4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
14 #include <linux/types.h>
15 #include <linux/page-flags.h>
16 #include <linux/buffer_head.h>
17 #include <linux/slab.h>
18 #include <linux/crc32.h>
19 #include <linux/magic.h>
20 #include <linux/kobject.h>
21 #include <linux/sched.h>
22 #include <linux/cred.h>
23 #include <linux/vmalloc.h>
24 #include <linux/bio.h>
25 #include <linux/blkdev.h>
26 #include <linux/quotaops.h>
27 #include <crypto/hash.h>
29 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_F2FS_FS_ENCRYPTION)
30 #include <linux/fscrypt.h>
32 #ifdef CONFIG_F2FS_CHECK_FS
33 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
35 #define f2fs_bug_on(sbi, condition) \
37 if (unlikely(condition)) { \
39 set_sbi_flag(sbi, SBI_NEED_FSCK); \
62 #ifdef CONFIG_F2FS_FAULT_INJECTION
63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
65 struct f2fs_fault_info {
67 unsigned int inject_rate;
68 unsigned int inject_type;
71 extern char *f2fs_fault_name[FAULT_MAX];
72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
78 #define F2FS_MOUNT_BG_GC 0x00000001
79 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
80 #define F2FS_MOUNT_DISCARD 0x00000004
81 #define F2FS_MOUNT_NOHEAP 0x00000008
82 #define F2FS_MOUNT_XATTR_USER 0x00000010
83 #define F2FS_MOUNT_POSIX_ACL 0x00000020
84 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
85 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
86 #define F2FS_MOUNT_INLINE_DATA 0x00000100
87 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
88 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
89 #define F2FS_MOUNT_NOBARRIER 0x00000800
90 #define F2FS_MOUNT_FASTBOOT 0x00001000
91 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
92 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000
93 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
94 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
95 #define F2FS_MOUNT_ADAPTIVE 0x00020000
96 #define F2FS_MOUNT_LFS 0x00040000
97 #define F2FS_MOUNT_USRQUOTA 0x00080000
98 #define F2FS_MOUNT_GRPQUOTA 0x00100000
99 #define F2FS_MOUNT_PRJQUOTA 0x00200000
100 #define F2FS_MOUNT_QUOTA 0x00400000
101 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
102 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
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 */
154 #define F2FS_HAS_FEATURE(sb, mask) \
155 ((F2FS_SB(sb)->raw_super->feature & cpu_to_le32(mask)) != 0)
156 #define F2FS_SET_FEATURE(sb, mask) \
157 (F2FS_SB(sb)->raw_super->feature |= cpu_to_le32(mask))
158 #define F2FS_CLEAR_FEATURE(sb, mask) \
159 (F2FS_SB(sb)->raw_super->feature &= ~cpu_to_le32(mask))
162 * Default values for user and/or group using reserved blocks
164 #define F2FS_DEF_RESUID 0
165 #define F2FS_DEF_RESGID 0
168 * For checkpoint manager
175 #define CP_UMOUNT 0x00000001
176 #define CP_FASTBOOT 0x00000002
177 #define CP_SYNC 0x00000004
178 #define CP_RECOVERY 0x00000008
179 #define CP_DISCARD 0x00000010
180 #define CP_TRIMMED 0x00000020
182 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
183 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
184 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
185 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
186 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
187 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
188 #define DEF_CP_INTERVAL 60 /* 60 secs */
189 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
199 * indicate meta/data type
211 /* for the list of ino */
213 ORPHAN_INO, /* for orphan ino list */
214 APPEND_INO, /* for append ino list */
215 UPDATE_INO, /* for update ino list */
216 TRANS_DIR_INO, /* for trasactions dir ino list */
217 FLUSH_INO, /* for multiple device flushing */
218 MAX_INO_ENTRY, /* max. list */
222 struct list_head list; /* list head */
223 nid_t ino; /* inode number */
224 unsigned int dirty_device; /* dirty device bitmap */
227 /* for the list of inodes to be GCed */
229 struct list_head list; /* list head */
230 struct inode *inode; /* vfs inode pointer */
233 struct fsync_node_entry {
234 struct list_head list; /* list head */
235 struct page *page; /* warm node page pointer */
236 unsigned int seq_id; /* sequence id */
239 /* for the bitmap indicate blocks to be discarded */
240 struct discard_entry {
241 struct list_head list; /* list head */
242 block_t start_blkaddr; /* start blockaddr of current segment */
243 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
246 /* default discard granularity of inner discard thread, unit: block count */
247 #define DEFAULT_DISCARD_GRANULARITY 16
249 /* max discard pend list number */
250 #define MAX_PLIST_NUM 512
251 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
252 (MAX_PLIST_NUM - 1) : (blk_num - 1))
255 D_PREP, /* initial */
256 D_PARTIAL, /* partially submitted */
257 D_SUBMIT, /* all submitted */
258 D_DONE, /* finished */
261 struct discard_info {
262 block_t lstart; /* logical start address */
263 block_t len; /* length */
264 block_t start; /* actual start address in dev */
268 struct rb_node rb_node; /* rb node located in rb-tree */
271 block_t lstart; /* logical start address */
272 block_t len; /* length */
273 block_t start; /* actual start address in dev */
275 struct discard_info di; /* discard info */
278 struct list_head list; /* command list */
279 struct completion wait; /* compleation */
280 struct block_device *bdev; /* bdev */
281 unsigned short ref; /* reference count */
282 unsigned char state; /* state */
283 unsigned char issuing; /* issuing discard */
284 int error; /* bio error */
285 spinlock_t lock; /* for state/bio_ref updating */
286 unsigned short bio_ref; /* bio reference count */
297 struct discard_policy {
298 int type; /* type of discard */
299 unsigned int min_interval; /* used for candidates exist */
300 unsigned int mid_interval; /* used for device busy */
301 unsigned int max_interval; /* used for candidates not exist */
302 unsigned int max_requests; /* # of discards issued per round */
303 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
304 bool io_aware; /* issue discard in idle time */
305 bool sync; /* submit discard with REQ_SYNC flag */
306 bool ordered; /* issue discard by lba order */
307 unsigned int granularity; /* discard granularity */
310 struct discard_cmd_control {
311 struct task_struct *f2fs_issue_discard; /* discard thread */
312 struct list_head entry_list; /* 4KB discard entry list */
313 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
314 struct list_head wait_list; /* store on-flushing entries */
315 struct list_head fstrim_list; /* in-flight discard from fstrim */
316 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
317 unsigned int discard_wake; /* to wake up discard thread */
318 struct mutex cmd_lock;
319 unsigned int nr_discards; /* # of discards in the list */
320 unsigned int max_discards; /* max. discards to be issued */
321 unsigned int discard_granularity; /* discard granularity */
322 unsigned int undiscard_blks; /* # of undiscard blocks */
323 unsigned int next_pos; /* next discard position */
324 atomic_t issued_discard; /* # of issued discard */
325 atomic_t issing_discard; /* # of issing discard */
326 atomic_t discard_cmd_cnt; /* # of cached cmd count */
327 struct rb_root root; /* root of discard rb-tree */
328 bool rbtree_check; /* config for consistence check */
331 /* for the list of fsync inodes, used only during recovery */
332 struct fsync_inode_entry {
333 struct list_head list; /* list head */
334 struct inode *inode; /* vfs inode pointer */
335 block_t blkaddr; /* block address locating the last fsync */
336 block_t last_dentry; /* block address locating the last dentry */
339 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
340 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
342 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
343 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
344 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
345 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
347 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
348 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
350 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
352 int before = nats_in_cursum(journal);
354 journal->n_nats = cpu_to_le16(before + i);
358 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
360 int before = sits_in_cursum(journal);
362 journal->n_sits = cpu_to_le16(before + i);
366 static inline bool __has_cursum_space(struct f2fs_journal *journal,
369 if (type == NAT_JOURNAL)
370 return size <= MAX_NAT_JENTRIES(journal);
371 return size <= MAX_SIT_JENTRIES(journal);
377 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
378 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
379 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
381 #define F2FS_IOCTL_MAGIC 0xf5
382 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
383 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
384 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
385 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
386 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
387 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
388 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
389 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
390 struct f2fs_defragment)
391 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
392 struct f2fs_move_range)
393 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
394 struct f2fs_flush_device)
395 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
396 struct f2fs_gc_range)
397 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
398 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
399 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
400 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
402 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
403 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
404 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
407 * should be same as XFS_IOC_GOINGDOWN.
408 * Flags for going down operation used by FS_IOC_GOINGDOWN
410 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
411 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
412 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
413 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
414 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
416 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
418 * ioctl commands in 32 bit emulation
420 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
421 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
422 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
425 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
426 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
428 struct f2fs_gc_range {
434 struct f2fs_defragment {
439 struct f2fs_move_range {
440 u32 dst_fd; /* destination fd */
441 u64 pos_in; /* start position in src_fd */
442 u64 pos_out; /* start position in dst_fd */
443 u64 len; /* size to move */
446 struct f2fs_flush_device {
447 u32 dev_num; /* device number to flush */
448 u32 segments; /* # of segments to flush */
451 /* for inline stuff */
452 #define DEF_INLINE_RESERVED_SIZE 1
453 static inline int get_extra_isize(struct inode *inode);
454 static inline int get_inline_xattr_addrs(struct inode *inode);
455 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
456 (CUR_ADDRS_PER_INODE(inode) - \
457 get_inline_xattr_addrs(inode) - \
458 DEF_INLINE_RESERVED_SIZE))
461 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
462 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
464 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \
465 BITS_PER_BYTE - 1) / BITS_PER_BYTE)
466 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
467 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
468 NR_INLINE_DENTRY(inode) + \
469 INLINE_DENTRY_BITMAP_SIZE(inode)))
472 * For INODE and NODE manager
474 /* for directory operations */
475 struct f2fs_dentry_ptr {
478 struct f2fs_dir_entry *dentry;
479 __u8 (*filename)[F2FS_SLOT_LEN];
484 static inline void make_dentry_ptr_block(struct inode *inode,
485 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
488 d->max = NR_DENTRY_IN_BLOCK;
489 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
490 d->bitmap = t->dentry_bitmap;
491 d->dentry = t->dentry;
492 d->filename = t->filename;
495 static inline void make_dentry_ptr_inline(struct inode *inode,
496 struct f2fs_dentry_ptr *d, void *t)
498 int entry_cnt = NR_INLINE_DENTRY(inode);
499 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
500 int reserved_size = INLINE_RESERVED_SIZE(inode);
504 d->nr_bitmap = bitmap_size;
506 d->dentry = t + bitmap_size + reserved_size;
507 d->filename = t + bitmap_size + reserved_size +
508 SIZE_OF_DIR_ENTRY * entry_cnt;
512 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
513 * as its node offset to distinguish from index node blocks.
