1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
11 #include <linux/uio.h>
12 #include <linux/types.h>
13 #include <linux/page-flags.h>
14 #include <linux/buffer_head.h>
15 #include <linux/slab.h>
16 #include <linux/crc32.h>
17 #include <linux/magic.h>
18 #include <linux/kobject.h>
19 #include <linux/sched.h>
20 #include <linux/cred.h>
21 #include <linux/vmalloc.h>
22 #include <linux/bio.h>
23 #include <linux/blkdev.h>
24 #include <linux/quotaops.h>
25 #include <crypto/hash.h>
27 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_F2FS_FS_ENCRYPTION)
28 #include <linux/fscrypt.h>
30 #ifdef CONFIG_F2FS_CHECK_FS
31 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
33 #define f2fs_bug_on(sbi, condition) \
35 if (unlikely(condition)) { \
37 set_sbi_flag(sbi, SBI_NEED_FSCK); \
61 #ifdef CONFIG_F2FS_FAULT_INJECTION
62 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
64 struct f2fs_fault_info {
66 unsigned int inject_rate;
67 unsigned int inject_type;
70 extern const char *f2fs_fault_name[FAULT_MAX];
71 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
77 #define F2FS_MOUNT_BG_GC 0x00000001
78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
79 #define F2FS_MOUNT_DISCARD 0x00000004
80 #define F2FS_MOUNT_NOHEAP 0x00000008
81 #define F2FS_MOUNT_XATTR_USER 0x00000010
82 #define F2FS_MOUNT_POSIX_ACL 0x00000020
83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
85 #define F2FS_MOUNT_INLINE_DATA 0x00000100
86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
88 #define F2FS_MOUNT_NOBARRIER 0x00000800
89 #define F2FS_MOUNT_FASTBOOT 0x00001000
90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
91 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000
92 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
93 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
94 #define F2FS_MOUNT_ADAPTIVE 0x00020000
95 #define F2FS_MOUNT_LFS 0x00040000
96 #define F2FS_MOUNT_USRQUOTA 0x00080000
97 #define F2FS_MOUNT_GRPQUOTA 0x00100000
98 #define F2FS_MOUNT_PRJQUOTA 0x00200000
99 #define F2FS_MOUNT_QUOTA 0x00400000
100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
104 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
105 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
106 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
107 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
109 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
110 typecheck(unsigned long long, b) && \
111 ((long long)((a) - (b)) > 0))
113 typedef u32 block_t; /*
114 * should not change u32, since it is the on-disk block
115 * address format, __le32.
119 struct f2fs_mount_info {
121 int write_io_size_bits; /* Write IO size bits */
122 block_t root_reserved_blocks; /* root reserved blocks */
123 kuid_t s_resuid; /* reserved blocks for uid */
124 kgid_t s_resgid; /* reserved blocks for gid */
125 int active_logs; /* # of active logs */
126 int inline_xattr_size; /* inline xattr size */
127 #ifdef CONFIG_F2FS_FAULT_INJECTION
128 struct f2fs_fault_info fault_info; /* For fault injection */
131 /* Names of quota files with journalled quota */
132 char *s_qf_names[MAXQUOTAS];
133 int s_jquota_fmt; /* Format of quota to use */
135 /* For which write hints are passed down to block layer */
137 int alloc_mode; /* segment allocation policy */
138 int fsync_mode; /* fsync policy */
139 bool test_dummy_encryption; /* test dummy encryption */
142 #define F2FS_FEATURE_ENCRYPT 0x0001
143 #define F2FS_FEATURE_BLKZONED 0x0002
144 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
145 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
146 #define F2FS_FEATURE_PRJQUOTA 0x0010
147 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
148 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
149 #define F2FS_FEATURE_QUOTA_INO 0x0080
150 #define F2FS_FEATURE_INODE_CRTIME 0x0100
151 #define F2FS_FEATURE_LOST_FOUND 0x0200
152 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */
153 #define F2FS_FEATURE_SB_CHKSUM 0x0800
155 #define __F2FS_HAS_FEATURE(raw_super, mask) \
156 ((raw_super->feature & cpu_to_le32(mask)) != 0)
157 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
158 #define F2FS_SET_FEATURE(sbi, mask) \
159 (sbi->raw_super->feature |= cpu_to_le32(mask))
160 #define F2FS_CLEAR_FEATURE(sbi, mask) \
161 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
164 * Default values for user and/or group using reserved blocks
166 #define F2FS_DEF_RESUID 0
167 #define F2FS_DEF_RESGID 0
170 * For checkpoint manager
177 #define CP_UMOUNT 0x00000001
178 #define CP_FASTBOOT 0x00000002
179 #define CP_SYNC 0x00000004
180 #define CP_RECOVERY 0x00000008
181 #define CP_DISCARD 0x00000010
182 #define CP_TRIMMED 0x00000020
183 #define CP_PAUSE 0x00000040
185 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
186 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
187 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
188 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
189 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
190 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
191 #define DEF_CP_INTERVAL 60 /* 60 secs */
192 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
193 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
203 * indicate meta/data type
216 /* for the list of ino */
218 ORPHAN_INO, /* for orphan ino list */
219 APPEND_INO, /* for append ino list */
220 UPDATE_INO, /* for update ino list */
221 TRANS_DIR_INO, /* for trasactions dir ino list */
222 FLUSH_INO, /* for multiple device flushing */
223 MAX_INO_ENTRY, /* max. list */
227 struct list_head list; /* list head */
228 nid_t ino; /* inode number */
229 unsigned int dirty_device; /* dirty device bitmap */
232 /* for the list of inodes to be GCed */
234 struct list_head list; /* list head */
235 struct inode *inode; /* vfs inode pointer */
238 struct fsync_node_entry {
239 struct list_head list; /* list head */
240 struct page *page; /* warm node page pointer */
241 unsigned int seq_id; /* sequence id */
244 /* for the bitmap indicate blocks to be discarded */
245 struct discard_entry {
246 struct list_head list; /* list head */
247 block_t start_blkaddr; /* start blockaddr of current segment */
248 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
251 /* default discard granularity of inner discard thread, unit: block count */
252 #define DEFAULT_DISCARD_GRANULARITY 16
254 /* max discard pend list number */
255 #define MAX_PLIST_NUM 512
256 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
257 (MAX_PLIST_NUM - 1) : (blk_num - 1))
260 D_PREP, /* initial */
261 D_PARTIAL, /* partially submitted */
262 D_SUBMIT, /* all submitted */
263 D_DONE, /* finished */
266 struct discard_info {
267 block_t lstart; /* logical start address */
268 block_t len; /* length */
269 block_t start; /* actual start address in dev */
273 struct rb_node rb_node; /* rb node located in rb-tree */
276 block_t lstart; /* logical start address */
277 block_t len; /* length */
278 block_t start; /* actual start address in dev */
280 struct discard_info di; /* discard info */
283 struct list_head list; /* command list */
284 struct completion wait; /* compleation */
285 struct block_device *bdev; /* bdev */
286 unsigned short ref; /* reference count */
287 unsigned char state; /* state */
288 unsigned char queued; /* queued discard */
289 int error; /* bio error */
290 spinlock_t lock; /* for state/bio_ref updating */
291 unsigned short bio_ref; /* bio reference count */
302 struct discard_policy {
303 int type; /* type of discard */
304 unsigned int min_interval; /* used for candidates exist */
305 unsigned int mid_interval; /* used for device busy */
306 unsigned int max_interval; /* used for candidates not exist */
307 unsigned int max_requests; /* # of discards issued per round */
308 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
309 bool io_aware; /* issue discard in idle time */
310 bool sync; /* submit discard with REQ_SYNC flag */
311 bool ordered; /* issue discard by lba order */
312 unsigned int granularity; /* discard granularity */
315 struct discard_cmd_control {
316 struct task_struct *f2fs_issue_discard; /* discard thread */
317 struct list_head entry_list; /* 4KB discard entry list */
318 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
319 struct list_head wait_list; /* store on-flushing entries */
320 struct list_head fstrim_list; /* in-flight discard from fstrim */
321 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
322 unsigned int discard_wake; /* to wake up discard thread */
323 struct mutex cmd_lock;
324 unsigned int nr_discards; /* # of discards in the list */
325 unsigned int max_discards; /* max. discards to be issued */
326 unsigned int discard_granularity; /* discard granularity */
327 unsigned int undiscard_blks; /* # of undiscard blocks */
328 unsigned int next_pos; /* next discard position */
329 atomic_t issued_discard; /* # of issued discard */
330 atomic_t queued_discard; /* # of queued discard */
331 atomic_t discard_cmd_cnt; /* # of cached cmd count */
332 struct rb_root_cached root; /* root of discard rb-tree */
333 bool rbtree_check; /* config for consistence check */
336 /* for the list of fsync inodes, used only during recovery */
337 struct fsync_inode_entry {
338 struct list_head list; /* list head */
339 struct inode *inode; /* vfs inode pointer */
340 block_t blkaddr; /* block address locating the last fsync */
341 block_t last_dentry; /* block address locating the last dentry */
344 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
345 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
347 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
348 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
349 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
350 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
352 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
353 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
355 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
357 int before = nats_in_cursum(journal);
359 journal->n_nats = cpu_to_le16(before + i);
363 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
365 int before = sits_in_cursum(journal);
367 journal->n_sits = cpu_to_le16(before + i);
371 static inline bool __has_cursum_space(struct f2fs_journal *journal,
374 if (type == NAT_JOURNAL)
375 return size <= MAX_NAT_JENTRIES(journal);
376 return size <= MAX_SIT_JENTRIES(journal);
382 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
383 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
384 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
386 #define F2FS_IOCTL_MAGIC 0xf5
387 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
388 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
389 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
390 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
391 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
392 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
393 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
394 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
395 struct f2fs_defragment)
396 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
397 struct f2fs_move_range)
398 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
399 struct f2fs_flush_device)
400 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
401 struct f2fs_gc_range)
402 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
403 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
404 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
405 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
407 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
408 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
409 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
412 * should be same as XFS_IOC_GOINGDOWN.
