1 /* SPDX-License-Identifier: GPL-2.0 */
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
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 <linux/part_stat.h>
26 #include <crypto/hash.h>
28 #include <linux/fscrypt.h>
29 #include <linux/fsverity.h>
31 #ifdef CONFIG_F2FS_CHECK_FS
32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
34 #define f2fs_bug_on(sbi, condition) \
36 if (unlikely(condition)) { \
38 set_sbi_flag(sbi, SBI_NEED_FSCK); \
62 #ifdef CONFIG_F2FS_FAULT_INJECTION
63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
65 struct f2fs_fault_info {
67 unsigned int inject_rate;
68 unsigned int inject_type;
71 extern const char *f2fs_fault_name[FAULT_MAX];
72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
79 #define F2FS_MOUNT_DISCARD 0x00000004
80 #define F2FS_MOUNT_NOHEAP 0x00000008
81 #define F2FS_MOUNT_XATTR_USER 0x00000010
82 #define F2FS_MOUNT_POSIX_ACL 0x00000020
83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
85 #define F2FS_MOUNT_INLINE_DATA 0x00000100
86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
88 #define F2FS_MOUNT_NOBARRIER 0x00000800
89 #define F2FS_MOUNT_FASTBOOT 0x00001000
90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
93 #define F2FS_MOUNT_USRQUOTA 0x00080000
94 #define F2FS_MOUNT_GRPQUOTA 0x00100000
95 #define F2FS_MOUNT_PRJQUOTA 0x00200000
96 #define F2FS_MOUNT_QUOTA 0x00400000
97 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
98 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
99 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
100 #define F2FS_MOUNT_NORECOVERY 0x04000000
101 #define F2FS_MOUNT_ATGC 0x08000000
103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
109 typecheck(unsigned long long, b) && \
110 ((long long)((a) - (b)) > 0))
112 typedef u32 block_t; /*
113 * should not change u32, since it is the on-disk block
114 * address format, __le32.
118 #define COMPRESS_EXT_NUM 16
120 struct f2fs_mount_info {
122 int write_io_size_bits; /* Write IO size bits */
123 block_t root_reserved_blocks; /* root reserved blocks */
124 kuid_t s_resuid; /* reserved blocks for uid */
125 kgid_t s_resgid; /* reserved blocks for gid */
126 int active_logs; /* # of active logs */
127 int inline_xattr_size; /* inline xattr size */
128 #ifdef CONFIG_F2FS_FAULT_INJECTION
129 struct f2fs_fault_info fault_info; /* For fault injection */
132 /* Names of quota files with journalled quota */
133 char *s_qf_names[MAXQUOTAS];
134 int s_jquota_fmt; /* Format of quota to use */
136 /* For which write hints are passed down to block layer */
138 int alloc_mode; /* segment allocation policy */
139 int fsync_mode; /* fsync policy */
140 int fs_mode; /* fs mode: LFS or ADAPTIVE */
141 int bggc_mode; /* bggc mode: off, on or sync */
142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
143 block_t unusable_cap_perc; /* percentage for cap */
144 block_t unusable_cap; /* Amount of space allowed to be
145 * unusable when disabling checkpoint
148 /* For compression */
149 unsigned char compress_algorithm; /* algorithm type */
150 unsigned char compress_log_size; /* cluster log size */
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 */
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
172 #define __F2FS_HAS_FEATURE(raw_super, mask) \
173 ((raw_super->feature & cpu_to_le32(mask)) != 0)
174 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
175 #define F2FS_SET_FEATURE(sbi, mask) \
176 (sbi->raw_super->feature |= cpu_to_le32(mask))
177 #define F2FS_CLEAR_FEATURE(sbi, mask) \
178 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
181 * Default values for user and/or group using reserved blocks
183 #define F2FS_DEF_RESUID 0
184 #define F2FS_DEF_RESGID 0
187 * For checkpoint manager
194 #define CP_UMOUNT 0x00000001
195 #define CP_FASTBOOT 0x00000002
196 #define CP_SYNC 0x00000004
197 #define CP_RECOVERY 0x00000008
198 #define CP_DISCARD 0x00000010
199 #define CP_TRIMMED 0x00000020
200 #define CP_PAUSE 0x00000040
201 #define CP_RESIZE 0x00000080
203 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
204 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
205 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
206 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
207 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
208 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
209 #define DEF_CP_INTERVAL 60 /* 60 secs */
210 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
211 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
212 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
213 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
223 * indicate meta/data type
232 DATA_GENERIC, /* check range only */
233 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
234 DATA_GENERIC_ENHANCE_READ, /*
235 * strong check on range and segment
236 * bitmap but no warning due to race
237 * condition of read on truncated area
240 DATA_GENERIC_ENHANCE_UPDATE, /*
241 * strong check on range and segment
242 * bitmap for update case
247 /* for the list of ino */
249 ORPHAN_INO, /* for orphan ino list */
250 APPEND_INO, /* for append ino list */
251 UPDATE_INO, /* for update ino list */
252 TRANS_DIR_INO, /* for trasactions dir ino list */
253 FLUSH_INO, /* for multiple device flushing */
254 MAX_INO_ENTRY, /* max. list */
258 struct list_head list; /* list head */
259 nid_t ino; /* inode number */
260 unsigned int dirty_device; /* dirty device bitmap */
263 /* for the list of inodes to be GCed */
265 struct list_head list; /* list head */
266 struct inode *inode; /* vfs inode pointer */
269 struct fsync_node_entry {
270 struct list_head list; /* list head */
271 struct page *page; /* warm node page pointer */
272 unsigned int seq_id; /* sequence id */
275 /* for the bitmap indicate blocks to be discarded */
276 struct discard_entry {
277 struct list_head list; /* list head */
278 block_t start_blkaddr; /* start blockaddr of current segment */
279 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
282 /* default discard granularity of inner discard thread, unit: block count */
283 #define DEFAULT_DISCARD_GRANULARITY 16
285 /* max discard pend list number */
286 #define MAX_PLIST_NUM 512
287 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
288 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
291 D_PREP, /* initial */
292 D_PARTIAL, /* partially submitted */
293 D_SUBMIT, /* all submitted */
294 D_DONE, /* finished */
297 struct discard_info {
298 block_t lstart; /* logical start address */
299 block_t len; /* length */
300 block_t start; /* actual start address in dev */
304 struct rb_node rb_node; /* rb node located in rb-tree */
307 block_t lstart; /* logical start address */
308 block_t len; /* length */
309 block_t start; /* actual start address in dev */
311 struct discard_info di; /* discard info */
314 struct list_head list; /* command list */
315 struct completion wait; /* compleation */
316 struct block_device *bdev; /* bdev */
317 unsigned short ref; /* reference count */
318 unsigned char state; /* state */
319 unsigned char queued; /* queued discard */
320 int error; /* bio error */
321 spinlock_t lock; /* for state/bio_ref updating */
322 unsigned short bio_ref; /* bio reference count */
333 struct discard_policy {
334 int type; /* type of discard */
335 unsigned int min_interval; /* used for candidates exist */
336 unsigned int mid_interval; /* used for device busy */
337 unsigned int max_interval; /* used for candidates not exist */
338 unsigned int max_requests; /* # of discards issued per round */
339 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
340 bool io_aware; /* issue discard in idle time */
341 bool sync; /* submit discard with REQ_SYNC flag */
342 bool ordered; /* issue discard by lba order */
343 bool timeout; /* discard timeout for put_super */
344 unsigned int granularity; /* discard granularity */
347 struct discard_cmd_control {
348 struct task_struct *f2fs_issue_discard; /* discard thread */
349 struct list_head entry_list; /* 4KB discard entry list */
350 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
351 struct list_head wait_list; /* store on-flushing entries */
352 struct list_head fstrim_list; /* in-flight discard from fstrim */
353 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
354 unsigned int discard_wake; /* to wake up discard thread */
355 struct mutex cmd_lock;
356 unsigned int nr_discards; /* # of discards in the list */
357 unsigned int max_discards; /* max. discards to be issued */
358 unsigned int discard_granularity; /* discard granularity */
359 unsigned int undiscard_blks; /* # of undiscard blocks */
360 unsigned int next_pos; /* next discard position */
361 atomic_t issued_discard; /* # of issued discard */
362 atomic_t queued_discard; /* # of queued discard */
363 atomic_t discard_cmd_cnt; /* # of cached cmd count */
364 struct rb_root_cached root; /* root of discard rb-tree */
365 bool rbtree_check; /* config for consistence check */
368 /* for the list of fsync inodes, used only during recovery */
369 struct fsync_inode_entry {
370 struct list_head list; /* list head */
371 struct inode *inode; /* vfs inode pointer */
372 block_t blkaddr; /* block address locating the last fsync */
373 block_t last_dentry; /* block address locating the last dentry */
376 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
377 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
379 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
380 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
381 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
382 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
384 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
385 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
387 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
389 int before = nats_in_cursum(journal);
391 journal->n_nats = cpu_to_le16(before + i);
395 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
397 int before = sits_in_cursum(journal);
399 journal->n_sits = cpu_to_le16(before + i);
403 static inline bool __has_cursum_space(struct f2fs_journal *journal,
406 if (type == NAT_JOURNAL)
407 return size <= MAX_NAT_JENTRIES(journal);
408 return size <= MAX_SIT_JENTRIES(journal);
411 /* for inline stuff */
412 #define DEF_INLINE_RESERVED_SIZE 1
413 static inline int get_extra_isize(struct inode *inode);
414 static inline int get_inline_xattr_addrs(struct inode *inode);
415 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
416 (CUR_ADDRS_PER_INODE(inode) - \
417 get_inline_xattr_addrs(inode) - \
418 DEF_INLINE_RESERVED_SIZE))
421 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
422 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
424 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
425 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
426 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
427 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
428 NR_INLINE_DENTRY(inode) + \
429 INLINE_DENTRY_BITMAP_SIZE(inode)))
432 * For INODE and NODE manager
434 /* for directory operations */
436 struct f2fs_filename {
438 * The filename the user specified. This is NULL for some
439 * filesystem-internal operations, e.g. converting an inline directory
440 * to a non-inline one, or roll-forward recovering an encrypted dentry.
442 const struct qstr *usr_fname;
445 * The on-disk filename. For encrypted directories, this is encrypted.
446 * This may be NULL for lookups in an encrypted dir without the key.
448 struct fscrypt_str disk_name;
450 /* The dirhash of this filename */
453 #ifdef CONFIG_FS_ENCRYPTION
455 * For lookups in encrypted directories: either the buffer backing
456 * disk_name, or a buffer that holds the decoded no-key name.
458 struct fscrypt_str crypto_buf;
460 #ifdef CONFIG_UNICODE
462 * For casefolded directories: the casefolded name, but it's left NULL
463 * if the original name is not valid Unicode or if the filesystem is
464 * doing an internal operation where usr_fname is also NULL. In these
465 * cases we fall back to treating the name as an opaque byte sequence.
467 struct fscrypt_str cf_name;
471 struct f2fs_dentry_ptr {
474 struct f2fs_dir_entry *dentry;
475 __u8 (*filename)[F2FS_SLOT_LEN];
480 static inline void make_dentry_ptr_block(struct inode *inode,
481 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
484 d->max = NR_DENTRY_IN_BLOCK;
485 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
486 d->bitmap = t->dentry_bitmap;
487 d->dentry = t->dentry;
488 d->filename = t->filename;
491 static inline void make_dentry_ptr_inline(struct inode *inode,
492 struct f2fs_dentry_ptr *d, void *t)
494 int entry_cnt = NR_INLINE_DENTRY(inode);
495 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
496 int reserved_size = INLINE_RESERVED_SIZE(inode);
500 d->nr_bitmap = bitmap_size;
502 d->dentry = t + bitmap_size + reserved_size;
503 d->filename = t + bitmap_size + reserved_size +
504 SIZE_OF_DIR_ENTRY * entry_cnt;
508 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
509 * as its node offset to distinguish from index node blocks.
510 * But some bits are used to mark the node block.
