1 /* SPDX-License-Identifier: GPL-2.0 */
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
11 #include <linux/uio.h>
12 #include <linux/types.h>
13 #include <linux/page-flags.h>
14 #include <linux/buffer_head.h>
15 #include <linux/slab.h>
16 #include <linux/crc32.h>
17 #include <linux/magic.h>
18 #include <linux/kobject.h>
19 #include <linux/sched.h>
20 #include <linux/cred.h>
21 #include <linux/vmalloc.h>
22 #include <linux/bio.h>
23 #include <linux/blkdev.h>
24 #include <linux/quotaops.h>
25 #include <linux/part_stat.h>
26 #include <crypto/hash.h>
28 #include <linux/fscrypt.h>
29 #include <linux/fsverity.h>
31 #ifdef CONFIG_F2FS_CHECK_FS
32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
34 #define f2fs_bug_on(sbi, condition) \
36 if (unlikely(condition)) { \
38 set_sbi_flag(sbi, SBI_NEED_FSCK); \
62 #ifdef CONFIG_F2FS_FAULT_INJECTION
63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
65 struct f2fs_fault_info {
67 unsigned int inject_rate;
68 unsigned int inject_type;
71 extern const char *f2fs_fault_name[FAULT_MAX];
72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
79 #define F2FS_MOUNT_DISCARD 0x00000004
80 #define F2FS_MOUNT_NOHEAP 0x00000008
81 #define F2FS_MOUNT_XATTR_USER 0x00000010
82 #define F2FS_MOUNT_POSIX_ACL 0x00000020
83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
85 #define F2FS_MOUNT_INLINE_DATA 0x00000100
86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
88 #define F2FS_MOUNT_NOBARRIER 0x00000800
89 #define F2FS_MOUNT_FASTBOOT 0x00001000
90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
93 #define F2FS_MOUNT_USRQUOTA 0x00080000
94 #define F2FS_MOUNT_GRPQUOTA 0x00100000
95 #define F2FS_MOUNT_PRJQUOTA 0x00200000
96 #define F2FS_MOUNT_QUOTA 0x00400000
97 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
98 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
99 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
100 #define F2FS_MOUNT_NORECOVERY 0x04000000
101 #define F2FS_MOUNT_ATGC 0x08000000
103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
109 typecheck(unsigned long long, b) && \
110 ((long long)((a) - (b)) > 0))
112 typedef u32 block_t; /*
113 * should not change u32, since it is the on-disk block
114 * address format, __le32.
118 #define COMPRESS_EXT_NUM 16
120 struct f2fs_mount_info {
122 int write_io_size_bits; /* Write IO size bits */
123 block_t root_reserved_blocks; /* root reserved blocks */
124 kuid_t s_resuid; /* reserved blocks for uid */
125 kgid_t s_resgid; /* reserved blocks for gid */
126 int active_logs; /* # of active logs */
127 int inline_xattr_size; /* inline xattr size */
128 #ifdef CONFIG_F2FS_FAULT_INJECTION
129 struct f2fs_fault_info fault_info; /* For fault injection */
132 /* Names of quota files with journalled quota */
133 char *s_qf_names[MAXQUOTAS];
134 int s_jquota_fmt; /* Format of quota to use */
136 /* For which write hints are passed down to block layer */
138 int alloc_mode; /* segment allocation policy */
139 int fsync_mode; /* fsync policy */
140 int fs_mode; /* fs mode: LFS or ADAPTIVE */
141 int bggc_mode; /* bggc mode: off, on or sync */
142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
143 block_t unusable_cap_perc; /* percentage for cap */
144 block_t unusable_cap; /* Amount of space allowed to be
145 * unusable when disabling checkpoint
148 /* For compression */
149 unsigned char compress_algorithm; /* algorithm type */
150 unsigned compress_log_size; /* cluster log size */
151 unsigned char compress_ext_cnt; /* extension count */
152 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
155 #define F2FS_FEATURE_ENCRYPT 0x0001
156 #define F2FS_FEATURE_BLKZONED 0x0002
157 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
158 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
159 #define F2FS_FEATURE_PRJQUOTA 0x0010
160 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
161 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
162 #define F2FS_FEATURE_QUOTA_INO 0x0080
163 #define F2FS_FEATURE_INODE_CRTIME 0x0100
164 #define F2FS_FEATURE_LOST_FOUND 0x0200
165 #define F2FS_FEATURE_VERITY 0x0400
166 #define F2FS_FEATURE_SB_CHKSUM 0x0800
167 #define F2FS_FEATURE_CASEFOLD 0x1000
168 #define F2FS_FEATURE_COMPRESSION 0x2000
170 #define __F2FS_HAS_FEATURE(raw_super, mask) \
171 ((raw_super->feature & cpu_to_le32(mask)) != 0)
172 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
173 #define F2FS_SET_FEATURE(sbi, mask) \
174 (sbi->raw_super->feature |= cpu_to_le32(mask))
175 #define F2FS_CLEAR_FEATURE(sbi, mask) \
176 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
179 * Default values for user and/or group using reserved blocks
181 #define F2FS_DEF_RESUID 0
182 #define F2FS_DEF_RESGID 0
185 * For checkpoint manager
192 #define CP_UMOUNT 0x00000001
193 #define CP_FASTBOOT 0x00000002
194 #define CP_SYNC 0x00000004
195 #define CP_RECOVERY 0x00000008
196 #define CP_DISCARD 0x00000010
197 #define CP_TRIMMED 0x00000020
198 #define CP_PAUSE 0x00000040
199 #define CP_RESIZE 0x00000080
201 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
202 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
203 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
204 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
205 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
206 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
207 #define DEF_CP_INTERVAL 60 /* 60 secs */
208 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
209 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
210 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
211 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
221 * indicate meta/data type
230 DATA_GENERIC, /* check range only */
231 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
232 DATA_GENERIC_ENHANCE_READ, /*
233 * strong check on range and segment
234 * bitmap but no warning due to race
235 * condition of read on truncated area
238 DATA_GENERIC_ENHANCE_UPDATE, /*
239 * strong check on range and segment
240 * bitmap for update case
245 /* for the list of ino */
247 ORPHAN_INO, /* for orphan ino list */
248 APPEND_INO, /* for append ino list */
249 UPDATE_INO, /* for update ino list */
250 TRANS_DIR_INO, /* for trasactions dir ino list */
251 FLUSH_INO, /* for multiple device flushing */
252 MAX_INO_ENTRY, /* max. list */
256 struct list_head list; /* list head */
257 nid_t ino; /* inode number */
258 unsigned int dirty_device; /* dirty device bitmap */
261 /* for the list of inodes to be GCed */
263 struct list_head list; /* list head */
264 struct inode *inode; /* vfs inode pointer */
267 struct fsync_node_entry {
268 struct list_head list; /* list head */
269 struct page *page; /* warm node page pointer */
270 unsigned int seq_id; /* sequence id */
273 /* for the bitmap indicate blocks to be discarded */
274 struct discard_entry {
275 struct list_head list; /* list head */
276 block_t start_blkaddr; /* start blockaddr of current segment */
277 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
280 /* default discard granularity of inner discard thread, unit: block count */
281 #define DEFAULT_DISCARD_GRANULARITY 16
283 /* max discard pend list number */
284 #define MAX_PLIST_NUM 512
285 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
286 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
289 D_PREP, /* initial */
290 D_PARTIAL, /* partially submitted */
291 D_SUBMIT, /* all submitted */
292 D_DONE, /* finished */
295 struct discard_info {
296 block_t lstart; /* logical start address */
297 block_t len; /* length */
298 block_t start; /* actual start address in dev */
302 struct rb_node rb_node; /* rb node located in rb-tree */
305 block_t lstart; /* logical start address */
306 block_t len; /* length */
307 block_t start; /* actual start address in dev */
309 struct discard_info di; /* discard info */
312 struct list_head list; /* command list */
313 struct completion wait; /* compleation */
314 struct block_device *bdev; /* bdev */
315 unsigned short ref; /* reference count */
316 unsigned char state; /* state */
317 unsigned char queued; /* queued discard */
318 int error; /* bio error */
319 spinlock_t lock; /* for state/bio_ref updating */
320 unsigned short bio_ref; /* bio reference count */
331 struct discard_policy {
332 int type; /* type of discard */
333 unsigned int min_interval; /* used for candidates exist */
334 unsigned int mid_interval; /* used for device busy */
335 unsigned int max_interval; /* used for candidates not exist */
336 unsigned int max_requests; /* # of discards issued per round */
337 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
338 bool io_aware; /* issue discard in idle time */
339 bool sync; /* submit discard with REQ_SYNC flag */
340 bool ordered; /* issue discard by lba order */
341 bool timeout; /* discard timeout for put_super */
342 unsigned int granularity; /* discard granularity */
345 struct discard_cmd_control {
346 struct task_struct *f2fs_issue_discard; /* discard thread */
347 struct list_head entry_list; /* 4KB discard entry list */
348 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
349 struct list_head wait_list; /* store on-flushing entries */
350 struct list_head fstrim_list; /* in-flight discard from fstrim */
351 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
352 unsigned int discard_wake; /* to wake up discard thread */
353 struct mutex cmd_lock;
354 unsigned int nr_discards; /* # of discards in the list */
355 unsigned int max_discards; /* max. discards to be issued */
356 unsigned int discard_granularity; /* discard granularity */
357 unsigned int undiscard_blks; /* # of undiscard blocks */
358 unsigned int next_pos; /* next discard position */
359 atomic_t issued_discard; /* # of issued discard */
360 atomic_t queued_discard; /* # of queued discard */
361 atomic_t discard_cmd_cnt; /* # of cached cmd count */
362 struct rb_root_cached root; /* root of discard rb-tree */
363 bool rbtree_check; /* config for consistence check */
366 /* for the list of fsync inodes, used only during recovery */
367 struct fsync_inode_entry {
368 struct list_head list; /* list head */
369 struct inode *inode; /* vfs inode pointer */
370 block_t blkaddr; /* block address locating the last fsync */
371 block_t last_dentry; /* block address locating the last dentry */
374 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
375 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
377 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
378 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
379 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
380 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
382 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
383 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
385 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
387 int before = nats_in_cursum(journal);
389 journal->n_nats = cpu_to_le16(before + i);
393 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
395 int before = sits_in_cursum(journal);
397 journal->n_sits = cpu_to_le16(before + i);
401 static inline bool __has_cursum_space(struct f2fs_journal *journal,
404 if (type == NAT_JOURNAL)
405 return size <= MAX_NAT_JENTRIES(journal);
406 return size <= MAX_SIT_JENTRIES(journal);
409 /* for inline stuff */
410 #define DEF_INLINE_RESERVED_SIZE 1
411 static inline int get_extra_isize(struct inode *inode);
412 static inline int get_inline_xattr_addrs(struct inode *inode);
413 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
414 (CUR_ADDRS_PER_INODE(inode) - \
415 get_inline_xattr_addrs(inode) - \
416 DEF_INLINE_RESERVED_SIZE))
419 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
420 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
422 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
423 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
424 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
425 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
426 NR_INLINE_DENTRY(inode) + \
427 INLINE_DENTRY_BITMAP_SIZE(inode)))
430 * For INODE and NODE manager
432 /* for directory operations */
434 struct f2fs_filename {
436 * The filename the user specified. This is NULL for some
437 * filesystem-internal operations, e.g. converting an inline directory
438 * to a non-inline one, or roll-forward recovering an encrypted dentry.
440 const struct qstr *usr_fname;
443 * The on-disk filename. For encrypted directories, this is encrypted.
444 * This may be NULL for lookups in an encrypted dir without the key.
446 struct fscrypt_str disk_name;
448 /* The dirhash of this filename */
451 #ifdef CONFIG_FS_ENCRYPTION
453 * For lookups in encrypted directories: either the buffer backing
454 * disk_name, or a buffer that holds the decoded no-key name.
456 struct fscrypt_str crypto_buf;
458 #ifdef CONFIG_UNICODE
460 * For casefolded directories: the casefolded name, but it's left NULL
461 * if the original name is not valid Unicode or if the filesystem is
462 * doing an internal operation where usr_fname is also NULL. In these
463 * cases we fall back to treating the name as an opaque byte sequence.
