summaryrefslogtreecommitdiff
path: root/fs/f2fs/extent_cache.c
blob: 342af24b2f8cf7d8237ff05b0db6cd92e51f5e8c (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
// SPDX-License-Identifier: GPL-2.0
/*
 * f2fs extent cache support
 *
 * Copyright (c) 2015 Motorola Mobility
 * Copyright (c) 2015 Samsung Electronics
 * Authors: Jaegeuk Kim <jaegeuk@kernel.org>
 *          Chao Yu <chao2.yu@samsung.com>
 *
 * block_age-based extent cache added by:
 * Copyright (c) 2022 xiaomi Co., Ltd.
 *             http://www.xiaomi.com/
 */

#include <linux/fs.h>
#include <linux/f2fs_fs.h>

#include "f2fs.h"
#include "node.h"
#include <trace/events/f2fs.h>

static void __set_extent_info(struct extent_info *ei,
				unsigned int fofs, unsigned int len,
				block_t blk, bool keep_clen,
				unsigned long age, unsigned long last_blocks,
				enum extent_type type)
{
	ei->fofs = fofs;
	ei->len = len;

	if (type == EX_READ) {
		ei->blk = blk;
		if (keep_clen)
			return;
#ifdef CONFIG_F2FS_FS_COMPRESSION
		ei->c_len = 0;
#endif
	} else if (type == EX_BLOCK_AGE) {
		ei->age = age;
		ei->last_blocks = last_blocks;
	}
}

static bool __may_read_extent_tree(struct inode *inode)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);

	if (!test_opt(sbi, READ_EXTENT_CACHE))
		return false;
	if (is_inode_flag_set(inode, FI_NO_EXTENT))
		return false;
	if (is_inode_flag_set(inode, FI_COMPRESSED_FILE) &&
			 !f2fs_sb_has_readonly(sbi))
		return false;
	return S_ISREG(inode->i_mode);
}

static bool __may_age_extent_tree(struct inode *inode)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);

	if (!test_opt(sbi, AGE_EXTENT_CACHE))
		return false;
	/* don't cache block age info for cold file */
	if (is_inode_flag_set(inode, FI_COMPRESSED_FILE))
		return false;
	if (file_is_cold(inode))
		return false;

	return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
}

static bool __init_may_extent_tree(struct inode *inode, enum extent_type type)
{
	if (type == EX_READ)
		return __may_read_extent_tree(inode);
	else if (type == EX_BLOCK_AGE)
		return __may_age_extent_tree(inode);
	return false;
}

static bool __may_extent_tree(struct inode *inode, enum extent_type type)
{
	/*
	 * for recovered files during mount do not create extents
	 * if shrinker is not registered.
	 */
	if (list_empty(&F2FS_I_SB(inode)->s_list))
		return false;

	return __init_may_extent_tree(inode, type);
}

static void __try_update_largest_extent(struct extent_tree *et,
						struct extent_node *en)
{
	if (et->type != EX_READ)
		return;
	if (en->ei.len <= et->largest.len)
		return;

	et->largest = en->ei;
	et->largest_updated = true;
}

static bool __is_extent_mergeable(struct extent_info *back,
		struct extent_info *front, enum extent_type type)
{
	if (type == EX_READ) {
#ifdef CONFIG_F2FS_FS_COMPRESSION
		if (back->c_len && back->len != back->c_len)
			return false;
		if (front->c_len && front->len != front->c_len)
			return false;
#endif
		return (back->fofs + back->len == front->fofs &&
				back->blk + back->len == front->blk);
	} else if (type == EX_BLOCK_AGE) {
		return (back->fofs + back->len == front->fofs &&
			abs(back->age - front->age) <= SAME_AGE_REGION &&
			abs(back->last_blocks - front->last_blocks) <=
							SAME_AGE_REGION);
	}
	return false;
}

static bool __is_back_mergeable(struct extent_info *cur,
		struct extent_info *back, enum extent_type type)
{
	return __is_extent_mergeable(back, cur, type);
}

static bool __is_front_mergeable(struct extent_info *cur,
		struct extent_info *front, enum extent_type type)
{
	return __is_extent_mergeable(cur, front, type);
}

static struct rb_entry *__lookup_rb_tree_fast(struct rb_entry *cached_re,
							unsigned int ofs)
{
	if (cached_re) {
		if (cached_re->ofs <= ofs &&
				cached_re->ofs + cached_re->len > ofs) {
			return cached_re;
		}
	}
	return NULL;
}

static struct rb_entry *__lookup_rb_tree_slow(struct rb_root_cached *root,
							unsigned int ofs)
{
	struct rb_node *node = root->rb_root.rb_node;
	struct rb_entry *re;

	while (node) {
		re = rb_entry(node, struct rb_entry, rb_node);

