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/sched/mm.h>
22 #include <linux/vmalloc.h>
23 #include <linux/bio.h>
24 #include <linux/blkdev.h>
25 #include <linux/quotaops.h>
26 #include <linux/part_stat.h>
27 #include <crypto/hash.h>
29 #include <linux/fscrypt.h>
30 #include <linux/fsverity.h>
34 #ifdef CONFIG_F2FS_CHECK_FS
35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
37 #define f2fs_bug_on(sbi, condition) \
39 if (WARN_ON(condition)) \
40 set_sbi_flag(sbi, SBI_NEED_FSCK); \
49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */
67 #ifdef CONFIG_F2FS_FAULT_INJECTION
68 #define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0))
70 struct f2fs_fault_info {
72 unsigned int inject_rate;
73 unsigned int inject_type;
76 extern const char *f2fs_fault_name[FAULT_MAX];
77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))
79 /* maximum retry count for injected failure */
80 #define DEFAULT_FAILURE_RETRY_COUNT 8
82 #define DEFAULT_FAILURE_RETRY_COUNT 1
88 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001
89 #define F2FS_MOUNT_DISCARD 0x00000002
90 #define F2FS_MOUNT_NOHEAP 0x00000004
91 #define F2FS_MOUNT_XATTR_USER 0x00000008
92 #define F2FS_MOUNT_POSIX_ACL 0x00000010
93 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020
94 #define F2FS_MOUNT_INLINE_XATTR 0x00000040
95 #define F2FS_MOUNT_INLINE_DATA 0x00000080
96 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100
97 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200
98 #define F2FS_MOUNT_NOBARRIER 0x00000400
99 #define F2FS_MOUNT_FASTBOOT 0x00000800
100 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000
101 #define F2FS_MOUNT_DATA_FLUSH 0x00002000
102 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000
103 #define F2FS_MOUNT_USRQUOTA 0x00008000
104 #define F2FS_MOUNT_GRPQUOTA 0x00010000
105 #define F2FS_MOUNT_PRJQUOTA 0x00020000
106 #define F2FS_MOUNT_QUOTA 0x00040000
107 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000
108 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000
109 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000
110 #define F2FS_MOUNT_NORECOVERY 0x00400000
111 #define F2FS_MOUNT_ATGC 0x00800000
112 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000
113 #define F2FS_MOUNT_GC_MERGE 0x02000000
114 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000
115 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000
117 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
118 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
119 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
120 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
122 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
123 typecheck(unsigned long long, b) && \
124 ((long long)((a) - (b)) > 0))
126 typedef u32 block_t; /*
127 * should not change u32, since it is the on-disk block
128 * address format, __le32.
132 #define COMPRESS_EXT_NUM 16
135 * An implementation of an rwsem that is explicitly unfair to readers. This
136 * prevents priority inversion when a low-priority reader acquires the read lock
137 * while sleeping on the write lock but the write lock is needed by
138 * higher-priority clients.
142 struct rw_semaphore internal_rwsem;
143 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
144 wait_queue_head_t read_waiters;
148 struct f2fs_mount_info {
150 int write_io_size_bits; /* Write IO size bits */
151 block_t root_reserved_blocks; /* root reserved blocks */
152 kuid_t s_resuid; /* reserved blocks for uid */
153 kgid_t s_resgid; /* reserved blocks for gid */
154 int active_logs; /* # of active logs */
155 int inline_xattr_size; /* inline xattr size */
156 #ifdef CONFIG_F2FS_FAULT_INJECTION
157 struct f2fs_fault_info fault_info; /* For fault injection */
160 /* Names of quota files with journalled quota */
161 char *s_qf_names[MAXQUOTAS];
162 int s_jquota_fmt; /* Format of quota to use */
164 /* For which write hints are passed down to block layer */
165 int alloc_mode; /* segment allocation policy */
166 int fsync_mode; /* fsync policy */
167 int fs_mode; /* fs mode: LFS or ADAPTIVE */
168 int bggc_mode; /* bggc mode: off, on or sync */
169 int memory_mode; /* memory mode */
170 int errors; /* errors parameter */
172 * discard command's offset/size should
173 * be aligned to this unit: block,
176 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
177 block_t unusable_cap_perc; /* percentage for cap */
178 block_t unusable_cap; /* Amount of space allowed to be
179 * unusable when disabling checkpoint
182 /* For compression */
183 unsigned char compress_algorithm; /* algorithm type */
184 unsigned char compress_log_size; /* cluster log size */
185 unsigned char compress_level; /* compress level */
186 bool compress_chksum; /* compressed data chksum */
187 unsigned char compress_ext_cnt; /* extension count */
188 unsigned char nocompress_ext_cnt; /* nocompress extension count */
189 int compress_mode; /* compression mode */
190 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
191 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
194 #define F2FS_FEATURE_ENCRYPT 0x00000001
195 #define F2FS_FEATURE_BLKZONED 0x00000002
196 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004
197 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008
198 #define F2FS_FEATURE_PRJQUOTA 0x00000010
199 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020
200 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040
201 #define F2FS_FEATURE_QUOTA_INO 0x00000080
202 #define F2FS_FEATURE_INODE_CRTIME 0x00000100
203 #define F2FS_FEATURE_LOST_FOUND 0x00000200
204 #define F2FS_FEATURE_VERITY 0x00000400
205 #define F2FS_FEATURE_SB_CHKSUM 0x00000800
206 #define F2FS_FEATURE_CASEFOLD 0x00001000
207 #define F2FS_FEATURE_COMPRESSION 0x00002000
208 #define F2FS_FEATURE_RO 0x00004000
210 #define __F2FS_HAS_FEATURE(raw_super, mask) \
211 ((raw_super->feature & cpu_to_le32(mask)) != 0)
212 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
215 * Default values for user and/or group using reserved blocks
217 #define F2FS_DEF_RESUID 0
218 #define F2FS_DEF_RESGID 0
221 * For checkpoint manager
228 #define CP_UMOUNT 0x00000001
229 #define CP_FASTBOOT 0x00000002
230 #define CP_SYNC 0x00000004
231 #define CP_RECOVERY 0x00000008
232 #define CP_DISCARD 0x00000010
233 #define CP_TRIMMED 0x00000020
234 #define CP_PAUSE 0x00000040
235 #define CP_RESIZE 0x00000080
237 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
238 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
239 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
240 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
241 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
242 #define DEF_CP_INTERVAL 60 /* 60 secs */
243 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
244 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
245 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
246 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
256 * indicate meta/data type
265 DATA_GENERIC, /* check range only */
266 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
267 DATA_GENERIC_ENHANCE_READ, /*
268 * strong check on range and segment
269 * bitmap but no warning due to race
270 * condition of read on truncated area
273 DATA_GENERIC_ENHANCE_UPDATE, /*
274 * strong check on range and segment
275 * bitmap for update case
280 /* for the list of ino */
282 ORPHAN_INO, /* for orphan ino list */
283 APPEND_INO, /* for append ino list */
284 UPDATE_INO, /* for update ino list */
285 TRANS_DIR_INO, /* for transactions dir ino list */
286 FLUSH_INO, /* for multiple device flushing */
287 MAX_INO_ENTRY, /* max. list */
291 struct list_head list; /* list head */
292 nid_t ino; /* inode number */
293 unsigned int dirty_device; /* dirty device bitmap */
296 /* for the list of inodes to be GCed */
298 struct list_head list; /* list head */
299 struct inode *inode; /* vfs inode pointer */
302 struct fsync_node_entry {
303 struct list_head list; /* list head */
304 struct page *page; /* warm node page pointer */
305 unsigned int seq_id; /* sequence id */
309 struct completion wait; /* completion for checkpoint done */
310 struct llist_node llnode; /* llist_node to be linked in wait queue */
311 int ret; /* return code of checkpoint */
312 ktime_t queue_time; /* request queued time */
315 struct ckpt_req_control {
316 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
317 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
318 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
319 atomic_t issued_ckpt; /* # of actually issued ckpts */
320 atomic_t total_ckpt; /* # of total ckpts */
321 atomic_t queued_ckpt; /* # of queued ckpts */
322 struct llist_head issue_list; /* list for command issue */
323 spinlock_t stat_lock; /* lock for below checkpoint time stats */
324 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
325 unsigned int peak_time; /* peak wait time in msec until now */
328 /* for the bitmap indicate blocks to be discarded */
329 struct discard_entry {
330 struct list_head list; /* list head */
331 block_t start_blkaddr; /* start blockaddr of current segment */
332 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
335 /* minimum discard granularity, unit: block count */
336 #define MIN_DISCARD_GRANULARITY 1
337 /* default discard granularity of inner discard thread, unit: block count */
338 #define DEFAULT_DISCARD_GRANULARITY 16
339 /* default maximum discard granularity of ordered discard, unit: block count */
340 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16
342 /* max discard pend list number */
343 #define MAX_PLIST_NUM 512
344 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
345 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
348 D_PREP, /* initial */
349 D_PARTIAL, /* partially submitted */
350 D_SUBMIT, /* all submitted */
351 D_DONE, /* finished */
354 struct discard_info {
355 block_t lstart; /* logical start address */
356 block_t len; /* length */
357 block_t start; /* actual start address in dev */
361 struct rb_node rb_node; /* rb node located in rb-tree */
362 struct discard_info di; /* discard info */
363 struct list_head list; /* command list */
364 struct completion wait; /* compleation */
365 struct block_device *bdev; /* bdev */
366 unsigned short ref; /* reference count */
367 unsigned char state; /* state */
368 unsigned char queued; /* queued discard */
369 int error; /* bio error */
370 spinlock_t lock; /* for state/bio_ref updating */
371 unsigned short bio_ref; /* bio reference count */
383 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */
384 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */
385 DPOLICY_IO_AWARE_MAX,
388 struct discard_policy {
389 int type; /* type of discard */
390 unsigned int min_interval; /* used for candidates exist */
391 unsigned int mid_interval; /* used for device busy */
392 unsigned int max_interval; /* used for candidates not exist */
393 unsigned int max_requests; /* # of discards issued per round */
394 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
395 bool io_aware; /* issue discard in idle time */
396 bool sync; /* submit discard with REQ_SYNC flag */
397 bool ordered; /* issue discard by lba order */
398 bool timeout; /* discard timeout for put_super */
399 unsigned int granularity; /* discard granularity */
402 struct discard_cmd_control {
403 struct task_struct *f2fs_issue_discard; /* discard thread */
404 struct list_head entry_list; /* 4KB discard entry list */
405 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
406 struct list_head wait_list; /* store on-flushing entries */
407 struct list_head fstrim_list; /* in-flight discard from fstrim */
408 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
409 struct mutex cmd_lock;
410 unsigned int nr_discards; /* # of discards in the list */
411 unsigned int max_discards; /* max. discards to be issued */
412 unsigned int max_discard_request; /* max. discard request per round */
413 unsigned int min_discard_issue_time; /* min. interval between discard issue */
414 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */
415 unsigned int max_discard_issue_time; /* max. interval between discard issue */
416 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */
417 unsigned int discard_urgent_util; /* utilization which issue discard proactively */
418 unsigned int discard_granularity; /* discard granularity */
419 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */
420 unsigned int discard_io_aware; /* io_aware policy */
421 unsigned int undiscard_blks; /* # of undiscard blocks */
422 unsigned int next_pos; /* next discard position */
423 atomic_t issued_discard; /* # of issued discard */
424 atomic_t queued_discard; /* # of queued discard */
425 atomic_t discard_cmd_cnt; /* # of cached cmd count */
426 struct rb_root_cached root; /* root of discard rb-tree */
427 bool rbtree_check; /* config for consistence check */
428 bool discard_wake; /* to wake up discard thread */
431 /* for the list of fsync inodes, used only during recovery */
432 struct fsync_inode_entry {
433 struct list_head list; /* list head */
434 struct inode *inode; /* vfs inode pointer */
435 block_t blkaddr; /* block address locating the last fsync */
436 block_t last_dentry; /* block address locating the last dentry */
439 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
440 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
442 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
443 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
444 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
445 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
447 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
448 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
450 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
452 int before = nats_in_cursum(journal);
454 journal->n_nats = cpu_to_le16(before + i);
458 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
460 int before = sits_in_cursum(journal);
462 journal->n_sits = cpu_to_le16(before + i);
466 static inline bool __has_cursum_space(struct f2fs_journal *journal,
469 if (type == NAT_JOURNAL)
470 return size <= MAX_NAT_JENTRIES(journal);
471 return size <= MAX_SIT_JENTRIES(journal);
474 /* for inline stuff */
475 #define DEF_INLINE_RESERVED_SIZE 1
476 static inline int get_extra_isize(struct inode *inode);
477 static inline int get_inline_xattr_addrs(struct inode *inode);
478 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
479 (CUR_ADDRS_PER_INODE(inode) - \
480 get_inline_xattr_addrs(inode) - \
481 DEF_INLINE_RESERVED_SIZE))
484 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
485 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
487 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
488 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
489 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
490 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
491 NR_INLINE_DENTRY(inode) + \
492 INLINE_DENTRY_BITMAP_SIZE(inode)))
495 * For INODE and NODE manager
497 /* for directory operations */
499 struct f2fs_filename {
501 * The filename the user specified. This is NULL for some
502 * filesystem-internal operations, e.g. converting an inline directory
503 * to a non-inline one, or roll-forward recovering an encrypted dentry.
505 const struct qstr *usr_fname;
508 * The on-disk filename. For encrypted directories, this is encrypted.
509 * This may be NULL for lookups in an encrypted dir without the key.
511 struct fscrypt_str disk_name;
513 /* The dirhash of this filename */
516 #ifdef CONFIG_FS_ENCRYPTION
518 * For lookups in encrypted directories: either the buffer backing
519 * disk_name, or a buffer that holds the decoded no-key name.
521 struct fscrypt_str crypto_buf;
523 #if IS_ENABLED(CONFIG_UNICODE)
525 * For casefolded directories: the casefolded name, but it's left NULL
526 * if the original name is not valid Unicode, if the original name is
527 * "." or "..", if the directory is both casefolded and encrypted and
528 * its encryption key is unavailable, or if the filesystem is doing an
529 * internal operation where usr_fname is also NULL. In all these cases
530 * we fall back to treating the name as an opaque byte sequence.