413 * Flags for going down operation used by FS_IOC_GOINGDOWN
415 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
416 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
417 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
418 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
419 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
420 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
422 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
424 * ioctl commands in 32 bit emulation
426 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
427 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
428 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
431 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
432 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
434 struct f2fs_gc_range {
440 struct f2fs_defragment {
445 struct f2fs_move_range {
446 u32 dst_fd; /* destination fd */
447 u64 pos_in; /* start position in src_fd */
448 u64 pos_out; /* start position in dst_fd */
449 u64 len; /* size to move */
452 struct f2fs_flush_device {
453 u32 dev_num; /* device number to flush */
454 u32 segments; /* # of segments to flush */
457 /* for inline stuff */
458 #define DEF_INLINE_RESERVED_SIZE 1
459 #define DEF_MIN_INLINE_SIZE 1
460 static inline int get_extra_isize(struct inode *inode);
461 static inline int get_inline_xattr_addrs(struct inode *inode);
462 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
463 (CUR_ADDRS_PER_INODE(inode) - \
464 get_inline_xattr_addrs(inode) - \
465 DEF_INLINE_RESERVED_SIZE))
468 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
469 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
471 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \
472 BITS_PER_BYTE - 1) / BITS_PER_BYTE)
473 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
474 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
475 NR_INLINE_DENTRY(inode) + \
476 INLINE_DENTRY_BITMAP_SIZE(inode)))
479 * For INODE and NODE manager
481 /* for directory operations */
482 struct f2fs_dentry_ptr {
485 struct f2fs_dir_entry *dentry;
486 __u8 (*filename)[F2FS_SLOT_LEN];
491 static inline void make_dentry_ptr_block(struct inode *inode,
492 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
495 d->max = NR_DENTRY_IN_BLOCK;
496 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
497 d->bitmap = t->dentry_bitmap;
498 d->dentry = t->dentry;
499 d->filename = t->filename;
502 static inline void make_dentry_ptr_inline(struct inode *inode,
503 struct f2fs_dentry_ptr *d, void *t)
505 int entry_cnt = NR_INLINE_DENTRY(inode);
506 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
507 int reserved_size = INLINE_RESERVED_SIZE(inode);
511 d->nr_bitmap = bitmap_size;
513 d->dentry = t + bitmap_size + reserved_size;
514 d->filename = t + bitmap_size + reserved_size +
515 SIZE_OF_DIR_ENTRY * entry_cnt;
519 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
520 * as its node offset to distinguish from index node blocks.
521 * But some bits are used to mark the node block.
523 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
526 ALLOC_NODE, /* allocate a new node page if needed */
527 LOOKUP_NODE, /* look up a node without readahead */
529 * look up a node with readahead called
534 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
536 /* maximum retry quota flush count */
537 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
539 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
541 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
543 /* for in-memory extent cache entry */
544 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
546 /* number of extent info in extent cache we try to shrink */
547 #define EXTENT_CACHE_SHRINK_NUMBER 128
550 struct rb_node rb_node; /* rb node located in rb-tree */
551 unsigned int ofs; /* start offset of the entry */
552 unsigned int len; /* length of the entry */
556 unsigned int fofs; /* start offset in a file */
557 unsigned int len; /* length of the extent */
558 u32 blk; /* start block address of the extent */
562 struct rb_node rb_node; /* rb node located in rb-tree */
563 struct extent_info ei; /* extent info */
564 struct list_head list; /* node in global extent list of sbi */
565 struct extent_tree *et; /* extent tree pointer */
569 nid_t ino; /* inode number */
570 struct rb_root_cached root; /* root of extent info rb-tree */
571 struct extent_node *cached_en; /* recently accessed extent node */
572 struct extent_info largest; /* largested extent info */
573 struct list_head list; /* to be used by sbi->zombie_list */
574 rwlock_t lock; /* protect extent info rb-tree */
575 atomic_t node_cnt; /* # of extent node in rb-tree*/
576 bool largest_updated; /* largest extent updated */
580 * This structure is taken from ext4_map_blocks.
582 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
584 #define F2FS_MAP_NEW (1 << BH_New)
585 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
586 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
587 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
590 struct f2fs_map_blocks {
594 unsigned int m_flags;
595 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
596 pgoff_t *m_next_extent; /* point to next possible extent */
598 bool m_may_create; /* indicate it is from write path */
601 /* for flag in get_data_block */
603 F2FS_GET_BLOCK_DEFAULT,
604 F2FS_GET_BLOCK_FIEMAP,
607 F2FS_GET_BLOCK_PRE_DIO,
608 F2FS_GET_BLOCK_PRE_AIO,
609 F2FS_GET_BLOCK_PRECACHE,
613 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
615 #define FADVISE_COLD_BIT 0x01
616 #define FADVISE_LOST_PINO_BIT 0x02
617 #define FADVISE_ENCRYPT_BIT 0x04
618 #define FADVISE_ENC_NAME_BIT 0x08
619 #define FADVISE_KEEP_SIZE_BIT 0x10
620 #define FADVISE_HOT_BIT 0x20
621 #define FADVISE_VERITY_BIT 0x40 /* reserved */
623 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
625 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
626 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
627 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
628 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
629 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
630 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
631 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
632 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
633 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
634 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
635 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
636 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
637 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
638 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
639 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
640 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
642 #define DEF_DIR_LEVEL 0
650 struct f2fs_inode_info {
651 struct inode vfs_inode; /* serve a vfs inode */
652 unsigned long i_flags; /* keep an inode flags for ioctl */
653 unsigned char i_advise; /* use to give file attribute hints */
654 unsigned char i_dir_level; /* use for dentry level for large dir */
655 unsigned int i_current_depth; /* only for directory depth */
656 /* for gc failure statistic */
657 unsigned int i_gc_failures[MAX_GC_FAILURE];
658 unsigned int i_pino; /* parent inode number */
659 umode_t i_acl_mode; /* keep file acl mode temporarily */
661 /* Use below internally in f2fs*/
662 unsigned long flags; /* use to pass per-file flags */
663 struct rw_semaphore i_sem; /* protect fi info */
664 atomic_t dirty_pages; /* # of dirty pages */
665 f2fs_hash_t chash; /* hash value of given file name */
666 unsigned int clevel; /* maximum level of given file name */
667 struct task_struct *task; /* lookup and create consistency */
668 struct task_struct *cp_task; /* separate cp/wb IO stats*/
669 nid_t i_xattr_nid; /* node id that contains xattrs */
670 loff_t last_disk_size; /* lastly written file size */
673 struct dquot *i_dquot[MAXQUOTAS];
675 /* quota space reservation, managed internally by quota code */
676 qsize_t i_reserved_quota;
678 struct list_head dirty_list; /* dirty list for dirs and files */
679 struct list_head gdirty_list; /* linked in global dirty list */
680 struct list_head inmem_ilist; /* list for inmem inodes */
681 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
682 struct task_struct *inmem_task; /* store inmemory task */
683 struct mutex inmem_lock; /* lock for inmemory pages */
684 struct extent_tree *extent_tree; /* cached extent_tree entry */
686 /* avoid racing between foreground op and gc */
687 struct rw_semaphore i_gc_rwsem[2];
688 struct rw_semaphore i_mmap_sem;
689 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
691 int i_extra_isize; /* size of extra space located in i_addr */
692 kprojid_t i_projid; /* id for project quota */
693 int i_inline_xattr_size; /* inline xattr size */
694 struct timespec64 i_crtime; /* inode creation time */
695 struct timespec64 i_disk_time[4];/* inode disk times */
698 static inline void get_extent_info(struct extent_info *ext,
699 struct f2fs_extent *i_ext)
701 ext->fofs = le32_to_cpu(i_ext->fofs);
702 ext->blk = le32_to_cpu(i_ext->blk);
703 ext->len = le32_to_cpu(i_ext->len);
706 static inline void set_raw_extent(struct extent_info *ext,
707 struct f2fs_extent *i_ext)
709 i_ext->fofs = cpu_to_le32(ext->fofs);
710 i_ext->blk = cpu_to_le32(ext->blk);
711 i_ext->len = cpu_to_le32(ext->len);
714 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
715 u32 blk, unsigned int len)
722 static inline bool __is_discard_mergeable(struct discard_info *back,
723 struct discard_info *front, unsigned int max_len)
725 return (back->lstart + back->len == front->lstart) &&
726 (back->len + front->len <= max_len);
729 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
730 struct discard_info *back, unsigned int max_len)
732 return __is_discard_mergeable(back, cur, max_len);
735 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
736 struct discard_info *front, unsigned int max_len)
738 return __is_discard_mergeable(cur, front, max_len);
741 static inline bool __is_extent_mergeable(struct extent_info *back,
742 struct extent_info *front)
744 return (back->fofs + back->len == front->fofs &&
745 back->blk + back->len == front->blk);
748 static inline bool __is_back_mergeable(struct extent_info *cur,
749 struct extent_info *back)
751 return __is_extent_mergeable(back, cur);
754 static inline bool __is_front_mergeable(struct extent_info *cur,
755 struct extent_info *front)
757 return __is_extent_mergeable(cur, front);
760 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
761 static inline void __try_update_largest_extent(struct extent_tree *et,
762 struct extent_node *en)
764 if (en->ei.len > et->largest.len) {
765 et->largest = en->ei;
766 et->largest_updated = true;
771 * For free nid management
774 FREE_NID, /* newly added to free nid list */
775 PREALLOC_NID, /* it is preallocated */
779 struct f2fs_nm_info {
780 block_t nat_blkaddr; /* base disk address of NAT */
781 nid_t max_nid; /* maximum possible node ids */
782 nid_t available_nids; /* # of available node ids */
783 nid_t next_scan_nid; /* the next nid to be scanned */
784 unsigned int ram_thresh; /* control the memory footprint */
785 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
786 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
788 /* NAT cache management */
789 struct radix_tree_root nat_root;/* root of the nat entry cache */
790 struct radix_tree_root nat_set_root;/* root of the nat set cache */
791 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
792 struct list_head nat_entries; /* cached nat entry list (clean) */
793 spinlock_t nat_list_lock; /* protect clean nat entry list */
794 unsigned int nat_cnt; /* the # of cached nat entries */
795 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
796 unsigned int nat_blocks; /* # of nat blocks */
798 /* free node ids management */
799 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
800 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
801 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
802 spinlock_t nid_list_lock; /* protect nid lists ops */
803 struct mutex build_lock; /* lock for build free nids */
804 unsigned char **free_nid_bitmap;
805 unsigned char *nat_block_bitmap;
806 unsigned short *free_nid_count; /* free nid count of NAT block */
809 char *nat_bitmap; /* NAT bitmap pointer */
811 unsigned int nat_bits_blocks; /* # of nat bits blocks */
812 unsigned char *nat_bits; /* NAT bits blocks */
813 unsigned char *full_nat_bits; /* full NAT pages */
814 unsigned char *empty_nat_bits; /* empty NAT pages */
815 #ifdef CONFIG_F2FS_CHECK_FS
816 char *nat_bitmap_mir; /* NAT bitmap mirror */
818 int bitmap_size; /* bitmap size */
822 * this structure is used as one of function parameters.
