1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
11 #include <linux/uio.h>
12 #include <linux/types.h>
13 #include <linux/page-flags.h>
14 #include <linux/buffer_head.h>
15 #include <linux/slab.h>
16 #include <linux/crc32.h>
17 #include <linux/magic.h>
18 #include <linux/kobject.h>
19 #include <linux/sched.h>
20 #include <linux/cred.h>
21 #include <linux/vmalloc.h>
22 #include <linux/bio.h>
23 #include <linux/blkdev.h>
24 #include <linux/quotaops.h>
25 #include <crypto/hash.h>
27 #include <linux/fscrypt.h>
29 #ifdef CONFIG_F2FS_CHECK_FS
30 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
32 #define f2fs_bug_on(sbi, condition) \
34 if (unlikely(condition)) { \
36 set_sbi_flag(sbi, SBI_NEED_FSCK); \
60 #ifdef CONFIG_F2FS_FAULT_INJECTION
61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
63 struct f2fs_fault_info {
65 unsigned int inject_rate;
66 unsigned int inject_type;
69 extern const char *f2fs_fault_name[FAULT_MAX];
70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
76 #define F2FS_MOUNT_BG_GC 0x00000001
77 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
78 #define F2FS_MOUNT_DISCARD 0x00000004
79 #define F2FS_MOUNT_NOHEAP 0x00000008
80 #define F2FS_MOUNT_XATTR_USER 0x00000010
81 #define F2FS_MOUNT_POSIX_ACL 0x00000020
82 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
83 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
84 #define F2FS_MOUNT_INLINE_DATA 0x00000100
85 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
86 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
87 #define F2FS_MOUNT_NOBARRIER 0x00000800
88 #define F2FS_MOUNT_FASTBOOT 0x00001000
89 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
90 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000
91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
93 #define F2FS_MOUNT_ADAPTIVE 0x00020000
94 #define F2FS_MOUNT_LFS 0x00040000
95 #define F2FS_MOUNT_USRQUOTA 0x00080000
96 #define F2FS_MOUNT_GRPQUOTA 0x00100000
97 #define F2FS_MOUNT_PRJQUOTA 0x00200000
98 #define F2FS_MOUNT_QUOTA 0x00400000
99 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
100 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
101 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
109 typecheck(unsigned long long, b) && \
110 ((long long)((a) - (b)) > 0))
112 typedef u32 block_t; /*
113 * should not change u32, since it is the on-disk block
114 * address format, __le32.
118 struct f2fs_mount_info {
120 int write_io_size_bits; /* Write IO size bits */
121 block_t root_reserved_blocks; /* root reserved blocks */
122 kuid_t s_resuid; /* reserved blocks for uid */
123 kgid_t s_resgid; /* reserved blocks for gid */
124 int active_logs; /* # of active logs */
125 int inline_xattr_size; /* inline xattr size */
126 #ifdef CONFIG_F2FS_FAULT_INJECTION
127 struct f2fs_fault_info fault_info; /* For fault injection */
130 /* Names of quota files with journalled quota */
131 char *s_qf_names[MAXQUOTAS];
132 int s_jquota_fmt; /* Format of quota to use */
134 /* For which write hints are passed down to block layer */
136 int alloc_mode; /* segment allocation policy */
137 int fsync_mode; /* fsync policy */
138 bool test_dummy_encryption; /* test dummy encryption */
141 #define F2FS_FEATURE_ENCRYPT 0x0001
142 #define F2FS_FEATURE_BLKZONED 0x0002
143 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
144 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
145 #define F2FS_FEATURE_PRJQUOTA 0x0010
146 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
147 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
148 #define F2FS_FEATURE_QUOTA_INO 0x0080
149 #define F2FS_FEATURE_INODE_CRTIME 0x0100
150 #define F2FS_FEATURE_LOST_FOUND 0x0200
151 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */
152 #define F2FS_FEATURE_SB_CHKSUM 0x0800
154 #define __F2FS_HAS_FEATURE(raw_super, mask) \
155 ((raw_super->feature & cpu_to_le32(mask)) != 0)
156 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
157 #define F2FS_SET_FEATURE(sbi, mask) \
158 (sbi->raw_super->feature |= cpu_to_le32(mask))
159 #define F2FS_CLEAR_FEATURE(sbi, mask) \
160 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
163 * Default values for user and/or group using reserved blocks
165 #define F2FS_DEF_RESUID 0
166 #define F2FS_DEF_RESGID 0
169 * For checkpoint manager
176 #define CP_UMOUNT 0x00000001
177 #define CP_FASTBOOT 0x00000002
178 #define CP_SYNC 0x00000004
179 #define CP_RECOVERY 0x00000008
180 #define CP_DISCARD 0x00000010
181 #define CP_TRIMMED 0x00000020
182 #define CP_PAUSE 0x00000040
184 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
185 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
186 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
187 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
188 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
189 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
190 #define DEF_CP_INTERVAL 60 /* 60 secs */
191 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
192 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
202 * indicate meta/data type
215 /* for the list of ino */
217 ORPHAN_INO, /* for orphan ino list */
218 APPEND_INO, /* for append ino list */
219 UPDATE_INO, /* for update ino list */
220 TRANS_DIR_INO, /* for trasactions dir ino list */
221 FLUSH_INO, /* for multiple device flushing */
222 MAX_INO_ENTRY, /* max. list */
226 struct list_head list; /* list head */
227 nid_t ino; /* inode number */
228 unsigned int dirty_device; /* dirty device bitmap */
231 /* for the list of inodes to be GCed */
233 struct list_head list; /* list head */
234 struct inode *inode; /* vfs inode pointer */
237 struct fsync_node_entry {
238 struct list_head list; /* list head */
239 struct page *page; /* warm node page pointer */
240 unsigned int seq_id; /* sequence id */
243 /* for the bitmap indicate blocks to be discarded */
244 struct discard_entry {
245 struct list_head list; /* list head */
246 block_t start_blkaddr; /* start blockaddr of current segment */
247 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
250 /* default discard granularity of inner discard thread, unit: block count */
251 #define DEFAULT_DISCARD_GRANULARITY 16
253 /* max discard pend list number */
254 #define MAX_PLIST_NUM 512
255 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
256 (MAX_PLIST_NUM - 1) : (blk_num - 1))
259 D_PREP, /* initial */
260 D_PARTIAL, /* partially submitted */
261 D_SUBMIT, /* all submitted */
262 D_DONE, /* finished */
265 struct discard_info {
266 block_t lstart; /* logical start address */
267 block_t len; /* length */
268 block_t start; /* actual start address in dev */
272 struct rb_node rb_node; /* rb node located in rb-tree */
275 block_t lstart; /* logical start address */
276 block_t len; /* length */
277 block_t start; /* actual start address in dev */
279 struct discard_info di; /* discard info */
282 struct list_head list; /* command list */
283 struct completion wait; /* compleation */
284 struct block_device *bdev; /* bdev */
285 unsigned short ref; /* reference count */
286 unsigned char state; /* state */
287 unsigned char queued; /* queued discard */
288 int error; /* bio error */
289 spinlock_t lock; /* for state/bio_ref updating */
290 unsigned short bio_ref; /* bio reference count */
301 struct discard_policy {
302 int type; /* type of discard */
303 unsigned int min_interval; /* used for candidates exist */
304 unsigned int mid_interval; /* used for device busy */
305 unsigned int max_interval; /* used for candidates not exist */
306 unsigned int max_requests; /* # of discards issued per round */
307 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
308 bool io_aware; /* issue discard in idle time */
309 bool sync; /* submit discard with REQ_SYNC flag */
310 bool ordered; /* issue discard by lba order */
311 unsigned int granularity; /* discard granularity */
314 struct discard_cmd_control {
315 struct task_struct *f2fs_issue_discard; /* discard thread */
316 struct list_head entry_list; /* 4KB discard entry list */
317 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
318 struct list_head wait_list; /* store on-flushing entries */
319 struct list_head fstrim_list; /* in-flight discard from fstrim */
320 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
321 unsigned int discard_wake; /* to wake up discard thread */
322 struct mutex cmd_lock;
323 unsigned int nr_discards; /* # of discards in the list */
324 unsigned int max_discards; /* max. discards to be issued */
325 unsigned int discard_granularity; /* discard granularity */
326 unsigned int undiscard_blks; /* # of undiscard blocks */
327 unsigned int next_pos; /* next discard position */
328 atomic_t issued_discard; /* # of issued discard */
329 atomic_t queued_discard; /* # of queued discard */
330 atomic_t discard_cmd_cnt; /* # of cached cmd count */
331 struct rb_root_cached root; /* root of discard rb-tree */
332 bool rbtree_check; /* config for consistence check */
335 /* for the list of fsync inodes, used only during recovery */
336 struct fsync_inode_entry {
337 struct list_head list; /* list head */
338 struct inode *inode; /* vfs inode pointer */
339 block_t blkaddr; /* block address locating the last fsync */
340 block_t last_dentry; /* block address locating the last dentry */
343 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
344 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
346 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
347 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
348 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
349 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
351 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
352 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
354 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
356 int before = nats_in_cursum(journal);
358 journal->n_nats = cpu_to_le16(before + i);
362 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
364 int before = sits_in_cursum(journal);
366 journal->n_sits = cpu_to_le16(before + i);
370 static inline bool __has_cursum_space(struct f2fs_journal *journal,
373 if (type == NAT_JOURNAL)
374 return size <= MAX_NAT_JENTRIES(journal);
375 return size <= MAX_SIT_JENTRIES(journal);
381 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
382 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
383 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
385 #define F2FS_IOCTL_MAGIC 0xf5
386 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
387 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
388 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
389 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
390 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
391 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
392 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
393 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
394 struct f2fs_defragment)
395 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
396 struct f2fs_move_range)
397 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
398 struct f2fs_flush_device)
399 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
400 struct f2fs_gc_range)
401 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
402 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
403 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
404 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
406 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
407 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
408 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
411 * should be same as XFS_IOC_GOINGDOWN.
