summaryrefslogtreecommitdiffstats
path: root/fs/ocfs2/dlm/dlmdomain.c
blob: e8ecf8c3dbe7f1382316f768f2366b81acf77c90 (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
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * dlmdomain.c
 *
 * defines domain join / leave apis
 *
 * Copyright (C) 2004 Oracle.  All rights reserved.
 *
 * 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., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/utsname.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/delay.h>
#include <linux/err.h>

#include "cluster/heartbeat.h"
#include "cluster/nodemanager.h"
#include "cluster/tcp.h"

#include "dlmapi.h"
#include "dlmcommon.h"

#include "dlmdomain.h"

#include "dlmver.h"

#define MLOG_MASK_PREFIX (ML_DLM|ML_DLM_DOMAIN)
#include "cluster/masklog.h"

/*
 * ocfs2 node maps are array of long int, which limits to send them freely
 * across the wire due to endianness issues. To workaround this, we convert
 * long ints to byte arrays. Following 3 routines are helper functions to
 * set/test/copy bits within those array of bytes
 */
static inline void byte_set_bit(u8 nr, u8 map[])
{
	map[nr >> 3] |= (1UL << (nr & 7));
}

static inline int byte_test_bit(u8 nr, u8 map[])
{
	return ((1UL << (nr & 7)) & (map[nr >> 3])) != 0;
}

static inline void byte_copymap(u8 dmap[], unsigned long smap[],
			unsigned int sz)
{
	unsigned int nn;

	if (!sz)
		return;

	memset(dmap, 0, ((sz + 7) >> 3));
	for (nn = 0 ; nn < sz; nn++)
		if (test_bit(nn, smap))
			byte_set_bit(nn, dmap);
}

static void dlm_free_pagevec(void **vec, int pages)
{
	while (pages--)
		free_page((unsigned long)vec[pages]);
	kfree(vec);
}

static void **dlm_alloc_pagevec(int pages)
{
	void **vec = kmalloc(pages * sizeof(void *), GFP_KERNEL);
	int i;

	if (!vec)
		return NULL;

	for (i = 0; i < pages; i++)
		if (!(vec[i] = (void *)__get_free_page(GFP_KERNEL)))
			goto out_free;

	mlog(0, "Allocated DLM hash pagevec; %d pages (%lu expected), %lu buckets per page\n",
	     pages, (unsigned long)DLM_HASH_PAGES,
	     (unsigned long)DLM_BUCKETS_PER_PAGE);
	return vec;
out_free:
	dlm_free_pagevec(vec, i);
	return NULL;
}

/*
 *
 * spinlock lock ordering: if multiple locks are needed, obey this ordering:
 *    dlm_domain_lock
 *    struct dlm_ctxt->spinlock
 *    struct dlm_lock_resource->spinlock
 *    struct dlm_ctxt->master_lock
 *    struct dlm_ctxt->ast_lock
 *    dlm_master_list_entry->spinlock
 *    dlm_lock->spinlock
 *
 */

DEFINE_SPINLOCK(dlm_domain_lock);
LIST_HEAD(dlm_domains);
static DECLARE_WAIT_QUEUE_HEAD(dlm_domain_events);

#define DLM_DOMAIN_BACKOFF_MS 200

static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
				  void **ret_data);
static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data,
				     void **ret_data);
static int dlm_cancel_join_handler(struct o2net_msg *msg, u32 len, void *data,
				   void **ret_data);
static int dlm_exit_domain_handler(struct o2net_msg *msg, u32 len, void *data,
				   void **ret_data);

static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm);

void __dlm_unhash_lockres(struct dlm_lock_resource *lockres)
{
	hlist_del_init(&lockres->hash_node);
	dlm_lockres_put(lockres);
}

void __dlm_insert_lockres(struct dlm_ctxt *dlm,
		       struct dlm_lock_resource *res)
{
	struct hlist_head *bucket;
	struct qstr *q;

	assert_spin_locked(&dlm->spinlock);

	q = &res->lockname;
	bucket = dlm_lockres_hash(dlm, q->hash);

	/* get a reference for our hashtable */
	dlm_lockres_get(res);

	hlist_add_head(&res->hash_node, bucket);
}

struct dlm_lock_resource * __dlm_lookup_lockres_full(struct dlm_ctxt *dlm,
						     const char *name,
						     unsigned int len,
						     unsigned int hash)
{
	struct hlist_head *bucket;
	struct hlist_node *list;

	mlog_entry("%.*s\n", len, name);

	assert_spin_locked(&dlm->spinlock);

	bucket = dlm_lockres_hash(dlm, hash);

	hlist_for_each(list, bucket) {
		struct dlm_lock_resource *res = hlist_entry(list,
			struct dlm_lock_resource, hash_node);
		if (res->lockname.name[0] != name[0])
			continue;
		if (unlikely(res->lockname.len != len))
			continue;
		if (memcmp(res->lockname.name + 1, name + 1, len - 1))
			continue;
		dlm_lockres_get(res);
		return res;
	}
	return NULL;
}

/* intended to be called by functions which do not care about lock
 * resources which are being purged (most net _handler functions).
 * this will return NULL for any lock resource which is found but
 * currently in the process of dropping its mastery reference.
 * use __dlm_lookup_lockres_full when you need the lock resource
 * regardless (e.g. dlm_get_lock_resource) */
struct dlm_lock_resource * __dlm_lookup_lockres(struct dlm_ctxt *dlm,
						const char *name,
						unsigned int len,
						unsigned int hash)
{
	struct dlm_lock_resource *res = NULL;

	mlog_entry("%.*s\n", len, name);

	assert_spin_locked(&dlm->spinlock);

	res = __dlm_lookup_lockres_full(dlm, name, len, hash);
	if (res) {
		spin_lock(&res->spinlock);
		if (res->state & DLM_LOCK_RES_DROPPING_REF) {
			spin_unlock(&res->spinlock);
			dlm_lockres_put(res);
			return NULL;
		}
		spin_unlock(&res->spinlock);
	}

	return res;
}

struct dlm_lock_resource * dlm_lookup_lockres(struct dlm_ctxt *dlm,
				    const char *name,
				    unsigned int len)
{
	struct dlm_lock_resource *res;
	unsigned int hash = dlm_lockid_hash(name, len);

	spin_lock(&dlm->spinlock);
	res = __dlm_lookup_lockres(dlm, name, len, hash);
	spin_unlock(&dlm->spinlock);
	return res;
}

static struct dlm_ctxt * __dlm_lookup_domain_full(const char *domain, int len)
{
	struct dlm_ctxt *tmp = NULL;
	struct list_head *iter;

	assert_spin_locked(&dlm_domain_lock);