514 * But some bits are used to mark the node block.
516 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
519 ALLOC_NODE, /* allocate a new node page if needed */
520 LOOKUP_NODE, /* look up a node without readahead */
522 * look up a node with readahead called
527 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
529 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
531 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
533 /* for in-memory extent cache entry */
534 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
536 /* number of extent info in extent cache we try to shrink */
537 #define EXTENT_CACHE_SHRINK_NUMBER 128
540 struct rb_node rb_node; /* rb node located in rb-tree */
541 unsigned int ofs; /* start offset of the entry */
542 unsigned int len; /* length of the entry */
546 unsigned int fofs; /* start offset in a file */
547 unsigned int len; /* length of the extent */
548 u32 blk; /* start block address of the extent */
552 struct rb_node rb_node;
559 struct extent_info ei; /* extent info */
562 struct list_head list; /* node in global extent list of sbi */
563 struct extent_tree *et; /* extent tree pointer */
567 nid_t ino; /* inode number */
568 struct rb_root root; /* root of extent info rb-tree */
569 struct extent_node *cached_en; /* recently accessed extent node */
570 struct extent_info largest; /* largested extent info */
571 struct list_head list; /* to be used by sbi->zombie_list */
572 rwlock_t lock; /* protect extent info rb-tree */
573 atomic_t node_cnt; /* # of extent node in rb-tree*/
574 bool largest_updated; /* largest extent updated */
578 * This structure is taken from ext4_map_blocks.
580 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
582 #define F2FS_MAP_NEW (1 << BH_New)
583 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
584 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
585 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
588 struct f2fs_map_blocks {
592 unsigned int m_flags;
593 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
594 pgoff_t *m_next_extent; /* point to next possible extent */
598 /* for flag in get_data_block */
600 F2FS_GET_BLOCK_DEFAULT,
601 F2FS_GET_BLOCK_FIEMAP,
604 F2FS_GET_BLOCK_PRE_DIO,
605 F2FS_GET_BLOCK_PRE_AIO,
606 F2FS_GET_BLOCK_PRECACHE,
610 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
612 #define FADVISE_COLD_BIT 0x01
613 #define FADVISE_LOST_PINO_BIT 0x02
614 #define FADVISE_ENCRYPT_BIT 0x04
615 #define FADVISE_ENC_NAME_BIT 0x08
616 #define FADVISE_KEEP_SIZE_BIT 0x10
617 #define FADVISE_HOT_BIT 0x20
618 #define FADVISE_VERITY_BIT 0x40 /* reserved */
620 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
622 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
623 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
624 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
625 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
626 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
627 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
628 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
629 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
630 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
631 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
632 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
633 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
634 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
635 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
636 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
637 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
639 #define DEF_DIR_LEVEL 0
647 struct f2fs_inode_info {
648 struct inode vfs_inode; /* serve a vfs inode */
649 unsigned long i_flags; /* keep an inode flags for ioctl */
650 unsigned char i_advise; /* use to give file attribute hints */
651 unsigned char i_dir_level; /* use for dentry level for large dir */
652 unsigned int i_current_depth; /* only for directory depth */
653 /* for gc failure statistic */
654 unsigned int i_gc_failures[MAX_GC_FAILURE];
655 unsigned int i_pino; /* parent inode number */
656 umode_t i_acl_mode; /* keep file acl mode temporarily */
658 /* Use below internally in f2fs*/
659 unsigned long flags; /* use to pass per-file flags */
660 struct rw_semaphore i_sem; /* protect fi info */
661 atomic_t dirty_pages; /* # of dirty pages */
662 f2fs_hash_t chash; /* hash value of given file name */
663 unsigned int clevel; /* maximum level of given file name */
664 struct task_struct *task; /* lookup and create consistency */
665 struct task_struct *cp_task; /* separate cp/wb IO stats*/
666 nid_t i_xattr_nid; /* node id that contains xattrs */
667 loff_t last_disk_size; /* lastly written file size */
670 struct dquot *i_dquot[MAXQUOTAS];
672 /* quota space reservation, managed internally by quota code */
673 qsize_t i_reserved_quota;
675 struct list_head dirty_list; /* dirty list for dirs and files */
676 struct list_head gdirty_list; /* linked in global dirty list */
677 struct list_head inmem_ilist; /* list for inmem inodes */
678 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
679 struct task_struct *inmem_task; /* store inmemory task */
680 struct mutex inmem_lock; /* lock for inmemory pages */
681 struct extent_tree *extent_tree; /* cached extent_tree entry */
683 /* avoid racing between foreground op and gc */
684 struct rw_semaphore i_gc_rwsem[2];
685 struct rw_semaphore i_mmap_sem;
686 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
688 int i_extra_isize; /* size of extra space located in i_addr */
689 kprojid_t i_projid; /* id for project quota */
690 int i_inline_xattr_size; /* inline xattr size */
691 struct timespec64 i_crtime; /* inode creation time */
692 struct timespec64 i_disk_time[4];/* inode disk times */
695 static inline void get_extent_info(struct extent_info *ext,
696 struct f2fs_extent *i_ext)
698 ext->fofs = le32_to_cpu(i_ext->fofs);
699 ext->blk = le32_to_cpu(i_ext->blk);
700 ext->len = le32_to_cpu(i_ext->len);
703 static inline void set_raw_extent(struct extent_info *ext,
704 struct f2fs_extent *i_ext)
706 i_ext->fofs = cpu_to_le32(ext->fofs);
707 i_ext->blk = cpu_to_le32(ext->blk);
708 i_ext->len = cpu_to_le32(ext->len);
711 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
712 u32 blk, unsigned int len)
719 static inline bool __is_discard_mergeable(struct discard_info *back,
720 struct discard_info *front, unsigned int max_len)
722 return (back->lstart + back->len == front->lstart) &&
723 (back->len + front->len <= max_len);
726 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
727 struct discard_info *back, unsigned int max_len)
729 return __is_discard_mergeable(back, cur, max_len);
732 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
733 struct discard_info *front, unsigned int max_len)
735 return __is_discard_mergeable(cur, front, max_len);
738 static inline bool __is_extent_mergeable(struct extent_info *back,
739 struct extent_info *front)
741 return (back->fofs + back->len == front->fofs &&
742 back->blk + back->len == front->blk);
745 static inline bool __is_back_mergeable(struct extent_info *cur,
746 struct extent_info *back)
748 return __is_extent_mergeable(back, cur);
751 static inline bool __is_front_mergeable(struct extent_info *cur,
752 struct extent_info *front)
754 return __is_extent_mergeable(cur, front);
757 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
758 static inline void __try_update_largest_extent(struct extent_tree *et,
759 struct extent_node *en)
761 if (en->ei.len > et->largest.len) {
762 et->largest = en->ei;
763 et->largest_updated = true;
768 * For free nid management
771 FREE_NID, /* newly added to free nid list */
772 PREALLOC_NID, /* it is preallocated */
776 struct f2fs_nm_info {
777 block_t nat_blkaddr; /* base disk address of NAT */
778 nid_t max_nid; /* maximum possible node ids */
779 nid_t available_nids; /* # of available node ids */
780 nid_t next_scan_nid; /* the next nid to be scanned */
781 unsigned int ram_thresh; /* control the memory footprint */
782 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
783 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
785 /* NAT cache management */
786 struct radix_tree_root nat_root;/* root of the nat entry cache */
787 struct radix_tree_root nat_set_root;/* root of the nat set cache */
788 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
789 struct list_head nat_entries; /* cached nat entry list (clean) */
790 spinlock_t nat_list_lock; /* protect clean nat entry list */
791 unsigned int nat_cnt; /* the # of cached nat entries */
792 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
793 unsigned int nat_blocks; /* # of nat blocks */
795 /* free node ids management */
796 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
797 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
798 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
799 spinlock_t nid_list_lock; /* protect nid lists ops */
800 struct mutex build_lock; /* lock for build free nids */
801 unsigned char **free_nid_bitmap;
802 unsigned char *nat_block_bitmap;
803 unsigned short *free_nid_count; /* free nid count of NAT block */
806 char *nat_bitmap; /* NAT bitmap pointer */
808 unsigned int nat_bits_blocks; /* # of nat bits blocks */
809 unsigned char *nat_bits; /* NAT bits blocks */
810 unsigned char *full_nat_bits; /* full NAT pages */
811 unsigned char *empty_nat_bits; /* empty NAT pages */
812 #ifdef CONFIG_F2FS_CHECK_FS
813 char *nat_bitmap_mir; /* NAT bitmap mirror */
815 int bitmap_size; /* bitmap size */
819 * this structure is used as one of function parameters.
820 * all the information are dedicated to a given direct node block determined
821 * by the data offset in a file.
823 struct dnode_of_data {
824 struct inode *inode; /* vfs inode pointer */
825 struct page *inode_page; /* its inode page, NULL is possible */
826 struct page *node_page; /* cached direct node page */
827 nid_t nid; /* node id of the direct node block */
828 unsigned int ofs_in_node; /* data offset in the node page */
829 bool inode_page_locked; /* inode page is locked or not */
830 bool node_changed; /* is node block changed */
831 char cur_level; /* level of hole node page */
832 char max_level; /* level of current page located */
833 block_t data_blkaddr; /* block address of the node block */
836 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
837 struct page *ipage, struct page *npage, nid_t nid)
839 memset(dn, 0, sizeof(*dn));
841 dn->inode_page = ipage;
842 dn->node_page = npage;
849 * By default, there are 6 active log areas across the whole main area.
