1 /* SPDX-License-Identifier: GPL-2.0 */
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
11 #include <linux/uio.h>
12 #include <linux/types.h>
13 #include <linux/page-flags.h>
14 #include <linux/buffer_head.h>
15 #include <linux/slab.h>
16 #include <linux/crc32.h>
17 #include <linux/magic.h>
18 #include <linux/kobject.h>
19 #include <linux/sched.h>
20 #include <linux/cred.h>
21 #include <linux/sched/mm.h>
22 #include <linux/vmalloc.h>
23 #include <linux/bio.h>
24 #include <linux/blkdev.h>
25 #include <linux/quotaops.h>
26 #include <linux/part_stat.h>
27 #include <crypto/hash.h>
29 #include <linux/fscrypt.h>
30 #include <linux/fsverity.h>
34 #ifdef CONFIG_F2FS_CHECK_FS
35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
37 #define f2fs_bug_on(sbi, condition) \
39 if (WARN_ON(condition)) \
40 set_sbi_flag(sbi, SBI_NEED_FSCK); \
49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */
66 #ifdef CONFIG_F2FS_FAULT_INJECTION
67 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
69 struct f2fs_fault_info {
71 unsigned int inject_rate;
72 unsigned int inject_type;
75 extern const char *f2fs_fault_name[FAULT_MAX];
76 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
82 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
83 #define F2FS_MOUNT_DISCARD 0x00000004
84 #define F2FS_MOUNT_NOHEAP 0x00000008
85 #define F2FS_MOUNT_XATTR_USER 0x00000010
86 #define F2FS_MOUNT_POSIX_ACL 0x00000020
87 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
88 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
89 #define F2FS_MOUNT_INLINE_DATA 0x00000100
90 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
91 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
92 #define F2FS_MOUNT_NOBARRIER 0x00000800
93 #define F2FS_MOUNT_FASTBOOT 0x00001000
94 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
95 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
96 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
97 #define F2FS_MOUNT_USRQUOTA 0x00080000
98 #define F2FS_MOUNT_GRPQUOTA 0x00100000
99 #define F2FS_MOUNT_PRJQUOTA 0x00200000
100 #define F2FS_MOUNT_QUOTA 0x00400000
101 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
102 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
103 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
104 #define F2FS_MOUNT_NORECOVERY 0x04000000
105 #define F2FS_MOUNT_ATGC 0x08000000
106 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000
107 #define F2FS_MOUNT_GC_MERGE 0x20000000
108 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000
110 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
111 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
112 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
113 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
115 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
116 typecheck(unsigned long long, b) && \
117 ((long long)((a) - (b)) > 0))
119 typedef u32 block_t; /*
120 * should not change u32, since it is the on-disk block
121 * address format, __le32.
125 #define COMPRESS_EXT_NUM 16
128 * An implementation of an rwsem that is explicitly unfair to readers. This
129 * prevents priority inversion when a low-priority reader acquires the read lock
130 * while sleeping on the write lock but the write lock is needed by
131 * higher-priority clients.
135 struct rw_semaphore internal_rwsem;
136 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
137 wait_queue_head_t read_waiters;
141 struct f2fs_mount_info {
143 int write_io_size_bits; /* Write IO size bits */
144 block_t root_reserved_blocks; /* root reserved blocks */
145 kuid_t s_resuid; /* reserved blocks for uid */
146 kgid_t s_resgid; /* reserved blocks for gid */
147 int active_logs; /* # of active logs */
148 int inline_xattr_size; /* inline xattr size */
149 #ifdef CONFIG_F2FS_FAULT_INJECTION
150 struct f2fs_fault_info fault_info; /* For fault injection */
153 /* Names of quota files with journalled quota */
154 char *s_qf_names[MAXQUOTAS];
155 int s_jquota_fmt; /* Format of quota to use */
157 /* For which write hints are passed down to block layer */
158 int alloc_mode; /* segment allocation policy */
159 int fsync_mode; /* fsync policy */
160 int fs_mode; /* fs mode: LFS or ADAPTIVE */
161 int bggc_mode; /* bggc mode: off, on or sync */
162 int memory_mode; /* memory mode */
164 * discard command's offset/size should
165 * be aligned to this unit: block,
168 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
169 block_t unusable_cap_perc; /* percentage for cap */
170 block_t unusable_cap; /* Amount of space allowed to be
171 * unusable when disabling checkpoint
174 /* For compression */
175 unsigned char compress_algorithm; /* algorithm type */
176 unsigned char compress_log_size; /* cluster log size */
177 unsigned char compress_level; /* compress level */
178 bool compress_chksum; /* compressed data chksum */
179 unsigned char compress_ext_cnt; /* extension count */
180 unsigned char nocompress_ext_cnt; /* nocompress extension count */
181 int compress_mode; /* compression mode */
182 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
183 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
186 #define F2FS_FEATURE_ENCRYPT 0x0001
187 #define F2FS_FEATURE_BLKZONED 0x0002
188 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
189 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
190 #define F2FS_FEATURE_PRJQUOTA 0x0010
191 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
192 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
193 #define F2FS_FEATURE_QUOTA_INO 0x0080
194 #define F2FS_FEATURE_INODE_CRTIME 0x0100
195 #define F2FS_FEATURE_LOST_FOUND 0x0200
196 #define F2FS_FEATURE_VERITY 0x0400
197 #define F2FS_FEATURE_SB_CHKSUM 0x0800
198 #define F2FS_FEATURE_CASEFOLD 0x1000
199 #define F2FS_FEATURE_COMPRESSION 0x2000
200 #define F2FS_FEATURE_RO 0x4000
202 #define __F2FS_HAS_FEATURE(raw_super, mask) \
203 ((raw_super->feature & cpu_to_le32(mask)) != 0)
204 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
205 #define F2FS_SET_FEATURE(sbi, mask) \
206 (sbi->raw_super->feature |= cpu_to_le32(mask))
207 #define F2FS_CLEAR_FEATURE(sbi, mask) \
208 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
211 * Default values for user and/or group using reserved blocks
213 #define F2FS_DEF_RESUID 0
214 #define F2FS_DEF_RESGID 0
217 * For checkpoint manager
224 #define CP_UMOUNT 0x00000001
225 #define CP_FASTBOOT 0x00000002
226 #define CP_SYNC 0x00000004
227 #define CP_RECOVERY 0x00000008
228 #define CP_DISCARD 0x00000010
229 #define CP_TRIMMED 0x00000020
230 #define CP_PAUSE 0x00000040
231 #define CP_RESIZE 0x00000080
233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
238 #define DEF_CP_INTERVAL 60 /* 60 secs */
239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
252 * indicate meta/data type
261 DATA_GENERIC, /* check range only */
262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
263 DATA_GENERIC_ENHANCE_READ, /*
264 * strong check on range and segment
265 * bitmap but no warning due to race
266 * condition of read on truncated area
269 DATA_GENERIC_ENHANCE_UPDATE, /*
270 * strong check on range and segment
271 * bitmap for update case
276 /* for the list of ino */
278 ORPHAN_INO, /* for orphan ino list */
279 APPEND_INO, /* for append ino list */
280 UPDATE_INO, /* for update ino list */
281 TRANS_DIR_INO, /* for transactions dir ino list */
282 FLUSH_INO, /* for multiple device flushing */
283 MAX_INO_ENTRY, /* max. list */
287 struct list_head list; /* list head */
288 nid_t ino; /* inode number */
289 unsigned int dirty_device; /* dirty device bitmap */
292 /* for the list of inodes to be GCed */
294 struct list_head list; /* list head */
295 struct inode *inode; /* vfs inode pointer */
298 struct fsync_node_entry {
299 struct list_head list; /* list head */
300 struct page *page; /* warm node page pointer */
301 unsigned int seq_id; /* sequence id */
305 struct completion wait; /* completion for checkpoint done */
306 struct llist_node llnode; /* llist_node to be linked in wait queue */
307 int ret; /* return code of checkpoint */
308 ktime_t queue_time; /* request queued time */
311 struct ckpt_req_control {
312 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
313 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
314 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
315 atomic_t issued_ckpt; /* # of actually issued ckpts */
316 atomic_t total_ckpt; /* # of total ckpts */
317 atomic_t queued_ckpt; /* # of queued ckpts */
318 struct llist_head issue_list; /* list for command issue */
319 spinlock_t stat_lock; /* lock for below checkpoint time stats */
320 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
321 unsigned int peak_time; /* peak wait time in msec until now */
324 /* for the bitmap indicate blocks to be discarded */
325 struct discard_entry {
326 struct list_head list; /* list head */
327 block_t start_blkaddr; /* start blockaddr of current segment */
328 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
331 /* default discard granularity of inner discard thread, unit: block count */
332 #define DEFAULT_DISCARD_GRANULARITY 16
334 /* max discard pend list number */
335 #define MAX_PLIST_NUM 512
336 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
337 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
340 D_PREP, /* initial */
341 D_PARTIAL, /* partially submitted */
342 D_SUBMIT, /* all submitted */
343 D_DONE, /* finished */
346 struct discard_info {
347 block_t lstart; /* logical start address */
348 block_t len; /* length */
349 block_t start; /* actual start address in dev */
353 struct rb_node rb_node; /* rb node located in rb-tree */
356 block_t lstart; /* logical start address */
357 block_t len; /* length */
358 block_t start; /* actual start address in dev */
360 struct discard_info di; /* discard info */
363 struct list_head list; /* command list */
364 struct completion wait; /* compleation */
365 struct block_device *bdev; /* bdev */
366 unsigned short ref; /* reference count */
367 unsigned char state; /* state */
368 unsigned char queued; /* queued discard */
369 int error; /* bio error */
370 spinlock_t lock; /* for state/bio_ref updating */
371 unsigned short bio_ref; /* bio reference count */
382 struct discard_policy {
383 int type; /* type of discard */
384 unsigned int min_interval; /* used for candidates exist */
385 unsigned int mid_interval; /* used for device busy */
386 unsigned int max_interval; /* used for candidates not exist */
387 unsigned int max_requests; /* # of discards issued per round */
388 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
389 bool io_aware; /* issue discard in idle time */
390 bool sync; /* submit discard with REQ_SYNC flag */
391 bool ordered; /* issue discard by lba order */
392 bool timeout; /* discard timeout for put_super */
393 unsigned int granularity; /* discard granularity */
396 struct discard_cmd_control {
397 struct task_struct *f2fs_issue_discard; /* discard thread */
398 struct list_head entry_list; /* 4KB discard entry list */
399 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
400 struct list_head wait_list; /* store on-flushing entries */
401 struct list_head fstrim_list; /* in-flight discard from fstrim */
402 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
403 unsigned int discard_wake; /* to wake up discard thread */
404 struct mutex cmd_lock;
405 unsigned int nr_discards; /* # of discards in the list */
406 unsigned int max_discards; /* max. discards to be issued */
407 unsigned int max_discard_request; /* max. discard request per round */
408 unsigned int min_discard_issue_time; /* min. interval between discard issue */
409 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */
410 unsigned int max_discard_issue_time; /* max. interval between discard issue */
411 unsigned int discard_granularity; /* discard granularity */
412 unsigned int undiscard_blks; /* # of undiscard blocks */
413 unsigned int next_pos; /* next discard position */
414 atomic_t issued_discard; /* # of issued discard */
415 atomic_t queued_discard; /* # of queued discard */
416 atomic_t discard_cmd_cnt; /* # of cached cmd count */
417 struct rb_root_cached root; /* root of discard rb-tree */
418 bool rbtree_check; /* config for consistence check */
421 /* for the list of fsync inodes, used only during recovery */
422 struct fsync_inode_entry {
423 struct list_head list; /* list head */
424 struct inode *inode; /* vfs inode pointer */
425 block_t blkaddr; /* block address locating the last fsync */
426 block_t last_dentry; /* block address locating the last dentry */
429 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
430 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
432 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
433 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
434 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
435 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
437 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
438 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
440 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
442 int before = nats_in_cursum(journal);
444 journal->n_nats = cpu_to_le16(before + i);
448 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
450 int before = sits_in_cursum(journal);
452 journal->n_sits = cpu_to_le16(before + i);
456 static inline bool __has_cursum_space(struct f2fs_journal *journal,
459 if (type == NAT_JOURNAL)
460 return size <= MAX_NAT_JENTRIES(journal);
461 return size <= MAX_SIT_JENTRIES(journal);
464 /* for inline stuff */
465 #define DEF_INLINE_RESERVED_SIZE 1
466 static inline int get_extra_isize(struct inode *inode);
467 static inline int get_inline_xattr_addrs(struct inode *inode);
468 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
469 (CUR_ADDRS_PER_INODE(inode) - \
470 get_inline_xattr_addrs(inode) - \
471 DEF_INLINE_RESERVED_SIZE))
474 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
475 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
477 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
478 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
479 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
480 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
481 NR_INLINE_DENTRY(inode) + \
482 INLINE_DENTRY_BITMAP_SIZE(inode)))
485 * For INODE and NODE manager
487 /* for directory operations */
489 struct f2fs_filename {
491 * The filename the user specified. This is NULL for some
492 * filesystem-internal operations, e.g. converting an inline directory
493 * to a non-inline one, or roll-forward recovering an encrypted dentry.
495 const struct qstr *usr_fname;
498 * The on-disk filename. For encrypted directories, this is encrypted.
499 * This may be NULL for lookups in an encrypted dir without the key.
501 struct fscrypt_str disk_name;
503 /* The dirhash of this filename */
506 #ifdef CONFIG_FS_ENCRYPTION
508 * For lookups in encrypted directories: either the buffer backing
509 * disk_name, or a buffer that holds the decoded no-key name.
511 struct fscrypt_str crypto_buf;
513 #if IS_ENABLED(CONFIG_UNICODE)
515 * For casefolded directories: the casefolded name, but it's left NULL
516 * if the original name is not valid Unicode, if the original name is
517 * "." or "..", if the directory is both casefolded and encrypted and
518 * its encryption key is unavailable, or if the filesystem is doing an
519 * internal operation where usr_fname is also NULL. In all these cases
520 * we fall back to treating the name as an opaque byte sequence.