	/* tmp->name here is always NULL terminated,
	 * but domain may not be! */
	list_for_each(iter, &dlm_domains) {
		tmp = list_entry (iter, struct dlm_ctxt, list);
		if (strlen(tmp->name) == len &&
		    memcmp(tmp->name, domain, len)==0)
			break;
		tmp = NULL;
	}

	return tmp;
}

/* For null terminated domain strings ONLY */
static struct dlm_ctxt * __dlm_lookup_domain(const char *domain)
{
	assert_spin_locked(&dlm_domain_lock);

	return __dlm_lookup_domain_full(domain, strlen(domain));
}


/* returns true on one of two conditions:
 * 1) the domain does not exist
 * 2) the domain exists and it's state is "joined" */
static int dlm_wait_on_domain_helper(const char *domain)
{
	int ret = 0;
	struct dlm_ctxt *tmp = NULL;

	spin_lock(&dlm_domain_lock);

	tmp = __dlm_lookup_domain(domain);
	if (!tmp)
		ret = 1;
	else if (tmp->dlm_state == DLM_CTXT_JOINED)
		ret = 1;

	spin_unlock(&dlm_domain_lock);
	return ret;
}

static void dlm_free_ctxt_mem(struct dlm_ctxt *dlm)
{
	if (dlm->lockres_hash)
		dlm_free_pagevec((void **)dlm->lockres_hash, DLM_HASH_PAGES);

	if (dlm->name)
		kfree(dlm->name);

	kfree(dlm);
}

/* A little strange - this function will be called while holding
 * dlm_domain_lock and is expected to be holding it on the way out. We
 * will however drop and reacquire it multiple times */
static void dlm_ctxt_release(struct kref *kref)
{
	struct dlm_ctxt *dlm;

	dlm = container_of(kref, struct dlm_ctxt, dlm_refs);

	BUG_ON(dlm->num_joins);
	BUG_ON(dlm->dlm_state == DLM_CTXT_JOINED);

	/* we may still be in the list if we hit an error during join. */
	list_del_init(&dlm->list);

	spin_unlock(&dlm_domain_lock);

	mlog(0, "freeing memory from domain %s\n", dlm->name);

	wake_up(&dlm_domain_events);

	dlm_free_ctxt_mem(dlm);

	spin_lock(&dlm_domain_lock);
}

void dlm_put(struct dlm_ctxt *dlm)
{
	spin_lock(&dlm_domain_lock);
	kref_put(&dlm->dlm_refs, dlm_ctxt_release);
	spin_unlock(&dlm_domain_lock);
}

static void __dlm_get(struct dlm_ctxt *dlm)
{
	kref_get(&dlm->dlm_refs);
}

/* given a questionable reference to a dlm object, gets a reference if
 * it can find it in the list, otherwise returns NULL in which case
 * you shouldn't trust your pointer. */
struct dlm_ctxt *dlm_grab(struct dlm_ctxt *dlm)
{
	struct list_head *iter;
	struct dlm_ctxt *target = NULL;

	spin_lock(&dlm_domain_lock);

	list_for_each(iter, &dlm_domains) {
		target = list_entry (iter, struct dlm_ctxt, list);

		if (target == dlm) {
			__dlm_get(target);
			break;
		}

		target = NULL;
	}

	spin_unlock(&dlm_domain_lock);

	return target;
}

int dlm_domain_fully_joined(struct dlm_ctxt *dlm)
{
	int ret;

	spin_lock(&dlm_domain_lock);
	ret = (dlm->dlm_state == DLM_CTXT_JOINED) ||
		(dlm->dlm_state == DLM_CTXT_IN_SHUTDOWN);
	spin_unlock(&dlm_domain_lock);

	return ret;
}

static void dlm_destroy_dlm_worker(struct dlm_ctxt *dlm)
{
	if (dlm->dlm_worker) {
		flush_workqueue(dlm->dlm_worker);
		destroy_workqueue(dlm->dlm_worker);
		dlm->dlm_worker = NULL;
	}
}

static void dlm_complete_dlm_shutdown(struct dlm_ctxt *dlm)
{
	dlm_unregister_domain_handlers(dlm);
	dlm_complete_thread(dlm);
	dlm_complete_recovery_thread(dlm);
	dlm_destroy_dlm_worker(dlm);

	/* We've left the domain. Now we can take ourselves out of the
	 * list and allow the kref stuff to help us free the
	 * memory. */
	spin_lock(&dlm_domain_lock);
	list_del_init(&dlm->list);
	spin_unlock(&dlm_domain_lock);

	/* Wake up anyone waiting for us to remove this domain */
	wake_up(&dlm_domain_events);
}

static int dlm_migrate_all_locks(struct dlm_ctxt *dlm)
{
	int i, num, n, ret = 0;
	struct dlm_lock_resource *res;
	struct hlist_node *iter;
	struct hlist_head *bucket;
	int dropped;

	mlog(0, "Migrating locks from domain %s\n", dlm->name);

	num = 0;
	spin_lock(&dlm->spinlock);
	for (i = 0; i < DLM_HASH_BUCKETS; i++) {
redo_bucket:
		n = 0;
		bucket = dlm_lockres_hash(dlm, i);
		iter = bucket->first;
		while (iter) {
			n++;
			res = hlist_entry(iter, struct dlm_lock_resource,
					  hash_node);
			dlm_lockres_get(res);
			/* migrate, if necessary.  this will drop the dlm
			 * spinlock and retake it if it does migration. */
			dropped = dlm_empty_lockres(dlm, res);

			spin_lock(&res->spinlock);
			__dlm_lockres_calc_usage(dlm, res);
			iter = res->hash_node.next;
			spin_unlock(&res->spinlock);

			dlm_lockres_put(res);

			cond_resched_lock(&dlm->spinlock);

			if (dropped)
				goto redo_bucket;
		}
		num += n;
		mlog(0, "%s: touched %d lockreses in bucket %d "
		     "(tot=%d)\n", dlm->name, n, i, num);
	}
	spin_unlock(&dlm->spinlock);
	wake_up(&dlm->dlm_thread_wq);