850 * When considering hot and cold data separation to reduce cleaning overhead,
851 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
853 * In the current design, you should not change the numbers intentionally.
854 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
855 * logs individually according to the underlying devices. (default: 6)
856 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
857 * data and 8 for node logs.
859 #define NR_CURSEG_DATA_TYPE (3)
860 #define NR_CURSEG_NODE_TYPE (3)
861 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
864 CURSEG_HOT_DATA = 0, /* directory entry blocks */
865 CURSEG_WARM_DATA, /* data blocks */
866 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
867 CURSEG_HOT_NODE, /* direct node blocks of directory files */
868 CURSEG_WARM_NODE, /* direct node blocks of normal files */
869 CURSEG_COLD_NODE, /* indirect node blocks */
874 struct completion wait;
875 struct llist_node llnode;
880 struct flush_cmd_control {
881 struct task_struct *f2fs_issue_flush; /* flush thread */
882 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
883 atomic_t issued_flush; /* # of issued flushes */
884 atomic_t issing_flush; /* # of issing flushes */
885 struct llist_head issue_list; /* list for command issue */
886 struct llist_node *dispatch_list; /* list for command dispatch */
889 struct f2fs_sm_info {
890 struct sit_info *sit_info; /* whole segment information */
891 struct free_segmap_info *free_info; /* free segment information */
892 struct dirty_seglist_info *dirty_info; /* dirty segment information */
893 struct curseg_info *curseg_array; /* active segment information */
895 struct rw_semaphore curseg_lock; /* for preventing curseg change */
897 block_t seg0_blkaddr; /* block address of 0'th segment */
898 block_t main_blkaddr; /* start block address of main area */
899 block_t ssa_blkaddr; /* start block address of SSA area */
901 unsigned int segment_count; /* total # of segments */
902 unsigned int main_segments; /* # of segments in main area */
903 unsigned int reserved_segments; /* # of reserved segments */
904 unsigned int ovp_segments; /* # of overprovision segments */
906 /* a threshold to reclaim prefree segments */
907 unsigned int rec_prefree_segments;
909 /* for batched trimming */
910 unsigned int trim_sections; /* # of sections to trim */
912 struct list_head sit_entry_set; /* sit entry set list */
914 unsigned int ipu_policy; /* in-place-update policy */
915 unsigned int min_ipu_util; /* in-place-update threshold */
916 unsigned int min_fsync_blocks; /* threshold for fsync */
917 unsigned int min_seq_blocks; /* threshold for sequential blocks */
918 unsigned int min_hot_blocks; /* threshold for hot block allocation */
919 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
921 /* for flush command control */
922 struct flush_cmd_control *fcc_info;
924 /* for discard command control */
925 struct discard_cmd_control *dcc_info;
932 * COUNT_TYPE for monitoring
934 * f2fs monitors the number of several block types such as on-writeback,
935 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
937 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
952 * The below are the page types of bios used in submit_bio().
953 * The available types are:
954 * DATA User data pages. It operates as async mode.
955 * NODE Node pages. It operates as async mode.
956 * META FS metadata pages such as SIT, NAT, CP.
957 * NR_PAGE_TYPE The number of page types.
958 * META_FLUSH Make sure the previous pages are written
959 * with waiting the bio's completion
960 * ... Only can be used with META.
962 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
969 INMEM, /* the below types are used by tracepoints only. */
978 HOT = 0, /* must be zero for meta bio */
984 enum need_lock_type {
990 enum cp_reason_type {
1004 APP_DIRECT_IO, /* app direct IOs */
1005 APP_BUFFERED_IO, /* app buffered IOs */
1006 APP_WRITE_IO, /* app write IOs */
1007 APP_MAPPED_IO, /* app mapped IOs */
1008 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1009 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1010 FS_META_IO, /* meta IOs from kworker/reclaimer */
1011 FS_GC_DATA_IO, /* data IOs from forground gc */
1012 FS_GC_NODE_IO, /* node IOs from forground gc */
1013 FS_CP_DATA_IO, /* data IOs from checkpoint */
1014 FS_CP_NODE_IO, /* node IOs from checkpoint */
1015 FS_CP_META_IO, /* meta IOs from checkpoint */
1016 FS_DISCARD, /* discard */
1020 struct f2fs_io_info {
1021 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1022 nid_t ino; /* inode number */
1023 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1024 enum temp_type temp; /* contains HOT/WARM/COLD */
1025 int op; /* contains REQ_OP_ */
1026 int op_flags; /* req_flag_bits */
1027 block_t new_blkaddr; /* new block address to be written */
1028 block_t old_blkaddr; /* old block address before Cow */
1029 struct page *page; /* page to be written */
1030 struct page *encrypted_page; /* encrypted page */
1031 struct list_head list; /* serialize IOs */
1032 bool submitted; /* indicate IO submission */
1033 int need_lock; /* indicate we need to lock cp_rwsem */
1034 bool in_list; /* indicate fio is in io_list */
1035 bool is_meta; /* indicate borrow meta inode mapping or not */
1036 bool retry; /* need to reallocate block address */
1037 enum iostat_type io_type; /* io type */
1038 struct writeback_control *io_wbc; /* writeback control */
1039 unsigned char version; /* version of the node */
1042 #define is_read_io(rw) ((rw) == READ)
1043 struct f2fs_bio_info {
1044 struct f2fs_sb_info *sbi; /* f2fs superblock */
1045 struct bio *bio; /* bios to merge */
1046 sector_t last_block_in_bio; /* last block number */
1047 struct f2fs_io_info fio; /* store buffered io info. */
1048 struct rw_semaphore io_rwsem; /* blocking op for bio */
1049 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1050 struct list_head io_list; /* track fios */
1053 #define FDEV(i) (sbi->devs[i])
1054 #define RDEV(i) (raw_super->devs[i])
1055 struct f2fs_dev_info {
1056 struct block_device *bdev;
1057 char path[MAX_PATH_LEN];
1058 unsigned int total_segments;
1061 #ifdef CONFIG_BLK_DEV_ZONED
1062 unsigned int nr_blkz; /* Total number of zones */
1063 u8 *blkz_type; /* Array of zones type */
1068 DIR_INODE, /* for dirty dir inode */
1069 FILE_INODE, /* for dirty regular/symlink inode */
1070 DIRTY_META, /* for all dirtied inode metadata */
1071 ATOMIC_FILE, /* for all atomic files */
1075 /* for inner inode cache management */
1076 struct inode_management {
1077 struct radix_tree_root ino_root; /* ino entry array */
1078 spinlock_t ino_lock; /* for ino entry lock */
1079 struct list_head ino_list; /* inode list head */
1080 unsigned long ino_num; /* number of entries */
1083 /* For s_flag in struct f2fs_sb_info */
1085 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1086 SBI_IS_CLOSE, /* specify unmounting */
1087 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1088 SBI_POR_DOING, /* recovery is doing or not */
1089 SBI_NEED_SB_WRITE, /* need to recover superblock */
1090 SBI_NEED_CP, /* need to checkpoint */
1091 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1092 SBI_IS_RECOVERED, /* recovered orphan/data */
1109 WHINT_MODE_OFF, /* not pass down write hints */
1110 WHINT_MODE_USER, /* try to pass down hints given by users */
1111 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1115 ALLOC_MODE_DEFAULT, /* stay default */
1116 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1120 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1121 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1122 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1125 #ifdef CONFIG_F2FS_FS_ENCRYPTION
1126 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1127 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1129 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1132 struct f2fs_sb_info {
1133 struct super_block *sb; /* pointer to VFS super block */
1134 struct proc_dir_entry *s_proc; /* proc entry */
1135 struct f2fs_super_block *raw_super; /* raw super block pointer */
1136 struct rw_semaphore sb_lock; /* lock for raw super block */
1137 int valid_super_block; /* valid super block no */
1138 unsigned long s_flag; /* flags for sbi */
1139 struct mutex writepages; /* mutex for writepages() */
1141 #ifdef CONFIG_BLK_DEV_ZONED
1142 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1143 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1146 /* for node-related operations */
1147 struct f2fs_nm_info *nm_info; /* node manager */
1148 struct inode *node_inode; /* cache node blocks */
1150 /* for segment-related operations */
1151 struct f2fs_sm_info *sm_info; /* segment manager */
1153 /* for bio operations */
1154 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1155 struct mutex wio_mutex[NR_PAGE_TYPE - 1][NR_TEMP_TYPE];
1156 /* bio ordering for NODE/DATA */
1157 /* keep migration IO order for LFS mode */
1158 struct rw_semaphore io_order_lock;
1159 mempool_t *write_io_dummy; /* Dummy pages */
1161 /* for checkpoint */
1162 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1163 int cur_cp_pack; /* remain current cp pack */
1164 spinlock_t cp_lock; /* for flag in ckpt */
1165 struct inode *meta_inode; /* cache meta blocks */
1166 struct mutex cp_mutex; /* checkpoint procedure lock */
1167 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1168 struct rw_semaphore node_write; /* locking node writes */
1169 struct rw_semaphore node_change; /* locking node change */
1170 wait_queue_head_t cp_wait;
1171 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1172 long interval_time[MAX_TIME]; /* to store thresholds */
1174 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1176 spinlock_t fsync_node_lock; /* for node entry lock */
1177 struct list_head fsync_node_list; /* node list head */
1178 unsigned int fsync_seg_id; /* sequence id */
1179 unsigned int fsync_node_num; /* number of node entries */
1181 /* for orphan inode, use 0'th array */
1182 unsigned int max_orphans; /* max orphan inodes */
1184 /* for inode management */
1185 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1186 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1188 /* for extent tree cache */
1189 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1190 struct mutex extent_tree_lock; /* locking extent radix tree */
1191 struct list_head extent_list; /* lru list for shrinker */
1192 spinlock_t extent_lock; /* locking extent lru list */
1193 atomic_t total_ext_tree; /* extent tree count */
1194 struct list_head zombie_list; /* extent zombie tree list */
1195 atomic_t total_zombie_tree; /* extent zombie tree count */
1196 atomic_t total_ext_node; /* extent info count */
1198 /* basic filesystem units */
1199 unsigned int log_sectors_per_block; /* log2 sectors per block */
1200 unsigned int log_blocksize; /* log2 block size */
1201 unsigned int blocksize; /* block size */
1202 unsigned int root_ino_num; /* root inode number*/
1203 unsigned int node_ino_num; /* node inode number*/
1204 unsigned int meta_ino_num; /* meta inode number*/
1205 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1206 unsigned int blocks_per_seg; /* blocks per segment */
1207 unsigned int segs_per_sec; /* segments per section */
1208 unsigned int secs_per_zone; /* sections per zone */
1209 unsigned int total_sections; /* total section count */
1210 unsigned int total_node_count; /* total node block count */
1211 unsigned int total_valid_node_count; /* valid node block count */
1212 loff_t max_file_blocks; /* max block index of file */
1213 int dir_level; /* directory level */
1214 unsigned int trigger_ssr_threshold; /* threshold to trigger ssr */
1215 int readdir_ra; /* readahead inode in readdir */
1217 block_t user_block_count; /* # of user blocks */
1218 block_t total_valid_block_count; /* # of valid blocks */
1219 block_t discard_blks; /* discard command candidats */
1220 block_t last_valid_block_count; /* for recovery */
1221 block_t reserved_blocks; /* configurable reserved blocks */
1222 block_t current_reserved_blocks; /* current reserved blocks */
1224 unsigned int nquota_files; /* # of quota sysfile */
1226 u32 s_next_generation; /* for NFS support */
1228 /* # of pages, see count_type */
1229 atomic_t nr_pages[NR_COUNT_TYPE];
1230 /* # of allocated blocks */
1231 struct percpu_counter alloc_valid_block_count;
1233 /* writeback control */
1234 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1236 /* valid inode count */
1237 struct percpu_counter total_valid_inode_count;
1239 struct f2fs_mount_info mount_opt; /* mount options */
1241 /* for cleaning operations */
1242 struct mutex gc_mutex; /* mutex for GC */
1243 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1244 unsigned int cur_victim_sec; /* current victim section num */
1245 unsigned int gc_mode; /* current GC state */
1246 /* for skip statistic */
1247 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1248 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1250 /* threshold for gc trials on pinned files */
1251 u64 gc_pin_file_threshold;
1253 /* maximum # of trials to find a victim segment for SSR and GC */
1254 unsigned int max_victim_search;
1257 * for stat information.
