blob: 7cad04739756639ef5c7598ff602432dbe71e98f [file] [log] [blame]
b.liue9582032025-04-17 19:18:16 +08001/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * fs/f2fs/f2fs.h
4 *
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
7 */
8#ifndef _LINUX_F2FS_H
9#define _LINUX_F2FS_H
10
11#include <linux/uio.h>
12#include <linux/types.h>
13#include <linux/page-flags.h>
14#include <linux/buffer_head.h>
15#include <linux/slab.h>
16#include <linux/crc32.h>
17#include <linux/magic.h>
18#include <linux/kobject.h>
19#include <linux/sched.h>
20#include <linux/cred.h>
21#include <linux/vmalloc.h>
22#include <linux/bio.h>
23#include <linux/blkdev.h>
24#include <linux/quotaops.h>
25#include <crypto/hash.h>
26
27#include <linux/fscrypt.h>
28#include <linux/fsverity.h>
29
30#ifdef CONFIG_F2FS_CHECK_FS
31#define f2fs_bug_on(sbi, condition) BUG_ON(condition)
32#else
33#define f2fs_bug_on(sbi, condition) \
34 do { \
35 if (WARN_ON(condition)) \
36 set_sbi_flag(sbi, SBI_NEED_FSCK); \
37 } while (0)
38#endif
39
40enum {
41 FAULT_KMALLOC,
42 FAULT_KVMALLOC,
43 FAULT_PAGE_ALLOC,
44 FAULT_PAGE_GET,
45 FAULT_ALLOC_NID,
46 FAULT_ORPHAN,
47 FAULT_BLOCK,
48 FAULT_DIR_DEPTH,
49 FAULT_EVICT_INODE,
50 FAULT_TRUNCATE,
51 FAULT_READ_IO,
52 FAULT_CHECKPOINT,
53 FAULT_DISCARD,
54 FAULT_WRITE_IO,
55 FAULT_MAX,
56};
57
58#ifdef CONFIG_F2FS_FAULT_INJECTION
59#define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
60
61struct f2fs_fault_info {
62 atomic_t inject_ops;
63 unsigned int inject_rate;
64 unsigned int inject_type;
65};
66
67extern const char *f2fs_fault_name[FAULT_MAX];
68#define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
69#endif
70
71/*
72 * For mount options
73 */
74#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
75#define F2FS_MOUNT_DISCARD 0x00000004
76#define F2FS_MOUNT_NOHEAP 0x00000008
77#define F2FS_MOUNT_XATTR_USER 0x00000010
78#define F2FS_MOUNT_POSIX_ACL 0x00000020
79#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
80#define F2FS_MOUNT_INLINE_XATTR 0x00000080
81#define F2FS_MOUNT_INLINE_DATA 0x00000100
82#define F2FS_MOUNT_INLINE_DENTRY 0x00000200
83#define F2FS_MOUNT_FLUSH_MERGE 0x00000400
84#define F2FS_MOUNT_NOBARRIER 0x00000800
85#define F2FS_MOUNT_FASTBOOT 0x00001000
86#define F2FS_MOUNT_EXTENT_CACHE 0x00002000
87#define F2FS_MOUNT_DATA_FLUSH 0x00008000
88#define F2FS_MOUNT_FAULT_INJECTION 0x00010000
89#define F2FS_MOUNT_USRQUOTA 0x00080000
90#define F2FS_MOUNT_GRPQUOTA 0x00100000
91#define F2FS_MOUNT_PRJQUOTA 0x00200000
92#define F2FS_MOUNT_QUOTA 0x00400000
93#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
94#define F2FS_MOUNT_RESERVE_ROOT 0x01000000
95#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
96#define F2FS_MOUNT_NORECOVERY 0x04000000
97#define F2FS_MOUNT_ATGC 0x08000000
98#define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000
99#define F2FS_MOUNT_GC_MERGE 0x20000000
100#define F2FS_MOUNT_COMPRESS_CACHE 0x40000000
101
102#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
103#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
104#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
105#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
106
107#define ver_after(a, b) (typecheck(unsigned long long, a) && \
108 typecheck(unsigned long long, b) && \
109 ((long long)((a) - (b)) > 0))
110
111typedef u32 block_t; /*
112 * should not change u32, since it is the on-disk block
113 * address format, __le32.
114 */
115typedef u32 nid_t;
116
117#define COMPRESS_EXT_NUM 16
118
119struct f2fs_mount_info {
120 unsigned int opt;
121 int write_io_size_bits; /* Write IO size bits */
122 block_t root_reserved_blocks; /* root reserved blocks */
123 kuid_t s_resuid; /* reserved blocks for uid */
124 kgid_t s_resgid; /* reserved blocks for gid */
125 int active_logs; /* # of active logs */
126 int inline_xattr_size; /* inline xattr size */
127#ifdef CONFIG_F2FS_FAULT_INJECTION
128 struct f2fs_fault_info fault_info; /* For fault injection */
129#endif
130#ifdef CONFIG_QUOTA
131 /* Names of quota files with journalled quota */
132 char *s_qf_names[MAXQUOTAS];
133 int s_jquota_fmt; /* Format of quota to use */
134#endif
135 /* For which write hints are passed down to block layer */
136 int whint_mode;
137 int alloc_mode; /* segment allocation policy */
138 int fsync_mode; /* fsync policy */
139 int fs_mode; /* fs mode: LFS or ADAPTIVE */
140 int bggc_mode; /* bggc mode: off, on or sync */
141 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
142 block_t unusable_cap_perc; /* percentage for cap */
143 block_t unusable_cap; /* Amount of space allowed to be
144 * unusable when disabling checkpoint
145 */
146
147 /* For compression */
148 unsigned char compress_algorithm; /* algorithm type */
149 unsigned char compress_log_size; /* cluster log size */
150 unsigned char compress_level; /* compress level */
151 bool compress_chksum; /* compressed data chksum */
152 unsigned char compress_ext_cnt; /* extension count */
153 int compress_mode; /* compression mode */
154 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
155};
156
157#define F2FS_FEATURE_ENCRYPT 0x0001
158#define F2FS_FEATURE_BLKZONED 0x0002
159#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
160#define F2FS_FEATURE_EXTRA_ATTR 0x0008
161#define F2FS_FEATURE_PRJQUOTA 0x0010
162#define F2FS_FEATURE_INODE_CHKSUM 0x0020
163#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
164#define F2FS_FEATURE_QUOTA_INO 0x0080
165#define F2FS_FEATURE_INODE_CRTIME 0x0100
166#define F2FS_FEATURE_LOST_FOUND 0x0200
167#define F2FS_FEATURE_VERITY 0x0400
168#define F2FS_FEATURE_SB_CHKSUM 0x0800
169#define F2FS_FEATURE_CASEFOLD 0x1000
170#define F2FS_FEATURE_COMPRESSION 0x2000
171#define F2FS_FEATURE_RO 0x4000
172
173#define __F2FS_HAS_FEATURE(raw_super, mask) \
174 ((raw_super->feature & cpu_to_le32(mask)) != 0)
175#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
176#define F2FS_SET_FEATURE(sbi, mask) \
177 (sbi->raw_super->feature |= cpu_to_le32(mask))
178#define F2FS_CLEAR_FEATURE(sbi, mask) \
179 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
180
181/*
182 * Default values for user and/or group using reserved blocks
183 */
184#define F2FS_DEF_RESUID 0
185#define F2FS_DEF_RESGID 0
186
187/*
188 * For checkpoint manager
189 */
190enum {
191 NAT_BITMAP,
192 SIT_BITMAP
193};
194
195#define CP_UMOUNT 0x00000001
196#define CP_FASTBOOT 0x00000002
197#define CP_SYNC 0x00000004
198#define CP_RECOVERY 0x00000008
199#define CP_DISCARD 0x00000010
200#define CP_TRIMMED 0x00000020
201#define CP_PAUSE 0x00000040
202#define CP_RESIZE 0x00000080
203
204#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
205#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
206#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
207#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
208#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
209#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
210#define DEF_CP_INTERVAL 60 /* 60 secs */
211#define DEF_IDLE_INTERVAL 5 /* 5 secs */
212#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
213#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
214#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
215
216struct cp_control {
217 int reason;
218 __u64 trim_start;
219 __u64 trim_end;
220 __u64 trim_minlen;
221};
222
223/*
224 * indicate meta/data type
225 */
226enum {
227 META_CP,
228 META_NAT,
229 META_SIT,
230 META_SSA,
231 META_MAX,
232 META_POR,
233 DATA_GENERIC, /* check range only */
234 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
235 DATA_GENERIC_ENHANCE_READ, /*
236 * strong check on range and segment
237 * bitmap but no warning due to race
238 * condition of read on truncated area
239 * by extent_cache
240 */
241 DATA_GENERIC_ENHANCE_UPDATE, /*
242 * strong check on range and segment
243 * bitmap for update case
244 */
245 META_GENERIC,
246};
247
248/* for the list of ino */
249enum {
250 ORPHAN_INO, /* for orphan ino list */
251 APPEND_INO, /* for append ino list */
252 UPDATE_INO, /* for update ino list */
253 TRANS_DIR_INO, /* for transactions dir ino list */
254 XATTR_DIR_INO, /* for xattr updated dir ino list */
255 FLUSH_INO, /* for multiple device flushing */
256 MAX_INO_ENTRY, /* max. list */
257};
258
259struct ino_entry {
260 struct list_head list; /* list head */
261 nid_t ino; /* inode number */
262 unsigned int dirty_device; /* dirty device bitmap */
263};
264
265/* for the list of inodes to be GCed */
266struct inode_entry {
267 struct list_head list; /* list head */
268 struct inode *inode; /* vfs inode pointer */
269};
270
271struct fsync_node_entry {
272 struct list_head list; /* list head */
273 struct page *page; /* warm node page pointer */
274 unsigned int seq_id; /* sequence id */
275};
276
277struct ckpt_req {
278 struct completion wait; /* completion for checkpoint done */
279 struct llist_node llnode; /* llist_node to be linked in wait queue */
280 int ret; /* return code of checkpoint */
281 ktime_t queue_time; /* request queued time */
282};
283
284struct ckpt_req_control {
285 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
286 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
287 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
288 atomic_t issued_ckpt; /* # of actually issued ckpts */
289 atomic_t total_ckpt; /* # of total ckpts */
290 atomic_t queued_ckpt; /* # of queued ckpts */
291 struct llist_head issue_list; /* list for command issue */
292 spinlock_t stat_lock; /* lock for below checkpoint time stats */
293 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
294 unsigned int peak_time; /* peak wait time in msec until now */
295};
296
297/* for the bitmap indicate blocks to be discarded */
298struct discard_entry {
299 struct list_head list; /* list head */
300 block_t start_blkaddr; /* start blockaddr of current segment */
301 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
302};
303
304/* default discard granularity of inner discard thread, unit: block count */
305#define DEFAULT_DISCARD_GRANULARITY 16
306
307/* max discard pend list number */
308#define MAX_PLIST_NUM 512
309#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
310 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
311
312enum {
313 D_PREP, /* initial */
314 D_PARTIAL, /* partially submitted */
315 D_SUBMIT, /* all submitted */
316 D_DONE, /* finished */
317};
318
319struct discard_info {
320 block_t lstart; /* logical start address */
321 block_t len; /* length */
322 block_t start; /* actual start address in dev */
323};
324
325struct discard_cmd {
326 struct rb_node rb_node; /* rb node located in rb-tree */
327 union {
328 struct {
329 block_t lstart; /* logical start address */
330 block_t len; /* length */
331 block_t start; /* actual start address in dev */
332 };
333 struct discard_info di; /* discard info */
334
335 };
336 struct list_head list; /* command list */
337 struct completion wait; /* compleation */
338 struct block_device *bdev; /* bdev */
339 unsigned short ref; /* reference count */
340 unsigned char state; /* state */
341 unsigned char queued; /* queued discard */
342 int error; /* bio error */
343 spinlock_t lock; /* for state/bio_ref updating */
344 unsigned short bio_ref; /* bio reference count */
345};
346
347enum {
348 DPOLICY_BG,
349 DPOLICY_FORCE,
350 DPOLICY_FSTRIM,
351 DPOLICY_UMOUNT,
352 MAX_DPOLICY,
353};
354
355struct discard_policy {
356 int type; /* type of discard */
357 unsigned int min_interval; /* used for candidates exist */
358 unsigned int mid_interval; /* used for device busy */
359 unsigned int max_interval; /* used for candidates not exist */
360 unsigned int max_requests; /* # of discards issued per round */
361 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
362 bool io_aware; /* issue discard in idle time */
363 bool sync; /* submit discard with REQ_SYNC flag */
364 bool ordered; /* issue discard by lba order */
365 bool timeout; /* discard timeout for put_super */
366 unsigned int granularity; /* discard granularity */
367};
368
369struct discard_cmd_control {
370 struct task_struct *f2fs_issue_discard; /* discard thread */
371 struct list_head entry_list; /* 4KB discard entry list */
372 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
373 struct list_head wait_list; /* store on-flushing entries */
374 struct list_head fstrim_list; /* in-flight discard from fstrim */
375 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
376 unsigned int discard_wake; /* to wake up discard thread */
377 struct mutex cmd_lock;
378 unsigned int nr_discards; /* # of discards in the list */
379 unsigned int max_discards; /* max. discards to be issued */
380 unsigned int discard_granularity; /* discard granularity */
381 unsigned int undiscard_blks; /* # of undiscard blocks */
382 unsigned int next_pos; /* next discard position */
383 atomic_t issued_discard; /* # of issued discard */
384 atomic_t queued_discard; /* # of queued discard */
385 atomic_t discard_cmd_cnt; /* # of cached cmd count */
386 struct rb_root_cached root; /* root of discard rb-tree */
387 bool rbtree_check; /* config for consistence check */
388};
389
390/* for the list of fsync inodes, used only during recovery */
391struct fsync_inode_entry {
392 struct list_head list; /* list head */
393 struct inode *inode; /* vfs inode pointer */
394 block_t blkaddr; /* block address locating the last fsync */
395 block_t last_dentry; /* block address locating the last dentry */
396};
397
398#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
399#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
400
401#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
402#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
403#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
404#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
405
406#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
407#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
408
409static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
410{
411 int before = nats_in_cursum(journal);
412
413 journal->n_nats = cpu_to_le16(before + i);
414 return before;
415}
416
417static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
418{
419 int before = sits_in_cursum(journal);
420
421 journal->n_sits = cpu_to_le16(before + i);
422 return before;
423}
424
425static inline bool __has_cursum_space(struct f2fs_journal *journal,
426 int size, int type)
427{
428 if (type == NAT_JOURNAL)
429 return size <= MAX_NAT_JENTRIES(journal);
430 return size <= MAX_SIT_JENTRIES(journal);
431}
432
433/* for inline stuff */
434#define DEF_INLINE_RESERVED_SIZE 1
435static inline int get_extra_isize(struct inode *inode);
436static inline int get_inline_xattr_addrs(struct inode *inode);
437#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
438 (CUR_ADDRS_PER_INODE(inode) - \
439 get_inline_xattr_addrs(inode) - \
440 DEF_INLINE_RESERVED_SIZE))
441
442/* for inline dir */
443#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
444 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
445 BITS_PER_BYTE + 1))
446#define INLINE_DENTRY_BITMAP_SIZE(inode) \
447 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
448#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
449 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
450 NR_INLINE_DENTRY(inode) + \
451 INLINE_DENTRY_BITMAP_SIZE(inode)))
452
453/*
454 * For INODE and NODE manager
455 */
456/* for directory operations */
457
458struct f2fs_filename {
459 /*
460 * The filename the user specified. This is NULL for some
461 * filesystem-internal operations, e.g. converting an inline directory
462 * to a non-inline one, or roll-forward recovering an encrypted dentry.
463 */
464 const struct qstr *usr_fname;
465
466 /*
467 * The on-disk filename. For encrypted directories, this is encrypted.
468 * This may be NULL for lookups in an encrypted dir without the key.
469 */
470 struct fscrypt_str disk_name;
471
472 /* The dirhash of this filename */
473 f2fs_hash_t hash;
474
475#ifdef CONFIG_FS_ENCRYPTION
476 /*
477 * For lookups in encrypted directories: either the buffer backing
478 * disk_name, or a buffer that holds the decoded no-key name.
479 */
480 struct fscrypt_str crypto_buf;
481#endif
482#ifdef CONFIG_UNICODE
483 /*
484 * For casefolded directories: the casefolded name, but it's left NULL
485 * if the original name is not valid Unicode, if the directory is both
486 * casefolded and encrypted and its encryption key is unavailable, or if
487 * the filesystem is doing an internal operation where usr_fname is also
488 * NULL. In all these cases we fall back to treating the name as an
489 * opaque byte sequence.
