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>
24 #define F2FS_MOUNT_BG_GC 0x00000001
25 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
26 #define F2FS_MOUNT_DISCARD 0x00000004
27 #define F2FS_MOUNT_NOHEAP 0x00000008
28 #define F2FS_MOUNT_XATTR_USER 0x00000010
29 #define F2FS_MOUNT_POSIX_ACL 0x00000020
30 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
32 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
33 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
34 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
36 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
37 typecheck(unsigned long long, b) && \
38 ((long long)((a) - (b)) > 0))
40 typedef u32 block_t; /*
41 * should not change u32, since it is the on-disk block
42 * address format, __le32.
46 struct f2fs_mount_info {
50 static inline __u32 f2fs_crc32(void *buff, size_t len)
52 return crc32_le(F2FS_SUPER_MAGIC, buff, len);
55 static inline bool f2fs_crc_valid(__u32 blk_crc, void *buff, size_t buff_size)
57 return f2fs_crc32(buff, buff_size) == blk_crc;
61 * For checkpoint manager
68 /* for the list of orphan inodes */
69 struct orphan_inode_entry {
70 struct list_head list; /* list head */
71 nid_t ino; /* inode number */
74 /* for the list of directory inodes */
75 struct dir_inode_entry {
76 struct list_head list; /* list head */
77 struct inode *inode; /* vfs inode pointer */
80 /* for the list of fsync inodes, used only during recovery */
81 struct fsync_inode_entry {
82 struct list_head list; /* list head */
83 struct inode *inode; /* vfs inode pointer */
84 block_t blkaddr; /* block address locating the last inode */
87 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
88 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
90 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
91 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
92 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
93 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
95 static inline int update_nats_in_cursum(struct f2fs_summary_block *rs, int i)
97 int before = nats_in_cursum(rs);
98 rs->n_nats = cpu_to_le16(before + i);
102 static inline int update_sits_in_cursum(struct f2fs_summary_block *rs, int i)
104 int before = sits_in_cursum(rs);
105 rs->n_sits = cpu_to_le16(before + i);
112 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
113 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
115 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
117 * ioctl commands in 32 bit emulation
119 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
120 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
124 * For INODE and NODE manager
126 #define XATTR_NODE_OFFSET (-1) /*
127 * store xattrs to one node block per
128 * file keeping -1 as its node offset to
129 * distinguish from index node blocks.
132 ALLOC_NODE, /* allocate a new node page if needed */
133 LOOKUP_NODE, /* look up a node without readahead */
135 * look up a node with readahead called
136 * by get_datablock_ro.
140 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
142 /* for in-memory extent cache entry */
144 rwlock_t ext_lock; /* rwlock for consistency */
145 unsigned int fofs; /* start offset in a file */
146 u32 blk_addr; /* start block address of the extent */
147 unsigned int len; /* length of the extent */
151 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
153 #define FADVISE_COLD_BIT 0x01
154 #define FADVISE_CP_BIT 0x02
156 struct f2fs_inode_info {
157 struct inode vfs_inode; /* serve a vfs inode */
158 unsigned long i_flags; /* keep an inode flags for ioctl */
159 unsigned char i_advise; /* use to give file attribute hints */
160 unsigned int i_current_depth; /* use only in directory structure */
161 unsigned int i_pino; /* parent inode number */
162 umode_t i_acl_mode; /* keep file acl mode temporarily */
164 /* Use below internally in f2fs*/
165 unsigned long flags; /* use to pass per-file flags */
166 atomic_t dirty_dents; /* # of dirty dentry pages */
167 f2fs_hash_t chash; /* hash value of given file name */
168 unsigned int clevel; /* maximum level of given file name */
169 nid_t i_xattr_nid; /* node id that contains xattrs */
170 struct extent_info ext; /* in-memory extent cache entry */
173 static inline void get_extent_info(struct extent_info *ext,
174 struct f2fs_extent i_ext)
176 write_lock(&ext->ext_lock);
177 ext->fofs = le32_to_cpu(i_ext.fofs);
178 ext->blk_addr = le32_to_cpu(i_ext.blk_addr);