		if (ofs < re->ofs)
			node = node->rb_left;
		else if (ofs >= re->ofs + re->len)
			node = node->rb_right;
		else
			return re;
	}
	return NULL;
}

struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
				struct rb_entry *cached_re, unsigned int ofs)
{
	struct rb_entry *re;

	re = __lookup_rb_tree_fast(cached_re, ofs);
	if (!re)
		return __lookup_rb_tree_slow(root, ofs);

	return re;
}

struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
					struct rb_root_cached *root,
					struct rb_node **parent,
					unsigned long long key, bool *leftmost)
{
	struct rb_node **p = &root->rb_root.rb_node;
	struct rb_entry *re;

	while (*p) {
		*parent = *p;
		re = rb_entry(*parent, struct rb_entry, rb_node);

		if (key < re->key) {
			p = &(*p)->rb_left;
		} else {
			p = &(*p)->rb_right;
			*leftmost = false;
		}
	}

	return p;
}

struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
				struct rb_root_cached *root,
				struct rb_node **parent,
				unsigned int ofs, bool *leftmost)
{
	struct rb_node **p = &root->rb_root.rb_node;
	struct rb_entry *re;

	while (*p) {
		*parent = *p;
		re = rb_entry(*parent, struct rb_entry, rb_node);

		if (ofs < re->ofs) {
			p = &(*p)->rb_left;
		} else if (ofs >= re->ofs + re->len) {
			p = &(*p)->rb_right;
			*leftmost = false;
		} else {
			f2fs_bug_on(sbi, 1);
		}
	}

	return p;
}

/*
 * lookup rb entry in position of @ofs in rb-tree,
 * if hit, return the entry, otherwise, return NULL
 * @prev_ex: extent before ofs
 * @next_ex: extent after ofs
 * @insert_p: insert point for new extent at ofs
 * in order to simpfy the insertion after.
 * tree must stay unchanged between lookup and insertion.
 */
struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
				struct rb_entry *cached_re,
				unsigned int ofs,
				struct rb_entry **prev_entry,
				struct rb_entry **next_entry,
				struct rb_node ***insert_p,
				struct rb_node **insert_parent,
				bool force, bool *leftmost)
{
	struct rb_node **pnode = &root->rb_root.rb_node;
	struct rb_node *parent = NULL, *tmp_node;
	struct rb_entry *re = cached_re;

	*insert_p = NULL;
	*insert_parent = NULL;
	*prev_entry = NULL;
	*next_entry = NULL;

	if (RB_EMPTY_ROOT(&root->rb_root))
		return NULL;

	if (re) {
		if (re->ofs <= ofs && re->ofs + re->len > ofs)
			goto lookup_neighbors;
	}

	if (leftmost)
		*leftmost = true;

	while (*pnode) {
		parent = *pnode;
		re = rb_entry(*pnode, struct rb_entry, rb_node);

		if (ofs < re->ofs) {
			pnode = &(*pnode)->rb_left;
		} else if (ofs >= re->ofs + re->len) {
			pnode = &(*pnode)->rb_right;
			if (leftmost)
				*leftmost = false;
		} else {
			goto lookup_neighbors;
		}
	}

	*insert_p = pnode;
	*insert_parent = parent;

	re = rb_entry(parent, struct rb_entry, rb_node);
	tmp_node = parent;
	if (parent && ofs > re->ofs)
		tmp_node = rb_next(parent);
	*next_entry = rb_entry_safe(tmp_node, struct rb_entry, rb_node);

	tmp_node = parent;
	if (parent && ofs < re->ofs)
		tmp_node = rb_prev(parent);
	*prev_entry = rb_entry_safe(tmp_node, struct rb_entry, rb_node);
	return NULL;

lookup_neighbors:
	if (ofs == re->ofs || force) {
		/* lookup prev node for merging backward later */
		tmp_node = rb_prev(&re->rb_node);
		*prev_entry = rb_entry_safe(tmp_node, struct rb_entry, rb_node);
	}
	if (ofs == re->ofs + re->len - 1 || force) {
		/* lookup next node for merging frontward later */
		tmp_node = rb_next(&re->rb_node);
		*next_entry = rb_entry_safe(tmp_node, struct rb_entry, rb_node);
	}
	return re;
}

bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
				struct rb_root_cached *root, bool check_key)
{
#ifdef CONFIG_F2FS_CHECK_FS
	struct rb_node *cur = rb_first_cached(root), *next;
	struct rb_entry *cur_re, *next_re;

	if (!cur)
		return true;

	while (cur) {
		next = rb_next(cur);
		if (!next)
			return true;

		cur_re = rb_entry(cur, struct rb_entry, rb_node);
		next_re = rb_entry(next, struct rb_entry, rb_node);

		if (check_key) {
			if (cur_re->key > next_re->key) {
				f2fs_info(sbi, "inconsistent rbtree, "
					"cur(%llu) next(%llu)",
					cur_re->key, next_re->key);
				return false;
			}
			goto next;
		}

