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_io_info
{
368 enum page_type type
; /* contains DATA/NODE/META/META_FLUSH */
369 int rw
; /* contains R/RS/W/WS */
370 int rw_flag
; /* contains REQ_META/REQ_PRIO */
373 #define is_read_io(rw) (((rw) & 1) == READ)
374 struct f2fs_bio_info
{
375 struct f2fs_sb_info
*sbi
; /* f2fs superblock */
376 struct bio
*bio
; /* bios to merge */
377 sector_t last_block_in_bio
; /* last block number */
378 struct f2fs_io_info fio
; /* store buffered io info. */
379 struct mutex io_mutex
; /* mutex for bio */
382 struct f2fs_sb_info
{
383 struct super_block
*sb
; /* pointer to VFS super block */
384 struct proc_dir_entry
*s_proc
; /* proc entry */
385 struct buffer_head
*raw_super_buf
; /* buffer head of raw sb */
386 struct f2fs_super_block
*raw_super
; /* raw super block pointer */
387 int s_dirty
; /* dirty flag for checkpoint */
389 /* for node-related operations */
390 struct f2fs_nm_info
*nm_info
; /* node manager */
391 struct inode
*node_inode
; /* cache node blocks */
393 /* for segment-related operations */
394 struct f2fs_sm_info
*sm_info
; /* segment manager */
396 /* for bio operations */
397 struct f2fs_bio_info read_io
; /* for read bios */
398 struct f2fs_bio_info write_io
[NR_PAGE_TYPE
]; /* for write bios */
401 struct f2fs_checkpoint
*ckpt
; /* raw checkpoint pointer */
402 struct inode
*meta_inode
; /* cache meta blocks */
403 struct mutex cp_mutex
; /* checkpoint procedure lock */
404 struct rw_semaphore cp_rwsem
; /* blocking FS operations */
405 struct mutex node_write
; /* locking node writes */
406 struct mutex writepages
; /* mutex for writepages() */
407 bool por_doing
; /* recovery is doing or not */
408 bool on_build_free_nids
; /* build_free_nids is doing */
409 wait_queue_head_t cp_wait
;
411 /* for orphan inode management */
412 struct list_head orphan_inode_list
; /* orphan inode list */
413 struct mutex orphan_inode_mutex
; /* for orphan inode list */
414 unsigned int n_orphans
; /* # of orphan inodes */
416 /* for directory inode management */
417 struct list_head dir_inode_list
; /* dir inode list */
418 spinlock_t dir_inode_lock
; /* for dir inode list lock */
420 /* basic file system units */
421 unsigned int log_sectors_per_block
; /* log2 sectors per block */
422 unsigned int log_blocksize
; /* log2 block size */
423 unsigned int blocksize
; /* block size */
424 unsigned int root_ino_num
; /* root inode number*/
425 unsigned int node_ino_num
; /* node inode number*/
426 unsigned int meta_ino_num
; /* meta inode number*/
427 unsigned int log_blocks_per_seg
; /* log2 blocks per segment */
428 unsigned int blocks_per_seg
; /* blocks per segment */
429 unsigned int segs_per_sec
; /* segments per section */
430 unsigned int secs_per_zone
; /* sections per zone */
431 unsigned int total_sections
; /* total section count */
432 unsigned int total_node_count
; /* total node block count */
433 unsigned int total_valid_node_count
; /* valid node block count */
434 unsigned int total_valid_inode_count
; /* valid inode count */
435 int active_logs
; /* # of active logs */
437 block_t user_block_count
; /* # of user blocks */
438 block_t total_valid_block_count
; /* # of valid blocks */
439 block_t alloc_valid_block_count
; /* # of allocated blocks */
440 block_t last_valid_block_count
; /* for recovery */
441 u32 s_next_generation
; /* for NFS support */
442 atomic_t nr_pages
[NR_COUNT_TYPE
]; /* # of pages, see count_type */
444 struct f2fs_mount_info mount_opt
; /* mount options */
446 /* for cleaning operations */
447 struct mutex gc_mutex
; /* mutex for GC */
448 struct f2fs_gc_kthread
*gc_thread
; /* GC thread */
449 unsigned int cur_victim_sec
; /* current victim section num */
452 * for stat information.
453 * one is for the LFS mode, and the other is for the SSR mode.