823 * all the information are dedicated to a given direct node block determined
824 * by the data offset in a file.
826 struct dnode_of_data {
827 struct inode *inode; /* vfs inode pointer */
828 struct page *inode_page; /* its inode page, NULL is possible */
829 struct page *node_page; /* cached direct node page */
830 nid_t nid; /* node id of the direct node block */
831 unsigned int ofs_in_node; /* data offset in the node page */
832 bool inode_page_locked; /* inode page is locked or not */
833 bool node_changed; /* is node block changed */
834 char cur_level; /* level of hole node page */
835 char max_level; /* level of current page located */
836 block_t data_blkaddr; /* block address of the node block */
839 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
840 struct page *ipage, struct page *npage, nid_t nid)
842 memset(dn, 0, sizeof(*dn));
844 dn->inode_page = ipage;
845 dn->node_page = npage;
852 * By default, there are 6 active log areas across the whole main area.
853 * When considering hot and cold data separation to reduce cleaning overhead,
854 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
856 * In the current design, you should not change the numbers intentionally.
857 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
858 * logs individually according to the underlying devices. (default: 6)
859 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
860 * data and 8 for node logs.
862 #define NR_CURSEG_DATA_TYPE (3)
863 #define NR_CURSEG_NODE_TYPE (3)
864 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
867 CURSEG_HOT_DATA = 0, /* directory entry blocks */
868 CURSEG_WARM_DATA, /* data blocks */
869 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
870 CURSEG_HOT_NODE, /* direct node blocks of directory files */
871 CURSEG_WARM_NODE, /* direct node blocks of normal files */
872 CURSEG_COLD_NODE, /* indirect node blocks */
877 struct completion wait;
878 struct llist_node llnode;
883 struct flush_cmd_control {
884 struct task_struct *f2fs_issue_flush; /* flush thread */
885 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
886 atomic_t issued_flush; /* # of issued flushes */
887 atomic_t queued_flush; /* # of queued flushes */
888 struct llist_head issue_list; /* list for command issue */
889 struct llist_node *dispatch_list; /* list for command dispatch */
892 struct f2fs_sm_info {
893 struct sit_info *sit_info; /* whole segment information */
894 struct free_segmap_info *free_info; /* free segment information */
895 struct dirty_seglist_info *dirty_info; /* dirty segment information */
896 struct curseg_info *curseg_array; /* active segment information */
898 struct rw_semaphore curseg_lock; /* for preventing curseg change */
900 block_t seg0_blkaddr; /* block address of 0'th segment */
901 block_t main_blkaddr; /* start block address of main area */
902 block_t ssa_blkaddr; /* start block address of SSA area */
904 unsigned int segment_count; /* total # of segments */
905 unsigned int main_segments; /* # of segments in main area */
906 unsigned int reserved_segments; /* # of reserved segments */
907 unsigned int ovp_segments; /* # of overprovision segments */
909 /* a threshold to reclaim prefree segments */
910 unsigned int rec_prefree_segments;
912 /* for batched trimming */
913 unsigned int trim_sections; /* # of sections to trim */
915 struct list_head sit_entry_set; /* sit entry set list */
917 unsigned int ipu_policy; /* in-place-update policy */
918 unsigned int min_ipu_util; /* in-place-update threshold */
919 unsigned int min_fsync_blocks; /* threshold for fsync */
920 unsigned int min_seq_blocks; /* threshold for sequential blocks */
921 unsigned int min_hot_blocks; /* threshold for hot block allocation */
922 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
924 /* for flush command control */
925 struct flush_cmd_control *fcc_info;
927 /* for discard command control */
928 struct discard_cmd_control *dcc_info;
935 * COUNT_TYPE for monitoring
937 * f2fs monitors the number of several block types such as on-writeback,
938 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
940 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
960 * The below are the page types of bios used in submit_bio().
961 * The available types are:
962 * DATA User data pages. It operates as async mode.
963 * NODE Node pages. It operates as async mode.
964 * META FS metadata pages such as SIT, NAT, CP.
965 * NR_PAGE_TYPE The number of page types.
966 * META_FLUSH Make sure the previous pages are written
967 * with waiting the bio's completion
968 * ... Only can be used with META.
970 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
977 INMEM, /* the below types are used by tracepoints only. */
986 HOT = 0, /* must be zero for meta bio */
992 enum need_lock_type {
998 enum cp_reason_type {
1012 APP_DIRECT_IO, /* app direct IOs */
1013 APP_BUFFERED_IO, /* app buffered IOs */
1014 APP_WRITE_IO, /* app write IOs */
1015 APP_MAPPED_IO, /* app mapped IOs */
1016 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1017 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1018 FS_META_IO, /* meta IOs from kworker/reclaimer */
1019 FS_GC_DATA_IO, /* data IOs from forground gc */
1020 FS_GC_NODE_IO, /* node IOs from forground gc */
1021 FS_CP_DATA_IO, /* data IOs from checkpoint */
1022 FS_CP_NODE_IO, /* node IOs from checkpoint */
1023 FS_CP_META_IO, /* meta IOs from checkpoint */
1024 FS_DISCARD, /* discard */
1028 struct f2fs_io_info {
1029 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1030 nid_t ino; /* inode number */
1031 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1032 enum temp_type temp; /* contains HOT/WARM/COLD */
1033 int op; /* contains REQ_OP_ */
1034 int op_flags; /* req_flag_bits */
1035 block_t new_blkaddr; /* new block address to be written */
1036 block_t old_blkaddr; /* old block address before Cow */
1037 struct page *page; /* page to be written */
1038 struct page *encrypted_page; /* encrypted page */
1039 struct list_head list; /* serialize IOs */
1040 bool submitted; /* indicate IO submission */
1041 int need_lock; /* indicate we need to lock cp_rwsem */
1042 bool in_list; /* indicate fio is in io_list */
1043 bool is_meta; /* indicate borrow meta inode mapping or not */
1044 bool retry; /* need to reallocate block address */
1045 enum iostat_type io_type; /* io type */
1046 struct writeback_control *io_wbc; /* writeback control */
1047 unsigned char version; /* version of the node */
1050 #define is_read_io(rw) ((rw) == READ)
1051 struct f2fs_bio_info {
1052 struct f2fs_sb_info *sbi; /* f2fs superblock */
1053 struct bio *bio; /* bios to merge */
1054 sector_t last_block_in_bio; /* last block number */
1055 struct f2fs_io_info fio; /* store buffered io info. */
1056 struct rw_semaphore io_rwsem; /* blocking op for bio */
1057 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1058 struct list_head io_list; /* track fios */
1061 #define FDEV(i) (sbi->devs[i])
1062 #define RDEV(i) (raw_super->devs[i])
1063 struct f2fs_dev_info {
1064 struct block_device *bdev;
1065 char path[MAX_PATH_LEN];
1066 unsigned int total_segments;
1069 #ifdef CONFIG_BLK_DEV_ZONED
1070 unsigned int nr_blkz; /* Total number of zones */
1071 u8 *blkz_type; /* Array of zones type */
1076 DIR_INODE, /* for dirty dir inode */
1077 FILE_INODE, /* for dirty regular/symlink inode */
1078 DIRTY_META, /* for all dirtied inode metadata */
1079 ATOMIC_FILE, /* for all atomic files */
1083 /* for inner inode cache management */
1084 struct inode_management {
1085 struct radix_tree_root ino_root; /* ino entry array */
1086 spinlock_t ino_lock; /* for ino entry lock */
1087 struct list_head ino_list; /* inode list head */
1088 unsigned long ino_num; /* number of entries */
1091 /* For s_flag in struct f2fs_sb_info */
1093 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1094 SBI_IS_CLOSE, /* specify unmounting */
1095 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1096 SBI_POR_DOING, /* recovery is doing or not */
1097 SBI_NEED_SB_WRITE, /* need to recover superblock */
1098 SBI_NEED_CP, /* need to checkpoint */
1099 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1100 SBI_IS_RECOVERED, /* recovered orphan/data */
1101 SBI_CP_DISABLED, /* CP was disabled last mount */
1102 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1103 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1104 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1124 WHINT_MODE_OFF, /* not pass down write hints */
1125 WHINT_MODE_USER, /* try to pass down hints given by users */
1126 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1130 ALLOC_MODE_DEFAULT, /* stay default */
1131 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1135 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1136 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1137 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1140 #ifdef CONFIG_F2FS_FS_ENCRYPTION
1141 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1142 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1144 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1147 struct f2fs_sb_info {
1148 struct super_block *sb; /* pointer to VFS super block */
1149 struct proc_dir_entry *s_proc; /* proc entry */
1150 struct f2fs_super_block *raw_super; /* raw super block pointer */
1151 struct rw_semaphore sb_lock; /* lock for raw super block */
1152 int valid_super_block; /* valid super block no */
1153 unsigned long s_flag; /* flags for sbi */
1154 struct mutex writepages; /* mutex for writepages() */
1156 #ifdef CONFIG_BLK_DEV_ZONED
1157 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1158 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1161 /* for node-related operations */
1162 struct f2fs_nm_info *nm_info; /* node manager */
1163 struct inode *node_inode; /* cache node blocks */
1165 /* for segment-related operations */
1166 struct f2fs_sm_info *sm_info; /* segment manager */
1168 /* for bio operations */
1169 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1170 /* keep migration IO order for LFS mode */
1171 struct rw_semaphore io_order_lock;
1172 mempool_t *write_io_dummy; /* Dummy pages */
1174 /* for checkpoint */
1175 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1176 int cur_cp_pack; /* remain current cp pack */
1177 spinlock_t cp_lock; /* for flag in ckpt */
1178 struct inode *meta_inode; /* cache meta blocks */
1179 struct mutex cp_mutex; /* checkpoint procedure lock */
1180 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1181 struct rw_semaphore node_write; /* locking node writes */
1182 struct rw_semaphore node_change; /* locking node change */
1183 wait_queue_head_t cp_wait;
1184 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1185 long interval_time[MAX_TIME]; /* to store thresholds */
1187 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1189 spinlock_t fsync_node_lock; /* for node entry lock */
1190 struct list_head fsync_node_list; /* node list head */
1191 unsigned int fsync_seg_id; /* sequence id */
1192 unsigned int fsync_node_num; /* number of node entries */
1194 /* for orphan inode, use 0'th array */
1195 unsigned int max_orphans; /* max orphan inodes */
1197 /* for inode management */
1198 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1199 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1201 /* for extent tree cache */
1202 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1203 struct mutex extent_tree_lock; /* locking extent radix tree */
1204 struct list_head extent_list; /* lru list for shrinker */
