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
91 /* for the list of orphan inodes */
92 struct orphan_inode_entry
{
93 struct list_head list
; /* list head */
94 nid_t ino
; /* inode number */
97 /* for the list of directory inodes */
98 struct dir_inode_entry
{
99 struct list_head list
; /* list head */
100 struct inode
*inode
; /* vfs inode pointer */
103 /* for the list of blockaddresses to be discarded */
104 struct discard_entry
{
105 struct list_head list
; /* list head */
106 block_t blkaddr
; /* block address to be discarded */
107 int len
; /* # of consecutive blocks of the discard */
110 /* for the list of fsync inodes, used only during recovery */
111 struct fsync_inode_entry
{
112 struct list_head list
; /* list head */
113 struct inode
*inode
; /* vfs inode pointer */
114 block_t blkaddr
; /* block address locating the last inode */
117 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
118 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
120 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
121 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
122 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
123 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
125 static inline int update_nats_in_cursum(struct f2fs_summary_block
*rs
, int i
)
127 int before
= nats_in_cursum(rs
);
128 rs
->n_nats
= cpu_to_le16(before
+ i
);
132 static inline int update_sits_in_cursum(struct f2fs_summary_block
*rs
, int i
)
134 int before
= sits_in_cursum(rs
);
135 rs
->n_sits
= cpu_to_le16(before
+ i
);
142 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
143 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
145 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
147 * ioctl commands in 32 bit emulation
149 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
150 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
154 * For INODE and NODE manager
157 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
158 * as its node offset to distinguish from index node blocks.
159 * But some bits are used to mark the node block.
161 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
164 ALLOC_NODE
, /* allocate a new node page if needed */
165 LOOKUP_NODE
, /* look up a node without readahead */
167 * look up a node with readahead called
168 * by get_datablock_ro.
172 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
174 /* for in-memory extent cache entry */
175 #define F2FS_MIN_EXTENT_LEN 16 /* minimum extent length */
178 rwlock_t ext_lock
; /* rwlock for consistency */
179 unsigned int fofs
; /* start offset in a file */
180 u32 blk_addr
; /* start block address of the extent */
181 unsigned int len
; /* length of the extent */
185 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
187 #define FADVISE_COLD_BIT 0x01
188 #define FADVISE_LOST_PINO_BIT 0x02
190 struct f2fs_inode_info
{
191 struct inode vfs_inode
; /* serve a vfs inode */
192 unsigned long i_flags
; /* keep an inode flags for ioctl */
193 unsigned char i_advise
; /* use to give file attribute hints */
194 unsigned int i_current_depth
; /* use only in directory structure */
195 unsigned int i_pino
; /* parent inode number */
196 umode_t i_acl_mode
; /* keep file acl mode temporarily */
198 /* Use below internally in f2fs*/
199 unsigned long flags
; /* use to pass per-file flags */
200 atomic_t dirty_dents
; /* # of dirty dentry pages */
201 f2fs_hash_t chash
; /* hash value of given file name */
202 unsigned int clevel
; /* maximum level of given file name */
203 nid_t i_xattr_nid
; /* node id that contains xattrs */
204 unsigned long long xattr_ver
; /* cp version of xattr modification */
205 struct extent_info ext
; /* in-memory extent cache entry */
208 static inline void get_extent_info(struct extent_info
*ext
,
209 struct f2fs_extent i_ext
)
211 write_lock(&ext
->ext_lock
);
212 ext
->fofs
= le32_to_cpu(i_ext
.fofs
);
213 ext
->blk_addr
= le32_to_cpu(i_ext
.blk_addr
);
214 ext
->len
= le32_to_cpu(i_ext
.len
);
215 write_unlock(&ext
->ext_lock
);
218 static inline void set_raw_extent(struct extent_info
*ext
,
219 struct f2fs_extent
*i_ext
)
221 read_lock(&ext
->ext_lock
);
222 i_ext
->fofs
= cpu_to_le32(ext
->fofs
);
223 i_ext
->blk_addr
= cpu_to_le32(ext
->blk_addr
);
224 i_ext
->len
= cpu_to_le32(ext
->len
);
225 read_unlock(&ext
->ext_lock
);
228 struct f2fs_nm_info
{
229 block_t nat_blkaddr
; /* base disk address of NAT */
230 nid_t max_nid
; /* maximum possible node ids */
231 nid_t next_scan_nid
; /* the next nid to be scanned */
233 /* NAT cache management */
234 struct radix_tree_root nat_root
;/* root of the nat entry cache */
235 rwlock_t nat_tree_lock
; /* protect nat_tree_lock */
236 unsigned int nat_cnt
; /* the # of cached nat entries */
237 struct list_head nat_entries
; /* cached nat entry list (clean) */
238 struct list_head dirty_nat_entries
; /* cached nat entry list (dirty) */
240 /* free node ids management */
241 struct list_head free_nid_list
; /* a list for free nids */
242 spinlock_t free_nid_list_lock
; /* protect free nid list */
243 unsigned int fcnt
; /* the number of free node id */
244 struct mutex build_lock
; /* lock for build free nids */
247 char *nat_bitmap
; /* NAT bitmap pointer */
248 int bitmap_size
; /* bitmap size */
252 * this structure is used as one of function parameters.
253 * all the information are dedicated to a given direct node block determined
254 * by the data offset in a file.
