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 atomic_t dirty_dents; /* # of dirty dentry pages */
214 f2fs_hash_t chash; /* hash value of given file name */
215 unsigned int clevel; /* maximum level of given file name */
216 nid_t i_xattr_nid; /* node id that contains xattrs */
217 unsigned long long xattr_ver; /* cp version of xattr modification */
218 struct extent_info ext; /* in-memory extent cache entry */
221 static inline void get_extent_info(struct extent_info *ext,
222 struct f2fs_extent i_ext)
224 write_lock(&ext->ext_lock);
225 ext->fofs = le32_to_cpu(i_ext.fofs);
226 ext->blk_addr = le32_to_cpu(i_ext.blk_addr);
227 ext->len = le32_to_cpu(i_ext.len);
228 write_unlock(&ext->ext_lock);
231 static inline void set_raw_extent(struct extent_info *ext,
232 struct f2fs_extent *i_ext)
234 read_lock(&ext->ext_lock);
235 i_ext->fofs = cpu_to_le32(ext->fofs);
236 i_ext->blk_addr = cpu_to_le32(ext->blk_addr);
237 i_ext->len = cpu_to_le32(ext->len);
238 read_unlock(&ext->ext_lock);
241 struct f2fs_nm_info {
242 block_t nat_blkaddr; /* base disk address of NAT */
243 nid_t max_nid; /* maximum possible node ids */
244 nid_t next_scan_nid; /* the next nid to be scanned */
246 /* NAT cache management */
247 struct radix_tree_root nat_root;/* root of the nat entry cache */
248 rwlock_t nat_tree_lock; /* protect nat_tree_lock */
249 unsigned int nat_cnt; /* the # of cached nat entries */
250 struct list_head nat_entries; /* cached nat entry list (clean) */
251 struct list_head dirty_nat_entries; /* cached nat entry list (dirty) */
253 /* free node ids management */
254 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
255 struct list_head free_nid_list; /* a list for free nids */
256 spinlock_t free_nid_list_lock; /* protect free nid list */
257 unsigned int fcnt; /* the number of free node id */
258 struct mutex build_lock; /* lock for build free nids */
261 char *nat_bitmap; /* NAT bitmap pointer */
262 int bitmap_size; /* bitmap size */
266 * this structure is used as one of function parameters.
267 * all the information are dedicated to a given direct node block determined
268 * by the data offset in a file.
270 struct dnode_of_data {
271 struct inode *inode; /* vfs inode pointer */
272 struct page *inode_page; /* its inode page, NULL is possible */
273 struct page *node_page; /* cached direct node page */
274 nid_t nid; /* node id of the direct node block */
275 unsigned int ofs_in_node; /* data offset in the node page */
276 bool inode_page_locked; /* inode page is locked or not */
277 block_t data_blkaddr; /* block address of the node block */
280 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
281 struct page *ipage, struct page *npage, nid_t nid)
283 memset(dn, 0, sizeof(*dn));
285 dn->inode_page = ipage;
286 dn->node_page = npage;
293 * By default, there are 6 active log areas across the whole main area.
294 * When considering hot and cold data separation to reduce cleaning overhead,
295 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
297 * In the current design, you should not change the numbers intentionally.
298 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
299 * logs individually according to the underlying devices. (default: 6)
300 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
301 * data and 8 for node logs.