1205 spinlock_t extent_lock; /* locking extent lru list */
1206 atomic_t total_ext_tree; /* extent tree count */
1207 struct list_head zombie_list; /* extent zombie tree list */
1208 atomic_t total_zombie_tree; /* extent zombie tree count */
1209 atomic_t total_ext_node; /* extent info count */
1211 /* basic filesystem units */
1212 unsigned int log_sectors_per_block; /* log2 sectors per block */
1213 unsigned int log_blocksize; /* log2 block size */
1214 unsigned int blocksize; /* block size */
1215 unsigned int root_ino_num; /* root inode number*/
1216 unsigned int node_ino_num; /* node inode number*/
1217 unsigned int meta_ino_num; /* meta inode number*/
1218 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1219 unsigned int blocks_per_seg; /* blocks per segment */
1220 unsigned int segs_per_sec; /* segments per section */
1221 unsigned int secs_per_zone; /* sections per zone */
1222 unsigned int total_sections; /* total section count */
1223 unsigned int total_node_count; /* total node block count */
1224 unsigned int total_valid_node_count; /* valid node block count */
1225 loff_t max_file_blocks; /* max block index of file */
1226 int dir_level; /* directory level */
1227 int readdir_ra; /* readahead inode in readdir */
1229 block_t user_block_count; /* # of user blocks */
1230 block_t total_valid_block_count; /* # of valid blocks */
1231 block_t discard_blks; /* discard command candidats */
1232 block_t last_valid_block_count; /* for recovery */
1233 block_t reserved_blocks; /* configurable reserved blocks */
1234 block_t current_reserved_blocks; /* current reserved blocks */
1236 /* Additional tracking for no checkpoint mode */
1237 block_t unusable_block_count; /* # of blocks saved by last cp */
1239 unsigned int nquota_files; /* # of quota sysfile */
1241 u32 s_next_generation; /* for NFS support */
1243 /* # of pages, see count_type */
1244 atomic_t nr_pages[NR_COUNT_TYPE];
1245 /* # of allocated blocks */
1246 struct percpu_counter alloc_valid_block_count;
1248 /* writeback control */
1249 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1251 /* valid inode count */
1252 struct percpu_counter total_valid_inode_count;
1254 struct f2fs_mount_info mount_opt; /* mount options */
1256 /* for cleaning operations */
1257 struct mutex gc_mutex; /* mutex for GC */
1258 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1259 unsigned int cur_victim_sec; /* current victim section num */
1260 unsigned int gc_mode; /* current GC state */
1261 unsigned int next_victim_seg[2]; /* next segment in victim section */
1262 /* for skip statistic */
1263 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1264 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1266 /* threshold for gc trials on pinned files */
1267 u64 gc_pin_file_threshold;
1269 /* maximum # of trials to find a victim segment for SSR and GC */
1270 unsigned int max_victim_search;
1271 /* migration granularity of garbage collection, unit: segment */
1272 unsigned int migration_granularity;
1275 * for stat information.
1276 * one is for the LFS mode, and the other is for the SSR mode.
1278 #ifdef CONFIG_F2FS_STAT_FS
1279 struct f2fs_stat_info *stat_info; /* FS status information */
1280 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1281 unsigned int segment_count[2]; /* # of allocated segments */
1282 unsigned int block_count[2]; /* # of allocated blocks */
1283 atomic_t inplace_count; /* # of inplace update */
1284 atomic64_t total_hit_ext; /* # of lookup extent cache */
1285 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1286 atomic64_t read_hit_largest; /* # of hit largest extent node */
1287 atomic64_t read_hit_cached; /* # of hit cached extent node */
1288 atomic_t inline_xattr; /* # of inline_xattr inodes */
1289 atomic_t inline_inode; /* # of inline_data inodes */
1290 atomic_t inline_dir; /* # of inline_dentry inodes */
1291 atomic_t aw_cnt; /* # of atomic writes */
1292 atomic_t vw_cnt; /* # of volatile writes */
1293 atomic_t max_aw_cnt; /* max # of atomic writes */
1294 atomic_t max_vw_cnt; /* max # of volatile writes */
1295 int bg_gc; /* background gc calls */
1296 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1297 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1298 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1300 spinlock_t stat_lock; /* lock for stat operations */
1302 /* For app/fs IO statistics */
1303 spinlock_t iostat_lock;
1304 unsigned long long write_iostat[NR_IO_TYPE];
1307 /* For sysfs suppport */
1308 struct kobject s_kobj;
1309 struct completion s_kobj_unregister;
1311 /* For shrinker support */
1312 struct list_head s_list;
1313 int s_ndevs; /* number of devices */
1314 struct f2fs_dev_info *devs; /* for device list */
1315 unsigned int dirty_device; /* for checkpoint data flush */
1316 spinlock_t dev_lock; /* protect dirty_device */
1317 struct mutex umount_mutex;
1318 unsigned int shrinker_run_no;
1320 /* For write statistics */
1321 u64 sectors_written_start;
1324 /* Reference to checksum algorithm driver via cryptoapi */
1325 struct crypto_shash *s_chksum_driver;
1327 /* Precomputed FS UUID checksum for seeding other checksums */
1328 __u32 s_chksum_seed;
1331 struct f2fs_private_dio {
1332 struct inode *inode;
1334 bio_end_io_t *orig_end_io;
1338 #ifdef CONFIG_F2FS_FAULT_INJECTION
1339 #define f2fs_show_injection_info(type) \
1340 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pF\n", \
1341 KERN_INFO, f2fs_fault_name[type], \
1342 __func__, __builtin_return_address(0))
1343 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1345 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1347 if (!ffi->inject_rate)
1350 if (!IS_FAULT_SET(ffi, type))
1353 atomic_inc(&ffi->inject_ops);
1354 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1355 atomic_set(&ffi->inject_ops, 0);
1361 #define f2fs_show_injection_info(type) do { } while (0)
1362 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1368 /* For write statistics. Suppose sector size is 512 bytes,
1369 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1371 #define BD_PART_WRITTEN(s) \
1372 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1373 (s)->sectors_written_start) >> 1)
1375 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1377 unsigned long now = jiffies;
1379 sbi->last_time[type] = now;
1381 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1382 if (type == REQ_TIME) {
1383 sbi->last_time[DISCARD_TIME] = now;
1384 sbi->last_time[GC_TIME] = now;
1388 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1390 unsigned long interval = sbi->interval_time[type] * HZ;
1392 return time_after(jiffies, sbi->last_time[type] + interval);
1395 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1398 unsigned long interval = sbi->interval_time[type] * HZ;
1399 unsigned int wait_ms = 0;
1402 delta = (sbi->last_time[type] + interval) - jiffies;
1404 wait_ms = jiffies_to_msecs(delta);
1412 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1413 const void *address, unsigned int length)
1416 struct shash_desc shash;
1421 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1423 desc.shash.tfm = sbi->s_chksum_driver;
1424 desc.shash.flags = 0;
1425 *(u32 *)desc.ctx = crc;
1427 err = crypto_shash_update(&desc.shash, address, length);
1430 return *(u32 *)desc.ctx;
1433 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1434 unsigned int length)
1436 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1439 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1440 void *buf, size_t buf_size)
1442 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1445 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1446 const void *address, unsigned int length)
1448 return __f2fs_crc32(sbi, crc, address, length);
1451 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1453 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1456 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1458 return sb->s_fs_info;
1461 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1463 return F2FS_SB(inode->i_sb);
1466 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1468 return F2FS_I_SB(mapping->host);
1471 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1473 return F2FS_M_SB(page->mapping);
1476 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1478 return (struct f2fs_super_block *)(sbi->raw_super);
1481 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1483 return (struct f2fs_checkpoint *)(sbi->ckpt);
1486 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1488 return (struct f2fs_node *)page_address(page);
1491 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1493 return &((struct f2fs_node *)page_address(page))->i;
1496 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1498 return (struct f2fs_nm_info *)(sbi->nm_info);
1501 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1503 return (struct f2fs_sm_info *)(sbi->sm_info);
1506 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1508 return (struct sit_info *)(SM_I(sbi)->sit_info);
1511 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1513 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1516 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1518 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1521 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1523 return sbi->meta_inode->i_mapping;
1526 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1528 return sbi->node_inode->i_mapping;
1531 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1533 return test_bit(type, &sbi->s_flag);
1536 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1538 set_bit(type, &sbi->s_flag);
1541 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1543 clear_bit(type, &sbi->s_flag);
1546 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1548 return le64_to_cpu(cp->checkpoint_ver);
1551 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1553 if (type < F2FS_MAX_QUOTAS)
1554 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1558 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1560 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1561 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1564 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1566 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1568 return ckpt_flags & f;
1571 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1573 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1576 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1578 unsigned int ckpt_flags;
1580 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1582 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1585 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1587 unsigned long flags;
1589 spin_lock_irqsave(&sbi->cp_lock, flags);
1590 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1591 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1594 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1596 unsigned int ckpt_flags;
1598 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1600 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1603 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1605 unsigned long flags;
1607 spin_lock_irqsave(&sbi->cp_lock, flags);
1608 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1609 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1612 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1614 unsigned long flags;
1617 * In order to re-enable nat_bits we need to call fsck.f2fs by
1618 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1619 * so let's rely on regular fsck or unclean shutdown.