465 struct fscrypt_str cf_name;
469 struct f2fs_dentry_ptr {
472 struct f2fs_dir_entry *dentry;
473 __u8 (*filename)[F2FS_SLOT_LEN];
478 static inline void make_dentry_ptr_block(struct inode *inode,
479 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
482 d->max = NR_DENTRY_IN_BLOCK;
483 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
484 d->bitmap = t->dentry_bitmap;
485 d->dentry = t->dentry;
486 d->filename = t->filename;
489 static inline void make_dentry_ptr_inline(struct inode *inode,
490 struct f2fs_dentry_ptr *d, void *t)
492 int entry_cnt = NR_INLINE_DENTRY(inode);
493 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
494 int reserved_size = INLINE_RESERVED_SIZE(inode);
498 d->nr_bitmap = bitmap_size;
500 d->dentry = t + bitmap_size + reserved_size;
501 d->filename = t + bitmap_size + reserved_size +
502 SIZE_OF_DIR_ENTRY * entry_cnt;
506 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
507 * as its node offset to distinguish from index node blocks.
508 * But some bits are used to mark the node block.
510 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
513 ALLOC_NODE, /* allocate a new node page if needed */
514 LOOKUP_NODE, /* look up a node without readahead */
516 * look up a node with readahead called
521 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
523 /* congestion wait timeout value, default: 20ms */
524 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
526 /* maximum retry quota flush count */
527 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
529 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
531 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
533 /* for in-memory extent cache entry */
534 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
536 /* number of extent info in extent cache we try to shrink */
537 #define EXTENT_CACHE_SHRINK_NUMBER 128
540 struct rb_node rb_node; /* rb node located in rb-tree */
543 unsigned int ofs; /* start offset of the entry */
544 unsigned int len; /* length of the entry */
546 unsigned long long key; /* 64-bits key */
551 unsigned int fofs; /* start offset in a file */
552 unsigned int len; /* length of the extent */
553 u32 blk; /* start block address of the extent */
557 struct rb_node rb_node; /* rb node located in rb-tree */
558 struct extent_info ei; /* extent info */
559 struct list_head list; /* node in global extent list of sbi */
560 struct extent_tree *et; /* extent tree pointer */
564 nid_t ino; /* inode number */
565 struct rb_root_cached root; /* root of extent info rb-tree */
566 struct extent_node *cached_en; /* recently accessed extent node */
567 struct extent_info largest; /* largested extent info */
568 struct list_head list; /* to be used by sbi->zombie_list */
569 rwlock_t lock; /* protect extent info rb-tree */
570 atomic_t node_cnt; /* # of extent node in rb-tree*/
571 bool largest_updated; /* largest extent updated */
575 * This structure is taken from ext4_map_blocks.
577 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
579 #define F2FS_MAP_NEW (1 << BH_New)
580 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
581 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
582 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
585 struct f2fs_map_blocks {
589 unsigned int m_flags;
590 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
591 pgoff_t *m_next_extent; /* point to next possible extent */
593 bool m_may_create; /* indicate it is from write path */
596 /* for flag in get_data_block */
598 F2FS_GET_BLOCK_DEFAULT,
599 F2FS_GET_BLOCK_FIEMAP,
602 F2FS_GET_BLOCK_PRE_DIO,
603 F2FS_GET_BLOCK_PRE_AIO,
604 F2FS_GET_BLOCK_PRECACHE,
608 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
610 #define FADVISE_COLD_BIT 0x01
611 #define FADVISE_LOST_PINO_BIT 0x02
612 #define FADVISE_ENCRYPT_BIT 0x04
613 #define FADVISE_ENC_NAME_BIT 0x08
614 #define FADVISE_KEEP_SIZE_BIT 0x10
615 #define FADVISE_HOT_BIT 0x20
616 #define FADVISE_VERITY_BIT 0x40
618 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
620 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
621 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
622 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
623 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
624 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
625 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
626 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
627 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
628 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
629 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
630 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
631 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
632 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
633 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
634 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
635 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
636 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
637 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
639 #define DEF_DIR_LEVEL 0
647 /* used for f2fs_inode_info->flags */
649 FI_NEW_INODE, /* indicate newly allocated inode */
650 FI_DIRTY_INODE, /* indicate inode is dirty or not */
651 FI_AUTO_RECOVER, /* indicate inode is recoverable */
652 FI_DIRTY_DIR, /* indicate directory has dirty pages */
653 FI_INC_LINK, /* need to increment i_nlink */
654 FI_ACL_MODE, /* indicate acl mode */
655 FI_NO_ALLOC, /* should not allocate any blocks */
656 FI_FREE_NID, /* free allocated nide */
657 FI_NO_EXTENT, /* not to use the extent cache */
658 FI_INLINE_XATTR, /* used for inline xattr */
659 FI_INLINE_DATA, /* used for inline data*/
660 FI_INLINE_DENTRY, /* used for inline dentry */
661 FI_APPEND_WRITE, /* inode has appended data */
662 FI_UPDATE_WRITE, /* inode has in-place-update data */
663 FI_NEED_IPU, /* used for ipu per file */
664 FI_ATOMIC_FILE, /* indicate atomic file */
665 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
666 FI_VOLATILE_FILE, /* indicate volatile file */
667 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
668 FI_DROP_CACHE, /* drop dirty page cache */
669 FI_DATA_EXIST, /* indicate data exists */
670 FI_INLINE_DOTS, /* indicate inline dot dentries */
671 FI_DO_DEFRAG, /* indicate defragment is running */
672 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
673 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
674 FI_HOT_DATA, /* indicate file is hot */
675 FI_EXTRA_ATTR, /* indicate file has extra attribute */
676 FI_PROJ_INHERIT, /* indicate file inherits projectid */
677 FI_PIN_FILE, /* indicate file should not be gced */
678 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
679 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
680 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
681 FI_MMAP_FILE, /* indicate file was mmapped */
682 FI_MAX, /* max flag, never be used */
685 struct f2fs_inode_info {
686 struct inode vfs_inode; /* serve a vfs inode */
687 unsigned long i_flags; /* keep an inode flags for ioctl */
688 unsigned char i_advise; /* use to give file attribute hints */
689 unsigned char i_dir_level; /* use for dentry level for large dir */
690 unsigned int i_current_depth; /* only for directory depth */
691 /* for gc failure statistic */
692 unsigned int i_gc_failures[MAX_GC_FAILURE];
693 unsigned int i_pino; /* parent inode number */
694 umode_t i_acl_mode; /* keep file acl mode temporarily */
696 /* Use below internally in f2fs*/
697 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
698 struct rw_semaphore i_sem; /* protect fi info */
699 atomic_t dirty_pages; /* # of dirty pages */
700 f2fs_hash_t chash; /* hash value of given file name */
701 unsigned int clevel; /* maximum level of given file name */
702 struct task_struct *task; /* lookup and create consistency */
703 struct task_struct *cp_task; /* separate cp/wb IO stats*/
704 struct task_struct *wb_task; /* indicate inode is in context of writeback */
705 nid_t i_xattr_nid; /* node id that contains xattrs */
706 loff_t last_disk_size; /* lastly written file size */
707 spinlock_t i_size_lock; /* protect last_disk_size */
710 struct dquot *i_dquot[MAXQUOTAS];
712 /* quota space reservation, managed internally by quota code */
713 qsize_t i_reserved_quota;
715 struct list_head dirty_list; /* dirty list for dirs and files */
716 struct list_head gdirty_list; /* linked in global dirty list */
717 struct list_head inmem_ilist; /* list for inmem inodes */
718 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
719 struct task_struct *inmem_task; /* store inmemory task */
720 struct mutex inmem_lock; /* lock for inmemory pages */
721 pgoff_t ra_offset; /* ongoing readahead offset */
722 struct extent_tree *extent_tree; /* cached extent_tree entry */
724 /* avoid racing between foreground op and gc */
725 struct rw_semaphore i_gc_rwsem[2];
726 struct rw_semaphore i_mmap_sem;
727 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
729 int i_extra_isize; /* size of extra space located in i_addr */
730 kprojid_t i_projid; /* id for project quota */
731 int i_inline_xattr_size; /* inline xattr size */
732 struct timespec64 i_crtime; /* inode creation time */
733 struct timespec64 i_disk_time[4];/* inode disk times */
735 /* for file compress */
736 atomic_t i_compr_blocks; /* # of compressed blocks */
737 unsigned char i_compress_algorithm; /* algorithm type */
738 unsigned char i_log_cluster_size; /* log of cluster size */
739 unsigned int i_cluster_size; /* cluster size */
742 static inline void get_extent_info(struct extent_info *ext,
743 struct f2fs_extent *i_ext)
745 ext->fofs = le32_to_cpu(i_ext->fofs);
746 ext->blk = le32_to_cpu(i_ext->blk);
747 ext->len = le32_to_cpu(i_ext->len);
750 static inline void set_raw_extent(struct extent_info *ext,
751 struct f2fs_extent *i_ext)
753 i_ext->fofs = cpu_to_le32(ext->fofs);
754 i_ext->blk = cpu_to_le32(ext->blk);
755 i_ext->len = cpu_to_le32(ext->len);
758 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
759 u32 blk, unsigned int len)
766 static inline bool __is_discard_mergeable(struct discard_info *back,
767 struct discard_info *front, unsigned int max_len)
769 return (back->lstart + back->len == front->lstart) &&
770 (back->len + front->len <= max_len);
773 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
774 struct discard_info *back, unsigned int max_len)
776 return __is_discard_mergeable(back, cur, max_len);
779 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
780 struct discard_info *front, unsigned int max_len)
782 return __is_discard_mergeable(cur, front, max_len);
785 static inline bool __is_extent_mergeable(struct extent_info *back,
786 struct extent_info *front)
788 return (back->fofs + back->len == front->fofs &&
789 back->blk + back->len == front->blk);
792 static inline bool __is_back_mergeable(struct extent_info *cur,
793 struct extent_info *back)
795 return __is_extent_mergeable(back, cur);
798 static inline bool __is_front_mergeable(struct extent_info *cur,
799 struct extent_info *front)
801 return __is_extent_mergeable(cur, front);
804 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
805 static inline void __try_update_largest_extent(struct extent_tree *et,
806 struct extent_node *en)
808 if (en->ei.len > et->largest.len) {
809 et->largest = en->ei;
810 et->largest_updated = true;
815 * For free nid management
818 FREE_NID, /* newly added to free nid list */
819 PREALLOC_NID, /* it is preallocated */
830 struct f2fs_nm_info {
831 block_t nat_blkaddr; /* base disk address of NAT */
832 nid_t max_nid; /* maximum possible node ids */
833 nid_t available_nids; /* # of available node ids */
834 nid_t next_scan_nid; /* the next nid to be scanned */
835 unsigned int ram_thresh; /* control the memory footprint */
836 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
837 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
839 /* NAT cache management */
840 struct radix_tree_root nat_root;/* root of the nat entry cache */
841 struct radix_tree_root nat_set_root;/* root of the nat set cache */
842 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
843 struct list_head nat_entries; /* cached nat entry list (clean) */
844 spinlock_t nat_list_lock; /* protect clean nat entry list */
845 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
846 unsigned int nat_blocks; /* # of nat blocks */
848 /* free node ids management */
849 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
850 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
851 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
852 spinlock_t nid_list_lock; /* protect nid lists ops */
853 struct mutex build_lock; /* lock for build free nids */
854 unsigned char **free_nid_bitmap;
855 unsigned char *nat_block_bitmap;
856 unsigned short *free_nid_count; /* free nid count of NAT block */
859 char *nat_bitmap; /* NAT bitmap pointer */
861 unsigned int nat_bits_blocks; /* # of nat bits blocks */
862 unsigned char *nat_bits; /* NAT bits blocks */
863 unsigned char *full_nat_bits; /* full NAT pages */
864 unsigned char *empty_nat_bits; /* empty NAT pages */
865 #ifdef CONFIG_F2FS_CHECK_FS
866 char *nat_bitmap_mir; /* NAT bitmap mirror */
868 int bitmap_size; /* bitmap size */
872 * this structure is used as one of function parameters.
873 * all the information are dedicated to a given direct node block determined
874 * by the data offset in a file.
876 struct dnode_of_data {
877 struct inode *inode; /* vfs inode pointer */
878 struct page *inode_page; /* its inode page, NULL is possible */
879 struct page *node_page; /* cached direct node page */
880 nid_t nid; /* node id of the direct node block */
881 unsigned int ofs_in_node; /* data offset in the node page */
882 bool inode_page_locked; /* inode page is locked or not */
883 bool node_changed; /* is node block changed */
884 char cur_level; /* level of hole node page */
885 char max_level; /* level of current page located */
886 block_t data_blkaddr; /* block address of the node block */
889 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
890 struct page *ipage, struct page *npage, nid_t nid)
892 memset(dn, 0, sizeof(*dn));
894 dn->inode_page = ipage;
895 dn->node_page = npage;
902 * By default, there are 6 active log areas across the whole main area.