490 */
491 struct fscrypt_str cf_name;
492#endif
493};
494
495struct f2fs_dentry_ptr {
496 struct inode *inode;
497 void *bitmap;
498 struct f2fs_dir_entry *dentry;
499 __u8 (*filename)[F2FS_SLOT_LEN];
500 int max;
501 int nr_bitmap;
502};
503
504static inline void make_dentry_ptr_block(struct inode *inode,
505 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
506{
507 d->inode = inode;
508 d->max = NR_DENTRY_IN_BLOCK;
509 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
510 d->bitmap = t->dentry_bitmap;
511 d->dentry = t->dentry;
512 d->filename = t->filename;
513}
514
515static inline void make_dentry_ptr_inline(struct inode *inode,
516 struct f2fs_dentry_ptr *d, void *t)
517{
518 int entry_cnt = NR_INLINE_DENTRY(inode);
519 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
520 int reserved_size = INLINE_RESERVED_SIZE(inode);
521
522 d->inode = inode;
523 d->max = entry_cnt;
524 d->nr_bitmap = bitmap_size;
525 d->bitmap = t;
526 d->dentry = t + bitmap_size + reserved_size;
527 d->filename = t + bitmap_size + reserved_size +
528 SIZE_OF_DIR_ENTRY * entry_cnt;
529}
530
531/*
532 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
533 * as its node offset to distinguish from index node blocks.
534 * But some bits are used to mark the node block.
535 */
536#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
537 >> OFFSET_BIT_SHIFT)
538enum {
539 ALLOC_NODE, /* allocate a new node page if needed */
540 LOOKUP_NODE, /* look up a node without readahead */
541 LOOKUP_NODE_RA, /*
542 * look up a node with readahead called
543 * by get_data_block.
544 */
545};
546
547#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
548
549/* congestion wait timeout value, default: 20ms */
550#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
551
552/* maximum retry quota flush count */
553#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
554
555#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
556
557#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
558
559/* for in-memory extent cache entry */
560#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
561
562/* number of extent info in extent cache we try to shrink */
563#define EXTENT_CACHE_SHRINK_NUMBER 128
564
565struct rb_entry {
566 struct rb_node rb_node; /* rb node located in rb-tree */
567 union {
568 struct {
569 unsigned int ofs; /* start offset of the entry */
570 unsigned int len; /* length of the entry */
571 };
572 unsigned long long key; /* 64-bits key */
573 } __packed;
574};
575
576struct extent_info {
577 unsigned int fofs; /* start offset in a file */
578 unsigned int len; /* length of the extent */
579 u32 blk; /* start block address of the extent */
580};
581
582struct extent_node {
583 struct rb_node rb_node; /* rb node located in rb-tree */
584 struct extent_info ei; /* extent info */
585 struct list_head list; /* node in global extent list of sbi */
586 struct extent_tree *et; /* extent tree pointer */
587};
588
589struct extent_tree {
590 nid_t ino; /* inode number */
591 struct rb_root_cached root; /* root of extent info rb-tree */
592 struct extent_node *cached_en; /* recently accessed extent node */
593 struct extent_info largest; /* largested extent info */
594 struct list_head list; /* to be used by sbi->zombie_list */
595 rwlock_t lock; /* protect extent info rb-tree */
596 atomic_t node_cnt; /* # of extent node in rb-tree*/
597 bool largest_updated; /* largest extent updated */
598};
599
600/*
601 * This structure is taken from ext4_map_blocks.
602 *
603 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
604 */
605#define F2FS_MAP_NEW (1 << BH_New)
606#define F2FS_MAP_MAPPED (1 << BH_Mapped)
607#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
608#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
609 F2FS_MAP_UNWRITTEN)
610
611struct f2fs_map_blocks {
612 block_t m_pblk;
613 block_t m_lblk;
614 unsigned int m_len;
615 unsigned int m_flags;
616 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
617 pgoff_t *m_next_extent; /* point to next possible extent */
618 int m_seg_type;
619 bool m_may_create; /* indicate it is from write path */
620};
621
622/* for flag in get_data_block */
623enum {
624 F2FS_GET_BLOCK_DEFAULT,
625 F2FS_GET_BLOCK_FIEMAP,
626 F2FS_GET_BLOCK_BMAP,
627 F2FS_GET_BLOCK_DIO,
628 F2FS_GET_BLOCK_PRE_DIO,
629 F2FS_GET_BLOCK_PRE_AIO,
630 F2FS_GET_BLOCK_PRECACHE,
631};
632
633/*
634 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
635 */
636#define FADVISE_COLD_BIT 0x01
637#define FADVISE_LOST_PINO_BIT 0x02
638#define FADVISE_ENCRYPT_BIT 0x04
639#define FADVISE_ENC_NAME_BIT 0x08
640#define FADVISE_KEEP_SIZE_BIT 0x10
641#define FADVISE_HOT_BIT 0x20
642#define FADVISE_VERITY_BIT 0x40
643
644#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
645
646#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
647#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
648#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
649
650#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
651#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
652#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
653
654#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
655#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
656
657#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
658#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
659
660#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
661#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
662
663#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
664#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
665#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
666
667#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
668#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
669
670#define DEF_DIR_LEVEL 0
671
672enum {
673 GC_FAILURE_PIN,
674 GC_FAILURE_ATOMIC,
675 MAX_GC_FAILURE
676};
677
678/* used for f2fs_inode_info->flags */
679enum {
680 FI_NEW_INODE, /* indicate newly allocated inode */
681 FI_DIRTY_INODE, /* indicate inode is dirty or not */
682 FI_AUTO_RECOVER, /* indicate inode is recoverable */
683 FI_DIRTY_DIR, /* indicate directory has dirty pages */
684 FI_INC_LINK, /* need to increment i_nlink */
685 FI_ACL_MODE, /* indicate acl mode */
686 FI_NO_ALLOC, /* should not allocate any blocks */
687 FI_FREE_NID, /* free allocated nide */
688 FI_NO_EXTENT, /* not to use the extent cache */
689 FI_INLINE_XATTR, /* used for inline xattr */
690 FI_INLINE_DATA, /* used for inline data*/
691 FI_INLINE_DENTRY, /* used for inline dentry */
692 FI_APPEND_WRITE, /* inode has appended data */
693 FI_UPDATE_WRITE, /* inode has in-place-update data */
694 FI_NEED_IPU, /* used for ipu per file */
695 FI_ATOMIC_FILE, /* indicate atomic file */
696 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
697 FI_VOLATILE_FILE, /* indicate volatile file */
698 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
699 FI_DROP_CACHE, /* drop dirty page cache */
700 FI_DATA_EXIST, /* indicate data exists */
701 FI_INLINE_DOTS, /* indicate inline dot dentries */
702 FI_DO_DEFRAG, /* indicate defragment is running */
703 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
704 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
705 FI_HOT_DATA, /* indicate file is hot */
706 FI_EXTRA_ATTR, /* indicate file has extra attribute */
707 FI_PROJ_INHERIT, /* indicate file inherits projectid */
708 FI_PIN_FILE, /* indicate file should not be gced */
709 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
710 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
711 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
712 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
713 FI_MMAP_FILE, /* indicate file was mmapped */
714 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
715 FI_COMPRESS_RELEASED, /* compressed blocks were released */
716 FI_ALIGNED_WRITE, /* enable aligned write */
717 FI_MAX, /* max flag, never be used */
718};
719
720struct f2fs_inode_info {
721 struct inode vfs_inode; /* serve a vfs inode */
722 unsigned long i_flags; /* keep an inode flags for ioctl */
723 unsigned char i_advise; /* use to give file attribute hints */
724 unsigned char i_dir_level; /* use for dentry level for large dir */
725 unsigned int i_current_depth; /* only for directory depth */
726 /* for gc failure statistic */
727 unsigned int i_gc_failures[MAX_GC_FAILURE];
728 unsigned int i_pino; /* parent inode number */
729 umode_t i_acl_mode; /* keep file acl mode temporarily */
730
731 /* Use below internally in f2fs*/
732 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
733 struct rw_semaphore i_sem; /* protect fi info */
734 atomic_t dirty_pages; /* # of dirty pages */
735 f2fs_hash_t chash; /* hash value of given file name */
736 unsigned int clevel; /* maximum level of given file name */
737 struct task_struct *task; /* lookup and create consistency */
738 struct task_struct *cp_task; /* separate cp/wb IO stats*/
739 nid_t i_xattr_nid; /* node id that contains xattrs */
740 loff_t last_disk_size; /* lastly written file size */
741 spinlock_t i_size_lock; /* protect last_disk_size */
742
743#ifdef CONFIG_QUOTA
744 struct dquot *i_dquot[MAXQUOTAS];
745
746 /* quota space reservation, managed internally by quota code */
747 qsize_t i_reserved_quota;
748#endif
749 struct list_head dirty_list; /* dirty list for dirs and files */
750 struct list_head gdirty_list; /* linked in global dirty list */
751 struct list_head inmem_ilist; /* list for inmem inodes */
752 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
753 struct task_struct *inmem_task; /* store inmemory task */
754 struct mutex inmem_lock; /* lock for inmemory pages */
755 pgoff_t ra_offset; /* ongoing readahead offset */
756 struct extent_tree *extent_tree; /* cached extent_tree entry */
757
758 /* avoid racing between foreground op and gc */
759 struct rw_semaphore i_gc_rwsem[2];
760 struct rw_semaphore i_mmap_sem;
761 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
762
763 int i_extra_isize; /* size of extra space located in i_addr */
764 kprojid_t i_projid; /* id for project quota */
765 int i_inline_xattr_size; /* inline xattr size */
766 struct timespec64 i_crtime; /* inode creation time */
767 struct timespec64 i_disk_time[4];/* inode disk times */
768
769 /* for file compress */
770 atomic_t i_compr_blocks; /* # of compressed blocks */
771 unsigned char i_compress_algorithm; /* algorithm type */
772 unsigned char i_log_cluster_size; /* log of cluster size */
773 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
774 unsigned short i_compress_flag; /* compress flag */
775 unsigned int i_cluster_size; /* cluster size */
776};
777
778static inline void get_extent_info(struct extent_info *ext,
779 struct f2fs_extent *i_ext)
780{
781 ext->fofs = le32_to_cpu(i_ext->fofs);
782 ext->blk = le32_to_cpu(i_ext->blk);
783 ext->len = le32_to_cpu(i_ext->len);
784}
785
786static inline void set_raw_extent(struct extent_info *ext,
787 struct f2fs_extent *i_ext)
788{
789 i_ext->fofs = cpu_to_le32(ext->fofs);
790 i_ext->blk = cpu_to_le32(ext->blk);
791 i_ext->len = cpu_to_le32(ext->len);
792}
793
794static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
795 u32 blk, unsigned int len)
796{
797 ei->fofs = fofs;
798 ei->blk = blk;
799 ei->len = len;
800}
801
802static inline bool __is_discard_mergeable(struct discard_info *back,
803 struct discard_info *front, unsigned int max_len)
804{
805 return (back->lstart + back->len == front->lstart) &&
806 (back->len + front->len <= max_len);
807}
808
809static inline bool __is_discard_back_mergeable(struct discard_info *cur,
810 struct discard_info *back, unsigned int max_len)
811{
812 return __is_discard_mergeable(back, cur, max_len);
813}
814
815static inline bool __is_discard_front_mergeable(struct discard_info *cur,
816 struct discard_info *front, unsigned int max_len)
817{
818 return __is_discard_mergeable(cur, front, max_len);
819}
820
821static inline bool __is_extent_mergeable(struct extent_info *back,
822 struct extent_info *front)
823{
824 return (back->fofs + back->len == front->fofs &&
825 back->blk + back->len == front->blk);
826}
827
828static inline bool __is_back_mergeable(struct extent_info *cur,
829 struct extent_info *back)
830{
831 return __is_extent_mergeable(back, cur);
832}
833
834static inline bool __is_front_mergeable(struct extent_info *cur,
835 struct extent_info *front)
836{
837 return __is_extent_mergeable(cur, front);
838}
839
840extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
841static inline void __try_update_largest_extent(struct extent_tree *et,
842 struct extent_node *en)
843{
844 if (en->ei.len > et->largest.len) {
845 et->largest = en->ei;
846 et->largest_updated = true;
847 }
848}
849
850/*
851 * For free nid management
852 */
853enum nid_state {
854 FREE_NID, /* newly added to free nid list */
855 PREALLOC_NID, /* it is preallocated */
856 MAX_NID_STATE,
857};
858
859enum nat_state {
860 TOTAL_NAT,
861 DIRTY_NAT,
862 RECLAIMABLE_NAT,
863 MAX_NAT_STATE,
864};
865
866struct f2fs_nm_info {
867 block_t nat_blkaddr; /* base disk address of NAT */
868 nid_t max_nid; /* maximum possible node ids */
869 nid_t available_nids; /* # of available node ids */
870 nid_t next_scan_nid; /* the next nid to be scanned */
871 unsigned int ram_thresh; /* control the memory footprint */
872 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
873 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
874
875 /* NAT cache management */
876 struct radix_tree_root nat_root;/* root of the nat entry cache */
877 struct radix_tree_root nat_set_root;/* root of the nat set cache */
878 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */
879 struct list_head nat_entries; /* cached nat entry list (clean) */
880 spinlock_t nat_list_lock; /* protect clean nat entry list */
881 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
882 unsigned int nat_blocks; /* # of nat blocks */
883
884 /* free node ids management */
885 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
886 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
887 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
888 spinlock_t nid_list_lock; /* protect nid lists ops */
889 struct mutex build_lock; /* lock for build free nids */
890 unsigned char **free_nid_bitmap;
891 unsigned char *nat_block_bitmap;
892 unsigned short *free_nid_count; /* free nid count of NAT block */
893
894 /* for checkpoint */
895 char *nat_bitmap; /* NAT bitmap pointer */
896
897 unsigned int nat_bits_blocks; /* # of nat bits blocks */
898 unsigned char *nat_bits; /* NAT bits blocks */
899 unsigned char *full_nat_bits; /* full NAT pages */
900 unsigned char *empty_nat_bits; /* empty NAT pages */
901#ifdef CONFIG_F2FS_CHECK_FS
902 char *nat_bitmap_mir; /* NAT bitmap mirror */
903#endif
904 int bitmap_size; /* bitmap size */
905};
906
907/*
908 * this structure is used as one of function parameters.
909 * all the information are dedicated to a given direct node block determined
910 * by the data offset in a file.
911 */
912struct dnode_of_data {
913 struct inode *inode; /* vfs inode pointer */
914 struct page *inode_page; /* its inode page, NULL is possible */
915 struct page *node_page; /* cached direct node page */
916 nid_t nid; /* node id of the direct node block */
917 unsigned int ofs_in_node; /* data offset in the node page */
918 bool inode_page_locked; /* inode page is locked or not */
919 bool node_changed; /* is node block changed */
920 char cur_level; /* level of hole node page */
921 char max_level; /* level of current page located */
922 block_t data_blkaddr; /* block address of the node block */
923};
924
925static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
926 struct page *ipage, struct page *npage, nid_t nid)
927{
928 memset(dn, 0, sizeof(*dn));
929 dn->inode = inode;
930 dn->inode_page = ipage;
931 dn->node_page = npage;
932 dn->nid = nid;
933}
934
935/*
936 * For SIT manager
937 *
938 * By default, there are 6 active log areas across the whole main area.
939 * When considering hot and cold data separation to reduce cleaning overhead,
940 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
941 * respectively.
942 * In the current design, you should not change the numbers intentionally.
943 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
944 * logs individually according to the underlying devices. (default: 6)
945 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
946 * data and 8 for node logs.