	/* let the dlm thread take care of purging, keep scanning until
	 * nothing remains in the hash */
	if (num) {
		mlog(0, "%s: %d lock resources in hash last pass\n",
		     dlm->name, num);
		ret = -EAGAIN;
	}
	mlog(0, "DONE Migrating locks from domain %s\n", dlm->name);
	return ret;
}

static int dlm_no_joining_node(struct dlm_ctxt *dlm)
{
	int ret;

	spin_lock(&dlm->spinlock);
	ret = dlm->joining_node == DLM_LOCK_RES_OWNER_UNKNOWN;
	spin_unlock(&dlm->spinlock);

	return ret;
}

static void dlm_mark_domain_leaving(struct dlm_ctxt *dlm)
{
	/* Yikes, a double spinlock! I need domain_lock for the dlm
	 * state and the dlm spinlock for join state... Sorry! */
again:
	spin_lock(&dlm_domain_lock);
	spin_lock(&dlm->spinlock);

	if (dlm->joining_node != DLM_LOCK_RES_OWNER_UNKNOWN) {
		mlog(0, "Node %d is joining, we wait on it.\n",
			  dlm->joining_node);
		spin_unlock(&dlm->spinlock);
		spin_unlock(&dlm_domain_lock);

		wait_event(dlm->dlm_join_events, dlm_no_joining_node(dlm));
		goto again;
	}

	dlm->dlm_state = DLM_CTXT_LEAVING;
	spin_unlock(&dlm->spinlock);
	spin_unlock(&dlm_domain_lock);
}

static void __dlm_print_nodes(struct dlm_ctxt *dlm)
{
	int node = -1;

	assert_spin_locked(&dlm->spinlock);

	printk(KERN_INFO "ocfs2_dlm: Nodes in domain (\"%s\"): ", dlm->name);

	while ((node = find_next_bit(dlm->domain_map, O2NM_MAX_NODES,
				     node + 1)) < O2NM_MAX_NODES) {
		printk("%d ", node);
	}
	printk("\n");
}

static int dlm_exit_domain_handler(struct o2net_msg *msg, u32 len, void *data,
				   void **ret_data)
{
	struct dlm_ctxt *dlm = data;
	unsigned int node;
	struct dlm_exit_domain *exit_msg = (struct dlm_exit_domain *) msg->buf;

	mlog_entry("%p %u %p", msg, len, data);

	if (!dlm_grab(dlm))
		return 0;

	node = exit_msg->node_idx;

	printk(KERN_INFO "ocfs2_dlm: Node %u leaves domain %s\n", node, dlm->name);

	spin_lock(&dlm->spinlock);
	clear_bit(node, dlm->domain_map);
	__dlm_print_nodes(dlm);

	/* notify anything attached to the heartbeat events */
	dlm_hb_event_notify_attached(dlm, node, 0);

	spin_unlock(&dlm->spinlock);

	dlm_put(dlm);

	return 0;
}

static int dlm_send_one_domain_exit(struct dlm_ctxt *dlm,
				    unsigned int node)
{
	int status;
	struct dlm_exit_domain leave_msg;

	mlog(0, "Asking node %u if we can leave the domain %s me = %u\n",
		  node, dlm->name, dlm->node_num);

	memset(&leave_msg, 0, sizeof(leave_msg));
	leave_msg.node_idx = dlm->node_num;

	status = o2net_send_message(DLM_EXIT_DOMAIN_MSG, dlm->key,
				    &leave_msg, sizeof(leave_msg), node,
				    NULL);

	mlog(0, "status return %d from o2net_send_message\n", status);

	return status;
}


static void dlm_leave_domain(struct dlm_ctxt *dlm)
{
	int node, clear_node, status;

	/* At this point we've migrated away all our locks and won't
	 * accept mastership of new ones. The dlm is responsible for
	 * almost nothing now. We make sure not to confuse any joining
	 * nodes and then commence shutdown procedure. */

	spin_lock(&dlm->spinlock);
	/* Clear ourselves from the domain map */
	clear_bit(dlm->node_num, dlm->domain_map);
	while ((node = find_next_bit(dlm->domain_map, O2NM_MAX_NODES,
				     0)) < O2NM_MAX_NODES) {
		/* Drop the dlm spinlock. This is safe wrt the domain_map.
		 * -nodes cannot be added now as the
		 *   query_join_handlers knows to respond with OK_NO_MAP
		 * -we catch the right network errors if a node is
		 *   removed from the map while we're sending him the
		 *   exit message. */
		spin_unlock(&dlm->spinlock);

		clear_node = 1;

		status = dlm_send_one_domain_exit(dlm, node);
		if (status < 0 &&
		    status != -ENOPROTOOPT &&
		    status != -ENOTCONN) {
			mlog(ML_NOTICE, "Error %d sending domain exit message "
			     "to node %d\n", status, node);

			/* Not sure what to do here but lets sleep for
			 * a bit in case this was a transient
			 * error... */
			msleep(DLM_DOMAIN_BACKOFF_MS);
			clear_node = 0;
		}

		spin_lock(&dlm->spinlock);
		/* If we're not clearing the node bit then we intend
		 * to loop back around to try again. */
		if (clear_node)
			clear_bit(node, dlm->domain_map);
	}
	spin_unlock(&dlm->spinlock);
}

int dlm_joined(struct dlm_ctxt *dlm)
{
	int ret = 0;

	spin_lock(&dlm_domain_lock);

	if (dlm->dlm_state == DLM_CTXT_JOINED)
		ret = 1;

	spin_unlock(&dlm_domain_lock);

	return ret;
}

int dlm_shutting_down(struct dlm_ctxt *dlm)
{
	int ret = 0;

	spin_lock(&dlm_domain_lock);

	if (dlm->dlm_state == DLM_CTXT_IN_SHUTDOWN)
		ret = 1;

	spin_unlock(&dlm_domain_lock);

	return ret;
}

void dlm_unregister_domain(struct dlm_ctxt *dlm)
{
	int leave = 0;

	spin_lock(&dlm_domain_lock);
	BUG_ON(dlm->dlm_state != DLM_CTXT_JOINED);
	BUG_ON(!dlm->num_joins);

	dlm->num_joins--;
	if (!dlm->num_joins) {
		/* We mark it "in shutdown" now so new register
		 * requests wait until we've completely left the
		 * domain. Don't use DLM_CTXT_LEAVING yet as we still
		 * want new domain joins to communicate with us at
		 * least until we've completed migration of our
		 * resources. */
		dlm->dlm_state = DLM_CTXT_IN_SHUTDOWN;
		leave = 1;
	}
	spin_unlock(&dlm_domain_lock);

	if (leave) {
		mlog(0, "shutting down domain %s\n", dlm->name);

