summaryrefslogtreecommitdiff
path: root/fs/nfs/nfs42xattr.c
blob: b6e3d8f77b9101b527893d74a78caf1a2fb3db9e (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
// SPDX-License-Identifier: GPL-2.0

/*
 * Copyright 2019, 2020 Amazon.com, Inc. or its affiliates. All rights reserved.
 *
 * User extended attribute client side cache functions.
 *
 * Author: Frank van der Linden <fllinden@amazon.com>
 */
#include <linux/errno.h>
#include <linux/nfs_fs.h>
#include <linux/hashtable.h>
#include <linux/refcount.h>
#include <uapi/linux/xattr.h>

#include "nfs4_fs.h"
#include "internal.h"

/*
 * User extended attributes client side caching is implemented by having
 * a cache structure attached to NFS inodes. This structure is allocated
 * when needed, and freed when the cache is zapped.
 *
 * The cache structure contains as hash table of entries, and a pointer
 * to a special-cased entry for the listxattr cache.
 *
 * Accessing and allocating / freeing the caches is done via reference
 * counting. The cache entries use a similar refcounting scheme.
 *
 * This makes freeing a cache, both from the shrinker and from the
 * zap cache path, easy. It also means that, in current use cases,
 * the large majority of inodes will not waste any memory, as they
 * will never have any user extended attributes assigned to them.
 *
 * Attribute entries are hashed in to a simple hash table. They are
 * also part of an LRU.
 *
 * There are three shrinkers.
 *
 * Two shrinkers deal with the cache entries themselves: one for
 * large entries (> PAGE_SIZE), and one for smaller entries. The
 * shrinker for the larger entries works more aggressively than
 * those for the smaller entries.
 *
 * The other shrinker frees the cache structures themselves.
 */

/*
 * 64 buckets is a good default. There is likely no reasonable
 * workload that uses more than even 64 user extended attributes.
 * You can certainly add a lot more - but you get what you ask for
 * in those circumstances.
 */
#define NFS4_XATTR_HASH_SIZE	64

#define NFSDBG_FACILITY	NFSDBG_XATTRCACHE

struct nfs4_xattr_cache;
struct nfs4_xattr_entry;

struct nfs4_xattr_bucket {
	spinlock_t lock;
	struct hlist_head hlist;
	struct nfs4_xattr_cache *cache;
	bool draining;
};

struct nfs4_xattr_cache {
	struct kref ref;
	struct nfs4_xattr_bucket buckets[NFS4_XATTR_HASH_SIZE];
	struct list_head lru;
	struct list_head dispose;
	atomic_long_t nent;
	spinlock_t listxattr_lock;
	struct inode *inode;
	struct nfs4_xattr_entry *listxattr;
};

struct nfs4_xattr_entry {
	struct kref ref;
	struct hlist_node hnode;
	struct list_head lru;
	struct list_head dispose;
	char *xattr_name;
	void *xattr_value;
	size_t xattr_size;
	struct nfs4_xattr_bucket *bucket;
	uint32_t flags;
};

#define	NFS4_XATTR_ENTRY_EXTVAL	0x0001

/*
 * LRU list of NFS inodes that have xattr caches.
 */
static struct list_lru nfs4_xattr_cache_lru;
static struct list_lru nfs4_xattr_entry_lru;
static struct list_lru nfs4_xattr_large_entry_lru;

static struct kmem_cache *nfs4_xattr_cache_cachep;

/*
 * Hashing helper functions.
 */
static void
nfs4_xattr_hash_init(struct nfs4_xattr_cache *cache)
{
	unsigned int i;

	for (i = 0; i < NFS4_XATTR_HASH_SIZE; i++) {
		INIT_HLIST_HEAD(&cache->buckets[i].hlist);
		spin_lock_init(&cache->buckets[i].lock);
		cache->buckets[i].cache = cache;
		cache->buckets[i].draining = false;
	}
}

/*
 * Locking order:
 * 1. inode i_lock or bucket lock
 * 2. list_lru lock (taken by list_lru_* functions)
 */

