4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
14 #include <linux/types.h>
15 #include <linux/page-flags.h>
16 #include <linux/buffer_head.h>
17 #include <linux/slab.h>
18 #include <linux/crc32.h>
19 #include <linux/magic.h>
20 #include <linux/kobject.h>
21 #include <linux/sched.h>
23 #ifdef CONFIG_F2FS_CHECK_FS
24 #define f2fs_bug_on(condition) BUG_ON(condition)
25 #define f2fs_down_write(x, y) down_write_nest_lock(x, y)
27 #define f2fs_bug_on(condition)
28 #define f2fs_down_write(x, y) down_write(x)
34 #define F2FS_MOUNT_BG_GC 0x00000001
35 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
36 #define F2FS_MOUNT_DISCARD 0x00000004
37 #define F2FS_MOUNT_NOHEAP 0x00000008
38 #define F2FS_MOUNT_XATTR_USER 0x00000010
39 #define F2FS_MOUNT_POSIX_ACL 0x00000020
40 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
41 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
42 #define F2FS_MOUNT_INLINE_DATA 0x00000100
44 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
45 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
46 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
48 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
49 typecheck(unsigned long long, b) && \
50 ((long long)((a) - (b)) > 0))
52 typedef u32 block_t; /*
53 * should not change u32, since it is the on-disk block
54 * address format, __le32.
58 struct f2fs_mount_info {
62 #define CRCPOLY_LE 0xedb88320
64 static inline __u32 f2fs_crc32(void *buf, size_t len)
66 unsigned char *p = (unsigned char *)buf;
67 __u32 crc = F2FS_SUPER_MAGIC;
72 for (i = 0; i < 8; i++)
73 crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);
78 static inline bool f2fs_crc_valid(__u32 blk_crc, void *buf, size_t buf_size)
80 return f2fs_crc32(buf, buf_size) == blk_crc;
84 * For checkpoint manager
92 * For CP/NAT/SIT readahead
100 /* for the list of orphan inodes */
101 struct orphan_inode_entry {
102 struct list_head list; /* list head */
103 nid_t ino; /* inode number */
106 /* for the list of directory inodes */
107 struct dir_inode_entry {
108 struct list_head list; /* list head */
109 struct inode *inode; /* vfs inode pointer */
112 /* for the list of blockaddresses to be discarded */
113 struct discard_entry {
114 struct list_head list; /* list head */
115 block_t blkaddr; /* block address to be discarded */
116 int len; /* # of consecutive blocks of the discard */
119 /* for the list of fsync inodes, used only during recovery */
120 struct fsync_inode_entry {
121 struct list_head list; /* list head */
122 struct inode *inode; /* vfs inode pointer */
123 block_t blkaddr; /* block address locating the last inode */
126 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
127 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
129 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
130 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
131 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
132 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
134 static inline int update_nats_in_cursum(struct f2fs_summary_block *rs, int i)
136 int before = nats_in_cursum(rs);
137 rs->n_nats = cpu_to_le16(before + i);
141 static inline int update_sits_in_cursum(struct f2fs_summary_block *rs, int i)
143 int before = sits_in_cursum(rs);
144 rs->n_sits = cpu_to_le16(before + i);
151 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
152 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
154 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
156 * ioctl commands in 32 bit emulation
158 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
159 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
163 * For INODE and NODE manager
166 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
167 * as its node offset to distinguish from index node blocks.
168 * But some bits are used to mark the node block.