		/* We changed dlm state, notify the thread */
		dlm_kick_thread(dlm, NULL);

		while (dlm_migrate_all_locks(dlm)) {
			mlog(0, "%s: more migration to do\n", dlm->name);
		}
		dlm_mark_domain_leaving(dlm);
		dlm_leave_domain(dlm);
		dlm_complete_dlm_shutdown(dlm);
	}
	dlm_put(dlm);
}
EXPORT_SYMBOL_GPL(dlm_unregister_domain);

static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data,
				  void **ret_data)
{
	struct dlm_query_join_request *query;
	enum dlm_query_join_response response;
	struct dlm_ctxt *dlm = NULL;
	u8 nodenum;

	query = (struct dlm_query_join_request *) msg->buf;

	mlog(0, "node %u wants to join domain %s\n", query->node_idx,
		  query->domain);

	/*
	 * If heartbeat doesn't consider the node live, tell it
	 * to back off and try again.  This gives heartbeat a chance
	 * to catch up.
	 */
	if (!o2hb_check_node_heartbeating(query->node_idx)) {
		mlog(0, "node %u is not in our live map yet\n",
		     query->node_idx);

		response = JOIN_DISALLOW;
		goto respond;
	}

	response = JOIN_OK_NO_MAP;

	spin_lock(&dlm_domain_lock);
	dlm = __dlm_lookup_domain_full(query->domain, query->name_len);
	if (!dlm)
		goto unlock_respond;

	/*
	 * There is a small window where the joining node may not see the
	 * node(s) that just left but still part of the cluster. DISALLOW
	 * join request if joining node has different node map.
	 */
	nodenum=0;
	while (nodenum < O2NM_MAX_NODES) {
		if (test_bit(nodenum, dlm->domain_map)) {
			if (!byte_test_bit(nodenum, query->node_map)) {
				mlog(0, "disallow join as node %u does not "
				     "have node %u in its nodemap\n",
				     query->node_idx, nodenum);
				response = JOIN_DISALLOW;
				goto unlock_respond;
			}
		}
		nodenum++;
	}

	/* Once the dlm ctxt is marked as leaving then we don't want
	 * to be put in someone's domain map. 
	 * Also, explicitly disallow joining at certain troublesome
	 * times (ie. during recovery). */
	if (dlm && dlm->dlm_state != DLM_CTXT_LEAVING) {
		int bit = query->node_idx;
		spin_lock(&dlm->spinlock);

		if (dlm->dlm_state == DLM_CTXT_NEW &&
		    dlm->joining_node == DLM_LOCK_RES_OWNER_UNKNOWN) {
			/*If this is a brand new context and we
			 * haven't started our join process yet, then
			 * the other node won the race. */
			response = JOIN_OK_NO_MAP;
		} else if (dlm->joining_node != DLM_LOCK_RES_OWNER_UNKNOWN) {
			/* Disallow parallel joins. */
			response = JOIN_DISALLOW;
		} else if (dlm->reco.state & DLM_RECO_STATE_ACTIVE) {
			mlog(0, "node %u trying to join, but recovery "
			     "is ongoing.\n", bit);
			response = JOIN_DISALLOW;
		} else if (test_bit(bit, dlm->recovery_map)) {
			mlog(0, "node %u trying to join, but it "
			     "still needs recovery.\n", bit);
			response = JOIN_DISALLOW;
		} else if (test_bit(bit, dlm->domain_map)) {
			mlog(0, "node %u trying to join, but it "
			     "is still in the domain! needs recovery?\n",
			     bit);
			response = JOIN_DISALLOW;
		} else {
			/* Alright we're fully a part of this domain
			 * so we keep some state as to who's joining
			 * and indicate to him that needs to be fixed
			 * up. */
			response = JOIN_OK;
			__dlm_set_joining_node(dlm, query->node_idx);
		}

		spin_unlock(&dlm->spinlock);
	}
unlock_respond:
	spin_unlock(&dlm_domain_lock);

respond:
	mlog(0, "We respond with %u\n", response);

	return response;
}

static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data,
				     void **ret_data)
{
	struct dlm_assert_joined *assert;
	struct dlm_ctxt *dlm = NULL;

	assert = (struct dlm_assert_joined *) msg->buf;

	mlog(0, "node %u asserts join on domain %s\n", assert->node_idx,
		  assert->domain);

	spin_lock(&dlm_domain_lock);
	dlm = __dlm_lookup_domain_full(assert->domain, assert->name_len);
	/* XXX should we consider no dlm ctxt an error? */
	if (dlm) {
		spin_lock(&dlm->spinlock);

		/* Alright, this node has officially joined our
		 * domain. Set him in the map and clean up our
		 * leftover join state. */
		BUG_ON(dlm->joining_node != assert->node_idx);
		set_bit(assert->node_idx, dlm->domain_map);
		__dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN);

		printk(KERN_INFO "ocfs2_dlm: Node %u joins domain %s\n",
		       assert->node_idx, dlm->name);
		__dlm_print_nodes(dlm);