947 */
948#define NR_CURSEG_DATA_TYPE (3)
949#define NR_CURSEG_NODE_TYPE (3)
950#define NR_CURSEG_INMEM_TYPE (2)
951#define NR_CURSEG_RO_TYPE (2)
952#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
953#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
954
955enum {
956 CURSEG_HOT_DATA = 0, /* directory entry blocks */
957 CURSEG_WARM_DATA, /* data blocks */
958 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
959 CURSEG_HOT_NODE, /* direct node blocks of directory files */
960 CURSEG_WARM_NODE, /* direct node blocks of normal files */
961 CURSEG_COLD_NODE, /* indirect node blocks */
962 NR_PERSISTENT_LOG, /* number of persistent log */
963 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
964 /* pinned file that needs consecutive block address */
965 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
966 NO_CHECK_TYPE, /* number of persistent & inmem log */
967};
968
969struct flush_cmd {
970 struct completion wait;
971 struct llist_node llnode;
972 nid_t ino;
973 int ret;
974};
975
976struct flush_cmd_control {
977 struct task_struct *f2fs_issue_flush; /* flush thread */
978 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
979 atomic_t issued_flush; /* # of issued flushes */
980 atomic_t queued_flush; /* # of queued flushes */
981 struct llist_head issue_list; /* list for command issue */
982 struct llist_node *dispatch_list; /* list for command dispatch */
983};
984
985struct f2fs_sm_info {
986 struct sit_info *sit_info; /* whole segment information */
987 struct free_segmap_info *free_info; /* free segment information */
988 struct dirty_seglist_info *dirty_info; /* dirty segment information */
989 struct curseg_info *curseg_array; /* active segment information */
990
991 struct rw_semaphore curseg_lock; /* for preventing curseg change */
992
993 block_t seg0_blkaddr; /* block address of 0'th segment */
994 block_t main_blkaddr; /* start block address of main area */
995 block_t ssa_blkaddr; /* start block address of SSA area */
996
997 unsigned int segment_count; /* total # of segments */
998 unsigned int main_segments; /* # of segments in main area */
999 unsigned int reserved_segments; /* # of reserved segments */
1000 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1001 unsigned int ovp_segments; /* # of overprovision segments */
1002
1003 /* a threshold to reclaim prefree segments */
1004 unsigned int rec_prefree_segments;
1005
1006 /* for batched trimming */
1007 unsigned int trim_sections; /* # of sections to trim */
1008
1009 struct list_head sit_entry_set; /* sit entry set list */
1010
1011 unsigned int ipu_policy; /* in-place-update policy */
1012 unsigned int min_ipu_util; /* in-place-update threshold */
1013 unsigned int min_fsync_blocks; /* threshold for fsync */
1014 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1015 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1016 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1017
1018 /* for flush command control */
1019 struct flush_cmd_control *fcc_info;
1020
1021 /* for discard command control */
1022 struct discard_cmd_control *dcc_info;
1023};
1024
1025/*
1026 * For superblock
1027 */
1028/*
1029 * COUNT_TYPE for monitoring
1030 *
1031 * f2fs monitors the number of several block types such as on-writeback,
1032 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1033 */
1034#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1035enum count_type {
1036 F2FS_DIRTY_DENTS,
1037 F2FS_DIRTY_DATA,
1038 F2FS_DIRTY_QDATA,
1039 F2FS_DIRTY_NODES,
1040 F2FS_DIRTY_META,
1041 F2FS_INMEM_PAGES,
1042 F2FS_DIRTY_IMETA,
1043 F2FS_WB_CP_DATA,
1044 F2FS_WB_DATA,
1045 F2FS_RD_DATA,
1046 F2FS_RD_NODE,
1047 F2FS_RD_META,
1048 F2FS_DIO_WRITE,
1049 F2FS_DIO_READ,
1050 NR_COUNT_TYPE,
1051};
1052
1053/*
1054 * The below are the page types of bios used in submit_bio().
1055 * The available types are:
1056 * DATA User data pages. It operates as async mode.
1057 * NODE Node pages. It operates as async mode.
1058 * META FS metadata pages such as SIT, NAT, CP.
1059 * NR_PAGE_TYPE The number of page types.
1060 * META_FLUSH Make sure the previous pages are written
1061 * with waiting the bio's completion
1062 * ... Only can be used with META.
1063 */
1064#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1065enum page_type {
1066 DATA,
1067 NODE,
1068 META,
1069 NR_PAGE_TYPE,
1070 META_FLUSH,
1071 INMEM, /* the below types are used by tracepoints only. */
1072 INMEM_DROP,
1073 INMEM_INVALIDATE,
1074 INMEM_REVOKE,
1075 IPU,
1076 OPU,
1077};
1078
1079enum temp_type {
1080 HOT = 0, /* must be zero for meta bio */
1081 WARM,
1082 COLD,
1083 NR_TEMP_TYPE,
1084};
1085
1086enum need_lock_type {
1087 LOCK_REQ = 0,
1088 LOCK_DONE,
1089 LOCK_RETRY,
1090};
1091
1092enum cp_reason_type {
1093 CP_NO_NEEDED,
1094 CP_NON_REGULAR,
1095 CP_COMPRESSED,
1096 CP_HARDLINK,
1097 CP_SB_NEED_CP,
1098 CP_WRONG_PINO,
1099 CP_NO_SPC_ROLL,
1100 CP_NODE_NEED_CP,
1101 CP_FASTBOOT_MODE,
1102 CP_SPEC_LOG_NUM,
1103 CP_RECOVER_DIR,
1104 CP_XATTR_DIR,
1105};
1106
1107enum iostat_type {
1108 /* WRITE IO */
1109 APP_DIRECT_IO, /* app direct write IOs */
1110 APP_BUFFERED_IO, /* app buffered write IOs */
1111 APP_WRITE_IO, /* app write IOs */
1112 APP_MAPPED_IO, /* app mapped IOs */
1113 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1114 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1115 FS_META_IO, /* meta IOs from kworker/reclaimer */
1116 FS_GC_DATA_IO, /* data IOs from forground gc */
1117 FS_GC_NODE_IO, /* node IOs from forground gc */
1118 FS_CP_DATA_IO, /* data IOs from checkpoint */
1119 FS_CP_NODE_IO, /* node IOs from checkpoint */
1120 FS_CP_META_IO, /* meta IOs from checkpoint */
1121
1122 /* READ IO */
1123 APP_DIRECT_READ_IO, /* app direct read IOs */
1124 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1125 APP_READ_IO, /* app read IOs */
1126 APP_MAPPED_READ_IO, /* app mapped read IOs */
1127 FS_DATA_READ_IO, /* data read IOs */
1128 FS_GDATA_READ_IO, /* data read IOs from background gc */
1129 FS_CDATA_READ_IO, /* compressed data read IOs */
1130 FS_NODE_READ_IO, /* node read IOs */
1131 FS_META_READ_IO, /* meta read IOs */
1132
1133 /* other */
1134 FS_DISCARD, /* discard */
1135 NR_IO_TYPE,
1136};
1137
1138struct f2fs_io_info {
1139 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1140 nid_t ino; /* inode number */
1141 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1142 enum temp_type temp; /* contains HOT/WARM/COLD */
1143 int op; /* contains REQ_OP_ */
1144 int op_flags; /* req_flag_bits */
1145 block_t new_blkaddr; /* new block address to be written */
1146 block_t old_blkaddr; /* old block address before Cow */
1147 struct page *page; /* page to be written */
1148 struct page *encrypted_page; /* encrypted page */
1149 struct page *compressed_page; /* compressed page */
1150 struct list_head list; /* serialize IOs */
1151 bool submitted; /* indicate IO submission */
1152 int need_lock; /* indicate we need to lock cp_rwsem */
1153 bool in_list; /* indicate fio is in io_list */
1154 bool is_por; /* indicate IO is from recovery or not */
1155 bool retry; /* need to reallocate block address */
1156 int compr_blocks; /* # of compressed block addresses */
1157 bool encrypted; /* indicate file is encrypted */
1158 enum iostat_type io_type; /* io type */
1159 struct writeback_control *io_wbc; /* writeback control */
1160 struct bio **bio; /* bio for ipu */
1161 sector_t *last_block; /* last block number in bio */
1162 unsigned char version; /* version of the node */
1163};
1164
1165struct bio_entry {
1166 struct bio *bio;
1167 struct list_head list;
1168};
1169
1170#define is_read_io(rw) ((rw) == READ)
1171struct f2fs_bio_info {
1172 struct f2fs_sb_info *sbi; /* f2fs superblock */
1173 struct bio *bio; /* bios to merge */
1174 sector_t last_block_in_bio; /* last block number */
1175 struct f2fs_io_info fio; /* store buffered io info. */
1176 struct rw_semaphore io_rwsem; /* blocking op for bio */
1177 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1178 struct list_head io_list; /* track fios */
1179 struct list_head bio_list; /* bio entry list head */
1180 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1181};
1182
1183#define FDEV(i) (sbi->devs[i])
1184#define RDEV(i) (raw_super->devs[i])
1185struct f2fs_dev_info {
1186 struct block_device *bdev;
1187 char path[MAX_PATH_LEN];
1188 unsigned int total_segments;
1189 block_t start_blk;
1190 block_t end_blk;
1191#ifdef CONFIG_BLK_DEV_ZONED
1192 unsigned int nr_blkz; /* Total number of zones */
1193 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1194 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */
1195#endif
1196};
1197
1198enum inode_type {
1199 DIR_INODE, /* for dirty dir inode */
1200 FILE_INODE, /* for dirty regular/symlink inode */
1201 DIRTY_META, /* for all dirtied inode metadata */
1202 ATOMIC_FILE, /* for all atomic files */
1203 NR_INODE_TYPE,
1204};
1205
1206/* for inner inode cache management */
1207struct inode_management {
1208 struct radix_tree_root ino_root; /* ino entry array */
1209 spinlock_t ino_lock; /* for ino entry lock */
1210 struct list_head ino_list; /* inode list head */
1211 unsigned long ino_num; /* number of entries */
1212};
1213
1214/* for GC_AT */
1215struct atgc_management {
1216 bool atgc_enabled; /* ATGC is enabled or not */
1217 struct rb_root_cached root; /* root of victim rb-tree */
1218 struct list_head victim_list; /* linked with all victim entries */
1219 unsigned int victim_count; /* victim count in rb-tree */
1220 unsigned int candidate_ratio; /* candidate ratio */
1221 unsigned int max_candidate_count; /* max candidate count */
1222 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1223 unsigned long long age_threshold; /* age threshold */
1224};
1225
1226/* For s_flag in struct f2fs_sb_info */
1227enum {
1228 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1229 SBI_IS_CLOSE, /* specify unmounting */
1230 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1231 SBI_POR_DOING, /* recovery is doing or not */
1232 SBI_NEED_SB_WRITE, /* need to recover superblock */
1233 SBI_NEED_CP, /* need to checkpoint */
1234 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1235 SBI_IS_RECOVERED, /* recovered orphan/data */
1236 SBI_CP_DISABLED, /* CP was disabled last mount */
1237 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1238 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1239 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1240 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1241 SBI_IS_RESIZEFS, /* resizefs is in process */
1242};
1243
1244enum {
1245 CP_TIME,
1246 REQ_TIME,
1247 DISCARD_TIME,
1248 GC_TIME,
1249 DISABLE_TIME,
1250 UMOUNT_DISCARD_TIMEOUT,
1251 MAX_TIME,
1252};
1253
1254enum {
1255 GC_NORMAL,
1256 GC_IDLE_CB,
1257 GC_IDLE_GREEDY,
1258 GC_IDLE_AT,
1259 GC_URGENT_HIGH,
1260 GC_URGENT_LOW,
1261};
1262
1263enum {
1264 BGGC_MODE_ON, /* background gc is on */
1265 BGGC_MODE_OFF, /* background gc is off */
1266 BGGC_MODE_SYNC, /*
1267 * background gc is on, migrating blocks
1268 * like foreground gc
1269 */
1270};
1271
1272enum {
1273 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1274 FS_MODE_LFS, /* use lfs allocation only */
1275};
1276
1277enum {
1278 WHINT_MODE_OFF, /* not pass down write hints */
1279 WHINT_MODE_USER, /* try to pass down hints given by users */
1280 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1281};
1282
1283enum {
1284 ALLOC_MODE_DEFAULT, /* stay default */
1285 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1286};
1287
1288enum fsync_mode {
1289 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1290 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1291 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1292};
1293
1294enum {
1295 COMPR_MODE_FS, /*
1296 * automatically compress compression
1297 * enabled files
1298 */
1299 COMPR_MODE_USER, /*
1300 * automatical compression is disabled.
1301 * user can control the file compression
1302 * using ioctls
1303 */
1304};
1305
1306static inline int f2fs_test_bit(unsigned int nr, char *addr);
1307static inline void f2fs_set_bit(unsigned int nr, char *addr);
1308static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1309
1310/*
1311 * Layout of f2fs page.private:
1312 *
1313 * Layout A: lowest bit should be 1
1314 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1315 * bit 0 PAGE_PRIVATE_NOT_POINTER
1316 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE
1317 * bit 2 PAGE_PRIVATE_DUMMY_WRITE
1318 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION
1319 * bit 4 PAGE_PRIVATE_INLINE_INODE
1320 * bit 5 PAGE_PRIVATE_REF_RESOURCE
1321 * bit 6- f2fs private data
1322 *
1323 * Layout B: lowest bit should be 0
1324 * page.private is a wrapped pointer.
1325 */
1326enum {
1327 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1328 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */
1329 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1330 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1331 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1332 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1333 PAGE_PRIVATE_MAX
1334};
1335
1336#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
1337static inline bool page_private_##name(struct page *page) \
1338{ \
1339 return PagePrivate(page) && \
1340 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
1341 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1342}
1343
1344#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
1345static inline void set_page_private_##name(struct page *page) \
1346{ \
1347 if (!PagePrivate(page)) { \
1348 get_page(page); \
1349 SetPagePrivate(page); \
1350 set_page_private(page, 0); \
1351 } \
1352 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
1353 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1354}
1355
1356#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
1357static inline void clear_page_private_##name(struct page *page) \
1358{ \
1359 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1360 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \
1361 set_page_private(page, 0); \
1362 if (PagePrivate(page)) { \
1363 ClearPagePrivate(page); \
1364 put_page(page); \
1365 }\
1366 } \
1367}
1368
1369PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
1370PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE);
1371PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
1372PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
1373PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);
1374PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
1375
1376PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
1377PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
1378PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
1379PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);
1380PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
1381
1382PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
1383PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
1384PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
1385PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);
1386PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
1387
1388static inline unsigned long get_page_private_data(struct page *page)
1389{
1390 unsigned long data = page_private(page);
1391
1392 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
1393 return 0;
1394 return data >> PAGE_PRIVATE_MAX;
1395}
1396
1397static inline void set_page_private_data(struct page *page, unsigned long data)
1398{
1399 if (!PagePrivate(page)) {
1400 get_page(page);
1401 SetPagePrivate(page);
1402 set_page_private(page, 0);
1403 }
1404 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
1405 page_private(page) |= data << PAGE_PRIVATE_MAX;
1406}
1407
1408static inline void clear_page_private_data(struct page *page)
1409{
1410 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1;
1411 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) {
1412 set_page_private(page, 0);
1413 if (PagePrivate(page)) {
1414 ClearPagePrivate(page);
1415 put_page(page);
1416 }
1417 }
1418}
1419
1420/* For compression */
1421enum compress_algorithm_type {
1422 COMPRESS_LZO,
1423 COMPRESS_LZ4,
1424 COMPRESS_ZSTD,
1425 COMPRESS_LZORLE,
1426 COMPRESS_MAX,
1427};
1428
1429enum compress_flag {
1430 COMPRESS_CHKSUM,
1431 COMPRESS_MAX_FLAG,
1432};
1433
1434#define COMPRESS_WATERMARK 20
1435#define COMPRESS_PERCENT 20
1436
1437#define COMPRESS_DATA_RESERVED_SIZE 4
1438struct compress_data {
1439 __le32 clen; /* compressed data size */
1440 __le32 chksum; /* compressed data chksum */
1441 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1442 u8 cdata[]; /* compressed data */
1443};
1444
1445#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1446
1447#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1448
1449#define COMPRESS_LEVEL_OFFSET 8
1450
1451/* compress context */
1452struct compress_ctx {
1453 struct inode *inode; /* inode the context belong to */
1454 pgoff_t cluster_idx; /* cluster index number */
1455 unsigned int cluster_size; /* page count in cluster */
1456 unsigned int log_cluster_size; /* log of cluster size */
1457 struct page **rpages; /* pages store raw data in cluster */
1458 unsigned int nr_rpages; /* total page number in rpages */
1459 struct page **cpages; /* pages store compressed data in cluster */
1460 unsigned int nr_cpages; /* total page number in cpages */
1461 void *rbuf; /* virtual mapped address on rpages */
1462 struct compress_data *cbuf; /* virtual mapped address on cpages */
1463 size_t rlen; /* valid data length in rbuf */
1464 size_t clen; /* valid data length in cbuf */
1465 void *private; /* payload buffer for specified compression algorithm */
1466 void *private2; /* extra payload buffer */
1467};
1468
1469/* compress context for write IO path */
1470struct compress_io_ctx {
1471 u32 magic; /* magic number to indicate page is compressed */
1472 struct inode *inode; /* inode the context belong to */
1473 struct page **rpages; /* pages store raw data in cluster */
1474 unsigned int nr_rpages; /* total page number in rpages */
1475 atomic_t pending_pages; /* in-flight compressed page count */
1476};
1477
1478/* Context for decompressing one cluster on the read IO path */
1479struct decompress_io_ctx {
1480 u32 magic; /* magic number to indicate page is compressed */
1481 struct inode *inode; /* inode the context belong to */
1482 pgoff_t cluster_idx; /* cluster index number */
1483 unsigned int cluster_size; /* page count in cluster */
1484 unsigned int log_cluster_size; /* log of cluster size */
1485 struct page **rpages; /* pages store raw data in cluster */
1486 unsigned int nr_rpages; /* total page number in rpages */
1487 struct page **cpages; /* pages store compressed data in cluster */
1488 unsigned int nr_cpages; /* total page number in cpages */
1489 struct page **tpages; /* temp pages to pad holes in cluster */
1490 void *rbuf; /* virtual mapped address on rpages */
1491 struct compress_data *cbuf; /* virtual mapped address on cpages */
1492 size_t rlen; /* valid data length in rbuf */
1493 size_t clen; /* valid data length in cbuf */
1494
1495 /*
1496 * The number of compressed pages remaining to be read in this cluster.