/*
 * Wrapper functions to add a cache entry to the right LRU.
 */
static bool
nfs4_xattr_entry_lru_add(struct nfs4_xattr_entry *entry)
{
	struct list_lru *lru;

	lru = (entry->flags & NFS4_XATTR_ENTRY_EXTVAL) ?
	    &nfs4_xattr_large_entry_lru : &nfs4_xattr_entry_lru;

	return list_lru_add_obj(lru, &entry->lru);
}

static bool
nfs4_xattr_entry_lru_del(struct nfs4_xattr_entry *entry)
{
	struct list_lru *lru;

	lru = (entry->flags & NFS4_XATTR_ENTRY_EXTVAL) ?
	    &nfs4_xattr_large_entry_lru : &nfs4_xattr_entry_lru;

	return list_lru_del_obj(lru, &entry->lru);
}

/*
 * This function allocates cache entries. They are the normal
 * extended attribute name/value pairs, but may also be a listxattr
 * cache. Those allocations use the same entry so that they can be
 * treated as one by the memory shrinker.
 *
 * xattr cache entries are allocated together with names. If the
 * value fits in to one page with the entry structure and the name,
 * it will also be part of the same allocation (kmalloc). This is
 * expected to be the vast majority of cases. Larger allocations
 * have a value pointer that is allocated separately by kvmalloc.
 *
 * Parameters:
 *
 * @name:  Name of the extended attribute. NULL for listxattr cache
 *         entry.
 * @value: Value of attribute, or listxattr cache. NULL if the
 *         value is to be copied from pages instead.
 * @pages: Pages to copy the value from, if not NULL. Passed in to
 *	   make it easier to copy the value after an RPC, even if
 *	   the value will not be passed up to application (e.g.
 *	   for a 'query' getxattr with NULL buffer).
 * @len:   Length of the value. Can be 0 for zero-length attributes.
 *         @value and @pages will be NULL if @len is 0.
 */
static struct nfs4_xattr_entry *
nfs4_xattr_alloc_entry(const char *name, const void *value,
		       struct page **pages, size_t len)
{
	struct nfs4_xattr_entry *entry;
	void *valp;
	char *namep;
	size_t alloclen, slen;
	char *buf;
	uint32_t flags;

	BUILD_BUG_ON(sizeof(struct nfs4_xattr_entry) +
	    XATTR_NAME_MAX + 1 > PAGE_SIZE);

	alloclen = sizeof(struct nfs4_xattr_entry);
	if (name != NULL) {
		slen = strlen(name) + 1;
		alloclen += slen;
	} else
		slen = 0;

	if (alloclen + len <= PAGE_SIZE) {
		alloclen += len;
		flags = 0;
	} else {
		flags = NFS4_XATTR_ENTRY_EXTVAL;
	}

	buf = kmalloc(alloclen, GFP_KERNEL);
	if (buf == NULL)
		return NULL;
	entry = (struct nfs4_xattr_entry *)buf;

	if (name != NULL) {
		namep = buf + sizeof(struct nfs4_xattr_entry);
		memcpy(namep, name, slen);
	} else {
		namep = NULL;
	}


	if (flags & NFS4_XATTR_ENTRY_EXTVAL) {
		valp = kvmalloc(len, GFP_KERNEL);
		if (valp == NULL) {
			kfree(buf);
			return NULL;
		}
	} else if (len != 0) {
		valp = buf + sizeof(struct nfs4_xattr_entry) + slen;
	} else
		valp = NULL;

	if (valp != NULL) {
		if (value != NULL)
			memcpy(valp, value, len);
		else
			_copy_from_pages(valp, pages, 0, len);
	}

	entry->flags = flags;
	entry->xattr_value = valp;
	kref_init(&entry->ref);
	entry->xattr_name = namep;
	entry->xattr_size = len;
	entry->bucket = NULL;
	INIT_LIST_HEAD(&entry->lru);
	INIT_LIST_HEAD(&entry->dispose);
	INIT_HLIST_NODE(&entry->hnode);

	return entry;
}

static void
nfs4_xattr_free_entry(struct nfs4_xattr_entry *entry)
{
	if (entry->flags & NFS4_XATTR_ENTRY_EXTVAL)
		kvfree(entry->xattr_value);
	kfree(entry);
}