179 ext->len = le32_to_cpu(i_ext.len);
180 write_unlock(&ext->ext_lock);
183 static inline void set_raw_extent(struct extent_info *ext,
184 struct f2fs_extent *i_ext)
186 read_lock(&ext->ext_lock);
187 i_ext->fofs = cpu_to_le32(ext->fofs);
188 i_ext->blk_addr = cpu_to_le32(ext->blk_addr);
189 i_ext->len = cpu_to_le32(ext->len);
190 read_unlock(&ext->ext_lock);
193 struct f2fs_nm_info {
194 block_t nat_blkaddr; /* base disk address of NAT */
195 nid_t max_nid; /* maximum possible node ids */
196 nid_t next_scan_nid; /* the next nid to be scanned */
198 /* NAT cache management */
199 struct radix_tree_root nat_root;/* root of the nat entry cache */
200 rwlock_t nat_tree_lock; /* protect nat_tree_lock */
201 unsigned int nat_cnt; /* the # of cached nat entries */
202 struct list_head nat_entries; /* cached nat entry list (clean) */
203 struct list_head dirty_nat_entries; /* cached nat entry list (dirty) */
205 /* free node ids management */
206 struct list_head free_nid_list; /* a list for free nids */
207 spinlock_t free_nid_list_lock; /* protect free nid list */
208 unsigned int fcnt; /* the number of free node id */
209 struct mutex build_lock; /* lock for build free nids */
212 char *nat_bitmap; /* NAT bitmap pointer */
213 int bitmap_size; /* bitmap size */
217 * this structure is used as one of function parameters.
218 * all the information are dedicated to a given direct node block determined
219 * by the data offset in a file.
221 struct dnode_of_data {
222 struct inode *inode; /* vfs inode pointer */
223 struct page *inode_page; /* its inode page, NULL is possible */
224 struct page *node_page; /* cached direct node page */
225 nid_t nid; /* node id of the direct node block */
226 unsigned int ofs_in_node; /* data offset in the node page */
227 bool inode_page_locked; /* inode page is locked or not */
228 block_t data_blkaddr; /* block address of the node block */
231 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
232 struct page *ipage, struct page *npage, nid_t nid)
234 memset(dn, 0, sizeof(*dn));
236 dn->inode_page = ipage;
237 dn->node_page = npage;
244 * By default, there are 6 active log areas across the whole main area.
245 * When considering hot and cold data separation to reduce cleaning overhead,
246 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
248 * In the current design, you should not change the numbers intentionally.
249 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
250 * logs individually according to the underlying devices. (default: 6)
251 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
252 * data and 8 for node logs.
254 #define NR_CURSEG_DATA_TYPE (3)
255 #define NR_CURSEG_NODE_TYPE (3)
256 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
259 CURSEG_HOT_DATA = 0, /* directory entry blocks */
260 CURSEG_WARM_DATA, /* data blocks */
261 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
262 CURSEG_HOT_NODE, /* direct node blocks of directory files */
263 CURSEG_WARM_NODE, /* direct node blocks of normal files */
264 CURSEG_COLD_NODE, /* indirect node blocks */
268 struct f2fs_sm_info {
269 struct sit_info *sit_info; /* whole segment information */
270 struct free_segmap_info *free_info; /* free segment information */
271 struct dirty_seglist_info *dirty_info; /* dirty segment information */
272 struct curseg_info *curseg_array; /* active segment information */
274 struct list_head wblist_head; /* list of under-writeback pages */
275 spinlock_t wblist_lock; /* lock for checkpoint */
277 block_t seg0_blkaddr; /* block address of 0'th segment */
278 block_t main_blkaddr; /* start block address of main area */
279 block_t ssa_blkaddr; /* start block address of SSA area */
281 unsigned int segment_count; /* total # of segments */
282 unsigned int main_segments; /* # of segments in main area */
283 unsigned int reserved_segments; /* # of reserved segments */
284 unsigned int ovp_segments; /* # of overprovision segments */
288 * For directory operation
290 #define NODE_DIR1_BLOCK (ADDRS_PER_INODE + 1)
291 #define NODE_DIR2_BLOCK (ADDRS_PER_INODE + 2)
292 #define NODE_IND1_BLOCK (ADDRS_PER_INODE + 3)
293 #define NODE_IND2_BLOCK (ADDRS_PER_INODE + 4)
294 #define NODE_DIND_BLOCK (ADDRS_PER_INODE + 5)
300 * COUNT_TYPE for monitoring
302 * f2fs monitors the number of several block types such as on-writeback,
303 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
314 * Uses as sbi->fs_lock[NR_GLOBAL_LOCKS].