256 struct dnode_of_data
{
257 struct inode
*inode
; /* vfs inode pointer */
258 struct page
*inode_page
; /* its inode page, NULL is possible */
259 struct page
*node_page
; /* cached direct node page */
260 nid_t nid
; /* node id of the direct node block */
261 unsigned int ofs_in_node
; /* data offset in the node page */
262 bool inode_page_locked
; /* inode page is locked or not */
263 block_t data_blkaddr
; /* block address of the node block */
266 static inline void set_new_dnode(struct dnode_of_data
*dn
, struct inode
*inode
,
267 struct page
*ipage
, struct page
*npage
, nid_t nid
)
269 memset(dn
, 0, sizeof(*dn
));
271 dn
->inode_page
= ipage
;
272 dn
->node_page
= npage
;
279 * By default, there are 6 active log areas across the whole main area.
280 * When considering hot and cold data separation to reduce cleaning overhead,
281 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
283 * In the current design, you should not change the numbers intentionally.
284 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
285 * logs individually according to the underlying devices. (default: 6)
286 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
287 * data and 8 for node logs.
289 #define NR_CURSEG_DATA_TYPE (3)
290 #define NR_CURSEG_NODE_TYPE (3)
291 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
294 CURSEG_HOT_DATA
= 0, /* directory entry blocks */
295 CURSEG_WARM_DATA
, /* data blocks */
296 CURSEG_COLD_DATA
, /* multimedia or GCed data blocks */
297 CURSEG_HOT_NODE
, /* direct node blocks of directory files */
298 CURSEG_WARM_NODE
, /* direct node blocks of normal files */
299 CURSEG_COLD_NODE
, /* indirect node blocks */
303 struct f2fs_sm_info
{
304 struct sit_info
*sit_info
; /* whole segment information */
305 struct free_segmap_info
*free_info
; /* free segment information */
306 struct dirty_seglist_info
*dirty_info
; /* dirty segment information */
307 struct curseg_info
*curseg_array
; /* active segment information */
309 struct list_head wblist_head
; /* list of under-writeback pages */
310 spinlock_t wblist_lock
; /* lock for checkpoint */
312 block_t seg0_blkaddr
; /* block address of 0'th segment */
313 block_t main_blkaddr
; /* start block address of main area */
314 block_t ssa_blkaddr
; /* start block address of SSA area */
316 unsigned int segment_count
; /* total # of segments */
317 unsigned int main_segments
; /* # of segments in main area */
318 unsigned int reserved_segments
; /* # of reserved segments */
319 unsigned int ovp_segments
; /* # of overprovision segments */
321 /* a threshold to reclaim prefree segments */
322 unsigned int rec_prefree_segments
;
324 /* for small discard management */
325 struct list_head discard_list
; /* 4KB discard list */
326 int nr_discards
; /* # of discards in the list */
327 int max_discards
; /* max. discards to be issued */
334 * COUNT_TYPE for monitoring
336 * f2fs monitors the number of several block types such as on-writeback,
337 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
348 * The below are the page types of bios used in submti_bio().
349 * The available types are:
350 * DATA User data pages. It operates as async mode.
351 * NODE Node pages. It operates as async mode.
352 * META FS metadata pages such as SIT, NAT, CP.
353 * NR_PAGE_TYPE The number of page types.
354 * META_FLUSH Make sure the previous pages are written
355 * with waiting the bio's completion
356 * ... Only can be used with META.
358 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
367 struct f2fs_bio_info
{
368 struct bio
*bio
; /* bios to merge */
369 sector_t last_block_in_bio
; /* last block number */
370 struct mutex io_mutex
; /* mutex for bio */
373 struct f2fs_sb_info
{
374 struct super_block
*sb
; /* pointer to VFS super block */
375 struct proc_dir_entry
*s_proc
; /* proc entry */
376 struct buffer_head
*raw_super_buf
; /* buffer head of raw sb */
377 struct f2fs_super_block
*raw_super
; /* raw super block pointer */
378 int s_dirty
; /* dirty flag for checkpoint */
380 /* for node-related operations */
381 struct f2fs_nm_info
*nm_info
; /* node manager */
382 struct inode
*node_inode
; /* cache node blocks */
384 /* for segment-related operations */
385 struct f2fs_sm_info
*sm_info
; /* segment manager */
387 /* for bio operations */
388 struct f2fs_bio_info read_io
; /* for read bios */
389 struct f2fs_bio_info write_io
[NR_PAGE_TYPE
]; /* for write bios */
392 struct f2fs_checkpoint
*ckpt
; /* raw checkpoint pointer */
393 struct inode
*meta_inode
; /* cache meta blocks */
394 struct mutex cp_mutex
; /* checkpoint procedure lock */
395 struct rw_semaphore cp_rwsem
; /* blocking FS operations */
396 struct mutex node_write
; /* locking node writes */
397 struct mutex writepages
; /* mutex for writepages() */
398 bool por_doing
; /* recovery is doing or not */
399 bool on_build_free_nids
; /* build_free_nids is doing */
400 wait_queue_head_t cp_wait
;
402 /* for orphan inode management */
403 struct list_head orphan_inode_list
; /* orphan inode list */
404 struct mutex orphan_inode_mutex