static void
nfs4_xattr_free_entry_cb(struct kref *kref)
{
	struct nfs4_xattr_entry *entry;

	entry = container_of(kref, struct nfs4_xattr_entry, ref);

	if (WARN_ON(!list_empty(&entry->lru)))
		return;

	nfs4_xattr_free_entry(entry);
}

static void
nfs4_xattr_free_cache_cb(struct kref *kref)
{
	struct nfs4_xattr_cache *cache;
	int i;

	cache = container_of(kref, struct nfs4_xattr_cache, ref);

	for (i = 0; i < NFS4_XATTR_HASH_SIZE; i++) {
		if (WARN_ON(!hlist_empty(&cache->buckets[i].hlist)))
			return;
		cache->buckets[i].draining = false;
	}

	cache->listxattr = NULL;

	kmem_cache_free(nfs4_xattr_cache_cachep, cache);

}

static struct nfs4_xattr_cache *
nfs4_xattr_alloc_cache(void)
{
	struct nfs4_xattr_cache *cache;

	cache = kmem_cache_alloc(nfs4_xattr_cache_cachep, GFP_KERNEL);
	if (cache == NULL)
		return NULL;

	kref_init(&cache->ref);
	atomic_long_set(&cache->nent, 0);

	return cache;
}

/*
 * Set the listxattr cache, which is a special-cased cache entry.
 * The special value ERR_PTR(-ESTALE) is used to indicate that
 * the cache is being drained - this prevents a new listxattr
 * cache from being added to what is now a stale cache.
 */
static int
nfs4_xattr_set_listcache(struct nfs4_xattr_cache *cache,
			 struct nfs4_xattr_entry *new)
{
	struct nfs4_xattr_entry *old;
	int ret = 1;

	spin_lock(&cache->listxattr_lock);

	old = cache->listxattr;

	if (old == ERR_PTR(-ESTALE)) {
		ret = 0;
		goto out;
	}

	cache->listxattr = new;
	if (new != NULL && new != ERR_PTR(-ESTALE))
		nfs4_xattr_entry_lru_add(new);

	if (old != NULL) {
		nfs4_xattr_entry_lru_del(old);
		kref_put(&old->ref, nfs4_xattr_free_entry_cb);
	}
out:
	spin_unlock(&cache->listxattr_lock);

	return ret;
}

/*
 * Unlink a cache from its parent inode, clearing out an invalid
 * cache. Must be called with i_lock held.
 */
static struct nfs4_xattr_cache *
nfs4_xattr_cache_unlink(struct inode *inode)
{
	struct nfs_inode *nfsi;
	struct nfs4_xattr_cache *oldcache;

	nfsi = NFS_I(inode);

	oldcache = nfsi->xattr_cache;
	if (oldcache != NULL) {
		list_lru_del_obj(&nfs4_xattr_cache_lru, &oldcache->lru);
		oldcache->inode = NULL;
	}
	nfsi->xattr_cache = NULL;
	nfsi->cache_validity &= ~NFS_INO_INVALID_XATTR;

	return oldcache;

}

/*
 * Discard a cache. Called by get_cache() if there was an old,
 * invalid cache. Can also be called from a shrinker callback.
 *
 * The cache is dead, it has already been unlinked from its inode,
 * and no longer appears on the cache LRU list.
 *
 * Mark all buckets as draining, so that no new entries are added. This
 * could still happen in the unlikely, but possible case that another
 * thread had grabbed a reference before it was unlinked from the inode,
 * and is still holding it for an add operation.
 *
 * Remove all entries from the LRU lists, so that there is no longer
 * any way to 'find' this cache. Then, remove the entries from the hash
 * table.
 *
 * At that point, the cache will remain empty and can be freed when the final
 * reference drops, which is very likely the kref_put at the end of
 * this function, or the one called immediately afterwards in the
 * shrinker callback.
 */
static void
nfs4_xattr_discard_cache(struct nfs4_xattr_cache *cache)
{
	unsigned int i;
	struct nfs4_xattr_entry *entry;
	struct nfs4_xattr_bucket *bucket;
	struct hlist_node *n;

	nfs4_xattr_set_listcache(cache, ERR_PTR(-ESTALE));

	for (i = 0; i < NFS4_XATTR_HASH_SIZE; i++) {
		bucket = &cache->buckets[i];

		spin_lock(&bucket->lock);
		bucket->draining = true;
		hlist_for_each_entry_safe(entry, n, &bucket->hlist, hnode) {
			nfs4_xattr_entry_lru_del(entry);
			hlist_del_init(&entry->hnode);
			kref_put(&entry->ref, nfs4_xattr_free_entry_cb);
		}
		spin_unlock(&bucket->lock);
	}

	atomic_long_set(&cache->nent, 0);