512 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
515 ALLOC_NODE, /* allocate a new node page if needed */
516 LOOKUP_NODE, /* look up a node without readahead */
518 * look up a node with readahead called
523 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
525 /* congestion wait timeout value, default: 20ms */
526 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
528 /* maximum retry quota flush count */
529 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
531 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
533 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
535 /* for in-memory extent cache entry */
536 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
538 /* number of extent info in extent cache we try to shrink */
539 #define EXTENT_CACHE_SHRINK_NUMBER 128
542 struct rb_node rb_node; /* rb node located in rb-tree */
545 unsigned int ofs; /* start offset of the entry */
546 unsigned int len; /* length of the entry */
548 unsigned long long key; /* 64-bits key */
553 unsigned int fofs; /* start offset in a file */
554 unsigned int len; /* length of the extent */
555 u32 blk; /* start block address of the extent */
559 struct rb_node rb_node; /* rb node located in rb-tree */
560 struct extent_info ei; /* extent info */
561 struct list_head list; /* node in global extent list of sbi */
562 struct extent_tree *et; /* extent tree pointer */
566 nid_t ino; /* inode number */
567 struct rb_root_cached root; /* root of extent info rb-tree */
568 struct extent_node *cached_en; /* recently accessed extent node */
569 struct extent_info largest; /* largested extent info */
570 struct list_head list; /* to be used by sbi->zombie_list */
571 rwlock_t lock; /* protect extent info rb-tree */
572 atomic_t node_cnt; /* # of extent node in rb-tree*/
573 bool largest_updated; /* largest extent updated */
577 * This structure is taken from ext4_map_blocks.
579 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
581 #define F2FS_MAP_NEW (1 << BH_New)
582 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
583 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
584 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
587 struct f2fs_map_blocks {
591 unsigned int m_flags;
592 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
593 pgoff_t *m_next_extent; /* point to next possible extent */
595 bool m_may_create; /* indicate it is from write path */
598 /* for flag in get_data_block */
600 F2FS_GET_BLOCK_DEFAULT,
601 F2FS_GET_BLOCK_FIEMAP,
604 F2FS_GET_BLOCK_PRE_DIO,
605 F2FS_GET_BLOCK_PRE_AIO,
606 F2FS_GET_BLOCK_PRECACHE,
610 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
612 #define FADVISE_COLD_BIT 0x01
613 #define FADVISE_LOST_PINO_BIT 0x02
614 #define FADVISE_ENCRYPT_BIT 0x04
615 #define FADVISE_ENC_NAME_BIT 0x08
616 #define FADVISE_KEEP_SIZE_BIT 0x10
617 #define FADVISE_HOT_BIT 0x20
618 #define FADVISE_VERITY_BIT 0x40
620 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
622 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
623 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
624 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
625 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
626 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
627 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
628 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
629 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
630 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
631 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
632 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
633 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
634 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
635 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
636 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
637 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
638 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
639 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
641 #define DEF_DIR_LEVEL 0
649 /* used for f2fs_inode_info->flags */
651 FI_NEW_INODE, /* indicate newly allocated inode */
652 FI_DIRTY_INODE, /* indicate inode is dirty or not */
653 FI_AUTO_RECOVER, /* indicate inode is recoverable */
654 FI_DIRTY_DIR, /* indicate directory has dirty pages */
655 FI_INC_LINK, /* need to increment i_nlink */
656 FI_ACL_MODE, /* indicate acl mode */
657 FI_NO_ALLOC, /* should not allocate any blocks */
658 FI_FREE_NID, /* free allocated nide */
659 FI_NO_EXTENT, /* not to use the extent cache */
660 FI_INLINE_XATTR, /* used for inline xattr */
661 FI_INLINE_DATA, /* used for inline data*/
662 FI_INLINE_DENTRY, /* used for inline dentry */
663 FI_APPEND_WRITE, /* inode has appended data */
664 FI_UPDATE_WRITE, /* inode has in-place-update data */
665 FI_NEED_IPU, /* used for ipu per file */
666 FI_ATOMIC_FILE, /* indicate atomic file */
667 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
668 FI_VOLATILE_FILE, /* indicate volatile file */
669 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
670 FI_DROP_CACHE, /* drop dirty page cache */
671 FI_DATA_EXIST, /* indicate data exists */
672 FI_INLINE_DOTS, /* indicate inline dot dentries */
673 FI_DO_DEFRAG, /* indicate defragment is running */
674 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
675 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
676 FI_HOT_DATA, /* indicate file is hot */
677 FI_EXTRA_ATTR, /* indicate file has extra attribute */
678 FI_PROJ_INHERIT, /* indicate file inherits projectid */
679 FI_PIN_FILE, /* indicate file should not be gced */
680 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
681 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
682 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
683 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
684 FI_MMAP_FILE, /* indicate file was mmapped */
685 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
686 FI_COMPRESS_RELEASED, /* compressed blocks were released */
687 FI_MAX, /* max flag, never be used */
690 struct f2fs_inode_info {
691 struct inode vfs_inode; /* serve a vfs inode */
692 unsigned long i_flags; /* keep an inode flags for ioctl */
693 unsigned char i_advise; /* use to give file attribute hints */
694 unsigned char i_dir_level; /* use for dentry level for large dir */
695 unsigned int i_current_depth; /* only for directory depth */
696 /* for gc failure statistic */
697 unsigned int i_gc_failures[MAX_GC_FAILURE];
698 unsigned int i_pino; /* parent inode number */
699 umode_t i_acl_mode; /* keep file acl mode temporarily */
701 /* Use below internally in f2fs*/
702 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
703 struct rw_semaphore i_sem; /* protect fi info */
704 atomic_t dirty_pages; /* # of dirty pages */
705 f2fs_hash_t chash; /* hash value of given file name */
706 unsigned int clevel; /* maximum level of given file name */
707 struct task_struct *task; /* lookup and create consistency */
708 struct task_struct *cp_task; /* separate cp/wb IO stats*/
709 struct task_struct *wb_task; /* indicate inode is in context of writeback */
710 nid_t i_xattr_nid; /* node id that contains xattrs */
711 loff_t last_disk_size; /* lastly written file size */
712 spinlock_t i_size_lock; /* protect last_disk_size */
715 struct dquot *i_dquot[MAXQUOTAS];
717 /* quota space reservation, managed internally by quota code */
718 qsize_t i_reserved_quota;
720 struct list_head dirty_list; /* dirty list for dirs and files */
721 struct list_head gdirty_list; /* linked in global dirty list */
722 struct list_head inmem_ilist; /* list for inmem inodes */
723 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
724 struct task_struct *inmem_task; /* store inmemory task */
725 struct mutex inmem_lock; /* lock for inmemory pages */
726 pgoff_t ra_offset; /* ongoing readahead offset */
727 struct extent_tree *extent_tree; /* cached extent_tree entry */
729 /* avoid racing between foreground op and gc */
730 struct rw_semaphore i_gc_rwsem[2];
731 struct rw_semaphore i_mmap_sem;
732 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
734 int i_extra_isize; /* size of extra space located in i_addr */
735 kprojid_t i_projid; /* id for project quota */
736 int i_inline_xattr_size; /* inline xattr size */
737 struct timespec64 i_crtime; /* inode creation time */
738 struct timespec64 i_disk_time[4];/* inode disk times */
740 /* for file compress */
741 atomic_t i_compr_blocks; /* # of compressed blocks */
742 unsigned char i_compress_algorithm; /* algorithm type */
743 unsigned char i_log_cluster_size; /* log of cluster size */
744 unsigned short i_compress_flag; /* compress flag */
745 unsigned int i_cluster_size; /* cluster size */
748 static inline void get_extent_info(struct extent_info *ext,
749 struct f2fs_extent *i_ext)
751 ext->fofs = le32_to_cpu(i_ext->fofs);
752 ext->blk = le32_to_cpu(i_ext->blk);
753 ext->len = le32_to_cpu(i_ext->len);
756 static inline void set_raw_extent(struct extent_info *ext,
757 struct f2fs_extent *i_ext)
759 i_ext->fofs = cpu_to_le32(ext->fofs);
760 i_ext->blk = cpu_to_le32(ext->blk);
761 i_ext->len = cpu_to_le32(ext->len);
764 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
765 u32 blk, unsigned int len)
772 static inline bool __is_discard_mergeable(struct discard_info *back,
773 struct discard_info *front, unsigned int max_len)
775 return (back->lstart + back->len == front->lstart) &&
776 (back->len + front->len <= max_len);
779 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
780 struct discard_info *back, unsigned int max_len)
782 return __is_discard_mergeable(back, cur, max_len);
785 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
786 struct discard_info *front, unsigned int max_len)
788 return __is_discard_mergeable(cur, front, max_len);
791 static inline bool __is_extent_mergeable(struct extent_info *back,
792 struct extent_info *front)
794 return (back->fofs + back->len == front->fofs &&
795 back->blk + back->len == front->blk);
798 static inline bool __is_back_mergeable(struct extent_info *cur,
799 struct extent_info *back)
801 return __is_extent_mergeable(back, cur);
804 static inline bool __is_front_mergeable(struct extent_info *cur,
805 struct extent_info *front)
807 return __is_extent_mergeable(cur, front);
810 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
811 static inline void __try_update_largest_extent(struct extent_tree *et,
812 struct extent_node *en)
814 if (en->ei.len > et->largest.len) {
815 et->largest = en->ei;
816 et->largest_updated = true;
821 * For free nid management
824 FREE_NID, /* newly added to free nid list */
825 PREALLOC_NID, /* it is preallocated */
836 struct f2fs_nm_info {
837 block_t nat_blkaddr; /* base disk address of NAT */
838 nid_t max_nid; /* maximum possible node ids */
839 nid_t available_nids; /* # of available node ids */
840 nid_t next_scan_nid; /* the next nid to be scanned */
841 unsigned int ram_thresh; /* control the memory footprint */
842 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
843 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
845 /* NAT cache management */
846 struct radix_tree_root nat_root;/* root of the nat entry cache */
847 struct radix_tree_root nat_set_root;/* root of the nat set cache */
848 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
849 struct list_head nat_entries; /* cached nat entry list (clean) */
850 spinlock_t nat_list_lock; /* protect clean nat entry list */
851 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
852 unsigned int nat_blocks; /* # of nat blocks */
854 /* free node ids management */
855 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
856 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
857 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
858 spinlock_t nid_list_lock; /* protect nid lists ops */
859 struct mutex build_lock; /* lock for build free nids */
860 unsigned char **free_nid_bitmap;
861 unsigned char *nat_block_bitmap;
862 unsigned short *free_nid_count; /* free nid count of NAT block */
865 char *nat_bitmap; /* NAT bitmap pointer */
867 unsigned int nat_bits_blocks; /* # of nat bits blocks */
868 unsigned char *nat_bits; /* NAT bits blocks */
869 unsigned char *full_nat_bits; /* full NAT pages */
870 unsigned char *empty_nat_bits; /* empty NAT pages */
871 #ifdef CONFIG_F2FS_CHECK_FS
872 char *nat_bitmap_mir; /* NAT bitmap mirror */
874 int bitmap_size; /* bitmap size */
878 * this structure is used as one of function parameters.
879 * all the information are dedicated to a given direct node block determined
880 * by the data offset in a file.
882 struct dnode_of_data {
883 struct inode *inode; /* vfs inode pointer */
884 struct page *inode_page; /* its inode page, NULL is possible */
885 struct page *node_page; /* cached direct node page */
886 nid_t nid; /* node id of the direct node block */
887 unsigned int ofs_in_node; /* data offset in the node page */
888 bool inode_page_locked; /* inode page is locked or not */
889 bool node_changed; /* is node block changed */
890 char cur_level; /* level of hole node page */
891 char max_level; /* level of current page located */
892 block_t data_blkaddr; /* block address of the node block */
895 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
896 struct page *ipage, struct page *npage, nid_t nid)
898 memset(dn, 0, sizeof(*dn));
900 dn->inode_page = ipage;
901 dn->node_page = npage;
908 * By default, there are 6 active log areas across the whole main area.
909 * When considering hot and cold data separation to reduce cleaning overhead,
910 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
912 * In the current design, you should not change the numbers intentionally.
913 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
914 * logs individually according to the underlying devices. (default: 6)
915 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
916 * data and 8 for node logs.