		if (cur_re->ofs + cur_re->len > next_re->ofs) {
			f2fs_info(sbi, "inconsistent rbtree, cur(%u, %u) next(%u, %u)",
				  cur_re->ofs, cur_re->len,
				  next_re->ofs, next_re->len);
			return false;
		}
next:
		cur = next;
	}
#endif
	return true;
}

static struct kmem_cache *extent_tree_slab;
static struct kmem_cache *extent_node_slab;

static struct extent_node *__attach_extent_node(struct f2fs_sb_info *sbi,
				struct extent_tree *et, struct extent_info *ei,
				struct rb_node *parent, struct rb_node **p,
				bool leftmost)
{
	struct extent_tree_info *eti = &sbi->extent_tree[et->type];
	struct extent_node *en;

	en = f2fs_kmem_cache_alloc(extent_node_slab, GFP_ATOMIC, false, sbi);
	if (!en)
		return NULL;

	en->ei = *ei;
	INIT_LIST_HEAD(&en->list);
	en->et = et;

	rb_link_node(&en->rb_node, parent, p);
	rb_insert_color_cached(&en->rb_node, &et->root, leftmost);
	atomic_inc(&et->node_cnt);
	atomic_inc(&eti->total_ext_node);
	return en;
}

static void __detach_extent_node(struct f2fs_sb_info *sbi,
				struct extent_tree *et, struct extent_node *en)
{
	struct extent_tree_info *eti = &sbi->extent_tree[et->type];

	rb_erase_cached(&en->rb_node, &et->root);
	atomic_dec(&et->node_cnt);
	atomic_dec(&eti->total_ext_node);

	if (et->cached_en == en)
		et->cached_en = NULL;
	kmem_cache_free(extent_node_slab, en);
}

/*
 * Flow to release an extent_node:
 * 1. list_del_init
 * 2. __detach_extent_node
 * 3. kmem_cache_free.
 */
static void __release_extent_node(struct f2fs_sb_info *sbi,
			struct extent_tree *et, struct extent_node *en)
{
	struct extent_tree_info *eti = &sbi->extent_tree[et->type];

	spin_lock(&eti->extent_lock);
	f2fs_bug_on(sbi, list_empty(&en->list));
	list_del_init(&en->list);
	spin_unlock(&eti->extent_lock);

	__detach_extent_node(sbi, et, en);
}

static struct extent_tree *__grab_extent_tree(struct inode *inode,
						enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree_info *eti = &sbi->extent_tree[type];
	struct extent_tree *et;
	nid_t ino = inode->i_ino;

	mutex_lock(&eti->extent_tree_lock);
	et = radix_tree_lookup(&eti->extent_tree_root, ino);
	if (!et) {
		et = f2fs_kmem_cache_alloc(extent_tree_slab,
					GFP_NOFS, true, NULL);
		f2fs_radix_tree_insert(&eti->extent_tree_root, ino, et);
		memset(et, 0, sizeof(struct extent_tree));
		et->ino = ino;
		et->type = type;
		et->root = RB_ROOT_CACHED;
		et->cached_en = NULL;
		rwlock_init(&et->lock);
		INIT_LIST_HEAD(&et->list);
		atomic_set(&et->node_cnt, 0);
		atomic_inc(&eti->total_ext_tree);
	} else {
		atomic_dec(&eti->total_zombie_tree);
		list_del_init(&et->list);
	}
	mutex_unlock(&eti->extent_tree_lock);

	/* never died until evict_inode */
	F2FS_I(inode)->extent_tree[type] = et;

	return et;
}

static unsigned int __free_extent_tree(struct f2fs_sb_info *sbi,
					struct extent_tree *et)
{
	struct rb_node *node, *next;
	struct extent_node *en;
	unsigned int count = atomic_read(&et->node_cnt);

	node = rb_first_cached(&et->root);
	while (node) {
		next = rb_next(node);
		en = rb_entry(node, struct extent_node, rb_node);
		__release_extent_node(sbi, et, en);
		node = next;
	}

	return count - atomic_read(&et->node_cnt);
}

static void __drop_largest_extent(struct extent_tree *et,
					pgoff_t fofs, unsigned int len)
{
	if (fofs < et->largest.fofs + et->largest.len &&
			fofs + len > et->largest.fofs) {
		et->largest.len = 0;
		et->largest_updated = true;
	}
}

void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree_info *eti = &sbi->extent_tree[EX_READ];
	struct f2fs_extent *i_ext = &F2FS_INODE(ipage)->i_ext;
	struct extent_tree *et;
	struct extent_node *en;
	struct extent_info ei;

	if (!__may_extent_tree(inode, EX_READ)) {
		/* drop largest read extent */
		if (i_ext && i_ext->len) {
			f2fs_wait_on_page_writeback(ipage, NODE, true, true);
			i_ext->len = 0;
			set_page_dirty(ipage);
		}
		goto out;
	}

