4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
14 #include <linux/types.h>
15 #include <linux/page-flags.h>
16 #include <linux/buffer_head.h>
17 #include <linux/slab.h>
18 #include <linux/crc32.h>
19 #include <linux/magic.h>
20 #include <linux/kobject.h>
21 #include <linux/sched.h>
22 #include <linux/vmalloc.h>
23 #include <linux/bio.h>
24 #include <linux/blkdev.h>
25 #ifdef CONFIG_F2FS_FS_ENCRYPTION
26 #include <linux/fscrypt_supp.h>
28 #include <linux/fscrypt_notsupp.h>
30 #include <crypto/hash.h>
32 #ifdef CONFIG_F2FS_CHECK_FS
33 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
35 #define f2fs_bug_on(sbi, condition) \
37 if (unlikely(condition)) { \
39 set_sbi_flag(sbi, SBI_NEED_FSCK); \
44 #ifdef CONFIG_F2FS_FAULT_INJECTION
58 struct f2fs_fault_info {
60 unsigned int inject_rate;
61 unsigned int inject_type;
64 extern char *fault_name[FAULT_MAX];
65 #define IS_FAULT_SET(fi, type) (fi->inject_type & (1 << (type)))
71 #define F2FS_MOUNT_BG_GC 0x00000001
72 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
73 #define F2FS_MOUNT_DISCARD 0x00000004
74 #define F2FS_MOUNT_NOHEAP 0x00000008
75 #define F2FS_MOUNT_XATTR_USER 0x00000010
76 #define F2FS_MOUNT_POSIX_ACL 0x00000020
77 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
78 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
79 #define F2FS_MOUNT_INLINE_DATA 0x00000100
80 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200
81 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400
82 #define F2FS_MOUNT_NOBARRIER 0x00000800
83 #define F2FS_MOUNT_FASTBOOT 0x00001000
84 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000
85 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000
86 #define F2FS_MOUNT_DATA_FLUSH 0x00008000
87 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
88 #define F2FS_MOUNT_ADAPTIVE 0x00020000
89 #define F2FS_MOUNT_LFS 0x00040000
91 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
92 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
93 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
95 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
96 typecheck(unsigned long long, b) && \
97 ((long long)((a) - (b)) > 0))
99 typedef u32 block_t; /*
100 * should not change u32, since it is the on-disk block
101 * address format, __le32.
105 struct f2fs_mount_info {
109 #define F2FS_FEATURE_ENCRYPT 0x0001
110 #define F2FS_FEATURE_BLKZONED 0x0002
112 #define F2FS_HAS_FEATURE(sb, mask) \
113 ((F2FS_SB(sb)->raw_super->feature & cpu_to_le32(mask)) != 0)
114 #define F2FS_SET_FEATURE(sb, mask) \
115 F2FS_SB(sb)->raw_super->feature |= cpu_to_le32(mask)
116 #define F2FS_CLEAR_FEATURE(sb, mask) \
117 F2FS_SB(sb)->raw_super->feature &= ~cpu_to_le32(mask)
120 * For checkpoint manager
135 #define DEF_BATCHED_TRIM_SECTIONS 2
136 #define BATCHED_TRIM_SEGMENTS(sbi) \
137 (SM_I(sbi)->trim_sections * (sbi)->segs_per_sec)
138 #define BATCHED_TRIM_BLOCKS(sbi) \
139 (BATCHED_TRIM_SEGMENTS(sbi) << (sbi)->log_blocks_per_seg)
140 #define DEF_CP_INTERVAL 60 /* 60 secs */
141 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
152 * For CP/NAT/SIT/SSA readahead
162 /* for the list of ino */
164 ORPHAN_INO, /* for orphan ino list */
165 APPEND_INO, /* for append ino list */
166 UPDATE_INO, /* for update ino list */
167 MAX_INO_ENTRY, /* max. list */
171 struct list_head list; /* list head */
172 nid_t ino; /* inode number */
175 /* for the list of inodes to be GCed */
177 struct list_head list; /* list head */
178 struct inode *inode; /* vfs inode pointer */
181 /* for the list of blockaddresses to be discarded */
182 struct discard_entry {
183 struct list_head list; /* list head */
184 block_t blkaddr; /* block address to be discarded */
185 int len; /* # of consecutive blocks of the discard */
189 struct list_head list;
191 struct completion event;
195 /* for the list of fsync inodes, used only during recovery */
196 struct fsync_inode_entry {
197 struct list_head list; /* list head */
198 struct inode *inode; /* vfs inode pointer */
199 block_t blkaddr; /* block address locating the last fsync */
200 block_t last_dentry; /* block address locating the last dentry */
203 #define nats_in_cursum(jnl) (le16_to_cpu(jnl->n_nats))
204 #define sits_in_cursum(jnl) (le16_to_cpu(jnl->n_sits))
206 #define nat_in_journal(jnl, i) (jnl->nat_j.entries[i].ne)
207 #define nid_in_journal(jnl, i) (jnl->nat_j.entries[i].nid)
208 #define sit_in_journal(jnl, i) (jnl->sit_j.entries[i].se)
209 #define segno_in_journal(jnl, i) (jnl->sit_j.entries[i].segno)
211 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
212 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
214 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
216 int before = nats_in_cursum(journal);
217 journal->n_nats = cpu_to_le16(before + i);
221 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
223 int before = sits_in_cursum(journal);
224 journal->n_sits = cpu_to_le16(before + i);
228 static inline bool __has_cursum_space(struct f2fs_journal *journal,
231 if (type == NAT_JOURNAL)
232 return size <= MAX_NAT_JENTRIES(journal);
233 return size <= MAX_SIT_JENTRIES(journal);
239 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
240 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
241 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
243 #define F2FS_IOCTL_MAGIC 0xf5
244 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
245 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
246 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
247 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
248 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
249 #define F2FS_IOC_GARBAGE_COLLECT _IO(F2FS_IOCTL_MAGIC, 6)
250 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
251 #define F2FS_IOC_DEFRAGMENT _IO(F2FS_IOCTL_MAGIC, 8)
252 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
253 struct f2fs_move_range)
255 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
256 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
257 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
260 * should be same as XFS_IOC_GOINGDOWN.
261 * Flags for going down operation used by FS_IOC_GOINGDOWN
263 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
264 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
265 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
266 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
267 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
269 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
271 * ioctl commands in 32 bit emulation
273 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
274 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
275 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
278 struct f2fs_defragment {
283 struct f2fs_move_range {
284 u32 dst_fd; /* destination fd */
285 u64 pos_in; /* start position in src_fd */
286 u64 pos_out; /* start position in dst_fd */
287 u64 len; /* size to move */
291 * For INODE and NODE manager
293 /* for directory operations */
294 struct f2fs_dentry_ptr {
297 struct f2fs_dir_entry *dentry;
298 __u8 (*filename)[F2FS_SLOT_LEN];
302 static inline void make_dentry_ptr(struct inode *inode,
303 struct f2fs_dentry_ptr *d, void *src, int type)
308 struct f2fs_dentry_block *t = (struct f2fs_dentry_block *)src;
309 d->max = NR_DENTRY_IN_BLOCK;
310 d->bitmap = &t->dentry_bitmap;
311 d->dentry = t->dentry;
312 d->filename = t->filename;
314 struct f2fs_inline_dentry *t = (struct f2fs_inline_dentry *)src;
315 d->max = NR_INLINE_DENTRY;
316 d->bitmap = &t->dentry_bitmap;
317 d->dentry = t->dentry;
318 d->filename = t->filename;
323 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
324 * as its node offset to distinguish from index node blocks.
325 * But some bits are used to mark the node block.
327 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
330 ALLOC_NODE, /* allocate a new node page if needed */
331 LOOKUP_NODE, /* look up a node without readahead */
333 * look up a node with readahead called
338 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
340 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
342 /* vector size for gang look-up from extent cache that consists of radix tree */
343 #define EXT_TREE_VEC_SIZE 64
345 /* for in-memory extent cache entry */
346 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
348 /* number of extent info in extent cache we try to shrink */
349 #define EXTENT_CACHE_SHRINK_NUMBER 128
352 unsigned int fofs; /* start offset in a file */
353 u32 blk; /* start block address of the extent */
354 unsigned int len; /* length of the extent */
358 struct rb_node rb_node; /* rb node located in rb-tree */
359 struct list_head list; /* node in global extent list of sbi */
360 struct extent_info ei; /* extent info */
361 struct extent_tree *et; /* extent tree pointer */
365 nid_t ino; /* inode number */
366 struct rb_root root; /* root of extent info rb-tree */
367 struct extent_node *cached_en; /* recently accessed extent node */
368 struct extent_info largest; /* largested extent info */
369 struct list_head list; /* to be used by sbi->zombie_list */
370 rwlock_t lock; /* protect extent info rb-tree */
371 atomic_t node_cnt; /* # of extent node in rb-tree*/
375 * This structure is taken from ext4_map_blocks.