532 struct fscrypt_str cf_name;
536 struct f2fs_dentry_ptr {
539 struct f2fs_dir_entry *dentry;
540 __u8 (*filename)[F2FS_SLOT_LEN];
545 static inline void make_dentry_ptr_block(struct inode *inode,
546 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
549 d->max = NR_DENTRY_IN_BLOCK;
550 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
551 d->bitmap = t->dentry_bitmap;
552 d->dentry = t->dentry;
553 d->filename = t->filename;
556 static inline void make_dentry_ptr_inline(struct inode *inode,
557 struct f2fs_dentry_ptr *d, void *t)
559 int entry_cnt = NR_INLINE_DENTRY(inode);
560 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
561 int reserved_size = INLINE_RESERVED_SIZE(inode);
565 d->nr_bitmap = bitmap_size;
567 d->dentry = t + bitmap_size + reserved_size;
568 d->filename = t + bitmap_size + reserved_size +
569 SIZE_OF_DIR_ENTRY * entry_cnt;
573 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
574 * as its node offset to distinguish from index node blocks.
575 * But some bits are used to mark the node block.
577 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
580 ALLOC_NODE, /* allocate a new node page if needed */
581 LOOKUP_NODE, /* look up a node without readahead */
583 * look up a node with readahead called
588 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
590 /* congestion wait timeout value, default: 20ms */
591 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
593 /* maximum retry quota flush count */
594 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
596 /* maximum retry of EIO'ed page */
597 #define MAX_RETRY_PAGE_EIO 100
599 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
601 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
603 /* dirty segments threshold for triggering CP */
604 #define DEFAULT_DIRTY_THRESHOLD 4
606 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS
607 #define RECOVERY_MIN_RA_BLOCKS 1
609 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */
611 /* for in-memory extent cache entry */
612 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
614 /* number of extent info in extent cache we try to shrink */
615 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128
617 /* number of age extent info in extent cache we try to shrink */
618 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128
619 #define LAST_AGE_WEIGHT 30
620 #define SAME_AGE_REGION 1024
623 * Define data block with age less than 1GB as hot data
624 * define data block with age less than 10GB but more than 1GB as warm data
626 #define DEF_HOT_DATA_AGE_THRESHOLD 262144
627 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440
629 /* extent cache type */
637 unsigned int fofs; /* start offset in a file */
638 unsigned int len; /* length of the extent */
640 /* read extent_cache */
642 /* start block address of the extent */
644 #ifdef CONFIG_F2FS_FS_COMPRESSION
645 /* physical extent length of compressed blocks */
649 /* block age extent_cache */
651 /* block age of the extent */
652 unsigned long long age;
653 /* last total blocks allocated */
654 unsigned long long last_blocks;
660 struct rb_node rb_node; /* rb node located in rb-tree */
661 struct extent_info ei; /* extent info */
662 struct list_head list; /* node in global extent list of sbi */
663 struct extent_tree *et; /* extent tree pointer */
667 nid_t ino; /* inode number */
668 enum extent_type type; /* keep the extent tree type */
669 struct rb_root_cached root; /* root of extent info rb-tree */
670 struct extent_node *cached_en; /* recently accessed extent node */
671 struct list_head list; /* to be used by sbi->zombie_list */
672 rwlock_t lock; /* protect extent info rb-tree */
673 atomic_t node_cnt; /* # of extent node in rb-tree*/
674 bool largest_updated; /* largest extent updated */
675 struct extent_info largest; /* largest cached extent for EX_READ */
678 struct extent_tree_info {
679 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
680 struct mutex extent_tree_lock; /* locking extent radix tree */
681 struct list_head extent_list; /* lru list for shrinker */
682 spinlock_t extent_lock; /* locking extent lru list */
683 atomic_t total_ext_tree; /* extent tree count */
684 struct list_head zombie_list; /* extent zombie tree list */
685 atomic_t total_zombie_tree; /* extent zombie tree count */
686 atomic_t total_ext_node; /* extent info count */
690 * State of block returned by f2fs_map_blocks.
692 #define F2FS_MAP_NEW (1U << 0)
693 #define F2FS_MAP_MAPPED (1U << 1)
694 #define F2FS_MAP_DELALLOC (1U << 2)
695 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
698 struct f2fs_map_blocks {
699 struct block_device *m_bdev; /* for multi-device dio */
703 unsigned int m_flags;
704 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
705 pgoff_t *m_next_extent; /* point to next possible extent */
707 bool m_may_create; /* indicate it is from write path */
708 bool m_multidev_dio; /* indicate it allows multi-device dio */
711 /* for flag in get_data_block */
713 F2FS_GET_BLOCK_DEFAULT,
714 F2FS_GET_BLOCK_FIEMAP,
717 F2FS_GET_BLOCK_PRE_DIO,
718 F2FS_GET_BLOCK_PRE_AIO,
719 F2FS_GET_BLOCK_PRECACHE,
723 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
725 #define FADVISE_COLD_BIT 0x01
726 #define FADVISE_LOST_PINO_BIT 0x02
727 #define FADVISE_ENCRYPT_BIT 0x04
728 #define FADVISE_ENC_NAME_BIT 0x08
729 #define FADVISE_KEEP_SIZE_BIT 0x10
730 #define FADVISE_HOT_BIT 0x20
731 #define FADVISE_VERITY_BIT 0x40
732 #define FADVISE_TRUNC_BIT 0x80
734 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
736 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
737 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
738 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
740 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
741 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
742 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
744 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
745 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
747 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
748 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
750 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
751 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
753 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
754 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
755 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
757 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
758 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
760 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)
761 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)
762 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)
764 #define DEF_DIR_LEVEL 0
771 /* used for f2fs_inode_info->flags */
773 FI_NEW_INODE, /* indicate newly allocated inode */
774 FI_DIRTY_INODE, /* indicate inode is dirty or not */
775 FI_AUTO_RECOVER, /* indicate inode is recoverable */
776 FI_DIRTY_DIR, /* indicate directory has dirty pages */
777 FI_INC_LINK, /* need to increment i_nlink */
778 FI_ACL_MODE, /* indicate acl mode */
779 FI_NO_ALLOC, /* should not allocate any blocks */
780 FI_FREE_NID, /* free allocated nide */
781 FI_NO_EXTENT, /* not to use the extent cache */
782 FI_INLINE_XATTR, /* used for inline xattr */
783 FI_INLINE_DATA, /* used for inline data*/
784 FI_INLINE_DENTRY, /* used for inline dentry */
785 FI_APPEND_WRITE, /* inode has appended data */
786 FI_UPDATE_WRITE, /* inode has in-place-update data */
787 FI_NEED_IPU, /* used for ipu per file */
788 FI_ATOMIC_FILE, /* indicate atomic file */
789 FI_DATA_EXIST, /* indicate data exists */
790 FI_INLINE_DOTS, /* indicate inline dot dentries */
791 FI_SKIP_WRITES, /* should skip data page writeback */
792 FI_OPU_WRITE, /* used for opu per file */
793 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
794 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */
795 FI_HOT_DATA, /* indicate file is hot */
796 FI_EXTRA_ATTR, /* indicate file has extra attribute */
797 FI_PROJ_INHERIT, /* indicate file inherits projectid */
798 FI_PIN_FILE, /* indicate file should not be gced */
799 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
800 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
801 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
802 FI_MMAP_FILE, /* indicate file was mmapped */
803 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
804 FI_COMPRESS_RELEASED, /* compressed blocks were released */
805 FI_ALIGNED_WRITE, /* enable aligned write */
806 FI_COW_FILE, /* indicate COW file */
807 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */
808 FI_ATOMIC_REPLACE, /* indicate atomic replace */
809 FI_MAX, /* max flag, never be used */
812 struct f2fs_inode_info {
813 struct inode vfs_inode; /* serve a vfs inode */
814 unsigned long i_flags; /* keep an inode flags for ioctl */
815 unsigned char i_advise; /* use to give file attribute hints */
816 unsigned char i_dir_level; /* use for dentry level for large dir */
817 unsigned int i_current_depth; /* only for directory depth */
818 /* for gc failure statistic */
819 unsigned int i_gc_failures[MAX_GC_FAILURE];
820 unsigned int i_pino; /* parent inode number */
821 umode_t i_acl_mode; /* keep file acl mode temporarily */
823 /* Use below internally in f2fs*/
824 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
825 struct f2fs_rwsem i_sem; /* protect fi info */
826 atomic_t dirty_pages; /* # of dirty pages */
827 f2fs_hash_t chash; /* hash value of given file name */
828 unsigned int clevel; /* maximum level of given file name */
829 struct task_struct *task; /* lookup and create consistency */
830 struct task_struct *cp_task; /* separate cp/wb IO stats*/
831 struct task_struct *wb_task; /* indicate inode is in context of writeback */
832 nid_t i_xattr_nid; /* node id that contains xattrs */
833 loff_t last_disk_size; /* lastly written file size */
834 spinlock_t i_size_lock; /* protect last_disk_size */
837 struct dquot __rcu *i_dquot[MAXQUOTAS];
839 /* quota space reservation, managed internally by quota code */
840 qsize_t i_reserved_quota;
842 struct list_head dirty_list; /* dirty list for dirs and files */
843 struct list_head gdirty_list; /* linked in global dirty list */
844 struct task_struct *atomic_write_task; /* store atomic write task */
845 struct extent_tree *extent_tree[NR_EXTENT_CACHES];
846 /* cached extent_tree entry */
847 struct inode *cow_inode; /* copy-on-write inode for atomic write */
849 /* avoid racing between foreground op and gc */
850 struct f2fs_rwsem i_gc_rwsem[2];
851 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */
853 int i_extra_isize; /* size of extra space located in i_addr */
854 kprojid_t i_projid; /* id for project quota */
855 int i_inline_xattr_size; /* inline xattr size */
856 struct timespec64 i_crtime; /* inode creation time */
857 struct timespec64 i_disk_time[3];/* inode disk times */
859 /* for file compress */
860 atomic_t i_compr_blocks; /* # of compressed blocks */
861 unsigned char i_compress_algorithm; /* algorithm type */
862 unsigned char i_log_cluster_size; /* log of cluster size */
863 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
864 unsigned char i_compress_flag; /* compress flag */
865 unsigned int i_cluster_size; /* cluster size */
867 unsigned int atomic_write_cnt;
868 loff_t original_i_size; /* original i_size before atomic write */
871 static inline void get_read_extent_info(struct extent_info *ext,
872 struct f2fs_extent *i_ext)
874 ext->fofs = le32_to_cpu(i_ext->fofs);
875 ext->blk = le32_to_cpu(i_ext->blk);
876 ext->len = le32_to_cpu(i_ext->len);
879 static inline void set_raw_read_extent(struct extent_info *ext,
880 struct f2fs_extent *i_ext)
882 i_ext->fofs = cpu_to_le32(ext->fofs);
883 i_ext->blk = cpu_to_le32(ext->blk);
884 i_ext->len = cpu_to_le32(ext->len);
887 static inline bool __is_discard_mergeable(struct discard_info *back,
888 struct discard_info *front, unsigned int max_len)
890 return (back->lstart + back->len == front->lstart) &&
891 (back->len + front->len <= max_len);
894 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
895 struct discard_info *back, unsigned int max_len)
897 return __is_discard_mergeable(back, cur, max_len);
900 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
901 struct discard_info *front, unsigned int max_len)
903 return __is_discard_mergeable(cur, front, max_len);
907 * For free nid management
910 FREE_NID, /* newly added to free nid list */
911 PREALLOC_NID, /* it is preallocated */
922 struct f2fs_nm_info {
923 block_t nat_blkaddr; /* base disk address of NAT */
924 nid_t max_nid; /* maximum possible node ids */
925 nid_t available_nids; /* # of available node ids */
926 nid_t next_scan_nid; /* the next nid to be scanned */
927 nid_t max_rf_node_blocks; /* max # of nodes for recovery */
928 unsigned int ram_thresh; /* control the memory footprint */
929 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
930 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
932 /* NAT cache management */
933 struct radix_tree_root nat_root;/* root of the nat entry cache */
934 struct radix_tree_root nat_set_root;/* root of the nat set cache */
935 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */
936 struct list_head nat_entries; /* cached nat entry list (clean) */
937 spinlock_t nat_list_lock; /* protect clean nat entry list */
938 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
939 unsigned int nat_blocks; /* # of nat blocks */
941 /* free node ids management */
942 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
943 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
944 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
945 spinlock_t nid_list_lock; /* protect nid lists ops */
946 struct mutex build_lock; /* lock for build free nids */
947 unsigned char **free_nid_bitmap;
948 unsigned char *nat_block_bitmap;
949 unsigned short *free_nid_count; /* free nid count of NAT block */
952 char *nat_bitmap; /* NAT bitmap pointer */
954 unsigned int nat_bits_blocks; /* # of nat bits blocks */
955 unsigned char *nat_bits; /* NAT bits blocks */
956 unsigned char *full_nat_bits; /* full NAT pages */
957 unsigned char *empty_nat_bits; /* empty NAT pages */
958 #ifdef CONFIG_F2FS_CHECK_FS
959 char *nat_bitmap_mir; /* NAT bitmap mirror */
961 int bitmap_size; /* bitmap size */
965 * this structure is used as one of function parameters.
966 * all the information are dedicated to a given direct node block determined
967 * by the data offset in a file.
969 struct dnode_of_data {
970 struct inode *inode; /* vfs inode pointer */
971 struct page *inode_page; /* its inode page, NULL is possible */
972 struct page *node_page; /* cached direct node page */
973 nid_t nid; /* node id of the direct node block */
974 unsigned int ofs_in_node; /* data offset in the node page */
975 bool inode_page_locked; /* inode page is locked or not */
976 bool node_changed; /* is node block changed */
977 char cur_level; /* level of hole node page */
978 char max_level; /* level of current page located */
979 block_t data_blkaddr; /* block address of the node block */
982 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
983 struct page *ipage, struct page *npage, nid_t nid)
985 memset(dn, 0, sizeof(*dn));
987 dn->inode_page = ipage;
988 dn->node_page = npage;
995 * By default, there are 6 active log areas across the whole main area.
996 * When considering hot and cold data separation to reduce cleaning overhead,
997 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
999 * In the current design, you should not change the numbers intentionally.
1000 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
1001 * logs individually according to the underlying devices. (default: 6)
1002 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
1003 * data and 8 for node logs.