522 struct fscrypt_str cf_name;
526 struct f2fs_dentry_ptr {
529 struct f2fs_dir_entry *dentry;
530 __u8 (*filename)[F2FS_SLOT_LEN];
535 static inline void make_dentry_ptr_block(struct inode *inode,
536 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
539 d->max = NR_DENTRY_IN_BLOCK;
540 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
541 d->bitmap = t->dentry_bitmap;
542 d->dentry = t->dentry;
543 d->filename = t->filename;
546 static inline void make_dentry_ptr_inline(struct inode *inode,
547 struct f2fs_dentry_ptr *d, void *t)
549 int entry_cnt = NR_INLINE_DENTRY(inode);
550 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
551 int reserved_size = INLINE_RESERVED_SIZE(inode);
555 d->nr_bitmap = bitmap_size;
557 d->dentry = t + bitmap_size + reserved_size;
558 d->filename = t + bitmap_size + reserved_size +
559 SIZE_OF_DIR_ENTRY * entry_cnt;
563 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
564 * as its node offset to distinguish from index node blocks.
565 * But some bits are used to mark the node block.
567 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
570 ALLOC_NODE, /* allocate a new node page if needed */
571 LOOKUP_NODE, /* look up a node without readahead */
573 * look up a node with readahead called
578 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
580 /* congestion wait timeout value, default: 20ms */
581 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
583 /* maximum retry quota flush count */
584 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
586 /* maximum retry of EIO'ed page */
587 #define MAX_RETRY_PAGE_EIO 100
589 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
591 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
593 /* dirty segments threshold for triggering CP */
594 #define DEFAULT_DIRTY_THRESHOLD 4
596 /* for in-memory extent cache entry */
597 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
599 /* number of extent info in extent cache we try to shrink */
600 #define EXTENT_CACHE_SHRINK_NUMBER 128
602 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS
603 #define RECOVERY_MIN_RA_BLOCKS 1
605 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */
608 struct rb_node rb_node; /* rb node located in rb-tree */
611 unsigned int ofs; /* start offset of the entry */
612 unsigned int len; /* length of the entry */
614 unsigned long long key; /* 64-bits key */
619 unsigned int fofs; /* start offset in a file */
620 unsigned int len; /* length of the extent */
621 u32 blk; /* start block address of the extent */
622 #ifdef CONFIG_F2FS_FS_COMPRESSION
623 unsigned int c_len; /* physical extent length of compressed blocks */
628 struct rb_node rb_node; /* rb node located in rb-tree */
629 struct extent_info ei; /* extent info */
630 struct list_head list; /* node in global extent list of sbi */
631 struct extent_tree *et; /* extent tree pointer */
635 nid_t ino; /* inode number */
636 struct rb_root_cached root; /* root of extent info rb-tree */
637 struct extent_node *cached_en; /* recently accessed extent node */
638 struct extent_info largest; /* largested extent info */
639 struct list_head list; /* to be used by sbi->zombie_list */
640 rwlock_t lock; /* protect extent info rb-tree */
641 atomic_t node_cnt; /* # of extent node in rb-tree*/
642 bool largest_updated; /* largest extent updated */
646 * This structure is taken from ext4_map_blocks.
648 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
650 #define F2FS_MAP_NEW (1 << BH_New)
651 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
652 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
653 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
656 struct f2fs_map_blocks {
657 struct block_device *m_bdev; /* for multi-device dio */
661 unsigned int m_flags;
662 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
663 pgoff_t *m_next_extent; /* point to next possible extent */
665 bool m_may_create; /* indicate it is from write path */
666 bool m_multidev_dio; /* indicate it allows multi-device dio */
669 /* for flag in get_data_block */
671 F2FS_GET_BLOCK_DEFAULT,
672 F2FS_GET_BLOCK_FIEMAP,
675 F2FS_GET_BLOCK_PRE_DIO,
676 F2FS_GET_BLOCK_PRE_AIO,
677 F2FS_GET_BLOCK_PRECACHE,
681 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
683 #define FADVISE_COLD_BIT 0x01
684 #define FADVISE_LOST_PINO_BIT 0x02
685 #define FADVISE_ENCRYPT_BIT 0x04
686 #define FADVISE_ENC_NAME_BIT 0x08
687 #define FADVISE_KEEP_SIZE_BIT 0x10
688 #define FADVISE_HOT_BIT 0x20
689 #define FADVISE_VERITY_BIT 0x40
690 #define FADVISE_TRUNC_BIT 0x80
692 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
694 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
695 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
696 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
698 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
699 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
700 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
702 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
703 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
705 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
706 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
708 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
709 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
711 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
712 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
713 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
715 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
716 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
718 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)
719 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)
720 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)
722 #define DEF_DIR_LEVEL 0
729 /* used for f2fs_inode_info->flags */
731 FI_NEW_INODE, /* indicate newly allocated inode */
732 FI_DIRTY_INODE, /* indicate inode is dirty or not */
733 FI_AUTO_RECOVER, /* indicate inode is recoverable */
734 FI_DIRTY_DIR, /* indicate directory has dirty pages */
735 FI_INC_LINK, /* need to increment i_nlink */
736 FI_ACL_MODE, /* indicate acl mode */
737 FI_NO_ALLOC, /* should not allocate any blocks */
738 FI_FREE_NID, /* free allocated nide */
739 FI_NO_EXTENT, /* not to use the extent cache */
740 FI_INLINE_XATTR, /* used for inline xattr */
741 FI_INLINE_DATA, /* used for inline data*/
742 FI_INLINE_DENTRY, /* used for inline dentry */
743 FI_APPEND_WRITE, /* inode has appended data */
744 FI_UPDATE_WRITE, /* inode has in-place-update data */
745 FI_NEED_IPU, /* used for ipu per file */
746 FI_ATOMIC_FILE, /* indicate atomic file */
747 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
748 FI_DROP_CACHE, /* drop dirty page cache */
749 FI_DATA_EXIST, /* indicate data exists */
750 FI_INLINE_DOTS, /* indicate inline dot dentries */
751 FI_SKIP_WRITES, /* should skip data page writeback */
752 FI_OPU_WRITE, /* used for opu per file */
753 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
754 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */
755 FI_HOT_DATA, /* indicate file is hot */
756 FI_EXTRA_ATTR, /* indicate file has extra attribute */
757 FI_PROJ_INHERIT, /* indicate file inherits projectid */
758 FI_PIN_FILE, /* indicate file should not be gced */
759 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
760 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
761 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
762 FI_MMAP_FILE, /* indicate file was mmapped */
763 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
764 FI_COMPRESS_RELEASED, /* compressed blocks were released */
765 FI_ALIGNED_WRITE, /* enable aligned write */
766 FI_COW_FILE, /* indicate COW file */
767 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */
768 FI_MAX, /* max flag, never be used */
771 struct f2fs_inode_info {
772 struct inode vfs_inode; /* serve a vfs inode */
773 unsigned long i_flags; /* keep an inode flags for ioctl */
774 unsigned char i_advise; /* use to give file attribute hints */
775 unsigned char i_dir_level; /* use for dentry level for large dir */
776 unsigned int i_current_depth; /* only for directory depth */
777 /* for gc failure statistic */
778 unsigned int i_gc_failures[MAX_GC_FAILURE];
779 unsigned int i_pino; /* parent inode number */
780 umode_t i_acl_mode; /* keep file acl mode temporarily */
782 /* Use below internally in f2fs*/
783 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
784 struct f2fs_rwsem i_sem; /* protect fi info */
785 atomic_t dirty_pages; /* # of dirty pages */
786 f2fs_hash_t chash; /* hash value of given file name */
787 unsigned int clevel; /* maximum level of given file name */
788 struct task_struct *task; /* lookup and create consistency */
789 struct task_struct *cp_task; /* separate cp/wb IO stats*/
790 struct task_struct *wb_task; /* indicate inode is in context of writeback */
791 nid_t i_xattr_nid; /* node id that contains xattrs */
792 loff_t last_disk_size; /* lastly written file size */
793 spinlock_t i_size_lock; /* protect last_disk_size */
796 struct dquot *i_dquot[MAXQUOTAS];
798 /* quota space reservation, managed internally by quota code */
799 qsize_t i_reserved_quota;
801 struct list_head dirty_list; /* dirty list for dirs and files */
802 struct list_head gdirty_list; /* linked in global dirty list */
803 struct task_struct *atomic_write_task; /* store atomic write task */
804 struct extent_tree *extent_tree; /* cached extent_tree entry */
805 struct inode *cow_inode; /* copy-on-write inode for atomic write */
807 /* avoid racing between foreground op and gc */
808 struct f2fs_rwsem i_gc_rwsem[2];
809 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */
811 int i_extra_isize; /* size of extra space located in i_addr */
812 kprojid_t i_projid; /* id for project quota */
813 int i_inline_xattr_size; /* inline xattr size */
814 struct timespec64 i_crtime; /* inode creation time */
815 struct timespec64 i_disk_time[4];/* inode disk times */
817 /* for file compress */
818 atomic_t i_compr_blocks; /* # of compressed blocks */
819 unsigned char i_compress_algorithm; /* algorithm type */
820 unsigned char i_log_cluster_size; /* log of cluster size */
821 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
822 unsigned short i_compress_flag; /* compress flag */
823 unsigned int i_cluster_size; /* cluster size */
825 unsigned int atomic_write_cnt;
826 loff_t original_i_size; /* original i_size before atomic write */
829 static inline void get_extent_info(struct extent_info *ext,
830 struct f2fs_extent *i_ext)
832 ext->fofs = le32_to_cpu(i_ext->fofs);
833 ext->blk = le32_to_cpu(i_ext->blk);
834 ext->len = le32_to_cpu(i_ext->len);
837 static inline void set_raw_extent(struct extent_info *ext,
838 struct f2fs_extent *i_ext)
840 i_ext->fofs = cpu_to_le32(ext->fofs);
841 i_ext->blk = cpu_to_le32(ext->blk);
842 i_ext->len = cpu_to_le32(ext->len);
845 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
846 u32 blk, unsigned int len)
851 #ifdef CONFIG_F2FS_FS_COMPRESSION
856 static inline bool __is_discard_mergeable(struct discard_info *back,
857 struct discard_info *front, unsigned int max_len)
859 return (back->lstart + back->len == front->lstart) &&
860 (back->len + front->len <= max_len);
863 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
864 struct discard_info *back, unsigned int max_len)
866 return __is_discard_mergeable(back, cur, max_len);
869 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
870 struct discard_info *front, unsigned int max_len)
872 return __is_discard_mergeable(cur, front, max_len);
875 static inline bool __is_extent_mergeable(struct extent_info *back,
876 struct extent_info *front)
878 #ifdef CONFIG_F2FS_FS_COMPRESSION
879 if (back->c_len && back->len != back->c_len)
881 if (front->c_len && front->len != front->c_len)
884 return (back->fofs + back->len == front->fofs &&
885 back->blk + back->len == front->blk);
888 static inline bool __is_back_mergeable(struct extent_info *cur,
889 struct extent_info *back)
891 return __is_extent_mergeable(back, cur);
894 static inline bool __is_front_mergeable(struct extent_info *cur,
895 struct extent_info *front)
897 return __is_extent_mergeable(cur, front);
900 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
901 static inline void __try_update_largest_extent(struct extent_tree *et,
902 struct extent_node *en)
904 if (en->ei.len > et->largest.len) {
905 et->largest = en->ei;
906 et->largest_updated = true;
911 * For free nid management
914 FREE_NID, /* newly added to free nid list */
915 PREALLOC_NID, /* it is preallocated */
926 struct f2fs_nm_info {
927 block_t nat_blkaddr; /* base disk address of NAT */
928 nid_t max_nid; /* maximum possible node ids */
929 nid_t available_nids; /* # of available node ids */
930 nid_t next_scan_nid; /* the next nid to be scanned */
931 nid_t max_rf_node_blocks; /* max # of nodes for recovery */
932 unsigned int ram_thresh; /* control the memory footprint */
933 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
934 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
936 /* NAT cache management */
937 struct radix_tree_root nat_root;/* root of the nat entry cache */
938 struct radix_tree_root nat_set_root;/* root of the nat set cache */
939 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */
940 struct list_head nat_entries; /* cached nat entry list (clean) */
941 spinlock_t nat_list_lock; /* protect clean nat entry list */
942 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
943 unsigned int nat_blocks; /* # of nat blocks */
945 /* free node ids management */
946 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
947 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
948 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
949 spinlock_t nid_list_lock; /* protect nid lists ops */
950 struct mutex build_lock; /* lock for build free nids */
951 unsigned char **free_nid_bitmap;
952 unsigned char *nat_block_bitmap;
953 unsigned short *free_nid_count; /* free nid count of NAT block */
956 char *nat_bitmap; /* NAT bitmap pointer */
958 unsigned int nat_bits_blocks; /* # of nat bits blocks */
959 unsigned char *nat_bits; /* NAT bits blocks */
960 unsigned char *full_nat_bits; /* full NAT pages */
961 unsigned char *empty_nat_bits; /* empty NAT pages */
962 #ifdef CONFIG_F2FS_CHECK_FS
963 char *nat_bitmap_mir; /* NAT bitmap mirror */
965 int bitmap_size; /* bitmap size */
969 * this structure is used as one of function parameters.
970 * all the information are dedicated to a given direct node block determined
971 * by the data offset in a file.
973 struct dnode_of_data {
974 struct inode *inode; /* vfs inode pointer */
975 struct page *inode_page; /* its inode page, NULL is possible */
976 struct page *node_page; /* cached direct node page */
977 nid_t nid; /* node id of the direct node block */
978 unsigned int ofs_in_node; /* data offset in the node page */
979 bool inode_page_locked; /* inode page is locked or not */
980 bool node_changed; /* is node block changed */
981 char cur_level; /* level of hole node page */
982 char max_level; /* level of current page located */
983 block_t data_blkaddr; /* block address of the node block */
986 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
987 struct page *ipage, struct page *npage, nid_t nid)
989 memset(dn, 0, sizeof(*dn));
991 dn->inode_page = ipage;
992 dn->node_page = npage;
999 * By default, there are 6 active log areas across the whole main area.