918 #define NR_CURSEG_DATA_TYPE (3)
919 #define NR_CURSEG_NODE_TYPE (3)
920 #define NR_CURSEG_INMEM_TYPE (2)
921 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
922 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
925 CURSEG_HOT_DATA = 0, /* directory entry blocks */
926 CURSEG_WARM_DATA, /* data blocks */
927 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
928 CURSEG_HOT_NODE, /* direct node blocks of directory files */
929 CURSEG_WARM_NODE, /* direct node blocks of normal files */
930 CURSEG_COLD_NODE, /* indirect node blocks */
931 NR_PERSISTENT_LOG, /* number of persistent log */
932 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
933 /* pinned file that needs consecutive block address */
934 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
935 NO_CHECK_TYPE, /* number of persistent & inmem log */
939 struct completion wait;
940 struct llist_node llnode;
945 struct flush_cmd_control {
946 struct task_struct *f2fs_issue_flush; /* flush thread */
947 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
948 atomic_t issued_flush; /* # of issued flushes */
949 atomic_t queued_flush; /* # of queued flushes */
950 struct llist_head issue_list; /* list for command issue */
951 struct llist_node *dispatch_list; /* list for command dispatch */
954 struct f2fs_sm_info {
955 struct sit_info *sit_info; /* whole segment information */
956 struct free_segmap_info *free_info; /* free segment information */
957 struct dirty_seglist_info *dirty_info; /* dirty segment information */
958 struct curseg_info *curseg_array; /* active segment information */
960 struct rw_semaphore curseg_lock; /* for preventing curseg change */
962 block_t seg0_blkaddr; /* block address of 0'th segment */
963 block_t main_blkaddr; /* start block address of main area */
964 block_t ssa_blkaddr; /* start block address of SSA area */
966 unsigned int segment_count; /* total # of segments */
967 unsigned int main_segments; /* # of segments in main area */
968 unsigned int reserved_segments; /* # of reserved segments */
969 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
970 unsigned int ovp_segments; /* # of overprovision segments */
972 /* a threshold to reclaim prefree segments */
973 unsigned int rec_prefree_segments;
975 /* for batched trimming */
976 unsigned int trim_sections; /* # of sections to trim */
978 struct list_head sit_entry_set; /* sit entry set list */
980 unsigned int ipu_policy; /* in-place-update policy */
981 unsigned int min_ipu_util; /* in-place-update threshold */
982 unsigned int min_fsync_blocks; /* threshold for fsync */
983 unsigned int min_seq_blocks; /* threshold for sequential blocks */
984 unsigned int min_hot_blocks; /* threshold for hot block allocation */
985 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
987 /* for flush command control */
988 struct flush_cmd_control *fcc_info;
990 /* for discard command control */
991 struct discard_cmd_control *dcc_info;
998 * COUNT_TYPE for monitoring
1000 * f2fs monitors the number of several block types such as on-writeback,
1001 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1003 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1023 * The below are the page types of bios used in submit_bio().
1024 * The available types are:
1025 * DATA User data pages. It operates as async mode.
1026 * NODE Node pages. It operates as async mode.
1027 * META FS metadata pages such as SIT, NAT, CP.
1028 * NR_PAGE_TYPE The number of page types.
1029 * META_FLUSH Make sure the previous pages are written
1030 * with waiting the bio's completion
1031 * ... Only can be used with META.
1033 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1036 NODE = 1, /* should not change this */
1040 INMEM, /* the below types are used by tracepoints only. */
1049 HOT = 0, /* must be zero for meta bio */
1055 enum need_lock_type {
1061 enum cp_reason_type {
1077 APP_DIRECT_IO, /* app direct write IOs */
1078 APP_BUFFERED_IO, /* app buffered write IOs */
1079 APP_WRITE_IO, /* app write IOs */
1080 APP_MAPPED_IO, /* app mapped IOs */
1081 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1082 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1083 FS_META_IO, /* meta IOs from kworker/reclaimer */
1084 FS_GC_DATA_IO, /* data IOs from forground gc */
1085 FS_GC_NODE_IO, /* node IOs from forground gc */
1086 FS_CP_DATA_IO, /* data IOs from checkpoint */
1087 FS_CP_NODE_IO, /* node IOs from checkpoint */
1088 FS_CP_META_IO, /* meta IOs from checkpoint */
1091 APP_DIRECT_READ_IO, /* app direct read IOs */
1092 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1093 APP_READ_IO, /* app read IOs */
1094 APP_MAPPED_READ_IO, /* app mapped read IOs */
1095 FS_DATA_READ_IO, /* data read IOs */
1096 FS_GDATA_READ_IO, /* data read IOs from background gc */
1097 FS_CDATA_READ_IO, /* compressed data read IOs */
1098 FS_NODE_READ_IO, /* node read IOs */
1099 FS_META_READ_IO, /* meta read IOs */
1102 FS_DISCARD, /* discard */
1106 struct f2fs_io_info {
1107 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1108 nid_t ino; /* inode number */
1109 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1110 enum temp_type temp; /* contains HOT/WARM/COLD */
1111 int op; /* contains REQ_OP_ */
1112 int op_flags; /* req_flag_bits */
1113 block_t new_blkaddr; /* new block address to be written */
1114 block_t old_blkaddr; /* old block address before Cow */
1115 struct page *page; /* page to be written */
1116 struct page *encrypted_page; /* encrypted page */
1117 struct page *compressed_page; /* compressed page */
1118 struct list_head list; /* serialize IOs */
1119 bool submitted; /* indicate IO submission */
1120 int need_lock; /* indicate we need to lock cp_rwsem */
1121 bool in_list; /* indicate fio is in io_list */
1122 bool is_por; /* indicate IO is from recovery or not */
1123 bool retry; /* need to reallocate block address */
1124 int compr_blocks; /* # of compressed block addresses */
1125 bool encrypted; /* indicate file is encrypted */
1126 enum iostat_type io_type; /* io type */
1127 struct writeback_control *io_wbc; /* writeback control */
1128 struct bio **bio; /* bio for ipu */
1129 sector_t *last_block; /* last block number in bio */
1130 unsigned char version; /* version of the node */
1135 struct list_head list;
1138 #define is_read_io(rw) ((rw) == READ)
1139 struct f2fs_bio_info {
1140 struct f2fs_sb_info *sbi; /* f2fs superblock */
1141 struct bio *bio; /* bios to merge */
1142 sector_t last_block_in_bio; /* last block number */
1143 struct f2fs_io_info fio; /* store buffered io info. */
1144 struct rw_semaphore io_rwsem; /* blocking op for bio */
1145 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1146 struct list_head io_list; /* track fios */
1147 struct list_head bio_list; /* bio entry list head */
1148 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1151 #define FDEV(i) (sbi->devs[i])
1152 #define RDEV(i) (raw_super->devs[i])
1153 struct f2fs_dev_info {
1154 struct block_device *bdev;
1155 char path[MAX_PATH_LEN];
1156 unsigned int total_segments;
1159 #ifdef CONFIG_BLK_DEV_ZONED
1160 unsigned int nr_blkz; /* Total number of zones */
1161 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1166 DIR_INODE, /* for dirty dir inode */
1167 FILE_INODE, /* for dirty regular/symlink inode */
1168 DIRTY_META, /* for all dirtied inode metadata */
1169 ATOMIC_FILE, /* for all atomic files */
1173 /* for inner inode cache management */
1174 struct inode_management {
1175 struct radix_tree_root ino_root; /* ino entry array */
1176 spinlock_t ino_lock; /* for ino entry lock */
1177 struct list_head ino_list; /* inode list head */
1178 unsigned long ino_num; /* number of entries */
1182 struct atgc_management {
1183 bool atgc_enabled; /* ATGC is enabled or not */
1184 struct rb_root_cached root; /* root of victim rb-tree */
1185 struct list_head victim_list; /* linked with all victim entries */
1186 unsigned int victim_count; /* victim count in rb-tree */
1187 unsigned int candidate_ratio; /* candidate ratio */
1188 unsigned int max_candidate_count; /* max candidate count */
1189 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1190 unsigned long long age_threshold; /* age threshold */
1193 /* For s_flag in struct f2fs_sb_info */
1195 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1196 SBI_IS_CLOSE, /* specify unmounting */
1197 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1198 SBI_POR_DOING, /* recovery is doing or not */
1199 SBI_NEED_SB_WRITE, /* need to recover superblock */
1200 SBI_NEED_CP, /* need to checkpoint */
1201 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1202 SBI_IS_RECOVERED, /* recovered orphan/data */
1203 SBI_CP_DISABLED, /* CP was disabled last mount */
1204 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1205 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1206 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1207 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1208 SBI_IS_RESIZEFS, /* resizefs is in process */
1217 UMOUNT_DISCARD_TIMEOUT,
1231 BGGC_MODE_ON, /* background gc is on */
1232 BGGC_MODE_OFF, /* background gc is off */
1234 * background gc is on, migrating blocks
1235 * like foreground gc
1240 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1241 FS_MODE_LFS, /* use lfs allocation only */
1245 WHINT_MODE_OFF, /* not pass down write hints */
1246 WHINT_MODE_USER, /* try to pass down hints given by users */
1247 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1251 ALLOC_MODE_DEFAULT, /* stay default */
1252 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1256 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1257 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1258 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1263 * automatically compress compression
1267 * automatical compression is disabled.
1268 * user can control the file compression
1274 * this value is set in page as a private data which indicate that
1275 * the page is atomically written, and it is in inmem_pages list.
1277 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1278 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1280 #define IS_ATOMIC_WRITTEN_PAGE(page) \
1281 (page_private(page) == ATOMIC_WRITTEN_PAGE)
1282 #define IS_DUMMY_WRITTEN_PAGE(page) \
1283 (page_private(page) == DUMMY_WRITTEN_PAGE)
1285 #ifdef CONFIG_F2FS_IO_TRACE
1286 #define IS_IO_TRACED_PAGE(page) \
1287 (page_private(page) > 0 && \
1288 page_private(page) < (unsigned long)PID_MAX_LIMIT)
1290 #define IS_IO_TRACED_PAGE(page) (0)
1293 /* For compression */
1294 enum compress_algorithm_type {
1302 enum compress_flag {
1307 #define COMPRESS_DATA_RESERVED_SIZE 4
1308 struct compress_data {
1309 __le32 clen; /* compressed data size */
1310 __le32 chksum; /* compressed data chksum */
1311 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1312 u8 cdata[]; /* compressed data */
1315 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1317 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1319 /* compress context */
1320 struct compress_ctx {
1321 struct inode *inode; /* inode the context belong to */
1322 pgoff_t cluster_idx; /* cluster index number */
1323 unsigned int cluster_size; /* page count in cluster */
1324 unsigned int log_cluster_size; /* log of cluster size */
1325 struct page **rpages; /* pages store raw data in cluster */
1326 unsigned int nr_rpages; /* total page number in rpages */
1327 struct page **cpages; /* pages store compressed data in cluster */
1328 unsigned int nr_cpages; /* total page number in cpages */
1329 void *rbuf; /* virtual mapped address on rpages */
1330 struct compress_data *cbuf; /* virtual mapped address on cpages */
1331 size_t rlen; /* valid data length in rbuf */
1332 size_t clen; /* valid data length in cbuf */
1333 void *private; /* payload buffer for specified compression algorithm */
1334 void *private2; /* extra payload buffer */
1337 /* compress context for write IO path */
1338 struct compress_io_ctx {
1339 u32 magic; /* magic number to indicate page is compressed */
1340 struct inode *inode; /* inode the context belong to */
1341 struct page **rpages; /* pages store raw data in cluster */
1342 unsigned int nr_rpages; /* total page number in rpages */
1343 atomic_t pending_pages; /* in-flight compressed page count */
1346 /* decompress io context for read IO path */
1347 struct decompress_io_ctx {
1348 u32 magic; /* magic number to indicate page is compressed */
1349 struct inode *inode; /* inode the context belong to */
1350 pgoff_t cluster_idx; /* cluster index number */
1351 unsigned int cluster_size; /* page count in cluster */
1352 unsigned int log_cluster_size; /* log of cluster size */
1353 struct page **rpages; /* pages store raw data in cluster */
1354 unsigned int nr_rpages; /* total page number in rpages */
1355 struct page **cpages; /* pages store compressed data in cluster */
1356 unsigned int nr_cpages; /* total page number in cpages */
1357 struct page **tpages; /* temp pages to pad holes in cluster */
1358 void *rbuf; /* virtual mapped address on rpages */
1359 struct compress_data *cbuf; /* virtual mapped address on cpages */
1360 size_t rlen; /* valid data length in rbuf */
1361 size_t clen; /* valid data length in cbuf */
1362 atomic_t pending_pages; /* in-flight compressed page count */
1363 atomic_t verity_pages; /* in-flight page count for verity */
1364 bool failed; /* indicate IO error during decompression */
1365 void *private; /* payload buffer for specified decompression algorithm */
1366 void *private2; /* extra payload buffer */
1369 #define NULL_CLUSTER ((unsigned int)(~0))
1370 #define MIN_COMPRESS_LOG_SIZE 2
1371 #define MAX_COMPRESS_LOG_SIZE 8
1372 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1374 struct f2fs_sb_info {
1375 struct super_block *sb; /* pointer to VFS super block */
1376 struct proc_dir_entry *s_proc; /* proc entry */
1377 struct f2fs_super_block *raw_super; /* raw super block pointer */
1378 struct rw_semaphore sb_lock; /* lock for raw super block */
1379 int valid_super_block; /* valid super block no */
1380 unsigned long s_flag; /* flags for sbi */
1381 struct mutex writepages; /* mutex for writepages() */
1383 #ifdef CONFIG_BLK_DEV_ZONED
1384 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1385 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1388 /* for node-related operations */
1389 struct f2fs_nm_info *nm_info; /* node manager */
1390 struct inode *node_inode; /* cache node blocks */
1392 /* for segment-related operations */
1393 struct f2fs_sm_info *sm_info; /* segment manager */
