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
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 */
329 unsigned int ipu_policy; /* in-place-update policy */
330 unsigned int min_ipu_util; /* in-place-update threshold */
337 * COUNT_TYPE for monitoring
339 * f2fs monitors the number of several block types such as on-writeback,
340 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
351 * The below are the page types of bios used in submti_bio().
352 * The available types are:
353 * DATA User data pages. It operates as async mode.
354 * NODE Node pages. It operates as async mode.
355 * META FS metadata pages such as SIT, NAT, CP.
356 * NR_PAGE_TYPE The number of page types.
357 * META_FLUSH Make sure the previous pages are written
358 * with waiting the bio's completion
359 * ... Only can be used with META.
361 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
370 struct f2fs_io_info {
371 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
372 int rw; /* contains R/RS/W/WS with REQ_META/REQ_PRIO */
375 #define is_read_io(rw) (((rw) & 1) == READ)
376 struct f2fs_bio_info {
377 struct f2fs_sb_info *sbi; /* f2fs superblock */
378 struct bio *bio; /* bios to merge */
379 sector_t last_block_in_bio; /* last block number */
380 struct f2fs_io_info fio; /* store buffered io info. */
381 struct mutex io_mutex; /* mutex for bio */
384 struct f2fs_sb_info {
385 struct super_block *sb; /* pointer to VFS super block */
386 struct proc_dir_entry *s_proc; /* proc entry */
387 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
388 struct f2fs_super_block *raw_super; /* raw super block pointer */
389 int s_dirty; /* dirty flag for checkpoint */
391 /* for node-related operations */
392 struct f2fs_nm_info *nm_info; /* node manager */
393 struct inode *node_inode; /* cache node blocks */
395 /* for segment-related operations */
396 struct f2fs_sm_info *sm_info; /* segment manager */
398 /* for bio operations */
399 struct f2fs_bio_info read_io; /* for read bios */
400 struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */
403 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
404 struct inode *meta_inode; /* cache meta blocks */
405 struct mutex cp_mutex; /* checkpoint procedure lock */
406 struct rw_semaphore cp_rwsem; /* blocking FS operations */
407 struct mutex node_write; /* locking node writes */
408 struct mutex writepages; /* mutex for writepages() */
409 bool por_doing; /* recovery is doing or not */
410 bool on_build_free_nids; /* build_free_nids is doing */
411 wait_queue_head_t cp_wait;
413 /* for orphan inode management */
414 struct list_head orphan_inode_list; /* orphan inode list */
415 struct mutex orphan_inode_mutex; /* for orphan inode list */
416 unsigned int n_orphans; /* # of orphan inodes */
417 unsigned int max_orphans; /* max orphan inodes */
419 /* for directory inode management */
420 struct list_head dir_inode_list; /* dir inode list */
421 spinlock_t dir_inode_lock; /* for dir inode list lock */
423 /* basic file system units */
424 unsigned int log_sectors_per_block; /* log2 sectors per block */
425 unsigned int log_blocksize; /* log2 block size */
426 unsigned int blocksize; /* block size */
427 unsigned int root_ino_num; /* root inode number*/
428 unsigned int node_ino_num; /* node inode number*/
429 unsigned int meta_ino_num; /* meta inode number*/
430 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
431 unsigned int blocks_per_seg; /* blocks per segment */
432 unsigned int segs_per_sec; /* segments per section */
433 unsigned int secs_per_zone; /* sections per zone */
434 unsigned int total_sections; /* total section count */
435 unsigned int total_node_count; /* total node block count */
436 unsigned int total_valid_node_count; /* valid node block count */
437 unsigned int total_valid_inode_count; /* valid inode count */
438 int active_logs; /* # of active logs */
440 block_t user_block_count; /* # of user blocks */
441 block_t total_valid_block_count; /* # of valid blocks */
442 block_t alloc_valid_block_count; /* # of allocated blocks */
443 block_t last_valid_block_count; /* for recovery */
444 u32 s_next_generation; /* for NFS support */
445 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
447 struct f2fs_mount_info mount_opt; /* mount options */
449 /* for cleaning operations */
450 struct mutex gc_mutex; /* mutex for GC */
451 struct f2fs_gc_kthread *gc_thread; /* GC thread */
452 unsigned int cur_victim_sec; /* current victim section num */
454 /* maximum # of trials to find a victim segment for SSR and GC */
455 unsigned int max_victim_search;
458 * for stat information.