1623 spin_lock_irqsave(&sbi->cp_lock, flags);
1624 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1625 kvfree(NM_I(sbi)->nat_bits);
1626 NM_I(sbi)->nat_bits = NULL;
1628 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1631 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1632 struct cp_control *cpc)
1634 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1636 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1639 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1641 down_read(&sbi->cp_rwsem);
1644 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1646 return down_read_trylock(&sbi->cp_rwsem);
1649 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1651 up_read(&sbi->cp_rwsem);
1654 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1656 down_write(&sbi->cp_rwsem);
1659 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1661 up_write(&sbi->cp_rwsem);
1664 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1666 int reason = CP_SYNC;
1668 if (test_opt(sbi, FASTBOOT))
1669 reason = CP_FASTBOOT;
1670 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1675 static inline bool __remain_node_summaries(int reason)
1677 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1680 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1682 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1683 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1687 * Check whether the inode has blocks or not
1689 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1691 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1693 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1696 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1698 return ofs == XATTR_NODE_OFFSET;
1701 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1702 struct inode *inode, bool cap)
1706 if (!test_opt(sbi, RESERVE_ROOT))
1708 if (IS_NOQUOTA(inode))
1710 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1712 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1713 in_group_p(F2FS_OPTION(sbi).s_resgid))
1715 if (cap && capable(CAP_SYS_RESOURCE))
1720 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1721 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1722 struct inode *inode, blkcnt_t *count)
1724 blkcnt_t diff = 0, release = 0;
1725 block_t avail_user_block_count;
1728 ret = dquot_reserve_block(inode, *count);
1732 if (time_to_inject(sbi, FAULT_BLOCK)) {
1733 f2fs_show_injection_info(FAULT_BLOCK);
1739 * let's increase this in prior to actual block count change in order
1740 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1742 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1744 spin_lock(&sbi->stat_lock);
1745 sbi->total_valid_block_count += (block_t)(*count);
1746 avail_user_block_count = sbi->user_block_count -
1747 sbi->current_reserved_blocks;
1749 if (!__allow_reserved_blocks(sbi, inode, true))
1750 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1751 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1752 avail_user_block_count -= sbi->unusable_block_count;
1753 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1754 diff = sbi->total_valid_block_count - avail_user_block_count;
1759 sbi->total_valid_block_count -= diff;
1761 spin_unlock(&sbi->stat_lock);
1765 spin_unlock(&sbi->stat_lock);
1767 if (unlikely(release)) {
1768 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1769 dquot_release_reservation_block(inode, release);
1771 f2fs_i_blocks_write(inode, *count, true, true);
1775 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1776 dquot_release_reservation_block(inode, release);
1780 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1781 struct inode *inode,
1784 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1786 spin_lock(&sbi->stat_lock);
1787 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1788 f2fs_bug_on(sbi, inode->i_blocks < sectors);
1789 sbi->total_valid_block_count -= (block_t)count;
1790 if (sbi->reserved_blocks &&
1791 sbi->current_reserved_blocks < sbi->reserved_blocks)
1792 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1793 sbi->current_reserved_blocks + count);
1794 spin_unlock(&sbi->stat_lock);
1795 f2fs_i_blocks_write(inode, count, false, true);
1798 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1800 atomic_inc(&sbi->nr_pages[count_type]);
1802 if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES ||
1803 count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA ||
1804 count_type == F2FS_RD_DATA || count_type == F2FS_RD_NODE ||
1805 count_type == F2FS_RD_META)
1808 set_sbi_flag(sbi, SBI_IS_DIRTY);
1811 static inline void inode_inc_dirty_pages(struct inode *inode)
1813 atomic_inc(&F2FS_I(inode)->dirty_pages);
1814 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1815 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1816 if (IS_NOQUOTA(inode))
1817 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1820 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1822 atomic_dec(&sbi->nr_pages[count_type]);
1825 static inline void inode_dec_dirty_pages(struct inode *inode)
1827 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1828 !S_ISLNK(inode->i_mode))
1831 atomic_dec(&F2FS_I(inode)->dirty_pages);
1832 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1833 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1834 if (IS_NOQUOTA(inode))
1835 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1838 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1840 return atomic_read(&sbi->nr_pages[count_type]);
1843 static inline int get_dirty_pages(struct inode *inode)
1845 return atomic_read(&F2FS_I(inode)->dirty_pages);
1848 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1850 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1851 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1852 sbi->log_blocks_per_seg;
1854 return segs / sbi->segs_per_sec;
1857 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1859 return sbi->total_valid_block_count;
1862 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1864 return sbi->discard_blks;
1867 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1869 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1871 /* return NAT or SIT bitmap */
1872 if (flag == NAT_BITMAP)
1873 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1874 else if (flag == SIT_BITMAP)
1875 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1880 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1882 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1885 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1887 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1890 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1891 offset = (flag == SIT_BITMAP) ?
1892 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1893 return &ckpt->sit_nat_version_bitmap + offset;
1896 if (__cp_payload(sbi) > 0) {
1897 if (flag == NAT_BITMAP)
1898 return &ckpt->sit_nat_version_bitmap;
1900 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1902 offset = (flag == NAT_BITMAP) ?
1903 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1904 return &ckpt->sit_nat_version_bitmap + offset;
1908 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1910 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1912 if (sbi->cur_cp_pack == 2)
1913 start_addr += sbi->blocks_per_seg;
1917 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1919 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1921 if (sbi->cur_cp_pack == 1)
1922 start_addr += sbi->blocks_per_seg;
1926 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1928 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1931 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1933 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1936 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
1937 struct inode *inode, bool is_inode)
1939 block_t valid_block_count;
1940 unsigned int valid_node_count;
1945 err = dquot_alloc_inode(inode);
1950 err = dquot_reserve_block(inode, 1);
1955 if (time_to_inject(sbi, FAULT_BLOCK)) {
1956 f2fs_show_injection_info(FAULT_BLOCK);
1960 spin_lock(&sbi->stat_lock);
1962 valid_block_count = sbi->total_valid_block_count +
1963 sbi->current_reserved_blocks + 1;
1965 if (!__allow_reserved_blocks(sbi, inode, false))
1966 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
1967 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1968 valid_block_count += sbi->unusable_block_count;
1970 if (unlikely(valid_block_count > sbi->user_block_count)) {
1971 spin_unlock(&sbi->stat_lock);
1975 valid_node_count = sbi->total_valid_node_count + 1;
1976 if (unlikely(valid_node_count > sbi->total_node_count)) {
1977 spin_unlock(&sbi->stat_lock);
1981 sbi->total_valid_node_count++;
1982 sbi->total_valid_block_count++;
1983 spin_unlock(&sbi->stat_lock);
1987 f2fs_mark_inode_dirty_sync(inode, true);
1989 f2fs_i_blocks_write(inode, 1, true, true);
1992 percpu_counter_inc(&sbi->alloc_valid_block_count);
1998 dquot_free_inode(inode);
2000 dquot_release_reservation_block(inode, 1);
2005 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2006 struct inode *inode, bool is_inode)
2008 spin_lock(&sbi->stat_lock);
2010 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2011 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2012 f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
2014 sbi->total_valid_node_count--;
2015 sbi->total_valid_block_count--;
2016 if (sbi->reserved_blocks &&
2017 sbi->current_reserved_blocks < sbi->reserved_blocks)
2018 sbi->current_reserved_blocks++;
2020 spin_unlock(&sbi->stat_lock);
2023 dquot_free_inode(inode);
2025 f2fs_i_blocks_write(inode, 1, false, true);
2028 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2030 return sbi->total_valid_node_count;
2033 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2035 percpu_counter_inc(&sbi->total_valid_inode_count);
2038 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2040 percpu_counter_dec(&sbi->total_valid_inode_count);
2043 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2045 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2048 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2049 pgoff_t index, bool for_write)
2053 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2055 page = find_get_page_flags(mapping, index,
2056 FGP_LOCK | FGP_ACCESSED);
2058 page = find_lock_page(mapping, index);
2062 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2063 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2069 return grab_cache_page(mapping, index);
2070 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2073 static inline struct page *f2fs_pagecache_get_page(
2074 struct address_space *mapping, pgoff_t index,
2075 int fgp_flags, gfp_t gfp_mask)
2077 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2078 f2fs_show_injection_info(FAULT_PAGE_GET);
2082 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2085 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2087 char *src_kaddr = kmap(src);
2088 char *dst_kaddr = kmap(dst);
2090 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2095 static inline void f2fs_put_page(struct page *page, int unlock)
2101 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2107 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2110 f2fs_put_page(dn->node_page, 1);
2111 if (dn->inode_page && dn->node_page != dn->inode_page)
2112 f2fs_put_page(dn->inode_page, 0);
2113 dn->node_page = NULL;
2114 dn->inode_page = NULL;
2117 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2120 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2123 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2128 entry = kmem_cache_alloc(cachep, flags);
2130 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2134 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2135 int npages, bool no_fail)
2140 /* No failure on bio allocation */
2141 bio = bio_alloc(GFP_NOIO, npages);
2143 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2146 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2147 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2151 return bio_alloc(GFP_KERNEL, npages);
2154 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2156 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2157 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2158 get_pages(sbi, F2FS_WB_CP_DATA) ||
2159 get_pages(sbi, F2FS_DIO_READ) ||
2160 get_pages(sbi, F2FS_DIO_WRITE) ||
2161 atomic_read(&SM_I(sbi)->dcc_info->queued_discard) ||
2162 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2164 return f2fs_time_over(sbi, type);
2167 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2168 unsigned long index, void *item)
2170 while (radix_tree_insert(root, index, item))
2174 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2176 static inline bool IS_INODE(struct page *page)
2178 struct f2fs_node *p = F2FS_NODE(page);
2180 return RAW_IS_INODE(p);
2183 static inline int offset_in_addr(struct f2fs_inode *i)
2185 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2186 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2189 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2191 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2194 static inline int f2fs_has_extra_attr(struct inode *inode);