1395 /* for bio operations */
1396 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1397 /* keep migration IO order for LFS mode */
1398 struct rw_semaphore io_order_lock;
1399 mempool_t *write_io_dummy; /* Dummy pages */
1401 /* for checkpoint */
1402 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1403 int cur_cp_pack; /* remain current cp pack */
1404 spinlock_t cp_lock; /* for flag in ckpt */
1405 struct inode *meta_inode; /* cache meta blocks */
1406 struct mutex cp_mutex; /* checkpoint procedure lock */
1407 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1408 struct rw_semaphore node_write; /* locking node writes */
1409 struct rw_semaphore node_change; /* locking node change */
1410 wait_queue_head_t cp_wait;
1411 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1412 long interval_time[MAX_TIME]; /* to store thresholds */
1414 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1416 spinlock_t fsync_node_lock; /* for node entry lock */
1417 struct list_head fsync_node_list; /* node list head */
1418 unsigned int fsync_seg_id; /* sequence id */
1419 unsigned int fsync_node_num; /* number of node entries */
1421 /* for orphan inode, use 0'th array */
1422 unsigned int max_orphans; /* max orphan inodes */
1424 /* for inode management */
1425 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1426 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1427 struct mutex flush_lock; /* for flush exclusion */
1429 /* for extent tree cache */
1430 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1431 struct mutex extent_tree_lock; /* locking extent radix tree */
1432 struct list_head extent_list; /* lru list for shrinker */
1433 spinlock_t extent_lock; /* locking extent lru list */
1434 atomic_t total_ext_tree; /* extent tree count */
1435 struct list_head zombie_list; /* extent zombie tree list */
1436 atomic_t total_zombie_tree; /* extent zombie tree count */
1437 atomic_t total_ext_node; /* extent info count */
1439 /* basic filesystem units */
1440 unsigned int log_sectors_per_block; /* log2 sectors per block */
1441 unsigned int log_blocksize; /* log2 block size */
1442 unsigned int blocksize; /* block size */
1443 unsigned int root_ino_num; /* root inode number*/
1444 unsigned int node_ino_num; /* node inode number*/
1445 unsigned int meta_ino_num; /* meta inode number*/
1446 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1447 unsigned int blocks_per_seg; /* blocks per segment */
1448 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1449 unsigned int segs_per_sec; /* segments per section */
1450 unsigned int secs_per_zone; /* sections per zone */
1451 unsigned int total_sections; /* total section count */
1452 unsigned int total_node_count; /* total node block count */
1453 unsigned int total_valid_node_count; /* valid node block count */
1454 loff_t max_file_blocks; /* max block index of file */
1455 int dir_level; /* directory level */
1456 int readdir_ra; /* readahead inode in readdir */
1458 block_t user_block_count; /* # of user blocks */
1459 block_t total_valid_block_count; /* # of valid blocks */
1460 block_t discard_blks; /* discard command candidats */
1461 block_t last_valid_block_count; /* for recovery */
1462 block_t reserved_blocks; /* configurable reserved blocks */
1463 block_t current_reserved_blocks; /* current reserved blocks */
1465 /* Additional tracking for no checkpoint mode */
1466 block_t unusable_block_count; /* # of blocks saved by last cp */
1468 unsigned int nquota_files; /* # of quota sysfile */
1469 struct rw_semaphore quota_sem; /* blocking cp for flags */
1471 /* # of pages, see count_type */
1472 atomic_t nr_pages[NR_COUNT_TYPE];
1473 /* # of allocated blocks */
1474 struct percpu_counter alloc_valid_block_count;
1476 /* writeback control */
1477 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1479 /* valid inode count */
1480 struct percpu_counter total_valid_inode_count;
1482 struct f2fs_mount_info mount_opt; /* mount options */
1484 /* for cleaning operations */
1485 struct rw_semaphore gc_lock; /*
1486 * semaphore for GC, avoid
1487 * race between GC and GC or CP
1489 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1490 struct atgc_management am; /* atgc management */
1491 unsigned int cur_victim_sec; /* current victim section num */
1492 unsigned int gc_mode; /* current GC state */
1493 unsigned int next_victim_seg[2]; /* next segment in victim section */
1495 /* for skip statistic */
1496 unsigned int atomic_files; /* # of opened atomic file */
1497 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1498 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1500 /* threshold for gc trials on pinned files */
1501 u64 gc_pin_file_threshold;
1502 struct rw_semaphore pin_sem;
1504 /* maximum # of trials to find a victim segment for SSR and GC */
1505 unsigned int max_victim_search;
1506 /* migration granularity of garbage collection, unit: segment */
1507 unsigned int migration_granularity;
1510 * for stat information.
1511 * one is for the LFS mode, and the other is for the SSR mode.
1513 #ifdef CONFIG_F2FS_STAT_FS
1514 struct f2fs_stat_info *stat_info; /* FS status information */
1515 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1516 unsigned int segment_count[2]; /* # of allocated segments */
1517 unsigned int block_count[2]; /* # of allocated blocks */
1518 atomic_t inplace_count; /* # of inplace update */
1519 atomic64_t total_hit_ext; /* # of lookup extent cache */
1520 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1521 atomic64_t read_hit_largest; /* # of hit largest extent node */
1522 atomic64_t read_hit_cached; /* # of hit cached extent node */
1523 atomic_t inline_xattr; /* # of inline_xattr inodes */
1524 atomic_t inline_inode; /* # of inline_data inodes */
1525 atomic_t inline_dir; /* # of inline_dentry inodes */
1526 atomic_t compr_inode; /* # of compressed inodes */
1527 atomic64_t compr_blocks; /* # of compressed blocks */
1528 atomic_t vw_cnt; /* # of volatile writes */
1529 atomic_t max_aw_cnt; /* max # of atomic writes */
1530 atomic_t max_vw_cnt; /* max # of volatile writes */
1531 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1532 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1533 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1535 spinlock_t stat_lock; /* lock for stat operations */
1537 /* For app/fs IO statistics */
1538 spinlock_t iostat_lock;
1539 unsigned long long rw_iostat[NR_IO_TYPE];
1540 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1542 unsigned long iostat_next_period;
1543 unsigned int iostat_period_ms;
1545 /* to attach REQ_META|REQ_FUA flags */
1546 unsigned int data_io_flag;
1547 unsigned int node_io_flag;
1549 /* For sysfs suppport */
1550 struct kobject s_kobj;
1551 struct completion s_kobj_unregister;
1553 /* For shrinker support */
1554 struct list_head s_list;
1555 int s_ndevs; /* number of devices */
1556 struct f2fs_dev_info *devs; /* for device list */
1557 unsigned int dirty_device; /* for checkpoint data flush */
1558 spinlock_t dev_lock; /* protect dirty_device */
1559 struct mutex umount_mutex;
1560 unsigned int shrinker_run_no;
1562 /* For write statistics */
1563 u64 sectors_written_start;
1566 /* Reference to checksum algorithm driver via cryptoapi */
1567 struct crypto_shash *s_chksum_driver;
1569 /* Precomputed FS UUID checksum for seeding other checksums */
1570 __u32 s_chksum_seed;
1572 struct workqueue_struct *post_read_wq; /* post read workqueue */
1574 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1575 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1577 #ifdef CONFIG_F2FS_FS_COMPRESSION
1578 struct kmem_cache *page_array_slab; /* page array entry */
1579 unsigned int page_array_slab_size; /* default page array slab size */
1583 struct f2fs_private_dio {
1584 struct inode *inode;
1586 bio_end_io_t *orig_end_io;
1590 #ifdef CONFIG_F2FS_FAULT_INJECTION
1591 #define f2fs_show_injection_info(sbi, type) \
1592 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1593 KERN_INFO, sbi->sb->s_id, \
1594 f2fs_fault_name[type], \
1595 __func__, __builtin_return_address(0))
1596 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1598 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1600 if (!ffi->inject_rate)
1603 if (!IS_FAULT_SET(ffi, type))
1606 atomic_inc(&ffi->inject_ops);
1607 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1608 atomic_set(&ffi->inject_ops, 0);
1614 #define f2fs_show_injection_info(sbi, type) do { } while (0)
1615 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1622 * Test if the mounted volume is a multi-device volume.
1623 * - For a single regular disk volume, sbi->s_ndevs is 0.
1624 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1625 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1627 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1629 return sbi->s_ndevs > 1;
1632 /* For write statistics. Suppose sector size is 512 bytes,
1633 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1635 #define BD_PART_WRITTEN(s) \
1636 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1637 (s)->sectors_written_start) >> 1)
1639 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1641 unsigned long now = jiffies;
1643 sbi->last_time[type] = now;
1645 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1646 if (type == REQ_TIME) {
1647 sbi->last_time[DISCARD_TIME] = now;
1648 sbi->last_time[GC_TIME] = now;
1652 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1654 unsigned long interval = sbi->interval_time[type] * HZ;
1656 return time_after(jiffies, sbi->last_time[type] + interval);
1659 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1662 unsigned long interval = sbi->interval_time[type] * HZ;
1663 unsigned int wait_ms = 0;
1666 delta = (sbi->last_time[type] + interval) - jiffies;
1668 wait_ms = jiffies_to_msecs(delta);
1676 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1677 const void *address, unsigned int length)
1680 struct shash_desc shash;
1685 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1687 desc.shash.tfm = sbi->s_chksum_driver;
1688 *(u32 *)desc.ctx = crc;
1690 err = crypto_shash_update(&desc.shash, address, length);
1693 return *(u32 *)desc.ctx;
1696 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1697 unsigned int length)
1699 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1702 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1703 void *buf, size_t buf_size)
1705 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1708 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1709 const void *address, unsigned int length)
1711 return __f2fs_crc32(sbi, crc, address, length);
1714 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1716 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1719 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1721 return sb->s_fs_info;
1724 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1726 return F2FS_SB(inode->i_sb);
1729 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1731 return F2FS_I_SB(mapping->host);
1734 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1736 return F2FS_M_SB(page_file_mapping(page));
1739 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1741 return (struct f2fs_super_block *)(sbi->raw_super);
1744 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1746 return (struct f2fs_checkpoint *)(sbi->ckpt);
1749 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1751 return (struct f2fs_node *)page_address(page);
1754 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1756 return &((struct f2fs_node *)page_address(page))->i;
1759 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1761 return (struct f2fs_nm_info *)(sbi->nm_info);
1764 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1766 return (struct f2fs_sm_info *)(sbi->sm_info);
1769 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1771 return (struct sit_info *)(SM_I(sbi)->sit_info);
1774 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1776 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1779 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1781 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1784 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1786 return sbi->meta_inode->i_mapping;
1789 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1791 return sbi->node_inode->i_mapping;
1794 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1796 return test_bit(type, &sbi->s_flag);
1799 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1801 set_bit(type, &sbi->s_flag);
1804 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1806 clear_bit(type, &sbi->s_flag);
1809 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1811 return le64_to_cpu(cp->checkpoint_ver);
1814 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1816 if (type < F2FS_MAX_QUOTAS)
1817 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1821 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1823 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1824 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1827 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1829 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1831 return ckpt_flags & f;
1834 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1836 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1839 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1841 unsigned int ckpt_flags;
1843 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1845 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1848 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1850 unsigned long flags;
1852 spin_lock_irqsave(&sbi->cp_lock, flags);
1853 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1854 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1857 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1859 unsigned int ckpt_flags;
1861 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1863 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1866 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1868 unsigned long flags;
1870 spin_lock_irqsave(&sbi->cp_lock, flags);
1871 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1872 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1875 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1877 unsigned long flags;
1878 unsigned char *nat_bits;
1881 * In order to re-enable nat_bits we need to call fsck.f2fs by
1882 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1883 * so let's rely on regular fsck or unclean shutdown.