903 * When considering hot and cold data separation to reduce cleaning overhead,
904 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
906 * In the current design, you should not change the numbers intentionally.
907 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
908 * logs individually according to the underlying devices. (default: 6)
909 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
910 * data and 8 for node logs.
912 #define NR_CURSEG_DATA_TYPE (3)
913 #define NR_CURSEG_NODE_TYPE (3)
914 #define NR_CURSEG_INMEM_TYPE (2)
915 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
916 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
919 CURSEG_HOT_DATA = 0, /* directory entry blocks */
920 CURSEG_WARM_DATA, /* data blocks */
921 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
922 CURSEG_HOT_NODE, /* direct node blocks of directory files */
923 CURSEG_WARM_NODE, /* direct node blocks of normal files */
924 CURSEG_COLD_NODE, /* indirect node blocks */
925 NR_PERSISTENT_LOG, /* number of persistent log */
926 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
927 /* pinned file that needs consecutive block address */
928 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
929 NO_CHECK_TYPE, /* number of persistent & inmem log */
933 struct completion wait;
934 struct llist_node llnode;
939 struct flush_cmd_control {
940 struct task_struct *f2fs_issue_flush; /* flush thread */
941 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
942 atomic_t issued_flush; /* # of issued flushes */
943 atomic_t queued_flush; /* # of queued flushes */
944 struct llist_head issue_list; /* list for command issue */
945 struct llist_node *dispatch_list; /* list for command dispatch */
948 struct f2fs_sm_info {
949 struct sit_info *sit_info; /* whole segment information */
950 struct free_segmap_info *free_info; /* free segment information */
951 struct dirty_seglist_info *dirty_info; /* dirty segment information */
952 struct curseg_info *curseg_array; /* active segment information */
954 struct rw_semaphore curseg_lock; /* for preventing curseg change */
956 block_t seg0_blkaddr; /* block address of 0'th segment */
957 block_t main_blkaddr; /* start block address of main area */
958 block_t ssa_blkaddr; /* start block address of SSA area */
960 unsigned int segment_count; /* total # of segments */
961 unsigned int main_segments; /* # of segments in main area */
962 unsigned int reserved_segments; /* # of reserved segments */
963 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
964 unsigned int ovp_segments; /* # of overprovision segments */
966 /* a threshold to reclaim prefree segments */
967 unsigned int rec_prefree_segments;
969 /* for batched trimming */
970 unsigned int trim_sections; /* # of sections to trim */
972 struct list_head sit_entry_set; /* sit entry set list */
974 unsigned int ipu_policy; /* in-place-update policy */
975 unsigned int min_ipu_util; /* in-place-update threshold */
976 unsigned int min_fsync_blocks; /* threshold for fsync */
977 unsigned int min_seq_blocks; /* threshold for sequential blocks */
978 unsigned int min_hot_blocks; /* threshold for hot block allocation */
979 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
981 /* for flush command control */
982 struct flush_cmd_control *fcc_info;
984 /* for discard command control */
985 struct discard_cmd_control *dcc_info;
992 * COUNT_TYPE for monitoring
994 * f2fs monitors the number of several block types such as on-writeback,
995 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
997 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1017 * The below are the page types of bios used in submit_bio().
1018 * The available types are:
1019 * DATA User data pages. It operates as async mode.
1020 * NODE Node pages. It operates as async mode.
1021 * META FS metadata pages such as SIT, NAT, CP.
1022 * NR_PAGE_TYPE The number of page types.
1023 * META_FLUSH Make sure the previous pages are written
1024 * with waiting the bio's completion
1025 * ... Only can be used with META.
1027 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1030 NODE = 1, /* should not change this */
1034 INMEM, /* the below types are used by tracepoints only. */
1043 HOT = 0, /* must be zero for meta bio */
1049 enum need_lock_type {
1055 enum cp_reason_type {
1071 APP_DIRECT_IO, /* app direct write IOs */
1072 APP_BUFFERED_IO, /* app buffered write IOs */
1073 APP_WRITE_IO, /* app write IOs */
1074 APP_MAPPED_IO, /* app mapped IOs */
1075 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1076 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1077 FS_META_IO, /* meta IOs from kworker/reclaimer */
1078 FS_GC_DATA_IO, /* data IOs from forground gc */
1079 FS_GC_NODE_IO, /* node IOs from forground gc */
1080 FS_CP_DATA_IO, /* data IOs from checkpoint */
1081 FS_CP_NODE_IO, /* node IOs from checkpoint */
1082 FS_CP_META_IO, /* meta IOs from checkpoint */
1085 APP_DIRECT_READ_IO, /* app direct read IOs */
1086 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1087 APP_READ_IO, /* app read IOs */
1088 APP_MAPPED_READ_IO, /* app mapped read IOs */
1089 FS_DATA_READ_IO, /* data read IOs */
1090 FS_GDATA_READ_IO, /* data read IOs from background gc */
1091 FS_CDATA_READ_IO, /* compressed data read IOs */
1092 FS_NODE_READ_IO, /* node read IOs */
1093 FS_META_READ_IO, /* meta read IOs */
1096 FS_DISCARD, /* discard */
1100 struct f2fs_io_info {
1101 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1102 nid_t ino; /* inode number */
1103 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1104 enum temp_type temp; /* contains HOT/WARM/COLD */
1105 int op; /* contains REQ_OP_ */
1106 int op_flags; /* req_flag_bits */
1107 block_t new_blkaddr; /* new block address to be written */
1108 block_t old_blkaddr; /* old block address before Cow */
1109 struct page *page; /* page to be written */
1110 struct page *encrypted_page; /* encrypted page */
1111 struct page *compressed_page; /* compressed page */
1112 struct list_head list; /* serialize IOs */
1113 bool submitted; /* indicate IO submission */
1114 int need_lock; /* indicate we need to lock cp_rwsem */
1115 bool in_list; /* indicate fio is in io_list */
1116 bool is_por; /* indicate IO is from recovery or not */
1117 bool retry; /* need to reallocate block address */
1118 int compr_blocks; /* # of compressed block addresses */
1119 bool encrypted; /* indicate file is encrypted */
1120 enum iostat_type io_type; /* io type */
1121 struct writeback_control *io_wbc; /* writeback control */
1122 struct bio **bio; /* bio for ipu */
1123 sector_t *last_block; /* last block number in bio */
1124 unsigned char version; /* version of the node */
1129 struct list_head list;
1132 #define is_read_io(rw) ((rw) == READ)
1133 struct f2fs_bio_info {
1134 struct f2fs_sb_info *sbi; /* f2fs superblock */
1135 struct bio *bio; /* bios to merge */
1136 sector_t last_block_in_bio; /* last block number */
1137 struct f2fs_io_info fio; /* store buffered io info. */
1138 struct rw_semaphore io_rwsem; /* blocking op for bio */
1139 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1140 struct list_head io_list; /* track fios */
1141 struct list_head bio_list; /* bio entry list head */
1142 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1145 #define FDEV(i) (sbi->devs[i])
1146 #define RDEV(i) (raw_super->devs[i])
1147 struct f2fs_dev_info {
1148 struct block_device *bdev;
1149 char path[MAX_PATH_LEN];
1150 unsigned int total_segments;
1153 #ifdef CONFIG_BLK_DEV_ZONED
1154 unsigned int nr_blkz; /* Total number of zones */
1155 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1160 DIR_INODE, /* for dirty dir inode */
1161 FILE_INODE, /* for dirty regular/symlink inode */
1162 DIRTY_META, /* for all dirtied inode metadata */
1163 ATOMIC_FILE, /* for all atomic files */
1167 /* for inner inode cache management */
1168 struct inode_management {
1169 struct radix_tree_root ino_root; /* ino entry array */
1170 spinlock_t ino_lock; /* for ino entry lock */
1171 struct list_head ino_list; /* inode list head */
1172 unsigned long ino_num; /* number of entries */
1176 struct atgc_management {
1177 bool atgc_enabled; /* ATGC is enabled or not */
1178 struct rb_root_cached root; /* root of victim rb-tree */
1179 struct list_head victim_list; /* linked with all victim entries */
1180 unsigned int victim_count; /* victim count in rb-tree */
1181 unsigned int candidate_ratio; /* candidate ratio */
1182 unsigned int max_candidate_count; /* max candidate count */
1183 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1184 unsigned long long age_threshold; /* age threshold */
1187 /* For s_flag in struct f2fs_sb_info */
1189 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1190 SBI_IS_CLOSE, /* specify unmounting */
1191 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1192 SBI_POR_DOING, /* recovery is doing or not */
1193 SBI_NEED_SB_WRITE, /* need to recover superblock */
1194 SBI_NEED_CP, /* need to checkpoint */
1195 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1196 SBI_IS_RECOVERED, /* recovered orphan/data */
1197 SBI_CP_DISABLED, /* CP was disabled last mount */
1198 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1199 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1200 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1201 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1202 SBI_IS_RESIZEFS, /* resizefs is in process */
1211 UMOUNT_DISCARD_TIMEOUT,
1225 BGGC_MODE_ON, /* background gc is on */
1226 BGGC_MODE_OFF, /* background gc is off */
1228 * background gc is on, migrating blocks
1229 * like foreground gc
1234 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1235 FS_MODE_LFS, /* use lfs allocation only */
1239 WHINT_MODE_OFF, /* not pass down write hints */
1240 WHINT_MODE_USER, /* try to pass down hints given by users */
1241 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1245 ALLOC_MODE_DEFAULT, /* stay default */
1246 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1250 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1251 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1252 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1256 * this value is set in page as a private data which indicate that
1257 * the page is atomically written, and it is in inmem_pages list.