412 * Flags for going down operation used by FS_IOC_GOINGDOWN
414 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
415 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
416 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
417 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
418 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
419 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
421 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
423 * ioctl commands in 32 bit emulation
425 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
426 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
427 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
430 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
431 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
433 struct f2fs_gc_range {
439 struct f2fs_defragment {
444 struct f2fs_move_range {
445 u32 dst_fd; /* destination fd */
446 u64 pos_in; /* start position in src_fd */
447 u64 pos_out; /* start position in dst_fd */
448 u64 len; /* size to move */
451 struct f2fs_flush_device {
452 u32 dev_num; /* device number to flush */
453 u32 segments; /* # of segments to flush */
456 /* for inline stuff */
457 #define DEF_INLINE_RESERVED_SIZE 1
458 #define DEF_MIN_INLINE_SIZE 1
459 static inline int get_extra_isize(struct inode *inode);
460 static inline int get_inline_xattr_addrs(struct inode *inode);
461 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
462 (CUR_ADDRS_PER_INODE(inode) - \
463 get_inline_xattr_addrs(inode) - \
464 DEF_INLINE_RESERVED_SIZE))
467 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
468 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
470 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \
471 BITS_PER_BYTE - 1) / BITS_PER_BYTE)
472 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
473 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
474 NR_INLINE_DENTRY(inode) + \
475 INLINE_DENTRY_BITMAP_SIZE(inode)))
478 * For INODE and NODE manager
480 /* for directory operations */
481 struct f2fs_dentry_ptr {
484 struct f2fs_dir_entry *dentry;
485 __u8 (*filename)[F2FS_SLOT_LEN];
490 static inline void make_dentry_ptr_block(struct inode *inode,
491 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
494 d->max = NR_DENTRY_IN_BLOCK;
495 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
496 d->bitmap = t->dentry_bitmap;
497 d->dentry = t->dentry;
498 d->filename = t->filename;
501 static inline void make_dentry_ptr_inline(struct inode *inode,
502 struct f2fs_dentry_ptr *d, void *t)
504 int entry_cnt = NR_INLINE_DENTRY(inode);
505 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
506 int reserved_size = INLINE_RESERVED_SIZE(inode);
510 d->nr_bitmap = bitmap_size;
512 d->dentry = t + bitmap_size + reserved_size;
513 d->filename = t + bitmap_size + reserved_size +
514 SIZE_OF_DIR_ENTRY * entry_cnt;
518 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
519 * as its node offset to distinguish from index node blocks.
520 * But some bits are used to mark the node block.
522 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
525 ALLOC_NODE, /* allocate a new node page if needed */
526 LOOKUP_NODE, /* look up a node without readahead */
528 * look up a node with readahead called
533 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
535 /* maximum retry quota flush count */
536 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
538 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
540 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
542 /* for in-memory extent cache entry */
543 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
545 /* number of extent info in extent cache we try to shrink */
546 #define EXTENT_CACHE_SHRINK_NUMBER 128
549 struct rb_node rb_node; /* rb node located in rb-tree */
550 unsigned int ofs; /* start offset of the entry */
551 unsigned int len; /* length of the entry */
555 unsigned int fofs; /* start offset in a file */
556 unsigned int len; /* length of the extent */
557 u32 blk; /* start block address of the extent */
561 struct rb_node rb_node; /* rb node located in rb-tree */
562 struct extent_info ei; /* extent info */
563 struct list_head list; /* node in global extent list of sbi */
564 struct extent_tree *et; /* extent tree pointer */
568 nid_t ino; /* inode number */
569 struct rb_root_cached root; /* root of extent info rb-tree */
570 struct extent_node *cached_en; /* recently accessed extent node */
571 struct extent_info largest; /* largested extent info */
572 struct list_head list; /* to be used by sbi->zombie_list */
573 rwlock_t lock; /* protect extent info rb-tree */
574 atomic_t node_cnt; /* # of extent node in rb-tree*/
575 bool largest_updated; /* largest extent updated */
579 * This structure is taken from ext4_map_blocks.
581 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
583 #define F2FS_MAP_NEW (1 << BH_New)
584 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
585 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
586 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
589 struct f2fs_map_blocks {
593 unsigned int m_flags;
594 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
595 pgoff_t *m_next_extent; /* point to next possible extent */
597 bool m_may_create; /* indicate it is from write path */
600 /* for flag in get_data_block */
602 F2FS_GET_BLOCK_DEFAULT,
603 F2FS_GET_BLOCK_FIEMAP,
606 F2FS_GET_BLOCK_PRE_DIO,
607 F2FS_GET_BLOCK_PRE_AIO,
608 F2FS_GET_BLOCK_PRECACHE,
612 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
614 #define FADVISE_COLD_BIT 0x01
615 #define FADVISE_LOST_PINO_BIT 0x02
616 #define FADVISE_ENCRYPT_BIT 0x04
617 #define FADVISE_ENC_NAME_BIT 0x08
618 #define FADVISE_KEEP_SIZE_BIT 0x10
619 #define FADVISE_HOT_BIT 0x20
620 #define FADVISE_VERITY_BIT 0x40 /* reserved */
622 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
624 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
625 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
626 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
627 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
628 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
629 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
630 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
631 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
632 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
633 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
634 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
635 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
636 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
637 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
638 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
639 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
641 #define DEF_DIR_LEVEL 0
649 struct f2fs_inode_info {
650 struct inode vfs_inode; /* serve a vfs inode */
651 unsigned long i_flags; /* keep an inode flags for ioctl */
652 unsigned char i_advise; /* use to give file attribute hints */
653 unsigned char i_dir_level; /* use for dentry level for large dir */
654 unsigned int i_current_depth; /* only for directory depth */
655 /* for gc failure statistic */
656 unsigned int i_gc_failures[MAX_GC_FAILURE];
657 unsigned int i_pino; /* parent inode number */
658 umode_t i_acl_mode; /* keep file acl mode temporarily */
660 /* Use below internally in f2fs*/
661 unsigned long flags; /* use to pass per-file flags */
662 struct rw_semaphore i_sem; /* protect fi info */
663 atomic_t dirty_pages; /* # of dirty pages */
664 f2fs_hash_t chash; /* hash value of given file name */
665 unsigned int clevel; /* maximum level of given file name */
666 struct task_struct *task; /* lookup and create consistency */
667 struct task_struct *cp_task; /* separate cp/wb IO stats*/
668 nid_t i_xattr_nid; /* node id that contains xattrs */
669 loff_t last_disk_size; /* lastly written file size */
672 struct dquot *i_dquot[MAXQUOTAS];
674 /* quota space reservation, managed internally by quota code */
675 qsize_t i_reserved_quota;
677 struct list_head dirty_list; /* dirty list for dirs and files */
678 struct list_head gdirty_list; /* linked in global dirty list */
679 struct list_head inmem_ilist; /* list for inmem inodes */
680 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
681 struct task_struct *inmem_task; /* store inmemory task */
682 struct mutex inmem_lock; /* lock for inmemory pages */
683 struct extent_tree *extent_tree; /* cached extent_tree entry */
685 /* avoid racing between foreground op and gc */
686 struct rw_semaphore i_gc_rwsem[2];
687 struct rw_semaphore i_mmap_sem;
688 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
690 int i_extra_isize; /* size of extra space located in i_addr */
691 kprojid_t i_projid; /* id for project quota */
692 int i_inline_xattr_size; /* inline xattr size */
693 struct timespec64 i_crtime; /* inode creation time */
694 struct timespec64 i_disk_time[4];/* inode disk times */
697 static inline void get_extent_info(struct extent_info *ext,
698 struct f2fs_extent *i_ext)
700 ext->fofs = le32_to_cpu(i_ext->fofs);
701 ext->blk = le32_to_cpu(i_ext->blk);
702 ext->len = le32_to_cpu(i_ext->len);
705 static inline void set_raw_extent(struct extent_info *ext,
706 struct f2fs_extent *i_ext)
708 i_ext->fofs = cpu_to_le32(ext->fofs);
709 i_ext->blk = cpu_to_le32(ext->blk);
710 i_ext->len = cpu_to_le32(ext->len);
713 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
714 u32 blk, unsigned int len)
721 static inline bool __is_discard_mergeable(struct discard_info *back,
722 struct discard_info *front, unsigned int max_len)
724 return (back->lstart + back->len == front->lstart) &&
725 (back->len + front->len <= max_len);
728 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
729 struct discard_info *back, unsigned int max_len)
731 return __is_discard_mergeable(back, cur, max_len);
734 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
735 struct discard_info *front, unsigned int max_len)
737 return __is_discard_mergeable(cur, front, max_len);
740 static inline bool __is_extent_mergeable(struct extent_info *back,
741 struct extent_info *front)
743 return (back->fofs + back->len == front->fofs &&
744 back->blk + back->len == front->blk);
747 static inline bool __is_back_mergeable(struct extent_info *cur,
748 struct extent_info *back)
750 return __is_extent_mergeable(back, cur);
753 static inline bool __is_front_mergeable(struct extent_info *cur,
754 struct extent_info *front)
756 return __is_extent_mergeable(cur, front);
759 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
760 static inline void __try_update_largest_extent(struct extent_tree *et,
761 struct extent_node *en)
763 if (en->ei.len > et->largest.len) {
764 et->largest = en->ei;
765 et->largest_updated = true;
770 * For free nid management
773 FREE_NID, /* newly added to free nid list */
774 PREALLOC_NID, /* it is preallocated */
778 struct f2fs_nm_info {
779 block_t nat_blkaddr; /* base disk address of NAT */
780 nid_t max_nid; /* maximum possible node ids */
781 nid_t available_nids; /* # of available node ids */
782 nid_t next_scan_nid; /* the next nid to be scanned */
783 unsigned int ram_thresh; /* control the memory footprint */
784 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
785 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
787 /* NAT cache management */
788 struct radix_tree_root nat_root;/* root of the nat entry cache */
789 struct radix_tree_root nat_set_root;/* root of the nat set cache */
790 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
791 struct list_head nat_entries; /* cached nat entry list (clean) */
792 spinlock_t nat_list_lock; /* protect clean nat entry list */
793 unsigned int nat_cnt; /* the # of cached nat entries */
794 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
795 unsigned int nat_blocks; /* # of nat blocks */
797 /* free node ids management */
798 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
799 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
800 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
801 spinlock_t nid_list_lock; /* protect nid lists ops */
802 struct mutex build_lock; /* lock for build free nids */
803 unsigned char **free_nid_bitmap;
804 unsigned char *nat_block_bitmap;
805 unsigned short *free_nid_count; /* free nid count of NAT block */
808 char *nat_bitmap; /* NAT bitmap pointer */
810 unsigned int nat_bits_blocks; /* # of nat bits blocks */
811 unsigned char *nat_bits; /* NAT bits blocks */
812 unsigned char *full_nat_bits; /* full NAT pages */
813 unsigned char *empty_nat_bits; /* empty NAT pages */
814 #ifdef CONFIG_F2FS_CHECK_FS
815 char *nat_bitmap_mir; /* NAT bitmap mirror */
817 int bitmap_size; /* bitmap size */
821 * this structure is used as one of function parameters.
822 * all the information are dedicated to a given direct node block determined
823 * by the data offset in a file.
825 struct dnode_of_data {
826 struct inode *inode; /* vfs inode pointer */
827 struct page *inode_page; /* its inode page, NULL is possible */
828 struct page *node_page; /* cached direct node page */
829 nid_t nid; /* node id of the direct node block */
830 unsigned int ofs_in_node; /* data offset in the node page */
831 bool inode_page_locked; /* inode page is locked or not */
832 bool node_changed; /* is node block changed */
833 char cur_level; /* level of hole node page */
834 char max_level; /* level of current page located */
835 block_t data_blkaddr; /* block address of the node block */
838 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
839 struct page *ipage, struct page *npage, nid_t nid)
841 memset(dn, 0, sizeof(*dn));
843 dn->inode_page = ipage;
844 dn->node_page = npage;
851 * By default, there are 6 active log areas across the whole main area.
852 * When considering hot and cold data separation to reduce cleaning overhead,
853 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
855 * In the current design, you should not change the numbers intentionally.
856 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
857 * logs individually according to the underlying devices. (default: 6)
858 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
859 * data and 8 for node logs.
861 #define NR_CURSEG_DATA_TYPE (3)
862 #define NR_CURSEG_NODE_TYPE (3)
863 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
866 CURSEG_HOT_DATA = 0, /* directory entry blocks */
867 CURSEG_WARM_DATA, /* data blocks */
868 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
869 CURSEG_HOT_NODE, /* direct node blocks of directory files */
870 CURSEG_WARM_NODE, /* direct node blocks of normal files */
871 CURSEG_COLD_NODE, /* indirect node blocks */
876 struct completion wait;
877 struct llist_node llnode;
882 struct flush_cmd_control {
883 struct task_struct *f2fs_issue_flush; /* flush thread */
884 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
885 atomic_t issued_flush; /* # of issued flushes */
886 atomic_t queued_flush; /* # of queued flushes */
887 struct llist_head issue_list; /* list for command issue */
888 struct llist_node *dispatch_list; /* list for command dispatch */
891 struct f2fs_sm_info {
892 struct sit_info *sit_info; /* whole segment information */
893 struct free_segmap_info *free_info; /* free segment information */
894 struct dirty_seglist_info *dirty_info; /* dirty segment information */
895 struct curseg_info *curseg_array; /* active segment information */
897 struct rw_semaphore curseg_lock; /* for preventing curseg change */
899 block_t seg0_blkaddr; /* block address of 0'th segment */
900 block_t main_blkaddr; /* start block address of main area */
901 block_t ssa_blkaddr; /* start block address of SSA area */
903 unsigned int segment_count; /* total # of segments */
904 unsigned int main_segments; /* # of segments in main area */
905 unsigned int reserved_segments; /* # of reserved segments */
906 unsigned int ovp_segments; /* # of overprovision segments */
908 /* a threshold to reclaim prefree segments */
909 unsigned int rec_prefree_segments;
911 /* for batched trimming */
912 unsigned int trim_sections; /* # of sections to trim */
914 struct list_head sit_entry_set; /* sit entry set list */
916 unsigned int ipu_policy; /* in-place-update policy */
917 unsigned int min_ipu_util; /* in-place-update threshold */
918 unsigned int min_fsync_blocks; /* threshold for fsync */
919 unsigned int min_seq_blocks; /* threshold for sequential blocks */
920 unsigned int min_hot_blocks; /* threshold for hot block allocation */
921 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
923 /* for flush command control */
924 struct flush_cmd_control *fcc_info;
926 /* for discard command control */
927 struct discard_cmd_control *dcc_info;
934 * COUNT_TYPE for monitoring
936 * f2fs monitors the number of several block types such as on-writeback,
937 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
939 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
959 * The below are the page types of bios used in submit_bio().
960 * The available types are:
961 * DATA User data pages. It operates as async mode.
962 * NODE Node pages. It operates as async mode.
963 * META FS metadata pages such as SIT, NAT, CP.
964 * NR_PAGE_TYPE The number of page types.
965 * META_FLUSH Make sure the previous pages are written
966 * with waiting the bio's completion
967 * ... Only can be used with META.
969 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
976 INMEM, /* the below types are used by tracepoints only. */
985 HOT = 0, /* must be zero for meta bio */
991 enum need_lock_type {
997 enum cp_reason_type {
1011 APP_DIRECT_IO, /* app direct IOs */
1012 APP_BUFFERED_IO, /* app buffered IOs */
1013 APP_WRITE_IO, /* app write IOs */
1014 APP_MAPPED_IO, /* app mapped IOs */
1015 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1016 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1017 FS_META_IO, /* meta IOs from kworker/reclaimer */
1018 FS_GC_DATA_IO, /* data IOs from forground gc */
1019 FS_GC_NODE_IO, /* node IOs from forground gc */
1020 FS_CP_DATA_IO, /* data IOs from checkpoint */
1021 FS_CP_NODE_IO, /* node IOs from checkpoint */
1022 FS_CP_META_IO, /* meta IOs from checkpoint */
1023 FS_DISCARD, /* discard */
1027 struct f2fs_io_info {
1028 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1029 nid_t ino; /* inode number */
1030 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1031 enum temp_type temp; /* contains HOT/WARM/COLD */
1032 int op; /* contains REQ_OP_ */
1033 int op_flags; /* req_flag_bits */
1034 block_t new_blkaddr; /* new block address to be written */
1035 block_t old_blkaddr; /* old block address before Cow */
1036 struct page *page; /* page to be written */
1037 struct page *encrypted_page; /* encrypted page */
1038 struct list_head list; /* serialize IOs */
1039 bool submitted; /* indicate IO submission */
1040 int need_lock; /* indicate we need to lock cp_rwsem */
1041 bool in_list; /* indicate fio is in io_list */
1042 bool is_meta; /* indicate borrow meta inode mapping or not */
1043 bool retry; /* need to reallocate block address */
1044 enum iostat_type io_type; /* io type */
1045 struct writeback_control *io_wbc; /* writeback control */
1046 unsigned char version; /* version of the node */
1049 #define is_read_io(rw) ((rw) == READ)
1050 struct f2fs_bio_info {
1051 struct f2fs_sb_info *sbi; /* f2fs superblock */
1052 struct bio *bio; /* bios to merge */
1053 sector_t last_block_in_bio; /* last block number */
1054 struct f2fs_io_info fio; /* store buffered io info. */
1055 struct rw_semaphore io_rwsem; /* blocking op for bio */
1056 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1057 struct list_head io_list; /* track fios */
1060 #define FDEV(i) (sbi->devs[i])
1061 #define RDEV(i) (raw_super->devs[i])
1062 struct f2fs_dev_info {
1063 struct block_device *bdev;
1064 char path[MAX_PATH_LEN];
1065 unsigned int total_segments;
1068 #ifdef CONFIG_BLK_DEV_ZONED
1069 unsigned int nr_blkz; /* Total number of zones */
1070 u8 *blkz_type; /* Array of zones type */
1075 DIR_INODE, /* for dirty dir inode */
1076 FILE_INODE, /* for dirty regular/symlink inode */
1077 DIRTY_META, /* for all dirtied inode metadata */
1078 ATOMIC_FILE, /* for all atomic files */
1082 /* for inner inode cache management */
1083 struct inode_management {
1084 struct radix_tree_root ino_root; /* ino entry array */
1085 spinlock_t ino_lock; /* for ino entry lock */
1086 struct list_head ino_list; /* inode list head */
1087 unsigned long ino_num; /* number of entries */
1090 /* For s_flag in struct f2fs_sb_info */
1092 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1093 SBI_IS_CLOSE, /* specify unmounting */
1094 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1095 SBI_POR_DOING, /* recovery is doing or not */
1096 SBI_NEED_SB_WRITE, /* need to recover superblock */
1097 SBI_NEED_CP, /* need to checkpoint */
1098 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1099 SBI_IS_RECOVERED, /* recovered orphan/data */
1100 SBI_CP_DISABLED, /* CP was disabled last mount */
1101 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1102 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1103 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1123 WHINT_MODE_OFF, /* not pass down write hints */
1124 WHINT_MODE_USER, /* try to pass down hints given by users */
1125 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1129 ALLOC_MODE_DEFAULT, /* stay default */
1130 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1134 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1135 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1136 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1139 #ifdef CONFIG_FS_ENCRYPTION
1140 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1141 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1143 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1146 struct f2fs_sb_info {
1147 struct super_block *sb; /* pointer to VFS super block */
1148 struct proc_dir_entry *s_proc; /* proc entry */
1149 struct f2fs_super_block *raw_super; /* raw super block pointer */
1150 struct rw_semaphore sb_lock; /* lock for raw super block */
1151 int valid_super_block; /* valid super block no */
1152 unsigned long s_flag; /* flags for sbi */
1153 struct mutex writepages; /* mutex for writepages() */
1155 #ifdef CONFIG_BLK_DEV_ZONED
1156 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1157 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1160 /* for node-related operations */
1161 struct f2fs_nm_info *nm_info; /* node manager */
1162 struct inode *node_inode; /* cache node blocks */
1164 /* for segment-related operations */
1165 struct f2fs_sm_info *sm_info; /* segment manager */
1167 /* for bio operations */
1168 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1169 /* keep migration IO order for LFS mode */
1170 struct rw_semaphore io_order_lock;
1171 mempool_t *write_io_dummy; /* Dummy pages */
1173 /* for checkpoint */
1174 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1175 int cur_cp_pack; /* remain current cp pack */
1176 spinlock_t cp_lock; /* for flag in ckpt */
1177 struct inode *meta_inode; /* cache meta blocks */
1178 struct mutex cp_mutex; /* checkpoint procedure lock */
1179 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1180 struct rw_semaphore node_write; /* locking node writes */
1181 struct rw_semaphore node_change; /* locking node change */
1182 wait_queue_head_t cp_wait;
1183 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1184 long interval_time[MAX_TIME]; /* to store thresholds */
1186 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1188 spinlock_t fsync_node_lock; /* for node entry lock */
1189 struct list_head fsync_node_list; /* node list head */
1190 unsigned int fsync_seg_id; /* sequence id */
1191 unsigned int fsync_node_num; /* number of node entries */
1193 /* for orphan inode, use 0'th array */
1194 unsigned int max_orphans; /* max orphan inodes */
1196 /* for inode management */
1197 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1198 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1200 /* for extent tree cache */
1201 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1202 struct mutex extent_tree_lock; /* locking extent radix tree */
1203 struct list_head extent_list; /* lru list for shrinker */
1204 spinlock_t extent_lock; /* locking extent lru list */
1205 atomic_t total_ext_tree; /* extent tree count */
1206 struct list_head zombie_list; /* extent zombie tree list */
1207 atomic_t total_zombie_tree; /* extent zombie tree count */
1208 atomic_t total_ext_node; /* extent info count */
1210 /* basic filesystem units */
1211 unsigned int log_sectors_per_block; /* log2 sectors per block */
1212 unsigned int log_blocksize; /* log2 block size */
1213 unsigned int blocksize; /* block size */
1214 unsigned int root_ino_num; /* root inode number*/
1215 unsigned int node_ino_num; /* node inode number*/
1216 unsigned int meta_ino_num; /* meta inode number*/
1217 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1218 unsigned int blocks_per_seg; /* blocks per segment */
1219 unsigned int segs_per_sec; /* segments per section */
1220 unsigned int secs_per_zone; /* sections per zone */
1221 unsigned int total_sections; /* total section count */
1222 unsigned int total_node_count; /* total node block count */
1223 unsigned int total_valid_node_count; /* valid node block count */
1224 loff_t max_file_blocks; /* max block index of file */
1225 int dir_level; /* directory level */
1226 int readdir_ra; /* readahead inode in readdir */
1228 block_t user_block_count; /* # of user blocks */
1229 block_t total_valid_block_count; /* # of valid blocks */
1230 block_t discard_blks; /* discard command candidats */
1231 block_t last_valid_block_count; /* for recovery */
1232 block_t reserved_blocks; /* configurable reserved blocks */
1233 block_t current_reserved_blocks; /* current reserved blocks */
1235 /* Additional tracking for no checkpoint mode */
1236 block_t unusable_block_count; /* # of blocks saved by last cp */
1238 unsigned int nquota_files; /* # of quota sysfile */
1240 u32 s_next_generation; /* for NFS support */
1242 /* # of pages, see count_type */
1243 atomic_t nr_pages[NR_COUNT_TYPE];
1244 /* # of allocated blocks */
1245 struct percpu_counter alloc_valid_block_count;
1247 /* writeback control */
1248 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1250 /* valid inode count */
1251 struct percpu_counter total_valid_inode_count;
1253 struct f2fs_mount_info mount_opt; /* mount options */
1255 /* for cleaning operations */
1256 struct mutex gc_mutex; /* mutex for GC */
1257 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1258 unsigned int cur_victim_sec; /* current victim section num */
1259 unsigned int gc_mode; /* current GC state */
1260 unsigned int next_victim_seg[2]; /* next segment in victim section */
1261 /* for skip statistic */
1262 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1263 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1265 /* threshold for gc trials on pinned files */
1266 u64 gc_pin_file_threshold;
1268 /* maximum # of trials to find a victim segment for SSR and GC */
1269 unsigned int max_victim_search;
1270 /* migration granularity of garbage collection, unit: segment */
1271 unsigned int migration_granularity;
1274 * for stat information.
1275 * one is for the LFS mode, and the other is for the SSR mode.
1277 #ifdef CONFIG_F2FS_STAT_FS
1278 struct f2fs_stat_info *stat_info; /* FS status information */
1279 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1280 unsigned int segment_count[2]; /* # of allocated segments */
1281 unsigned int block_count[2]; /* # of allocated blocks */
1282 atomic_t inplace_count; /* # of inplace update */
1283 atomic64_t total_hit_ext; /* # of lookup extent cache */
1284 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1285 atomic64_t read_hit_largest; /* # of hit largest extent node */
1286 atomic64_t read_hit_cached; /* # of hit cached extent node */
1287 atomic_t inline_xattr; /* # of inline_xattr inodes */
1288 atomic_t inline_inode; /* # of inline_data inodes */
1289 atomic_t inline_dir; /* # of inline_dentry inodes */
1290 atomic_t aw_cnt; /* # of atomic writes */
1291 atomic_t vw_cnt; /* # of volatile writes */
1292 atomic_t max_aw_cnt; /* max # of atomic writes */
1293 atomic_t max_vw_cnt; /* max # of volatile writes */
1294 int bg_gc; /* background gc calls */
1295 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1296 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1297 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1299 spinlock_t stat_lock; /* lock for stat operations */
1301 /* For app/fs IO statistics */
1302 spinlock_t iostat_lock;
1303 unsigned long long write_iostat[NR_IO_TYPE];
1306 /* For sysfs suppport */
1307 struct kobject s_kobj;
1308 struct completion s_kobj_unregister;
1310 /* For shrinker support */
1311 struct list_head s_list;
1312 int s_ndevs; /* number of devices */
1313 struct f2fs_dev_info *devs; /* for device list */
1314 unsigned int dirty_device; /* for checkpoint data flush */
1315 spinlock_t dev_lock; /* protect dirty_device */
1316 struct mutex umount_mutex;
1317 unsigned int shrinker_run_no;
1319 /* For write statistics */
1320 u64 sectors_written_start;
1323 /* Reference to checksum algorithm driver via cryptoapi */
1324 struct crypto_shash *s_chksum_driver;
1326 /* Precomputed FS UUID checksum for seeding other checksums */
1327 __u32 s_chksum_seed;
1330 struct f2fs_private_dio {
1331 struct inode *inode;
1333 bio_end_io_t *orig_end_io;
1337 #ifdef CONFIG_F2FS_FAULT_INJECTION
1338 #define f2fs_show_injection_info(type) \
1339 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pF\n", \
1340 KERN_INFO, f2fs_fault_name[type], \
1341 __func__, __builtin_return_address(0))
1342 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1344 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1346 if (!ffi->inject_rate)
1349 if (!IS_FAULT_SET(ffi, type))
1352 atomic_inc(&ffi->inject_ops);
1353 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1354 atomic_set(&ffi->inject_ops, 0);
1360 #define f2fs_show_injection_info(type) do { } while (0)
1361 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1367 /* For write statistics. Suppose sector size is 512 bytes,
1368 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1370 #define BD_PART_WRITTEN(s) \
1371 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1372 (s)->sectors_written_start) >> 1)
1374 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1376 unsigned long now = jiffies;
1378 sbi->last_time[type] = now;
1380 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1381 if (type == REQ_TIME) {
1382 sbi->last_time[DISCARD_TIME] = now;
1383 sbi->last_time[GC_TIME] = now;
1387 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1389 unsigned long interval = sbi->interval_time[type] * HZ;
1391 return time_after(jiffies, sbi->last_time[type] + interval);
1394 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1397 unsigned long interval = sbi->interval_time[type] * HZ;
1398 unsigned int wait_ms = 0;
1401 delta = (sbi->last_time[type] + interval) - jiffies;
1403 wait_ms = jiffies_to_msecs(delta);
1411 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1412 const void *address, unsigned int length)
1415 struct shash_desc shash;
1420 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1422 desc.shash.tfm = sbi->s_chksum_driver;
1423 desc.shash.flags = 0;
1424 *(u32 *)desc.ctx = crc;
1426 err = crypto_shash_update(&desc.shash, address, length);
1429 return *(u32 *)desc.ctx;
1432 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1433 unsigned int length)
1435 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1438 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1439 void *buf, size_t buf_size)
1441 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1444 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1445 const void *address, unsigned int length)
1447 return __f2fs_crc32(sbi, crc, address, length);
1450 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1452 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1455 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1457 return sb->s_fs_info;
1460 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1462 return F2FS_SB(inode->i_sb);
1465 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1467 return F2FS_I_SB(mapping->host);
1470 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1472 return F2FS_M_SB(page->mapping);
1475 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1477 return (struct f2fs_super_block *)(sbi->raw_super);
1480 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1482 return (struct f2fs_checkpoint *)(sbi->ckpt);
1485 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1487 return (struct f2fs_node *)page_address(page);
1490 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1492 return &((struct f2fs_node *)page_address(page))->i;
1495 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1497 return (struct f2fs_nm_info *)(sbi->nm_info);
1500 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1502 return (struct f2fs_sm_info *)(sbi->sm_info);
1505 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1507 return (struct sit_info *)(SM_I(sbi)->sit_info);
1510 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1512 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1515 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1517 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1520 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1522 return sbi->meta_inode->i_mapping;
1525 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1527 return sbi->node_inode->i_mapping;
1530 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1532 return test_bit(type, &sbi->s_flag);
1535 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1537 set_bit(type, &sbi->s_flag);
1540 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1542 clear_bit(type, &sbi->s_flag);
1545 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1547 return le64_to_cpu(cp->checkpoint_ver);
1550 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1552 if (type < F2FS_MAX_QUOTAS)
1553 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1557 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1559 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1560 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1563 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1565 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1567 return ckpt_flags & f;
1570 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1572 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1575 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1577 unsigned int ckpt_flags;
1579 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1581 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1584 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1586 unsigned long flags;
1588 spin_lock_irqsave(&sbi->cp_lock, flags);
1589 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1590 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1593 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1595 unsigned int ckpt_flags;
1597 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1599 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1602 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1604 unsigned long flags;
1606 spin_lock_irqsave(&sbi->cp_lock, flags);
1607 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1608 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1611 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1613 unsigned long flags;
1616 * In order to re-enable nat_bits we need to call fsck.f2fs by
1617 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1618 * so let's rely on regular fsck or unclean shutdown.
1622 spin_lock_irqsave(&sbi->cp_lock, flags);
1623 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1624 kvfree(NM_I(sbi)->nat_bits);
1625 NM_I(sbi)->nat_bits = NULL;
1627 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1630 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1631 struct cp_control *cpc)
1633 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1635 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1638 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1640 down_read(&sbi->cp_rwsem);
1643 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1645 return down_read_trylock(&sbi->cp_rwsem);
1648 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1650 up_read(&sbi->cp_rwsem);
1653 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1655 down_write(&sbi->cp_rwsem);
1658 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1660 up_write(&sbi->cp_rwsem);
1663 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1665 int reason = CP_SYNC;
1667 if (test_opt(sbi, FASTBOOT))
1668 reason = CP_FASTBOOT;
1669 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1674 static inline bool __remain_node_summaries(int reason)
1676 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1679 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1681 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1682 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1686 * Check whether the inode has blocks or not
1688 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1690 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1692 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1695 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1697 return ofs == XATTR_NODE_OFFSET;
1700 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1701 struct inode *inode, bool cap)
1705 if (!test_opt(sbi, RESERVE_ROOT))
1707 if (IS_NOQUOTA(inode))
1709 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1711 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1712 in_group_p(F2FS_OPTION(sbi).s_resgid))
1714 if (cap && capable(CAP_SYS_RESOURCE))
1719 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1720 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1721 struct inode *inode, blkcnt_t *count)
1723 blkcnt_t diff = 0, release = 0;
1724 block_t avail_user_block_count;
1727 ret = dquot_reserve_block(inode, *count);
1731 if (time_to_inject(sbi, FAULT_BLOCK)) {
1732 f2fs_show_injection_info(FAULT_BLOCK);
1738 * let's increase this in prior to actual block count change in order
1739 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1741 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1743 spin_lock(&sbi->stat_lock);
1744 sbi->total_valid_block_count += (block_t)(*count);
1745 avail_user_block_count = sbi->user_block_count -
1746 sbi->current_reserved_blocks;
1748 if (!__allow_reserved_blocks(sbi, inode, true))
1749 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1750 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1751 avail_user_block_count -= sbi->unusable_block_count;
1752 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1753 diff = sbi->total_valid_block_count - avail_user_block_count;
1758 sbi->total_valid_block_count -= diff;
1760 spin_unlock(&sbi->stat_lock);
1764 spin_unlock(&sbi->stat_lock);
1766 if (unlikely(release)) {
1767 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1768 dquot_release_reservation_block(inode, release);
1770 f2fs_i_blocks_write(inode, *count, true, true);
1774 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1775 dquot_release_reservation_block(inode, release);
1779 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1780 struct inode *inode,
1783 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1785 spin_lock(&sbi->stat_lock);
1786 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1787 f2fs_bug_on(sbi, inode->i_blocks < sectors);
1788 sbi->total_valid_block_count -= (block_t)count;
1789 if (sbi->reserved_blocks &&
1790 sbi->current_reserved_blocks < sbi->reserved_blocks)
1791 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1792 sbi->current_reserved_blocks + count);
1793 spin_unlock(&sbi->stat_lock);
1794 f2fs_i_blocks_write(inode, count, false, true);
1797 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1799 atomic_inc(&sbi->nr_pages[count_type]);
1801 if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES ||
1802 count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA ||
1803 count_type == F2FS_RD_DATA || count_type == F2FS_RD_NODE ||
1804 count_type == F2FS_RD_META)
1807 set_sbi_flag(sbi, SBI_IS_DIRTY);
1810 static inline void inode_inc_dirty_pages(struct inode *inode)
1812 atomic_inc(&F2FS_I(inode)->dirty_pages);
1813 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1814 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1815 if (IS_NOQUOTA(inode))
1816 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1819 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1821 atomic_dec(&sbi->nr_pages[count_type]);
1824 static inline void inode_dec_dirty_pages(struct inode *inode)
1826 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1827 !S_ISLNK(inode->i_mode))
1830 atomic_dec(&F2FS_I(inode)->dirty_pages);
1831 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1832 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1833 if (IS_NOQUOTA(inode))
1834 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1837 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1839 return atomic_read(&sbi->nr_pages[count_type]);
1842 static inline int get_dirty_pages(struct inode *inode)
1844 return atomic_read(&F2FS_I(inode)->dirty_pages);
1847 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1849 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1850 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1851 sbi->log_blocks_per_seg;
1853 return segs / sbi->segs_per_sec;
1856 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1858 return sbi->total_valid_block_count;
1861 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1863 return sbi->discard_blks;
1866 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1868 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1870 /* return NAT or SIT bitmap */
1871 if (flag == NAT_BITMAP)
1872 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1873 else if (flag == SIT_BITMAP)
1874 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1879 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1881 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1884 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1886 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1889 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1890 offset = (flag == SIT_BITMAP) ?
1891 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1892 return &ckpt->sit_nat_version_bitmap + offset;
1895 if (__cp_payload(sbi) > 0) {
1896 if (flag == NAT_BITMAP)
1897 return &ckpt->sit_nat_version_bitmap;
1899 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1901 offset = (flag == NAT_BITMAP) ?
1902 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1903 return &ckpt->sit_nat_version_bitmap + offset;
1907 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1909 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1911 if (sbi->cur_cp_pack == 2)
1912 start_addr += sbi->blocks_per_seg;
1916 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1918 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1920 if (sbi->cur_cp_pack == 1)
1921 start_addr += sbi->blocks_per_seg;
1925 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1927 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1930 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1932 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1935 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
1936 struct inode *inode, bool is_inode)
1938 block_t valid_block_count;
1939 unsigned int valid_node_count;
1944 err = dquot_alloc_inode(inode);
1949 err = dquot_reserve_block(inode, 1);
1954 if (time_to_inject(sbi, FAULT_BLOCK)) {
1955 f2fs_show_injection_info(FAULT_BLOCK);
1959 spin_lock(&sbi->stat_lock);
1961 valid_block_count = sbi->total_valid_block_count +
1962 sbi->current_reserved_blocks + 1;
1964 if (!__allow_reserved_blocks(sbi, inode, false))
1965 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
1966 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1967 valid_block_count += sbi->unusable_block_count;
1969 if (unlikely(valid_block_count > sbi->user_block_count)) {
1970 spin_unlock(&sbi->stat_lock);
1974 valid_node_count = sbi->total_valid_node_count + 1;
1975 if (unlikely(valid_node_count > sbi->total_node_count)) {
1976 spin_unlock(&sbi->stat_lock);
1980 sbi->total_valid_node_count++;
1981 sbi->total_valid_block_count++;
1982 spin_unlock(&sbi->stat_lock);
1986 f2fs_mark_inode_dirty_sync(inode, true);
1988 f2fs_i_blocks_write(inode, 1, true, true);
1991 percpu_counter_inc(&sbi->alloc_valid_block_count);
1997 dquot_free_inode(inode);
1999 dquot_release_reservation_block(inode, 1);
2004 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2005 struct inode *inode, bool is_inode)
2007 spin_lock(&sbi->stat_lock);
2009 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2010 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2011 f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
2013 sbi->total_valid_node_count--;
2014 sbi->total_valid_block_count--;
2015 if (sbi->reserved_blocks &&
2016 sbi->current_reserved_blocks < sbi->reserved_blocks)
2017 sbi->current_reserved_blocks++;
2019 spin_unlock(&sbi->stat_lock);
2022 dquot_free_inode(inode);
2024 f2fs_i_blocks_write(inode, 1, false, true);
2027 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2029 return sbi->total_valid_node_count;
2032 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2034 percpu_counter_inc(&sbi->total_valid_inode_count);
2037 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2039 percpu_counter_dec(&sbi->total_valid_inode_count);
2042 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2044 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2047 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2048 pgoff_t index, bool for_write)
2052 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2054 page = find_get_page_flags(mapping, index,
2055 FGP_LOCK | FGP_ACCESSED);
2057 page = find_lock_page(mapping, index);
2061 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2062 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2068 return grab_cache_page(mapping, index);
2069 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2072 static inline struct page *f2fs_pagecache_get_page(
2073 struct address_space *mapping, pgoff_t index,
2074 int fgp_flags, gfp_t gfp_mask)
2076 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2077 f2fs_show_injection_info(FAULT_PAGE_GET);
2081 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2084 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2086 char *src_kaddr = kmap(src);
2087 char *dst_kaddr = kmap(dst);
2089 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2094 static inline void f2fs_put_page(struct page *page, int unlock)
2100 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2106 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2109 f2fs_put_page(dn->node_page, 1);
2110 if (dn->inode_page && dn->node_page != dn->inode_page)
2111 f2fs_put_page(dn->inode_page, 0);
2112 dn->node_page = NULL;
2113 dn->inode_page = NULL;
2116 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2119 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2122 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2127 entry = kmem_cache_alloc(cachep, flags);
2129 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2133 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2134 int npages, bool no_fail)
2139 /* No failure on bio allocation */
2140 bio = bio_alloc(GFP_NOIO, npages);
2142 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2145 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2146 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2150 return bio_alloc(GFP_KERNEL, npages);
2153 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2155 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2156 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2157 get_pages(sbi, F2FS_WB_CP_DATA) ||
2158 get_pages(sbi, F2FS_DIO_READ) ||
2159 get_pages(sbi, F2FS_DIO_WRITE) ||
2160 atomic_read(&SM_I(sbi)->dcc_info->queued_discard) ||
2161 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2163 return f2fs_time_over(sbi, type);
2166 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2167 unsigned long index, void *item)
2169 while (radix_tree_insert(root, index, item))
2173 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2175 static inline bool IS_INODE(struct page *page)
2177 struct f2fs_node *p = F2FS_NODE(page);
2179 return RAW_IS_INODE(p);
2182 static inline int offset_in_addr(struct f2fs_inode *i)
2184 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2185 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2188 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2190 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2193 static inline int f2fs_has_extra_attr(struct inode *inode);
2194 static inline block_t datablock_addr(struct inode *inode,
2195 struct page *node_page, unsigned int offset)
2197 struct f2fs_node *raw_node;
2200 bool is_inode = IS_INODE(node_page);
2202 raw_node = F2FS_NODE(node_page);
2204 /* from GC path only */
2207 base = offset_in_addr(&raw_node->i);
2208 else if (f2fs_has_extra_attr(inode))
2209 base = get_extra_isize(inode);
2212 addr_array = blkaddr_in_node(raw_node);
2213 return le32_to_cpu(addr_array[base + offset]);
2216 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2221 mask = 1 << (7 - (nr & 0x07));
2222 return mask & *addr;
2225 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2230 mask = 1 << (7 - (nr & 0x07));
2234 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2239 mask = 1 << (7 - (nr & 0x07));
2243 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2249 mask = 1 << (7 - (nr & 0x07));
2255 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2261 mask = 1 << (7 - (nr & 0x07));
2267 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2272 mask = 1 << (7 - (nr & 0x07));
2279 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */
2280 #define F2FS_UNRM_FL 0x00000002 /* Undelete */
2281 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2282 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2283 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2284 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2285 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2286 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2287 /* Reserved for compression usage... */
2288 #define F2FS_DIRTY_FL 0x00000100
2289 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
2290 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */
2291 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */
2292 /* End compression flags --- maybe not all used */
2293 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2294 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */
2295 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
2296 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
2297 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2298 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
2299 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
2300 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */
2301 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */
2302 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
2303 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
2304 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2305 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
2307 #define F2FS_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
2308 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
2310 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */
2311 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \
2312 F2FS_IMMUTABLE_FL | \
2316 F2FS_PROJINHERIT_FL)
2318 /* Flags that should be inherited by new inodes from their parent. */
2319 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\
2320 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\
2321 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\
2322 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\
2323 F2FS_PROJINHERIT_FL)
2325 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2326 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL))
2328 /* Flags that are appropriate for non-directories/regular files. */
2329 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2331 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2335 else if (S_ISREG(mode))
2336 return flags & F2FS_REG_FLMASK;
2338 return flags & F2FS_OTHER_FLMASK;
2341 /* used for f2fs_inode_info->flags */
2343 FI_NEW_INODE, /* indicate newly allocated inode */
2344 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2345 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2346 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2347 FI_INC_LINK, /* need to increment i_nlink */
2348 FI_ACL_MODE, /* indicate acl mode */
2349 FI_NO_ALLOC, /* should not allocate any blocks */
2350 FI_FREE_NID, /* free allocated nide */
2351 FI_NO_EXTENT, /* not to use the extent cache */
2352 FI_INLINE_XATTR, /* used for inline xattr */
2353 FI_INLINE_DATA, /* used for inline data*/
2354 FI_INLINE_DENTRY, /* used for inline dentry */
2355 FI_APPEND_WRITE, /* inode has appended data */
2356 FI_UPDATE_WRITE, /* inode has in-place-update data */
2357 FI_NEED_IPU, /* used for ipu per file */
2358 FI_ATOMIC_FILE, /* indicate atomic file */
2359 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2360 FI_VOLATILE_FILE, /* indicate volatile file */
2361 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2362 FI_DROP_CACHE, /* drop dirty page cache */
2363 FI_DATA_EXIST, /* indicate data exists */
2364 FI_INLINE_DOTS, /* indicate inline dot dentries */
2365 FI_DO_DEFRAG, /* indicate defragment is running */
2366 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2367 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2368 FI_HOT_DATA, /* indicate file is hot */
2369 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2370 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2371 FI_PIN_FILE, /* indicate file should not be gced */
2372 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2375 static inline void __mark_inode_dirty_flag(struct inode *inode,
2379 case FI_INLINE_XATTR:
2380 case FI_INLINE_DATA:
2381 case FI_INLINE_DENTRY:
2387 case FI_INLINE_DOTS:
2389 f2fs_mark_inode_dirty_sync(inode, true);
2393 static inline void set_inode_flag(struct inode *inode, int flag)
2395 if (!test_bit(flag, &F2FS_I(inode)->flags))
2396 set_bit(flag, &F2FS_I(inode)->flags);
2397 __mark_inode_dirty_flag(inode, flag, true);
2400 static inline int is_inode_flag_set(struct inode *inode, int flag)
2402 return test_bit(flag, &F2FS_I(inode)->flags);
2405 static inline void clear_inode_flag(struct inode *inode, int flag)
2407 if (test_bit(flag, &F2FS_I(inode)->flags))
2408 clear_bit(flag, &F2FS_I(inode)->flags);
2409 __mark_inode_dirty_flag(inode, flag, false);
2412 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2414 F2FS_I(inode)->i_acl_mode = mode;
2415 set_inode_flag(inode, FI_ACL_MODE);
2416 f2fs_mark_inode_dirty_sync(inode, false);
2419 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2425 f2fs_mark_inode_dirty_sync(inode, true);
2428 static inline void f2fs_i_blocks_write(struct inode *inode,
2429 block_t diff, bool add, bool claim)
2431 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2432 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2434 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2437 dquot_claim_block(inode, diff);
2439 dquot_alloc_block_nofail(inode, diff);
2441 dquot_free_block(inode, diff);
2444 f2fs_mark_inode_dirty_sync(inode, true);
2445 if (clean || recover)
2446 set_inode_flag(inode, FI_AUTO_RECOVER);
2449 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2451 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2452 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2454 if (i_size_read(inode) == i_size)
2457 i_size_write(inode, i_size);
2458 f2fs_mark_inode_dirty_sync(inode, true);
2459 if (clean || recover)
2460 set_inode_flag(inode, FI_AUTO_RECOVER);
2463 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2465 F2FS_I(inode)->i_current_depth = depth;
2466 f2fs_mark_inode_dirty_sync(inode, true);
2469 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2472 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2473 f2fs_mark_inode_dirty_sync(inode, true);
2476 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2478 F2FS_I(inode)->i_xattr_nid = xnid;
2479 f2fs_mark_inode_dirty_sync(inode, true);
2482 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2484 F2FS_I(inode)->i_pino = pino;
2485 f2fs_mark_inode_dirty_sync(inode, true);
2488 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2490 struct f2fs_inode_info *fi = F2FS_I(inode);
2492 if (ri->i_inline & F2FS_INLINE_XATTR)
2493 set_bit(FI_INLINE_XATTR, &fi->flags);
2494 if (ri->i_inline & F2FS_INLINE_DATA)
2495 set_bit(FI_INLINE_DATA, &fi->flags);
2496 if (ri->i_inline & F2FS_INLINE_DENTRY)
2497 set_bit(FI_INLINE_DENTRY, &fi->flags);
2498 if (ri->i_inline & F2FS_DATA_EXIST)
2499 set_bit(FI_DATA_EXIST, &fi->flags);
2500 if (ri->i_inline & F2FS_INLINE_DOTS)
2501 set_bit(FI_INLINE_DOTS, &fi->flags);
2502 if (ri->i_inline & F2FS_EXTRA_ATTR)
2503 set_bit(FI_EXTRA_ATTR, &fi->flags);
2504 if (ri->i_inline & F2FS_PIN_FILE)
2505 set_bit(FI_PIN_FILE, &fi->flags);
2508 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2512 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2513 ri->i_inline |= F2FS_INLINE_XATTR;
2514 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2515 ri->i_inline |= F2FS_INLINE_DATA;
2516 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2517 ri->i_inline |= F2FS_INLINE_DENTRY;
2518 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2519 ri->i_inline |= F2FS_DATA_EXIST;
2520 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2521 ri->i_inline |= F2FS_INLINE_DOTS;
2522 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2523 ri->i_inline |= F2FS_EXTRA_ATTR;
2524 if (is_inode_flag_set(inode, FI_PIN_FILE))
2525 ri->i_inline |= F2FS_PIN_FILE;
2528 static inline int f2fs_has_extra_attr(struct inode *inode)
2530 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2533 static inline int f2fs_has_inline_xattr(struct inode *inode)
2535 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2538 static inline unsigned int addrs_per_inode(struct inode *inode)
2540 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
2543 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2545 struct f2fs_inode *ri = F2FS_INODE(page);
2547 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2548 get_inline_xattr_addrs(inode)]);
2551 static inline int inline_xattr_size(struct inode *inode)
2553 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2556 static inline int f2fs_has_inline_data(struct inode *inode)
2558 return is_inode_flag_set(inode, FI_INLINE_DATA);
2561 static inline int f2fs_exist_data(struct inode *inode)
2563 return is_inode_flag_set(inode, FI_DATA_EXIST);
2566 static inline int f2fs_has_inline_dots(struct inode *inode)
2568 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2571 static inline bool f2fs_is_pinned_file(struct inode *inode)
2573 return is_inode_flag_set(inode, FI_PIN_FILE);
2576 static inline bool f2fs_is_atomic_file(struct inode *inode)
2578 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2581 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2583 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2586 static inline bool f2fs_is_volatile_file(struct inode *inode)
2588 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2591 static inline bool f2fs_is_first_block_written(struct inode *inode)
2593 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2596 static inline bool f2fs_is_drop_cache(struct inode *inode)
2598 return is_inode_flag_set(inode, FI_DROP_CACHE);
2601 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2603 struct f2fs_inode *ri = F2FS_INODE(page);
2604 int extra_size = get_extra_isize(inode);
2606 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2609 static inline int f2fs_has_inline_dentry(struct inode *inode)
2611 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2614 static inline int is_file(struct inode *inode, int type)
2616 return F2FS_I(inode)->i_advise & type;
2619 static inline void set_file(struct inode *inode, int type)
2621 F2FS_I(inode)->i_advise |= type;
2622 f2fs_mark_inode_dirty_sync(inode, true);
2625 static inline void clear_file(struct inode *inode, int type)
2627 F2FS_I(inode)->i_advise &= ~type;
2628 f2fs_mark_inode_dirty_sync(inode, true);
2631 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2636 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2638 spin_lock(&sbi->inode_lock[DIRTY_META]);
2639 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2640 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2643 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2644 file_keep_isize(inode) ||
2645 i_size_read(inode) & ~PAGE_MASK)
2648 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2650 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2652 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2654 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2655 &F2FS_I(inode)->i_crtime))
2658 down_read(&F2FS_I(inode)->i_sem);
2659 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2660 up_read(&F2FS_I(inode)->i_sem);
2665 static inline bool f2fs_readonly(struct super_block *sb)
2667 return sb_rdonly(sb);
2670 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2672 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2675 static inline bool is_dot_dotdot(const struct qstr *str)
2677 if (str->len == 1 && str->name[0] == '.')
2680 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2686 static inline bool f2fs_may_extent_tree(struct inode *inode)
2688 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2690 if (!test_opt(sbi, EXTENT_CACHE) ||
2691 is_inode_flag_set(inode, FI_NO_EXTENT))
2695 * for recovered files during mount do not create extents
2696 * if shrinker is not registered.
2698 if (list_empty(&sbi->s_list))
2701 return S_ISREG(inode->i_mode);
2704 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2705 size_t size, gfp_t flags)
2709 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2710 f2fs_show_injection_info(FAULT_KMALLOC);
2714 ret = kmalloc(size, flags);
2718 return kvmalloc(size, flags);
2721 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2722 size_t size, gfp_t flags)
2724 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2727 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2728 size_t size, gfp_t flags)
2730 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2731 f2fs_show_injection_info(FAULT_KVMALLOC);
2735 return kvmalloc(size, flags);
2738 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2739 size_t size, gfp_t flags)
2741 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2744 static inline int get_extra_isize(struct inode *inode)
2746 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2749 static inline int get_inline_xattr_addrs(struct inode *inode)
2751 return F2FS_I(inode)->i_inline_xattr_size;
2754 #define f2fs_get_inode_mode(i) \
2755 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2756 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2758 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2759 (offsetof(struct f2fs_inode, i_extra_end) - \
2760 offsetof(struct f2fs_inode, i_extra_isize)) \
2762 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2763 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2764 ((offsetof(typeof(*f2fs_inode), field) + \
2765 sizeof((f2fs_inode)->field)) \
2766 <= (F2FS_OLD_ATTRIBUTE_SIZE + extra_isize)) \
2768 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2772 spin_lock(&sbi->iostat_lock);
2773 for (i = 0; i < NR_IO_TYPE; i++)
2774 sbi->write_iostat[i] = 0;
2775 spin_unlock(&sbi->iostat_lock);
2778 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2779 enum iostat_type type, unsigned long long io_bytes)
2781 if (!sbi->iostat_enable)
2783 spin_lock(&sbi->iostat_lock);
2784 sbi->write_iostat[type] += io_bytes;
2786 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2787 sbi->write_iostat[APP_BUFFERED_IO] =
2788 sbi->write_iostat[APP_WRITE_IO] -
2789 sbi->write_iostat[APP_DIRECT_IO];
2790 spin_unlock(&sbi->iostat_lock);
2793 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2795 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO(fio->type) == META && \
2796 (!is_read_io(fio->op) || fio->is_meta))
2798 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2799 block_t blkaddr, int type);
2800 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2801 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2802 block_t blkaddr, int type)
2804 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2805 f2fs_msg(sbi->sb, KERN_ERR,
2806 "invalid blkaddr: %u, type: %d, run fsck to fix.",
2808 f2fs_bug_on(sbi, 1);
2812 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2814 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2819 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2822 if (!__is_valid_data_blkaddr(blkaddr))
2824 verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2831 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2832 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2833 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock,
2835 int f2fs_truncate(struct inode *inode);
2836 int f2fs_getattr(const struct path *path, struct kstat *stat,
2837 u32 request_mask, unsigned int flags);
2838 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2839 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2840 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2841 int f2fs_precache_extents(struct inode *inode);
2842 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2843 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2844 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
2845 int f2fs_pin_file_control(struct inode *inode, bool inc);
2850 void f2fs_set_inode_flags(struct inode *inode);
2851 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2852 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2853 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2854 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2855 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2856 void f2fs_update_inode(struct inode *inode, struct page *node_page);
2857 void f2fs_update_inode_page(struct inode *inode);
2858 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2859 void f2fs_evict_inode(struct inode *inode);
2860 void f2fs_handle_failed_inode(struct inode *inode);
2865 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2866 bool hot, bool set);
2867 struct dentry *f2fs_get_parent(struct dentry *child);
2872 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2873 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2874 f2fs_hash_t namehash, int *max_slots,
2875 struct f2fs_dentry_ptr *d);
2876 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2877 unsigned int start_pos, struct fscrypt_str *fstr);
2878 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2879 struct f2fs_dentry_ptr *d);
2880 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2881 const struct qstr *new_name,
2882 const struct qstr *orig_name, struct page *dpage);
2883 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2884 unsigned int current_depth);
2885 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2886 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2887 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2888 struct fscrypt_name *fname, struct page **res_page);
2889 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2890 const struct qstr *child, struct page **res_page);
2891 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2892 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2893 struct page **page);
2894 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2895 struct page *page, struct inode *inode);
2896 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2897 const struct qstr *name, f2fs_hash_t name_hash,
2898 unsigned int bit_pos);
2899 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2900 const struct qstr *orig_name,
2901 struct inode *inode, nid_t ino, umode_t mode);
2902 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2903 struct inode *inode, nid_t ino, umode_t mode);
2904 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2905 struct inode *inode, nid_t ino, umode_t mode);
2906 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2907 struct inode *dir, struct inode *inode);
2908 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2909 bool f2fs_empty_dir(struct inode *dir);
2911 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2913 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2914 inode, inode->i_ino, inode->i_mode);
2920 int f2fs_inode_dirtied(struct inode *inode, bool sync);
2921 void f2fs_inode_synced(struct inode *inode);
2922 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
2923 int f2fs_quota_sync(struct super_block *sb, int type);
2924 void f2fs_quota_off_umount(struct super_block *sb);
2925 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
2926 int f2fs_sync_fs(struct super_block *sb, int sync);
2927 extern __printf(3, 4)
2928 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2929 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
2934 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info,
2935 struct fscrypt_name *fname);
2940 struct dnode_of_data;
2943 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
2944 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
2945 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
2946 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
2947 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
2948 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
2949 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
2950 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
2951 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
2952 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
2953 struct node_info *ni);
2954 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
2955 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
2956 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
2957 int f2fs_truncate_xattr_node(struct inode *inode);
2958 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
2959 unsigned int seq_id);
2960 int f2fs_remove_inode_page(struct inode *inode);
2961 struct page *f2fs_new_inode_page(struct inode *inode);
2962 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
2963 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
2964 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
2965 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
2966 int f2fs_move_node_page(struct page *node_page, int gc_type);
2967 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
2968 struct writeback_control *wbc, bool atomic,
2969 unsigned int *seq_id);
2970 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
2971 struct writeback_control *wbc,
2972 bool do_balance, enum iostat_type io_type);
2973 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
2974 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
2975 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
2976 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
2977 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
2978 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
2979 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
2980 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
2981 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
2982 unsigned int segno, struct f2fs_summary_block *sum);
2983 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2984 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
2985 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
2986 int __init f2fs_create_node_manager_caches(void);
2987 void f2fs_destroy_node_manager_caches(void);
2992 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
2993 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
2994 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
2995 void f2fs_drop_inmem_pages(struct inode *inode);
2996 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
2997 int f2fs_commit_inmem_pages(struct inode *inode);
2998 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
2999 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
3000 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3001 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3002 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3003 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3004 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3005 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3006 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3007 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3008 bool f2fs_wait_discard_bios(struct f2fs_sb_info *sbi);
3009 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3010 struct cp_control *cpc);
3011 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3012 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi);
3013 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3014 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3015 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3016 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3017 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3018 struct cp_control *cpc);
3019 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3020 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3022 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3023 enum iostat_type io_type);
3024 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3025 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3026 struct f2fs_io_info *fio);
3027 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3028 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3029 block_t old_blkaddr, block_t new_blkaddr,
3030 bool recover_curseg, bool recover_newaddr);
3031 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3032 block_t old_addr, block_t new_addr,
3033 unsigned char version, bool recover_curseg,
3034 bool recover_newaddr);
3035 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3036 block_t old_blkaddr, block_t *new_blkaddr,
3037 struct f2fs_summary *sum, int type,
3038 struct f2fs_io_info *fio, bool add_list);
3039 void f2fs_wait_on_page_writeback(struct page *page,
3040 enum page_type type, bool ordered, bool locked);
3041 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3042 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3044 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3045 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3046 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3047 unsigned int val, int alloc);
3048 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3049 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3050 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3051 int __init f2fs_create_segment_manager_caches(void);
3052 void f2fs_destroy_segment_manager_caches(void);
3053 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3054 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3055 enum page_type type, enum temp_type temp);
3060 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3061 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3062 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3063 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3064 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3065 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3066 block_t blkaddr, int type);
3067 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3068 int type, bool sync);
3069 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3070 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3071 long nr_to_write, enum iostat_type io_type);
3072 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3073 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3074 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3075 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3076 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3077 unsigned int devidx, int type);
3078 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3079 unsigned int devidx, int type);
3080 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3081 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3082 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3083 void f2fs_add_orphan_inode(struct inode *inode);
3084 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3085 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3086 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3087 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3088 void f2fs_remove_dirty_inode(struct inode *inode);
3089 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3090 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3091 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3092 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3093 int __init f2fs_create_checkpoint_caches(void);
3094 void f2fs_destroy_checkpoint_caches(void);
3099 int f2fs_init_post_read_processing(void);
3100 void f2fs_destroy_post_read_processing(void);
3101 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3102 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3103 struct inode *inode, struct page *page,
3104 nid_t ino, enum page_type type);
3105 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3106 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3107 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3108 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3109 block_t blk_addr, struct bio *bio);
3110 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3111 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3112 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3113 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3114 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3115 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3116 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3117 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3118 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3119 int op_flags, bool for_write);
3120 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3121 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3123 struct page *f2fs_get_new_data_page(struct inode *inode,
3124 struct page *ipage, pgoff_t index, bool new_i_size);
3125 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3126 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3127 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3128 int create, int flag);
3129 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3130 u64 start, u64 len);
