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>
23 #ifdef CONFIG_F2FS_CHECK_FS
24 #define f2fs_bug_on(condition) BUG_ON(condition)
25 #define f2fs_down_write(x, y) down_write_nest_lock(x, y)
27 #define f2fs_bug_on(condition)
28 #define f2fs_down_write(x, y) down_write(x)
34 #define F2FS_MOUNT_BG_GC 0x00000001
35 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
36 #define F2FS_MOUNT_DISCARD 0x00000004
37 #define F2FS_MOUNT_NOHEAP 0x00000008
38 #define F2FS_MOUNT_XATTR_USER 0x00000010
39 #define F2FS_MOUNT_POSIX_ACL 0x00000020
40 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
41 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
42 #define F2FS_MOUNT_INLINE_DATA 0x00000100
44 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
45 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
46 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
48 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
49 typecheck(unsigned long long, b) && \
50 ((long long)((a) - (b)) > 0))
52 typedef u32 block_t; /*
53 * should not change u32, since it is the on-disk block
54 * address format, __le32.
58 struct f2fs_mount_info {
62 #define CRCPOLY_LE 0xedb88320
64 static inline __u32 f2fs_crc32(void *buf, size_t len)
66 unsigned char *p = (unsigned char *)buf;
67 __u32 crc = F2FS_SUPER_MAGIC;
72 for (i = 0; i < 8; i++)
73 crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);
78 static inline bool f2fs_crc_valid(__u32 blk_crc, void *buf, size_t buf_size)
80 return f2fs_crc32(buf, buf_size) == blk_crc;
84 * For checkpoint manager
92 * For CP/NAT/SIT/SSA readahead
101 /* for the list of orphan inodes */
102 struct orphan_inode_entry {
103 struct list_head list; /* list head */
104 nid_t ino; /* inode number */
107 /* for the list of directory inodes */
108 struct dir_inode_entry {
109 struct list_head list; /* list head */
110 struct inode *inode; /* vfs inode pointer */
113 /* for the list of blockaddresses to be discarded */
114 struct discard_entry {
115 struct list_head list; /* list head */
116 block_t blkaddr; /* block address to be discarded */
117 int len; /* # of consecutive blocks of the discard */
120 /* for the list of fsync inodes, used only during recovery */
121 struct fsync_inode_entry {
122 struct list_head list; /* list head */
123 struct inode *inode; /* vfs inode pointer */
124 block_t blkaddr; /* block address locating the last inode */
127 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
128 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
130 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
131 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
132 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
133 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
135 static inline int update_nats_in_cursum(struct f2fs_summary_block *rs, int i)
137 int before = nats_in_cursum(rs);
138 rs->n_nats = cpu_to_le16(before + i);
142 static inline int update_sits_in_cursum(struct f2fs_summary_block *rs, int i)
144 int before = sits_in_cursum(rs);
145 rs->n_sits = cpu_to_le16(before + i);
152 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
153 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
155 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
157 * ioctl commands in 32 bit emulation
159 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
160 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
164 * For INODE and NODE manager
167 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
168 * as its node offset to distinguish from index node blocks.
169 * But some bits are used to mark the node block.
171 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
174 ALLOC_NODE, /* allocate a new node page if needed */
175 LOOKUP_NODE, /* look up a node without readahead */
177 * look up a node with readahead called
182 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
184 /* for in-memory extent cache entry */
185 #define F2FS_MIN_EXTENT_LEN 16 /* minimum extent length */
188 rwlock_t ext_lock; /* rwlock for consistency */
189 unsigned int fofs; /* start offset in a file */
190 u32 blk_addr; /* start block address of the extent */
191 unsigned int len; /* length of the extent */
195 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
197 #define FADVISE_COLD_BIT 0x01
198 #define FADVISE_LOST_PINO_BIT 0x02
200 #define DEF_DIR_LEVEL 0
202 struct f2fs_inode_info {
203 struct inode vfs_inode; /* serve a vfs inode */
204 unsigned long i_flags; /* keep an inode flags for ioctl */
205 unsigned char i_advise; /* use to give file attribute hints */
206 unsigned char i_dir_level; /* use for dentry level for large dir */
207 unsigned int i_current_depth; /* use only in directory structure */
208 unsigned int i_pino; /* parent inode number */
209 umode_t i_acl_mode; /* keep file acl mode temporarily */
211 /* Use below internally in f2fs*/
212 unsigned long flags; /* use to pass per-file flags */
213 struct rw_semaphore i_sem; /* protect fi info */
214 atomic_t dirty_dents; /* # of dirty dentry pages */
215 f2fs_hash_t chash; /* hash value of given file name */
216 unsigned int clevel; /* maximum level of given file name */