	et = __grab_extent_tree(inode, EX_READ);

	if (!i_ext || !i_ext->len)
		goto out;

	get_read_extent_info(&ei, i_ext);

	write_lock(&et->lock);
	if (atomic_read(&et->node_cnt))
		goto unlock_out;

	en = __attach_extent_node(sbi, et, &ei, NULL,
				&et->root.rb_root.rb_node, true);
	if (en) {
		et->largest = en->ei;
		et->cached_en = en;

		spin_lock(&eti->extent_lock);
		list_add_tail(&en->list, &eti->extent_list);
		spin_unlock(&eti->extent_lock);
	}
unlock_out:
	write_unlock(&et->lock);
out:
	if (!F2FS_I(inode)->extent_tree[EX_READ])
		set_inode_flag(inode, FI_NO_EXTENT);
}

void f2fs_init_age_extent_tree(struct inode *inode)
{
	if (!__init_may_extent_tree(inode, EX_BLOCK_AGE))
		return;
	__grab_extent_tree(inode, EX_BLOCK_AGE);
}

void f2fs_init_extent_tree(struct inode *inode)
{
	/* initialize read cache */
	if (__init_may_extent_tree(inode, EX_READ))
		__grab_extent_tree(inode, EX_READ);

	/* initialize block age cache */
	if (__init_may_extent_tree(inode, EX_BLOCK_AGE))
		__grab_extent_tree(inode, EX_BLOCK_AGE);
}

static bool __lookup_extent_tree(struct inode *inode, pgoff_t pgofs,
			struct extent_info *ei, enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree_info *eti = &sbi->extent_tree[type];
	struct extent_tree *et = F2FS_I(inode)->extent_tree[type];
	struct extent_node *en;
	bool ret = false;

	if (!et)
		return false;

	trace_f2fs_lookup_extent_tree_start(inode, pgofs, type);

	read_lock(&et->lock);

	if (type == EX_READ &&
			et->largest.fofs <= pgofs &&
			et->largest.fofs + et->largest.len > pgofs) {
		*ei = et->largest;
		ret = true;
		stat_inc_largest_node_hit(sbi);
		goto out;
	}

	en = (struct extent_node *)f2fs_lookup_rb_tree(&et->root,
				(struct rb_entry *)et->cached_en, pgofs);
	if (!en)
		goto out;

	if (en == et->cached_en)
		stat_inc_cached_node_hit(sbi, type);
	else
		stat_inc_rbtree_node_hit(sbi, type);

	*ei = en->ei;
	spin_lock(&eti->extent_lock);
	if (!list_empty(&en->list)) {
		list_move_tail(&en->list, &eti->extent_list);
		et->cached_en = en;
	}
	spin_unlock(&eti->extent_lock);
	ret = true;
out:
	stat_inc_total_hit(sbi, type);
	read_unlock(&et->lock);

	if (type == EX_READ)
		trace_f2fs_lookup_read_extent_tree_end(inode, pgofs, ei);
	else if (type == EX_BLOCK_AGE)
		trace_f2fs_lookup_age_extent_tree_end(inode, pgofs, ei);
	return ret;
}

static struct extent_node *__try_merge_extent_node(struct f2fs_sb_info *sbi,
				struct extent_tree *et, struct extent_info *ei,
				struct extent_node *prev_ex,
				struct extent_node *next_ex)
{
	struct extent_tree_info *eti = &sbi->extent_tree[et->type];
	struct extent_node *en = NULL;

	if (prev_ex && __is_back_mergeable(ei, &prev_ex->ei, et->type)) {
		prev_ex->ei.len += ei->len;
		ei = &prev_ex->ei;
		en = prev_ex;
	}

	if (next_ex && __is_front_mergeable(ei, &next_ex->ei, et->type)) {
		next_ex->ei.fofs = ei->fofs;
		next_ex->ei.len += ei->len;
		if (et->type == EX_READ)
			next_ex->ei.blk = ei->blk;
		if (en)
			__release_extent_node(sbi, et, prev_ex);

		en = next_ex;
	}

	if (!en)
		return NULL;

	__try_update_largest_extent(et, en);

	spin_lock(&eti->extent_lock);
	if (!list_empty(&en->list)) {
		list_move_tail(&en->list, &eti->extent_list);
		et->cached_en = en;
	}
	spin_unlock(&eti->extent_lock);
	return en;
}

static struct extent_node *__insert_extent_tree(struct f2fs_sb_info *sbi,
				struct extent_tree *et, struct extent_info *ei,
				struct rb_node **insert_p,
				struct rb_node *insert_parent,
				bool leftmost)
{
	struct extent_tree_info *eti = &sbi->extent_tree[et->type];
	struct rb_node **p;
	struct rb_node *parent = NULL;
	struct extent_node *en = NULL;

	if (insert_p && insert_parent) {
		parent = insert_parent;
		p = insert_p;
		goto do_insert;
	}

	leftmost = true;

	p = f2fs_lookup_rb_tree_for_insert(sbi, &et->root, &parent,
						ei->fofs, &leftmost);
do_insert:
	en = __attach_extent_node(sbi, et, ei, parent, p, leftmost);
	if (!en)
		return NULL;