459 * one is for the LFS mode, and the other is for the SSR mode.
461 #ifdef CONFIG_F2FS_STAT_FS
462 struct f2fs_stat_info *stat_info; /* FS status information */
463 unsigned int segment_count[2]; /* # of allocated segments */
464 unsigned int block_count[2]; /* # of allocated blocks */
465 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
466 int inline_inode; /* # of inline_data inodes */
467 int bg_gc; /* background gc calls */
468 unsigned int n_dirty_dirs; /* # of dir inodes */
470 unsigned int last_victim[2]; /* last victim segment # */
471 spinlock_t stat_lock; /* lock for stat operations */
473 /* For sysfs suppport */
474 struct kobject s_kobj;
475 struct completion s_kobj_unregister;
481 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
483 return container_of(inode, struct f2fs_inode_info, vfs_inode);
486 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
488 return sb->s_fs_info;
491 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
493 return (struct f2fs_super_block *)(sbi->raw_super);
496 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
498 return (struct f2fs_checkpoint *)(sbi->ckpt);
501 static inline struct f2fs_node *F2FS_NODE(struct page *page)
503 return (struct f2fs_node *)page_address(page);
506 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
508 return &((struct f2fs_node *)page_address(page))->i;
511 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
513 return (struct f2fs_nm_info *)(sbi->nm_info);
516 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
518 return (struct f2fs_sm_info *)(sbi->sm_info);
521 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
523 return (struct sit_info *)(SM_I(sbi)->sit_info);
526 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
528 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
531 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
533 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
536 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
541 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
546 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
548 return le64_to_cpu(cp->checkpoint_ver);
551 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
553 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
554 return ckpt_flags & f;
557 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
559 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
561 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
564 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
566 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
568 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
571 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
573 down_read(&sbi->cp_rwsem);
576 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
578 up_read(&sbi->cp_rwsem);
581 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
583 f2fs_down_write(&sbi->cp_rwsem, &sbi->cp_mutex);
586 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
588 up_write(&sbi->cp_rwsem);
592 * Check whether the given nid is within node id range.
594 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
596 WARN_ON((nid >= NM_I(sbi)->max_nid));
597 if (unlikely(nid >= NM_I(sbi)->max_nid))
602 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
605 * Check whether the inode has blocks or not
607 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
609 if (F2FS_I(inode)->i_xattr_nid)
610 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1);
612 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS);
615 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
616 struct inode *inode, blkcnt_t count)
618 block_t valid_block_count;
620 spin_lock(&sbi->stat_lock);
622 sbi->total_valid_block_count + (block_t)count;
623 if (unlikely(valid_block_count > sbi->user_block_count)) {
624 spin_unlock(&sbi->stat_lock);
627 inode->i_blocks += count;
628 sbi->total_valid_block_count = valid_block_count;
629 sbi->alloc_valid_block_count += (block_t)count;
630 spin_unlock(&sbi->stat_lock);
634 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
638 spin_lock(&sbi->stat_lock);
639 f2fs_bug_on(sbi->total_valid_block_count < (block_t) count);
640 f2fs_bug_on(inode->i_blocks < count);
641 inode->i_blocks -= count;
642 sbi->total_valid_block_count -= (block_t)count;
643 spin_unlock(&sbi->stat_lock);
646 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
648 atomic_inc(&sbi->nr_pages[count_type]);
649 F2FS_SET_SB_DIRT(sbi);
652 static inline void inode_inc_dirty_dents(struct inode *inode)
654 atomic_inc(&F2FS_I(inode)->dirty_dents);
657 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
659 atomic_dec(&sbi->nr_pages[count_type]);
662 static inline void inode_dec_dirty_dents(struct inode *inode)
664 atomic_dec(&F2FS_I(inode)->dirty_dents);
667 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
669 return atomic_read(&sbi->nr_pages[count_type]);
672 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
674 unsigned int pages_per_sec = sbi->segs_per_sec *
675 (1 << sbi->log_blocks_per_seg);
676 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
677 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
680 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
683 spin_lock(&sbi->stat_lock);
684 ret = sbi->total_valid_block_count;
685 spin_unlock(&sbi->stat_lock);
689 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
691 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
693 /* return NAT or SIT bitmap */
694 if (flag == NAT_BITMAP)
695 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
696 else if (flag == SIT_BITMAP)
697 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
702 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
704 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
705 int offset = (flag == NAT_BITMAP) ?