1000 * When considering hot and cold data separation to reduce cleaning overhead,
1001 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
1003 * In the current design, you should not change the numbers intentionally.
1004 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
1005 * logs individually according to the underlying devices. (default: 6)
1006 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
1007 * data and 8 for node logs.
1009 #define NR_CURSEG_DATA_TYPE (3)
1010 #define NR_CURSEG_NODE_TYPE (3)
1011 #define NR_CURSEG_INMEM_TYPE (2)
1012 #define NR_CURSEG_RO_TYPE (2)
1013 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
1014 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
1017 CURSEG_HOT_DATA = 0, /* directory entry blocks */
1018 CURSEG_WARM_DATA, /* data blocks */
1019 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
1020 CURSEG_HOT_NODE, /* direct node blocks of directory files */
1021 CURSEG_WARM_NODE, /* direct node blocks of normal files */
1022 CURSEG_COLD_NODE, /* indirect node blocks */
1023 NR_PERSISTENT_LOG, /* number of persistent log */
1024 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
1025 /* pinned file that needs consecutive block address */
1026 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
1027 NO_CHECK_TYPE, /* number of persistent & inmem log */
1031 struct completion wait;
1032 struct llist_node llnode;
1037 struct flush_cmd_control {
1038 struct task_struct *f2fs_issue_flush; /* flush thread */
1039 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
1040 atomic_t issued_flush; /* # of issued flushes */
1041 atomic_t queued_flush; /* # of queued flushes */
1042 struct llist_head issue_list; /* list for command issue */
1043 struct llist_node *dispatch_list; /* list for command dispatch */
1046 struct f2fs_sm_info {
1047 struct sit_info *sit_info; /* whole segment information */
1048 struct free_segmap_info *free_info; /* free segment information */
1049 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1050 struct curseg_info *curseg_array; /* active segment information */
1052 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */
1054 block_t seg0_blkaddr; /* block address of 0'th segment */
1055 block_t main_blkaddr; /* start block address of main area */
1056 block_t ssa_blkaddr; /* start block address of SSA area */
1058 unsigned int segment_count; /* total # of segments */
1059 unsigned int main_segments; /* # of segments in main area */
1060 unsigned int reserved_segments; /* # of reserved segments */
1061 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1062 unsigned int ovp_segments; /* # of overprovision segments */
1064 /* a threshold to reclaim prefree segments */
1065 unsigned int rec_prefree_segments;
1067 /* for batched trimming */
1068 unsigned int trim_sections; /* # of sections to trim */
1070 struct list_head sit_entry_set; /* sit entry set list */
1072 unsigned int ipu_policy; /* in-place-update policy */
1073 unsigned int min_ipu_util; /* in-place-update threshold */
1074 unsigned int min_fsync_blocks; /* threshold for fsync */
1075 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1076 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1077 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1079 /* for flush command control */
1080 struct flush_cmd_control *fcc_info;
1082 /* for discard command control */
1083 struct discard_cmd_control *dcc_info;
1090 * COUNT_TYPE for monitoring
1092 * f2fs monitors the number of several block types such as on-writeback,
1093 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1095 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1114 * The below are the page types of bios used in submit_bio().
1115 * The available types are:
1116 * DATA User data pages. It operates as async mode.
1117 * NODE Node pages. It operates as async mode.
1118 * META FS metadata pages such as SIT, NAT, CP.
1119 * NR_PAGE_TYPE The number of page types.
1120 * META_FLUSH Make sure the previous pages are written
1121 * with waiting the bio's completion
1122 * ... Only can be used with META.
1124 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1127 NODE = 1, /* should not change this */
1131 IPU, /* the below types are used by tracepoints only. */
1136 HOT = 0, /* must be zero for meta bio */
1142 enum need_lock_type {
1148 enum cp_reason_type {
1164 APP_DIRECT_IO, /* app direct write IOs */
1165 APP_BUFFERED_IO, /* app buffered write IOs */
1166 APP_WRITE_IO, /* app write IOs */
1167 APP_MAPPED_IO, /* app mapped IOs */
1168 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */
1169 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */
1170 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1171 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */
1172 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1173 FS_META_IO, /* meta IOs from kworker/reclaimer */
1174 FS_GC_DATA_IO, /* data IOs from forground gc */
1175 FS_GC_NODE_IO, /* node IOs from forground gc */
1176 FS_CP_DATA_IO, /* data IOs from checkpoint */
1177 FS_CP_NODE_IO, /* node IOs from checkpoint */
1178 FS_CP_META_IO, /* meta IOs from checkpoint */
1181 APP_DIRECT_READ_IO, /* app direct read IOs */
1182 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1183 APP_READ_IO, /* app read IOs */
1184 APP_MAPPED_READ_IO, /* app mapped read IOs */
1185 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */
1186 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */
1187 FS_DATA_READ_IO, /* data read IOs */
1188 FS_GDATA_READ_IO, /* data read IOs from background gc */
1189 FS_CDATA_READ_IO, /* compressed data read IOs */
1190 FS_NODE_READ_IO, /* node read IOs */
1191 FS_META_READ_IO, /* meta read IOs */
1194 FS_DISCARD, /* discard */
1198 struct f2fs_io_info {
1199 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1200 nid_t ino; /* inode number */
1201 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1202 enum temp_type temp; /* contains HOT/WARM/COLD */
1203 enum req_op op; /* contains REQ_OP_ */
1204 blk_opf_t op_flags; /* req_flag_bits */
1205 block_t new_blkaddr; /* new block address to be written */
1206 block_t old_blkaddr; /* old block address before Cow */
1207 struct page *page; /* page to be written */
1208 struct page *encrypted_page; /* encrypted page */
1209 struct page *compressed_page; /* compressed page */
1210 struct list_head list; /* serialize IOs */
1211 bool submitted; /* indicate IO submission */
1212 int need_lock; /* indicate we need to lock cp_rwsem */
1213 bool in_list; /* indicate fio is in io_list */
1214 bool is_por; /* indicate IO is from recovery or not */
1215 bool retry; /* need to reallocate block address */
1216 int compr_blocks; /* # of compressed block addresses */
1217 bool encrypted; /* indicate file is encrypted */
1218 bool post_read; /* require post read */
1219 enum iostat_type io_type; /* io type */
1220 struct writeback_control *io_wbc; /* writeback control */
1221 struct bio **bio; /* bio for ipu */
1222 sector_t *last_block; /* last block number in bio */
1223 unsigned char version; /* version of the node */
1228 struct list_head list;
1231 #define is_read_io(rw) ((rw) == READ)
1232 struct f2fs_bio_info {
1233 struct f2fs_sb_info *sbi; /* f2fs superblock */
1234 struct bio *bio; /* bios to merge */
1235 sector_t last_block_in_bio; /* last block number */
1236 struct f2fs_io_info fio; /* store buffered io info. */
1237 struct f2fs_rwsem io_rwsem; /* blocking op for bio */
1238 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1239 struct list_head io_list; /* track fios */
1240 struct list_head bio_list; /* bio entry list head */
1241 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */
1244 #define FDEV(i) (sbi->devs[i])
1245 #define RDEV(i) (raw_super->devs[i])
1246 struct f2fs_dev_info {
1247 struct block_device *bdev;
1248 char path[MAX_PATH_LEN];
1249 unsigned int total_segments;
1252 #ifdef CONFIG_BLK_DEV_ZONED
1253 unsigned int nr_blkz; /* Total number of zones */
1254 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1259 DIR_INODE, /* for dirty dir inode */
1260 FILE_INODE, /* for dirty regular/symlink inode */
1261 DIRTY_META, /* for all dirtied inode metadata */
1265 /* for inner inode cache management */
1266 struct inode_management {
1267 struct radix_tree_root ino_root; /* ino entry array */
1268 spinlock_t ino_lock; /* for ino entry lock */
1269 struct list_head ino_list; /* inode list head */
1270 unsigned long ino_num; /* number of entries */
1274 struct atgc_management {
1275 bool atgc_enabled; /* ATGC is enabled or not */
1276 struct rb_root_cached root; /* root of victim rb-tree */
1277 struct list_head victim_list; /* linked with all victim entries */
1278 unsigned int victim_count; /* victim count in rb-tree */
1279 unsigned int candidate_ratio; /* candidate ratio */
1280 unsigned int max_candidate_count; /* max candidate count */
1281 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1282 unsigned long long age_threshold; /* age threshold */
1285 struct f2fs_gc_control {
1286 unsigned int victim_segno; /* target victim segment number */
1287 int init_gc_type; /* FG_GC or BG_GC */
1288 bool no_bg_gc; /* check the space and stop bg_gc */
1289 bool should_migrate_blocks; /* should migrate blocks */
1290 bool err_gc_skipped; /* return EAGAIN if GC skipped */
1291 unsigned int nr_free_secs; /* # of free sections to do GC */
1294 /* For s_flag in struct f2fs_sb_info */
1296 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1297 SBI_IS_CLOSE, /* specify unmounting */
1298 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1299 SBI_POR_DOING, /* recovery is doing or not */
1300 SBI_NEED_SB_WRITE, /* need to recover superblock */
1301 SBI_NEED_CP, /* need to checkpoint */
1302 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1303 SBI_IS_RECOVERED, /* recovered orphan/data */
1304 SBI_CP_DISABLED, /* CP was disabled last mount */
1305 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1306 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1307 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1308 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1309 SBI_IS_RESIZEFS, /* resizefs is in process */
1310 SBI_IS_FREEZING, /* freezefs is in process */
1319 UMOUNT_DISCARD_TIMEOUT,
1335 BGGC_MODE_ON, /* background gc is on */
1336 BGGC_MODE_OFF, /* background gc is off */
1338 * background gc is on, migrating blocks
1339 * like foreground gc
1344 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1345 FS_MODE_LFS, /* use lfs allocation only */
1346 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */
1347 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */
1351 ALLOC_MODE_DEFAULT, /* stay default */
1352 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1356 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1357 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1358 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1363 * automatically compress compression
1367 * automatical compression is disabled.
1368 * user can control the file compression
1374 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1375 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1376 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1380 MEMORY_MODE_NORMAL, /* memory mode for normal devices */
1381 MEMORY_MODE_LOW, /* memory mode for low memry devices */
1386 static inline int f2fs_test_bit(unsigned int nr, char *addr);
1387 static inline void f2fs_set_bit(unsigned int nr, char *addr);
1388 static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1391 * Layout of f2fs page.private:
1393 * Layout A: lowest bit should be 1
1394 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1395 * bit 0 PAGE_PRIVATE_NOT_POINTER
1396 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE
1397 * bit 2 PAGE_PRIVATE_DUMMY_WRITE
1398 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION
1399 * bit 4 PAGE_PRIVATE_INLINE_INODE
1400 * bit 5 PAGE_PRIVATE_REF_RESOURCE
1401 * bit 6- f2fs private data
1403 * Layout B: lowest bit should be 0
1404 * page.private is a wrapped pointer.
1407 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1408 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */
1409 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1410 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1411 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1412 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1416 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \
1417 static inline bool page_private_##name(struct page *page) \
1419 return PagePrivate(page) && \
1420 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
1421 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1424 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \
1425 static inline void set_page_private_##name(struct page *page) \
1427 if (!PagePrivate(page)) { \
1429 SetPagePrivate(page); \
1430 set_page_private(page, 0); \
1432 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
1433 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1436 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
1437 static inline void clear_page_private_##name(struct page *page) \
1439 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1440 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \
1441 set_page_private(page, 0); \
1442 if (PagePrivate(page)) { \
1443 ClearPagePrivate(page); \
1449 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
1450 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE);
1451 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
1452 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
1453 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);
1454 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
1456 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
1457 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
1458 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
1459 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);
1460 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
1462 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
1463 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
1464 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
1465 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);
1466 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
1468 static inline unsigned long get_page_private_data(struct page *page)
1470 unsigned long data = page_private(page);
1472 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
1474 return data >> PAGE_PRIVATE_MAX;
1477 static inline void set_page_private_data(struct page *page, unsigned long data)
1479 if (!PagePrivate(page)) {
1481 SetPagePrivate(page);
1482 set_page_private(page, 0);
1484 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
1485 page_private(page) |= data << PAGE_PRIVATE_MAX;
1488 static inline void clear_page_private_data(struct page *page)
1490 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1;
1491 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) {
1492 set_page_private(page, 0);
1493 if (PagePrivate(page)) {
1494 ClearPagePrivate(page);
1500 /* For compression */
1501 enum compress_algorithm_type {
1509 enum compress_flag {
1514 #define COMPRESS_WATERMARK 20
1515 #define COMPRESS_PERCENT 20
1517 #define COMPRESS_DATA_RESERVED_SIZE 4
1518 struct compress_data {
1519 __le32 clen; /* compressed data size */
1520 __le32 chksum; /* compressed data chksum */
1521 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1522 u8 cdata[]; /* compressed data */
1525 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1527 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1529 #define COMPRESS_LEVEL_OFFSET 8
1531 /* compress context */
1532 struct compress_ctx {
1533 struct inode *inode; /* inode the context belong to */
1534 pgoff_t cluster_idx; /* cluster index number */
1535 unsigned int cluster_size; /* page count in cluster */
1536 unsigned int log_cluster_size; /* log of cluster size */
1537 struct page **rpages; /* pages store raw data in cluster */
1538 unsigned int nr_rpages; /* total page number in rpages */
1539 struct page **cpages; /* pages store compressed data in cluster */
1540 unsigned int nr_cpages; /* total page number in cpages */
1541 unsigned int valid_nr_cpages; /* valid page number in cpages */
1542 void *rbuf; /* virtual mapped address on rpages */
1543 struct compress_data *cbuf; /* virtual mapped address on cpages */
1544 size_t rlen; /* valid data length in rbuf */
1545 size_t clen; /* valid data length in cbuf */
1546 void *private; /* payload buffer for specified compression algorithm */
1547 void *private2; /* extra payload buffer */
1550 /* compress context for write IO path */
1551 struct compress_io_ctx {
1552 u32 magic; /* magic number to indicate page is compressed */
1553 struct inode *inode; /* inode the context belong to */
1554 struct page **rpages; /* pages store raw data in cluster */
1555 unsigned int nr_rpages; /* total page number in rpages */
1556 atomic_t pending_pages; /* in-flight compressed page count */
1559 /* Context for decompressing one cluster on the read IO path */
1560 struct decompress_io_ctx {
1561 u32 magic; /* magic number to indicate page is compressed */
1562 struct inode *inode; /* inode the context belong to */
1563 pgoff_t cluster_idx; /* cluster index number */
1564 unsigned int cluster_size; /* page count in cluster */
1565 unsigned int log_cluster_size; /* log of cluster size */
1566 struct page **rpages; /* pages store raw data in cluster */
1567 unsigned int nr_rpages; /* total page number in rpages */
1568 struct page **cpages; /* pages store compressed data in cluster */
1569 unsigned int nr_cpages; /* total page number in cpages */
1570 struct page **tpages; /* temp pages to pad holes in cluster */
1571 void *rbuf; /* virtual mapped address on rpages */
1572 struct compress_data *cbuf; /* virtual mapped address on cpages */
1573 size_t rlen; /* valid data length in rbuf */
1574 size_t clen; /* valid data length in cbuf */
1577 * The number of compressed pages remaining to be read in this cluster.