; /* for orphan inode list */
405 unsigned int n_orphans
; /* # of orphan inodes */
407 /* for directory inode management */
408 struct list_head dir_inode_list
; /* dir inode list */
409 spinlock_t dir_inode_lock
; /* for dir inode list lock */
411 /* basic file system units */
412 unsigned int log_sectors_per_block
; /* log2 sectors per block */
413 unsigned int log_blocksize
; /* log2 block size */
414 unsigned int blocksize
; /* block size */
415 unsigned int root_ino_num
; /* root inode number*/
416 unsigned int node_ino_num
; /* node inode number*/
417 unsigned int meta_ino_num
; /* meta inode number*/
418 unsigned int log_blocks_per_seg
; /* log2 blocks per segment */
419 unsigned int blocks_per_seg
; /* blocks per segment */
420 unsigned int segs_per_sec
; /* segments per section */
421 unsigned int secs_per_zone
; /* sections per zone */
422 unsigned int total_sections
; /* total section count */
423 unsigned int total_node_count
; /* total node block count */
424 unsigned int total_valid_node_count
; /* valid node block count */
425 unsigned int total_valid_inode_count
; /* valid inode count */
426 int active_logs
; /* # of active logs */
428 block_t user_block_count
; /* # of user blocks */
429 block_t total_valid_block_count
; /* # of valid blocks */
430 block_t alloc_valid_block_count
; /* # of allocated blocks */
431 block_t last_valid_block_count
; /* for recovery */
432 u32 s_next_generation
; /* for NFS support */
433 atomic_t nr_pages
[NR_COUNT_TYPE
]; /* # of pages, see count_type */
435 struct f2fs_mount_info mount_opt
; /* mount options */
437 /* for cleaning operations */
438 struct mutex gc_mutex
; /* mutex for GC */
439 struct f2fs_gc_kthread
*gc_thread
; /* GC thread */
440 unsigned int cur_victim_sec
; /* current victim section num */
443 * for stat information.
444 * one is for the LFS mode, and the other is for the SSR mode.
446 #ifdef CONFIG_F2FS_STAT_FS
447 struct f2fs_stat_info
*stat_info
; /* FS status information */
448 unsigned int segment_count
[2]; /* # of allocated segments */
449 unsigned int block_count
[2]; /* # of allocated blocks */
450 int total_hit_ext
, read_hit_ext
; /* extent cache hit ratio */
451 int bg_gc
; /* background gc calls */
452 unsigned int n_dirty_dirs
; /* # of dir inodes */
454 unsigned int last_victim
[2]; /* last victim segment # */
455 spinlock_t stat_lock
; /* lock for stat operations */
457 /* For sysfs suppport */
458 struct kobject s_kobj
;
459 struct completion s_kobj_unregister
;
465 static inline struct f2fs_inode_info
*F2FS_I(struct inode
*inode
)
467 return container_of(inode
, struct f2fs_inode_info
, vfs_inode
);
470 static inline struct f2fs_sb_info
*F2FS_SB(struct super_block
*sb
)
472 return sb
->s_fs_info
;
475 static inline struct f2fs_super_block
*F2FS_RAW_SUPER(struct f2fs_sb_info
*sbi
)
477 return (struct f2fs_super_block
*)(sbi
->raw_super
);
480 static inline struct f2fs_checkpoint
*F2FS_CKPT(struct f2fs_sb_info
*sbi
)
482 return (struct f2fs_checkpoint
*)(sbi
->ckpt
);
485 static inline struct f2fs_node
*F2FS_NODE(struct page
*page
)
487 return (struct f2fs_node
*)page_address(page
);
490 static inline struct f2fs_nm_info
*NM_I(struct f2fs_sb_info
*sbi
)
492 return (struct f2fs_nm_info
*)(sbi
->nm_info
);
495 static inline struct f2fs_sm_info
*SM_I(struct f2fs_sb_info
*sbi
)
497 return (struct f2fs_sm_info
*)(sbi
->sm_info
);
500 static inline struct sit_info
*SIT_I(struct f2fs_sb_info
*sbi
)
502 return (struct sit_info
*)(SM_I(sbi
)->sit_info
);
505 static inline struct free_segmap_info
*FREE_I(struct f2fs_sb_info
*sbi
)
507 return (struct free_segmap_info
*)(SM_I(sbi
)->free_info
);
510 static inline struct dirty_seglist_info
*DIRTY_I(struct f2fs_sb_info
*sbi
)
512 return (struct dirty_seglist_info
*)(SM_I(sbi
)->dirty_info
);
515 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info
*sbi
)
520 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info
*sbi
)
525 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint
*cp
)
527 return le64_to_cpu(cp
->checkpoint_ver
);
530 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
532 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
533 return ckpt_flags
& f
;
536 static inline void set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
538 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
540 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
543 static inline void clear_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
545 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
547 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
550 static inline void f2fs_lock_op(struct f2fs_sb_info
*sbi
)
552 down_read(&sbi
->cp_rwsem
);
555 static inline void f2fs_unlock_op(struct f2fs_sb_info
*sbi
)
557 up_read(&sbi
->cp_rwsem
);
560 static inline void f2fs_lock_all(struct f2fs_sb_info
*sbi
)
562 f2fs_down_write(&sbi
->cp_rwsem
, &sbi
->cp_mutex
);
565 static inline void f2fs_unlock_all(struct f2fs_sb_info
*sbi
)
567 up_write(&sbi
->cp_rwsem
);
571 * Check whether the given nid is within node id range.