315 * The checkpoint procedure blocks all the locks in this fs_lock array.
316 * Some FS operations grab free locks, and if there is no free lock,
317 * then wait to grab a lock in a round-robin manner.
319 #define NR_GLOBAL_LOCKS 8
322 * The below are the page types of bios used in submti_bio().
323 * The available types are:
324 * DATA User data pages. It operates as async mode.
325 * NODE Node pages. It operates as async mode.
326 * META FS metadata pages such as SIT, NAT, CP.
327 * NR_PAGE_TYPE The number of page types.
328 * META_FLUSH Make sure the previous pages are written
329 * with waiting the bio's completion
330 * ... Only can be used with META.
340 struct f2fs_sb_info {
341 struct super_block *sb; /* pointer to VFS super block */
342 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
343 struct f2fs_super_block *raw_super; /* raw super block pointer */
344 int s_dirty; /* dirty flag for checkpoint */
346 /* for node-related operations */
347 struct f2fs_nm_info *nm_info; /* node manager */
348 struct inode *node_inode; /* cache node blocks */
350 /* for segment-related operations */
351 struct f2fs_sm_info *sm_info; /* segment manager */
352 struct bio *bio[NR_PAGE_TYPE]; /* bios to merge */
353 sector_t last_block_in_bio[NR_PAGE_TYPE]; /* last block number */
354 struct rw_semaphore bio_sem; /* IO semaphore */
357 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
358 struct inode *meta_inode; /* cache meta blocks */
359 struct mutex cp_mutex; /* checkpoint procedure lock */
360 struct mutex fs_lock[NR_GLOBAL_LOCKS]; /* blocking FS operations */
361 struct mutex node_write; /* locking node writes */
362 struct mutex writepages; /* mutex for writepages() */
363 unsigned char next_lock_num; /* round-robin global locks */
364 int por_doing; /* recovery is doing or not */
365 int on_build_free_nids; /* build_free_nids is doing */
367 /* for orphan inode management */
368 struct list_head orphan_inode_list; /* orphan inode list */
369 struct mutex orphan_inode_mutex; /* for orphan inode list */
370 unsigned int n_orphans; /* # of orphan inodes */
372 /* for directory inode management */
373 struct list_head dir_inode_list; /* dir inode list */
374 spinlock_t dir_inode_lock; /* for dir inode list lock */
376 /* basic file system units */
377 unsigned int log_sectors_per_block; /* log2 sectors per block */
378 unsigned int log_blocksize; /* log2 block size */
379 unsigned int blocksize; /* block size */
380 unsigned int root_ino_num; /* root inode number*/
381 unsigned int node_ino_num; /* node inode number*/
382 unsigned int meta_ino_num; /* meta inode number*/
383 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
384 unsigned int blocks_per_seg; /* blocks per segment */
385 unsigned int segs_per_sec; /* segments per section */
386 unsigned int secs_per_zone; /* sections per zone */
387 unsigned int total_sections; /* total section count */
388 unsigned int total_node_count; /* total node block count */
389 unsigned int total_valid_node_count; /* valid node block count */
390 unsigned int total_valid_inode_count; /* valid inode count */
391 int active_logs; /* # of active logs */
393 block_t user_block_count; /* # of user blocks */
394 block_t total_valid_block_count; /* # of valid blocks */
395 block_t alloc_valid_block_count; /* # of allocated blocks */
396 block_t last_valid_block_count; /* for recovery */
397 u32 s_next_generation; /* for NFS support */
398 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
400 struct f2fs_mount_info mount_opt; /* mount options */
402 /* for cleaning operations */
403 struct mutex gc_mutex; /* mutex for GC */
404 struct f2fs_gc_kthread *gc_thread; /* GC thread */
405 unsigned int cur_victim_sec; /* current victim section num */
408 * for stat information.