170 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
173 ALLOC_NODE, /* allocate a new node page if needed */
174 LOOKUP_NODE, /* look up a node without readahead */
176 * look up a node with readahead called
181 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
183 /* for in-memory extent cache entry */
184 #define F2FS_MIN_EXTENT_LEN 16 /* minimum extent length */
187 rwlock_t ext_lock; /* rwlock for consistency */
188 unsigned int fofs; /* start offset in a file */
189 u32 blk_addr; /* start block address of the extent */
190 unsigned int len; /* length of the extent */
194 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
196 #define FADVISE_COLD_BIT 0x01
197 #define FADVISE_LOST_PINO_BIT 0x02
199 struct f2fs_inode_info {
200 struct inode vfs_inode; /* serve a vfs inode */
201 unsigned long i_flags; /* keep an inode flags for ioctl */
202 unsigned char i_advise; /* use to give file attribute hints */
203 unsigned int i_current_depth; /* use only in directory structure */
204 unsigned int i_pino; /* parent inode number */
205 umode_t i_acl_mode; /* keep file acl mode temporarily */
207 /* Use below internally in f2fs*/
208 unsigned long flags; /* use to pass per-file flags */
209 atomic_t dirty_dents; /* # of dirty dentry pages */
210 f2fs_hash_t chash; /* hash value of given file name */
211 unsigned int clevel; /* maximum level of given file name */
212 nid_t i_xattr_nid; /* node id that contains xattrs */
213 unsigned long long xattr_ver; /* cp version of xattr modification */
214 struct extent_info ext; /* in-memory extent cache entry */
217 static inline void get_extent_info(struct extent_info *ext,
218 struct f2fs_extent i_ext)
220 write_lock(&ext->ext_lock);
221 ext->fofs = le32_to_cpu(i_ext.fofs);
222 ext->blk_addr = le32_to_cpu(i_ext.blk_addr);
223 ext->len = le32_to_cpu(i_ext.len);
224 write_unlock(&ext->ext_lock);
227 static inline void set_raw_extent(struct extent_info *ext,
228 struct f2fs_extent *i_ext)
230 read_lock(&ext->ext_lock);
231 i_ext->fofs = cpu_to_le32(ext->fofs);
232 i_ext->blk_addr = cpu_to_le32(ext->blk_addr);
233 i_ext->len = cpu_to_le32(ext->len);
234 read_unlock(&ext->ext_lock);
237 struct f2fs_nm_info {
238 block_t nat_blkaddr; /* base disk address of NAT */
239 nid_t max_nid; /* maximum possible node ids */
240 nid_t next_scan_nid; /* the next nid to be scanned */
242 /* NAT cache management */
243 struct radix_tree_root nat_root;/* root of the nat entry cache */
244 rwlock_t nat_tree_lock; /* protect nat_tree_lock */
245 unsigned int nat_cnt; /* the # of cached nat entries */
246 struct list_head nat_entries; /* cached nat entry list (clean) */
247 struct list_head dirty_nat_entries; /* cached nat entry list (dirty) */
249 /* free node ids management */
250 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
251 struct list_head free_nid_list; /* a list for free nids */
252 spinlock_t free_nid_list_lock; /* protect free nid list */
253 unsigned int fcnt; /* the number of free node id */
254 struct mutex build_lock; /* lock for build free nids */
257 char *nat_bitmap; /* NAT bitmap pointer */
258 int bitmap_size; /* bitmap size */
262 * this structure is used as one of function parameters.
263 * all the information are dedicated to a given direct node block determined
264 * by the data offset in a file.
266 struct dnode_of_data {
267 struct inode *inode; /* vfs inode pointer */
268 struct page *inode_page; /* its inode page, NULL is possible */
269 struct page *node_page; /* cached direct node page */
270 nid_t nid; /* node id of the direct node block */
271 unsigned int ofs_in_node; /* data offset in the node page */
272 bool inode_page_locked; /* inode page is locked or not */
273 block_t data_blkaddr; /* block address of the node block */
276 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
277 struct page *ipage, struct page *npage, nid_t nid)
279 memset(dn, 0, sizeof(*dn));
281 dn->inode_page = ipage;
282 dn->node_page = npage;
289 * By default, there are 6 active log areas across the whole main area.
290 * When considering hot and cold data separation to reduce cleaning overhead,
291 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
293 * In the current design, you should not change the numbers intentionally.
294 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
295 * logs individually according to the underlying devices. (default: 6)
296 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
297 * data and 8 for node logs.
299 #define NR_CURSEG_DATA_TYPE (3)
300 #define NR_CURSEG_NODE_TYPE (3)
301 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
304 CURSEG_HOT_DATA = 0, /* directory entry blocks */
305 CURSEG_WARM_DATA, /* data blocks */
306 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
307 CURSEG_HOT_NODE, /* direct node blocks of directory files */
308 CURSEG_WARM_NODE, /* direct node blocks of normal files */
309 CURSEG_COLD_NODE, /* indirect node blocks */
313 struct f2fs_sm_info {
314 struct sit_info *sit_info; /* whole segment information */
315 struct free_segmap_info *free_info; /* free segment information */
316 struct dirty_seglist_info *dirty_info; /* dirty segment information */
317 struct curseg_info *curseg_array; /* active segment information */
319 struct list_head wblist_head; /* list of under-writeback pages */
320 spinlock_t wblist_lock; /* lock for checkpoint */
322 block_t seg0_blkaddr; /* block address of 0'th segment */
323 block_t main_blkaddr; /* start block address of main area */
324 block_t ssa_blkaddr; /* start block address of SSA area */
326 unsigned int segment_count; /* total # of segments */
327 unsigned int main_segments; /* # of segments in main area */
328 unsigned int reserved_segments; /* # of reserved segments */
329 unsigned int ovp_segments; /* # of overprovision segments */
331 /* a threshold to reclaim prefree segments */
332 unsigned int rec_prefree_segments;
334 /* for small discard management */
335 struct list_head discard_list; /* 4KB discard list */
336 int nr_discards; /* # of discards in the list */
337 int max_discards; /* max. discards to be issued */
339 unsigned int ipu_policy; /* in-place-update policy */
340 unsigned int min_ipu_util; /* in-place-update threshold */
347 * COUNT_TYPE for monitoring
349 * f2fs monitors the number of several block types such as on-writeback,
350 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
361 * The below are the page types of bios used in submti_bio().
