OPLOCK);
	share_mode_entry_to_message(msg, &sme);
	messaging_send_buf(smbd_messaging_context(),
			  sme.pid,
			  MSG_SMB_BREAK_REQUEST,
			  (uint8_t *)msg,
			  MSG_SMB_SHARE_MODE_ENTRY_SIZE);

	/*
	 * We could still receive an OPLOCK_REVOKED or OPLOCK_BREAK_TO_NONE
	 * message, so keep the oplock_callback_id around.
	 */
}

/**
 * Revoke an oplock from an unresponsive client.
 *
 * The kernel will send this message when it times out waiting for a level 1
 * oplock break to be acknowledged by the client.  The oplock is then
 * immediately removed.
 */
static void oplock_revoked_handler(uint64_t id)
{
	struct onefs_callback_record *cb;
	files_struct *fsp = NULL;

	DEBUG(10, ("oplock_revoked_handler called for id %llu\n", id));

	if (!(cb = onefs_find_cb(id, ONEFS_OPEN_FILE))) {
		DEBUG(3, ("oplock_revoked_handler: could not find "
			  "callback id %llu\n", id));
		return;
	}

	fsp = cb->data.fsp;

	SMB_ASSERT(fsp->oplock_timeout == NULL);

	DEBUG(0,("Level 1 oplock break failed for file %s. Forcefully "
		 "revoking oplock\n", fsp_str_dbg(fsp)));

	remove_oplock(fsp);

	/*
	 * cb record is cleaned up in fsp ext data destructor on close, so
	 * leave it in the list.
	 */
}

/**
 * Asynchronous ifs_createfile callback
 *
 * If ifs_createfile had to asynchronously break any oplocks, this function is
 * called when the kernel sends an event that the open can be retried.
 */
static void semlock_available_handler(uint64_t id)
{
	struct onefs_callback_record *cb;

	DEBUG(10, ("semlock_available_handler called: %llu\n", id));

	if (!(cb = onefs_find_cb(id, ONEFS_WAITING_FOR_OPLOCK))) {
		DEBUG(5, ("semlock_available_handler: Did not find callback "
			  "%llu\n", id));
		return;
	}

	DEBUG(10, ("Got semlock available for mid %llu\n",
		(unsigned long long)cb->data.mid));

	/* Paranoia check */
	if (!(open_was_deferred(cb->data.mid))) {
		char *msg;
		if (asprintf(&msg, "Semlock available on an open that wasn't "
			     "deferred: %s\n",
			      onefs_cb_record_str_dbg(cb)) != -1) {
			smb_panic(msg);
		}
		smb_panic("Semlock available on an open that wasn't "
			  "deferred\n");
	}

	schedule_deferred_open_smb_message(cb->data.mid);

	/* Cleanup the callback record since the open will be retried. */
	destroy_onefs_callback_record(id);

	return;
}

/**
 * Asynchronous ifs_createfile failure callback
 *
 * If ifs_createfile had to asynchronously break any oplocks, but an error was
 * encountered in the kernel, the open will be retried with the state->failed
 * set to true.  This will prompt the open path to send an INTERNAL_ERROR
 * error message to the client.
 */
static void semlock_async_failure_handler(uint64_t id)
{
	struct onefs_callback_record *cb;
	struct deferred_open_record *state;

	DEBUG(1, ("semlock_async_failure_handler called: %llu\n", id));

	if (!(cb = onefs_find_cb(id, ONEFS_WAITING_FOR_OPLOCK))) {
		DEBUG(5, ("semlock_async_failure_handler: Did not find callback "
			  "%llu\n", id));
		return;
	}

	DEBUG(1, ("Got semlock_async_failure message for mid %llu\n",
		(unsigned long long)cb->data.mid));

	/* Paranoia check */
	if (!(open_was_deferred(cb->data.mid))) {
		char *msg;
		if (asprintf(&msg, "Semlock failure on an open that wasn't "
			     "deferred: %s\n",
			      onefs_cb_record_str_dbg(cb)) != -1) {
			smb_panic(msg);
		}
		smb_panic("Semlock failure on an open that wasn't deferred\n");
	}

	/* Find the actual deferred open record. */
	if (!get_open_deferred_message_state(cb->data.mid, NULL, &state)) {
		DEBUG(0, ("Could not find deferred request for "
			  "mid %d\n", cb->data.mid));
		destroy_onefs_callback_record(id);
		return;
	}

	/* Update to failed so the client can be notified on retried open. */
	state->failed = true;