377 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
379 #define F2FS_MAP_NEW (1 << BH_New)
380 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
381 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
382 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
385 struct f2fs_map_blocks {
389 unsigned int m_flags;
390 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
393 /* for flag in get_data_block */
394 #define F2FS_GET_BLOCK_READ 0
395 #define F2FS_GET_BLOCK_DIO 1
396 #define F2FS_GET_BLOCK_FIEMAP 2
397 #define F2FS_GET_BLOCK_BMAP 3
398 #define F2FS_GET_BLOCK_PRE_DIO 4
399 #define F2FS_GET_BLOCK_PRE_AIO 5
402 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
404 #define FADVISE_COLD_BIT 0x01
405 #define FADVISE_LOST_PINO_BIT 0x02
406 #define FADVISE_ENCRYPT_BIT 0x04
407 #define FADVISE_ENC_NAME_BIT 0x08
408 #define FADVISE_KEEP_SIZE_BIT 0x10
410 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
411 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
412 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
413 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
414 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
415 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
416 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
417 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
418 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
419 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
420 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
421 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
422 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
424 #define DEF_DIR_LEVEL 0
426 struct f2fs_inode_info {
427 struct inode vfs_inode; /* serve a vfs inode */
428 unsigned long i_flags; /* keep an inode flags for ioctl */
429 unsigned char i_advise; /* use to give file attribute hints */
430 unsigned char i_dir_level; /* use for dentry level for large dir */
431 unsigned int i_current_depth; /* use only in directory structure */
432 unsigned int i_pino; /* parent inode number */
433 umode_t i_acl_mode; /* keep file acl mode temporarily */
435 /* Use below internally in f2fs*/
436 unsigned long flags; /* use to pass per-file flags */
437 struct rw_semaphore i_sem; /* protect fi info */
438 atomic_t dirty_pages; /* # of dirty pages */
439 f2fs_hash_t chash; /* hash value of given file name */
440 unsigned int clevel; /* maximum level of given file name */
441 nid_t i_xattr_nid; /* node id that contains xattrs */
442 unsigned long long xattr_ver; /* cp version of xattr modification */
443 loff_t last_disk_size; /* lastly written file size */
445 struct list_head dirty_list; /* dirty list for dirs and files */
446 struct list_head gdirty_list; /* linked in global dirty list */
447 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
448 struct mutex inmem_lock; /* lock for inmemory pages */
449 struct extent_tree *extent_tree; /* cached extent_tree entry */
450 struct rw_semaphore dio_rwsem[2];/* avoid racing between dio and gc */
453 static inline void get_extent_info(struct extent_info *ext,
454 struct f2fs_extent *i_ext)
456 ext->fofs = le32_to_cpu(i_ext->fofs);
457 ext->blk = le32_to_cpu(i_ext->blk);
458 ext->len = le32_to_cpu(i_ext->len);
461 static inline void set_raw_extent(struct extent_info *ext,
462 struct f2fs_extent *i_ext)
464 i_ext->fofs = cpu_to_le32(ext->fofs);
465 i_ext->blk = cpu_to_le32(ext->blk);
466 i_ext->len = cpu_to_le32(ext->len);
469 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
470 u32 blk, unsigned int len)
477 static inline bool __is_extent_same(struct extent_info *ei1,
478 struct extent_info *ei2)
480 return (ei1->fofs == ei2->fofs && ei1->blk == ei2->blk &&
481 ei1->len == ei2->len);
484 static inline bool __is_extent_mergeable(struct extent_info *back,
485 struct extent_info *front)
487 return (back->fofs + back->len == front->fofs &&
488 back->blk + back->len == front->blk);
491 static inline bool __is_back_mergeable(struct extent_info *cur,
492 struct extent_info *back)
494 return __is_extent_mergeable(back, cur);
497 static inline bool __is_front_mergeable(struct extent_info *cur,
498 struct extent_info *front)
500 return __is_extent_mergeable(cur, front);
503 extern void f2fs_mark_inode_dirty_sync(struct inode *, bool);
504 static inline void __try_update_largest_extent(struct inode *inode,
505 struct extent_tree *et, struct extent_node *en)
507 if (en->ei.len > et->largest.len) {
508 et->largest = en->ei;
509 f2fs_mark_inode_dirty_sync(inode, true);
519 struct f2fs_nm_info {
520 block_t nat_blkaddr; /* base disk address of NAT */
521 nid_t max_nid; /* maximum possible node ids */
522 nid_t available_nids; /* # of available node ids */
523 nid_t next_scan_nid; /* the next nid to be scanned */
524 unsigned int ram_thresh; /* control the memory footprint */
525 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
526 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
528 /* NAT cache management */
529 struct radix_tree_root nat_root;/* root of the nat entry cache */
530 struct radix_tree_root nat_set_root;/* root of the nat set cache */
531 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
532 struct list_head nat_entries; /* cached nat entry list (clean) */
533 unsigned int nat_cnt; /* the # of cached nat entries */
534 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
536 /* free node ids management */
537 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
538 struct list_head nid_list[MAX_NID_LIST];/* lists for free nids */
539 unsigned int nid_cnt[MAX_NID_LIST]; /* the number of free node id */
540 spinlock_t nid_list_lock; /* protect nid lists ops */
541 struct mutex build_lock; /* lock for build free nids */
544 char *nat_bitmap; /* NAT bitmap pointer */
545 int bitmap_size; /* bitmap size */
549 * this structure is used as one of function parameters.
550 * all the information are dedicated to a given direct node block determined
551 * by the data offset in a file.
553 struct dnode_of_data {
554 struct inode *inode; /* vfs inode pointer */
555 struct page *inode_page; /* its inode page, NULL is possible */
556 struct page *node_page; /* cached direct node page */
557 nid_t nid; /* node id of the direct node block */
558 unsigned int ofs_in_node; /* data offset in the node page */
559 bool inode_page_locked; /* inode page is locked or not */
560 bool node_changed; /* is node block changed */
561 char cur_level; /* level of hole node page */
562 char max_level; /* level of current page located */
563 block_t data_blkaddr; /* block address of the node block */
566 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
567 struct page *ipage, struct page *npage, nid_t nid)
569 memset(dn, 0, sizeof(*dn));
571 dn->inode_page = ipage;
572 dn->node_page = npage;
579 * By default, there are 6 active log areas across the whole main area.
580 * When considering hot and cold data separation to reduce cleaning overhead,
581 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
583 * In the current design, you should not change the numbers intentionally.
584 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
585 * logs individually according to the underlying devices. (default: 6)
586 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
587 * data and 8 for node logs.
589 #define NR_CURSEG_DATA_TYPE (3)
590 #define NR_CURSEG_NODE_TYPE (3)
591 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
594 CURSEG_HOT_DATA = 0, /* directory entry blocks */
595 CURSEG_WARM_DATA, /* data blocks */
596 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
597 CURSEG_HOT_NODE, /* direct node blocks of directory files */
598 CURSEG_WARM_NODE, /* direct node blocks of normal files */
599 CURSEG_COLD_NODE, /* indirect node blocks */
604 struct completion wait;
605 struct llist_node llnode;
609 struct flush_cmd_control {
610 struct task_struct *f2fs_issue_flush; /* flush thread */
611 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
612 atomic_t submit_flush; /* # of issued flushes */
613 struct llist_head issue_list; /* list for command issue */
614 struct llist_node *dispatch_list; /* list for command dispatch */
617 struct f2fs_sm_info {
618 struct sit_info *sit_info; /* whole segment information */
619 struct free_segmap_info *free_info; /* free segment information */
620 struct dirty_seglist_info *dirty_info; /* dirty segment information */
621 struct curseg_info *curseg_array; /* active segment information */
623 block_t seg0_blkaddr; /* block address of 0'th segment */
624 block_t main_blkaddr; /* start block address of main area */
625 block_t ssa_blkaddr; /* start block address of SSA area */
627 unsigned int segment_count; /* total # of segments */
628 unsigned int main_segments; /* # of segments in main area */
629 unsigned int reserved_segments; /* # of reserved segments */
630 unsigned int ovp_segments; /* # of overprovision segments */
632 /* a threshold to reclaim prefree segments */
633 unsigned int rec_prefree_segments;
635 /* for small discard management */
636 struct list_head discard_list; /* 4KB discard list */
637 struct list_head wait_list; /* linked with issued discard bio */
638 int nr_discards; /* # of discards in the list */
639 int max_discards; /* max. discards to be issued */
641 /* for batched trimming */
642 unsigned int trim_sections; /* # of sections to trim */
644 struct list_head sit_entry_set; /* sit entry set list */
646 unsigned int ipu_policy; /* in-place-update policy */
647 unsigned int min_ipu_util; /* in-place-update threshold */
648 unsigned int min_fsync_blocks; /* threshold for fsync */
650 /* for flush command control */
651 struct flush_cmd_control *cmd_control_info;
659 * COUNT_TYPE for monitoring
661 * f2fs monitors the number of several block types such as on-writeback,
662 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
664 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
678 * The below are the page types of bios used in submit_bio().