1258 * one is for the LFS mode, and the other is for the SSR mode.
1260 #ifdef CONFIG_F2FS_STAT_FS
1261 struct f2fs_stat_info *stat_info; /* FS status information */
1262 unsigned int segment_count[2]; /* # of allocated segments */
1263 unsigned int block_count[2]; /* # of allocated blocks */
1264 atomic_t inplace_count; /* # of inplace update */
1265 atomic64_t total_hit_ext; /* # of lookup extent cache */
1266 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1267 atomic64_t read_hit_largest; /* # of hit largest extent node */
1268 atomic64_t read_hit_cached; /* # of hit cached extent node */
1269 atomic_t inline_xattr; /* # of inline_xattr inodes */
1270 atomic_t inline_inode; /* # of inline_data inodes */
1271 atomic_t inline_dir; /* # of inline_dentry inodes */
1272 atomic_t aw_cnt; /* # of atomic writes */
1273 atomic_t vw_cnt; /* # of volatile writes */
1274 atomic_t max_aw_cnt; /* max # of atomic writes */
1275 atomic_t max_vw_cnt; /* max # of volatile writes */
1276 int bg_gc; /* background gc calls */
1277 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1279 spinlock_t stat_lock; /* lock for stat operations */
1281 /* For app/fs IO statistics */
1282 spinlock_t iostat_lock;
1283 unsigned long long write_iostat[NR_IO_TYPE];
1286 /* For sysfs suppport */
1287 struct kobject s_kobj;
1288 struct completion s_kobj_unregister;
1290 /* For shrinker support */
1291 struct list_head s_list;
1292 int s_ndevs; /* number of devices */
1293 struct f2fs_dev_info *devs; /* for device list */
1294 unsigned int dirty_device; /* for checkpoint data flush */
1295 spinlock_t dev_lock; /* protect dirty_device */
1296 struct mutex umount_mutex;
1297 unsigned int shrinker_run_no;
1299 /* For write statistics */
1300 u64 sectors_written_start;
1303 /* Reference to checksum algorithm driver via cryptoapi */
1304 struct crypto_shash *s_chksum_driver;
1306 /* Precomputed FS UUID checksum for seeding other checksums */
1307 __u32 s_chksum_seed;
1310 #ifdef CONFIG_F2FS_FAULT_INJECTION
1311 #define f2fs_show_injection_info(type) \
1312 printk("%sF2FS-fs : inject %s in %s of %pF\n", \
1313 KERN_INFO, f2fs_fault_name[type], \
1314 __func__, __builtin_return_address(0))
1315 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1317 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1319 if (!ffi->inject_rate)
1322 if (!IS_FAULT_SET(ffi, type))
1325 atomic_inc(&ffi->inject_ops);
1326 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1327 atomic_set(&ffi->inject_ops, 0);
1333 #define f2fs_show_injection_info(type) do { } while (0)
1334 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1341 * Test if the mounted volume is a multi-device volume.
1342 * - For a single regular disk volume, sbi->s_ndevs is 0.
1343 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1344 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1346 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1348 return sbi->s_ndevs > 1;
1351 /* For write statistics. Suppose sector size is 512 bytes,
1352 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1354 #define BD_PART_WRITTEN(s) \
1355 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1356 (s)->sectors_written_start) >> 1)
1358 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1360 sbi->last_time[type] = jiffies;
1363 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1365 unsigned long interval = sbi->interval_time[type] * HZ;
1367 return time_after(jiffies, sbi->last_time[type] + interval);
1370 static inline bool is_idle(struct f2fs_sb_info *sbi)
1372 struct block_device *bdev = sbi->sb->s_bdev;
1373 struct request_queue *q = bdev_get_queue(bdev);
1374 struct request_list *rl = &q->root_rl;
1376 if (rl->count[BLK_RW_SYNC] || rl->count[BLK_RW_ASYNC])
1379 return f2fs_time_over(sbi, REQ_TIME);
1385 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1386 const void *address, unsigned int length)
1389 struct shash_desc shash;
1394 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1396 desc.shash.tfm = sbi->s_chksum_driver;
1397 desc.shash.flags = 0;
1398 *(u32 *)desc.ctx = crc;
1400 err = crypto_shash_update(&desc.shash, address, length);
1403 return *(u32 *)desc.ctx;
1406 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1407 unsigned int length)
1409 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1412 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1413 void *buf, size_t buf_size)
1415 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1418 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1419 const void *address, unsigned int length)
1421 return __f2fs_crc32(sbi, crc, address, length);
1424 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1426 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1429 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1431 return sb->s_fs_info;
1434 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1436 return F2FS_SB(inode->i_sb);
1439 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1441 return F2FS_I_SB(mapping->host);
1444 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1446 return F2FS_M_SB(page->mapping);
1449 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1451 return (struct f2fs_super_block *)(sbi->raw_super);
1454 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1456 return (struct f2fs_checkpoint *)(sbi->ckpt);
1459 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1461 return (struct f2fs_node *)page_address(page);
1464 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1466 return &((struct f2fs_node *)page_address(page))->i;
1469 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1471 return (struct f2fs_nm_info *)(sbi->nm_info);
1474 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1476 return (struct f2fs_sm_info *)(sbi->sm_info);
1479 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1481 return (struct sit_info *)(SM_I(sbi)->sit_info);
1484 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1486 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1489 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1491 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1494 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1496 return sbi->meta_inode->i_mapping;
1499 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1501 return sbi->node_inode->i_mapping;
1504 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1506 return test_bit(type, &sbi->s_flag);
1509 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1511 set_bit(type, &sbi->s_flag);
1514 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1516 clear_bit(type, &sbi->s_flag);
1519 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1521 return le64_to_cpu(cp->checkpoint_ver);
1524 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1526 if (type < F2FS_MAX_QUOTAS)
1527 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1531 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1533 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1534 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1537 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1539 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1541 return ckpt_flags & f;
1544 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1546 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1549 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1551 unsigned int ckpt_flags;
1553 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1555 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1558 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1560 unsigned long flags;
1562 spin_lock_irqsave(&sbi->cp_lock, flags);
1563 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1564 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1567 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1569 unsigned int ckpt_flags;
1571 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1573 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1576 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1578 unsigned long flags;
1580 spin_lock_irqsave(&sbi->cp_lock, flags);
1581 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1582 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1585 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1587 unsigned long flags;
1589 set_sbi_flag(sbi, SBI_NEED_FSCK);
1592 spin_lock_irqsave(&sbi->cp_lock, flags);
1593 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1594 kfree(NM_I(sbi)->nat_bits);
1595 NM_I(sbi)->nat_bits = NULL;
1597 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1600 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1601 struct cp_control *cpc)
1603 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1605 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1608 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1610 down_read(&sbi->cp_rwsem);
1613 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1615 return down_read_trylock(&sbi->cp_rwsem);
1618 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1620 up_read(&sbi->cp_rwsem);
1623 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1625 down_write(&sbi->cp_rwsem);
1628 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1630 up_write(&sbi->cp_rwsem);
1633 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1635 int reason = CP_SYNC;
1637 if (test_opt(sbi, FASTBOOT))
1638 reason = CP_FASTBOOT;
1639 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1644 static inline bool __remain_node_summaries(int reason)
1646 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1649 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1651 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1652 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1656 * Check whether the inode has blocks or not
1658 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1660 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1662 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1665 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1667 return ofs == XATTR_NODE_OFFSET;
1670 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1671 struct inode *inode, bool cap)
1675 if (!test_opt(sbi, RESERVE_ROOT))
1677 if (IS_NOQUOTA(inode))
1679 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1681 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1682 in_group_p(F2FS_OPTION(sbi).s_resgid))
1684 if (cap && capable(CAP_SYS_RESOURCE))
1689 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1690 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1691 struct inode *inode, blkcnt_t *count)
1693 blkcnt_t diff = 0, release = 0;
1694 block_t avail_user_block_count;
1697 ret = dquot_reserve_block(inode, *count);
1701 if (time_to_inject(sbi, FAULT_BLOCK)) {
1702 f2fs_show_injection_info(FAULT_BLOCK);
1708 * let's increase this in prior to actual block count change in order
1709 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1711 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1713 spin_lock(&sbi->stat_lock);
1714 sbi->total_valid_block_count += (block_t)(*count);
1715 avail_user_block_count = sbi->user_block_count -
1716 sbi->current_reserved_blocks;
1718 if (!__allow_reserved_blocks(sbi, inode, true))
1719 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1721 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1722 diff = sbi->total_valid_block_count - avail_user_block_count;
1727 sbi->total_valid_block_count -= diff;
1729 spin_unlock(&sbi->stat_lock);