573 static inline int check_nid_range(struct f2fs_sb_info
*sbi
, nid_t nid
)
575 WARN_ON((nid
>= NM_I(sbi
)->max_nid
));
576 if (nid
>= NM_I(sbi
)->max_nid
)
581 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
584 * Check whether the inode has blocks or not
586 static inline int F2FS_HAS_BLOCKS(struct inode
*inode
)
588 if (F2FS_I(inode
)->i_xattr_nid
)
589 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
+ 1);
591 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
);
594 static inline bool inc_valid_block_count(struct f2fs_sb_info
*sbi
,
595 struct inode
*inode
, blkcnt_t count
)
597 block_t valid_block_count
;
599 spin_lock(&sbi
->stat_lock
);
601 sbi
->total_valid_block_count
+ (block_t
)count
;
602 if (valid_block_count
> sbi
->user_block_count
) {
603 spin_unlock(&sbi
->stat_lock
);
606 inode
->i_blocks
+= count
;
607 sbi
->total_valid_block_count
= valid_block_count
;
608 sbi
->alloc_valid_block_count
+= (block_t
)count
;
609 spin_unlock(&sbi
->stat_lock
);
613 static inline void dec_valid_block_count(struct f2fs_sb_info
*sbi
,
617 spin_lock(&sbi
->stat_lock
);
618 f2fs_bug_on(sbi
->total_valid_block_count
< (block_t
) count
);
619 f2fs_bug_on(inode
->i_blocks
< count
);
620 inode
->i_blocks
-= count
;
621 sbi
->total_valid_block_count
-= (block_t
)count
;
622 spin_unlock(&sbi
->stat_lock
);
625 static inline void inc_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
627 atomic_inc(&sbi
->nr_pages
[count_type
]);
628 F2FS_SET_SB_DIRT(sbi
);
631 static inline void inode_inc_dirty_dents(struct inode
*inode
)
633 atomic_inc(&F2FS_I(inode
)->dirty_dents
);
636 static inline void dec_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
638 atomic_dec(&sbi
->nr_pages
[count_type
]);
641 static inline void inode_dec_dirty_dents(struct inode
*inode
)
643 atomic_dec(&F2FS_I(inode
)->dirty_dents
);
646 static inline int get_pages(struct f2fs_sb_info
*sbi
, int count_type
)
648 return atomic_read(&sbi
->nr_pages
[count_type
]);
651 static inline int get_blocktype_secs(struct f2fs_sb_info
*sbi
, int block_type
)
653 unsigned int pages_per_sec
= sbi
->segs_per_sec
*
654 (1 << sbi
->log_blocks_per_seg
);
655 return ((get_pages(sbi
, block_type
) + pages_per_sec
- 1)
656 >> sbi
->log_blocks_per_seg
) / sbi
->segs_per_sec
;
659 static inline block_t
valid_user_blocks(struct f2fs_sb_info
*sbi
)
662 spin_lock(&sbi
->stat_lock
);
663 ret
= sbi
->total_valid_block_count
;
664 spin_unlock(&sbi
->stat_lock
);
668 static inline unsigned long __bitmap_size(struct f2fs_sb_info
*sbi
, int flag
)
670 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
672 /* return NAT or SIT bitmap */
673 if (flag
== NAT_BITMAP
)
674 return le32_to_cpu(ckpt
->nat_ver_bitmap_bytesize
);
675 else if (flag
== SIT_BITMAP
)
676 return le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
);
681 static inline void *__bitmap_ptr(struct f2fs_sb_info
*sbi
, int flag
)
683 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
684 int offset
= (flag
== NAT_BITMAP
) ?
