summaryrefslogtreecommitdiff
path: root/drivers/infiniband/core/cache.c
blob: 5a76611e684a3f5d7d4d0870b88d7dd190da25a1 (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
/*
 * Copyright (c) 2004 Topspin Communications.  All rights reserved.
 * Copyright (c) 2005 Intel Corporation. All rights reserved.
 * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
 * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/module.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/netdevice.h>
#include <net/addrconf.h>

#include <rdma/ib_cache.h>

#include "core_priv.h"

struct ib_pkey_cache {
	int             table_len;
	u16             table[];
};

struct ib_update_work {
	struct work_struct work;
	struct ib_event event;
	bool enforce_security;
};

union ib_gid zgid;
EXPORT_SYMBOL(zgid);

enum gid_attr_find_mask {
	GID_ATTR_FIND_MASK_GID          = 1UL << 0,
	GID_ATTR_FIND_MASK_NETDEV	= 1UL << 1,
	GID_ATTR_FIND_MASK_DEFAULT	= 1UL << 2,
	GID_ATTR_FIND_MASK_GID_TYPE	= 1UL << 3,
};

enum gid_table_entry_state {
	GID_TABLE_ENTRY_INVALID		= 1,
	GID_TABLE_ENTRY_VALID		= 2,
	/*
	 * Indicates that entry is pending to be removed, there may
	 * be active users of this GID entry.
	 * When last user of the GID entry releases reference to it,
	 * GID entry is detached from the table.
	 */
	GID_TABLE_ENTRY_PENDING_DEL	= 3,
};

struct roce_gid_ndev_storage {
	struct rcu_head rcu_head;
	struct net_device *ndev;
};

struct ib_gid_table_entry {
	struct kref			kref;
	struct work_struct		del_work;
	struct ib_gid_attr		attr;
	void				*context;
	/* Store the ndev pointer to release reference later on in
	 * call_rcu context because by that time gid_table_entry
	 * and attr might be already freed. So keep a copy of it.
	 * ndev_storage is freed by rcu callback.
	 */
	struct roce_gid_ndev_storage	*ndev_storage;
	enum gid_table_entry_state	state;
};

struct ib_gid_table {
	int				sz;
	/* In RoCE, adding a GID to the table requires:
	 * (a) Find if this GID is already exists.
	 * (b) Find a free space.
	 * (c) Write the new GID
	 *
	 * Delete requires different set of operations:
	 * (a) Find the GID
	 * (b) Delete it.
	 *
	 **/
	/* Any writer to data_vec must hold this lock and the write side of
	 * rwlock. Readers must hold only rwlock. All writers must be in a
	 * sleepable context.
	 */
	struct mutex			lock;
	/* rwlock protects data_vec[ix]->state and entry pointer.
	 */
	rwlock_t			rwlock;
	struct ib_gid_table_entry	**data_vec;
	/* bit field, each bit indicates the index of default GID */
	u32				default_gid_indices;
};

static void dispatch_gid_change_event(struct ib_device *ib_dev, u8 port)
{
	struct ib_event event;

	event.device		= ib_dev;
	event.element.port_num	= port;
	event.event		= IB_EVENT_GID_CHANGE;

	ib_dispatch_event_clients(&event);
}

static const char * const gid_type_str[] = {
	[IB_GID_TYPE_IB]	= "IB/RoCE v1",
	[IB_GID_TYPE_ROCE_UDP_ENCAP]	= "RoCE v2",
};

const char *ib_cache_gid_type_str(enum ib_gid_type gid_type)
{
	if (gid_type < ARRAY_SIZE(gid_type_str) && gid_type_str[gid_type])
		return gid_type_str[gid_type];

	return "Invalid GID type";
}
EXPORT_SYMBOL(ib_cache_gid_type_str);

/** rdma_is_zero_gid - Check if given GID is zero or not.
 * @gid:	GID to check
 * Returns true if given GID is zero, returns false otherwise.
 */
bool rdma_is_zero_gid(const union ib_gid *gid)
{
	return !memcmp(gid, &zgid, sizeof(*gid));
}
EXPORT_SYMBOL(rdma_is_zero_gid);

/** is_gid_index_default - Check if a given index belongs to
 * reserved default GIDs or not.
 * @table:	GID table pointer
 * @index:	Index to check in GID table
 * Returns true if index is one of the reserved default GID index otherwise
 * returns false.
 */
static bool is_gid_index_default(const struct ib_gid_table *table,
				 unsigned int index)
{
	return index < 32 && (BIT(index) & table->default_gid_indices);
}

int ib_cache_gid_parse_type_str(const char *buf)
{
	unsigned int i;
	size_t len;
	int err = -EINVAL;

	len = strlen(buf);
	if (len == 0)
		return -EINVAL;

	if (buf[len - 1] == '\n')
		len--;

	for (i = 0; i < ARRAY_SIZE(gid_type_str); ++i)
		if (gid_type_str[i] && !strncmp(buf, gid_type_str[i], len) &&
		    len == strlen(gid_type_str[i])) {
			err = i;
			break;
		}

	return err;
}
EXPORT_SYMBOL(ib_cache_gid_parse_type_str);

static struct ib_gid_table *rdma_gid_table(struct ib_device *device, u8 port)
{
	return device->port_data[port].cache.gid;
}

static bool is_gid_entry_free(const struct ib_gid_table_entry *entry)
{
	return !entry;
}

static bool is_gid_entry_valid(const struct ib_gid_table_entry *entry)
{
	return entry && entry->state == GID_TABLE_ENTRY_VALID;
}

static void schedule_free_gid(struct kref *kref)
{
	struct ib_gid_table_entry *entry =
			container_of(kref, struct ib_gid_table_entry, kref);

	queue_work(ib_wq, &entry->del_work);
}

static void put_gid_ndev(struct rcu_head *head)
{
	struct roce_gid_ndev_storage *storage =
		container_of(head, struct roce_gid_ndev_storage, rcu_head);