1733 spin_unlock(&sbi->stat_lock);
1735 if (unlikely(release)) {
1736 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1737 dquot_release_reservation_block(inode, release);
1739 f2fs_i_blocks_write(inode, *count, true, true);
1743 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1744 dquot_release_reservation_block(inode, release);
1748 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
1749 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1750 struct inode *inode,
1753 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1755 spin_lock(&sbi->stat_lock);
1756 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1757 sbi->total_valid_block_count -= (block_t)count;
1758 if (sbi->reserved_blocks &&
1759 sbi->current_reserved_blocks < sbi->reserved_blocks)
1760 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1761 sbi->current_reserved_blocks + count);
1762 spin_unlock(&sbi->stat_lock);
1763 if (unlikely(inode->i_blocks < sectors)) {
1764 f2fs_msg(sbi->sb, KERN_WARNING,
1765 "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
1767 (unsigned long long)inode->i_blocks,
1768 (unsigned long long)sectors);
1769 set_sbi_flag(sbi, SBI_NEED_FSCK);
1772 f2fs_i_blocks_write(inode, count, false, true);
1775 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1777 atomic_inc(&sbi->nr_pages[count_type]);
1779 if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES ||
1780 count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA)
1783 set_sbi_flag(sbi, SBI_IS_DIRTY);
1786 static inline void inode_inc_dirty_pages(struct inode *inode)
1788 atomic_inc(&F2FS_I(inode)->dirty_pages);
1789 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1790 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1791 if (IS_NOQUOTA(inode))
1792 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1795 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1797 atomic_dec(&sbi->nr_pages[count_type]);
1800 static inline void inode_dec_dirty_pages(struct inode *inode)
1802 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1803 !S_ISLNK(inode->i_mode))
1806 atomic_dec(&F2FS_I(inode)->dirty_pages);
1807 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1808 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1809 if (IS_NOQUOTA(inode))
1810 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1813 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1815 return atomic_read(&sbi->nr_pages[count_type]);
1818 static inline int get_dirty_pages(struct inode *inode)
1820 return atomic_read(&F2FS_I(inode)->dirty_pages);
1823 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1825 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1826 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1827 sbi->log_blocks_per_seg;
1829 return segs / sbi->segs_per_sec;
1832 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1834 return sbi->total_valid_block_count;
1837 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1839 return sbi->discard_blks;
1842 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1844 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1846 /* return NAT or SIT bitmap */
1847 if (flag == NAT_BITMAP)
1848 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1849 else if (flag == SIT_BITMAP)
1850 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1855 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1857 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1860 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1862 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1865 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1866 offset = (flag == SIT_BITMAP) ?
1867 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1868 return &ckpt->sit_nat_version_bitmap + offset;
1871 if (__cp_payload(sbi) > 0) {
1872 if (flag == NAT_BITMAP)
1873 return &ckpt->sit_nat_version_bitmap;
1875 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1877 offset = (flag == NAT_BITMAP) ?
1878 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1879 return &ckpt->sit_nat_version_bitmap + offset;
1883 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1885 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1887 if (sbi->cur_cp_pack == 2)
1888 start_addr += sbi->blocks_per_seg;
1892 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1894 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1896 if (sbi->cur_cp_pack == 1)
1897 start_addr += sbi->blocks_per_seg;
1901 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1903 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1906 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1908 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1911 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
1912 struct inode *inode, bool is_inode)
1914 block_t valid_block_count;
1915 unsigned int valid_node_count;
1916 bool quota = inode && !is_inode;
1919 int ret = dquot_reserve_block(inode, 1);
1924 if (time_to_inject(sbi, FAULT_BLOCK)) {
1925 f2fs_show_injection_info(FAULT_BLOCK);
1929 spin_lock(&sbi->stat_lock);
1931 valid_block_count = sbi->total_valid_block_count +
1932 sbi->current_reserved_blocks + 1;
1934 if (!__allow_reserved_blocks(sbi, inode, false))
1935 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
1937 if (unlikely(valid_block_count > sbi->user_block_count)) {
1938 spin_unlock(&sbi->stat_lock);
1942 valid_node_count = sbi->total_valid_node_count + 1;
1943 if (unlikely(valid_node_count > sbi->total_node_count)) {
1944 spin_unlock(&sbi->stat_lock);
1948 sbi->total_valid_node_count++;
1949 sbi->total_valid_block_count++;
1950 spin_unlock(&sbi->stat_lock);
1954 f2fs_mark_inode_dirty_sync(inode, true);
1956 f2fs_i_blocks_write(inode, 1, true, true);
1959 percpu_counter_inc(&sbi->alloc_valid_block_count);
1964 dquot_release_reservation_block(inode, 1);
1968 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
1969 struct inode *inode, bool is_inode)
1971 spin_lock(&sbi->stat_lock);
1973 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
1974 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
1975 f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
1977 sbi->total_valid_node_count--;
1978 sbi->total_valid_block_count--;
1979 if (sbi->reserved_blocks &&
1980 sbi->current_reserved_blocks < sbi->reserved_blocks)
1981 sbi->current_reserved_blocks++;
1983 spin_unlock(&sbi->stat_lock);
1986 f2fs_i_blocks_write(inode, 1, false, true);
1989 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
1991 return sbi->total_valid_node_count;
1994 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
1996 percpu_counter_inc(&sbi->total_valid_inode_count);
1999 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2001 percpu_counter_dec(&sbi->total_valid_inode_count);
2004 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2006 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2009 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2010 pgoff_t index, bool for_write)
2014 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2016 page = find_get_page_flags(mapping, index,
2017 FGP_LOCK | FGP_ACCESSED);
2019 page = find_lock_page(mapping, index);
2023 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2024 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2030 return grab_cache_page(mapping, index);
2031 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2034 static inline struct page *f2fs_pagecache_get_page(
2035 struct address_space *mapping, pgoff_t index,
2036 int fgp_flags, gfp_t gfp_mask)
2038 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2039 f2fs_show_injection_info(FAULT_PAGE_GET);
2043 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2046 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2048 char *src_kaddr = kmap(src);
2049 char *dst_kaddr = kmap(dst);
2051 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2056 static inline void f2fs_put_page(struct page *page, int unlock)
2062 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2068 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2071 f2fs_put_page(dn->node_page, 1);
2072 if (dn->inode_page && dn->node_page != dn->inode_page)
2073 f2fs_put_page(dn->inode_page, 0);
2074 dn->node_page = NULL;
2075 dn->inode_page = NULL;
2078 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2081 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2084 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2089 entry = kmem_cache_alloc(cachep, flags);
2091 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2095 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2096 int npages, bool no_fail)
2101 /* No failure on bio allocation */
2102 bio = bio_alloc(GFP_NOIO, npages);
2104 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2107 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2108 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2112 return bio_alloc(GFP_KERNEL, npages);
2115 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2116 unsigned long index, void *item)
2118 while (radix_tree_insert(root, index, item))
2122 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2124 static inline bool IS_INODE(struct page *page)
2126 struct f2fs_node *p = F2FS_NODE(page);
2128 return RAW_IS_INODE(p);
2131 static inline int offset_in_addr(struct f2fs_inode *i)
2133 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2134 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2137 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2139 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2142 static inline int f2fs_has_extra_attr(struct inode *inode);
2143 static inline block_t datablock_addr(struct inode *inode,
2144 struct page *node_page, unsigned int offset)
2146 struct f2fs_node *raw_node;
2149 bool is_inode = IS_INODE(node_page);
2151 raw_node = F2FS_NODE(node_page);
2153 /* from GC path only */
2156 base = offset_in_addr(&raw_node->i);
2157 else if (f2fs_has_extra_attr(inode))
2158 base = get_extra_isize(inode);
2161 addr_array = blkaddr_in_node(raw_node);
2162 return le32_to_cpu(addr_array[base + offset]);
2165 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2170 mask = 1 << (7 - (nr & 0x07));
2171 return mask & *addr;
2174 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2179 mask = 1 << (7 - (nr & 0x07));
2183 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2188 mask = 1 << (7 - (nr & 0x07));
2192 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2198 mask = 1 << (7 - (nr & 0x07));
2204 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2210 mask = 1 << (7 - (nr & 0x07));
2216 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2221 mask = 1 << (7 - (nr & 0x07));
2228 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */
2229 #define F2FS_UNRM_FL 0x00000002 /* Undelete */
2230 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2231 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2232 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2233 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2234 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2235 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2236 /* Reserved for compression usage... */
2237 #define F2FS_DIRTY_FL 0x00000100
2238 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
2239 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */
2240 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */
2241 /* End compression flags --- maybe not all used */