685 le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
) : 0;
686 return &ckpt
->sit_nat_version_bitmap
+ offset
;
689 static inline block_t
__start_cp_addr(struct f2fs_sb_info
*sbi
)
692 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
693 unsigned long long ckpt_version
= cur_cp_version(ckpt
);
695 start_addr
= le32_to_cpu(F2FS_RAW_SUPER(sbi
)->cp_blkaddr
);
698 * odd numbered checkpoint should at cp segment 0
699 * and even segent must be at cp segment 1
701 if (!(ckpt_version
& 1))
702 start_addr
+= sbi
->blocks_per_seg
;
707 static inline block_t
__start_sum_addr(struct f2fs_sb_info
*sbi
)
709 return le32_to_cpu(F2FS_CKPT(sbi
)->cp_pack_start_sum
);
712 static inline bool inc_valid_node_count(struct f2fs_sb_info
*sbi
,
715 block_t valid_block_count
;
716 unsigned int valid_node_count
;
718 spin_lock(&sbi
->stat_lock
);
720 valid_block_count
= sbi
->total_valid_block_count
+ 1;
721 if (valid_block_count
> sbi
->user_block_count
) {
722 spin_unlock(&sbi
->stat_lock
);
726 valid_node_count
= sbi
->total_valid_node_count
+ 1;
727 if (valid_node_count
> sbi
->total_node_count
) {
728 spin_unlock(&sbi
->stat_lock
);
735 sbi
->alloc_valid_block_count
++;
736 sbi
->total_valid_node_count
++;
737 sbi
->total_valid_block_count
++;
738 spin_unlock(&sbi
->stat_lock
);
743 static inline void dec_valid_node_count(struct f2fs_sb_info
*sbi
,
746 spin_lock(&sbi
->stat_lock
);
748 f2fs_bug_on(!sbi
->total_valid_block_count
);
749 f2fs_bug_on(!sbi
->total_valid_node_count
);
750 f2fs_bug_on(!inode
->i_blocks
);
753 sbi
->total_valid_node_count
--;
754 sbi
->total_valid_block_count
--;
756 spin_unlock(&sbi
->stat_lock
);
759 static inline unsigned int valid_node_count(struct f2fs_sb_info
*sbi
)
762 spin_lock(&sbi
->stat_lock
);
763 ret
= sbi
->total_valid_node_count
;
764 spin_unlock(&sbi
->stat_lock
);
768 static inline void inc_valid_inode_count(struct f2fs_sb_info
*sbi
)
770 spin_lock(&sbi
->stat_lock
);
771 f2fs_bug_on(sbi
->total_valid_inode_count
== sbi
->total_node_count
);
772 sbi
->total_valid_inode_count
++;
773 spin_unlock(&sbi
->stat_lock
);
776 static inline void dec_valid_inode_count(struct f2fs_sb_info
*sbi
)
778 spin_lock(&sbi
->stat_lock
);
779 f2fs_bug_on(!sbi
->total_valid_inode_count
);
780 sbi
->total_valid_inode_count
--;
781 spin_unlock(&sbi
->stat_lock
);
784 static inline unsigned int valid_inode_count(struct f2fs_sb_info
*sbi
)
787 spin_lock(&sbi
->stat_lock
);
788 ret
= sbi
->total_valid_inode_count
;
789 spin_unlock(&sbi
->stat_lock
);
793 static inline void f2fs_put_page(struct page
*page
, int unlock
)
795 if (!page
|| IS_ERR(page
))
799 f2fs_bug_on(!PageLocked(page
));
802 page_cache_release(page
);
805 static inline void f2fs_put_dnode(struct dnode_of_data
*dn
)
808 f2fs_put_page(dn
->node_page
, 1);
809 if (dn
->inode_page
&& dn
->node_page
!= dn
->inode_page
)
810 f2fs_put_page(dn
->inode_page
, 0);
811 dn
->node_page
= NULL
;
812 dn
->inode_page
= NULL
;
815 static inline struct kmem_cache
*f2fs_kmem_cache_create(const char *name
,
816 size_t size
, void (*ctor
)(void *))
818 return kmem_cache_create(name
, size
, 0, SLAB_RECLAIM_ACCOUNT
, ctor
);
821 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache
*cachep
,
826 entry
= kmem_cache_alloc(cachep
, flags
);
835 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
837 static inline bool IS_INODE(struct page
*page
)
839 struct f2fs_node
*p
= F2FS_NODE(page
);
840 return RAW_IS_INODE(p
);
843 static inline __le32
*blkaddr_in_node(struct f2fs_node
*node
)
845 return RAW_IS_INODE(node
) ? node
->i
.i_addr
: node
->dn
.addr
;
848 static inline block_t
datablock_addr(struct page
*node_page
,
851 struct f2fs_node
*raw_node
;
853 raw_node
= F2FS_NODE(node_page
);
854 addr_array
= blkaddr_in_node(raw_node
);
855 return le32_to_cpu(addr_array
[offset
]);
858 static inline int f2fs_test_bit(unsigned int nr
, char *addr
)
863 mask
= 1 << (7 - (nr
& 0x07));
867 static inline int f2fs_set_bit(unsigned int nr
, char *addr
)
873 mask
= 1 << (7 - (nr
& 0x07));
879 static inline int f2fs_clear_bit(unsigned int nr
, char *addr
)
885 mask
= 1 << (7 - (nr
& 0x07));
891 /* used for f2fs_inode_info->flags */
893 FI_NEW_INODE
, /* indicate newly allocated inode */
894 FI_DIRTY_INODE
, /* indicate inode is dirty or not */
895 FI_INC_LINK
, /* need to increment i_nlink */
896 FI_ACL_MODE
, /* indicate acl mode */
897 FI_NO_ALLOC
, /* should not allocate any blocks */
898 FI_UPDATE_DIR
, /* should update inode block for consistency */
899 FI_DELAY_IPUT
, /* used for the recovery */
900 FI_NO_EXTENT