1887 spin_lock_irqsave(&sbi->cp_lock, flags);
1888 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1889 nat_bits = NM_I(sbi)->nat_bits;
1890 NM_I(sbi)->nat_bits = NULL;
1892 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1897 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1898 struct cp_control *cpc)
1900 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1902 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1905 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1907 down_read(&sbi->cp_rwsem);
1910 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1912 return down_read_trylock(&sbi->cp_rwsem);
1915 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1917 up_read(&sbi->cp_rwsem);
1920 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1922 down_write(&sbi->cp_rwsem);
1925 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1927 up_write(&sbi->cp_rwsem);
1930 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1932 int reason = CP_SYNC;
1934 if (test_opt(sbi, FASTBOOT))
1935 reason = CP_FASTBOOT;
1936 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1941 static inline bool __remain_node_summaries(int reason)
1943 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1946 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1948 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1949 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1953 * Check whether the inode has blocks or not
1955 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1957 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1959 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1962 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1964 return ofs == XATTR_NODE_OFFSET;
1967 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1968 struct inode *inode, bool cap)
1972 if (!test_opt(sbi, RESERVE_ROOT))
1974 if (IS_NOQUOTA(inode))
1976 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1978 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1979 in_group_p(F2FS_OPTION(sbi).s_resgid))
1981 if (cap && capable(CAP_SYS_RESOURCE))
1986 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1987 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1988 struct inode *inode, blkcnt_t *count)
1990 blkcnt_t diff = 0, release = 0;
1991 block_t avail_user_block_count;
1994 ret = dquot_reserve_block(inode, *count);
1998 if (time_to_inject(sbi, FAULT_BLOCK)) {
1999 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2005 * let's increase this in prior to actual block count change in order
2006 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2008 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2010 spin_lock(&sbi->stat_lock);
2011 sbi->total_valid_block_count += (block_t)(*count);
2012 avail_user_block_count = sbi->user_block_count -
2013 sbi->current_reserved_blocks;
2015 if (!__allow_reserved_blocks(sbi, inode, true))
2016 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2018 if (F2FS_IO_ALIGNED(sbi))
2019 avail_user_block_count -= sbi->blocks_per_seg *
2020 SM_I(sbi)->additional_reserved_segments;
2022 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2023 if (avail_user_block_count > sbi->unusable_block_count)
2024 avail_user_block_count -= sbi->unusable_block_count;
2026 avail_user_block_count = 0;
2028 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2029 diff = sbi->total_valid_block_count - avail_user_block_count;
2034 sbi->total_valid_block_count -= diff;
2036 spin_unlock(&sbi->stat_lock);
2040 spin_unlock(&sbi->stat_lock);
2042 if (unlikely(release)) {
2043 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2044 dquot_release_reservation_block(inode, release);
2046 f2fs_i_blocks_write(inode, *count, true, true);
2050 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2052 dquot_release_reservation_block(inode, release);
2057 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2059 #define f2fs_err(sbi, fmt, ...) \
2060 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2061 #define f2fs_warn(sbi, fmt, ...) \
2062 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2063 #define f2fs_notice(sbi, fmt, ...) \
2064 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2065 #define f2fs_info(sbi, fmt, ...) \
2066 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2067 #define f2fs_debug(sbi, fmt, ...) \
2068 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2070 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2071 struct inode *inode,
2074 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2076 spin_lock(&sbi->stat_lock);
2077 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2078 sbi->total_valid_block_count -= (block_t)count;
2079 if (sbi->reserved_blocks &&
2080 sbi->current_reserved_blocks < sbi->reserved_blocks)
2081 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2082 sbi->current_reserved_blocks + count);
2083 spin_unlock(&sbi->stat_lock);
2084 if (unlikely(inode->i_blocks < sectors)) {
2085 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2087 (unsigned long long)inode->i_blocks,
2088 (unsigned long long)sectors);
2089 set_sbi_flag(sbi, SBI_NEED_FSCK);
2092 f2fs_i_blocks_write(inode, count, false, true);
2095 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2097 atomic_inc(&sbi->nr_pages[count_type]);
2099 if (count_type == F2FS_DIRTY_DENTS ||
2100 count_type == F2FS_DIRTY_NODES ||
2101 count_type == F2FS_DIRTY_META ||
2102 count_type == F2FS_DIRTY_QDATA ||
2103 count_type == F2FS_DIRTY_IMETA)
2104 set_sbi_flag(sbi, SBI_IS_DIRTY);
2107 static inline void inode_inc_dirty_pages(struct inode *inode)
2109 atomic_inc(&F2FS_I(inode)->dirty_pages);
2110 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2111 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2112 if (IS_NOQUOTA(inode))
2113 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2116 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2118 atomic_dec(&sbi->nr_pages[count_type]);
2121 static inline void inode_dec_dirty_pages(struct inode *inode)
2123 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2124 !S_ISLNK(inode->i_mode))
2127 atomic_dec(&F2FS_I(inode)->dirty_pages);
2128 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2129 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2130 if (IS_NOQUOTA(inode))
2131 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2134 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2136 return atomic_read(&sbi->nr_pages[count_type]);
2139 static inline int get_dirty_pages(struct inode *inode)
2141 return atomic_read(&F2FS_I(inode)->dirty_pages);
2144 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2146 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2147 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2148 sbi->log_blocks_per_seg;
2150 return segs / sbi->segs_per_sec;
2153 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2155 return sbi->total_valid_block_count;
2158 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2160 return sbi->discard_blks;
2163 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2165 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2167 /* return NAT or SIT bitmap */
2168 if (flag == NAT_BITMAP)
2169 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2170 else if (flag == SIT_BITMAP)
2171 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2176 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2178 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2181 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2183 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2186 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2187 offset = (flag == SIT_BITMAP) ?
2188 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2190 * if large_nat_bitmap feature is enabled, leave checksum
2191 * protection for all nat/sit bitmaps.
2193 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2196 if (__cp_payload(sbi) > 0) {
2197 if (flag == NAT_BITMAP)
2198 return &ckpt->sit_nat_version_bitmap;
2200 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2202 offset = (flag == NAT_BITMAP) ?
2203 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2204 return &ckpt->sit_nat_version_bitmap + offset;
2208 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2210 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2212 if (sbi->cur_cp_pack == 2)
2213 start_addr += sbi->blocks_per_seg;
2217 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2219 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2221 if (sbi->cur_cp_pack == 1)
2222 start_addr += sbi->blocks_per_seg;
2226 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2228 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2231 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2233 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2236 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2237 struct inode *inode, bool is_inode)
2239 block_t valid_block_count;
2240 unsigned int valid_node_count, user_block_count;
2245 err = dquot_alloc_inode(inode);
2250 err = dquot_reserve_block(inode, 1);
2255 if (time_to_inject(sbi, FAULT_BLOCK)) {
2256 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2260 spin_lock(&sbi->stat_lock);
2262 valid_block_count = sbi->total_valid_block_count +
2263 sbi->current_reserved_blocks + 1;
2265 if (!__allow_reserved_blocks(sbi, inode, false))
2266 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2268 if (F2FS_IO_ALIGNED(sbi))
2269 valid_block_count += sbi->blocks_per_seg *
2270 SM_I(sbi)->additional_reserved_segments;
2272 user_block_count = sbi->user_block_count;
2273 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2274 user_block_count -= sbi->unusable_block_count;
2276 if (unlikely(valid_block_count > user_block_count)) {
2277 spin_unlock(&sbi->stat_lock);
2281 valid_node_count = sbi->total_valid_node_count + 1;
2282 if (unlikely(valid_node_count > sbi->total_node_count)) {
2283 spin_unlock(&sbi->stat_lock);
2287 sbi->total_valid_node_count++;
2288 sbi->total_valid_block_count++;
2289 spin_unlock(&sbi->stat_lock);
2293 f2fs_mark_inode_dirty_sync(inode, true);
2295 f2fs_i_blocks_write(inode, 1, true, true);
2298 percpu_counter_inc(&sbi->alloc_valid_block_count);
2304 dquot_free_inode(inode);
2306 dquot_release_reservation_block(inode, 1);
2311 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2312 struct inode *inode, bool is_inode)
2314 spin_lock(&sbi->stat_lock);
2316 if (unlikely(!sbi->total_valid_block_count ||
2317 !sbi->total_valid_node_count)) {
2318 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2319 sbi->total_valid_block_count,
2320 sbi->total_valid_node_count);
2321 set_sbi_flag(sbi, SBI_NEED_FSCK);
2323 sbi->total_valid_block_count--;
2324 sbi->total_valid_node_count--;
2327 if (sbi->reserved_blocks &&
2328 sbi->current_reserved_blocks < sbi->reserved_blocks)
2329 sbi->current_reserved_blocks++;
2331 spin_unlock(&sbi->stat_lock);
2334 dquot_free_inode(inode);
2336 if (unlikely(inode->i_blocks == 0)) {
2337 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2339 (unsigned long long)inode->i_blocks);
2340 set_sbi_flag(sbi, SBI_NEED_FSCK);
2343 f2fs_i_blocks_write(inode, 1, false, true);
2347 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2349 return sbi->total_valid_node_count;
2352 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2354 percpu_counter_inc(&sbi->total_valid_inode_count);
2357 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2359 percpu_counter_dec(&sbi->total_valid_inode_count);
2362 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2364 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2367 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2368 pgoff_t index, bool for_write)
2372 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2374 page = find_get_page_flags(mapping, index,
2375 FGP_LOCK | FGP_ACCESSED);
2377 page = find_lock_page(mapping, index);
2381 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2382 f2fs_show_injection_info(F2FS_M_SB(mapping),
2389 return grab_cache_page(mapping, index);
2390 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2393 static inline struct page *f2fs_pagecache_get_page(
2394 struct address_space *mapping, pgoff_t index,
2395 int fgp_flags, gfp_t gfp_mask)
2397 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2398 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2402 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2405 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2407 char *src_kaddr = kmap(src);
2408 char *dst_kaddr = kmap(dst);
2410 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2415 static inline void f2fs_put_page(struct page *page, int unlock)
2421 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2427 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2430 f2fs_put_page(dn->node_page, 1);
2431 if (dn->inode_page && dn->node_page != dn->inode_page)
2432 f2fs_put_page(dn->inode_page, 0);
2433 dn->node_page = NULL;
2434 dn->inode_page = NULL;
2437 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2440 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2443 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2448 entry = kmem_cache_alloc(cachep, flags);
2450 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2454 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2456 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2457 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2458 get_pages(sbi, F2FS_WB_CP_DATA) ||