	WARN_ON(!storage->ndev);
	/* At this point its safe to release netdev reference,
	 * as all callers working on gid_attr->ndev are done
	 * using this netdev.
	 */
	dev_put(storage->ndev);
	kfree(storage);
}

static void free_gid_entry_locked(struct ib_gid_table_entry *entry)
{
	struct ib_device *device = entry->attr.device;
	u8 port_num = entry->attr.port_num;
	struct ib_gid_table *table = rdma_gid_table(device, port_num);

	dev_dbg(&device->dev, "%s port=%d index=%d gid %pI6\n", __func__,
		port_num, entry->attr.index, entry->attr.gid.raw);

	write_lock_irq(&table->rwlock);

	/*
	 * The only way to avoid overwriting NULL in table is
	 * by comparing if it is same entry in table or not!
	 * If new entry in table is added by the time we free here,
	 * don't overwrite the table entry.
	 */
	if (entry == table->data_vec[entry->attr.index])
		table->data_vec[entry->attr.index] = NULL;
	/* Now this index is ready to be allocated */
	write_unlock_irq(&table->rwlock);

	if (entry->ndev_storage)
		call_rcu(&entry->ndev_storage->rcu_head, put_gid_ndev);
	kfree(entry);
}

static void free_gid_entry(struct kref *kref)
{
	struct ib_gid_table_entry *entry =
			container_of(kref, struct ib_gid_table_entry, kref);

	free_gid_entry_locked(entry);
}

/**
 * free_gid_work - Release reference to the GID entry
 * @work: Work structure to refer to GID entry which needs to be
 * deleted.
 *
 * free_gid_work() frees the entry from the HCA's hardware table
 * if provider supports it. It releases reference to netdevice.
 */
static void free_gid_work(struct work_struct *work)
{
	struct ib_gid_table_entry *entry =
		container_of(work, struct ib_gid_table_entry, del_work);
	struct ib_device *device = entry->attr.device;
	u8 port_num = entry->attr.port_num;
	struct ib_gid_table *table = rdma_gid_table(device, port_num);

	mutex_lock(&table->lock);
	free_gid_entry_locked(entry);
	mutex_unlock(&table->lock);
}

static struct ib_gid_table_entry *
alloc_gid_entry(const struct ib_gid_attr *attr)
{
	struct ib_gid_table_entry *entry;
	struct net_device *ndev;

	entry = kzalloc(sizeof(*entry), GFP_KERNEL);
	if (!entry)
		return NULL;

	ndev = rcu_dereference_protected(attr->ndev, 1);
	if (ndev) {
		entry->ndev_storage = kzalloc(sizeof(*entry->ndev_storage),
					      GFP_KERNEL);
		if (!entry->ndev_storage) {
			kfree(entry);
			return NULL;
		}
		dev_hold(ndev);
		entry->ndev_storage->ndev = ndev;
	}
	kref_init(&entry->kref);
	memcpy(&entry->attr, attr, sizeof(*attr));
	INIT_WORK(&entry->del_work, free_gid_work);
	entry->state = GID_TABLE_ENTRY_INVALID;
	return entry;
}

static void store_gid_entry(struct ib_gid_table *table,
			    struct ib_gid_table_entry *entry)
{
	entry->state = GID_TABLE_ENTRY_VALID;

	dev_dbg(&entry->attr.device->dev, "%s port=%d index=%d gid %pI6\n",
		__func__, entry->attr.port_num, entry->attr.index,
		entry->attr.gid.raw);

	lockdep_assert_held(&table->lock);
	write_lock_irq(&table->rwlock);
	table->data_vec[entry->attr.index] = entry;
	write_unlock_irq(&table->rwlock);
}

static void get_gid_entry(struct ib_gid_table_entry *entry)
{
	kref_get(&entry->kref);
}

static void put_gid_entry(struct ib_gid_table_entry *entry)
{
	kref_put(&entry->kref, schedule_free_gid);
}

static void put_gid_entry_locked(struct ib_gid_table_entry *entry)
{
	kref_put(&entry->kref, free_gid_entry);
}

static int add_roce_gid(struct ib_gid_table_entry *entry)
{
	const struct ib_gid_attr *attr = &entry->attr;
	int ret;

	if (!attr->ndev) {
		dev_err(&attr->device->dev, "%s NULL netdev port=%d index=%d\n",
			__func__, attr->port_num, attr->index);
		return -EINVAL;
	}
	if (rdma_cap_roce_gid_table(attr->device, attr->port_num)) {
		ret = attr->device->ops.add_gid(attr, &entry->context);
		if (ret) {
			dev_err(&attr->device->dev,
				"%s GID add failed port=%d index=%d\n",
				__func__, attr->port_num, attr->index);
			return ret;
		}
	}
	return 0;
}

/**
 * del_gid - Delete GID table entry
 *
 * @ib_dev:	IB device whose GID entry to be deleted
 * @port:	Port number of the IB device
 * @table:	GID table of the IB device for a port
 * @ix:		GID entry index to delete
 *
 */
static void del_gid(struct ib_device *ib_dev, u8 port,
		    struct ib_gid_table *table, int ix)
{
	struct roce_gid_ndev_storage *ndev_storage;
	struct ib_gid_table_entry *entry;

	lockdep_assert_held(&table->lock);

	dev_dbg(&ib_dev->dev, "%s port=%d index=%d gid %pI6\n", __func__, port,
		ix, table->data_vec[ix]->attr.gid.raw);

	write_lock_irq(&table->rwlock);
	entry = table->data_vec[ix];
	entry->state = GID_TABLE_ENTRY_PENDING_DEL;
	/*
	 * For non RoCE protocol, GID entry slot is ready to use.
	 */
	if (!rdma_protocol_roce(ib_dev, port))
		table->data_vec[ix] = NULL;
	write_unlock_irq(&table->rwlock);

	ndev_storage = entry->ndev_storage;
	if (ndev_storage) {
		entry->ndev_storage = NULL;
		rcu_assign_pointer(entry->attr.ndev, NULL);
		call_rcu(&ndev_storage->rcu_head, put_gid_ndev);
	}

	if (rdma_cap_roce_gid_table(ib_dev, port))
		ib_dev->ops.del_gid(&entry->attr, &entry->context);

	put_gid_entry_locked(entry);
}

/**
 * add_modify_gid - Add or modify GID table entry
 *
 * @table:	GID table in which GID to be added or modified
 * @attr:	Attributes of the GID
 *
 * Returns 0 on success or appropriate error code. It accepts zero
 * GID addition for non RoCE ports for HCA's who report them as valid
 * GID. However such zero GIDs are not added to the cache.
 */
static int add_modify_gid(struct ib_gid_table *table,
			  const struct ib_gid_attr *attr)
{
	struct ib_gid_table_entry *entry;
	int ret = 0;