, /* not to use the extent cache */
901 FI_INLINE_XATTR
, /* used for inline xattr */
902 FI_INLINE_DATA
, /* used for inline data*/
905 static inline void set_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
907 set_bit(flag
, &fi
->flags
);
910 static inline int is_inode_flag_set(struct f2fs_inode_info
*fi
, int flag
)
912 return test_bit(flag
, &fi
->flags
);
915 static inline void clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
917 clear_bit(flag
, &fi
->flags
);
920 static inline void set_acl_inode(struct f2fs_inode_info
*fi
, umode_t mode
)
922 fi
->i_acl_mode
= mode
;
923 set_inode_flag(fi
, FI_ACL_MODE
);
926 static inline int cond_clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
928 if (is_inode_flag_set(fi
, FI_ACL_MODE
)) {
929 clear_inode_flag(fi
, FI_ACL_MODE
);
935 static inline void get_inline_info(struct f2fs_inode_info
*fi
,
936 struct f2fs_inode
*ri
)
938 if (ri
->i_inline
& F2FS_INLINE_XATTR
)
939 set_inode_flag(fi
, FI_INLINE_XATTR
);
940 if (ri
->i_inline
& F2FS_INLINE_DATA
)
941 set_inode_flag(fi
, FI_INLINE_DATA
);
944 static inline void set_raw_inline(struct f2fs_inode_info
*fi
,
945 struct f2fs_inode
*ri
)
949 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
950 ri
->i_inline
|= F2FS_INLINE_XATTR
;
951 if (is_inode_flag_set(fi
, FI_INLINE_DATA
))
952 ri
->i_inline
|= F2FS_INLINE_DATA
;
955 static inline unsigned int addrs_per_inode(struct f2fs_inode_info
*fi
)
957 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
958 return DEF_ADDRS_PER_INODE
- F2FS_INLINE_XATTR_ADDRS
;
959 return DEF_ADDRS_PER_INODE
;
962 static inline void *inline_xattr_addr(struct page
*page
)
964 struct f2fs_inode
*ri
;
965 ri
= (struct f2fs_inode
*)page_address(page
);
966 return (void *)&(ri
->i_addr
[DEF_ADDRS_PER_INODE
-
967 F2FS_INLINE_XATTR_ADDRS
]);
970 static inline int inline_xattr_size(struct inode
*inode
)
972 if (is_inode_flag_set(F2FS_I(inode
), FI_INLINE_XATTR
))
973 return F2FS_INLINE_XATTR_ADDRS
<< 2;
978 static inline void *inline_data_addr(struct page
*page
)
980 struct f2fs_inode
*ri
;
981 ri
= (struct f2fs_inode
*)page_address(page
);
982 return (void *)&(ri
->i_addr
[1]);
985 static inline int f2fs_readonly(struct super_block
*sb
)
987 return sb
->s_flags
& MS_RDONLY
;
993 int f2fs_sync_file(struct file
*, loff_t
, loff_t
, int);
994 void truncate_data_blocks(struct dnode_of_data
*);
995 void f2fs_truncate(struct inode
*);
996 int f2fs_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
997 int f2fs_setattr(struct dentry
*, struct iattr
*);
998 int truncate_hole(struct inode
*, pgoff_t
, pgoff_t
);
999 int truncate_data_blocks_range(struct dnode_of_data
*, int);
1000 long f2fs_ioctl(struct file
*, unsigned int, unsigned long);
1001 long f2fs_compat_ioctl(struct file
*, unsigned int, unsigned long);
1006 void f2fs_set_inode_flags(struct inode
*);
1007 struct inode
*f2fs_iget(struct super_block
*, unsigned long);
1008 int try_to_free_nats(struct f2fs_sb_info
*, int);
1009 void update_inode(struct inode
*, struct page
*);
1010 int update_inode_page(struct inode
*);
1011 int f2fs_write_inode(struct inode
*, struct writeback_control
*);
1012 void f2fs_evict_inode(struct inode
*);
1017 struct dentry
*f2fs_get_parent(struct dentry
*child
);
1022 struct f2fs_dir_entry
*f2fs_find_entry(struct inode
*, struct qstr
*,
1024 struct f2fs_dir_entry
*f2fs_parent_dir(struct inode
*, struct page
**);
1025 ino_t
f2fs_inode_by_name(struct inode
*, struct qstr
*);
1026 void f2fs_set_link(struct inode
*, struct f2fs_dir_entry
*,
1027 struct page
*, struct inode
*);
1028 int update_dent_inode(struct inode
*, const struct qstr
*);
1029 int __f2fs_add_link(struct inode
*, const struct qstr
*, struct inode
*);
1030 void f2fs_delete_entry(struct f2fs_dir_entry
*, struct page
*, struct inode
*);
1031 int f2fs_make_empty(struct inode
*, struct inode
*);
1032 bool f2fs_empty_dir(struct inode
*);
1034 static inline int f2fs_add_link(struct dentry
*dentry
, struct inode
*inode
)
1036 return __f2fs_add_link(dentry
->d_parent
->d_inode
, &dentry
->d_name
,
1043 int f2fs_sync_fs(struct super_block
*, int);
1044 extern __printf(3, 4)
1045 void f2fs_msg(struct super_block
*, const char *, const char *, ...);
1050 f2fs_hash_t
f2fs_dentry_hash(const char *, size_t);
1055 struct dnode_of_data
;
1058 int is_checkpointed_node(struct f2fs_sb_info
*, nid_t
);
1059 void get_node_info(struct f2fs_sb_info
*, nid_t
, struct node_info
*);
1060 int get_dnode_of_data(struct dnode_of_data
*, pgoff_t
, int);
1061 int truncate_inode_blocks(struct inode
*, pgoff_t
);