1259 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1260 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1262 #define IS_ATOMIC_WRITTEN_PAGE(page) \
1263 (page_private(page) == ATOMIC_WRITTEN_PAGE)
1264 #define IS_DUMMY_WRITTEN_PAGE(page) \
1265 (page_private(page) == DUMMY_WRITTEN_PAGE)
1267 #ifdef CONFIG_F2FS_IO_TRACE
1268 #define IS_IO_TRACED_PAGE(page) \
1269 (page_private(page) > 0 && \
1270 page_private(page) < (unsigned long)PID_MAX_LIMIT)
1272 #define IS_IO_TRACED_PAGE(page) (0)
1275 /* For compression */
1276 enum compress_algorithm_type {
1284 #define COMPRESS_DATA_RESERVED_SIZE 5
1285 struct compress_data {
1286 __le32 clen; /* compressed data size */
1287 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1288 u8 cdata[]; /* compressed data */
1291 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1293 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1295 /* compress context */
1296 struct compress_ctx {
1297 struct inode *inode; /* inode the context belong to */
1298 pgoff_t cluster_idx; /* cluster index number */
1299 unsigned int cluster_size; /* page count in cluster */
1300 unsigned int log_cluster_size; /* log of cluster size */
1301 struct page **rpages; /* pages store raw data in cluster */
1302 unsigned int nr_rpages; /* total page number in rpages */
1303 struct page **cpages; /* pages store compressed data in cluster */
1304 unsigned int nr_cpages; /* total page number in cpages */
1305 void *rbuf; /* virtual mapped address on rpages */
1306 struct compress_data *cbuf; /* virtual mapped address on cpages */
1307 size_t rlen; /* valid data length in rbuf */
1308 size_t clen; /* valid data length in cbuf */
1309 void *private; /* payload buffer for specified compression algorithm */
1310 void *private2; /* extra payload buffer */
1313 /* compress context for write IO path */
1314 struct compress_io_ctx {
1315 u32 magic; /* magic number to indicate page is compressed */
1316 struct inode *inode; /* inode the context belong to */
1317 struct page **rpages; /* pages store raw data in cluster */
1318 unsigned int nr_rpages; /* total page number in rpages */
1319 atomic_t pending_pages; /* in-flight compressed page count */
1322 /* decompress io context for read IO path */
1323 struct decompress_io_ctx {
1324 u32 magic; /* magic number to indicate page is compressed */
1325 struct inode *inode; /* inode the context belong to */
1326 pgoff_t cluster_idx; /* cluster index number */
1327 unsigned int cluster_size; /* page count in cluster */
1328 unsigned int log_cluster_size; /* log of cluster size */
1329 struct page **rpages; /* pages store raw data in cluster */
1330 unsigned int nr_rpages; /* total page number in rpages */
1331 struct page **cpages; /* pages store compressed data in cluster */
1332 unsigned int nr_cpages; /* total page number in cpages */
1333 struct page **tpages; /* temp pages to pad holes in cluster */
1334 void *rbuf; /* virtual mapped address on rpages */
1335 struct compress_data *cbuf; /* virtual mapped address on cpages */
1336 size_t rlen; /* valid data length in rbuf */
1337 size_t clen; /* valid data length in cbuf */
1338 atomic_t pending_pages; /* in-flight compressed page count */
1339 atomic_t verity_pages; /* in-flight page count for verity */
1340 bool failed; /* indicate IO error during decompression */
1341 void *private; /* payload buffer for specified decompression algorithm */
1342 void *private2; /* extra payload buffer */
1345 #define NULL_CLUSTER ((unsigned int)(~0))
1346 #define MIN_COMPRESS_LOG_SIZE 2
1347 #define MAX_COMPRESS_LOG_SIZE 8
1348 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1350 struct f2fs_sb_info {
1351 struct super_block *sb; /* pointer to VFS super block */
1352 struct proc_dir_entry *s_proc; /* proc entry */
1353 struct f2fs_super_block *raw_super; /* raw super block pointer */
1354 struct rw_semaphore sb_lock; /* lock for raw super block */
1355 int valid_super_block; /* valid super block no */
1356 unsigned long s_flag; /* flags for sbi */
1357 struct mutex writepages; /* mutex for writepages() */
1359 #ifdef CONFIG_BLK_DEV_ZONED
1360 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1361 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1364 /* for node-related operations */
1365 struct f2fs_nm_info *nm_info; /* node manager */
1366 struct inode *node_inode; /* cache node blocks */
1368 /* for segment-related operations */
1369 struct f2fs_sm_info *sm_info; /* segment manager */
1371 /* for bio operations */
1372 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1373 /* keep migration IO order for LFS mode */
1374 struct rw_semaphore io_order_lock;
1375 mempool_t *write_io_dummy; /* Dummy pages */
1377 /* for checkpoint */
1378 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1379 int cur_cp_pack; /* remain current cp pack */
1380 spinlock_t cp_lock; /* for flag in ckpt */
1381 struct inode *meta_inode; /* cache meta blocks */
1382 struct mutex cp_mutex; /* checkpoint procedure lock */
1383 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1384 struct rw_semaphore node_write; /* locking node writes */
1385 struct rw_semaphore node_change; /* locking node change */
1386 wait_queue_head_t cp_wait;
1387 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1388 long interval_time[MAX_TIME]; /* to store thresholds */
1390 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1392 spinlock_t fsync_node_lock; /* for node entry lock */
1393 struct list_head fsync_node_list; /* node list head */
1394 unsigned int fsync_seg_id; /* sequence id */
1395 unsigned int fsync_node_num; /* number of node entries */
1397 /* for orphan inode, use 0'th array */
1398 unsigned int max_orphans; /* max orphan inodes */
1400 /* for inode management */
1401 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1402 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1403 struct mutex flush_lock; /* for flush exclusion */
1405 /* for extent tree cache */
1406 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1407 struct mutex extent_tree_lock; /* locking extent radix tree */
1408 struct list_head extent_list; /* lru list for shrinker */
1409 spinlock_t extent_lock; /* locking extent lru list */
1410 atomic_t total_ext_tree; /* extent tree count */
1411 struct list_head zombie_list; /* extent zombie tree list */
1412 atomic_t total_zombie_tree; /* extent zombie tree count */
1413 atomic_t total_ext_node; /* extent info count */
1415 /* basic filesystem units */
1416 unsigned int log_sectors_per_block; /* log2 sectors per block */
1417 unsigned int log_blocksize; /* log2 block size */
1418 unsigned int blocksize; /* block size */
1419 unsigned int root_ino_num; /* root inode number*/
1420 unsigned int node_ino_num; /* node inode number*/
1421 unsigned int meta_ino_num; /* meta inode number*/
1422 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1423 unsigned int blocks_per_seg; /* blocks per segment */
1424 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1425 unsigned int segs_per_sec; /* segments per section */
1426 unsigned int secs_per_zone; /* sections per zone */
1427 unsigned int total_sections; /* total section count */
1428 unsigned int total_node_count; /* total node block count */
1429 unsigned int total_valid_node_count; /* valid node block count */
1430 loff_t max_file_blocks; /* max block index of file */
1431 int dir_level; /* directory level */
1432 int readdir_ra; /* readahead inode in readdir */
1434 block_t user_block_count; /* # of user blocks */
1435 block_t total_valid_block_count; /* # of valid blocks */
1436 block_t discard_blks; /* discard command candidats */
1437 block_t last_valid_block_count; /* for recovery */
1438 block_t reserved_blocks; /* configurable reserved blocks */
1439 block_t current_reserved_blocks; /* current reserved blocks */
1441 /* Additional tracking for no checkpoint mode */
1442 block_t unusable_block_count; /* # of blocks saved by last cp */
1444 unsigned int nquota_files; /* # of quota sysfile */
1445 struct rw_semaphore quota_sem; /* blocking cp for flags */
1447 /* # of pages, see count_type */
1448 atomic_t nr_pages[NR_COUNT_TYPE];
1449 /* # of allocated blocks */
1450 struct percpu_counter alloc_valid_block_count;
1452 /* writeback control */
1453 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1455 /* valid inode count */
1456 struct percpu_counter total_valid_inode_count;
1458 struct f2fs_mount_info mount_opt; /* mount options */
1460 /* for cleaning operations */
1461 struct rw_semaphore gc_lock; /*
1462 * semaphore for GC, avoid
1463 * race between GC and GC or CP
1465 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1466 struct atgc_management am; /* atgc management */
1467 unsigned int cur_victim_sec; /* current victim section num */
1468 unsigned int gc_mode; /* current GC state */
1469 unsigned int next_victim_seg[2]; /* next segment in victim section */
1471 /* for skip statistic */
1472 unsigned int atomic_files; /* # of opened atomic file */
1473 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1474 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1476 /* threshold for gc trials on pinned files */
1477 u64 gc_pin_file_threshold;
1478 struct rw_semaphore pin_sem;
1480 /* maximum # of trials to find a victim segment for SSR and GC */
1481 unsigned int max_victim_search;
1482 /* migration granularity of garbage collection, unit: segment */
1483 unsigned int migration_granularity;
1486 * for stat information.
1487 * one is for the LFS mode, and the other is for the SSR mode.
1489 #ifdef CONFIG_F2FS_STAT_FS
1490 struct f2fs_stat_info *stat_info; /* FS status information */
1491 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1492 unsigned int segment_count[2]; /* # of allocated segments */
1493 unsigned int block_count[2]; /* # of allocated blocks */
1494 atomic_t inplace_count; /* # of inplace update */
1495 atomic64_t total_hit_ext; /* # of lookup extent cache */
1496 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1497 atomic64_t read_hit_largest; /* # of hit largest extent node */
1498 atomic64_t read_hit_cached; /* # of hit cached extent node */
1499 atomic_t inline_xattr; /* # of inline_xattr inodes */
1500 atomic_t inline_inode; /* # of inline_data inodes */
1501 atomic_t inline_dir; /* # of inline_dentry inodes */
1502 atomic_t compr_inode; /* # of compressed inodes */
1503 atomic64_t compr_blocks; /* # of compressed blocks */
1504 atomic_t vw_cnt; /* # of volatile writes */
1505 atomic_t max_aw_cnt; /* max # of atomic writes */
1506 atomic_t max_vw_cnt; /* max # of volatile writes */
1507 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1508 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1509 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1511 spinlock_t stat_lock; /* lock for stat operations */
1513 /* For app/fs IO statistics */
1514 spinlock_t iostat_lock;
1515 unsigned long long rw_iostat[NR_IO_TYPE];
1516 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1518 unsigned long iostat_next_period;
1519 unsigned int iostat_period_ms;
1521 /* to attach REQ_META|REQ_FUA flags */
1522 unsigned int data_io_flag;
1523 unsigned int node_io_flag;
1525 /* For sysfs suppport */
1526 struct kobject s_kobj;
1527 struct completion s_kobj_unregister;
1529 /* For shrinker support */
1530 struct list_head s_list;
1531 int s_ndevs; /* number of devices */
1532 struct f2fs_dev_info *devs; /* for device list */
1533 unsigned int dirty_device; /* for checkpoint data flush */
1534 spinlock_t dev_lock; /* protect dirty_device */
1535 struct mutex umount_mutex;
1536 unsigned int shrinker_run_no;
1538 /* For write statistics */
1539 u64 sectors_written_start;
1542 /* Reference to checksum algorithm driver via cryptoapi */
1543 struct crypto_shash *s_chksum_driver;
1545 /* Precomputed FS UUID checksum for seeding other checksums */
1546 __u32 s_chksum_seed;
1548 struct workqueue_struct *post_read_wq; /* post read workqueue */
1550 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1551 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1553 #ifdef CONFIG_F2FS_FS_COMPRESSION
1554 struct kmem_cache *page_array_slab; /* page array entry */
1555 unsigned int page_array_slab_size; /* default page array slab size */
1559 struct f2fs_private_dio {
1560 struct inode *inode;
1562 bio_end_io_t *orig_end_io;
1566 #ifdef CONFIG_F2FS_FAULT_INJECTION
1567 #define f2fs_show_injection_info(sbi, type) \
1568 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1569 KERN_INFO, sbi->sb->s_id, \
1570 f2fs_fault_name[type], \
1571 __func__, __builtin_return_address(0))
1572 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1574 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1576 if (!ffi->inject_rate)
1579 if (!IS_FAULT_SET(ffi, type))
1582 atomic_inc(&ffi->inject_ops);
1583 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1584 atomic_set(&ffi->inject_ops, 0);
1590 #define f2fs_show_injection_info(sbi, type) do { } while (0)
1591 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1598 * Test if the mounted volume is a multi-device volume.
1599 * - For a single regular disk volume, sbi->s_ndevs is 0.
1600 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1601 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1603 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1605 return sbi->s_ndevs > 1;
1608 /* For write statistics. Suppose sector size is 512 bytes,
1609 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1611 #define BD_PART_WRITTEN(s) \
1612 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1613 (s)->sectors_written_start) >> 1)
1615 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1617 unsigned long now = jiffies;
1619 sbi->last_time[type] = now;
1621 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1622 if (type == REQ_TIME) {
1623 sbi->last_time[DISCARD_TIME] = now;
1624 sbi->last_time[GC_TIME] = now;
1628 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1630 unsigned long interval = sbi->interval_time[type] * HZ;
1632 return time_after(jiffies, sbi->last_time[type] + interval);
1635 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1638 unsigned long interval = sbi->interval_time[type] * HZ;
1639 unsigned int wait_ms = 0;
1642 delta = (sbi->last_time[type] + interval) - jiffies;
1644 wait_ms = jiffies_to_msecs(delta);
1652 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1653 const void *address, unsigned int length)
1656 struct shash_desc shash;
1661 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1663 desc.shash.tfm = sbi->s_chksum_driver;
1664 *(u32 *)desc.ctx = crc;
1666 err = crypto_shash_update(&desc.shash, address, length);
1669 return *(u32 *)desc.ctx;
1672 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1673 unsigned int length)
1675 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1678 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1679 void *buf, size_t buf_size)
1681 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1684 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1685 const void *address, unsigned int length)
1687 return __f2fs_crc32(sbi, crc, address, length);
1690 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1692 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1695 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1697 return sb->s_fs_info;
1700 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1702 return F2FS_SB(inode->i_sb);
1705 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1707 return F2FS_I_SB(mapping->host);
1710 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1712 return F2FS_M_SB(page_file_mapping(page));
1715 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1717 return (struct f2fs_super_block *)(sbi->raw_super);
1720 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1722 return (struct f2fs_checkpoint *)(sbi->ckpt);
1725 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1727 return (struct f2fs_node *)page_address(page);
1730 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1732 return &((struct f2fs_node *)page_address(page))->i;
1735 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1737 return (struct f2fs_nm_info *)(sbi->nm_info);
1740 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1742 return (struct f2fs_sm_info *)(sbi->sm_info);
1745 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1747 return (struct sit_info *)(SM_I(sbi)->sit_info);
1750 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1752 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1755 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1757 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1760 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1762 return sbi->meta_inode->i_mapping;
1765 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1767 return sbi->node_inode->i_mapping;
1770 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1772 return test_bit(type, &sbi->s_flag);
1775 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1777 set_bit(type, &sbi->s_flag);
1780 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1782 clear_bit(type, &sbi->s_flag);
1785 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1787 return le64_to_cpu(cp->checkpoint_ver);
1790 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1792 if (type < F2FS_MAX_QUOTAS)
1793 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1797 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1799 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1800 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1803 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1805 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1807 return ckpt_flags & f;
1810 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1812 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1815 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1817 unsigned int ckpt_flags;
1819 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1821 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1824 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1826 unsigned long flags;
1828 spin_lock_irqsave(&sbi->cp_lock, flags);
1829 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1830 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1833 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1835 unsigned int ckpt_flags;
1837 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1839 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1842 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1844 unsigned long flags;
1846 spin_lock_irqsave(&sbi->cp_lock, flags);
1847 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1848 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1851 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1853 unsigned long flags;
1854 unsigned char *nat_bits;
1857 * In order to re-enable nat_bits we need to call fsck.f2fs by
1858 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1859 * so let's rely on regular fsck or unclean shutdown.