		/* notify anything attached to the heartbeat events */
		dlm_hb_event_notify_attached(dlm, assert->node_idx, 1);

		spin_unlock(&dlm->spinlock);
	}
	spin_unlock(&dlm_domain_lock);

	return 0;
}

static int dlm_cancel_join_handler(struct o2net_msg *msg, u32 len, void *data,
				   void **ret_data)
{
	struct dlm_cancel_join *cancel;
	struct dlm_ctxt *dlm = NULL;

	cancel = (struct dlm_cancel_join *) msg->buf;

	mlog(0, "node %u cancels join on domain %s\n", cancel->node_idx,
		  cancel->domain);

	spin_lock(&dlm_domain_lock);
	dlm = __dlm_lookup_domain_full(cancel->domain, cancel->name_len);

	if (dlm) {
		spin_lock(&dlm->spinlock);

		/* Yikes, this guy wants to cancel his join. No
		 * problem, we simply cleanup our join state. */
		BUG_ON(dlm->joining_node != cancel->node_idx);
		__dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN);

		spin_unlock(&dlm->spinlock);
	}
	spin_unlock(&dlm_domain_lock);

	return 0;
}

static int dlm_send_one_join_cancel(struct dlm_ctxt *dlm,
				    unsigned int node)
{
	int status;
	struct dlm_cancel_join cancel_msg;

	memset(&cancel_msg, 0, sizeof(cancel_msg));
	cancel_msg.node_idx = dlm->node_num;
	cancel_msg.name_len = strlen(dlm->name);
	memcpy(cancel_msg.domain, dlm->name, cancel_msg.name_len);

	status = o2net_send_message(DLM_CANCEL_JOIN_MSG, DLM_MOD_KEY,
				    &cancel_msg, sizeof(cancel_msg), node,
				    NULL);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

bail:
	return status;
}

/* map_size should be in bytes. */
static int dlm_send_join_cancels(struct dlm_ctxt *dlm,
				 unsigned long *node_map,
				 unsigned int map_size)
{
	int status, tmpstat;
	unsigned int node;

	if (map_size != (BITS_TO_LONGS(O2NM_MAX_NODES) *
			 sizeof(unsigned long))) {
		mlog(ML_ERROR,
		     "map_size %u != BITS_TO_LONGS(O2NM_MAX_NODES) %u\n",
		     map_size, BITS_TO_LONGS(O2NM_MAX_NODES));
		return -EINVAL;
	}

	status = 0;
	node = -1;
	while ((node = find_next_bit(node_map, O2NM_MAX_NODES,
				     node + 1)) < O2NM_MAX_NODES) {
		if (node == dlm->node_num)
			continue;

		tmpstat = dlm_send_one_join_cancel(dlm, node);
		if (tmpstat) {
			mlog(ML_ERROR, "Error return %d cancelling join on "
			     "node %d\n", tmpstat, node);
			if (!status)
				status = tmpstat;
		}
	}

	if (status)
		mlog_errno(status);
	return status;
}

static int dlm_request_join(struct dlm_ctxt *dlm,
			    int node,
			    enum dlm_query_join_response *response)
{
	int status, retval;
	struct dlm_query_join_request join_msg;

	mlog(0, "querying node %d\n", node);

	memset(&join_msg, 0, sizeof(join_msg));
	join_msg.node_idx = dlm->node_num;
	join_msg.name_len = strlen(dlm->name);
	memcpy(join_msg.domain, dlm->name, join_msg.name_len);

	/* copy live node map to join message */
	byte_copymap(join_msg.node_map, dlm->live_nodes_map, O2NM_MAX_NODES);

	status = o2net_send_message(DLM_QUERY_JOIN_MSG, DLM_MOD_KEY, &join_msg,
				    sizeof(join_msg), node, &retval);
	if (status < 0 && status != -ENOPROTOOPT) {
		mlog_errno(status);
		goto bail;
	}

	/* -ENOPROTOOPT from the net code means the other side isn't
	    listening for our message type -- that's fine, it means
	    his dlm isn't up, so we can consider him a 'yes' but not
	    joined into the domain.  */
	if (status == -ENOPROTOOPT) {
		status = 0;
		*response = JOIN_OK_NO_MAP;
	} else if (retval == JOIN_DISALLOW ||
		   retval == JOIN_OK ||
		   retval == JOIN_OK_NO_MAP) {
		*response = retval;
	} else {
		status = -EINVAL;
		mlog(ML_ERROR, "invalid response %d from node %u\n", retval,
		     node);
	}

	mlog(0, "status %d, node %d response is %d\n", status, node,
		  *response);

bail:
	return status;
}

static int dlm_send_one_join_assert(struct dlm_ctxt *dlm,
				    unsigned int node)
{
	int status;
	struct dlm_assert_joined assert_msg;

	mlog(0, "Sending join assert to node %u\n", node);

	memset(&assert_msg, 0, sizeof(assert_msg));
	assert_msg.node_idx = dlm->node_num;
	assert_msg.name_len = strlen(dlm->name);
	memcpy(assert_msg.domain, dlm->name, assert_msg.name_len);

	status = o2net_send_message(DLM_ASSERT_JOINED_MSG, DLM_MOD_KEY,
				    &assert_msg, sizeof(assert_msg), node,
				    NULL);
	if (status < 0)
		mlog_errno(status);

	return status;
}

static void dlm_send_join_asserts(struct dlm_ctxt *dlm,
				  unsigned long *node_map)
{
	int status, node, live;

	status = 0;
	node = -1;
	while ((node = find_next_bit(node_map, O2NM_MAX_NODES,
				     node + 1)) < O2NM_MAX_NODES) {
		if (node == dlm->node_num)
			continue;

		do {
			/* It is very important that this message be
			 * received so we spin until either the node
			 * has died or it gets the message. */
			status = dlm_send_one_join_assert(dlm, node);

			spin_lock(&dlm->spinlock);
			live = test_bit(node, dlm->live_nodes_map);
			spin_unlock(&dlm->spinlock);

			if (status) {
				mlog(ML_ERROR, "Error return %d asserting "
				     "join on node %d\n", status, node);