679 * The available types are:
680 * DATA User data pages. It operates as async mode.
681 * NODE Node pages. It operates as async mode.
682 * META FS metadata pages such as SIT, NAT, CP.
683 * NR_PAGE_TYPE The number of page types.
684 * META_FLUSH Make sure the previous pages are written
685 * with waiting the bio's completion
686 * ... Only can be used with META.
688 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
695 INMEM, /* the below types are used by tracepoints only. */
702 struct f2fs_io_info {
703 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
704 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
705 int op; /* contains REQ_OP_ */
706 int op_flags; /* req_flag_bits */
707 block_t new_blkaddr; /* new block address to be written */
708 block_t old_blkaddr; /* old block address before Cow */
709 struct page *page; /* page to be written */
710 struct page *encrypted_page; /* encrypted page */
713 #define is_read_io(rw) (rw == READ)
714 struct f2fs_bio_info {
715 struct f2fs_sb_info *sbi; /* f2fs superblock */
716 struct bio *bio; /* bios to merge */
717 sector_t last_block_in_bio; /* last block number */
718 struct f2fs_io_info fio; /* store buffered io info. */
719 struct rw_semaphore io_rwsem; /* blocking op for bio */
722 #define FDEV(i) (sbi->devs[i])
723 #define RDEV(i) (raw_super->devs[i])
724 struct f2fs_dev_info {
725 struct block_device *bdev;
726 char path[MAX_PATH_LEN];
727 unsigned int total_segments;
730 #ifdef CONFIG_BLK_DEV_ZONED
731 unsigned int nr_blkz; /* Total number of zones */
732 u8 *blkz_type; /* Array of zones type */
737 DIR_INODE, /* for dirty dir inode */
738 FILE_INODE, /* for dirty regular/symlink inode */
739 DIRTY_META, /* for all dirtied inode metadata */
743 /* for inner inode cache management */
744 struct inode_management {
745 struct radix_tree_root ino_root; /* ino entry array */
746 spinlock_t ino_lock; /* for ino entry lock */
747 struct list_head ino_list; /* inode list head */
748 unsigned long ino_num; /* number of entries */
751 /* For s_flag in struct f2fs_sb_info */
753 SBI_IS_DIRTY, /* dirty flag for checkpoint */
754 SBI_IS_CLOSE, /* specify unmounting */
755 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
756 SBI_POR_DOING, /* recovery is doing or not */
757 SBI_NEED_SB_WRITE, /* need to recover superblock */
758 SBI_NEED_CP, /* need to checkpoint */
767 struct f2fs_sb_info {
768 struct super_block *sb; /* pointer to VFS super block */
769 struct proc_dir_entry *s_proc; /* proc entry */
770 struct f2fs_super_block *raw_super; /* raw super block pointer */
771 int valid_super_block; /* valid super block no */
772 unsigned long s_flag; /* flags for sbi */
774 #ifdef CONFIG_BLK_DEV_ZONED
775 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
776 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
779 /* for node-related operations */
780 struct f2fs_nm_info *nm_info; /* node manager */
781 struct inode *node_inode; /* cache node blocks */
783 /* for segment-related operations */
784 struct f2fs_sm_info *sm_info; /* segment manager */
786 /* for bio operations */
787 struct f2fs_bio_info read_io; /* for read bios */
788 struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */
789 struct mutex wio_mutex[NODE + 1]; /* bio ordering for NODE/DATA */
792 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
793 int cur_cp_pack; /* remain current cp pack */
794 spinlock_t cp_lock; /* for flag in ckpt */
795 struct inode *meta_inode; /* cache meta blocks */
796 struct mutex cp_mutex; /* checkpoint procedure lock */
797 struct rw_semaphore cp_rwsem; /* blocking FS operations */
798 struct rw_semaphore node_write; /* locking node writes */
799 wait_queue_head_t cp_wait;
800 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
801 long interval_time[MAX_TIME]; /* to store thresholds */
803 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
805 /* for orphan inode, use 0'th array */
806 unsigned int max_orphans; /* max orphan inodes */
808 /* for inode management */
809 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
810 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
812 /* for extent tree cache */
813 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
814 struct rw_semaphore extent_tree_lock; /* locking extent radix tree */
815 struct list_head extent_list; /* lru list for shrinker */
816 spinlock_t extent_lock; /* locking extent lru list */
817 atomic_t total_ext_tree; /* extent tree count */
818 struct list_head zombie_list; /* extent zombie tree list */
819 atomic_t total_zombie_tree; /* extent zombie tree count */
820 atomic_t total_ext_node; /* extent info count */
822 /* basic filesystem units */
823 unsigned int log_sectors_per_block; /* log2 sectors per block */
824 unsigned int log_blocksize; /* log2 block size */
825 unsigned int blocksize; /* block size */
826 unsigned int root_ino_num; /* root inode number*/
827 unsigned int node_ino_num; /* node inode number*/
828 unsigned int meta_ino_num; /* meta inode number*/
829 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
830 unsigned int blocks_per_seg; /* blocks per segment */
831 unsigned int segs_per_sec; /* segments per section */
832 unsigned int secs_per_zone; /* sections per zone */
833 unsigned int total_sections; /* total section count */
834 unsigned int total_node_count; /* total node block count */
835 unsigned int total_valid_node_count; /* valid node block count */
836 loff_t max_file_blocks; /* max block index of file */
837 int active_logs; /* # of active logs */
838 int dir_level; /* directory level */
840 block_t user_block_count; /* # of user blocks */
841 block_t total_valid_block_count; /* # of valid blocks */
842 block_t discard_blks; /* discard command candidats */
843 block_t last_valid_block_count; /* for recovery */
844 u32 s_next_generation; /* for NFS support */
846 /* # of pages, see count_type */
847 atomic_t nr_pages[NR_COUNT_TYPE];
848 /* # of allocated blocks */
849 struct percpu_counter alloc_valid_block_count;
851 /* valid inode count */
852 struct percpu_counter total_valid_inode_count;
854 struct f2fs_mount_info mount_opt; /* mount options */
856 /* for cleaning operations */
857 struct mutex gc_mutex; /* mutex for GC */
858 struct f2fs_gc_kthread *gc_thread; /* GC thread */
859 unsigned int cur_victim_sec; /* current victim section num */
861 /* maximum # of trials to find a victim segment for SSR and GC */
862 unsigned int max_victim_search;
865 * for stat information.
866 * one is for the LFS mode, and the other is for the SSR mode.
868 #ifdef CONFIG_F2FS_STAT_FS
869 struct f2fs_stat_info *stat_info; /* FS status information */
870 unsigned int segment_count[2]; /* # of allocated segments */
871 unsigned int block_count[2]; /* # of allocated blocks */
872 atomic_t inplace_count; /* # of inplace update */
873 atomic64_t total_hit_ext; /* # of lookup extent cache */
874 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
875 atomic64_t read_hit_largest; /* # of hit largest extent node */
876 atomic64_t read_hit_cached; /* # of hit cached extent node */
877 atomic_t inline_xattr; /* # of inline_xattr inodes */
878 atomic_t inline_inode; /* # of inline_data inodes */
879 atomic_t inline_dir; /* # of inline_dentry inodes */
880 int bg_gc; /* background gc calls */
881 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
883 unsigned int last_victim[2]; /* last victim segment # */
884 spinlock_t stat_lock; /* lock for stat operations */
886 /* For sysfs suppport */
887 struct kobject s_kobj;
888 struct completion s_kobj_unregister;
890 /* For shrinker support */
891 struct list_head s_list;
892 int s_ndevs; /* number of devices */
893 struct f2fs_dev_info *devs; /* for device list */
894 struct mutex umount_mutex;
895 unsigned int shrinker_run_no;
897 /* For write statistics */
898 u64 sectors_written_start;
901 /* Reference to checksum algorithm driver via cryptoapi */
902 struct crypto_shash *s_chksum_driver;
904 /* For fault injection */
905 #ifdef CONFIG_F2FS_FAULT_INJECTION
906 struct f2fs_fault_info fault_info;
910 #ifdef CONFIG_F2FS_FAULT_INJECTION
911 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
913 struct f2fs_fault_info *ffi = &sbi->fault_info;
915 if (!ffi->inject_rate)
918 if (!IS_FAULT_SET(ffi, type))
921 atomic_inc(&ffi->inject_ops);
922 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
923 atomic_set(&ffi->inject_ops, 0);
924 printk("%sF2FS-fs : inject %s in %pF\n",
927 __builtin_return_address(0));
934 /* For write statistics. Suppose sector size is 512 bytes,
935 * and the return value is in kbytes. s is of struct f2fs_sb_info.