1578 * This is initially nr_cpages. It is decremented by 1 each time a page
1579 * has been read (or failed to be read). When it reaches 0, the cluster
1580 * is decompressed (or an error is reported).
1582 * If an error occurs before all the pages have been submitted for I/O,
1583 * then this will never reach 0. In this case the I/O submitter is
1584 * responsible for calling f2fs_decompress_end_io() instead.
1586 atomic_t remaining_pages;
1589 * Number of references to this decompress_io_ctx.
1591 * One reference is held for I/O completion. This reference is dropped
1592 * after the pagecache pages are updated and unlocked -- either after
1593 * decompression (and verity if enabled), or after an error.
1595 * In addition, each compressed page holds a reference while it is in a
1596 * bio. These references are necessary prevent compressed pages from
1597 * being freed while they are still in a bio.
1601 bool failed; /* IO error occurred before decompression? */
1602 bool need_verity; /* need fs-verity verification after decompression? */
1603 void *private; /* payload buffer for specified decompression algorithm */
1604 void *private2; /* extra payload buffer */
1605 struct work_struct verity_work; /* work to verify the decompressed pages */
1606 struct work_struct free_work; /* work for late free this structure itself */
1609 #define NULL_CLUSTER ((unsigned int)(~0))
1610 #define MIN_COMPRESS_LOG_SIZE 2
1611 #define MAX_COMPRESS_LOG_SIZE 8
1612 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1614 struct f2fs_sb_info {
1615 struct super_block *sb; /* pointer to VFS super block */
1616 struct proc_dir_entry *s_proc; /* proc entry */
1617 struct f2fs_super_block *raw_super; /* raw super block pointer */
1618 struct f2fs_rwsem sb_lock; /* lock for raw super block */
1619 int valid_super_block; /* valid super block no */
1620 unsigned long s_flag; /* flags for sbi */
1621 struct mutex writepages; /* mutex for writepages() */
1623 #ifdef CONFIG_BLK_DEV_ZONED
1624 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1625 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1628 /* for node-related operations */
1629 struct f2fs_nm_info *nm_info; /* node manager */
1630 struct inode *node_inode; /* cache node blocks */
1632 /* for segment-related operations */
1633 struct f2fs_sm_info *sm_info; /* segment manager */
1635 /* for bio operations */
1636 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1637 /* keep migration IO order for LFS mode */
1638 struct f2fs_rwsem io_order_lock;
1639 mempool_t *write_io_dummy; /* Dummy pages */
1640 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */
1641 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */
1643 /* for checkpoint */
1644 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1645 int cur_cp_pack; /* remain current cp pack */
1646 spinlock_t cp_lock; /* for flag in ckpt */
1647 struct inode *meta_inode; /* cache meta blocks */
1648 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */
1649 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */
1650 struct f2fs_rwsem node_write; /* locking node writes */
1651 struct f2fs_rwsem node_change; /* locking node change */
1652 wait_queue_head_t cp_wait;
1653 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1654 long interval_time[MAX_TIME]; /* to store thresholds */
1655 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1657 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1659 spinlock_t fsync_node_lock; /* for node entry lock */
1660 struct list_head fsync_node_list; /* node list head */
1661 unsigned int fsync_seg_id; /* sequence id */
1662 unsigned int fsync_node_num; /* number of node entries */
1664 /* for orphan inode, use 0'th array */
1665 unsigned int max_orphans; /* max orphan inodes */
1667 /* for inode management */
1668 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1669 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1670 struct mutex flush_lock; /* for flush exclusion */
1672 /* for extent tree cache */
1673 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1674 struct mutex extent_tree_lock; /* locking extent radix tree */
1675 struct list_head extent_list; /* lru list for shrinker */
1676 spinlock_t extent_lock; /* locking extent lru list */
1677 atomic_t total_ext_tree; /* extent tree count */
1678 struct list_head zombie_list; /* extent zombie tree list */
1679 atomic_t total_zombie_tree; /* extent zombie tree count */
1680 atomic_t total_ext_node; /* extent info count */
1682 /* basic filesystem units */
1683 unsigned int log_sectors_per_block; /* log2 sectors per block */
1684 unsigned int log_blocksize; /* log2 block size */
1685 unsigned int blocksize; /* block size */
1686 unsigned int root_ino_num; /* root inode number*/
1687 unsigned int node_ino_num; /* node inode number*/
1688 unsigned int meta_ino_num; /* meta inode number*/
1689 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1690 unsigned int blocks_per_seg; /* blocks per segment */
1691 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1692 unsigned int segs_per_sec; /* segments per section */
1693 unsigned int secs_per_zone; /* sections per zone */
1694 unsigned int total_sections; /* total section count */
1695 unsigned int total_node_count; /* total node block count */
1696 unsigned int total_valid_node_count; /* valid node block count */
1697 int dir_level; /* directory level */
1698 int readdir_ra; /* readahead inode in readdir */
1699 u64 max_io_bytes; /* max io bytes to merge IOs */
1701 block_t user_block_count; /* # of user blocks */
1702 block_t total_valid_block_count; /* # of valid blocks */
1703 block_t discard_blks; /* discard command candidats */
1704 block_t last_valid_block_count; /* for recovery */
1705 block_t reserved_blocks; /* configurable reserved blocks */
1706 block_t current_reserved_blocks; /* current reserved blocks */
1708 /* Additional tracking for no checkpoint mode */
1709 block_t unusable_block_count; /* # of blocks saved by last cp */
1711 unsigned int nquota_files; /* # of quota sysfile */
1712 struct f2fs_rwsem quota_sem; /* blocking cp for flags */
1714 /* # of pages, see count_type */
1715 atomic_t nr_pages[NR_COUNT_TYPE];
1716 /* # of allocated blocks */
1717 struct percpu_counter alloc_valid_block_count;
1718 /* # of node block writes as roll forward recovery */
1719 struct percpu_counter rf_node_block_count;
1721 /* writeback control */
1722 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1724 /* valid inode count */
1725 struct percpu_counter total_valid_inode_count;
1727 struct f2fs_mount_info mount_opt; /* mount options */
1729 /* for cleaning operations */
1730 struct f2fs_rwsem gc_lock; /*
1731 * semaphore for GC, avoid
1732 * race between GC and GC or CP
1734 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1735 struct atgc_management am; /* atgc management */
1736 unsigned int cur_victim_sec; /* current victim section num */
1737 unsigned int gc_mode; /* current GC state */
1738 unsigned int next_victim_seg[2]; /* next segment in victim section */
1739 spinlock_t gc_urgent_high_lock;
1740 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */
1742 /* for skip statistic */
1743 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1745 /* threshold for gc trials on pinned files */
1746 u64 gc_pin_file_threshold;
1747 struct f2fs_rwsem pin_sem;
1749 /* maximum # of trials to find a victim segment for SSR and GC */
1750 unsigned int max_victim_search;
1751 /* migration granularity of garbage collection, unit: segment */
1752 unsigned int migration_granularity;
1755 * for stat information.
1756 * one is for the LFS mode, and the other is for the SSR mode.
1758 #ifdef CONFIG_F2FS_STAT_FS
1759 struct f2fs_stat_info *stat_info; /* FS status information */
1760 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1761 unsigned int segment_count[2]; /* # of allocated segments */
1762 unsigned int block_count[2]; /* # of allocated blocks */
1763 atomic_t inplace_count; /* # of inplace update */
1764 atomic64_t total_hit_ext; /* # of lookup extent cache */
1765 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1766 atomic64_t read_hit_largest; /* # of hit largest extent node */
1767 atomic64_t read_hit_cached; /* # of hit cached extent node */
1768 atomic_t inline_xattr; /* # of inline_xattr inodes */
1769 atomic_t inline_inode; /* # of inline_data inodes */
1770 atomic_t inline_dir; /* # of inline_dentry inodes */
1771 atomic_t compr_inode; /* # of compressed inodes */
1772 atomic64_t compr_blocks; /* # of compressed blocks */
1773 atomic_t swapfile_inode; /* # of swapfile inodes */
1774 atomic_t atomic_files; /* # of opened atomic file */
1775 atomic_t max_aw_cnt; /* max # of atomic writes */
1776 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1777 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1778 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1780 spinlock_t stat_lock; /* lock for stat operations */
1782 /* to attach REQ_META|REQ_FUA flags */
1783 unsigned int data_io_flag;
1784 unsigned int node_io_flag;
1786 /* For sysfs support */
1787 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1788 struct completion s_kobj_unregister;
1790 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1791 struct completion s_stat_kobj_unregister;
1793 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1794 struct completion s_feature_list_kobj_unregister;
1796 /* For shrinker support */
1797 struct list_head s_list;
1798 struct mutex umount_mutex;
1799 unsigned int shrinker_run_no;
1801 /* For multi devices */
1802 int s_ndevs; /* number of devices */
1803 struct f2fs_dev_info *devs; /* for device list */
1804 unsigned int dirty_device; /* for checkpoint data flush */
1805 spinlock_t dev_lock; /* protect dirty_device */
1806 bool aligned_blksize; /* all devices has the same logical blksize */
1808 /* For write statistics */
1809 u64 sectors_written_start;
1812 /* Reference to checksum algorithm driver via cryptoapi */
1813 struct crypto_shash *s_chksum_driver;
1815 /* Precomputed FS UUID checksum for seeding other checksums */
1816 __u32 s_chksum_seed;
1818 struct workqueue_struct *post_read_wq; /* post read workqueue */
1820 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
1821 spinlock_t error_lock; /* protect errors array */
1822 bool error_dirty; /* errors of sb is dirty */
1824 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1825 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1827 /* For reclaimed segs statistics per each GC mode */
1828 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1829 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1831 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1833 int max_fragment_chunk; /* max chunk size for block fragmentation mode */
1834 int max_fragment_hole; /* max hole size for block fragmentation mode */
1836 /* For atomic write statistics */
1837 atomic64_t current_atomic_write;
1838 s64 peak_atomic_write;
1839 u64 committed_atomic_block;
1840 u64 revoked_atomic_block;
1842 #ifdef CONFIG_F2FS_FS_COMPRESSION
1843 struct kmem_cache *page_array_slab; /* page array entry */
1844 unsigned int page_array_slab_size; /* default page array slab size */
1846 /* For runtime compression statistics */
1847 u64 compr_written_block;
1848 u64 compr_saved_block;
1849 u32 compr_new_inode;
1851 /* For compressed block cache */
1852 struct inode *compress_inode; /* cache compressed blocks */
1853 unsigned int compress_percent; /* cache page percentage */
1854 unsigned int compress_watermark; /* cache page watermark */
1855 atomic_t compress_page_hit; /* cache hit count */
1858 #ifdef CONFIG_F2FS_IOSTAT
1859 /* For app/fs IO statistics */
1860 spinlock_t iostat_lock;
1861 unsigned long long rw_iostat[NR_IO_TYPE];
1862 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1864 unsigned long iostat_next_period;
1865 unsigned int iostat_period_ms;
1867 /* For io latency related statistics info in one iostat period */
1868 spinlock_t iostat_lat_lock;
1869 struct iostat_lat_info *iostat_io_lat;
1873 #ifdef CONFIG_F2FS_FAULT_INJECTION
1874 #define f2fs_show_injection_info(sbi, type) \
1875 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1876 KERN_INFO, sbi->sb->s_id, \
1877 f2fs_fault_name[type], \
1878 __func__, __builtin_return_address(0))
1879 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1881 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1883 if (!ffi->inject_rate)
1886 if (!IS_FAULT_SET(ffi, type))
1889 atomic_inc(&ffi->inject_ops);
1890 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1891 atomic_set(&ffi->inject_ops, 0);
1897 #define f2fs_show_injection_info(sbi, type) do { } while (0)
1898 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1905 * Test if the mounted volume is a multi-device volume.
1906 * - For a single regular disk volume, sbi->s_ndevs is 0.