				/* give us some time between errors... */
				if (live)
					msleep(DLM_DOMAIN_BACKOFF_MS);
			}
		} while (status && live);
	}
}

struct domain_join_ctxt {
	unsigned long live_map[BITS_TO_LONGS(O2NM_MAX_NODES)];
	unsigned long yes_resp_map[BITS_TO_LONGS(O2NM_MAX_NODES)];
};

static int dlm_should_restart_join(struct dlm_ctxt *dlm,
				   struct domain_join_ctxt *ctxt,
				   enum dlm_query_join_response response)
{
	int ret;

	if (response == JOIN_DISALLOW) {
		mlog(0, "Latest response of disallow -- should restart\n");
		return 1;
	}

	spin_lock(&dlm->spinlock);
	/* For now, we restart the process if the node maps have
	 * changed at all */
	ret = memcmp(ctxt->live_map, dlm->live_nodes_map,
		     sizeof(dlm->live_nodes_map));
	spin_unlock(&dlm->spinlock);

	if (ret)
		mlog(0, "Node maps changed -- should restart\n");

	return ret;
}

static int dlm_try_to_join_domain(struct dlm_ctxt *dlm)
{
	int status = 0, tmpstat, node;
	struct domain_join_ctxt *ctxt;
	enum dlm_query_join_response response;

	mlog_entry("%p", dlm);

	ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
	if (!ctxt) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	/* group sem locking should work for us here -- we're already
	 * registered for heartbeat events so filling this should be
	 * atomic wrt getting those handlers called. */
	o2hb_fill_node_map(dlm->live_nodes_map, sizeof(dlm->live_nodes_map));

	spin_lock(&dlm->spinlock);
	memcpy(ctxt->live_map, dlm->live_nodes_map, sizeof(ctxt->live_map));

	__dlm_set_joining_node(dlm, dlm->node_num);

	spin_unlock(&dlm->spinlock);

	node = -1;
	while ((node = find_next_bit(ctxt->live_map, O2NM_MAX_NODES,
				     node + 1)) < O2NM_MAX_NODES) {
		if (node == dlm->node_num)
			continue;

		status = dlm_request_join(dlm, node, &response);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		/* Ok, either we got a response or the node doesn't have a
		 * dlm up. */
		if (response == JOIN_OK)
			set_bit(node, ctxt->yes_resp_map);

		if (dlm_should_restart_join(dlm, ctxt, response)) {
			status = -EAGAIN;
			goto bail;
		}
	}

	mlog(0, "Yay, done querying nodes!\n");

	/* Yay, everyone agree's we can join the domain. My domain is
	 * comprised of all nodes who were put in the
	 * yes_resp_map. Copy that into our domain map and send a join
	 * assert message to clean up everyone elses state. */
	spin_lock(&dlm->spinlock);
	memcpy(dlm->domain_map, ctxt->yes_resp_map,
	       sizeof(ctxt->yes_resp_map));
	set_bit(dlm->node_num, dlm->domain_map);
	spin_unlock(&dlm->spinlock);

	dlm_send_join_asserts(dlm, ctxt->yes_resp_map);

	/* Joined state *must* be set before the joining node
	 * information, otherwise the query_join handler may read no
	 * current joiner but a state of NEW and tell joining nodes
	 * we're not in the domain. */
	spin_lock(&dlm_domain_lock);
	dlm->dlm_state = DLM_CTXT_JOINED;
	dlm->num_joins++;
	spin_unlock(&dlm_domain_lock);

bail:
	spin_lock(&dlm->spinlock);
	__dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN);
	if (!status)
		__dlm_print_nodes(dlm);
	spin_unlock(&dlm->spinlock);

	if (ctxt) {
		/* Do we need to send a cancel message to any nodes? */
		if (status < 0) {
			tmpstat = dlm_send_join_cancels(dlm,
							ctxt->yes_resp_map,
							sizeof(ctxt->yes_resp_map));
			if (tmpstat < 0)
				mlog_errno(tmpstat);
		}
		kfree(ctxt);
	}

	mlog(0, "returning %d\n", status);
	return status;
}

static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm)
{
	o2hb_unregister_callback(&dlm->dlm_hb_up);
	o2hb_unregister_callback(&dlm->dlm_hb_down);
	o2net_unregister_handler_list(&dlm->dlm_domain_handlers);
}

static int dlm_register_domain_handlers(struct dlm_ctxt *dlm)
{
	int status;

	mlog(0, "registering handlers.\n");

	o2hb_setup_callback(&dlm->dlm_hb_down, O2HB_NODE_DOWN_CB,
			    dlm_hb_node_down_cb, dlm, DLM_HB_NODE_DOWN_PRI);
	status = o2hb_register_callback(&dlm->dlm_hb_down);
	if (status)
		goto bail;

	o2hb_setup_callback(&dlm->dlm_hb_up, O2HB_NODE_UP_CB,
			    dlm_hb_node_up_cb, dlm, DLM_HB_NODE_UP_PRI);
	status = o2hb_register_callback(&dlm->dlm_hb_up);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_MASTER_REQUEST_MSG, dlm->key,
					sizeof(struct dlm_master_request),
					dlm_master_request_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_ASSERT_MASTER_MSG, dlm->key,
					sizeof(struct dlm_assert_master),
					dlm_assert_master_handler,
					dlm, dlm_assert_master_post_handler,
					&dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_CREATE_LOCK_MSG, dlm->key,
					sizeof(struct dlm_create_lock),
					dlm_create_lock_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_CONVERT_LOCK_MSG, dlm->key,
					DLM_CONVERT_LOCK_MAX_LEN,
					dlm_convert_lock_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_UNLOCK_LOCK_MSG, dlm->key,
					DLM_UNLOCK_LOCK_MAX_LEN,
					dlm_unlock_lock_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_PROXY_AST_MSG, dlm->key,
					DLM_PROXY_AST_MAX_LEN,
					dlm_proxy_ast_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_EXIT_DOMAIN_MSG, dlm->key,
					sizeof(struct dlm_exit_domain),
					dlm_exit_domain_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_DEREF_LOCKRES_MSG, dlm->key,
					sizeof(struct dlm_deref_lockres),
					dlm_deref_lockres_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_MIGRATE_REQUEST_MSG, dlm->key,
					sizeof(struct dlm_migrate_request),
					dlm_migrate_request_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_MIG_LOCKRES_MSG, dlm->key,
					DLM_MIG_LOCKRES_MAX_LEN,
					dlm_mig_lockres_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_MASTER_REQUERY_MSG, dlm->key,
					sizeof(struct dlm_master_requery),
					dlm_master_requery_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_LOCK_REQUEST_MSG, dlm->key,
					sizeof(struct dlm_lock_request),
					dlm_request_all_locks_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_RECO_DATA_DONE_MSG, dlm->key,
					sizeof(struct dlm_reco_data_done),
					dlm_reco_data_done_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_BEGIN_RECO_MSG, dlm->key,
					sizeof(struct dlm_begin_reco),
					dlm_begin_reco_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_FINALIZE_RECO_MSG, dlm->key,
					sizeof(struct dlm_finalize_reco),
					dlm_finalize_reco_handler,
					dlm, NULL, &dlm->dlm_domain_handlers);
	if (status)
		goto bail;

bail:
	if (status)
		dlm_unregister_domain_handlers(dlm);

	return status;
}

static int dlm_join_domain(struct dlm_ctxt *dlm)
{
	int status;