409 * one is for the LFS mode, and the other is for the SSR mode.
411 #ifdef CONFIG_F2FS_STAT_FS
412 struct f2fs_stat_info *stat_info; /* FS status information */
413 unsigned int segment_count[2]; /* # of allocated segments */
414 unsigned int block_count[2]; /* # of allocated blocks */
415 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
416 int bg_gc; /* background gc calls */
417 unsigned int n_dirty_dirs; /* # of dir inodes */
419 unsigned int last_victim[2]; /* last victim segment # */
420 spinlock_t stat_lock; /* lock for stat operations */
426 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
428 return container_of(inode, struct f2fs_inode_info, vfs_inode);
431 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
433 return sb->s_fs_info;
436 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
438 return (struct f2fs_super_block *)(sbi->raw_super);
441 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
443 return (struct f2fs_checkpoint *)(sbi->ckpt);
446 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
448 return (struct f2fs_nm_info *)(sbi->nm_info);
451 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
453 return (struct f2fs_sm_info *)(sbi->sm_info);
456 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
458 return (struct sit_info *)(SM_I(sbi)->sit_info);
461 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
463 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
466 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
468 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
471 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
476 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
481 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
483 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
484 return ckpt_flags & f;
487 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
489 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
491 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
494 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
496 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
498 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
501 static inline void mutex_lock_all(struct f2fs_sb_info *sbi)
505 for (i = 0; i < NR_GLOBAL_LOCKS; i++) {
507 * This is the only time we take multiple fs_lock[]
508 * instances; the order is immaterial since we
509 * always hold cp_mutex, which serializes multiple
512 mutex_lock_nest_lock(&sbi->fs_lock[i], &sbi->cp_mutex);
516 static inline void mutex_unlock_all(struct f2fs_sb_info *sbi)
519 for (; i < NR_GLOBAL_LOCKS; i++)
520 mutex_unlock(&sbi->fs_lock[i]);
523 static inline int mutex_lock_op(struct f2fs_sb_info *sbi)
525 unsigned char next_lock = sbi->next_lock_num % NR_GLOBAL_LOCKS;
528 for (; i < NR_GLOBAL_LOCKS; i++)
529 if (mutex_trylock(&sbi->fs_lock[i]))
532 mutex_lock(&sbi->fs_lock[next_lock]);
533 sbi->next_lock_num++;
537 static inline void mutex_unlock_op(struct f2fs_sb_info *sbi, int ilock)
541 BUG_ON(ilock >= NR_GLOBAL_LOCKS);
542 mutex_unlock(&sbi->fs_lock[ilock]);
546 * Check whether the given nid is within node id range.