	/*
	 * Invalidate any old entry in the table to make it safe to write to
	 * this index.
	 */
	if (is_gid_entry_valid(table->data_vec[attr->index]))
		del_gid(attr->device, attr->port_num, table, attr->index);

	/*
	 * Some HCA's report multiple GID entries with only one valid GID, and
	 * leave other unused entries as the zero GID. Convert zero GIDs to
	 * empty table entries instead of storing them.
	 */
	if (rdma_is_zero_gid(&attr->gid))
		return 0;

	entry = alloc_gid_entry(attr);
	if (!entry)
		return -ENOMEM;

	if (rdma_protocol_roce(attr->device, attr->port_num)) {
		ret = add_roce_gid(entry);
		if (ret)
			goto done;
	}

	store_gid_entry(table, entry);
	return 0;

done:
	put_gid_entry(entry);
	return ret;
}

/* rwlock should be read locked, or lock should be held */
static int find_gid(struct ib_gid_table *table, const union ib_gid *gid,
		    const struct ib_gid_attr *val, bool default_gid,
		    unsigned long mask, int *pempty)
{
	int i = 0;
	int found = -1;
	int empty = pempty ? -1 : 0;

	while (i < table->sz && (found < 0 || empty < 0)) {
		struct ib_gid_table_entry *data = table->data_vec[i];
		struct ib_gid_attr *attr;
		int curr_index = i;

		i++;

		/* find_gid() is used during GID addition where it is expected
		 * to return a free entry slot which is not duplicate.
		 * Free entry slot is requested and returned if pempty is set,
		 * so lookup free slot only if requested.
		 */
		if (pempty && empty < 0) {
			if (is_gid_entry_free(data) &&
			    default_gid ==
				is_gid_index_default(table, curr_index)) {
				/*
				 * Found an invalid (free) entry; allocate it.
				 * If default GID is requested, then our
				 * found slot must be one of the DEFAULT
				 * reserved slots or we fail.
				 * This ensures that only DEFAULT reserved
				 * slots are used for default property GIDs.
				 */
				empty = curr_index;
			}
		}

		/*
		 * Additionally find_gid() is used to find valid entry during
		 * lookup operation; so ignore the entries which are marked as
		 * pending for removal and the entries which are marked as
		 * invalid.
		 */
		if (!is_gid_entry_valid(data))
			continue;

		if (found >= 0)
			continue;

		attr = &data->attr;
		if (mask & GID_ATTR_FIND_MASK_GID_TYPE &&
		    attr->gid_type != val->gid_type)
			continue;

		if (mask & GID_ATTR_FIND_MASK_GID &&
		    memcmp(gid, &data->attr.gid, sizeof(*gid)))
			continue;

		if (mask & GID_ATTR_FIND_MASK_NETDEV &&
		    attr->ndev != val->ndev)
			continue;

		if (mask & GID_ATTR_FIND_MASK_DEFAULT &&
		    is_gid_index_default(table, curr_index) != default_gid)
			continue;

		found = curr_index;
	}

	if (pempty)
		*pempty = empty;

	return found;
}

static void make_default_gid(struct  net_device *dev, union ib_gid *gid)
{
	gid->global.subnet_prefix = cpu_to_be64(0xfe80000000000000LL);
	addrconf_ifid_eui48(&gid->raw[8], dev);
}

static int __ib_cache_gid_add(struct ib_device *ib_dev, u8 port,
			      union ib_gid *gid, struct ib_gid_attr *attr,
			      unsigned long mask, bool default_gid)
{
	struct ib_gid_table *table;
	int ret = 0;
	int empty;
	int ix;

	/* Do not allow adding zero GID in support of
	 * IB spec version 1.3 section 4.1.1 point (6) and
	 * section 12.7.10 and section 12.7.20
	 */
	if (rdma_is_zero_gid(gid))
		return -EINVAL;

	table = rdma_gid_table(ib_dev, port);

	mutex_lock(&table->lock);

	ix = find_gid(table, gid, attr, default_gid, mask, &empty);
	if (ix >= 0)
		goto out_unlock;

	if (empty < 0) {
		ret = -ENOSPC;
		goto out_unlock;
	}
	attr->device = ib_dev;
	attr->index = empty;
	attr->port_num = port;
	attr->gid = *gid;
	ret = add_modify_gid(table, attr);
	if (!ret)
		dispatch_gid_change_event(ib_dev, port);

out_unlock:
	mutex_unlock(&table->lock);
	if (ret)
		pr_warn("%s: unable to add gid %pI6 error=%d\n",
			__func__, gid->raw, ret);
	return ret;
}

int ib_cache_gid_add(struct ib_device *ib_dev, u8 port,
		     union ib_gid *gid, struct ib_gid_attr *attr)
{
	unsigned long mask = GID_ATTR_FIND_MASK_GID |
			     GID_ATTR_FIND_MASK_GID_TYPE |
			     GID_ATTR_FIND_MASK_NETDEV;

	return __ib_cache_gid_add(ib_dev, port, gid, attr, mask, false);
}

static int
_ib_cache_gid_del(struct ib_device *ib_dev, u8 port,
		  union ib_gid *gid, struct ib_gid_attr *attr,
		  unsigned long mask, bool default_gid)
{
	struct ib_gid_table *table;
	int ret = 0;
	int ix;