	__try_update_largest_extent(et, en);

	/* update in global extent list */
	spin_lock(&eti->extent_lock);
	list_add_tail(&en->list, &eti->extent_list);
	et->cached_en = en;
	spin_unlock(&eti->extent_lock);
	return en;
}

static void __update_extent_tree_range(struct inode *inode,
			struct extent_info *tei, enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree *et = F2FS_I(inode)->extent_tree[type];
	struct extent_node *en = NULL, *en1 = NULL;
	struct extent_node *prev_en = NULL, *next_en = NULL;
	struct extent_info ei, dei, prev;
	struct rb_node **insert_p = NULL, *insert_parent = NULL;
	unsigned int fofs = tei->fofs, len = tei->len;
	unsigned int end = fofs + len;
	bool updated = false;
	bool leftmost = false;

	if (!et)
		return;

	if (type == EX_READ)
		trace_f2fs_update_read_extent_tree_range(inode, fofs, len,
						tei->blk, 0);
	else if (type == EX_BLOCK_AGE)
		trace_f2fs_update_age_extent_tree_range(inode, fofs, len,
						tei->age, tei->last_blocks);

	write_lock(&et->lock);

	if (type == EX_READ) {
		if (is_inode_flag_set(inode, FI_NO_EXTENT)) {
			write_unlock(&et->lock);
			return;
		}

		prev = et->largest;
		dei.len = 0;

		/*
		 * drop largest extent before lookup, in case it's already
		 * been shrunk from extent tree
		 */
		__drop_largest_extent(et, fofs, len);
	}

	/* 1. lookup first extent node in range [fofs, fofs + len - 1] */
	en = (struct extent_node *)f2fs_lookup_rb_tree_ret(&et->root,
					(struct rb_entry *)et->cached_en, fofs,
					(struct rb_entry **)&prev_en,
					(struct rb_entry **)&next_en,
					&insert_p, &insert_parent, false,
					&leftmost);
	if (!en)
		en = next_en;

	/* 2. invlidate all extent nodes in range [fofs, fofs + len - 1] */
	while (en && en->ei.fofs < end) {
		unsigned int org_end;
		int parts = 0;	/* # of parts current extent split into */

		next_en = en1 = NULL;

		dei = en->ei;
		org_end = dei.fofs + dei.len;
		f2fs_bug_on(sbi, fofs >= org_end);

		if (fofs > dei.fofs && (type != EX_READ ||
				fofs - dei.fofs >= F2FS_MIN_EXTENT_LEN)) {
			en->ei.len = fofs - en->ei.fofs;
			prev_en = en;
			parts = 1;
		}

		if (end < org_end && (type != EX_READ ||
				org_end - end >= F2FS_MIN_EXTENT_LEN)) {
			if (parts) {
				__set_extent_info(&ei,
					end, org_end - end,
					end - dei.fofs + dei.blk, false,
					dei.age, dei.last_blocks,
					type);
				en1 = __insert_extent_tree(sbi, et, &ei,
							NULL, NULL, true);
				next_en = en1;
			} else {
				__set_extent_info(&en->ei,
					end, en->ei.len - (end - dei.fofs),
					en->ei.blk + (end - dei.fofs), true,
					dei.age, dei.last_blocks,
					type);
				next_en = en;
			}
			parts++;
		}

		if (!next_en) {
			struct rb_node *node = rb_next(&en->rb_node);

			next_en = rb_entry_safe(node, struct extent_node,
						rb_node);
		}

		if (parts)
			__try_update_largest_extent(et, en);
		else
			__release_extent_node(sbi, et, en);

		/*
		 * if original extent is split into zero or two parts, extent
		 * tree has been altered by deletion or insertion, therefore
		 * invalidate pointers regard to tree.
		 */
		if (parts != 1) {
			insert_p = NULL;
			insert_parent = NULL;
		}
		en = next_en;
	}

	if (type == EX_BLOCK_AGE)
		goto update_age_extent_cache;

	/* 3. update extent in read extent cache */
	BUG_ON(type != EX_READ);

	if (tei->blk) {
		__set_extent_info(&ei, fofs, len, tei->blk, false,
				  0, 0, EX_READ);
		if (!__try_merge_extent_node(sbi, et, &ei, prev_en, next_en))
			__insert_extent_tree(sbi, et, &ei,
					insert_p, insert_parent, leftmost);