548 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
550 WARN_ON((nid >= NM_I(sbi)->max_nid));
551 if (nid >= NM_I(sbi)->max_nid)
556 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
559 * Check whether the inode has blocks or not
561 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
563 if (F2FS_I(inode)->i_xattr_nid)
564 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1);
566 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS);
569 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
570 struct inode *inode, blkcnt_t count)
572 block_t valid_block_count;
574 spin_lock(&sbi->stat_lock);
576 sbi->total_valid_block_count + (block_t)count;
577 if (valid_block_count > sbi->user_block_count) {
578 spin_unlock(&sbi->stat_lock);
581 inode->i_blocks += count;
582 sbi->total_valid_block_count = valid_block_count;
583 sbi->alloc_valid_block_count += (block_t)count;
584 spin_unlock(&sbi->stat_lock);
588 static inline int dec_valid_block_count(struct f2fs_sb_info *sbi,
592 spin_lock(&sbi->stat_lock);
593 BUG_ON(sbi->total_valid_block_count < (block_t) count);
594 BUG_ON(inode->i_blocks < count);
595 inode->i_blocks -= count;
596 sbi->total_valid_block_count -= (block_t)count;
597 spin_unlock(&sbi->stat_lock);
601 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
603 atomic_inc(&sbi->nr_pages[count_type]);
604 F2FS_SET_SB_DIRT(sbi);
607 static inline void inode_inc_dirty_dents(struct inode *inode)
609 atomic_inc(&F2FS_I(inode)->dirty_dents);
612 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
614 atomic_dec(&sbi->nr_pages[count_type]);
617 static inline void inode_dec_dirty_dents(struct inode *inode)
619 atomic_dec(&F2FS_I(inode)->dirty_dents);
622 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
624 return atomic_read(&sbi->nr_pages[count_type]);
627 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
629 unsigned int pages_per_sec = sbi->segs_per_sec *
630 (1 << sbi->log_blocks_per_seg);
631 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
632 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
635 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
638 spin_lock(&sbi->stat_lock);
639 ret = sbi->total_valid_block_count;
640 spin_unlock(&sbi->stat_lock);
644 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
646 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
648 /* return NAT or SIT bitmap */
649 if (flag == NAT_BITMAP)
650 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
651 else if (flag == SIT_BITMAP)
652 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
657 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
659 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
660 int offset = (flag == NAT_BITMAP) ?
661 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
662 return &ckpt->sit_nat_version_bitmap + offset;
665 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
668 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
669 unsigned long long ckpt_version = le64_to_cpu(ckpt->checkpoint_ver);
671 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
674 * odd numbered checkpoint should at cp segment 0
675 * and even segent must be at cp segment 1
677 if (!(ckpt_version & 1))
678 start_addr += sbi->blocks_per_seg;
683 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
685 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
688 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
692 block_t valid_block_count;
693 unsigned int valid_node_count;
695 spin_lock(&sbi->stat_lock);
697 valid_block_count = sbi->total_valid_block_count + (block_t)count;
698 sbi->alloc_valid_block_count += (block_t)count;
699 valid_node_count = sbi->total_valid_node_count + count;
701 if (valid_block_count > sbi->user_block_count) {
702 spin_unlock(&sbi->stat_lock);
706 if (valid_node_count > sbi->total_node_count) {
707 spin_unlock(&sbi->stat_lock);
712 inode->i_blocks += count;
713 sbi->total_valid_node_count = valid_node_count;
714 sbi->total_valid_block_count = valid_block_count;
715 spin_unlock(&sbi->stat_lock);
720 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
724 spin_lock(&sbi->stat_lock);
726 BUG_ON(sbi->total_valid_block_count < count);
727 BUG_ON(sbi->total_valid_node_count < count);