455 #ifdef CONFIG_F2FS_STAT_FS
456 struct f2fs_stat_info
*stat_info
; /* FS status information */
457 unsigned int segment_count
[2]; /* # of allocated segments */
458 unsigned int block_count
[2]; /* # of allocated blocks */
459 int total_hit_ext
, read_hit_ext
; /* extent cache hit ratio */
460 int bg_gc
; /* background gc calls */
461 unsigned int n_dirty_dirs
; /* # of dir inodes */
463 unsigned int last_victim
[2]; /* last victim segment # */
464 spinlock_t stat_lock
; /* lock for stat operations */
466 /* For sysfs suppport */
467 struct kobject s_kobj
;
468 struct completion s_kobj_unregister
;
474 static inline struct f2fs_inode_info
*F2FS_I(struct inode
*inode
)
476 return container_of(inode
, struct f2fs_inode_info
, vfs_inode
);
479 static inline struct f2fs_sb_info
*F2FS_SB(struct super_block
*sb
)
481 return sb
->s_fs_info
;
484 static inline struct f2fs_super_block
*F2FS_RAW_SUPER(struct f2fs_sb_info
*sbi
)
486 return (struct f2fs_super_block
*)(sbi
->raw_super
);
489 static inline struct f2fs_checkpoint
*F2FS_CKPT(struct f2fs_sb_info
*sbi
)
491 return (struct f2fs_checkpoint
*)(sbi
->ckpt
);
494 static inline struct f2fs_node
*F2FS_NODE(struct page
*page
)
496 return (struct f2fs_node
*)page_address(page
);
499 static inline struct f2fs_nm_info
*NM_I(struct f2fs_sb_info
*sbi
)
501 return (struct f2fs_nm_info
*)(sbi
->nm_info
);
504 static inline struct f2fs_sm_info
*SM_I(struct f2fs_sb_info
*sbi
)
506 return (struct f2fs_sm_info
*)(sbi
->sm_info
);
509 static inline struct sit_info
*SIT_I(struct f2fs_sb_info
*sbi
)
511 return (struct sit_info
*)(SM_I(sbi
)->sit_info
);
514 static inline struct free_segmap_info
*FREE_I(struct f2fs_sb_info
*sbi
)
516 return (struct free_segmap_info
*)(SM_I(sbi
)->free_info
);
519 static inline struct dirty_seglist_info
*DIRTY_I(struct f2fs_sb_info
*sbi
)
521 return (struct dirty_seglist_info
*)(SM_I(sbi
)->dirty_info
);
524 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info
*sbi
)
529 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info
*sbi
)
534 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint
*cp
)
536 return le64_to_cpu(cp
->checkpoint_ver
);
539 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
541 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
542 return ckpt_flags
& f
;
545 static inline void set_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
547 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
549 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
552 static inline void clear_ckpt_flags(struct f2fs_checkpoint
*cp
, unsigned int f
)
554 unsigned int ckpt_flags
= le32_to_cpu(cp
->ckpt_flags
);
556 cp
->ckpt_flags
= cpu_to_le32(ckpt_flags
);
559 static inline void f2fs_lock_op(struct f2fs_sb_info
*sbi
)
561 down_read(&sbi
->cp_rwsem
);
564 static inline void f2fs_unlock_op(struct f2fs_sb_info
*sbi
)
566 up_read(&sbi
->cp_rwsem
);
569 static inline void f2fs_lock_all(struct f2fs_sb_info
*sbi
)
571 f2fs_down_write(&sbi
->cp_rwsem
, &sbi
->cp_mutex
);
574 static inline void f2fs_unlock_all(struct f2fs_sb_info
*sbi
)
576 up_write(&sbi
->cp_rwsem
);
580 * Check whether the given nid is within node id range.