2459 get_pages(sbi, F2FS_DIO_READ) ||
2460 get_pages(sbi, F2FS_DIO_WRITE))
2463 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2464 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2467 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2468 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2473 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2475 if (sbi->gc_mode == GC_URGENT_HIGH)
2478 if (is_inflight_io(sbi, type))
2481 if (sbi->gc_mode == GC_URGENT_LOW &&
2482 (type == DISCARD_TIME || type == GC_TIME))
2485 return f2fs_time_over(sbi, type);
2488 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2489 unsigned long index, void *item)
2491 while (radix_tree_insert(root, index, item))
2495 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2497 static inline bool IS_INODE(struct page *page)
2499 struct f2fs_node *p = F2FS_NODE(page);
2501 return RAW_IS_INODE(p);
2504 static inline int offset_in_addr(struct f2fs_inode *i)
2506 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2507 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2510 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2512 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2515 static inline int f2fs_has_extra_attr(struct inode *inode);
2516 static inline block_t data_blkaddr(struct inode *inode,
2517 struct page *node_page, unsigned int offset)
2519 struct f2fs_node *raw_node;
2522 bool is_inode = IS_INODE(node_page);
2524 raw_node = F2FS_NODE(node_page);
2528 /* from GC path only */
2529 base = offset_in_addr(&raw_node->i);
2530 else if (f2fs_has_extra_attr(inode))
2531 base = get_extra_isize(inode);
2534 addr_array = blkaddr_in_node(raw_node);
2535 return le32_to_cpu(addr_array[base + offset]);
2538 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2540 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2543 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2548 mask = 1 << (7 - (nr & 0x07));
2549 return mask & *addr;
2552 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2557 mask = 1 << (7 - (nr & 0x07));
2561 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2566 mask = 1 << (7 - (nr & 0x07));
2570 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2576 mask = 1 << (7 - (nr & 0x07));
2582 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2588 mask = 1 << (7 - (nr & 0x07));
2594 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2599 mask = 1 << (7 - (nr & 0x07));
2604 * On-disk inode flags (f2fs_inode::i_flags)
2606 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2607 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2608 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2609 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2610 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2611 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2612 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2613 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2614 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2615 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2616 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2618 /* Flags that should be inherited by new inodes from their parent. */
2619 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2620 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2621 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2623 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2624 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2627 /* Flags that are appropriate for non-directories/regular files. */
2628 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2630 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2634 else if (S_ISREG(mode))
2635 return flags & F2FS_REG_FLMASK;
2637 return flags & F2FS_OTHER_FLMASK;
2640 static inline void __mark_inode_dirty_flag(struct inode *inode,
2644 case FI_INLINE_XATTR:
2645 case FI_INLINE_DATA:
2646 case FI_INLINE_DENTRY:
2652 case FI_INLINE_DOTS:
2654 case FI_COMPRESS_RELEASED:
2655 f2fs_mark_inode_dirty_sync(inode, true);
2659 static inline void set_inode_flag(struct inode *inode, int flag)
2661 set_bit(flag, F2FS_I(inode)->flags);
2662 __mark_inode_dirty_flag(inode, flag, true);
2665 static inline int is_inode_flag_set(struct inode *inode, int flag)
2667 return test_bit(flag, F2FS_I(inode)->flags);
2670 static inline void clear_inode_flag(struct inode *inode, int flag)
2672 clear_bit(flag, F2FS_I(inode)->flags);
2673 __mark_inode_dirty_flag(inode, flag, false);
2676 static inline bool f2fs_verity_in_progress(struct inode *inode)
2678 return IS_ENABLED(CONFIG_FS_VERITY) &&
2679 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2682 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2684 F2FS_I(inode)->i_acl_mode = mode;
2685 set_inode_flag(inode, FI_ACL_MODE);
2686 f2fs_mark_inode_dirty_sync(inode, false);
2689 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2695 f2fs_mark_inode_dirty_sync(inode, true);
2698 static inline void f2fs_i_blocks_write(struct inode *inode,
2699 block_t diff, bool add, bool claim)
2701 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2702 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2704 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2707 dquot_claim_block(inode, diff);
2709 dquot_alloc_block_nofail(inode, diff);
2711 dquot_free_block(inode, diff);
2714 f2fs_mark_inode_dirty_sync(inode, true);
2715 if (clean || recover)
2716 set_inode_flag(inode, FI_AUTO_RECOVER);
2719 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2721 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2722 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2724 if (i_size_read(inode) == i_size)
2727 i_size_write(inode, i_size);
2728 f2fs_mark_inode_dirty_sync(inode, true);
2729 if (clean || recover)
2730 set_inode_flag(inode, FI_AUTO_RECOVER);
2733 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2735 F2FS_I(inode)->i_current_depth = depth;
2736 f2fs_mark_inode_dirty_sync(inode, true);
2739 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2742 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2743 f2fs_mark_inode_dirty_sync(inode, true);
2746 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2748 F2FS_I(inode)->i_xattr_nid = xnid;
2749 f2fs_mark_inode_dirty_sync(inode, true);
2752 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2754 F2FS_I(inode)->i_pino = pino;
2755 f2fs_mark_inode_dirty_sync(inode, true);
2758 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2760 struct f2fs_inode_info *fi = F2FS_I(inode);
2762 if (ri->i_inline & F2FS_INLINE_XATTR)
2763 set_bit(FI_INLINE_XATTR, fi->flags);
2764 if (ri->i_inline & F2FS_INLINE_DATA)
2765 set_bit(FI_INLINE_DATA, fi->flags);
2766 if (ri->i_inline & F2FS_INLINE_DENTRY)
2767 set_bit(FI_INLINE_DENTRY, fi->flags);
2768 if (ri->i_inline & F2FS_DATA_EXIST)
2769 set_bit(FI_DATA_EXIST, fi->flags);
2770 if (ri->i_inline & F2FS_INLINE_DOTS)
2771 set_bit(FI_INLINE_DOTS, fi->flags);
2772 if (ri->i_inline & F2FS_EXTRA_ATTR)
2773 set_bit(FI_EXTRA_ATTR, fi->flags);
2774 if (ri->i_inline & F2FS_PIN_FILE)
2775 set_bit(FI_PIN_FILE, fi->flags);
2776 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
2777 set_bit(FI_COMPRESS_RELEASED, fi->flags);
2780 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2784 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2785 ri->i_inline |= F2FS_INLINE_XATTR;
2786 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2787 ri->i_inline |= F2FS_INLINE_DATA;
2788 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2789 ri->i_inline |= F2FS_INLINE_DENTRY;
2790 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2791 ri->i_inline |= F2FS_DATA_EXIST;
2792 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2793 ri->i_inline |= F2FS_INLINE_DOTS;
2794 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2795 ri->i_inline |= F2FS_EXTRA_ATTR;
2796 if (is_inode_flag_set(inode, FI_PIN_FILE))
2797 ri->i_inline |= F2FS_PIN_FILE;
2798 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
2799 ri->i_inline |= F2FS_COMPRESS_RELEASED;
2802 static inline int f2fs_has_extra_attr(struct inode *inode)
2804 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2807 static inline int f2fs_has_inline_xattr(struct inode *inode)
2809 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2812 static inline int f2fs_compressed_file(struct inode *inode)
2814 return S_ISREG(inode->i_mode) &&
2815 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2818 static inline bool f2fs_need_compress_data(struct inode *inode)
2820 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
2822 if (!f2fs_compressed_file(inode))
2825 if (compress_mode == COMPR_MODE_FS)
2827 else if (compress_mode == COMPR_MODE_USER &&
2828 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
2834 static inline unsigned int addrs_per_inode(struct inode *inode)
2836 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2837 get_inline_xattr_addrs(inode);
2839 if (!f2fs_compressed_file(inode))
2841 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2844 static inline unsigned int addrs_per_block(struct inode *inode)
2846 if (!f2fs_compressed_file(inode))
2847 return DEF_ADDRS_PER_BLOCK;
2848 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2851 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2853 struct f2fs_inode *ri = F2FS_INODE(page);
2855 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2856 get_inline_xattr_addrs(inode)]);
2859 static inline int inline_xattr_size(struct inode *inode)
2861 if (f2fs_has_inline_xattr(inode))
2862 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2866 static inline int f2fs_has_inline_data(struct inode *inode)
2868 return is_inode_flag_set(inode, FI_INLINE_DATA);
2871 static inline int f2fs_exist_data(struct inode *inode)
2873 return is_inode_flag_set(inode, FI_DATA_EXIST);
2876 static inline int f2fs_has_inline_dots(struct inode *inode)
2878 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2881 static inline int f2fs_is_mmap_file(struct inode *inode)
2883 return is_inode_flag_set(inode, FI_MMAP_FILE);
2886 static inline bool f2fs_is_pinned_file(struct inode *inode)
2888 return is_inode_flag_set(inode, FI_PIN_FILE);
2891 static inline bool f2fs_is_atomic_file(struct inode *inode)
2893 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2896 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2898 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2901 static inline bool f2fs_is_volatile_file(struct inode *inode)
2903 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2906 static inline bool f2fs_is_first_block_written(struct inode *inode)
2908 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2911 static inline bool f2fs_is_drop_cache(struct inode *inode)
2913 return is_inode_flag_set(inode, FI_DROP_CACHE);
2916 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2918 struct f2fs_inode *ri = F2FS_INODE(page);
2919 int extra_size = get_extra_isize(inode);
2921 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2924 static inline int f2fs_has_inline_dentry(struct inode *inode)
2926 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2929 static inline int is_file(struct inode *inode, int type)
2931 return F2FS_I(inode)->i_advise & type;
2934 static inline void set_file(struct inode *inode, int type)
2936 F2FS_I(inode)->i_advise |= type;
2937 f2fs_mark_inode_dirty_sync(inode, true);
2940 static inline void clear_file(struct inode *inode, int type)
2942 F2FS_I(inode)->i_advise &= ~type;
2943 f2fs_mark_inode_dirty_sync(inode, true);
2946 static inline bool f2fs_is_time_consistent(struct inode *inode)
2948 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2950 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2952 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2954 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2955 &F2FS_I(inode)->i_crtime))
2960 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2965 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2967 spin_lock(&sbi->inode_lock[DIRTY_META]);
2968 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2969 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2972 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2973 file_keep_isize(inode) ||
2974 i_size_read(inode) & ~PAGE_MASK)
2977 if (!f2fs_is_time_consistent(inode))
2980 spin_lock(&F2FS_I(inode)->i_size_lock);
2981 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2982 spin_unlock(&F2FS_I(inode)->i_size_lock);
2987 static inline bool f2fs_readonly(struct super_block *sb)
2989 return sb_rdonly(sb);
2992 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2994 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2997 static inline bool is_dot_dotdot(const u8 *name, size_t len)
2999 if (len == 1 && name[0] == '.')
3002 if (len == 2 && name[0] == '.' && name[1] == '.')
3008 static inline bool f2fs_may_extent_tree(struct inode *inode)
3010 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3012 if (!test_opt(sbi, EXTENT_CACHE) ||
3013 is_inode_flag_set(inode, FI_NO_EXTENT) ||
3014 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
3018 * for recovered files during mount do not create extents
3019 * if shrinker is not registered.