217 nid_t i_xattr_nid; /* node id that contains xattrs */
218 unsigned long long xattr_ver; /* cp version of xattr modification */
219 struct extent_info ext; /* in-memory extent cache entry */
222 static inline void get_extent_info(struct extent_info *ext,
223 struct f2fs_extent i_ext)
225 write_lock(&ext->ext_lock);
226 ext->fofs = le32_to_cpu(i_ext.fofs);
227 ext->blk_addr = le32_to_cpu(i_ext.blk_addr);
228 ext->len = le32_to_cpu(i_ext.len);
229 write_unlock(&ext->ext_lock);
232 static inline void set_raw_extent(struct extent_info *ext,
233 struct f2fs_extent *i_ext)
235 read_lock(&ext->ext_lock);
236 i_ext->fofs = cpu_to_le32(ext->fofs);
237 i_ext->blk_addr = cpu_to_le32(ext->blk_addr);
238 i_ext->len = cpu_to_le32(ext->len);
239 read_unlock(&ext->ext_lock);
242 struct f2fs_nm_info {
243 block_t nat_blkaddr; /* base disk address of NAT */
244 nid_t max_nid; /* maximum possible node ids */
245 nid_t next_scan_nid; /* the next nid to be scanned */
246 unsigned int ram_thresh; /* control the memory footprint */
248 /* NAT cache management */
249 struct radix_tree_root nat_root;/* root of the nat entry cache */
250 rwlock_t nat_tree_lock; /* protect nat_tree_lock */
251 unsigned int nat_cnt; /* the # of cached nat entries */
252 struct list_head nat_entries; /* cached nat entry list (clean) */
253 struct list_head dirty_nat_entries; /* cached nat entry list (dirty) */
255 /* free node ids management */
256 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
257 struct list_head free_nid_list; /* a list for free nids */
258 spinlock_t free_nid_list_lock; /* protect free nid list */
259 unsigned int fcnt; /* the number of free node id */
260 struct mutex build_lock; /* lock for build free nids */
263 char *nat_bitmap; /* NAT bitmap pointer */
264 int bitmap_size; /* bitmap size */
268 * this structure is used as one of function parameters.
269 * all the information are dedicated to a given direct node block determined
270 * by the data offset in a file.
272 struct dnode_of_data {
273 struct inode *inode; /* vfs inode pointer */
274 struct page *inode_page; /* its inode page, NULL is possible */
275 struct page *node_page; /* cached direct node page */
276 nid_t nid; /* node id of the direct node block */
277 unsigned int ofs_in_node; /* data offset in the node page */
278 bool inode_page_locked; /* inode page is locked or not */
279 block_t data_blkaddr; /* block address of the node block */
282 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
283 struct page *ipage, struct page *npage, nid_t nid)
285 memset(dn, 0, sizeof(*dn));
287 dn->inode_page = ipage;
288 dn->node_page = npage;
295 * By default, there are 6 active log areas across the whole main area.
296 * When considering hot and cold data separation to reduce cleaning overhead,
297 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
299 * In the current design, you should not change the numbers intentionally.
300 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
301 * logs individually according to the underlying devices. (default: 6)
302 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
303 * data and 8 for node logs.
305 #define NR_CURSEG_DATA_TYPE (3)
306 #define NR_CURSEG_NODE_TYPE (3)
307 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
310 CURSEG_HOT_DATA = 0, /* directory entry blocks */
311 CURSEG_WARM_DATA, /* data blocks */
312 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
313 CURSEG_HOT_NODE, /* direct node blocks of directory files */
314 CURSEG_WARM_NODE, /* direct node blocks of normal files */
315 CURSEG_COLD_NODE, /* indirect node blocks */
319 struct f2fs_sm_info {
320 struct sit_info *sit_info; /* whole segment information */
321 struct free_segmap_info *free_info; /* free segment information */
322 struct dirty_seglist_info *dirty_info; /* dirty segment information */
323 struct curseg_info *curseg_array; /* active segment information */
325 struct list_head wblist_head; /* list of under-writeback pages */
326 spinlock_t wblist_lock; /* lock for checkpoint */
328 block_t seg0_blkaddr; /* block address of 0'th segment */
329 block_t main_blkaddr; /* start block address of main area */
330 block_t ssa_blkaddr; /* start block address of SSA area */
332 unsigned int segment_count; /* total # of segments */
333 unsigned int main_segments; /* # of segments in main area */
334 unsigned int reserved_segments; /* # of reserved segments */
335 unsigned int ovp_segments; /* # of overprovision segments */
337 /* a threshold to reclaim prefree segments */
338 unsigned int rec_prefree_segments;
340 /* for small discard management */
341 struct list_head discard_list; /* 4KB discard list */
342 int nr_discards; /* # of discards in the list */
343 int max_discards; /* max. discards to be issued */
345 unsigned int ipu_policy; /* in-place-update policy */
346 unsigned int min_ipu_util; /* in-place-update threshold */
353 * COUNT_TYPE for monitoring
355 * f2fs monitors the number of several block types such as on-writeback,
356 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
367 * The below are the page types of bios used in submti_bio().