1907 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1908 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1910 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1912 return sbi->s_ndevs > 1;
1915 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1917 unsigned long now = jiffies;
1919 sbi->last_time[type] = now;
1921 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1922 if (type == REQ_TIME) {
1923 sbi->last_time[DISCARD_TIME] = now;
1924 sbi->last_time[GC_TIME] = now;
1928 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1930 unsigned long interval = sbi->interval_time[type] * HZ;
1932 return time_after(jiffies, sbi->last_time[type] + interval);
1935 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1938 unsigned long interval = sbi->interval_time[type] * HZ;
1939 unsigned int wait_ms = 0;
1942 delta = (sbi->last_time[type] + interval) - jiffies;
1944 wait_ms = jiffies_to_msecs(delta);
1952 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1953 const void *address, unsigned int length)
1956 struct shash_desc shash;
1961 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1963 desc.shash.tfm = sbi->s_chksum_driver;
1964 *(u32 *)desc.ctx = crc;
1966 err = crypto_shash_update(&desc.shash, address, length);
1969 return *(u32 *)desc.ctx;
1972 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1973 unsigned int length)
1975 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1978 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1979 void *buf, size_t buf_size)
1981 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1984 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1985 const void *address, unsigned int length)
1987 return __f2fs_crc32(sbi, crc, address, length);
1990 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1992 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1995 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1997 return sb->s_fs_info;
2000 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
2002 return F2FS_SB(inode->i_sb);
2005 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
2007 return F2FS_I_SB(mapping->host);
2010 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
2012 return F2FS_M_SB(page_file_mapping(page));
2015 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
2017 return (struct f2fs_super_block *)(sbi->raw_super);
2020 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
2022 return (struct f2fs_checkpoint *)(sbi->ckpt);
2025 static inline struct f2fs_node *F2FS_NODE(struct page *page)
2027 return (struct f2fs_node *)page_address(page);
2030 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
2032 return &((struct f2fs_node *)page_address(page))->i;
2035 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
2037 return (struct f2fs_nm_info *)(sbi->nm_info);
2040 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
2042 return (struct f2fs_sm_info *)(sbi->sm_info);
2045 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
2047 return (struct sit_info *)(SM_I(sbi)->sit_info);
2050 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
2052 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
2055 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
2057 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
2060 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
2062 return sbi->meta_inode->i_mapping;
2065 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
2067 return sbi->node_inode->i_mapping;
2070 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
2072 return test_bit(type, &sbi->s_flag);
2075 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2077 set_bit(type, &sbi->s_flag);
2080 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2082 clear_bit(type, &sbi->s_flag);
2085 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
2087 return le64_to_cpu(cp->checkpoint_ver);
2090 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
2092 if (type < F2FS_MAX_QUOTAS)
2093 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
2097 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2099 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2100 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2103 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2105 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2107 return ckpt_flags & f;
2110 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2112 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2115 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2117 unsigned int ckpt_flags;
2119 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2121 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2124 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2126 unsigned long flags;
2128 spin_lock_irqsave(&sbi->cp_lock, flags);
2129 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2130 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2133 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2135 unsigned int ckpt_flags;
2137 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2139 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2142 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2144 unsigned long flags;
2146 spin_lock_irqsave(&sbi->cp_lock, flags);
2147 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2148 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2151 #define init_f2fs_rwsem(sem) \
2153 static struct lock_class_key __key; \
2155 __init_f2fs_rwsem((sem), #sem, &__key); \
2158 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,
2159 const char *sem_name, struct lock_class_key *key)
2161 __init_rwsem(&sem->internal_rwsem, sem_name, key);
2162 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2163 init_waitqueue_head(&sem->read_waiters);
2167 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)
2169 return rwsem_is_locked(&sem->internal_rwsem);
2172 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)
2174 return rwsem_is_contended(&sem->internal_rwsem);
2177 static inline void f2fs_down_read(struct f2fs_rwsem *sem)
2179 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2180 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));
2182 down_read(&sem->internal_rwsem);
2186 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)
2188 return down_read_trylock(&sem->internal_rwsem);
2191 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2192 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)
2194 down_read_nested(&sem->internal_rwsem, subclass);
2197 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)
2200 static inline void f2fs_up_read(struct f2fs_rwsem *sem)
2202 up_read(&sem->internal_rwsem);
2205 static inline void f2fs_down_write(struct f2fs_rwsem *sem)
2207 down_write(&sem->internal_rwsem);
2210 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)
2212 return down_write_trylock(&sem->internal_rwsem);
2215 static inline void f2fs_up_write(struct f2fs_rwsem *sem)
2217 up_write(&sem->internal_rwsem);
2218 #ifdef CONFIG_F2FS_UNFAIR_RWSEM
2219 wake_up_all(&sem->read_waiters);
2223 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2225 f2fs_down_read(&sbi->cp_rwsem);
2228 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2230 if (time_to_inject(sbi, FAULT_LOCK_OP)) {
2231 f2fs_show_injection_info(sbi, FAULT_LOCK_OP);
2234 return f2fs_down_read_trylock(&sbi->cp_rwsem);
2237 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2239 f2fs_up_read(&sbi->cp_rwsem);
2242 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2244 f2fs_down_write(&sbi->cp_rwsem);
2247 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2249 f2fs_up_write(&sbi->cp_rwsem);
2252 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2254 int reason = CP_SYNC;
2256 if (test_opt(sbi, FASTBOOT))
2257 reason = CP_FASTBOOT;
2258 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2263 static inline bool __remain_node_summaries(int reason)
2265 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2268 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2270 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2271 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2275 * Check whether the inode has blocks or not
2277 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2279 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2281 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2284 static inline bool f2fs_has_xattr_block(unsigned int ofs)
2286 return ofs == XATTR_NODE_OFFSET;
2289 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2290 struct inode *inode, bool cap)
2294 if (!test_opt(sbi, RESERVE_ROOT))
2296 if (IS_NOQUOTA(inode))
2298 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2300 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2301 in_group_p(F2FS_OPTION(sbi).s_resgid))
2303 if (cap && capable(CAP_SYS_RESOURCE))
2308 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2309 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2310 struct inode *inode, blkcnt_t *count)
2312 blkcnt_t diff = 0, release = 0;
2313 block_t avail_user_block_count;
2316 ret = dquot_reserve_block(inode, *count);
2320 if (time_to_inject(sbi, FAULT_BLOCK)) {
2321 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2327 * let's increase this in prior to actual block count change in order
2328 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2330 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2332 spin_lock(&sbi->stat_lock);
2333 sbi->total_valid_block_count += (block_t)(*count);
2334 avail_user_block_count = sbi->user_block_count -
2335 sbi->current_reserved_blocks;
2337 if (!__allow_reserved_blocks(sbi, inode, true))
2338 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2340 if (F2FS_IO_ALIGNED(sbi))
2341 avail_user_block_count -= sbi->blocks_per_seg *
2342 SM_I(sbi)->additional_reserved_segments;
2344 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2345 if (avail_user_block_count > sbi->unusable_block_count)
2346 avail_user_block_count -= sbi->unusable_block_count;
2348 avail_user_block_count = 0;
2350 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2351 diff = sbi->total_valid_block_count - avail_user_block_count;
2356 sbi->total_valid_block_count -= diff;
2358 spin_unlock(&sbi->stat_lock);
2362 spin_unlock(&sbi->stat_lock);
2364 if (unlikely(release)) {
2365 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2366 dquot_release_reservation_block(inode, release);
2368 f2fs_i_blocks_write(inode, *count, true, true);
2372 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2374 dquot_release_reservation_block(inode, release);
2379 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2381 #define f2fs_err(sbi, fmt, ...) \
2382 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2383 #define f2fs_warn(sbi, fmt, ...) \
2384 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2385 #define f2fs_notice(sbi, fmt, ...) \
2386 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2387 #define f2fs_info(sbi, fmt, ...) \
2388 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2389 #define f2fs_debug(sbi, fmt, ...) \
2390 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2392 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2393 struct inode *inode,
2396 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2398 spin_lock(&sbi->stat_lock);
2399 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2400 sbi->total_valid_block_count -= (block_t)count;
2401 if (sbi->reserved_blocks &&
2402 sbi->current_reserved_blocks < sbi->reserved_blocks)
2403 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2404 sbi->current_reserved_blocks + count);
2405 spin_unlock(&sbi->stat_lock);
2406 if (unlikely(inode->i_blocks < sectors)) {
2407 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2409 (unsigned long long)inode->i_blocks,
2410 (unsigned long long)sectors);
2411 set_sbi_flag(sbi, SBI_NEED_FSCK);
2414 f2fs_i_blocks_write(inode, count, false, true);
2417 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2419 atomic_inc(&sbi->nr_pages[count_type]);
2421 if (count_type == F2FS_DIRTY_DENTS ||
2422 count_type == F2FS_DIRTY_NODES ||
2423 count_type == F2FS_DIRTY_META ||
2424 count_type == F2FS_DIRTY_QDATA ||
2425 count_type == F2FS_DIRTY_IMETA)
2426 set_sbi_flag(sbi, SBI_IS_DIRTY);
2429 static inline void inode_inc_dirty_pages(struct inode *inode)
2431 atomic_inc(&F2FS_I(inode)->dirty_pages);
2432 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2433 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2434 if (IS_NOQUOTA(inode))
2435 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2438 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2440 atomic_dec(&sbi->nr_pages[count_type]);
2443 static inline void inode_dec_dirty_pages(struct inode *inode)
2445 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2446 !S_ISLNK(inode->i_mode))
2449 atomic_dec(&F2FS_I(inode)->dirty_pages);
2450 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2451 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2452 if (IS_NOQUOTA(inode))
2453 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2456 static inline void inc_atomic_write_cnt(struct inode *inode)
2458 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2459 struct f2fs_inode_info *fi = F2FS_I(inode);
2462 fi->atomic_write_cnt++;
2463 atomic64_inc(&sbi->current_atomic_write);
2464 current_write = atomic64_read(&sbi->current_atomic_write);
2465 if (current_write > sbi->peak_atomic_write)
2466 sbi->peak_atomic_write = current_write;
2469 static inline void release_atomic_write_cnt(struct inode *inode)
2471 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2472 struct f2fs_inode_info *fi = F2FS_I(inode);
2474 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write);
2475 fi->atomic_write_cnt = 0;
2478 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2480 return atomic_read(&sbi->nr_pages[count_type]);
2483 static inline int get_dirty_pages(struct inode *inode)
2485 return atomic_read(&F2FS_I(inode)->dirty_pages);
2488 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2490 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2491 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2492 sbi->log_blocks_per_seg;
2494 return segs / sbi->segs_per_sec;
2497 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2499 return sbi->total_valid_block_count;
2502 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2504 return sbi->discard_blks;
2507 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2509 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2511 /* return NAT or SIT bitmap */
2512 if (flag == NAT_BITMAP)
2513 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2514 else if (flag == SIT_BITMAP)
2515 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2520 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2522 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2525 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2527 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2528 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2531 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2532 offset = (flag == SIT_BITMAP) ?
2533 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2535 * if large_nat_bitmap feature is enabled, leave checksum
2536 * protection for all nat/sit bitmaps.
2538 return tmp_ptr + offset + sizeof(__le32);
2541 if (__cp_payload(sbi) > 0) {
2542 if (flag == NAT_BITMAP)
2545 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2547 offset = (flag == NAT_BITMAP) ?