362 * The available types are:
363 * DATA User data pages. It operates as async mode.
364 * NODE Node pages. It operates as async mode.
365 * META FS metadata pages such as SIT, NAT, CP.
366 * NR_PAGE_TYPE The number of page types.
367 * META_FLUSH Make sure the previous pages are written
368 * with waiting the bio's completion
369 * ... Only can be used with META.
371 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
380 struct f2fs_io_info {
381 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
382 int rw; /* contains R/RS/W/WS with REQ_META/REQ_PRIO */
385 #define is_read_io(rw) (((rw) & 1) == READ)
386 struct f2fs_bio_info {
387 struct f2fs_sb_info *sbi; /* f2fs superblock */
388 struct bio *bio; /* bios to merge */
389 sector_t last_block_in_bio; /* last block number */
390 struct f2fs_io_info fio; /* store buffered io info. */
391 struct mutex io_mutex; /* mutex for bio */
394 struct f2fs_sb_info {
395 struct super_block *sb; /* pointer to VFS super block */
396 struct proc_dir_entry *s_proc; /* proc entry */
397 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
398 struct f2fs_super_block *raw_super; /* raw super block pointer */
399 int s_dirty; /* dirty flag for checkpoint */
401 /* for node-related operations */
402 struct f2fs_nm_info *nm_info; /* node manager */
403 struct inode *node_inode; /* cache node blocks */
405 /* for segment-related operations */
406 struct f2fs_sm_info *sm_info; /* segment manager */
408 /* for bio operations */
409 struct f2fs_bio_info read_io; /* for read bios */
410 struct f2fs_bio_info write_io[NR_PAGE_TYPE]; /* for write bios */
411 struct completion *wait_io; /* for completion bios */
414 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
415 struct inode *meta_inode; /* cache meta blocks */
416 struct mutex cp_mutex; /* checkpoint procedure lock */
417 struct rw_semaphore cp_rwsem; /* blocking FS operations */
418 struct mutex node_write; /* locking node writes */
419 struct mutex writepages; /* mutex for writepages() */
420 bool por_doing; /* recovery is doing or not */
421 wait_queue_head_t cp_wait;
423 /* for orphan inode management */
424 struct list_head orphan_inode_list; /* orphan inode list */
425 spinlock_t orphan_inode_lock; /* for orphan inode list */
426 unsigned int n_orphans; /* # of orphan inodes */
427 unsigned int max_orphans; /* max orphan inodes */
429 /* for directory inode management */
430 struct list_head dir_inode_list; /* dir inode list */
431 spinlock_t dir_inode_lock; /* for dir inode list lock */
433 /* basic file system units */
434 unsigned int log_sectors_per_block; /* log2 sectors per block */
435 unsigned int log_blocksize; /* log2 block size */
436 unsigned int blocksize; /* block size */
437 unsigned int root_ino_num; /* root inode number*/
438 unsigned int node_ino_num; /* node inode number*/
439 unsigned int meta_ino_num; /* meta inode number*/
440 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
441 unsigned int blocks_per_seg; /* blocks per segment */
442 unsigned int segs_per_sec; /* segments per section */
443 unsigned int secs_per_zone; /* sections per zone */
444 unsigned int total_sections; /* total section count */
445 unsigned int total_node_count; /* total node block count */
446 unsigned int total_valid_node_count; /* valid node block count */
447 unsigned int total_valid_inode_count; /* valid inode count */
448 int active_logs; /* # of active logs */
450 block_t user_block_count; /* # of user blocks */
451 block_t total_valid_block_count; /* # of valid blocks */
452 block_t alloc_valid_block_count; /* # of allocated blocks */
453 block_t last_valid_block_count; /* for recovery */
454 u32 s_next_generation; /* for NFS support */
455 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
457 struct f2fs_mount_info mount_opt; /* mount options */
459 /* for cleaning operations */
460 struct mutex gc_mutex; /* mutex for GC */
461 struct f2fs_gc_kthread *gc_thread; /* GC thread */
462 unsigned int cur_victim_sec; /* current victim section num */
464 /* maximum # of trials to find a victim segment for SSR and GC */
465 unsigned int max_victim_search;
468 * for stat information.