368 * The available types are:
369 * DATA User data pages. It operates as async mode.
370 * NODE Node pages. It operates as async mode.
371 * META FS metadata pages such as SIT, NAT, CP.
372 * NR_PAGE_TYPE The number of page types.
373 * META_FLUSH Make sure the previous pages are written
374 * with waiting the bio's completion
375 * ... Only can be used with META.
377 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
386 struct f2fs_io_info {
387 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
388 int rw; /* contains R/RS/W/WS with REQ_META/REQ_PRIO */
391 #define is_read_io(rw) (((rw) & 1) == READ)
392 struct f2fs_bio_info {
393 struct f2fs_sb_info *sbi; /* f2fs superblock */
394 struct bio *bio; /* bios to merge */
395 sector_t last_block_in_bio; /* last block number */
396 struct f2fs_io_info fio; /* store buffered io info. */
397 struct rw_semaphore io_rwsem; /* blocking op for bio */
400 struct f2fs_sb_info {
401 struct super_block *sb; /* pointer to VFS super block */
402 struct proc_dir_entry *s_proc; /* proc entry */
403 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
404 struct f2fs_super_block *raw_super; /* raw super block pointer */
405 int s_dirty; /* dirty flag for checkpoint */
407 /* for node-related operations */
408 struct f2fs_nm_info *nm_info; /* node manager */
409 struct inode *node_inode; /* cache node blocks */
411 /* for segment-related operations */
412 struct f2fs_sm_info *sm_info; /* segment manager */
414 /* for bio operations */
415 struct f2fs_bio_info read_io; /* for read bios */
416 struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */
417 struct completion *wait_io; /* for completion bios */
420 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
421 struct inode *meta_inode; /* cache meta blocks */
422 struct mutex cp_mutex; /* checkpoint procedure lock */
423 struct rw_semaphore cp_rwsem; /* blocking FS operations */
424 struct mutex node_write; /* locking node writes */
425 struct mutex writepages; /* mutex for writepages() */
426 bool por_doing; /* recovery is doing or not */
427 wait_queue_head_t cp_wait;
429 /* for orphan inode management */
430 struct list_head orphan_inode_list; /* orphan inode list */
431 spinlock_t orphan_inode_lock; /* for orphan inode list */
432 unsigned int n_orphans; /* # of orphan inodes */
433 unsigned int max_orphans; /* max orphan inodes */
435 /* for directory inode management */
436 struct list_head dir_inode_list; /* dir inode list */
437 spinlock_t dir_inode_lock; /* for dir inode list lock */
439 /* basic file system units */
440 unsigned int log_sectors_per_block; /* log2 sectors per block */
441 unsigned int log_blocksize; /* log2 block size */
442 unsigned int blocksize; /* block size */
443 unsigned int root_ino_num; /* root inode number*/
444 unsigned int node_ino_num; /* node inode number*/
445 unsigned int meta_ino_num; /* meta inode number*/
446 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
447 unsigned int blocks_per_seg; /* blocks per segment */
448 unsigned int segs_per_sec; /* segments per section */
449 unsigned int secs_per_zone; /* sections per zone */
450 unsigned int total_sections; /* total section count */
451 unsigned int total_node_count; /* total node block count */
452 unsigned int total_valid_node_count; /* valid node block count */
453 unsigned int total_valid_inode_count; /* valid inode count */
454 int active_logs; /* # of active logs */
455 int dir_level; /* directory level */
457 block_t user_block_count; /* # of user blocks */
458 block_t total_valid_block_count; /* # of valid blocks */
459 block_t alloc_valid_block_count; /* # of allocated blocks */
460 block_t last_valid_block_count; /* for recovery */
461 u32 s_next_generation; /* for NFS support */
462 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
464 struct f2fs_mount_info mount_opt; /* mount options */
466 /* for cleaning operations */
467 struct mutex gc_mutex; /* mutex for GC */
468 struct f2fs_gc_kthread *gc_thread; /* GC thread */
469 unsigned int cur_victim_sec; /* current victim section num */
471 /* maximum # of trials to find a victim segment for SSR and GC */
472 unsigned int max_victim_search;
475 * for stat information.