1062 int truncate_xattr_node(struct inode
*, struct page
*);
1063 int wait_on_node_pages_writeback(struct f2fs_sb_info
*, nid_t
);
1064 void remove_inode_page(struct inode
*);
1065 struct page
*new_inode_page(struct inode
*, const struct qstr
*);
1066 struct page
*new_node_page(struct dnode_of_data
*, unsigned int, struct page
*);
1067 void ra_node_page(struct f2fs_sb_info
*, nid_t
);
1068 struct page
*get_node_page(struct f2fs_sb_info
*, pgoff_t
);
1069 struct page
*get_node_page_ra(struct page
*, int);
1070 void sync_inode_page(struct dnode_of_data
*);
1071 int sync_node_pages(struct f2fs_sb_info
*, nid_t
, struct writeback_control
*);
1072 bool alloc_nid(struct f2fs_sb_info
*, nid_t
*);
1073 void alloc_nid_done(struct f2fs_sb_info
*, nid_t
);
1074 void alloc_nid_failed(struct f2fs_sb_info
*, nid_t
);
1075 void recover_node_page(struct f2fs_sb_info
*, struct page
*,
1076 struct f2fs_summary
*, struct node_info
*, block_t
);
1077 int recover_inode_page(struct f2fs_sb_info
*, struct page
*);
1078 int restore_node_summary(struct f2fs_sb_info
*, unsigned int,
1079 struct f2fs_summary_block
*);
1080 void flush_nat_entries(struct f2fs_sb_info
*);
1081 int build_node_manager(struct f2fs_sb_info
*);
1082 void destroy_node_manager(struct f2fs_sb_info
*);
1083 int __init
create_node_manager_caches(void);
1084 void destroy_node_manager_caches(void);
1089 void f2fs_balance_fs(struct f2fs_sb_info
*);
1090 void f2fs_balance_fs_bg(struct f2fs_sb_info
*);
1091 void invalidate_blocks(struct f2fs_sb_info
*, block_t
);
1092 void clear_prefree_segments(struct f2fs_sb_info
*);
1093 int npages_for_summary_flush(struct f2fs_sb_info
*);
1094 void allocate_new_segments(struct f2fs_sb_info
*);
1095 struct page
*get_sum_page(struct f2fs_sb_info
*, unsigned int);
1096 struct bio
*f2fs_bio_alloc(struct block_device
*, int);
1097 void f2fs_submit_bio(struct f2fs_sb_info
*, enum page_type
, bool);
1098 void f2fs_wait_on_page_writeback(struct page
*, enum page_type
, bool);
1099 void write_meta_page(struct f2fs_sb_info
*, struct page
*);
1100 void write_node_page(struct f2fs_sb_info
*, struct page
*, unsigned int,
1101 block_t
, block_t
*);
1102 void write_data_page(struct inode
*, struct page
*, struct dnode_of_data
*,
1103 block_t
, block_t
*);
1104 void rewrite_data_page(struct f2fs_sb_info
*, struct page
*, block_t
);
1105 void recover_data_page(struct f2fs_sb_info
*, struct page
*,
1106 struct f2fs_summary
*, block_t
, block_t
);
1107 void rewrite_node_page(struct f2fs_sb_info
*, struct page
*,
1108 struct f2fs_summary
*, block_t
, block_t
);
1109 void write_data_summaries(struct f2fs_sb_info
*, block_t
);
1110 void write_node_summaries(struct f2fs_sb_info
*, block_t
);
1111 int lookup_journal_in_cursum(struct f2fs_summary_block
*,
1112 int, unsigned int, int);
1113 void flush_sit_entries(struct f2fs_sb_info
*);
1114 int build_segment_manager(struct f2fs_sb_info
*);
1115 void destroy_segment_manager(struct f2fs_sb_info
*);
1116 int __init
create_segment_manager_caches(void);
1117 void destroy_segment_manager_caches(void);
1122 struct page
*grab_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1123 struct page
*get_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1124 long sync_meta_pages(struct f2fs_sb_info
*, enum page_type
, long);
1125 int acquire_orphan_inode(struct f2fs_sb_info
*);
1126 void release_orphan_inode(struct f2fs_sb_info
*);
1127 void add_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1128 void remove_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1129 int recover_orphan_inodes(struct f2fs_sb_info
*);
1130 int get_valid_checkpoint(struct f2fs_sb_info
*);
1131 void set_dirty_dir_page(struct inode
*, struct page
*);
1132 void add_dirty_dir_inode(struct inode
*);
1133 void remove_dirty_dir_inode(struct inode
*);
1134 struct inode
*check_dirty_dir_inode(struct f2fs_sb_info
*, nid_t
);
1135 void sync_dirty_dir_inodes(struct f2fs_sb_info
*);
1136 void write_checkpoint(struct f2fs_sb_info
*, bool);
1137 void init_orphan_info(struct f2fs_sb_info
*);
1138 int __init
create_checkpoint_caches(void);
1139 void destroy_checkpoint_caches(void);
1144 int reserve_new_block(struct dnode_of_data
*);
1145 int f2fs_reserve_block(struct dnode_of_data
*, pgoff_t
);
1146 void update_extent_cache(block_t
, struct dnode_of_data
*);
1147 struct page
*find_data_page(struct inode
*, pgoff_t
, bool);
1148 struct page
*get_lock_data_page(struct inode
*, pgoff_t
);
1149 struct page
*get_new_data_page(struct inode
*, struct page
*, pgoff_t
, bool);
1150 int f2fs_readpage(struct f2fs_sb_info
*, struct page