	BUG_ON(!dlm);

	mlog(0, "Join domain %s\n", dlm->name);

	status = dlm_register_domain_handlers(dlm);
	if (status) {
		mlog_errno(status);
		goto bail;
	}

	status = dlm_launch_thread(dlm);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	status = dlm_launch_recovery_thread(dlm);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	dlm->dlm_worker = create_singlethread_workqueue("dlm_wq");
	if (!dlm->dlm_worker) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	do {
		unsigned int backoff;
		status = dlm_try_to_join_domain(dlm);

		/* If we're racing another node to the join, then we
		 * need to back off temporarily and let them
		 * complete. */
		if (status == -EAGAIN) {
			if (signal_pending(current)) {
				status = -ERESTARTSYS;
				goto bail;
			}

			/*
			 * <chip> After you!
			 * <dale> No, after you!
			 * <chip> I insist!
			 * <dale> But you first!
			 * ...
			 */
			backoff = (unsigned int)(jiffies & 0x3);
			backoff *= DLM_DOMAIN_BACKOFF_MS;
			mlog(0, "backoff %d\n", backoff);
			msleep(backoff);
		}
	} while (status == -EAGAIN);

	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	status = 0;
bail:
	wake_up(&dlm_domain_events);

	if (status) {
		dlm_unregister_domain_handlers(dlm);
		dlm_complete_thread(dlm);
		dlm_complete_recovery_thread(dlm);
		dlm_destroy_dlm_worker(dlm);
	}

	return status;
}

static struct dlm_ctxt *dlm_alloc_ctxt(const char *domain,
				u32 key)
{
	int i;
	struct dlm_ctxt *dlm = NULL;

	dlm = kzalloc(sizeof(*dlm), GFP_KERNEL);
	if (!dlm) {
		mlog_errno(-ENOMEM);
		goto leave;
	}

	dlm->name = kmalloc(strlen(domain) + 1, GFP_KERNEL);
	if (dlm->name == NULL) {
		mlog_errno(-ENOMEM);
		kfree(dlm);
		dlm = NULL;
		goto leave;
	}

	dlm->lockres_hash = (struct hlist_head **)dlm_alloc_pagevec(DLM_HASH_PAGES);
	if (!dlm->lockres_hash) {
		mlog_errno(-ENOMEM);
		kfree(dlm->name);
		kfree(dlm);
		dlm = NULL;
		goto leave;
	}

	for (i = 0; i < DLM_HASH_BUCKETS; i++)
		INIT_HLIST_HEAD(dlm_lockres_hash(dlm, i));

	strcpy(dlm->name, domain);
	dlm->key = key;
	dlm->node_num = o2nm_this_node();

	spin_lock_init(&dlm->spinlock);
	spin_lock_init(&dlm->master_lock);
	spin_lock_init(&dlm->ast_lock);
	INIT_LIST_HEAD(&dlm->list);
	INIT_LIST_HEAD(&dlm->dirty_list);
	INIT_LIST_HEAD(&dlm->reco.resources);
	INIT_LIST_HEAD(&dlm->reco.received);
	INIT_LIST_HEAD(&dlm->reco.node_data);
	INIT_LIST_HEAD(&dlm->purge_list);
	INIT_LIST_HEAD(&dlm->dlm_domain_handlers);
	dlm->reco.state = 0;

	INIT_LIST_HEAD(&dlm->pending_asts);
	INIT_LIST_HEAD(&dlm->pending_basts);

	mlog(0, "dlm->recovery_map=%p, &(dlm->recovery_map[0])=%p\n",
		  dlm->recovery_map, &(dlm->recovery_map[0]));

	memset(dlm->recovery_map, 0, sizeof(dlm->recovery_map));
	memset(dlm->live_nodes_map, 0, sizeof(dlm->live_nodes_map));
	memset(dlm->domain_map, 0, sizeof(dlm->domain_map));

	dlm->dlm_thread_task = NULL;
	dlm->dlm_reco_thread_task = NULL;
	dlm->dlm_worker = NULL;
	init_waitqueue_head(&dlm->dlm_thread_wq);
	init_waitqueue_head(&dlm->dlm_reco_thread_wq);
	init_waitqueue_head(&dlm->reco.event);
	init_waitqueue_head(&dlm->ast_wq);
	init_waitqueue_head(&dlm->migration_wq);
	INIT_LIST_HEAD(&dlm->master_list);
	INIT_LIST_HEAD(&dlm->mle_hb_events);

	dlm->joining_node = DLM_LOCK_RES_OWNER_UNKNOWN;
	init_waitqueue_head(&dlm->dlm_join_events);

	dlm->reco.new_master = O2NM_INVALID_NODE_NUM;
	dlm->reco.dead_node = O2NM_INVALID_NODE_NUM;
	atomic_set(&dlm->local_resources, 0);
	atomic_set(&dlm->remote_resources, 0);
	atomic_set(&dlm->unknown_resources, 0);

	spin_lock_init(&dlm->work_lock);
	INIT_LIST_HEAD(&dlm->work_list);
	INIT_WORK(&dlm->dispatched_work, dlm_dispatch_work);

	kref_init(&dlm->dlm_refs);
	dlm->dlm_state = DLM_CTXT_NEW;