476 * one is for the LFS mode, and the other is for the SSR mode.
478 #ifdef CONFIG_F2FS_STAT_FS
479 struct f2fs_stat_info *stat_info; /* FS status information */
480 unsigned int segment_count[2]; /* # of allocated segments */
481 unsigned int block_count[2]; /* # of allocated blocks */
482 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
483 int inline_inode; /* # of inline_data inodes */
484 int bg_gc; /* background gc calls */
485 unsigned int n_dirty_dirs; /* # of dir inodes */
487 unsigned int last_victim[2]; /* last victim segment # */
488 spinlock_t stat_lock; /* lock for stat operations */
490 /* For sysfs suppport */
491 struct kobject s_kobj;
492 struct completion s_kobj_unregister;
498 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
500 return container_of(inode, struct f2fs_inode_info, vfs_inode);
503 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
505 return sb->s_fs_info;
508 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
510 return (struct f2fs_super_block *)(sbi->raw_super);
513 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
515 return (struct f2fs_checkpoint *)(sbi->ckpt);
518 static inline struct f2fs_node *F2FS_NODE(struct page *page)
520 return (struct f2fs_node *)page_address(page);
523 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
525 return &((struct f2fs_node *)page_address(page))->i;
528 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
530 return (struct f2fs_nm_info *)(sbi->nm_info);
533 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
535 return (struct f2fs_sm_info *)(sbi->sm_info);
538 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
540 return (struct sit_info *)(SM_I(sbi)->sit_info);
543 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
545 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
548 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
550 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
553 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
555 return sbi->meta_inode->i_mapping;
558 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
560 return sbi->node_inode->i_mapping;
563 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
568 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
573 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
575 return le64_to_cpu(cp->checkpoint_ver);
578 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
580 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
581 return ckpt_flags & f;
584 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
586 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
588 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
591 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
593 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
595 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
598 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
600 down_read(&sbi->cp_rwsem);
603 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
605 up_read(&sbi->cp_rwsem);
608 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
610 f2fs_down_write(&sbi->cp_rwsem, &sbi->cp_mutex);
613 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
615 up_write(&sbi->cp_rwsem);
619 * Check whether the given nid is within node id range.
621 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
623 WARN_ON((nid >= NM_I(sbi)->max_nid));
624 if (unlikely(nid >= NM_I(sbi)->max_nid))
629 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
632 * Check whether the inode has blocks or not
634 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
636 if (F2FS_I(inode)->i_xattr_nid)
637 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1;
639 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS;
642 static inline bool f2fs_has_xattr_block(unsigned int ofs)
644 return ofs == XATTR_NODE_OFFSET;
647 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
648 struct inode *inode, blkcnt_t count)
650 block_t valid_block_count;
652 spin_lock(&sbi->stat_lock);
654 sbi->total_valid_block_count + (block_t)count;
655 if (unlikely(valid_block_count > sbi->user_block_count)) {
656 spin_unlock(&sbi->stat_lock);
659 inode->i_blocks += count;
660 sbi->total_valid_block_count = valid_block_count;
661 sbi->alloc_valid_block_count += (block_t)count;
662 spin_unlock(&sbi->stat_lock);
666 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
670 spin_lock(&sbi->stat_lock);
671 f2fs_bug_on(sbi->total_valid_block_count < (block_t) count);
672 f2fs_bug_on(inode->i_blocks < count);
673 inode->i_blocks -= count;
674 sbi->total_valid_block_count -= (block_t)count;
675 spin_unlock(&sbi->stat_lock);
678 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
680 atomic_inc(&sbi->nr_pages[count_type]);
681 F2FS_SET_SB_DIRT(sbi);
684 static inline void inode_inc_dirty_dents(struct inode *inode)
686 inc_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
687 atomic_inc(&F2FS_I(inode)->dirty_dents);
690 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
692 atomic_dec(&sbi->nr_pages[count_type]);
695 static inline void inode_dec_dirty_dents(struct inode *inode)
697 if (!S_ISDIR(inode->i_mode))
700 dec_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
701 atomic_dec(&F2FS_I(inode)->dirty_dents);
704 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
706 return atomic_read(&sbi->nr_pages[count_type]);
709 static inline int get_dirty_dents(struct inode *inode)
711 return atomic_read(&F2FS_I(inode)->dirty_dents);
714 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
716 unsigned int pages_per_sec = sbi->segs_per_sec *
717 (1 << sbi->log_blocks_per_seg);
718 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
719 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
722 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
724 return sbi->total_valid_block_count;
727 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
729 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
731 /* return NAT or SIT bitmap */
732 if (flag == NAT_BITMAP)
733 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
734 else if (flag == SIT_BITMAP)
735 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
740 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
742 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
743 int offset = (flag == NAT_BITMAP) ?