937 #define BD_PART_WRITTEN(s) \
938 (((u64)part_stat_read(s->sb->s_bdev->bd_part, sectors[1]) - \
939 s->sectors_written_start) >> 1)
941 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
943 sbi->last_time[type] = jiffies;
946 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
948 struct timespec ts = {sbi->interval_time[type], 0};
949 unsigned long interval = timespec_to_jiffies(&ts);
951 return time_after(jiffies, sbi->last_time[type] + interval);
954 static inline bool is_idle(struct f2fs_sb_info *sbi)
956 struct block_device *bdev = sbi->sb->s_bdev;
957 struct request_queue *q = bdev_get_queue(bdev);
958 struct request_list *rl = &q->root_rl;
960 if (rl->count[BLK_RW_SYNC] || rl->count[BLK_RW_ASYNC])
963 return f2fs_time_over(sbi, REQ_TIME);
969 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
972 SHASH_DESC_ON_STACK(shash, sbi->s_chksum_driver);
973 u32 *ctx = (u32 *)shash_desc_ctx(shash);
976 shash->tfm = sbi->s_chksum_driver;
978 *ctx = F2FS_SUPER_MAGIC;
980 err = crypto_shash_update(shash, address, length);
986 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
987 void *buf, size_t buf_size)
989 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
992 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
994 return container_of(inode, struct f2fs_inode_info, vfs_inode);
997 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
999 return sb->s_fs_info;
1002 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1004 return F2FS_SB(inode->i_sb);
1007 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1009 return F2FS_I_SB(mapping->host);
1012 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1014 return F2FS_M_SB(page->mapping);
1017 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1019 return (struct f2fs_super_block *)(sbi->raw_super);
1022 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1024 return (struct f2fs_checkpoint *)(sbi->ckpt);
1027 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1029 return (struct f2fs_node *)page_address(page);
1032 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1034 return &((struct f2fs_node *)page_address(page))->i;
1037 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1039 return (struct f2fs_nm_info *)(sbi->nm_info);
1042 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1044 return (struct f2fs_sm_info *)(sbi->sm_info);
1047 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1049 return (struct sit_info *)(SM_I(sbi)->sit_info);
1052 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1054 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1057 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1059 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1062 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1064 return sbi->meta_inode->i_mapping;
1067 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1069 return sbi->node_inode->i_mapping;
1072 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1074 return test_bit(type, &sbi->s_flag);
1077 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1079 set_bit(type, &sbi->s_flag);
1082 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1084 clear_bit(type, &sbi->s_flag);
1087 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1089 return le64_to_cpu(cp->checkpoint_ver);
1092 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1094 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1096 return ckpt_flags & f;
1099 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1101 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1104 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1106 unsigned int ckpt_flags;
1108 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1110 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1113 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1115 spin_lock(&sbi->cp_lock);
1116 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1117 spin_unlock(&sbi->cp_lock);
1120 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1122 unsigned int ckpt_flags;
1124 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1126 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1129 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1131 spin_lock(&sbi->cp_lock);
1132 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1133 spin_unlock(&sbi->cp_lock);
1136 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1138 down_read(&sbi->cp_rwsem);
1141 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1143 up_read(&sbi->cp_rwsem);
1146 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1148 down_write(&sbi->cp_rwsem);
1151 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1153 up_write(&sbi->cp_rwsem);
1156 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1158 int reason = CP_SYNC;
1160 if (test_opt(sbi, FASTBOOT))
1161 reason = CP_FASTBOOT;
1162 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1167 static inline bool __remain_node_summaries(int reason)
1169 return (reason == CP_UMOUNT || reason == CP_FASTBOOT);
1172 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1174 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1175 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1179 * Check whether the given nid is within node id range.
1181 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
1183 if (unlikely(nid < F2FS_ROOT_INO(sbi)))
1185 if (unlikely(nid >= NM_I(sbi)->max_nid))
1190 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
1193 * Check whether the inode has blocks or not
1195 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1197 if (F2FS_I(inode)->i_xattr_nid)
1198 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1;
1200 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS;
1203 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1205 return ofs == XATTR_NODE_OFFSET;
1208 static inline void f2fs_i_blocks_write(struct inode *, blkcnt_t, bool);
1209 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
1210 struct inode *inode, blkcnt_t *count)
1214 #ifdef CONFIG_F2FS_FAULT_INJECTION
1215 if (time_to_inject(sbi, FAULT_BLOCK))
1219 * let's increase this in prior to actual block count change in order
1220 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1222 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1224 spin_lock(&sbi->stat_lock);
1225 sbi->total_valid_block_count += (block_t)(*count);
1226 if (unlikely(sbi->total_valid_block_count > sbi->user_block_count)) {
1227 diff = sbi->total_valid_block_count - sbi->user_block_count;
1229 sbi->total_valid_block_count = sbi->user_block_count;
1231 spin_unlock(&sbi->stat_lock);
1232 percpu_counter_sub(&sbi->alloc_valid_block_count, diff);
1236 spin_unlock(&sbi->stat_lock);
1238 f2fs_i_blocks_write(inode, *count, true);
1242 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1243 struct inode *inode,
1246 spin_lock(&sbi->stat_lock);
1247 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1248 f2fs_bug_on(sbi, inode->i_blocks < count);
1249 sbi->total_valid_block_count -= (block_t)count;
1250 spin_unlock(&sbi->stat_lock);
1251 f2fs_i_blocks_write(inode, count, false);
1254 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1256 atomic_inc(&sbi->nr_pages[count_type]);
1258 if (count_type == F2FS_DIRTY_DATA || count_type == F2FS_INMEM_PAGES ||
1259 count_type == F2FS_WB_CP_DATA || count_type == F2FS_WB_DATA)
1262 set_sbi_flag(sbi, SBI_IS_DIRTY);
1265 static inline void inode_inc_dirty_pages(struct inode *inode)
1267 atomic_inc(&F2FS_I(inode)->dirty_pages);
1268 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1269 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1272 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1274 atomic_dec(&sbi->nr_pages[count_type]);
1277 static inline void inode_dec_dirty_pages(struct inode *inode)
1279 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1280 !S_ISLNK(inode->i_mode))
1283 atomic_dec(&F2FS_I(inode)->dirty_pages);
1284 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1285 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1288 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1290 return atomic_read(&sbi->nr_pages[count_type]);
1293 static inline int get_dirty_pages(struct inode *inode)
1295 return atomic_read(&F2FS_I(inode)->dirty_pages);
1298 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1300 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1301 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1302 sbi->log_blocks_per_seg;
1304 return segs / sbi->segs_per_sec;
1307 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1309 return sbi->total_valid_block_count;
1312 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1314 return sbi->discard_blks;
1317 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1319 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1321 /* return NAT or SIT bitmap */
1322 if (flag == NAT_BITMAP)
1323 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1324 else if (flag == SIT_BITMAP)
1325 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1330 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1332 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1335 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1337 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1340 if (__cp_payload(sbi) > 0) {
1341 if (flag == NAT_BITMAP)
1342 return &ckpt->sit_nat_version_bitmap;
1344 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1346 offset = (flag == NAT_BITMAP) ?