	INIT_LIST_HEAD(&dlm->dlm_eviction_callbacks);

	mlog(0, "context init: refcount %u\n",
		  atomic_read(&dlm->dlm_refs.refcount));

leave:
	return dlm;
}

/*
 * dlm_register_domain: one-time setup per "domain"
 */
struct dlm_ctxt * dlm_register_domain(const char *domain,
			       u32 key)
{
	int ret;
	struct dlm_ctxt *dlm = NULL;
	struct dlm_ctxt *new_ctxt = NULL;

	if (strlen(domain) > O2NM_MAX_NAME_LEN) {
		ret = -ENAMETOOLONG;
		mlog(ML_ERROR, "domain name length too long\n");
		goto leave;
	}

	if (!o2hb_check_local_node_heartbeating()) {
		mlog(ML_ERROR, "the local node has not been configured, or is "
		     "not heartbeating\n");
		ret = -EPROTO;
		goto leave;
	}

	mlog(0, "register called for domain \"%s\"\n", domain);

retry:
	dlm = NULL;
	if (signal_pending(current)) {
		ret = -ERESTARTSYS;
		mlog_errno(ret);
		goto leave;
	}

	spin_lock(&dlm_domain_lock);

	dlm = __dlm_lookup_domain(domain);
	if (dlm) {
		if (dlm->dlm_state != DLM_CTXT_JOINED) {
			spin_unlock(&dlm_domain_lock);

			mlog(0, "This ctxt is not joined yet!\n");
			wait_event_interruptible(dlm_domain_events,
						 dlm_wait_on_domain_helper(
							 domain));
			goto retry;
		}

		__dlm_get(dlm);
		dlm->num_joins++;

		spin_unlock(&dlm_domain_lock);

		ret = 0;
		goto leave;
	}

	/* doesn't exist */
	if (!new_ctxt) {
		spin_unlock(&dlm_domain_lock);

		new_ctxt = dlm_alloc_ctxt(domain, key);
		if (new_ctxt)
			goto retry;

		ret = -ENOMEM;
		mlog_errno(ret);
		goto leave;
	}

	/* a little variable switch-a-roo here... */
	dlm = new_ctxt;
	new_ctxt = NULL;

	/* add the new domain */
	list_add_tail(&dlm->list, &dlm_domains);
	spin_unlock(&dlm_domain_lock);

	ret = dlm_join_domain(dlm);
	if (ret) {
		mlog_errno(ret);
		dlm_put(dlm);
		goto leave;
	}

	ret = 0;
leave:
	if (new_ctxt)
		dlm_free_ctxt_mem(new_ctxt);

	if (ret < 0)
		dlm = ERR_PTR(ret);

	return dlm;
}
EXPORT_SYMBOL_GPL(dlm_register_domain);

static LIST_HEAD(dlm_join_handlers);

static void dlm_unregister_net_handlers(void)
{
	o2net_unregister_handler_list(&dlm_join_handlers);
}

static int dlm_register_net_handlers(void)
{
	int status = 0;

	status = o2net_register_handler(DLM_QUERY_JOIN_MSG, DLM_MOD_KEY,
					sizeof(struct dlm_query_join_request),
					dlm_query_join_handler,
					NULL, NULL, &dlm_join_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_ASSERT_JOINED_MSG, DLM_MOD_KEY,
					sizeof(struct dlm_assert_joined),
					dlm_assert_joined_handler,
					NULL, NULL, &dlm_join_handlers);
	if (status)
		goto bail;

	status = o2net_register_handler(DLM_CANCEL_JOIN_MSG, DLM_MOD_KEY,
					sizeof(struct dlm_cancel_join),
					dlm_cancel_join_handler,
					NULL, NULL, &dlm_join_handlers);

bail:
	if (status < 0)
		dlm_unregister_net_handlers();

	return status;
}

/* Domain eviction callback handling.
 *
 * The file system requires notification of node death *before* the
 * dlm completes it's recovery work, otherwise it may be able to
 * acquire locks on resources requiring recovery. Since the dlm can
 * evict a node from it's domain *before* heartbeat fires, a similar
 * mechanism is required. */

/* Eviction is not expected to happen often, so a per-domain lock is
 * not necessary. Eviction callbacks are allowed to sleep for short
 * periods of time. */
static DECLARE_RWSEM(dlm_callback_sem);

void dlm_fire_domain_eviction_callbacks(struct dlm_ctxt *dlm,
					int node_num)
{
	struct list_head *iter;
	struct dlm_eviction_cb *cb;

	down_read(&dlm_callback_sem);
	list_for_each(iter, &dlm->dlm_eviction_callbacks) {
		cb = list_entry(iter, struct dlm_eviction_cb, ec_item);

		cb->ec_func(node_num, cb->ec_data);
	}
	up_read(&dlm_callback_sem);
}

void dlm_setup_eviction_cb(struct dlm_eviction_cb *cb,
			   dlm_eviction_func *f,
			   void *data)
{
	INIT_LIST_HEAD(&cb->ec_item);
	cb->ec_func = f;
	cb->ec_data = data;
}
EXPORT_SYMBOL_GPL(dlm_setup_eviction_cb);

void dlm_register_eviction_cb(struct dlm_ctxt *dlm,
			      struct dlm_eviction_cb *cb)
{
	down_write(&dlm_callback_sem);
	list_add_tail(&cb->ec_item, &dlm->dlm_eviction_callbacks);
	up_write(&dlm_callback_sem);
}
EXPORT_SYMBOL_GPL(dlm_register_eviction_cb);

void dlm_unregister_eviction_cb(struct dlm_eviction_cb *cb)
{
	down_write(&dlm_callback_sem);
	list_del_init(&cb->ec_item);
	up_write(&dlm_callback_sem);
}
EXPORT_SYMBOL_GPL(dlm_unregister_eviction_cb);

static int __init dlm_init(void)
{
	int status;

	dlm_print_version();

	status = dlm_init_mle_cache();
	if (status)
		return -1;

	status = dlm_register_net_handlers();
	if (status) {
		dlm_destroy_mle_cache();
		return -1;
	}

	return 0;
}

static void __exit dlm_exit (void)
{
	dlm_unregister_net_handlers();
	dlm_destroy_mle_cache();
}

MODULE_AUTHOR("Oracle");
MODULE_LICENSE("GPL");

module_init(dlm_init);
module_exit(dlm_exit);