582 static inline int check_nid_range(struct f2fs_sb_info
*sbi
, nid_t nid
)
584 WARN_ON((nid
>= NM_I(sbi
)->max_nid
));
585 if (unlikely(nid
>= NM_I(sbi
)->max_nid
))
590 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
593 * Check whether the inode has blocks or not
595 static inline int F2FS_HAS_BLOCKS(struct inode
*inode
)
597 if (F2FS_I(inode
)->i_xattr_nid
)
598 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
+ 1);
600 return (inode
->i_blocks
> F2FS_DEFAULT_ALLOCATED_BLOCKS
);
603 static inline bool inc_valid_block_count(struct f2fs_sb_info
*sbi
,
604 struct inode
*inode
, blkcnt_t count
)
606 block_t valid_block_count
;
608 spin_lock(&sbi
->stat_lock
);
610 sbi
->total_valid_block_count
+ (block_t
)count
;
611 if (unlikely(valid_block_count
> sbi
->user_block_count
)) {
612 spin_unlock(&sbi
->stat_lock
);
615 inode
->i_blocks
+= count
;
616 sbi
->total_valid_block_count
= valid_block_count
;
617 sbi
->alloc_valid_block_count
+= (block_t
)count
;
618 spin_unlock(&sbi
->stat_lock
);
622 static inline void dec_valid_block_count(struct f2fs_sb_info
*sbi
,
626 spin_lock(&sbi
->stat_lock
);
627 f2fs_bug_on(sbi
->total_valid_block_count
< (block_t
) count
);
628 f2fs_bug_on(inode
->i_blocks
< count
);
629 inode
->i_blocks
-= count
;
630 sbi
->total_valid_block_count
-= (block_t
)count
;
631 spin_unlock(&sbi
->stat_lock
);
634 static inline void inc_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
636 atomic_inc(&sbi
->nr_pages
[count_type
]);
637 F2FS_SET_SB_DIRT(sbi
);
640 static inline void inode_inc_dirty_dents(struct inode
*inode
)
642 atomic_inc(&F2FS_I(inode
)->dirty_dents
);
645 static inline void dec_page_count(struct f2fs_sb_info
*sbi
, int count_type
)
647 atomic_dec(&sbi
->nr_pages
[count_type
]);
650 static inline void inode_dec_dirty_dents(struct inode
*inode
)
652 atomic_dec(&F2FS_I(inode
)->dirty_dents
);
655 static inline int get_pages(struct f2fs_sb_info
*sbi
, int count_type
)
657 return atomic_read(&sbi
->nr_pages
[count_type
]);
660 static inline int get_blocktype_secs(struct f2fs_sb_info
*sbi
, int block_type
)
662 unsigned int pages_per_sec
= sbi
->segs_per_sec
*
663 (1 << sbi
->log_blocks_per_seg
);
664 return ((get_pages(sbi
, block_type
) + pages_per_sec
- 1)
665 >> sbi
->log_blocks_per_seg
) / sbi
->segs_per_sec
;
668 static inline block_t
valid_user_blocks(struct f2fs_sb_info
*sbi
)
671 spin_lock(&sbi
->stat_lock
);
672 ret
= sbi
->total_valid_block_count
;
673 spin_unlock(&sbi
->stat_lock
);
677 static inline unsigned long __bitmap_size(struct f2fs_sb_info
*sbi
, int flag
)
679 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
681 /* return NAT or SIT bitmap */
682 if (flag
== NAT_BITMAP
)
683 return le32_to_cpu(ckpt
->nat_ver_bitmap_bytesize
);
684 else if (flag
== SIT_BITMAP
)
685 return le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
);
690 static inline void *__bitmap_ptr(struct f2fs_sb_info
*sbi
, int flag
)
692 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
693 int offset
= (flag
== NAT_BITMAP
) ?