1347 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1348 return &ckpt->sit_nat_version_bitmap + offset;
1352 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1354 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1356 if (sbi->cur_cp_pack == 2)
1357 start_addr += sbi->blocks_per_seg;
1361 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1363 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1365 if (sbi->cur_cp_pack == 1)
1366 start_addr += sbi->blocks_per_seg;
1370 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1372 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1375 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1377 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1380 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
1381 struct inode *inode)
1383 block_t valid_block_count;
1384 unsigned int valid_node_count;
1386 spin_lock(&sbi->stat_lock);
1388 valid_block_count = sbi->total_valid_block_count + 1;
1389 if (unlikely(valid_block_count > sbi->user_block_count)) {
1390 spin_unlock(&sbi->stat_lock);
1394 valid_node_count = sbi->total_valid_node_count + 1;
1395 if (unlikely(valid_node_count > sbi->total_node_count)) {
1396 spin_unlock(&sbi->stat_lock);
1401 f2fs_i_blocks_write(inode, 1, true);
1403 sbi->total_valid_node_count++;
1404 sbi->total_valid_block_count++;
1405 spin_unlock(&sbi->stat_lock);
1407 percpu_counter_inc(&sbi->alloc_valid_block_count);
1411 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
1412 struct inode *inode)
1414 spin_lock(&sbi->stat_lock);
1416 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
1417 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
1418 f2fs_bug_on(sbi, !inode->i_blocks);
1420 f2fs_i_blocks_write(inode, 1, false);
1421 sbi->total_valid_node_count--;
1422 sbi->total_valid_block_count--;
1424 spin_unlock(&sbi->stat_lock);
1427 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
1429 return sbi->total_valid_node_count;
1432 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
1434 percpu_counter_inc(&sbi->total_valid_inode_count);
1437 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
1439 percpu_counter_dec(&sbi->total_valid_inode_count);
1442 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
1444 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
1447 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
1448 pgoff_t index, bool for_write)
1450 #ifdef CONFIG_F2FS_FAULT_INJECTION
1451 struct page *page = find_lock_page(mapping, index);
1455 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))
1459 return grab_cache_page(mapping, index);
1460 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
1463 static inline void f2fs_copy_page(struct page *src, struct page *dst)
1465 char *src_kaddr = kmap(src);
1466 char *dst_kaddr = kmap(dst);
1468 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
1473 static inline void f2fs_put_page(struct page *page, int unlock)
1479 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
1485 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
1488 f2fs_put_page(dn->node_page, 1);
1489 if (dn->inode_page && dn->node_page != dn->inode_page)
1490 f2fs_put_page(dn->inode_page, 0);
1491 dn->node_page = NULL;
1492 dn->inode_page = NULL;
1495 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
1498 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
1501 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
1506 entry = kmem_cache_alloc(cachep, flags);
1508 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
1512 static inline struct bio *f2fs_bio_alloc(int npages)
1516 /* No failure on bio allocation */
1517 bio = bio_alloc(GFP_NOIO, npages);
1519 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
1523 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
1524 unsigned long index, void *item)
1526 while (radix_tree_insert(root, index, item))
1530 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
1532 static inline bool IS_INODE(struct page *page)
1534 struct f2fs_node *p = F2FS_NODE(page);
1535 return RAW_IS_INODE(p);
1538 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
1540 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
1543 static inline block_t datablock_addr(struct page *node_page,
1544 unsigned int offset)
1546 struct f2fs_node *raw_node;
1548 raw_node = F2FS_NODE(node_page);
1549 addr_array = blkaddr_in_node(raw_node);
1550 return le32_to_cpu(addr_array[offset]);
1553 static inline int f2fs_test_bit(unsigned int nr, char *addr)
1558 mask = 1 << (7 - (nr & 0x07));
1559 return mask & *addr;
1562 static inline void f2fs_set_bit(unsigned int nr, char *addr)
1567 mask = 1 << (7 - (nr & 0x07));
1571 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
1576 mask = 1 << (7 - (nr & 0x07));
1580 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
1586 mask = 1 << (7 - (nr & 0x07));
1592 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
1598 mask = 1 << (7 - (nr & 0x07));
1604 static inline void f2fs_change_bit(unsigned int nr, char *addr)
1609 mask = 1 << (7 - (nr & 0x07));
1613 /* used for f2fs_inode_info->flags */
1615 FI_NEW_INODE, /* indicate newly allocated inode */
1616 FI_DIRTY_INODE, /* indicate inode is dirty or not */
1617 FI_AUTO_RECOVER, /* indicate inode is recoverable */
1618 FI_DIRTY_DIR, /* indicate directory has dirty pages */
1619 FI_INC_LINK, /* need to increment i_nlink */
1620 FI_ACL_MODE, /* indicate acl mode */
1621 FI_NO_ALLOC, /* should not allocate any blocks */
1622 FI_FREE_NID, /* free allocated nide */
1623 FI_NO_EXTENT, /* not to use the extent cache */
1624 FI_INLINE_XATTR, /* used for inline xattr */
1625 FI_INLINE_DATA, /* used for inline data*/
1626 FI_INLINE_DENTRY, /* used for inline dentry */
1627 FI_APPEND_WRITE, /* inode has appended data */
1628 FI_UPDATE_WRITE, /* inode has in-place-update data */
1629 FI_NEED_IPU, /* used for ipu per file */
1630 FI_ATOMIC_FILE, /* indicate atomic file */
1631 FI_VOLATILE_FILE, /* indicate volatile file */
1632 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
1633 FI_DROP_CACHE, /* drop dirty page cache */
1634 FI_DATA_EXIST, /* indicate data exists */
1635 FI_INLINE_DOTS, /* indicate inline dot dentries */
1636 FI_DO_DEFRAG, /* indicate defragment is running */
1637 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
1640 static inline void __mark_inode_dirty_flag(struct inode *inode,
1644 case FI_INLINE_XATTR:
1645 case FI_INLINE_DATA:
1646 case FI_INLINE_DENTRY:
1650 case FI_INLINE_DOTS:
1651 f2fs_mark_inode_dirty_sync(inode, true);
1655 static inline void set_inode_flag(struct inode *inode, int flag)
1657 if (!test_bit(flag, &F2FS_I(inode)->flags))
1658 set_bit(flag, &F2FS_I(inode)->flags);
1659 __mark_inode_dirty_flag(inode, flag, true);
1662 static inline int is_inode_flag_set(struct inode *inode, int flag)
1664 return test_bit(flag, &F2FS_I(inode)->flags);
1667 static inline void clear_inode_flag(struct inode *inode, int flag)
1669 if (test_bit(flag, &F2FS_I(inode)->flags))
1670 clear_bit(flag, &F2FS_I(inode)->flags);
1671 __mark_inode_dirty_flag(inode, flag, false);
1674 static inline void set_acl_inode(struct inode *inode, umode_t mode)
1676 F2FS_I(inode)->i_acl_mode = mode;
1677 set_inode_flag(inode, FI_ACL_MODE);
1678 f2fs_mark_inode_dirty_sync(inode, false);
1681 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
1687 f2fs_mark_inode_dirty_sync(inode, true);
1690 static inline void f2fs_i_blocks_write(struct inode *inode,
1691 blkcnt_t diff, bool add)
1693 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
1694 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
1696 inode->i_blocks = add ? inode->i_blocks + diff :
1697 inode->i_blocks - diff;
1698 f2fs_mark_inode_dirty_sync(inode, true);
1699 if (clean || recover)
1700 set_inode_flag(inode, FI_AUTO_RECOVER);
1703 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
1705 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
1706 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
1708 if (i_size_read(inode) == i_size)
1711 i_size_write(inode, i_size);
1712 f2fs_mark_inode_dirty_sync(inode, true);
1713 if (clean || recover)
1714 set_inode_flag(inode, FI_AUTO_RECOVER);
1717 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
1719 F2FS_I(inode)->i_current_depth = depth;
1720 f2fs_mark_inode_dirty_sync(inode, true);
1723 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
1725 F2FS_I(inode)->i_xattr_nid = xnid;
1726 f2fs_mark_inode_dirty_sync(inode, true);
1729 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
1731 F2FS_I(inode)->i_pino = pino;
1732 f2fs_mark_inode_dirty_sync(inode, true);
1735 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
1737 struct f2fs_inode_info *fi = F2FS_I(inode);
1739 if (ri->i_inline & F2FS_INLINE_XATTR)
1740 set_bit(FI_INLINE_XATTR, &fi->flags);
1741 if (ri->i_inline & F2FS_INLINE_DATA)
1742 set_bit(FI_INLINE_DATA, &fi->flags);
1743 if (ri->i_inline & F2FS_INLINE_DENTRY)
1744 set_bit(FI_INLINE_DENTRY, &fi->flags);
1745 if (ri->i_inline & F2FS_DATA_EXIST)
1746 set_bit(FI_DATA_EXIST, &fi->flags);
1747 if (ri->i_inline & F2FS_INLINE_DOTS)
1748 set_bit(FI_INLINE_DOTS, &fi->flags);
1751 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
1755 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
1756 ri->i_inline |= F2FS_INLINE_XATTR;
1757 if (is_inode_flag_set(inode, FI_INLINE_DATA))
1758 ri->i_inline |= F2FS_INLINE_DATA;
1759 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
1760 ri->i_inline |= F2FS_INLINE_DENTRY;
1761 if (is_inode_flag_set(inode, FI_DATA_EXIST))
1762 ri->i_inline |= F2FS_DATA_EXIST;
1763 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
1764 ri->i_inline |= F2FS_INLINE_DOTS;
1767 static inline int f2fs_has_inline_xattr(struct inode *inode)
1769 return is_inode_flag_set(inode, FI_INLINE_XATTR);
1772 static inline unsigned int addrs_per_inode(struct inode *inode)
1774 if (f2fs_has_inline_xattr(inode))
1775 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
1776 return DEF_ADDRS_PER_INODE;
1779 static inline void *inline_xattr_addr(struct page *page)
1781 struct f2fs_inode *ri = F2FS_INODE(page);
1782 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
1783 F2FS_INLINE_XATTR_ADDRS]);
1786 static inline int inline_xattr_size(struct inode *inode)
1788 if (f2fs_has_inline_xattr(inode))
1789 return F2FS_INLINE_XATTR_ADDRS << 2;
1794 static inline int f2fs_has_inline_data(struct inode *inode)
1796 return is_inode_flag_set(inode, FI_INLINE_DATA);
1799 static inline void f2fs_clear_inline_inode(struct inode *inode)
1801 clear_inode_flag(inode, FI_INLINE_DATA);
1802 clear_inode_flag(inode, FI_DATA_EXIST);
1805 static inline int f2fs_exist_data(struct inode *inode)
1807 return is_inode_flag_set(inode, FI_DATA_EXIST);
1810 static inline int f2fs_has_inline_dots(struct inode *inode)
1812 return is_inode_flag_set(inode, FI_INLINE_DOTS);
1815 static inline bool f2fs_is_atomic_file(struct inode *inode)
1817 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
1820 static inline bool f2fs_is_volatile_file(struct inode *inode)
1822 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
1825 static inline bool f2fs_is_first_block_written(struct inode *inode)
1827 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
1830 static inline bool f2fs_is_drop_cache(struct inode *inode)
1832 return is_inode_flag_set(inode, FI_DROP_CACHE);
1835 static inline void *inline_data_addr(struct page *page)
1837 struct f2fs_inode *ri = F2FS_INODE(page);
1838 return (void *)&(ri->i_addr[1]);
1841 static inline int f2fs_has_inline_dentry(struct inode *inode)
1843 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
1846 static inline void f2fs_dentry_kunmap(struct inode *dir, struct page *page)
1848 if (!f2fs_has_inline_dentry(dir))
1852 static inline int is_file(struct inode *inode, int type)
1854 return F2FS_I(inode)->i_advise & type;
1857 static inline void set_file(struct inode *inode, int type)
1859 F2FS_I(inode)->i_advise |= type;
1860 f2fs_mark_inode_dirty_sync(inode, true);
1863 static inline void clear_file(struct inode *inode, int type)
1865 F2FS_I(inode)->i_advise &= ~type;
1866 f2fs_mark_inode_dirty_sync(inode, true);
1869 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
1872 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1875 spin_lock(&sbi->inode_lock[DIRTY_META]);
1876 ret = list_empty(&F2FS_I(inode)->gdirty_list);
1877 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1880 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
1881 file_keep_isize(inode) ||
1882 i_size_read(inode) & PAGE_MASK)
1884 return F2FS_I(inode)->last_disk_size == i_size_read(inode);
1887 static inline int f2fs_readonly(struct super_block *sb)
1889 return sb->s_flags & MS_RDONLY;
1892 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
1894 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
1897 static inline bool is_dot_dotdot(const struct qstr *str)
1899 if (str->len == 1 && str->name[0] == '.')