1497 * This is initially nr_cpages. It is decremented by 1 each time a page
1498 * has been read (or failed to be read). When it reaches 0, the cluster
1499 * is decompressed (or an error is reported).
1500 *
1501 * If an error occurs before all the pages have been submitted for I/O,
1502 * then this will never reach 0. In this case the I/O submitter is
1503 * responsible for calling f2fs_decompress_end_io() instead.
1504 */
1505 atomic_t remaining_pages;
1506
1507 /*
1508 * Number of references to this decompress_io_ctx.
1509 *
1510 * One reference is held for I/O completion. This reference is dropped
1511 * after the pagecache pages are updated and unlocked -- either after
1512 * decompression (and verity if enabled), or after an error.
1513 *
1514 * In addition, each compressed page holds a reference while it is in a
1515 * bio. These references are necessary prevent compressed pages from
1516 * being freed while they are still in a bio.
1517 */
1518 refcount_t refcnt;
1519
1520 bool failed; /* IO error occurred before decompression? */
1521 bool need_verity; /* need fs-verity verification after decompression? */
1522 void *private; /* payload buffer for specified decompression algorithm */
1523 void *private2; /* extra payload buffer */
1524 struct work_struct verity_work; /* work to verify the decompressed pages */
1525};
1526
1527#define NULL_CLUSTER ((unsigned int)(~0))
1528#define MIN_COMPRESS_LOG_SIZE 2
1529#define MAX_COMPRESS_LOG_SIZE 8
1530#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1531
1532struct f2fs_sb_info {
1533 struct super_block *sb; /* pointer to VFS super block */
1534 struct proc_dir_entry *s_proc; /* proc entry */
1535 struct f2fs_super_block *raw_super; /* raw super block pointer */
1536 struct rw_semaphore sb_lock; /* lock for raw super block */
1537 int valid_super_block; /* valid super block no */
1538 unsigned long s_flag; /* flags for sbi */
1539 struct mutex writepages; /* mutex for writepages() */
1540
1541#ifdef CONFIG_BLK_DEV_ZONED
1542 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1543 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1544#endif
1545
1546 /* for node-related operations */
1547 struct f2fs_nm_info *nm_info; /* node manager */
1548 struct inode *node_inode; /* cache node blocks */
1549
1550 /* for segment-related operations */
1551 struct f2fs_sm_info *sm_info; /* segment manager */
1552
1553 /* for bio operations */
1554 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1555 /* keep migration IO order for LFS mode */
1556 struct rw_semaphore io_order_lock;
1557 mempool_t *write_io_dummy; /* Dummy pages */
1558
1559 /* for checkpoint */
1560 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1561 int cur_cp_pack; /* remain current cp pack */
1562 spinlock_t cp_lock; /* for flag in ckpt */
1563 struct inode *meta_inode; /* cache meta blocks */
1564 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */
1565 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1566 struct rw_semaphore node_write; /* locking node writes */
1567 struct rw_semaphore node_change; /* locking node change */
1568 wait_queue_head_t cp_wait;
1569 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1570 long interval_time[MAX_TIME]; /* to store thresholds */
1571 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1572
1573 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1574
1575 spinlock_t fsync_node_lock; /* for node entry lock */
1576 struct list_head fsync_node_list; /* node list head */
1577 unsigned int fsync_seg_id; /* sequence id */
1578 unsigned int fsync_node_num; /* number of node entries */
1579
1580 /* for orphan inode, use 0'th array */
1581 unsigned int max_orphans; /* max orphan inodes */
1582
1583 /* for inode management */
1584 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1585 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1586 struct mutex flush_lock; /* for flush exclusion */
1587
1588 /* for extent tree cache */
1589 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1590 struct mutex extent_tree_lock; /* locking extent radix tree */
1591 struct list_head extent_list; /* lru list for shrinker */
1592 spinlock_t extent_lock; /* locking extent lru list */
1593 atomic_t total_ext_tree; /* extent tree count */
1594 struct list_head zombie_list; /* extent zombie tree list */
1595 atomic_t total_zombie_tree; /* extent zombie tree count */
1596 atomic_t total_ext_node; /* extent info count */
1597
1598 /* basic filesystem units */
1599 unsigned int log_sectors_per_block; /* log2 sectors per block */
1600 unsigned int log_blocksize; /* log2 block size */
1601 unsigned int blocksize; /* block size */
1602 unsigned int root_ino_num; /* root inode number*/
1603 unsigned int node_ino_num; /* node inode number*/
1604 unsigned int meta_ino_num; /* meta inode number*/
1605 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1606 unsigned int blocks_per_seg; /* blocks per segment */
1607 unsigned int segs_per_sec; /* segments per section */
1608 unsigned int secs_per_zone; /* sections per zone */
1609 unsigned int total_sections; /* total section count */
1610 unsigned int total_node_count; /* total node block count */
1611 unsigned int total_valid_node_count; /* valid node block count */
1612 int dir_level; /* directory level */
1613 int readdir_ra; /* readahead inode in readdir */
1614 u64 max_io_bytes; /* max io bytes to merge IOs */
1615
1616 block_t user_block_count; /* # of user blocks */
1617 block_t total_valid_block_count; /* # of valid blocks */
1618 block_t discard_blks; /* discard command candidats */
1619 block_t last_valid_block_count; /* for recovery */
1620 block_t reserved_blocks; /* configurable reserved blocks */
1621 block_t current_reserved_blocks; /* current reserved blocks */
1622
1623 /* Additional tracking for no checkpoint mode */
1624 block_t unusable_block_count; /* # of blocks saved by last cp */
1625
1626 unsigned int nquota_files; /* # of quota sysfile */
1627 struct rw_semaphore quota_sem; /* blocking cp for flags */
1628
1629 /* # of pages, see count_type */
1630 atomic_t nr_pages[NR_COUNT_TYPE];
1631 /* # of allocated blocks */
1632 struct percpu_counter alloc_valid_block_count;
1633
1634 /* writeback control */
1635 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1636
1637 /* valid inode count */
1638 struct percpu_counter total_valid_inode_count;
1639
1640 struct f2fs_mount_info mount_opt; /* mount options */
1641
1642 /* for cleaning operations */
1643 struct rw_semaphore gc_lock; /*
1644 * semaphore for GC, avoid
1645 * race between GC and GC or CP
1646 */
1647 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1648 struct atgc_management am; /* atgc management */
1649 unsigned int cur_victim_sec; /* current victim section num */
1650 unsigned int gc_mode; /* current GC state */
1651 unsigned int next_victim_seg[2]; /* next segment in victim section */
1652
1653 /* for skip statistic */
1654 unsigned int atomic_files; /* # of opened atomic file */
1655 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1656 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1657
1658 /* threshold for gc trials on pinned files */
1659 u64 gc_pin_file_threshold;
1660 struct rw_semaphore pin_sem;
1661
1662 /* maximum # of trials to find a victim segment for SSR and GC */
1663 unsigned int max_victim_search;
1664 /* migration granularity of garbage collection, unit: segment */
1665 unsigned int migration_granularity;
1666
1667 /*
1668 * for stat information.
1669 * one is for the LFS mode, and the other is for the SSR mode.
1670 */
1671#ifdef CONFIG_F2FS_STAT_FS
1672 struct f2fs_stat_info *stat_info; /* FS status information */
1673 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1674 unsigned int segment_count[2]; /* # of allocated segments */
1675 unsigned int block_count[2]; /* # of allocated blocks */
1676 atomic_t inplace_count; /* # of inplace update */
1677 atomic64_t total_hit_ext; /* # of lookup extent cache */
1678 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1679 atomic64_t read_hit_largest; /* # of hit largest extent node */
1680 atomic64_t read_hit_cached; /* # of hit cached extent node */
1681 atomic_t inline_xattr; /* # of inline_xattr inodes */
1682 atomic_t inline_inode; /* # of inline_data inodes */
1683 atomic_t inline_dir; /* # of inline_dentry inodes */
1684 atomic_t compr_inode; /* # of compressed inodes */
1685 atomic64_t compr_blocks; /* # of compressed blocks */
1686 atomic_t vw_cnt; /* # of volatile writes */
1687 atomic_t max_aw_cnt; /* max # of atomic writes */
1688 atomic_t max_vw_cnt; /* max # of volatile writes */
1689 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1690 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1691 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1692#endif
1693 spinlock_t stat_lock; /* lock for stat operations */
1694
1695 /* For app/fs IO statistics */
1696 spinlock_t iostat_lock;
1697 unsigned long long rw_iostat[NR_IO_TYPE];
1698 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1699 bool iostat_enable;
1700 unsigned long iostat_next_period;
1701 unsigned int iostat_period_ms;
1702
1703 /* to attach REQ_META|REQ_FUA flags */
1704 unsigned int data_io_flag;
1705 unsigned int node_io_flag;
1706
1707 /* For sysfs suppport */
1708 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1709 struct completion s_kobj_unregister;
1710
1711 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1712 struct completion s_stat_kobj_unregister;
1713
1714 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1715 struct completion s_feature_list_kobj_unregister;
1716
1717 /* For shrinker support */
1718 struct list_head s_list;
1719 int s_ndevs; /* number of devices */
1720 struct f2fs_dev_info *devs; /* for device list */
1721 unsigned int dirty_device; /* for checkpoint data flush */
1722 spinlock_t dev_lock; /* protect dirty_device */
1723 struct mutex umount_mutex;
1724 unsigned int shrinker_run_no;
1725
1726 /* For write statistics */
1727 u64 sectors_written_start;
1728 u64 kbytes_written;
1729
1730 /* Reference to checksum algorithm driver via cryptoapi */
1731 struct crypto_shash *s_chksum_driver;
1732
1733 /* Precomputed FS UUID checksum for seeding other checksums */
1734 __u32 s_chksum_seed;
1735
1736 struct workqueue_struct *post_read_wq; /* post read workqueue */
1737
1738 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1739 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1740
1741#ifdef CONFIG_F2FS_FS_COMPRESSION
1742 struct kmem_cache *page_array_slab; /* page array entry */
1743 unsigned int page_array_slab_size; /* default page array slab size */
1744
1745 /* For runtime compression statistics */
1746 u64 compr_written_block;
1747 u64 compr_saved_block;
1748 u32 compr_new_inode;
1749
1750 /* For compressed block cache */
1751 struct inode *compress_inode; /* cache compressed blocks */
1752 unsigned int compress_percent; /* cache page percentage */
1753 unsigned int compress_watermark; /* cache page watermark */
1754 atomic_t compress_page_hit; /* cache hit count */
1755#endif
1756};
1757
1758struct f2fs_private_dio {
1759 struct inode *inode;
1760 void *orig_private;
1761 bio_end_io_t *orig_end_io;
1762 bool write;
1763};
1764
1765#ifdef CONFIG_F2FS_FAULT_INJECTION
1766#define f2fs_show_injection_info(sbi, type) \
1767 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1768 KERN_INFO, sbi->sb->s_id, \
1769 f2fs_fault_name[type], \
1770 __func__, __builtin_return_address(0))
1771static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1772{
1773 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1774
1775 if (!ffi->inject_rate)
1776 return false;
1777
1778 if (!IS_FAULT_SET(ffi, type))
1779 return false;
1780
1781 atomic_inc(&ffi->inject_ops);
1782 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1783 atomic_set(&ffi->inject_ops, 0);
1784 return true;
1785 }
1786 return false;
1787}
1788#else
1789#define f2fs_show_injection_info(sbi, type) do { } while (0)
1790static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1791{
1792 return false;
1793}
1794#endif
1795
1796/*
1797 * Test if the mounted volume is a multi-device volume.
1798 * - For a single regular disk volume, sbi->s_ndevs is 0.
1799 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1800 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1801 */
1802static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1803{
1804 return sbi->s_ndevs > 1;
1805}
1806
1807static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1808{
1809 unsigned long now = jiffies;
1810
1811 sbi->last_time[type] = now;
1812
1813 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1814 if (type == REQ_TIME) {
1815 sbi->last_time[DISCARD_TIME] = now;
1816 sbi->last_time[GC_TIME] = now;
1817 }
1818}
1819
1820static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1821{
1822 unsigned long interval = sbi->interval_time[type] * HZ;
1823
1824 return time_after(jiffies, sbi->last_time[type] + interval);
1825}
1826
1827static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1828 int type)
1829{
1830 unsigned long interval = sbi->interval_time[type] * HZ;
1831 unsigned int wait_ms = 0;
1832 long delta;
1833
1834 delta = (sbi->last_time[type] + interval) - jiffies;
1835 if (delta > 0)
1836 wait_ms = jiffies_to_msecs(delta);
1837
1838 return wait_ms;
1839}
1840
1841/*
1842 * Inline functions
1843 */
1844static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1845 const void *address, unsigned int length)
1846{
1847 struct {
1848 struct shash_desc shash;
1849 char ctx[4];
1850 } desc;
1851 int err;
1852
1853 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1854
1855 desc.shash.tfm = sbi->s_chksum_driver;
1856 *(u32 *)desc.ctx = crc;
1857
1858 err = crypto_shash_update(&desc.shash, address, length);
1859 BUG_ON(err);
1860
1861 return *(u32 *)desc.ctx;
1862}
1863
1864static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1865 unsigned int length)
1866{
1867 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1868}
1869
1870static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1871 void *buf, size_t buf_size)
1872{
1873 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1874}
1875
1876static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1877 const void *address, unsigned int length)
1878{
1879 return __f2fs_crc32(sbi, crc, address, length);
1880}
1881
1882static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1883{
1884 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1885}
1886
1887static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1888{
1889 return sb->s_fs_info;
1890}
1891
1892static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1893{
1894 return F2FS_SB(inode->i_sb);
1895}
1896
1897static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1898{
1899 return F2FS_I_SB(mapping->host);
1900}
1901
1902static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1903{
1904 return F2FS_M_SB(page_file_mapping(page));
1905}
1906
1907static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1908{
1909 return (struct f2fs_super_block *)(sbi->raw_super);
1910}
1911
1912static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1913{
1914 return (struct f2fs_checkpoint *)(sbi->ckpt);
1915}
1916
1917static inline struct f2fs_node *F2FS_NODE(struct page *page)
1918{
1919 return (struct f2fs_node *)page_address(page);
1920}
1921
1922static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1923{
1924 return &((struct f2fs_node *)page_address(page))->i;
1925}
1926
1927static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1928{
1929 return (struct f2fs_nm_info *)(sbi->nm_info);
1930}
1931
1932static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1933{
1934 return (struct f2fs_sm_info *)(sbi->sm_info);
1935}
1936
1937static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1938{
1939 return (struct sit_info *)(SM_I(sbi)->sit_info);
1940}
1941
1942static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1943{
1944 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1945}
1946
1947static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1948{
1949 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1950}
1951
1952static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1953{
1954 return sbi->meta_inode->i_mapping;
1955}
1956
1957static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1958{
1959 return sbi->node_inode->i_mapping;
1960}
1961
1962static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1963{
1964 return test_bit(type, &sbi->s_flag);
1965}
1966
1967static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1968{
1969 set_bit(type, &sbi->s_flag);
1970}
1971
1972static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1973{
1974 clear_bit(type, &sbi->s_flag);
1975}
1976
1977static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1978{
1979 return le64_to_cpu(cp->checkpoint_ver);
1980}
1981
1982static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1983{
1984 if (type < F2FS_MAX_QUOTAS)
1985 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1986 return 0;
1987}
1988
1989static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1990{
1991 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1992 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1993}
1994
1995static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1996{
1997 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1998
1999 return ckpt_flags & f;
2000}
2001
2002static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2003{
2004 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2005}
2006
2007static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2008{
2009 unsigned int ckpt_flags;
2010
2011 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2012 ckpt_flags |= f;
2013 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2014}
2015
2016static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2017{
2018 unsigned long flags;
2019
2020 spin_lock_irqsave(&sbi->cp_lock, flags);
2021 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2022 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2023}
2024
2025static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2026{
2027 unsigned int ckpt_flags;
2028
2029 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2030 ckpt_flags &= (~f);
2031 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2032}
2033
2034static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2035{
2036 unsigned long flags;
2037
2038 spin_lock_irqsave(&sbi->cp_lock, flags);
2039 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2040 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2041}
2042
2043static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
2044{
2045 unsigned long flags;
2046 unsigned char *nat_bits;
2047
2048 /*
2049 * In order to re-enable nat_bits we need to call fsck.f2fs by
2050 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
2051 * so let's rely on regular fsck or unclean shutdown.