706 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
707 return &ckpt->sit_nat_version_bitmap + offset;
710 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
713 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
714 unsigned long long ckpt_version = cur_cp_version(ckpt);
716 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
719 * odd numbered checkpoint should at cp segment 0
720 * and even segent must be at cp segment 1
722 if (!(ckpt_version & 1))
723 start_addr += sbi->blocks_per_seg;
728 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
730 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
733 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
736 block_t valid_block_count;
737 unsigned int valid_node_count;
739 spin_lock(&sbi->stat_lock);
741 valid_block_count = sbi->total_valid_block_count + 1;
742 if (unlikely(valid_block_count > sbi->user_block_count)) {
743 spin_unlock(&sbi->stat_lock);
747 valid_node_count = sbi->total_valid_node_count + 1;
748 if (unlikely(valid_node_count > sbi->total_node_count)) {
749 spin_unlock(&sbi->stat_lock);
756 sbi->alloc_valid_block_count++;
757 sbi->total_valid_node_count++;
758 sbi->total_valid_block_count++;
759 spin_unlock(&sbi->stat_lock);
764 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
767 spin_lock(&sbi->stat_lock);
769 f2fs_bug_on(!sbi->total_valid_block_count);
770 f2fs_bug_on(!sbi->total_valid_node_count);
771 f2fs_bug_on(!inode->i_blocks);
774 sbi->total_valid_node_count--;
775 sbi->total_valid_block_count--;
777 spin_unlock(&sbi->stat_lock);
780 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
783 spin_lock(&sbi->stat_lock);
784 ret = sbi->total_valid_node_count;
785 spin_unlock(&sbi->stat_lock);
789 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
791 spin_lock(&sbi->stat_lock);
792 f2fs_bug_on(sbi->total_valid_inode_count == sbi->total_node_count);
793 sbi->total_valid_inode_count++;
794 spin_unlock(&sbi->stat_lock);
797 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
799 spin_lock(&sbi->stat_lock);
800 f2fs_bug_on(!sbi->total_valid_inode_count);
801 sbi->total_valid_inode_count--;
802 spin_unlock(&sbi->stat_lock);
805 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
808 spin_lock(&sbi->stat_lock);
809 ret = sbi->total_valid_inode_count;
810 spin_unlock(&sbi->stat_lock);
814 static inline void f2fs_put_page(struct page *page, int unlock)
820 f2fs_bug_on(!PageLocked(page));
823 page_cache_release(page);
826 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
829 f2fs_put_page(dn->node_page, 1);
830 if (dn->inode_page && dn->node_page != dn->inode_page)
831 f2fs_put_page(dn->inode_page, 0);
832 dn->node_page = NULL;
833 dn->inode_page = NULL;
836 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
837 size_t size, void (*ctor)(void *))
839 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, ctor);
842 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
847 entry = kmem_cache_alloc(cachep, flags);
856 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
858 static inline bool IS_INODE(struct page *page)
860 struct f2fs_node *p = F2FS_NODE(page);
861 return RAW_IS_INODE(p);
864 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
866 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
869 static inline block_t datablock_addr(struct page *node_page,
872 struct f2fs_node *raw_node;
874 raw_node = F2FS_NODE(node_page);
875 addr_array = blkaddr_in_node(raw_node);
876 return le32_to_cpu(addr_array[offset]);
879 static inline int f2fs_test_bit(unsigned int nr, char *addr)
884 mask = 1 << (7 - (nr & 0x07));
888 static inline int f2fs_set_bit(unsigned int nr, char *addr)
894 mask = 1 << (7 - (nr & 0x07));
900 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
906 mask = 1 << (7 - (nr & 0x07));
912 /* used for f2fs_inode_info->flags */
914 FI_NEW_INODE, /* indicate newly allocated inode */
915 FI_DIRTY_INODE, /* indicate inode is dirty or not */
916 FI_INC_LINK, /* need to increment i_nlink */
917 FI_ACL_MODE, /* indicate acl mode */
918 FI_NO_ALLOC, /* should not allocate any blocks */
919 FI_UPDATE_DIR, /* should update inode block for consistency */
920 FI_DELAY_IPUT, /* used for the recovery */
921 FI_NO_EXTENT, /* not to use the extent cache */
922 FI_INLINE_XATTR, /* used for inline xattr */
923 FI_INLINE_DATA, /* used for inline data*/
926 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
928 set_bit(flag, &fi->flags);
931 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
933 return test_bit(flag, &fi->flags);