303 #define NR_CURSEG_DATA_TYPE (3)
304 #define NR_CURSEG_NODE_TYPE (3)
305 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
308 CURSEG_HOT_DATA = 0, /* directory entry blocks */
309 CURSEG_WARM_DATA, /* data blocks */
310 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
311 CURSEG_HOT_NODE, /* direct node blocks of directory files */
312 CURSEG_WARM_NODE, /* direct node blocks of normal files */
313 CURSEG_COLD_NODE, /* indirect node blocks */
317 struct f2fs_sm_info {
318 struct sit_info *sit_info; /* whole segment information */
319 struct free_segmap_info *free_info; /* free segment information */
320 struct dirty_seglist_info *dirty_info; /* dirty segment information */
321 struct curseg_info *curseg_array; /* active segment information */
323 struct list_head wblist_head; /* list of under-writeback pages */
324 spinlock_t wblist_lock; /* lock for checkpoint */
326 block_t seg0_blkaddr; /* block address of 0'th segment */
327 block_t main_blkaddr; /* start block address of main area */
328 block_t ssa_blkaddr; /* start block address of SSA area */
330 unsigned int segment_count; /* total # of segments */
331 unsigned int main_segments; /* # of segments in main area */
332 unsigned int reserved_segments; /* # of reserved segments */
333 unsigned int ovp_segments; /* # of overprovision segments */
335 /* a threshold to reclaim prefree segments */
336 unsigned int rec_prefree_segments;
338 /* for small discard management */
339 struct list_head discard_list; /* 4KB discard list */
340 int nr_discards; /* # of discards in the list */
341 int max_discards; /* max. discards to be issued */
343 unsigned int ipu_policy; /* in-place-update policy */
344 unsigned int min_ipu_util; /* in-place-update threshold */
351 * COUNT_TYPE for monitoring
353 * f2fs monitors the number of several block types such as on-writeback,
354 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
365 * The below are the page types of bios used in submti_bio().
366 * The available types are:
367 * DATA User data pages. It operates as async mode.
368 * NODE Node pages. It operates as async mode.
369 * META FS metadata pages such as SIT, NAT, CP.
370 * NR_PAGE_TYPE The number of page types.
371 * META_FLUSH Make sure the previous pages are written
372 * with waiting the bio's completion
373 * ... Only can be used with META.
375 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
384 struct f2fs_io_info {
385 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
386 int rw; /* contains R/RS/W/WS with REQ_META/REQ_PRIO */
389 #define is_read_io(rw) (((rw) & 1) == READ)
390 struct f2fs_bio_info {
391 struct f2fs_sb_info *sbi; /* f2fs superblock */
392 struct bio *bio; /* bios to merge */
393 sector_t last_block_in_bio; /* last block number */
394 struct f2fs_io_info fio; /* store buffered io info. */
395 struct mutex io_mutex; /* mutex for bio */
398 struct f2fs_sb_info {
399 struct super_block *sb; /* pointer to VFS super block */
400 struct proc_dir_entry *s_proc; /* proc entry */
401 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
402 struct f2fs_super_block *raw_super; /* raw super block pointer */
403 int s_dirty; /* dirty flag for checkpoint */
405 /* for node-related operations */
406 struct f2fs_nm_info *nm_info; /* node manager */
407 struct inode *node_inode; /* cache node blocks */
409 /* for segment-related operations */
410 struct f2fs_sm_info *sm_info; /* segment manager */
412 /* for bio operations */
413 struct f2fs_bio_info read_io; /* for read bios */
414 struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */
415 struct completion *wait_io; /* for completion bios */
418 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
419 struct inode *meta_inode; /* cache meta blocks */
420 struct mutex cp_mutex; /* checkpoint procedure lock */
421 struct rw_semaphore cp_rwsem; /* blocking FS operations */
422 struct mutex node_write; /* locking node writes */
423 struct mutex writepages; /* mutex for writepages() */
424 bool por_doing; /* recovery is doing or not */
425 wait_queue_head_t cp_wait;
427 /* for orphan inode management */
428 struct list_head orphan_inode_list; /* orphan inode list */
429 spinlock_t orphan_inode_lock; /* for orphan inode list */
430 unsigned int n_orphans; /* # of orphan inodes */
431 unsigned int max_orphans; /* max orphan inodes */
433 /* for directory inode management */
434 struct list_head dir_inode_list; /* dir inode list */
435 spinlock_t dir_inode_lock; /* for dir inode list lock */
437 /* basic file system units */
438 unsigned int log_sectors_per_block; /* log2 sectors per block */
439 unsigned int log_blocksize; /* log2 block size */
440 unsigned int blocksize; /* block size */