2052 */
2053
2054 if (lock)
2055 spin_lock_irqsave(&sbi->cp_lock, flags);
2056 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
2057 nat_bits = NM_I(sbi)->nat_bits;
2058 NM_I(sbi)->nat_bits = NULL;
2059 if (lock)
2060 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2061
2062 kvfree(nat_bits);
2063}
2064
2065static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
2066 struct cp_control *cpc)
2067{
2068 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
2069
2070 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
2071}
2072
2073static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2074{
2075 down_read(&sbi->cp_rwsem);
2076}
2077
2078static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2079{
2080 return down_read_trylock(&sbi->cp_rwsem);
2081}
2082
2083static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2084{
2085 up_read(&sbi->cp_rwsem);
2086}
2087
2088static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2089{
2090 down_write(&sbi->cp_rwsem);
2091}
2092
2093static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2094{
2095 up_write(&sbi->cp_rwsem);
2096}
2097
2098static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2099{
2100 int reason = CP_SYNC;
2101
2102 if (test_opt(sbi, FASTBOOT))
2103 reason = CP_FASTBOOT;
2104 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2105 reason = CP_UMOUNT;
2106 return reason;
2107}
2108
2109static inline bool __remain_node_summaries(int reason)
2110{
2111 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2112}
2113
2114static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2115{
2116 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2117 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2118}
2119
2120/*
2121 * Check whether the inode has blocks or not
2122 */
2123static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2124{
2125 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2126
2127 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2128}
2129
2130static inline bool f2fs_has_xattr_block(unsigned int ofs)
2131{
2132 return ofs == XATTR_NODE_OFFSET;
2133}
2134
2135static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2136 struct inode *inode, bool cap)
2137{
2138 if (!inode)
2139 return true;
2140 if (!test_opt(sbi, RESERVE_ROOT))
2141 return false;
2142 if (IS_NOQUOTA(inode))
2143 return true;
2144 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2145 return true;
2146 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2147 in_group_p(F2FS_OPTION(sbi).s_resgid))
2148 return true;
2149 if (cap && capable(CAP_SYS_RESOURCE))
2150 return true;
2151 return false;
2152}
2153
2154static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2155static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2156 struct inode *inode, blkcnt_t *count)
2157{
2158 blkcnt_t diff = 0, release = 0;
2159 block_t avail_user_block_count;
2160 int ret;
2161
2162 ret = dquot_reserve_block(inode, *count);
2163 if (ret)
2164 return ret;
2165
2166 if (time_to_inject(sbi, FAULT_BLOCK)) {
2167 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2168 release = *count;
2169 goto release_quota;
2170 }
2171
2172 /*
2173 * let's increase this in prior to actual block count change in order
2174 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2175 */
2176 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2177
2178 spin_lock(&sbi->stat_lock);
2179 sbi->total_valid_block_count += (block_t)(*count);
2180 avail_user_block_count = sbi->user_block_count -
2181 sbi->current_reserved_blocks;
2182
2183 if (!__allow_reserved_blocks(sbi, inode, true))
2184 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2185
2186 if (F2FS_IO_ALIGNED(sbi))
2187 avail_user_block_count -= sbi->blocks_per_seg *
2188 SM_I(sbi)->additional_reserved_segments;
2189
2190 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2191 if (avail_user_block_count > sbi->unusable_block_count)
2192 avail_user_block_count -= sbi->unusable_block_count;
2193 else
2194 avail_user_block_count = 0;
2195 }
2196 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2197 diff = sbi->total_valid_block_count - avail_user_block_count;
2198 if (diff > *count)
2199 diff = *count;
2200 *count -= diff;
2201 release = diff;
2202 sbi->total_valid_block_count -= diff;
2203 if (!*count) {
2204 spin_unlock(&sbi->stat_lock);
2205 goto enospc;
2206 }
2207 }
2208 spin_unlock(&sbi->stat_lock);
2209
2210 if (unlikely(release)) {
2211 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2212 dquot_release_reservation_block(inode, release);
2213 }
2214 f2fs_i_blocks_write(inode, *count, true, true);
2215 return 0;
2216
2217enospc:
2218 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2219release_quota:
2220 dquot_release_reservation_block(inode, release);
2221 return -ENOSPC;
2222}
2223
2224__printf(2, 3)
2225void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2226
2227#define f2fs_err(sbi, fmt, ...) \
2228 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2229#define f2fs_warn(sbi, fmt, ...) \
2230 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2231#define f2fs_notice(sbi, fmt, ...) \
2232 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2233#define f2fs_info(sbi, fmt, ...) \
2234 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2235#define f2fs_debug(sbi, fmt, ...) \
2236 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2237
2238static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2239 struct inode *inode,
2240 block_t count)
2241{
2242 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2243
2244 spin_lock(&sbi->stat_lock);
2245 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2246 sbi->total_valid_block_count -= (block_t)count;
2247 if (sbi->reserved_blocks &&
2248 sbi->current_reserved_blocks < sbi->reserved_blocks)
2249 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2250 sbi->current_reserved_blocks + count);
2251 spin_unlock(&sbi->stat_lock);
2252 if (unlikely(inode->i_blocks < sectors)) {
2253 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2254 inode->i_ino,
2255 (unsigned long long)inode->i_blocks,
2256 (unsigned long long)sectors);
2257 set_sbi_flag(sbi, SBI_NEED_FSCK);
2258 return;
2259 }
2260 f2fs_i_blocks_write(inode, count, false, true);
2261}
2262
2263static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2264{
2265 atomic_inc(&sbi->nr_pages[count_type]);
2266
2267 if (count_type == F2FS_DIRTY_DENTS ||
2268 count_type == F2FS_DIRTY_NODES ||
2269 count_type == F2FS_DIRTY_META ||
2270 count_type == F2FS_DIRTY_QDATA ||
2271 count_type == F2FS_DIRTY_IMETA)
2272 set_sbi_flag(sbi, SBI_IS_DIRTY);
2273}
2274
2275static inline void inode_inc_dirty_pages(struct inode *inode)
2276{
2277 atomic_inc(&F2FS_I(inode)->dirty_pages);
2278 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2279 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2280 if (IS_NOQUOTA(inode))
2281 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2282}
2283
2284static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2285{
2286 atomic_dec(&sbi->nr_pages[count_type]);
2287}
2288
2289static inline void inode_dec_dirty_pages(struct inode *inode)
2290{
2291 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2292 !S_ISLNK(inode->i_mode))
2293 return;
2294
2295 atomic_dec(&F2FS_I(inode)->dirty_pages);
2296 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2297 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2298 if (IS_NOQUOTA(inode))
2299 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2300}
2301
2302static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2303{
2304 return atomic_read(&sbi->nr_pages[count_type]);
2305}
2306
2307static inline int get_dirty_pages(struct inode *inode)
2308{
2309 return atomic_read(&F2FS_I(inode)->dirty_pages);
2310}
2311
2312static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2313{
2314 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2315 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2316 sbi->log_blocks_per_seg;
2317
2318 return segs / sbi->segs_per_sec;
2319}
2320
2321static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2322{
2323 return sbi->total_valid_block_count;
2324}
2325
2326static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2327{
2328 return sbi->discard_blks;
2329}
2330
2331static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2332{
2333 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2334
2335 /* return NAT or SIT bitmap */
2336 if (flag == NAT_BITMAP)
2337 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2338 else if (flag == SIT_BITMAP)
2339 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2340
2341 return 0;
2342}
2343
2344static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2345{
2346 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2347}
2348
2349static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2350{
2351 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2352 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2353 int offset;
2354
2355 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2356 offset = (flag == SIT_BITMAP) ?
2357 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2358 /*
2359 * if large_nat_bitmap feature is enabled, leave checksum
2360 * protection for all nat/sit bitmaps.
2361 */
2362 return tmp_ptr + offset + sizeof(__le32);
2363 }
2364
2365 if (__cp_payload(sbi) > 0) {
2366 if (flag == NAT_BITMAP)
2367 return &ckpt->sit_nat_version_bitmap;
2368 else
2369 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2370 } else {
2371 offset = (flag == NAT_BITMAP) ?
2372 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2373 return tmp_ptr + offset;
2374 }
2375}
2376
2377static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2378{
2379 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2380
2381 if (sbi->cur_cp_pack == 2)
2382 start_addr += sbi->blocks_per_seg;
2383 return start_addr;
2384}
2385
2386static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2387{
2388 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2389
2390 if (sbi->cur_cp_pack == 1)
2391 start_addr += sbi->blocks_per_seg;
2392 return start_addr;
2393}
2394
2395static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2396{
2397 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2398}
2399
2400static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2401{
2402 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2403}
2404
2405static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2406 struct inode *inode, bool is_inode)
2407{
2408 block_t valid_block_count;
2409 unsigned int valid_node_count, user_block_count;
2410 int err;
2411
2412 if (is_inode) {
2413 if (inode) {
2414 err = dquot_alloc_inode(inode);
2415 if (err)
2416 return err;
2417 }
2418 } else {
2419 err = dquot_reserve_block(inode, 1);
2420 if (err)
2421 return err;
2422 }
2423
2424 if (time_to_inject(sbi, FAULT_BLOCK)) {
2425 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2426 goto enospc;
2427 }
2428
2429 spin_lock(&sbi->stat_lock);
2430
2431 valid_block_count = sbi->total_valid_block_count +
2432 sbi->current_reserved_blocks + 1;
2433
2434 if (!__allow_reserved_blocks(sbi, inode, false))
2435 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2436
2437 if (F2FS_IO_ALIGNED(sbi))
2438 valid_block_count += sbi->blocks_per_seg *
2439 SM_I(sbi)->additional_reserved_segments;
2440
2441 user_block_count = sbi->user_block_count;
2442 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2443 user_block_count -= sbi->unusable_block_count;
2444
2445 if (unlikely(valid_block_count > user_block_count)) {
2446 spin_unlock(&sbi->stat_lock);
2447 goto enospc;
2448 }
2449
2450 valid_node_count = sbi->total_valid_node_count + 1;
2451 if (unlikely(valid_node_count > sbi->total_node_count)) {
2452 spin_unlock(&sbi->stat_lock);
2453 goto enospc;
2454 }
2455
2456 sbi->total_valid_node_count++;
2457 sbi->total_valid_block_count++;
2458 spin_unlock(&sbi->stat_lock);
2459
2460 if (inode) {
2461 if (is_inode)
2462 f2fs_mark_inode_dirty_sync(inode, true);
2463 else
2464 f2fs_i_blocks_write(inode, 1, true, true);
2465 }
2466
2467 percpu_counter_inc(&sbi->alloc_valid_block_count);
2468 return 0;
2469
2470enospc:
2471 if (is_inode) {
2472 if (inode)
2473 dquot_free_inode(inode);
2474 } else {
2475 dquot_release_reservation_block(inode, 1);
2476 }
2477 return -ENOSPC;
2478}
2479
2480static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2481 struct inode *inode, bool is_inode)
2482{
2483 spin_lock(&sbi->stat_lock);
2484
2485 if (unlikely(!sbi->total_valid_block_count ||
2486 !sbi->total_valid_node_count)) {
2487 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2488 sbi->total_valid_block_count,
2489 sbi->total_valid_node_count);
2490 set_sbi_flag(sbi, SBI_NEED_FSCK);
2491 } else {
2492 sbi->total_valid_block_count--;
2493 sbi->total_valid_node_count--;
2494 }
2495
2496 if (sbi->reserved_blocks &&
2497 sbi->current_reserved_blocks < sbi->reserved_blocks)
2498 sbi->current_reserved_blocks++;
2499
2500 spin_unlock(&sbi->stat_lock);
2501
2502 if (is_inode) {
2503 dquot_free_inode(inode);
2504 } else {
2505 if (unlikely(inode->i_blocks == 0)) {
2506 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2507 inode->i_ino,
2508 (unsigned long long)inode->i_blocks);
2509 set_sbi_flag(sbi, SBI_NEED_FSCK);
2510 return;
2511 }
2512 f2fs_i_blocks_write(inode, 1, false, true);
2513 }
2514}
2515
2516static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2517{
2518 return sbi->total_valid_node_count;
2519}
2520
2521static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2522{
2523 percpu_counter_inc(&sbi->total_valid_inode_count);
2524}
2525
2526static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2527{
2528 percpu_counter_dec(&sbi->total_valid_inode_count);
2529}
2530
2531static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2532{
2533 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2534}
2535
2536static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2537 pgoff_t index, bool for_write)
2538{
2539 struct page *page;
2540
2541 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2542 if (!for_write)
2543 page = find_get_page_flags(mapping, index,
2544 FGP_LOCK | FGP_ACCESSED);
2545 else
2546 page = find_lock_page(mapping, index);
2547 if (page)
2548 return page;
2549
2550 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2551 f2fs_show_injection_info(F2FS_M_SB(mapping),
2552 FAULT_PAGE_ALLOC);
2553 return NULL;
2554 }
2555 }
2556
2557 if (!for_write)
2558 return grab_cache_page(mapping, index);
2559 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2560}
2561
2562static inline struct page *f2fs_pagecache_get_page(
2563 struct address_space *mapping, pgoff_t index,
2564 int fgp_flags, gfp_t gfp_mask)
2565{
2566 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2567 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2568 return NULL;
2569 }
2570
2571 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2572}
2573
2574static inline void f2fs_copy_page(struct page *src, struct page *dst)
2575{
2576 char *src_kaddr = kmap(src);
2577 char *dst_kaddr = kmap(dst);
2578
2579 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2580 kunmap(dst);
2581 kunmap(src);
2582}
2583
2584static inline void f2fs_put_page(struct page *page, int unlock)
2585{
2586 if (!page)
2587 return;
2588
2589 if (unlock) {
2590 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2591 unlock_page(page);
2592 }
2593 put_page(page);
2594}
2595
2596static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2597{
2598 if (dn->node_page)
2599 f2fs_put_page(dn->node_page, 1);
2600 if (dn->inode_page && dn->node_page != dn->inode_page)
2601 f2fs_put_page(dn->inode_page, 0);
2602 dn->node_page = NULL;
2603 dn->inode_page = NULL;
2604}
2605
2606static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2607 size_t size)
2608{
2609 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2610}
2611
2612static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2613 gfp_t flags)
2614{
2615 void *entry;
2616
2617 entry = kmem_cache_alloc(cachep, flags);
2618 if (!entry)
2619 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2620 return entry;
2621}
2622
2623static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2624{
2625 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2626 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2627 get_pages(sbi, F2FS_WB_CP_DATA) ||
2628 get_pages(sbi, F2FS_DIO_READ) ||
2629 get_pages(sbi, F2FS_DIO_WRITE))
2630 return true;
2631
2632 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2633 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2634 return true;
2635
2636 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2637 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2638 return true;
2639 return false;
2640}
2641
2642static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2643{
2644 if (sbi->gc_mode == GC_URGENT_HIGH)
2645 return true;
2646
2647 if (is_inflight_io(sbi, type))
2648 return false;
2649
2650 if (sbi->gc_mode == GC_URGENT_LOW &&
2651 (type == DISCARD_TIME || type == GC_TIME))
2652 return true;
2653
2654 return f2fs_time_over(sbi, type);
2655}
2656
2657static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2658 unsigned long index, void *item)
2659{
2660 while (radix_tree_insert(root, index, item))
2661 cond_resched();
2662}
2663
2664#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2665
2666static inline bool IS_INODE(struct page *page)
2667{
2668 struct f2fs_node *p = F2FS_NODE(page);
2669
2670 return RAW_IS_INODE(p);
2671}
2672
2673static inline int offset_in_addr(struct f2fs_inode *i)
2674{
2675 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2676 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2677}
2678
2679static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2680{
2681 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2682}
2683
2684static inline int f2fs_has_extra_attr(struct inode *inode);