1863 spin_lock_irqsave(&sbi->cp_lock, flags);
1864 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1865 nat_bits = NM_I(sbi)->nat_bits;
1866 NM_I(sbi)->nat_bits = NULL;
1868 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1873 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1874 struct cp_control *cpc)
1876 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1878 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1881 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1883 down_read(&sbi->cp_rwsem);
1886 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1888 return down_read_trylock(&sbi->cp_rwsem);
1891 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1893 up_read(&sbi->cp_rwsem);
1896 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1898 down_write(&sbi->cp_rwsem);
1901 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1903 up_write(&sbi->cp_rwsem);
1906 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1908 int reason = CP_SYNC;
1910 if (test_opt(sbi, FASTBOOT))
1911 reason = CP_FASTBOOT;
1912 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1917 static inline bool __remain_node_summaries(int reason)
1919 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1922 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1924 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1925 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1929 * Check whether the inode has blocks or not
1931 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1933 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1935 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1938 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1940 return ofs == XATTR_NODE_OFFSET;
1943 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1944 struct inode *inode, bool cap)
1948 if (!test_opt(sbi, RESERVE_ROOT))
1950 if (IS_NOQUOTA(inode))
1952 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1954 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1955 in_group_p(F2FS_OPTION(sbi).s_resgid))
1957 if (cap && capable(CAP_SYS_RESOURCE))
1962 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1963 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1964 struct inode *inode, blkcnt_t *count)
1966 blkcnt_t diff = 0, release = 0;
1967 block_t avail_user_block_count;
1970 ret = dquot_reserve_block(inode, *count);
1974 if (time_to_inject(sbi, FAULT_BLOCK)) {
1975 f2fs_show_injection_info(sbi, FAULT_BLOCK);
1981 * let's increase this in prior to actual block count change in order
1982 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1984 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1986 spin_lock(&sbi->stat_lock);
1987 sbi->total_valid_block_count += (block_t)(*count);
1988 avail_user_block_count = sbi->user_block_count -
1989 sbi->current_reserved_blocks;
1991 if (!__allow_reserved_blocks(sbi, inode, true))
1992 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1994 if (F2FS_IO_ALIGNED(sbi))
1995 avail_user_block_count -= sbi->blocks_per_seg *
1996 SM_I(sbi)->additional_reserved_segments;
1998 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1999 if (avail_user_block_count > sbi->unusable_block_count)
2000 avail_user_block_count -= sbi->unusable_block_count;
2002 avail_user_block_count = 0;
2004 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2005 diff = sbi->total_valid_block_count - avail_user_block_count;
2010 sbi->total_valid_block_count -= diff;
2012 spin_unlock(&sbi->stat_lock);
2016 spin_unlock(&sbi->stat_lock);
2018 if (unlikely(release)) {
2019 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2020 dquot_release_reservation_block(inode, release);
2022 f2fs_i_blocks_write(inode, *count, true, true);
2026 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2028 dquot_release_reservation_block(inode, release);
2033 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2035 #define f2fs_err(sbi, fmt, ...) \
2036 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2037 #define f2fs_warn(sbi, fmt, ...) \
2038 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2039 #define f2fs_notice(sbi, fmt, ...) \
2040 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2041 #define f2fs_info(sbi, fmt, ...) \
2042 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2043 #define f2fs_debug(sbi, fmt, ...) \
2044 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2046 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2047 struct inode *inode,
2050 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2052 spin_lock(&sbi->stat_lock);
2053 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2054 sbi->total_valid_block_count -= (block_t)count;
2055 if (sbi->reserved_blocks &&
2056 sbi->current_reserved_blocks < sbi->reserved_blocks)
2057 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2058 sbi->current_reserved_blocks + count);
2059 spin_unlock(&sbi->stat_lock);
2060 if (unlikely(inode->i_blocks < sectors)) {
2061 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2063 (unsigned long long)inode->i_blocks,
2064 (unsigned long long)sectors);
2065 set_sbi_flag(sbi, SBI_NEED_FSCK);
2068 f2fs_i_blocks_write(inode, count, false, true);
2071 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2073 atomic_inc(&sbi->nr_pages[count_type]);
2075 if (count_type == F2FS_DIRTY_DENTS ||
2076 count_type == F2FS_DIRTY_NODES ||
2077 count_type == F2FS_DIRTY_META ||
2078 count_type == F2FS_DIRTY_QDATA ||
2079 count_type == F2FS_DIRTY_IMETA)
2080 set_sbi_flag(sbi, SBI_IS_DIRTY);
2083 static inline void inode_inc_dirty_pages(struct inode *inode)
2085 atomic_inc(&F2FS_I(inode)->dirty_pages);
2086 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2087 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2088 if (IS_NOQUOTA(inode))
2089 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2092 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2094 atomic_dec(&sbi->nr_pages[count_type]);
2097 static inline void inode_dec_dirty_pages(struct inode *inode)
2099 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2100 !S_ISLNK(inode->i_mode))
2103 atomic_dec(&F2FS_I(inode)->dirty_pages);
2104 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2105 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2106 if (IS_NOQUOTA(inode))
2107 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2110 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2112 return atomic_read(&sbi->nr_pages[count_type]);
2115 static inline int get_dirty_pages(struct inode *inode)
2117 return atomic_read(&F2FS_I(inode)->dirty_pages);
2120 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2122 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2123 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2124 sbi->log_blocks_per_seg;
2126 return segs / sbi->segs_per_sec;
2129 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2131 return sbi->total_valid_block_count;
2134 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2136 return sbi->discard_blks;
2139 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2141 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2143 /* return NAT or SIT bitmap */
2144 if (flag == NAT_BITMAP)
2145 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2146 else if (flag == SIT_BITMAP)
2147 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2152 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2154 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2157 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2159 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2162 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2163 offset = (flag == SIT_BITMAP) ?
2164 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2166 * if large_nat_bitmap feature is enabled, leave checksum
2167 * protection for all nat/sit bitmaps.
2169 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2172 if (__cp_payload(sbi) > 0) {
2173 if (flag == NAT_BITMAP)
2174 return &ckpt->sit_nat_version_bitmap;
2176 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2178 offset = (flag == NAT_BITMAP) ?
2179 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2180 return &ckpt->sit_nat_version_bitmap + offset;
2184 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2186 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2188 if (sbi->cur_cp_pack == 2)
2189 start_addr += sbi->blocks_per_seg;
2193 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2195 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2197 if (sbi->cur_cp_pack == 1)
2198 start_addr += sbi->blocks_per_seg;
2202 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2204 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2207 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2209 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2212 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2213 struct inode *inode, bool is_inode)
2215 block_t valid_block_count;
2216 unsigned int valid_node_count, user_block_count;
2221 err = dquot_alloc_inode(inode);
2226 err = dquot_reserve_block(inode, 1);
2231 if (time_to_inject(sbi, FAULT_BLOCK)) {
2232 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2236 spin_lock(&sbi->stat_lock);
2238 valid_block_count = sbi->total_valid_block_count +
2239 sbi->current_reserved_blocks + 1;
2241 if (!__allow_reserved_blocks(sbi, inode, false))
2242 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2244 if (F2FS_IO_ALIGNED(sbi))
2245 valid_block_count += sbi->blocks_per_seg *
2246 SM_I(sbi)->additional_reserved_segments;
2248 user_block_count = sbi->user_block_count;
2249 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2250 user_block_count -= sbi->unusable_block_count;
2252 if (unlikely(valid_block_count > user_block_count)) {
2253 spin_unlock(&sbi->stat_lock);
2257 valid_node_count = sbi->total_valid_node_count + 1;
2258 if (unlikely(valid_node_count > sbi->total_node_count)) {
2259 spin_unlock(&sbi->stat_lock);
2263 sbi->total_valid_node_count++;
2264 sbi->total_valid_block_count++;
2265 spin_unlock(&sbi->stat_lock);
2269 f2fs_mark_inode_dirty_sync(inode, true);
2271 f2fs_i_blocks_write(inode, 1, true, true);
2274 percpu_counter_inc(&sbi->alloc_valid_block_count);
2280 dquot_free_inode(inode);
2282 dquot_release_reservation_block(inode, 1);
2287 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2288 struct inode *inode, bool is_inode)
2290 spin_lock(&sbi->stat_lock);
2292 if (unlikely(!sbi->total_valid_block_count ||
2293 !sbi->total_valid_node_count)) {
2294 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2295 sbi->total_valid_block_count,
2296 sbi->total_valid_node_count);
2297 set_sbi_flag(sbi, SBI_NEED_FSCK);
2299 sbi->total_valid_block_count--;
2300 sbi->total_valid_node_count--;
2303 if (sbi->reserved_blocks &&
2304 sbi->current_reserved_blocks < sbi->reserved_blocks)
2305 sbi->current_reserved_blocks++;
2307 spin_unlock(&sbi->stat_lock);
2310 dquot_free_inode(inode);
2312 if (unlikely(inode->i_blocks == 0)) {
2313 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2315 (unsigned long long)inode->i_blocks);
2316 set_sbi_flag(sbi, SBI_NEED_FSCK);
2319 f2fs_i_blocks_write(inode, 1, false, true);
2323 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2325 return sbi->total_valid_node_count;
2328 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2330 percpu_counter_inc(&sbi->total_valid_inode_count);
2333 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2335 percpu_counter_dec(&sbi->total_valid_inode_count);
2338 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2340 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2343 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2344 pgoff_t index, bool for_write)
2348 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2350 page = find_get_page_flags(mapping, index,
2351 FGP_LOCK | FGP_ACCESSED);
2353 page = find_lock_page(mapping, index);
2357 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2358 f2fs_show_injection_info(F2FS_M_SB(mapping),
2365 return grab_cache_page(mapping, index);
2366 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2369 static inline struct page *f2fs_pagecache_get_page(
2370 struct address_space *mapping, pgoff_t index,
2371 int fgp_flags, gfp_t gfp_mask)
2373 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2374 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2378 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2381 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2383 char *src_kaddr = kmap(src);
2384 char *dst_kaddr = kmap(dst);
2386 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2391 static inline void f2fs_put_page(struct page *page, int unlock)
2397 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2403 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2406 f2fs_put_page(dn->node_page, 1);
2407 if (dn->inode_page && dn->node_page != dn->inode_page)
2408 f2fs_put_page(dn->inode_page, 0);
2409 dn->node_page = NULL;
2410 dn->inode_page = NULL;