	table = rdma_gid_table(ib_dev, port);

	mutex_lock(&table->lock);

	ix = find_gid(table, gid, attr, default_gid, mask, NULL);
	if (ix < 0) {
		ret = -EINVAL;
		goto out_unlock;
	}

	del_gid(ib_dev, port, table, ix);
	dispatch_gid_change_event(ib_dev, port);

out_unlock:
	mutex_unlock(&table->lock);
	if (ret)
		pr_debug("%s: can't delete gid %pI6 error=%d\n",
			 __func__, gid->raw, ret);
	return ret;
}

int ib_cache_gid_del(struct ib_device *ib_dev, u8 port,
		     union ib_gid *gid, struct ib_gid_attr *attr)
{
	unsigned long mask = GID_ATTR_FIND_MASK_GID	  |
			     GID_ATTR_FIND_MASK_GID_TYPE |
			     GID_ATTR_FIND_MASK_DEFAULT  |
			     GID_ATTR_FIND_MASK_NETDEV;

	return _ib_cache_gid_del(ib_dev, port, gid, attr, mask, false);
}

int ib_cache_gid_del_all_netdev_gids(struct ib_device *ib_dev, u8 port,
				     struct net_device *ndev)
{
	struct ib_gid_table *table;
	int ix;
	bool deleted = false;

	table = rdma_gid_table(ib_dev, port);

	mutex_lock(&table->lock);

	for (ix = 0; ix < table->sz; ix++) {
		if (is_gid_entry_valid(table->data_vec[ix]) &&
		    table->data_vec[ix]->attr.ndev == ndev) {
			del_gid(ib_dev, port, table, ix);
			deleted = true;
		}
	}

	mutex_unlock(&table->lock);

	if (deleted)
		dispatch_gid_change_event(ib_dev, port);

	return 0;
}

/**
 * rdma_find_gid_by_port - Returns the GID entry attributes when it finds
 * a valid GID entry for given search parameters. It searches for the specified
 * GID value in the local software cache.
 * @device: The device to query.
 * @gid: The GID value to search for.
 * @gid_type: The GID type to search for.
 * @port_num: The port number of the device where the GID value should be
 *   searched.
 * @ndev: In RoCE, the net device of the device. NULL means ignore.
 *
 * Returns sgid attributes if the GID is found with valid reference or
 * returns ERR_PTR for the error.
 * The caller must invoke rdma_put_gid_attr() to release the reference.
 */
const struct ib_gid_attr *
rdma_find_gid_by_port(struct ib_device *ib_dev,
		      const union ib_gid *gid,
		      enum ib_gid_type gid_type,
		      u8 port, struct net_device *ndev)
{
	int local_index;
	struct ib_gid_table *table;
	unsigned long mask = GID_ATTR_FIND_MASK_GID |
			     GID_ATTR_FIND_MASK_GID_TYPE;
	struct ib_gid_attr val = {.ndev = ndev, .gid_type = gid_type};
	const struct ib_gid_attr *attr;
	unsigned long flags;

	if (!rdma_is_port_valid(ib_dev, port))
		return ERR_PTR(-ENOENT);

	table = rdma_gid_table(ib_dev, port);

	if (ndev)
		mask |= GID_ATTR_FIND_MASK_NETDEV;

	read_lock_irqsave(&table->rwlock, flags);
	local_index = find_gid(table, gid, &val, false, mask, NULL);
	if (local_index >= 0) {
		get_gid_entry(table->data_vec[local_index]);
		attr = &table->data_vec[local_index]->attr;
		read_unlock_irqrestore(&table->rwlock, flags);
		return attr;
	}

	read_unlock_irqrestore(&table->rwlock, flags);
	return ERR_PTR(-ENOENT);
}
EXPORT_SYMBOL(rdma_find_gid_by_port);

/**
 * rdma_find_gid_by_filter - Returns the GID table attribute where a
 * specified GID value occurs
 * @device: The device to query.
 * @gid: The GID value to search for.
 * @port: The port number of the device where the GID value could be
 *   searched.
 * @filter: The filter function is executed on any matching GID in the table.
 *   If the filter function returns true, the corresponding index is returned,
 *   otherwise, we continue searching the GID table. It's guaranteed that
 *   while filter is executed, ndev field is valid and the structure won't
 *   change. filter is executed in an atomic context. filter must not be NULL.
 *
 * rdma_find_gid_by_filter() searches for the specified GID value
 * of which the filter function returns true in the port's GID table.
 *
 */
const struct ib_gid_attr *rdma_find_gid_by_filter(
	struct ib_device *ib_dev, const union ib_gid *gid, u8 port,
	bool (*filter)(const union ib_gid *gid, const struct ib_gid_attr *,
		       void *),
	void *context)
{
	const struct ib_gid_attr *res = ERR_PTR(-ENOENT);
	struct ib_gid_table *table;
	unsigned long flags;
	unsigned int i;

	if (!rdma_is_port_valid(ib_dev, port))
		return ERR_PTR(-EINVAL);

	table = rdma_gid_table(ib_dev, port);

	read_lock_irqsave(&table->rwlock, flags);
	for (i = 0; i < table->sz; i++) {
		struct ib_gid_table_entry *entry = table->data_vec[i];

		if (!is_gid_entry_valid(entry))
			continue;

		if (memcmp(gid, &entry->attr.gid, sizeof(*gid)))
			continue;

		if (filter(gid, &entry->attr, context)) {
			get_gid_entry(entry);
			res = &entry->attr;
			break;
		}
	}
	read_unlock_irqrestore(&table->rwlock, flags);
	return res;
}

static struct ib_gid_table *alloc_gid_table(int sz)
{
	struct ib_gid_table *table = kzalloc(sizeof(*table), GFP_KERNEL);

	if (!table)
		return NULL;

	table->data_vec = kcalloc(sz, sizeof(*table->data_vec), GFP_KERNEL);
	if (!table->data_vec)
		goto err_free_table;

	mutex_init(&table->lock);

	table->sz = sz;
	rwlock_init(&table->rwlock);
	return table;