2685static inline block_t data_blkaddr(struct inode *inode,
2686 struct page *node_page, unsigned int offset)
2687{
2688 struct f2fs_node *raw_node;
2689 __le32 *addr_array;
2690 int base = 0;
2691 bool is_inode = IS_INODE(node_page);
2692
2693 raw_node = F2FS_NODE(node_page);
2694
2695 if (is_inode) {
2696 if (!inode)
2697 /* from GC path only */
2698 base = offset_in_addr(&raw_node->i);
2699 else if (f2fs_has_extra_attr(inode))
2700 base = get_extra_isize(inode);
2701 }
2702
2703 addr_array = blkaddr_in_node(raw_node);
2704 return le32_to_cpu(addr_array[base + offset]);
2705}
2706
2707static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2708{
2709 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2710}
2711
2712static inline int f2fs_test_bit(unsigned int nr, char *addr)
2713{
2714 int mask;
2715
2716 addr += (nr >> 3);
2717 mask = 1 << (7 - (nr & 0x07));
2718 return mask & *addr;
2719}
2720
2721static inline void f2fs_set_bit(unsigned int nr, char *addr)
2722{
2723 int mask;
2724
2725 addr += (nr >> 3);
2726 mask = 1 << (7 - (nr & 0x07));
2727 *addr |= mask;
2728}
2729
2730static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2731{
2732 int mask;
2733
2734 addr += (nr >> 3);
2735 mask = 1 << (7 - (nr & 0x07));
2736 *addr &= ~mask;
2737}
2738
2739static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2740{
2741 int mask;
2742 int ret;
2743
2744 addr += (nr >> 3);
2745 mask = 1 << (7 - (nr & 0x07));
2746 ret = mask & *addr;
2747 *addr |= mask;
2748 return ret;
2749}
2750
2751static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2752{
2753 int mask;
2754 int ret;
2755
2756 addr += (nr >> 3);
2757 mask = 1 << (7 - (nr & 0x07));
2758 ret = mask & *addr;
2759 *addr &= ~mask;
2760 return ret;
2761}
2762
2763static inline void f2fs_change_bit(unsigned int nr, char *addr)
2764{
2765 int mask;
2766
2767 addr += (nr >> 3);
2768 mask = 1 << (7 - (nr & 0x07));
2769 *addr ^= mask;
2770}
2771
2772/*
2773 * On-disk inode flags (f2fs_inode::i_flags)
2774 */
2775#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2776#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2777#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2778#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2779#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2780#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2781#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2782#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2783#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2784#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2785#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2786
2787/* Flags that should be inherited by new inodes from their parent. */
2788#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2789 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2790 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2791
2792/* Flags that are appropriate for regular files (all but dir-specific ones). */
2793#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2794 F2FS_CASEFOLD_FL))
2795
2796/* Flags that are appropriate for non-directories/regular files. */
2797#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2798
2799static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2800{
2801 if (S_ISDIR(mode))
2802 return flags;
2803 else if (S_ISREG(mode))
2804 return flags & F2FS_REG_FLMASK;
2805 else
2806 return flags & F2FS_OTHER_FLMASK;
2807}
2808
2809static inline void __mark_inode_dirty_flag(struct inode *inode,
2810 int flag, bool set)
2811{
2812 switch (flag) {
2813 case FI_INLINE_XATTR:
2814 case FI_INLINE_DATA:
2815 case FI_INLINE_DENTRY:
2816 case FI_NEW_INODE:
2817 if (set)
2818 return;
2819 /* fall through */
2820 case FI_DATA_EXIST:
2821 case FI_INLINE_DOTS:
2822 case FI_PIN_FILE:
2823 case FI_COMPRESS_RELEASED:
2824 f2fs_mark_inode_dirty_sync(inode, true);
2825 }
2826}
2827
2828static inline void set_inode_flag(struct inode *inode, int flag)
2829{
2830 set_bit(flag, F2FS_I(inode)->flags);
2831 __mark_inode_dirty_flag(inode, flag, true);
2832}
2833
2834static inline int is_inode_flag_set(struct inode *inode, int flag)
2835{
2836 return test_bit(flag, F2FS_I(inode)->flags);
2837}
2838
2839static inline void clear_inode_flag(struct inode *inode, int flag)
2840{
2841 clear_bit(flag, F2FS_I(inode)->flags);
2842 __mark_inode_dirty_flag(inode, flag, false);
2843}
2844
2845static inline bool f2fs_verity_in_progress(struct inode *inode)
2846{
2847 return IS_ENABLED(CONFIG_FS_VERITY) &&
2848 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2849}
2850
2851static inline void set_acl_inode(struct inode *inode, umode_t mode)
2852{
2853 F2FS_I(inode)->i_acl_mode = mode;
2854 set_inode_flag(inode, FI_ACL_MODE);
2855 f2fs_mark_inode_dirty_sync(inode, false);
2856}
2857
2858static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2859{
2860 if (inc)
2861 inc_nlink(inode);
2862 else
2863 drop_nlink(inode);
2864 f2fs_mark_inode_dirty_sync(inode, true);
2865}
2866
2867static inline void f2fs_i_blocks_write(struct inode *inode,
2868 block_t diff, bool add, bool claim)
2869{
2870 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2871 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2872
2873 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2874 if (add) {
2875 if (claim)
2876 dquot_claim_block(inode, diff);
2877 else
2878 dquot_alloc_block_nofail(inode, diff);
2879 } else {
2880 dquot_free_block(inode, diff);
2881 }
2882
2883 f2fs_mark_inode_dirty_sync(inode, true);
2884 if (clean || recover)
2885 set_inode_flag(inode, FI_AUTO_RECOVER);
2886}
2887
2888static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2889{
2890 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2891 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2892
2893 if (i_size_read(inode) == i_size)
2894 return;
2895
2896 i_size_write(inode, i_size);
2897 f2fs_mark_inode_dirty_sync(inode, true);
2898 if (clean || recover)
2899 set_inode_flag(inode, FI_AUTO_RECOVER);
2900}
2901
2902static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2903{
2904 F2FS_I(inode)->i_current_depth = depth;
2905 f2fs_mark_inode_dirty_sync(inode, true);
2906}
2907
2908static inline void f2fs_i_gc_failures_write(struct inode *inode,
2909 unsigned int count)
2910{
2911 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2912 f2fs_mark_inode_dirty_sync(inode, true);
2913}
2914
2915static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2916{
2917 F2FS_I(inode)->i_xattr_nid = xnid;
2918 f2fs_mark_inode_dirty_sync(inode, true);
2919}
2920
2921static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2922{
2923 F2FS_I(inode)->i_pino = pino;
2924 f2fs_mark_inode_dirty_sync(inode, true);
2925}
2926
2927static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2928{
2929 struct f2fs_inode_info *fi = F2FS_I(inode);
2930
2931 if (ri->i_inline & F2FS_INLINE_XATTR)
2932 set_bit(FI_INLINE_XATTR, fi->flags);
2933 if (ri->i_inline & F2FS_INLINE_DATA)
2934 set_bit(FI_INLINE_DATA, fi->flags);
2935 if (ri->i_inline & F2FS_INLINE_DENTRY)
2936 set_bit(FI_INLINE_DENTRY, fi->flags);
2937 if (ri->i_inline & F2FS_DATA_EXIST)
2938 set_bit(FI_DATA_EXIST, fi->flags);
2939 if (ri->i_inline & F2FS_INLINE_DOTS)
2940 set_bit(FI_INLINE_DOTS, fi->flags);
2941 if (ri->i_inline & F2FS_EXTRA_ATTR)
2942 set_bit(FI_EXTRA_ATTR, fi->flags);
2943 if (ri->i_inline & F2FS_PIN_FILE)
2944 set_bit(FI_PIN_FILE, fi->flags);
2945 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
2946 set_bit(FI_COMPRESS_RELEASED, fi->flags);
2947}
2948
2949static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2950{
2951 ri->i_inline = 0;
2952
2953 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2954 ri->i_inline |= F2FS_INLINE_XATTR;
2955 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2956 ri->i_inline |= F2FS_INLINE_DATA;
2957 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2958 ri->i_inline |= F2FS_INLINE_DENTRY;
2959 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2960 ri->i_inline |= F2FS_DATA_EXIST;
2961 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2962 ri->i_inline |= F2FS_INLINE_DOTS;
2963 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2964 ri->i_inline |= F2FS_EXTRA_ATTR;
2965 if (is_inode_flag_set(inode, FI_PIN_FILE))
2966 ri->i_inline |= F2FS_PIN_FILE;
2967 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
2968 ri->i_inline |= F2FS_COMPRESS_RELEASED;
2969}
2970
2971static inline int f2fs_has_extra_attr(struct inode *inode)
2972{
2973 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2974}
2975
2976static inline int f2fs_has_inline_xattr(struct inode *inode)
2977{
2978 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2979}
2980
2981static inline int f2fs_compressed_file(struct inode *inode)
2982{
2983 return S_ISREG(inode->i_mode) &&
2984 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2985}
2986
2987static inline bool f2fs_need_compress_data(struct inode *inode)
2988{
2989 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
2990
2991 if (!f2fs_compressed_file(inode))
2992 return false;
2993
2994 if (compress_mode == COMPR_MODE_FS)
2995 return true;
2996 else if (compress_mode == COMPR_MODE_USER &&
2997 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
2998 return true;
2999
3000 return false;
3001}
3002
3003static inline unsigned int addrs_per_inode(struct inode *inode)
3004{
3005 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3006 get_inline_xattr_addrs(inode);
3007
3008 if (!f2fs_compressed_file(inode))
3009 return addrs;
3010 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3011}
3012
3013static inline unsigned int addrs_per_block(struct inode *inode)
3014{
3015 if (!f2fs_compressed_file(inode))
3016 return DEF_ADDRS_PER_BLOCK;
3017 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3018}
3019
3020static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3021{
3022 struct f2fs_inode *ri = F2FS_INODE(page);
3023
3024 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3025 get_inline_xattr_addrs(inode)]);
3026}
3027
3028static inline int inline_xattr_size(struct inode *inode)
3029{
3030 if (f2fs_has_inline_xattr(inode))
3031 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3032 return 0;
3033}
3034
3035static inline int f2fs_has_inline_data(struct inode *inode)
3036{
3037 return is_inode_flag_set(inode, FI_INLINE_DATA);
3038}
3039
3040static inline int f2fs_exist_data(struct inode *inode)
3041{
3042 return is_inode_flag_set(inode, FI_DATA_EXIST);
3043}
3044
3045static inline int f2fs_has_inline_dots(struct inode *inode)
3046{
3047 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3048}
3049
3050static inline int f2fs_is_mmap_file(struct inode *inode)
3051{
3052 return is_inode_flag_set(inode, FI_MMAP_FILE);
3053}
3054
3055static inline bool f2fs_is_pinned_file(struct inode *inode)
3056{
3057 return is_inode_flag_set(inode, FI_PIN_FILE);
3058}
3059
3060static inline bool f2fs_is_atomic_file(struct inode *inode)
3061{
3062 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3063}
3064
3065static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
3066{
3067 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
3068}
3069
3070static inline bool f2fs_is_volatile_file(struct inode *inode)
3071{
3072 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
3073}
3074
3075static inline bool f2fs_is_first_block_written(struct inode *inode)
3076{
3077 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
3078}
3079
3080static inline bool f2fs_is_drop_cache(struct inode *inode)
3081{
3082 return is_inode_flag_set(inode, FI_DROP_CACHE);
3083}
3084
3085static inline void *inline_data_addr(struct inode *inode, struct page *page)
3086{
3087 struct f2fs_inode *ri = F2FS_INODE(page);
3088 int extra_size = get_extra_isize(inode);
3089
3090 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
3091}
3092
3093static inline int f2fs_has_inline_dentry(struct inode *inode)
3094{
3095 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3096}
3097
3098static inline int is_file(struct inode *inode, int type)
3099{
3100 return F2FS_I(inode)->i_advise & type;
3101}
3102
3103static inline void set_file(struct inode *inode, int type)
3104{
3105 F2FS_I(inode)->i_advise |= type;
3106 f2fs_mark_inode_dirty_sync(inode, true);
3107}
3108
3109static inline void clear_file(struct inode *inode, int type)
3110{
3111 F2FS_I(inode)->i_advise &= ~type;
3112 f2fs_mark_inode_dirty_sync(inode, true);
3113}
3114
3115static inline bool f2fs_is_time_consistent(struct inode *inode)
3116{
3117 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
3118 return false;
3119 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
3120 return false;
3121 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
3122 return false;
3123 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
3124 &F2FS_I(inode)->i_crtime))
3125 return false;
3126 return true;
3127}
3128
3129static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3130{
3131 bool ret;
3132
3133 if (dsync) {
3134 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3135
3136 spin_lock(&sbi->inode_lock[DIRTY_META]);
3137 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3138 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3139 return ret;
3140 }
3141 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3142 file_keep_isize(inode) ||
3143 i_size_read(inode) & ~PAGE_MASK)
3144 return false;
3145
3146 if (!f2fs_is_time_consistent(inode))
3147 return false;
3148
3149 spin_lock(&F2FS_I(inode)->i_size_lock);
3150 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3151 spin_unlock(&F2FS_I(inode)->i_size_lock);
3152
3153 return ret;
3154}
3155
3156static inline bool f2fs_readonly(struct super_block *sb)
3157{
3158 return sb_rdonly(sb);
3159}
3160
3161static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3162{
3163 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3164}
3165
3166static inline bool is_dot_dotdot(const u8 *name, size_t len)
3167{
3168 if (len == 1 && name[0] == '.')
3169 return true;
3170
3171 if (len == 2 && name[0] == '.' && name[1] == '.')
3172 return true;
3173
3174 return false;
3175}
3176
3177static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3178 size_t size, gfp_t flags)
3179{
3180 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3181 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3182 return NULL;
3183 }
3184
3185 return kmalloc(size, flags);
3186}
3187
3188static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3189 size_t size, gfp_t flags)
3190{
3191 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3192}
3193
3194static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3195 size_t size, gfp_t flags)
3196{
3197 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3198 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3199 return NULL;
3200 }
3201
3202 return kvmalloc(size, flags);
3203}
3204
3205static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3206 size_t size, gfp_t flags)
3207{
3208 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3209}
3210
3211static inline int get_extra_isize(struct inode *inode)
3212{
3213 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3214}
3215
3216static inline int get_inline_xattr_addrs(struct inode *inode)
3217{
3218 return F2FS_I(inode)->i_inline_xattr_size;
3219}
3220
3221#define f2fs_get_inode_mode(i) \
3222 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3223 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3224
3225#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3226 (offsetof(struct f2fs_inode, i_extra_end) - \
3227 offsetof(struct f2fs_inode, i_extra_isize)) \
3228
3229#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3230#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3231 ((offsetof(typeof(*(f2fs_inode)), field) + \
3232 sizeof((f2fs_inode)->field)) \
3233 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3234
3235#define DEFAULT_IOSTAT_PERIOD_MS 3000
3236#define MIN_IOSTAT_PERIOD_MS 100
3237/* maximum period of iostat tracing is 1 day */
3238#define MAX_IOSTAT_PERIOD_MS 8640000
3239
3240static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3241{
3242 int i;
3243
3244 spin_lock(&sbi->iostat_lock);
3245 for (i = 0; i < NR_IO_TYPE; i++) {
3246 sbi->rw_iostat[i] = 0;
3247 sbi->prev_rw_iostat[i] = 0;
3248 }
3249 spin_unlock(&sbi->iostat_lock);
3250}
3251
3252extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3253
3254static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3255 enum iostat_type type, unsigned long long io_bytes)
3256{
3257 if (!sbi->iostat_enable)
3258 return;
3259 spin_lock(&sbi->iostat_lock);
3260 sbi->rw_iostat[type] += io_bytes;
3261
3262 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3263 sbi->rw_iostat[APP_BUFFERED_IO] =
3264 sbi->rw_iostat[APP_WRITE_IO] -
3265 sbi->rw_iostat[APP_DIRECT_IO];
3266
3267 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3268 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3269 sbi->rw_iostat[APP_READ_IO] -
3270 sbi->rw_iostat[APP_DIRECT_READ_IO];
3271 spin_unlock(&sbi->iostat_lock);
3272
3273 f2fs_record_iostat(sbi);
3274}
3275
3276#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3277
3278#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3279
3280bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3281 block_t blkaddr, int type);
3282static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3283 block_t blkaddr, int type)
3284{
3285 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3286 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3287 blkaddr, type);
3288 f2fs_bug_on(sbi, 1);
3289 }
3290}
3291
3292static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3293{
3294 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3295 blkaddr == COMPRESS_ADDR)
3296 return false;
3297 return true;
3298}
3299
3300/**
3301 * attach_page_private - Attach private data to a page.
3302 * @page: Page to attach data to.
3303 * @data: Data to attach to page.