	/* Schedule deferred open for immediate retry. */
	schedule_deferred_open_smb_message(cb->data.mid);

	/* Cleanup the callback record here since the open will be retried. */
	destroy_onefs_callback_record(id);

	return;
}

/**
 * OneFS acquires all oplocks via ifs_createfile, so this is a no-op.
 */
static bool onefs_set_kernel_oplock(struct kernel_oplocks *_ctx,
				    files_struct *fsp, int oplock_type) {
	return true;
}

/**
 * Release the kernel oplock.
 */
static void onefs_release_kernel_oplock(struct kernel_oplocks *_ctx,
					files_struct *fsp, int oplock_type)
{
	enum oplock_type oplock = onefs_samba_oplock_to_oplock(oplock_type);

	DEBUG(10, ("onefs_release_kernel_oplock: Releasing %s to type %s\n",
		   fsp_str_dbg(fsp), onefs_oplock_str(oplock)));

	if (fsp->fh->fd == -1) {
		DEBUG(1, ("no fd\n"));
		return;
	}

	/* Downgrade oplock to either SHARED or NONE. */
	if (ifs_oplock_downgrade(fsp->fh->fd, oplock)) {
		DEBUG(1,("ifs_oplock_downgrade failed: %s\n",
			 strerror(errno)));
	}
}

/**
 * Wrap ifs_semlock_write so it is only called on operations that aren't
 * already contended in the kernel.
 */
static void onefs_semlock_write(int fd, enum level2_contention_type type,
				enum semlock_operation semlock_op)
{
	int ret;

	switch (type) {
	case LEVEL2_CONTEND_ALLOC_GROW:
	case LEVEL2_CONTEND_POSIX_BRL:
		DEBUG(10, ("Taking %d write semlock for cmd %d on fd: %d\n",
			   semlock_op, type, fd));
		ret = ifs_semlock_write(fd, semlock_op);
		if (ret) {
			DEBUG(0,("ifs_semlock_write failed taking %d write "
				 "semlock for cmd %d on fd: %d: %s",
				 semlock_op, type, fd, strerror(errno)));
		}
		break;
	default:
		DEBUG(10, ("Skipping write semlock for cmd %d on fd: %d\n",
			   type, fd));
	}
}

/**
 * Contend level 2 oplocks in the kernel and smbd.
 *
 * Taking a write semlock will contend all level 2 oplocks in all smbds across
 * the cluster except the fsp's own level 2 oplock.  This lack of
 * self-contention is a limitation of the current OneFS kernel oplocks
 * implementation.  Luckily it is easy to contend our own level 2 oplock by
 * checking the the fsp's oplock_type.  If it's a level2, send a break message
 * to the client and remove the oplock.
 */
static void onefs_contend_level2_oplocks_begin(files_struct *fsp,
					       enum level2_contention_type type)
{
	/* Take care of level 2 kernel contention. */
	onefs_semlock_write(fsp->fh->fd, type, SEMLOCK_LOCK);

	/* Take care of level 2 self contention. */
	if (LEVEL_II_OPLOCK_TYPE(fsp->oplock_type))
		break_level2_to_none_async(fsp);
}

/**
 * Unlock the write semlock when the level 2 contending operation ends.
 */
static void onefs_contend_level2_oplocks_end(files_struct *fsp,
					     enum level2_contention_type type)
{
	/* Take care of level 2 kernel contention. */
	onefs_semlock_write(fsp->fh->fd, type, SEMLOCK_UNLOCK);
}

/**
 * Return string value of onefs oplock types.
 */
const char *onefs_oplock_str(enum oplock_type onefs_oplock_type)
{
	switch (onefs_oplock_type) {
	case OPLOCK_NONE:
		return "OPLOCK_NONE";
	case OPLOCK_EXCLUSIVE:
		return "OPLOCK_EXCLUSIVE";
	case OPLOCK_BATCH:
		return "OPLOCK_BATCH";
	case OPLOCK_SHARED:
		return "OPLOCK_SHARED";
	default:
		break;
	}
	return "UNKNOWN";
}

/**
 * Convert from onefs to samba oplock.
 */
int onefs_oplock_to_samba_oplock(enum oplock_type onefs_oplock)
{
	switch (onefs_oplock) {
	case OPLOCK_NONE:
		return NO_OPLOCK;
	case OPLOCK_EXCLUSIVE:
		return EXCLUSIVE_OPLOCK;
	case OPLOCK_BATCH:
		return BATCH_OPLOCK;
	case OPLOCK_SHARED:
		return LEVEL_II_OPLOCK;
	default:
		DEBUG(0, ("unknown oplock type %d found\n", onefs_oplock));
		break;
	}
	return NO_OPLOCK;
}