err_free_table:
	kfree(table);
	return NULL;
}

static void release_gid_table(struct ib_device *device,
			      struct ib_gid_table *table)
{
	bool leak = false;
	int i;

	if (!table)
		return;

	for (i = 0; i < table->sz; i++) {
		if (is_gid_entry_free(table->data_vec[i]))
			continue;
		if (kref_read(&table->data_vec[i]->kref) > 1) {
			dev_err(&device->dev,
				"GID entry ref leak for index %d ref=%d\n", i,
				kref_read(&table->data_vec[i]->kref));
			leak = true;
		}
	}
	if (leak)
		return;

	mutex_destroy(&table->lock);
	kfree(table->data_vec);
	kfree(table);
}

static void cleanup_gid_table_port(struct ib_device *ib_dev, u8 port,
				   struct ib_gid_table *table)
{
	int i;

	if (!table)
		return;

	mutex_lock(&table->lock);
	for (i = 0; i < table->sz; ++i) {
		if (is_gid_entry_valid(table->data_vec[i]))
			del_gid(ib_dev, port, table, i);
	}
	mutex_unlock(&table->lock);
}

void ib_cache_gid_set_default_gid(struct ib_device *ib_dev, u8 port,
				  struct net_device *ndev,
				  unsigned long gid_type_mask,
				  enum ib_cache_gid_default_mode mode)
{
	union ib_gid gid = { };
	struct ib_gid_attr gid_attr;
	unsigned int gid_type;
	unsigned long mask;

	mask = GID_ATTR_FIND_MASK_GID_TYPE |
	       GID_ATTR_FIND_MASK_DEFAULT |
	       GID_ATTR_FIND_MASK_NETDEV;
	memset(&gid_attr, 0, sizeof(gid_attr));
	gid_attr.ndev = ndev;

	for (gid_type = 0; gid_type < IB_GID_TYPE_SIZE; ++gid_type) {
		if (1UL << gid_type & ~gid_type_mask)
			continue;

		gid_attr.gid_type = gid_type;

		if (mode == IB_CACHE_GID_DEFAULT_MODE_SET) {
			make_default_gid(ndev, &gid);
			__ib_cache_gid_add(ib_dev, port, &gid,
					   &gid_attr, mask, true);
		} else if (mode == IB_CACHE_GID_DEFAULT_MODE_DELETE) {
			_ib_cache_gid_del(ib_dev, port, &gid,
					  &gid_attr, mask, true);
		}
	}
}

static void gid_table_reserve_default(struct ib_device *ib_dev, u8 port,
				      struct ib_gid_table *table)
{
	unsigned int i;
	unsigned long roce_gid_type_mask;
	unsigned int num_default_gids;

	roce_gid_type_mask = roce_gid_type_mask_support(ib_dev, port);
	num_default_gids = hweight_long(roce_gid_type_mask);
	/* Reserve starting indices for default GIDs */
	for (i = 0; i < num_default_gids && i < table->sz; i++)
		table->default_gid_indices |= BIT(i);
}


static void gid_table_release_one(struct ib_device *ib_dev)
{
	unsigned int p;

	rdma_for_each_port (ib_dev, p) {
		release_gid_table(ib_dev, ib_dev->port_data[p].cache.gid);
		ib_dev->port_data[p].cache.gid = NULL;
	}
}

static int _gid_table_setup_one(struct ib_device *ib_dev)
{
	struct ib_gid_table *table;
	unsigned int rdma_port;

	rdma_for_each_port (ib_dev, rdma_port) {
		table = alloc_gid_table(
			ib_dev->port_data[rdma_port].immutable.gid_tbl_len);
		if (!table)
			goto rollback_table_setup;

		gid_table_reserve_default(ib_dev, rdma_port, table);
		ib_dev->port_data[rdma_port].cache.gid = table;
	}
	return 0;

rollback_table_setup:
	gid_table_release_one(ib_dev);
	return -ENOMEM;
}

static void gid_table_cleanup_one(struct ib_device *ib_dev)
{
	unsigned int p;

	rdma_for_each_port (ib_dev, p)
		cleanup_gid_table_port(ib_dev, p,
				       ib_dev->port_data[p].cache.gid);
}

static int gid_table_setup_one(struct ib_device *ib_dev)
{
	int err;

	err = _gid_table_setup_one(ib_dev);

	if (err)
		return err;

	rdma_roce_rescan_device(ib_dev);

	return err;
}

/**
 * rdma_query_gid - Read the GID content from the GID software cache
 * @device:		Device to query the GID
 * @port_num:		Port number of the device
 * @index:		Index of the GID table entry to read
 * @gid:		Pointer to GID where to store the entry's GID
 *
 * rdma_query_gid() only reads the GID entry content for requested device,
 * port and index. It reads for IB, RoCE and iWarp link layers.  It doesn't
 * hold any reference to the GID table entry in the HCA or software cache.
 *
 * Returns 0 on success or appropriate error code.
 *
 */
int rdma_query_gid(struct ib_device *device, u8 port_num,
		   int index, union ib_gid *gid)
{
	struct ib_gid_table *table;
	unsigned long flags;
	int res = -EINVAL;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	table = rdma_gid_table(device, port_num);
	read_lock_irqsave(&table->rwlock, flags);

	if (index < 0 || index >= table->sz ||
	    !is_gid_entry_valid(table->data_vec[index]))
		goto done;

	memcpy(gid, &table->data_vec[index]->attr.gid, sizeof(*gid));
	res = 0;

done:
	read_unlock_irqrestore(&table->rwlock, flags);
	return res;
}
EXPORT_SYMBOL(rdma_query_gid);

/**
 * rdma_read_gid_hw_context - Read the HW GID context from GID attribute
 * @attr:		Potinter to the GID attribute
 *
 * rdma_read_gid_hw_context() reads the drivers GID HW context corresponding
 * to the SGID attr. Callers are required to already be holding the reference
 * to an existing GID entry.
 *
 * Returns the HW GID context
 *
 */
void *rdma_read_gid_hw_context(const struct ib_gid_attr *attr)
{
	return container_of(attr, struct ib_gid_table_entry, attr)->context;
}
EXPORT_SYMBOL(rdma_read_gid_hw_context);

