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 (WARN_ON(condition)) \
37 set_sbi_flag(sbi, SBI_NEED_FSCK); \
59 #ifdef CONFIG_F2FS_FAULT_INJECTION
60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
62 struct f2fs_fault_info {
64 unsigned int inject_rate;
65 unsigned int inject_type;
68 extern const char *f2fs_fault_name[FAULT_MAX];
69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
76 #define F2FS_MOUNT_DISCARD 0x00000004
77 #define F2FS_MOUNT_NOHEAP 0x00000008
78 #define F2FS_MOUNT_XATTR_USER 0x00000010
79 #define F2FS_MOUNT_POSIX_ACL 0x00000020
80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
82 #define F2FS_MOUNT_INLINE_DATA 0x00000100
83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
85 #define F2FS_MOUNT_NOBARRIER 0x00000800
86 #define F2FS_MOUNT_FASTBOOT 0x00001000
87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
90 #define F2FS_MOUNT_USRQUOTA 0x00080000
91 #define F2FS_MOUNT_GRPQUOTA 0x00100000
92 #define F2FS_MOUNT_PRJQUOTA 0x00200000
93 #define F2FS_MOUNT_QUOTA 0x00400000
94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
97 #define F2FS_MOUNT_NORECOVERY 0x04000000
98 #define F2FS_MOUNT_ATGC 0x08000000
99 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000
101 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
102 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
103 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
104 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
106 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
107 typecheck(unsigned long long, b) && \
108 ((long long)((a) - (b)) > 0))
110 typedef u32 block_t; /*
111 * should not change u32, since it is the on-disk block
112 * address format, __le32.
116 #define COMPRESS_EXT_NUM 16
118 struct f2fs_mount_info {
120 int write_io_size_bits; /* Write IO size bits */
121 block_t root_reserved_blocks; /* root reserved blocks */
122 kuid_t s_resuid; /* reserved blocks for uid */
123 kgid_t s_resgid; /* reserved blocks for gid */
124 int active_logs; /* # of active logs */
125 int inline_xattr_size; /* inline xattr size */
126 #ifdef CONFIG_F2FS_FAULT_INJECTION
127 struct f2fs_fault_info fault_info; /* For fault injection */
130 /* Names of quota files with journalled quota */
131 char *s_qf_names[MAXQUOTAS];
132 int s_jquota_fmt; /* Format of quota to use */
134 /* For which write hints are passed down to block layer */
136 int alloc_mode; /* segment allocation policy */
137 int fsync_mode; /* fsync policy */
138 int fs_mode; /* fs mode: LFS or ADAPTIVE */
139 int bggc_mode; /* bggc mode: off, on or sync */
140 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
141 block_t unusable_cap_perc; /* percentage for cap */
142 block_t unusable_cap; /* Amount of space allowed to be
143 * unusable when disabling checkpoint
146 /* For compression */
147 unsigned char compress_algorithm; /* algorithm type */
148 unsigned char compress_log_size; /* cluster log size */
149 unsigned char compress_level; /* compress level */
150 bool compress_chksum; /* compressed data chksum */
151 unsigned char compress_ext_cnt; /* extension count */
152 int compress_mode; /* compression mode */
153 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
156 #define F2FS_FEATURE_ENCRYPT 0x0001
157 #define F2FS_FEATURE_BLKZONED 0x0002
158 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
159 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
160 #define F2FS_FEATURE_PRJQUOTA 0x0010
161 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
162 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
163 #define F2FS_FEATURE_QUOTA_INO 0x0080
164 #define F2FS_FEATURE_INODE_CRTIME 0x0100
165 #define F2FS_FEATURE_LOST_FOUND 0x0200
166 #define F2FS_FEATURE_VERITY 0x0400
167 #define F2FS_FEATURE_SB_CHKSUM 0x0800
168 #define F2FS_FEATURE_CASEFOLD 0x1000
169 #define F2FS_FEATURE_COMPRESSION 0x2000
171 #define __F2FS_HAS_FEATURE(raw_super, mask) \
172 ((raw_super->feature & cpu_to_le32(mask)) != 0)
173 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
174 #define F2FS_SET_FEATURE(sbi, mask) \
175 (sbi->raw_super->feature |= cpu_to_le32(mask))
176 #define F2FS_CLEAR_FEATURE(sbi, mask) \
177 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
180 * Default values for user and/or group using reserved blocks
182 #define F2FS_DEF_RESUID 0
183 #define F2FS_DEF_RESGID 0
186 * For checkpoint manager
193 #define CP_UMOUNT 0x00000001
194 #define CP_FASTBOOT 0x00000002
195 #define CP_SYNC 0x00000004
196 #define CP_RECOVERY 0x00000008
197 #define CP_DISCARD 0x00000010
198 #define CP_TRIMMED 0x00000020
199 #define CP_PAUSE 0x00000040
200 #define CP_RESIZE 0x00000080
202 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
203 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
204 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
205 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
206 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
207 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
208 #define DEF_CP_INTERVAL 60 /* 60 secs */
209 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
210 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
211 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
212 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
222 * indicate meta/data type
231 DATA_GENERIC, /* check range only */
232 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
233 DATA_GENERIC_ENHANCE_READ, /*
234 * strong check on range and segment
235 * bitmap but no warning due to race
236 * condition of read on truncated area
242 /* for the list of ino */
244 ORPHAN_INO, /* for orphan ino list */
245 APPEND_INO, /* for append ino list */
246 UPDATE_INO, /* for update ino list */
247 TRANS_DIR_INO, /* for trasactions dir ino list */
248 FLUSH_INO, /* for multiple device flushing */
249 MAX_INO_ENTRY, /* max. list */
253 struct list_head list; /* list head */
254 nid_t ino; /* inode number */
255 unsigned int dirty_device; /* dirty device bitmap */
258 /* for the list of inodes to be GCed */
260 struct list_head list; /* list head */
261 struct inode *inode; /* vfs inode pointer */
264 struct fsync_node_entry {
265 struct list_head list; /* list head */
266 struct page *page; /* warm node page pointer */
267 unsigned int seq_id; /* sequence id */
271 struct completion wait; /* completion for checkpoint done */
272 struct llist_node llnode; /* llist_node to be linked in wait queue */
273 int ret; /* return code of checkpoint */
274 ktime_t queue_time; /* request queued time */
277 struct ckpt_req_control {
278 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
279 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
280 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
281 atomic_t issued_ckpt; /* # of actually issued ckpts */
282 atomic_t total_ckpt; /* # of total ckpts */
283 atomic_t queued_ckpt; /* # of queued ckpts */
284 struct llist_head issue_list; /* list for command issue */
285 spinlock_t stat_lock; /* lock for below checkpoint time stats */
286 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
287 unsigned int peak_time; /* peak wait time in msec until now */
290 /* for the bitmap indicate blocks to be discarded */
291 struct discard_entry {
292 struct list_head list; /* list head */
293 block_t start_blkaddr; /* start blockaddr of current segment */
294 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
297 /* default discard granularity of inner discard thread, unit: block count */
298 #define DEFAULT_DISCARD_GRANULARITY 16
300 /* max discard pend list number */
301 #define MAX_PLIST_NUM 512
302 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
303 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
306 D_PREP, /* initial */
307 D_PARTIAL, /* partially submitted */
308 D_SUBMIT, /* all submitted */
309 D_DONE, /* finished */
312 struct discard_info {
313 block_t lstart; /* logical start address */
314 block_t len; /* length */
315 block_t start; /* actual start address in dev */
319 struct rb_node rb_node; /* rb node located in rb-tree */
322 block_t lstart; /* logical start address */
323 block_t len; /* length */
324 block_t start; /* actual start address in dev */
326 struct discard_info di; /* discard info */
329 struct list_head list; /* command list */
330 struct completion wait; /* compleation */
331 struct block_device *bdev; /* bdev */
332 unsigned short ref; /* reference count */
333 unsigned char state; /* state */
334 unsigned char queued; /* queued discard */
335 int error; /* bio error */
336 spinlock_t lock; /* for state/bio_ref updating */
337 unsigned short bio_ref; /* bio reference count */
348 struct discard_policy {
349 int type; /* type of discard */
350 unsigned int min_interval; /* used for candidates exist */
351 unsigned int mid_interval; /* used for device busy */
352 unsigned int max_interval; /* used for candidates not exist */
353 unsigned int max_requests; /* # of discards issued per round */
354 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
355 bool io_aware; /* issue discard in idle time */
356 bool sync; /* submit discard with REQ_SYNC flag */
357 bool ordered; /* issue discard by lba order */
358 bool timeout; /* discard timeout for put_super */
359 unsigned int granularity; /* discard granularity */
362 struct discard_cmd_control {
363 struct task_struct *f2fs_issue_discard; /* discard thread */
364 struct list_head entry_list; /* 4KB discard entry list */
365 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
366 struct list_head wait_list; /* store on-flushing entries */
367 struct list_head fstrim_list; /* in-flight discard from fstrim */
368 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
369 unsigned int discard_wake; /* to wake up discard thread */
370 struct mutex cmd_lock;
371 unsigned int nr_discards; /* # of discards in the list */
372 unsigned int max_discards; /* max. discards to be issued */
373 unsigned int discard_granularity; /* discard granularity */
374 unsigned int undiscard_blks; /* # of undiscard blocks */
375 unsigned int next_pos; /* next discard position */
376 atomic_t issued_discard; /* # of issued discard */
377 atomic_t queued_discard; /* # of queued discard */
378 atomic_t discard_cmd_cnt; /* # of cached cmd count */
379 struct rb_root_cached root; /* root of discard rb-tree */
380 bool rbtree_check; /* config for consistence check */
383 /* for the list of fsync inodes, used only during recovery */
384 struct fsync_inode_entry {
385 struct list_head list; /* list head */
386 struct inode *inode; /* vfs inode pointer */
387 block_t blkaddr; /* block address locating the last fsync */
388 block_t last_dentry; /* block address locating the last dentry */
391 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
392 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
394 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
395 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
396 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
397 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
399 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
400 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
402 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
404 int before = nats_in_cursum(journal);
406 journal->n_nats = cpu_to_le16(before + i);
410 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
412 int before = sits_in_cursum(journal);
414 journal->n_sits = cpu_to_le16(before + i);
418 static inline bool __has_cursum_space(struct f2fs_journal *journal,
421 if (type == NAT_JOURNAL)
422 return size <= MAX_NAT_JENTRIES(journal);
423 return size <= MAX_SIT_JENTRIES(journal);
426 /* for inline stuff */
427 #define DEF_INLINE_RESERVED_SIZE 1
428 static inline int get_extra_isize(struct inode *inode);
429 static inline int get_inline_xattr_addrs(struct inode *inode);
430 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
431 (CUR_ADDRS_PER_INODE(inode) - \
432 get_inline_xattr_addrs(inode) - \
433 DEF_INLINE_RESERVED_SIZE))
436 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
437 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
439 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
440 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
441 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
442 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
443 NR_INLINE_DENTRY(inode) + \
444 INLINE_DENTRY_BITMAP_SIZE(inode)))
447 * For INODE and NODE manager
449 /* for directory operations */
451 struct f2fs_filename {
453 * The filename the user specified. This is NULL for some
454 * filesystem-internal operations, e.g. converting an inline directory
455 * to a non-inline one, or roll-forward recovering an encrypted dentry.
457 const struct qstr *usr_fname;
460 * The on-disk filename. For encrypted directories, this is encrypted.
461 * This may be NULL for lookups in an encrypted dir without the key.
463 struct fscrypt_str disk_name;
465 /* The dirhash of this filename */
468 #ifdef CONFIG_FS_ENCRYPTION
470 * For lookups in encrypted directories: either the buffer backing
471 * disk_name, or a buffer that holds the decoded no-key name.
473 struct fscrypt_str crypto_buf;
475 #ifdef CONFIG_UNICODE
477 * For casefolded directories: the casefolded name, but it's left NULL
478 * if the original name is not valid Unicode, if the directory is both
479 * casefolded and encrypted and its encryption key is unavailable, or if
480 * the filesystem is doing an internal operation where usr_fname is also
481 * NULL. In all these cases we fall back to treating the name as an
482 * opaque byte sequence.