2548 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2549 return tmp_ptr + offset;
2553 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2555 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2557 if (sbi->cur_cp_pack == 2)
2558 start_addr += sbi->blocks_per_seg;
2562 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2564 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2566 if (sbi->cur_cp_pack == 1)
2567 start_addr += sbi->blocks_per_seg;
2571 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2573 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2576 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2578 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2581 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2582 struct inode *inode, bool is_inode)
2584 block_t valid_block_count;
2585 unsigned int valid_node_count, user_block_count;
2590 err = dquot_alloc_inode(inode);
2595 err = dquot_reserve_block(inode, 1);
2600 if (time_to_inject(sbi, FAULT_BLOCK)) {
2601 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2605 spin_lock(&sbi->stat_lock);
2607 valid_block_count = sbi->total_valid_block_count +
2608 sbi->current_reserved_blocks + 1;
2610 if (!__allow_reserved_blocks(sbi, inode, false))
2611 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2613 if (F2FS_IO_ALIGNED(sbi))
2614 valid_block_count += sbi->blocks_per_seg *
2615 SM_I(sbi)->additional_reserved_segments;
2617 user_block_count = sbi->user_block_count;
2618 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2619 user_block_count -= sbi->unusable_block_count;
2621 if (unlikely(valid_block_count > user_block_count)) {
2622 spin_unlock(&sbi->stat_lock);
2626 valid_node_count = sbi->total_valid_node_count + 1;
2627 if (unlikely(valid_node_count > sbi->total_node_count)) {
2628 spin_unlock(&sbi->stat_lock);
2632 sbi->total_valid_node_count++;
2633 sbi->total_valid_block_count++;
2634 spin_unlock(&sbi->stat_lock);
2638 f2fs_mark_inode_dirty_sync(inode, true);
2640 f2fs_i_blocks_write(inode, 1, true, true);
2643 percpu_counter_inc(&sbi->alloc_valid_block_count);
2649 dquot_free_inode(inode);
2651 dquot_release_reservation_block(inode, 1);
2656 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2657 struct inode *inode, bool is_inode)
2659 spin_lock(&sbi->stat_lock);
2661 if (unlikely(!sbi->total_valid_block_count ||
2662 !sbi->total_valid_node_count)) {
2663 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2664 sbi->total_valid_block_count,
2665 sbi->total_valid_node_count);
2666 set_sbi_flag(sbi, SBI_NEED_FSCK);
2668 sbi->total_valid_block_count--;
2669 sbi->total_valid_node_count--;
2672 if (sbi->reserved_blocks &&
2673 sbi->current_reserved_blocks < sbi->reserved_blocks)
2674 sbi->current_reserved_blocks++;
2676 spin_unlock(&sbi->stat_lock);
2679 dquot_free_inode(inode);
2681 if (unlikely(inode->i_blocks == 0)) {
2682 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2684 (unsigned long long)inode->i_blocks);
2685 set_sbi_flag(sbi, SBI_NEED_FSCK);
2688 f2fs_i_blocks_write(inode, 1, false, true);
2692 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2694 return sbi->total_valid_node_count;
2697 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2699 percpu_counter_inc(&sbi->total_valid_inode_count);
2702 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2704 percpu_counter_dec(&sbi->total_valid_inode_count);
2707 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2709 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2712 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2713 pgoff_t index, bool for_write)
2718 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2720 page = find_get_page_flags(mapping, index,
2721 FGP_LOCK | FGP_ACCESSED);
2723 page = find_lock_page(mapping, index);
2727 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2728 f2fs_show_injection_info(F2FS_M_SB(mapping),
2735 return grab_cache_page(mapping, index);
2737 flags = memalloc_nofs_save();
2738 page = grab_cache_page_write_begin(mapping, index);
2739 memalloc_nofs_restore(flags);
2744 static inline struct page *f2fs_pagecache_get_page(
2745 struct address_space *mapping, pgoff_t index,
2746 int fgp_flags, gfp_t gfp_mask)
2748 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2749 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2753 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2756 static inline void f2fs_put_page(struct page *page, int unlock)
2762 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2768 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2771 f2fs_put_page(dn->node_page, 1);
2772 if (dn->inode_page && dn->node_page != dn->inode_page)
2773 f2fs_put_page(dn->inode_page, 0);
2774 dn->node_page = NULL;
2775 dn->inode_page = NULL;
2778 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2781 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2784 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
2789 entry = kmem_cache_alloc(cachep, flags);
2791 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2795 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2796 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
2799 return f2fs_kmem_cache_alloc_nofail(cachep, flags);
2801 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) {
2802 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC);
2806 return kmem_cache_alloc(cachep, flags);
2809 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2811 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2812 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2813 get_pages(sbi, F2FS_WB_CP_DATA) ||
2814 get_pages(sbi, F2FS_DIO_READ) ||
2815 get_pages(sbi, F2FS_DIO_WRITE))
2818 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2819 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2822 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2823 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2828 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2830 if (sbi->gc_mode == GC_URGENT_HIGH)
2833 if (is_inflight_io(sbi, type))
2836 if (sbi->gc_mode == GC_URGENT_MID)
2839 if (sbi->gc_mode == GC_URGENT_LOW &&
2840 (type == DISCARD_TIME || type == GC_TIME))
2843 return f2fs_time_over(sbi, type);
2846 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2847 unsigned long index, void *item)
2849 while (radix_tree_insert(root, index, item))
2853 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2855 static inline bool IS_INODE(struct page *page)
2857 struct f2fs_node *p = F2FS_NODE(page);
2859 return RAW_IS_INODE(p);
2862 static inline int offset_in_addr(struct f2fs_inode *i)
2864 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2865 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2868 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2870 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2873 static inline int f2fs_has_extra_attr(struct inode *inode);
2874 static inline block_t data_blkaddr(struct inode *inode,
2875 struct page *node_page, unsigned int offset)
2877 struct f2fs_node *raw_node;
2880 bool is_inode = IS_INODE(node_page);
2882 raw_node = F2FS_NODE(node_page);
2886 /* from GC path only */
2887 base = offset_in_addr(&raw_node->i);
2888 else if (f2fs_has_extra_attr(inode))
2889 base = get_extra_isize(inode);
2892 addr_array = blkaddr_in_node(raw_node);
2893 return le32_to_cpu(addr_array[base + offset]);
2896 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2898 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2901 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2906 mask = 1 << (7 - (nr & 0x07));
2907 return mask & *addr;
2910 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2915 mask = 1 << (7 - (nr & 0x07));
2919 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2924 mask = 1 << (7 - (nr & 0x07));
2928 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2934 mask = 1 << (7 - (nr & 0x07));
2940 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2946 mask = 1 << (7 - (nr & 0x07));
2952 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2957 mask = 1 << (7 - (nr & 0x07));
2962 * On-disk inode flags (f2fs_inode::i_flags)
2964 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2965 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2966 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2967 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2968 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2969 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2970 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2971 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2972 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2973 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2974 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2976 /* Flags that should be inherited by new inodes from their parent. */
2977 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2978 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2981 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2982 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2985 /* Flags that are appropriate for non-directories/regular files. */
2986 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2988 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2992 else if (S_ISREG(mode))
2993 return flags & F2FS_REG_FLMASK;
2995 return flags & F2FS_OTHER_FLMASK;
2998 static inline void __mark_inode_dirty_flag(struct inode *inode,
3002 case FI_INLINE_XATTR:
3003 case FI_INLINE_DATA:
3004 case FI_INLINE_DENTRY:
3010 case FI_INLINE_DOTS:
3012 case FI_COMPRESS_RELEASED:
3013 f2fs_mark_inode_dirty_sync(inode, true);
3017 static inline void set_inode_flag(struct inode *inode, int flag)
3019 set_bit(flag, F2FS_I(inode)->flags);
3020 __mark_inode_dirty_flag(inode, flag, true);
3023 static inline int is_inode_flag_set(struct inode *inode, int flag)
3025 return test_bit(flag, F2FS_I(inode)->flags);
3028 static inline void clear_inode_flag(struct inode *inode, int flag)
3030 clear_bit(flag, F2FS_I(inode)->flags);
3031 __mark_inode_dirty_flag(inode, flag, false);
3034 static inline bool f2fs_verity_in_progress(struct inode *inode)
3036 return IS_ENABLED(CONFIG_FS_VERITY) &&
3037 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
3040 static inline void set_acl_inode(struct inode *inode, umode_t mode)
3042 F2FS_I(inode)->i_acl_mode = mode;
3043 set_inode_flag(inode, FI_ACL_MODE);
3044 f2fs_mark_inode_dirty_sync(inode, false);
3047 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
3053 f2fs_mark_inode_dirty_sync(inode, true);
3056 static inline void f2fs_i_blocks_write(struct inode *inode,
3057 block_t diff, bool add, bool claim)
3059 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3060 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3062 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
3065 dquot_claim_block(inode, diff);
3067 dquot_alloc_block_nofail(inode, diff);
3069 dquot_free_block(inode, diff);
3072 f2fs_mark_inode_dirty_sync(inode, true);
3073 if (clean || recover)
3074 set_inode_flag(inode, FI_AUTO_RECOVER);
3077 static inline bool f2fs_is_atomic_file(struct inode *inode);
3079 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
3081 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3082 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3084 if (i_size_read(inode) == i_size)
3087 i_size_write(inode, i_size);
3089 if (f2fs_is_atomic_file(inode))
3092 f2fs_mark_inode_dirty_sync(inode, true);
3093 if (clean || recover)
3094 set_inode_flag(inode, FI_AUTO_RECOVER);
3097 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
3099 F2FS_I(inode)->i_current_depth = depth;
3100 f2fs_mark_inode_dirty_sync(inode, true);
3103 static inline void f2fs_i_gc_failures_write(struct inode *inode,
3106 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
3107 f2fs_mark_inode_dirty_sync(inode, true);
3110 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
3112 F2FS_I(inode)->i_xattr_nid = xnid;
3113 f2fs_mark_inode_dirty_sync(inode, true);
3116 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
3118 F2FS_I(inode)->i_pino = pino;
3119 f2fs_mark_inode_dirty_sync(inode, true);
3122 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
3124 struct f2fs_inode_info *fi = F2FS_I(inode);
3126 if (ri->i_inline & F2FS_INLINE_XATTR)
3127 set_bit(FI_INLINE_XATTR, fi->flags);
3128 if (ri->i_inline & F2FS_INLINE_DATA)
3129 set_bit(FI_INLINE_DATA, fi->flags);
3130 if (ri->i_inline & F2FS_INLINE_DENTRY)
3131 set_bit(FI_INLINE_DENTRY, fi->flags);
3132 if (ri->i_inline & F2FS_DATA_EXIST)
3133 set_bit(FI_DATA_EXIST, fi->flags);
3134 if (ri->i_inline & F2FS_INLINE_DOTS)
3135 set_bit(FI_INLINE_DOTS, fi->flags);
3136 if (ri->i_inline & F2FS_EXTRA_ATTR)
3137 set_bit(FI_EXTRA_ATTR, fi->flags);
3138 if (ri->i_inline & F2FS_PIN_FILE)
3139 set_bit(FI_PIN_FILE, fi->flags);
3140 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
3141 set_bit(FI_COMPRESS_RELEASED, fi->flags);
3144 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
3148 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
3149 ri->i_inline |= F2FS_INLINE_XATTR;
3150 if (is_inode_flag_set(inode, FI_INLINE_DATA))
3151 ri->i_inline |= F2FS_INLINE_DATA;
3152 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
3153 ri->i_inline |= F2FS_INLINE_DENTRY;
3154 if (is_inode_flag_set(inode, FI_DATA_EXIST))
3155 ri->i_inline |= F2FS_DATA_EXIST;
3156 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
3157 ri->i_inline |= F2FS_INLINE_DOTS;
3158 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
3159 ri->i_inline |= F2FS_EXTRA_ATTR;
3160 if (is_inode_flag_set(inode, FI_PIN_FILE))
3161 ri->i_inline |= F2FS_PIN_FILE;
3162 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
3163 ri->i_inline |= F2FS_COMPRESS_RELEASED;
3166 static inline int f2fs_has_extra_attr(struct inode *inode)
3168 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
3171 static inline int f2fs_has_inline_xattr(struct inode *inode)
3173 return is_inode_flag_set(inode, FI_INLINE_XATTR);
3176 static inline int f2fs_compressed_file(struct inode *inode)
3178 return S_ISREG(inode->i_mode) &&
3179 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
3182 static inline bool f2fs_need_compress_data(struct inode *inode)
3184 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
3186 if (!f2fs_compressed_file(inode))
3189 if (compress_mode == COMPR_MODE_FS)
3191 else if (compress_mode == COMPR_MODE_USER &&
3192 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
3198 static inline unsigned int addrs_per_inode(struct inode *inode)
3200 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3201 get_inline_xattr_addrs(inode);
3203 if (!f2fs_compressed_file(inode))
3205 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3208 static inline unsigned int addrs_per_block(struct inode *inode)
3210 if (!f2fs_compressed_file(inode))
3211 return DEF_ADDRS_PER_BLOCK;
3212 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3215 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3217 struct f2fs_inode *ri = F2FS_INODE(page);
3219 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3220 get_inline_xattr_addrs(inode)]);
3223 static inline int inline_xattr_size(struct inode *inode)
3225 if (f2fs_has_inline_xattr(inode))
3226 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3231 * Notice: check inline_data flag without inode page lock is unsafe.
3232 * It could change at any time by f2fs_convert_inline_page().
3234 static inline int f2fs_has_inline_data(struct inode *inode)
3236 return is_inode_flag_set(inode, FI_INLINE_DATA);
3239 static inline int f2fs_exist_data(struct inode *inode)
3241 return is_inode_flag_set(inode, FI_DATA_EXIST);
3244 static inline int f2fs_has_inline_dots(struct inode *inode)
3246 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3249 static inline int f2fs_is_mmap_file(struct inode *inode)
3251 return is_inode_flag_set(inode, FI_MMAP_FILE);
3254 static inline bool f2fs_is_pinned_file(struct inode *inode)
3256 return is_inode_flag_set(inode, FI_PIN_FILE);
3259 static inline bool f2fs_is_atomic_file(struct inode *inode)
3261 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3264 static inline bool f2fs_is_cow_file(struct inode *inode)
3266 return is_inode_flag_set(inode, FI_COW_FILE);
3269 static inline bool f2fs_is_first_block_written(struct inode *inode)
3271 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
3274 static inline bool f2fs_is_drop_cache(struct inode *inode)
3276 return is_inode_flag_set(inode, FI_DROP_CACHE);
3279 static inline void *inline_data_addr(struct inode *inode, struct page *page)
3281 struct f2fs_inode *ri = F2FS_INODE(page);
3282 int extra_size = get_extra_isize(inode);
3284 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
3287 static inline int f2fs_has_inline_dentry(struct inode *inode)
3289 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3292 static inline int is_file(struct inode *inode, int type)
3294 return F2FS_I(inode)->i_advise & type;
3297 static inline void set_file(struct inode *inode, int type)
3299 if (is_file(inode, type))
3301 F2FS_I(inode)->i_advise |= type;
3302 f2fs_mark_inode_dirty_sync(inode, true);
3305 static inline void clear_file(struct inode *inode, int type)
3307 if (!is_file(inode, type))
3309 F2FS_I(inode)->i_advise &= ~type;
3310 f2fs_mark_inode_dirty_sync(inode, true);
3313 static inline bool f2fs_is_time_consistent(struct inode *inode)
3315 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
3317 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
3319 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
3321 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
3322 &F2FS_I(inode)->i_crtime))
3327 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3332 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3334 spin_lock(&sbi->inode_lock[DIRTY_META]);
3335 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3336 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3339 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3340 file_keep_isize(inode) ||
3341 i_size_read(inode) & ~PAGE_MASK)
3344 if (!f2fs_is_time_consistent(inode))
3347 spin_lock(&F2FS_I(inode)->i_size_lock);
3348 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3349 spin_unlock(&F2FS_I(inode)->i_size_lock);
3354 static inline bool f2fs_readonly(struct super_block *sb)
3356 return sb_rdonly(sb);
3359 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3361 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3364 static inline bool is_dot_dotdot(const u8 *name, size_t len)
3366 if (len == 1 && name[0] == '.')
3369 if (len == 2 && name[0] == '.' && name[1] == '.')