728 BUG_ON(inode->i_blocks < count);
730 inode->i_blocks -= count;
731 sbi->total_valid_node_count -= count;
732 sbi->total_valid_block_count -= (block_t)count;
734 spin_unlock(&sbi->stat_lock);
737 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
740 spin_lock(&sbi->stat_lock);
741 ret = sbi->total_valid_node_count;
742 spin_unlock(&sbi->stat_lock);
746 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
748 spin_lock(&sbi->stat_lock);
749 BUG_ON(sbi->total_valid_inode_count == sbi->total_node_count);
750 sbi->total_valid_inode_count++;
751 spin_unlock(&sbi->stat_lock);
754 static inline int dec_valid_inode_count(struct f2fs_sb_info *sbi)
756 spin_lock(&sbi->stat_lock);
757 BUG_ON(!sbi->total_valid_inode_count);
758 sbi->total_valid_inode_count--;
759 spin_unlock(&sbi->stat_lock);
763 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
766 spin_lock(&sbi->stat_lock);
767 ret = sbi->total_valid_inode_count;
768 spin_unlock(&sbi->stat_lock);
772 static inline void f2fs_put_page(struct page *page, int unlock)
774 if (!page || IS_ERR(page))
778 BUG_ON(!PageLocked(page));
781 page_cache_release(page);
784 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
787 f2fs_put_page(dn->node_page, 1);
788 if (dn->inode_page && dn->node_page != dn->inode_page)
789 f2fs_put_page(dn->inode_page, 0);
790 dn->node_page = NULL;
791 dn->inode_page = NULL;
794 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
795 size_t size, void (*ctor)(void *))
797 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, ctor);
800 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
802 static inline bool IS_INODE(struct page *page)
804 struct f2fs_node *p = (struct f2fs_node *)page_address(page);
805 return RAW_IS_INODE(p);
808 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
810 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
813 static inline block_t datablock_addr(struct page *node_page,
816 struct f2fs_node *raw_node;
818 raw_node = (struct f2fs_node *)page_address(node_page);
819 addr_array = blkaddr_in_node(raw_node);
820 return le32_to_cpu(addr_array[offset]);
823 static inline int f2fs_test_bit(unsigned int nr, char *addr)
828 mask = 1 << (7 - (nr & 0x07));
832 static inline int f2fs_set_bit(unsigned int nr, char *addr)
838 mask = 1 << (7 - (nr & 0x07));
844 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
850 mask = 1 << (7 - (nr & 0x07));
856 /* used for f2fs_inode_info->flags */
858 FI_NEW_INODE, /* indicate newly allocated inode */
859 FI_DIRTY_INODE, /* indicate inode is dirty or not */
860 FI_INC_LINK, /* need to increment i_nlink */
861 FI_ACL_MODE, /* indicate acl mode */
862 FI_NO_ALLOC, /* should not allocate any blocks */
863 FI_UPDATE_DIR, /* should update inode block for consistency */
864 FI_DELAY_IPUT, /* used for the recovery */
867 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
869 set_bit(flag, &fi->flags);
872 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
874 return test_bit(flag, &fi->flags);
877 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
879 clear_bit(flag, &fi->flags);
882 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
884 fi->i_acl_mode = mode;
885 set_inode_flag(fi, FI_ACL_MODE);
888 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
890 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
891 clear_inode_flag(fi, FI_ACL_MODE);
897 static inline int f2fs_readonly(struct super_block *sb)
899 return sb->s_flags & MS_RDONLY;
905 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
906 void truncate_data_blocks(struct dnode_of_data *);
907 void f2fs_truncate(struct inode *);
908 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
909 int f2fs_setattr(struct dentry *, struct iattr *);
910 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
911 int truncate_data_blocks_range(struct dnode_of_data *, int);
912 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
913 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
918 void f2fs_set_inode_flags(struct inode *);
919 struct inode *f2fs_iget(struct super_block *, unsigned long);
920 void update_inode(struct inode *, struct page *);
921 int update_inode_page(struct inode *);
922 int f2fs_write_inode(struct inode *, struct writeback_control *);