441 unsigned int root_ino_num; /* root inode number*/
442 unsigned int node_ino_num; /* node inode number*/
443 unsigned int meta_ino_num; /* meta inode number*/
444 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
445 unsigned int blocks_per_seg; /* blocks per segment */
446 unsigned int segs_per_sec; /* segments per section */
447 unsigned int secs_per_zone; /* sections per zone */
448 unsigned int total_sections; /* total section count */
449 unsigned int total_node_count; /* total node block count */
450 unsigned int total_valid_node_count; /* valid node block count */
451 unsigned int total_valid_inode_count; /* valid inode count */
452 int active_logs; /* # of active logs */
453 int dir_level; /* directory level */
455 block_t user_block_count; /* # of user blocks */
456 block_t total_valid_block_count; /* # of valid blocks */
457 block_t alloc_valid_block_count; /* # of allocated blocks */
458 block_t last_valid_block_count; /* for recovery */
459 u32 s_next_generation; /* for NFS support */
460 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
462 struct f2fs_mount_info mount_opt; /* mount options */
464 /* for cleaning operations */
465 struct mutex gc_mutex; /* mutex for GC */
466 struct f2fs_gc_kthread *gc_thread; /* GC thread */
467 unsigned int cur_victim_sec; /* current victim section num */
469 /* maximum # of trials to find a victim segment for SSR and GC */
470 unsigned int max_victim_search;
473 * for stat information.
474 * one is for the LFS mode, and the other is for the SSR mode.
476 #ifdef CONFIG_F2FS_STAT_FS
477 struct f2fs_stat_info *stat_info; /* FS status information */
478 unsigned int segment_count[2]; /* # of allocated segments */
479 unsigned int block_count[2]; /* # of allocated blocks */
480 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
481 int inline_inode; /* # of inline_data inodes */
482 int bg_gc; /* background gc calls */
483 unsigned int n_dirty_dirs; /* # of dir inodes */
485 unsigned int last_victim[2]; /* last victim segment # */
486 spinlock_t stat_lock; /* lock for stat operations */
488 /* For sysfs suppport */
489 struct kobject s_kobj;
490 struct completion s_kobj_unregister;
496 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
498 return container_of(inode, struct f2fs_inode_info, vfs_inode);
501 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
503 return sb->s_fs_info;
506 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
508 return (struct f2fs_super_block *)(sbi->raw_super);
511 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
513 return (struct f2fs_checkpoint *)(sbi->ckpt);
516 static inline struct f2fs_node *F2FS_NODE(struct page *page)
518 return (struct f2fs_node *)page_address(page);
521 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
523 return &((struct f2fs_node *)page_address(page))->i;
526 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
528 return (struct f2fs_nm_info *)(sbi->nm_info);
531 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
533 return (struct f2fs_sm_info *)(sbi->sm_info);
536 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
538 return (struct sit_info *)(SM_I(sbi)->sit_info);
541 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
543 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
546 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
548 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
551 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
553 return sbi->meta_inode->i_mapping;
556 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
558 return sbi->node_inode->i_mapping;
561 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
566 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
571 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
573 return le64_to_cpu(cp->checkpoint_ver);
576 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
578 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
579 return ckpt_flags & f;
582 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
584 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
586 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
589 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
591 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
593 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
596 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
598 down_read(&sbi->cp_rwsem);
601 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
603 up_read(&sbi->cp_rwsem);
606 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
608 f2fs_down_write(&sbi->cp_rwsem, &sbi->cp_mutex);
611 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
613 up_write(&sbi->cp_rwsem);
617 * Check whether the given nid is within node id range.