1005 #define NR_CURSEG_DATA_TYPE (3)
1006 #define NR_CURSEG_NODE_TYPE (3)
1007 #define NR_CURSEG_INMEM_TYPE (2)
1008 #define NR_CURSEG_RO_TYPE (2)
1009 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
1010 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
1013 CURSEG_HOT_DATA = 0, /* directory entry blocks */
1014 CURSEG_WARM_DATA, /* data blocks */
1015 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
1016 CURSEG_HOT_NODE, /* direct node blocks of directory files */
1017 CURSEG_WARM_NODE, /* direct node blocks of normal files */
1018 CURSEG_COLD_NODE, /* indirect node blocks */
1019 NR_PERSISTENT_LOG, /* number of persistent log */
1020 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
1021 /* pinned file that needs consecutive block address */
1022 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
1023 NO_CHECK_TYPE, /* number of persistent & inmem log */
1027 struct completion wait;
1028 struct llist_node llnode;
1033 struct flush_cmd_control {
1034 struct task_struct *f2fs_issue_flush; /* flush thread */
1035 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
1036 atomic_t issued_flush; /* # of issued flushes */
1037 atomic_t queued_flush; /* # of queued flushes */
1038 struct llist_head issue_list; /* list for command issue */
1039 struct llist_node *dispatch_list; /* list for command dispatch */
1042 struct f2fs_sm_info {
1043 struct sit_info *sit_info; /* whole segment information */
1044 struct free_segmap_info *free_info; /* free segment information */
1045 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1046 struct curseg_info *curseg_array; /* active segment information */
1048 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */
1050 block_t seg0_blkaddr; /* block address of 0'th segment */
1051 block_t main_blkaddr; /* start block address of main area */
1052 block_t ssa_blkaddr; /* start block address of SSA area */
1054 unsigned int segment_count; /* total # of segments */
1055 unsigned int main_segments; /* # of segments in main area */
1056 unsigned int reserved_segments; /* # of reserved segments */
1057 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1058 unsigned int ovp_segments; /* # of overprovision segments */
1060 /* a threshold to reclaim prefree segments */
1061 unsigned int rec_prefree_segments;
1063 struct list_head sit_entry_set; /* sit entry set list */
1065 unsigned int ipu_policy; /* in-place-update policy */
1066 unsigned int min_ipu_util; /* in-place-update threshold */
1067 unsigned int min_fsync_blocks; /* threshold for fsync */
1068 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1069 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1070 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1072 /* for flush command control */
1073 struct flush_cmd_control *fcc_info;
1075 /* for discard command control */
1076 struct discard_cmd_control *dcc_info;
1083 * COUNT_TYPE for monitoring
1085 * f2fs monitors the number of several block types such as on-writeback,
1086 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1088 #define WB_DATA_TYPE(p, f) \
1089 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1108 * The below are the page types of bios used in submit_bio().
1109 * The available types are:
1110 * DATA User data pages. It operates as async mode.
1111 * NODE Node pages. It operates as async mode.
1112 * META FS metadata pages such as SIT, NAT, CP.
1113 * NR_PAGE_TYPE The number of page types.
1114 * META_FLUSH Make sure the previous pages are written
1115 * with waiting the bio's completion
1116 * ... Only can be used with META.
1118 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1121 NODE = 1, /* should not change this */
1125 IPU, /* the below types are used by tracepoints only. */
1130 HOT = 0, /* must be zero for meta bio */
1136 enum need_lock_type {
1142 enum cp_reason_type {
1158 APP_DIRECT_IO, /* app direct write IOs */
1159 APP_BUFFERED_IO, /* app buffered write IOs */
1160 APP_WRITE_IO, /* app write IOs */
1161 APP_MAPPED_IO, /* app mapped IOs */
1162 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */
1163 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */
1164 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1165 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */
1166 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1167 FS_META_IO, /* meta IOs from kworker/reclaimer */
1168 FS_GC_DATA_IO, /* data IOs from forground gc */
1169 FS_GC_NODE_IO, /* node IOs from forground gc */
1170 FS_CP_DATA_IO, /* data IOs from checkpoint */
1171 FS_CP_NODE_IO, /* node IOs from checkpoint */
1172 FS_CP_META_IO, /* meta IOs from checkpoint */
1175 APP_DIRECT_READ_IO, /* app direct read IOs */
1176 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1177 APP_READ_IO, /* app read IOs */
1178 APP_MAPPED_READ_IO, /* app mapped read IOs */
1179 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */
1180 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */
1181 FS_DATA_READ_IO, /* data read IOs */
1182 FS_GDATA_READ_IO, /* data read IOs from background gc */
1183 FS_CDATA_READ_IO, /* compressed data read IOs */
1184 FS_NODE_READ_IO, /* node read IOs */
1185 FS_META_READ_IO, /* meta read IOs */
1188 FS_DISCARD_IO, /* discard */
1189 FS_FLUSH_IO, /* flush */
1190 FS_ZONE_RESET_IO, /* zone reset */
1194 struct f2fs_io_info {
1195 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1196 nid_t ino; /* inode number */
1197 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1198 enum temp_type temp; /* contains HOT/WARM/COLD */
1199 enum req_op op; /* contains REQ_OP_ */
1200 blk_opf_t op_flags; /* req_flag_bits */
1201 block_t new_blkaddr; /* new block address to be written */
1202 block_t old_blkaddr; /* old block address before Cow */
1203 struct page *page; /* page to be written */
1204 struct page *encrypted_page; /* encrypted page */
1205 struct page *compressed_page; /* compressed page */
1206 struct list_head list; /* serialize IOs */
1207 unsigned int compr_blocks; /* # of compressed block addresses */
1208 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */
1209 unsigned int version:8; /* version of the node */
1210 unsigned int submitted:1; /* indicate IO submission */
1211 unsigned int in_list:1; /* indicate fio is in io_list */
1212 unsigned int is_por:1; /* indicate IO is from recovery or not */
1213 unsigned int retry:1; /* need to reallocate block address */
1214 unsigned int encrypted:1; /* indicate file is encrypted */
1215 unsigned int post_read:1; /* require post read */
1216 enum iostat_type io_type; /* io type */
1217 struct writeback_control *io_wbc; /* writeback control */
1218 struct bio **bio; /* bio for ipu */
1219 sector_t *last_block; /* last block number in bio */
1224 struct list_head list;
1227 #define is_read_io(rw) ((rw) == READ)
1228 struct f2fs_bio_info {
1229 struct f2fs_sb_info *sbi; /* f2fs superblock */
1230 struct bio *bio; /* bios to merge */
1231 sector_t last_block_in_bio; /* last block number */
1232 struct f2fs_io_info fio; /* store buffered io info. */
1233 #ifdef CONFIG_BLK_DEV_ZONED
1234 struct completion zone_wait; /* condition value for the previous open zone to close */
1235 struct bio *zone_pending_bio; /* pending bio for the previous zone */
1236 void *bi_private; /* previous bi_private for pending bio */
1238 struct f2fs_rwsem io_rwsem; /* blocking op for bio */
1239 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1240 struct list_head io_list; /* track fios */
1241 struct list_head bio_list; /* bio entry list head */
1242 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */
1245 #define FDEV(i) (sbi->devs[i])
1246 #define RDEV(i) (raw_super->devs[i])
1247 struct f2fs_dev_info {
1248 struct bdev_handle *bdev_handle;
1249 struct block_device *bdev;
1250 char path[MAX_PATH_LEN];
1251 unsigned int total_segments;
1254 #ifdef CONFIG_BLK_DEV_ZONED
1255 unsigned int nr_blkz; /* Total number of zones */
1256 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1261 DIR_INODE, /* for dirty dir inode */
1262 FILE_INODE, /* for dirty regular/symlink inode */
1263 DIRTY_META, /* for all dirtied inode metadata */
1267 /* for inner inode cache management */
1268 struct inode_management {
1269 struct radix_tree_root ino_root; /* ino entry array */
1270 spinlock_t ino_lock; /* for ino entry lock */
1271 struct list_head ino_list; /* inode list head */
1272 unsigned long ino_num; /* number of entries */
1276 struct atgc_management {
1277 bool atgc_enabled; /* ATGC is enabled or not */
1278 struct rb_root_cached root; /* root of victim rb-tree */
1279 struct list_head victim_list; /* linked with all victim entries */
1280 unsigned int victim_count; /* victim count in rb-tree */
1281 unsigned int candidate_ratio; /* candidate ratio */
1282 unsigned int max_candidate_count; /* max candidate count */
1283 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1284 unsigned long long age_threshold; /* age threshold */
1287 struct f2fs_gc_control {
1288 unsigned int victim_segno; /* target victim segment number */
1289 int init_gc_type; /* FG_GC or BG_GC */
1290 bool no_bg_gc; /* check the space and stop bg_gc */
1291 bool should_migrate_blocks; /* should migrate blocks */
1292 bool err_gc_skipped; /* return EAGAIN if GC skipped */
1293 unsigned int nr_free_secs; /* # of free sections to do GC */
1297 * For s_flag in struct f2fs_sb_info
1298 * Modification on enum should be synchronized with s_flag array
1301 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1302 SBI_IS_CLOSE, /* specify unmounting */
1303 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1304 SBI_POR_DOING, /* recovery is doing or not */
1305 SBI_NEED_SB_WRITE, /* need to recover superblock */
1306 SBI_NEED_CP, /* need to checkpoint */
1307 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1308 SBI_IS_RECOVERED, /* recovered orphan/data */
1309 SBI_CP_DISABLED, /* CP was disabled last mount */
1310 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1311 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1312 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1313 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1314 SBI_IS_RESIZEFS, /* resizefs is in process */
1315 SBI_IS_FREEZING, /* freezefs is in process */
1316 SBI_IS_WRITABLE, /* remove ro mountoption transiently */
1326 UMOUNT_DISCARD_TIMEOUT,
1330 /* Note that you need to keep synchronization with this gc_mode_names array */
1343 BGGC_MODE_ON, /* background gc is on */
1344 BGGC_MODE_OFF, /* background gc is off */
1346 * background gc is on, migrating blocks
1347 * like foreground gc
1352 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1353 FS_MODE_LFS, /* use lfs allocation only */
1354 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */
1355 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */
1359 ALLOC_MODE_DEFAULT, /* stay default */
1360 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1364 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1365 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1366 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1371 * automatically compress compression
1375 * automatical compression is disabled.
1376 * user can control the file compression
1382 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1383 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1384 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1388 MEMORY_MODE_NORMAL, /* memory mode for normal devices */
1389 MEMORY_MODE_LOW, /* memory mode for low memry devices */
1392 enum errors_option {
1393 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */
1394 MOUNT_ERRORS_CONTINUE, /* continue on errors */
1395 MOUNT_ERRORS_PANIC, /* panic on errors */
1402 TOTAL_CALL = FOREGROUND,
1405 static inline int f2fs_test_bit(unsigned int nr, char *addr);
1406 static inline void f2fs_set_bit(unsigned int nr, char *addr);
1407 static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1410 * Layout of f2fs page.private:
1412 * Layout A: lowest bit should be 1
1413 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1414 * bit 0 PAGE_PRIVATE_NOT_POINTER
1415 * bit 1 PAGE_PRIVATE_DUMMY_WRITE
1416 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION
1417 * bit 3 PAGE_PRIVATE_INLINE_INODE
1418 * bit 4 PAGE_PRIVATE_REF_RESOURCE
1419 * bit 5- f2fs private data
1421 * Layout B: lowest bit should be 0
1422 * page.private is a wrapped pointer.
1425 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1426 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1427 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1428 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1429 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1433 /* For compression */
1434 enum compress_algorithm_type {
1442 enum compress_flag {
1447 #define COMPRESS_WATERMARK 20
1448 #define COMPRESS_PERCENT 20
1450 #define COMPRESS_DATA_RESERVED_SIZE 4
1451 struct compress_data {
1452 __le32 clen; /* compressed data size */
1453 __le32 chksum; /* compressed data chksum */
1454 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1455 u8 cdata[]; /* compressed data */
1458 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1460 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1462 #define F2FS_ZSTD_DEFAULT_CLEVEL 1
1464 #define COMPRESS_LEVEL_OFFSET 8
1466 /* compress context */
1467 struct compress_ctx {
1468 struct inode *inode; /* inode the context belong to */
1469 pgoff_t cluster_idx; /* cluster index number */
1470 unsigned int cluster_size; /* page count in cluster */
1471 unsigned int log_cluster_size; /* log of cluster size */
1472 struct page **rpages; /* pages store raw data in cluster */
1473 unsigned int nr_rpages; /* total page number in rpages */
1474 struct page **cpages; /* pages store compressed data in cluster */
1475 unsigned int nr_cpages; /* total page number in cpages */
1476 unsigned int valid_nr_cpages; /* valid page number in cpages */
1477 void *rbuf; /* virtual mapped address on rpages */
1478 struct compress_data *cbuf; /* virtual mapped address on cpages */
1479 size_t rlen; /* valid data length in rbuf */
1480 size_t clen; /* valid data length in cbuf */
1481 void *private; /* payload buffer for specified compression algorithm */
1482 void *private2; /* extra payload buffer */
1485 /* compress context for write IO path */
1486 struct compress_io_ctx {
1487 u32 magic; /* magic number to indicate page is compressed */
1488 struct inode *inode; /* inode the context belong to */
1489 struct page **rpages; /* pages store raw data in cluster */
1490 unsigned int nr_rpages; /* total page number in rpages */
1491 atomic_t pending_pages; /* in-flight compressed page count */
1494 /* Context for decompressing one cluster on the read IO path */
1495 struct decompress_io_ctx {
1496 u32 magic; /* magic number to indicate page is compressed */
1497 struct inode *inode; /* inode the context belong to */
1498 pgoff_t cluster_idx; /* cluster index number */
1499 unsigned int cluster_size; /* page count in cluster */
1500 unsigned int log_cluster_size; /* log of cluster size */
1501 struct page **rpages; /* pages store raw data in cluster */
1502 unsigned int nr_rpages; /* total page number in rpages */
1503 struct page **cpages; /* pages store compressed data in cluster */
1504 unsigned int nr_cpages; /* total page number in cpages */
1505 struct page **tpages; /* temp pages to pad holes in cluster */
1506 void *rbuf; /* virtual mapped address on rpages */
1507 struct compress_data *cbuf; /* virtual mapped address on cpages */
1508 size_t rlen; /* valid data length in rbuf */
1509 size_t clen; /* valid data length in cbuf */
1512 * The number of compressed pages remaining to be read in this cluster.
1513 * This is initially nr_cpages. It is decremented by 1 each time a page
1514 * has been read (or failed to be read). When it reaches 0, the cluster
1515 * is decompressed (or an error is reported).
1517 * If an error occurs before all the pages have been submitted for I/O,
1518 * then this will never reach 0. In this case the I/O submitter is
1519 * responsible for calling f2fs_decompress_end_io() instead.
1521 atomic_t remaining_pages;
1524 * Number of references to this decompress_io_ctx.
1526 * One reference is held for I/O completion. This reference is dropped
1527 * after the pagecache pages are updated and unlocked -- either after
1528 * decompression (and verity if enabled), or after an error.