/**
 * rdma_find_gid - Returns SGID attributes if the matching GID is found.
 * @device: The device to query.
 * @gid: The GID value to search for.
 * @gid_type: The GID type to search for.
 * @ndev: In RoCE, the net device of the device. NULL means ignore.
 *
 * rdma_find_gid() searches for the specified GID value in the software cache.
 *
 * Returns GID attributes if a valid GID is found or returns ERR_PTR for the
 * error. The caller must invoke rdma_put_gid_attr() to release the reference.
 *
 */
const struct ib_gid_attr *rdma_find_gid(struct ib_device *device,
					const union ib_gid *gid,
					enum ib_gid_type gid_type,
					struct net_device *ndev)
{
	unsigned long mask = GID_ATTR_FIND_MASK_GID |
			     GID_ATTR_FIND_MASK_GID_TYPE;
	struct ib_gid_attr gid_attr_val = {.ndev = ndev, .gid_type = gid_type};
	unsigned int p;

	if (ndev)
		mask |= GID_ATTR_FIND_MASK_NETDEV;

	rdma_for_each_port(device, p) {
		struct ib_gid_table *table;
		unsigned long flags;
		int index;

		table = device->port_data[p].cache.gid;
		read_lock_irqsave(&table->rwlock, flags);
		index = find_gid(table, gid, &gid_attr_val, false, mask, NULL);
		if (index >= 0) {
			const struct ib_gid_attr *attr;

			get_gid_entry(table->data_vec[index]);
			attr = &table->data_vec[index]->attr;
			read_unlock_irqrestore(&table->rwlock, flags);
			return attr;
		}
		read_unlock_irqrestore(&table->rwlock, flags);
	}

	return ERR_PTR(-ENOENT);
}
EXPORT_SYMBOL(rdma_find_gid);

int ib_get_cached_pkey(struct ib_device *device,
		       u8                port_num,
		       int               index,
		       u16              *pkey)
{
	struct ib_pkey_cache *cache;
	unsigned long flags;
	int ret = 0;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);

	cache = device->port_data[port_num].cache.pkey;

	if (!cache || index < 0 || index >= cache->table_len)
		ret = -EINVAL;
	else
		*pkey = cache->table[index];

	read_unlock_irqrestore(&device->cache_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_get_cached_pkey);

int ib_get_cached_subnet_prefix(struct ib_device *device,
				u8                port_num,
				u64              *sn_pfx)
{
	unsigned long flags;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);
	*sn_pfx = device->port_data[port_num].cache.subnet_prefix;
	read_unlock_irqrestore(&device->cache_lock, flags);

	return 0;
}
EXPORT_SYMBOL(ib_get_cached_subnet_prefix);

int ib_find_cached_pkey(struct ib_device *device,
			u8                port_num,
			u16               pkey,
			u16              *index)
{
	struct ib_pkey_cache *cache;
	unsigned long flags;
	int i;
	int ret = -ENOENT;
	int partial_ix = -1;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);

	cache = device->port_data[port_num].cache.pkey;
	if (!cache) {
		ret = -EINVAL;
		goto err;
	}

	*index = -1;

	for (i = 0; i < cache->table_len; ++i)
		if ((cache->table[i] & 0x7fff) == (pkey & 0x7fff)) {
			if (cache->table[i] & 0x8000) {
				*index = i;
				ret = 0;
				break;
			} else
				partial_ix = i;
		}

	if (ret && partial_ix >= 0) {
		*index = partial_ix;
		ret = 0;
	}

err:
	read_unlock_irqrestore(&device->cache_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_find_cached_pkey);

int ib_find_exact_cached_pkey(struct ib_device *device,
			      u8                port_num,
			      u16               pkey,
			      u16              *index)
{
	struct ib_pkey_cache *cache;
	unsigned long flags;
	int i;
	int ret = -ENOENT;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);

	cache = device->port_data[port_num].cache.pkey;
	if (!cache) {
		ret = -EINVAL;
		goto err;
	}

	*index = -1;

	for (i = 0; i < cache->table_len; ++i)
		if (cache->table[i] == pkey) {
			*index = i;
			ret = 0;
			break;
		}

err:
	read_unlock_irqrestore(&device->cache_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_find_exact_cached_pkey);

int ib_get_cached_lmc(struct ib_device *device,
		      u8                port_num,
		      u8                *lmc)
{
	unsigned long flags;
	int ret = 0;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);
	*lmc = device->port_data[port_num].cache.lmc;
	read_unlock_irqrestore(&device->cache_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_get_cached_lmc);

int ib_get_cached_port_state(struct ib_device   *device,
			     u8                  port_num,
			     enum ib_port_state *port_state)
{
	unsigned long flags;
	int ret = 0;

	if (!rdma_is_port_valid(device, port_num))
		return -EINVAL;

	read_lock_irqsave(&device->cache_lock, flags);
	*port_state = device->port_data[port_num].cache.port_state;
	read_unlock_irqrestore(&device->cache_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_get_cached_port_state);

/**
 * rdma_get_gid_attr - Returns GID attributes for a port of a device
 * at a requested gid_index, if a valid GID entry exists.
 * @device:		The device to query.
 * @port_num:		The port number on the device where the GID value
 *			is to be queried.
 * @index:		Index of the GID table entry whose attributes are to
 *                      be queried.
 *
 * rdma_get_gid_attr() acquires reference count of gid attributes from the
 * cached GID table. Caller must invoke rdma_put_gid_attr() to release
 * reference to gid attribute regardless of link layer.
 *
 * Returns pointer to valid gid attribute or ERR_PTR for the appropriate error
 * code.
 */
const struct ib_gid_attr *
rdma_get_gid_attr(struct ib_device *device, u8 port_num, int index)
{
	const struct ib_gid_attr *attr = ERR_PTR(-EINVAL);
	struct ib_gid_table *table;
	unsigned long flags;

	if (!rdma_is_port_valid(device, port_num))
		return ERR_PTR(-EINVAL);

	table = rdma_gid_table(device, port_num);
	if (index < 0 || index >= table->sz)
		return ERR_PTR(-EINVAL);

	read_lock_irqsave(&table->rwlock, flags);
	if (!is_gid_entry_valid(table->data_vec[index]))
		goto done;

	get_gid_entry(table->data_vec[index]);
	attr = &table->data_vec[index]->attr;
done:
	read_unlock_irqrestore(&table->rwlock, flags);
	return attr;
}
EXPORT_SYMBOL(rdma_get_gid_attr);