469 * one is for the LFS mode, and the other is for the SSR mode.
471 #ifdef CONFIG_F2FS_STAT_FS
472 struct f2fs_stat_info *stat_info; /* FS status information */
473 unsigned int segment_count[2]; /* # of allocated segments */
474 unsigned int block_count[2]; /* # of allocated blocks */
475 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
476 int inline_inode; /* # of inline_data inodes */
477 int bg_gc; /* background gc calls */
478 unsigned int n_dirty_dirs; /* # of dir inodes */
480 unsigned int last_victim[2]; /* last victim segment # */
481 spinlock_t stat_lock; /* lock for stat operations */
483 /* For sysfs suppport */
484 struct kobject s_kobj;
485 struct completion s_kobj_unregister;
491 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
493 return container_of(inode, struct f2fs_inode_info, vfs_inode);
496 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
498 return sb->s_fs_info;
501 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
503 return (struct f2fs_super_block *)(sbi->raw_super);
506 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
508 return (struct f2fs_checkpoint *)(sbi->ckpt);
511 static inline struct f2fs_node *F2FS_NODE(struct page *page)
513 return (struct f2fs_node *)page_address(page);
516 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
518 return &((struct f2fs_node *)page_address(page))->i;
521 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
523 return (struct f2fs_nm_info *)(sbi->nm_info);
526 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
528 return (struct f2fs_sm_info *)(sbi->sm_info);
531 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
533 return (struct sit_info *)(SM_I(sbi)->sit_info);
536 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
538 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
541 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
543 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
546 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
548 return sbi->meta_inode->i_mapping;
551 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
553 return sbi->node_inode->i_mapping;
556 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
561 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
566 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
568 return le64_to_cpu(cp->checkpoint_ver);
571 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
573 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
574 return ckpt_flags & f;
577 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
579 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
581 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
584 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
586 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
588 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
591 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
593 down_read(&sbi->cp_rwsem);
596 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
598 up_read(&sbi->cp_rwsem);
601 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
603 f2fs_down_write(&sbi->cp_rwsem, &sbi->cp_mutex);
606 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
608 up_write(&sbi->cp_rwsem);
612 * Check whether the given nid is within node id range.
614 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
616 WARN_ON((nid >= NM_I(sbi)->max_nid));
617 if (unlikely(nid >= NM_I(sbi)->max_nid))
622 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
625 * Check whether the inode has blocks or not
627 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
629 if (F2FS_I(inode)->i_xattr_nid)
630 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1;
632 return inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS;
635 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
636 struct inode *inode, blkcnt_t count)
638 block_t valid_block_count;
640 spin_lock(&sbi->stat_lock);
642 sbi->total_valid_block_count + (block_t)count;
643 if (unlikely(valid_block_count > sbi->user_block_count)) {
644 spin_unlock(&sbi->stat_lock);
647 inode->i_blocks += count;
648 sbi->total_valid_block_count = valid_block_count;
649 sbi->alloc_valid_block_count += (block_t)count;
650 spin_unlock(&sbi->stat_lock);
654 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
658 spin_lock(&sbi->stat_lock);
659 f2fs_bug_on(sbi->total_valid_block_count < (block_t) count);
660 f2fs_bug_on(inode->i_blocks < count);
661 inode->i_blocks -= count;
662 sbi->total_valid_block_count -= (block_t)count;
663 spin_unlock(&sbi->stat_lock);
666 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
668 atomic_inc(&sbi->nr_pages[count_type]);
669 F2FS_SET_SB_DIRT(sbi);
672 static inline void inode_inc_dirty_dents(struct inode *inode)
674 inc_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
675 atomic_inc(&F2FS_I(inode)->dirty_dents);
678 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
680 atomic_dec(&sbi->nr_pages[count_type]);
683 static inline void inode_dec_dirty_dents(struct inode *inode)
685 if (!S_ISDIR(inode->i_mode))
688 dec_page_count(F2FS_SB(inode->i_sb), F2FS_DIRTY_DENTS);
689 atomic_dec(&F2FS_I(inode)->dirty_dents);
692 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
694 return atomic_read(&sbi->nr_pages[count_type]);
697 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
699 unsigned int pages_per_sec = sbi->segs_per_sec *
700 (1 << sbi->log_blocks_per_seg);
701 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
702 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
705 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
707 return sbi->total_valid_block_count;
710 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
712 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
714 /* return NAT or SIT bitmap */
715 if (flag == NAT_BITMAP)
716 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
717 else if (flag == SIT_BITMAP)
718 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
723 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
725 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
726 int offset = (flag == NAT_BITMAP) ?