1530 * In addition, each compressed page holds a reference while it is in a
1531 * bio. These references are necessary prevent compressed pages from
1532 * being freed while they are still in a bio.
1536 bool failed; /* IO error occurred before decompression? */
1537 bool need_verity; /* need fs-verity verification after decompression? */
1538 void *private; /* payload buffer for specified decompression algorithm */
1539 void *private2; /* extra payload buffer */
1540 struct work_struct verity_work; /* work to verify the decompressed pages */
1541 struct work_struct free_work; /* work for late free this structure itself */
1544 #define NULL_CLUSTER ((unsigned int)(~0))
1545 #define MIN_COMPRESS_LOG_SIZE 2
1546 #define MAX_COMPRESS_LOG_SIZE 8
1547 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1549 struct f2fs_sb_info {
1550 struct super_block *sb; /* pointer to VFS super block */
1551 struct proc_dir_entry *s_proc; /* proc entry */
1552 struct f2fs_super_block *raw_super; /* raw super block pointer */
1553 struct f2fs_rwsem sb_lock; /* lock for raw super block */
1554 int valid_super_block; /* valid super block no */
1555 unsigned long s_flag; /* flags for sbi */
1556 struct mutex writepages; /* mutex for writepages() */
1558 #ifdef CONFIG_BLK_DEV_ZONED
1559 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1562 /* for node-related operations */
1563 struct f2fs_nm_info *nm_info; /* node manager */
1564 struct inode *node_inode; /* cache node blocks */
1566 /* for segment-related operations */
1567 struct f2fs_sm_info *sm_info; /* segment manager */
1569 /* for bio operations */
1570 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1571 /* keep migration IO order for LFS mode */
1572 struct f2fs_rwsem io_order_lock;
1573 mempool_t *write_io_dummy; /* Dummy pages */
1574 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */
1575 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */
1577 /* for checkpoint */
1578 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1579 int cur_cp_pack; /* remain current cp pack */
1580 spinlock_t cp_lock; /* for flag in ckpt */
1581 struct inode *meta_inode; /* cache meta blocks */
1582 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */
1583 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */
1584 struct f2fs_rwsem node_write; /* locking node writes */
1585 struct f2fs_rwsem node_change; /* locking node change */
1586 wait_queue_head_t cp_wait;
1587 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1588 long interval_time[MAX_TIME]; /* to store thresholds */
1589 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1591 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1593 spinlock_t fsync_node_lock; /* for node entry lock */
1594 struct list_head fsync_node_list; /* node list head */
1595 unsigned int fsync_seg_id; /* sequence id */
1596 unsigned int fsync_node_num; /* number of node entries */
1598 /* for orphan inode, use 0'th array */
1599 unsigned int max_orphans; /* max orphan inodes */
1601 /* for inode management */
1602 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1603 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1604 struct mutex flush_lock; /* for flush exclusion */
1606 /* for extent tree cache */
1607 struct extent_tree_info extent_tree[NR_EXTENT_CACHES];
1608 atomic64_t allocated_data_blocks; /* for block age extent_cache */
1610 /* The threshold used for hot and warm data seperation*/
1611 unsigned int hot_data_age_threshold;
1612 unsigned int warm_data_age_threshold;
1613 unsigned int last_age_weight;
1615 /* basic filesystem units */
1616 unsigned int log_sectors_per_block; /* log2 sectors per block */
1617 unsigned int log_blocksize; /* log2 block size */
1618 unsigned int blocksize; /* block size */
1619 unsigned int root_ino_num; /* root inode number*/
1620 unsigned int node_ino_num; /* node inode number*/
1621 unsigned int meta_ino_num; /* meta inode number*/
1622 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1623 unsigned int blocks_per_seg; /* blocks per segment */
1624 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1625 unsigned int segs_per_sec; /* segments per section */
1626 unsigned int secs_per_zone; /* sections per zone */
1627 unsigned int total_sections; /* total section count */
1628 unsigned int total_node_count; /* total node block count */
1629 unsigned int total_valid_node_count; /* valid node block count */
1630 int dir_level; /* directory level */
1631 bool readdir_ra; /* readahead inode in readdir */
1632 u64 max_io_bytes; /* max io bytes to merge IOs */
1634 block_t user_block_count; /* # of user blocks */
1635 block_t total_valid_block_count; /* # of valid blocks */
1636 block_t discard_blks; /* discard command candidats */
1637 block_t last_valid_block_count; /* for recovery */
1638 block_t reserved_blocks; /* configurable reserved blocks */
1639 block_t current_reserved_blocks; /* current reserved blocks */
1641 /* Additional tracking for no checkpoint mode */
1642 block_t unusable_block_count; /* # of blocks saved by last cp */
1644 unsigned int nquota_files; /* # of quota sysfile */
1645 struct f2fs_rwsem quota_sem; /* blocking cp for flags */
1647 /* # of pages, see count_type */
1648 atomic_t nr_pages[NR_COUNT_TYPE];
1649 /* # of allocated blocks */
1650 struct percpu_counter alloc_valid_block_count;
1651 /* # of node block writes as roll forward recovery */
1652 struct percpu_counter rf_node_block_count;
1654 /* writeback control */
1655 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1657 /* valid inode count */
1658 struct percpu_counter total_valid_inode_count;
1660 struct f2fs_mount_info mount_opt; /* mount options */
1662 /* for cleaning operations */
1663 struct f2fs_rwsem gc_lock; /*
1664 * semaphore for GC, avoid
1665 * race between GC and GC or CP
1667 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1668 struct atgc_management am; /* atgc management */
1669 unsigned int cur_victim_sec; /* current victim section num */
1670 unsigned int gc_mode; /* current GC state */
1671 unsigned int next_victim_seg[2]; /* next segment in victim section */
1672 spinlock_t gc_remaining_trials_lock;
1673 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */
1674 unsigned int gc_remaining_trials;
1676 /* for skip statistic */
1677 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1679 /* threshold for gc trials on pinned files */
1680 u64 gc_pin_file_threshold;
1681 struct f2fs_rwsem pin_sem;
1683 /* maximum # of trials to find a victim segment for SSR and GC */
1684 unsigned int max_victim_search;
1685 /* migration granularity of garbage collection, unit: segment */
1686 unsigned int migration_granularity;
1689 * for stat information.
1690 * one is for the LFS mode, and the other is for the SSR mode.
1692 #ifdef CONFIG_F2FS_STAT_FS
1693 struct f2fs_stat_info *stat_info; /* FS status information */
1694 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1695 unsigned int segment_count[2]; /* # of allocated segments */
1696 unsigned int block_count[2]; /* # of allocated blocks */
1697 atomic_t inplace_count; /* # of inplace update */
1698 /* # of lookup extent cache */
1699 atomic64_t total_hit_ext[NR_EXTENT_CACHES];
1700 /* # of hit rbtree extent node */
1701 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES];
1702 /* # of hit cached extent node */
1703 atomic64_t read_hit_cached[NR_EXTENT_CACHES];
1704 /* # of hit largest extent node in read extent cache */
1705 atomic64_t read_hit_largest;
1706 atomic_t inline_xattr; /* # of inline_xattr inodes */
1707 atomic_t inline_inode; /* # of inline_data inodes */
1708 atomic_t inline_dir; /* # of inline_dentry inodes */
1709 atomic_t compr_inode; /* # of compressed inodes */
1710 atomic64_t compr_blocks; /* # of compressed blocks */
1711 atomic_t swapfile_inode; /* # of swapfile inodes */
1712 atomic_t atomic_files; /* # of opened atomic file */
1713 atomic_t max_aw_cnt; /* max # of atomic writes */
1714 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1715 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1716 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1717 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */
1719 spinlock_t stat_lock; /* lock for stat operations */
1721 /* to attach REQ_META|REQ_FUA flags */
1722 unsigned int data_io_flag;
1723 unsigned int node_io_flag;
1725 /* For sysfs support */
1726 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1727 struct completion s_kobj_unregister;
1729 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1730 struct completion s_stat_kobj_unregister;
1732 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1733 struct completion s_feature_list_kobj_unregister;
1735 /* For shrinker support */
1736 struct list_head s_list;
1737 struct mutex umount_mutex;
1738 unsigned int shrinker_run_no;
1740 /* For multi devices */
1741 int s_ndevs; /* number of devices */
1742 struct f2fs_dev_info *devs; /* for device list */
1743 unsigned int dirty_device; /* for checkpoint data flush */
1744 spinlock_t dev_lock; /* protect dirty_device */
1745 bool aligned_blksize; /* all devices has the same logical blksize */
1747 /* For write statistics */
1748 u64 sectors_written_start;
1751 /* Reference to checksum algorithm driver via cryptoapi */
1752 struct crypto_shash *s_chksum_driver;
1754 /* Precomputed FS UUID checksum for seeding other checksums */
1755 __u32 s_chksum_seed;
1757 struct workqueue_struct *post_read_wq; /* post read workqueue */
1760 * If we are in irq context, let's update error information into
1761 * on-disk superblock in the work.
1763 struct work_struct s_error_work;
1764 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
1765 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */
1766 spinlock_t error_lock; /* protect errors/stop_reason array */
1767 bool error_dirty; /* errors of sb is dirty */
1769 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1770 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1772 /* For reclaimed segs statistics per each GC mode */
1773 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1774 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1776 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1778 int max_fragment_chunk; /* max chunk size for block fragmentation mode */
1779 int max_fragment_hole; /* max hole size for block fragmentation mode */
1781 /* For atomic write statistics */
1782 atomic64_t current_atomic_write;
1783 s64 peak_atomic_write;
1784 u64 committed_atomic_block;
1785 u64 revoked_atomic_block;
1787 #ifdef CONFIG_F2FS_FS_COMPRESSION
1788 struct kmem_cache *page_array_slab; /* page array entry */
1789 unsigned int page_array_slab_size; /* default page array slab size */
1791 /* For runtime compression statistics */
1792 u64 compr_written_block;
1793 u64 compr_saved_block;
1794 u32 compr_new_inode;
1796 /* For compressed block cache */
1797 struct inode *compress_inode; /* cache compressed blocks */
1798 unsigned int compress_percent; /* cache page percentage */
1799 unsigned int compress_watermark; /* cache page watermark */
1800 atomic_t compress_page_hit; /* cache hit count */
1803 #ifdef CONFIG_F2FS_IOSTAT
1804 /* For app/fs IO statistics */
1805 spinlock_t iostat_lock;
1806 unsigned long long iostat_count[NR_IO_TYPE];
1807 unsigned long long iostat_bytes[NR_IO_TYPE];
1808 unsigned long long prev_iostat_bytes[NR_IO_TYPE];
1810 unsigned long iostat_next_period;
1811 unsigned int iostat_period_ms;
1813 /* For io latency related statistics info in one iostat period */
1814 spinlock_t iostat_lat_lock;
1815 struct iostat_lat_info *iostat_io_lat;
1819 #ifdef CONFIG_F2FS_FAULT_INJECTION
1820 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \
1821 __builtin_return_address(0))
1822 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type,
1823 const char *func, const char *parent_func)
1825 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1827 if (!ffi->inject_rate)
1830 if (!IS_FAULT_SET(ffi, type))
1833 atomic_inc(&ffi->inject_ops);
1834 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1835 atomic_set(&ffi->inject_ops, 0);
1836 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n",
1837 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type],
1844 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1851 * Test if the mounted volume is a multi-device volume.
1852 * - For a single regular disk volume, sbi->s_ndevs is 0.