694 le32_to_cpu(ckpt
->sit_ver_bitmap_bytesize
) : 0;
695 return &ckpt
->sit_nat_version_bitmap
+ offset
;
698 static inline block_t
__start_cp_addr(struct f2fs_sb_info
*sbi
)
701 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
702 unsigned long long ckpt_version
= cur_cp_version(ckpt
);
704 start_addr
= le32_to_cpu(F2FS_RAW_SUPER(sbi
)->cp_blkaddr
);
707 * odd numbered checkpoint should at cp segment 0
708 * and even segent must be at cp segment 1
710 if (!(ckpt_version
& 1))
711 start_addr
+= sbi
->blocks_per_seg
;
716 static inline block_t
__start_sum_addr(struct f2fs_sb_info
*sbi
)
718 return le32_to_cpu(F2FS_CKPT(sbi
)->cp_pack_start_sum
);
721 static inline bool inc_valid_node_count(struct f2fs_sb_info
*sbi
,
724 block_t valid_block_count
;
725 unsigned int valid_node_count
;
727 spin_lock(&sbi
->stat_lock
);
729 valid_block_count
= sbi
->total_valid_block_count
+ 1;
730 if (unlikely(valid_block_count
> sbi
->user_block_count
)) {
731 spin_unlock(&sbi
->stat_lock
);
735 valid_node_count
= sbi
->total_valid_node_count
+ 1;
736 if (unlikely(valid_node_count
> sbi
->total_node_count
)) {
737 spin_unlock(&sbi
->stat_lock
);
744 sbi
->alloc_valid_block_count
++;
745 sbi
->total_valid_node_count
++;
746 sbi
->total_valid_block_count
++;
747 spin_unlock(&sbi
->stat_lock
);
752 static inline void dec_valid_node_count(struct f2fs_sb_info
*sbi
,
755 spin_lock(&sbi
->stat_lock
);
757 f2fs_bug_on(!sbi
->total_valid_block_count
);
758 f2fs_bug_on(!sbi
->total_valid_node_count
);
759 f2fs_bug_on(!inode
->i_blocks
);
762 sbi
->total_valid_node_count
--;
763 sbi
->total_valid_block_count
--;
765 spin_unlock(&sbi
->stat_lock
);
768 static inline unsigned int valid_node_count(struct f2fs_sb_info
*sbi
)
771 spin_lock(&sbi
->stat_lock
);
772 ret
= sbi
->total_valid_node_count
;
773 spin_unlock(&sbi
->stat_lock
);
777 static inline void inc_valid_inode_count(struct f2fs_sb_info
*sbi
)
779 spin_lock(&sbi
->stat_lock
);
780 f2fs_bug_on(sbi
->total_valid_inode_count
== sbi
->total_node_count
);
781 sbi
->total_valid_inode_count
++;
782 spin_unlock(&sbi
->stat_lock
);
785 static inline void dec_valid_inode_count(struct f2fs_sb_info
*sbi
)
787 spin_lock(&sbi
->stat_lock
);
788 f2fs_bug_on(!sbi
->total_valid_inode_count
);
789 sbi
->total_valid_inode_count
--;
790 spin_unlock(&sbi
->stat_lock
);
793 static inline unsigned int valid_inode_count(struct f2fs_sb_info
*sbi
)
796 spin_lock(&sbi
->stat_lock
);
797 ret
= sbi
->total_valid_inode_count
;
798 spin_unlock(&sbi
->stat_lock
);
802 static inline void f2fs_put_page(struct page
*page
, int unlock
)
808 f2fs_bug_on(!PageLocked(page
));
811 page_cache_release(page
);
814 static inline void f2fs_put_dnode(struct dnode_of_data
*dn
)
817 f2fs_put_page(dn
->node_page
, 1);
818 if (dn
->inode_page
&& dn
->node_page
!= dn
->inode_page
)
819 f2fs_put_page(dn
->inode_page
, 0);
820 dn
->node_page
= NULL
;
821 dn
->inode_page
= NULL
;
824 static inline struct kmem_cache
*f2fs_kmem_cache_create(const char *name
,
825 size_t size
, void (*ctor
)(void *))
827 return kmem_cache_create(name
, size
, 0, SLAB_RECLAIM_ACCOUNT
, ctor
);
830 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache
*cachep
,
835 entry
= kmem_cache_alloc(cachep
, flags
);
844 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
846 static inline bool IS_INODE(struct page
*page
)
848 struct f2fs_node
*p
= F2FS_NODE(page
);
849 return RAW_IS_INODE(p
);
852 static inline __le32
*blkaddr_in_node(struct f2fs_node
*node
)
854 return RAW_IS_INODE(node
) ? node
->i
.i_addr
: node
->dn
.addr
;
857 static inline block_t
datablock_addr(struct page
*node_page
,