2242 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2243 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */
2244 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
2245 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
2246 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2247 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
2248 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
2249 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */
2250 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */
2251 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
2252 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
2253 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2254 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
2256 #define F2FS_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
2257 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
2259 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */
2260 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \
2261 F2FS_IMMUTABLE_FL | \
2265 F2FS_PROJINHERIT_FL)
2267 /* Flags that should be inherited by new inodes from their parent. */
2268 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\
2269 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\
2270 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\
2271 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\
2272 F2FS_PROJINHERIT_FL)
2274 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2275 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL))
2277 /* Flags that are appropriate for non-directories/regular files. */
2278 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2280 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2284 else if (S_ISREG(mode))
2285 return flags & F2FS_REG_FLMASK;
2287 return flags & F2FS_OTHER_FLMASK;
2290 /* used for f2fs_inode_info->flags */
2292 FI_NEW_INODE, /* indicate newly allocated inode */
2293 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2294 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2295 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2296 FI_INC_LINK, /* need to increment i_nlink */
2297 FI_ACL_MODE, /* indicate acl mode */
2298 FI_NO_ALLOC, /* should not allocate any blocks */
2299 FI_FREE_NID, /* free allocated nide */
2300 FI_NO_EXTENT, /* not to use the extent cache */
2301 FI_INLINE_XATTR, /* used for inline xattr */
2302 FI_INLINE_DATA, /* used for inline data*/
2303 FI_INLINE_DENTRY, /* used for inline dentry */
2304 FI_APPEND_WRITE, /* inode has appended data */
2305 FI_UPDATE_WRITE, /* inode has in-place-update data */
2306 FI_NEED_IPU, /* used for ipu per file */
2307 FI_ATOMIC_FILE, /* indicate atomic file */
2308 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2309 FI_VOLATILE_FILE, /* indicate volatile file */
2310 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2311 FI_DROP_CACHE, /* drop dirty page cache */
2312 FI_DATA_EXIST, /* indicate data exists */
2313 FI_INLINE_DOTS, /* indicate inline dot dentries */
2314 FI_DO_DEFRAG, /* indicate defragment is running */
2315 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2316 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2317 FI_HOT_DATA, /* indicate file is hot */
2318 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2319 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2320 FI_PIN_FILE, /* indicate file should not be gced */
2321 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2324 static inline void __mark_inode_dirty_flag(struct inode *inode,
2328 case FI_INLINE_XATTR:
2329 case FI_INLINE_DATA:
2330 case FI_INLINE_DENTRY:
2335 case FI_INLINE_DOTS:
2337 f2fs_mark_inode_dirty_sync(inode, true);
2341 static inline void set_inode_flag(struct inode *inode, int flag)
2343 if (!test_bit(flag, &F2FS_I(inode)->flags))
2344 set_bit(flag, &F2FS_I(inode)->flags);
2345 __mark_inode_dirty_flag(inode, flag, true);
2348 static inline int is_inode_flag_set(struct inode *inode, int flag)
2350 return test_bit(flag, &F2FS_I(inode)->flags);
2353 static inline void clear_inode_flag(struct inode *inode, int flag)
2355 if (test_bit(flag, &F2FS_I(inode)->flags))
2356 clear_bit(flag, &F2FS_I(inode)->flags);
2357 __mark_inode_dirty_flag(inode, flag, false);
2360 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2362 F2FS_I(inode)->i_acl_mode = mode;
2363 set_inode_flag(inode, FI_ACL_MODE);
2364 f2fs_mark_inode_dirty_sync(inode, false);
2367 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2373 f2fs_mark_inode_dirty_sync(inode, true);
2376 static inline void f2fs_i_blocks_write(struct inode *inode,
2377 block_t diff, bool add, bool claim)
2379 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2380 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2382 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2385 dquot_claim_block(inode, diff);
2387 dquot_alloc_block_nofail(inode, diff);
2389 dquot_free_block(inode, diff);
2392 f2fs_mark_inode_dirty_sync(inode, true);
2393 if (clean || recover)
2394 set_inode_flag(inode, FI_AUTO_RECOVER);
2397 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2399 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2400 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2402 if (i_size_read(inode) == i_size)
2405 i_size_write(inode, i_size);
2406 f2fs_mark_inode_dirty_sync(inode, true);
2407 if (clean || recover)
2408 set_inode_flag(inode, FI_AUTO_RECOVER);
2411 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2413 F2FS_I(inode)->i_current_depth = depth;
2414 f2fs_mark_inode_dirty_sync(inode, true);
2417 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2420 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2421 f2fs_mark_inode_dirty_sync(inode, true);
2424 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2426 F2FS_I(inode)->i_xattr_nid = xnid;
2427 f2fs_mark_inode_dirty_sync(inode, true);
2430 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2432 F2FS_I(inode)->i_pino = pino;
2433 f2fs_mark_inode_dirty_sync(inode, true);
2436 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2438 struct f2fs_inode_info *fi = F2FS_I(inode);
2440 if (ri->i_inline & F2FS_INLINE_XATTR)
2441 set_bit(FI_INLINE_XATTR, &fi->flags);
2442 if (ri->i_inline & F2FS_INLINE_DATA)
2443 set_bit(FI_INLINE_DATA, &fi->flags);
2444 if (ri->i_inline & F2FS_INLINE_DENTRY)
2445 set_bit(FI_INLINE_DENTRY, &fi->flags);
2446 if (ri->i_inline & F2FS_DATA_EXIST)
2447 set_bit(FI_DATA_EXIST, &fi->flags);
2448 if (ri->i_inline & F2FS_INLINE_DOTS)
2449 set_bit(FI_INLINE_DOTS, &fi->flags);
2450 if (ri->i_inline & F2FS_EXTRA_ATTR)
2451 set_bit(FI_EXTRA_ATTR, &fi->flags);
2452 if (ri->i_inline & F2FS_PIN_FILE)
2453 set_bit(FI_PIN_FILE, &fi->flags);
2456 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2460 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2461 ri->i_inline |= F2FS_INLINE_XATTR;
2462 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2463 ri->i_inline |= F2FS_INLINE_DATA;
2464 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2465 ri->i_inline |= F2FS_INLINE_DENTRY;
2466 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2467 ri->i_inline |= F2FS_DATA_EXIST;
2468 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2469 ri->i_inline |= F2FS_INLINE_DOTS;
2470 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2471 ri->i_inline |= F2FS_EXTRA_ATTR;
2472 if (is_inode_flag_set(inode, FI_PIN_FILE))
2473 ri->i_inline |= F2FS_PIN_FILE;
2476 static inline int f2fs_has_extra_attr(struct inode *inode)
2478 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2481 static inline int f2fs_has_inline_xattr(struct inode *inode)
2483 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2486 static inline unsigned int addrs_per_inode(struct inode *inode)
2488 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
2491 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2493 struct f2fs_inode *ri = F2FS_INODE(page);
2495 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2496 get_inline_xattr_addrs(inode)]);
2499 static inline int inline_xattr_size(struct inode *inode)
2501 if (f2fs_has_inline_xattr(inode))
2502 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2506 static inline int f2fs_has_inline_data(struct inode *inode)
2508 return is_inode_flag_set(inode, FI_INLINE_DATA);
2511 static inline int f2fs_exist_data(struct inode *inode)
2513 return is_inode_flag_set(inode, FI_DATA_EXIST);
2516 static inline int f2fs_has_inline_dots(struct inode *inode)
2518 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2521 static inline bool f2fs_is_pinned_file(struct inode *inode)
2523 return is_inode_flag_set(inode, FI_PIN_FILE);
2526 static inline bool f2fs_is_atomic_file(struct inode *inode)
2528 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2531 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2533 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2536 static inline bool f2fs_is_volatile_file(struct inode *inode)
2538 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2541 static inline bool f2fs_is_first_block_written(struct inode *inode)
2543 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2546 static inline bool f2fs_is_drop_cache(struct inode *inode)
2548 return is_inode_flag_set(inode, FI_DROP_CACHE);
2551 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2553 struct f2fs_inode *ri = F2FS_INODE(page);
2554 int extra_size = get_extra_isize(inode);
2556 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2559 static inline int f2fs_has_inline_dentry(struct inode *inode)
2561 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2564 static inline int is_file(struct inode *inode, int type)
2566 return F2FS_I(inode)->i_advise & type;
2569 static inline void set_file(struct inode *inode, int type)
2571 F2FS_I(inode)->i_advise |= type;
2572 f2fs_mark_inode_dirty_sync(inode, true);
2575 static inline void clear_file(struct inode *inode, int type)
2577 F2FS_I(inode)->i_advise &= ~type;
2578 f2fs_mark_inode_dirty_sync(inode, true);
2581 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2586 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2588 spin_lock(&sbi->inode_lock[DIRTY_META]);
2589 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2590 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2593 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2594 file_keep_isize(inode) ||
2595 i_size_read(inode) & ~PAGE_MASK)
2598 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2600 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2602 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2604 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2605 &F2FS_I(inode)->i_crtime))
2608 down_read(&F2FS_I(inode)->i_sem);
2609 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2610 up_read(&F2FS_I(inode)->i_sem);
2615 static inline bool f2fs_readonly(struct super_block *sb)
2617 return sb_rdonly(sb);
2620 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2622 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2625 static inline bool is_dot_dotdot(const struct qstr *str)
2627 if (str->len == 1 && str->name[0] == '.')
2630 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2636 static inline bool f2fs_may_extent_tree(struct inode *inode)
2638 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2640 if (!test_opt(sbi, EXTENT_CACHE) ||
2641 is_inode_flag_set(inode, FI_NO_EXTENT))
2645 * for recovered files during mount do not create extents
2646 * if shrinker is not registered.