1853 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1854 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1856 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1858 return sbi->s_ndevs > 1;
1861 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1863 unsigned long now = jiffies;
1865 sbi->last_time[type] = now;
1867 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1868 if (type == REQ_TIME) {
1869 sbi->last_time[DISCARD_TIME] = now;
1870 sbi->last_time[GC_TIME] = now;
1874 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1876 unsigned long interval = sbi->interval_time[type] * HZ;
1878 return time_after(jiffies, sbi->last_time[type] + interval);
1881 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1884 unsigned long interval = sbi->interval_time[type] * HZ;
1885 unsigned int wait_ms = 0;
1888 delta = (sbi->last_time[type] + interval) - jiffies;
1890 wait_ms = jiffies_to_msecs(delta);
1898 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1899 const void *address, unsigned int length)
1902 struct shash_desc shash;
1907 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1909 desc.shash.tfm = sbi->s_chksum_driver;
1910 *(u32 *)desc.ctx = crc;
1912 err = crypto_shash_update(&desc.shash, address, length);
1915 return *(u32 *)desc.ctx;
1918 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1919 unsigned int length)
1921 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1924 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1925 void *buf, size_t buf_size)
1927 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1930 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1931 const void *address, unsigned int length)
1933 return __f2fs_crc32(sbi, crc, address, length);
1936 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1938 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1941 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1943 return sb->s_fs_info;
1946 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1948 return F2FS_SB(inode->i_sb);
1951 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1953 return F2FS_I_SB(mapping->host);
1956 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1958 return F2FS_M_SB(page_file_mapping(page));
1961 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1963 return (struct f2fs_super_block *)(sbi->raw_super);
1966 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1968 return (struct f2fs_checkpoint *)(sbi->ckpt);
1971 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1973 return (struct f2fs_node *)page_address(page);
1976 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1978 return &((struct f2fs_node *)page_address(page))->i;
1981 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1983 return (struct f2fs_nm_info *)(sbi->nm_info);
1986 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1988 return (struct f2fs_sm_info *)(sbi->sm_info);
1991 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1993 return (struct sit_info *)(SM_I(sbi)->sit_info);
1996 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1998 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
2001 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
2003 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
2006 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
2008 return sbi->meta_inode->i_mapping;
2011 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
2013 return sbi->node_inode->i_mapping;
2016 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
2018 return test_bit(type, &sbi->s_flag);
2021 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2023 set_bit(type, &sbi->s_flag);
2026 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2028 clear_bit(type, &sbi->s_flag);
2031 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
2033 return le64_to_cpu(cp->checkpoint_ver);
2036 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
2038 if (type < F2FS_MAX_QUOTAS)
2039 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
2043 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2045 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2046 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2049 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2051 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2053 return ckpt_flags & f;
2056 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2058 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2061 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2063 unsigned int ckpt_flags;
2065 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2067 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2070 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2072 unsigned long flags;
2074 spin_lock_irqsave(&sbi->cp_lock, flags);
2075 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2076 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2079 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2081 unsigned int ckpt_flags;
2083 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2085 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2088 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2090 unsigned long flags;
2092 spin_lock_irqsave(&sbi->cp_lock, flags);
2093 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2094 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2097 #define init_f2fs_rwsem(sem) \
2099 static struct lock_class_key __key; \
2101 __init_f2fs_rwsem((sem), #sem, &__key); \
2104 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,
2105 const char *sem_name, struct lock_class_key *key)
2107 __init_rwsem(&sem->internal_rwsem, sem_name, key);
2108 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2109 init_waitqueue_head(&sem->read_waiters);
2113 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)
2115 return rwsem_is_locked(&sem->internal_rwsem);
2118 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)
2120 return rwsem_is_contended(&sem->internal_rwsem);
2123 static inline void f2fs_down_read(struct f2fs_rwsem *sem)
2125 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2126 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));
2128 down_read(&sem->internal_rwsem);
2132 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)
2134 return down_read_trylock(&sem->internal_rwsem);
2137 static inline void f2fs_up_read(struct f2fs_rwsem *sem)
2139 up_read(&sem->internal_rwsem);
2142 static inline void f2fs_down_write(struct f2fs_rwsem *sem)
2144 down_write(&sem->internal_rwsem);
2147 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2148 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)
2150 down_read_nested(&sem->internal_rwsem, subclass);
2153 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass)
2155 down_write_nested(&sem->internal_rwsem, subclass);
2158 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)
2159 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem)
2162 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)
2164 return down_write_trylock(&sem->internal_rwsem);
2167 static inline void f2fs_up_write(struct f2fs_rwsem *sem)
2169 up_write(&sem->internal_rwsem);
2170 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2171 wake_up_all(&sem->read_waiters);
2175 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2177 f2fs_down_read(&sbi->cp_rwsem);
2180 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2182 if (time_to_inject(sbi, FAULT_LOCK_OP))
2184 return f2fs_down_read_trylock(&sbi->cp_rwsem);
2187 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2189 f2fs_up_read(&sbi->cp_rwsem);
2192 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2194 f2fs_down_write(&sbi->cp_rwsem);
2197 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2199 f2fs_up_write(&sbi->cp_rwsem);
2202 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2204 int reason = CP_SYNC;
2206 if (test_opt(sbi, FASTBOOT))
2207 reason = CP_FASTBOOT;
2208 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2213 static inline bool __remain_node_summaries(int reason)
2215 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2218 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2220 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2221 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2225 * Check whether the inode has blocks or not
2227 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2229 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2231 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2234 static inline bool f2fs_has_xattr_block(unsigned int ofs)
2236 return ofs == XATTR_NODE_OFFSET;
2239 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2240 struct inode *inode, bool cap)
2244 if (!test_opt(sbi, RESERVE_ROOT))
2246 if (IS_NOQUOTA(inode))
2248 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2250 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2251 in_group_p(F2FS_OPTION(sbi).s_resgid))
2253 if (cap && capable(CAP_SYS_RESOURCE))
2258 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2259 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2260 struct inode *inode, blkcnt_t *count, bool partial)
2262 blkcnt_t diff = 0, release = 0;
2263 block_t avail_user_block_count;
2266 ret = dquot_reserve_block(inode, *count);
2270 if (time_to_inject(sbi, FAULT_BLOCK)) {
2276 * let's increase this in prior to actual block count change in order
2277 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2279 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2281 spin_lock(&sbi->stat_lock);
2282 sbi->total_valid_block_count += (block_t)(*count);
2283 avail_user_block_count = sbi->user_block_count -
2284 sbi->current_reserved_blocks;
2286 if (!__allow_reserved_blocks(sbi, inode, true))
2287 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2289 if (F2FS_IO_ALIGNED(sbi))
2290 avail_user_block_count -= sbi->blocks_per_seg *
2291 SM_I(sbi)->additional_reserved_segments;
2293 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2294 if (avail_user_block_count > sbi->unusable_block_count)
2295 avail_user_block_count -= sbi->unusable_block_count;
2297 avail_user_block_count = 0;
2299 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2301 spin_unlock(&sbi->stat_lock);
2305 diff = sbi->total_valid_block_count - avail_user_block_count;
2310 sbi->total_valid_block_count -= diff;
2312 spin_unlock(&sbi->stat_lock);
2316 spin_unlock(&sbi->stat_lock);
2318 if (unlikely(release)) {
2319 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2320 dquot_release_reservation_block(inode, release);
2322 f2fs_i_blocks_write(inode, *count, true, true);
2326 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2328 dquot_release_reservation_block(inode, release);
2333 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2335 #define f2fs_err(sbi, fmt, ...) \
2336 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2337 #define f2fs_warn(sbi, fmt, ...) \
2338 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2339 #define f2fs_notice(sbi, fmt, ...) \
2340 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2341 #define f2fs_info(sbi, fmt, ...) \
2342 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2343 #define f2fs_debug(sbi, fmt, ...) \
2344 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2346 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \
2347 static inline bool page_private_##name(struct page *page) \
2349 return PagePrivate(page) && \
2350 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
2351 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2354 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \
2355 static inline void set_page_private_##name(struct page *page) \
2357 if (!PagePrivate(page)) \
2358 attach_page_private(page, (void *)0); \
2359 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
2360 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2363 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
2364 static inline void clear_page_private_##name(struct page *page) \
2366 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2367 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \
2368 detach_page_private(page); \
2371 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
2372 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
2373 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
2374 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
2376 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
2377 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
2378 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
2379 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
2381 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
2382 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
2383 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
2384 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
2386 static inline unsigned long get_page_private_data(struct page *page)
2388 unsigned long data = page_private(page);
2390 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
2392 return data >> PAGE_PRIVATE_MAX;
2395 static inline void set_page_private_data(struct page *page, unsigned long data)
2397 if (!PagePrivate(page))
2398 attach_page_private(page, (void *)0);
2399 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
2400 page_private(page) |= data << PAGE_PRIVATE_MAX;
2403 static inline void clear_page_private_data(struct page *page)
2405 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0);
2406 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER))
2407 detach_page_private(page);
2410 static inline void clear_page_private_all(struct page *page)
2412 clear_page_private_data(page);
2413 clear_page_private_reference(page);
2414 clear_page_private_gcing(page);
2415 clear_page_private_inline(page);
2417 f2fs_bug_on(F2FS_P_SB(page), page_private(page));
2420 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2421 struct inode *inode,
2424 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2426 spin_lock(&sbi->stat_lock);
2427 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2428 sbi->total_valid_block_count -= (block_t)count;
2429 if (sbi->reserved_blocks &&
2430 sbi->current_reserved_blocks < sbi->reserved_blocks)
2431 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2432 sbi->current_reserved_blocks + count);
2433 spin_unlock(&sbi->stat_lock);
2434 if (unlikely(inode->i_blocks < sectors)) {
2435 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2437 (unsigned long long)inode->i_blocks,
2438 (unsigned long long)sectors);
2439 set_sbi_flag(sbi, SBI_NEED_FSCK);
2442 f2fs_i_blocks_write(inode, count, false, true);
2445 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2447 atomic_inc(&sbi->nr_pages[count_type]);
2449 if (count_type == F2FS_DIRTY_DENTS ||
2450 count_type == F2FS_DIRTY_NODES ||
2451 count_type == F2FS_DIRTY_META ||
2452 count_type == F2FS_DIRTY_QDATA ||
2453 count_type == F2FS_DIRTY_IMETA)
2454 set_sbi_flag(sbi, SBI_IS_DIRTY);
2457 static inline void inode_inc_dirty_pages(struct inode *inode)
2459 atomic_inc(&F2FS_I(inode)->dirty_pages);
2460 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2461 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2462 if (IS_NOQUOTA(inode))
2463 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2466 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2468 atomic_dec(&sbi->nr_pages[count_type]);
2471 static inline void inode_dec_dirty_pages(struct inode *inode)
2473 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2474 !S_ISLNK(inode->i_mode))
2477 atomic_dec(&F2FS_I(inode)->dirty_pages);
2478 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2479 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2480 if (IS_NOQUOTA(inode))
2481 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2484 static inline void inc_atomic_write_cnt(struct inode *inode)
2486 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2487 struct f2fs_inode_info *fi = F2FS_I(inode);
2490 fi->atomic_write_cnt++;
2491 atomic64_inc(&sbi->current_atomic_write);
2492 current_write = atomic64_read(&sbi->current_atomic_write);
2493 if (current_write > sbi->peak_atomic_write)
2494 sbi->peak_atomic_write = current_write;
2497 static inline void release_atomic_write_cnt(struct inode *inode)
2499 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2500 struct f2fs_inode_info *fi = F2FS_I(inode);
2502 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write);
2503 fi->atomic_write_cnt = 0;
2506 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2508 return atomic_read(&sbi->nr_pages[count_type]);
2511 static inline int get_dirty_pages(struct inode *inode)
2513 return atomic_read(&F2FS_I(inode)->dirty_pages);
2516 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2518 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2519 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2520 sbi->log_blocks_per_seg;
2522 return segs / sbi->segs_per_sec;
2525 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2527 return sbi->total_valid_block_count;
2530 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2532 return sbi->discard_blks;
2535 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2537 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2539 /* return NAT or SIT bitmap */
2540 if (flag == NAT_BITMAP)
2541 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2542 else if (flag == SIT_BITMAP)
2543 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2548 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2550 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2553 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2555 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2556 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2559 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2560 offset = (flag == SIT_BITMAP) ?
2561 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2563 * if large_nat_bitmap feature is enabled, leave checksum
2564 * protection for all nat/sit bitmaps.
2566 return tmp_ptr + offset + sizeof(__le32);
2569 if (__cp_payload(sbi) > 0) {
2570 if (flag == NAT_BITMAP)
2573 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2575 offset = (flag == NAT_BITMAP) ?
2576 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2577 return tmp_ptr + offset;