	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
}

/*
 * Get a referenced copy of the cache structure. Avoid doing allocs
 * while holding i_lock. Which means that we do some optimistic allocation,
 * and might have to free the result in rare cases.
 *
 * This function only checks the NFS_INO_INVALID_XATTR cache validity bit
 * and acts accordingly, replacing the cache when needed. For the read case
 * (!add), this means that the caller must make sure that the cache
 * is valid before caling this function. getxattr and listxattr call
 * revalidate_inode to do this. The attribute cache timeout (for the
 * non-delegated case) is expected to be dealt with in the revalidate
 * call.
 */

static struct nfs4_xattr_cache *
nfs4_xattr_get_cache(struct inode *inode, int add)
{
	struct nfs_inode *nfsi;
	struct nfs4_xattr_cache *cache, *oldcache, *newcache;

	nfsi = NFS_I(inode);

	cache = oldcache = NULL;

	spin_lock(&inode->i_lock);

	if (nfsi->cache_validity & NFS_INO_INVALID_XATTR)
		oldcache = nfs4_xattr_cache_unlink(inode);
	else
		cache = nfsi->xattr_cache;

	if (cache != NULL)
		kref_get(&cache->ref);

	spin_unlock(&inode->i_lock);

	if (add && cache == NULL) {
		newcache = NULL;

		cache = nfs4_xattr_alloc_cache();
		if (cache == NULL)
			goto out;

		spin_lock(&inode->i_lock);
		if (nfsi->cache_validity & NFS_INO_INVALID_XATTR) {
			/*
			 * The cache was invalidated again. Give up,
			 * since what we want to enter is now likely
			 * outdated anyway.
			 */
			spin_unlock(&inode->i_lock);
			kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
			cache = NULL;
			goto out;
		}

		/*
		 * Check if someone beat us to it.
		 */
		if (nfsi->xattr_cache != NULL) {
			newcache = nfsi->xattr_cache;
			kref_get(&newcache->ref);
		} else {
			kref_get(&cache->ref);
			nfsi->xattr_cache = cache;
			cache->inode = inode;
			list_lru_add_obj(&nfs4_xattr_cache_lru, &cache->lru);
		}

		spin_unlock(&inode->i_lock);

		/*
		 * If there was a race, throw away the cache we just
		 * allocated, and use the new one allocated by someone
		 * else.
		 */
		if (newcache != NULL) {
			kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
			cache = newcache;
		}
	}

out:
	/*
	 * Discard the now orphaned old cache.
	 */
	if (oldcache != NULL)
		nfs4_xattr_discard_cache(oldcache);

	return cache;
}

static inline struct nfs4_xattr_bucket *
nfs4_xattr_hash_bucket(struct nfs4_xattr_cache *cache, const char *name)
{
	return &cache->buckets[jhash(name, strlen(name), 0) &
	    (ARRAY_SIZE(cache->buckets) - 1)];
}

static struct nfs4_xattr_entry *
nfs4_xattr_get_entry(struct nfs4_xattr_bucket *bucket, const char *name)
{
	struct nfs4_xattr_entry *entry;

	entry = NULL;

	hlist_for_each_entry(entry, &bucket->hlist, hnode) {
		if (!strcmp(entry->xattr_name, name))
			break;
	}

	return entry;
}

static int
nfs4_xattr_hash_add(struct nfs4_xattr_cache *cache,
		    struct nfs4_xattr_entry *entry)
{
	struct nfs4_xattr_bucket *bucket;
	struct nfs4_xattr_entry *oldentry = NULL;
	int ret = 1;