/**
 * rdma_put_gid_attr - Release reference to the GID attribute
 * @attr:		Pointer to the GID attribute whose reference
 *			needs to be released.
 *
 * rdma_put_gid_attr() must be used to release reference whose
 * reference is acquired using rdma_get_gid_attr() or any APIs
 * which returns a pointer to the ib_gid_attr regardless of link layer
 * of IB or RoCE.
 *
 */
void rdma_put_gid_attr(const struct ib_gid_attr *attr)
{
	struct ib_gid_table_entry *entry =
		container_of(attr, struct ib_gid_table_entry, attr);

	put_gid_entry(entry);
}
EXPORT_SYMBOL(rdma_put_gid_attr);

/**
 * rdma_hold_gid_attr - Get reference to existing GID attribute
 *
 * @attr:		Pointer to the GID attribute whose reference
 *			needs to be taken.
 *
 * Increase the reference count to a GID attribute to keep it from being
 * freed. Callers are required to already be holding a reference to attribute.
 *
 */
void rdma_hold_gid_attr(const struct ib_gid_attr *attr)
{
	struct ib_gid_table_entry *entry =
		container_of(attr, struct ib_gid_table_entry, attr);

	get_gid_entry(entry);
}
EXPORT_SYMBOL(rdma_hold_gid_attr);

/**
 * rdma_read_gid_attr_ndev_rcu - Read GID attribute netdevice
 * which must be in UP state.
 *
 * @attr:Pointer to the GID attribute
 *
 * Returns pointer to netdevice if the netdevice was attached to GID and
 * netdevice is in UP state. Caller must hold RCU lock as this API
 * reads the netdev flags which can change while netdevice migrates to
 * different net namespace. Returns ERR_PTR with error code otherwise.
 *
 */
struct net_device *rdma_read_gid_attr_ndev_rcu(const struct ib_gid_attr *attr)
{
	struct ib_gid_table_entry *entry =
			container_of(attr, struct ib_gid_table_entry, attr);
	struct ib_device *device = entry->attr.device;
	struct net_device *ndev = ERR_PTR(-ENODEV);
	u8 port_num = entry->attr.port_num;
	struct ib_gid_table *table;
	unsigned long flags;
	bool valid;

	table = rdma_gid_table(device, port_num);

	read_lock_irqsave(&table->rwlock, flags);
	valid = is_gid_entry_valid(table->data_vec[attr->index]);
	if (valid) {
		ndev = rcu_dereference(attr->ndev);
		if (!ndev ||
		    (ndev && ((READ_ONCE(ndev->flags) & IFF_UP) == 0)))
			ndev = ERR_PTR(-ENODEV);
	}
	read_unlock_irqrestore(&table->rwlock, flags);
	return ndev;
}
EXPORT_SYMBOL(rdma_read_gid_attr_ndev_rcu);

static int get_lower_dev_vlan(struct net_device *lower_dev,
			      struct netdev_nested_priv *priv)
{
	u16 *vlan_id = (u16 *)priv->data;

	if (is_vlan_dev(lower_dev))
		*vlan_id = vlan_dev_vlan_id(lower_dev);

	/* We are interested only in first level vlan device, so
	 * always return 1 to stop iterating over next level devices.
	 */
	return 1;
}

/**
 * rdma_read_gid_l2_fields - Read the vlan ID and source MAC address
 *			     of a GID entry.
 *
 * @attr:	GID attribute pointer whose L2 fields to be read
 * @vlan_id:	Pointer to vlan id to fill up if the GID entry has
 *		vlan id. It is optional.
 * @smac:	Pointer to smac to fill up for a GID entry. It is optional.
 *
 * rdma_read_gid_l2_fields() returns 0 on success and returns vlan id
 * (if gid entry has vlan) and source MAC, or returns error.
 */
int rdma_read_gid_l2_fields(const struct ib_gid_attr *attr,
			    u16 *vlan_id, u8 *smac)
{
	struct netdev_nested_priv priv = {
		.data = (void *)vlan_id,
	};
	struct net_device *ndev;

	rcu_read_lock();
	ndev = rcu_dereference(attr->ndev);
	if (!ndev) {
		rcu_read_unlock();
		return -ENODEV;
	}
	if (smac)
		ether_addr_copy(smac, ndev->dev_addr);
	if (vlan_id) {
		*vlan_id = 0xffff;
		if (is_vlan_dev(ndev)) {
			*vlan_id = vlan_dev_vlan_id(ndev);
		} else {
			/* If the netdev is upper device and if it's lower
			 * device is vlan device, consider vlan id of the
			 * the lower vlan device for this gid entry.
			 */
			netdev_walk_all_lower_dev_rcu(attr->ndev,
					get_lower_dev_vlan, &priv);
		}
	}
	rcu_read_unlock();
	return 0;
}
EXPORT_SYMBOL(rdma_read_gid_l2_fields);

static int config_non_roce_gid_cache(struct ib_device *device,
				     u8 port, int gid_tbl_len)
{
	struct ib_gid_attr gid_attr = {};
	struct ib_gid_table *table;
	int ret = 0;
	int i;

	gid_attr.device = device;
	gid_attr.port_num = port;
	table = rdma_gid_table(device, port);