860 struct f2fs_node
*raw_node
;
862 raw_node
= F2FS_NODE(node_page
);
863 addr_array
= blkaddr_in_node(raw_node
);
864 return le32_to_cpu(addr_array
[offset
]);
867 static inline int f2fs_test_bit(unsigned int nr
, char *addr
)
872 mask
= 1 << (7 - (nr
& 0x07));
876 static inline int f2fs_set_bit(unsigned int nr
, char *addr
)
882 mask
= 1 << (7 - (nr
& 0x07));
888 static inline int f2fs_clear_bit(unsigned int nr
, char *addr
)
894 mask
= 1 << (7 - (nr
& 0x07));
900 /* used for f2fs_inode_info->flags */
902 FI_NEW_INODE
, /* indicate newly allocated inode */
903 FI_DIRTY_INODE
, /* indicate inode is dirty or not */
904 FI_INC_LINK
, /* need to increment i_nlink */
905 FI_ACL_MODE
, /* indicate acl mode */
906 FI_NO_ALLOC
, /* should not allocate any blocks */
907 FI_UPDATE_DIR
, /* should update inode block for consistency */
908 FI_DELAY_IPUT
, /* used for the recovery */
909 FI_NO_EXTENT
, /* not to use the extent cache */
910 FI_INLINE_XATTR
, /* used for inline xattr */
911 FI_INLINE_DATA
, /* used for inline data*/
914 static inline void set_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
916 set_bit(flag
, &fi
->flags
);
919 static inline int is_inode_flag_set(struct f2fs_inode_info
*fi
, int flag
)
921 return test_bit(flag
, &fi
->flags
);
924 static inline void clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
926 clear_bit(flag
, &fi
->flags
);
929 static inline void set_acl_inode(struct f2fs_inode_info
*fi
, umode_t mode
)
931 fi
->i_acl_mode
= mode
;
932 set_inode_flag(fi
, FI_ACL_MODE
);
935 static inline int cond_clear_inode_flag(struct f2fs_inode_info
*fi
, int flag
)
937 if (is_inode_flag_set(fi
, FI_ACL_MODE
)) {
938 clear_inode_flag(fi
, FI_ACL_MODE
);
944 static inline void get_inline_info(struct f2fs_inode_info
*fi
,
945 struct f2fs_inode
*ri
)
947 if (ri
->i_inline
& F2FS_INLINE_XATTR
)
948 set_inode_flag(fi
, FI_INLINE_XATTR
);
949 if (ri
->i_inline
& F2FS_INLINE_DATA
)
950 set_inode_flag(fi
, FI_INLINE_DATA
);
953 static inline void set_raw_inline(struct f2fs_inode_info
*fi
,
954 struct f2fs_inode
*ri
)
958 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
959 ri
->i_inline
|= F2FS_INLINE_XATTR
;
960 if (is_inode_flag_set(fi
, FI_INLINE_DATA
))
961 ri
->i_inline
|= F2FS_INLINE_DATA
;
964 static inline unsigned int addrs_per_inode(struct f2fs_inode_info
*fi
)
966 if (is_inode_flag_set(fi
, FI_INLINE_XATTR
))
967 return DEF_ADDRS_PER_INODE
- F2FS_INLINE_XATTR_ADDRS
;
968 return DEF_ADDRS_PER_INODE
;
971 static inline void *inline_xattr_addr(struct page
*page
)
973 struct f2fs_inode
*ri
;
974 ri
= (struct f2fs_inode
*)page_address(page
);
975 return (void *)&(ri
->i_addr
[DEF_ADDRS_PER_INODE
-
976 F2FS_INLINE_XATTR_ADDRS
]);
979 static inline int inline_xattr_size(struct inode
*inode
)
981 if (is_inode_flag_set(F2FS_I(inode
), FI_INLINE_XATTR
))
982 return F2FS_INLINE_XATTR_ADDRS
<< 2;
987 static inline void *inline_data_addr(struct page
*page
)
989 struct f2fs_inode
*ri
;
990 ri
= (struct f2fs_inode
*)page_address(page
);
991 return (void *)&(ri
->i_addr
[1]);
994 static inline int f2fs_readonly(struct super_block
*sb
)
996 return sb
->s_flags
& MS_RDONLY
;
1002 int f2fs_sync_file(struct file
*, loff_t
, loff_t
, int);
1003 void truncate_data_blocks(struct dnode_of_data
*);
1004 void f2fs_truncate(struct inode
*);
1005 int f2fs_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
1006 int f2fs_setattr(struct dentry
*, struct iattr
*);
1007 int truncate_hole(struct inode
*, pgoff_t
, pgoff_t
);
1008 int truncate_data_blocks_range(struct dnode_of_data
*, int);
1009 long f2fs_ioctl(struct file
*, unsigned int, unsigned long);