2581 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2583 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2585 if (sbi->cur_cp_pack == 2)
2586 start_addr += sbi->blocks_per_seg;
2590 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2592 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2594 if (sbi->cur_cp_pack == 1)
2595 start_addr += sbi->blocks_per_seg;
2599 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2601 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2604 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2606 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2609 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
2610 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2611 struct inode *inode, bool is_inode)
2613 block_t valid_block_count;
2614 unsigned int valid_node_count, user_block_count;
2619 err = dquot_alloc_inode(inode);
2624 err = dquot_reserve_block(inode, 1);
2629 if (time_to_inject(sbi, FAULT_BLOCK))
2632 spin_lock(&sbi->stat_lock);
2634 valid_block_count = sbi->total_valid_block_count +
2635 sbi->current_reserved_blocks + 1;
2637 if (!__allow_reserved_blocks(sbi, inode, false))
2638 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2640 if (F2FS_IO_ALIGNED(sbi))
2641 valid_block_count += sbi->blocks_per_seg *
2642 SM_I(sbi)->additional_reserved_segments;
2644 user_block_count = sbi->user_block_count;
2645 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2646 user_block_count -= sbi->unusable_block_count;
2648 if (unlikely(valid_block_count > user_block_count)) {
2649 spin_unlock(&sbi->stat_lock);
2653 valid_node_count = sbi->total_valid_node_count + 1;
2654 if (unlikely(valid_node_count > sbi->total_node_count)) {
2655 spin_unlock(&sbi->stat_lock);
2659 sbi->total_valid_node_count++;
2660 sbi->total_valid_block_count++;
2661 spin_unlock(&sbi->stat_lock);
2665 f2fs_mark_inode_dirty_sync(inode, true);
2667 f2fs_i_blocks_write(inode, 1, true, true);
2670 percpu_counter_inc(&sbi->alloc_valid_block_count);
2676 dquot_free_inode(inode);
2678 dquot_release_reservation_block(inode, 1);
2683 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2684 struct inode *inode, bool is_inode)
2686 spin_lock(&sbi->stat_lock);
2688 if (unlikely(!sbi->total_valid_block_count ||
2689 !sbi->total_valid_node_count)) {
2690 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2691 sbi->total_valid_block_count,
2692 sbi->total_valid_node_count);
2693 set_sbi_flag(sbi, SBI_NEED_FSCK);
2695 sbi->total_valid_block_count--;
2696 sbi->total_valid_node_count--;
2699 if (sbi->reserved_blocks &&
2700 sbi->current_reserved_blocks < sbi->reserved_blocks)
2701 sbi->current_reserved_blocks++;
2703 spin_unlock(&sbi->stat_lock);
2706 dquot_free_inode(inode);
2708 if (unlikely(inode->i_blocks == 0)) {
2709 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2711 (unsigned long long)inode->i_blocks);
2712 set_sbi_flag(sbi, SBI_NEED_FSCK);
2715 f2fs_i_blocks_write(inode, 1, false, true);
2719 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2721 return sbi->total_valid_node_count;
2724 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2726 percpu_counter_inc(&sbi->total_valid_inode_count);
2729 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2731 percpu_counter_dec(&sbi->total_valid_inode_count);
2734 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2736 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2739 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2740 pgoff_t index, bool for_write)
2745 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2747 page = find_get_page_flags(mapping, index,
2748 FGP_LOCK | FGP_ACCESSED);
2750 page = find_lock_page(mapping, index);
2754 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))
2759 return grab_cache_page(mapping, index);
2761 flags = memalloc_nofs_save();
2762 page = grab_cache_page_write_begin(mapping, index);
2763 memalloc_nofs_restore(flags);
2768 static inline struct page *f2fs_pagecache_get_page(
2769 struct address_space *mapping, pgoff_t index,
2770 fgf_t fgp_flags, gfp_t gfp_mask)
2772 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET))
2775 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2778 static inline void f2fs_put_page(struct page *page, int unlock)
2784 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2790 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2793 f2fs_put_page(dn->node_page, 1);
2794 if (dn->inode_page && dn->node_page != dn->inode_page)
2795 f2fs_put_page(dn->inode_page, 0);
2796 dn->node_page = NULL;
2797 dn->inode_page = NULL;
2800 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2803 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2806 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
2811 entry = kmem_cache_alloc(cachep, flags);
2813 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2817 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2818 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
2821 return f2fs_kmem_cache_alloc_nofail(cachep, flags);
2823 if (time_to_inject(sbi, FAULT_SLAB_ALLOC))
2826 return kmem_cache_alloc(cachep, flags);
2829 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2831 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2832 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2833 get_pages(sbi, F2FS_WB_CP_DATA) ||
2834 get_pages(sbi, F2FS_DIO_READ) ||
2835 get_pages(sbi, F2FS_DIO_WRITE))
2838 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2839 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2842 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2843 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2848 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2850 if (sbi->gc_mode == GC_URGENT_HIGH)
2853 if (is_inflight_io(sbi, type))
2856 if (sbi->gc_mode == GC_URGENT_MID)
2859 if (sbi->gc_mode == GC_URGENT_LOW &&
2860 (type == DISCARD_TIME || type == GC_TIME))
2863 return f2fs_time_over(sbi, type);
2866 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2867 unsigned long index, void *item)
2869 while (radix_tree_insert(root, index, item))
2873 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2875 static inline bool IS_INODE(struct page *page)
2877 struct f2fs_node *p = F2FS_NODE(page);
2879 return RAW_IS_INODE(p);
2882 static inline int offset_in_addr(struct f2fs_inode *i)
2884 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2885 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2888 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2890 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2893 static inline int f2fs_has_extra_attr(struct inode *inode);
2894 static inline block_t data_blkaddr(struct inode *inode,
2895 struct page *node_page, unsigned int offset)
2897 struct f2fs_node *raw_node;
2900 bool is_inode = IS_INODE(node_page);
2902 raw_node = F2FS_NODE(node_page);
2906 /* from GC path only */
2907 base = offset_in_addr(&raw_node->i);
2908 else if (f2fs_has_extra_attr(inode))
2909 base = get_extra_isize(inode);
2912 addr_array = blkaddr_in_node(raw_node);
2913 return le32_to_cpu(addr_array[base + offset]);
2916 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2918 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2921 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2926 mask = BIT(7 - (nr & 0x07));
2927 return mask & *addr;
2930 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2935 mask = BIT(7 - (nr & 0x07));
2939 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2944 mask = BIT(7 - (nr & 0x07));
2948 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2954 mask = BIT(7 - (nr & 0x07));
2960 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2966 mask = BIT(7 - (nr & 0x07));
2972 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2977 mask = BIT(7 - (nr & 0x07));
2982 * On-disk inode flags (f2fs_inode::i_flags)
2984 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2985 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2986 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2987 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2988 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2989 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2990 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2991 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2992 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2993 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2994 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2996 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL)
2998 /* Flags that should be inherited by new inodes from their parent. */
2999 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
3000 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
3003 /* Flags that are appropriate for regular files (all but dir-specific ones). */
3004 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
3007 /* Flags that are appropriate for non-directories/regular files. */
3008 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
3010 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
3014 else if (S_ISREG(mode))
3015 return flags & F2FS_REG_FLMASK;
3017 return flags & F2FS_OTHER_FLMASK;
3020 static inline void __mark_inode_dirty_flag(struct inode *inode,
3024 case FI_INLINE_XATTR:
3025 case FI_INLINE_DATA:
3026 case FI_INLINE_DENTRY:
3032 case FI_INLINE_DOTS:
3034 case FI_COMPRESS_RELEASED:
3035 case FI_ATOMIC_COMMITTED:
3036 f2fs_mark_inode_dirty_sync(inode, true);
3040 static inline void set_inode_flag(struct inode *inode, int flag)
3042 set_bit(flag, F2FS_I(inode)->flags);
3043 __mark_inode_dirty_flag(inode, flag, true);
3046 static inline int is_inode_flag_set(struct inode *inode, int flag)
3048 return test_bit(flag, F2FS_I(inode)->flags);
3051 static inline void clear_inode_flag(struct inode *inode, int flag)
3053 clear_bit(flag, F2FS_I(inode)->flags);
3054 __mark_inode_dirty_flag(inode, flag, false);
3057 static inline bool f2fs_verity_in_progress(struct inode *inode)
3059 return IS_ENABLED(CONFIG_FS_VERITY) &&
3060 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
3063 static inline void set_acl_inode(struct inode *inode, umode_t mode)
3065 F2FS_I(inode)->i_acl_mode = mode;
3066 set_inode_flag(inode, FI_ACL_MODE);
3067 f2fs_mark_inode_dirty_sync(inode, false);
3070 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
3076 f2fs_mark_inode_dirty_sync(inode, true);
3079 static inline void f2fs_i_blocks_write(struct inode *inode,
3080 block_t diff, bool add, bool claim)
3082 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3083 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3085 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
3088 dquot_claim_block(inode, diff);
3090 dquot_alloc_block_nofail(inode, diff);
3092 dquot_free_block(inode, diff);
3095 f2fs_mark_inode_dirty_sync(inode, true);
3096 if (clean || recover)
3097 set_inode_flag(inode, FI_AUTO_RECOVER);
3100 static inline bool f2fs_is_atomic_file(struct inode *inode);
3102 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
3104 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3105 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3107 if (i_size_read(inode) == i_size)
3110 i_size_write(inode, i_size);
3112 if (f2fs_is_atomic_file(inode))
3115 f2fs_mark_inode_dirty_sync(inode, true);
3116 if (clean || recover)
3117 set_inode_flag(inode, FI_AUTO_RECOVER);
3120 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
3122 F2FS_I(inode)->i_current_depth = depth;
3123 f2fs_mark_inode_dirty_sync(inode, true);
3126 static inline void f2fs_i_gc_failures_write(struct inode *inode,
3129 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
3130 f2fs_mark_inode_dirty_sync(inode, true);
3133 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
3135 F2FS_I(inode)->i_xattr_nid = xnid;
3136 f2fs_mark_inode_dirty_sync(inode, true);
3139 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
3141 F2FS_I(inode)->i_pino = pino;
3142 f2fs_mark_inode_dirty_sync(inode, true);
3145 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
3147 struct f2fs_inode_info *fi = F2FS_I(inode);
3149 if (ri->i_inline & F2FS_INLINE_XATTR)
3150 set_bit(FI_INLINE_XATTR, fi->flags);
3151 if (ri->i_inline & F2FS_INLINE_DATA)
3152 set_bit(FI_INLINE_DATA, fi->flags);
3153 if (ri->i_inline & F2FS_INLINE_DENTRY)
3154 set_bit(FI_INLINE_DENTRY, fi->flags);
3155 if (ri->i_inline & F2FS_DATA_EXIST)
3156 set_bit(FI_DATA_EXIST, fi->flags);
3157 if (ri->i_inline & F2FS_INLINE_DOTS)
3158 set_bit(FI_INLINE_DOTS, fi->flags);
3159 if (ri->i_inline & F2FS_EXTRA_ATTR)
3160 set_bit(FI_EXTRA_ATTR, fi->flags);
3161 if (ri->i_inline & F2FS_PIN_FILE)
3162 set_bit(FI_PIN_FILE, fi->flags);
3163 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
3164 set_bit(FI_COMPRESS_RELEASED, fi->flags);
3167 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
3171 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
3172 ri->i_inline |= F2FS_INLINE_XATTR;
3173 if (is_inode_flag_set(inode, FI_INLINE_DATA))
3174 ri->i_inline |= F2FS_INLINE_DATA;
3175 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
3176 ri->i_inline |= F2FS_INLINE_DENTRY;
3177 if (is_inode_flag_set(inode, FI_DATA_EXIST))
3178 ri->i_inline |= F2FS_DATA_EXIST;
3179 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
3180 ri->i_inline |= F2FS_INLINE_DOTS;
3181 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
3182 ri->i_inline |= F2FS_EXTRA_ATTR;
3183 if (is_inode_flag_set(inode, FI_PIN_FILE))
3184 ri->i_inline |= F2FS_PIN_FILE;
3185 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
3186 ri->i_inline |= F2FS_COMPRESS_RELEASED;
3189 static inline int f2fs_has_extra_attr(struct inode *inode)
3191 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
3194 static inline int f2fs_has_inline_xattr(struct inode *inode)
3196 return is_inode_flag_set(inode, FI_INLINE_XATTR);
3199 static inline int f2fs_compressed_file(struct inode *inode)
3201 return S_ISREG(inode->i_mode) &&
3202 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
3205 static inline bool f2fs_need_compress_data(struct inode *inode)
3207 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
3209 if (!f2fs_compressed_file(inode))
3212 if (compress_mode == COMPR_MODE_FS)
3214 else if (compress_mode == COMPR_MODE_USER &&
3215 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
3221 static inline unsigned int addrs_per_inode(struct inode *inode)
3223 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3224 get_inline_xattr_addrs(inode);
3226 if (!f2fs_compressed_file(inode))
3228 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3231 static inline unsigned int addrs_per_block(struct inode *inode)
3233 if (!f2fs_compressed_file(inode))
3234 return DEF_ADDRS_PER_BLOCK;
3235 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3238 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3240 struct f2fs_inode *ri = F2FS_INODE(page);
3242 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3243 get_inline_xattr_addrs(inode)]);
3246 static inline int inline_xattr_size(struct inode *inode)
3248 if (f2fs_has_inline_xattr(inode))
3249 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3254 * Notice: check inline_data flag without inode page lock is unsafe.
3255 * It could change at any time by f2fs_convert_inline_page().
3257 static inline int f2fs_has_inline_data(struct inode *inode)
3259 return is_inode_flag_set(inode, FI_INLINE_DATA);
3262 static inline int f2fs_exist_data(struct inode *inode)
3264 return is_inode_flag_set(inode, FI_DATA_EXIST);
3267 static inline int f2fs_has_inline_dots(struct inode *inode)
3269 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3272 static inline int f2fs_is_mmap_file(struct inode *inode)
3274 return is_inode_flag_set(inode, FI_MMAP_FILE);
3277 static inline bool f2fs_is_pinned_file(struct inode *inode)
3279 return is_inode_flag_set(inode, FI_PIN_FILE);
3282 static inline bool f2fs_is_atomic_file(struct inode *inode)
3284 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3287 static inline bool f2fs_is_cow_file(struct inode *inode)
3289 return is_inode_flag_set(inode, FI_COW_FILE);
3292 static inline __le32 *get_dnode_addr(struct inode *inode,
3293 struct page *node_page);
3294 static inline void *inline_data_addr(struct inode *inode, struct page *page)
3296 __le32 *addr = get_dnode_addr(inode, page);
3298 return (void *)(addr + DEF_INLINE_RESERVED_SIZE);
3301 static inline int f2fs_has_inline_dentry(struct inode *inode)
3303 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3306 static inline int is_file(struct inode *inode, int type)
3308 return F2FS_I(inode)->i_advise & type;
3311 static inline void set_file(struct inode *inode, int type)
3313 if (is_file(inode, type))
3315 F2FS_I(inode)->i_advise |= type;
3316 f2fs_mark_inode_dirty_sync(inode, true);
3319 static inline void clear_file(struct inode *inode, int type)
3321 if (!is_file(inode, type))
3323 F2FS_I(inode)->i_advise &= ~type;
3324 f2fs_mark_inode_dirty_sync(inode, true);
3327 static inline bool f2fs_is_time_consistent(struct inode *inode)
3329 struct timespec64 ts = inode_get_atime(inode);
3331 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts))
3333 ts = inode_get_ctime(inode);
3334 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts))
3336 ts = inode_get_mtime(inode);
3337 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts))
3342 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3347 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3349 spin_lock(&sbi->inode_lock[DIRTY_META]);
3350 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3351 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3354 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3355 file_keep_isize(inode) ||
3356 i_size_read(inode) & ~PAGE_MASK)
3359 if (!f2fs_is_time_consistent(inode))
3362 spin_lock(&F2FS_I(inode)->i_size_lock);
3363 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3364 spin_unlock(&F2FS_I(inode)->i_size_lock);
3369 static inline bool f2fs_readonly(struct super_block *sb)
3371 return sb_rdonly(sb);
3374 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3376 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3379 static inline bool is_dot_dotdot(const u8 *name, size_t len)
3381 if (len == 1 && name[0] == '.')
3384 if (len == 2 && name[0] == '.' && name[1] == '.')