		/* give up extent_cache, if split and small updates happen */
		if (dei.len >= 1 &&
				prev.len < F2FS_MIN_EXTENT_LEN &&
				et->largest.len < F2FS_MIN_EXTENT_LEN) {
			et->largest.len = 0;
			et->largest_updated = true;
			set_inode_flag(inode, FI_NO_EXTENT);
		}
	}

	if (is_inode_flag_set(inode, FI_NO_EXTENT))
		__free_extent_tree(sbi, et);

	if (et->largest_updated) {
		et->largest_updated = false;
		updated = true;
	}
	goto out_read_extent_cache;
update_age_extent_cache:
	if (!tei->last_blocks)
		goto out_read_extent_cache;

	__set_extent_info(&ei, fofs, len, 0, false,
			tei->age, tei->last_blocks, EX_BLOCK_AGE);
	if (!__try_merge_extent_node(sbi, et, &ei, prev_en, next_en))
		__insert_extent_tree(sbi, et, &ei,
					insert_p, insert_parent, leftmost);
out_read_extent_cache:
	write_unlock(&et->lock);

	if (updated)
		f2fs_mark_inode_dirty_sync(inode, true);
}

#ifdef CONFIG_F2FS_FS_COMPRESSION
void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,
				pgoff_t fofs, block_t blkaddr, unsigned int llen,
				unsigned int c_len)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree *et = F2FS_I(inode)->extent_tree[EX_READ];
	struct extent_node *en = NULL;
	struct extent_node *prev_en = NULL, *next_en = NULL;
	struct extent_info ei;
	struct rb_node **insert_p = NULL, *insert_parent = NULL;
	bool leftmost = false;

	trace_f2fs_update_read_extent_tree_range(inode, fofs, llen,
						blkaddr, c_len);

	/* it is safe here to check FI_NO_EXTENT w/o et->lock in ro image */
	if (is_inode_flag_set(inode, FI_NO_EXTENT))
		return;

	write_lock(&et->lock);

	en = (struct extent_node *)f2fs_lookup_rb_tree_ret(&et->root,
				(struct rb_entry *)et->cached_en, fofs,
				(struct rb_entry **)&prev_en,
				(struct rb_entry **)&next_en,
				&insert_p, &insert_parent, false,
				&leftmost);
	if (en)
		goto unlock_out;

	__set_extent_info(&ei, fofs, llen, blkaddr, true, 0, 0, EX_READ);
	ei.c_len = c_len;

	if (!__try_merge_extent_node(sbi, et, &ei, prev_en, next_en))
		__insert_extent_tree(sbi, et, &ei,
				insert_p, insert_parent, leftmost);
unlock_out:
	write_unlock(&et->lock);
}
#endif

static unsigned long long __calculate_block_age(unsigned long long new,
						unsigned long long old)
{
	unsigned long long diff;

	diff = (new >= old) ? new - (new - old) : new + (old - new);

	return div_u64(diff * LAST_AGE_WEIGHT, 100);
}

/* This returns a new age and allocated blocks in ei */
static int __get_new_block_age(struct inode *inode, struct extent_info *ei,
						block_t blkaddr)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	loff_t f_size = i_size_read(inode);
	unsigned long long cur_blocks =
				atomic64_read(&sbi->allocated_data_blocks);
	struct extent_info tei = *ei;	/* only fofs and len are valid */

	/*
	 * When I/O is not aligned to a PAGE_SIZE, update will happen to the last
	 * file block even in seq write. So don't record age for newly last file
	 * block here.
	 */
	if ((f_size >> PAGE_SHIFT) == ei->fofs && f_size & (PAGE_SIZE - 1) &&
			blkaddr == NEW_ADDR)
		return -EINVAL;

	if (__lookup_extent_tree(inode, ei->fofs, &tei, EX_BLOCK_AGE)) {
		unsigned long long cur_age;

		if (cur_blocks >= tei.last_blocks)
			cur_age = cur_blocks - tei.last_blocks;
		else
			/* allocated_data_blocks overflow */
			cur_age = ULLONG_MAX - tei.last_blocks + cur_blocks;

		if (tei.age)
			ei->age = __calculate_block_age(cur_age, tei.age);
		else
			ei->age = cur_age;
		ei->last_blocks = cur_blocks;
		WARN_ON(ei->age > cur_blocks);
		return 0;
	}

	f2fs_bug_on(sbi, blkaddr == NULL_ADDR);