619 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
621 WARN_ON((nid >= NM_I(sbi)->max_nid));
622 if (unlikely(nid >= NM_I(sbi)->max_nid))
627 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
630 * Check whether the inode has blocks or not
632 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
634 if (F2FS_I(inode)->i_xattr_nid)
635 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1;
637 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS;
640 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
641 struct inode *inode, blkcnt_t count)
643 block_t valid_block_count;
645 spin_lock(&sbi->stat_lock);
647 sbi->total_valid_block_count + (block_t)count;
648 if (unlikely(valid_block_count > sbi->user_block_count)) {
649 spin_unlock(&sbi->stat_lock);
652 inode->i_blocks += count;
653 sbi->total_valid_block_count = valid_block_count;
654 sbi->alloc_valid_block_count += (block_t)count;
655 spin_unlock(&sbi->stat_lock);
659 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
663 spin_lock(&sbi->stat_lock);
664 f2fs_bug_on(sbi->total_valid_block_count < (block_t) count);
665 f2fs_bug_on(inode->i_blocks < count);
666 inode->i_blocks -= count;
667 sbi->total_valid_block_count -= (block_t)count;
668 spin_unlock(&sbi->stat_lock);
671 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
673 atomic_inc(&sbi->nr_pages[count_type]);
674 F2FS_SET_SB_DIRT(sbi);
677 static inline void inode_inc_dirty_dents(struct inode *inode)
679 inc_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
680 atomic_inc(&F2FS_I(inode)->dirty_dents);
683 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
685 atomic_dec(&sbi->nr_pages[count_type]);
688 static inline void inode_dec_dirty_dents(struct inode *inode)
690 if (!S_ISDIR(inode->i_mode))
693 dec_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
694 atomic_dec(&F2FS_I(inode)->dirty_dents);
697 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
699 return atomic_read(&sbi->nr_pages[count_type]);
702 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
704 unsigned int pages_per_sec = sbi->segs_per_sec *
705 (1 << sbi->log_blocks_per_seg);
706 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
707 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
710 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
712 return sbi->total_valid_block_count;
715 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
717 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
719 /* return NAT or SIT bitmap */
720 if (flag == NAT_BITMAP)
721 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
722 else if (flag == SIT_BITMAP)
723 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
728 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
730 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
731 int offset = (flag == NAT_BITMAP) ?
732 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
733 return &ckpt->sit_nat_version_bitmap + offset;
736 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
739 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
740 unsigned long long ckpt_version = cur_cp_version(ckpt);
742 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
745 * odd numbered checkpoint should at cp segment 0
746 * and even segent must be at cp segment 1
748 if (!(ckpt_version & 1))
749 start_addr += sbi->blocks_per_seg;
754 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
756 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
759 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
762 block_t valid_block_count;
763 unsigned int valid_node_count;
765 spin_lock(&sbi->stat_lock);
767 valid_block_count = sbi->total_valid_block_count + 1;
768 if (unlikely(valid_block_count > sbi->user_block_count)) {
769 spin_unlock(&sbi->stat_lock);
773 valid_node_count = sbi->total_valid_node_count + 1;
774 if (unlikely(valid_node_count > sbi->total_node_count)) {
775 spin_unlock(&sbi->stat_lock);
782 sbi->alloc_valid_block_count++;
783 sbi->total_valid_node_count++;
784 sbi->total_valid_block_count++;
785 spin_unlock(&sbi->stat_lock);
790 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
793 spin_lock(&sbi->stat_lock);
795 f2fs_bug_on(!sbi->total_valid_block_count);
796 f2fs_bug_on(!sbi->total_valid_node_count);
797 f2fs_bug_on(!inode->i_blocks);