2413 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2416 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2419 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2424 entry = kmem_cache_alloc(cachep, flags);
2426 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2430 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2432 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2433 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2434 get_pages(sbi, F2FS_WB_CP_DATA) ||
2435 get_pages(sbi, F2FS_DIO_READ) ||
2436 get_pages(sbi, F2FS_DIO_WRITE))
2439 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2440 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2443 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2444 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2449 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2451 if (sbi->gc_mode == GC_URGENT_HIGH)
2454 if (is_inflight_io(sbi, type))
2457 if (sbi->gc_mode == GC_URGENT_LOW &&
2458 (type == DISCARD_TIME || type == GC_TIME))
2461 return f2fs_time_over(sbi, type);
2464 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2465 unsigned long index, void *item)
2467 while (radix_tree_insert(root, index, item))
2471 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2473 static inline bool IS_INODE(struct page *page)
2475 struct f2fs_node *p = F2FS_NODE(page);
2477 return RAW_IS_INODE(p);
2480 static inline int offset_in_addr(struct f2fs_inode *i)
2482 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2483 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2486 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2488 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2491 static inline int f2fs_has_extra_attr(struct inode *inode);
2492 static inline block_t data_blkaddr(struct inode *inode,
2493 struct page *node_page, unsigned int offset)
2495 struct f2fs_node *raw_node;
2498 bool is_inode = IS_INODE(node_page);
2500 raw_node = F2FS_NODE(node_page);
2504 /* from GC path only */
2505 base = offset_in_addr(&raw_node->i);
2506 else if (f2fs_has_extra_attr(inode))
2507 base = get_extra_isize(inode);
2510 addr_array = blkaddr_in_node(raw_node);
2511 return le32_to_cpu(addr_array[base + offset]);
2514 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2516 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2519 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2524 mask = 1 << (7 - (nr & 0x07));
2525 return mask & *addr;
2528 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2533 mask = 1 << (7 - (nr & 0x07));
2537 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2542 mask = 1 << (7 - (nr & 0x07));
2546 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2552 mask = 1 << (7 - (nr & 0x07));
2558 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2564 mask = 1 << (7 - (nr & 0x07));
2570 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2575 mask = 1 << (7 - (nr & 0x07));
2580 * On-disk inode flags (f2fs_inode::i_flags)
2582 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2583 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2584 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2585 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2586 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2587 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2588 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2589 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2590 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2591 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2592 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2594 /* Flags that should be inherited by new inodes from their parent. */
2595 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2596 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2597 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2599 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2600 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2603 /* Flags that are appropriate for non-directories/regular files. */
2604 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2606 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2610 else if (S_ISREG(mode))
2611 return flags & F2FS_REG_FLMASK;
2613 return flags & F2FS_OTHER_FLMASK;
2616 static inline void __mark_inode_dirty_flag(struct inode *inode,
2620 case FI_INLINE_XATTR:
2621 case FI_INLINE_DATA:
2622 case FI_INLINE_DENTRY:
2628 case FI_INLINE_DOTS:
2630 f2fs_mark_inode_dirty_sync(inode, true);
2634 static inline void set_inode_flag(struct inode *inode, int flag)
2636 set_bit(flag, F2FS_I(inode)->flags);
2637 __mark_inode_dirty_flag(inode, flag, true);
2640 static inline int is_inode_flag_set(struct inode *inode, int flag)
2642 return test_bit(flag, F2FS_I(inode)->flags);
2645 static inline void clear_inode_flag(struct inode *inode, int flag)
2647 clear_bit(flag, F2FS_I(inode)->flags);
2648 __mark_inode_dirty_flag(inode, flag, false);
2651 static inline bool f2fs_verity_in_progress(struct inode *inode)
2653 return IS_ENABLED(CONFIG_FS_VERITY) &&
2654 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2657 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2659 F2FS_I(inode)->i_acl_mode = mode;
2660 set_inode_flag(inode, FI_ACL_MODE);
2661 f2fs_mark_inode_dirty_sync(inode, false);
2664 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2670 f2fs_mark_inode_dirty_sync(inode, true);
2673 static inline void f2fs_i_blocks_write(struct inode *inode,
2674 block_t diff, bool add, bool claim)
2676 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2677 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2679 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2682 dquot_claim_block(inode, diff);
2684 dquot_alloc_block_nofail(inode, diff);
2686 dquot_free_block(inode, diff);
2689 f2fs_mark_inode_dirty_sync(inode, true);
2690 if (clean || recover)
2691 set_inode_flag(inode, FI_AUTO_RECOVER);
2694 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2696 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2697 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2699 if (i_size_read(inode) == i_size)
2702 i_size_write(inode, i_size);
2703 f2fs_mark_inode_dirty_sync(inode, true);
2704 if (clean || recover)
2705 set_inode_flag(inode, FI_AUTO_RECOVER);
2708 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2710 F2FS_I(inode)->i_current_depth = depth;
2711 f2fs_mark_inode_dirty_sync(inode, true);
2714 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2717 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2718 f2fs_mark_inode_dirty_sync(inode, true);
2721 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2723 F2FS_I(inode)->i_xattr_nid = xnid;
2724 f2fs_mark_inode_dirty_sync(inode, true);
2727 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2729 F2FS_I(inode)->i_pino = pino;
2730 f2fs_mark_inode_dirty_sync(inode, true);
2733 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2735 struct f2fs_inode_info *fi = F2FS_I(inode);
2737 if (ri->i_inline & F2FS_INLINE_XATTR)
2738 set_bit(FI_INLINE_XATTR, fi->flags);
2739 if (ri->i_inline & F2FS_INLINE_DATA)
2740 set_bit(FI_INLINE_DATA, fi->flags);
2741 if (ri->i_inline & F2FS_INLINE_DENTRY)
2742 set_bit(FI_INLINE_DENTRY, fi->flags);
2743 if (ri->i_inline & F2FS_DATA_EXIST)
2744 set_bit(FI_DATA_EXIST, fi->flags);
2745 if (ri->i_inline & F2FS_INLINE_DOTS)
2746 set_bit(FI_INLINE_DOTS, fi->flags);
2747 if (ri->i_inline & F2FS_EXTRA_ATTR)
2748 set_bit(FI_EXTRA_ATTR, fi->flags);
2749 if (ri->i_inline & F2FS_PIN_FILE)
2750 set_bit(FI_PIN_FILE, fi->flags);
2753 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2757 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2758 ri->i_inline |= F2FS_INLINE_XATTR;
2759 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2760 ri->i_inline |= F2FS_INLINE_DATA;
2761 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2762 ri->i_inline |= F2FS_INLINE_DENTRY;
2763 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2764 ri->i_inline |= F2FS_DATA_EXIST;
2765 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2766 ri->i_inline |= F2FS_INLINE_DOTS;
2767 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2768 ri->i_inline |= F2FS_EXTRA_ATTR;
2769 if (is_inode_flag_set(inode, FI_PIN_FILE))
2770 ri->i_inline |= F2FS_PIN_FILE;
2773 static inline int f2fs_has_extra_attr(struct inode *inode)
2775 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2778 static inline int f2fs_has_inline_xattr(struct inode *inode)
2780 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2783 static inline int f2fs_compressed_file(struct inode *inode)
2785 return S_ISREG(inode->i_mode) &&
2786 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2789 static inline unsigned int addrs_per_inode(struct inode *inode)
2791 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2792 get_inline_xattr_addrs(inode);
2794 if (!f2fs_compressed_file(inode))
2796 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2799 static inline unsigned int addrs_per_block(struct inode *inode)
2801 if (!f2fs_compressed_file(inode))
2802 return DEF_ADDRS_PER_BLOCK;
2803 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2806 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2808 struct f2fs_inode *ri = F2FS_INODE(page);
2810 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2811 get_inline_xattr_addrs(inode)]);
2814 static inline int inline_xattr_size(struct inode *inode)
2816 if (f2fs_has_inline_xattr(inode))
2817 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2821 static inline int f2fs_has_inline_data(struct inode *inode)
2823 return is_inode_flag_set(inode, FI_INLINE_DATA);
2826 static inline int f2fs_exist_data(struct inode *inode)
2828 return is_inode_flag_set(inode, FI_DATA_EXIST);
2831 static inline int f2fs_has_inline_dots(struct inode *inode)
2833 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2836 static inline int f2fs_is_mmap_file(struct inode *inode)
2838 return is_inode_flag_set(inode, FI_MMAP_FILE);
2841 static inline bool f2fs_is_pinned_file(struct inode *inode)
2843 return is_inode_flag_set(inode, FI_PIN_FILE);
2846 static inline bool f2fs_is_atomic_file(struct inode *inode)
2848 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2851 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2853 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2856 static inline bool f2fs_is_volatile_file(struct inode *inode)
2858 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2861 static inline bool f2fs_is_first_block_written(struct inode *inode)
2863 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2866 static inline bool f2fs_is_drop_cache(struct inode *inode)
2868 return is_inode_flag_set(inode, FI_DROP_CACHE);
2871 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2873 struct f2fs_inode *ri = F2FS_INODE(page);
2874 int extra_size = get_extra_isize(inode);
2876 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2879 static inline int f2fs_has_inline_dentry(struct inode *inode)
2881 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2884 static inline int is_file(struct inode *inode, int type)
2886 return F2FS_I(inode)->i_advise & type;
2889 static inline void set_file(struct inode *inode, int type)
2891 F2FS_I(inode)->i_advise |= type;
2892 f2fs_mark_inode_dirty_sync(inode, true);
2895 static inline void clear_file(struct inode *inode, int type)
2897 F2FS_I(inode)->i_advise &= ~type;
2898 f2fs_mark_inode_dirty_sync(inode, true);
2901 static inline bool f2fs_is_time_consistent(struct inode *inode)
2903 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2905 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2907 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2909 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2910 &F2FS_I(inode)->i_crtime))
2915 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2920 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2922 spin_lock(&sbi->inode_lock[DIRTY_META]);
2923 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2924 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2927 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2928 file_keep_isize(inode) ||
2929 i_size_read(inode) & ~PAGE_MASK)
2932 if (!f2fs_is_time_consistent(inode))
2935 spin_lock(&F2FS_I(inode)->i_size_lock);
2936 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2937 spin_unlock(&F2FS_I(inode)->i_size_lock);
2942 static inline bool f2fs_readonly(struct super_block *sb)
2944 return sb_rdonly(sb);
2947 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2949 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2952 static inline bool is_dot_dotdot(const u8 *name, size_t len)
2954 if (len == 1 && name[0] == '.')
2957 if (len == 2 && name[0] == '.' && name[1] == '.')
2963 static inline bool f2fs_may_extent_tree(struct inode *inode)
2965 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2967 if (!test_opt(sbi, EXTENT_CACHE) ||
2968 is_inode_flag_set(inode, FI_NO_EXTENT) ||
2969 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
2973 * for recovered files during mount do not create extents
2974 * if shrinker is not registered.