727 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
728 return &ckpt->sit_nat_version_bitmap + offset;
731 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
734 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
735 unsigned long long ckpt_version = cur_cp_version(ckpt);
737 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
740 * odd numbered checkpoint should at cp segment 0
741 * and even segent must be at cp segment 1
743 if (!(ckpt_version & 1))
744 start_addr += sbi->blocks_per_seg;
749 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
751 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
754 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
757 block_t valid_block_count;
758 unsigned int valid_node_count;
760 spin_lock(&sbi->stat_lock);
762 valid_block_count = sbi->total_valid_block_count + 1;
763 if (unlikely(valid_block_count > sbi->user_block_count)) {
764 spin_unlock(&sbi->stat_lock);
768 valid_node_count = sbi->total_valid_node_count + 1;
769 if (unlikely(valid_node_count > sbi->total_node_count)) {
770 spin_unlock(&sbi->stat_lock);
777 sbi->alloc_valid_block_count++;
778 sbi->total_valid_node_count++;
779 sbi->total_valid_block_count++;
780 spin_unlock(&sbi->stat_lock);
785 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
788 spin_lock(&sbi->stat_lock);
790 f2fs_bug_on(!sbi->total_valid_block_count);
791 f2fs_bug_on(!sbi->total_valid_node_count);
792 f2fs_bug_on(!inode->i_blocks);
795 sbi->total_valid_node_count--;
796 sbi->total_valid_block_count--;
798 spin_unlock(&sbi->stat_lock);
801 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
803 return sbi->total_valid_node_count;
806 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
808 spin_lock(&sbi->stat_lock);
809 f2fs_bug_on(sbi->total_valid_inode_count == sbi->total_node_count);
810 sbi->total_valid_inode_count++;
811 spin_unlock(&sbi->stat_lock);
814 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
816 spin_lock(&sbi->stat_lock);
817 f2fs_bug_on(!sbi->total_valid_inode_count);
818 sbi->total_valid_inode_count--;
819 spin_unlock(&sbi->stat_lock);
822 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
824 return sbi->total_valid_inode_count;
827 static inline void f2fs_put_page(struct page *page, int unlock)
833 f2fs_bug_on(!PageLocked(page));
836 page_cache_release(page);
839 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
842 f2fs_put_page(dn->node_page, 1);
843 if (dn->inode_page && dn->node_page != dn->inode_page)
844 f2fs_put_page(dn->inode_page, 0);
845 dn->node_page = NULL;
846 dn->inode_page = NULL;
849 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
850 size_t size, void (*ctor)(void *))
852 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, ctor);
855 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
860 entry = kmem_cache_alloc(cachep, flags);
869 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
871 static inline bool IS_INODE(struct page *page)
873 struct f2fs_node *p = F2FS_NODE(page);
874 return RAW_IS_INODE(p);
877 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
879 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
882 static inline block_t datablock_addr(struct page *node_page,
885 struct f2fs_node *raw_node;
887 raw_node = F2FS_NODE(node_page);
888 addr_array = blkaddr_in_node(raw_node);
889 return le32_to_cpu(addr_array[offset]);
892 static inline int f2fs_test_bit(unsigned int nr, char *addr)
897 mask = 1 << (7 - (nr & 0x07));
901 static inline int f2fs_set_bit(unsigned int nr, char *addr)
907 mask = 1 << (7 - (nr & 0x07));
913 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
919 mask = 1 << (7 - (nr & 0x07));
925 /* used for f2fs_inode_info->flags */
927 FI_NEW_INODE, /* indicate newly allocated inode */
928 FI_DIRTY_INODE, /* indicate inode is dirty or not */
929 FI_INC_LINK, /* need to increment i_nlink */
930 FI_ACL_MODE, /* indicate acl mode */
931 FI_NO_ALLOC, /* should not allocate any blocks */
932 FI_UPDATE_DIR, /* should update inode block for consistency */
933 FI_DELAY_IPUT, /* used for the recovery */
934 FI_NO_EXTENT, /* not to use the extent cache */
935 FI_INLINE_XATTR, /* used for inline xattr */
936 FI_INLINE_DATA, /* used for inline data*/
939 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
941 set_bit(flag, &fi->flags);
944 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
946 return test_bit(flag, &fi->flags);
949 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
951 clear_bit(flag, &fi->flags);
954 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
956 fi->i_acl_mode = mode;
957 set_inode_flag(fi, FI_ACL_MODE);
960 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