2648 if (list_empty(&sbi->s_list))
2651 return S_ISREG(inode->i_mode);
2654 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2655 size_t size, gfp_t flags)
2657 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2658 f2fs_show_injection_info(FAULT_KMALLOC);
2662 return kmalloc(size, flags);
2665 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2666 size_t size, gfp_t flags)
2668 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2671 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2672 size_t size, gfp_t flags)
2674 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2675 f2fs_show_injection_info(FAULT_KVMALLOC);
2679 return kvmalloc(size, flags);
2682 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2683 size_t size, gfp_t flags)
2685 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2688 static inline int get_extra_isize(struct inode *inode)
2690 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2693 static inline int get_inline_xattr_addrs(struct inode *inode)
2695 return F2FS_I(inode)->i_inline_xattr_size;
2698 #define f2fs_get_inode_mode(i) \
2699 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2700 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2702 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2703 (offsetof(struct f2fs_inode, i_extra_end) - \
2704 offsetof(struct f2fs_inode, i_extra_isize)) \
2706 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2707 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2708 ((offsetof(typeof(*f2fs_inode), field) + \
2709 sizeof((f2fs_inode)->field)) \
2710 <= (F2FS_OLD_ATTRIBUTE_SIZE + extra_isize)) \
2712 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2716 spin_lock(&sbi->iostat_lock);
2717 for (i = 0; i < NR_IO_TYPE; i++)
2718 sbi->write_iostat[i] = 0;
2719 spin_unlock(&sbi->iostat_lock);
2722 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2723 enum iostat_type type, unsigned long long io_bytes)
2725 if (!sbi->iostat_enable)
2727 spin_lock(&sbi->iostat_lock);
2728 sbi->write_iostat[type] += io_bytes;
2730 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2731 sbi->write_iostat[APP_BUFFERED_IO] =
2732 sbi->write_iostat[APP_WRITE_IO] -
2733 sbi->write_iostat[APP_DIRECT_IO];
2734 spin_unlock(&sbi->iostat_lock);
2737 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO(fio->type) == META && \
2738 (!is_read_io(fio->op) || fio->is_meta))
2740 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2741 block_t blkaddr, int type);
2742 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2743 block_t blkaddr, int type)
2745 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2746 f2fs_msg(sbi->sb, KERN_ERR,
2747 "invalid blkaddr: %u, type: %d, run fsck to fix.",
2749 f2fs_bug_on(sbi, 1);
2753 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2755 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2760 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2763 if (!__is_valid_data_blkaddr(blkaddr))
2765 verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2772 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2773 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2774 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
2775 int f2fs_truncate(struct inode *inode);
2776 int f2fs_getattr(const struct path *path, struct kstat *stat,
2777 u32 request_mask, unsigned int flags);
2778 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2779 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2780 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2781 int f2fs_precache_extents(struct inode *inode);
2782 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2783 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2784 int f2fs_pin_file_control(struct inode *inode, bool inc);
2789 void f2fs_set_inode_flags(struct inode *inode);
2790 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2791 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2792 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2793 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2794 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2795 void f2fs_update_inode(struct inode *inode, struct page *node_page);
2796 void f2fs_update_inode_page(struct inode *inode);
2797 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2798 void f2fs_evict_inode(struct inode *inode);
2799 void f2fs_handle_failed_inode(struct inode *inode);
2804 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2805 bool hot, bool set);
2806 struct dentry *f2fs_get_parent(struct dentry *child);
2811 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2812 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2813 f2fs_hash_t namehash, int *max_slots,
2814 struct f2fs_dentry_ptr *d);
2815 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2816 unsigned int start_pos, struct fscrypt_str *fstr);
2817 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2818 struct f2fs_dentry_ptr *d);
2819 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2820 const struct qstr *new_name,
2821 const struct qstr *orig_name, struct page *dpage);
2822 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2823 unsigned int current_depth);
2824 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2825 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2826 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2827 struct fscrypt_name *fname, struct page **res_page);
2828 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2829 const struct qstr *child, struct page **res_page);
2830 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2831 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2832 struct page **page);
2833 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2834 struct page *page, struct inode *inode);
2835 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2836 const struct qstr *name, f2fs_hash_t name_hash,
2837 unsigned int bit_pos);
2838 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2839 const struct qstr *orig_name,
2840 struct inode *inode, nid_t ino, umode_t mode);
2841 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2842 struct inode *inode, nid_t ino, umode_t mode);
2843 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2844 struct inode *inode, nid_t ino, umode_t mode);
2845 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2846 struct inode *dir, struct inode *inode);
2847 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2848 bool f2fs_empty_dir(struct inode *dir);
2850 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2852 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2853 inode, inode->i_ino, inode->i_mode);
2859 int f2fs_inode_dirtied(struct inode *inode, bool sync);
2860 void f2fs_inode_synced(struct inode *inode);
2861 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
2862 void f2fs_quota_off_umount(struct super_block *sb);
2863 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
2864 int f2fs_sync_fs(struct super_block *sb, int sync);
2865 extern __printf(3, 4)
2866 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2867 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
2872 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info,
2873 struct fscrypt_name *fname);
2878 struct dnode_of_data;
2881 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
2882 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
2883 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
2884 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
2885 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
2886 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
2887 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
2888 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
2889 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
2890 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
2891 struct node_info *ni);
2892 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
2893 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
2894 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
2895 int f2fs_truncate_xattr_node(struct inode *inode);
2896 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
2897 unsigned int seq_id);
2898 int f2fs_remove_inode_page(struct inode *inode);
2899 struct page *f2fs_new_inode_page(struct inode *inode);
2900 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
2901 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
2902 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
2903 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
2904 void f2fs_move_node_page(struct page *node_page, int gc_type);
2905 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
2906 struct writeback_control *wbc, bool atomic,
2907 unsigned int *seq_id);
2908 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
2909 struct writeback_control *wbc,
2910 bool do_balance, enum iostat_type io_type);
2911 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
2912 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
2913 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
2914 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
2915 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
2916 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
2917 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
2918 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
2919 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
2920 unsigned int segno, struct f2fs_summary_block *sum);
2921 void f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2922 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
2923 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
2924 int __init f2fs_create_node_manager_caches(void);
2925 void f2fs_destroy_node_manager_caches(void);
2930 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
2931 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
2932 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
2933 void f2fs_drop_inmem_pages(struct inode *inode);
2934 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
2935 int f2fs_commit_inmem_pages(struct inode *inode);
2936 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
2937 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
2938 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
2939 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
2940 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
2941 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
2942 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
2943 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
2944 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
2945 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
2946 bool f2fs_wait_discard_bios(struct f2fs_sb_info *sbi);
2947 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
2948 struct cp_control *cpc);
2949 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
2950 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
2951 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
2952 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
2953 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
2954 struct cp_control *cpc);
2955 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
2956 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
2958 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
2959 enum iostat_type io_type);
2960 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
2961 void f2fs_outplace_write_data(struct dnode_of_data *dn,
2962 struct f2fs_io_info *fio);
2963 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
2964 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
2965 block_t old_blkaddr, block_t new_blkaddr,
2966 bool recover_curseg, bool recover_newaddr);
2967 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
2968 block_t old_addr, block_t new_addr,
2969 unsigned char version, bool recover_curseg,
2970 bool recover_newaddr);
2971 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
2972 block_t old_blkaddr, block_t *new_blkaddr,
2973 struct f2fs_summary *sum, int type,
2974 struct f2fs_io_info *fio, bool add_list);
2975 void f2fs_wait_on_page_writeback(struct page *page,
2976 enum page_type type, bool ordered);
2977 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
2978 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
2979 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
2980 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
2981 unsigned int val, int alloc);
2982 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2983 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
2984 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
2985 int __init f2fs_create_segment_manager_caches(void);
2986 void f2fs_destroy_segment_manager_caches(void);
2987 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
2988 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
2989 enum page_type type, enum temp_type temp);
2994 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
2995 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
2996 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
2997 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
2998 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
2999 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3000 block_t blkaddr, int type);
3001 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3002 int type, bool sync);
3003 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3004 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3005 long nr_to_write, enum iostat_type io_type);
3006 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3007 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3008 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3009 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3010 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3011 unsigned int devidx, int type);
3012 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3013 unsigned int devidx, int type);
3014 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3015 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3016 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3017 void f2fs_add_orphan_inode(struct inode *inode);
3018 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3019 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3020 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3021 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3022 void f2fs_remove_dirty_inode(struct inode *inode);
3023 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3024 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3025 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3026 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3027 int __init f2fs_create_checkpoint_caches(void);
3028 void f2fs_destroy_checkpoint_caches(void);
3033 int f2fs_init_post_read_processing(void);
3034 void f2fs_destroy_post_read_processing(void);
3035 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3036 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3037 struct inode *inode, nid_t ino, pgoff_t idx,
3038 enum page_type type);
3039 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3040 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3041 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3042 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3043 block_t blk_addr, struct bio *bio);
3044 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3045 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3046 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3047 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3048 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3049 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3050 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3051 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3052 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3053 int op_flags, bool for_write);