	mutex_lock(&table->lock);
	for (i = 0; i < gid_tbl_len; ++i) {
		if (!device->ops.query_gid)
			continue;
		ret = device->ops.query_gid(device, port, i, &gid_attr.gid);
		if (ret) {
			dev_warn(&device->dev,
				 "query_gid failed (%d) for index %d\n", ret,
				 i);
			goto err;
		}
		gid_attr.index = i;
		add_modify_gid(table, &gid_attr);
	}
err:
	mutex_unlock(&table->lock);
	return ret;
}

static int
ib_cache_update(struct ib_device *device, u8 port, bool enforce_security)
{
	struct ib_port_attr       *tprops = NULL;
	struct ib_pkey_cache      *pkey_cache = NULL, *old_pkey_cache;
	int                        i;
	int                        ret;

	if (!rdma_is_port_valid(device, port))
		return -EINVAL;

	tprops = kmalloc(sizeof *tprops, GFP_KERNEL);
	if (!tprops)
		return -ENOMEM;

	ret = ib_query_port(device, port, tprops);
	if (ret) {
		dev_warn(&device->dev, "ib_query_port failed (%d)\n", ret);
		goto err;
	}

	if (!rdma_protocol_roce(device, port)) {
		ret = config_non_roce_gid_cache(device, port,
						tprops->gid_tbl_len);
		if (ret)
			goto err;
	}

	if (tprops->pkey_tbl_len) {
		pkey_cache = kmalloc(struct_size(pkey_cache, table,
						 tprops->pkey_tbl_len),
				     GFP_KERNEL);
		if (!pkey_cache) {
			ret = -ENOMEM;
			goto err;
		}

		pkey_cache->table_len = tprops->pkey_tbl_len;

		for (i = 0; i < pkey_cache->table_len; ++i) {
			ret = ib_query_pkey(device, port, i,
					    pkey_cache->table + i);
			if (ret) {
				dev_warn(&device->dev,
					 "ib_query_pkey failed (%d) for index %d\n",
					 ret, i);
				goto err;
			}
		}
	}

	write_lock_irq(&device->cache_lock);

	old_pkey_cache = device->port_data[port].cache.pkey;

	device->port_data[port].cache.pkey = pkey_cache;
	device->port_data[port].cache.lmc = tprops->lmc;
	device->port_data[port].cache.port_state = tprops->state;

	device->port_data[port].cache.subnet_prefix = tprops->subnet_prefix;
	write_unlock_irq(&device->cache_lock);

	if (enforce_security)
		ib_security_cache_change(device,
					 port,
					 tprops->subnet_prefix);

	kfree(old_pkey_cache);
	kfree(tprops);
	return 0;

err:
	kfree(pkey_cache);
	kfree(tprops);
	return ret;
}

static void ib_cache_event_task(struct work_struct *_work)
{
	struct ib_update_work *work =
		container_of(_work, struct ib_update_work, work);
	int ret;

	/* Before distributing the cache update event, first sync
	 * the cache.
	 */
	ret = ib_cache_update(work->event.device, work->event.element.port_num,
			      work->enforce_security);

	/* GID event is notified already for individual GID entries by
	 * dispatch_gid_change_event(). Hence, notifiy for rest of the
	 * events.
	 */
	if (!ret && work->event.event != IB_EVENT_GID_CHANGE)
		ib_dispatch_event_clients(&work->event);

	kfree(work);
}

static void ib_generic_event_task(struct work_struct *_work)
{
	struct ib_update_work *work =
		container_of(_work, struct ib_update_work, work);

	ib_dispatch_event_clients(&work->event);
	kfree(work);
}

static bool is_cache_update_event(const struct ib_event *event)
{
	return (event->event == IB_EVENT_PORT_ERR    ||
		event->event == IB_EVENT_PORT_ACTIVE ||
		event->event == IB_EVENT_LID_CHANGE  ||
		event->event == IB_EVENT_PKEY_CHANGE ||
		event->event == IB_EVENT_CLIENT_REREGISTER ||
		event->event == IB_EVENT_GID_CHANGE);
}

/**
 * ib_dispatch_event - Dispatch an asynchronous event
 * @event:Event to dispatch
 *
 * Low-level drivers must call ib_dispatch_event() to dispatch the
 * event to all registered event handlers when an asynchronous event
 * occurs.
 */
void ib_dispatch_event(const struct ib_event *event)
{
	struct ib_update_work *work;

	work = kzalloc(sizeof(*work), GFP_ATOMIC);
	if (!work)
		return;

	if (is_cache_update_event(event))
		INIT_WORK(&work->work, ib_cache_event_task);
	else
		INIT_WORK(&work->work, ib_generic_event_task);

	work->event = *event;
	if (event->event == IB_EVENT_PKEY_CHANGE ||
	    event->event == IB_EVENT_GID_CHANGE)
		work->enforce_security = true;

	queue_work(ib_wq, &work->work);
}
EXPORT_SYMBOL(ib_dispatch_event);

int ib_cache_setup_one(struct ib_device *device)
{
	unsigned int p;
	int err;

	rwlock_init(&device->cache_lock);

	err = gid_table_setup_one(device);
	if (err)
		return err;

	rdma_for_each_port (device, p) {
		err = ib_cache_update(device, p, true);
		if (err)
			return err;
	}

	return 0;
}

void ib_cache_release_one(struct ib_device *device)
{
	unsigned int p;

	/*
	 * The release function frees all the cache elements.
	 * This function should be called as part of freeing
	 * all the device's resources when the cache could no
	 * longer be accessed.
	 */
	rdma_for_each_port (device, p)
		kfree(device->port_data[p].cache.pkey);

	gid_table_release_one(device);
}

void ib_cache_cleanup_one(struct ib_device *device)
{
	/* The cleanup function waits for all in-progress workqueue
	 * elements and cleans up the GID cache. This function should be
	 * called after the device was removed from the devices list and
	 * all clients were removed, so the cache exists but is
	 * non-functional and shouldn't be updated anymore.
	 */
	flush_workqueue(ib_wq);
	gid_table_cleanup_one(device);

	/*
	 * Flush the wq second time for any pending GID delete work.
	 */
	flush_workqueue(ib_wq);
}