1010 long f2fs_compat_ioctl(struct file
*, unsigned int, unsigned long);
1015 void f2fs_set_inode_flags(struct inode
*);
1016 struct inode
*f2fs_iget(struct super_block
*, unsigned long);
1017 int try_to_free_nats(struct f2fs_sb_info
*, int);
1018 void update_inode(struct inode
*, struct page
*);
1019 int update_inode_page(struct inode
*);
1020 int f2fs_write_inode(struct inode
*, struct writeback_control
*);
1021 void f2fs_evict_inode(struct inode
*);
1026 struct dentry
*f2fs_get_parent(struct dentry
*child
);
1031 struct f2fs_dir_entry
*f2fs_find_entry(struct inode
*, struct qstr
*,
1033 struct f2fs_dir_entry
*f2fs_parent_dir(struct inode
*, struct page
**);
1034 ino_t
f2fs_inode_by_name(struct inode
*, struct qstr
*);
1035 void f2fs_set_link(struct inode
*, struct f2fs_dir_entry
*,
1036 struct page
*, struct inode
*);
1037 int update_dent_inode(struct inode
*, const struct qstr
*);
1038 int __f2fs_add_link(struct inode
*, const struct qstr
*, struct inode
*);
1039 void f2fs_delete_entry(struct f2fs_dir_entry
*, struct page
*, struct inode
*);
1040 int f2fs_make_empty(struct inode
*, struct inode
*);
1041 bool f2fs_empty_dir(struct inode
*);
1043 static inline int f2fs_add_link(struct dentry
*dentry
, struct inode
*inode
)
1045 return __f2fs_add_link(dentry
->d_parent
->d_inode
, &dentry
->d_name
,
1052 int f2fs_sync_fs(struct super_block
*, int);
1053 extern __printf(3, 4)
1054 void f2fs_msg(struct super_block
*, const char *, const char *, ...);
1059 f2fs_hash_t
f2fs_dentry_hash(const char *, size_t);
1064 struct dnode_of_data
;
1067 int is_checkpointed_node(struct f2fs_sb_info
*, nid_t
);
1068 void get_node_info(struct f2fs_sb_info
*, nid_t
, struct node_info
*);
1069 int get_dnode_of_data(struct dnode_of_data
*, pgoff_t
, int);
1070 int truncate_inode_blocks(struct inode
*, pgoff_t
);
1071 int truncate_xattr_node(struct inode
*, struct page
*);
1072 int wait_on_node_pages_writeback(struct f2fs_sb_info
*, nid_t
);
1073 void remove_inode_page(struct inode
*);
1074 struct page
*new_inode_page(struct inode
*, const struct qstr
*);
1075 struct page
*new_node_page(struct dnode_of_data
*, unsigned int, struct page
*);
1076 void ra_node_page(struct f2fs_sb_info
*, nid_t
);
1077 struct page
*get_node_page(struct f2fs_sb_info
*, pgoff_t
);
1078 struct page
*get_node_page_ra(struct page
*, int);
1079 void sync_inode_page(struct dnode_of_data
*);
1080 int sync_node_pages(struct f2fs_sb_info
*, nid_t
, struct writeback_control
*);
1081 bool alloc_nid(struct f2fs_sb_info
*, nid_t
*);
1082 void alloc_nid_done(struct f2fs_sb_info
*, nid_t
);
1083 void alloc_nid_failed(struct f2fs_sb_info
*, nid_t
);
1084 void recover_node_page(struct f2fs_sb_info
*, struct page
*,
1085 struct f2fs_summary
*, struct node_info
*, block_t
);
1086 int recover_inode_page(struct f2fs_sb_info
*, struct page
*);
1087 int restore_node_summary(struct f2fs_sb_info
*, unsigned int,
1088 struct f2fs_summary_block
*);
1089 void flush_nat_entries(struct f2fs_sb_info
*);
1090 int build_node_manager(struct f2fs_sb_info
*);
1091 void destroy_node_manager(struct f2fs_sb_info
*);
1092 int __init
create_node_manager_caches(void);
1093 void destroy_node_manager_caches(void);
1098 void f2fs_balance_fs(struct f2fs_sb_info
*);
1099 void f2fs_balance_fs_bg(struct f2fs_sb_info
*);
1100 void invalidate_blocks(struct f2fs_sb_info
*, block_t
);
1101 void clear_prefree_segments(struct f2fs_sb_info
*);
1102 int npages_for_summary_flush(struct f2fs_sb_info
*);
1103 void allocate_new_segments(struct f2fs_sb_info
*);
1104 struct page
*get_sum_page(struct f2fs_sb_info
*, unsigned int);
1105 void write_meta_page(struct f2fs_sb_info