744 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
745 return &ckpt->sit_nat_version_bitmap + offset;
748 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
751 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
752 unsigned long long ckpt_version = cur_cp_version(ckpt);
754 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
757 * odd numbered checkpoint should at cp segment 0
758 * and even segent must be at cp segment 1
760 if (!(ckpt_version & 1))
761 start_addr += sbi->blocks_per_seg;
766 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
768 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
771 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
774 block_t valid_block_count;
775 unsigned int valid_node_count;
777 spin_lock(&sbi->stat_lock);
779 valid_block_count = sbi->total_valid_block_count + 1;
780 if (unlikely(valid_block_count > sbi->user_block_count)) {
781 spin_unlock(&sbi->stat_lock);
785 valid_node_count = sbi->total_valid_node_count + 1;
786 if (unlikely(valid_node_count > sbi->total_node_count)) {
787 spin_unlock(&sbi->stat_lock);
794 sbi->alloc_valid_block_count++;
795 sbi->total_valid_node_count++;
796 sbi->total_valid_block_count++;
797 spin_unlock(&sbi->stat_lock);
802 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
805 spin_lock(&sbi->stat_lock);
807 f2fs_bug_on(!sbi->total_valid_block_count);
808 f2fs_bug_on(!sbi->total_valid_node_count);
809 f2fs_bug_on(!inode->i_blocks);
812 sbi->total_valid_node_count--;
813 sbi->total_valid_block_count--;
815 spin_unlock(&sbi->stat_lock);
818 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
820 return sbi->total_valid_node_count;
823 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
825 spin_lock(&sbi->stat_lock);
826 f2fs_bug_on(sbi->total_valid_inode_count == sbi->total_node_count);
827 sbi->total_valid_inode_count++;
828 spin_unlock(&sbi->stat_lock);
831 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
833 spin_lock(&sbi->stat_lock);
834 f2fs_bug_on(!sbi->total_valid_inode_count);
835 sbi->total_valid_inode_count--;
836 spin_unlock(&sbi->stat_lock);
839 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
841 return sbi->total_valid_inode_count;
844 static inline void f2fs_put_page(struct page *page, int unlock)
850 f2fs_bug_on(!PageLocked(page));
853 page_cache_release(page);
856 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
859 f2fs_put_page(dn->node_page, 1);
860 if (dn->inode_page && dn->node_page != dn->inode_page)
861 f2fs_put_page(dn->inode_page, 0);
862 dn->node_page = NULL;
863 dn->inode_page = NULL;
866 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
869 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
872 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
877 entry = kmem_cache_alloc(cachep, flags);
886 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
888 static inline bool IS_INODE(struct page *page)
890 struct f2fs_node *p = F2FS_NODE(page);
891 return RAW_IS_INODE(p);
894 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
896 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
899 static inline block_t datablock_addr(struct page *node_page,
902 struct f2fs_node *raw_node;
904 raw_node = F2FS_NODE(node_page);
905 addr_array = blkaddr_in_node(raw_node);
906 return le32_to_cpu(addr_array[offset]);
909 static inline int f2fs_test_bit(unsigned int nr, char *addr)
914 mask = 1 << (7 - (nr & 0x07));
918 static inline int f2fs_set_bit(unsigned int nr, char *addr)
924 mask = 1 << (7 - (nr & 0x07));
930 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
936 mask = 1 << (7 - (nr & 0x07));
942 /* used for f2fs_inode_info->flags */
944 FI_NEW_INODE, /* indicate newly allocated inode */
945 FI_DIRTY_INODE, /* indicate inode is dirty or not */
946 FI_INC_LINK, /* need to increment i_nlink */
947 FI_ACL_MODE, /* indicate acl mode */
948 FI_NO_ALLOC, /* should not allocate any blocks */
949 FI_UPDATE_DIR, /* should update inode block for consistency */
950 FI_DELAY_IPUT, /* used for the recovery */
951 FI_NO_EXTENT, /* not to use the extent cache */
952 FI_INLINE_XATTR, /* used for inline xattr */
953 FI_INLINE_DATA, /* used for inline data*/
956 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
958 set_bit(flag, &fi->flags);
961 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
963 return test_bit(flag, &fi->flags);
966 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
968 clear_bit(flag, &fi->flags);
971 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
973 fi->i_acl_mode = mode;
974 set_inode_flag(fi, FI_ACL_MODE);
977 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
979 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