/**
 * Convert from samba to onefs oplock.
 */
enum oplock_type onefs_samba_oplock_to_oplock(int samba_oplock_type)
{
	if (BATCH_OPLOCK_TYPE(samba_oplock_type)) return OPLOCK_BATCH;
	if (EXCLUSIVE_OPLOCK_TYPE(samba_oplock_type)) return OPLOCK_EXCLUSIVE;
	if (LEVEL_II_OPLOCK_TYPE(samba_oplock_type)) return OPLOCK_SHARED;
	return OPLOCK_NONE;
}

/**
 * Oplock event handler.
 *
 * Call into the event system dispatcher to handle each event.
 */
static void onefs_oplocks_read_fde_handler(struct event_context *ev,
					   struct fd_event *fde,
					   uint16_t flags,
					   void *private_data)
{
	struct onefs_oplocks_context *ctx =
	    talloc_get_type(private_data, struct onefs_oplocks_context);

	if (oplocks_event_dispatcher(ctx->onefs_ops)) {
		DEBUG(0, ("oplocks_event_dispatcher failed: %s\n",
			  strerror(errno)));
	}
}

/**
 * Setup kernel oplocks
 */
static const struct kernel_oplocks_ops onefs_koplocks_ops = {
	.set_oplock			= onefs_set_kernel_oplock,
	.release_oplock			= onefs_release_kernel_oplock,
	.contend_level2_oplocks_begin	= onefs_contend_level2_oplocks_begin,
	.contend_level2_oplocks_end	= onefs_contend_level2_oplocks_end,
};

static const struct oplocks_event_ops onefs_dispatch_ops = {
	.oplock_break_to_none = oplock_break_to_none_handler,
	.oplock_break_to_level_two = oplock_break_to_level_two_handler,
	.oplock_revoked = oplock_revoked_handler,
	.semlock_available = semlock_available_handler,
	.semlock_async_failure = semlock_async_failure_handler,
};

struct kernel_oplocks *onefs_init_kernel_oplocks(TALLOC_CTX *mem_ctx)
{
	struct kernel_oplocks *_ctx = NULL;
	struct onefs_oplocks_context *ctx = NULL;
        struct procoptions po = PROCOPTIONS_INIT;

	DEBUG(10, ("onefs_init_kernel_oplocks called\n"));

	/* Set the non-blocking proc flag */
	po.po_flags_on |= P_NON_BLOCKING_SEMLOCK;
	if (setprocoptions(&po) != 0) {
		DEBUG(0, ("setprocoptions failed: %s.\n", strerror(errno)));
		return NULL;
	}

	/* Setup the oplock contexts */
	_ctx = talloc_zero(mem_ctx, struct kernel_oplocks);
	if (!_ctx) {
		return NULL;
	}

	ctx = talloc_zero(_ctx, struct onefs_oplocks_context);
	if (!ctx) {
		goto err_out;
	}

	_ctx->ops = &onefs_koplocks_ops;
	_ctx->flags = (KOPLOCKS_LEVEL2_SUPPORTED |
		       KOPLOCKS_DEFERRED_OPEN_NOTIFICATION |
		       KOPLOCKS_TIMEOUT_NOTIFICATION |
		       KOPLOCKS_OPLOCK_BROKEN_NOTIFICATION);
	_ctx->private_data = ctx;
	ctx->ctx = _ctx;
	ctx->onefs_ops = &onefs_dispatch_ops;

	/* Register an kernel event channel for oplocks */
	ctx->onefs_event_fd = oplocks_event_register();
	if (ctx->onefs_event_fd == -1) {
		DEBUG(0, ("oplocks_event_register failed: %s\n",
			   strerror(errno)));
		goto err_out;
	}

	DEBUG(10, ("oplock event_fd = %d\n", ctx->onefs_event_fd));

	/* Register the oplock event_fd with samba's event system */
	ctx->read_fde = event_add_fd(smbd_event_context(),
				     ctx,
				     ctx->onefs_event_fd,
				     EVENT_FD_READ,
				     onefs_oplocks_read_fde_handler,
				     ctx);
	return _ctx;

 err_out:
	talloc_free(_ctx);
	return NULL;
}

#else
 void oplock_onefs_dummy(void);
 void oplock_onefs_dummy(void) {}
#endif /* HAVE_ONEFS */