2195 static inline block_t datablock_addr(struct inode *inode,
2196 struct page *node_page, unsigned int offset)
2198 struct f2fs_node *raw_node;
2201 bool is_inode = IS_INODE(node_page);
2203 raw_node = F2FS_NODE(node_page);
2205 /* from GC path only */
2208 base = offset_in_addr(&raw_node->i);
2209 else if (f2fs_has_extra_attr(inode))
2210 base = get_extra_isize(inode);
2213 addr_array = blkaddr_in_node(raw_node);
2214 return le32_to_cpu(addr_array[base + offset]);
2217 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2222 mask = 1 << (7 - (nr & 0x07));
2223 return mask & *addr;
2226 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2231 mask = 1 << (7 - (nr & 0x07));
2235 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2240 mask = 1 << (7 - (nr & 0x07));
2244 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2250 mask = 1 << (7 - (nr & 0x07));
2256 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2262 mask = 1 << (7 - (nr & 0x07));
2268 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2273 mask = 1 << (7 - (nr & 0x07));
2280 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */
2281 #define F2FS_UNRM_FL 0x00000002 /* Undelete */
2282 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2283 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2284 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2285 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2286 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2287 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2288 /* Reserved for compression usage... */
2289 #define F2FS_DIRTY_FL 0x00000100
2290 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
2291 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */
2292 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */
2293 /* End compression flags --- maybe not all used */
2294 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2295 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */
2296 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
2297 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
2298 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2299 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
2300 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
2301 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */
2302 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */
2303 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
2304 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
2305 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2306 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
2308 #define F2FS_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
2309 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
2311 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */
2312 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \
2313 F2FS_IMMUTABLE_FL | \
2317 F2FS_PROJINHERIT_FL)
2319 /* Flags that should be inherited by new inodes from their parent. */
2320 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\
2321 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\
2322 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\
2323 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\
2324 F2FS_PROJINHERIT_FL)
2326 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2327 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL))
2329 /* Flags that are appropriate for non-directories/regular files. */
2330 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2332 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2336 else if (S_ISREG(mode))
2337 return flags & F2FS_REG_FLMASK;
2339 return flags & F2FS_OTHER_FLMASK;
2342 /* used for f2fs_inode_info->flags */
2344 FI_NEW_INODE, /* indicate newly allocated inode */
2345 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2346 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2347 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2348 FI_INC_LINK, /* need to increment i_nlink */
2349 FI_ACL_MODE, /* indicate acl mode */
2350 FI_NO_ALLOC, /* should not allocate any blocks */
2351 FI_FREE_NID, /* free allocated nide */
2352 FI_NO_EXTENT, /* not to use the extent cache */
2353 FI_INLINE_XATTR, /* used for inline xattr */
2354 FI_INLINE_DATA, /* used for inline data*/
2355 FI_INLINE_DENTRY, /* used for inline dentry */
2356 FI_APPEND_WRITE, /* inode has appended data */
2357 FI_UPDATE_WRITE, /* inode has in-place-update data */
2358 FI_NEED_IPU, /* used for ipu per file */
2359 FI_ATOMIC_FILE, /* indicate atomic file */
2360 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2361 FI_VOLATILE_FILE, /* indicate volatile file */
2362 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2363 FI_DROP_CACHE, /* drop dirty page cache */
2364 FI_DATA_EXIST, /* indicate data exists */
2365 FI_INLINE_DOTS, /* indicate inline dot dentries */
2366 FI_DO_DEFRAG, /* indicate defragment is running */
2367 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2368 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2369 FI_HOT_DATA, /* indicate file is hot */
2370 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2371 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2372 FI_PIN_FILE, /* indicate file should not be gced */
2373 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2376 static inline void __mark_inode_dirty_flag(struct inode *inode,
2380 case FI_INLINE_XATTR:
2381 case FI_INLINE_DATA:
2382 case FI_INLINE_DENTRY:
2388 case FI_INLINE_DOTS:
2390 f2fs_mark_inode_dirty_sync(inode, true);
2394 static inline void set_inode_flag(struct inode *inode, int flag)
2396 if (!test_bit(flag, &F2FS_I(inode)->flags))
2397 set_bit(flag, &F2FS_I(inode)->flags);
2398 __mark_inode_dirty_flag(inode, flag, true);
2401 static inline int is_inode_flag_set(struct inode *inode, int flag)
2403 return test_bit(flag, &F2FS_I(inode)->flags);
2406 static inline void clear_inode_flag(struct inode *inode, int flag)
2408 if (test_bit(flag, &F2FS_I(inode)->flags))
2409 clear_bit(flag, &F2FS_I(inode)->flags);
2410 __mark_inode_dirty_flag(inode, flag, false);
2413 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2415 F2FS_I(inode)->i_acl_mode = mode;
2416 set_inode_flag(inode, FI_ACL_MODE);
2417 f2fs_mark_inode_dirty_sync(inode, false);
2420 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2426 f2fs_mark_inode_dirty_sync(inode, true);
2429 static inline void f2fs_i_blocks_write(struct inode *inode,
2430 block_t diff, bool add, bool claim)
2432 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2433 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2435 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2438 dquot_claim_block(inode, diff);
2440 dquot_alloc_block_nofail(inode, diff);
2442 dquot_free_block(inode, diff);
2445 f2fs_mark_inode_dirty_sync(inode, true);
2446 if (clean || recover)
2447 set_inode_flag(inode, FI_AUTO_RECOVER);
2450 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2452 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2453 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2455 if (i_size_read(inode) == i_size)
2458 i_size_write(inode, i_size);
2459 f2fs_mark_inode_dirty_sync(inode, true);
2460 if (clean || recover)
2461 set_inode_flag(inode, FI_AUTO_RECOVER);
2464 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2466 F2FS_I(inode)->i_current_depth = depth;
2467 f2fs_mark_inode_dirty_sync(inode, true);
2470 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2473 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2474 f2fs_mark_inode_dirty_sync(inode, true);
2477 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2479 F2FS_I(inode)->i_xattr_nid = xnid;
2480 f2fs_mark_inode_dirty_sync(inode, true);
2483 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2485 F2FS_I(inode)->i_pino = pino;
2486 f2fs_mark_inode_dirty_sync(inode, true);
2489 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2491 struct f2fs_inode_info *fi = F2FS_I(inode);
2493 if (ri->i_inline & F2FS_INLINE_XATTR)
2494 set_bit(FI_INLINE_XATTR, &fi->flags);
2495 if (ri->i_inline & F2FS_INLINE_DATA)
2496 set_bit(FI_INLINE_DATA, &fi->flags);
2497 if (ri->i_inline & F2FS_INLINE_DENTRY)
2498 set_bit(FI_INLINE_DENTRY, &fi->flags);
2499 if (ri->i_inline & F2FS_DATA_EXIST)
2500 set_bit(FI_DATA_EXIST, &fi->flags);
2501 if (ri->i_inline & F2FS_INLINE_DOTS)
2502 set_bit(FI_INLINE_DOTS, &fi->flags);
2503 if (ri->i_inline & F2FS_EXTRA_ATTR)
2504 set_bit(FI_EXTRA_ATTR, &fi->flags);
2505 if (ri->i_inline & F2FS_PIN_FILE)
2506 set_bit(FI_PIN_FILE, &fi->flags);
2509 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2513 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2514 ri->i_inline |= F2FS_INLINE_XATTR;
2515 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2516 ri->i_inline |= F2FS_INLINE_DATA;
2517 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2518 ri->i_inline |= F2FS_INLINE_DENTRY;
2519 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2520 ri->i_inline |= F2FS_DATA_EXIST;
2521 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2522 ri->i_inline |= F2FS_INLINE_DOTS;
2523 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2524 ri->i_inline |= F2FS_EXTRA_ATTR;
2525 if (is_inode_flag_set(inode, FI_PIN_FILE))
2526 ri->i_inline |= F2FS_PIN_FILE;
2529 static inline int f2fs_has_extra_attr(struct inode *inode)
2531 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2534 static inline int f2fs_has_inline_xattr(struct inode *inode)
2536 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2539 static inline unsigned int addrs_per_inode(struct inode *inode)
2541 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
2544 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2546 struct f2fs_inode *ri = F2FS_INODE(page);
2548 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2549 get_inline_xattr_addrs(inode)]);
2552 static inline int inline_xattr_size(struct inode *inode)
2554 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2557 static inline int f2fs_has_inline_data(struct inode *inode)
2559 return is_inode_flag_set(inode, FI_INLINE_DATA);
2562 static inline int f2fs_exist_data(struct inode *inode)
2564 return is_inode_flag_set(inode, FI_DATA_EXIST);
2567 static inline int f2fs_has_inline_dots(struct inode *inode)
2569 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2572 static inline bool f2fs_is_pinned_file(struct inode *inode)
2574 return is_inode_flag_set(inode, FI_PIN_FILE);
2577 static inline bool f2fs_is_atomic_file(struct inode *inode)
2579 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2582 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2584 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2587 static inline bool f2fs_is_volatile_file(struct inode *inode)
2589 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2592 static inline bool f2fs_is_first_block_written(struct inode *inode)
2594 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2597 static inline bool f2fs_is_drop_cache(struct inode *inode)
2599 return is_inode_flag_set(inode, FI_DROP_CACHE);
2602 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2604 struct f2fs_inode *ri = F2FS_INODE(page);
2605 int extra_size = get_extra_isize(inode);
2607 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2610 static inline int f2fs_has_inline_dentry(struct inode *inode)
2612 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2615 static inline int is_file(struct inode *inode, int type)
2617 return F2FS_I(inode)->i_advise & type;
2620 static inline void set_file(struct inode *inode, int type)
2622 F2FS_I(inode)->i_advise |= type;
2623 f2fs_mark_inode_dirty_sync(inode, true);
2626 static inline void clear_file(struct inode *inode, int type)
2628 F2FS_I(inode)->i_advise &= ~type;
2629 f2fs_mark_inode_dirty_sync(inode, true);
2632 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2637 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2639 spin_lock(&sbi->inode_lock[DIRTY_META]);
2640 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2641 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2644 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2645 file_keep_isize(inode) ||
2646 i_size_read(inode) & ~PAGE_MASK)
2649 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2651 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2653 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2655 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2656 &F2FS_I(inode)->i_crtime))
2659 down_read(&F2FS_I(inode)->i_sem);
2660 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2661 up_read(&F2FS_I(inode)->i_sem);
2666 static inline bool f2fs_readonly(struct super_block *sb)
2668 return sb_rdonly(sb);
2671 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2673 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2676 static inline bool is_dot_dotdot(const struct qstr *str)
2678 if (str->len == 1 && str->name[0] == '.')
2681 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2687 static inline bool f2fs_may_extent_tree(struct inode *inode)
2689 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2691 if (!test_opt(sbi, EXTENT_CACHE) ||
2692 is_inode_flag_set(inode, FI_NO_EXTENT))
2696 * for recovered files during mount do not create extents
2697 * if shrinker is not registered.