980 clear_inode_flag(fi, FI_ACL_MODE);
986 static inline void get_inline_info(struct f2fs_inode_info *fi,
987 struct f2fs_inode *ri)
989 if (ri->i_inline & F2FS_INLINE_XATTR)
990 set_inode_flag(fi, FI_INLINE_XATTR);
991 if (ri->i_inline & F2FS_INLINE_DATA)
992 set_inode_flag(fi, FI_INLINE_DATA);
995 static inline void set_raw_inline(struct f2fs_inode_info *fi,
996 struct f2fs_inode *ri)
1000 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
1001 ri->i_inline |= F2FS_INLINE_XATTR;
1002 if (is_inode_flag_set(fi, FI_INLINE_DATA))
1003 ri->i_inline |= F2FS_INLINE_DATA;
1006 static inline int f2fs_has_inline_xattr(struct inode *inode)
1008 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_XATTR);
1011 static inline unsigned int addrs_per_inode(struct f2fs_inode_info *fi)
1013 if (f2fs_has_inline_xattr(&fi->vfs_inode))
1014 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
1015 return DEF_ADDRS_PER_INODE;
1018 static inline void *inline_xattr_addr(struct page *page)
1020 struct f2fs_inode *ri = F2FS_INODE(page);
1021 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
1022 F2FS_INLINE_XATTR_ADDRS]);
1025 static inline int inline_xattr_size(struct inode *inode)
1027 if (f2fs_has_inline_xattr(inode))
1028 return F2FS_INLINE_XATTR_ADDRS << 2;
1033 static inline int f2fs_has_inline_data(struct inode *inode)
1035 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_DATA);
1038 static inline void *inline_data_addr(struct page *page)
1040 struct f2fs_inode *ri = F2FS_INODE(page);
1041 return (void *)&(ri->i_addr[1]);
1044 static inline int f2fs_readonly(struct super_block *sb)
1046 return sb->s_flags & MS_RDONLY;
1049 static inline void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi)
1051 set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
1052 sbi->sb->s_flags |= MS_RDONLY;
1055 #define get_inode_mode(i) \
1056 ((is_inode_flag_set(F2FS_I(i), FI_ACL_MODE)) ? \
1057 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
1062 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
1063 void truncate_data_blocks(struct dnode_of_data *);
1064 int truncate_blocks(struct inode *, u64);
1065 void f2fs_truncate(struct inode *);
1066 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1067 int f2fs_setattr(struct dentry *, struct iattr *);
1068 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
1069 int truncate_data_blocks_range(struct dnode_of_data *, int);
1070 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
1071 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
1076 void f2fs_set_inode_flags(struct inode *);
1077 struct inode *f2fs_iget(struct super_block *, unsigned long);
1078 int try_to_free_nats(struct f2fs_sb_info *, int);
1079 void update_inode(struct inode *, struct page *);
1080 void update_inode_page(struct inode *);
1081 int f2fs_write_inode(struct inode *, struct writeback_control *);
1082 void f2fs_evict_inode(struct inode *);
1087 struct dentry *f2fs_get_parent(struct dentry *child);
1092 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
1094 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
1095 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
1096 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
1097 struct page *, struct inode *);
1098 int update_dent_inode(struct inode *, const struct qstr *);
1099 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
1100 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
1101 int f2fs_make_empty(struct inode *, struct inode *);
1102 bool f2fs_empty_dir(struct inode *);
1104 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
1106 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
1113 int f2fs_sync_fs(struct super_block *, int);
1114 extern __printf(3, 4)
1115 void f2fs_msg(struct super_block *, const char *, const char *, ...);
1120 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
1125 struct dnode_of_data;
1128 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
1129 bool fsync_mark_done(struct f2fs_sb_info *, nid_t);
1130 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
1131 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
1132 int truncate_inode_blocks(struct inode *, pgoff_t);
1133 int truncate_xattr_node(struct inode *, struct page *);
1134 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
1135 void remove_inode_page(struct inode *);
1136 struct page *new_inode_page(struct inode *, const struct qstr *);
1137 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
1138 void ra_node_page(struct f2fs_sb_info *, nid_t);
1139 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
1140 struct page *get_node_page_ra(struct page *, int);