936 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
938 clear_bit(flag, &fi->flags);
941 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
943 fi->i_acl_mode = mode;
944 set_inode_flag(fi, FI_ACL_MODE);
947 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
949 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
950 clear_inode_flag(fi, FI_ACL_MODE);
956 static inline void get_inline_info(struct f2fs_inode_info *fi,
957 struct f2fs_inode *ri)
959 if (ri->i_inline & F2FS_INLINE_XATTR)
960 set_inode_flag(fi, FI_INLINE_XATTR);
961 if (ri->i_inline & F2FS_INLINE_DATA)
962 set_inode_flag(fi, FI_INLINE_DATA);
965 static inline void set_raw_inline(struct f2fs_inode_info *fi,
966 struct f2fs_inode *ri)
970 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
971 ri->i_inline |= F2FS_INLINE_XATTR;
972 if (is_inode_flag_set(fi, FI_INLINE_DATA))
973 ri->i_inline |= F2FS_INLINE_DATA;
976 static inline unsigned int addrs_per_inode(struct f2fs_inode_info *fi)
978 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
979 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
980 return DEF_ADDRS_PER_INODE;
983 static inline void *inline_xattr_addr(struct page *page)
985 struct f2fs_inode *ri;
986 ri = (struct f2fs_inode *)page_address(page);
987 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
988 F2FS_INLINE_XATTR_ADDRS]);
991 static inline int inline_xattr_size(struct inode *inode)
993 if (is_inode_flag_set(F2FS_I(inode), FI_INLINE_XATTR))
994 return F2FS_INLINE_XATTR_ADDRS << 2;
999 static inline int f2fs_has_inline_data(struct inode *inode)
1001 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_DATA);
1004 static inline void *inline_data_addr(struct page *page)
1006 struct f2fs_inode *ri;
1007 ri = (struct f2fs_inode *)page_address(page);
1008 return (void *)&(ri->i_addr[1]);
1011 static inline int f2fs_readonly(struct super_block *sb)
1013 return sb->s_flags & MS_RDONLY;
1019 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
1020 void truncate_data_blocks(struct dnode_of_data *);
1021 int truncate_blocks(struct inode *, u64);
1022 void f2fs_truncate(struct inode *);
1023 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1024 int f2fs_setattr(struct dentry *, struct iattr *);
1025 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
1026 int truncate_data_blocks_range(struct dnode_of_data *, int);
1027 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
1028 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
1033 void f2fs_set_inode_flags(struct inode *);
1034 struct inode *f2fs_iget(struct super_block *, unsigned long);
1035 int try_to_free_nats(struct f2fs_sb_info *, int);
1036 void update_inode(struct inode *, struct page *);
1037 int update_inode_page(struct inode *);
1038 int f2fs_write_inode(struct inode *, struct writeback_control *);
1039 void f2fs_evict_inode(struct inode *);
1044 struct dentry *f2fs_get_parent(struct dentry *child);
1049 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
1051 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
1052 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
1053 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
1054 struct page *, struct inode *);
1055 int update_dent_inode(struct inode *, const struct qstr *);
1056 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
1057 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
1058 int f2fs_make_empty(struct inode *, struct inode *);
1059 bool f2fs_empty_dir(struct inode *);
1061 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
1063 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
1070 int f2fs_sync_fs(struct super_block *, int);
1071 extern __printf(3, 4)
1072 void f2fs_msg(struct super_block *, const char *, const char *, ...);
1077 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
1082 struct dnode_of_data;
1085 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
1086 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
1087 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
1088 int truncate_inode_blocks(struct inode *, pgoff_t);
1089 int truncate_xattr_node(struct inode *, struct page *);
1090 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
1091 void remove_inode_page(struct inode *);
1092 struct page *new_inode_page(struct inode *, const struct qstr *);
1093 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
1094 void ra_node_page(struct f2fs_sb_info *, nid_t);
1095 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
1096 struct page *get_node_page_ra(struct page *, int);