2699 if (list_empty(&sbi->s_list))
2702 return S_ISREG(inode->i_mode);
2705 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2706 size_t size, gfp_t flags)
2710 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2711 f2fs_show_injection_info(FAULT_KMALLOC);
2715 ret = kmalloc(size, flags);
2719 return kvmalloc(size, flags);
2722 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2723 size_t size, gfp_t flags)
2725 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2728 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2729 size_t size, gfp_t flags)
2731 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2732 f2fs_show_injection_info(FAULT_KVMALLOC);
2736 return kvmalloc(size, flags);
2739 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2740 size_t size, gfp_t flags)
2742 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2745 static inline int get_extra_isize(struct inode *inode)
2747 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2750 static inline int get_inline_xattr_addrs(struct inode *inode)
2752 return F2FS_I(inode)->i_inline_xattr_size;
2755 #define f2fs_get_inode_mode(i) \
2756 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2757 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2759 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2760 (offsetof(struct f2fs_inode, i_extra_end) - \
2761 offsetof(struct f2fs_inode, i_extra_isize)) \
2763 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2764 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2765 ((offsetof(typeof(*f2fs_inode), field) + \
2766 sizeof((f2fs_inode)->field)) \
2767 <= (F2FS_OLD_ATTRIBUTE_SIZE + extra_isize)) \
2769 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2773 spin_lock(&sbi->iostat_lock);
2774 for (i = 0; i < NR_IO_TYPE; i++)
2775 sbi->write_iostat[i] = 0;
2776 spin_unlock(&sbi->iostat_lock);
2779 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2780 enum iostat_type type, unsigned long long io_bytes)
2782 if (!sbi->iostat_enable)
2784 spin_lock(&sbi->iostat_lock);
2785 sbi->write_iostat[type] += io_bytes;
2787 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2788 sbi->write_iostat[APP_BUFFERED_IO] =
2789 sbi->write_iostat[APP_WRITE_IO] -
2790 sbi->write_iostat[APP_DIRECT_IO];
2791 spin_unlock(&sbi->iostat_lock);
2794 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2796 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO(fio->type) == META && \
2797 (!is_read_io(fio->op) || fio->is_meta))
2799 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2800 block_t blkaddr, int type);
2801 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2802 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2803 block_t blkaddr, int type)
2805 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2806 f2fs_msg(sbi->sb, KERN_ERR,
2807 "invalid blkaddr: %u, type: %d, run fsck to fix.",
2809 f2fs_bug_on(sbi, 1);
2813 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2815 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2820 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2823 if (!__is_valid_data_blkaddr(blkaddr))
2825 verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2832 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2833 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2834 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock,
2836 int f2fs_truncate(struct inode *inode);
2837 int f2fs_getattr(const struct path *path, struct kstat *stat,
2838 u32 request_mask, unsigned int flags);
2839 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2840 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2841 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2842 int f2fs_precache_extents(struct inode *inode);
2843 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2844 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2845 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
2846 int f2fs_pin_file_control(struct inode *inode, bool inc);
2851 void f2fs_set_inode_flags(struct inode *inode);
2852 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2853 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2854 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2855 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2856 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2857 void f2fs_update_inode(struct inode *inode, struct page *node_page);
2858 void f2fs_update_inode_page(struct inode *inode);
2859 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2860 void f2fs_evict_inode(struct inode *inode);
2861 void f2fs_handle_failed_inode(struct inode *inode);
2866 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2867 bool hot, bool set);
2868 struct dentry *f2fs_get_parent(struct dentry *child);
2873 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2874 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2875 f2fs_hash_t namehash, int *max_slots,
2876 struct f2fs_dentry_ptr *d);
2877 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2878 unsigned int start_pos, struct fscrypt_str *fstr);
2879 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2880 struct f2fs_dentry_ptr *d);
2881 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2882 const struct qstr *new_name,
2883 const struct qstr *orig_name, struct page *dpage);
2884 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2885 unsigned int current_depth);
2886 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2887 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2888 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2889 struct fscrypt_name *fname, struct page **res_page);
2890 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2891 const struct qstr *child, struct page **res_page);
2892 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2893 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2894 struct page **page);
2895 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2896 struct page *page, struct inode *inode);
2897 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2898 const struct qstr *name, f2fs_hash_t name_hash,
2899 unsigned int bit_pos);
2900 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2901 const struct qstr *orig_name,
2902 struct inode *inode, nid_t ino, umode_t mode);
2903 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2904 struct inode *inode, nid_t ino, umode_t mode);
2905 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2906 struct inode *inode, nid_t ino, umode_t mode);
2907 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2908 struct inode *dir, struct inode *inode);
2909 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2910 bool f2fs_empty_dir(struct inode *dir);
2912 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2914 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2915 inode, inode->i_ino, inode->i_mode);
2921 int f2fs_inode_dirtied(struct inode *inode, bool sync);
2922 void f2fs_inode_synced(struct inode *inode);
2923 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
2924 int f2fs_quota_sync(struct super_block *sb, int type);
2925 void f2fs_quota_off_umount(struct super_block *sb);
2926 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
2927 int f2fs_sync_fs(struct super_block *sb, int sync);
2928 extern __printf(3, 4)
2929 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2930 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
2935 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info,
2936 struct fscrypt_name *fname);
2941 struct dnode_of_data;
2944 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
2945 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
2946 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
2947 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
2948 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
2949 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
2950 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
2951 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
2952 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
2953 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
2954 struct node_info *ni);
2955 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
2956 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
2957 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
2958 int f2fs_truncate_xattr_node(struct inode *inode);
2959 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
2960 unsigned int seq_id);
2961 int f2fs_remove_inode_page(struct inode *inode);
2962 struct page *f2fs_new_inode_page(struct inode *inode);
2963 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
2964 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
2965 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
2966 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
2967 int f2fs_move_node_page(struct page *node_page, int gc_type);
2968 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
2969 struct writeback_control *wbc, bool atomic,
2970 unsigned int *seq_id);
2971 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
2972 struct writeback_control *wbc,
2973 bool do_balance, enum iostat_type io_type);
2974 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
2975 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
2976 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
2977 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
2978 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
2979 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
2980 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
2981 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
2982 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
2983 unsigned int segno, struct f2fs_summary_block *sum);
2984 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2985 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
2986 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
2987 int __init f2fs_create_node_manager_caches(void);
2988 void f2fs_destroy_node_manager_caches(void);
2993 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
2994 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
2995 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
2996 void f2fs_drop_inmem_pages(struct inode *inode);
2997 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
2998 int f2fs_commit_inmem_pages(struct inode *inode);
2999 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3000 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
3001 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3002 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3003 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3004 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3005 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3006 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3007 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3008 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3009 bool f2fs_wait_discard_bios(struct f2fs_sb_info *sbi);
3010 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3011 struct cp_control *cpc);
3012 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3013 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi);
3014 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3015 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3016 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3017 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3018 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3019 struct cp_control *cpc);
3020 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3021 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3023 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3024 enum iostat_type io_type);
3025 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3026 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3027 struct f2fs_io_info *fio);
3028 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3029 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3030 block_t old_blkaddr, block_t new_blkaddr,
3031 bool recover_curseg, bool recover_newaddr);
3032 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3033 block_t old_addr, block_t new_addr,
3034 unsigned char version, bool recover_curseg,
3035 bool recover_newaddr);
3036 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3037 block_t old_blkaddr, block_t *new_blkaddr,
3038 struct f2fs_summary *sum, int type,
3039 struct f2fs_io_info *fio, bool add_list);
3040 void f2fs_wait_on_page_writeback(struct page *page,
3041 enum page_type type, bool ordered, bool locked);
3042 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3043 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3045 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3046 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3047 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3048 unsigned int val, int alloc);
3049 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3050 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3051 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3052 int __init f2fs_create_segment_manager_caches(void);
3053 void f2fs_destroy_segment_manager_caches(void);
3054 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3055 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3056 enum page_type type, enum temp_type temp);
3061 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3062 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3063 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3064 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3065 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3066 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3067 block_t blkaddr, int type);
3068 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3069 int type, bool sync);
3070 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3071 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3072 long nr_to_write, enum iostat_type io_type);
3073 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3074 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3075 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3076 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3077 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3078 unsigned int devidx, int type);
3079 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3080 unsigned int devidx, int type);
3081 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3082 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3083 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3084 void f2fs_add_orphan_inode(struct inode *inode);
3085 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3086 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3087 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3088 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3089 void f2fs_remove_dirty_inode(struct inode *inode);
3090 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3091 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3092 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3093 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3094 int __init f2fs_create_checkpoint_caches(void);
3095 void f2fs_destroy_checkpoint_caches(void);
3100 int f2fs_init_post_read_processing(void);
3101 void f2fs_destroy_post_read_processing(void);
3102 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3103 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3104 struct inode *inode, struct page *page,
3105 nid_t ino, enum page_type type);
3106 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3107 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3108 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3109 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3110 block_t blk_addr, struct bio *bio);
3111 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3112 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3113 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3114 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3115 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3116 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3117 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3118 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3119 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3120 int op_flags, bool for_write);
3121 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3122 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3124 struct page *f2fs_get_new_data_page(struct inode *inode,
3125 struct page *ipage, pgoff_t index, bool new_i_size);
3126 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3127 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3128 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3129 int create, int flag);
3130 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3131 u64 start, u64 len);
3132 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3133 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3134 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3135 unsigned int length);