3304 *
3305 * Attaching private data to a page increments the page's reference count.
3306 * The data must be detached before the page will be freed.
3307 */
3308static inline void attach_page_private(struct page *page, void *data)
3309{
3310 get_page(page);
3311 set_page_private(page, (unsigned long)data);
3312 SetPagePrivate(page);
3313}
3314
3315/**
3316 * detach_page_private - Detach private data from a page.
3317 * @page: Page to detach data from.
3318 *
3319 * Removes the data that was previously attached to the page and decrements
3320 * the refcount on the page.
3321 *
3322 * Return: Data that was attached to the page.
3323 */
3324static inline void *detach_page_private(struct page *page)
3325{
3326 void *data = (void *)page_private(page);
3327
3328 if (!PagePrivate(page))
3329 return NULL;
3330 ClearPagePrivate(page);
3331 set_page_private(page, 0);
3332 put_page(page);
3333
3334 return data;
3335}
3336
3337/*
3338 * file.c
3339 */
3340int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3341void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3342int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3343int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3344int f2fs_truncate(struct inode *inode);
3345int f2fs_getattr(const struct path *path, struct kstat *stat,
3346 u32 request_mask, unsigned int flags);
3347int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3348int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3349void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3350int f2fs_precache_extents(struct inode *inode);
3351long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3352long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3353int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3354int f2fs_pin_file_control(struct inode *inode, bool inc);
3355
3356/*
3357 * inode.c
3358 */
3359void f2fs_set_inode_flags(struct inode *inode);
3360bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3361void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3362struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3363struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3364int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3365void f2fs_update_inode(struct inode *inode, struct page *node_page);
3366void f2fs_update_inode_page(struct inode *inode);
3367int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3368void f2fs_evict_inode(struct inode *inode);
3369void f2fs_handle_failed_inode(struct inode *inode);
3370
3371/*
3372 * namei.c
3373 */
3374int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3375 bool hot, bool set);
3376struct dentry *f2fs_get_parent(struct dentry *child);
3377
3378/*
3379 * dir.c
3380 */
3381unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3382int f2fs_init_casefolded_name(const struct inode *dir,
3383 struct f2fs_filename *fname);
3384int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3385 int lookup, struct f2fs_filename *fname);
3386int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3387 struct f2fs_filename *fname);
3388void f2fs_free_filename(struct f2fs_filename *fname);
3389struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3390 const struct f2fs_filename *fname, int *max_slots);
3391int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3392 unsigned int start_pos, struct fscrypt_str *fstr);
3393void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3394 struct f2fs_dentry_ptr *d);
3395struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3396 const struct f2fs_filename *fname, struct page *dpage);
3397void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3398 unsigned int current_depth);
3399int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3400void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3401struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3402 const struct f2fs_filename *fname,
3403 struct page **res_page);
3404struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3405 const struct qstr *child, struct page **res_page);
3406struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3407ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3408 struct page **page);
3409void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3410 struct page *page, struct inode *inode);
3411bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3412 const struct f2fs_filename *fname);
3413void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3414 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3415 unsigned int bit_pos);
3416int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3417 struct inode *inode, nid_t ino, umode_t mode);
3418int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3419 struct inode *inode, nid_t ino, umode_t mode);
3420int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3421 struct inode *inode, nid_t ino, umode_t mode);
3422void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3423 struct inode *dir, struct inode *inode);
3424int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3425bool f2fs_empty_dir(struct inode *dir);
3426
3427static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3428{
3429 if (fscrypt_is_nokey_name(dentry))
3430 return -ENOKEY;
3431 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3432 inode, inode->i_ino, inode->i_mode);
3433}
3434
3435/*
3436 * super.c
3437 */
3438int f2fs_inode_dirtied(struct inode *inode, bool sync);
3439void f2fs_inode_synced(struct inode *inode);
3440int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3441int f2fs_quota_sync(struct super_block *sb, int type);
3442loff_t max_file_blocks(struct inode *inode);
3443void f2fs_quota_off_umount(struct super_block *sb);
3444int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3445int f2fs_sync_fs(struct super_block *sb, int sync);
3446int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3447
3448/*
3449 * hash.c
3450 */
3451void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3452
3453/*
3454 * node.c
3455 */
3456struct node_info;
3457
3458int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3459bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3460bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3461void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3462void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3463void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3464int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3465bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3466bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3467int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3468 struct node_info *ni);
3469pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3470int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3471int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3472int f2fs_truncate_xattr_node(struct inode *inode);
3473int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3474 unsigned int seq_id);
3475int f2fs_remove_inode_page(struct inode *inode);
3476struct page *f2fs_new_inode_page(struct inode *inode);
3477struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3478void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3479struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3480struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3481int f2fs_move_node_page(struct page *node_page, int gc_type);
3482void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3483int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3484 struct writeback_control *wbc, bool atomic,
3485 unsigned int *seq_id);
3486int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3487 struct writeback_control *wbc,
3488 bool do_balance, enum iostat_type io_type);
3489int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3490bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3491void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3492void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3493int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3494int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3495int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3496int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3497int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3498 unsigned int segno, struct f2fs_summary_block *sum);
3499int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3500int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3501void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3502int __init f2fs_create_node_manager_caches(void);
3503void f2fs_destroy_node_manager_caches(void);
3504
3505/*
3506 * segment.c
3507 */
3508bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3509void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3510void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3511void f2fs_drop_inmem_pages(struct inode *inode);
3512void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3513int f2fs_commit_inmem_pages(struct inode *inode);
3514void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3515void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3516int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3517int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3518int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3519void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3520void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3521bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3522void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3523void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3524bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3525void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3526 struct cp_control *cpc);
3527void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3528block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3529int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3530void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3531int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3532bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3533void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3534void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3535void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3536void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3537 unsigned int *newseg, bool new_sec, int dir);
3538void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3539 unsigned int start, unsigned int end);
3540void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3541void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3542int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3543bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3544 struct cp_control *cpc);
3545struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3546void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3547 block_t blk_addr);
3548void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3549 enum iostat_type io_type);
3550void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3551void f2fs_outplace_write_data(struct dnode_of_data *dn,
3552 struct f2fs_io_info *fio);
3553int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3554void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3555 block_t old_blkaddr, block_t new_blkaddr,
3556 bool recover_curseg, bool recover_newaddr,
3557 bool from_gc);
3558void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3559 block_t old_addr, block_t new_addr,
3560 unsigned char version, bool recover_curseg,
3561 bool recover_newaddr);
3562void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3563 block_t old_blkaddr, block_t *new_blkaddr,
3564 struct f2fs_summary *sum, int type,
3565 struct f2fs_io_info *fio);
3566void f2fs_wait_on_page_writeback(struct page *page,
3567 enum page_type type, bool ordered, bool locked);
3568void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3569void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3570 block_t len);
3571void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3572void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3573int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3574 unsigned int val, int alloc);
3575void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3576int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3577int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3578int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3579void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3580int __init f2fs_create_segment_manager_caches(void);
3581void f2fs_destroy_segment_manager_caches(void);
3582int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3583enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3584 enum page_type type, enum temp_type temp);
3585unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3586 unsigned int segno);
3587unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3588 unsigned int segno);
3589
3590/*
3591 * checkpoint.c
3592 */
3593void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3594struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3595struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3596struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3597struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3598bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3599 block_t blkaddr, int type);
3600int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3601 int type, bool sync);
3602void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3603long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3604 long nr_to_write, enum iostat_type io_type);
3605void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3606void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3607void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3608bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3609void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3610 unsigned int devidx, int type);
3611bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3612 unsigned int devidx, int type);
3613int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3614int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3615void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3616void f2fs_add_orphan_inode(struct inode *inode);
3617void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3618int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3619int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3620void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3621void f2fs_remove_dirty_inode(struct inode *inode);
3622int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3623void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3624u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3625int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3626void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3627int __init f2fs_create_checkpoint_caches(void);
3628void f2fs_destroy_checkpoint_caches(void);
3629int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3630int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3631void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3632void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3633
3634/*
3635 * data.c
3636 */
3637int __init f2fs_init_bioset(void);
3638void f2fs_destroy_bioset(void);
3639int f2fs_init_bio_entry_cache(void);
3640void f2fs_destroy_bio_entry_cache(void);
3641void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3642 struct bio *bio, enum page_type type);
3643void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3644void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3645 struct inode *inode, struct page *page,
3646 nid_t ino, enum page_type type);
3647void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3648 struct bio **bio, struct page *page);
3649void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3650int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3651int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3652void f2fs_submit_page_write(struct f2fs_io_info *fio);
3653struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3654 block_t blk_addr, struct bio *bio);
3655int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3656void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3657void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3658int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3659int f2fs_reserve_new_block(struct dnode_of_data *dn);
3660int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3661int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3662int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3663int f2fs_mpage_readpages(struct address_space *mapping,
3664 struct list_head *pages, struct page *page,
3665 unsigned nr_pages, bool is_readahead);
3666struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3667 int op_flags, bool for_write);
3668struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3669struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3670 bool for_write);
3671struct page *f2fs_get_new_data_page(struct inode *inode,
3672 struct page *ipage, pgoff_t index, bool new_i_size);
3673int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3674void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3675int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3676 int create, int flag);
3677int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3678 u64 start, u64 len);
3679int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3680bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3681bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3682int f2fs_write_single_data_page(struct page *page, int *submitted,
3683 struct bio **bio, sector_t *last_block,
3684 struct writeback_control *wbc,
3685 enum iostat_type io_type,
3686 int compr_blocks, bool allow_balance);
3687void f2fs_invalidate_page(struct page *page, unsigned int offset,
3688 unsigned int length);
3689int f2fs_release_page(struct page *page, gfp_t wait);
3690#ifdef CONFIG_MIGRATION
3691int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3692 struct page *page, enum migrate_mode mode);
3693#endif
3694bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3695void f2fs_clear_page_cache_dirty_tag(struct page *page);
3696int f2fs_init_post_read_processing(void);
3697void f2fs_destroy_post_read_processing(void);
3698int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3699void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3700
3701/*
3702 * gc.c
3703 */
3704int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3705void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3706block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3707int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force,
3708 unsigned int segno);
3709void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3710int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3711int __init f2fs_create_garbage_collection_cache(void);
3712void f2fs_destroy_garbage_collection_cache(void);
3713
3714/*
3715 * recovery.c
3716 */
3717int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3718bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3719int __init f2fs_create_recovery_cache(void);
3720void f2fs_destroy_recovery_cache(void);
3721
3722/*
3723 * debug.c
3724 */
3725#ifdef CONFIG_F2FS_STAT_FS
3726struct f2fs_stat_info {
3727 struct list_head stat_list;
3728 struct f2fs_sb_info *sbi;
3729 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3730 int main_area_segs, main_area_sections, main_area_zones;
3731 unsigned long long hit_largest, hit_cached, hit_rbtree;
3732 unsigned long long hit_total, total_ext;
3733 int ext_tree, zombie_tree, ext_node;
3734 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3735 int ndirty_data, ndirty_qdata;
3736 int inmem_pages;
3737 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3738 int nats, dirty_nats, sits, dirty_sits;
3739 int free_nids, avail_nids, alloc_nids;
3740 int total_count, utilization;
3741 int bg_gc, nr_wb_cp_data, nr_wb_data;
3742 int nr_rd_data, nr_rd_node, nr_rd_meta;
3743 int nr_dio_read, nr_dio_write;
3744 unsigned int io_skip_bggc, other_skip_bggc;
3745 int nr_flushing, nr_flushed, flush_list_empty;
3746 int nr_discarding, nr_discarded;
3747 int nr_discard_cmd;
3748 unsigned int undiscard_blks;
3749 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3750 unsigned int cur_ckpt_time, peak_ckpt_time;
3751 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3752 int compr_inode;
3753 unsigned long long compr_blocks;
3754 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3755 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3756 unsigned int bimodal, avg_vblocks;
3757 int util_free, util_valid, util_invalid;
3758 int rsvd_segs, overp_segs;
3759 int dirty_count, node_pages, meta_pages, compress_pages;
3760 int compress_page_hit;
3761 int prefree_count, call_count, cp_count, bg_cp_count;
3762 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3763 int bg_node_segs, bg_data_segs;
3764 int tot_blks, data_blks, node_blks;
3765 int bg_data_blks, bg_node_blks;
3766 unsigned long long skipped_atomic_files[2];
3767 int curseg[NR_CURSEG_TYPE];
3768 int cursec[NR_CURSEG_TYPE];
3769 int curzone[NR_CURSEG_TYPE];
3770 unsigned int dirty_seg[NR_CURSEG_TYPE];
3771 unsigned int full_seg[NR_CURSEG_TYPE];
3772 unsigned int valid_blks[NR_CURSEG_TYPE];
3773
3774 unsigned int meta_count[META_MAX];
3775 unsigned int segment_count[2];
3776 unsigned int block_count[2];
3777 unsigned int inplace_count;
3778 unsigned long long base_mem, cache_mem, page_mem;
3779};
3780
3781static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3782{
3783 return (struct f2fs_stat_info *)sbi->stat_info;
3784}
3785
3786#define stat_inc_cp_count(si) ((si)->cp_count++)
3787#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3788#define stat_inc_call_count(si) ((si)->call_count++)
3789#define stat_inc_bggc_count(si) ((si)->bg_gc++)
3790#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3791#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3792#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3793#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3794#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3795#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3796#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3797#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3798#define stat_inc_inline_xattr(inode) \
3799 do { \
3800 if (f2fs_has_inline_xattr(inode)) \
3801 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3802 } while (0)
3803#define stat_dec_inline_xattr(inode) \
3804 do { \
3805 if (f2fs_has_inline_xattr(inode)) \
3806 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3807 } while (0)
3808#define stat_inc_inline_inode(inode) \
3809 do { \
3810 if (f2fs_has_inline_data(inode)) \
3811 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3812 } while (0)
3813#define stat_dec_inline_inode(inode) \
3814 do { \
3815 if (f2fs_has_inline_data(inode)) \
3816 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3817 } while (0)
3818#define stat_inc_inline_dir(inode) \
3819 do { \
3820 if (f2fs_has_inline_dentry(inode)) \
3821 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3822 } while (0)
3823#define stat_dec_inline_dir(inode) \