*, struct page
*);
1106 void write_node_page(struct f2fs_sb_info
*, struct page
*, unsigned int,
1107 block_t
, block_t
*);
1108 void write_data_page(struct page
*, struct dnode_of_data
*, block_t
*,
1109 struct f2fs_io_info
*);
1110 void rewrite_data_page(struct page
*, block_t
, struct f2fs_io_info
*);
1111 void recover_data_page(struct f2fs_sb_info
*, struct page
*,
1112 struct f2fs_summary
*, block_t
, block_t
);
1113 void rewrite_node_page(struct f2fs_sb_info
*, struct page
*,
1114 struct f2fs_summary
*, block_t
, block_t
);
1115 void f2fs_wait_on_page_writeback(struct page
*, enum page_type
, bool);
1116 void write_data_summaries(struct f2fs_sb_info
*, block_t
);
1117 void write_node_summaries(struct f2fs_sb_info
*, block_t
);
1118 int lookup_journal_in_cursum(struct f2fs_summary_block
*,
1119 int, unsigned int, int);
1120 void flush_sit_entries(struct f2fs_sb_info
*);
1121 int build_segment_manager(struct f2fs_sb_info
*);
1122 void destroy_segment_manager(struct f2fs_sb_info
*);
1123 int __init
create_segment_manager_caches(void);
1124 void destroy_segment_manager_caches(void);
1129 struct page
*grab_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1130 struct page
*get_meta_page(struct f2fs_sb_info
*, pgoff_t
);
1131 long sync_meta_pages(struct f2fs_sb_info
*, enum page_type
, long);
1132 int acquire_orphan_inode(struct f2fs_sb_info
*);
1133 void release_orphan_inode(struct f2fs_sb_info
*);
1134 void add_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1135 void remove_orphan_inode(struct f2fs_sb_info
*, nid_t
);
1136 void recover_orphan_inodes(struct f2fs_sb_info
*);
1137 int get_valid_checkpoint(struct f2fs_sb_info
*);
1138 void set_dirty_dir_page(struct inode
*, struct page
*);
1139 void add_dirty_dir_inode(struct inode
*);
1140 void remove_dirty_dir_inode(struct inode
*);
1141 struct inode
*check_dirty_dir_inode(struct f2fs_sb_info
*, nid_t
);
1142 void sync_dirty_dir_inodes(struct f2fs_sb_info
*);
1143 void write_checkpoint(struct f2fs_sb_info
*, bool);
1144 void init_orphan_info(struct f2fs_sb_info
*);
1145 int __init
create_checkpoint_caches(void);
1146 void destroy_checkpoint_caches(void);
1151 void f2fs_submit_merged_bio(struct f2fs_sb_info
*, enum page_type
, int);
1152 int f2fs_submit_page_bio(struct f2fs_sb_info
*, struct page
*, block_t
, int);
1153 void f2fs_submit_page_mbio(struct f2fs_sb_info
*, struct page
*, block_t
,
1154 struct f2fs_io_info
*);
1155 int reserve_new_block(struct dnode_of_data
*);
1156 int f2fs_reserve_block(struct dnode_of_data
*, pgoff_t
);
1157 void update_extent_cache(block_t
, struct dnode_of_data
*);
1158 struct page
*find_data_page(struct inode
*, pgoff_t
, bool);
1159 struct page
*get_lock_data_page(struct inode
*, pgoff_t
);
1160 struct page
*get_new_data_page(struct inode
*, struct page
*, pgoff_t
, bool);
1161 int do_write_data_page(struct page
*, struct f2fs_io_info
*);
1166 int start_gc_thread(struct f2fs_sb_info
*);
1167 void stop_gc_thread(struct f2fs_sb_info
*);
1168 block_t
start_bidx_of_node(unsigned int, struct f2fs_inode_info
*);
1169 int f2fs_gc(struct f2fs_sb_info
*);
1170 void build_gc_manager(struct f2fs_sb_info
*);
1171 int __init
create_gc_caches(void);
1172 void destroy_gc_caches(void);
1177 int recover_fsync_data(struct f2fs_sb_info
*);
1178 bool space_for_roll_forward(struct f2fs_sb_info
*);
1183 #ifdef CONFIG_F2FS_STAT_FS
1184 struct f2fs_stat_info
{
1185 struct list_head stat_list
;
1186 struct f2fs_sb_info
*sbi
;
1187 struct mutex stat_lock
;
1188 int all_area_segs
, sit_area_segs
, nat_area_segs
, ssa_area_segs
;
1189 int main_area_segs
, main_area_sections