1097 void sync_inode_page(struct dnode_of_data *);
1098 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
1099 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
1100 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
1101 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
1102 void recover_node_page(struct f2fs_sb_info *, struct page *,
1103 struct f2fs_summary *, struct node_info *, block_t);
1104 int recover_inode_page(struct f2fs_sb_info *, struct page *);
1105 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
1106 struct f2fs_summary_block *);
1107 void flush_nat_entries(struct f2fs_sb_info *);
1108 int build_node_manager(struct f2fs_sb_info *);
1109 void destroy_node_manager(struct f2fs_sb_info *);
1110 int __init create_node_manager_caches(void);
1111 void destroy_node_manager_caches(void);
1116 void f2fs_balance_fs(struct f2fs_sb_info *);
1117 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
1118 void invalidate_blocks(struct f2fs_sb_info *, block_t);
1119 void clear_prefree_segments(struct f2fs_sb_info *);
1120 int npages_for_summary_flush(struct f2fs_sb_info *);
1121 void allocate_new_segments(struct f2fs_sb_info *);
1122 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1123 void write_meta_page(struct f2fs_sb_info *, struct page *);
1124 void write_node_page(struct f2fs_sb_info *, struct page *,
1125 struct f2fs_io_info *, unsigned int, block_t, block_t *);
1126 void write_data_page(struct page *, struct dnode_of_data *, block_t *,
1127 struct f2fs_io_info *);
1128 void rewrite_data_page(struct page *, block_t, struct f2fs_io_info *);
1129 void recover_data_page(struct f2fs_sb_info *, struct page *,
1130 struct f2fs_summary *, block_t, block_t);
1131 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1132 struct f2fs_summary *, block_t, block_t);
1133 void allocate_data_block(struct f2fs_sb_info *, struct page *,
1134 block_t, block_t *, struct f2fs_summary *, int);
1135 void f2fs_wait_on_page_writeback(struct page *, enum page_type, bool);
1136 void write_data_summaries(struct f2fs_sb_info *, block_t);
1137 void write_node_summaries(struct f2fs_sb_info *, block_t);
1138 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1139 int, unsigned int, int);
1140 void flush_sit_entries(struct f2fs_sb_info *);
1141 int build_segment_manager(struct f2fs_sb_info *);
1142 void destroy_segment_manager(struct f2fs_sb_info *);
1143 int __init create_segment_manager_caches(void);
1144 void destroy_segment_manager_caches(void);
1149 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1150 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1151 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1152 int acquire_orphan_inode(struct f2fs_sb_info *);
1153 void release_orphan_inode(struct f2fs_sb_info *);
1154 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1155 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1156 void recover_orphan_inodes(struct f2fs_sb_info *);
1157 int get_valid_checkpoint(struct f2fs_sb_info *);
1158 void set_dirty_dir_page(struct inode *, struct page *);
1159 void add_dirty_dir_inode(struct inode *);
1160 void remove_dirty_dir_inode(struct inode *);
1161 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1162 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1163 void write_checkpoint(struct f2fs_sb_info *, bool);
1164 void init_orphan_info(struct f2fs_sb_info *);
1165 int __init create_checkpoint_caches(void);
1166 void destroy_checkpoint_caches(void);
1171 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
1172 int f2fs_submit_page_bio(struct f2fs_sb_info *, struct page *, block_t, int);
1173 void f2fs_submit_page_mbio(struct f2fs_sb_info *, struct page *, block_t,
1174 struct f2fs_io_info *);
1175 int reserve_new_block(struct dnode_of_data *);
1176 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
1177 void update_extent_cache(block_t, struct dnode_of_data *);
1178 struct page *find_data_page(struct inode *, pgoff_t, bool);
1179 struct page *get_lock_data_page(struct inode *, pgoff_t);
1180 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1181 int do_write_data_page(struct page *, struct f2fs_io_info *);
1186 int start_gc_thread(struct f2fs_sb_info *);
1187 void stop_gc_thread(struct f2fs_sb_info *);
1188 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
1189 int f2fs_gc(struct f2fs_sb_info *);
1190 void build_gc_manager(struct f2fs_sb_info *);
1191 int __init create_gc_caches(void);
1192 void destroy_gc_caches(void);
1197 int recover_fsync_data(struct f2fs_sb_info *);
