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>
26 #define F2FS_MOUNT_BG_GC 0x00000001
27 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
28 #define F2FS_MOUNT_DISCARD 0x00000004
29 #define F2FS_MOUNT_NOHEAP 0x00000008
30 #define F2FS_MOUNT_XATTR_USER 0x00000010
31 #define F2FS_MOUNT_POSIX_ACL 0x00000020
32 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
33 #define F2FS_MOUNT_INLINE_XATTR 0x00000080
35 #define clear_opt(sbi, option) (sbi->mount_opt.opt &= ~F2FS_MOUNT_##option)
36 #define set_opt(sbi, option) (sbi->mount_opt.opt |= F2FS_MOUNT_##option)
37 #define test_opt(sbi, option) (sbi->mount_opt.opt & F2FS_MOUNT_##option)
39 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
40 typecheck(unsigned long long, b) && \
41 ((long long)((a) - (b)) > 0))
43 typedef u32 block_t; /*
44 * should not change u32, since it is the on-disk block
45 * address format, __le32.
49 struct f2fs_mount_info {
53 #define CRCPOLY_LE 0xedb88320
55 static inline __u32 f2fs_crc32(void *buf, size_t len)
57 unsigned char *p = (unsigned char *)buf;
58 __u32 crc = F2FS_SUPER_MAGIC;
63 for (i = 0; i < 8; i++)
64 crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);
69 static inline bool f2fs_crc_valid(__u32 blk_crc, void *buf, size_t buf_size)
71 return f2fs_crc32(buf, buf_size) == blk_crc;
75 * For checkpoint manager
82 /* for the list of orphan inodes */
83 struct orphan_inode_entry {
84 struct list_head list; /* list head */
85 nid_t ino; /* inode number */
88 /* for the list of directory inodes */
89 struct dir_inode_entry {
90 struct list_head list; /* list head */
91 struct inode *inode; /* vfs inode pointer */
94 /* for the list of fsync inodes, used only during recovery */
95 struct fsync_inode_entry {
96 struct list_head list; /* list head */
97 struct inode *inode; /* vfs inode pointer */
98 block_t blkaddr; /* block address locating the last inode */
101 #define nats_in_cursum(sum) (le16_to_cpu(sum->n_nats))
102 #define sits_in_cursum(sum) (le16_to_cpu(sum->n_sits))
104 #define nat_in_journal(sum, i) (sum->nat_j.entries[i].ne)
105 #define nid_in_journal(sum, i) (sum->nat_j.entries[i].nid)
106 #define sit_in_journal(sum, i) (sum->sit_j.entries[i].se)
107 #define segno_in_journal(sum, i) (sum->sit_j.entries[i].segno)
109 static inline int update_nats_in_cursum(struct f2fs_summary_block *rs, int i)
111 int before = nats_in_cursum(rs);
112 rs->n_nats = cpu_to_le16(before + i);
116 static inline int update_sits_in_cursum(struct f2fs_summary_block *rs, int i)
118 int before = sits_in_cursum(rs);
119 rs->n_sits = cpu_to_le16(before + i);
126 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
127 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
129 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
131 * ioctl commands in 32 bit emulation
133 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
134 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
138 * For INODE and NODE manager
141 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
142 * as its node offset to distinguish from index node blocks.
143 * But some bits are used to mark the node block.
145 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
148 ALLOC_NODE, /* allocate a new node page if needed */
149 LOOKUP_NODE, /* look up a node without readahead */
151 * look up a node with readahead called
152 * by get_datablock_ro.
156 #define F2FS_LINK_MAX 32000 /* maximum link count per file */
158 /* for in-memory extent cache entry */
160 rwlock_t ext_lock; /* rwlock for consistency */
161 unsigned int fofs; /* start offset in a file */
162 u32 blk_addr; /* start block address of the extent */
163 unsigned int len; /* length of the extent */
167 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
169 #define FADVISE_COLD_BIT 0x01
170 #define FADVISE_LOST_PINO_BIT 0x02
172 struct f2fs_inode_info {
173 struct inode vfs_inode; /* serve a vfs inode */
174 unsigned long i_flags; /* keep an inode flags for ioctl */
175 unsigned char i_advise; /* use to give file attribute hints */
176 unsigned int i_current_depth; /* use only in directory structure */
177 unsigned int i_pino; /* parent inode number */
178 umode_t i_acl_mode; /* keep file acl mode temporarily */
180 /* Use below internally in f2fs*/
181 unsigned long flags; /* use to pass per-file flags */
182 atomic_t dirty_dents; /* # of dirty dentry pages */
183 f2fs_hash_t chash; /* hash value of given file name */
184 unsigned int clevel; /* maximum level of given file name */
185 nid_t i_xattr_nid; /* node id that contains xattrs */
186 unsigned long long xattr_ver; /* cp version of xattr modification */
187 struct extent_info ext; /* in-memory extent cache entry */
190 static inline void get_extent_info(struct extent_info *ext,
191 struct f2fs_extent i_ext)
193 write_lock(&ext->ext_lock);
194 ext->fofs = le32_to_cpu(i_ext.fofs);
195 ext->blk_addr = le32_to_cpu(i_ext.blk_addr);
196 ext->len = le32_to_cpu(i_ext.len);
197 write_unlock(&ext->ext_lock);