1141 void sync_inode_page(struct dnode_of_data *);
1142 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
1143 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
1144 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
1145 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
1146 void recover_node_page(struct f2fs_sb_info *, struct page *,
1147 struct f2fs_summary *, struct node_info *, block_t);
1148 bool recover_xattr_data(struct inode *, struct page *, block_t);
1149 int recover_inode_page(struct f2fs_sb_info *, struct page *);
1150 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
1151 struct f2fs_summary_block *);
1152 void flush_nat_entries(struct f2fs_sb_info *);
1153 int build_node_manager(struct f2fs_sb_info *);
1154 void destroy_node_manager(struct f2fs_sb_info *);
1155 int __init create_node_manager_caches(void);
1156 void destroy_node_manager_caches(void);
1161 void f2fs_balance_fs(struct f2fs_sb_info *);
1162 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
1163 void invalidate_blocks(struct f2fs_sb_info *, block_t);
1164 void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t);
1165 void clear_prefree_segments(struct f2fs_sb_info *);
1166 int npages_for_summary_flush(struct f2fs_sb_info *);
1167 void allocate_new_segments(struct f2fs_sb_info *);
1168 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1169 void write_meta_page(struct f2fs_sb_info *, struct page *);
1170 void write_node_page(struct f2fs_sb_info *, struct page *,
1171 struct f2fs_io_info *, unsigned int, block_t, block_t *);
1172 void write_data_page(struct page *, struct dnode_of_data *, block_t *,
1173 struct f2fs_io_info *);
1174 void rewrite_data_page(struct page *, block_t, struct f2fs_io_info *);
1175 void recover_data_page(struct f2fs_sb_info *, struct page *,
1176 struct f2fs_summary *, block_t, block_t);
1177 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1178 struct f2fs_summary *, block_t, block_t);
1179 void allocate_data_block(struct f2fs_sb_info *, struct page *,
1180 block_t, block_t *, struct f2fs_summary *, int);
1181 void f2fs_wait_on_page_writeback(struct page *, enum page_type);
1182 void write_data_summaries(struct f2fs_sb_info *, block_t);
1183 void write_node_summaries(struct f2fs_sb_info *, block_t);
1184 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1185 int, unsigned int, int);
1186 void flush_sit_entries(struct f2fs_sb_info *);
1187 int build_segment_manager(struct f2fs_sb_info *);
1188 void destroy_segment_manager(struct f2fs_sb_info *);
1189 int __init create_segment_manager_caches(void);
1190 void destroy_segment_manager_caches(void);
1195 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1196 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1197 int ra_meta_pages(struct f2fs_sb_info *, int, int, int);
1198 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1199 int acquire_orphan_inode(struct f2fs_sb_info *);
1200 void release_orphan_inode(struct f2fs_sb_info *);
1201 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1202 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1203 void recover_orphan_inodes(struct f2fs_sb_info *);
1204 int get_valid_checkpoint(struct f2fs_sb_info *);
1205 void set_dirty_dir_page(struct inode *, struct page *);
1206 void add_dirty_dir_inode(struct inode *);
1207 void remove_dirty_dir_inode(struct inode *);
1208 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1209 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1210 void write_checkpoint(struct f2fs_sb_info *, bool);
1211 void init_orphan_info(struct f2fs_sb_info *);
1212 int __init create_checkpoint_caches(void);
1213 void destroy_checkpoint_caches(void);
1218 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
1219 int f2fs_submit_page_bio(struct f2fs_sb_info *, struct page *, block_t, int);
1220 void f2fs_submit_page_mbio(struct f2fs_sb_info *, struct page *, block_t,
1221 struct f2fs_io_info *);
1222 int reserve_new_block(struct dnode_of_data *);
1223 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
1224 void update_extent_cache(block_t, struct dnode_of_data *);
1225 struct page *find_data_page(struct inode *, pgoff_t, bool);
1226 struct page *get_lock_data_page(struct inode *, pgoff_t);
1227 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1228 int do_write_data_page(struct page *, struct f2fs_io_info *);
1233 int start_gc_thread(struct f2fs_sb_info *);
1234 void stop_gc_thread(struct f2fs_sb_info *);
1235 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
1236 int f2fs_gc(struct f2fs_sb_info *);
1237 void build_gc_manager(struct f2fs_sb_info *);
1238 int __init create_gc_caches(void);
1239 void destroy_gc_caches(void);