	/* the data block was allocated for the first time */
	if (blkaddr == NEW_ADDR)
		goto out;

	if (__is_valid_data_blkaddr(blkaddr) &&
	    !f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC_ENHANCE)) {
		f2fs_bug_on(sbi, 1);
		return -EINVAL;
	}
out:
	/*
	 * init block age with zero, this can happen when the block age extent
	 * was reclaimed due to memory constraint or system reboot
	 */
	ei->age = 0;
	ei->last_blocks = cur_blocks;
	return 0;
}

static void __update_extent_cache(struct dnode_of_data *dn, enum extent_type type)
{
	struct extent_info ei = {};

	if (!__may_extent_tree(dn->inode, type))
		return;

	ei.fofs = f2fs_start_bidx_of_node(ofs_of_node(dn->node_page), dn->inode) +
								dn->ofs_in_node;
	ei.len = 1;

	if (type == EX_READ) {
		if (dn->data_blkaddr == NEW_ADDR)
			ei.blk = NULL_ADDR;
		else
			ei.blk = dn->data_blkaddr;
	} else if (type == EX_BLOCK_AGE) {
		if (__get_new_block_age(dn->inode, &ei, dn->data_blkaddr))
			return;
	}
	__update_extent_tree_range(dn->inode, &ei, type);
}

static unsigned int __shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink,
					enum extent_type type)
{
	struct extent_tree_info *eti = &sbi->extent_tree[type];
	struct extent_tree *et, *next;
	struct extent_node *en;
	unsigned int node_cnt = 0, tree_cnt = 0;
	int remained;

	if (!atomic_read(&eti->total_zombie_tree))
		goto free_node;

	if (!mutex_trylock(&eti->extent_tree_lock))
		goto out;

	/* 1. remove unreferenced extent tree */
	list_for_each_entry_safe(et, next, &eti->zombie_list, list) {
		if (atomic_read(&et->node_cnt)) {
			write_lock(&et->lock);
			node_cnt += __free_extent_tree(sbi, et);
			write_unlock(&et->lock);
		}
		f2fs_bug_on(sbi, atomic_read(&et->node_cnt));
		list_del_init(&et->list);
		radix_tree_delete(&eti->extent_tree_root, et->ino);
		kmem_cache_free(extent_tree_slab, et);
		atomic_dec(&eti->total_ext_tree);
		atomic_dec(&eti->total_zombie_tree);
		tree_cnt++;

		if (node_cnt + tree_cnt >= nr_shrink)
			goto unlock_out;
		cond_resched();
	}
	mutex_unlock(&eti->extent_tree_lock);

free_node:
	/* 2. remove LRU extent entries */
	if (!mutex_trylock(&eti->extent_tree_lock))
		goto out;

	remained = nr_shrink - (node_cnt + tree_cnt);

	spin_lock(&eti->extent_lock);
	for (; remained > 0; remained--) {
		if (list_empty(&eti->extent_list))
			break;
		en = list_first_entry(&eti->extent_list,
					struct extent_node, list);
		et = en->et;
		if (!write_trylock(&et->lock)) {
			/* refresh this extent node's position in extent list */
			list_move_tail(&en->list, &eti->extent_list);
			continue;
		}

		list_del_init(&en->list);
		spin_unlock(&eti->extent_lock);

		__detach_extent_node(sbi, et, en);

		write_unlock(&et->lock);
		node_cnt++;
		spin_lock(&eti->extent_lock);
	}
	spin_unlock(&eti->extent_lock);

unlock_out:
	mutex_unlock(&eti->extent_tree_lock);
out:
	trace_f2fs_shrink_extent_tree(sbi, node_cnt, tree_cnt, type);

	return node_cnt + tree_cnt;
}

/* read extent cache operations */
bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,
				struct extent_info *ei)
{
	if (!__may_extent_tree(inode, EX_READ))
		return false;

	return __lookup_extent_tree(inode, pgofs, ei, EX_READ);
}

void f2fs_update_read_extent_cache(struct dnode_of_data *dn)
{
	return __update_extent_cache(dn, EX_READ);
}

void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,
				pgoff_t fofs, block_t blkaddr, unsigned int len)
{
	struct extent_info ei = {
		.fofs = fofs,
		.len = len,
		.blk = blkaddr,
	};

	if (!__may_extent_tree(dn->inode, EX_READ))
		return;

	__update_extent_tree_range(dn->inode, &ei, EX_READ);
}

unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
{
	if (!test_opt(sbi, READ_EXTENT_CACHE))
		return 0;

	return __shrink_extent_tree(sbi, nr_shrink, EX_READ);
}

/* block age extent cache operations */
bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,
				struct extent_info *ei)
{
	if (!__may_extent_tree(inode, EX_BLOCK_AGE))
		return false;

	return __lookup_extent_tree(inode, pgofs, ei, EX_BLOCK_AGE);
}

void f2fs_update_age_extent_cache(struct dnode_of_data *dn)
{
	return __update_extent_cache(dn, EX_BLOCK_AGE);
}

void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,
				pgoff_t fofs, unsigned int len)
{
	struct extent_info ei = {
		.fofs = fofs,
		.len = len,
	};

	if (!__may_extent_tree(dn->inode, EX_BLOCK_AGE))
		return;