3824 do { \
3825 if (f2fs_has_inline_dentry(inode)) \
3826 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3827 } while (0)
3828#define stat_inc_compr_inode(inode) \
3829 do { \
3830 if (f2fs_compressed_file(inode)) \
3831 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3832 } while (0)
3833#define stat_dec_compr_inode(inode) \
3834 do { \
3835 if (f2fs_compressed_file(inode)) \
3836 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3837 } while (0)
3838#define stat_add_compr_blocks(inode, blocks) \
3839 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3840#define stat_sub_compr_blocks(inode, blocks) \
3841 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3842#define stat_inc_meta_count(sbi, blkaddr) \
3843 do { \
3844 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3845 atomic_inc(&(sbi)->meta_count[META_CP]); \
3846 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3847 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3848 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3849 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3850 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3851 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3852 } while (0)
3853#define stat_inc_seg_type(sbi, curseg) \
3854 ((sbi)->segment_count[(curseg)->alloc_type]++)
3855#define stat_inc_block_count(sbi, curseg) \
3856 ((sbi)->block_count[(curseg)->alloc_type]++)
3857#define stat_inc_inplace_blocks(sbi) \
3858 (atomic_inc(&(sbi)->inplace_count))
3859#define stat_update_max_atomic_write(inode) \
3860 do { \
3861 int cur = F2FS_I_SB(inode)->atomic_files; \
3862 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3863 if (cur > max) \
3864 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3865 } while (0)
3866#define stat_inc_volatile_write(inode) \
3867 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3868#define stat_dec_volatile_write(inode) \
3869 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3870#define stat_update_max_volatile_write(inode) \
3871 do { \
3872 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3873 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3874 if (cur > max) \
3875 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3876 } while (0)
3877#define stat_inc_seg_count(sbi, type, gc_type) \
3878 do { \
3879 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3880 si->tot_segs++; \
3881 if ((type) == SUM_TYPE_DATA) { \
3882 si->data_segs++; \
3883 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3884 } else { \
3885 si->node_segs++; \
3886 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3887 } \
3888 } while (0)
3889
3890#define stat_inc_tot_blk_count(si, blks) \
3891 ((si)->tot_blks += (blks))
3892
3893#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3894 do { \
3895 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3896 stat_inc_tot_blk_count(si, blks); \
3897 si->data_blks += (blks); \
3898 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3899 } while (0)
3900
3901#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3902 do { \
3903 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3904 stat_inc_tot_blk_count(si, blks); \
3905 si->node_blks += (blks); \
3906 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3907 } while (0)
3908
3909int f2fs_build_stats(struct f2fs_sb_info *sbi);
3910void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3911void __init f2fs_create_root_stats(void);
3912void f2fs_destroy_root_stats(void);
3913void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3914#else
3915#define stat_inc_cp_count(si) do { } while (0)
3916#define stat_inc_bg_cp_count(si) do { } while (0)
3917#define stat_inc_call_count(si) do { } while (0)
3918#define stat_inc_bggc_count(si) do { } while (0)
3919#define stat_io_skip_bggc_count(sbi) do { } while (0)
3920#define stat_other_skip_bggc_count(sbi) do { } while (0)
3921#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3922#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3923#define stat_inc_total_hit(sbi) do { } while (0)
3924#define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3925#define stat_inc_largest_node_hit(sbi) do { } while (0)
3926#define stat_inc_cached_node_hit(sbi) do { } while (0)
3927#define stat_inc_inline_xattr(inode) do { } while (0)
3928#define stat_dec_inline_xattr(inode) do { } while (0)
3929#define stat_inc_inline_inode(inode) do { } while (0)
3930#define stat_dec_inline_inode(inode) do { } while (0)
3931#define stat_inc_inline_dir(inode) do { } while (0)
3932#define stat_dec_inline_dir(inode) do { } while (0)
3933#define stat_inc_compr_inode(inode) do { } while (0)
3934#define stat_dec_compr_inode(inode) do { } while (0)
3935#define stat_add_compr_blocks(inode, blocks) do { } while (0)
3936#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3937#define stat_update_max_atomic_write(inode) do { } while (0)
3938#define stat_inc_volatile_write(inode) do { } while (0)
3939#define stat_dec_volatile_write(inode) do { } while (0)
3940#define stat_update_max_volatile_write(inode) do { } while (0)
3941#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3942#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3943#define stat_inc_block_count(sbi, curseg) do { } while (0)
3944#define stat_inc_inplace_blocks(sbi) do { } while (0)
3945#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3946#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3947#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3948#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3949
3950static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3951static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3952static inline void __init f2fs_create_root_stats(void) { }
3953static inline void f2fs_destroy_root_stats(void) { }
3954static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3955#endif
3956
3957extern const struct file_operations f2fs_dir_operations;
3958extern const struct file_operations f2fs_file_operations;
3959extern const struct inode_operations f2fs_file_inode_operations;
3960extern const struct address_space_operations f2fs_dblock_aops;
3961extern const struct address_space_operations f2fs_node_aops;
3962extern const struct address_space_operations f2fs_meta_aops;
3963extern const struct inode_operations f2fs_dir_inode_operations;
3964extern const struct inode_operations f2fs_symlink_inode_operations;
3965extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3966extern const struct inode_operations f2fs_special_inode_operations;
3967extern struct kmem_cache *f2fs_inode_entry_slab;
3968
3969/*
3970 * inline.c
3971 */
3972bool f2fs_may_inline_data(struct inode *inode);
3973bool f2fs_may_inline_dentry(struct inode *inode);
3974void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3975void f2fs_truncate_inline_inode(struct inode *inode,
3976 struct page *ipage, u64 from);
3977int f2fs_read_inline_data(struct inode *inode, struct page *page);
3978int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3979int f2fs_convert_inline_inode(struct inode *inode);
3980int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3981int f2fs_write_inline_data(struct inode *inode, struct page *page);
3982int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3983struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3984 const struct f2fs_filename *fname,
3985 struct page **res_page);
3986int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3987 struct page *ipage);
3988int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3989 struct inode *inode, nid_t ino, umode_t mode);
3990void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3991 struct page *page, struct inode *dir,
3992 struct inode *inode);
3993bool f2fs_empty_inline_dir(struct inode *dir);
3994int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3995 struct fscrypt_str *fstr);
3996int f2fs_inline_data_fiemap(struct inode *inode,
3997 struct fiemap_extent_info *fieinfo,
3998 __u64 start, __u64 len);
3999
4000/*
4001 * shrinker.c
4002 */
4003unsigned long f2fs_shrink_count(struct shrinker *shrink,
4004 struct shrink_control *sc);
4005unsigned long f2fs_shrink_scan(struct shrinker *shrink,
4006 struct shrink_control *sc);
4007void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
4008void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
4009
4010/*
4011 * extent_cache.c
4012 */
4013struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
4014 struct rb_entry *cached_re, unsigned int ofs);
4015struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
4016 struct rb_root_cached *root,
4017 struct rb_node **parent,
4018 unsigned long long key, bool *left_most);
4019struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
4020 struct rb_root_cached *root,
4021 struct rb_node **parent,
4022 unsigned int ofs, bool *leftmost);
4023struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
4024 struct rb_entry *cached_re, unsigned int ofs,
4025 struct rb_entry **prev_entry, struct rb_entry **next_entry,
4026 struct rb_node ***insert_p, struct rb_node **insert_parent,
4027 bool force, bool *leftmost);
4028bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
4029 struct rb_root_cached *root, bool check_key);
4030unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
4031void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
4032void f2fs_drop_extent_tree(struct inode *inode);
4033unsigned int f2fs_destroy_extent_node(struct inode *inode);
4034void f2fs_destroy_extent_tree(struct inode *inode);
4035bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
4036 struct extent_info *ei);
4037void f2fs_update_extent_cache(struct dnode_of_data *dn);
4038void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
4039 pgoff_t fofs, block_t blkaddr, unsigned int len);
4040void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4041int __init f2fs_create_extent_cache(void);
4042void f2fs_destroy_extent_cache(void);
4043
4044/*
4045 * sysfs.c
4046 */
4047int __init f2fs_init_sysfs(void);
4048void f2fs_exit_sysfs(void);
4049int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4050void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4051
4052/* verity.c */
4053extern const struct fsverity_operations f2fs_verityops;
4054
4055/*
4056 * crypto support
4057 */
4058static inline bool f2fs_encrypted_file(struct inode *inode)
4059{
4060 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4061}
4062
4063static inline void f2fs_set_encrypted_inode(struct inode *inode)
4064{
4065#ifdef CONFIG_FS_ENCRYPTION
4066 file_set_encrypt(inode);
4067 f2fs_set_inode_flags(inode);
4068#endif
4069}
4070
4071/*
4072 * Returns true if the reads of the inode's data need to undergo some
4073 * postprocessing step, like decryption or authenticity verification.
4074 */
4075static inline bool f2fs_post_read_required(struct inode *inode)
4076{
4077 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4078 f2fs_compressed_file(inode);
4079}
4080
4081/*
4082 * compress.c
4083 */
4084#ifdef CONFIG_F2FS_FS_COMPRESSION
4085bool f2fs_is_compressed_page(struct page *page);
4086struct page *f2fs_compress_control_page(struct page *page);
4087int f2fs_prepare_compress_overwrite(struct inode *inode,
4088 struct page **pagep, pgoff_t index, void **fsdata);
4089bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4090 pgoff_t index, unsigned copied);
4091int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4092void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4093bool f2fs_is_compress_backend_ready(struct inode *inode);
4094int f2fs_init_compress_mempool(void);
4095void f2fs_destroy_compress_mempool(void);
4096void f2fs_decompress_cluster(struct decompress_io_ctx *dic);
4097void f2fs_end_read_compressed_page(struct page *page, bool failed,
4098 block_t blkaddr);
4099bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4100bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4101void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4102int f2fs_write_multi_pages(struct compress_ctx *cc,
4103 int *submitted,
4104 struct writeback_control *wbc,
4105 enum iostat_type io_type);
4106int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4107int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4108 unsigned nr_pages, sector_t *last_block_in_bio,
4109 bool is_readahead, bool for_write);
4110struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4111void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed);
4112void f2fs_put_page_dic(struct page *page);
4113int f2fs_init_compress_ctx(struct compress_ctx *cc);
4114void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4115void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4116int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4117void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4118int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4119void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4120int __init f2fs_init_compress_cache(void);
4121void f2fs_destroy_compress_cache(void);
4122struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4123void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4124void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4125 nid_t ino, block_t blkaddr);
4126bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4127 block_t blkaddr);
4128void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4129#define inc_compr_inode_stat(inode) \
4130 do { \
4131 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4132 sbi->compr_new_inode++; \
4133 } while (0)
4134#define add_compr_block_stat(inode, blocks) \
4135 do { \
4136 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4137 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4138 sbi->compr_written_block += blocks; \
4139 sbi->compr_saved_block += diff; \
4140 } while (0)
4141#else
4142static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4143static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4144{
4145 if (!f2fs_compressed_file(inode))
4146 return true;
4147 /* not support compression */
4148 return false;
4149}
4150static inline struct page *f2fs_compress_control_page(struct page *page)
4151{
4152 WARN_ON_ONCE(1);
4153 return ERR_PTR(-EINVAL);
4154}
4155static inline int f2fs_init_compress_mempool(void) { return 0; }
4156static inline void f2fs_destroy_compress_mempool(void) { }
4157static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { }
4158static inline void f2fs_end_read_compressed_page(struct page *page,
4159 bool failed, block_t blkaddr)
4160{
4161 WARN_ON_ONCE(1);
4162}
4163static inline void f2fs_put_page_dic(struct page *page)
4164{
4165 WARN_ON_ONCE(1);
4166}
4167static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4168static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4169static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4170static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4171static inline int __init f2fs_init_compress_cache(void) { return 0; }
4172static inline void f2fs_destroy_compress_cache(void) { }
4173static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4174 block_t blkaddr) { }
4175static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4176 struct page *page, nid_t ino, block_t blkaddr) { }
4177static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4178 struct page *page, block_t blkaddr) { return false; }
4179static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4180 nid_t ino) { }
4181#define inc_compr_inode_stat(inode) do { } while (0)
4182#endif
4183
4184static inline void set_compress_context(struct inode *inode)
4185{
4186 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4187
4188 F2FS_I(inode)->i_compress_algorithm =
4189 F2FS_OPTION(sbi).compress_algorithm;
4190 F2FS_I(inode)->i_log_cluster_size =
4191 F2FS_OPTION(sbi).compress_log_size;
4192 F2FS_I(inode)->i_compress_flag =
4193 F2FS_OPTION(sbi).compress_chksum ?
4194 1 << COMPRESS_CHKSUM : 0;
4195 F2FS_I(inode)->i_cluster_size =
4196 1 << F2FS_I(inode)->i_log_cluster_size;
4197 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 &&
4198 F2FS_OPTION(sbi).compress_level)
4199 F2FS_I(inode)->i_compress_flag |=
4200 F2FS_OPTION(sbi).compress_level <<
4201 COMPRESS_LEVEL_OFFSET;
4202 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4203 set_inode_flag(inode, FI_COMPRESSED_FILE);
4204 stat_inc_compr_inode(inode);
4205 inc_compr_inode_stat(inode);
4206 f2fs_mark_inode_dirty_sync(inode, true);
4207}
4208
4209static inline bool f2fs_disable_compressed_file(struct inode *inode)
4210{
4211 struct f2fs_inode_info *fi = F2FS_I(inode);
4212
4213 if (!f2fs_compressed_file(inode))
4214 return true;
4215 if (S_ISREG(inode->i_mode) &&
4216 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks)))
4217 return false;
4218
4219 fi->i_flags &= ~F2FS_COMPR_FL;
4220 stat_dec_compr_inode(inode);
4221 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4222 f2fs_mark_inode_dirty_sync(inode, true);
4223 return true;
4224}
4225
4226#define F2FS_FEATURE_FUNCS(name, flagname) \
4227static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4228{ \
4229 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4230}
4231
4232F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4233F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4234F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4235F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4236F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4237F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4238F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4239F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4240F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4241F2FS_FEATURE_FUNCS(verity, VERITY);
4242F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4243F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4244F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4245F2FS_FEATURE_FUNCS(readonly, RO);
4246
4247static inline bool f2fs_may_extent_tree(struct inode *inode)
4248{
4249 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4250
4251 if (!test_opt(sbi, EXTENT_CACHE) ||
4252 is_inode_flag_set(inode, FI_NO_EXTENT) ||
4253 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) &&
4254 !f2fs_sb_has_readonly(sbi)))
4255 return false;
4256
4257 /*
4258 * for recovered files during mount do not create extents
4259 * if shrinker is not registered.
4260 */
4261 if (list_empty(&sbi->s_list))
4262 return false;
4263
4264 return S_ISREG(inode->i_mode);
4265}
4266
4267#ifdef CONFIG_BLK_DEV_ZONED
4268static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4269 block_t blkaddr)
4270{
4271 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4272
4273 return test_bit(zno, FDEV(devi).blkz_seq);
4274}
4275#endif
4276
4277static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4278{
4279 return f2fs_sb_has_blkzoned(sbi);
4280}
4281
4282static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4283{
4284 return blk_queue_discard(bdev_get_queue(bdev)) ||
4285 bdev_is_zoned(bdev);
4286}
4287
4288static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4289{
4290 int i;
4291
4292 if (!f2fs_is_multi_device(sbi))
4293 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4294
4295 for (i = 0; i < sbi->s_ndevs; i++)
4296 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4297 return true;
4298 return false;
4299}
4300
4301static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4302{
4303 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4304 f2fs_hw_should_discard(sbi);
4305}
4306
4307static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4308{
4309 int i;
4310
4311 if (!f2fs_is_multi_device(sbi))
4312 return bdev_read_only(sbi->sb->s_bdev);
4313
4314 for (i = 0; i < sbi->s_ndevs; i++)
4315 if (bdev_read_only(FDEV(i).bdev))
4316 return true;
4317 return false;
4318}
4319
4320static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4321{
4322 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4323}
4324
4325static inline bool f2fs_may_compress(struct inode *inode)
4326{
4327 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4328 f2fs_is_atomic_file(inode) ||
4329 f2fs_is_volatile_file(inode))
4330 return false;
4331 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4332}
4333
4334static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4335 u64 blocks, bool add)
4336{
4337 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4338 struct f2fs_inode_info *fi = F2FS_I(inode);
4339
4340 /* don't update i_compr_blocks if saved blocks were released */
4341 if (!add && !atomic_read(&fi->i_compr_blocks))
4342 return;
4343
4344 if (add) {
4345 atomic_add(diff, &fi->i_compr_blocks);
4346 stat_add_compr_blocks(inode, diff);
4347 } else {
4348 atomic_sub(diff, &fi->i_compr_blocks);
4349 stat_sub_compr_blocks(inode, diff);
4350 }
4351 f2fs_mark_inode_dirty_sync(inode, true);
4352}
4353
4354static inline int block_unaligned_IO(struct inode *inode,
4355 struct kiocb *iocb, struct iov_iter *iter)
4356{
4357 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4358 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4359 loff_t offset = iocb->ki_pos;
4360 unsigned long align = offset | iov_iter_alignment(iter);
4361
4362 return align & blocksize_mask;
4363}
4364
4365static inline int allow_outplace_dio(struct inode *inode,
4366 struct kiocb *iocb, struct iov_iter *iter)
4367{
4368 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4369 int rw = iov_iter_rw(iter);
4370
4371 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4372 !block_unaligned_IO(inode, iocb, iter));
4373}
4374
4375static inline bool f2fs_force_buffered_io(struct inode *inode,
4376 struct kiocb *iocb, struct iov_iter *iter)
4377{
4378 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4379 int rw = iov_iter_rw(iter);
4380
4381 if (!fscrypt_dio_supported(iocb, iter))
4382 return true;
4383 if (fsverity_active(inode))
4384 return true;
4385 if (f2fs_compressed_file(inode))
4386 return true;
4387 if (f2fs_is_multi_device(sbi))
4388 return true;
4389 /*
4390 * for blkzoned device, fallback direct IO to buffered IO, so
4391 * all IOs can be serialized by log-structured write.
4392 */
4393 if (f2fs_sb_has_blkzoned(sbi))
4394 return true;
4395 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4396 if (block_unaligned_IO(inode, iocb, iter))
4397 return true;
4398 if (F2FS_IO_ALIGNED(sbi))
4399 return true;
4400 }
4401 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
4402 return true;
4403
4404 return false;
4405}
4406
4407static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4408{
4409 return fsverity_active(inode) &&
4410 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4411}
4412
4413#ifdef CONFIG_F2FS_FAULT_INJECTION
4414extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4415 unsigned int type);
4416#else
4417#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4418#endif
4419
4420static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4421{
4422#ifdef CONFIG_QUOTA
4423 if (f2fs_sb_has_quota_ino(sbi))
4424 return true;
4425 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4426 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4427 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4428 return true;
4429#endif
4430 return false;
4431}
4432
4433#define EFSBADCRC EBADMSG /* Bad CRC detected */
4434#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4435
4436#endif /* _LINUX_F2FS_H */