1244 int recover_fsync_data(struct f2fs_sb_info *);
1245 bool space_for_roll_forward(struct f2fs_sb_info *);
1250 #ifdef CONFIG_F2FS_STAT_FS
1251 struct f2fs_stat_info {
1252 struct list_head stat_list;
1253 struct f2fs_sb_info *sbi;
1254 struct mutex stat_lock;
1255 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1256 int main_area_segs, main_area_sections, main_area_zones;
1257 int hit_ext, total_ext;
1258 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1259 int nats, sits, fnids;
1260 int total_count, utilization;
1261 int bg_gc, inline_inode;
1262 unsigned int valid_count, valid_node_count, valid_inode_count;
1263 unsigned int bimodal, avg_vblocks;
1264 int util_free, util_valid, util_invalid;
1265 int rsvd_segs, overp_segs;
1266 int dirty_count, node_pages, meta_pages;
1267 int prefree_count, call_count, cp_count;
1268 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1269 int tot_blks, data_blks, node_blks;
1270 int curseg[NR_CURSEG_TYPE];
1271 int cursec[NR_CURSEG_TYPE];
1272 int curzone[NR_CURSEG_TYPE];
1274 unsigned int segment_count[2];
1275 unsigned int block_count[2];
1276 unsigned base_mem, cache_mem;
1279 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
1281 return (struct f2fs_stat_info *)sbi->stat_info;
1284 #define stat_inc_cp_count(si) ((si)->cp_count++)
1285 #define stat_inc_call_count(si) ((si)->call_count++)
1286 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1287 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1288 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1289 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1290 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1291 #define stat_inc_inline_inode(inode) \
1293 if (f2fs_has_inline_data(inode)) \
1294 ((F2FS_SB(inode->i_sb))->inline_inode++); \
1296 #define stat_dec_inline_inode(inode) \
1298 if (f2fs_has_inline_data(inode)) \
1299 ((F2FS_SB(inode->i_sb))->inline_inode--); \
1302 #define stat_inc_seg_type(sbi, curseg) \
1303 ((sbi)->segment_count[(curseg)->alloc_type]++)
1304 #define stat_inc_block_count(sbi, curseg) \
1305 ((sbi)->block_count[(curseg)->alloc_type]++)
1307 #define stat_inc_seg_count(sbi, type) \
1309 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1311 if (type == SUM_TYPE_DATA) \
1317 #define stat_inc_tot_blk_count(si, blks) \
1318 (si->tot_blks += (blks))
1320 #define stat_inc_data_blk_count(sbi, blks) \
1322 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1323 stat_inc_tot_blk_count(si, blks); \
1324 si->data_blks += (blks); \
1327 #define stat_inc_node_blk_count(sbi, blks) \
1329 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1330 stat_inc_tot_blk_count(si, blks); \
1331 si->node_blks += (blks); \
1334 int f2fs_build_stats(struct f2fs_sb_info *);
1335 void f2fs_destroy_stats(struct f2fs_sb_info *);
1336 void __init f2fs_create_root_stats(void);
1337 void f2fs_destroy_root_stats(void);
1339 #define stat_inc_cp_count(si)
1340 #define stat_inc_call_count(si)
1341 #define stat_inc_bggc_count(si)
1342 #define stat_inc_dirty_dir(sbi)
1343 #define stat_dec_dirty_dir(sbi)
1344 #define stat_inc_total_hit(sb)
1345 #define stat_inc_read_hit(sb)
1346 #define stat_inc_inline_inode(inode)
1347 #define stat_dec_inline_inode(inode)
1348 #define stat_inc_seg_type(sbi, curseg)
1349 #define stat_inc_block_count(sbi, curseg)
1350 #define stat_inc_seg_count(si, type)
1351 #define stat_inc_tot_blk_count(si, blks)
1352 #define stat_inc_data_blk_count(si, blks)
1353 #define stat_inc_node_blk_count(sbi, blks)
1355 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1356 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1357 static inline void __init f2fs_create_root_stats(void) { }
1358 static inline void f2fs_destroy_root_stats(void) { }
1361 extern const struct file_operations f2fs_dir_operations;
1362 extern const struct file_operations f2fs_file_operations;
1363 extern const struct inode_operations f2fs_file_inode_operations;
1364 extern const struct address_space_operations f2fs_dblock_aops;
1365 extern const struct address_space_operations f2fs_node_aops;
1366 extern const struct address_space_operations f2fs_meta_aops;
1367 extern const struct inode_operations f2fs_dir_inode_operations;
1368 extern const struct inode_operations f2fs_symlink_inode_operations;
1369 extern const struct inode_operations f2fs_special_inode_operations;
1374 bool f2fs_may_inline(struct inode *);
1375 int f2fs_read_inline_data(struct inode *, struct page *);
1376 int f2fs_convert_inline_data(struct inode *, pgoff_t);
1377 int f2fs_write_inline_data(struct inode *, struct page *, unsigned int);
1378 int recover_inline_data(struct inode *, struct page *);