484 struct fscrypt_str cf_name;
488 struct f2fs_dentry_ptr {
491 struct f2fs_dir_entry *dentry;
492 __u8 (*filename)[F2FS_SLOT_LEN];
497 static inline void make_dentry_ptr_block(struct inode *inode,
498 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
501 d->max = NR_DENTRY_IN_BLOCK;
502 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
503 d->bitmap = t->dentry_bitmap;
504 d->dentry = t->dentry;
505 d->filename = t->filename;
508 static inline void make_dentry_ptr_inline(struct inode *inode,
509 struct f2fs_dentry_ptr *d, void *t)
511 int entry_cnt = NR_INLINE_DENTRY(inode);
512 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
513 int reserved_size = INLINE_RESERVED_SIZE(inode);
517 d->nr_bitmap = bitmap_size;
519 d->dentry = t + bitmap_size + reserved_size;
520 d->filename = t + bitmap_size + reserved_size +
521 SIZE_OF_DIR_ENTRY * entry_cnt;
525 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
526 * as its node offset to distinguish from index node blocks.
527 * But some bits are used to mark the node block.
529 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
532 ALLOC_NODE, /* allocate a new node page if needed */
533 LOOKUP_NODE, /* look up a node without readahead */
535 * look up a node with readahead called
540 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
542 /* congestion wait timeout value, default: 20ms */
543 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
545 /* maximum retry quota flush count */
546 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
548 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
550 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
552 /* for in-memory extent cache entry */
553 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
555 /* number of extent info in extent cache we try to shrink */
556 #define EXTENT_CACHE_SHRINK_NUMBER 128
559 struct rb_node rb_node; /* rb node located in rb-tree */
562 unsigned int ofs; /* start offset of the entry */
563 unsigned int len; /* length of the entry */
565 unsigned long long key; /* 64-bits key */
570 unsigned int fofs; /* start offset in a file */
571 unsigned int len; /* length of the extent */
572 u32 blk; /* start block address of the extent */
576 struct rb_node rb_node; /* rb node located in rb-tree */
577 struct extent_info ei; /* extent info */
578 struct list_head list; /* node in global extent list of sbi */
579 struct extent_tree *et; /* extent tree pointer */
583 nid_t ino; /* inode number */
584 struct rb_root_cached root; /* root of extent info rb-tree */
585 struct extent_node *cached_en; /* recently accessed extent node */
586 struct extent_info largest; /* largested extent info */
587 struct list_head list; /* to be used by sbi->zombie_list */
588 rwlock_t lock; /* protect extent info rb-tree */
589 atomic_t node_cnt; /* # of extent node in rb-tree*/
590 bool largest_updated; /* largest extent updated */
594 * This structure is taken from ext4_map_blocks.
596 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
598 #define F2FS_MAP_NEW (1 << BH_New)
599 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
600 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
601 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
604 struct f2fs_map_blocks {
608 unsigned int m_flags;
609 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
610 pgoff_t *m_next_extent; /* point to next possible extent */
612 bool m_may_create; /* indicate it is from write path */
615 /* for flag in get_data_block */
617 F2FS_GET_BLOCK_DEFAULT,
618 F2FS_GET_BLOCK_FIEMAP,
621 F2FS_GET_BLOCK_PRE_DIO,
622 F2FS_GET_BLOCK_PRE_AIO,
623 F2FS_GET_BLOCK_PRECACHE,
627 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
629 #define FADVISE_COLD_BIT 0x01
630 #define FADVISE_LOST_PINO_BIT 0x02
631 #define FADVISE_ENCRYPT_BIT 0x04
632 #define FADVISE_ENC_NAME_BIT 0x08
633 #define FADVISE_KEEP_SIZE_BIT 0x10
634 #define FADVISE_HOT_BIT 0x20
635 #define FADVISE_VERITY_BIT 0x40
637 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
639 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
640 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
641 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
642 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
643 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
644 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
645 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
646 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
647 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
648 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
649 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
650 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
651 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
652 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
653 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
654 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
655 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
656 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
658 #define DEF_DIR_LEVEL 0
666 /* used for f2fs_inode_info->flags */
668 FI_NEW_INODE, /* indicate newly allocated inode */
669 FI_DIRTY_INODE, /* indicate inode is dirty or not */
670 FI_AUTO_RECOVER, /* indicate inode is recoverable */
671 FI_DIRTY_DIR, /* indicate directory has dirty pages */
672 FI_INC_LINK, /* need to increment i_nlink */
673 FI_ACL_MODE, /* indicate acl mode */
674 FI_NO_ALLOC, /* should not allocate any blocks */
675 FI_FREE_NID, /* free allocated nide */
676 FI_NO_EXTENT, /* not to use the extent cache */
677 FI_INLINE_XATTR, /* used for inline xattr */
678 FI_INLINE_DATA, /* used for inline data*/
679 FI_INLINE_DENTRY, /* used for inline dentry */
680 FI_APPEND_WRITE, /* inode has appended data */
681 FI_UPDATE_WRITE, /* inode has in-place-update data */
682 FI_NEED_IPU, /* used for ipu per file */
683 FI_ATOMIC_FILE, /* indicate atomic file */
684 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
685 FI_VOLATILE_FILE, /* indicate volatile file */
686 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
687 FI_DROP_CACHE, /* drop dirty page cache */
688 FI_DATA_EXIST, /* indicate data exists */
689 FI_INLINE_DOTS, /* indicate inline dot dentries */
690 FI_DO_DEFRAG, /* indicate defragment is running */
691 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
692 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
693 FI_HOT_DATA, /* indicate file is hot */
694 FI_EXTRA_ATTR, /* indicate file has extra attribute */
695 FI_PROJ_INHERIT, /* indicate file inherits projectid */
696 FI_PIN_FILE, /* indicate file should not be gced */
697 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
698 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
699 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
700 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
701 FI_MMAP_FILE, /* indicate file was mmapped */
702 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
703 FI_MAX, /* max flag, never be used */
706 struct f2fs_inode_info {
707 struct inode vfs_inode; /* serve a vfs inode */
708 unsigned long i_flags; /* keep an inode flags for ioctl */
709 unsigned char i_advise; /* use to give file attribute hints */
710 unsigned char i_dir_level; /* use for dentry level for large dir */
711 unsigned int i_current_depth; /* only for directory depth */
712 /* for gc failure statistic */
713 unsigned int i_gc_failures[MAX_GC_FAILURE];
714 unsigned int i_pino; /* parent inode number */
715 umode_t i_acl_mode; /* keep file acl mode temporarily */
717 /* Use below internally in f2fs*/
718 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
719 struct rw_semaphore i_sem; /* protect fi info */
720 atomic_t dirty_pages; /* # of dirty pages */
721 f2fs_hash_t chash; /* hash value of given file name */
722 unsigned int clevel; /* maximum level of given file name */
723 struct task_struct *task; /* lookup and create consistency */
724 struct task_struct *cp_task; /* separate cp/wb IO stats*/
725 nid_t i_xattr_nid; /* node id that contains xattrs */
726 loff_t last_disk_size; /* lastly written file size */
727 spinlock_t i_size_lock; /* protect last_disk_size */
730 struct dquot *i_dquot[MAXQUOTAS];
732 /* quota space reservation, managed internally by quota code */
733 qsize_t i_reserved_quota;
735 struct list_head dirty_list; /* dirty list for dirs and files */
736 struct list_head gdirty_list; /* linked in global dirty list */
737 struct list_head inmem_ilist; /* list for inmem inodes */
738 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
739 struct task_struct *inmem_task; /* store inmemory task */
740 struct mutex inmem_lock; /* lock for inmemory pages */
741 struct extent_tree *extent_tree; /* cached extent_tree entry */
743 /* avoid racing between foreground op and gc */
744 struct rw_semaphore i_gc_rwsem[2];
745 struct rw_semaphore i_mmap_sem;
746 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
748 int i_extra_isize; /* size of extra space located in i_addr */
749 kprojid_t i_projid; /* id for project quota */
750 int i_inline_xattr_size; /* inline xattr size */
751 struct timespec64 i_crtime; /* inode creation time */
752 struct timespec64 i_disk_time[4];/* inode disk times */
754 /* for file compress */
755 atomic_t i_compr_blocks; /* # of compressed blocks */
756 unsigned char i_compress_algorithm; /* algorithm type */
757 unsigned char i_log_cluster_size; /* log of cluster size */
758 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
759 unsigned short i_compress_flag; /* compress flag */
760 unsigned int i_cluster_size; /* cluster size */
763 static inline void get_extent_info(struct extent_info *ext,
764 struct f2fs_extent *i_ext)
766 ext->fofs = le32_to_cpu(i_ext->fofs);
767 ext->blk = le32_to_cpu(i_ext->blk);
768 ext->len = le32_to_cpu(i_ext->len);
771 static inline void set_raw_extent(struct extent_info *ext,
772 struct f2fs_extent *i_ext)
774 i_ext->fofs = cpu_to_le32(ext->fofs);
775 i_ext->blk = cpu_to_le32(ext->blk);
776 i_ext->len = cpu_to_le32(ext->len);
779 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
780 u32 blk, unsigned int len)
787 static inline bool __is_discard_mergeable(struct discard_info *back,
788 struct discard_info *front, unsigned int max_len)
790 return (back->lstart + back->len == front->lstart) &&
791 (back->len + front->len <= max_len);
794 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
795 struct discard_info *back, unsigned int max_len)
797 return __is_discard_mergeable(back, cur, max_len);
800 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
801 struct discard_info *front, unsigned int max_len)
803 return __is_discard_mergeable(cur, front, max_len);
806 static inline bool __is_extent_mergeable(struct extent_info *back,
807 struct extent_info *front)
809 return (back->fofs + back->len == front->fofs &&
810 back->blk + back->len == front->blk);
813 static inline bool __is_back_mergeable(struct extent_info *cur,
814 struct extent_info *back)
816 return __is_extent_mergeable(back, cur);
819 static inline bool __is_front_mergeable(struct extent_info *cur,
820 struct extent_info *front)
822 return __is_extent_mergeable(cur, front);
825 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
826 static inline void __try_update_largest_extent(struct extent_tree *et,
827 struct extent_node *en)
829 if (en->ei.len > et->largest.len) {
830 et->largest = en->ei;
831 et->largest_updated = true;
836 * For free nid management
839 FREE_NID, /* newly added to free nid list */
840 PREALLOC_NID, /* it is preallocated */
851 struct f2fs_nm_info {
852 block_t nat_blkaddr; /* base disk address of NAT */
853 nid_t max_nid; /* maximum possible node ids */
854 nid_t available_nids; /* # of available node ids */
855 nid_t next_scan_nid; /* the next nid to be scanned */
856 unsigned int ram_thresh; /* control the memory footprint */
857 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
858 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
860 /* NAT cache management */
861 struct radix_tree_root nat_root;/* root of the nat entry cache */
862 struct radix_tree_root nat_set_root;/* root of the nat set cache */
863 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
864 struct list_head nat_entries; /* cached nat entry list (clean) */
865 spinlock_t nat_list_lock; /* protect clean nat entry list */
866 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
867 unsigned int nat_blocks; /* # of nat blocks */
869 /* free node ids management */
870 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
871 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
872 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
873 spinlock_t nid_list_lock; /* protect nid lists ops */
874 struct mutex build_lock; /* lock for build free nids */
875 unsigned char **free_nid_bitmap;
876 unsigned char *nat_block_bitmap;
877 unsigned short *free_nid_count; /* free nid count of NAT block */
880 char *nat_bitmap; /* NAT bitmap pointer */
882 unsigned int nat_bits_blocks; /* # of nat bits blocks */
883 unsigned char *nat_bits; /* NAT bits blocks */
884 unsigned char *full_nat_bits; /* full NAT pages */
885 unsigned char *empty_nat_bits; /* empty NAT pages */
886 #ifdef CONFIG_F2FS_CHECK_FS
887 char *nat_bitmap_mir; /* NAT bitmap mirror */
889 int bitmap_size; /* bitmap size */
893 * this structure is used as one of function parameters.
894 * all the information are dedicated to a given direct node block determined
895 * by the data offset in a file.
897 struct dnode_of_data {
898 struct inode *inode; /* vfs inode pointer */
899 struct page *inode_page; /* its inode page, NULL is possible */
900 struct page *node_page; /* cached direct node page */
901 nid_t nid; /* node id of the direct node block */
902 unsigned int ofs_in_node; /* data offset in the node page */
903 bool inode_page_locked; /* inode page is locked or not */
904 bool node_changed; /* is node block changed */
905 char cur_level; /* level of hole node page */
906 char max_level; /* level of current page located */
907 block_t data_blkaddr; /* block address of the node block */
910 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
911 struct page *ipage, struct page *npage, nid_t nid)
913 memset(dn, 0, sizeof(*dn));
915 dn->inode_page = ipage;
916 dn->node_page = npage;
923 * By default, there are 6 active log areas across the whole main area.
924 * When considering hot and cold data separation to reduce cleaning overhead,
925 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
927 * In the current design, you should not change the numbers intentionally.
928 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
929 * logs individually according to the underlying devices. (default: 6)
930 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
931 * data and 8 for node logs.
933 #define NR_CURSEG_DATA_TYPE (3)
934 #define NR_CURSEG_NODE_TYPE (3)
935 #define NR_CURSEG_INMEM_TYPE (2)
936 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
937 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
940 CURSEG_HOT_DATA = 0, /* directory entry blocks */
941 CURSEG_WARM_DATA, /* data blocks */
942 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
943 CURSEG_HOT_NODE, /* direct node blocks of directory files */
944 CURSEG_WARM_NODE, /* direct node blocks of normal files */
945 CURSEG_COLD_NODE, /* indirect node blocks */
946 NR_PERSISTENT_LOG, /* number of persistent log */
947 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
948 /* pinned file that needs consecutive block address */
949 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
950 NO_CHECK_TYPE, /* number of persistent & inmem log */
954 struct completion wait;
955 struct llist_node llnode;
960 struct flush_cmd_control {
961 struct task_struct *f2fs_issue_flush; /* flush thread */
962 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
963 atomic_t issued_flush; /* # of issued flushes */
964 atomic_t queued_flush; /* # of queued flushes */
965 struct llist_head issue_list; /* list for command issue */
966 struct llist_node *dispatch_list; /* list for command dispatch */
969 struct f2fs_sm_info {
970 struct sit_info *sit_info; /* whole segment information */
971 struct free_segmap_info *free_info; /* free segment information */
972 struct dirty_seglist_info *dirty_info; /* dirty segment information */
973 struct curseg_info *curseg_array; /* active segment information */
975 struct rw_semaphore curseg_lock; /* for preventing curseg change */
977 block_t seg0_blkaddr; /* block address of 0'th segment */
978 block_t main_blkaddr; /* start block address of main area */
979 block_t ssa_blkaddr; /* start block address of SSA area */
981 unsigned int segment_count; /* total # of segments */
982 unsigned int main_segments; /* # of segments in main area */
983 unsigned int reserved_segments; /* # of reserved segments */
984 unsigned int ovp_segments; /* # of overprovision segments */
986 /* a threshold to reclaim prefree segments */
987 unsigned int rec_prefree_segments;
989 /* for batched trimming */
990 unsigned int trim_sections; /* # of sections to trim */
992 struct list_head sit_entry_set; /* sit entry set list */
994 unsigned int ipu_policy; /* in-place-update policy */
995 unsigned int min_ipu_util; /* in-place-update threshold */
996 unsigned int min_fsync_blocks; /* threshold for fsync */
997 unsigned int min_seq_blocks; /* threshold for sequential blocks */
998 unsigned int min_hot_blocks; /* threshold for hot block allocation */
999 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1001 /* for flush command control */
1002 struct flush_cmd_control *fcc_info;
1004 /* for discard command control */
1005 struct discard_cmd_control *dcc_info;
1012 * COUNT_TYPE for monitoring
1014 * f2fs monitors the number of several block types such as on-writeback,
1015 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1017 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1037 * The below are the page types of bios used in submit_bio().
1038 * The available types are:
1039 * DATA User data pages. It operates as async mode.
1040 * NODE Node pages. It operates as async mode.
1041 * META FS metadata pages such as SIT, NAT, CP.
1042 * NR_PAGE_TYPE The number of page types.
1043 * META_FLUSH Make sure the previous pages are written
1044 * with waiting the bio's completion
1045 * ... Only can be used with META.
1047 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1054 INMEM, /* the below types are used by tracepoints only. */
1063 HOT = 0, /* must be zero for meta bio */
1069 enum need_lock_type {
1075 enum cp_reason_type {
1091 APP_DIRECT_IO, /* app direct write IOs */
1092 APP_BUFFERED_IO, /* app buffered write IOs */
1093 APP_WRITE_IO, /* app write IOs */
1094 APP_MAPPED_IO, /* app mapped IOs */
1095 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1096 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1097 FS_META_IO, /* meta IOs from kworker/reclaimer */
1098 FS_GC_DATA_IO, /* data IOs from forground gc */
1099 FS_GC_NODE_IO, /* node IOs from forground gc */
1100 FS_CP_DATA_IO, /* data IOs from checkpoint */
1101 FS_CP_NODE_IO, /* node IOs from checkpoint */
1102 FS_CP_META_IO, /* meta IOs from checkpoint */
1105 APP_DIRECT_READ_IO, /* app direct read IOs */
1106 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1107 APP_READ_IO, /* app read IOs */
1108 APP_MAPPED_READ_IO, /* app mapped read IOs */
1109 FS_DATA_READ_IO, /* data read IOs */
1110 FS_GDATA_READ_IO, /* data read IOs from background gc */
1111 FS_CDATA_READ_IO, /* compressed data read IOs */
1112 FS_NODE_READ_IO, /* node read IOs */
1113 FS_META_READ_IO, /* meta read IOs */
1116 FS_DISCARD, /* discard */
1120 struct f2fs_io_info {
1121 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1122 nid_t ino; /* inode number */
1123 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1124 enum temp_type temp; /* contains HOT/WARM/COLD */
1125 int op; /* contains REQ_OP_ */
1126 int op_flags; /* req_flag_bits */
1127 block_t new_blkaddr; /* new block address to be written */
1128 block_t old_blkaddr; /* old block address before Cow */
1129 struct page *page; /* page to be written */
1130 struct page *encrypted_page; /* encrypted page */
1131 struct page *compressed_page; /* compressed page */
1132 struct list_head list; /* serialize IOs */
1133 bool submitted; /* indicate IO submission */
1134 int need_lock; /* indicate we need to lock cp_rwsem */
1135 bool in_list; /* indicate fio is in io_list */
1136 bool is_por; /* indicate IO is from recovery or not */
1137 bool retry; /* need to reallocate block address */
1138 int compr_blocks; /* # of compressed block addresses */
1139 bool encrypted; /* indicate file is encrypted */
1140 enum iostat_type io_type; /* io type */
1141 struct writeback_control *io_wbc; /* writeback control */
1142 struct bio **bio; /* bio for ipu */
1143 sector_t *last_block; /* last block number in bio */
1144 unsigned char version; /* version of the node */
1149 struct list_head list;
1152 #define is_read_io(rw) ((rw) == READ)
1153 struct f2fs_bio_info {
1154 struct f2fs_sb_info *sbi; /* f2fs superblock */
1155 struct bio *bio; /* bios to merge */
1156 sector_t last_block_in_bio; /* last block number */
1157 struct f2fs_io_info fio; /* store buffered io info. */
1158 struct rw_semaphore io_rwsem; /* blocking op for bio */
1159 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1160 struct list_head io_list; /* track fios */
1161 struct list_head bio_list; /* bio entry list head */
1162 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1165 #define FDEV(i) (sbi->devs[i])
1166 #define RDEV(i) (raw_super->devs[i])
1167 struct f2fs_dev_info {
1168 struct block_device *bdev;
1169 char path[MAX_PATH_LEN];
1170 unsigned int total_segments;
1173 #ifdef CONFIG_BLK_DEV_ZONED
1174 unsigned int nr_blkz; /* Total number of zones */
1175 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1176 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */
1181 DIR_INODE, /* for dirty dir inode */
1182 FILE_INODE, /* for dirty regular/symlink inode */
1183 DIRTY_META, /* for all dirtied inode metadata */
1184 ATOMIC_FILE, /* for all atomic files */
1188 /* for inner inode cache management */
1189 struct inode_management {
1190 struct radix_tree_root ino_root; /* ino entry array */
1191 spinlock_t ino_lock; /* for ino entry lock */
1192 struct list_head ino_list; /* inode list head */
1193 unsigned long ino_num; /* number of entries */
1197 struct atgc_management {
1198 bool atgc_enabled; /* ATGC is enabled or not */
1199 struct rb_root_cached root; /* root of victim rb-tree */
1200 struct list_head victim_list; /* linked with all victim entries */
1201 unsigned int victim_count; /* victim count in rb-tree */
1202 unsigned int candidate_ratio; /* candidate ratio */
1203 unsigned int max_candidate_count; /* max candidate count */
1204 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1205 unsigned long long age_threshold; /* age threshold */
1208 /* For s_flag in struct f2fs_sb_info */
1210 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1211 SBI_IS_CLOSE, /* specify unmounting */
1212 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1213 SBI_POR_DOING, /* recovery is doing or not */
1214 SBI_NEED_SB_WRITE, /* need to recover superblock */
1215 SBI_NEED_CP, /* need to checkpoint */
1216 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1217 SBI_IS_RECOVERED, /* recovered orphan/data */
1218 SBI_CP_DISABLED, /* CP was disabled last mount */
1219 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1220 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1221 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1222 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1223 SBI_IS_RESIZEFS, /* resizefs is in process */
1232 UMOUNT_DISCARD_TIMEOUT,
1246 BGGC_MODE_ON, /* background gc is on */
1247 BGGC_MODE_OFF, /* background gc is off */
1249 * background gc is on, migrating blocks
1250 * like foreground gc
1255 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1256 FS_MODE_LFS, /* use lfs allocation only */
1260 WHINT_MODE_OFF, /* not pass down write hints */
1261 WHINT_MODE_USER, /* try to pass down hints given by users */
1262 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1266 ALLOC_MODE_DEFAULT, /* stay default */
1267 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1271 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1272 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1273 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1278 * automatically compress compression
1282 * automatical compression is disabled.
1283 * user can control the file compression
1289 * this value is set in page as a private data which indicate that
1290 * the page is atomically written, and it is in inmem_pages list.
1292 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1293 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1295 #define IS_ATOMIC_WRITTEN_PAGE(page) \
1296 (page_private(page) == ATOMIC_WRITTEN_PAGE)
1297 #define IS_DUMMY_WRITTEN_PAGE(page) \
1298 (page_private(page) == DUMMY_WRITTEN_PAGE)
1300 #ifdef CONFIG_F2FS_IO_TRACE
1301 #define IS_IO_TRACED_PAGE(page) \
1302 (page_private(page) > 0 && \
1303 page_private(page) < (unsigned long)PID_MAX_LIMIT)
1305 #define IS_IO_TRACED_PAGE(page) (0)
1308 /* For compression */
1309 enum compress_algorithm_type {
1317 enum compress_flag {
1322 #define COMPRESS_DATA_RESERVED_SIZE 4
1323 struct compress_data {
1324 __le32 clen; /* compressed data size */
1325 __le32 chksum; /* compressed data chksum */
1326 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1327 u8 cdata[]; /* compressed data */
1330 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1332 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1334 #define COMPRESS_LEVEL_OFFSET 8
1336 /* compress context */
1337 struct compress_ctx {
1338 struct inode *inode; /* inode the context belong to */
1339 pgoff_t cluster_idx; /* cluster index number */
1340 unsigned int cluster_size; /* page count in cluster */
1341 unsigned int log_cluster_size; /* log of cluster size */
1342 struct page **rpages; /* pages store raw data in cluster */
1343 unsigned int nr_rpages; /* total page number in rpages */
1344 struct page **cpages; /* pages store compressed data in cluster */
1345 unsigned int nr_cpages; /* total page number in cpages */
1346 void *rbuf; /* virtual mapped address on rpages */
1347 struct compress_data *cbuf; /* virtual mapped address on cpages */
1348 size_t rlen; /* valid data length in rbuf */
1349 size_t clen; /* valid data length in cbuf */
1350 void *private; /* payload buffer for specified compression algorithm */
1351 void *private2; /* extra payload buffer */
1354 /* compress context for write IO path */
1355 struct compress_io_ctx {
1356 u32 magic; /* magic number to indicate page is compressed */
1357 struct inode *inode; /* inode the context belong to */
1358 struct page **rpages; /* pages store raw data in cluster */
1359 unsigned int nr_rpages; /* total page number in rpages */
1360 atomic_t pending_pages; /* in-flight compressed page count */
1363 /* Context for decompressing one cluster on the read IO path */
1364 struct decompress_io_ctx {
1365 u32 magic; /* magic number to indicate page is compressed */
1366 struct inode *inode; /* inode the context belong to */
1367 pgoff_t cluster_idx; /* cluster index number */
1368 unsigned int cluster_size; /* page count in cluster */
1369 unsigned int log_cluster_size; /* log of cluster size */
1370 struct page **rpages; /* pages store raw data in cluster */
1371 unsigned int nr_rpages; /* total page number in rpages */
1372 struct page **cpages; /* pages store compressed data in cluster */
1373 unsigned int nr_cpages; /* total page number in cpages */
1374 struct page **tpages; /* temp pages to pad holes in cluster */
1375 void *rbuf; /* virtual mapped address on rpages */
1376 struct compress_data *cbuf; /* virtual mapped address on cpages */
1377 size_t rlen; /* valid data length in rbuf */
1378 size_t clen; /* valid data length in cbuf */
1381 * The number of compressed pages remaining to be read in this cluster.
1382 * This is initially nr_cpages. It is decremented by 1 each time a page
1383 * has been read (or failed to be read). When it reaches 0, the cluster
1384 * is decompressed (or an error is reported).
1386 * If an error occurs before all the pages have been submitted for I/O,
1387 * then this will never reach 0. In this case the I/O submitter is
1388 * responsible for calling f2fs_decompress_end_io() instead.
1390 atomic_t remaining_pages;
1393 * Number of references to this decompress_io_ctx.
1395 * One reference is held for I/O completion. This reference is dropped
1396 * after the pagecache pages are updated and unlocked -- either after
1397 * decompression (and verity if enabled), or after an error.
1399 * In addition, each compressed page holds a reference while it is in a
1400 * bio. These references are necessary prevent compressed pages from
1401 * being freed while they are still in a bio.
1405 bool failed; /* IO error occurred before decompression? */
1406 bool need_verity; /* need fs-verity verification after decompression? */
1407 void *private; /* payload buffer for specified decompression algorithm */
1408 void *private2; /* extra payload buffer */
1409 struct work_struct verity_work; /* work to verify the decompressed pages */
1412 #define NULL_CLUSTER ((unsigned int)(~0))
1413 #define MIN_COMPRESS_LOG_SIZE 2
1414 #define MAX_COMPRESS_LOG_SIZE 8
1415 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1417 struct f2fs_sb_info {
1418 struct super_block *sb; /* pointer to VFS super block */
1419 struct proc_dir_entry *s_proc; /* proc entry */
1420 struct f2fs_super_block *raw_super; /* raw super block pointer */
1421 struct rw_semaphore sb_lock; /* lock for raw super block */
1422 int valid_super_block; /* valid super block no */
1423 unsigned long s_flag; /* flags for sbi */
1424 struct mutex writepages; /* mutex for writepages() */
1426 #ifdef CONFIG_BLK_DEV_ZONED
1427 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1428 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1431 /* for node-related operations */
1432 struct f2fs_nm_info *nm_info; /* node manager */
1433 struct inode *node_inode; /* cache node blocks */
1435 /* for segment-related operations */
1436 struct f2fs_sm_info *sm_info; /* segment manager */
1438 /* for bio operations */
1439 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1440 /* keep migration IO order for LFS mode */
1441 struct rw_semaphore io_order_lock;
1442 mempool_t *write_io_dummy; /* Dummy pages */
1444 /* for checkpoint */
1445 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1446 int cur_cp_pack; /* remain current cp pack */
1447 spinlock_t cp_lock; /* for flag in ckpt */
1448 struct inode *meta_inode; /* cache meta blocks */
1449 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */
1450 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1451 struct rw_semaphore node_write; /* locking node writes */
1452 struct rw_semaphore node_change; /* locking node change */
1453 wait_queue_head_t cp_wait;
1454 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1455 long interval_time[MAX_TIME]; /* to store thresholds */
1456 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1458 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1460 spinlock_t fsync_node_lock; /* for node entry lock */
1461 struct list_head fsync_node_list; /* node list head */
1462 unsigned int fsync_seg_id; /* sequence id */
1463 unsigned int fsync_node_num; /* number of node entries */
1465 /* for orphan inode, use 0'th array */
1466 unsigned int max_orphans; /* max orphan inodes */
1468 /* for inode management */
1469 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1470 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1471 struct mutex flush_lock; /* for flush exclusion */
1473 /* for extent tree cache */
1474 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1475 struct mutex extent_tree_lock; /* locking extent radix tree */
1476 struct list_head extent_list; /* lru list for shrinker */
1477 spinlock_t extent_lock; /* locking extent lru list */
1478 atomic_t total_ext_tree; /* extent tree count */
1479 struct list_head zombie_list; /* extent zombie tree list */
1480 atomic_t total_zombie_tree; /* extent zombie tree count */
1481 atomic_t total_ext_node; /* extent info count */
1483 /* basic filesystem units */
1484 unsigned int log_sectors_per_block; /* log2 sectors per block */
1485 unsigned int log_blocksize; /* log2 block size */
1486 unsigned int blocksize; /* block size */
1487 unsigned int root_ino_num; /* root inode number*/
1488 unsigned int node_ino_num; /* node inode number*/
1489 unsigned int meta_ino_num; /* meta inode number*/
1490 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1491 unsigned int blocks_per_seg; /* blocks per segment */
1492 unsigned int segs_per_sec; /* segments per section */
1493 unsigned int secs_per_zone; /* sections per zone */
1494 unsigned int total_sections; /* total section count */
1495 unsigned int total_node_count; /* total node block count */
1496 unsigned int total_valid_node_count; /* valid node block count */
1497 int dir_level; /* directory level */
1498 int readdir_ra; /* readahead inode in readdir */
1499 u64 max_io_bytes; /* max io bytes to merge IOs */
1501 block_t user_block_count; /* # of user blocks */
1502 block_t total_valid_block_count; /* # of valid blocks */
1503 block_t discard_blks; /* discard command candidats */
1504 block_t last_valid_block_count; /* for recovery */
1505 block_t reserved_blocks; /* configurable reserved blocks */
1506 block_t current_reserved_blocks; /* current reserved blocks */
1508 /* Additional tracking for no checkpoint mode */
1509 block_t unusable_block_count; /* # of blocks saved by last cp */
1511 unsigned int nquota_files; /* # of quota sysfile */
1512 struct rw_semaphore quota_sem; /* blocking cp for flags */
1514 /* # of pages, see count_type */
1515 atomic_t nr_pages[NR_COUNT_TYPE];
1516 /* # of allocated blocks */
1517 struct percpu_counter alloc_valid_block_count;
1519 /* writeback control */
1520 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1522 /* valid inode count */
1523 struct percpu_counter total_valid_inode_count;
1525 struct f2fs_mount_info mount_opt; /* mount options */
1527 /* for cleaning operations */
1528 struct rw_semaphore gc_lock; /*
1529 * semaphore for GC, avoid
1530 * race between GC and GC or CP
1532 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1533 struct atgc_management am; /* atgc management */
1534 unsigned int cur_victim_sec; /* current victim section num */
1535 unsigned int gc_mode; /* current GC state */
1536 unsigned int next_victim_seg[2]; /* next segment in victim section */
1538 /* for skip statistic */
1539 unsigned int atomic_files; /* # of opened atomic file */
1540 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1541 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1543 /* threshold for gc trials on pinned files */
1544 u64 gc_pin_file_threshold;
1545 struct rw_semaphore pin_sem;
1547 /* maximum # of trials to find a victim segment for SSR and GC */
1548 unsigned int max_victim_search;
1549 /* migration granularity of garbage collection, unit: segment */
1550 unsigned int migration_granularity;
1553 * for stat information.
1554 * one is for the LFS mode, and the other is for the SSR mode.
1556 #ifdef CONFIG_F2FS_STAT_FS
1557 struct f2fs_stat_info *stat_info; /* FS status information */
1558 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1559 unsigned int segment_count[2]; /* # of allocated segments */
1560 unsigned int block_count[2]; /* # of allocated blocks */
1561 atomic_t inplace_count; /* # of inplace update */
1562 atomic64_t total_hit_ext; /* # of lookup extent cache */
1563 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1564 atomic64_t read_hit_largest; /* # of hit largest extent node */
1565 atomic64_t read_hit_cached; /* # of hit cached extent node */
1566 atomic_t inline_xattr; /* # of inline_xattr inodes */
1567 atomic_t inline_inode; /* # of inline_data inodes */
1568 atomic_t inline_dir; /* # of inline_dentry inodes */
1569 atomic_t compr_inode; /* # of compressed inodes */
1570 atomic64_t compr_blocks; /* # of compressed blocks */
1571 atomic_t vw_cnt; /* # of volatile writes */
1572 atomic_t max_aw_cnt; /* max # of atomic writes */
1573 atomic_t max_vw_cnt; /* max # of volatile writes */
1574 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1575 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1576 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1578 spinlock_t stat_lock; /* lock for stat operations */
1580 /* For app/fs IO statistics */
1581 spinlock_t iostat_lock;
1582 unsigned long long rw_iostat[NR_IO_TYPE];
1583 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1585 unsigned long iostat_next_period;
1586 unsigned int iostat_period_ms;
1588 /* to attach REQ_META|REQ_FUA flags */
1589 unsigned int data_io_flag;
1590 unsigned int node_io_flag;
1592 /* For sysfs suppport */
1593 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1594 struct completion s_kobj_unregister;
1596 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1597 struct completion s_stat_kobj_unregister;
1599 /* For shrinker support */
1600 struct list_head s_list;
1601 int s_ndevs; /* number of devices */
1602 struct f2fs_dev_info *devs; /* for device list */
1603 unsigned int dirty_device; /* for checkpoint data flush */
1604 spinlock_t dev_lock; /* protect dirty_device */
1605 struct mutex umount_mutex;
1606 unsigned int shrinker_run_no;
1608 /* For write statistics */
1609 u64 sectors_written_start;
1612 /* Reference to checksum algorithm driver via cryptoapi */
1613 struct crypto_shash *s_chksum_driver;
1615 /* Precomputed FS UUID checksum for seeding other checksums */
1616 __u32 s_chksum_seed;
1618 struct workqueue_struct *post_read_wq; /* post read workqueue */
1620 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1621 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1623 #ifdef CONFIG_F2FS_FS_COMPRESSION
1624 struct kmem_cache *page_array_slab; /* page array entry */
1625 unsigned int page_array_slab_size; /* default page array slab size */
1629 struct f2fs_private_dio {
1630 struct inode *inode;
1632 bio_end_io_t *orig_end_io;
1636 #ifdef CONFIG_F2FS_FAULT_INJECTION
1637 #define f2fs_show_injection_info(sbi, type) \
1638 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1639 KERN_INFO, sbi->sb->s_id, \
1640 f2fs_fault_name[type], \
1641 __func__, __builtin_return_address(0))
1642 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1644 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1646 if (!ffi->inject_rate)
1649 if (!IS_FAULT_SET(ffi, type))
1652 atomic_inc(&ffi->inject_ops);
1653 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1654 atomic_set(&ffi->inject_ops, 0);
1660 #define f2fs_show_injection_info(sbi, type) do { } while (0)
1661 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1668 * Test if the mounted volume is a multi-device volume.
1669 * - For a single regular disk volume, sbi->s_ndevs is 0.
1670 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1671 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1673 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1675 return sbi->s_ndevs > 1;
1678 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1680 unsigned long now = jiffies;
1682 sbi->last_time[type] = now;
1684 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1685 if (type == REQ_TIME) {
1686 sbi->last_time[DISCARD_TIME] = now;
1687 sbi->last_time[GC_TIME] = now;
1691 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1693 unsigned long interval = sbi->interval_time[type] * HZ;
1695 return time_after(jiffies, sbi->last_time[type] + interval);
1698 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1701 unsigned long interval = sbi->interval_time[type] * HZ;
1702 unsigned int wait_ms = 0;
1705 delta = (sbi->last_time[type] + interval) - jiffies;
1707 wait_ms = jiffies_to_msecs(delta);
1715 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1716 const void *address, unsigned int length)
1719 struct shash_desc shash;
1724 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1726 desc.shash.tfm = sbi->s_chksum_driver;
1727 *(u32 *)desc.ctx = crc;
1729 err = crypto_shash_update(&desc.shash, address, length);
1732 return *(u32 *)desc.ctx;
1735 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1736 unsigned int length)
1738 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1741 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1742 void *buf, size_t buf_size)
1744 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1747 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1748 const void *address, unsigned int length)
1750 return __f2fs_crc32(sbi, crc, address, length);
1753 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1755 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1758 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1760 return sb->s_fs_info;
1763 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1765 return F2FS_SB(inode->i_sb);
1768 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1770 return F2FS_I_SB(mapping->host);
1773 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1775 return F2FS_M_SB(page_file_mapping(page));
1778 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1780 return (struct f2fs_super_block *)(sbi->raw_super);
1783 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1785 return (struct f2fs_checkpoint *)(sbi->ckpt);
1788 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1790 return (struct f2fs_node *)page_address(page);
1793 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1795 return &((struct f2fs_node *)page_address(page))->i;
1798 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1800 return (struct f2fs_nm_info *)(sbi->nm_info);
1803 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1805 return (struct f2fs_sm_info *)(sbi->sm_info);
1808 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1810 return (struct sit_info *)(SM_I(sbi)->sit_info);
1813 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1815 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1818 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1820 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1823 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1825 return sbi->meta_inode->i_mapping;
1828 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1830 return sbi->node_inode->i_mapping;
1833 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1835 return test_bit(type, &sbi->s_flag);
1838 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1840 set_bit(type, &sbi->s_flag);
1843 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1845 clear_bit(type, &sbi->s_flag);
1848 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1850 return le64_to_cpu(cp->checkpoint_ver);
1853 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1855 if (type < F2FS_MAX_QUOTAS)
1856 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1860 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1862 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1863 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1866 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1868 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1870 return ckpt_flags & f;
1873 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1875 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1878 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1880 unsigned int ckpt_flags;
1882 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1884 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1887 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1889 unsigned long flags;
1891 spin_lock_irqsave(&sbi->cp_lock, flags);
1892 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1893 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1896 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1898 unsigned int ckpt_flags;
1900 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1902 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1905 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1907 unsigned long flags;
1909 spin_lock_irqsave(&sbi->cp_lock, flags);
1910 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1911 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1914 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1916 unsigned long flags;
1917 unsigned char *nat_bits;
1920 * In order to re-enable nat_bits we need to call fsck.f2fs by
1921 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1922 * so let's rely on regular fsck or unclean shutdown.
1926 spin_lock_irqsave(&sbi->cp_lock, flags);
1927 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1928 nat_bits = NM_I(sbi)->nat_bits;
1929 NM_I(sbi)->nat_bits = NULL;
1931 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1936 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1937 struct cp_control *cpc)
1939 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1941 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1944 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1946 down_read(&sbi->cp_rwsem);
1949 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1951 return down_read_trylock(&sbi->cp_rwsem);
1954 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1956 up_read(&sbi->cp_rwsem);
1959 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1961 down_write(&sbi->cp_rwsem);
1964 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1966 up_write(&sbi->cp_rwsem);
1969 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1971 int reason = CP_SYNC;
1973 if (test_opt(sbi, FASTBOOT))
1974 reason = CP_FASTBOOT;
1975 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1980 static inline bool __remain_node_summaries(int reason)
1982 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1985 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1987 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1988 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1992 * Check whether the inode has blocks or not
1994 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1996 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1998 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2001 static inline bool f2fs_has_xattr_block(unsigned int ofs)
2003 return ofs == XATTR_NODE_OFFSET;
2006 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2007 struct inode *inode, bool cap)
2011 if (!test_opt(sbi, RESERVE_ROOT))
2013 if (IS_NOQUOTA(inode))
2015 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2017 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2018 in_group_p(F2FS_OPTION(sbi).s_resgid))
2020 if (cap && capable(CAP_SYS_RESOURCE))
2025 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2026 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2027 struct inode *inode, blkcnt_t *count)
2029 blkcnt_t diff = 0, release = 0;
2030 block_t avail_user_block_count;
2033 ret = dquot_reserve_block(inode, *count);
2037 if (time_to_inject(sbi, FAULT_BLOCK)) {
2038 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2044 * let's increase this in prior to actual block count change in order
2045 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2047 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2049 spin_lock(&sbi->stat_lock);
2050 sbi->total_valid_block_count += (block_t)(*count);
2051 avail_user_block_count = sbi->user_block_count -
2052 sbi->current_reserved_blocks;
2054 if (!__allow_reserved_blocks(sbi, inode, true))
2055 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2056 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2057 if (avail_user_block_count > sbi->unusable_block_count)
2058 avail_user_block_count -= sbi->unusable_block_count;
2060 avail_user_block_count = 0;
2062 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2063 diff = sbi->total_valid_block_count - avail_user_block_count;
2068 sbi->total_valid_block_count -= diff;
2070 spin_unlock(&sbi->stat_lock);
2074 spin_unlock(&sbi->stat_lock);
2076 if (unlikely(release)) {
2077 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2078 dquot_release_reservation_block(inode, release);
2080 f2fs_i_blocks_write(inode, *count, true, true);
2084 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2086 dquot_release_reservation_block(inode, release);
2091 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2093 #define f2fs_err(sbi, fmt, ...) \
2094 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2095 #define f2fs_warn(sbi, fmt, ...) \
2096 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2097 #define f2fs_notice(sbi, fmt, ...) \
2098 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2099 #define f2fs_info(sbi, fmt, ...) \
2100 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2101 #define f2fs_debug(sbi, fmt, ...) \
2102 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2104 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2105 struct inode *inode,
2108 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2110 spin_lock(&sbi->stat_lock);
2111 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2112 sbi->total_valid_block_count -= (block_t)count;
2113 if (sbi->reserved_blocks &&
2114 sbi->current_reserved_blocks < sbi->reserved_blocks)
2115 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2116 sbi->current_reserved_blocks + count);
2117 spin_unlock(&sbi->stat_lock);
2118 if (unlikely(inode->i_blocks < sectors)) {
2119 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2121 (unsigned long long)inode->i_blocks,
2122 (unsigned long long)sectors);
2123 set_sbi_flag(sbi, SBI_NEED_FSCK);
2126 f2fs_i_blocks_write(inode, count, false, true);
2129 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2131 atomic_inc(&sbi->nr_pages[count_type]);
2133 if (count_type == F2FS_DIRTY_DENTS ||
2134 count_type == F2FS_DIRTY_NODES ||
2135 count_type == F2FS_DIRTY_META ||
2136 count_type == F2FS_DIRTY_QDATA ||
2137 count_type == F2FS_DIRTY_IMETA)
2138 set_sbi_flag(sbi, SBI_IS_DIRTY);
2141 static inline void inode_inc_dirty_pages(struct inode *inode)
2143 atomic_inc(&F2FS_I(inode)->dirty_pages);
2144 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2145 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2146 if (IS_NOQUOTA(inode))
2147 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2150 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2152 atomic_dec(&sbi->nr_pages[count_type]);
2155 static inline void inode_dec_dirty_pages(struct inode *inode)
2157 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2158 !S_ISLNK(inode->i_mode))
2161 atomic_dec(&F2FS_I(inode)->dirty_pages);
2162 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2163 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2164 if (IS_NOQUOTA(inode))
2165 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2168 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2170 return atomic_read(&sbi->nr_pages[count_type]);
2173 static inline int get_dirty_pages(struct inode *inode)
2175 return atomic_read(&F2FS_I(inode)->dirty_pages);
2178 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2180 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2181 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2182 sbi->log_blocks_per_seg;
2184 return segs / sbi->segs_per_sec;
2187 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2189 return sbi->total_valid_block_count;
2192 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2194 return sbi->discard_blks;
2197 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2199 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2201 /* return NAT or SIT bitmap */
2202 if (flag == NAT_BITMAP)
2203 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2204 else if (flag == SIT_BITMAP)
2205 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2210 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2212 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2215 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2217 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2220 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2221 offset = (flag == SIT_BITMAP) ?
2222 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2224 * if large_nat_bitmap feature is enabled, leave checksum
2225 * protection for all nat/sit bitmaps.
2227 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2230 if (__cp_payload(sbi) > 0) {
2231 if (flag == NAT_BITMAP)
2232 return &ckpt->sit_nat_version_bitmap;
2234 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2236 offset = (flag == NAT_BITMAP) ?
2237 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2238 return &ckpt->sit_nat_version_bitmap + offset;
2242 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2244 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2246 if (sbi->cur_cp_pack == 2)
2247 start_addr += sbi->blocks_per_seg;
2251 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2253 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2255 if (sbi->cur_cp_pack == 1)
2256 start_addr += sbi->blocks_per_seg;
2260 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2262 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2265 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2267 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2270 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2271 struct inode *inode, bool is_inode)
2273 block_t valid_block_count;
2274 unsigned int valid_node_count, user_block_count;
2279 err = dquot_alloc_inode(inode);
2284 err = dquot_reserve_block(inode, 1);
2289 if (time_to_inject(sbi, FAULT_BLOCK)) {
2290 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2294 spin_lock(&sbi->stat_lock);
2296 valid_block_count = sbi->total_valid_block_count +
2297 sbi->current_reserved_blocks + 1;
2299 if (!__allow_reserved_blocks(sbi, inode, false))
2300 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2301 user_block_count = sbi->user_block_count;
2302 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2303 user_block_count -= sbi->unusable_block_count;
2305 if (unlikely(valid_block_count > user_block_count)) {
2306 spin_unlock(&sbi->stat_lock);
2310 valid_node_count = sbi->total_valid_node_count + 1;
2311 if (unlikely(valid_node_count > sbi->total_node_count)) {
2312 spin_unlock(&sbi->stat_lock);
2316 sbi->total_valid_node_count++;
2317 sbi->total_valid_block_count++;
2318 spin_unlock(&sbi->stat_lock);
2322 f2fs_mark_inode_dirty_sync(inode, true);
2324 f2fs_i_blocks_write(inode, 1, true, true);
2327 percpu_counter_inc(&sbi->alloc_valid_block_count);
2333 dquot_free_inode(inode);
2335 dquot_release_reservation_block(inode, 1);
2340 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2341 struct inode *inode, bool is_inode)
2343 spin_lock(&sbi->stat_lock);
2345 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2346 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2348 sbi->total_valid_node_count--;
2349 sbi->total_valid_block_count--;
2350 if (sbi->reserved_blocks &&
2351 sbi->current_reserved_blocks < sbi->reserved_blocks)
2352 sbi->current_reserved_blocks++;
2354 spin_unlock(&sbi->stat_lock);
2357 dquot_free_inode(inode);
2359 if (unlikely(inode->i_blocks == 0)) {
2360 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2362 (unsigned long long)inode->i_blocks);
2363 set_sbi_flag(sbi, SBI_NEED_FSCK);
2366 f2fs_i_blocks_write(inode, 1, false, true);
2370 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2372 return sbi->total_valid_node_count;
2375 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2377 percpu_counter_inc(&sbi->total_valid_inode_count);
2380 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2382 percpu_counter_dec(&sbi->total_valid_inode_count);
2385 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2387 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2390 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2391 pgoff_t index, bool for_write)
2395 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2397 page = find_get_page_flags(mapping, index,
2398 FGP_LOCK | FGP_ACCESSED);
2400 page = find_lock_page(mapping, index);
2404 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2405 f2fs_show_injection_info(F2FS_M_SB(mapping),
2412 return grab_cache_page(mapping, index);
2413 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2416 static inline struct page *f2fs_pagecache_get_page(
2417 struct address_space *mapping, pgoff_t index,
2418 int fgp_flags, gfp_t gfp_mask)
2420 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2421 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2425 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2428 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2430 char *src_kaddr = kmap(src);
2431 char *dst_kaddr = kmap(dst);
2433 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2438 static inline void f2fs_put_page(struct page *page, int unlock)
2444 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2450 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2453 f2fs_put_page(dn->node_page, 1);
2454 if (dn->inode_page && dn->node_page != dn->inode_page)
2455 f2fs_put_page(dn->inode_page, 0);
2456 dn->node_page = NULL;
2457 dn->inode_page = NULL;
2460 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2463 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2466 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2471 entry = kmem_cache_alloc(cachep, flags);
2473 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2477 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2479 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2480 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2481 get_pages(sbi, F2FS_WB_CP_DATA) ||
2482 get_pages(sbi, F2FS_DIO_READ) ||
2483 get_pages(sbi, F2FS_DIO_WRITE))
2486 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2487 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2490 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2491 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2496 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2498 if (sbi->gc_mode == GC_URGENT_HIGH)
2501 if (is_inflight_io(sbi, type))
2504 if (sbi->gc_mode == GC_URGENT_LOW &&
2505 (type == DISCARD_TIME || type == GC_TIME))
2508 return f2fs_time_over(sbi, type);
2511 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2512 unsigned long index, void *item)
2514 while (radix_tree_insert(root, index, item))
2518 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2520 static inline bool IS_INODE(struct page *page)
2522 struct f2fs_node *p = F2FS_NODE(page);
2524 return RAW_IS_INODE(p);
2527 static inline int offset_in_addr(struct f2fs_inode *i)
2529 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2530 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2533 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2535 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2538 static inline int f2fs_has_extra_attr(struct inode *inode);
2539 static inline block_t data_blkaddr(struct inode *inode,
2540 struct page *node_page, unsigned int offset)
2542 struct f2fs_node *raw_node;
2545 bool is_inode = IS_INODE(node_page);
2547 raw_node = F2FS_NODE(node_page);
2551 /* from GC path only */
2552 base = offset_in_addr(&raw_node->i);
2553 else if (f2fs_has_extra_attr(inode))
2554 base = get_extra_isize(inode);
2557 addr_array = blkaddr_in_node(raw_node);
2558 return le32_to_cpu(addr_array[base + offset]);
2561 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2563 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2566 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2571 mask = 1 << (7 - (nr & 0x07));
2572 return mask & *addr;
2575 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2580 mask = 1 << (7 - (nr & 0x07));
2584 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2589 mask = 1 << (7 - (nr & 0x07));
2593 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2599 mask = 1 << (7 - (nr & 0x07));
2605 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2611 mask = 1 << (7 - (nr & 0x07));
2617 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2622 mask = 1 << (7 - (nr & 0x07));
2627 * On-disk inode flags (f2fs_inode::i_flags)
2629 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2630 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2631 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2632 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2633 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2634 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2635 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2636 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2637 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2638 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2639 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2641 /* Flags that should be inherited by new inodes from their parent. */
2642 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2643 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2644 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2646 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2647 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2650 /* Flags that are appropriate for non-directories/regular files. */
2651 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2653 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2657 else if (S_ISREG(mode))
2658 return flags & F2FS_REG_FLMASK;
2660 return flags & F2FS_OTHER_FLMASK;
2663 static inline void __mark_inode_dirty_flag(struct inode *inode,
2667 case FI_INLINE_XATTR:
2668 case FI_INLINE_DATA:
2669 case FI_INLINE_DENTRY:
2675 case FI_INLINE_DOTS:
2677 f2fs_mark_inode_dirty_sync(inode, true);
2681 static inline void set_inode_flag(struct inode *inode, int flag)
2683 set_bit(flag, F2FS_I(inode)->flags);
2684 __mark_inode_dirty_flag(inode, flag, true);
2687 static inline int is_inode_flag_set(struct inode *inode, int flag)
2689 return test_bit(flag, F2FS_I(inode)->flags);
2692 static inline void clear_inode_flag(struct inode *inode, int flag)
2694 clear_bit(flag, F2FS_I(inode)->flags);
2695 __mark_inode_dirty_flag(inode, flag, false);
2698 static inline bool f2fs_verity_in_progress(struct inode *inode)
2700 return IS_ENABLED(CONFIG_FS_VERITY) &&
2701 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2704 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2706 F2FS_I(inode)->i_acl_mode = mode;
2707 set_inode_flag(inode, FI_ACL_MODE);
2708 f2fs_mark_inode_dirty_sync(inode, false);
2711 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2717 f2fs_mark_inode_dirty_sync(inode, true);
2720 static inline void f2fs_i_blocks_write(struct inode *inode,
2721 block_t diff, bool add, bool claim)
2723 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2724 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2726 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2729 dquot_claim_block(inode, diff);
2731 dquot_alloc_block_nofail(inode, diff);
2733 dquot_free_block(inode, diff);
2736 f2fs_mark_inode_dirty_sync(inode, true);
2737 if (clean || recover)
2738 set_inode_flag(inode, FI_AUTO_RECOVER);
2741 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2743 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2744 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2746 if (i_size_read(inode) == i_size)
2749 i_size_write(inode, i_size);
2750 f2fs_mark_inode_dirty_sync(inode, true);
2751 if (clean || recover)
2752 set_inode_flag(inode, FI_AUTO_RECOVER);
2755 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2757 F2FS_I(inode)->i_current_depth = depth;
2758 f2fs_mark_inode_dirty_sync(inode, true);
2761 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2764 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2765 f2fs_mark_inode_dirty_sync(inode, true);
2768 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2770 F2FS_I(inode)->i_xattr_nid = xnid;
2771 f2fs_mark_inode_dirty_sync(inode, true);
2774 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2776 F2FS_I(inode)->i_pino = pino;
2777 f2fs_mark_inode_dirty_sync(inode, true);
2780 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2782 struct f2fs_inode_info *fi = F2FS_I(inode);
2784 if (ri->i_inline & F2FS_INLINE_XATTR)
2785 set_bit(FI_INLINE_XATTR, fi->flags);
2786 if (ri->i_inline & F2FS_INLINE_DATA)
2787 set_bit(FI_INLINE_DATA, fi->flags);
2788 if (ri->i_inline & F2FS_INLINE_DENTRY)
2789 set_bit(FI_INLINE_DENTRY, fi->flags);
2790 if (ri->i_inline & F2FS_DATA_EXIST)
2791 set_bit(FI_DATA_EXIST, fi->flags);
2792 if (ri->i_inline & F2FS_INLINE_DOTS)
2793 set_bit(FI_INLINE_DOTS, fi->flags);
2794 if (ri->i_inline & F2FS_EXTRA_ATTR)
2795 set_bit(FI_EXTRA_ATTR, fi->flags);
2796 if (ri->i_inline & F2FS_PIN_FILE)
2797 set_bit(FI_PIN_FILE, fi->flags);
2800 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2804 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2805 ri->i_inline |= F2FS_INLINE_XATTR;
2806 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2807 ri->i_inline |= F2FS_INLINE_DATA;
2808 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2809 ri->i_inline |= F2FS_INLINE_DENTRY;
2810 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2811 ri->i_inline |= F2FS_DATA_EXIST;
2812 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2813 ri->i_inline |= F2FS_INLINE_DOTS;
2814 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2815 ri->i_inline |= F2FS_EXTRA_ATTR;
2816 if (is_inode_flag_set(inode, FI_PIN_FILE))
2817 ri->i_inline |= F2FS_PIN_FILE;
2820 static inline int f2fs_has_extra_attr(struct inode *inode)
2822 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2825 static inline int f2fs_has_inline_xattr(struct inode *inode)
2827 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2830 static inline int f2fs_compressed_file(struct inode *inode)
2832 return S_ISREG(inode->i_mode) &&
2833 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2836 static inline bool f2fs_need_compress_data(struct inode *inode)
2838 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
2840 if (!f2fs_compressed_file(inode))
2843 if (compress_mode == COMPR_MODE_FS)
2845 else if (compress_mode == COMPR_MODE_USER &&
2846 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
2852 static inline unsigned int addrs_per_inode(struct inode *inode)
2854 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2855 get_inline_xattr_addrs(inode);
2857 if (!f2fs_compressed_file(inode))
2859 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2862 static inline unsigned int addrs_per_block(struct inode *inode)
2864 if (!f2fs_compressed_file(inode))
2865 return DEF_ADDRS_PER_BLOCK;
2866 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2869 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2871 struct f2fs_inode *ri = F2FS_INODE(page);
2873 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2874 get_inline_xattr_addrs(inode)]);
2877 static inline int inline_xattr_size(struct inode *inode)
2879 if (f2fs_has_inline_xattr(inode))
2880 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2884 static inline int f2fs_has_inline_data(struct inode *inode)
2886 return is_inode_flag_set(inode, FI_INLINE_DATA);
2889 static inline int f2fs_exist_data(struct inode *inode)
2891 return is_inode_flag_set(inode, FI_DATA_EXIST);
2894 static inline int f2fs_has_inline_dots(struct inode *inode)
2896 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2899 static inline int f2fs_is_mmap_file(struct inode *inode)
2901 return is_inode_flag_set(inode, FI_MMAP_FILE);
2904 static inline bool f2fs_is_pinned_file(struct inode *inode)
2906 return is_inode_flag_set(inode, FI_PIN_FILE);
2909 static inline bool f2fs_is_atomic_file(struct inode *inode)
2911 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2914 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2916 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2919 static inline bool f2fs_is_volatile_file(struct inode *inode)
2921 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2924 static inline bool f2fs_is_first_block_written(struct inode *inode)
2926 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2929 static inline bool f2fs_is_drop_cache(struct inode *inode)
2931 return is_inode_flag_set(inode, FI_DROP_CACHE);
2934 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2936 struct f2fs_inode *ri = F2FS_INODE(page);
2937 int extra_size = get_extra_isize(inode);
2939 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2942 static inline int f2fs_has_inline_dentry(struct inode *inode)
2944 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2947 static inline int is_file(struct inode *inode, int type)
2949 return F2FS_I(inode)->i_advise & type;
2952 static inline void set_file(struct inode *inode, int type)
2954 F2FS_I(inode)->i_advise |= type;
2955 f2fs_mark_inode_dirty_sync(inode, true);
2958 static inline void clear_file(struct inode *inode, int type)
2960 F2FS_I(inode)->i_advise &= ~type;
2961 f2fs_mark_inode_dirty_sync(inode, true);
2964 static inline bool f2fs_is_time_consistent(struct inode *inode)
2966 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2968 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2970 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2972 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2973 &F2FS_I(inode)->i_crtime))
2978 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2983 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2985 spin_lock(&sbi->inode_lock[DIRTY_META]);
2986 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2987 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2990 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2991 file_keep_isize(inode) ||
2992 i_size_read(inode) & ~PAGE_MASK)
2995 if (!f2fs_is_time_consistent(inode))
2998 spin_lock(&F2FS_I(inode)->i_size_lock);
2999 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3000 spin_unlock(&F2FS_I(inode)->i_size_lock);
3005 static inline bool f2fs_readonly(struct super_block *sb)
3007 return sb_rdonly(sb);
3010 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3012 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3015 static inline bool is_dot_dotdot(const u8 *name, size_t len)
3017 if (len == 1 && name[0] == '.')
3020 if (len == 2 && name[0] == '.' && name[1] == '.')
3026 static inline bool f2fs_may_extent_tree(struct inode *inode)
3028 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3030 if (!test_opt(sbi, EXTENT_CACHE) ||
3031 is_inode_flag_set(inode, FI_NO_EXTENT) ||
3032 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
3036 * for recovered files during mount do not create extents
3037 * if shrinker is not registered.
3039 if (list_empty(&sbi->s_list))
3042 return S_ISREG(inode->i_mode);
3045 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3046 size_t size, gfp_t flags)
3048 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3049 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3053 return kmalloc(size, flags);
3056 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3057 size_t size, gfp_t flags)
3059 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3062 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3063 size_t size, gfp_t flags)
3065 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3066 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3070 return kvmalloc(size, flags);
3073 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3074 size_t size, gfp_t flags)
3076 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3079 static inline int get_extra_isize(struct inode *inode)
3081 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3084 static inline int get_inline_xattr_addrs(struct inode *inode)
3086 return F2FS_I(inode)->i_inline_xattr_size;
3089 #define f2fs_get_inode_mode(i) \
3090 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3091 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3093 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3094 (offsetof(struct f2fs_inode, i_extra_end) - \
3095 offsetof(struct f2fs_inode, i_extra_isize)) \
3097 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3098 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3099 ((offsetof(typeof(*(f2fs_inode)), field) + \
3100 sizeof((f2fs_inode)->field)) \
3101 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3103 #define DEFAULT_IOSTAT_PERIOD_MS 3000
3104 #define MIN_IOSTAT_PERIOD_MS 100
3105 /* maximum period of iostat tracing is 1 day */
3106 #define MAX_IOSTAT_PERIOD_MS 8640000
3108 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3112 spin_lock(&sbi->iostat_lock);
3113 for (i = 0; i < NR_IO_TYPE; i++) {
3114 sbi->rw_iostat[i] = 0;
3115 sbi->prev_rw_iostat[i] = 0;
3117 spin_unlock(&sbi->iostat_lock);
3120 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3122 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3123 enum iostat_type type, unsigned long long io_bytes)
3125 if (!sbi->iostat_enable)
3127 spin_lock(&sbi->iostat_lock);
3128 sbi->rw_iostat[type] += io_bytes;
3130 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3131 sbi->rw_iostat[APP_BUFFERED_IO] =
3132 sbi->rw_iostat[APP_WRITE_IO] -
3133 sbi->rw_iostat[APP_DIRECT_IO];
3135 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3136 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3137 sbi->rw_iostat[APP_READ_IO] -
3138 sbi->rw_iostat[APP_DIRECT_READ_IO];
3139 spin_unlock(&sbi->iostat_lock);
3141 f2fs_record_iostat(sbi);
3144 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3146 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3148 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3149 block_t blkaddr, int type);
3150 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3151 block_t blkaddr, int type)
3153 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3154 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3156 f2fs_bug_on(sbi, 1);
3160 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3162 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3163 blkaddr == COMPRESS_ADDR)
3168 static inline void f2fs_set_page_private(struct page *page,
3171 if (PagePrivate(page))
3174 attach_page_private(page, (void *)data);
3177 static inline void f2fs_clear_page_private(struct page *page)
3179 detach_page_private(page);
3185 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3186 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3187 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3188 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3189 int f2fs_truncate(struct inode *inode);
3190 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path,
3191 struct kstat *stat, u32 request_mask, unsigned int flags);
3192 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
3193 struct iattr *attr);
3194 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3195 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3196 int f2fs_precache_extents(struct inode *inode);
3197 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3198 int f2fs_fileattr_set(struct user_namespace *mnt_userns,
3199 struct dentry *dentry, struct fileattr *fa);
3200 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3201 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3202 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3203 int f2fs_pin_file_control(struct inode *inode, bool inc);
3208 void f2fs_set_inode_flags(struct inode *inode);
3209 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3210 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3211 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3212 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3213 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3214 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3215 void f2fs_update_inode_page(struct inode *inode);
3216 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3217 void f2fs_evict_inode(struct inode *inode);
3218 void f2fs_handle_failed_inode(struct inode *inode);
3223 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3224 bool hot, bool set);
3225 struct dentry *f2fs_get_parent(struct dentry *child);
3230 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3231 int f2fs_init_casefolded_name(const struct inode *dir,
3232 struct f2fs_filename *fname);
3233 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3234 int lookup, struct f2fs_filename *fname);
3235 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3236 struct f2fs_filename *fname);
3237 void f2fs_free_filename(struct f2fs_filename *fname);
3238 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3239 const struct f2fs_filename *fname, int *max_slots);
3240 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3241 unsigned int start_pos, struct fscrypt_str *fstr);
3242 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3243 struct f2fs_dentry_ptr *d);
3244 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3245 const struct f2fs_filename *fname, struct page *dpage);
3246 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3247 unsigned int current_depth);
3248 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3249 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3250 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3251 const struct f2fs_filename *fname,
3252 struct page **res_page);
3253 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3254 const struct qstr *child, struct page **res_page);
3255 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3256 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3257 struct page **page);
3258 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3259 struct page *page, struct inode *inode);
3260 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3261 const struct f2fs_filename *fname);
3262 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3263 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3264 unsigned int bit_pos);
3265 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3266 struct inode *inode, nid_t ino, umode_t mode);
3267 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3268 struct inode *inode, nid_t ino, umode_t mode);
3269 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3270 struct inode *inode, nid_t ino, umode_t mode);
3271 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3272 struct inode *dir, struct inode *inode);
3273 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3274 bool f2fs_empty_dir(struct inode *dir);
3276 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3278 if (fscrypt_is_nokey_name(dentry))
3280 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3281 inode, inode->i_ino, inode->i_mode);
3287 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3288 void f2fs_inode_synced(struct inode *inode);
3289 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3290 int f2fs_quota_sync(struct super_block *sb, int type);
3291 loff_t max_file_blocks(struct inode *inode);
3292 void f2fs_quota_off_umount(struct super_block *sb);
3293 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3294 int f2fs_sync_fs(struct super_block *sb, int sync);
3295 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3300 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3305 struct dnode_of_data;
3308 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3309 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3310 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3311 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3312 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3313 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3314 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3315 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3316 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3317 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3318 struct node_info *ni);
3319 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3320 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3321 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3322 int f2fs_truncate_xattr_node(struct inode *inode);
3323 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3324 unsigned int seq_id);
3325 int f2fs_remove_inode_page(struct inode *inode);
3326 struct page *f2fs_new_inode_page(struct inode *inode);
3327 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3328 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3329 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3330 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3331 int f2fs_move_node_page(struct page *node_page, int gc_type);
3332 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3333 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3334 struct writeback_control *wbc, bool atomic,
3335 unsigned int *seq_id);
3336 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3337 struct writeback_control *wbc,
3338 bool do_balance, enum iostat_type io_type);
3339 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3340 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3341 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3342 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3343 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3344 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3345 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3346 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3347 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3348 unsigned int segno, struct f2fs_summary_block *sum);
3349 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3350 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3351 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3352 int __init f2fs_create_node_manager_caches(void);
3353 void f2fs_destroy_node_manager_caches(void);
3358 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3359 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3360 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3361 void f2fs_drop_inmem_pages(struct inode *inode);
3362 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3363 int f2fs_commit_inmem_pages(struct inode *inode);
3364 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3365 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3366 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3367 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3368 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3369 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3370 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3371 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3372 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3373 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3374 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3375 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3376 struct cp_control *cpc);
3377 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3378 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3379 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3380 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3381 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3382 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3383 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3384 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3385 void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3386 unsigned int *newseg, bool new_sec, int dir);
3387 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3388 unsigned int start, unsigned int end);
3389 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type);
3390 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3391 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3392 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3393 struct cp_control *cpc);
3394 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3395 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3397 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3398 enum iostat_type io_type);
3399 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3400 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3401 struct f2fs_io_info *fio);
3402 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3403 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3404 block_t old_blkaddr, block_t new_blkaddr,
3405 bool recover_curseg, bool recover_newaddr,
3407 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3408 block_t old_addr, block_t new_addr,
3409 unsigned char version, bool recover_curseg,
3410 bool recover_newaddr);
3411 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3412 block_t old_blkaddr, block_t *new_blkaddr,
3413 struct f2fs_summary *sum, int type,
3414 struct f2fs_io_info *fio);
3415 void f2fs_wait_on_page_writeback(struct page *page,
3416 enum page_type type, bool ordered, bool locked);
3417 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3418 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3420 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3421 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3422 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3423 unsigned int val, int alloc);
3424 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3425 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3426 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3427 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3428 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3429 int __init f2fs_create_segment_manager_caches(void);
3430 void f2fs_destroy_segment_manager_caches(void);
3431 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3432 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3433 enum page_type type, enum temp_type temp);
3434 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3435 unsigned int segno);
3436 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3437 unsigned int segno);
3442 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3443 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3444 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3445 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3446 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3447 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3448 block_t blkaddr, int type);
3449 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3450 int type, bool sync);
3451 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3452 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3453 long nr_to_write, enum iostat_type io_type);
3454 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3455 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3456 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3457 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3458 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3459 unsigned int devidx, int type);
3460 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3461 unsigned int devidx, int type);
3462 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3463 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3464 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3465 void f2fs_add_orphan_inode(struct inode *inode);
3466 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3467 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3468 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3469 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3470 void f2fs_remove_dirty_inode(struct inode *inode);
3471 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3472 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3473 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3474 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3475 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3476 int __init f2fs_create_checkpoint_caches(void);
3477 void f2fs_destroy_checkpoint_caches(void);
3478 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3479 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3480 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3481 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3486 int __init f2fs_init_bioset(void);
3487 void f2fs_destroy_bioset(void);
3488 int f2fs_init_bio_entry_cache(void);
3489 void f2fs_destroy_bio_entry_cache(void);
3490 void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3491 struct bio *bio, enum page_type type);
3492 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3493 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3494 struct inode *inode, struct page *page,
3495 nid_t ino, enum page_type type);
3496 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3497 struct bio **bio, struct page *page);
3498 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3499 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3500 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3501 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3502 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3503 block_t blk_addr, struct bio *bio);
3504 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3505 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3506 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3507 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3508 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3509 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3510 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3511 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3512 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3513 int op_flags, bool for_write);
3514 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3515 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3517 struct page *f2fs_get_new_data_page(struct inode *inode,
3518 struct page *ipage, pgoff_t index, bool new_i_size);
3519 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3520 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3521 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3522 int create, int flag);
3523 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3524 u64 start, u64 len);
3525 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3526 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3527 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3528 int f2fs_write_single_data_page(struct page *page, int *submitted,
3529 struct bio **bio, sector_t *last_block,
3530 struct writeback_control *wbc,
3531 enum iostat_type io_type,
3532 int compr_blocks, bool allow_balance);
3533 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3534 unsigned int length);
3535 int f2fs_release_page(struct page *page, gfp_t wait);
3536 #ifdef CONFIG_MIGRATION
3537 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3538 struct page *page, enum migrate_mode mode);
3540 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3541 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3542 int f2fs_init_post_read_processing(void);
3543 void f2fs_destroy_post_read_processing(void);
3544 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3545 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3550 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3551 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3552 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3553 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3554 unsigned int segno);
3555 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3556 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3557 int __init f2fs_create_garbage_collection_cache(void);
3558 void f2fs_destroy_garbage_collection_cache(void);
3563 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3564 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3569 #ifdef CONFIG_F2FS_STAT_FS
3570 struct f2fs_stat_info {
3571 struct list_head stat_list;
3572 struct f2fs_sb_info *sbi;
3573 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3574 int main_area_segs, main_area_sections, main_area_zones;
3575 unsigned long long hit_largest, hit_cached, hit_rbtree;
3576 unsigned long long hit_total, total_ext;
3577 int ext_tree, zombie_tree, ext_node;
3578 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3579 int ndirty_data, ndirty_qdata;
3581 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3582 int nats, dirty_nats, sits, dirty_sits;
3583 int free_nids, avail_nids, alloc_nids;
3584 int total_count, utilization;
3585 int bg_gc, nr_wb_cp_data, nr_wb_data;
3586 int nr_rd_data, nr_rd_node, nr_rd_meta;
3587 int nr_dio_read, nr_dio_write;
3588 unsigned int io_skip_bggc, other_skip_bggc;
3589 int nr_flushing, nr_flushed, flush_list_empty;
3590 int nr_discarding, nr_discarded;
3592 unsigned int undiscard_blks;
3593 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3594 unsigned int cur_ckpt_time, peak_ckpt_time;
3595 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3597 unsigned long long compr_blocks;
3598 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3599 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3600 unsigned int bimodal, avg_vblocks;
3601 int util_free, util_valid, util_invalid;
3602 int rsvd_segs, overp_segs;
3603 int dirty_count, node_pages, meta_pages;
3604 int prefree_count, call_count, cp_count, bg_cp_count;
3605 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3606 int bg_node_segs, bg_data_segs;
3607 int tot_blks, data_blks, node_blks;
3608 int bg_data_blks, bg_node_blks;
3609 unsigned long long skipped_atomic_files[2];
3610 int curseg[NR_CURSEG_TYPE];
3611 int cursec[NR_CURSEG_TYPE];
3612 int curzone[NR_CURSEG_TYPE];
3613 unsigned int dirty_seg[NR_CURSEG_TYPE];
3614 unsigned int full_seg[NR_CURSEG_TYPE];
3615 unsigned int valid_blks[NR_CURSEG_TYPE];
3617 unsigned int meta_count[META_MAX];
3618 unsigned int segment_count[2];
3619 unsigned int block_count[2];
3620 unsigned int inplace_count;
3621 unsigned long long base_mem, cache_mem, page_mem;
3624 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3626 return (struct f2fs_stat_info *)sbi->stat_info;
3629 #define stat_inc_cp_count(si) ((si)->cp_count++)
3630 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3631 #define stat_inc_call_count(si) ((si)->call_count++)
3632 #define stat_inc_bggc_count(si) ((si)->bg_gc++)
3633 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3634 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3635 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3636 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3637 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3638 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3639 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3640 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3641 #define stat_inc_inline_xattr(inode) \
3643 if (f2fs_has_inline_xattr(inode)) \
3644 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3646 #define stat_dec_inline_xattr(inode) \
3648 if (f2fs_has_inline_xattr(inode)) \
3649 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3651 #define stat_inc_inline_inode(inode) \
3653 if (f2fs_has_inline_data(inode)) \
3654 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3656 #define stat_dec_inline_inode(inode) \
3658 if (f2fs_has_inline_data(inode)) \
3659 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3661 #define stat_inc_inline_dir(inode) \
3663 if (f2fs_has_inline_dentry(inode)) \
3664 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3666 #define stat_dec_inline_dir(inode) \
3668 if (f2fs_has_inline_dentry(inode)) \
3669 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3671 #define stat_inc_compr_inode(inode) \
3673 if (f2fs_compressed_file(inode)) \
3674 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3676 #define stat_dec_compr_inode(inode) \
3678 if (f2fs_compressed_file(inode)) \
3679 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3681 #define stat_add_compr_blocks(inode, blocks) \
3682 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3683 #define stat_sub_compr_blocks(inode, blocks) \
3684 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3685 #define stat_inc_meta_count(sbi, blkaddr) \
3687 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3688 atomic_inc(&(sbi)->meta_count[META_CP]); \
3689 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3690 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3691 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3692 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3693 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3694 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3696 #define stat_inc_seg_type(sbi, curseg) \
3697 ((sbi)->segment_count[(curseg)->alloc_type]++)
3698 #define stat_inc_block_count(sbi, curseg) \
3699 ((sbi)->block_count[(curseg)->alloc_type]++)
3700 #define stat_inc_inplace_blocks(sbi) \
3701 (atomic_inc(&(sbi)->inplace_count))
3702 #define stat_update_max_atomic_write(inode) \
3704 int cur = F2FS_I_SB(inode)->atomic_files; \
3705 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3707 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3709 #define stat_inc_volatile_write(inode) \
3710 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3711 #define stat_dec_volatile_write(inode) \
3712 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3713 #define stat_update_max_volatile_write(inode) \
3715 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3716 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3718 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3720 #define stat_inc_seg_count(sbi, type, gc_type) \
3722 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3724 if ((type) == SUM_TYPE_DATA) { \
3726 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3729 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3733 #define stat_inc_tot_blk_count(si, blks) \
3734 ((si)->tot_blks += (blks))
3736 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3738 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3739 stat_inc_tot_blk_count(si, blks); \
3740 si->data_blks += (blks); \
3741 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3744 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3746 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3747 stat_inc_tot_blk_count(si, blks); \
3748 si->node_blks += (blks); \
3749 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3752 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3753 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3754 void __init f2fs_create_root_stats(void);
3755 void f2fs_destroy_root_stats(void);
3756 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3758 #define stat_inc_cp_count(si) do { } while (0)
3759 #define stat_inc_bg_cp_count(si) do { } while (0)
3760 #define stat_inc_call_count(si) do { } while (0)
3761 #define stat_inc_bggc_count(si) do { } while (0)
3762 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3763 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3764 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3765 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3766 #define stat_inc_total_hit(sbi) do { } while (0)
3767 #define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3768 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3769 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3770 #define stat_inc_inline_xattr(inode) do { } while (0)
3771 #define stat_dec_inline_xattr(inode) do { } while (0)
3772 #define stat_inc_inline_inode(inode) do { } while (0)
3773 #define stat_dec_inline_inode(inode) do { } while (0)
3774 #define stat_inc_inline_dir(inode) do { } while (0)
3775 #define stat_dec_inline_dir(inode) do { } while (0)
3776 #define stat_inc_compr_inode(inode) do { } while (0)
3777 #define stat_dec_compr_inode(inode) do { } while (0)
3778 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
3779 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3780 #define stat_update_max_atomic_write(inode) do { } while (0)
3781 #define stat_inc_volatile_write(inode) do { } while (0)
3782 #define stat_dec_volatile_write(inode) do { } while (0)
3783 #define stat_update_max_volatile_write(inode) do { } while (0)
3784 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3785 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3786 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3787 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3788 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3789 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3790 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3791 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3793 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3794 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3795 static inline void __init f2fs_create_root_stats(void) { }
3796 static inline void f2fs_destroy_root_stats(void) { }
3797 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3800 extern const struct file_operations f2fs_dir_operations;
3801 extern const struct file_operations f2fs_file_operations;
3802 extern const struct inode_operations f2fs_file_inode_operations;
3803 extern const struct address_space_operations f2fs_dblock_aops;
3804 extern const struct address_space_operations f2fs_node_aops;
3805 extern const struct address_space_operations f2fs_meta_aops;
3806 extern const struct inode_operations f2fs_dir_inode_operations;
3807 extern const struct inode_operations f2fs_symlink_inode_operations;
3808 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3809 extern const struct inode_operations f2fs_special_inode_operations;
3810 extern struct kmem_cache *f2fs_inode_entry_slab;
3815 bool f2fs_may_inline_data(struct inode *inode);
3816 bool f2fs_may_inline_dentry(struct inode *inode);
3817 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3818 void f2fs_truncate_inline_inode(struct inode *inode,
3819 struct page *ipage, u64 from);
3820 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3821 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3822 int f2fs_convert_inline_inode(struct inode *inode);
3823 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3824 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3825 int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3826 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3827 const struct f2fs_filename *fname,
3828 struct page **res_page);
3829 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3830 struct page *ipage);
3831 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3832 struct inode *inode, nid_t ino, umode_t mode);
3833 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3834 struct page *page, struct inode *dir,
3835 struct inode *inode);
3836 bool f2fs_empty_inline_dir(struct inode *dir);
3837 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3838 struct fscrypt_str *fstr);
3839 int f2fs_inline_data_fiemap(struct inode *inode,
3840 struct fiemap_extent_info *fieinfo,
3841 __u64 start, __u64 len);
3846 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3847 struct shrink_control *sc);
3848 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3849 struct shrink_control *sc);
3850 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3851 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3856 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3857 struct rb_entry *cached_re, unsigned int ofs);
3858 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
3859 struct rb_root_cached *root,
3860 struct rb_node **parent,
3861 unsigned long long key, bool *left_most);
3862 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3863 struct rb_root_cached *root,
3864 struct rb_node **parent,
3865 unsigned int ofs, bool *leftmost);
3866 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3867 struct rb_entry *cached_re, unsigned int ofs,
3868 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3869 struct rb_node ***insert_p, struct rb_node **insert_parent,
3870 bool force, bool *leftmost);
3871 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3872 struct rb_root_cached *root, bool check_key);
3873 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3874 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3875 void f2fs_drop_extent_tree(struct inode *inode);
3876 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3877 void f2fs_destroy_extent_tree(struct inode *inode);
3878 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3879 struct extent_info *ei);
3880 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3881 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3882 pgoff_t fofs, block_t blkaddr, unsigned int len);
3883 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3884 int __init f2fs_create_extent_cache(void);
3885 void f2fs_destroy_extent_cache(void);
3890 int __init f2fs_init_sysfs(void);
3891 void f2fs_exit_sysfs(void);
3892 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3893 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3896 extern const struct fsverity_operations f2fs_verityops;
3901 static inline bool f2fs_encrypted_file(struct inode *inode)
3903 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3906 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3908 #ifdef CONFIG_FS_ENCRYPTION
3909 file_set_encrypt(inode);
3910 f2fs_set_inode_flags(inode);
3915 * Returns true if the reads of the inode's data need to undergo some
3916 * postprocessing step, like decryption or authenticity verification.
3918 static inline bool f2fs_post_read_required(struct inode *inode)
3920 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3921 f2fs_compressed_file(inode);
3927 #ifdef CONFIG_F2FS_FS_COMPRESSION
3928 bool f2fs_is_compressed_page(struct page *page);
3929 struct page *f2fs_compress_control_page(struct page *page);
3930 int f2fs_prepare_compress_overwrite(struct inode *inode,
3931 struct page **pagep, pgoff_t index, void **fsdata);
3932 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3933 pgoff_t index, unsigned copied);
3934 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3935 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3936 bool f2fs_is_compress_backend_ready(struct inode *inode);
3937 int f2fs_init_compress_mempool(void);
3938 void f2fs_destroy_compress_mempool(void);
3939 void f2fs_end_read_compressed_page(struct page *page, bool failed);
3940 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3941 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3942 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3943 int f2fs_write_multi_pages(struct compress_ctx *cc,
3945 struct writeback_control *wbc,
3946 enum iostat_type io_type);
3947 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3948 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3949 unsigned nr_pages, sector_t *last_block_in_bio,
3950 bool is_readahead, bool for_write);
3951 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3952 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed);
3953 void f2fs_put_page_dic(struct page *page);
3954 int f2fs_init_compress_ctx(struct compress_ctx *cc);
3955 void f2fs_destroy_compress_ctx(struct compress_ctx *cc);
3956 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3957 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
3958 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
3959 int __init f2fs_init_compress_cache(void);
3960 void f2fs_destroy_compress_cache(void);
3962 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3963 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3965 if (!f2fs_compressed_file(inode))
3967 /* not support compression */
3970 static inline struct page *f2fs_compress_control_page(struct page *page)
3973 return ERR_PTR(-EINVAL);
3975 static inline int f2fs_init_compress_mempool(void) { return 0; }
3976 static inline void f2fs_destroy_compress_mempool(void) { }
3977 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed)
3981 static inline void f2fs_put_page_dic(struct page *page)
3985 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
3986 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
3987 static inline int __init f2fs_init_compress_cache(void) { return 0; }
3988 static inline void f2fs_destroy_compress_cache(void) { }
3991 static inline void set_compress_context(struct inode *inode)
3993 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3995 F2FS_I(inode)->i_compress_algorithm =
3996 F2FS_OPTION(sbi).compress_algorithm;
3997 F2FS_I(inode)->i_log_cluster_size =
3998 F2FS_OPTION(sbi).compress_log_size;
3999 F2FS_I(inode)->i_compress_flag =
4000 F2FS_OPTION(sbi).compress_chksum ?
4001 1 << COMPRESS_CHKSUM : 0;
4002 F2FS_I(inode)->i_cluster_size =
4003 1 << F2FS_I(inode)->i_log_cluster_size;
4004 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 &&
4005 F2FS_OPTION(sbi).compress_level)
4006 F2FS_I(inode)->i_compress_flag |=
4007 F2FS_OPTION(sbi).compress_level <<
4008 COMPRESS_LEVEL_OFFSET;
4009 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4010 set_inode_flag(inode, FI_COMPRESSED_FILE);
4011 stat_inc_compr_inode(inode);
4012 f2fs_mark_inode_dirty_sync(inode, true);
4015 static inline bool f2fs_disable_compressed_file(struct inode *inode)
4017 struct f2fs_inode_info *fi = F2FS_I(inode);
4019 if (!f2fs_compressed_file(inode))
4021 if (S_ISREG(inode->i_mode) &&
4022 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks)))
4025 fi->i_flags &= ~F2FS_COMPR_FL;
4026 stat_dec_compr_inode(inode);
4027 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4028 f2fs_mark_inode_dirty_sync(inode, true);
4032 #define F2FS_FEATURE_FUNCS(name, flagname) \
4033 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4035 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4038 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4039 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4040 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4041 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4042 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4043 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4044 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4045 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4046 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4047 F2FS_FEATURE_FUNCS(verity, VERITY);
4048 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4049 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4050 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4052 #ifdef CONFIG_BLK_DEV_ZONED
4053 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4056 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4058 return test_bit(zno, FDEV(devi).blkz_seq);
4062 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4064 return f2fs_sb_has_blkzoned(sbi);
4067 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4069 return blk_queue_discard(bdev_get_queue(bdev)) ||
4070 bdev_is_zoned(bdev);
4073 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4077 if (!f2fs_is_multi_device(sbi))
4078 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4080 for (i = 0; i < sbi->s_ndevs; i++)
4081 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4086 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4088 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4089 f2fs_hw_should_discard(sbi);
4092 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4096 if (!f2fs_is_multi_device(sbi))
4097 return bdev_read_only(sbi->sb->s_bdev);
4099 for (i = 0; i < sbi->s_ndevs; i++)
4100 if (bdev_read_only(FDEV(i).bdev))
4105 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4107 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4110 static inline bool f2fs_may_compress(struct inode *inode)
4112 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4113 f2fs_is_atomic_file(inode) ||
4114 f2fs_is_volatile_file(inode))
4116 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4119 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4120 u64 blocks, bool add)
4122 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4123 struct f2fs_inode_info *fi = F2FS_I(inode);
4125 /* don't update i_compr_blocks if saved blocks were released */
4126 if (!add && !atomic_read(&fi->i_compr_blocks))
4130 atomic_add(diff, &fi->i_compr_blocks);
4131 stat_add_compr_blocks(inode, diff);
4133 atomic_sub(diff, &fi->i_compr_blocks);
4134 stat_sub_compr_blocks(inode, diff);
4136 f2fs_mark_inode_dirty_sync(inode, true);
4139 static inline int block_unaligned_IO(struct inode *inode,
4140 struct kiocb *iocb, struct iov_iter *iter)
4142 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4143 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4144 loff_t offset = iocb->ki_pos;
4145 unsigned long align = offset | iov_iter_alignment(iter);
4147 return align & blocksize_mask;
4150 static inline int allow_outplace_dio(struct inode *inode,
4151 struct kiocb *iocb, struct iov_iter *iter)
4153 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4154 int rw = iov_iter_rw(iter);
4156 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4157 !block_unaligned_IO(inode, iocb, iter));
4160 static inline bool f2fs_force_buffered_io(struct inode *inode,
4161 struct kiocb *iocb, struct iov_iter *iter)
4163 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4164 int rw = iov_iter_rw(iter);
4166 if (f2fs_post_read_required(inode))
4168 if (f2fs_is_multi_device(sbi))
4171 * for blkzoned device, fallback direct IO to buffered IO, so
4172 * all IOs can be serialized by log-structured write.
4174 if (f2fs_sb_has_blkzoned(sbi))
4176 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4177 if (block_unaligned_IO(inode, iocb, iter))
4179 if (F2FS_IO_ALIGNED(sbi))
4182 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4183 !IS_SWAPFILE(inode))
4189 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4191 return fsverity_active(inode) &&
4192 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4195 #ifdef CONFIG_F2FS_FAULT_INJECTION
4196 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4199 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4202 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4205 if (f2fs_sb_has_quota_ino(sbi))
4207 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4208 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4209 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4215 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4216 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4218 #endif /* _LINUX_F2FS_H */