962 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
963 clear_inode_flag(fi, FI_ACL_MODE);
969 static inline void get_inline_info(struct f2fs_inode_info *fi,
970 struct f2fs_inode *ri)
972 if (ri->i_inline & F2FS_INLINE_XATTR)
973 set_inode_flag(fi, FI_INLINE_XATTR);
974 if (ri->i_inline & F2FS_INLINE_DATA)
975 set_inode_flag(fi, FI_INLINE_DATA);
978 static inline void set_raw_inline(struct f2fs_inode_info *fi,
979 struct f2fs_inode *ri)
983 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
984 ri->i_inline |= F2FS_INLINE_XATTR;
985 if (is_inode_flag_set(fi, FI_INLINE_DATA))
986 ri->i_inline |= F2FS_INLINE_DATA;
989 static inline unsigned int addrs_per_inode(struct f2fs_inode_info *fi)
991 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
992 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
993 return DEF_ADDRS_PER_INODE;
996 static inline void *inline_xattr_addr(struct page *page)
998 struct f2fs_inode *ri;
999 ri = (struct f2fs_inode *)page_address(page);
1000 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
1001 F2FS_INLINE_XATTR_ADDRS]);
1004 static inline int inline_xattr_size(struct inode *inode)
1006 if (is_inode_flag_set(F2FS_I(inode), FI_INLINE_XATTR))
1007 return F2FS_INLINE_XATTR_ADDRS << 2;
1012 static inline int f2fs_has_inline_data(struct inode *inode)
1014 return is_inode_flag_set(F2FS_I(inode), FI_INLINE_DATA);
1017 static inline void *inline_data_addr(struct page *page)
1019 struct f2fs_inode *ri;
1020 ri = (struct f2fs_inode *)page_address(page);
1021 return (void *)&(ri->i_addr[1]);
1024 static inline int f2fs_readonly(struct super_block *sb)
1026 return sb->s_flags & MS_RDONLY;
1029 static inline void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi)
1031 set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
1032 sbi->sb->s_flags |= MS_RDONLY;
1035 #define get_inode_mode(i) \
1036 ((is_inode_flag_set(F2FS_I(i), FI_ACL_MODE)) ? \
1037 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
1042 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
1043 void truncate_data_blocks(struct dnode_of_data *);
1044 int truncate_blocks(struct inode *, u64);
1045 void f2fs_truncate(struct inode *);
1046 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1047 int f2fs_setattr(struct dentry *, struct iattr *);
1048 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
1049 int truncate_data_blocks_range(struct dnode_of_data *, int);
1050 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
1051 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
1056 void f2fs_set_inode_flags(struct inode *);
1057 struct inode *f2fs_iget(struct super_block *, unsigned long);
1058 int try_to_free_nats(struct f2fs_sb_info *, int);
1059 void update_inode(struct inode *, struct page *);
1060 void update_inode_page(struct inode *);
1061 int f2fs_write_inode(struct inode *, struct writeback_control *);
1062 void f2fs_evict_inode(struct inode *);
1067 struct dentry *f2fs_get_parent(struct dentry *child);
1072 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
1074 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
1075 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
1076 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
1077 struct page *, struct inode *);
1078 int update_dent_inode(struct inode *, const struct qstr *);
1079 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
1080 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
1081 int f2fs_make_empty(struct inode *, struct inode *);
1082 bool f2fs_empty_dir(struct inode *);
1084 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
1086 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
1093 int f2fs_sync_fs(struct super_block *, int);
1094 extern __printf(3, 4)
1095 void f2fs_msg(struct super_block *, const char *, const char *, ...);
1100 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
1105 struct dnode_of_data;
1108 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
1109 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
1110 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
1111 int truncate_inode_blocks(struct inode *, pgoff_t);
1112 int truncate_xattr_node(struct inode *, struct page *);
1113 int wait_on_node_pages_writeback(struct f2fs_sb_info *, nid_t);
1114 void remove_inode_page(struct inode *);
1115 struct page *new_inode_page(struct inode *, const struct qstr *);
1116 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
1117 void ra_node_page(struct f2fs_sb_info *, nid_t);
1118 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
1119 struct page *get_node_page_ra(struct page *, int);
1120 void sync_inode_page(struct dnode_of_data *);
1121 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
1122 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
1123 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
1124 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
1125 void recover_node_page(struct f2fs_sb_info *, struct page *,
1126 struct f2fs_summary *, struct node_info *, block_t);
1127 bool recover_xattr_data(struct inode *, struct page *, block_t);
1128 int recover_inode_page(struct f2fs_sb_info *, struct page *);
1129 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
1130 struct f2fs_summary_block *);
1131 void flush_nat_entries(struct f2fs_sb_info *);
1132 int build_node_manager(struct f2fs_sb_info *);
1133 void destroy_node_manager(struct f2fs_sb_info *);
1134 int __init create_node_manager_caches(void);
1135 void destroy_node_manager_caches(void);
1140 void f2fs_balance_fs(struct f2fs_sb_info *);
1141 void f2fs_balance_fs_bg(struct f2fs_sb_info *);
1142 void invalidate_blocks(struct f2fs_sb_info *, block_t);
1143 void refresh_sit_entry(struct f2fs_sb_info *, block_t, block_t);
1144 void clear_prefree_segments(struct f2fs_sb_info *);
1145 int npages_for_summary_flush(struct f2fs_sb_info *);
1146 void allocate_new_segments(struct f2fs_sb_info *);
1147 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1148 void write_meta_page(struct f2fs_sb_info *, struct page *);
1149 void write_node_page(struct f2fs_sb_info *, struct page *,
1150 struct f2fs_io_info *, unsigned int, block_t, block_t *);
1151 void write_data_page(struct page *, struct dnode_of_data *, block_t *,
1152 struct f2fs_io_info *);
1153 void rewrite_data_page(struct page *, block_t, struct f2fs_io_info *);
1154 void recover_data_page(struct f2fs_sb_info *, struct page *,
1155 struct f2fs_summary *, block_t, block_t);
1156 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1157 struct f2fs_summary *, block_t, block_t);
1158 void allocate_data_block(struct f2fs_sb_info *, struct page *,
1159 block_t, block_t *, struct f2fs_summary *, int);
1160 void f2fs_wait_on_page_writeback(struct page *, enum page_type);
1161 void write_data_summaries(struct f2fs_sb_info *, block_t);
1162 void write_node_summaries(struct f2fs_sb_info *, block_t);
1163 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1164 int, unsigned int, int);
1165 void flush_sit_entries(struct f2fs_sb_info *);
1166 int build_segment_manager(struct f2fs_sb_info *);
1167 void destroy_segment_manager(struct f2fs_sb_info *);
1168 int __init create_segment_manager_caches(void);
1169 void destroy_segment_manager_caches(void);
1174 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1175 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1176 int ra_meta_pages(struct f2fs_sb_info *, int, int, int);
1177 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1178 int acquire_orphan_inode(struct f2fs_sb_info *);
1179 void release_orphan_inode(struct f2fs_sb_info *);
1180 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1181 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1182 void recover_orphan_inodes(struct f2fs_sb_info *);
1183 int get_valid_checkpoint(struct f2fs_sb_info *);
1184 void set_dirty_dir_page(struct inode *, struct page *);
1185 void add_dirty_dir_inode(struct inode *);
1186 void remove_dirty_dir_inode(struct inode *);
1187 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1188 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1189 void write_checkpoint(struct f2fs_sb_info *, bool);
1190 void init_orphan_info(struct f2fs_sb_info *);
1191 int __init create_checkpoint_caches(void);
1192 void destroy_checkpoint_caches(void);
1197 void f2fs_submit_merged_bio(struct f2fs_sb_info *, enum page_type, int);
1198 int f2fs_submit_page_bio(struct f2fs_sb_info *, struct page *, block_t, int);
1199 void f2fs_submit_page_mbio(struct f2fs_sb_info *, struct page *, block_t,
1200 struct f2fs_io_info *);
1201 int reserve_new_block(struct dnode_of_data *);
1202 int f2fs_reserve_block(struct dnode_of_data *, pgoff_t);
1203 void update_extent_cache(block_t, struct dnode_of_data *);
1204 struct page *find_data_page(struct inode *, pgoff_t, bool);
1205 struct page *get_lock_data_page(struct inode *, pgoff_t);
1206 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1207 int do_write_data_page(struct page *, struct f2fs_io_info *);
1212 int start_gc_thread(struct f2fs_sb_info *);
1213 void stop_gc_thread(struct f2fs_sb_info *);
1214 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
1215 int f2fs_gc(struct f2fs_sb_info *);
1216 void build_gc_manager(struct f2fs_sb_info *);
1217 int __init create_gc_caches(void);
1218 void destroy_gc_caches(void);
1223 int recover_fsync_data(struct f2fs_sb_info *);
1224 bool space_for_roll_forward(struct f2fs_sb_info *);
1229 #ifdef CONFIG_F2FS_STAT_FS
1230 struct f2fs_stat_info {
1231 struct list_head stat_list;
1232 struct f2fs_sb_info *sbi;
1233 struct mutex stat_lock;
1234 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1235 int main_area_segs, main_area_sections, main_area_zones;
1236 int hit_ext, total_ext;
1237 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1238 int nats, sits, fnids;
1239 int total_count, utilization;
1240 int bg_gc, inline_inode;
1241 unsigned int valid_count, valid_node_count, valid_inode_count;
1242 unsigned int bimodal, avg_vblocks;
1243 int util_free, util_valid, util_invalid;
1244 int rsvd_segs, overp_segs;
1245 int dirty_count, node_pages, meta_pages;
1246 int prefree_count, call_count, cp_count;
1247 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1248 int tot_blks, data_blks, node_blks;
1249 int curseg[NR_CURSEG_TYPE];
1250 int cursec[NR_CURSEG_TYPE];
1251 int curzone[NR_CURSEG_TYPE];
1253 unsigned int segment_count[2];
1254 unsigned int block_count[2];
1255 unsigned base_mem, cache_mem;
1258 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
1260 return (struct f2fs_stat_info *)sbi->stat_info;
1263 #define stat_inc_cp_count(si) ((si)->cp_count++)
1264 #define stat_inc_call_count(si) ((si)->call_count++)
1265 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1266 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1267 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1268 #define stat_inc_total_hit(sb) ((F2FS_SB(sb))->total_hit_ext++)
1269 #define stat_inc_read_hit(sb) ((F2FS_SB(sb))->read_hit_ext++)
1270 #define stat_inc_inline_inode(inode) \
1272 if (f2fs_has_inline_data(inode)) \
1273 ((F2FS_SB(inode->i_sb))->inline_inode++); \
1275 #define stat_dec_inline_inode(inode) \
1277 if (f2fs_has_inline_data(inode)) \
1278 ((F2FS_SB(inode->i_sb))->inline_inode--); \
1281 #define stat_inc_seg_type(sbi, curseg) \
1282 ((sbi)->segment_count[(curseg)->alloc_type]++)
1283 #define stat_inc_block_count(sbi, curseg) \
1284 ((sbi)->block_count[(curseg)->alloc_type]++)
1286 #define stat_inc_seg_count(sbi, type) \
1288 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1290 if (type == SUM_TYPE_DATA) \
1296 #define stat_inc_tot_blk_count(si, blks) \
1297 (si->tot_blks += (blks))
1299 #define stat_inc_data_blk_count(sbi, blks) \
1301 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1302 stat_inc_tot_blk_count(si, blks); \
1303 si->data_blks += (blks); \
1306 #define stat_inc_node_blk_count(sbi, blks) \
1308 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1309 stat_inc_tot_blk_count(si, blks); \
1310 si->node_blks += (blks); \
1313 int f2fs_build_stats(struct f2fs_sb_info *);
1314 void f2fs_destroy_stats(struct f2fs_sb_info *);
1315 void __init f2fs_create_root_stats(void);
1316 void f2fs_destroy_root_stats(void);
1318 #define stat_inc_cp_count(si)
1319 #define stat_inc_call_count(si)
1320 #define stat_inc_bggc_count(si)
1321 #define stat_inc_dirty_dir(sbi)
1322 #define stat_dec_dirty_dir(sbi)
1323 #define stat_inc_total_hit(sb)
1324 #define stat_inc_read_hit(sb)
1325 #define stat_inc_inline_inode(inode)
1326 #define stat_dec_inline_inode(inode)
1327 #define stat_inc_seg_type(sbi, curseg)
1328 #define stat_inc_block_count(sbi, curseg)
1329 #define stat_inc_seg_count(si, type)
1330 #define stat_inc_tot_blk_count(si, blks)
1331 #define stat_inc_data_blk_count(si, blks)
1332 #define stat_inc_node_blk_count(sbi, blks)
1334 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1335 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1336 static inline void __init f2fs_create_root_stats(void) { }
1337 static inline void f2fs_destroy_root_stats(void) { }
1340 extern const struct file_operations f2fs_dir_operations;
1341 extern const struct file_operations f2fs_file_operations;
1342 extern const struct inode_operations f2fs_file_inode_operations;
1343 extern const struct address_space_operations f2fs_dblock_aops;
1344 extern const struct address_space_operations f2fs_node_aops;
1345 extern const struct address_space_operations f2fs_meta_aops;
1346 extern const struct inode_operations f2fs_dir_inode_operations;
1347 extern const struct inode_operations f2fs_symlink_inode_operations;
1348 extern const struct inode_operations f2fs_special_inode_operations;
1353 bool f2fs_may_inline(struct inode *);
1354 int f2fs_read_inline_data(struct inode *, struct page *);
1355 int f2fs_convert_inline_data(struct inode *, pgoff_t);
1356 int f2fs_write_inline_data(struct inode *, struct page *, unsigned int);
1357 int recover_inline_data(struct inode *, struct page *);