2976 if (list_empty(&sbi->s_list))
2979 return S_ISREG(inode->i_mode);
2982 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2983 size_t size, gfp_t flags)
2985 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2986 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
2990 return kmalloc(size, flags);
2993 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2994 size_t size, gfp_t flags)
2996 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2999 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3000 size_t size, gfp_t flags)
3002 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3003 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3007 return kvmalloc(size, flags);
3010 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3011 size_t size, gfp_t flags)
3013 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3016 static inline int get_extra_isize(struct inode *inode)
3018 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3021 static inline int get_inline_xattr_addrs(struct inode *inode)
3023 return F2FS_I(inode)->i_inline_xattr_size;
3026 #define f2fs_get_inode_mode(i) \
3027 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3028 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3030 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3031 (offsetof(struct f2fs_inode, i_extra_end) - \
3032 offsetof(struct f2fs_inode, i_extra_isize)) \
3034 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3035 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3036 ((offsetof(typeof(*(f2fs_inode)), field) + \
3037 sizeof((f2fs_inode)->field)) \
3038 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3040 #define DEFAULT_IOSTAT_PERIOD_MS 3000
3041 #define MIN_IOSTAT_PERIOD_MS 100
3042 /* maximum period of iostat tracing is 1 day */
3043 #define MAX_IOSTAT_PERIOD_MS 8640000
3045 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3049 spin_lock(&sbi->iostat_lock);
3050 for (i = 0; i < NR_IO_TYPE; i++) {
3051 sbi->rw_iostat[i] = 0;
3052 sbi->prev_rw_iostat[i] = 0;
3054 spin_unlock(&sbi->iostat_lock);
3057 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3059 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3060 enum iostat_type type, unsigned long long io_bytes)
3062 if (!sbi->iostat_enable)
3064 spin_lock(&sbi->iostat_lock);
3065 sbi->rw_iostat[type] += io_bytes;
3067 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3068 sbi->rw_iostat[APP_BUFFERED_IO] =
3069 sbi->rw_iostat[APP_WRITE_IO] -
3070 sbi->rw_iostat[APP_DIRECT_IO];
3072 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3073 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3074 sbi->rw_iostat[APP_READ_IO] -
3075 sbi->rw_iostat[APP_DIRECT_READ_IO];
3076 spin_unlock(&sbi->iostat_lock);
3078 f2fs_record_iostat(sbi);
3081 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3083 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3085 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3086 block_t blkaddr, int type);
3087 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3088 block_t blkaddr, int type)
3090 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3091 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3093 f2fs_bug_on(sbi, 1);
3097 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3099 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3100 blkaddr == COMPRESS_ADDR)
3105 static inline void f2fs_set_page_private(struct page *page,
3108 if (PagePrivate(page))
3111 attach_page_private(page, (void *)data);
3114 static inline void f2fs_clear_page_private(struct page *page)
3116 detach_page_private(page);
3122 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3123 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3124 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3125 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3126 int f2fs_truncate(struct inode *inode);
3127 int f2fs_getattr(const struct path *path, struct kstat *stat,
3128 u32 request_mask, unsigned int flags);
3129 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3130 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3131 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3132 int f2fs_precache_extents(struct inode *inode);
3133 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3134 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3135 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3136 int f2fs_pin_file_control(struct inode *inode, bool inc);
3141 void f2fs_set_inode_flags(struct inode *inode);
3142 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3143 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3144 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3145 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3146 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3147 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3148 void f2fs_update_inode_page(struct inode *inode);
3149 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3150 void f2fs_evict_inode(struct inode *inode);
3151 void f2fs_handle_failed_inode(struct inode *inode);
3156 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3157 bool hot, bool set);
3158 struct dentry *f2fs_get_parent(struct dentry *child);
3163 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3164 int f2fs_init_casefolded_name(const struct inode *dir,
3165 struct f2fs_filename *fname);
3166 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3167 int lookup, struct f2fs_filename *fname);
3168 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3169 struct f2fs_filename *fname);
3170 void f2fs_free_filename(struct f2fs_filename *fname);
3171 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3172 const struct f2fs_filename *fname, int *max_slots);
3173 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3174 unsigned int start_pos, struct fscrypt_str *fstr);
3175 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3176 struct f2fs_dentry_ptr *d);
3177 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3178 const struct f2fs_filename *fname, struct page *dpage);
3179 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3180 unsigned int current_depth);
3181 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3182 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3183 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3184 const struct f2fs_filename *fname,
3185 struct page **res_page);
3186 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3187 const struct qstr *child, struct page **res_page);
3188 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3189 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3190 struct page **page);
3191 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3192 struct page *page, struct inode *inode);
3193 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3194 const struct f2fs_filename *fname);
3195 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3196 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3197 unsigned int bit_pos);
3198 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3199 struct inode *inode, nid_t ino, umode_t mode);
3200 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3201 struct inode *inode, nid_t ino, umode_t mode);
3202 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3203 struct inode *inode, nid_t ino, umode_t mode);
3204 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3205 struct inode *dir, struct inode *inode);
3206 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3207 bool f2fs_empty_dir(struct inode *dir);
3209 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3211 if (fscrypt_is_nokey_name(dentry))
3213 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3214 inode, inode->i_ino, inode->i_mode);
3220 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3221 void f2fs_inode_synced(struct inode *inode);
3222 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3223 int f2fs_quota_sync(struct super_block *sb, int type);
3224 void f2fs_quota_off_umount(struct super_block *sb);
3225 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3226 int f2fs_sync_fs(struct super_block *sb, int sync);
3227 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3232 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3237 struct dnode_of_data;
3240 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3241 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3242 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3243 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3244 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3245 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3246 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3247 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3248 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3249 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3250 struct node_info *ni);
3251 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3252 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3253 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3254 int f2fs_truncate_xattr_node(struct inode *inode);
3255 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3256 unsigned int seq_id);
3257 int f2fs_remove_inode_page(struct inode *inode);
3258 struct page *f2fs_new_inode_page(struct inode *inode);
3259 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3260 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3261 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3262 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3263 int f2fs_move_node_page(struct page *node_page, int gc_type);
3264 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3265 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3266 struct writeback_control *wbc, bool atomic,
3267 unsigned int *seq_id);
3268 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3269 struct writeback_control *wbc,
3270 bool do_balance, enum iostat_type io_type);
3271 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3272 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3273 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3274 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3275 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3276 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3277 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3278 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3279 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3280 unsigned int segno, struct f2fs_summary_block *sum);
3281 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3282 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3283 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3284 int __init f2fs_create_node_manager_caches(void);
3285 void f2fs_destroy_node_manager_caches(void);
3290 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3291 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3292 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3293 void f2fs_drop_inmem_pages(struct inode *inode);
3294 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3295 int f2fs_commit_inmem_pages(struct inode *inode);
3296 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3297 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3298 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3299 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3300 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3301 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3302 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3303 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3304 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3305 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3306 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3307 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3308 struct cp_control *cpc);
3309 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3310 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3311 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3312 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3313 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3314 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3315 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3316 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3317 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3318 void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3319 unsigned int *newseg, bool new_sec, int dir);
3320 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3321 unsigned int start, unsigned int end);
3322 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type);
3323 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3324 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3325 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3326 struct cp_control *cpc);
3327 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3328 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3330 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3331 enum iostat_type io_type);
3332 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3333 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3334 struct f2fs_io_info *fio);
3335 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3336 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3337 block_t old_blkaddr, block_t new_blkaddr,
3338 bool recover_curseg, bool recover_newaddr,
3340 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3341 block_t old_addr, block_t new_addr,
3342 unsigned char version, bool recover_curseg,
3343 bool recover_newaddr);
3344 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3345 block_t old_blkaddr, block_t *new_blkaddr,
3346 struct f2fs_summary *sum, int type,
3347 struct f2fs_io_info *fio);
3348 void f2fs_wait_on_page_writeback(struct page *page,
3349 enum page_type type, bool ordered, bool locked);
3350 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3351 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3353 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3354 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3355 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3356 unsigned int val, int alloc);
3357 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3358 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3359 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3360 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3361 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3362 int __init f2fs_create_segment_manager_caches(void);
3363 void f2fs_destroy_segment_manager_caches(void);
3364 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3365 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3366 enum page_type type, enum temp_type temp);
3367 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3368 unsigned int segno);
3369 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3370 unsigned int segno);
3375 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3376 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3377 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3378 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3379 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3380 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3381 block_t blkaddr, int type);
3382 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3383 int type, bool sync);
3384 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3385 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3386 long nr_to_write, enum iostat_type io_type);
3387 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3388 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3389 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3390 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3391 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3392 unsigned int devidx, int type);
3393 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3394 unsigned int devidx, int type);
3395 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3396 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3397 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3398 void f2fs_add_orphan_inode(struct inode *inode);
3399 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3400 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3401 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3402 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3403 void f2fs_remove_dirty_inode(struct inode *inode);
3404 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3406 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3407 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3408 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3409 int __init f2fs_create_checkpoint_caches(void);
3410 void f2fs_destroy_checkpoint_caches(void);
3415 int __init f2fs_init_bioset(void);
3416 void f2fs_destroy_bioset(void);
3417 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio);
3418 int f2fs_init_bio_entry_cache(void);
3419 void f2fs_destroy_bio_entry_cache(void);
3420 void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3421 struct bio *bio, enum page_type type);
3422 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3423 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3424 struct inode *inode, struct page *page,
3425 nid_t ino, enum page_type type);
3426 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3427 struct bio **bio, struct page *page);
3428 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3429 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3430 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3431 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3432 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3433 block_t blk_addr, struct bio *bio);
3434 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3435 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3436 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3437 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3438 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3439 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3440 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3441 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3442 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3443 int op_flags, bool for_write);
3444 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3445 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3447 struct page *f2fs_get_new_data_page(struct inode *inode,
3448 struct page *ipage, pgoff_t index, bool new_i_size);
3449 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3450 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3451 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3452 int create, int flag);
3453 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3454 u64 start, u64 len);
3455 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3456 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3457 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3458 int f2fs_write_single_data_page(struct page *page, int *submitted,
3459 struct bio **bio, sector_t *last_block,
3460 struct writeback_control *wbc,
3461 enum iostat_type io_type,
3462 int compr_blocks, bool allow_balance);
3463 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3464 unsigned int length);
3465 int f2fs_release_page(struct page *page, gfp_t wait);
3466 #ifdef CONFIG_MIGRATION
3467 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3468 struct page *page, enum migrate_mode mode);
3470 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3471 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3472 int f2fs_init_post_read_processing(void);
3473 void f2fs_destroy_post_read_processing(void);
3474 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3475 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3480 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3481 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3482 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3483 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force,
3484 unsigned int segno);
3485 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3486 int f2fs_resize_fs(struct file *filp, __u64 block_count);
3487 int __init f2fs_create_garbage_collection_cache(void);
3488 void f2fs_destroy_garbage_collection_cache(void);
3493 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3494 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3495 int __init f2fs_create_recovery_cache(void);
3496 void f2fs_destroy_recovery_cache(void);
3501 #ifdef CONFIG_F2FS_STAT_FS
3502 struct f2fs_stat_info {
3503 struct list_head stat_list;
3504 struct f2fs_sb_info *sbi;
3505 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3506 int main_area_segs, main_area_sections, main_area_zones;
3507 unsigned long long hit_largest, hit_cached, hit_rbtree;
3508 unsigned long long hit_total, total_ext;
3509 int ext_tree, zombie_tree, ext_node;
3510 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3511 int ndirty_data, ndirty_qdata;
3513 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3514 int nats, dirty_nats, sits, dirty_sits;
3515 int free_nids, avail_nids, alloc_nids;
3516 int total_count, utilization;
3517 int bg_gc, nr_wb_cp_data, nr_wb_data;
3518 int nr_rd_data, nr_rd_node, nr_rd_meta;
3519 int nr_dio_read, nr_dio_write;