800 sbi->total_valid_node_count--;
801 sbi->total_valid_block_count--;
803 spin_unlock(&sbi->stat_lock);
806 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
808 return sbi->total_valid_node_count;
811 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
813 spin_lock(&sbi->stat_lock);
814 f2fs_bug_on(sbi->total_valid_inode_count == sbi->total_node_count);
815 sbi->total_valid_inode_count++;
816 spin_unlock(&sbi->stat_lock);
819 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
821 spin_lock(&sbi->stat_lock);
822 f2fs_bug_on(!sbi->total_valid_inode_count);
823 sbi->total_valid_inode_count--;
824 spin_unlock(&sbi->stat_lock);
827 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
829 return sbi->total_valid_inode_count;
832 static inline void f2fs_put_page(struct page *page, int unlock)
838 f2fs_bug_on(!PageLocked(page));
841 page_cache_release(page);
844 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
847 f2fs_put_page(dn->node_page, 1);
848 if (dn->inode_page && dn->node_page != dn->inode_page)
849 f2fs_put_page(dn->inode_page, 0);
850 dn->node_page = NULL;
851 dn->inode_page = NULL;
854 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
857 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
860 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
865 entry = kmem_cache_alloc(cachep, flags);
874 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
876 static inline bool IS_INODE(struct page *page)
878 struct f2fs_node *p = F2FS_NODE(page);
879 return RAW_IS_INODE(p);
882 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
884 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
887 static inline block_t datablock_addr(struct page *node_page,
890 struct f2fs_node *raw_node;
892 raw_node = F2FS_NODE(node_page);
893 addr_array = blkaddr_in_node(raw_node);
894 return le32_to_cpu(addr_array[offset]);
897 static inline int f2fs_test_bit(unsigned int nr, char *addr)
902 mask = 1 << (7 - (nr & 0x07));
906 static inline int f2fs_set_bit(unsigned int nr, char *addr)
912 mask = 1 << (7 - (nr & 0x07));
918 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
924 mask = 1 << (7 - (nr & 0x07));
930 /* used for f2fs_inode_info->flags */
932 FI_NEW_INODE, /* indicate newly allocated inode */
933 FI_DIRTY_INODE, /* indicate inode is dirty or not */
934 FI_INC_LINK, /* need to increment i_nlink */
935 FI_ACL_MODE, /* indicate acl mode */
936 FI_NO_ALLOC, /* should not allocate any blocks */
937 FI_UPDATE_DIR, /* should update inode block for consistency */
938 FI_DELAY_IPUT, /* used for the recovery */
939 FI_NO_EXTENT, /* not to use the extent cache */
940 FI_INLINE_XATTR, /* used for inline xattr */
941 FI_INLINE_DATA, /* used for inline data*/
944 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
946 set_bit(flag, &fi->flags);
949 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
951 return test_bit(flag, &fi->flags);
954 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
956 clear_bit(flag, &fi->flags);
959 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
961 fi->i_acl_mode = mode;
962 set_inode_flag(fi, FI_ACL_MODE);
965 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
967 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
968 clear_inode_flag(fi, FI_ACL_MODE);
974 static inline void get_inline_info(struct f2fs_inode_info *fi,
975 struct f2fs_inode *ri)
977 if (ri->i_inline & F2FS_INLINE_XATTR)
978 set_inode_flag(fi, FI_INLINE_XATTR);
979 if (ri->i_inline & F2FS_INLINE_DATA)
980 set_inode_flag(fi, FI_INLINE_DATA);
983 static inline void set_raw_inline(struct f2fs_inode_info *fi,
984 struct f2fs_inode *ri)
988 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
989 ri->i_inline |= F2FS_INLINE_XATTR;
990 if (is_inode_flag_set(fi, FI_INLINE_DATA))
991 ri->i_inline |= F2FS_INLINE_DATA;
994 static inline int f2fs_has_inline_xattr(struct inode *inode)
996 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_XATTR);
999 static inline unsigned int addrs_per_inode(struct f2fs_inode_info *fi)
1001 if (f2fs_has_inline_xattr(&fi->vfs_inode))
1002 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
1003 return DEF_ADDRS_PER_INODE;
1006 static inline void *inline_xattr_addr(struct page *page)
1008 struct f2fs_inode *ri = F2FS_INODE(page);
1009 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