	bucket = nfs4_xattr_hash_bucket(cache, entry->xattr_name);
	entry->bucket = bucket;

	spin_lock(&bucket->lock);

	if (bucket->draining) {
		ret = 0;
		goto out;
	}

	oldentry = nfs4_xattr_get_entry(bucket, entry->xattr_name);
	if (oldentry != NULL) {
		hlist_del_init(&oldentry->hnode);
		nfs4_xattr_entry_lru_del(oldentry);
	} else {
		atomic_long_inc(&cache->nent);
	}

	hlist_add_head(&entry->hnode, &bucket->hlist);
	nfs4_xattr_entry_lru_add(entry);

out:
	spin_unlock(&bucket->lock);

	if (oldentry != NULL)
		kref_put(&oldentry->ref, nfs4_xattr_free_entry_cb);

	return ret;
}

static void
nfs4_xattr_hash_remove(struct nfs4_xattr_cache *cache, const char *name)
{
	struct nfs4_xattr_bucket *bucket;
	struct nfs4_xattr_entry *entry;

	bucket = nfs4_xattr_hash_bucket(cache, name);

	spin_lock(&bucket->lock);

	entry = nfs4_xattr_get_entry(bucket, name);
	if (entry != NULL) {
		hlist_del_init(&entry->hnode);
		nfs4_xattr_entry_lru_del(entry);
		atomic_long_dec(&cache->nent);
	}

	spin_unlock(&bucket->lock);

	if (entry != NULL)
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);
}

static struct nfs4_xattr_entry *
nfs4_xattr_hash_find(struct nfs4_xattr_cache *cache, const char *name)
{
	struct nfs4_xattr_bucket *bucket;
	struct nfs4_xattr_entry *entry;

	bucket = nfs4_xattr_hash_bucket(cache, name);

	spin_lock(&bucket->lock);

	entry = nfs4_xattr_get_entry(bucket, name);
	if (entry != NULL)
		kref_get(&entry->ref);

	spin_unlock(&bucket->lock);

	return entry;
}

/*
 * Entry point to retrieve an entry from the cache.
 */
ssize_t nfs4_xattr_cache_get(struct inode *inode, const char *name, char *buf,
			 ssize_t buflen)
{
	struct nfs4_xattr_cache *cache;
	struct nfs4_xattr_entry *entry;
	ssize_t ret;

	cache = nfs4_xattr_get_cache(inode, 0);
	if (cache == NULL)
		return -ENOENT;

	ret = 0;
	entry = nfs4_xattr_hash_find(cache, name);

	if (entry != NULL) {
		dprintk("%s: cache hit '%s', len %lu\n", __func__,
		    entry->xattr_name, (unsigned long)entry->xattr_size);
		if (buflen == 0) {
			/* Length probe only */
			ret = entry->xattr_size;
		} else if (buflen < entry->xattr_size)
			ret = -ERANGE;
		else {
			memcpy(buf, entry->xattr_value, entry->xattr_size);
			ret = entry->xattr_size;
		}
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);
	} else {
		dprintk("%s: cache miss '%s'\n", __func__, name);
		ret = -ENOENT;
	}

	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);

	return ret;
}

/*
 * Retrieve a cached list of xattrs from the cache.
 */
ssize_t nfs4_xattr_cache_list(struct inode *inode, char *buf, ssize_t buflen)
{
	struct nfs4_xattr_cache *cache;
	struct nfs4_xattr_entry *entry;
	ssize_t ret;

	cache = nfs4_xattr_get_cache(inode, 0);
	if (cache == NULL)
		return -ENOENT;

	spin_lock(&cache->listxattr_lock);

	entry = cache->listxattr;

	if (entry != NULL && entry != ERR_PTR(-ESTALE)) {
		if (buflen == 0) {
			/* Length probe only */
			ret = entry->xattr_size;
		} else if (entry->xattr_size > buflen)
			ret = -ERANGE;
		else {
			memcpy(buf, entry->xattr_value, entry->xattr_size);
			ret = entry->xattr_size;
		}
	} else {
		ret = -ENOENT;
	}

	spin_unlock(&cache->listxattr_lock);