*, block_t
, int);
1151 void f2fs_submit_read_bio(struct f2fs_sb_info
*, int);
1152 void submit_read_page(struct f2fs_sb_info
*, struct page
*, block_t
, int);
1153 int do_write_data_page(struct page
*);
1158 int start_gc_thread(struct f2fs_sb_info
*);
1159 void stop_gc_thread(struct f2fs_sb_info
*);
1160 block_t
start_bidx_of_node(unsigned int, struct f2fs_inode_info
*);
1161 int f2fs_gc(struct f2fs_sb_info
*);
1162 void build_gc_manager(struct f2fs_sb_info
*);
1163 int __init
create_gc_caches(void);
1164 void destroy_gc_caches(void);
1169 int recover_fsync_data(struct f2fs_sb_info
*);
1170 bool space_for_roll_forward(struct f2fs_sb_info
*);
1175 #ifdef CONFIG_F2FS_STAT_FS
1176 struct f2fs_stat_info
{
1177 struct list_head stat_list
;
1178 struct f2fs_sb_info
*sbi
;
1179 struct mutex stat_lock
;
1180 int all_area_segs
, sit_area_segs
, nat_area_segs
, ssa_area_segs
;
1181 int main_area_segs
, main_area_sections
, main_area_zones
;
1182 int hit_ext
, total_ext
;
1183 int ndirty_node
, ndirty_dent
, ndirty_dirs
, ndirty_meta
;
1184 int nats
, sits
, fnids
;
1185 int total_count
, utilization
;
1187 unsigned int valid_count
, valid_node_count
, valid_inode_count
;
1188 unsigned int bimodal
, avg_vblocks
;
1189 int util_free
, util_valid
, util_invalid
;
1190 int rsvd_segs
, overp_segs
;
1191 int dirty_count
, node_pages
, meta_pages
;
1192 int prefree_count
, call_count
;
1193 int tot_segs
, node_segs
, data_segs
, free_segs
, free_secs
;
1194 int tot_blks
, data_blks
, node_blks
;
1195 int curseg
[NR_CURSEG_TYPE
];
1196 int cursec
[NR_CURSEG_TYPE
];
1197 int curzone
[NR_CURSEG_TYPE
];
1199 unsigned int segment_count
[2];
1200 unsigned int block_count
[2];
1201 unsigned base_mem
, cache_mem
;
1204 static inline struct f2fs_stat_info
*F2FS_STAT(struct f2fs_sb_info
*sbi
)
1206 return (struct f2fs_stat_info
*)sbi
->stat_info
;
1209 #define stat_inc_call_count(si) ((si)->call_count++)
1210 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1211 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1212 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1213 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1214 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1215 #define stat_inc_seg_type(sbi, curseg) \
1216 ((sbi)->segment_count[(curseg)->alloc_type]++)
1217 #define stat_inc_block_count(sbi, curseg) \
1218 ((sbi)->block_count[(curseg)->alloc_type]++)
1220 #define stat_inc_seg_count(sbi, type) \
1222 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1224 if (type == SUM_TYPE_DATA) \
1230 #define stat_inc_tot_blk_count(si, blks) \
1231 (si->tot_blks += (blks))
1233 #define stat_inc_data_blk_count(sbi, blks) \
1235 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1236 stat_inc_tot_blk_count(si, blks); \
1237 si->data_blks += (blks); \
1240 #define stat_inc_node_blk_count(sbi, blks) \
1242 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1243 stat_inc_tot_blk_count(si, blks); \
1244 si->node_blks += (blks); \
1247 int f2fs_build_stats(struct f2fs_sb_info
*);
1248 void f2fs_destroy_stats(struct f2fs_sb_info
*);
1249 void __init
f2fs_create_root_stats(void);
1250 void f2fs_destroy_root_stats(void);
1252 #define stat_inc_call_count(si)
1253 #define stat_inc_bggc_count(si)
1254 #define stat_inc_dirty_dir(sbi)
1255 #define stat_dec_dirty_dir(sbi)
1256 #define stat_inc_total_hit(sb)
1257 #define stat_inc_read_hit(sb)
1258 #define stat_inc_seg_type(sbi, curseg)
1259 #define stat_inc_block_count(sbi, curseg)
1260 #define stat_inc_seg_count(si, type)
1261 #define stat_inc_tot_blk_count(si, blks)
1262 #define stat_inc_data_blk_count(si, blks)
1263 #define stat_inc_node_blk_count(sbi, blks)
1265 static inline int f2fs_build_stats(struct f2fs_sb_info
*sbi
) { return 0; }
1266 static inline void f2fs_destroy_stats(struct f2fs_sb_info
*sbi
) { }
1267 static inline void __init
f2fs_create_root_stats(void) { }
1268 static inline void f2fs_destroy_root_stats(void) { }
1271 extern const struct file_operations f2fs_dir_operations
;
1272 extern const struct file_operations f2fs_file_operations
;
1273 extern const struct inode_operations f2fs_file_inode_operations
;
1274 extern const struct address_space_operations f2fs_dblock_aops
;
1275 extern const struct address_space_operations f2fs_node_aops
;
1276 extern const struct address_space_operations f2fs_meta_aops
;
1277 extern const struct inode_operations f2fs_dir_inode_operations
;
1278 extern const struct inode_operations f2fs_symlink_inode_operations
;
1279 extern const struct inode_operations f2fs_special_inode_operations
;