3054 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3055 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3057 struct page *f2fs_get_new_data_page(struct inode *inode,
3058 struct page *ipage, pgoff_t index, bool new_i_size);
3059 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3060 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3061 int create, int flag);
3062 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3063 u64 start, u64 len);
3064 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3065 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3066 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3067 unsigned int length);
3068 int f2fs_release_page(struct page *page, gfp_t wait);
3069 #ifdef CONFIG_MIGRATION
3070 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3071 struct page *page, enum migrate_mode mode);
3073 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3074 void f2fs_clear_radix_tree_dirty_tag(struct page *page);
3079 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3080 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3081 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3082 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3083 unsigned int segno);
3084 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3089 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3090 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3095 #ifdef CONFIG_F2FS_STAT_FS
3096 struct f2fs_stat_info {
3097 struct list_head stat_list;
3098 struct f2fs_sb_info *sbi;
3099 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3100 int main_area_segs, main_area_sections, main_area_zones;
3101 unsigned long long hit_largest, hit_cached, hit_rbtree;
3102 unsigned long long hit_total, total_ext;
3103 int ext_tree, zombie_tree, ext_node;
3104 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3105 int ndirty_data, ndirty_qdata;
3107 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3108 int nats, dirty_nats, sits, dirty_sits;
3109 int free_nids, avail_nids, alloc_nids;
3110 int total_count, utilization;
3111 int bg_gc, nr_wb_cp_data, nr_wb_data;
3112 int nr_flushing, nr_flushed, flush_list_empty;
3113 int nr_discarding, nr_discarded;
3115 unsigned int undiscard_blks;
3116 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3117 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3118 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3119 unsigned int bimodal, avg_vblocks;
3120 int util_free, util_valid, util_invalid;
3121 int rsvd_segs, overp_segs;
3122 int dirty_count, node_pages, meta_pages;
3123 int prefree_count, call_count, cp_count, bg_cp_count;
3124 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3125 int bg_node_segs, bg_data_segs;
3126 int tot_blks, data_blks, node_blks;
3127 int bg_data_blks, bg_node_blks;
3128 unsigned long long skipped_atomic_files[2];
3129 int curseg[NR_CURSEG_TYPE];
3130 int cursec[NR_CURSEG_TYPE];
3131 int curzone[NR_CURSEG_TYPE];
3133 unsigned int segment_count[2];
3134 unsigned int block_count[2];
3135 unsigned int inplace_count;
3136 unsigned long long base_mem, cache_mem, page_mem;
3139 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3141 return (struct f2fs_stat_info *)sbi->stat_info;
3144 #define stat_inc_cp_count(si) ((si)->cp_count++)
3145 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3146 #define stat_inc_call_count(si) ((si)->call_count++)
3147 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3148 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3149 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3150 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3151 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3152 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3153 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3154 #define stat_inc_inline_xattr(inode) \
3156 if (f2fs_has_inline_xattr(inode)) \
3157 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3159 #define stat_dec_inline_xattr(inode) \
3161 if (f2fs_has_inline_xattr(inode)) \
3162 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3164 #define stat_inc_inline_inode(inode) \
3166 if (f2fs_has_inline_data(inode)) \
3167 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3169 #define stat_dec_inline_inode(inode) \
3171 if (f2fs_has_inline_data(inode)) \
3172 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3174 #define stat_inc_inline_dir(inode) \
3176 if (f2fs_has_inline_dentry(inode)) \
3177 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3179 #define stat_dec_inline_dir(inode) \
3181 if (f2fs_has_inline_dentry(inode)) \
3182 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3184 #define stat_inc_seg_type(sbi, curseg) \
3185 ((sbi)->segment_count[(curseg)->alloc_type]++)
3186 #define stat_inc_block_count(sbi, curseg) \
3187 ((sbi)->block_count[(curseg)->alloc_type]++)
3188 #define stat_inc_inplace_blocks(sbi) \
3189 (atomic_inc(&(sbi)->inplace_count))
3190 #define stat_inc_atomic_write(inode) \
3191 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3192 #define stat_dec_atomic_write(inode) \
3193 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3194 #define stat_update_max_atomic_write(inode) \
3196 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3197 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3199 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3201 #define stat_inc_volatile_write(inode) \
3202 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3203 #define stat_dec_volatile_write(inode) \
3204 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3205 #define stat_update_max_volatile_write(inode) \
3207 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3208 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3210 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3212 #define stat_inc_seg_count(sbi, type, gc_type) \
3214 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3216 if ((type) == SUM_TYPE_DATA) { \
3218 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3221 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3225 #define stat_inc_tot_blk_count(si, blks) \
3226 ((si)->tot_blks += (blks))
3228 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3230 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3231 stat_inc_tot_blk_count(si, blks); \
3232 si->data_blks += (blks); \
3233 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3236 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3238 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3239 stat_inc_tot_blk_count(si, blks); \
3240 si->node_blks += (blks); \
3241 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3244 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3245 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3246 int __init f2fs_create_root_stats(void);
3247 void f2fs_destroy_root_stats(void);
3249 #define stat_inc_cp_count(si) do { } while (0)
3250 #define stat_inc_bg_cp_count(si) do { } while (0)
3251 #define stat_inc_call_count(si) do { } while (0)
3252 #define stat_inc_bggc_count(si) do { } while (0)
3253 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3254 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3255 #define stat_inc_total_hit(sb) do { } while (0)
3256 #define stat_inc_rbtree_node_hit(sb) do { } while (0)
3257 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3258 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3259 #define stat_inc_inline_xattr(inode) do { } while (0)
3260 #define stat_dec_inline_xattr(inode) do { } while (0)
3261 #define stat_inc_inline_inode(inode) do { } while (0)
3262 #define stat_dec_inline_inode(inode) do { } while (0)
3263 #define stat_inc_inline_dir(inode) do { } while (0)
3264 #define stat_dec_inline_dir(inode) do { } while (0)
3265 #define stat_inc_atomic_write(inode) do { } while (0)
3266 #define stat_dec_atomic_write(inode) do { } while (0)
3267 #define stat_update_max_atomic_write(inode) do { } while (0)
3268 #define stat_inc_volatile_write(inode) do { } while (0)
3269 #define stat_dec_volatile_write(inode) do { } while (0)
3270 #define stat_update_max_volatile_write(inode) do { } while (0)
3271 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3272 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3273 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3274 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3275 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3276 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3277 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3279 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3280 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3281 static inline int __init f2fs_create_root_stats(void) { return 0; }
3282 static inline void f2fs_destroy_root_stats(void) { }
3285 extern const struct file_operations f2fs_dir_operations;
3286 extern const struct file_operations f2fs_file_operations;
3287 extern const struct inode_operations f2fs_file_inode_operations;
3288 extern const struct address_space_operations f2fs_dblock_aops;
3289 extern const struct address_space_operations f2fs_node_aops;
3290 extern const struct address_space_operations f2fs_meta_aops;
3291 extern const struct inode_operations f2fs_dir_inode_operations;
3292 extern const struct inode_operations f2fs_symlink_inode_operations;
3293 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3294 extern const struct inode_operations f2fs_special_inode_operations;
3295 extern struct kmem_cache *f2fs_inode_entry_slab;
3300 bool f2fs_may_inline_data(struct inode *inode);
3301 bool f2fs_may_inline_dentry(struct inode *inode);
3302 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3303 void f2fs_truncate_inline_inode(struct inode *inode,
3304 struct page *ipage, u64 from);
3305 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3306 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3307 int f2fs_convert_inline_inode(struct inode *inode);
3308 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3309 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3310 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3311 struct fscrypt_name *fname, struct page **res_page);
3312 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3313 struct page *ipage);
3314 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3315 const struct qstr *orig_name,
3316 struct inode *inode, nid_t ino, umode_t mode);
3317 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3318 struct page *page, struct inode *dir,
3319 struct inode *inode);
3320 bool f2fs_empty_inline_dir(struct inode *dir);
3321 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3322 struct fscrypt_str *fstr);
3323 int f2fs_inline_data_fiemap(struct inode *inode,
3324 struct fiemap_extent_info *fieinfo,
3325 __u64 start, __u64 len);
3330 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3331 struct shrink_control *sc);
3332 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3333 struct shrink_control *sc);
3334 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3335 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3340 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root *root,
3341 struct rb_entry *cached_re, unsigned int ofs);
3342 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3343 struct rb_root *root, struct rb_node **parent,
3345 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root *root,
3346 struct rb_entry *cached_re, unsigned int ofs,
3347 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3348 struct rb_node ***insert_p, struct rb_node **insert_parent,
3350 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3351 struct rb_root *root);
3352 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3353 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3354 void f2fs_drop_extent_tree(struct inode *inode);
3355 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3356 void f2fs_destroy_extent_tree(struct inode *inode);
3357 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3358 struct extent_info *ei);
3359 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3360 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3361 pgoff_t fofs, block_t blkaddr, unsigned int len);
3362 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3363 int __init f2fs_create_extent_cache(void);
3364 void f2fs_destroy_extent_cache(void);
3369 int __init f2fs_init_sysfs(void);
3370 void f2fs_exit_sysfs(void);
3371 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3372 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3377 static inline bool f2fs_encrypted_inode(struct inode *inode)
3379 return file_is_encrypt(inode);
3382 static inline bool f2fs_encrypted_file(struct inode *inode)
3384 return f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode);
3387 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3389 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3390 file_set_encrypt(inode);
3391 f2fs_set_inode_flags(inode);
3396 * Returns true if the reads of the inode's data need to undergo some
3397 * postprocessing step, like decryption or authenticity verification.
3399 static inline bool f2fs_post_read_required(struct inode *inode)
3401 return f2fs_encrypted_file(inode);
3404 #define F2FS_FEATURE_FUNCS(name, flagname) \
3405 static inline int f2fs_sb_has_##name(struct super_block *sb) \
3407 return F2FS_HAS_FEATURE(sb, F2FS_FEATURE_##flagname); \
3410 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3411 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3412 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3413 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3414 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3415 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3416 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3417 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3418 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3420 #ifdef CONFIG_BLK_DEV_ZONED
3421 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3422 struct block_device *bdev, block_t blkaddr)
3424 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3427 for (i = 0; i < sbi->s_ndevs; i++)
3428 if (FDEV(i).bdev == bdev)
3429 return FDEV(i).blkz_type[zno];
3434 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3436 return f2fs_sb_has_blkzoned(sbi->sb);
3439 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3441 return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
3444 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3446 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3447 f2fs_hw_should_discard(sbi);
3450 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3452 clear_opt(sbi, ADAPTIVE);
3453 clear_opt(sbi, LFS);
3456 case F2FS_MOUNT_ADAPTIVE:
3457 set_opt(sbi, ADAPTIVE);
3459 case F2FS_MOUNT_LFS:
3465 static inline bool f2fs_may_encrypt(struct inode *inode)
3467 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3468 umode_t mode = inode->i_mode;
3470 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3476 static inline bool f2fs_force_buffered_io(struct inode *inode, int rw)
3478 return (f2fs_post_read_required(inode) ||
3479 (rw == WRITE && test_opt(F2FS_I_SB(inode), LFS)) ||
3480 f2fs_is_multi_device(F2FS_I_SB(inode)));
3483 #ifdef CONFIG_F2FS_FAULT_INJECTION
3484 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3487 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3492 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3493 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */