1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
11 #include <linux/uio.h>
12 #include <linux/types.h>
13 #include <linux/page-flags.h>
14 #include <linux/buffer_head.h>
15 #include <linux/slab.h>
16 #include <linux/crc32.h>
17 #include <linux/magic.h>
18 #include <linux/kobject.h>
19 #include <linux/sched.h>
20 #include <linux/cred.h>
21 #include <linux/vmalloc.h>
22 #include <linux/bio.h>
23 #include <linux/blkdev.h>
24 #include <linux/quotaops.h>
25 #include <crypto/hash.h>
27 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_F2FS_FS_ENCRYPTION)
28 #include <linux/fscrypt.h>
30 #ifdef CONFIG_F2FS_CHECK_FS
31 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
33 #define f2fs_bug_on(sbi, condition) \
35 if (unlikely(condition)) { \
37 set_sbi_flag(sbi, SBI_NEED_FSCK); \
61 #ifdef CONFIG_F2FS_FAULT_INJECTION
62 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
64 struct f2fs_fault_info {
66 unsigned int inject_rate;
67 unsigned int inject_type;
70 extern const char *f2fs_fault_name[FAULT_MAX];
71 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
77 #define F2FS_MOUNT_BG_GC 0x00000001
78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
79 #define F2FS_MOUNT_DISCARD 0x00000004
80 #define F2FS_MOUNT_NOHEAP 0x00000008
81 #define F2FS_MOUNT_XATTR_USER 0x00000010
82 #define F2FS_MOUNT_POSIX_ACL 0x00000020
83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
85 #define F2FS_MOUNT_INLINE_DATA 0x00000100
86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
88 #define F2FS_MOUNT_NOBARRIER 0x00000800
89 #define F2FS_MOUNT_FASTBOOT 0x00001000
90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
91 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000
92 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
93 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
94 #define F2FS_MOUNT_ADAPTIVE 0x00020000
95 #define F2FS_MOUNT_LFS 0x00040000
96 #define F2FS_MOUNT_USRQUOTA 0x00080000
97 #define F2FS_MOUNT_GRPQUOTA 0x00100000
98 #define F2FS_MOUNT_PRJQUOTA 0x00200000
99 #define F2FS_MOUNT_QUOTA 0x00400000
100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
104 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
105 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
106 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
107 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
109 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
110 typecheck(unsigned long long, b) && \
111 ((long long)((a) - (b)) > 0))
113 typedef u32 block_t; /*
114 * should not change u32, since it is the on-disk block
115 * address format, __le32.
119 struct f2fs_mount_info {
121 int write_io_size_bits; /* Write IO size bits */
122 block_t root_reserved_blocks; /* root reserved blocks */
123 kuid_t s_resuid; /* reserved blocks for uid */
124 kgid_t s_resgid; /* reserved blocks for gid */
125 int active_logs; /* # of active logs */
126 int inline_xattr_size; /* inline xattr size */
127 #ifdef CONFIG_F2FS_FAULT_INJECTION
128 struct f2fs_fault_info fault_info; /* For fault injection */
131 /* Names of quota files with journalled quota */
132 char *s_qf_names[MAXQUOTAS];
133 int s_jquota_fmt; /* Format of quota to use */
135 /* For which write hints are passed down to block layer */
137 int alloc_mode; /* segment allocation policy */
138 int fsync_mode; /* fsync policy */
139 bool test_dummy_encryption; /* test dummy encryption */
142 #define F2FS_FEATURE_ENCRYPT 0x0001
143 #define F2FS_FEATURE_BLKZONED 0x0002
144 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
145 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
146 #define F2FS_FEATURE_PRJQUOTA 0x0010
147 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
148 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
149 #define F2FS_FEATURE_QUOTA_INO 0x0080
150 #define F2FS_FEATURE_INODE_CRTIME 0x0100
151 #define F2FS_FEATURE_LOST_FOUND 0x0200
152 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */
153 #define F2FS_FEATURE_SB_CHKSUM 0x0800
155 #define __F2FS_HAS_FEATURE(raw_super, mask) \
156 ((raw_super->feature & cpu_to_le32(mask)) != 0)
157 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
158 #define F2FS_SET_FEATURE(sbi, mask) \
159 (sbi->raw_super->feature |= cpu_to_le32(mask))
160 #define F2FS_CLEAR_FEATURE(sbi, mask) \
161 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
164 * Default values for user and/or group using reserved blocks
166 #define F2FS_DEF_RESUID 0
167 #define F2FS_DEF_RESGID 0
170 * For checkpoint manager
177 #define CP_UMOUNT 0x00000001
178 #define CP_FASTBOOT 0x00000002
179 #define CP_SYNC 0x00000004
180 #define CP_RECOVERY 0x00000008
181 #define CP_DISCARD 0x00000010
182 #define CP_TRIMMED 0x00000020
183 #define CP_PAUSE 0x00000040
185 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
186 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
187 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
188 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
189 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
190 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
191 #define DEF_CP_INTERVAL 60 /* 60 secs */
192 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
193 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
194 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
195 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
205 * indicate meta/data type
218 /* for the list of ino */
220 ORPHAN_INO, /* for orphan ino list */
221 APPEND_INO, /* for append ino list */
222 UPDATE_INO, /* for update ino list */
223 TRANS_DIR_INO, /* for trasactions dir ino list */
224 FLUSH_INO, /* for multiple device flushing */
225 MAX_INO_ENTRY, /* max. list */
229 struct list_head list; /* list head */
230 nid_t ino; /* inode number */
231 unsigned int dirty_device; /* dirty device bitmap */
234 /* for the list of inodes to be GCed */
236 struct list_head list; /* list head */
237 struct inode *inode; /* vfs inode pointer */
240 struct fsync_node_entry {
241 struct list_head list; /* list head */
242 struct page *page; /* warm node page pointer */
243 unsigned int seq_id; /* sequence id */
246 /* for the bitmap indicate blocks to be discarded */
247 struct discard_entry {
248 struct list_head list; /* list head */
249 block_t start_blkaddr; /* start blockaddr of current segment */
250 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
253 /* default discard granularity of inner discard thread, unit: block count */
254 #define DEFAULT_DISCARD_GRANULARITY 16
256 /* max discard pend list number */
257 #define MAX_PLIST_NUM 512
258 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
259 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
262 D_PREP, /* initial */
263 D_PARTIAL, /* partially submitted */
264 D_SUBMIT, /* all submitted */
265 D_DONE, /* finished */
268 struct discard_info {
269 block_t lstart; /* logical start address */
270 block_t len; /* length */
271 block_t start; /* actual start address in dev */
275 struct rb_node rb_node; /* rb node located in rb-tree */
278 block_t lstart; /* logical start address */
279 block_t len; /* length */
280 block_t start; /* actual start address in dev */
282 struct discard_info di; /* discard info */
285 struct list_head list; /* command list */
286 struct completion wait; /* compleation */
287 struct block_device *bdev; /* bdev */
288 unsigned short ref; /* reference count */
289 unsigned char state; /* state */
290 unsigned char queued; /* queued discard */
291 int error; /* bio error */
292 spinlock_t lock; /* for state/bio_ref updating */
293 unsigned short bio_ref; /* bio reference count */
304 struct discard_policy {
305 int type; /* type of discard */
306 unsigned int min_interval; /* used for candidates exist */
307 unsigned int mid_interval; /* used for device busy */
308 unsigned int max_interval; /* used for candidates not exist */
309 unsigned int max_requests; /* # of discards issued per round */
310 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
311 bool io_aware; /* issue discard in idle time */
312 bool sync; /* submit discard with REQ_SYNC flag */
313 bool ordered; /* issue discard by lba order */
314 unsigned int granularity; /* discard granularity */
315 int timeout; /* discard timeout for put_super */
318 struct discard_cmd_control {
319 struct task_struct *f2fs_issue_discard; /* discard thread */
320 struct list_head entry_list; /* 4KB discard entry list */
321 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
322 struct list_head wait_list; /* store on-flushing entries */
323 struct list_head fstrim_list; /* in-flight discard from fstrim */
324 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
325 unsigned int discard_wake; /* to wake up discard thread */
326 struct mutex cmd_lock;
327 unsigned int nr_discards; /* # of discards in the list */
328 unsigned int max_discards; /* max. discards to be issued */
329 unsigned int discard_granularity; /* discard granularity */
330 unsigned int undiscard_blks; /* # of undiscard blocks */
331 unsigned int next_pos; /* next discard position */
332 atomic_t issued_discard; /* # of issued discard */
333 atomic_t queued_discard; /* # of queued discard */
334 atomic_t discard_cmd_cnt; /* # of cached cmd count */
335 struct rb_root_cached root; /* root of discard rb-tree */
336 bool rbtree_check; /* config for consistence check */
339 /* for the list of fsync inodes, used only during recovery */
340 struct fsync_inode_entry {
341 struct list_head list; /* list head */
342 struct inode *inode; /* vfs inode pointer */
343 block_t blkaddr; /* block address locating the last fsync */
344 block_t last_dentry; /* block address locating the last dentry */
347 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
348 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
350 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
351 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
352 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
353 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
355 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
356 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
358 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
360 int before = nats_in_cursum(journal);
362 journal->n_nats = cpu_to_le16(before + i);
366 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
368 int before = sits_in_cursum(journal);
370 journal->n_sits = cpu_to_le16(before + i);
374 static inline bool __has_cursum_space(struct f2fs_journal *journal,
377 if (type == NAT_JOURNAL)
378 return size <= MAX_NAT_JENTRIES(journal);
379 return size <= MAX_SIT_JENTRIES(journal);
385 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
386 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
387 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
389 #define F2FS_IOCTL_MAGIC 0xf5
390 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
391 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
392 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
393 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
394 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
395 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
396 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
397 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
398 struct f2fs_defragment)
399 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
400 struct f2fs_move_range)
401 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
402 struct f2fs_flush_device)
403 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
404 struct f2fs_gc_range)
405 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
406 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
407 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
408 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
410 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
411 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
412 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
415 * should be same as XFS_IOC_GOINGDOWN.
416 * Flags for going down operation used by FS_IOC_GOINGDOWN
418 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
419 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
420 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
421 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
422 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
423 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
425 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
427 * ioctl commands in 32 bit emulation
429 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
430 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
431 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
434 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
435 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
437 struct f2fs_gc_range {
443 struct f2fs_defragment {
448 struct f2fs_move_range {
449 u32 dst_fd; /* destination fd */
450 u64 pos_in; /* start position in src_fd */
451 u64 pos_out; /* start position in dst_fd */
452 u64 len; /* size to move */
455 struct f2fs_flush_device {
456 u32 dev_num; /* device number to flush */
457 u32 segments; /* # of segments to flush */
460 /* for inline stuff */
461 #define DEF_INLINE_RESERVED_SIZE 1
462 static inline int get_extra_isize(struct inode *inode);
463 static inline int get_inline_xattr_addrs(struct inode *inode);
464 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
465 (CUR_ADDRS_PER_INODE(inode) - \
466 get_inline_xattr_addrs(inode) - \
467 DEF_INLINE_RESERVED_SIZE))
470 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
471 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
473 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \
474 BITS_PER_BYTE - 1) / BITS_PER_BYTE)
475 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
477 NR_INLINE_DENTRY(inode) + \
478 INLINE_DENTRY_BITMAP_SIZE(inode)))
481 * For INODE and NODE manager
483 /* for directory operations */
484 struct f2fs_dentry_ptr {
487 struct f2fs_dir_entry *dentry;
488 __u8 (*filename)[F2FS_SLOT_LEN];
493 static inline void make_dentry_ptr_block(struct inode *inode,
494 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
497 d->max = NR_DENTRY_IN_BLOCK;
498 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
499 d->bitmap = t->dentry_bitmap;
500 d->dentry = t->dentry;
501 d->filename = t->filename;
504 static inline void make_dentry_ptr_inline(struct inode *inode,
505 struct f2fs_dentry_ptr *d, void *t)
507 int entry_cnt = NR_INLINE_DENTRY(inode);
508 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
509 int reserved_size = INLINE_RESERVED_SIZE(inode);
513 d->nr_bitmap = bitmap_size;
515 d->dentry = t + bitmap_size + reserved_size;
516 d->filename = t + bitmap_size + reserved_size +
517 SIZE_OF_DIR_ENTRY * entry_cnt;
521 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
522 * as its node offset to distinguish from index node blocks.
523 * But some bits are used to mark the node block.
525 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
528 ALLOC_NODE, /* allocate a new node page if needed */
529 LOOKUP_NODE, /* look up a node without readahead */
531 * look up a node with readahead called
536 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
538 /* maximum retry quota flush count */
539 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
541 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
543 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
545 /* for in-memory extent cache entry */
546 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
548 /* number of extent info in extent cache we try to shrink */
549 #define EXTENT_CACHE_SHRINK_NUMBER 128
552 struct rb_node rb_node; /* rb node located in rb-tree */
553 unsigned int ofs; /* start offset of the entry */
554 unsigned int len; /* length of the entry */
558 unsigned int fofs; /* start offset in a file */
559 unsigned int len; /* length of the extent */
560 u32 blk; /* start block address of the extent */
564 struct rb_node rb_node; /* rb node located in rb-tree */
565 struct extent_info ei; /* extent info */
566 struct list_head list; /* node in global extent list of sbi */
567 struct extent_tree *et; /* extent tree pointer */
571 nid_t ino; /* inode number */
572 struct rb_root_cached root; /* root of extent info rb-tree */
573 struct extent_node *cached_en; /* recently accessed extent node */
574 struct extent_info largest; /* largested extent info */
575 struct list_head list; /* to be used by sbi->zombie_list */
576 rwlock_t lock; /* protect extent info rb-tree */
577 atomic_t node_cnt; /* # of extent node in rb-tree*/
578 bool largest_updated; /* largest extent updated */
582 * This structure is taken from ext4_map_blocks.
584 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
586 #define F2FS_MAP_NEW (1 << BH_New)
587 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
588 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
589 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
592 struct f2fs_map_blocks {
596 unsigned int m_flags;
597 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
598 pgoff_t *m_next_extent; /* point to next possible extent */
600 bool m_may_create; /* indicate it is from write path */
603 /* for flag in get_data_block */
605 F2FS_GET_BLOCK_DEFAULT,
606 F2FS_GET_BLOCK_FIEMAP,
609 F2FS_GET_BLOCK_PRE_DIO,
610 F2FS_GET_BLOCK_PRE_AIO,
611 F2FS_GET_BLOCK_PRECACHE,
615 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
617 #define FADVISE_COLD_BIT 0x01
618 #define FADVISE_LOST_PINO_BIT 0x02
619 #define FADVISE_ENCRYPT_BIT 0x04
620 #define FADVISE_ENC_NAME_BIT 0x08
621 #define FADVISE_KEEP_SIZE_BIT 0x10
622 #define FADVISE_HOT_BIT 0x20
623 #define FADVISE_VERITY_BIT 0x40 /* reserved */
625 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
627 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
628 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
629 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
630 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
631 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
632 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
633 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
634 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
635 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
636 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
637 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
638 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
639 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
640 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
641 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
642 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
644 #define DEF_DIR_LEVEL 0
652 struct f2fs_inode_info {
653 struct inode vfs_inode; /* serve a vfs inode */
654 unsigned long i_flags; /* keep an inode flags for ioctl */
655 unsigned char i_advise; /* use to give file attribute hints */
656 unsigned char i_dir_level; /* use for dentry level for large dir */
657 unsigned int i_current_depth; /* only for directory depth */
658 /* for gc failure statistic */
659 unsigned int i_gc_failures[MAX_GC_FAILURE];
660 unsigned int i_pino; /* parent inode number */
661 umode_t i_acl_mode; /* keep file acl mode temporarily */
663 /* Use below internally in f2fs*/
664 unsigned long flags; /* use to pass per-file flags */
665 struct rw_semaphore i_sem; /* protect fi info */
666 atomic_t dirty_pages; /* # of dirty pages */
667 f2fs_hash_t chash; /* hash value of given file name */
668 unsigned int clevel; /* maximum level of given file name */
669 struct task_struct *task; /* lookup and create consistency */
670 struct task_struct *cp_task; /* separate cp/wb IO stats*/
671 nid_t i_xattr_nid; /* node id that contains xattrs */
672 loff_t last_disk_size; /* lastly written file size */
675 struct dquot *i_dquot[MAXQUOTAS];
677 /* quota space reservation, managed internally by quota code */
678 qsize_t i_reserved_quota;
680 struct list_head dirty_list; /* dirty list for dirs and files */
681 struct list_head gdirty_list; /* linked in global dirty list */
682 struct list_head inmem_ilist; /* list for inmem inodes */
683 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
684 struct task_struct *inmem_task; /* store inmemory task */
685 struct mutex inmem_lock; /* lock for inmemory pages */
686 struct extent_tree *extent_tree; /* cached extent_tree entry */
688 /* avoid racing between foreground op and gc */
689 struct rw_semaphore i_gc_rwsem[2];
690 struct rw_semaphore i_mmap_sem;
691 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
693 int i_extra_isize; /* size of extra space located in i_addr */
694 kprojid_t i_projid; /* id for project quota */
695 int i_inline_xattr_size; /* inline xattr size */
696 struct timespec64 i_crtime; /* inode creation time */
697 struct timespec64 i_disk_time[4];/* inode disk times */
700 static inline void get_extent_info(struct extent_info *ext,
701 struct f2fs_extent *i_ext)
703 ext->fofs = le32_to_cpu(i_ext->fofs);
704 ext->blk = le32_to_cpu(i_ext->blk);
705 ext->len = le32_to_cpu(i_ext->len);
708 static inline void set_raw_extent(struct extent_info *ext,
709 struct f2fs_extent *i_ext)
711 i_ext->fofs = cpu_to_le32(ext->fofs);
712 i_ext->blk = cpu_to_le32(ext->blk);
713 i_ext->len = cpu_to_le32(ext->len);
716 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
717 u32 blk, unsigned int len)
724 static inline bool __is_discard_mergeable(struct discard_info *back,
725 struct discard_info *front, unsigned int max_len)
727 return (back->lstart + back->len == front->lstart) &&
728 (back->len + front->len <= max_len);
731 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
732 struct discard_info *back, unsigned int max_len)
734 return __is_discard_mergeable(back, cur, max_len);
737 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
738 struct discard_info *front, unsigned int max_len)
740 return __is_discard_mergeable(cur, front, max_len);
743 static inline bool __is_extent_mergeable(struct extent_info *back,
744 struct extent_info *front)
746 return (back->fofs + back->len == front->fofs &&
747 back->blk + back->len == front->blk);
750 static inline bool __is_back_mergeable(struct extent_info *cur,
751 struct extent_info *back)
753 return __is_extent_mergeable(back, cur);
756 static inline bool __is_front_mergeable(struct extent_info *cur,
757 struct extent_info *front)
759 return __is_extent_mergeable(cur, front);
762 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
763 static inline void __try_update_largest_extent(struct extent_tree *et,
764 struct extent_node *en)
766 if (en->ei.len > et->largest.len) {
767 et->largest = en->ei;
768 et->largest_updated = true;
773 * For free nid management
776 FREE_NID, /* newly added to free nid list */
777 PREALLOC_NID, /* it is preallocated */
781 struct f2fs_nm_info {
782 block_t nat_blkaddr; /* base disk address of NAT */
783 nid_t max_nid; /* maximum possible node ids */
784 nid_t available_nids; /* # of available node ids */
785 nid_t next_scan_nid; /* the next nid to be scanned */
786 unsigned int ram_thresh; /* control the memory footprint */
787 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
788 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
790 /* NAT cache management */
791 struct radix_tree_root nat_root;/* root of the nat entry cache */
792 struct radix_tree_root nat_set_root;/* root of the nat set cache */
793 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
794 struct list_head nat_entries; /* cached nat entry list (clean) */
795 spinlock_t nat_list_lock; /* protect clean nat entry list */
796 unsigned int nat_cnt; /* the # of cached nat entries */
797 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
798 unsigned int nat_blocks; /* # of nat blocks */
800 /* free node ids management */
801 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
802 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
803 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
804 spinlock_t nid_list_lock; /* protect nid lists ops */
805 struct mutex build_lock; /* lock for build free nids */
806 unsigned char **free_nid_bitmap;
807 unsigned char *nat_block_bitmap;
808 unsigned short *free_nid_count; /* free nid count of NAT block */
811 char *nat_bitmap; /* NAT bitmap pointer */
813 unsigned int nat_bits_blocks; /* # of nat bits blocks */
814 unsigned char *nat_bits; /* NAT bits blocks */
815 unsigned char *full_nat_bits; /* full NAT pages */
816 unsigned char *empty_nat_bits; /* empty NAT pages */
817 #ifdef CONFIG_F2FS_CHECK_FS
818 char *nat_bitmap_mir; /* NAT bitmap mirror */
820 int bitmap_size; /* bitmap size */
824 * this structure is used as one of function parameters.
825 * all the information are dedicated to a given direct node block determined
826 * by the data offset in a file.
828 struct dnode_of_data {
829 struct inode *inode; /* vfs inode pointer */
830 struct page *inode_page; /* its inode page, NULL is possible */
831 struct page *node_page; /* cached direct node page */
832 nid_t nid; /* node id of the direct node block */
833 unsigned int ofs_in_node; /* data offset in the node page */
834 bool inode_page_locked; /* inode page is locked or not */
835 bool node_changed; /* is node block changed */
836 char cur_level; /* level of hole node page */
837 char max_level; /* level of current page located */
838 block_t data_blkaddr; /* block address of the node block */
841 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
842 struct page *ipage, struct page *npage, nid_t nid)
844 memset(dn, 0, sizeof(*dn));
846 dn->inode_page = ipage;
847 dn->node_page = npage;
854 * By default, there are 6 active log areas across the whole main area.
855 * When considering hot and cold data separation to reduce cleaning overhead,
856 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
858 * In the current design, you should not change the numbers intentionally.
859 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
860 * logs individually according to the underlying devices. (default: 6)
861 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
862 * data and 8 for node logs.
864 #define NR_CURSEG_DATA_TYPE (3)
865 #define NR_CURSEG_NODE_TYPE (3)
866 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
869 CURSEG_HOT_DATA = 0, /* directory entry blocks */
870 CURSEG_WARM_DATA, /* data blocks */
871 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
872 CURSEG_HOT_NODE, /* direct node blocks of directory files */
873 CURSEG_WARM_NODE, /* direct node blocks of normal files */
874 CURSEG_COLD_NODE, /* indirect node blocks */
879 struct completion wait;
880 struct llist_node llnode;
885 struct flush_cmd_control {
886 struct task_struct *f2fs_issue_flush; /* flush thread */
887 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
888 atomic_t issued_flush; /* # of issued flushes */
889 atomic_t queued_flush; /* # of queued flushes */
890 struct llist_head issue_list; /* list for command issue */
891 struct llist_node *dispatch_list; /* list for command dispatch */
894 struct f2fs_sm_info {
895 struct sit_info *sit_info; /* whole segment information */
896 struct free_segmap_info *free_info; /* free segment information */
897 struct dirty_seglist_info *dirty_info; /* dirty segment information */
898 struct curseg_info *curseg_array; /* active segment information */
900 struct rw_semaphore curseg_lock; /* for preventing curseg change */
902 block_t seg0_blkaddr; /* block address of 0'th segment */
903 block_t main_blkaddr; /* start block address of main area */
904 block_t ssa_blkaddr; /* start block address of SSA area */
906 unsigned int segment_count; /* total # of segments */
907 unsigned int main_segments; /* # of segments in main area */
908 unsigned int reserved_segments; /* # of reserved segments */
909 unsigned int ovp_segments; /* # of overprovision segments */
911 /* a threshold to reclaim prefree segments */
912 unsigned int rec_prefree_segments;
914 /* for batched trimming */
915 unsigned int trim_sections; /* # of sections to trim */
917 struct list_head sit_entry_set; /* sit entry set list */
919 unsigned int ipu_policy; /* in-place-update policy */
920 unsigned int min_ipu_util; /* in-place-update threshold */
921 unsigned int min_fsync_blocks; /* threshold for fsync */
922 unsigned int min_seq_blocks; /* threshold for sequential blocks */
923 unsigned int min_hot_blocks; /* threshold for hot block allocation */
924 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
926 /* for flush command control */
927 struct flush_cmd_control *fcc_info;
929 /* for discard command control */
930 struct discard_cmd_control *dcc_info;
937 * COUNT_TYPE for monitoring
939 * f2fs monitors the number of several block types such as on-writeback,
940 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
942 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
962 * The below are the page types of bios used in submit_bio().
963 * The available types are:
964 * DATA User data pages. It operates as async mode.
965 * NODE Node pages. It operates as async mode.
966 * META FS metadata pages such as SIT, NAT, CP.
967 * NR_PAGE_TYPE The number of page types.
968 * META_FLUSH Make sure the previous pages are written
969 * with waiting the bio's completion
970 * ... Only can be used with META.
972 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
979 INMEM, /* the below types are used by tracepoints only. */
988 HOT = 0, /* must be zero for meta bio */
994 enum need_lock_type {
1000 enum cp_reason_type {
1014 APP_DIRECT_IO, /* app direct IOs */
1015 APP_BUFFERED_IO, /* app buffered IOs */
1016 APP_WRITE_IO, /* app write IOs */
1017 APP_MAPPED_IO, /* app mapped IOs */
1018 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1019 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1020 FS_META_IO, /* meta IOs from kworker/reclaimer */
1021 FS_GC_DATA_IO, /* data IOs from forground gc */
1022 FS_GC_NODE_IO, /* node IOs from forground gc */
1023 FS_CP_DATA_IO, /* data IOs from checkpoint */
1024 FS_CP_NODE_IO, /* node IOs from checkpoint */
1025 FS_CP_META_IO, /* meta IOs from checkpoint */
1026 FS_DISCARD, /* discard */
1030 struct f2fs_io_info {
1031 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1032 nid_t ino; /* inode number */
1033 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1034 enum temp_type temp; /* contains HOT/WARM/COLD */
1035 int op; /* contains REQ_OP_ */
1036 int op_flags; /* req_flag_bits */
1037 block_t new_blkaddr; /* new block address to be written */
1038 block_t old_blkaddr; /* old block address before Cow */
1039 struct page *page; /* page to be written */
1040 struct page *encrypted_page; /* encrypted page */
1041 struct list_head list; /* serialize IOs */
1042 bool submitted; /* indicate IO submission */
1043 int need_lock; /* indicate we need to lock cp_rwsem */
1044 bool in_list; /* indicate fio is in io_list */
1045 bool is_meta; /* indicate borrow meta inode mapping or not */
1046 bool retry; /* need to reallocate block address */
1047 enum iostat_type io_type; /* io type */
1048 struct writeback_control *io_wbc; /* writeback control */
1049 unsigned char version; /* version of the node */
1052 #define is_read_io(rw) ((rw) == READ)
1053 struct f2fs_bio_info {
1054 struct f2fs_sb_info *sbi; /* f2fs superblock */
1055 struct bio *bio; /* bios to merge */
1056 sector_t last_block_in_bio; /* last block number */
1057 struct f2fs_io_info fio; /* store buffered io info. */
1058 struct rw_semaphore io_rwsem; /* blocking op for bio */
1059 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1060 struct list_head io_list; /* track fios */
1063 #define FDEV(i) (sbi->devs[i])
1064 #define RDEV(i) (raw_super->devs[i])
1065 struct f2fs_dev_info {
1066 struct block_device *bdev;
1067 char path[MAX_PATH_LEN];
1068 unsigned int total_segments;
1071 #ifdef CONFIG_BLK_DEV_ZONED
1072 unsigned int nr_blkz; /* Total number of zones */
1073 u8 *blkz_type; /* Array of zones type */
1078 DIR_INODE, /* for dirty dir inode */
1079 FILE_INODE, /* for dirty regular/symlink inode */
1080 DIRTY_META, /* for all dirtied inode metadata */
1081 ATOMIC_FILE, /* for all atomic files */
1085 /* for inner inode cache management */
1086 struct inode_management {
1087 struct radix_tree_root ino_root; /* ino entry array */
1088 spinlock_t ino_lock; /* for ino entry lock */
1089 struct list_head ino_list; /* inode list head */
1090 unsigned long ino_num; /* number of entries */
1093 /* For s_flag in struct f2fs_sb_info */
1095 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1096 SBI_IS_CLOSE, /* specify unmounting */
1097 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1098 SBI_POR_DOING, /* recovery is doing or not */
1099 SBI_NEED_SB_WRITE, /* need to recover superblock */
1100 SBI_NEED_CP, /* need to checkpoint */
1101 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1102 SBI_IS_RECOVERED, /* recovered orphan/data */
1103 SBI_CP_DISABLED, /* CP was disabled last mount */
1104 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1105 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1106 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1107 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1116 UMOUNT_DISCARD_TIMEOUT,
1128 WHINT_MODE_OFF, /* not pass down write hints */
1129 WHINT_MODE_USER, /* try to pass down hints given by users */
1130 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1134 ALLOC_MODE_DEFAULT, /* stay default */
1135 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1139 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1140 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1141 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1144 #ifdef CONFIG_F2FS_FS_ENCRYPTION
1145 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1146 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1148 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1151 struct f2fs_sb_info {
1152 struct super_block *sb; /* pointer to VFS super block */
1153 struct proc_dir_entry *s_proc; /* proc entry */
1154 struct f2fs_super_block *raw_super; /* raw super block pointer */
1155 struct rw_semaphore sb_lock; /* lock for raw super block */
1156 int valid_super_block; /* valid super block no */
1157 unsigned long s_flag; /* flags for sbi */
1158 struct mutex writepages; /* mutex for writepages() */
1160 #ifdef CONFIG_BLK_DEV_ZONED
1161 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1162 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1165 /* for node-related operations */
1166 struct f2fs_nm_info *nm_info; /* node manager */
1167 struct inode *node_inode; /* cache node blocks */
1169 /* for segment-related operations */
1170 struct f2fs_sm_info *sm_info; /* segment manager */
1172 /* for bio operations */
1173 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1174 /* keep migration IO order for LFS mode */
1175 struct rw_semaphore io_order_lock;
1176 mempool_t *write_io_dummy; /* Dummy pages */
1178 /* for checkpoint */
1179 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1180 int cur_cp_pack; /* remain current cp pack */
1181 spinlock_t cp_lock; /* for flag in ckpt */
1182 struct inode *meta_inode; /* cache meta blocks */
1183 struct mutex cp_mutex; /* checkpoint procedure lock */
1184 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1185 struct rw_semaphore node_write; /* locking node writes */
1186 struct rw_semaphore node_change; /* locking node change */
1187 wait_queue_head_t cp_wait;
1188 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1189 long interval_time[MAX_TIME]; /* to store thresholds */
1191 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1193 spinlock_t fsync_node_lock; /* for node entry lock */
1194 struct list_head fsync_node_list; /* node list head */
1195 unsigned int fsync_seg_id; /* sequence id */
1196 unsigned int fsync_node_num; /* number of node entries */
1198 /* for orphan inode, use 0'th array */
1199 unsigned int max_orphans; /* max orphan inodes */
1201 /* for inode management */
1202 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1203 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1205 /* for extent tree cache */
1206 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1207 struct mutex extent_tree_lock; /* locking extent radix tree */
1208 struct list_head extent_list; /* lru list for shrinker */
1209 spinlock_t extent_lock; /* locking extent lru list */
1210 atomic_t total_ext_tree; /* extent tree count */
1211 struct list_head zombie_list; /* extent zombie tree list */
1212 atomic_t total_zombie_tree; /* extent zombie tree count */
1213 atomic_t total_ext_node; /* extent info count */
1215 /* basic filesystem units */
1216 unsigned int log_sectors_per_block; /* log2 sectors per block */
1217 unsigned int log_blocksize; /* log2 block size */
1218 unsigned int blocksize; /* block size */
1219 unsigned int root_ino_num; /* root inode number*/
1220 unsigned int node_ino_num; /* node inode number*/
1221 unsigned int meta_ino_num; /* meta inode number*/
1222 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1223 unsigned int blocks_per_seg; /* blocks per segment */
1224 unsigned int segs_per_sec; /* segments per section */
1225 unsigned int secs_per_zone; /* sections per zone */
1226 unsigned int total_sections; /* total section count */
1227 unsigned int total_node_count; /* total node block count */
1228 unsigned int total_valid_node_count; /* valid node block count */
1229 loff_t max_file_blocks; /* max block index of file */
1230 int dir_level; /* directory level */
1231 int readdir_ra; /* readahead inode in readdir */
1233 block_t user_block_count; /* # of user blocks */
1234 block_t total_valid_block_count; /* # of valid blocks */
1235 block_t discard_blks; /* discard command candidats */
1236 block_t last_valid_block_count; /* for recovery */
1237 block_t reserved_blocks; /* configurable reserved blocks */
1238 block_t current_reserved_blocks; /* current reserved blocks */
1240 /* Additional tracking for no checkpoint mode */
1241 block_t unusable_block_count; /* # of blocks saved by last cp */
1243 unsigned int nquota_files; /* # of quota sysfile */
1245 u32 s_next_generation; /* for NFS support */
1247 /* # of pages, see count_type */
1248 atomic_t nr_pages[NR_COUNT_TYPE];
1249 /* # of allocated blocks */
1250 struct percpu_counter alloc_valid_block_count;
1252 /* writeback control */
1253 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1255 /* valid inode count */
1256 struct percpu_counter total_valid_inode_count;
1258 struct f2fs_mount_info mount_opt; /* mount options */
1260 /* for cleaning operations */
1261 struct mutex gc_mutex; /* mutex for GC */
1262 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1263 unsigned int cur_victim_sec; /* current victim section num */
1264 unsigned int gc_mode; /* current GC state */
1265 unsigned int next_victim_seg[2]; /* next segment in victim section */
1266 /* for skip statistic */
1267 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1268 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1270 /* threshold for gc trials on pinned files */
1271 u64 gc_pin_file_threshold;
1273 /* maximum # of trials to find a victim segment for SSR and GC */
1274 unsigned int max_victim_search;
1275 /* migration granularity of garbage collection, unit: segment */
1276 unsigned int migration_granularity;
1279 * for stat information.
1280 * one is for the LFS mode, and the other is for the SSR mode.
1282 #ifdef CONFIG_F2FS_STAT_FS
1283 struct f2fs_stat_info *stat_info; /* FS status information */
1284 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1285 unsigned int segment_count[2]; /* # of allocated segments */
1286 unsigned int block_count[2]; /* # of allocated blocks */
1287 atomic_t inplace_count; /* # of inplace update */
1288 atomic64_t total_hit_ext; /* # of lookup extent cache */
1289 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1290 atomic64_t read_hit_largest; /* # of hit largest extent node */
1291 atomic64_t read_hit_cached; /* # of hit cached extent node */
1292 atomic_t inline_xattr; /* # of inline_xattr inodes */
1293 atomic_t inline_inode; /* # of inline_data inodes */
1294 atomic_t inline_dir; /* # of inline_dentry inodes */
1295 atomic_t aw_cnt; /* # of atomic writes */
1296 atomic_t vw_cnt; /* # of volatile writes */
1297 atomic_t max_aw_cnt; /* max # of atomic writes */
1298 atomic_t max_vw_cnt; /* max # of volatile writes */
1299 int bg_gc; /* background gc calls */
1300 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1301 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1302 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1304 spinlock_t stat_lock; /* lock for stat operations */
1306 /* For app/fs IO statistics */
1307 spinlock_t iostat_lock;
1308 unsigned long long write_iostat[NR_IO_TYPE];
1311 /* For sysfs suppport */
1312 struct kobject s_kobj;
1313 struct completion s_kobj_unregister;
1315 /* For shrinker support */
1316 struct list_head s_list;
1317 int s_ndevs; /* number of devices */
1318 struct f2fs_dev_info *devs; /* for device list */
1319 unsigned int dirty_device; /* for checkpoint data flush */
1320 spinlock_t dev_lock; /* protect dirty_device */
1321 struct mutex umount_mutex;
1322 unsigned int shrinker_run_no;
1324 /* For write statistics */
1325 u64 sectors_written_start;
1328 /* Reference to checksum algorithm driver via cryptoapi */
1329 struct crypto_shash *s_chksum_driver;
1331 /* Precomputed FS UUID checksum for seeding other checksums */
1332 __u32 s_chksum_seed;
1335 struct f2fs_private_dio {
1336 struct inode *inode;
1338 bio_end_io_t *orig_end_io;
1342 #ifdef CONFIG_F2FS_FAULT_INJECTION
1343 #define f2fs_show_injection_info(type) \
1344 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pF\n", \
1345 KERN_INFO, f2fs_fault_name[type], \
1346 __func__, __builtin_return_address(0))
1347 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1349 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1351 if (!ffi->inject_rate)
1354 if (!IS_FAULT_SET(ffi, type))
1357 atomic_inc(&ffi->inject_ops);
1358 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1359 atomic_set(&ffi->inject_ops, 0);
1365 #define f2fs_show_injection_info(type) do { } while (0)
1366 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1372 /* For write statistics. Suppose sector size is 512 bytes,
1373 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1375 #define BD_PART_WRITTEN(s) \
1376 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1377 (s)->sectors_written_start) >> 1)
1379 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1381 unsigned long now = jiffies;
1383 sbi->last_time[type] = now;
1385 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1386 if (type == REQ_TIME) {
1387 sbi->last_time[DISCARD_TIME] = now;
1388 sbi->last_time[GC_TIME] = now;
1392 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1394 unsigned long interval = sbi->interval_time[type] * HZ;
1396 return time_after(jiffies, sbi->last_time[type] + interval);
1399 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1402 unsigned long interval = sbi->interval_time[type] * HZ;
1403 unsigned int wait_ms = 0;
1406 delta = (sbi->last_time[type] + interval) - jiffies;
1408 wait_ms = jiffies_to_msecs(delta);
1416 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1417 const void *address, unsigned int length)
1420 struct shash_desc shash;
1425 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1427 desc.shash.tfm = sbi->s_chksum_driver;
1428 desc.shash.flags = 0;
1429 *(u32 *)desc.ctx = crc;
1431 err = crypto_shash_update(&desc.shash, address, length);
1434 return *(u32 *)desc.ctx;
1437 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1438 unsigned int length)
1440 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1443 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1444 void *buf, size_t buf_size)
1446 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1449 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1450 const void *address, unsigned int length)
1452 return __f2fs_crc32(sbi, crc, address, length);
1455 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1457 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1460 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1462 return sb->s_fs_info;
1465 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1467 return F2FS_SB(inode->i_sb);
1470 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1472 return F2FS_I_SB(mapping->host);
1475 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1477 return F2FS_M_SB(page->mapping);
1480 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1482 return (struct f2fs_super_block *)(sbi->raw_super);
1485 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1487 return (struct f2fs_checkpoint *)(sbi->ckpt);
1490 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1492 return (struct f2fs_node *)page_address(page);
1495 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1497 return &((struct f2fs_node *)page_address(page))->i;
1500 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1502 return (struct f2fs_nm_info *)(sbi->nm_info);
1505 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1507 return (struct f2fs_sm_info *)(sbi->sm_info);
1510 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1512 return (struct sit_info *)(SM_I(sbi)->sit_info);
1515 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1517 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1520 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1522 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1525 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1527 return sbi->meta_inode->i_mapping;
1530 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1532 return sbi->node_inode->i_mapping;
1535 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1537 return test_bit(type, &sbi->s_flag);
1540 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1542 set_bit(type, &sbi->s_flag);
1545 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1547 clear_bit(type, &sbi->s_flag);
1550 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1552 return le64_to_cpu(cp->checkpoint_ver);
1555 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1557 if (type < F2FS_MAX_QUOTAS)
1558 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1562 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1564 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1565 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1568 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1570 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1572 return ckpt_flags & f;
1575 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1577 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1580 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1582 unsigned int ckpt_flags;
1584 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1586 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1589 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1591 unsigned long flags;
1593 spin_lock_irqsave(&sbi->cp_lock, flags);
1594 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1595 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1598 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1600 unsigned int ckpt_flags;
1602 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1604 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1607 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1609 unsigned long flags;
1611 spin_lock_irqsave(&sbi->cp_lock, flags);
1612 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1613 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1616 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1618 unsigned long flags;
1621 * In order to re-enable nat_bits we need to call fsck.f2fs by
1622 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1623 * so let's rely on regular fsck or unclean shutdown.
1627 spin_lock_irqsave(&sbi->cp_lock, flags);
1628 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1629 kvfree(NM_I(sbi)->nat_bits);
1630 NM_I(sbi)->nat_bits = NULL;
1632 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1635 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1636 struct cp_control *cpc)
1638 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1640 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1643 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1645 down_read(&sbi->cp_rwsem);
1648 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1650 return down_read_trylock(&sbi->cp_rwsem);
1653 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1655 up_read(&sbi->cp_rwsem);
1658 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1660 down_write(&sbi->cp_rwsem);
1663 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1665 up_write(&sbi->cp_rwsem);
1668 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1670 int reason = CP_SYNC;
1672 if (test_opt(sbi, FASTBOOT))
1673 reason = CP_FASTBOOT;
1674 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1679 static inline bool __remain_node_summaries(int reason)
1681 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1684 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1686 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1687 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1691 * Check whether the inode has blocks or not
1693 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1695 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1697 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1700 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1702 return ofs == XATTR_NODE_OFFSET;
1705 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1706 struct inode *inode, bool cap)
1710 if (!test_opt(sbi, RESERVE_ROOT))
1712 if (IS_NOQUOTA(inode))
1714 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1716 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1717 in_group_p(F2FS_OPTION(sbi).s_resgid))
1719 if (cap && capable(CAP_SYS_RESOURCE))
1724 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1725 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1726 struct inode *inode, blkcnt_t *count)
1728 blkcnt_t diff = 0, release = 0;
1729 block_t avail_user_block_count;
1732 ret = dquot_reserve_block(inode, *count);
1736 if (time_to_inject(sbi, FAULT_BLOCK)) {
1737 f2fs_show_injection_info(FAULT_BLOCK);
1743 * let's increase this in prior to actual block count change in order
1744 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1746 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1748 spin_lock(&sbi->stat_lock);
1749 sbi->total_valid_block_count += (block_t)(*count);
1750 avail_user_block_count = sbi->user_block_count -
1751 sbi->current_reserved_blocks;
1753 if (!__allow_reserved_blocks(sbi, inode, true))
1754 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1755 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1756 avail_user_block_count -= sbi->unusable_block_count;
1757 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1758 diff = sbi->total_valid_block_count - avail_user_block_count;
1763 sbi->total_valid_block_count -= diff;
1765 spin_unlock(&sbi->stat_lock);
1769 spin_unlock(&sbi->stat_lock);
1771 if (unlikely(release)) {
1772 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1773 dquot_release_reservation_block(inode, release);
1775 f2fs_i_blocks_write(inode, *count, true, true);
1779 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1780 dquot_release_reservation_block(inode, release);
1784 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1785 struct inode *inode,
1788 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1790 spin_lock(&sbi->stat_lock);
1791 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1792 f2fs_bug_on(sbi, inode->i_blocks < sectors);
1793 sbi->total_valid_block_count -= (block_t)count;
1794 if (sbi->reserved_blocks &&
1795 sbi->current_reserved_blocks < sbi->reserved_blocks)
1796 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1797 sbi->current_reserved_blocks + count);
1798 spin_unlock(&sbi->stat_lock);
1799 f2fs_i_blocks_write(inode, count, false, true);
1802 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1804 atomic_inc(&sbi->nr_pages[count_type]);
1806 if (count_type == F2FS_DIRTY_DENTS ||
1807 count_type == F2FS_DIRTY_NODES ||
1808 count_type == F2FS_DIRTY_META ||
1809 count_type == F2FS_DIRTY_QDATA ||
1810 count_type == F2FS_DIRTY_IMETA)
1811 set_sbi_flag(sbi, SBI_IS_DIRTY);
1814 static inline void inode_inc_dirty_pages(struct inode *inode)
1816 atomic_inc(&F2FS_I(inode)->dirty_pages);
1817 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1818 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1819 if (IS_NOQUOTA(inode))
1820 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1823 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1825 atomic_dec(&sbi->nr_pages[count_type]);
1828 static inline void inode_dec_dirty_pages(struct inode *inode)
1830 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1831 !S_ISLNK(inode->i_mode))
1834 atomic_dec(&F2FS_I(inode)->dirty_pages);
1835 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1836 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1837 if (IS_NOQUOTA(inode))
1838 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1841 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1843 return atomic_read(&sbi->nr_pages[count_type]);
1846 static inline int get_dirty_pages(struct inode *inode)
1848 return atomic_read(&F2FS_I(inode)->dirty_pages);
1851 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1853 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1854 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1855 sbi->log_blocks_per_seg;
1857 return segs / sbi->segs_per_sec;
1860 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1862 return sbi->total_valid_block_count;
1865 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1867 return sbi->discard_blks;
1870 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1872 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1874 /* return NAT or SIT bitmap */
1875 if (flag == NAT_BITMAP)
1876 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1877 else if (flag == SIT_BITMAP)
1878 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1883 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1885 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1888 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1890 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1893 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1894 offset = (flag == SIT_BITMAP) ?
1895 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1896 return &ckpt->sit_nat_version_bitmap + offset;
1899 if (__cp_payload(sbi) > 0) {
1900 if (flag == NAT_BITMAP)
1901 return &ckpt->sit_nat_version_bitmap;
1903 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1905 offset = (flag == NAT_BITMAP) ?
1906 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1907 return &ckpt->sit_nat_version_bitmap + offset;
1911 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1913 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1915 if (sbi->cur_cp_pack == 2)
1916 start_addr += sbi->blocks_per_seg;
1920 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1922 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1924 if (sbi->cur_cp_pack == 1)
1925 start_addr += sbi->blocks_per_seg;
1929 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1931 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1934 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1936 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1939 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
1940 struct inode *inode, bool is_inode)
1942 block_t valid_block_count;
1943 unsigned int valid_node_count;
1948 err = dquot_alloc_inode(inode);
1953 err = dquot_reserve_block(inode, 1);
1958 if (time_to_inject(sbi, FAULT_BLOCK)) {
1959 f2fs_show_injection_info(FAULT_BLOCK);
1963 spin_lock(&sbi->stat_lock);
1965 valid_block_count = sbi->total_valid_block_count +
1966 sbi->current_reserved_blocks + 1;
1968 if (!__allow_reserved_blocks(sbi, inode, false))
1969 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
1970 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1971 valid_block_count += sbi->unusable_block_count;
1973 if (unlikely(valid_block_count > sbi->user_block_count)) {
1974 spin_unlock(&sbi->stat_lock);
1978 valid_node_count = sbi->total_valid_node_count + 1;
1979 if (unlikely(valid_node_count > sbi->total_node_count)) {
1980 spin_unlock(&sbi->stat_lock);
1984 sbi->total_valid_node_count++;
1985 sbi->total_valid_block_count++;
1986 spin_unlock(&sbi->stat_lock);
1990 f2fs_mark_inode_dirty_sync(inode, true);
1992 f2fs_i_blocks_write(inode, 1, true, true);
1995 percpu_counter_inc(&sbi->alloc_valid_block_count);
2001 dquot_free_inode(inode);
2003 dquot_release_reservation_block(inode, 1);
2008 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2009 struct inode *inode, bool is_inode)
2011 spin_lock(&sbi->stat_lock);
2013 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2014 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2015 f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
2017 sbi->total_valid_node_count--;
2018 sbi->total_valid_block_count--;
2019 if (sbi->reserved_blocks &&
2020 sbi->current_reserved_blocks < sbi->reserved_blocks)
2021 sbi->current_reserved_blocks++;
2023 spin_unlock(&sbi->stat_lock);
2026 dquot_free_inode(inode);
2028 f2fs_i_blocks_write(inode, 1, false, true);
2031 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2033 return sbi->total_valid_node_count;
2036 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2038 percpu_counter_inc(&sbi->total_valid_inode_count);
2041 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2043 percpu_counter_dec(&sbi->total_valid_inode_count);
2046 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2048 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2051 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2052 pgoff_t index, bool for_write)
2056 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2058 page = find_get_page_flags(mapping, index,
2059 FGP_LOCK | FGP_ACCESSED);
2061 page = find_lock_page(mapping, index);
2065 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2066 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2072 return grab_cache_page(mapping, index);
2073 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2076 static inline struct page *f2fs_pagecache_get_page(
2077 struct address_space *mapping, pgoff_t index,
2078 int fgp_flags, gfp_t gfp_mask)
2080 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2081 f2fs_show_injection_info(FAULT_PAGE_GET);
2085 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2088 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2090 char *src_kaddr = kmap(src);
2091 char *dst_kaddr = kmap(dst);
2093 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2098 static inline void f2fs_put_page(struct page *page, int unlock)
2104 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2110 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2113 f2fs_put_page(dn->node_page, 1);
2114 if (dn->inode_page && dn->node_page != dn->inode_page)
2115 f2fs_put_page(dn->inode_page, 0);
2116 dn->node_page = NULL;
2117 dn->inode_page = NULL;
2120 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2123 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2126 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2131 entry = kmem_cache_alloc(cachep, flags);
2133 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2137 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2138 int npages, bool no_fail)
2143 /* No failure on bio allocation */
2144 bio = bio_alloc(GFP_NOIO, npages);
2146 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2149 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2150 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2154 return bio_alloc(GFP_KERNEL, npages);
2157 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2159 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2160 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2161 get_pages(sbi, F2FS_WB_CP_DATA) ||
2162 get_pages(sbi, F2FS_DIO_READ) ||
2163 get_pages(sbi, F2FS_DIO_WRITE))
2166 if (SM_I(sbi) && SM_I(sbi)->dcc_info &&
2167 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2170 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2171 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2174 return f2fs_time_over(sbi, type);
2177 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2178 unsigned long index, void *item)
2180 while (radix_tree_insert(root, index, item))
2184 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2186 static inline bool IS_INODE(struct page *page)
2188 struct f2fs_node *p = F2FS_NODE(page);
2190 return RAW_IS_INODE(p);
2193 static inline int offset_in_addr(struct f2fs_inode *i)
2195 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2196 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2199 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2201 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2204 static inline int f2fs_has_extra_attr(struct inode *inode);
2205 static inline block_t datablock_addr(struct inode *inode,
2206 struct page *node_page, unsigned int offset)
2208 struct f2fs_node *raw_node;
2211 bool is_inode = IS_INODE(node_page);
2213 raw_node = F2FS_NODE(node_page);
2215 /* from GC path only */
2218 base = offset_in_addr(&raw_node->i);
2219 else if (f2fs_has_extra_attr(inode))
2220 base = get_extra_isize(inode);
2223 addr_array = blkaddr_in_node(raw_node);
2224 return le32_to_cpu(addr_array[base + offset]);
2227 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2232 mask = 1 << (7 - (nr & 0x07));
2233 return mask & *addr;
2236 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2241 mask = 1 << (7 - (nr & 0x07));
2245 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2250 mask = 1 << (7 - (nr & 0x07));
2254 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2260 mask = 1 << (7 - (nr & 0x07));
2266 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2272 mask = 1 << (7 - (nr & 0x07));
2278 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2283 mask = 1 << (7 - (nr & 0x07));
2290 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */
2291 #define F2FS_UNRM_FL 0x00000002 /* Undelete */
2292 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2293 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2294 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2295 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2296 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2297 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2298 /* Reserved for compression usage... */
2299 #define F2FS_DIRTY_FL 0x00000100
2300 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
2301 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */
2302 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */
2303 /* End compression flags --- maybe not all used */
2304 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2305 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */
2306 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
2307 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
2308 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2309 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
2310 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
2311 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */
2312 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */
2313 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
2314 #define F2FS_NOCOW_FL 0x00800000 /* Do not cow file */
2315 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
2316 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2317 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
2319 #define F2FS_FL_USER_VISIBLE 0x30CBDFFF /* User visible flags */
2320 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
2322 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */
2323 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \
2324 F2FS_IMMUTABLE_FL | \
2328 F2FS_PROJINHERIT_FL)
2330 /* Flags that should be inherited by new inodes from their parent. */
2331 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\
2332 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\
2333 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\
2334 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\
2335 F2FS_PROJINHERIT_FL)
2337 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2338 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL))
2340 /* Flags that are appropriate for non-directories/regular files. */
2341 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2343 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2347 else if (S_ISREG(mode))
2348 return flags & F2FS_REG_FLMASK;
2350 return flags & F2FS_OTHER_FLMASK;
2353 /* used for f2fs_inode_info->flags */
2355 FI_NEW_INODE, /* indicate newly allocated inode */
2356 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2357 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2358 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2359 FI_INC_LINK, /* need to increment i_nlink */
2360 FI_ACL_MODE, /* indicate acl mode */
2361 FI_NO_ALLOC, /* should not allocate any blocks */
2362 FI_FREE_NID, /* free allocated nide */
2363 FI_NO_EXTENT, /* not to use the extent cache */
2364 FI_INLINE_XATTR, /* used for inline xattr */
2365 FI_INLINE_DATA, /* used for inline data*/
2366 FI_INLINE_DENTRY, /* used for inline dentry */
2367 FI_APPEND_WRITE, /* inode has appended data */
2368 FI_UPDATE_WRITE, /* inode has in-place-update data */
2369 FI_NEED_IPU, /* used for ipu per file */
2370 FI_ATOMIC_FILE, /* indicate atomic file */
2371 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2372 FI_VOLATILE_FILE, /* indicate volatile file */
2373 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2374 FI_DROP_CACHE, /* drop dirty page cache */
2375 FI_DATA_EXIST, /* indicate data exists */
2376 FI_INLINE_DOTS, /* indicate inline dot dentries */
2377 FI_DO_DEFRAG, /* indicate defragment is running */
2378 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2379 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2380 FI_HOT_DATA, /* indicate file is hot */
2381 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2382 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2383 FI_PIN_FILE, /* indicate file should not be gced */
2384 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2387 static inline void __mark_inode_dirty_flag(struct inode *inode,
2391 case FI_INLINE_XATTR:
2392 case FI_INLINE_DATA:
2393 case FI_INLINE_DENTRY:
2399 case FI_INLINE_DOTS:
2401 f2fs_mark_inode_dirty_sync(inode, true);
2405 static inline void set_inode_flag(struct inode *inode, int flag)
2407 if (!test_bit(flag, &F2FS_I(inode)->flags))
2408 set_bit(flag, &F2FS_I(inode)->flags);
2409 __mark_inode_dirty_flag(inode, flag, true);
2412 static inline int is_inode_flag_set(struct inode *inode, int flag)
2414 return test_bit(flag, &F2FS_I(inode)->flags);
2417 static inline void clear_inode_flag(struct inode *inode, int flag)
2419 if (test_bit(flag, &F2FS_I(inode)->flags))
2420 clear_bit(flag, &F2FS_I(inode)->flags);
2421 __mark_inode_dirty_flag(inode, flag, false);
2424 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2426 F2FS_I(inode)->i_acl_mode = mode;
2427 set_inode_flag(inode, FI_ACL_MODE);
2428 f2fs_mark_inode_dirty_sync(inode, false);
2431 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2437 f2fs_mark_inode_dirty_sync(inode, true);
2440 static inline void f2fs_i_blocks_write(struct inode *inode,
2441 block_t diff, bool add, bool claim)
2443 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2444 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2446 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2449 dquot_claim_block(inode, diff);
2451 dquot_alloc_block_nofail(inode, diff);
2453 dquot_free_block(inode, diff);
2456 f2fs_mark_inode_dirty_sync(inode, true);
2457 if (clean || recover)
2458 set_inode_flag(inode, FI_AUTO_RECOVER);
2461 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2463 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2464 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2466 if (i_size_read(inode) == i_size)
2469 i_size_write(inode, i_size);
2470 f2fs_mark_inode_dirty_sync(inode, true);
2471 if (clean || recover)
2472 set_inode_flag(inode, FI_AUTO_RECOVER);
2475 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2477 F2FS_I(inode)->i_current_depth = depth;
2478 f2fs_mark_inode_dirty_sync(inode, true);
2481 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2484 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2485 f2fs_mark_inode_dirty_sync(inode, true);
2488 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2490 F2FS_I(inode)->i_xattr_nid = xnid;
2491 f2fs_mark_inode_dirty_sync(inode, true);
2494 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2496 F2FS_I(inode)->i_pino = pino;
2497 f2fs_mark_inode_dirty_sync(inode, true);
2500 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2502 struct f2fs_inode_info *fi = F2FS_I(inode);
2504 if (ri->i_inline & F2FS_INLINE_XATTR)
2505 set_bit(FI_INLINE_XATTR, &fi->flags);
2506 if (ri->i_inline & F2FS_INLINE_DATA)
2507 set_bit(FI_INLINE_DATA, &fi->flags);
2508 if (ri->i_inline & F2FS_INLINE_DENTRY)
2509 set_bit(FI_INLINE_DENTRY, &fi->flags);
2510 if (ri->i_inline & F2FS_DATA_EXIST)
2511 set_bit(FI_DATA_EXIST, &fi->flags);
2512 if (ri->i_inline & F2FS_INLINE_DOTS)
2513 set_bit(FI_INLINE_DOTS, &fi->flags);
2514 if (ri->i_inline & F2FS_EXTRA_ATTR)
2515 set_bit(FI_EXTRA_ATTR, &fi->flags);
2516 if (ri->i_inline & F2FS_PIN_FILE)
2517 set_bit(FI_PIN_FILE, &fi->flags);
2520 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2524 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2525 ri->i_inline |= F2FS_INLINE_XATTR;
2526 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2527 ri->i_inline |= F2FS_INLINE_DATA;
2528 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2529 ri->i_inline |= F2FS_INLINE_DENTRY;
2530 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2531 ri->i_inline |= F2FS_DATA_EXIST;
2532 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2533 ri->i_inline |= F2FS_INLINE_DOTS;
2534 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2535 ri->i_inline |= F2FS_EXTRA_ATTR;
2536 if (is_inode_flag_set(inode, FI_PIN_FILE))
2537 ri->i_inline |= F2FS_PIN_FILE;
2540 static inline int f2fs_has_extra_attr(struct inode *inode)
2542 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2545 static inline int f2fs_has_inline_xattr(struct inode *inode)
2547 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2550 static inline unsigned int addrs_per_inode(struct inode *inode)
2552 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
2555 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2557 struct f2fs_inode *ri = F2FS_INODE(page);
2559 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2560 get_inline_xattr_addrs(inode)]);
2563 static inline int inline_xattr_size(struct inode *inode)
2565 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2568 static inline int f2fs_has_inline_data(struct inode *inode)
2570 return is_inode_flag_set(inode, FI_INLINE_DATA);
2573 static inline int f2fs_exist_data(struct inode *inode)
2575 return is_inode_flag_set(inode, FI_DATA_EXIST);
2578 static inline int f2fs_has_inline_dots(struct inode *inode)
2580 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2583 static inline bool f2fs_is_pinned_file(struct inode *inode)
2585 return is_inode_flag_set(inode, FI_PIN_FILE);
2588 static inline bool f2fs_is_atomic_file(struct inode *inode)
2590 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2593 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2595 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2598 static inline bool f2fs_is_volatile_file(struct inode *inode)
2600 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2603 static inline bool f2fs_is_first_block_written(struct inode *inode)
2605 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2608 static inline bool f2fs_is_drop_cache(struct inode *inode)
2610 return is_inode_flag_set(inode, FI_DROP_CACHE);
2613 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2615 struct f2fs_inode *ri = F2FS_INODE(page);
2616 int extra_size = get_extra_isize(inode);
2618 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2621 static inline int f2fs_has_inline_dentry(struct inode *inode)
2623 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2626 static inline int is_file(struct inode *inode, int type)
2628 return F2FS_I(inode)->i_advise & type;
2631 static inline void set_file(struct inode *inode, int type)
2633 F2FS_I(inode)->i_advise |= type;
2634 f2fs_mark_inode_dirty_sync(inode, true);
2637 static inline void clear_file(struct inode *inode, int type)
2639 F2FS_I(inode)->i_advise &= ~type;
2640 f2fs_mark_inode_dirty_sync(inode, true);
2643 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2648 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2650 spin_lock(&sbi->inode_lock[DIRTY_META]);
2651 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2652 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2655 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2656 file_keep_isize(inode) ||
2657 i_size_read(inode) & ~PAGE_MASK)
2660 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2662 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2664 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2666 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2667 &F2FS_I(inode)->i_crtime))
2670 down_read(&F2FS_I(inode)->i_sem);
2671 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2672 up_read(&F2FS_I(inode)->i_sem);
2677 static inline bool f2fs_readonly(struct super_block *sb)
2679 return sb_rdonly(sb);
2682 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2684 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2687 static inline bool is_dot_dotdot(const struct qstr *str)
2689 if (str->len == 1 && str->name[0] == '.')
2692 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2698 static inline bool f2fs_may_extent_tree(struct inode *inode)
2700 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2702 if (!test_opt(sbi, EXTENT_CACHE) ||
2703 is_inode_flag_set(inode, FI_NO_EXTENT))
2707 * for recovered files during mount do not create extents
2708 * if shrinker is not registered.
2710 if (list_empty(&sbi->s_list))
2713 return S_ISREG(inode->i_mode);
2716 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2717 size_t size, gfp_t flags)
2721 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2722 f2fs_show_injection_info(FAULT_KMALLOC);
2726 ret = kmalloc(size, flags);
2730 return kvmalloc(size, flags);
2733 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2734 size_t size, gfp_t flags)
2736 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2739 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2740 size_t size, gfp_t flags)
2742 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2743 f2fs_show_injection_info(FAULT_KVMALLOC);
2747 return kvmalloc(size, flags);
2750 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2751 size_t size, gfp_t flags)
2753 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2756 static inline int get_extra_isize(struct inode *inode)
2758 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2761 static inline int get_inline_xattr_addrs(struct inode *inode)
2763 return F2FS_I(inode)->i_inline_xattr_size;
2766 #define f2fs_get_inode_mode(i) \
2767 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2768 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2770 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2771 (offsetof(struct f2fs_inode, i_extra_end) - \
2772 offsetof(struct f2fs_inode, i_extra_isize)) \
2774 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2775 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2776 ((offsetof(typeof(*(f2fs_inode)), field) + \
2777 sizeof((f2fs_inode)->field)) \
2778 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
2780 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2784 spin_lock(&sbi->iostat_lock);
2785 for (i = 0; i < NR_IO_TYPE; i++)
2786 sbi->write_iostat[i] = 0;
2787 spin_unlock(&sbi->iostat_lock);
2790 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2791 enum iostat_type type, unsigned long long io_bytes)
2793 if (!sbi->iostat_enable)
2795 spin_lock(&sbi->iostat_lock);
2796 sbi->write_iostat[type] += io_bytes;
2798 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2799 sbi->write_iostat[APP_BUFFERED_IO] =
2800 sbi->write_iostat[APP_WRITE_IO] -
2801 sbi->write_iostat[APP_DIRECT_IO];
2802 spin_unlock(&sbi->iostat_lock);
2805 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2807 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META && \
2808 (!is_read_io((fio)->op) || (fio)->is_meta))
2810 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2811 block_t blkaddr, int type);
2812 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2813 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2814 block_t blkaddr, int type)
2816 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2817 f2fs_msg(sbi->sb, KERN_ERR,
2818 "invalid blkaddr: %u, type: %d, run fsck to fix.",
2820 f2fs_bug_on(sbi, 1);
2824 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2826 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2831 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2834 if (!__is_valid_data_blkaddr(blkaddr))
2836 verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2843 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2844 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2845 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
2846 int f2fs_truncate(struct inode *inode);
2847 int f2fs_getattr(const struct path *path, struct kstat *stat,
2848 u32 request_mask, unsigned int flags);
2849 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2850 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2851 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2852 int f2fs_precache_extents(struct inode *inode);
2853 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2854 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2855 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
2856 int f2fs_pin_file_control(struct inode *inode, bool inc);
2861 void f2fs_set_inode_flags(struct inode *inode);
2862 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2863 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2864 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2865 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2866 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2867 void f2fs_update_inode(struct inode *inode, struct page *node_page);
2868 void f2fs_update_inode_page(struct inode *inode);
2869 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2870 void f2fs_evict_inode(struct inode *inode);
2871 void f2fs_handle_failed_inode(struct inode *inode);
2876 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2877 bool hot, bool set);
2878 struct dentry *f2fs_get_parent(struct dentry *child);
2883 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2884 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2885 f2fs_hash_t namehash, int *max_slots,
2886 struct f2fs_dentry_ptr *d);
2887 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2888 unsigned int start_pos, struct fscrypt_str *fstr);
2889 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2890 struct f2fs_dentry_ptr *d);
2891 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2892 const struct qstr *new_name,
2893 const struct qstr *orig_name, struct page *dpage);
2894 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2895 unsigned int current_depth);
2896 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2897 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2898 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2899 struct fscrypt_name *fname, struct page **res_page);
2900 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2901 const struct qstr *child, struct page **res_page);
2902 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2903 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2904 struct page **page);
2905 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2906 struct page *page, struct inode *inode);
2907 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2908 const struct qstr *name, f2fs_hash_t name_hash,
2909 unsigned int bit_pos);
2910 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2911 const struct qstr *orig_name,
2912 struct inode *inode, nid_t ino, umode_t mode);
2913 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2914 struct inode *inode, nid_t ino, umode_t mode);
2915 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2916 struct inode *inode, nid_t ino, umode_t mode);
2917 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2918 struct inode *dir, struct inode *inode);
2919 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2920 bool f2fs_empty_dir(struct inode *dir);
2922 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2924 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2925 inode, inode->i_ino, inode->i_mode);
2931 int f2fs_inode_dirtied(struct inode *inode, bool sync);
2932 void f2fs_inode_synced(struct inode *inode);
2933 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
2934 int f2fs_quota_sync(struct super_block *sb, int type);
2935 void f2fs_quota_off_umount(struct super_block *sb);
2936 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
2937 int f2fs_sync_fs(struct super_block *sb, int sync);
2938 extern __printf(3, 4)
2939 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2940 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
2945 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info,
2946 struct fscrypt_name *fname);
2951 struct dnode_of_data;
2954 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
2955 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
2956 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
2957 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
2958 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
2959 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
2960 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
2961 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
2962 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
2963 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
2964 struct node_info *ni);
2965 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
2966 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
2967 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
2968 int f2fs_truncate_xattr_node(struct inode *inode);
2969 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
2970 unsigned int seq_id);
2971 int f2fs_remove_inode_page(struct inode *inode);
2972 struct page *f2fs_new_inode_page(struct inode *inode);
2973 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
2974 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
2975 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
2976 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
2977 int f2fs_move_node_page(struct page *node_page, int gc_type);
2978 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
2979 struct writeback_control *wbc, bool atomic,
2980 unsigned int *seq_id);
2981 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
2982 struct writeback_control *wbc,
2983 bool do_balance, enum iostat_type io_type);
2984 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
2985 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
2986 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
2987 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
2988 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
2989 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
2990 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
2991 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
2992 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
2993 unsigned int segno, struct f2fs_summary_block *sum);
2994 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2995 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
2996 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
2997 int __init f2fs_create_node_manager_caches(void);
2998 void f2fs_destroy_node_manager_caches(void);
3003 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3004 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3005 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3006 void f2fs_drop_inmem_pages(struct inode *inode);
3007 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3008 int f2fs_commit_inmem_pages(struct inode *inode);
3009 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3010 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
3011 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3012 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3013 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3014 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3015 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3016 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3017 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3018 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3019 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3020 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3021 struct cp_control *cpc);
3022 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3023 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi);
3024 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3025 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3026 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3027 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3028 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3029 struct cp_control *cpc);
3030 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3031 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3033 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3034 enum iostat_type io_type);
3035 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3036 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3037 struct f2fs_io_info *fio);
3038 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3039 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3040 block_t old_blkaddr, block_t new_blkaddr,
3041 bool recover_curseg, bool recover_newaddr);
3042 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3043 block_t old_addr, block_t new_addr,
3044 unsigned char version, bool recover_curseg,
3045 bool recover_newaddr);
3046 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3047 block_t old_blkaddr, block_t *new_blkaddr,
3048 struct f2fs_summary *sum, int type,
3049 struct f2fs_io_info *fio, bool add_list);
3050 void f2fs_wait_on_page_writeback(struct page *page,
3051 enum page_type type, bool ordered, bool locked);
3052 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3053 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3055 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3056 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3057 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3058 unsigned int val, int alloc);
3059 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3060 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3061 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3062 int __init f2fs_create_segment_manager_caches(void);
3063 void f2fs_destroy_segment_manager_caches(void);
3064 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3065 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3066 enum page_type type, enum temp_type temp);
3071 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3072 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3073 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3074 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3075 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3076 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3077 block_t blkaddr, int type);
3078 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3079 int type, bool sync);
3080 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3081 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3082 long nr_to_write, enum iostat_type io_type);
3083 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3084 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3085 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3086 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3087 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3088 unsigned int devidx, int type);
3089 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3090 unsigned int devidx, int type);
3091 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3092 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3093 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3094 void f2fs_add_orphan_inode(struct inode *inode);
3095 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3096 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3097 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3098 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3099 void f2fs_remove_dirty_inode(struct inode *inode);
3100 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3101 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3102 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3103 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3104 int __init f2fs_create_checkpoint_caches(void);
3105 void f2fs_destroy_checkpoint_caches(void);
3110 int f2fs_init_post_read_processing(void);
3111 void f2fs_destroy_post_read_processing(void);
3112 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3113 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3114 struct inode *inode, struct page *page,
3115 nid_t ino, enum page_type type);
3116 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3117 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3118 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3119 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3120 block_t blk_addr, struct bio *bio);
3121 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3122 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3123 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3124 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3125 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3126 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3127 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3128 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3129 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3130 int op_flags, bool for_write);
3131 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3132 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3134 struct page *f2fs_get_new_data_page(struct inode *inode,
3135 struct page *ipage, pgoff_t index, bool new_i_size);
3136 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3137 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3138 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3139 int create, int flag);
3140 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3141 u64 start, u64 len);
3142 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3143 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3144 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3145 unsigned int length);
3146 int f2fs_release_page(struct page *page, gfp_t wait);
3147 #ifdef CONFIG_MIGRATION
3148 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3149 struct page *page, enum migrate_mode mode);
3151 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3152 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3157 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3158 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3159 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3160 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3161 unsigned int segno);
3162 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3167 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3168 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3173 #ifdef CONFIG_F2FS_STAT_FS
3174 struct f2fs_stat_info {
3175 struct list_head stat_list;
3176 struct f2fs_sb_info *sbi;
3177 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3178 int main_area_segs, main_area_sections, main_area_zones;
3179 unsigned long long hit_largest, hit_cached, hit_rbtree;
3180 unsigned long long hit_total, total_ext;
3181 int ext_tree, zombie_tree, ext_node;
3182 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3183 int ndirty_data, ndirty_qdata;
3185 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3186 int nats, dirty_nats, sits, dirty_sits;
3187 int free_nids, avail_nids, alloc_nids;
3188 int total_count, utilization;
3189 int bg_gc, nr_wb_cp_data, nr_wb_data;
3190 int nr_rd_data, nr_rd_node, nr_rd_meta;
3191 int nr_dio_read, nr_dio_write;
3192 unsigned int io_skip_bggc, other_skip_bggc;
3193 int nr_flushing, nr_flushed, flush_list_empty;
3194 int nr_discarding, nr_discarded;
3196 unsigned int undiscard_blks;
3197 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3198 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3199 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3200 unsigned int bimodal, avg_vblocks;
3201 int util_free, util_valid, util_invalid;
3202 int rsvd_segs, overp_segs;
3203 int dirty_count, node_pages, meta_pages;
3204 int prefree_count, call_count, cp_count, bg_cp_count;
3205 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3206 int bg_node_segs, bg_data_segs;
3207 int tot_blks, data_blks, node_blks;
3208 int bg_data_blks, bg_node_blks;
3209 unsigned long long skipped_atomic_files[2];
3210 int curseg[NR_CURSEG_TYPE];
3211 int cursec[NR_CURSEG_TYPE];
3212 int curzone[NR_CURSEG_TYPE];
3214 unsigned int meta_count[META_MAX];
3215 unsigned int segment_count[2];
3216 unsigned int block_count[2];
3217 unsigned int inplace_count;
3218 unsigned long long base_mem, cache_mem, page_mem;
3221 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3223 return (struct f2fs_stat_info *)sbi->stat_info;
3226 #define stat_inc_cp_count(si) ((si)->cp_count++)
3227 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3228 #define stat_inc_call_count(si) ((si)->call_count++)
3229 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3230 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3231 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3232 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3233 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3234 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3235 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3236 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3237 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3238 #define stat_inc_inline_xattr(inode) \
3240 if (f2fs_has_inline_xattr(inode)) \
3241 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3243 #define stat_dec_inline_xattr(inode) \
3245 if (f2fs_has_inline_xattr(inode)) \
3246 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3248 #define stat_inc_inline_inode(inode) \
3250 if (f2fs_has_inline_data(inode)) \
3251 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3253 #define stat_dec_inline_inode(inode) \
3255 if (f2fs_has_inline_data(inode)) \
3256 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3258 #define stat_inc_inline_dir(inode) \
3260 if (f2fs_has_inline_dentry(inode)) \
3261 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3263 #define stat_dec_inline_dir(inode) \
3265 if (f2fs_has_inline_dentry(inode)) \
3266 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3268 #define stat_inc_meta_count(sbi, blkaddr) \
3270 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3271 atomic_inc(&(sbi)->meta_count[META_CP]); \
3272 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3273 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3274 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3275 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3276 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3277 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3279 #define stat_inc_seg_type(sbi, curseg) \
3280 ((sbi)->segment_count[(curseg)->alloc_type]++)
3281 #define stat_inc_block_count(sbi, curseg) \
3282 ((sbi)->block_count[(curseg)->alloc_type]++)
3283 #define stat_inc_inplace_blocks(sbi) \
3284 (atomic_inc(&(sbi)->inplace_count))
3285 #define stat_inc_atomic_write(inode) \
3286 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3287 #define stat_dec_atomic_write(inode) \
3288 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3289 #define stat_update_max_atomic_write(inode) \
3291 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3292 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3294 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3296 #define stat_inc_volatile_write(inode) \
3297 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3298 #define stat_dec_volatile_write(inode) \
3299 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3300 #define stat_update_max_volatile_write(inode) \
3302 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3303 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3305 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3307 #define stat_inc_seg_count(sbi, type, gc_type) \
3309 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3311 if ((type) == SUM_TYPE_DATA) { \
3313 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3316 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3320 #define stat_inc_tot_blk_count(si, blks) \
3321 ((si)->tot_blks += (blks))
3323 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3325 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3326 stat_inc_tot_blk_count(si, blks); \
3327 si->data_blks += (blks); \
3328 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3331 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3333 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3334 stat_inc_tot_blk_count(si, blks); \
3335 si->node_blks += (blks); \
3336 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3339 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3340 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3341 void __init f2fs_create_root_stats(void);
3342 void f2fs_destroy_root_stats(void);
3344 #define stat_inc_cp_count(si) do { } while (0)
3345 #define stat_inc_bg_cp_count(si) do { } while (0)
3346 #define stat_inc_call_count(si) do { } while (0)
3347 #define stat_inc_bggc_count(si) do { } while (0)
3348 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3349 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3350 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3351 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3352 #define stat_inc_total_hit(sb) do { } while (0)
3353 #define stat_inc_rbtree_node_hit(sb) do { } while (0)
3354 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3355 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3356 #define stat_inc_inline_xattr(inode) do { } while (0)
3357 #define stat_dec_inline_xattr(inode) do { } while (0)
3358 #define stat_inc_inline_inode(inode) do { } while (0)
3359 #define stat_dec_inline_inode(inode) do { } while (0)
3360 #define stat_inc_inline_dir(inode) do { } while (0)
3361 #define stat_dec_inline_dir(inode) do { } while (0)
3362 #define stat_inc_atomic_write(inode) do { } while (0)
3363 #define stat_dec_atomic_write(inode) do { } while (0)
3364 #define stat_update_max_atomic_write(inode) do { } while (0)
3365 #define stat_inc_volatile_write(inode) do { } while (0)
3366 #define stat_dec_volatile_write(inode) do { } while (0)
3367 #define stat_update_max_volatile_write(inode) do { } while (0)
3368 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3369 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3370 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3371 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3372 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3373 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3374 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3375 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3377 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3378 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3379 static inline void __init f2fs_create_root_stats(void) { }
3380 static inline void f2fs_destroy_root_stats(void) { }
3383 extern const struct file_operations f2fs_dir_operations;
3384 extern const struct file_operations f2fs_file_operations;
3385 extern const struct inode_operations f2fs_file_inode_operations;
3386 extern const struct address_space_operations f2fs_dblock_aops;
3387 extern const struct address_space_operations f2fs_node_aops;
3388 extern const struct address_space_operations f2fs_meta_aops;
3389 extern const struct inode_operations f2fs_dir_inode_operations;
3390 extern const struct inode_operations f2fs_symlink_inode_operations;
3391 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3392 extern const struct inode_operations f2fs_special_inode_operations;
3393 extern struct kmem_cache *f2fs_inode_entry_slab;
3398 bool f2fs_may_inline_data(struct inode *inode);
3399 bool f2fs_may_inline_dentry(struct inode *inode);
3400 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3401 void f2fs_truncate_inline_inode(struct inode *inode,
3402 struct page *ipage, u64 from);
3403 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3404 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3405 int f2fs_convert_inline_inode(struct inode *inode);
3406 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3407 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3408 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3409 struct fscrypt_name *fname, struct page **res_page);
3410 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3411 struct page *ipage);
3412 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3413 const struct qstr *orig_name,
3414 struct inode *inode, nid_t ino, umode_t mode);
3415 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3416 struct page *page, struct inode *dir,
3417 struct inode *inode);
3418 bool f2fs_empty_inline_dir(struct inode *dir);
3419 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3420 struct fscrypt_str *fstr);
3421 int f2fs_inline_data_fiemap(struct inode *inode,
3422 struct fiemap_extent_info *fieinfo,
3423 __u64 start, __u64 len);
3428 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3429 struct shrink_control *sc);
3430 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3431 struct shrink_control *sc);
3432 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3433 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3438 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3439 struct rb_entry *cached_re, unsigned int ofs);
3440 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3441 struct rb_root_cached *root,
3442 struct rb_node **parent,
3443 unsigned int ofs, bool *leftmost);
3444 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3445 struct rb_entry *cached_re, unsigned int ofs,
3446 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3447 struct rb_node ***insert_p, struct rb_node **insert_parent,
3448 bool force, bool *leftmost);
3449 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3450 struct rb_root_cached *root);
3451 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3452 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3453 void f2fs_drop_extent_tree(struct inode *inode);
3454 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3455 void f2fs_destroy_extent_tree(struct inode *inode);
3456 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3457 struct extent_info *ei);
3458 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3459 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3460 pgoff_t fofs, block_t blkaddr, unsigned int len);
3461 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3462 int __init f2fs_create_extent_cache(void);
3463 void f2fs_destroy_extent_cache(void);
3468 int __init f2fs_init_sysfs(void);
3469 void f2fs_exit_sysfs(void);
3470 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3471 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3476 static inline bool f2fs_encrypted_inode(struct inode *inode)
3478 return file_is_encrypt(inode);
3481 static inline bool f2fs_encrypted_file(struct inode *inode)
3483 return f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode);
3486 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3488 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3489 file_set_encrypt(inode);
3490 f2fs_set_inode_flags(inode);
3495 * Returns true if the reads of the inode's data need to undergo some
3496 * postprocessing step, like decryption or authenticity verification.
3498 static inline bool f2fs_post_read_required(struct inode *inode)
3500 return f2fs_encrypted_file(inode);
3503 #define F2FS_FEATURE_FUNCS(name, flagname) \
3504 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3506 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3509 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3510 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3511 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3512 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3513 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3514 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3515 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3516 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3517 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3518 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3520 #ifdef CONFIG_BLK_DEV_ZONED
3521 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3522 struct block_device *bdev, block_t blkaddr)
3524 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3527 for (i = 0; i < sbi->s_ndevs; i++)
3528 if (FDEV(i).bdev == bdev)
3529 return FDEV(i).blkz_type[zno];
3534 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3536 return f2fs_sb_has_blkzoned(sbi);
3539 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3541 return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
3544 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3546 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3547 f2fs_hw_should_discard(sbi);
3550 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3552 clear_opt(sbi, ADAPTIVE);
3553 clear_opt(sbi, LFS);
3556 case F2FS_MOUNT_ADAPTIVE:
3557 set_opt(sbi, ADAPTIVE);
3559 case F2FS_MOUNT_LFS:
3565 static inline bool f2fs_may_encrypt(struct inode *inode)
3567 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3568 umode_t mode = inode->i_mode;
3570 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3576 static inline int block_unaligned_IO(struct inode *inode,
3577 struct kiocb *iocb, struct iov_iter *iter)
3579 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
3580 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
3581 loff_t offset = iocb->ki_pos;
3582 unsigned long align = offset | iov_iter_alignment(iter);
3584 return align & blocksize_mask;
3587 static inline int allow_outplace_dio(struct inode *inode,
3588 struct kiocb *iocb, struct iov_iter *iter)
3590 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3591 int rw = iov_iter_rw(iter);
3593 return (test_opt(sbi, LFS) && (rw == WRITE) &&
3594 !block_unaligned_IO(inode, iocb, iter));
3597 static inline bool f2fs_force_buffered_io(struct inode *inode,
3598 struct kiocb *iocb, struct iov_iter *iter)
3600 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3601 int rw = iov_iter_rw(iter);
3603 if (f2fs_post_read_required(inode))
3608 * for blkzoned device, fallback direct IO to buffered IO, so
3609 * all IOs can be serialized by log-structured write.
3611 if (f2fs_sb_has_blkzoned(sbi))
3613 if (test_opt(sbi, LFS) && (rw == WRITE) &&
3614 block_unaligned_IO(inode, iocb, iter))
3616 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
3622 #ifdef CONFIG_F2FS_FAULT_INJECTION
3623 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3626 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3629 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
3632 if (f2fs_sb_has_quota_ino(sbi))
3634 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
3635 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
3636 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])