	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);

	return ret;
}

/*
 * Add an xattr to the cache.
 *
 * This also invalidates the xattr list cache.
 */
void nfs4_xattr_cache_add(struct inode *inode, const char *name,
			  const char *buf, struct page **pages, ssize_t buflen)
{
	struct nfs4_xattr_cache *cache;
	struct nfs4_xattr_entry *entry;

	dprintk("%s: add '%s' len %lu\n", __func__,
	    name, (unsigned long)buflen);

	cache = nfs4_xattr_get_cache(inode, 1);
	if (cache == NULL)
		return;

	entry = nfs4_xattr_alloc_entry(name, buf, pages, buflen);
	if (entry == NULL)
		goto out;

	(void)nfs4_xattr_set_listcache(cache, NULL);

	if (!nfs4_xattr_hash_add(cache, entry))
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);

out:
	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
}


/*
 * Remove an xattr from the cache.
 *
 * This also invalidates the xattr list cache.
 */
void nfs4_xattr_cache_remove(struct inode *inode, const char *name)
{
	struct nfs4_xattr_cache *cache;

	dprintk("%s: remove '%s'\n", __func__, name);

	cache = nfs4_xattr_get_cache(inode, 0);
	if (cache == NULL)
		return;

	(void)nfs4_xattr_set_listcache(cache, NULL);
	nfs4_xattr_hash_remove(cache, name);

	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
}

/*
 * Cache listxattr output, replacing any possible old one.
 */
void nfs4_xattr_cache_set_list(struct inode *inode, const char *buf,
			       ssize_t buflen)
{
	struct nfs4_xattr_cache *cache;
	struct nfs4_xattr_entry *entry;

	cache = nfs4_xattr_get_cache(inode, 1);
	if (cache == NULL)
		return;

	entry = nfs4_xattr_alloc_entry(NULL, buf, NULL, buflen);
	if (entry == NULL)
		goto out;

	/*
	 * This is just there to be able to get to bucket->cache,
	 * which is obviously the same for all buckets, so just
	 * use bucket 0.
	 */
	entry->bucket = &cache->buckets[0];

	if (!nfs4_xattr_set_listcache(cache, entry))
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);

out:
	kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
}

/*
 * Zap the entire cache. Called when an inode is evicted.
 */
void nfs4_xattr_cache_zap(struct inode *inode)
{
	struct nfs4_xattr_cache *oldcache;

	spin_lock(&inode->i_lock);
	oldcache = nfs4_xattr_cache_unlink(inode);
	spin_unlock(&inode->i_lock);

	if (oldcache)
		nfs4_xattr_discard_cache(oldcache);
}

/*
 * The entry LRU is shrunk more aggressively than the cache LRU,
 * by settings @seeks to 1.
 *
 * Cache structures are freed only when they've become empty, after
 * pruning all but one entry.
 */

static unsigned long nfs4_xattr_cache_count(struct shrinker *shrink,
					    struct shrink_control *sc);
static unsigned long nfs4_xattr_entry_count(struct shrinker *shrink,
					    struct shrink_control *sc);
static unsigned long nfs4_xattr_cache_scan(struct shrinker *shrink,
					   struct shrink_control *sc);
static unsigned long nfs4_xattr_entry_scan(struct shrinker *shrink,
					   struct shrink_control *sc);

static struct shrinker *nfs4_xattr_cache_shrinker;
static struct shrinker *nfs4_xattr_entry_shrinker;
static struct shrinker *nfs4_xattr_large_entry_shrinker;

static enum lru_status
cache_lru_isolate(struct list_head *item,
	struct list_lru_one *lru, spinlock_t *lru_lock, void *arg)
{
	struct list_head *dispose = arg;
	struct inode *inode;
	struct nfs4_xattr_cache *cache = container_of(item,
	    struct nfs4_xattr_cache, lru);

	if (atomic_long_read(&cache->nent) > 1)
		return LRU_SKIP;