3520 unsigned int io_skip_bggc, other_skip_bggc;
3521 int nr_flushing, nr_flushed, flush_list_empty;
3522 int nr_discarding, nr_discarded;
3524 unsigned int undiscard_blks;
3525 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3527 unsigned long long compr_blocks;
3528 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3529 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3530 unsigned int bimodal, avg_vblocks;
3531 int util_free, util_valid, util_invalid;
3532 int rsvd_segs, overp_segs;
3533 int dirty_count, node_pages, meta_pages;
3534 int prefree_count, call_count, cp_count, bg_cp_count;
3535 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3536 int bg_node_segs, bg_data_segs;
3537 int tot_blks, data_blks, node_blks;
3538 int bg_data_blks, bg_node_blks;
3539 unsigned long long skipped_atomic_files[2];
3540 int curseg[NR_CURSEG_TYPE];
3541 int cursec[NR_CURSEG_TYPE];
3542 int curzone[NR_CURSEG_TYPE];
3543 unsigned int dirty_seg[NR_CURSEG_TYPE];
3544 unsigned int full_seg[NR_CURSEG_TYPE];
3545 unsigned int valid_blks[NR_CURSEG_TYPE];
3547 unsigned int meta_count[META_MAX];
3548 unsigned int segment_count[2];
3549 unsigned int block_count[2];
3550 unsigned int inplace_count;
3551 unsigned long long base_mem, cache_mem, page_mem;
3554 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3556 return (struct f2fs_stat_info *)sbi->stat_info;
3559 #define stat_inc_cp_count(si) ((si)->cp_count++)
3560 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3561 #define stat_inc_call_count(si) ((si)->call_count++)
3562 #define stat_inc_bggc_count(si) ((si)->bg_gc++)
3563 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3564 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3565 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3566 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3567 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3568 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3569 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3570 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3571 #define stat_inc_inline_xattr(inode) \
3573 if (f2fs_has_inline_xattr(inode)) \
3574 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3576 #define stat_dec_inline_xattr(inode) \
3578 if (f2fs_has_inline_xattr(inode)) \
3579 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3581 #define stat_inc_inline_inode(inode) \
3583 if (f2fs_has_inline_data(inode)) \
3584 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3586 #define stat_dec_inline_inode(inode) \
3588 if (f2fs_has_inline_data(inode)) \
3589 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3591 #define stat_inc_inline_dir(inode) \
3593 if (f2fs_has_inline_dentry(inode)) \
3594 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3596 #define stat_dec_inline_dir(inode) \
3598 if (f2fs_has_inline_dentry(inode)) \
3599 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3601 #define stat_inc_compr_inode(inode) \
3603 if (f2fs_compressed_file(inode)) \
3604 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3606 #define stat_dec_compr_inode(inode) \
3608 if (f2fs_compressed_file(inode)) \
3609 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3611 #define stat_add_compr_blocks(inode, blocks) \
3612 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3613 #define stat_sub_compr_blocks(inode, blocks) \
3614 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3615 #define stat_inc_meta_count(sbi, blkaddr) \
3617 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3618 atomic_inc(&(sbi)->meta_count[META_CP]); \
3619 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3620 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3621 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3622 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3623 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3624 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3626 #define stat_inc_seg_type(sbi, curseg) \
3627 ((sbi)->segment_count[(curseg)->alloc_type]++)
3628 #define stat_inc_block_count(sbi, curseg) \
3629 ((sbi)->block_count[(curseg)->alloc_type]++)
3630 #define stat_inc_inplace_blocks(sbi) \
3631 (atomic_inc(&(sbi)->inplace_count))
3632 #define stat_update_max_atomic_write(inode) \
3634 int cur = F2FS_I_SB(inode)->atomic_files; \
3635 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3637 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3639 #define stat_inc_volatile_write(inode) \
3640 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3641 #define stat_dec_volatile_write(inode) \
3642 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3643 #define stat_update_max_volatile_write(inode) \
3645 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3646 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3648 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3650 #define stat_inc_seg_count(sbi, type, gc_type) \
3652 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3654 if ((type) == SUM_TYPE_DATA) { \
3656 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3659 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3663 #define stat_inc_tot_blk_count(si, blks) \
3664 ((si)->tot_blks += (blks))
3666 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3668 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3669 stat_inc_tot_blk_count(si, blks); \
3670 si->data_blks += (blks); \
3671 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3674 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3676 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3677 stat_inc_tot_blk_count(si, blks); \
3678 si->node_blks += (blks); \
3679 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3682 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3683 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3684 void __init f2fs_create_root_stats(void);
3685 void f2fs_destroy_root_stats(void);
3686 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3688 #define stat_inc_cp_count(si) do { } while (0)
3689 #define stat_inc_bg_cp_count(si) do { } while (0)
3690 #define stat_inc_call_count(si) do { } while (0)
3691 #define stat_inc_bggc_count(si) do { } while (0)
3692 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3693 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3694 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3695 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3696 #define stat_inc_total_hit(sbi) do { } while (0)
3697 #define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3698 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3699 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3700 #define stat_inc_inline_xattr(inode) do { } while (0)
3701 #define stat_dec_inline_xattr(inode) do { } while (0)
3702 #define stat_inc_inline_inode(inode) do { } while (0)
3703 #define stat_dec_inline_inode(inode) do { } while (0)
3704 #define stat_inc_inline_dir(inode) do { } while (0)
3705 #define stat_dec_inline_dir(inode) do { } while (0)
3706 #define stat_inc_compr_inode(inode) do { } while (0)
3707 #define stat_dec_compr_inode(inode) do { } while (0)
3708 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
3709 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3710 #define stat_inc_atomic_write(inode) do { } while (0)
3711 #define stat_dec_atomic_write(inode) do { } while (0)
3712 #define stat_update_max_atomic_write(inode) do { } while (0)
3713 #define stat_inc_volatile_write(inode) do { } while (0)
3714 #define stat_dec_volatile_write(inode) do { } while (0)
3715 #define stat_update_max_volatile_write(inode) do { } while (0)
3716 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3717 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3718 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3719 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3720 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3721 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3722 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3723 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3725 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3726 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3727 static inline void __init f2fs_create_root_stats(void) { }
3728 static inline void f2fs_destroy_root_stats(void) { }
3729 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3732 extern const struct file_operations f2fs_dir_operations;
3733 #ifdef CONFIG_UNICODE
3734 extern const struct dentry_operations f2fs_dentry_ops;
3736 extern const struct file_operations f2fs_file_operations;
3737 extern const struct inode_operations f2fs_file_inode_operations;
3738 extern const struct address_space_operations f2fs_dblock_aops;
3739 extern const struct address_space_operations f2fs_node_aops;
3740 extern const struct address_space_operations f2fs_meta_aops;
3741 extern const struct inode_operations f2fs_dir_inode_operations;
3742 extern const struct inode_operations f2fs_symlink_inode_operations;
3743 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3744 extern const struct inode_operations f2fs_special_inode_operations;
3745 extern struct kmem_cache *f2fs_inode_entry_slab;
3750 bool f2fs_may_inline_data(struct inode *inode);
3751 bool f2fs_sanity_check_inline_data(struct inode *inode);
3752 bool f2fs_may_inline_dentry(struct inode *inode);
3753 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3754 void f2fs_truncate_inline_inode(struct inode *inode,
3755 struct page *ipage, u64 from);
3756 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3757 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3758 int f2fs_convert_inline_inode(struct inode *inode);
3759 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3760 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3761 int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3762 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3763 const struct f2fs_filename *fname,
3764 struct page **res_page);
3765 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3766 struct page *ipage);
3767 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3768 struct inode *inode, nid_t ino, umode_t mode);
3769 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3770 struct page *page, struct inode *dir,
3771 struct inode *inode);
3772 bool f2fs_empty_inline_dir(struct inode *dir);
3773 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3774 struct fscrypt_str *fstr);
3775 int f2fs_inline_data_fiemap(struct inode *inode,
3776 struct fiemap_extent_info *fieinfo,
3777 __u64 start, __u64 len);
3782 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3783 struct shrink_control *sc);
3784 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3785 struct shrink_control *sc);
3786 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3787 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3792 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3793 struct rb_entry *cached_re, unsigned int ofs);
3794 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
3795 struct rb_root_cached *root,
3796 struct rb_node **parent,
3797 unsigned long long key, bool *left_most);
3798 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3799 struct rb_root_cached *root,
3800 struct rb_node **parent,
3801 unsigned int ofs, bool *leftmost);
3802 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3803 struct rb_entry *cached_re, unsigned int ofs,
3804 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3805 struct rb_node ***insert_p, struct rb_node **insert_parent,
3806 bool force, bool *leftmost);
3807 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3808 struct rb_root_cached *root, bool check_key);
3809 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3810 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3811 void f2fs_drop_extent_tree(struct inode *inode);
3812 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3813 void f2fs_destroy_extent_tree(struct inode *inode);
3814 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3815 struct extent_info *ei);
3816 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3817 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3818 pgoff_t fofs, block_t blkaddr, unsigned int len);
3819 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3820 int __init f2fs_create_extent_cache(void);
3821 void f2fs_destroy_extent_cache(void);
3826 int __init f2fs_init_sysfs(void);
3827 void f2fs_exit_sysfs(void);
3828 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3829 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3832 extern const struct fsverity_operations f2fs_verityops;
3837 static inline bool f2fs_encrypted_file(struct inode *inode)
3839 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3842 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3844 #ifdef CONFIG_FS_ENCRYPTION
3845 file_set_encrypt(inode);
3846 f2fs_set_inode_flags(inode);
3851 * Returns true if the reads of the inode's data need to undergo some
3852 * postprocessing step, like decryption or authenticity verification.
3854 static inline bool f2fs_post_read_required(struct inode *inode)
3856 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3857 f2fs_compressed_file(inode);
3863 #ifdef CONFIG_F2FS_FS_COMPRESSION
3864 bool f2fs_is_compressed_page(struct page *page);
3865 struct page *f2fs_compress_control_page(struct page *page);
3866 int f2fs_prepare_compress_overwrite(struct inode *inode,
3867 struct page **pagep, pgoff_t index, void **fsdata);
3868 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3869 pgoff_t index, unsigned copied);
3870 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3871 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3872 bool f2fs_is_compress_backend_ready(struct inode *inode);
3873 int f2fs_init_compress_mempool(void);
3874 void f2fs_destroy_compress_mempool(void);
3875 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity);
3876 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3877 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3878 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3879 int f2fs_write_multi_pages(struct compress_ctx *cc,
3881 struct writeback_control *wbc,
3882 enum iostat_type io_type);
3883 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3884 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3885 unsigned nr_pages, sector_t *last_block_in_bio,
3886 bool is_readahead, bool for_write);
3887 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3888 void f2fs_free_dic(struct decompress_io_ctx *dic);
3889 void f2fs_decompress_end_io(struct page **rpages,
3890 unsigned int cluster_size, bool err, bool verity);
3891 int f2fs_init_compress_ctx(struct compress_ctx *cc);
3892 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
3893 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3894 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
3895 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
3896 int __init f2fs_init_compress_cache(void);
3897 void f2fs_destroy_compress_cache(void);
3899 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3900 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3902 if (!f2fs_compressed_file(inode))
3904 /* not support compression */
3907 static inline struct page *f2fs_compress_control_page(struct page *page)
3910 return ERR_PTR(-EINVAL);
3912 static inline int f2fs_init_compress_mempool(void) { return 0; }
3913 static inline void f2fs_destroy_compress_mempool(void) { }
3914 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
3915 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
3916 static inline int __init f2fs_init_compress_cache(void) { return 0; }
3917 static inline void f2fs_destroy_compress_cache(void) { }
3920 static inline void set_compress_context(struct inode *inode)
3922 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3924 F2FS_I(inode)->i_compress_algorithm =
3925 F2FS_OPTION(sbi).compress_algorithm;
3926 F2FS_I(inode)->i_log_cluster_size =
3927 F2FS_OPTION(sbi).compress_log_size;
3928 F2FS_I(inode)->i_cluster_size =
3929 1 << F2FS_I(inode)->i_log_cluster_size;
3930 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
3931 set_inode_flag(inode, FI_COMPRESSED_FILE);
3932 stat_inc_compr_inode(inode);
3933 f2fs_mark_inode_dirty_sync(inode, true);
3936 static inline bool f2fs_disable_compressed_file(struct inode *inode)
3938 struct f2fs_inode_info *fi = F2FS_I(inode);
3940 if (!f2fs_compressed_file(inode))
3942 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))
3945 fi->i_flags &= ~F2FS_COMPR_FL;
3946 stat_dec_compr_inode(inode);
3947 clear_inode_flag(inode, FI_COMPRESSED_FILE);
3948 f2fs_mark_inode_dirty_sync(inode, true);
3952 #define F2FS_FEATURE_FUNCS(name, flagname) \
3953 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3955 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3958 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3959 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3960 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3961 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3962 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3963 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3964 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3965 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3966 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3967 F2FS_FEATURE_FUNCS(verity, VERITY);
3968 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3969 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
3970 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
3972 #ifdef CONFIG_BLK_DEV_ZONED
3973 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
3976 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3978 return test_bit(zno, FDEV(devi).blkz_seq);
3982 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3984 return f2fs_sb_has_blkzoned(sbi);
3987 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
3989 return blk_queue_discard(bdev_get_queue(bdev)) ||
3990 bdev_is_zoned(bdev);
3993 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3997 if (!f2fs_is_multi_device(sbi))
3998 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4000 for (i = 0; i < sbi->s_ndevs; i++)
4001 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4006 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4008 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4009 f2fs_hw_should_discard(sbi);
4012 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4016 if (!f2fs_is_multi_device(sbi))
4017 return bdev_read_only(sbi->sb->s_bdev);
4019 for (i = 0; i < sbi->s_ndevs; i++)
4020 if (bdev_read_only(FDEV(i).bdev))
4025 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4027 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4030 static inline bool f2fs_may_compress(struct inode *inode)
4032 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4033 f2fs_is_atomic_file(inode) ||
4034 f2fs_is_volatile_file(inode))
4036 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4039 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4040 u64 blocks, bool add)
4042 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4043 struct f2fs_inode_info *fi = F2FS_I(inode);
4045 /* don't update i_compr_blocks if saved blocks were released */
4046 if (!add && !atomic_read(&fi->i_compr_blocks))
4050 atomic_add(diff, &fi->i_compr_blocks);
4051 stat_add_compr_blocks(inode, diff);
4053 atomic_sub(diff, &fi->i_compr_blocks);
4054 stat_sub_compr_blocks(inode, diff);
4056 f2fs_mark_inode_dirty_sync(inode, true);
4059 static inline int block_unaligned_IO(struct inode *inode,
4060 struct kiocb *iocb, struct iov_iter *iter)
4062 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4063 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4064 loff_t offset = iocb->ki_pos;
4065 unsigned long align = offset | iov_iter_alignment(iter);
4067 return align & blocksize_mask;
4070 static inline int allow_outplace_dio(struct inode *inode,
4071 struct kiocb *iocb, struct iov_iter *iter)
4073 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4074 int rw = iov_iter_rw(iter);
4076 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4077 !block_unaligned_IO(inode, iocb, iter));
4080 static inline bool f2fs_force_buffered_io(struct inode *inode,
4081 struct kiocb *iocb, struct iov_iter *iter)
4083 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4084 int rw = iov_iter_rw(iter);
4086 if (f2fs_post_read_required(inode))
4088 if (f2fs_is_multi_device(sbi))
4091 * for blkzoned device, fallback direct IO to buffered IO, so
4092 * all IOs can be serialized by log-structured write.
4094 if (f2fs_sb_has_blkzoned(sbi))
4096 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4097 if (block_unaligned_IO(inode, iocb, iter))
4099 if (F2FS_IO_ALIGNED(sbi))
4102 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4103 !IS_SWAPFILE(inode))
4109 #ifdef CONFIG_F2FS_FAULT_INJECTION
4110 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4113 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4116 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4119 if (f2fs_sb_has_quota_ino(sbi))
4121 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4122 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4123 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4129 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4130 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4132 #endif /* _LINUX_F2FS_H */