	__update_extent_tree_range(dn->inode, &ei, EX_BLOCK_AGE);
}

unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
{
	if (!test_opt(sbi, AGE_EXTENT_CACHE))
		return 0;

	return __shrink_extent_tree(sbi, nr_shrink, EX_BLOCK_AGE);
}

static unsigned int __destroy_extent_node(struct inode *inode,
					enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree *et = F2FS_I(inode)->extent_tree[type];
	unsigned int node_cnt = 0;

	if (!et || !atomic_read(&et->node_cnt))
		return 0;

	write_lock(&et->lock);
	node_cnt = __free_extent_tree(sbi, et);
	write_unlock(&et->lock);

	return node_cnt;
}

void f2fs_destroy_extent_node(struct inode *inode)
{
	__destroy_extent_node(inode, EX_READ);
	__destroy_extent_node(inode, EX_BLOCK_AGE);
}

static void __drop_extent_tree(struct inode *inode, enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree *et = F2FS_I(inode)->extent_tree[type];
	bool updated = false;

	if (!__may_extent_tree(inode, type))
		return;

	write_lock(&et->lock);
	__free_extent_tree(sbi, et);
	if (type == EX_READ) {
		set_inode_flag(inode, FI_NO_EXTENT);
		if (et->largest.len) {
			et->largest.len = 0;
			updated = true;
		}
	}
	write_unlock(&et->lock);
	if (updated)
		f2fs_mark_inode_dirty_sync(inode, true);
}

void f2fs_drop_extent_tree(struct inode *inode)
{
	__drop_extent_tree(inode, EX_READ);
	__drop_extent_tree(inode, EX_BLOCK_AGE);
}

static void __destroy_extent_tree(struct inode *inode, enum extent_type type)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct extent_tree_info *eti = &sbi->extent_tree[type];
	struct extent_tree *et = F2FS_I(inode)->extent_tree[type];
	unsigned int node_cnt = 0;

	if (!et)
		return;

	if (inode->i_nlink && !is_bad_inode(inode) &&
					atomic_read(&et->node_cnt)) {
		mutex_lock(&eti->extent_tree_lock);
		list_add_tail(&et->list, &eti->zombie_list);
		atomic_inc(&eti->total_zombie_tree);
		mutex_unlock(&eti->extent_tree_lock);
		return;
	}

	/* free all extent info belong to this extent tree */
	node_cnt = __destroy_extent_node(inode, type);

	/* delete extent tree entry in radix tree */
	mutex_lock(&eti->extent_tree_lock);
	f2fs_bug_on(sbi, atomic_read(&et->node_cnt));
	radix_tree_delete(&eti->extent_tree_root, inode->i_ino);
	kmem_cache_free(extent_tree_slab, et);
	atomic_dec(&eti->total_ext_tree);
	mutex_unlock(&eti->extent_tree_lock);

	F2FS_I(inode)->extent_tree[type] = NULL;

	trace_f2fs_destroy_extent_tree(inode, node_cnt, type);
}

void f2fs_destroy_extent_tree(struct inode *inode)
{
	__destroy_extent_tree(inode, EX_READ);
	__destroy_extent_tree(inode, EX_BLOCK_AGE);
}

static void __init_extent_tree_info(struct extent_tree_info *eti)
{
	INIT_RADIX_TREE(&eti->extent_tree_root, GFP_NOIO);
	mutex_init(&eti->extent_tree_lock);
	INIT_LIST_HEAD(&eti->extent_list);
	spin_lock_init(&eti->extent_lock);
	atomic_set(&eti->total_ext_tree, 0);
	INIT_LIST_HEAD(&eti->zombie_list);
	atomic_set(&eti->total_zombie_tree, 0);
	atomic_set(&eti->total_ext_node, 0);
}

void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi)
{
	__init_extent_tree_info(&sbi->extent_tree[EX_READ]);
	__init_extent_tree_info(&sbi->extent_tree[EX_BLOCK_AGE]);

	/* initialize for block age extents */
	atomic64_set(&sbi->allocated_data_blocks, 0);
	sbi->hot_data_age_threshold = DEF_HOT_DATA_AGE_THRESHOLD;
	sbi->warm_data_age_threshold = DEF_WARM_DATA_AGE_THRESHOLD;
}

int __init f2fs_create_extent_cache(void)
{
	extent_tree_slab = f2fs_kmem_cache_create("f2fs_extent_tree",
			sizeof(struct extent_tree));
	if (!extent_tree_slab)
		return -ENOMEM;
	extent_node_slab = f2fs_kmem_cache_create("f2fs_extent_node",
			sizeof(struct extent_node));
	if (!extent_node_slab) {
		kmem_cache_destroy(extent_tree_slab);
		return -ENOMEM;
	}
	return 0;
}

void f2fs_destroy_extent_cache(void)
{
	kmem_cache_destroy(extent_node_slab);
	kmem_cache_destroy(extent_tree_slab);
}