923 void f2fs_evict_inode(struct inode *);
928 struct dentry *f2fs_get_parent(struct dentry *child);
933 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
935 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
936 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
937 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
938 struct page *, struct inode *);
939 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
940 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
941 int f2fs_make_empty(struct inode *, struct inode *);
942 bool f2fs_empty_dir(struct inode *);
944 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
946 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
953 int f2fs_sync_fs(struct super_block *, int);
954 extern __printf(3, 4)
955 void f2fs_msg(struct super_block *, const char *, const char *, ...);
960 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
965 struct dnode_of_data;
968 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
969 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
970 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
971 int truncate_inode_blocks(struct inode *, pgoff_t);
972 int remove_inode_page(struct inode *);
973 struct page *new_inode_page(struct inode *, const struct qstr *);
974 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
975 void ra_node_page(struct f2fs_sb_info *, nid_t);
976 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
977 struct page *get_node_page_ra(struct page *, int);
978 void sync_inode_page(struct dnode_of_data *);
979 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
980 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
981 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
982 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
983 void recover_node_page(struct f2fs_sb_info *, struct page *,
984 struct f2fs_summary *, struct node_info *, block_t);
985 int recover_inode_page(struct f2fs_sb_info *, struct page *);
986 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
987 struct f2fs_summary_block *);
988 void flush_nat_entries(struct f2fs_sb_info *);
989 int build_node_manager(struct f2fs_sb_info *);
990 void destroy_node_manager(struct f2fs_sb_info *);
991 int __init create_node_manager_caches(void);
992 void destroy_node_manager_caches(void);
997 void f2fs_balance_fs(struct f2fs_sb_info *);
998 void invalidate_blocks(struct f2fs_sb_info *, block_t);
999 void locate_dirty_segment(struct f2fs_sb_info *, unsigned int);
1000 void clear_prefree_segments(struct f2fs_sb_info *);
1001 int npages_for_summary_flush(struct f2fs_sb_info *);
1002 void allocate_new_segments(struct f2fs_sb_info *);
1003 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1004 struct bio *f2fs_bio_alloc(struct block_device *, int);
1005 void f2fs_submit_bio(struct f2fs_sb_info *, enum page_type, bool sync);
1006 void write_meta_page(struct f2fs_sb_info *, struct page *);
1007 void write_node_page(struct f2fs_sb_info *, struct page *, unsigned int,
1008 block_t, block_t *);
1009 void write_data_page(struct inode *, struct page *, struct dnode_of_data*,
1010 block_t, block_t *);
1011 void rewrite_data_page(struct f2fs_sb_info *, struct page *, block_t);
1012 void recover_data_page(struct f2fs_sb_info *, struct page *,
1013 struct f2fs_summary *, block_t, block_t);
1014 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1015 struct f2fs_summary *, block_t, block_t);
1016 void write_data_summaries(struct f2fs_sb_info *, block_t);
1017 void write_node_summaries(struct f2fs_sb_info *, block_t);
1018 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1019 int, unsigned int, int);
1020 void flush_sit_entries(struct f2fs_sb_info *);
1021 int build_segment_manager(struct f2fs_sb_info *);
1022 void destroy_segment_manager(struct f2fs_sb_info *);
1027 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1028 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1029 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1030 int check_orphan_space(struct f2fs_sb_info *);
1031 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1032 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1033 int recover_orphan_inodes(struct f2fs_sb_info *);