3136 int f2fs_release_page(struct page *page, gfp_t wait);
3137 #ifdef CONFIG_MIGRATION
3138 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3139 struct page *page, enum migrate_mode mode);
3141 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3142 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3147 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3148 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3149 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3150 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3151 unsigned int segno);
3152 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3157 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3158 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3163 #ifdef CONFIG_F2FS_STAT_FS
3164 struct f2fs_stat_info {
3165 struct list_head stat_list;
3166 struct f2fs_sb_info *sbi;
3167 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3168 int main_area_segs, main_area_sections, main_area_zones;
3169 unsigned long long hit_largest, hit_cached, hit_rbtree;
3170 unsigned long long hit_total, total_ext;
3171 int ext_tree, zombie_tree, ext_node;
3172 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3173 int ndirty_data, ndirty_qdata;
3175 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3176 int nats, dirty_nats, sits, dirty_sits;
3177 int free_nids, avail_nids, alloc_nids;
3178 int total_count, utilization;
3179 int bg_gc, nr_wb_cp_data, nr_wb_data;
3180 int nr_rd_data, nr_rd_node, nr_rd_meta;
3181 int nr_dio_read, nr_dio_write;
3182 unsigned int io_skip_bggc, other_skip_bggc;
3183 int nr_flushing, nr_flushed, flush_list_empty;
3184 int nr_discarding, nr_discarded;
3186 unsigned int undiscard_blks;
3187 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3188 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3189 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3190 unsigned int bimodal, avg_vblocks;
3191 int util_free, util_valid, util_invalid;
3192 int rsvd_segs, overp_segs;
3193 int dirty_count, node_pages, meta_pages;
3194 int prefree_count, call_count, cp_count, bg_cp_count;
3195 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3196 int bg_node_segs, bg_data_segs;
3197 int tot_blks, data_blks, node_blks;
3198 int bg_data_blks, bg_node_blks;
3199 unsigned long long skipped_atomic_files[2];
3200 int curseg[NR_CURSEG_TYPE];
3201 int cursec[NR_CURSEG_TYPE];
3202 int curzone[NR_CURSEG_TYPE];
3204 unsigned int meta_count[META_MAX];
3205 unsigned int segment_count[2];
3206 unsigned int block_count[2];
3207 unsigned int inplace_count;
3208 unsigned long long base_mem, cache_mem, page_mem;
3211 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3213 return (struct f2fs_stat_info *)sbi->stat_info;
3216 #define stat_inc_cp_count(si) ((si)->cp_count++)
3217 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3218 #define stat_inc_call_count(si) ((si)->call_count++)
3219 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3220 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3221 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3222 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3223 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3224 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3225 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3226 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3227 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3228 #define stat_inc_inline_xattr(inode) \
3230 if (f2fs_has_inline_xattr(inode)) \
3231 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3233 #define stat_dec_inline_xattr(inode) \
3235 if (f2fs_has_inline_xattr(inode)) \
3236 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3238 #define stat_inc_inline_inode(inode) \
3240 if (f2fs_has_inline_data(inode)) \
3241 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3243 #define stat_dec_inline_inode(inode) \
3245 if (f2fs_has_inline_data(inode)) \
3246 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3248 #define stat_inc_inline_dir(inode) \
3250 if (f2fs_has_inline_dentry(inode)) \
3251 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3253 #define stat_dec_inline_dir(inode) \
3255 if (f2fs_has_inline_dentry(inode)) \
3256 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3258 #define stat_inc_meta_count(sbi, blkaddr) \
3260 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3261 atomic_inc(&(sbi)->meta_count[META_CP]); \
3262 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3263 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3264 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3265 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3266 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3267 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3269 #define stat_inc_seg_type(sbi, curseg) \
3270 ((sbi)->segment_count[(curseg)->alloc_type]++)
3271 #define stat_inc_block_count(sbi, curseg) \
3272 ((sbi)->block_count[(curseg)->alloc_type]++)
3273 #define stat_inc_inplace_blocks(sbi) \
3274 (atomic_inc(&(sbi)->inplace_count))
3275 #define stat_inc_atomic_write(inode) \
3276 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3277 #define stat_dec_atomic_write(inode) \
3278 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3279 #define stat_update_max_atomic_write(inode) \
3281 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3282 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3284 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3286 #define stat_inc_volatile_write(inode) \
3287 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3288 #define stat_dec_volatile_write(inode) \
3289 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3290 #define stat_update_max_volatile_write(inode) \
3292 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3293 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3295 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3297 #define stat_inc_seg_count(sbi, type, gc_type) \
3299 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3301 if ((type) == SUM_TYPE_DATA) { \
3303 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3306 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3310 #define stat_inc_tot_blk_count(si, blks) \
3311 ((si)->tot_blks += (blks))
3313 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3315 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3316 stat_inc_tot_blk_count(si, blks); \
3317 si->data_blks += (blks); \
3318 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3321 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3323 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3324 stat_inc_tot_blk_count(si, blks); \
3325 si->node_blks += (blks); \
3326 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3329 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3330 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3331 int __init f2fs_create_root_stats(void);
3332 void f2fs_destroy_root_stats(void);
3334 #define stat_inc_cp_count(si) do { } while (0)
3335 #define stat_inc_bg_cp_count(si) do { } while (0)
3336 #define stat_inc_call_count(si) do { } while (0)
3337 #define stat_inc_bggc_count(si) do { } while (0)
3338 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3339 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3340 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3341 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3342 #define stat_inc_total_hit(sb) do { } while (0)
3343 #define stat_inc_rbtree_node_hit(sb) do { } while (0)
3344 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3345 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3346 #define stat_inc_inline_xattr(inode) do { } while (0)
3347 #define stat_dec_inline_xattr(inode) do { } while (0)
3348 #define stat_inc_inline_inode(inode) do { } while (0)
3349 #define stat_dec_inline_inode(inode) do { } while (0)
3350 #define stat_inc_inline_dir(inode) do { } while (0)
3351 #define stat_dec_inline_dir(inode) do { } while (0)
3352 #define stat_inc_atomic_write(inode) do { } while (0)
3353 #define stat_dec_atomic_write(inode) do { } while (0)
3354 #define stat_update_max_atomic_write(inode) do { } while (0)
3355 #define stat_inc_volatile_write(inode) do { } while (0)
3356 #define stat_dec_volatile_write(inode) do { } while (0)
3357 #define stat_update_max_volatile_write(inode) do { } while (0)
3358 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3359 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3360 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3361 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3362 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3363 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3364 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3365 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3367 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3368 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3369 static inline int __init f2fs_create_root_stats(void) { return 0; }
3370 static inline void f2fs_destroy_root_stats(void) { }
3373 extern const struct file_operations f2fs_dir_operations;
3374 extern const struct file_operations f2fs_file_operations;
3375 extern const struct inode_operations f2fs_file_inode_operations;
3376 extern const struct address_space_operations f2fs_dblock_aops;
3377 extern const struct address_space_operations f2fs_node_aops;
3378 extern const struct address_space_operations f2fs_meta_aops;
3379 extern const struct inode_operations f2fs_dir_inode_operations;
3380 extern const struct inode_operations f2fs_symlink_inode_operations;
3381 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3382 extern const struct inode_operations f2fs_special_inode_operations;
3383 extern struct kmem_cache *f2fs_inode_entry_slab;
3388 bool f2fs_may_inline_data(struct inode *inode);
3389 bool f2fs_may_inline_dentry(struct inode *inode);
3390 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3391 void f2fs_truncate_inline_inode(struct inode *inode,
3392 struct page *ipage, u64 from);
3393 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3394 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3395 int f2fs_convert_inline_inode(struct inode *inode);
3396 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3397 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3398 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3399 struct fscrypt_name *fname, struct page **res_page);
3400 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3401 struct page *ipage);
3402 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3403 const struct qstr *orig_name,
3404 struct inode *inode, nid_t ino, umode_t mode);
3405 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3406 struct page *page, struct inode *dir,
3407 struct inode *inode);
3408 bool f2fs_empty_inline_dir(struct inode *dir);
3409 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3410 struct fscrypt_str *fstr);
3411 int f2fs_inline_data_fiemap(struct inode *inode,
3412 struct fiemap_extent_info *fieinfo,
3413 __u64 start, __u64 len);
3418 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3419 struct shrink_control *sc);
3420 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3421 struct shrink_control *sc);
3422 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3423 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3428 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3429 struct rb_entry *cached_re, unsigned int ofs);
3430 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3431 struct rb_root_cached *root,
3432 struct rb_node **parent,
3433 unsigned int ofs, bool *leftmost);
3434 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3435 struct rb_entry *cached_re, unsigned int ofs,
3436 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3437 struct rb_node ***insert_p, struct rb_node **insert_parent,
3438 bool force, bool *leftmost);
3439 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3440 struct rb_root_cached *root);
3441 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3442 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3443 void f2fs_drop_extent_tree(struct inode *inode);
3444 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3445 void f2fs_destroy_extent_tree(struct inode *inode);
3446 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3447 struct extent_info *ei);
3448 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3449 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3450 pgoff_t fofs, block_t blkaddr, unsigned int len);
3451 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3452 int __init f2fs_create_extent_cache(void);
3453 void f2fs_destroy_extent_cache(void);
3458 int __init f2fs_init_sysfs(void);
3459 void f2fs_exit_sysfs(void);
3460 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3461 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3466 static inline bool f2fs_encrypted_inode(struct inode *inode)
3468 return file_is_encrypt(inode);
3471 static inline bool f2fs_encrypted_file(struct inode *inode)
3473 return f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode);
3476 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3478 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3479 file_set_encrypt(inode);
3480 f2fs_set_inode_flags(inode);
3485 * Returns true if the reads of the inode's data need to undergo some
3486 * postprocessing step, like decryption or authenticity verification.
3488 static inline bool f2fs_post_read_required(struct inode *inode)
3490 return f2fs_encrypted_file(inode);
3493 #define F2FS_FEATURE_FUNCS(name, flagname) \
3494 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3496 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3499 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3500 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3501 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3502 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3503 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3504 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3505 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3506 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3507 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3508 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3510 #ifdef CONFIG_BLK_DEV_ZONED
3511 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3512 struct block_device *bdev, block_t blkaddr)
3514 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3517 for (i = 0; i < sbi->s_ndevs; i++)
3518 if (FDEV(i).bdev == bdev)
3519 return FDEV(i).blkz_type[zno];
3524 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3526 return f2fs_sb_has_blkzoned(sbi);
3529 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3531 return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
3534 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3536 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3537 f2fs_hw_should_discard(sbi);
3540 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3542 clear_opt(sbi, ADAPTIVE);
3543 clear_opt(sbi, LFS);
3546 case F2FS_MOUNT_ADAPTIVE:
3547 set_opt(sbi, ADAPTIVE);
3549 case F2FS_MOUNT_LFS:
3555 static inline bool f2fs_may_encrypt(struct inode *inode)
3557 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3558 umode_t mode = inode->i_mode;
3560 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3566 static inline int block_unaligned_IO(struct inode *inode,
3567 struct kiocb *iocb, struct iov_iter *iter)
3569 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
3570 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
3571 loff_t offset = iocb->ki_pos;
3572 unsigned long align = offset | iov_iter_alignment(iter);
3574 return align & blocksize_mask;
3577 static inline int allow_outplace_dio(struct inode *inode,
3578 struct kiocb *iocb, struct iov_iter *iter)
3580 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3581 int rw = iov_iter_rw(iter);
3583 return (test_opt(sbi, LFS) && (rw == WRITE) &&
3584 !block_unaligned_IO(inode, iocb, iter));
3587 static inline bool f2fs_force_buffered_io(struct inode *inode,
3588 struct kiocb *iocb, struct iov_iter *iter)
3590 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3591 int rw = iov_iter_rw(iter);
3593 if (f2fs_post_read_required(inode))
3598 * for blkzoned device, fallback direct IO to buffered IO, so
3599 * all IOs can be serialized by log-structured write.
3601 if (f2fs_sb_has_blkzoned(sbi))
3603 if (test_opt(sbi, LFS) && (rw == WRITE) &&
3604 block_unaligned_IO(inode, iocb, iter))
3606 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
3612 #ifdef CONFIG_F2FS_FAULT_INJECTION
3613 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3616 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3621 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
3624 if (f2fs_sb_has_quota_ino(sbi))
3626 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
3627 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
3628 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])