1198 bool space_for_roll_forward(struct f2fs_sb_info *);
1203 #ifdef CONFIG_F2FS_STAT_FS
1204 struct f2fs_stat_info {
1205 struct list_head stat_list;
1206 struct f2fs_sb_info *sbi;
1207 struct mutex stat_lock;
1208 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1209 int main_area_segs, main_area_sections, main_area_zones;
1210 int hit_ext, total_ext;
1211 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1212 int nats, sits, fnids;
1213 int total_count, utilization;
1214 int bg_gc, inline_inode;
1215 unsigned int valid_count, valid_node_count, valid_inode_count;
1216 unsigned int bimodal, avg_vblocks;
1217 int util_free, util_valid, util_invalid;
1218 int rsvd_segs, overp_segs;
1219 int dirty_count, node_pages, meta_pages;
1220 int prefree_count, call_count;
1221 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1222 int tot_blks, data_blks, node_blks;
1223 int curseg[NR_CURSEG_TYPE];
1224 int cursec[NR_CURSEG_TYPE];
1225 int curzone[NR_CURSEG_TYPE];
1227 unsigned int segment_count[2];
1228 unsigned int block_count[2];
1229 unsigned base_mem, cache_mem;
1232 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
1234 return (struct f2fs_stat_info*)sbi->stat_info;
1237 #define stat_inc_call_count(si) ((si)->call_count++)
1238 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1239 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1240 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1241 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1242 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1243 #define stat_inc_inline_inode(inode) \
1245 if (f2fs_has_inline_data(inode)) \
1246 ((F2FS_SB(inode->i_sb))->inline_inode++); \
1248 #define stat_dec_inline_inode(inode) \
1250 if (f2fs_has_inline_data(inode)) \
1251 ((F2FS_SB(inode->i_sb))->inline_inode--); \
1254 #define stat_inc_seg_type(sbi, curseg) \
1255 ((sbi)->segment_count[(curseg)->alloc_type]++)
1256 #define stat_inc_block_count(sbi, curseg) \
1257 ((sbi)->block_count[(curseg)->alloc_type]++)
1259 #define stat_inc_seg_count(sbi, type) \
1261 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1263 if (type == SUM_TYPE_DATA) \
1269 #define stat_inc_tot_blk_count(si, blks) \
1270 (si->tot_blks += (blks))
1272 #define stat_inc_data_blk_count(sbi, blks) \
1274 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1275 stat_inc_tot_blk_count(si, blks); \
1276 si->data_blks += (blks); \
1279 #define stat_inc_node_blk_count(sbi, blks) \
1281 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1282 stat_inc_tot_blk_count(si, blks); \
1283 si->node_blks += (blks); \
1286 int f2fs_build_stats(struct f2fs_sb_info *);
1287 void f2fs_destroy_stats(struct f2fs_sb_info *);
1288 void __init f2fs_create_root_stats(void);
1289 void f2fs_destroy_root_stats(void);
1291 #define stat_inc_call_count(si)
1292 #define stat_inc_bggc_count(si)
1293 #define stat_inc_dirty_dir(sbi)
1294 #define stat_dec_dirty_dir(sbi)
1295 #define stat_inc_total_hit(sb)
1296 #define stat_inc_read_hit(sb)
1297 #define stat_inc_inline_inode(inode)
1298 #define stat_dec_inline_inode(inode)
1299 #define stat_inc_seg_type(sbi, curseg)
1300 #define stat_inc_block_count(sbi, curseg)
1301 #define stat_inc_seg_count(si, type)
1302 #define stat_inc_tot_blk_count(si, blks)
1303 #define stat_inc_data_blk_count(si, blks)
1304 #define stat_inc_node_blk_count(sbi, blks)
1306 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1307 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1308 static inline void __init f2fs_create_root_stats(void) { }
1309 static inline void f2fs_destroy_root_stats(void) { }
1312 extern const struct file_operations f2fs_dir_operations;
1313 extern const struct file_operations f2fs_file_operations;
1314 extern const struct inode_operations f2fs_file_inode_operations;
1315 extern const struct address_space_operations f2fs_dblock_aops;
1316 extern const struct address_space_operations f2fs_node_aops;
1317 extern const struct address_space_operations f2fs_meta_aops;
1318 extern const struct inode_operations f2fs_dir_inode_operations;
1319 extern const struct inode_operations f2fs_symlink_inode_operations;
1320 extern const struct inode_operations f2fs_special_inode_operations;
1325 bool f2fs_may_inline(struct inode *);
1326 int f2fs_read_inline_data(struct inode *, struct page *);
1327 int f2fs_convert_inline_data(struct inode *, pgoff_t);
1328 int f2fs_write_inline_data(struct inode *, struct page *, unsigned int);
1329 int recover_inline_data(struct inode *, struct page *);