1902 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
1908 static inline bool f2fs_may_extent_tree(struct inode *inode)
1910 if (!test_opt(F2FS_I_SB(inode), EXTENT_CACHE) ||
1911 is_inode_flag_set(inode, FI_NO_EXTENT))
1914 return S_ISREG(inode->i_mode);
1917 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
1918 size_t size, gfp_t flags)
1920 #ifdef CONFIG_F2FS_FAULT_INJECTION
1921 if (time_to_inject(sbi, FAULT_KMALLOC))
1924 return kmalloc(size, flags);
1927 static inline void *f2fs_kvmalloc(size_t size, gfp_t flags)
1931 ret = kmalloc(size, flags | __GFP_NOWARN);
1933 ret = __vmalloc(size, flags, PAGE_KERNEL);
1937 static inline void *f2fs_kvzalloc(size_t size, gfp_t flags)
1941 ret = kzalloc(size, flags | __GFP_NOWARN);
1943 ret = __vmalloc(size, flags | __GFP_ZERO, PAGE_KERNEL);
1947 #define get_inode_mode(i) \
1948 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
1949 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
1951 /* get offset of first page in next direct node */
1952 #define PGOFS_OF_NEXT_DNODE(pgofs, inode) \
1953 ((pgofs < ADDRS_PER_INODE(inode)) ? ADDRS_PER_INODE(inode) : \
1954 (pgofs - ADDRS_PER_INODE(inode) + ADDRS_PER_BLOCK) / \
1955 ADDRS_PER_BLOCK * ADDRS_PER_BLOCK + ADDRS_PER_INODE(inode))
1960 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
1961 void truncate_data_blocks(struct dnode_of_data *);
1962 int truncate_blocks(struct inode *, u64, bool);
1963 int f2fs_truncate(struct inode *);
1964 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1965 int f2fs_setattr(struct dentry *, struct iattr *);
1966 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
1967 int truncate_data_blocks_range(struct dnode_of_data *, int);
1968 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
1969 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
1974 void f2fs_set_inode_flags(struct inode *);
1975 struct inode *f2fs_iget(struct super_block *, unsigned long);
1976 struct inode *f2fs_iget_retry(struct super_block *, unsigned long);
1977 int try_to_free_nats(struct f2fs_sb_info *, int);
1978 int update_inode(struct inode *, struct page *);
1979 int update_inode_page(struct inode *);
1980 int f2fs_write_inode(struct inode *, struct writeback_control *);
1981 void f2fs_evict_inode(struct inode *);
1982 void handle_failed_inode(struct inode *);
1987 struct dentry *f2fs_get_parent(struct dentry *child);
1992 void set_de_type(struct f2fs_dir_entry *, umode_t);
1993 unsigned char get_de_type(struct f2fs_dir_entry *);
1994 struct f2fs_dir_entry *find_target_dentry(struct fscrypt_name *,
1995 f2fs_hash_t, int *, struct f2fs_dentry_ptr *);
1996 int f2fs_fill_dentries(struct dir_context *, struct f2fs_dentry_ptr *,
1997 unsigned int, struct fscrypt_str *);
1998 void do_make_empty_dir(struct inode *, struct inode *,
1999 struct f2fs_dentry_ptr *);
2000 struct page *init_inode_metadata(struct inode *, struct inode *,
2001 const struct qstr *, const struct qstr *, struct page *);
2002 void update_parent_metadata(struct inode *, struct inode *, unsigned int);
2003 int room_for_filename(const void *, int, int);
2004 void f2fs_drop_nlink(struct inode *, struct inode *);
2005 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *, struct fscrypt_name *,
2007 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, const struct qstr *,
2009 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
2010 ino_t f2fs_inode_by_name(struct inode *, const struct qstr *, struct page **);
2011 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
2012 struct page *, struct inode *);
2013 int update_dent_inode(struct inode *, struct inode *, const struct qstr *);
2014 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *,
2015 const struct qstr *, f2fs_hash_t , unsigned int);
2016 int f2fs_add_regular_entry(struct inode *, const struct qstr *,
2017 const struct qstr *, struct inode *, nid_t, umode_t);
2018 int __f2fs_do_add_link(struct inode *, struct fscrypt_name*, struct inode *,
2020 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *, nid_t,
2022 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *,
2024 int f2fs_do_tmpfile(struct inode *, struct inode *);
2025 bool f2fs_empty_dir(struct inode *);
2027 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2029 return __f2fs_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2030 inode, inode->i_ino, inode->i_mode);
2036 int f2fs_inode_dirtied(struct inode *, bool);
2037 void f2fs_inode_synced(struct inode *);
2038 int f2fs_commit_super(struct f2fs_sb_info *, bool);
2039 int f2fs_sync_fs(struct super_block *, int);
2040 extern __printf(3, 4)
2041 void f2fs_msg(struct super_block *, const char *, const char *, ...);
2042 int sanity_check_ckpt(struct f2fs_sb_info *sbi);
2047 f2fs_hash_t f2fs_dentry_hash(const struct qstr *);
2052 struct dnode_of_data;
2055 bool available_free_memory(struct f2fs_sb_info *, int);
2056 int need_dentry_mark(struct f2fs_sb_info *, nid_t);
2057 bool is_checkpointed_node(struct f2fs_sb_info *, nid_t);
2058 bool need_inode_block_update(struct f2fs_sb_info *, nid_t);
2059 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
2060 pgoff_t get_next_page_offset(struct dnode_of_data *, pgoff_t);
2061 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
2062 int truncate_inode_blocks(struct inode *, pgoff_t);
2063 int truncate_xattr_node(struct inode *, struct page *);
2064 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
2065 int remove_inode_page(struct inode *);
2066 struct page *new_inode_page(struct inode *);
2067 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
2068 void ra_node_page(struct f2fs_sb_info *, nid_t);
2069 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
2070 struct page *get_node_page_ra(struct page *, int);
2071 void move_node_page(struct page *, int);
2072 int fsync_node_pages(struct f2fs_sb_info *, struct inode *,
2073 struct writeback_control *, bool);
2074 int sync_node_pages(struct f2fs_sb_info *, struct writeback_control *);
2075 void build_free_nids(struct f2fs_sb_info *, bool);
2076 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
2077 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
2078 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
2079 int try_to_free_nids(struct f2fs_sb_info *, int);
2080 void recover_inline_xattr(struct inode *, struct page *);
2081 void recover_xattr_data(struct inode *, struct page *, block_t);
2082 int recover_inode_page(struct f2fs_sb_info *, struct page *);
2083 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
2084 struct f2fs_summary_block *);
2085 void flush_nat_entries(struct f2fs_sb_info *);
2086 int build_node_manager(struct f2fs_sb_info *);
2087 void destroy_node_manager(struct f2fs_sb_info *);
2088 int __init create_node_manager_caches(void);
2089 void destroy_node_manager_caches(void);
2094 void register_inmem_page(struct inode *, struct page *);
2095 void drop_inmem_pages(struct inode *);
2096 int commit_inmem_pages(struct inode *);
2097 void f2fs_balance_fs(struct f2fs_sb_info *, bool);
2098 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
2099 int f2fs_issue_flush(struct f2fs_sb_info *);
2100 int create_flush_cmd_control(struct f2fs_sb_info *);
2101 void destroy_flush_cmd_control(struct f2fs_sb_info *, bool);
2102 void invalidate_blocks(struct f2fs_sb_info *, block_t);
2103 bool is_checkpointed_data(struct f2fs_sb_info *, block_t);
2104 void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t);
2105 void f2fs_wait_all_discard_bio(struct f2fs_sb_info *);
2106 void clear_prefree_segments(struct f2fs_sb_info *, struct cp_control *);