3021 if (list_empty(&sbi->s_list))
3024 return S_ISREG(inode->i_mode);
3027 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3028 size_t size, gfp_t flags)
3030 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3031 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3035 return kmalloc(size, flags);
3038 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3039 size_t size, gfp_t flags)
3041 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3044 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3045 size_t size, gfp_t flags)
3047 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3048 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3052 return kvmalloc(size, flags);
3055 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3056 size_t size, gfp_t flags)
3058 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3061 static inline int get_extra_isize(struct inode *inode)
3063 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3066 static inline int get_inline_xattr_addrs(struct inode *inode)
3068 return F2FS_I(inode)->i_inline_xattr_size;
3071 #define f2fs_get_inode_mode(i) \
3072 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3073 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3075 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3076 (offsetof(struct f2fs_inode, i_extra_end) - \
3077 offsetof(struct f2fs_inode, i_extra_isize)) \
3079 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3080 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3081 ((offsetof(typeof(*(f2fs_inode)), field) + \
3082 sizeof((f2fs_inode)->field)) \
3083 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3085 #define DEFAULT_IOSTAT_PERIOD_MS 3000
3086 #define MIN_IOSTAT_PERIOD_MS 100
3087 /* maximum period of iostat tracing is 1 day */
3088 #define MAX_IOSTAT_PERIOD_MS 8640000
3090 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3094 spin_lock(&sbi->iostat_lock);
3095 for (i = 0; i < NR_IO_TYPE; i++) {
3096 sbi->rw_iostat[i] = 0;
3097 sbi->prev_rw_iostat[i] = 0;
3099 spin_unlock(&sbi->iostat_lock);
3102 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3104 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3105 enum iostat_type type, unsigned long long io_bytes)
3107 if (!sbi->iostat_enable)
3109 spin_lock(&sbi->iostat_lock);
3110 sbi->rw_iostat[type] += io_bytes;
3112 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3113 sbi->rw_iostat[APP_BUFFERED_IO] =
3114 sbi->rw_iostat[APP_WRITE_IO] -
3115 sbi->rw_iostat[APP_DIRECT_IO];
3117 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3118 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3119 sbi->rw_iostat[APP_READ_IO] -
3120 sbi->rw_iostat[APP_DIRECT_READ_IO];
3121 spin_unlock(&sbi->iostat_lock);
3123 f2fs_record_iostat(sbi);
3126 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3128 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3130 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3131 block_t blkaddr, int type);
3132 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3133 block_t blkaddr, int type)
3135 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3136 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3138 f2fs_bug_on(sbi, 1);
3142 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3144 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3145 blkaddr == COMPRESS_ADDR)
3150 static inline void f2fs_set_page_private(struct page *page,
3153 if (PagePrivate(page))
3156 attach_page_private(page, (void *)data);
3159 static inline void f2fs_clear_page_private(struct page *page)
3161 detach_page_private(page);
3167 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3168 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3169 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3170 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3171 int f2fs_truncate(struct inode *inode);
3172 int f2fs_getattr(const struct path *path, struct kstat *stat,
3173 u32 request_mask, unsigned int flags);
3174 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3175 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3176 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3177 int f2fs_precache_extents(struct inode *inode);
3178 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3179 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3180 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3181 int f2fs_pin_file_control(struct inode *inode, bool inc);
3186 void f2fs_set_inode_flags(struct inode *inode);
3187 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3188 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3189 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3190 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3191 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3192 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3193 void f2fs_update_inode_page(struct inode *inode);
3194 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3195 void f2fs_evict_inode(struct inode *inode);
3196 void f2fs_handle_failed_inode(struct inode *inode);
3201 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3202 bool hot, bool set);
3203 struct dentry *f2fs_get_parent(struct dentry *child);
3208 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3209 int f2fs_init_casefolded_name(const struct inode *dir,
3210 struct f2fs_filename *fname);
3211 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3212 int lookup, struct f2fs_filename *fname);
3213 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3214 struct f2fs_filename *fname);
3215 void f2fs_free_filename(struct f2fs_filename *fname);
3216 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3217 const struct f2fs_filename *fname, int *max_slots);
3218 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3219 unsigned int start_pos, struct fscrypt_str *fstr);
3220 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3221 struct f2fs_dentry_ptr *d);
3222 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3223 const struct f2fs_filename *fname, struct page *dpage);
3224 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3225 unsigned int current_depth);
3226 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3227 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3228 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3229 const struct f2fs_filename *fname,
3230 struct page **res_page);
3231 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3232 const struct qstr *child, struct page **res_page);
3233 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3234 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3235 struct page **page);
3236 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3237 struct page *page, struct inode *inode);
3238 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3239 const struct f2fs_filename *fname);
3240 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3241 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3242 unsigned int bit_pos);
3243 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3244 struct inode *inode, nid_t ino, umode_t mode);
3245 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3246 struct inode *inode, nid_t ino, umode_t mode);
3247 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3248 struct inode *inode, nid_t ino, umode_t mode);
3249 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3250 struct inode *dir, struct inode *inode);
3251 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3252 bool f2fs_empty_dir(struct inode *dir);
3254 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3256 if (fscrypt_is_nokey_name(dentry))
3258 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3259 inode, inode->i_ino, inode->i_mode);
3265 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3266 void f2fs_inode_synced(struct inode *inode);
3267 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3268 int f2fs_quota_sync(struct super_block *sb, int type);
3269 void f2fs_quota_off_umount(struct super_block *sb);
3270 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3271 int f2fs_sync_fs(struct super_block *sb, int sync);
3272 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3277 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3282 struct dnode_of_data;
3285 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3286 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3287 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3288 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3289 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3290 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3291 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3292 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3293 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3294 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3295 struct node_info *ni);
3296 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3297 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3298 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3299 int f2fs_truncate_xattr_node(struct inode *inode);
3300 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3301 unsigned int seq_id);
3302 int f2fs_remove_inode_page(struct inode *inode);
3303 struct page *f2fs_new_inode_page(struct inode *inode);
3304 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3305 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3306 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3307 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3308 int f2fs_move_node_page(struct page *node_page, int gc_type);
3309 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3310 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3311 struct writeback_control *wbc, bool atomic,
3312 unsigned int *seq_id);
3313 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3314 struct writeback_control *wbc,
3315 bool do_balance, enum iostat_type io_type);
3316 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3317 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3318 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3319 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3320 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3321 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3322 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3323 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3324 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3325 unsigned int segno, struct f2fs_summary_block *sum);
3326 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3327 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3328 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3329 int __init f2fs_create_node_manager_caches(void);
3330 void f2fs_destroy_node_manager_caches(void);
3335 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3336 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3337 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3338 void f2fs_drop_inmem_pages(struct inode *inode);
3339 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3340 int f2fs_commit_inmem_pages(struct inode *inode);
3341 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3342 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3343 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3344 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3345 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3346 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3347 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3348 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3349 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3350 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3351 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3352 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3353 struct cp_control *cpc);
3354 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3355 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3356 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3357 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3358 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3359 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3360 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3361 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3362 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3363 void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3364 unsigned int *newseg, bool new_sec, int dir);
3365 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3366 unsigned int start, unsigned int end);
3367 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type);
3368 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3369 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3370 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3371 struct cp_control *cpc);
3372 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3373 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3375 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3376 enum iostat_type io_type);
3377 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3378 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3379 struct f2fs_io_info *fio);
3380 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3381 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3382 block_t old_blkaddr, block_t new_blkaddr,
3383 bool recover_curseg, bool recover_newaddr,
3385 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3386 block_t old_addr, block_t new_addr,
3387 unsigned char version, bool recover_curseg,
3388 bool recover_newaddr);
3389 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3390 block_t old_blkaddr, block_t *new_blkaddr,
3391 struct f2fs_summary *sum, int type,
3392 struct f2fs_io_info *fio);
3393 void f2fs_wait_on_page_writeback(struct page *page,
3394 enum page_type type, bool ordered, bool locked);
3395 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3396 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3398 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3399 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3400 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3401 unsigned int val, int alloc);
3402 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3403 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3404 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3405 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3406 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3407 int __init f2fs_create_segment_manager_caches(void);
3408 void f2fs_destroy_segment_manager_caches(void);
3409 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3410 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3411 enum page_type type, enum temp_type temp);
3412 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3413 unsigned int segno);
3414 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3415 unsigned int segno);
3420 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3421 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3422 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3423 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3424 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3425 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3426 block_t blkaddr, int type);
3427 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3428 int type, bool sync);
3429 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3430 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3431 long nr_to_write, enum iostat_type io_type);
3432 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3433 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3434 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3435 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3436 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3437 unsigned int devidx, int type);
3438 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3439 unsigned int devidx, int type);
3440 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3441 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3442 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3443 void f2fs_add_orphan_inode(struct inode *inode);
3444 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3445 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3446 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3447 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3448 void f2fs_remove_dirty_inode(struct inode *inode);
3449 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3451 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3452 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3453 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3454 int __init f2fs_create_checkpoint_caches(void);
3455 void f2fs_destroy_checkpoint_caches(void);
3460 int __init f2fs_init_bioset(void);
3461 void f2fs_destroy_bioset(void);
3462 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio);
3463 int f2fs_init_bio_entry_cache(void);
3464 void f2fs_destroy_bio_entry_cache(void);
3465 void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3466 struct bio *bio, enum page_type type);
3467 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3468 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3469 struct inode *inode, struct page *page,
3470 nid_t ino, enum page_type type);
3471 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3472 struct bio **bio, struct page *page);
3473 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3474 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3475 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3476 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3477 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3478 block_t blk_addr, struct bio *bio);
3479 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3480 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3481 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3482 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3483 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3484 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3485 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3486 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3487 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3488 int op_flags, bool for_write);
3489 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3490 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3492 struct page *f2fs_get_new_data_page(struct inode *inode,
3493 struct page *ipage, pgoff_t index, bool new_i_size);
3494 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3495 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3496 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3497 int create, int flag);
3498 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3499 u64 start, u64 len);
3500 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3501 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3502 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3503 int f2fs_write_single_data_page(struct page *page, int *submitted,
3504 struct bio **bio, sector_t *last_block,
3505 struct writeback_control *wbc,
3506 enum iostat_type io_type,
3507 int compr_blocks, bool allow_balance);
3508 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3509 unsigned int length);
3510 int f2fs_release_page(struct page *page, gfp_t wait);
3511 #ifdef CONFIG_MIGRATION
3512 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3513 struct page *page, enum migrate_mode mode);
3515 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3516 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3517 int f2fs_init_post_read_processing(void);
3518 void f2fs_destroy_post_read_processing(void);
3519 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3520 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3525 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3526 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3527 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3528 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force,
3529 unsigned int segno);
3530 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3531 int f2fs_resize_fs(struct file *filp, __u64 block_count);
3532 int __init f2fs_create_garbage_collection_cache(void);
3533 void f2fs_destroy_garbage_collection_cache(void);
3538 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3539 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3540 int __init f2fs_create_recovery_cache(void);
3541 void f2fs_destroy_recovery_cache(void);
3546 #ifdef CONFIG_F2FS_STAT_FS
3547 struct f2fs_stat_info {
3548 struct list_head stat_list;
3549 struct f2fs_sb_info *sbi;
3550 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3551 int main_area_segs, main_area_sections, main_area_zones;
3552 unsigned long long hit_largest, hit_cached, hit_rbtree;
3553 unsigned long long hit_total, total_ext;
3554 int ext_tree, zombie_tree, ext_node;
3555 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3556 int ndirty_data, ndirty_qdata;
3558 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3559 int nats, dirty_nats, sits, dirty_sits;
3560 int free_nids, avail_nids, alloc_nids;
3561 int total_count, utilization;
3562 int bg_gc, nr_wb_cp_data, nr_wb_data;
3563 int nr_rd_data, nr_rd_node, nr_rd_meta;
3564 int nr_dio_read, nr_dio_write;
3565 unsigned int io_skip_bggc, other_skip_bggc;