3375 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3376 size_t size, gfp_t flags)
3378 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3379 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3383 return kmalloc(size, flags);
3386 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3387 size_t size, gfp_t flags)
3389 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3392 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3393 size_t size, gfp_t flags)
3395 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3396 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3400 return kvmalloc(size, flags);
3403 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3404 size_t size, gfp_t flags)
3406 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3409 static inline int get_extra_isize(struct inode *inode)
3411 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3414 static inline int get_inline_xattr_addrs(struct inode *inode)
3416 return F2FS_I(inode)->i_inline_xattr_size;
3419 #define f2fs_get_inode_mode(i) \
3420 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3421 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3423 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3424 (offsetof(struct f2fs_inode, i_extra_end) - \
3425 offsetof(struct f2fs_inode, i_extra_isize)) \
3427 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3428 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3429 ((offsetof(typeof(*(f2fs_inode)), field) + \
3430 sizeof((f2fs_inode)->field)) \
3431 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3433 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3435 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3437 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3438 block_t blkaddr, int type);
3439 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3440 block_t blkaddr, int type)
3442 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3443 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3445 f2fs_bug_on(sbi, 1);
3449 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3451 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3452 blkaddr == COMPRESS_ADDR)
3460 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3461 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3462 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3463 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3464 int f2fs_truncate(struct inode *inode);
3465 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path,
3466 struct kstat *stat, u32 request_mask, unsigned int flags);
3467 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
3468 struct iattr *attr);
3469 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3470 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3471 int f2fs_precache_extents(struct inode *inode);
3472 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3473 int f2fs_fileattr_set(struct user_namespace *mnt_userns,
3474 struct dentry *dentry, struct fileattr *fa);
3475 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3476 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3477 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3478 int f2fs_pin_file_control(struct inode *inode, bool inc);
3483 void f2fs_set_inode_flags(struct inode *inode);
3484 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3485 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3486 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3487 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3488 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3489 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3490 void f2fs_update_inode_page(struct inode *inode);
3491 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3492 void f2fs_evict_inode(struct inode *inode);
3493 void f2fs_handle_failed_inode(struct inode *inode);
3498 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3499 bool hot, bool set);
3500 struct dentry *f2fs_get_parent(struct dentry *child);
3501 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir,
3502 struct inode **new_inode);
3507 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3508 int f2fs_init_casefolded_name(const struct inode *dir,
3509 struct f2fs_filename *fname);
3510 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3511 int lookup, struct f2fs_filename *fname);
3512 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3513 struct f2fs_filename *fname);
3514 void f2fs_free_filename(struct f2fs_filename *fname);
3515 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3516 const struct f2fs_filename *fname, int *max_slots);
3517 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3518 unsigned int start_pos, struct fscrypt_str *fstr);
3519 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3520 struct f2fs_dentry_ptr *d);
3521 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3522 const struct f2fs_filename *fname, struct page *dpage);
3523 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3524 unsigned int current_depth);
3525 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3526 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3527 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3528 const struct f2fs_filename *fname,
3529 struct page **res_page);
3530 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3531 const struct qstr *child, struct page **res_page);
3532 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3533 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3534 struct page **page);
3535 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3536 struct page *page, struct inode *inode);
3537 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3538 const struct f2fs_filename *fname);
3539 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3540 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3541 unsigned int bit_pos);
3542 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3543 struct inode *inode, nid_t ino, umode_t mode);
3544 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3545 struct inode *inode, nid_t ino, umode_t mode);
3546 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3547 struct inode *inode, nid_t ino, umode_t mode);
3548 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3549 struct inode *dir, struct inode *inode);
3550 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3551 bool f2fs_empty_dir(struct inode *dir);
3553 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3555 if (fscrypt_is_nokey_name(dentry))
3557 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3558 inode, inode->i_ino, inode->i_mode);
3564 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3565 void f2fs_inode_synced(struct inode *inode);
3566 int f2fs_dquot_initialize(struct inode *inode);
3567 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3568 int f2fs_quota_sync(struct super_block *sb, int type);
3569 loff_t max_file_blocks(struct inode *inode);
3570 void f2fs_quota_off_umount(struct super_block *sb);
3571 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason);
3572 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
3573 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3574 int f2fs_sync_fs(struct super_block *sb, int sync);
3575 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3580 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3587 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3588 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3589 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3590 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3591 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3592 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3593 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3594 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3595 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3596 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3597 struct node_info *ni, bool checkpoint_context);
3598 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3599 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3600 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3601 int f2fs_truncate_xattr_node(struct inode *inode);
3602 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3603 unsigned int seq_id);
3604 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
3605 int f2fs_remove_inode_page(struct inode *inode);
3606 struct page *f2fs_new_inode_page(struct inode *inode);
3607 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3608 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3609 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3610 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3611 int f2fs_move_node_page(struct page *node_page, int gc_type);
3612 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3613 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3614 struct writeback_control *wbc, bool atomic,
3615 unsigned int *seq_id);
3616 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3617 struct writeback_control *wbc,
3618 bool do_balance, enum iostat_type io_type);
3619 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3620 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3621 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3622 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3623 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3624 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3625 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3626 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3627 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3628 unsigned int segno, struct f2fs_summary_block *sum);
3629 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
3630 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3631 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3632 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3633 int __init f2fs_create_node_manager_caches(void);
3634 void f2fs_destroy_node_manager_caches(void);
3639 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3640 int f2fs_commit_atomic_write(struct inode *inode);
3641 void f2fs_abort_atomic_write(struct inode *inode, bool clean);
3642 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3643 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3644 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3645 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3646 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3647 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3648 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3649 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3650 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
3651 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3652 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3653 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3654 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3655 struct cp_control *cpc);
3656 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3657 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3658 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3659 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3660 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3661 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3662 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3663 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3664 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3665 void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3666 unsigned int *newseg, bool new_sec, int dir);
3667 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3668 unsigned int start, unsigned int end);
3669 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3670 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3671 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3672 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3673 struct cp_control *cpc);
3674 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3675 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3677 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3678 enum iostat_type io_type);
3679 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3680 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3681 struct f2fs_io_info *fio);
3682 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3683 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3684 block_t old_blkaddr, block_t new_blkaddr,
3685 bool recover_curseg, bool recover_newaddr,
3687 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3688 block_t old_addr, block_t new_addr,
3689 unsigned char version, bool recover_curseg,
3690 bool recover_newaddr);
3691 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3692 block_t old_blkaddr, block_t *new_blkaddr,
3693 struct f2fs_summary *sum, int type,
3694 struct f2fs_io_info *fio);
3695 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,
3696 block_t blkaddr, unsigned int blkcnt);
3697 void f2fs_wait_on_page_writeback(struct page *page,
3698 enum page_type type, bool ordered, bool locked);
3699 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3700 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3702 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3703 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3704 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3705 unsigned int val, int alloc);
3706 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3707 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3708 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3709 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3710 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3711 int __init f2fs_create_segment_manager_caches(void);
3712 void f2fs_destroy_segment_manager_caches(void);
3713 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3714 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3715 unsigned int segno);
3716 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3717 unsigned int segno);
3719 #define DEF_FRAGMENT_SIZE 4
3720 #define MIN_FRAGMENT_SIZE 1
3721 #define MAX_FRAGMENT_SIZE 512
3723 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)
3725 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||
3726 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;
3732 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
3733 unsigned char reason);
3734 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);
3735 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3736 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3737 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3738 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3739 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3740 block_t blkaddr, int type);
3741 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3742 int type, bool sync);
3743 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,
3744 unsigned int ra_blocks);
3745 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3746 long nr_to_write, enum iostat_type io_type);
3747 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3748 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3749 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3750 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3751 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3752 unsigned int devidx, int type);
3753 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3754 unsigned int devidx, int type);
3755 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3756 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3757 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3758 void f2fs_add_orphan_inode(struct inode *inode);
3759 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3760 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3761 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3762 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);
3763 void f2fs_remove_dirty_inode(struct inode *inode);
3764 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3766 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3767 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3768 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3769 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3770 int __init f2fs_create_checkpoint_caches(void);
3771 void f2fs_destroy_checkpoint_caches(void);
3772 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3773 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3774 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3775 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3780 int __init f2fs_init_bioset(void);
3781 void f2fs_destroy_bioset(void);
3782 int f2fs_init_bio_entry_cache(void);
3783 void f2fs_destroy_bio_entry_cache(void);
3784 void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3785 struct bio *bio, enum page_type type);
3786 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);
3787 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3788 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3789 struct inode *inode, struct page *page,
3790 nid_t ino, enum page_type type);
3791 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3792 struct bio **bio, struct page *page);
3793 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3794 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3795 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3796 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3797 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3798 block_t blk_addr, sector_t *sector);
3799 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3800 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3801 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3802 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3803 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3804 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3805 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3806 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3807 blk_opf_t op_flags, bool for_write);
3808 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3809 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3811 struct page *f2fs_get_new_data_page(struct inode *inode,
3812 struct page *ipage, pgoff_t index, bool new_i_size);
3813 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3814 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3815 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3816 int create, int flag);
3817 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3818 u64 start, u64 len);
3819 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3820 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3821 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3822 int f2fs_write_single_data_page(struct page *page, int *submitted,
3823 struct bio **bio, sector_t *last_block,
3824 struct writeback_control *wbc,
3825 enum iostat_type io_type,
3826 int compr_blocks, bool allow_balance);
3827 void f2fs_write_failed(struct inode *inode, loff_t to);
3828 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);
3829 bool f2fs_release_folio(struct folio *folio, gfp_t wait);
3830 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3831 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3832 int f2fs_init_post_read_processing(void);
3833 void f2fs_destroy_post_read_processing(void);
3834 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3835 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3836 extern const struct iomap_ops f2fs_iomap_ops;
3841 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3842 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3843 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3844 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);
3845 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3846 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3847 int __init f2fs_create_garbage_collection_cache(void);
3848 void f2fs_destroy_garbage_collection_cache(void);
3853 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3854 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3855 int __init f2fs_create_recovery_cache(void);
3856 void f2fs_destroy_recovery_cache(void);
3861 #ifdef CONFIG_F2FS_STAT_FS
3862 struct f2fs_stat_info {
3863 struct list_head stat_list;
3864 struct f2fs_sb_info *sbi;
3865 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3866 int main_area_segs, main_area_sections, main_area_zones;
3867 unsigned long long hit_largest, hit_cached, hit_rbtree;
3868 unsigned long long hit_total, total_ext;
3869 int ext_tree, zombie_tree, ext_node;
3870 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3871 int ndirty_data, ndirty_qdata;
3872 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3873 int nats, dirty_nats, sits, dirty_sits;
3874 int free_nids, avail_nids, alloc_nids;
3875 int total_count, utilization;
3876 int bg_gc, nr_wb_cp_data, nr_wb_data;
3877 int nr_rd_data, nr_rd_node, nr_rd_meta;
3878 int nr_dio_read, nr_dio_write;
3879 unsigned int io_skip_bggc, other_skip_bggc;
3880 int nr_flushing, nr_flushed, flush_list_empty;
3881 int nr_discarding, nr_discarded;
3883 unsigned int undiscard_blks;
3884 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3885 unsigned int cur_ckpt_time, peak_ckpt_time;
3886 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3887 int compr_inode, swapfile_inode;
3888 unsigned long long compr_blocks;
3889 int aw_cnt, max_aw_cnt;
3890 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3891 unsigned int bimodal, avg_vblocks;
3892 int util_free, util_valid, util_invalid;
3893 int rsvd_segs, overp_segs;
3894 int dirty_count, node_pages, meta_pages, compress_pages;
3895 int compress_page_hit;
3896 int prefree_count, call_count, cp_count, bg_cp_count;
3897 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3898 int bg_node_segs, bg_data_segs;
3899 int tot_blks, data_blks, node_blks;
3900 int bg_data_blks, bg_node_blks;
3901 int curseg[NR_CURSEG_TYPE];
3902 int cursec[NR_CURSEG_TYPE];
3903 int curzone[NR_CURSEG_TYPE];
3904 unsigned int dirty_seg[NR_CURSEG_TYPE];
3905 unsigned int full_seg[NR_CURSEG_TYPE];
3906 unsigned int valid_blks[NR_CURSEG_TYPE];
3908 unsigned int meta_count[META_MAX];
3909 unsigned int segment_count[2];
3910 unsigned int block_count[2];
3911 unsigned int inplace_count;
3912 unsigned long long base_mem, cache_mem, page_mem;
3915 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3917 return (struct f2fs_stat_info *)sbi->stat_info;
3920 #define stat_inc_cp_count(si) ((si)->cp_count++)
3921 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3922 #define stat_inc_call_count(si) ((si)->call_count++)
3923 #define stat_inc_bggc_count(si) ((si)->bg_gc++)