1034 int get_valid_checkpoint(struct f2fs_sb_info *);
1035 void set_dirty_dir_page(struct inode *, struct page *);
1036 void add_dirty_dir_inode(struct inode *);
1037 void remove_dirty_dir_inode(struct inode *);
1038 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1039 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1040 void write_checkpoint(struct f2fs_sb_info *, bool);
1041 void init_orphan_info(struct f2fs_sb_info *);
1042 int __init create_checkpoint_caches(void);
1043 void destroy_checkpoint_caches(void);
1048 int reserve_new_block(struct dnode_of_data *);
1049 void update_extent_cache(block_t, struct dnode_of_data *);
1050 struct page *find_data_page(struct inode *, pgoff_t, bool);
1051 struct page *get_lock_data_page(struct inode *, pgoff_t);
1052 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1053 int f2fs_readpage(struct f2fs_sb_info *, struct page *, block_t, int);
1054 int do_write_data_page(struct page *);
1059 int start_gc_thread(struct f2fs_sb_info *);
1060 void stop_gc_thread(struct f2fs_sb_info *);
1061 block_t start_bidx_of_node(unsigned int);
1062 int f2fs_gc(struct f2fs_sb_info *);
1063 void build_gc_manager(struct f2fs_sb_info *);
1064 int __init create_gc_caches(void);
1065 void destroy_gc_caches(void);
1070 int recover_fsync_data(struct f2fs_sb_info *);
1071 bool space_for_roll_forward(struct f2fs_sb_info *);
1076 #ifdef CONFIG_F2FS_STAT_FS
1077 struct f2fs_stat_info {
1078 struct list_head stat_list;
1079 struct f2fs_sb_info *sbi;
1080 struct mutex stat_lock;
1081 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1082 int main_area_segs, main_area_sections, main_area_zones;
1083 int hit_ext, total_ext;
1084 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1085 int nats, sits, fnids;
1086 int total_count, utilization;
1088 unsigned int valid_count, valid_node_count, valid_inode_count;
1089 unsigned int bimodal, avg_vblocks;
1090 int util_free, util_valid, util_invalid;
1091 int rsvd_segs, overp_segs;
1092 int dirty_count, node_pages, meta_pages;
1093 int prefree_count, call_count;
1094 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1095 int tot_blks, data_blks, node_blks;
1096 int curseg[NR_CURSEG_TYPE];
1097 int cursec[NR_CURSEG_TYPE];
1098 int curzone[NR_CURSEG_TYPE];
1100 unsigned int segment_count[2];
1101 unsigned int block_count[2];
1102 unsigned base_mem, cache_mem;
1105 #define stat_inc_call_count(si) ((si)->call_count++)
1107 #define stat_inc_seg_count(sbi, type) \
1109 struct f2fs_stat_info *si = sbi->stat_info; \
1111 if (type == SUM_TYPE_DATA) \
1117 #define stat_inc_tot_blk_count(si, blks) \
1118 (si->tot_blks += (blks))
1120 #define stat_inc_data_blk_count(sbi, blks) \
1122 struct f2fs_stat_info *si = sbi->stat_info; \
1123 stat_inc_tot_blk_count(si, blks); \
1124 si->data_blks += (blks); \
1127 #define stat_inc_node_blk_count(sbi, blks) \
1129 struct f2fs_stat_info *si = sbi->stat_info; \
1130 stat_inc_tot_blk_count(si, blks); \
1131 si->node_blks += (blks); \
1134 int f2fs_build_stats(struct f2fs_sb_info *);
1135 void f2fs_destroy_stats(struct f2fs_sb_info *);
1136 void __init f2fs_create_root_stats(void);
1137 void f2fs_destroy_root_stats(void);
1139 #define stat_inc_call_count(si)
1140 #define stat_inc_seg_count(si, type)
1141 #define stat_inc_tot_blk_count(si, blks)
1142 #define stat_inc_data_blk_count(si, blks)
1143 #define stat_inc_node_blk_count(sbi, blks)
1145 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1146 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1147 static inline void __init f2fs_create_root_stats(void) { }
1148 static inline void f2fs_destroy_root_stats(void) { }
1151 extern const struct file_operations f2fs_dir_operations;
1152 extern const struct file_operations f2fs_file_operations;
1153 extern const struct inode_operations f2fs_file_inode_operations;
1154 extern const struct address_space_operations f2fs_dblock_aops;
1155 extern const struct address_space_operations f2fs_node_aops;
1156 extern const struct address_space_operations f2fs_meta_aops;
1157 extern const struct inode_operations f2fs_dir_inode_operations;
1158 extern const struct inode_operations f2fs_symlink_inode_operations;
1159 extern const struct inode_operations f2fs_special_inode_operations;