	/*
	 * If a cache structure is on the LRU list, we know that
	 * its inode is valid. Try to lock it to break the link.
	 * Since we're inverting the lock order here, only try.
	 */
	inode = cache->inode;

	if (!spin_trylock(&inode->i_lock))
		return LRU_SKIP;

	kref_get(&cache->ref);

	cache->inode = NULL;
	NFS_I(inode)->xattr_cache = NULL;
	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_XATTR;
	list_lru_isolate(lru, &cache->lru);

	spin_unlock(&inode->i_lock);

	list_add_tail(&cache->dispose, dispose);
	return LRU_REMOVED;
}

static unsigned long
nfs4_xattr_cache_scan(struct shrinker *shrink, struct shrink_control *sc)
{
	LIST_HEAD(dispose);
	unsigned long freed;
	struct nfs4_xattr_cache *cache;

	freed = list_lru_shrink_walk(&nfs4_xattr_cache_lru, sc,
	    cache_lru_isolate, &dispose);
	while (!list_empty(&dispose)) {
		cache = list_first_entry(&dispose, struct nfs4_xattr_cache,
		    dispose);
		list_del_init(&cache->dispose);
		nfs4_xattr_discard_cache(cache);
		kref_put(&cache->ref, nfs4_xattr_free_cache_cb);
	}

	return freed;
}


static unsigned long
nfs4_xattr_cache_count(struct shrinker *shrink, struct shrink_control *sc)
{
	unsigned long count;

	count = list_lru_shrink_count(&nfs4_xattr_cache_lru, sc);
	return vfs_pressure_ratio(count);
}

static enum lru_status
entry_lru_isolate(struct list_head *item,
	struct list_lru_one *lru, spinlock_t *lru_lock, void *arg)
{
	struct list_head *dispose = arg;
	struct nfs4_xattr_bucket *bucket;
	struct nfs4_xattr_cache *cache;
	struct nfs4_xattr_entry *entry = container_of(item,
	    struct nfs4_xattr_entry, lru);

	bucket = entry->bucket;
	cache = bucket->cache;

	/*
	 * Unhook the entry from its parent (either a cache bucket
	 * or a cache structure if it's a listxattr buf), so that
	 * it's no longer found. Then add it to the isolate list,
	 * to be freed later.
	 *
	 * In both cases, we're reverting lock order, so use
	 * trylock and skip the entry if we can't get the lock.
	 */
	if (entry->xattr_name != NULL) {
		/* Regular cache entry */
		if (!spin_trylock(&bucket->lock))
			return LRU_SKIP;

		kref_get(&entry->ref);

		hlist_del_init(&entry->hnode);
		atomic_long_dec(&cache->nent);
		list_lru_isolate(lru, &entry->lru);

		spin_unlock(&bucket->lock);
	} else {
		/* Listxattr cache entry */
		if (!spin_trylock(&cache->listxattr_lock))
			return LRU_SKIP;

		kref_get(&entry->ref);

		cache->listxattr = NULL;
		list_lru_isolate(lru, &entry->lru);

		spin_unlock(&cache->listxattr_lock);
	}

	list_add_tail(&entry->dispose, dispose);
	return LRU_REMOVED;
}

static unsigned long
nfs4_xattr_entry_scan(struct shrinker *shrink, struct shrink_control *sc)
{
	LIST_HEAD(dispose);
	unsigned long freed;
	struct nfs4_xattr_entry *entry;
	struct list_lru *lru;

	lru = (shrink == nfs4_xattr_large_entry_shrinker) ?
	    &nfs4_xattr_large_entry_lru : &nfs4_xattr_entry_lru;

	freed = list_lru_shrink_walk(lru, sc, entry_lru_isolate, &dispose);

	while (!list_empty(&dispose)) {
		entry = list_first_entry(&dispose, struct nfs4_xattr_entry,
		    dispose);
		list_del_init(&entry->dispose);