1010 F2FS_INLINE_XATTR_ADDRS]);
1013 static inline int inline_xattr_size(struct inode *inode)
1015 if (f2fs_has_inline_xattr(inode))
1016 return F2FS_INLINE_XATTR_ADDRS << 2;
1021 static inline int f2fs_has_inline_data(struct inode *inode)
1023 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_DATA);
1026 static inline void *inline_data_addr(struct page *page)
1028 struct f2fs_inode *ri = F2FS_INODE(page);
1029 return (void *)&(ri->i_addr[1]);
1032 static inline int f2fs_readonly(struct super_block *sb)
1034 return sb->s_flags & MS_RDONLY;
1037 static inline void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi)
1039 set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
1040 sbi->sb->s_flags |= MS_RDONLY;
1043 #define get_inode_mode(i) \
1044 ((is_inode_flag_set(F2FS_I(i), FI_ACL_MODE)) ? \
1045 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
1050 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
1051 void truncate_data_blocks(struct dnode_of_data *);
1052 int truncate_blocks(struct inode *, u64);
1053 void f2fs_truncate(struct inode *);
1054 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1055 int f2fs_setattr(struct dentry *, struct iattr *);
1056 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
1057 int truncate_data_blocks_range(struct dnode_of_data *, int);
1058 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
1059 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
1064 void f2fs_set_inode_flags(struct inode *);
1065 struct inode *f2fs_iget(struct super_block *, unsigned long);
1066 int try_to_free_nats(struct f2fs_sb_info *, int);
1067 void update_inode(struct inode *, struct page *);
1068 void update_inode_page(struct inode *);
1069 int f2fs_write_inode(struct inode *, struct writeback_control *);
1070 void f2fs_evict_inode(struct inode *);
1075 struct dentry *f2fs_get_parent(struct dentry *child);
1080 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
1082 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
1083 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
1084 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
1085 struct page *, struct inode *);
1086 int update_dent_inode(struct inode *, const struct qstr *);
1087 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
1088 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
1089 int f2fs_make_empty(struct inode *, struct inode *);
1090 bool f2fs_empty_dir(struct inode *);
1092 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
1094 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
1101 int f2fs_sync_fs(struct super_block *, int);
1102 extern __printf(3, 4)
1103 void f2fs_msg(struct super_block *, const char *, const char *, ...);
1108 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
1113 struct dnode_of_data;
1116 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
1117 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
1118 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
1119 int truncate_inode_blocks(struct inode *, pgoff_t);
1120 int truncate_xattr_node(struct inode *, struct page *);
1121 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
1122 void remove_inode_page(struct inode *);
1123 struct page *new_inode_page(struct inode *, const struct qstr *);
1124 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
1125 void ra_node_page(struct f2fs_sb_info *, nid_t);
1126 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
1127 struct page *get_node_page_ra(struct page *, int);
1128 void sync_inode_page(struct dnode_of_data *);
1129 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
1130 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
1131 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
1132 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
1133 void recover_node_page(struct f2fs_sb_info *, struct page *,
1134 struct f2fs_summary *, struct node_info *, block_t);
1135 bool recover_xattr_data(struct inode *, struct page *, block_t);
1136 int recover_inode_page(struct f2fs_sb_info *, struct page *);
1137 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
1138 struct f2fs_summary_block *);
1139 void flush_nat_entries(struct f2fs_sb_info *);
1140 int build_node_manager(struct f2fs_sb_info *);