3924 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3925 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3926 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3927 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3928 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3929 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3930 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3931 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3932 #define stat_inc_inline_xattr(inode) \
3934 if (f2fs_has_inline_xattr(inode)) \
3935 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3937 #define stat_dec_inline_xattr(inode) \
3939 if (f2fs_has_inline_xattr(inode)) \
3940 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3942 #define stat_inc_inline_inode(inode) \
3944 if (f2fs_has_inline_data(inode)) \
3945 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3947 #define stat_dec_inline_inode(inode) \
3949 if (f2fs_has_inline_data(inode)) \
3950 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3952 #define stat_inc_inline_dir(inode) \
3954 if (f2fs_has_inline_dentry(inode)) \
3955 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3957 #define stat_dec_inline_dir(inode) \
3959 if (f2fs_has_inline_dentry(inode)) \
3960 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3962 #define stat_inc_compr_inode(inode) \
3964 if (f2fs_compressed_file(inode)) \
3965 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3967 #define stat_dec_compr_inode(inode) \
3969 if (f2fs_compressed_file(inode)) \
3970 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3972 #define stat_add_compr_blocks(inode, blocks) \
3973 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3974 #define stat_sub_compr_blocks(inode, blocks) \
3975 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3976 #define stat_inc_swapfile_inode(inode) \
3977 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))
3978 #define stat_dec_swapfile_inode(inode) \
3979 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))
3980 #define stat_inc_atomic_inode(inode) \
3981 (atomic_inc(&F2FS_I_SB(inode)->atomic_files))
3982 #define stat_dec_atomic_inode(inode) \
3983 (atomic_dec(&F2FS_I_SB(inode)->atomic_files))
3984 #define stat_inc_meta_count(sbi, blkaddr) \
3986 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3987 atomic_inc(&(sbi)->meta_count[META_CP]); \
3988 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3989 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3990 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3991 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3992 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3993 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3995 #define stat_inc_seg_type(sbi, curseg) \
3996 ((sbi)->segment_count[(curseg)->alloc_type]++)
3997 #define stat_inc_block_count(sbi, curseg) \
3998 ((sbi)->block_count[(curseg)->alloc_type]++)
3999 #define stat_inc_inplace_blocks(sbi) \
4000 (atomic_inc(&(sbi)->inplace_count))
4001 #define stat_update_max_atomic_write(inode) \
4003 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \
4004 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
4006 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
4008 #define stat_inc_seg_count(sbi, type, gc_type) \
4010 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4012 if ((type) == SUM_TYPE_DATA) { \
4014 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
4017 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
4021 #define stat_inc_tot_blk_count(si, blks) \
4022 ((si)->tot_blks += (blks))
4024 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
4026 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4027 stat_inc_tot_blk_count(si, blks); \
4028 si->data_blks += (blks); \
4029 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4032 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
4034 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4035 stat_inc_tot_blk_count(si, blks); \
4036 si->node_blks += (blks); \
4037 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4040 int f2fs_build_stats(struct f2fs_sb_info *sbi);
4041 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
4042 void __init f2fs_create_root_stats(void);
4043 void f2fs_destroy_root_stats(void);
4044 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
4046 #define stat_inc_cp_count(si) do { } while (0)
4047 #define stat_inc_bg_cp_count(si) do { } while (0)
4048 #define stat_inc_call_count(si) do { } while (0)
4049 #define stat_inc_bggc_count(si) do { } while (0)
4050 #define stat_io_skip_bggc_count(sbi) do { } while (0)
4051 #define stat_other_skip_bggc_count(sbi) do { } while (0)
4052 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
4053 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
4054 #define stat_inc_total_hit(sbi) do { } while (0)
4055 #define stat_inc_rbtree_node_hit(sbi) do { } while (0)
4056 #define stat_inc_largest_node_hit(sbi) do { } while (0)
4057 #define stat_inc_cached_node_hit(sbi) do { } while (0)
4058 #define stat_inc_inline_xattr(inode) do { } while (0)
4059 #define stat_dec_inline_xattr(inode) do { } while (0)
4060 #define stat_inc_inline_inode(inode) do { } while (0)
4061 #define stat_dec_inline_inode(inode) do { } while (0)
4062 #define stat_inc_inline_dir(inode) do { } while (0)
4063 #define stat_dec_inline_dir(inode) do { } while (0)
4064 #define stat_inc_compr_inode(inode) do { } while (0)
4065 #define stat_dec_compr_inode(inode) do { } while (0)
4066 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
4067 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
4068 #define stat_inc_swapfile_inode(inode) do { } while (0)
4069 #define stat_dec_swapfile_inode(inode) do { } while (0)
4070 #define stat_inc_atomic_inode(inode) do { } while (0)
4071 #define stat_dec_atomic_inode(inode) do { } while (0)
4072 #define stat_update_max_atomic_write(inode) do { } while (0)
4073 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
4074 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
4075 #define stat_inc_block_count(sbi, curseg) do { } while (0)
4076 #define stat_inc_inplace_blocks(sbi) do { } while (0)
4077 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
4078 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
4079 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
4080 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
4082 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
4083 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
4084 static inline void __init f2fs_create_root_stats(void) { }
4085 static inline void f2fs_destroy_root_stats(void) { }
4086 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
4089 extern const struct file_operations f2fs_dir_operations;
4090 extern const struct file_operations f2fs_file_operations;
4091 extern const struct inode_operations f2fs_file_inode_operations;
4092 extern const struct address_space_operations f2fs_dblock_aops;
4093 extern const struct address_space_operations f2fs_node_aops;
4094 extern const struct address_space_operations f2fs_meta_aops;
4095 extern const struct inode_operations f2fs_dir_inode_operations;
4096 extern const struct inode_operations f2fs_symlink_inode_operations;
4097 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
4098 extern const struct inode_operations f2fs_special_inode_operations;
4099 extern struct kmem_cache *f2fs_inode_entry_slab;
4104 bool f2fs_may_inline_data(struct inode *inode);
4105 bool f2fs_sanity_check_inline_data(struct inode *inode);
4106 bool f2fs_may_inline_dentry(struct inode *inode);
4107 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
4108 void f2fs_truncate_inline_inode(struct inode *inode,
4109 struct page *ipage, u64 from);
4110 int f2fs_read_inline_data(struct inode *inode, struct page *page);
4111 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
4112 int f2fs_convert_inline_inode(struct inode *inode);
4113 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
4114 int f2fs_write_inline_data(struct inode *inode, struct page *page);
4115 int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
4116 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
4117 const struct f2fs_filename *fname,
4118 struct page **res_page);
4119 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
4120 struct page *ipage);
4121 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
4122 struct inode *inode, nid_t ino, umode_t mode);
4123 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
4124 struct page *page, struct inode *dir,
4125 struct inode *inode);
4126 bool f2fs_empty_inline_dir(struct inode *dir);
4127 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
4128 struct fscrypt_str *fstr);
4129 int f2fs_inline_data_fiemap(struct inode *inode,
4130 struct fiemap_extent_info *fieinfo,
4131 __u64 start, __u64 len);
4136 unsigned long f2fs_shrink_count(struct shrinker *shrink,
4137 struct shrink_control *sc);
4138 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
4139 struct shrink_control *sc);
4140 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
4141 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
4146 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
4147 struct rb_entry *cached_re, unsigned int ofs);
4148 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
4149 struct rb_root_cached *root,
4150 struct rb_node **parent,
4151 unsigned long long key, bool *left_most);
4152 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
4153 struct rb_root_cached *root,
4154 struct rb_node **parent,
4155 unsigned int ofs, bool *leftmost);
4156 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
4157 struct rb_entry *cached_re, unsigned int ofs,
4158 struct rb_entry **prev_entry, struct rb_entry **next_entry,
4159 struct rb_node ***insert_p, struct rb_node **insert_parent,
4160 bool force, bool *leftmost);
4161 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
4162 struct rb_root_cached *root, bool check_key);
4163 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
4164 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
4165 void f2fs_drop_extent_tree(struct inode *inode);
4166 unsigned int f2fs_destroy_extent_node(struct inode *inode);
4167 void f2fs_destroy_extent_tree(struct inode *inode);
4168 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
4169 struct extent_info *ei);
4170 void f2fs_update_extent_cache(struct dnode_of_data *dn);
4171 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
4172 pgoff_t fofs, block_t blkaddr, unsigned int len);
4173 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4174 int __init f2fs_create_extent_cache(void);
4175 void f2fs_destroy_extent_cache(void);
4180 #define MIN_RA_MUL 2
4181 #define MAX_RA_MUL 256
4183 int __init f2fs_init_sysfs(void);
4184 void f2fs_exit_sysfs(void);
4185 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4186 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4189 extern const struct fsverity_operations f2fs_verityops;
4194 static inline bool f2fs_encrypted_file(struct inode *inode)
4196 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4199 static inline void f2fs_set_encrypted_inode(struct inode *inode)
4201 #ifdef CONFIG_FS_ENCRYPTION
4202 file_set_encrypt(inode);
4203 f2fs_set_inode_flags(inode);
4208 * Returns true if the reads of the inode's data need to undergo some
4209 * postprocessing step, like decryption or authenticity verification.
4211 static inline bool f2fs_post_read_required(struct inode *inode)
4213 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4214 f2fs_compressed_file(inode);
4220 #ifdef CONFIG_F2FS_FS_COMPRESSION
4221 bool f2fs_is_compressed_page(struct page *page);
4222 struct page *f2fs_compress_control_page(struct page *page);
4223 int f2fs_prepare_compress_overwrite(struct inode *inode,
4224 struct page **pagep, pgoff_t index, void **fsdata);
4225 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4226 pgoff_t index, unsigned copied);
4227 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4228 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4229 bool f2fs_is_compress_backend_ready(struct inode *inode);
4230 int f2fs_init_compress_mempool(void);
4231 void f2fs_destroy_compress_mempool(void);
4232 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);
4233 void f2fs_end_read_compressed_page(struct page *page, bool failed,
4234 block_t blkaddr, bool in_task);
4235 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4236 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4237 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,
4238 int index, int nr_pages, bool uptodate);
4239 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
4240 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4241 int f2fs_write_multi_pages(struct compress_ctx *cc,
4243 struct writeback_control *wbc,
4244 enum iostat_type io_type);
4245 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4246 void f2fs_update_extent_tree_range_compressed(struct inode *inode,
4247 pgoff_t fofs, block_t blkaddr, unsigned int llen,
4248 unsigned int c_len);
4249 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4250 unsigned nr_pages, sector_t *last_block_in_bio,
4251 bool is_readahead, bool for_write);
4252 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4253 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
4255 void f2fs_put_page_dic(struct page *page, bool in_task);
4256 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn);
4257 int f2fs_init_compress_ctx(struct compress_ctx *cc);
4258 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4259 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4260 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4261 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4262 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4263 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4264 int __init f2fs_init_compress_cache(void);
4265 void f2fs_destroy_compress_cache(void);
4266 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4267 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4268 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4269 nid_t ino, block_t blkaddr);
4270 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4272 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4273 #define inc_compr_inode_stat(inode) \
4275 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4276 sbi->compr_new_inode++; \
4278 #define add_compr_block_stat(inode, blocks) \
4280 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4281 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4282 sbi->compr_written_block += blocks; \
4283 sbi->compr_saved_block += diff; \
4286 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4287 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4289 if (!f2fs_compressed_file(inode))
4291 /* not support compression */
4294 static inline struct page *f2fs_compress_control_page(struct page *page)
4297 return ERR_PTR(-EINVAL);
4299 static inline int f2fs_init_compress_mempool(void) { return 0; }
4300 static inline void f2fs_destroy_compress_mempool(void) { }
4301 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,
4303 static inline void f2fs_end_read_compressed_page(struct page *page,
4304 bool failed, block_t blkaddr, bool in_task)
4308 static inline void f2fs_put_page_dic(struct page *page, bool in_task)
4312 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; }
4313 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
4314 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4315 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4316 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4317 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4318 static inline int __init f2fs_init_compress_cache(void) { return 0; }
4319 static inline void f2fs_destroy_compress_cache(void) { }
4320 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4321 block_t blkaddr) { }
4322 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4323 struct page *page, nid_t ino, block_t blkaddr) { }
4324 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4325 struct page *page, block_t blkaddr) { return false; }
4326 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4328 #define inc_compr_inode_stat(inode) do { } while (0)
4329 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode,
4330 pgoff_t fofs, block_t blkaddr, unsigned int llen,
4331 unsigned int c_len) { }
4334 static inline int set_compress_context(struct inode *inode)
4336 #ifdef CONFIG_F2FS_FS_COMPRESSION
4337 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4339 F2FS_I(inode)->i_compress_algorithm =
4340 F2FS_OPTION(sbi).compress_algorithm;
4341 F2FS_I(inode)->i_log_cluster_size =
4342 F2FS_OPTION(sbi).compress_log_size;
4343 F2FS_I(inode)->i_compress_flag =
4344 F2FS_OPTION(sbi).compress_chksum ?
4345 1 << COMPRESS_CHKSUM : 0;
4346 F2FS_I(inode)->i_cluster_size =
4347 1 << F2FS_I(inode)->i_log_cluster_size;
4348 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4349 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4350 F2FS_OPTION(sbi).compress_level)
4351 F2FS_I(inode)->i_compress_flag |=
4352 F2FS_OPTION(sbi).compress_level <<
4353 COMPRESS_LEVEL_OFFSET;
4354 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4355 set_inode_flag(inode, FI_COMPRESSED_FILE);
4356 stat_inc_compr_inode(inode);
4357 inc_compr_inode_stat(inode);
4358 f2fs_mark_inode_dirty_sync(inode, true);
4365 static inline bool f2fs_disable_compressed_file(struct inode *inode)
4367 struct f2fs_inode_info *fi = F2FS_I(inode);
4369 if (!f2fs_compressed_file(inode))
4371 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))
4374 fi->i_flags &= ~F2FS_COMPR_FL;
4375 stat_dec_compr_inode(inode);
4376 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4377 f2fs_mark_inode_dirty_sync(inode, true);
4381 #define F2FS_FEATURE_FUNCS(name, flagname) \
4382 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4384 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4387 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4388 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4389 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4390 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4391 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4392 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4393 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4394 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4395 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4396 F2FS_FEATURE_FUNCS(verity, VERITY);
4397 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4398 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4399 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4400 F2FS_FEATURE_FUNCS(readonly, RO);
4402 static inline bool f2fs_may_extent_tree(struct inode *inode)
4404 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4406 if (!test_opt(sbi, EXTENT_CACHE) ||
4407 is_inode_flag_set(inode, FI_NO_EXTENT) ||
4408 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) &&
4409 !f2fs_sb_has_readonly(sbi)))
4413 * for recovered files during mount do not create extents
4414 * if shrinker is not registered.
4416 if (list_empty(&sbi->s_list))
4419 return S_ISREG(inode->i_mode);
4422 #ifdef CONFIG_BLK_DEV_ZONED
4423 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4426 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4428 return test_bit(zno, FDEV(devi).blkz_seq);
4432 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4434 return f2fs_sb_has_blkzoned(sbi);
4437 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4439 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);
4442 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4446 if (!f2fs_is_multi_device(sbi))
4447 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4449 for (i = 0; i < sbi->s_ndevs; i++)
4450 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4455 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4457 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4458 f2fs_hw_should_discard(sbi);
4461 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4465 if (!f2fs_is_multi_device(sbi))
4466 return bdev_read_only(sbi->sb->s_bdev);
4468 for (i = 0; i < sbi->s_ndevs; i++)
4469 if (bdev_read_only(FDEV(i).bdev))
4474 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4476 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4479 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)
4481 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;
4484 static inline bool f2fs_may_compress(struct inode *inode)
4486 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4487 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode))
4489 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4492 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4493 u64 blocks, bool add)
4495 struct f2fs_inode_info *fi = F2FS_I(inode);
4496 int diff = fi->i_cluster_size - blocks;
4498 /* don't update i_compr_blocks if saved blocks were released */
4499 if (!add && !atomic_read(&fi->i_compr_blocks))
4503 atomic_add(diff, &fi->i_compr_blocks);
4504 stat_add_compr_blocks(inode, diff);
4506 atomic_sub(diff, &fi->i_compr_blocks);
4507 stat_sub_compr_blocks(inode, diff);
4509 f2fs_mark_inode_dirty_sync(inode, true);
4512 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,
4515 if (!f2fs_is_multi_device(sbi))
4517 if (flag != F2FS_GET_BLOCK_DIO)
4519 return sbi->aligned_blksize;
4522 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4524 return fsverity_active(inode) &&
4525 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4528 #ifdef CONFIG_F2FS_FAULT_INJECTION
4529 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4532 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4535 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4538 if (f2fs_sb_has_quota_ino(sbi))
4540 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4541 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4542 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4548 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4550 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4553 static inline void f2fs_io_schedule_timeout(long timeout)
4555 set_current_state(TASK_UNINTERRUPTIBLE);
4556 io_schedule_timeout(timeout);
4559 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs,
4560 enum page_type type)
4562 if (unlikely(f2fs_cp_error(sbi)))
4565 if (ofs == sbi->page_eio_ofs[type]) {
4566 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)
4567 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4569 sbi->page_eio_ofs[type] = ofs;
4570 sbi->page_eio_cnt[type] = 0;
4574 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4575 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4577 #endif /* _LINUX_F2FS_H */