3131 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3132 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3133 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3134 unsigned int length);
3135 int f2fs_release_page(struct page *page, gfp_t wait);
3136 #ifdef CONFIG_MIGRATION
3137 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3138 struct page *page, enum migrate_mode mode);
3140 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3141 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3146 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3147 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3148 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3149 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3150 unsigned int segno);
3151 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3156 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3157 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3162 #ifdef CONFIG_F2FS_STAT_FS
3163 struct f2fs_stat_info {
3164 struct list_head stat_list;
3165 struct f2fs_sb_info *sbi;
3166 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3167 int main_area_segs, main_area_sections, main_area_zones;
3168 unsigned long long hit_largest, hit_cached, hit_rbtree;
3169 unsigned long long hit_total, total_ext;
3170 int ext_tree, zombie_tree, ext_node;
3171 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3172 int ndirty_data, ndirty_qdata;
3174 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3175 int nats, dirty_nats, sits, dirty_sits;
3176 int free_nids, avail_nids, alloc_nids;
3177 int total_count, utilization;
3178 int bg_gc, nr_wb_cp_data, nr_wb_data;
3179 int nr_rd_data, nr_rd_node, nr_rd_meta;
3180 int nr_dio_read, nr_dio_write;
3181 unsigned int io_skip_bggc, other_skip_bggc;
3182 int nr_flushing, nr_flushed, flush_list_empty;
3183 int nr_discarding, nr_discarded;
3185 unsigned int undiscard_blks;
3186 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3187 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3188 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3189 unsigned int bimodal, avg_vblocks;
3190 int util_free, util_valid, util_invalid;
3191 int rsvd_segs, overp_segs;
3192 int dirty_count, node_pages, meta_pages;
3193 int prefree_count, call_count, cp_count, bg_cp_count;
3194 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3195 int bg_node_segs, bg_data_segs;
3196 int tot_blks, data_blks, node_blks;
3197 int bg_data_blks, bg_node_blks;
3198 unsigned long long skipped_atomic_files[2];
3199 int curseg[NR_CURSEG_TYPE];
3200 int cursec[NR_CURSEG_TYPE];
3201 int curzone[NR_CURSEG_TYPE];
3203 unsigned int meta_count[META_MAX];
3204 unsigned int segment_count[2];
3205 unsigned int block_count[2];
3206 unsigned int inplace_count;
3207 unsigned long long base_mem, cache_mem, page_mem;
3210 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3212 return (struct f2fs_stat_info *)sbi->stat_info;
3215 #define stat_inc_cp_count(si) ((si)->cp_count++)
3216 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3217 #define stat_inc_call_count(si) ((si)->call_count++)
3218 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3219 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3220 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3221 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3222 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3223 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3224 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3225 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3226 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3227 #define stat_inc_inline_xattr(inode) \
3229 if (f2fs_has_inline_xattr(inode)) \
3230 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3232 #define stat_dec_inline_xattr(inode) \
3234 if (f2fs_has_inline_xattr(inode)) \
3235 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3237 #define stat_inc_inline_inode(inode) \
3239 if (f2fs_has_inline_data(inode)) \
3240 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3242 #define stat_dec_inline_inode(inode) \
3244 if (f2fs_has_inline_data(inode)) \
3245 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3247 #define stat_inc_inline_dir(inode) \
3249 if (f2fs_has_inline_dentry(inode)) \
3250 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3252 #define stat_dec_inline_dir(inode) \
3254 if (f2fs_has_inline_dentry(inode)) \
3255 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3257 #define stat_inc_meta_count(sbi, blkaddr) \
3259 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3260 atomic_inc(&(sbi)->meta_count[META_CP]); \
3261 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3262 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3263 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3264 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3265 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3266 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3268 #define stat_inc_seg_type(sbi, curseg) \
3269 ((sbi)->segment_count[(curseg)->alloc_type]++)
3270 #define stat_inc_block_count(sbi, curseg) \
3271 ((sbi)->block_count[(curseg)->alloc_type]++)
3272 #define stat_inc_inplace_blocks(sbi) \
3273 (atomic_inc(&(sbi)->inplace_count))
3274 #define stat_inc_atomic_write(inode) \
3275 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3276 #define stat_dec_atomic_write(inode) \
3277 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3278 #define stat_update_max_atomic_write(inode) \
3280 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3281 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3283 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3285 #define stat_inc_volatile_write(inode) \
3286 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3287 #define stat_dec_volatile_write(inode) \
3288 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3289 #define stat_update_max_volatile_write(inode) \
3291 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3292 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3294 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3296 #define stat_inc_seg_count(sbi, type, gc_type) \
3298 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3300 if ((type) == SUM_TYPE_DATA) { \
3302 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3305 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3309 #define stat_inc_tot_blk_count(si, blks) \
3310 ((si)->tot_blks += (blks))
3312 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3314 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3315 stat_inc_tot_blk_count(si, blks); \
3316 si->data_blks += (blks); \
3317 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3320 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3322 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3323 stat_inc_tot_blk_count(si, blks); \
3324 si->node_blks += (blks); \
3325 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3328 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3329 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3330 void __init f2fs_create_root_stats(void);
3331 void f2fs_destroy_root_stats(void);
3333 #define stat_inc_cp_count(si) do { } while (0)
3334 #define stat_inc_bg_cp_count(si) do { } while (0)
3335 #define stat_inc_call_count(si) do { } while (0)
3336 #define stat_inc_bggc_count(si) do { } while (0)
3337 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3338 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3339 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3340 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3341 #define stat_inc_total_hit(sb) do { } while (0)
3342 #define stat_inc_rbtree_node_hit(sb) do { } while (0)
3343 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3344 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3345 #define stat_inc_inline_xattr(inode) do { } while (0)
3346 #define stat_dec_inline_xattr(inode) do { } while (0)
3347 #define stat_inc_inline_inode(inode) do { } while (0)
3348 #define stat_dec_inline_inode(inode) do { } while (0)
3349 #define stat_inc_inline_dir(inode) do { } while (0)
3350 #define stat_dec_inline_dir(inode) do { } while (0)
3351 #define stat_inc_atomic_write(inode) do { } while (0)
3352 #define stat_dec_atomic_write(inode) do { } while (0)
3353 #define stat_update_max_atomic_write(inode) do { } while (0)
3354 #define stat_inc_volatile_write(inode) do { } while (0)
3355 #define stat_dec_volatile_write(inode) do { } while (0)
3356 #define stat_update_max_volatile_write(inode) do { } while (0)
3357 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3358 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3359 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3360 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3361 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3362 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3363 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3364 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3366 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3367 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3368 static inline void __init f2fs_create_root_stats(void) { }
3369 static inline void f2fs_destroy_root_stats(void) { }
3372 extern const struct file_operations f2fs_dir_operations;
3373 extern const struct file_operations f2fs_file_operations;
3374 extern const struct inode_operations f2fs_file_inode_operations;
3375 extern const struct address_space_operations f2fs_dblock_aops;
3376 extern const struct address_space_operations f2fs_node_aops;
3377 extern const struct address_space_operations f2fs_meta_aops;
3378 extern const struct inode_operations f2fs_dir_inode_operations;
3379 extern const struct inode_operations f2fs_symlink_inode_operations;
3380 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3381 extern const struct inode_operations f2fs_special_inode_operations;
3382 extern struct kmem_cache *f2fs_inode_entry_slab;
3387 bool f2fs_may_inline_data(struct inode *inode);
3388 bool f2fs_may_inline_dentry(struct inode *inode);
3389 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3390 void f2fs_truncate_inline_inode(struct inode *inode,
3391 struct page *ipage, u64 from);
3392 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3393 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3394 int f2fs_convert_inline_inode(struct inode *inode);
3395 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3396 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3397 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3398 struct fscrypt_name *fname, struct page **res_page);
3399 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3400 struct page *ipage);
3401 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3402 const struct qstr *orig_name,
3403 struct inode *inode, nid_t ino, umode_t mode);
3404 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3405 struct page *page, struct inode *dir,
3406 struct inode *inode);
3407 bool f2fs_empty_inline_dir(struct inode *dir);
3408 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3409 struct fscrypt_str *fstr);
3410 int f2fs_inline_data_fiemap(struct inode *inode,
3411 struct fiemap_extent_info *fieinfo,
3412 __u64 start, __u64 len);
3417 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3418 struct shrink_control *sc);
3419 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3420 struct shrink_control *sc);
3421 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3422 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3427 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3428 struct rb_entry *cached_re, unsigned int ofs);
3429 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3430 struct rb_root_cached *root,
3431 struct rb_node **parent,
3432 unsigned int ofs, bool *leftmost);
3433 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3434 struct rb_entry *cached_re, unsigned int ofs,
3435 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3436 struct rb_node ***insert_p, struct rb_node **insert_parent,
3437 bool force, bool *leftmost);
3438 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3439 struct rb_root_cached *root);
3440 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3441 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3442 void f2fs_drop_extent_tree(struct inode *inode);
3443 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3444 void f2fs_destroy_extent_tree(struct inode *inode);
3445 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3446 struct extent_info *ei);
3447 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3448 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3449 pgoff_t fofs, block_t blkaddr, unsigned int len);
3450 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3451 int __init f2fs_create_extent_cache(void);
3452 void f2fs_destroy_extent_cache(void);
3457 int __init f2fs_init_sysfs(void);
3458 void f2fs_exit_sysfs(void);
3459 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3460 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3465 static inline bool f2fs_encrypted_file(struct inode *inode)
3467 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3470 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3472 #ifdef CONFIG_FS_ENCRYPTION
3473 file_set_encrypt(inode);
3474 f2fs_set_inode_flags(inode);
3479 * Returns true if the reads of the inode's data need to undergo some
3480 * postprocessing step, like decryption or authenticity verification.
3482 static inline bool f2fs_post_read_required(struct inode *inode)
3484 return f2fs_encrypted_file(inode);
3487 #define F2FS_FEATURE_FUNCS(name, flagname) \
3488 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3490 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3493 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3494 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3495 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3496 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3497 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3498 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3499 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3500 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3501 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3502 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3504 #ifdef CONFIG_BLK_DEV_ZONED
3505 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3506 struct block_device *bdev, block_t blkaddr)
3508 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3511 for (i = 0; i < sbi->s_ndevs; i++)
3512 if (FDEV(i).bdev == bdev)
3513 return FDEV(i).blkz_type[zno];
3518 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3520 return f2fs_sb_has_blkzoned(sbi);
3523 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3525 return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
3528 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3530 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3531 f2fs_hw_should_discard(sbi);
3534 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3536 clear_opt(sbi, ADAPTIVE);
3537 clear_opt(sbi, LFS);
3540 case F2FS_MOUNT_ADAPTIVE:
3541 set_opt(sbi, ADAPTIVE);
3543 case F2FS_MOUNT_LFS:
3549 static inline bool f2fs_may_encrypt(struct inode *inode)
3551 #ifdef CONFIG_FS_ENCRYPTION
3552 umode_t mode = inode->i_mode;
3554 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3560 static inline int block_unaligned_IO(struct inode *inode,
3561 struct kiocb *iocb, struct iov_iter *iter)
3563 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
3564 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
3565 loff_t offset = iocb->ki_pos;
3566 unsigned long align = offset | iov_iter_alignment(iter);
3568 return align & blocksize_mask;
3571 static inline int allow_outplace_dio(struct inode *inode,
3572 struct kiocb *iocb, struct iov_iter *iter)
3574 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3575 int rw = iov_iter_rw(iter);
3577 return (test_opt(sbi, LFS) && (rw == WRITE) &&
3578 !block_unaligned_IO(inode, iocb, iter));
3581 static inline bool f2fs_force_buffered_io(struct inode *inode,
3582 struct kiocb *iocb, struct iov_iter *iter)
3584 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3585 int rw = iov_iter_rw(iter);
3587 if (f2fs_post_read_required(inode))
3592 * for blkzoned device, fallback direct IO to buffered IO, so
3593 * all IOs can be serialized by log-structured write.
3595 if (f2fs_sb_has_blkzoned(sbi))
3597 if (test_opt(sbi, LFS) && (rw == WRITE) &&
3598 block_unaligned_IO(inode, iocb, iter))
3600 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
3606 #ifdef CONFIG_F2FS_FAULT_INJECTION
3607 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3610 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3615 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
3618 if (f2fs_sb_has_quota_ino(sbi))
3620 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
3621 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
3622 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])