3390 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3391 size_t size, gfp_t flags)
3393 if (time_to_inject(sbi, FAULT_KMALLOC))
3396 return kmalloc(size, flags);
3399 static inline void *f2fs_getname(struct f2fs_sb_info *sbi)
3401 if (time_to_inject(sbi, FAULT_KMALLOC))
3407 static inline void f2fs_putname(char *buf)
3412 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3413 size_t size, gfp_t flags)
3415 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3418 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3419 size_t size, gfp_t flags)
3421 if (time_to_inject(sbi, FAULT_KVMALLOC))
3424 return kvmalloc(size, flags);
3427 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3428 size_t size, gfp_t flags)
3430 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3433 static inline int get_extra_isize(struct inode *inode)
3435 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3438 static inline int get_inline_xattr_addrs(struct inode *inode)
3440 return F2FS_I(inode)->i_inline_xattr_size;
3443 static inline __le32 *get_dnode_addr(struct inode *inode,
3444 struct page *node_page)
3448 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode))
3449 base = get_extra_isize(inode);
3451 return blkaddr_in_node(F2FS_NODE(node_page)) + base;
3454 #define f2fs_get_inode_mode(i) \
3455 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3456 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3458 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32))
3460 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3461 (offsetof(struct f2fs_inode, i_extra_end) - \
3462 offsetof(struct f2fs_inode, i_extra_isize)) \
3464 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3465 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3466 ((offsetof(typeof(*(f2fs_inode)), field) + \
3467 sizeof((f2fs_inode)->field)) \
3468 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3470 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3472 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3474 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3475 block_t blkaddr, int type);
3476 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3477 block_t blkaddr, int type)
3479 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type))
3480 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3484 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3486 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3487 blkaddr == COMPRESS_ADDR)
3495 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3496 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3497 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3498 int f2fs_truncate(struct inode *inode);
3499 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path,
3500 struct kstat *stat, u32 request_mask, unsigned int flags);
3501 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
3502 struct iattr *attr);
3503 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3504 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3505 int f2fs_precache_extents(struct inode *inode);
3506 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3507 int f2fs_fileattr_set(struct mnt_idmap *idmap,
3508 struct dentry *dentry, struct fileattr *fa);
3509 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3510 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3511 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3512 int f2fs_pin_file_control(struct inode *inode, bool inc);
3517 void f2fs_set_inode_flags(struct inode *inode);
3518 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3519 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3520 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3521 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3522 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3523 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3524 void f2fs_update_inode_page(struct inode *inode);
3525 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3526 void f2fs_evict_inode(struct inode *inode);
3527 void f2fs_handle_failed_inode(struct inode *inode);
3532 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3533 bool hot, bool set);
3534 struct dentry *f2fs_get_parent(struct dentry *child);
3535 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
3536 struct inode **new_inode);
3541 int f2fs_init_casefolded_name(const struct inode *dir,
3542 struct f2fs_filename *fname);
3543 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3544 int lookup, struct f2fs_filename *fname);
3545 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3546 struct f2fs_filename *fname);
3547 void f2fs_free_filename(struct f2fs_filename *fname);
3548 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3549 const struct f2fs_filename *fname, int *max_slots);
3550 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3551 unsigned int start_pos, struct fscrypt_str *fstr);
3552 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3553 struct f2fs_dentry_ptr *d);
3554 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3555 const struct f2fs_filename *fname, struct page *dpage);
3556 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3557 unsigned int current_depth);
3558 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3559 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3560 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3561 const struct f2fs_filename *fname,
3562 struct page **res_page);
3563 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3564 const struct qstr *child, struct page **res_page);
3565 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3566 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3567 struct page **page);
3568 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3569 struct page *page, struct inode *inode);
3570 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3571 const struct f2fs_filename *fname);
3572 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3573 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3574 unsigned int bit_pos);
3575 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3576 struct inode *inode, nid_t ino, umode_t mode);
3577 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3578 struct inode *inode, nid_t ino, umode_t mode);
3579 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3580 struct inode *inode, nid_t ino, umode_t mode);
3581 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3582 struct inode *dir, struct inode *inode);
3583 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir,
3584 struct f2fs_filename *fname);
3585 bool f2fs_empty_dir(struct inode *dir);
3587 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3589 if (fscrypt_is_nokey_name(dentry))
3591 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3592 inode, inode->i_ino, inode->i_mode);
3598 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3599 void f2fs_inode_synced(struct inode *inode);
3600 int f2fs_dquot_initialize(struct inode *inode);
3601 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3602 int f2fs_quota_sync(struct super_block *sb, int type);
3603 loff_t max_file_blocks(struct inode *inode);
3604 void f2fs_quota_off_umount(struct super_block *sb);
3605 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag);
3606 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
3608 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
3609 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error);
3610 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3611 int f2fs_sync_fs(struct super_block *sb, int sync);
3612 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3617 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3624 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3625 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3626 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3627 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3628 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3629 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3630 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3631 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3632 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3633 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3634 struct node_info *ni, bool checkpoint_context);
3635 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3636 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3637 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3638 int f2fs_truncate_xattr_node(struct inode *inode);
3639 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3640 unsigned int seq_id);
3641 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
3642 int f2fs_remove_inode_page(struct inode *inode);
3643 struct page *f2fs_new_inode_page(struct inode *inode);
3644 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3645 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3646 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3647 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3648 int f2fs_move_node_page(struct page *node_page, int gc_type);
3649 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3650 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3651 struct writeback_control *wbc, bool atomic,
3652 unsigned int *seq_id);
3653 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3654 struct writeback_control *wbc,
3655 bool do_balance, enum iostat_type io_type);
3656 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3657 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3658 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3659 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3660 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3661 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3662 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3663 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3664 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3665 unsigned int segno, struct f2fs_summary_block *sum);
3666 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
3667 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3668 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3669 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3670 int __init f2fs_create_node_manager_caches(void);
3671 void f2fs_destroy_node_manager_caches(void);
3676 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3677 int f2fs_commit_atomic_write(struct inode *inode);
3678 void f2fs_abort_atomic_write(struct inode *inode, bool clean);
3679 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3680 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3681 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3682 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3683 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3684 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3685 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3686 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3687 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
3688 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3689 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3690 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3691 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3692 struct cp_control *cpc);
3693 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3694 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3695 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3696 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3697 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3698 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3699 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3700 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3701 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3702 void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3703 unsigned int *newseg, bool new_sec, int dir);
3704 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3705 unsigned int start, unsigned int end);
3706 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3707 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3708 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3709 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3710 struct cp_control *cpc);
3711 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3712 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3714 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3715 enum iostat_type io_type);
3716 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3717 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3718 struct f2fs_io_info *fio);
3719 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3720 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3721 block_t old_blkaddr, block_t new_blkaddr,
3722 bool recover_curseg, bool recover_newaddr,
3724 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3725 block_t old_addr, block_t new_addr,
3726 unsigned char version, bool recover_curseg,
3727 bool recover_newaddr);
3728 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3729 block_t old_blkaddr, block_t *new_blkaddr,
3730 struct f2fs_summary *sum, int type,
3731 struct f2fs_io_info *fio);
3732 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,
3733 block_t blkaddr, unsigned int blkcnt);
3734 void f2fs_wait_on_page_writeback(struct page *page,
3735 enum page_type type, bool ordered, bool locked);
3736 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3737 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3739 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3740 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3741 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3742 unsigned int val, int alloc);
3743 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3744 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3745 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3746 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3747 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3748 int __init f2fs_create_segment_manager_caches(void);
3749 void f2fs_destroy_segment_manager_caches(void);
3750 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3751 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3752 unsigned int segno);
3753 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3754 unsigned int segno);
3756 #define DEF_FRAGMENT_SIZE 4
3757 #define MIN_FRAGMENT_SIZE 1
3758 #define MAX_FRAGMENT_SIZE 512
3760 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)
3762 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||
3763 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;
3769 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
3770 unsigned char reason);
3771 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);
3772 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3773 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3774 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3775 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3776 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3777 block_t blkaddr, int type);
3778 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3779 int type, bool sync);
3780 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,
3781 unsigned int ra_blocks);
3782 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3783 long nr_to_write, enum iostat_type io_type);
3784 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3785 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3786 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3787 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3788 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3789 unsigned int devidx, int type);
3790 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3791 unsigned int devidx, int type);
3792 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3793 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3794 void f2fs_add_orphan_inode(struct inode *inode);
3795 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3796 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3797 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3798 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);
3799 void f2fs_remove_dirty_inode(struct inode *inode);
3800 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3802 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3803 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3804 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3805 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3806 int __init f2fs_create_checkpoint_caches(void);
3807 void f2fs_destroy_checkpoint_caches(void);
3808 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3809 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3810 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3811 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3816 int __init f2fs_init_bioset(void);
3817 void f2fs_destroy_bioset(void);
3818 bool f2fs_is_cp_guaranteed(struct page *page);
3819 int f2fs_init_bio_entry_cache(void);
3820 void f2fs_destroy_bio_entry_cache(void);
3821 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio,
3822 enum page_type type);
3823 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);
3824 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3825 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3826 struct inode *inode, struct page *page,
3827 nid_t ino, enum page_type type);
3828 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3829 struct bio **bio, struct page *page);
3830 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3831 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3832 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3833 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3834 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3835 block_t blk_addr, sector_t *sector);
3836 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3837 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3838 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3839 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3840 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3841 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index);
3842 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3843 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3844 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs);
3845 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index,
3846 pgoff_t *next_pgofs);
3847 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3849 struct page *f2fs_get_new_data_page(struct inode *inode,
3850 struct page *ipage, pgoff_t index, bool new_i_size);
3851 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3852 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag);
3853 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3854 u64 start, u64 len);
3855 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3856 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3857 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3858 int f2fs_write_single_data_page(struct page *page, int *submitted,
3859 struct bio **bio, sector_t *last_block,
3860 struct writeback_control *wbc,
3861 enum iostat_type io_type,
3862 int compr_blocks, bool allow_balance);
3863 void f2fs_write_failed(struct inode *inode, loff_t to);
3864 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);
3865 bool f2fs_release_folio(struct folio *folio, gfp_t wait);
3866 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3867 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3868 int f2fs_init_post_read_processing(void);
3869 void f2fs_destroy_post_read_processing(void);
3870 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3871 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3872 extern const struct iomap_ops f2fs_iomap_ops;
3877 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3878 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3879 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3880 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);
3881 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3882 int f2fs_resize_fs(struct file *filp, __u64 block_count);
3883 int __init f2fs_create_garbage_collection_cache(void);
3884 void f2fs_destroy_garbage_collection_cache(void);
3885 /* victim selection function for cleaning and SSR */
3886 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result,
3887 int gc_type, int type, char alloc_mode,
3888 unsigned long long age);
3893 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3894 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3895 int __init f2fs_create_recovery_cache(void);
3896 void f2fs_destroy_recovery_cache(void);
3901 #ifdef CONFIG_F2FS_STAT_FS
3902 struct f2fs_stat_info {
3903 struct list_head stat_list;
3904 struct f2fs_sb_info *sbi;
3905 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3906 int main_area_segs, main_area_sections, main_area_zones;
3907 unsigned long long hit_cached[NR_EXTENT_CACHES];
3908 unsigned long long hit_rbtree[NR_EXTENT_CACHES];
3909 unsigned long long total_ext[NR_EXTENT_CACHES];
3910 unsigned long long hit_total[NR_EXTENT_CACHES];
3911 int ext_tree[NR_EXTENT_CACHES];
3912 int zombie_tree[NR_EXTENT_CACHES];
3913 int ext_node[NR_EXTENT_CACHES];
3914 /* to count memory footprint */
3915 unsigned long long ext_mem[NR_EXTENT_CACHES];
3916 /* for read extent cache */
3917 unsigned long long hit_largest;
3918 /* for block age extent cache */
3919 unsigned long long allocated_data_blocks;
3920 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3921 int ndirty_data, ndirty_qdata;
3922 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3923 int nats, dirty_nats, sits, dirty_sits;
3924 int free_nids, avail_nids, alloc_nids;
3925 int total_count, utilization;
3926 int nr_wb_cp_data, nr_wb_data;
3927 int nr_rd_data, nr_rd_node, nr_rd_meta;
3928 int nr_dio_read, nr_dio_write;
3929 unsigned int io_skip_bggc, other_skip_bggc;
3930 int nr_flushing, nr_flushed, flush_list_empty;
3931 int nr_discarding, nr_discarded;
3933 unsigned int undiscard_blks;
3934 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3935 unsigned int cur_ckpt_time, peak_ckpt_time;
3936 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3937 int compr_inode, swapfile_inode;
3938 unsigned long long compr_blocks;
3939 int aw_cnt, max_aw_cnt;
3940 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3941 unsigned int bimodal, avg_vblocks;
3942 int util_free, util_valid, util_invalid;
3943 int rsvd_segs, overp_segs;
3944 int dirty_count, node_pages, meta_pages, compress_pages;
3945 int compress_page_hit;
3946 int prefree_count, free_segs, free_secs;
3947 int cp_call_count[MAX_CALL_TYPE], cp_count;
3948 int gc_call_count[MAX_CALL_TYPE];
3951 int tot_blks, data_blks, node_blks;
3952 int bg_data_blks, bg_node_blks;
3953 int curseg[NR_CURSEG_TYPE];
3954 int cursec[NR_CURSEG_TYPE];
3955 int curzone[NR_CURSEG_TYPE];
3956 unsigned int dirty_seg[NR_CURSEG_TYPE];
3957 unsigned int full_seg[NR_CURSEG_TYPE];
3958 unsigned int valid_blks[NR_CURSEG_TYPE];
3960 unsigned int meta_count[META_MAX];
3961 unsigned int segment_count[2];
3962 unsigned int block_count[2];
3963 unsigned int inplace_count;
3964 unsigned long long base_mem, cache_mem, page_mem;
3967 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3969 return (struct f2fs_stat_info *)sbi->stat_info;
3972 #define stat_inc_cp_call_count(sbi, foreground) \
3973 atomic_inc(&sbi->cp_call_count[(foreground)])
3974 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++)
3975 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3976 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3977 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3978 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3979 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type]))
3980 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type]))
3981 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3982 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type]))
3983 #define stat_inc_inline_xattr(inode) \
3985 if (f2fs_has_inline_xattr(inode)) \
3986 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3988 #define stat_dec_inline_xattr(inode) \
3990 if (f2fs_has_inline_xattr(inode)) \
3991 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3993 #define stat_inc_inline_inode(inode) \
3995 if (f2fs_has_inline_data(inode)) \
3996 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3998 #define stat_dec_inline_inode(inode) \
4000 if (f2fs_has_inline_data(inode)) \
4001 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
4003 #define stat_inc_inline_dir(inode) \
4005 if (f2fs_has_inline_dentry(inode)) \
4006 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
4008 #define stat_dec_inline_dir(inode) \
4010 if (f2fs_has_inline_dentry(inode)) \
4011 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
4013 #define stat_inc_compr_inode(inode) \
4015 if (f2fs_compressed_file(inode)) \
4016 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
4018 #define stat_dec_compr_inode(inode) \
4020 if (f2fs_compressed_file(inode)) \
4021 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
4023 #define stat_add_compr_blocks(inode, blocks) \
4024 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
4025 #define stat_sub_compr_blocks(inode, blocks) \
4026 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
4027 #define stat_inc_swapfile_inode(inode) \
4028 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))
4029 #define stat_dec_swapfile_inode(inode) \
4030 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))
4031 #define stat_inc_atomic_inode(inode) \
4032 (atomic_inc(&F2FS_I_SB(inode)->atomic_files))
4033 #define stat_dec_atomic_inode(inode) \
4034 (atomic_dec(&F2FS_I_SB(inode)->atomic_files))
4035 #define stat_inc_meta_count(sbi, blkaddr) \
4037 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
4038 atomic_inc(&(sbi)->meta_count[META_CP]); \
4039 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
4040 atomic_inc(&(sbi)->meta_count[META_SIT]); \
4041 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
4042 atomic_inc(&(sbi)->meta_count[META_NAT]); \
4043 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
4044 atomic_inc(&(sbi)->meta_count[META_SSA]); \
4046 #define stat_inc_seg_type(sbi, curseg) \
4047 ((sbi)->segment_count[(curseg)->alloc_type]++)
4048 #define stat_inc_block_count(sbi, curseg) \
4049 ((sbi)->block_count[(curseg)->alloc_type]++)
4050 #define stat_inc_inplace_blocks(sbi) \
4051 (atomic_inc(&(sbi)->inplace_count))
4052 #define stat_update_max_atomic_write(inode) \
4054 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \
4055 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
4057 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
4059 #define stat_inc_gc_call_count(sbi, foreground) \
4060 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++)
4061 #define stat_inc_gc_sec_count(sbi, type, gc_type) \
4062 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++)
4063 #define stat_inc_gc_seg_count(sbi, type, gc_type) \
4064 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++)
4066 #define stat_inc_tot_blk_count(si, blks) \
4067 ((si)->tot_blks += (blks))
4069 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
4071 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4072 stat_inc_tot_blk_count(si, blks); \
4073 si->data_blks += (blks); \
4074 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4077 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
4079 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4080 stat_inc_tot_blk_count(si, blks); \
4081 si->node_blks += (blks); \
4082 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4085 int f2fs_build_stats(struct f2fs_sb_info *sbi);
4086 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
4087 void __init f2fs_create_root_stats(void);
4088 void f2fs_destroy_root_stats(void);
4089 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
4091 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0)
4092 #define stat_inc_cp_count(sbi) do { } while (0)
4093 #define stat_io_skip_bggc_count(sbi) do { } while (0)
4094 #define stat_other_skip_bggc_count(sbi) do { } while (0)
4095 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
4096 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
4097 #define stat_inc_total_hit(sbi, type) do { } while (0)
4098 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0)
4099 #define stat_inc_largest_node_hit(sbi) do { } while (0)
4100 #define stat_inc_cached_node_hit(sbi, type) do { } while (0)
4101 #define stat_inc_inline_xattr(inode) do { } while (0)
4102 #define stat_dec_inline_xattr(inode) do { } while (0)
4103 #define stat_inc_inline_inode(inode) do { } while (0)
4104 #define stat_dec_inline_inode(inode) do { } while (0)
4105 #define stat_inc_inline_dir(inode) do { } while (0)
4106 #define stat_dec_inline_dir(inode) do { } while (0)
4107 #define stat_inc_compr_inode(inode) do { } while (0)
4108 #define stat_dec_compr_inode(inode) do { } while (0)
4109 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
4110 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
4111 #define stat_inc_swapfile_inode(inode) do { } while (0)
4112 #define stat_dec_swapfile_inode(inode) do { } while (0)
4113 #define stat_inc_atomic_inode(inode) do { } while (0)
4114 #define stat_dec_atomic_inode(inode) do { } while (0)
4115 #define stat_update_max_atomic_write(inode) do { } while (0)
4116 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
4117 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
4118 #define stat_inc_block_count(sbi, curseg) do { } while (0)
4119 #define stat_inc_inplace_blocks(sbi) do { } while (0)
4120 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0)
4121 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0)
4122 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0)
4123 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
4124 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
4125 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
4127 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
4128 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
4129 static inline void __init f2fs_create_root_stats(void) { }
4130 static inline void f2fs_destroy_root_stats(void) { }
4131 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
4134 extern const struct file_operations f2fs_dir_operations;
4135 extern const struct file_operations f2fs_file_operations;
4136 extern const struct inode_operations f2fs_file_inode_operations;
4137 extern const struct address_space_operations f2fs_dblock_aops;
4138 extern const struct address_space_operations f2fs_node_aops;
4139 extern const struct address_space_operations f2fs_meta_aops;
4140 extern const struct inode_operations f2fs_dir_inode_operations;
4141 extern const struct inode_operations f2fs_symlink_inode_operations;
4142 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
4143 extern const struct inode_operations f2fs_special_inode_operations;
4144 extern struct kmem_cache *f2fs_inode_entry_slab;
4149 bool f2fs_may_inline_data(struct inode *inode);
4150 bool f2fs_sanity_check_inline_data(struct inode *inode);
4151 bool f2fs_may_inline_dentry(struct inode *inode);
4152 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
4153 void f2fs_truncate_inline_inode(struct inode *inode,
4154 struct page *ipage, u64 from);
4155 int f2fs_read_inline_data(struct inode *inode, struct page *page);
4156 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
4157 int f2fs_convert_inline_inode(struct inode *inode);
4158 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
4159 int f2fs_write_inline_data(struct inode *inode, struct page *page);
4160 int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
4161 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
4162 const struct f2fs_filename *fname,
4163 struct page **res_page);
4164 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
4165 struct page *ipage);
4166 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
4167 struct inode *inode, nid_t ino, umode_t mode);
4168 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
4169 struct page *page, struct inode *dir,
4170 struct inode *inode);
4171 bool f2fs_empty_inline_dir(struct inode *dir);
4172 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
4173 struct fscrypt_str *fstr);
4174 int f2fs_inline_data_fiemap(struct inode *inode,
4175 struct fiemap_extent_info *fieinfo,
4176 __u64 start, __u64 len);
4181 unsigned long f2fs_shrink_count(struct shrinker *shrink,
4182 struct shrink_control *sc);
4183 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
4184 struct shrink_control *sc);
4185 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
4186 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
4191 bool sanity_check_extent_cache(struct inode *inode);
4192 void f2fs_init_extent_tree(struct inode *inode);
4193 void f2fs_drop_extent_tree(struct inode *inode);
4194 void f2fs_destroy_extent_node(struct inode *inode);
4195 void f2fs_destroy_extent_tree(struct inode *inode);
4196 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4197 int __init f2fs_create_extent_cache(void);
4198 void f2fs_destroy_extent_cache(void);
4200 /* read extent cache ops */
4201 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage);
4202 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,
4203 struct extent_info *ei);
4204 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index,
4206 void f2fs_update_read_extent_cache(struct dnode_of_data *dn);
4207 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,
4208 pgoff_t fofs, block_t blkaddr, unsigned int len);
4209 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi,
4212 /* block age extent cache ops */
4213 void f2fs_init_age_extent_tree(struct inode *inode);
4214 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,
4215 struct extent_info *ei);
4216 void f2fs_update_age_extent_cache(struct dnode_of_data *dn);
4217 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,
4218 pgoff_t fofs, unsigned int len);
4219 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi,
4225 #define MIN_RA_MUL 2
4226 #define MAX_RA_MUL 256
4228 int __init f2fs_init_sysfs(void);
4229 void f2fs_exit_sysfs(void);
4230 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4231 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4234 extern const struct fsverity_operations f2fs_verityops;
4239 static inline bool f2fs_encrypted_file(struct inode *inode)
4241 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4244 static inline void f2fs_set_encrypted_inode(struct inode *inode)
4246 #ifdef CONFIG_FS_ENCRYPTION
4247 file_set_encrypt(inode);
4248 f2fs_set_inode_flags(inode);
4253 * Returns true if the reads of the inode's data need to undergo some
4254 * postprocessing step, like decryption or authenticity verification.
4256 static inline bool f2fs_post_read_required(struct inode *inode)
4258 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4259 f2fs_compressed_file(inode);
4265 #ifdef CONFIG_F2FS_FS_COMPRESSION
4266 bool f2fs_is_compressed_page(struct page *page);
4267 struct page *f2fs_compress_control_page(struct page *page);
4268 int f2fs_prepare_compress_overwrite(struct inode *inode,
4269 struct page **pagep, pgoff_t index, void **fsdata);
4270 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4271 pgoff_t index, unsigned copied);
4272 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4273 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4274 bool f2fs_is_compress_backend_ready(struct inode *inode);
4275 bool f2fs_is_compress_level_valid(int alg, int lvl);
4276 int __init f2fs_init_compress_mempool(void);
4277 void f2fs_destroy_compress_mempool(void);
4278 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);
4279 void f2fs_end_read_compressed_page(struct page *page, bool failed,
4280 block_t blkaddr, bool in_task);
4281 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4282 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4283 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,
4284 int index, int nr_pages, bool uptodate);
4285 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
4286 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4287 int f2fs_write_multi_pages(struct compress_ctx *cc,
4289 struct writeback_control *wbc,
4290 enum iostat_type io_type);
4291 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4292 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,
4293 pgoff_t fofs, block_t blkaddr,
4294 unsigned int llen, unsigned int c_len);
4295 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4296 unsigned nr_pages, sector_t *last_block_in_bio,
4297 bool is_readahead, bool for_write);
4298 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4299 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
4301 void f2fs_put_page_dic(struct page *page, bool in_task);
4302 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn,
4303 unsigned int ofs_in_node);
4304 int f2fs_init_compress_ctx(struct compress_ctx *cc);
4305 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4306 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4307 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4308 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4309 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4310 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4311 int __init f2fs_init_compress_cache(void);
4312 void f2fs_destroy_compress_cache(void);
4313 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4314 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4315 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4316 nid_t ino, block_t blkaddr);
4317 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4319 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4320 #define inc_compr_inode_stat(inode) \
4322 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4323 sbi->compr_new_inode++; \
4325 #define add_compr_block_stat(inode, blocks) \
4327 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4328 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4329 sbi->compr_written_block += blocks; \
4330 sbi->compr_saved_block += diff; \
4333 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4334 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4336 if (!f2fs_compressed_file(inode))
4338 /* not support compression */
4341 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; }
4342 static inline struct page *f2fs_compress_control_page(struct page *page)
4345 return ERR_PTR(-EINVAL);
4347 static inline int __init f2fs_init_compress_mempool(void) { return 0; }
4348 static inline void f2fs_destroy_compress_mempool(void) { }
4349 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,
4351 static inline void f2fs_end_read_compressed_page(struct page *page,
4352 bool failed, block_t blkaddr, bool in_task)
4356 static inline void f2fs_put_page_dic(struct page *page, bool in_task)
4360 static inline unsigned int f2fs_cluster_blocks_are_contiguous(
4361 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; }
4362 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
4363 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4364 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4365 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4366 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4367 static inline int __init f2fs_init_compress_cache(void) { return 0; }
4368 static inline void f2fs_destroy_compress_cache(void) { }
4369 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4370 block_t blkaddr) { }
4371 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4372 struct page *page, nid_t ino, block_t blkaddr) { }
4373 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4374 struct page *page, block_t blkaddr) { return false; }
4375 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4377 #define inc_compr_inode_stat(inode) do { } while (0)
4378 static inline void f2fs_update_read_extent_tree_range_compressed(
4379 struct inode *inode,
4380 pgoff_t fofs, block_t blkaddr,
4381 unsigned int llen, unsigned int c_len) { }
4384 static inline int set_compress_context(struct inode *inode)
4386 #ifdef CONFIG_F2FS_FS_COMPRESSION
4387 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4389 F2FS_I(inode)->i_compress_algorithm =
4390 F2FS_OPTION(sbi).compress_algorithm;
4391 F2FS_I(inode)->i_log_cluster_size =
4392 F2FS_OPTION(sbi).compress_log_size;
4393 F2FS_I(inode)->i_compress_flag =
4394 F2FS_OPTION(sbi).compress_chksum ?
4395 BIT(COMPRESS_CHKSUM) : 0;
4396 F2FS_I(inode)->i_cluster_size =
4397 BIT(F2FS_I(inode)->i_log_cluster_size);
4398 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4399 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4400 F2FS_OPTION(sbi).compress_level)
4401 F2FS_I(inode)->i_compress_level =
4402 F2FS_OPTION(sbi).compress_level;
4403 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4404 set_inode_flag(inode, FI_COMPRESSED_FILE);
4405 stat_inc_compr_inode(inode);
4406 inc_compr_inode_stat(inode);
4407 f2fs_mark_inode_dirty_sync(inode, true);
4414 static inline bool f2fs_disable_compressed_file(struct inode *inode)
4416 struct f2fs_inode_info *fi = F2FS_I(inode);
4418 f2fs_down_write(&F2FS_I(inode)->i_sem);
4420 if (!f2fs_compressed_file(inode)) {
4421 f2fs_up_write(&F2FS_I(inode)->i_sem);
4424 if (f2fs_is_mmap_file(inode) ||
4425 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) {
4426 f2fs_up_write(&F2FS_I(inode)->i_sem);
4430 fi->i_flags &= ~F2FS_COMPR_FL;
4431 stat_dec_compr_inode(inode);
4432 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4433 f2fs_mark_inode_dirty_sync(inode, true);
4435 f2fs_up_write(&F2FS_I(inode)->i_sem);
4439 #define F2FS_FEATURE_FUNCS(name, flagname) \
4440 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4442 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4445 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4446 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4447 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4448 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4449 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4450 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4451 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4452 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4453 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4454 F2FS_FEATURE_FUNCS(verity, VERITY);
4455 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4456 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4457 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4458 F2FS_FEATURE_FUNCS(readonly, RO);
4460 #ifdef CONFIG_BLK_DEV_ZONED
4461 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4464 unsigned int zno = blkaddr / sbi->blocks_per_blkz;
4466 return test_bit(zno, FDEV(devi).blkz_seq);
4470 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi,
4471 struct block_device *bdev)
4475 if (!f2fs_is_multi_device(sbi))
4478 for (i = 0; i < sbi->s_ndevs; i++)
4479 if (FDEV(i).bdev == bdev)
4486 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4488 return f2fs_sb_has_blkzoned(sbi);
4491 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4493 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);
4496 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4500 if (!f2fs_is_multi_device(sbi))
4501 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4503 for (i = 0; i < sbi->s_ndevs; i++)
4504 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4509 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4511 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4512 f2fs_hw_should_discard(sbi);
4515 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4519 if (!f2fs_is_multi_device(sbi))
4520 return bdev_read_only(sbi->sb->s_bdev);
4522 for (i = 0; i < sbi->s_ndevs; i++)
4523 if (bdev_read_only(FDEV(i).bdev))
4528 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi)
4530 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi);
4533 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4535 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4538 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)
4540 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;
4543 static inline bool f2fs_may_compress(struct inode *inode)
4545 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4546 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) ||
4547 f2fs_is_mmap_file(inode))
4549 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4552 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4553 u64 blocks, bool add)
4555 struct f2fs_inode_info *fi = F2FS_I(inode);
4556 int diff = fi->i_cluster_size - blocks;
4558 /* don't update i_compr_blocks if saved blocks were released */
4559 if (!add && !atomic_read(&fi->i_compr_blocks))
4563 atomic_add(diff, &fi->i_compr_blocks);
4564 stat_add_compr_blocks(inode, diff);
4566 atomic_sub(diff, &fi->i_compr_blocks);
4567 stat_sub_compr_blocks(inode, diff);
4569 f2fs_mark_inode_dirty_sync(inode, true);
4572 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,
4575 if (!f2fs_is_multi_device(sbi))
4577 if (flag != F2FS_GET_BLOCK_DIO)
4579 return sbi->aligned_blksize;
4582 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4584 return fsverity_active(inode) &&
4585 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4588 #ifdef CONFIG_F2FS_FAULT_INJECTION
4589 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4592 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4595 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4598 if (f2fs_sb_has_quota_ino(sbi))
4600 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4601 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4602 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4608 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4610 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4613 static inline void f2fs_io_schedule_timeout(long timeout)
4615 set_current_state(TASK_UNINTERRUPTIBLE);
4616 io_schedule_timeout(timeout);
4619 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs,
4620 enum page_type type)
4622 if (unlikely(f2fs_cp_error(sbi)))
4625 if (ofs == sbi->page_eio_ofs[type]) {
4626 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)
4627 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4629 sbi->page_eio_ofs[type] = ofs;
4630 sbi->page_eio_cnt[type] = 0;
4634 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi)
4636 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb);
4639 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi,
4640 block_t blkaddr, unsigned int cnt)
4642 bool need_submit = false;
4648 page = find_get_page(META_MAPPING(sbi), blkaddr + i);
4650 if (PageWriteback(page))
4652 f2fs_put_page(page, 0);
4654 } while (++i < cnt && !need_submit);
4657 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode,
4660 truncate_inode_pages_range(META_MAPPING(sbi),
4661 F2FS_BLK_TO_BYTES((loff_t)blkaddr),
4662 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1)));
4665 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi,
4668 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1);
4669 f2fs_invalidate_compress_page(sbi, blkaddr);
4672 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4673 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4675 #endif /* _LINUX_F2FS_H */