		/*
		 * Drop two references: the one that we just grabbed
		 * in entry_lru_isolate, and the one that was set
		 * when the entry was first allocated.
		 */
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);
		kref_put(&entry->ref, nfs4_xattr_free_entry_cb);
	}

	return freed;
}

static unsigned long
nfs4_xattr_entry_count(struct shrinker *shrink, struct shrink_control *sc)
{
	unsigned long count;
	struct list_lru *lru;

	lru = (shrink == nfs4_xattr_large_entry_shrinker) ?
	    &nfs4_xattr_large_entry_lru : &nfs4_xattr_entry_lru;

	count = list_lru_shrink_count(lru, sc);
	return vfs_pressure_ratio(count);
}


static void nfs4_xattr_cache_init_once(void *p)
{
	struct nfs4_xattr_cache *cache = p;

	spin_lock_init(&cache->listxattr_lock);
	atomic_long_set(&cache->nent, 0);
	nfs4_xattr_hash_init(cache);
	cache->listxattr = NULL;
	INIT_LIST_HEAD(&cache->lru);
	INIT_LIST_HEAD(&cache->dispose);
}

typedef unsigned long (*count_objects_cb)(struct shrinker *s,
					  struct shrink_control *sc);
typedef unsigned long (*scan_objects_cb)(struct shrinker *s,
					 struct shrink_control *sc);

static int __init nfs4_xattr_shrinker_init(struct shrinker **shrinker,
					   struct list_lru *lru, const char *name,
					   count_objects_cb count,
					   scan_objects_cb scan, long batch, int seeks)
{
	int ret;

	*shrinker = shrinker_alloc(SHRINKER_MEMCG_AWARE, name);
	if (!*shrinker)
		return -ENOMEM;

	ret = list_lru_init_memcg(lru, *shrinker);
	if (ret) {
		shrinker_free(*shrinker);
		return ret;
	}

	(*shrinker)->count_objects = count;
	(*shrinker)->scan_objects = scan;
	(*shrinker)->batch = batch;
	(*shrinker)->seeks = seeks;

	shrinker_register(*shrinker);

	return ret;
}

static void nfs4_xattr_shrinker_destroy(struct shrinker *shrinker,
					struct list_lru *lru)
{
	shrinker_free(shrinker);
	list_lru_destroy(lru);
}

int __init nfs4_xattr_cache_init(void)
{
	int ret = 0;

	nfs4_xattr_cache_cachep = kmem_cache_create("nfs4_xattr_cache_cache",
	    sizeof(struct nfs4_xattr_cache), 0,
	    (SLAB_RECLAIM_ACCOUNT),
	    nfs4_xattr_cache_init_once);
	if (nfs4_xattr_cache_cachep == NULL)
		return -ENOMEM;

	ret = nfs4_xattr_shrinker_init(&nfs4_xattr_cache_shrinker,
				       &nfs4_xattr_cache_lru, "nfs-xattr_cache",
				       nfs4_xattr_cache_count,
				       nfs4_xattr_cache_scan, 0, DEFAULT_SEEKS);
	if (ret)
		goto out1;

	ret = nfs4_xattr_shrinker_init(&nfs4_xattr_entry_shrinker,
				       &nfs4_xattr_entry_lru, "nfs-xattr_entry",
				       nfs4_xattr_entry_count,
				       nfs4_xattr_entry_scan, 512, DEFAULT_SEEKS);
	if (ret)
		goto out2;

	ret = nfs4_xattr_shrinker_init(&nfs4_xattr_large_entry_shrinker,
				       &nfs4_xattr_large_entry_lru,
				       "nfs-xattr_large_entry",
				       nfs4_xattr_entry_count,
				       nfs4_xattr_entry_scan, 512, 1);
	if (!ret)
		return 0;

	nfs4_xattr_shrinker_destroy(nfs4_xattr_entry_shrinker,
				    &nfs4_xattr_entry_lru);
out2:
	nfs4_xattr_shrinker_destroy(nfs4_xattr_cache_shrinker,
				    &nfs4_xattr_cache_lru);
out1:
	kmem_cache_destroy(nfs4_xattr_cache_cachep);

	return ret;
}

void nfs4_xattr_cache_exit(void)
{
	nfs4_xattr_shrinker_destroy(nfs4_xattr_large_entry_shrinker,
				    &nfs4_xattr_large_entry_lru);
	nfs4_xattr_shrinker_destroy(nfs4_xattr_entry_shrinker,
				    &nfs4_xattr_entry_lru);
	nfs4_xattr_shrinker_destroy(nfs4_xattr_cache_shrinker,
				    &nfs4_xattr_cache_lru);
	kmem_cache_destroy(nfs4_xattr_cache_cachep);
}