, main_area_zones
;
1190 int hit_ext
, total_ext
;
1191 int ndirty_node
, ndirty_dent
, ndirty_dirs
, ndirty_meta
;
1192 int nats
, sits
, fnids
;
1193 int total_count
, utilization
;
1195 unsigned int valid_count
, valid_node_count
, valid_inode_count
;
1196 unsigned int bimodal
, avg_vblocks
;
1197 int util_free
, util_valid
, util_invalid
;
1198 int rsvd_segs
, overp_segs
;
1199 int dirty_count
, node_pages
, meta_pages
;
1200 int prefree_count
, call_count
;
1201 int tot_segs
, node_segs
, data_segs
, free_segs
, free_secs
;
1202 int tot_blks
, data_blks
, node_blks
;
1203 int curseg
[NR_CURSEG_TYPE
];
1204 int cursec
[NR_CURSEG_TYPE
];
1205 int curzone
[NR_CURSEG_TYPE
];
1207 unsigned int segment_count
[2];
1208 unsigned int block_count
[2];
1209 unsigned base_mem
, cache_mem
;
1212 static inline struct f2fs_stat_info
*F2FS_STAT(struct f2fs_sb_info
*sbi
)
1214 return (struct f2fs_stat_info
*)sbi
->stat_info
;
1217 #define stat_inc_call_count(si) ((si)->call_count++)
1218 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1219 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1220 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1221 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1222 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1223 #define stat_inc_seg_type(sbi, curseg) \
1224 ((sbi)->segment_count[(curseg)->alloc_type]++)
1225 #define stat_inc_block_count(sbi, curseg) \
1226 ((sbi)->block_count[(curseg)->alloc_type]++)
1228 #define stat_inc_seg_count(sbi, type) \
1230 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1232 if (type == SUM_TYPE_DATA) \
1238 #define stat_inc_tot_blk_count(si, blks) \
1239 (si->tot_blks += (blks))
1241 #define stat_inc_data_blk_count(sbi, blks) \
1243 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1244 stat_inc_tot_blk_count(si, blks); \
1245 si->data_blks += (blks); \
1248 #define stat_inc_node_blk_count(sbi, blks) \
1250 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1251 stat_inc_tot_blk_count(si, blks); \
1252 si->node_blks += (blks); \
1255 int f2fs_build_stats(struct f2fs_sb_info
*);
1256 void f2fs_destroy_stats(struct f2fs_sb_info
*);
1257 void __init
f2fs_create_root_stats(void);
1258 void f2fs_destroy_root_stats(void);
1260 #define stat_inc_call_count(si)
1261 #define stat_inc_bggc_count(si)
1262 #define stat_inc_dirty_dir(sbi)
1263 #define stat_dec_dirty_dir(sbi)
1264 #define stat_inc_total_hit(sb)
1265 #define stat_inc_read_hit(sb)
1266 #define stat_inc_seg_type(sbi, curseg)
1267 #define stat_inc_block_count(sbi, curseg)
1268 #define stat_inc_seg_count(si, type)
1269 #define stat_inc_tot_blk_count(si, blks)
1270 #define stat_inc_data_blk_count(si, blks)
1271 #define stat_inc_node_blk_count(sbi, blks)
1273 static inline int f2fs_build_stats(struct f2fs_sb_info
*sbi
) { return 0; }
1274 static inline void f2fs_destroy_stats(struct f2fs_sb_info
*sbi
) { }
1275 static inline void __init
f2fs_create_root_stats(void) { }
1276 static inline void f2fs_destroy_root_stats(void) { }
1279 extern const struct file_operations f2fs_dir_operations
;
1280 extern const struct file_operations f2fs_file_operations
;
1281 extern const struct inode_operations f2fs_file_inode_operations
;
1282 extern const struct address_space_operations f2fs_dblock_aops
;
1283 extern const struct address_space_operations f2fs_node_aops
;
1284 extern const struct address_space_operations f2fs_meta_aops
;
1285 extern const struct inode_operations f2fs_dir_inode_operations
;
1286 extern const struct inode_operations f2fs_symlink_inode_operations
;
1287 extern const struct inode_operations f2fs_special_inode_operations
;