2107 void release_discard_addrs(struct f2fs_sb_info *);
2108 int npages_for_summary_flush(struct f2fs_sb_info *, bool);
2109 void allocate_new_segments(struct f2fs_sb_info *);
2110 int f2fs_trim_fs(struct f2fs_sb_info *, struct fstrim_range *);
2111 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
2112 void update_meta_page(struct f2fs_sb_info *, void *, block_t);
2113 void write_meta_page(struct f2fs_sb_info *, struct page *);
2114 void write_node_page(unsigned int, struct f2fs_io_info *);
2115 void write_data_page(struct dnode_of_data *, struct f2fs_io_info *);
2116 void rewrite_data_page(struct f2fs_io_info *);
2117 void __f2fs_replace_block(struct f2fs_sb_info *, struct f2fs_summary *,
2118 block_t, block_t, bool, bool);
2119 void f2fs_replace_block(struct f2fs_sb_info *, struct dnode_of_data *,
2120 block_t, block_t, unsigned char, bool, bool);
2121 void allocate_data_block(struct f2fs_sb_info *, struct page *,
2122 block_t, block_t *, struct f2fs_summary *, int);
2123 void f2fs_wait_on_page_writeback(struct page *, enum page_type, bool);
2124 void f2fs_wait_on_encrypted_page_writeback(struct f2fs_sb_info *, block_t);
2125 void write_data_summaries(struct f2fs_sb_info *, block_t);
2126 void write_node_summaries(struct f2fs_sb_info *, block_t);
2127 int lookup_journal_in_cursum(struct f2fs_journal *, int, unsigned int, int);
2128 void flush_sit_entries(struct f2fs_sb_info *, struct cp_control *);
2129 int build_segment_manager(struct f2fs_sb_info *);
2130 void destroy_segment_manager(struct f2fs_sb_info *);
2131 int __init create_segment_manager_caches(void);
2132 void destroy_segment_manager_caches(void);
2137 void f2fs_stop_checkpoint(struct f2fs_sb_info *, bool);
2138 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
2139 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
2140 struct page *get_tmp_page(struct f2fs_sb_info *, pgoff_t);
2141 bool is_valid_blkaddr(struct f2fs_sb_info *, block_t, int);
2142 int ra_meta_pages(struct f2fs_sb_info *, block_t, int, int, bool);
2143 void ra_meta_pages_cond(struct f2fs_sb_info *, pgoff_t);
2144 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
2145 void add_ino_entry(struct f2fs_sb_info *, nid_t, int type);
2146 void remove_ino_entry(struct f2fs_sb_info *, nid_t, int type);
2147 void release_ino_entry(struct f2fs_sb_info *, bool);
2148 bool exist_written_data(struct f2fs_sb_info *, nid_t, int);
2149 int f2fs_sync_inode_meta(struct f2fs_sb_info *);
2150 int acquire_orphan_inode(struct f2fs_sb_info *);
2151 void release_orphan_inode(struct f2fs_sb_info *);
2152 void add_orphan_inode(struct inode *);
2153 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
2154 int recover_orphan_inodes(struct f2fs_sb_info *);
2155 int get_valid_checkpoint(struct f2fs_sb_info *);
2156 void update_dirty_page(struct inode *, struct page *);
2157 void remove_dirty_inode(struct inode *);
2158 int sync_dirty_inodes(struct f2fs_sb_info *, enum inode_type);
2159 int write_checkpoint(struct f2fs_sb_info *, struct cp_control *);
2160 void init_ino_entry_info(struct f2fs_sb_info *);
2161 int __init create_checkpoint_caches(void);
2162 void destroy_checkpoint_caches(void);
2167 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
2168 void f2fs_submit_merged_bio_cond(struct f2fs_sb_info *, struct inode *,
2169 struct page *, nid_t, enum page_type, int);
2170 void f2fs_flush_merged_bios(struct f2fs_sb_info *);
2171 int f2fs_submit_page_bio(struct f2fs_io_info *);
2172 void f2fs_submit_page_mbio(struct f2fs_io_info *);
2173 struct block_device *f2fs_target_device(struct f2fs_sb_info *,
2174 block_t, struct bio *);
2175 int f2fs_target_device_index(struct f2fs_sb_info *, block_t);
2176 void set_data_blkaddr(struct dnode_of_data *);
2177 void f2fs_update_data_blkaddr(struct dnode_of_data *, block_t);
2178 int reserve_new_blocks(struct dnode_of_data *, blkcnt_t);
2179 int reserve_new_block(struct dnode_of_data *);
2180 int f2fs_get_block(struct dnode_of_data *, pgoff_t);
2181 int f2fs_preallocate_blocks(struct kiocb *, struct iov_iter *);
2182 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
2183 struct page *get_read_data_page(struct inode *, pgoff_t, int, bool);
2184 struct page *find_data_page(struct inode *, pgoff_t);
2185 struct page *get_lock_data_page(struct inode *, pgoff_t, bool);
2186 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
2187 int do_write_data_page(struct f2fs_io_info *);
2188 int f2fs_map_blocks(struct inode *, struct f2fs_map_blocks *, int, int);
2189 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *, u64, u64);
2190 void f2fs_set_page_dirty_nobuffers(struct page *);
2191 void f2fs_invalidate_page(struct page *, unsigned int, unsigned int);
2192 int f2fs_release_page(struct page *, gfp_t);
2193 #ifdef CONFIG_MIGRATION
2194 int f2fs_migrate_page(struct address_space *, struct page *, struct page *,
2201 int start_gc_thread(struct f2fs_sb_info *);
2202 void stop_gc_thread(struct f2fs_sb_info *);
2203 block_t start_bidx_of_node(unsigned int, struct inode *);
2204 int f2fs_gc(struct f2fs_sb_info *, bool, bool);
2205 void build_gc_manager(struct f2fs_sb_info *);
2210 int recover_fsync_data(struct f2fs_sb_info *, bool);
2211 bool space_for_roll_forward(struct f2fs_sb_info *);
2216 #ifdef CONFIG_F2FS_STAT_FS
2217 struct f2fs_stat_info {
2218 struct list_head stat_list;
2219 struct f2fs_sb_info *sbi;
2220 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
2221 int main_area_segs, main_area_sections, main_area_zones;
2222 unsigned long long hit_largest, hit_cached, hit_rbtree;
2223 unsigned long long hit_total, total_ext;
2224 int ext_tree, zombie_tree, ext_node;
2225 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_data, ndirty_imeta;
2227 unsigned int ndirty_dirs, ndirty_files, ndirty_all;
2228 int nats, dirty_nats, sits, dirty_sits, free_nids, alloc_nids;
2229 int total_count, utilization;
2230 int bg_gc, nr_wb_cp_data, nr_wb_data;
2231 int inline_xattr, inline_inode, inline_dir, orphans;
2232 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
2233 unsigned int bimodal, avg_vblocks;
2234 int util_free, util_valid, util_invalid;
2235 int rsvd_segs, overp_segs;
2236 int dirty_count, node_pages, meta_pages;
2237 int prefree_count, call_count, cp_count, bg_cp_count;
2238 int tot_segs, node_segs, data_segs, free_segs, free_secs;
2239 int bg_node_segs, bg_data_segs;
2240 int tot_blks, data_blks, node_blks;
2241 int bg_data_blks, bg_node_blks;
2242 int curseg[NR_CURSEG_TYPE];
2243 int cursec[NR_CURSEG_TYPE];
2244 int curzone[NR_CURSEG_TYPE];
2246 unsigned int segment_count[2];
2247 unsigned int block_count[2];
2248 unsigned int inplace_count;
2249 unsigned long long base_mem, cache_mem, page_mem;
2252 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
2254 return (struct f2fs_stat_info *)sbi->stat_info;
2257 #define stat_inc_cp_count(si) ((si)->cp_count++)
2258 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
2259 #define stat_inc_call_count(si) ((si)->call_count++)
2260 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
2261 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
2262 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
2263 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
2264 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
2265 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
2266 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
2267 #define stat_inc_inline_xattr(inode) \
2269 if (f2fs_has_inline_xattr(inode)) \
2270 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
2272 #define stat_dec_inline_xattr(inode) \