1141 void destroy_node_manager(struct f2fs_sb_info *);
1142 int __init create_node_manager_caches(void);
1143 void destroy_node_manager_caches(void);
1148 void f2fs_balance_fs(struct f2fs_sb_info *);
1149 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
1150 void invalidate_blocks(struct f2fs_sb_info *, block_t);
1151 void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t);
1152 void clear_prefree_segments(struct f2fs_sb_info *);
1153 int npages_for_summary_flush(struct f2fs_sb_info *);
1154 void allocate_new_segments(struct f2fs_sb_info *);
1155 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1156 void write_meta_page(struct f2fs_sb_info *, struct page *);
1157 void write_node_page(struct f2fs_sb_info *, struct page *,
1158 struct f2fs_io_info *, unsigned int, block_t, block_t *);
1159 void write_data_page(struct page *, struct dnode_of_data *, block_t *,
1160 struct f2fs_io_info *);
1161 void rewrite_data_page(struct page *, block_t, struct f2fs_io_info *);
1162 void recover_data_page(struct f2fs_sb_info *, struct page *,
1163 struct f2fs_summary *, block_t, block_t);
1164 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1165 struct f2fs_summary *, block_t, block_t);
1166 void allocate_data_block(struct f2fs_sb_info *, struct page *,
1167 block_t, block_t *, struct f2fs_summary *, int);
1168 void f2fs_wait_on_page_writeback(struct page *, enum page_type);
1169 void write_data_summaries(struct f2fs_sb_info *, block_t);
1170 void write_node_summaries(struct f2fs_sb_info *, block_t);
1171 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1172 int, unsigned int, int);
1173 void flush_sit_entries(struct f2fs_sb_info *);
1174 int build_segment_manager(struct f2fs_sb_info *);
1175 void destroy_segment_manager(struct f2fs_sb_info *);
1176 int __init create_segment_manager_caches(void);
1177 void destroy_segment_manager_caches(void);
1182 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1183 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1184 int ra_meta_pages(struct f2fs_sb_info *, int, int, int);
1185 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1186 int acquire_orphan_inode(struct f2fs_sb_info *);
1187 void release_orphan_inode(struct f2fs_sb_info *);
1188 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1189 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1190 void recover_orphan_inodes(struct f2fs_sb_info *);
1191 int get_valid_checkpoint(struct f2fs_sb_info *);
1192 void set_dirty_dir_page(struct inode *, struct page *);
1193 void add_dirty_dir_inode(struct inode *);
1194 void remove_dirty_dir_inode(struct inode *);
1195 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1196 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1197 void write_checkpoint(struct f2fs_sb_info *, bool);
1198 void init_orphan_info(struct f2fs_sb_info *);
1199 int __init create_checkpoint_caches(void);
1200 void destroy_checkpoint_caches(void);
1205 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
1206 int f2fs_submit_page_bio(struct f2fs_sb_info *, struct page *, block_t, int);
1207 void f2fs_submit_page_mbio(struct f2fs_sb_info *, struct page *, block_t,
1208 struct f2fs_io_info *);
1209 int reserve_new_block(struct dnode_of_data *);
1210 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
1211 void update_extent_cache(block_t, struct dnode_of_data *);
1212 struct page *find_data_page(struct inode *, pgoff_t, bool);
1213 struct page *get_lock_data_page(struct inode *, pgoff_t);
1214 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1215 int do_write_data_page(struct page *, struct f2fs_io_info *);
1220 int start_gc_thread(struct f2fs_sb_info *);
1221 void stop_gc_thread(struct f2fs_sb_info *);
1222 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
1223 int f2fs_gc(struct f2fs_sb_info *);
1224 void build_gc_manager(struct f2fs_sb_info *);
1225 int __init create_gc_caches(void);
1226 void destroy_gc_caches(void);
1231 int recover_fsync_data(struct f2fs_sb_info *);
1232 bool space_for_roll_forward(struct f2fs_sb_info *);
1237 #ifdef CONFIG_F2FS_STAT_FS
1238 struct f2fs_stat_info {
1239 struct list_head stat_list;
1240 struct f2fs_sb_info *sbi;
1241 struct mutex stat_lock;
1242 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1243 int main_area_segs, main_area_sections, main_area_zones;
1244 int hit_ext, total_ext;
1245 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1246 int nats, sits, fnids;
1247 int total_count, utilization;
1248 int bg_gc, inline_inode;
1249 unsigned int valid_count, valid_node_count, valid_inode_count;
1250 unsigned int bimodal, avg_vblocks;
1251 int util_free, util_valid, util_invalid;
1252 int rsvd_segs, overp_segs;
1253 int dirty_count, node_pages, meta_pages;
1254 int prefree_count, call_count, cp_count;
1255 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1256 int tot_blks, data_blks, node_blks;
1257 int curseg[NR_CURSEG_TYPE];
1258 int cursec[NR_CURSEG_TYPE];
1259 int curzone[NR_CURSEG_TYPE];
1261 unsigned int segment_count[2];
1262 unsigned int block_count[2];
1263 unsigned base_mem, cache_mem;
1266 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
1268 return (struct f2fs_stat_info *)sbi->stat_info;
1271 #define stat_inc_cp_count(si) ((si)->cp_count++)
1272 #define stat_inc_call_count(si) ((si)->call_count++)
1273 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1274 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1275 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1276 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1277 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1278 #define stat_inc_inline_inode(inode) \
1280 if (f2fs_has_inline_data(inode)) \
1281 ((F2FS_SB(inode->i_sb))->inline_inode++); \
1283 #define stat_dec_inline_inode(inode) \
1285 if (f2fs_has_inline_data(inode)) \
1286 ((F2FS_SB(inode->i_sb))->inline_inode--); \
1289 #define stat_inc_seg_type(sbi, curseg) \
1290 ((sbi)->segment_count[(curseg)->alloc_type]++)
1291 #define stat_inc_block_count(sbi, curseg) \
1292 ((sbi)->block_count[(curseg)->alloc_type]++)
1294 #define stat_inc_seg_count(sbi, type) \
1296 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1298 if (type == SUM_TYPE_DATA) \
1304 #define stat_inc_tot_blk_count(si, blks) \
1305 (si->tot_blks += (blks))
1307 #define stat_inc_data_blk_count(sbi, blks) \
1309 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1310 stat_inc_tot_blk_count(si, blks); \
1311 si->data_blks += (blks); \
1314 #define stat_inc_node_blk_count(sbi, blks) \
1316 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1317 stat_inc_tot_blk_count(si, blks); \
1318 si->node_blks += (blks); \
1321 int f2fs_build_stats(struct f2fs_sb_info *);
1322 void f2fs_destroy_stats(struct f2fs_sb_info *);
1323 void __init f2fs_create_root_stats(void);
1324 void f2fs_destroy_root_stats(void);
1326 #define stat_inc_cp_count(si)
1327 #define stat_inc_call_count(si)
1328 #define stat_inc_bggc_count(si)
1329 #define stat_inc_dirty_dir(sbi)
1330 #define stat_dec_dirty_dir(sbi)
1331 #define stat_inc_total_hit(sb)
1332 #define stat_inc_read_hit(sb)
1333 #define stat_inc_inline_inode(inode)
1334 #define stat_dec_inline_inode(inode)
1335 #define stat_inc_seg_type(sbi, curseg)
1336 #define stat_inc_block_count(sbi, curseg)
1337 #define stat_inc_seg_count(si, type)
1338 #define stat_inc_tot_blk_count(si, blks)
1339 #define stat_inc_data_blk_count(si, blks)
1340 #define stat_inc_node_blk_count(sbi, blks)
1342 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1343 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1344 static inline void __init f2fs_create_root_stats(void) { }
1345 static inline void f2fs_destroy_root_stats(void) { }
1348 extern const struct file_operations f2fs_dir_operations;
1349 extern const struct file_operations f2fs_file_operations;
1350 extern const struct inode_operations f2fs_file_inode_operations;
1351 extern const struct address_space_operations f2fs_dblock_aops;
1352 extern const struct address_space_operations f2fs_node_aops;
1353 extern const struct address_space_operations f2fs_meta_aops;
1354 extern const struct inode_operations f2fs_dir_inode_operations;
1355 extern const struct inode_operations f2fs_symlink_inode_operations;
1356 extern const struct inode_operations f2fs_special_inode_operations;
1361 bool f2fs_may_inline(struct inode *);
1362 int f2fs_read_inline_data(struct inode *, struct page *);
1363 int f2fs_convert_inline_data(struct inode *, pgoff_t);
1364 int f2fs_write_inline_data(struct inode *, struct page *, unsigned int);
1365 int recover_inline_data(struct inode *, struct page *);