3566 int nr_flushing, nr_flushed, flush_list_empty;
3567 int nr_discarding, nr_discarded;
3569 unsigned int undiscard_blks;
3570 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3572 unsigned long long compr_blocks;
3573 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3574 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3575 unsigned int bimodal, avg_vblocks;
3576 int util_free, util_valid, util_invalid;
3577 int rsvd_segs, overp_segs;
3578 int dirty_count, node_pages, meta_pages;
3579 int prefree_count, call_count, cp_count, bg_cp_count;
3580 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3581 int bg_node_segs, bg_data_segs;
3582 int tot_blks, data_blks, node_blks;
3583 int bg_data_blks, bg_node_blks;
3584 unsigned long long skipped_atomic_files[2];
3585 int curseg[NR_CURSEG_TYPE];
3586 int cursec[NR_CURSEG_TYPE];
3587 int curzone[NR_CURSEG_TYPE];
3588 unsigned int dirty_seg[NR_CURSEG_TYPE];
3589 unsigned int full_seg[NR_CURSEG_TYPE];
3590 unsigned int valid_blks[NR_CURSEG_TYPE];
3592 unsigned int meta_count[META_MAX];
3593 unsigned int segment_count[2];
3594 unsigned int block_count[2];
3595 unsigned int inplace_count;
3596 unsigned long long base_mem, cache_mem, page_mem;
3599 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3601 return (struct f2fs_stat_info *)sbi->stat_info;
3604 #define stat_inc_cp_count(si) ((si)->cp_count++)
3605 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3606 #define stat_inc_call_count(si) ((si)->call_count++)
3607 #define stat_inc_bggc_count(si) ((si)->bg_gc++)
3608 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3609 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3610 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3611 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3612 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3613 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3614 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3615 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3616 #define stat_inc_inline_xattr(inode) \
3618 if (f2fs_has_inline_xattr(inode)) \
3619 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3621 #define stat_dec_inline_xattr(inode) \
3623 if (f2fs_has_inline_xattr(inode)) \
3624 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3626 #define stat_inc_inline_inode(inode) \
3628 if (f2fs_has_inline_data(inode)) \
3629 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3631 #define stat_dec_inline_inode(inode) \
3633 if (f2fs_has_inline_data(inode)) \
3634 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3636 #define stat_inc_inline_dir(inode) \
3638 if (f2fs_has_inline_dentry(inode)) \
3639 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3641 #define stat_dec_inline_dir(inode) \
3643 if (f2fs_has_inline_dentry(inode)) \
3644 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3646 #define stat_inc_compr_inode(inode) \
3648 if (f2fs_compressed_file(inode)) \
3649 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3651 #define stat_dec_compr_inode(inode) \
3653 if (f2fs_compressed_file(inode)) \
3654 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3656 #define stat_add_compr_blocks(inode, blocks) \
3657 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3658 #define stat_sub_compr_blocks(inode, blocks) \
3659 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3660 #define stat_inc_meta_count(sbi, blkaddr) \
3662 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3663 atomic_inc(&(sbi)->meta_count[META_CP]); \
3664 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3665 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3666 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3667 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3668 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3669 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3671 #define stat_inc_seg_type(sbi, curseg) \
3672 ((sbi)->segment_count[(curseg)->alloc_type]++)
3673 #define stat_inc_block_count(sbi, curseg) \
3674 ((sbi)->block_count[(curseg)->alloc_type]++)
3675 #define stat_inc_inplace_blocks(sbi) \
3676 (atomic_inc(&(sbi)->inplace_count))
3677 #define stat_update_max_atomic_write(inode) \
3679 int cur = F2FS_I_SB(inode)->atomic_files; \
3680 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3682 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3684 #define stat_inc_volatile_write(inode) \
3685 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3686 #define stat_dec_volatile_write(inode) \
3687 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3688 #define stat_update_max_volatile_write(inode) \
3690 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3691 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3693 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3695 #define stat_inc_seg_count(sbi, type, gc_type) \
3697 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3699 if ((type) == SUM_TYPE_DATA) { \
3701 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3704 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3708 #define stat_inc_tot_blk_count(si, blks) \
3709 ((si)->tot_blks += (blks))
3711 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3713 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3714 stat_inc_tot_blk_count(si, blks); \
3715 si->data_blks += (blks); \
3716 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3719 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3721 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3722 stat_inc_tot_blk_count(si, blks); \
3723 si->node_blks += (blks); \
3724 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3727 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3728 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3729 void __init f2fs_create_root_stats(void);
3730 void f2fs_destroy_root_stats(void);
3731 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3733 #define stat_inc_cp_count(si) do { } while (0)
3734 #define stat_inc_bg_cp_count(si) do { } while (0)
3735 #define stat_inc_call_count(si) do { } while (0)
3736 #define stat_inc_bggc_count(si) do { } while (0)
3737 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3738 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3739 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3740 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3741 #define stat_inc_total_hit(sbi) do { } while (0)
3742 #define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3743 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3744 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3745 #define stat_inc_inline_xattr(inode) do { } while (0)
3746 #define stat_dec_inline_xattr(inode) do { } while (0)
3747 #define stat_inc_inline_inode(inode) do { } while (0)
3748 #define stat_dec_inline_inode(inode) do { } while (0)
3749 #define stat_inc_inline_dir(inode) do { } while (0)
3750 #define stat_dec_inline_dir(inode) do { } while (0)
3751 #define stat_inc_compr_inode(inode) do { } while (0)
3752 #define stat_dec_compr_inode(inode) do { } while (0)
3753 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
3754 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3755 #define stat_inc_atomic_write(inode) do { } while (0)
3756 #define stat_dec_atomic_write(inode) do { } while (0)
3757 #define stat_update_max_atomic_write(inode) do { } while (0)
3758 #define stat_inc_volatile_write(inode) do { } while (0)
3759 #define stat_dec_volatile_write(inode) do { } while (0)
3760 #define stat_update_max_volatile_write(inode) do { } while (0)
3761 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3762 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3763 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3764 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3765 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3766 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3767 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3768 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3770 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3771 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3772 static inline void __init f2fs_create_root_stats(void) { }
3773 static inline void f2fs_destroy_root_stats(void) { }
3774 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3777 extern const struct file_operations f2fs_dir_operations;
3778 #ifdef CONFIG_UNICODE
3779 extern const struct dentry_operations f2fs_dentry_ops;
3781 extern const struct file_operations f2fs_file_operations;
3782 extern const struct inode_operations f2fs_file_inode_operations;
3783 extern const struct address_space_operations f2fs_dblock_aops;
3784 extern const struct address_space_operations f2fs_node_aops;
3785 extern const struct address_space_operations f2fs_meta_aops;
3786 extern const struct inode_operations f2fs_dir_inode_operations;
3787 extern const struct inode_operations f2fs_symlink_inode_operations;
3788 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3789 extern const struct inode_operations f2fs_special_inode_operations;
3790 extern struct kmem_cache *f2fs_inode_entry_slab;
3795 bool f2fs_may_inline_data(struct inode *inode);
3796 bool f2fs_sanity_check_inline_data(struct inode *inode);
3797 bool f2fs_may_inline_dentry(struct inode *inode);
3798 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3799 void f2fs_truncate_inline_inode(struct inode *inode,
3800 struct page *ipage, u64 from);
3801 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3802 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3803 int f2fs_convert_inline_inode(struct inode *inode);
3804 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3805 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3806 int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3807 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3808 const struct f2fs_filename *fname,
3809 struct page **res_page);
3810 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3811 struct page *ipage);
3812 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3813 struct inode *inode, nid_t ino, umode_t mode);
3814 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3815 struct page *page, struct inode *dir,
3816 struct inode *inode);
3817 bool f2fs_empty_inline_dir(struct inode *dir);
3818 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3819 struct fscrypt_str *fstr);
3820 int f2fs_inline_data_fiemap(struct inode *inode,
3821 struct fiemap_extent_info *fieinfo,
3822 __u64 start, __u64 len);
3827 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3828 struct shrink_control *sc);
3829 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3830 struct shrink_control *sc);
3831 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3832 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3837 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3838 struct rb_entry *cached_re, unsigned int ofs);
3839 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
3840 struct rb_root_cached *root,
3841 struct rb_node **parent,
3842 unsigned long long key, bool *left_most);
3843 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3844 struct rb_root_cached *root,
3845 struct rb_node **parent,
3846 unsigned int ofs, bool *leftmost);
3847 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3848 struct rb_entry *cached_re, unsigned int ofs,
3849 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3850 struct rb_node ***insert_p, struct rb_node **insert_parent,
3851 bool force, bool *leftmost);
3852 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3853 struct rb_root_cached *root, bool check_key);
3854 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3855 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3856 void f2fs_drop_extent_tree(struct inode *inode);
3857 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3858 void f2fs_destroy_extent_tree(struct inode *inode);
3859 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3860 struct extent_info *ei);
3861 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3862 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3863 pgoff_t fofs, block_t blkaddr, unsigned int len);
3864 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3865 int __init f2fs_create_extent_cache(void);
3866 void f2fs_destroy_extent_cache(void);
3871 int __init f2fs_init_sysfs(void);
3872 void f2fs_exit_sysfs(void);
3873 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3874 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3877 extern const struct fsverity_operations f2fs_verityops;
3882 static inline bool f2fs_encrypted_file(struct inode *inode)
3884 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3887 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3889 #ifdef CONFIG_FS_ENCRYPTION
3890 file_set_encrypt(inode);
3891 f2fs_set_inode_flags(inode);
3896 * Returns true if the reads of the inode's data need to undergo some
3897 * postprocessing step, like decryption or authenticity verification.
3899 static inline bool f2fs_post_read_required(struct inode *inode)
3901 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3902 f2fs_compressed_file(inode);
3908 #ifdef CONFIG_F2FS_FS_COMPRESSION
3909 bool f2fs_is_compressed_page(struct page *page);
3910 struct page *f2fs_compress_control_page(struct page *page);
3911 int f2fs_prepare_compress_overwrite(struct inode *inode,
3912 struct page **pagep, pgoff_t index, void **fsdata);
3913 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3914 pgoff_t index, unsigned copied);
3915 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3916 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3917 bool f2fs_is_compress_backend_ready(struct inode *inode);
3918 int f2fs_init_compress_mempool(void);
3919 void f2fs_destroy_compress_mempool(void);
3920 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity);
3921 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3922 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3923 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3924 int f2fs_write_multi_pages(struct compress_ctx *cc,
3926 struct writeback_control *wbc,
3927 enum iostat_type io_type);
3928 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3929 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3930 unsigned nr_pages, sector_t *last_block_in_bio,
3931 bool is_readahead, bool for_write);
3932 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3933 void f2fs_free_dic(struct decompress_io_ctx *dic);
3934 void f2fs_decompress_end_io(struct page **rpages,
3935 unsigned int cluster_size, bool err, bool verity);
3936 int f2fs_init_compress_ctx(struct compress_ctx *cc);
3937 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
3938 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3939 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
3940 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
3941 int __init f2fs_init_compress_cache(void);
3942 void f2fs_destroy_compress_cache(void);
3944 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3945 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3947 if (!f2fs_compressed_file(inode))
3949 /* not support compression */
3952 static inline struct page *f2fs_compress_control_page(struct page *page)
3955 return ERR_PTR(-EINVAL);
3957 static inline int f2fs_init_compress_mempool(void) { return 0; }
3958 static inline void f2fs_destroy_compress_mempool(void) { }
3959 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
3960 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
3961 static inline int __init f2fs_init_compress_cache(void) { return 0; }
3962 static inline void f2fs_destroy_compress_cache(void) { }
3965 static inline void set_compress_context(struct inode *inode)
3967 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3969 F2FS_I(inode)->i_compress_algorithm =
3970 F2FS_OPTION(sbi).compress_algorithm;
3971 F2FS_I(inode)->i_log_cluster_size =
3972 F2FS_OPTION(sbi).compress_log_size;
3973 F2FS_I(inode)->i_compress_flag =
3974 F2FS_OPTION(sbi).compress_chksum ?
3975 1 << COMPRESS_CHKSUM : 0;
3976 F2FS_I(inode)->i_cluster_size =
3977 1 << F2FS_I(inode)->i_log_cluster_size;
3978 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
3979 set_inode_flag(inode, FI_COMPRESSED_FILE);
3980 stat_inc_compr_inode(inode);
3981 f2fs_mark_inode_dirty_sync(inode, true);
3984 static inline bool f2fs_disable_compressed_file(struct inode *inode)
3986 struct f2fs_inode_info *fi = F2FS_I(inode);
3988 if (!f2fs_compressed_file(inode))
3990 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))
3993 fi->i_flags &= ~F2FS_COMPR_FL;
3994 stat_dec_compr_inode(inode);
3995 clear_inode_flag(inode, FI_COMPRESSED_FILE);
3996 f2fs_mark_inode_dirty_sync(inode, true);
4000 #define F2FS_FEATURE_FUNCS(name, flagname) \
4001 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4003 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4006 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4007 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4008 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4009 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4010 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4011 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4012 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4013 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4014 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4015 F2FS_FEATURE_FUNCS(verity, VERITY);
4016 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4017 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4018 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4020 #ifdef CONFIG_BLK_DEV_ZONED
4021 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4024 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4026 return test_bit(zno, FDEV(devi).blkz_seq);
4030 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4032 return f2fs_sb_has_blkzoned(sbi);
4035 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4037 return blk_queue_discard(bdev_get_queue(bdev)) ||
4038 bdev_is_zoned(bdev);
4041 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4045 if (!f2fs_is_multi_device(sbi))
4046 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4048 for (i = 0; i < sbi->s_ndevs; i++)
4049 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4054 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4056 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4057 f2fs_hw_should_discard(sbi);
4060 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4064 if (!f2fs_is_multi_device(sbi))
4065 return bdev_read_only(sbi->sb->s_bdev);
4067 for (i = 0; i < sbi->s_ndevs; i++)
4068 if (bdev_read_only(FDEV(i).bdev))
4073 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4075 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4078 static inline bool f2fs_may_compress(struct inode *inode)
4080 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4081 f2fs_is_atomic_file(inode) ||
4082 f2fs_is_volatile_file(inode))
4084 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4087 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4088 u64 blocks, bool add)
4090 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4091 struct f2fs_inode_info *fi = F2FS_I(inode);
4093 /* don't update i_compr_blocks if saved blocks were released */
4094 if (!add && !atomic_read(&fi->i_compr_blocks))
4098 atomic_add(diff, &fi->i_compr_blocks);
4099 stat_add_compr_blocks(inode, diff);
4101 atomic_sub(diff, &fi->i_compr_blocks);
4102 stat_sub_compr_blocks(inode, diff);
4104 f2fs_mark_inode_dirty_sync(inode, true);
4107 static inline int block_unaligned_IO(struct inode *inode,
4108 struct kiocb *iocb, struct iov_iter *iter)
4110 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4111 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4112 loff_t offset = iocb->ki_pos;
4113 unsigned long align = offset | iov_iter_alignment(iter);
4115 return align & blocksize_mask;
4118 static inline int allow_outplace_dio(struct inode *inode,
4119 struct kiocb *iocb, struct iov_iter *iter)
4121 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4122 int rw = iov_iter_rw(iter);
4124 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4125 !block_unaligned_IO(inode, iocb, iter));
4128 static inline bool f2fs_force_buffered_io(struct inode *inode,
4129 struct kiocb *iocb, struct iov_iter *iter)
4131 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4132 int rw = iov_iter_rw(iter);
4134 if (f2fs_post_read_required(inode))
4136 if (f2fs_is_multi_device(sbi))
4139 * for blkzoned device, fallback direct IO to buffered IO, so
4140 * all IOs can be serialized by log-structured write.
4142 if (f2fs_sb_has_blkzoned(sbi))
4144 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4145 if (block_unaligned_IO(inode, iocb, iter))
4147 if (F2FS_IO_ALIGNED(sbi))
4150 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4151 !IS_SWAPFILE(inode))
4157 #ifdef CONFIG_F2FS_FAULT_INJECTION
4158 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4161 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4164 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4167 if (f2fs_sb_has_quota_ino(sbi))
4169 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4170 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4171 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4177 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4178 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4180 #endif /* _LINUX_F2FS_H */