2274 if (f2fs_has_inline_xattr(inode)) \
2275 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
2277 #define stat_inc_inline_inode(inode) \
2279 if (f2fs_has_inline_data(inode)) \
2280 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
2282 #define stat_dec_inline_inode(inode) \
2284 if (f2fs_has_inline_data(inode)) \
2285 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
2287 #define stat_inc_inline_dir(inode) \
2289 if (f2fs_has_inline_dentry(inode)) \
2290 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
2292 #define stat_dec_inline_dir(inode) \
2294 if (f2fs_has_inline_dentry(inode)) \
2295 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
2297 #define stat_inc_seg_type(sbi, curseg) \
2298 ((sbi)->segment_count[(curseg)->alloc_type]++)
2299 #define stat_inc_block_count(sbi, curseg) \
2300 ((sbi)->block_count[(curseg)->alloc_type]++)
2301 #define stat_inc_inplace_blocks(sbi) \
2302 (atomic_inc(&(sbi)->inplace_count))
2303 #define stat_inc_seg_count(sbi, type, gc_type) \
2305 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
2307 if (type == SUM_TYPE_DATA) { \
2309 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
2312 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
2316 #define stat_inc_tot_blk_count(si, blks) \
2317 (si->tot_blks += (blks))
2319 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
2321 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
2322 stat_inc_tot_blk_count(si, blks); \
2323 si->data_blks += (blks); \
2324 si->bg_data_blks += (gc_type == BG_GC) ? (blks) : 0; \
2327 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
2329 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
2330 stat_inc_tot_blk_count(si, blks); \
2331 si->node_blks += (blks); \
2332 si->bg_node_blks += (gc_type == BG_GC) ? (blks) : 0; \
2335 int f2fs_build_stats(struct f2fs_sb_info *);
2336 void f2fs_destroy_stats(struct f2fs_sb_info *);
2337 int __init f2fs_create_root_stats(void);
2338 void f2fs_destroy_root_stats(void);
2340 #define stat_inc_cp_count(si)
2341 #define stat_inc_bg_cp_count(si)
2342 #define stat_inc_call_count(si)
2343 #define stat_inc_bggc_count(si)
2344 #define stat_inc_dirty_inode(sbi, type)
2345 #define stat_dec_dirty_inode(sbi, type)
2346 #define stat_inc_total_hit(sb)
2347 #define stat_inc_rbtree_node_hit(sb)
2348 #define stat_inc_largest_node_hit(sbi)
2349 #define stat_inc_cached_node_hit(sbi)
2350 #define stat_inc_inline_xattr(inode)
2351 #define stat_dec_inline_xattr(inode)
2352 #define stat_inc_inline_inode(inode)
2353 #define stat_dec_inline_inode(inode)
2354 #define stat_inc_inline_dir(inode)
2355 #define stat_dec_inline_dir(inode)
2356 #define stat_inc_seg_type(sbi, curseg)
2357 #define stat_inc_block_count(sbi, curseg)
2358 #define stat_inc_inplace_blocks(sbi)
2359 #define stat_inc_seg_count(sbi, type, gc_type)
2360 #define stat_inc_tot_blk_count(si, blks)
2361 #define stat_inc_data_blk_count(sbi, blks, gc_type)
2362 #define stat_inc_node_blk_count(sbi, blks, gc_type)
2364 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
2365 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
2366 static inline int __init f2fs_create_root_stats(void) { return 0; }
2367 static inline void f2fs_destroy_root_stats(void) { }
2370 extern const struct file_operations f2fs_dir_operations;
2371 extern const struct file_operations f2fs_file_operations;
2372 extern const struct inode_operations f2fs_file_inode_operations;
2373 extern const struct address_space_operations f2fs_dblock_aops;
2374 extern const struct address_space_operations f2fs_node_aops;
2375 extern const struct address_space_operations f2fs_meta_aops;
2376 extern const struct inode_operations f2fs_dir_inode_operations;
2377 extern const struct inode_operations f2fs_symlink_inode_operations;
2378 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
2379 extern const struct inode_operations f2fs_special_inode_operations;
2380 extern struct kmem_cache *inode_entry_slab;
2385 bool f2fs_may_inline_data(struct inode *);
2386 bool f2fs_may_inline_dentry(struct inode *);
2387 void read_inline_data(struct page *, struct page *);
2388 bool truncate_inline_inode(struct page *, u64);
2389 int f2fs_read_inline_data(struct inode *, struct page *);
2390 int f2fs_convert_inline_page(struct dnode_of_data *, struct page *);
2391 int f2fs_convert_inline_inode(struct inode *);
2392 int f2fs_write_inline_data(struct inode *, struct page *);
2393 bool recover_inline_data(struct inode *, struct page *);
2394 struct f2fs_dir_entry *find_in_inline_dir(struct inode *,
2395 struct fscrypt_name *, struct page **);
2396 int make_empty_inline_dir(struct inode *inode, struct inode *, struct page *);
2397 int f2fs_add_inline_entry(struct inode *, const struct qstr *,
2398 const struct qstr *, struct inode *, nid_t, umode_t);
2399 void f2fs_delete_inline_entry(struct f2fs_dir_entry *, struct page *,
2400 struct inode *, struct inode *);
2401 bool f2fs_empty_inline_dir(struct inode *);
2402 int f2fs_read_inline_dir(struct file *, struct dir_context *,
2403 struct fscrypt_str *);
2404 int f2fs_inline_data_fiemap(struct inode *,
2405 struct fiemap_extent_info *, __u64, __u64);
2410 unsigned long f2fs_shrink_count(struct shrinker *, struct shrink_control *);
2411 unsigned long f2fs_shrink_scan(struct shrinker *, struct shrink_control *);
2412 void f2fs_join_shrinker(struct f2fs_sb_info *);
2413 void f2fs_leave_shrinker(struct f2fs_sb_info *);
2418 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *, int);
2419 bool f2fs_init_extent_tree(struct inode *, struct f2fs_extent *);
2420 void f2fs_drop_extent_tree(struct inode *);
2421 unsigned int f2fs_destroy_extent_node(struct inode *);
2422 void f2fs_destroy_extent_tree(struct inode *);
2423 bool f2fs_lookup_extent_cache(struct inode *, pgoff_t, struct extent_info *);
2424 void f2fs_update_extent_cache(struct dnode_of_data *);
2425 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
2426 pgoff_t, block_t, unsigned int);
2427 void init_extent_cache_info(struct f2fs_sb_info *);
2428 int __init create_extent_cache(void);
2429 void destroy_extent_cache(void);
2434 static inline bool f2fs_encrypted_inode(struct inode *inode)
2436 return file_is_encrypt(inode);
2439 static inline void f2fs_set_encrypted_inode(struct inode *inode)
2441 #ifdef CONFIG_F2FS_FS_ENCRYPTION
2442 file_set_encrypt(inode);
2446 static inline bool f2fs_bio_encrypted(struct bio *bio)
2448 return bio->bi_private != NULL;
2451 static inline int f2fs_sb_has_crypto(struct super_block *sb)
2453 return F2FS_HAS_FEATURE(sb, F2FS_FEATURE_ENCRYPT);
2456 static inline int f2fs_sb_mounted_blkzoned(struct super_block *sb)
2458 return F2FS_HAS_FEATURE(sb, F2FS_FEATURE_BLKZONED);
2461 #ifdef CONFIG_BLK_DEV_ZONED
2462 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
2463 struct block_device *bdev, block_t blkaddr)
2465 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
2468 for (i = 0; i < sbi->s_ndevs; i++)
2469 if (FDEV(i).bdev == bdev)
2470 return FDEV(i).blkz_type[zno];
2475 static inline bool f2fs_discard_en(struct f2fs_sb_info *sbi)
2477 struct request_queue *q = bdev_get_queue(sbi->sb->s_bdev);
2479 return blk_queue_discard(q) || f2fs_sb_mounted_blkzoned(sbi->sb);
2482 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
2484 clear_opt(sbi, ADAPTIVE);
2485 clear_opt(sbi, LFS);
2488 case F2FS_MOUNT_ADAPTIVE:
2489 set_opt(sbi, ADAPTIVE);
2491 case F2FS_MOUNT_LFS:
2497 static inline bool f2fs_may_encrypt(struct inode *inode)
2499 #ifdef CONFIG_F2FS_FS_ENCRYPTION
2500 umode_t mode = inode->i_mode;
2502 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));