200 static inline void set_raw_extent(struct extent_info *ext,
201 struct f2fs_extent *i_ext)
203 read_lock(&ext->ext_lock);
204 i_ext->fofs = cpu_to_le32(ext->fofs);
205 i_ext->blk_addr = cpu_to_le32(ext->blk_addr);
206 i_ext->len = cpu_to_le32(ext->len);
207 read_unlock(&ext->ext_lock);
210 struct f2fs_nm_info {
211 block_t nat_blkaddr; /* base disk address of NAT */
212 nid_t max_nid; /* maximum possible node ids */
213 nid_t next_scan_nid; /* the next nid to be scanned */
215 /* NAT cache management */
216 struct radix_tree_root nat_root;/* root of the nat entry cache */
217 rwlock_t nat_tree_lock; /* protect nat_tree_lock */
218 unsigned int nat_cnt; /* the # of cached nat entries */
219 struct list_head nat_entries; /* cached nat entry list (clean) */
220 struct list_head dirty_nat_entries; /* cached nat entry list (dirty) */
222 /* free node ids management */
223 struct list_head free_nid_list; /* a list for free nids */
224 spinlock_t free_nid_list_lock; /* protect free nid list */
225 unsigned int fcnt; /* the number of free node id */
226 struct mutex build_lock; /* lock for build free nids */
229 char *nat_bitmap; /* NAT bitmap pointer */
230 int bitmap_size; /* bitmap size */
234 * this structure is used as one of function parameters.
235 * all the information are dedicated to a given direct node block determined
236 * by the data offset in a file.
238 struct dnode_of_data {
239 struct inode *inode; /* vfs inode pointer */
240 struct page *inode_page; /* its inode page, NULL is possible */
241 struct page *node_page; /* cached direct node page */
242 nid_t nid; /* node id of the direct node block */
243 unsigned int ofs_in_node; /* data offset in the node page */
244 bool inode_page_locked; /* inode page is locked or not */
245 block_t data_blkaddr; /* block address of the node block */
248 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
249 struct page *ipage, struct page *npage, nid_t nid)
251 memset(dn, 0, sizeof(*dn));
253 dn->inode_page = ipage;
254 dn->node_page = npage;
261 * By default, there are 6 active log areas across the whole main area.
262 * When considering hot and cold data separation to reduce cleaning overhead,
263 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
265 * In the current design, you should not change the numbers intentionally.
266 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
267 * logs individually according to the underlying devices. (default: 6)
268 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
269 * data and 8 for node logs.
271 #define NR_CURSEG_DATA_TYPE (3)
272 #define NR_CURSEG_NODE_TYPE (3)
273 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
276 CURSEG_HOT_DATA = 0, /* directory entry blocks */
277 CURSEG_WARM_DATA, /* data blocks */
278 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
279 CURSEG_HOT_NODE, /* direct node blocks of directory files */
280 CURSEG_WARM_NODE, /* direct node blocks of normal files */
281 CURSEG_COLD_NODE, /* indirect node blocks */
285 struct f2fs_sm_info {
286 struct sit_info *sit_info; /* whole segment information */
287 struct free_segmap_info *free_info; /* free segment information */
288 struct dirty_seglist_info *dirty_info; /* dirty segment information */
289 struct curseg_info *curseg_array; /* active segment information */
291 struct list_head wblist_head; /* list of under-writeback pages */
292 spinlock_t wblist_lock; /* lock for checkpoint */
294 block_t seg0_blkaddr; /* block address of 0'th segment */
295 block_t main_blkaddr; /* start block address of main area */
296 block_t ssa_blkaddr; /* start block address of SSA area */
298 unsigned int segment_count; /* total # of segments */
299 unsigned int main_segments; /* # of segments in main area */
300 unsigned int reserved_segments; /* # of reserved segments */
301 unsigned int ovp_segments; /* # of overprovision segments */
308 * COUNT_TYPE for monitoring
310 * f2fs monitors the number of several block types such as on-writeback,
311 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
322 * The below are the page types of bios used in submti_bio().
323 * The available types are:
324 * DATA User data pages. It operates as async mode.
325 * NODE Node pages. It operates as async mode.
326 * META FS metadata pages such as SIT, NAT, CP.
327 * NR_PAGE_TYPE The number of page types.
328 * META_FLUSH Make sure the previous pages are written
329 * with waiting the bio's completion
330 * ... Only can be used with META.
340 struct f2fs_sb_info {
341 struct super_block *sb; /* pointer to VFS super block */
342 struct proc_dir_entry *s_proc; /* proc entry */
343 struct buffer_head *raw_super_buf; /* buffer head of raw sb */
344 struct f2fs_super_block *raw_super; /* raw super block pointer */
345 int s_dirty; /* dirty flag for checkpoint */
347 /* for node-related operations */
348 struct f2fs_nm_info *nm_info; /* node manager */
349 struct inode *node_inode; /* cache node blocks */
351 /* for segment-related operations */
352 struct f2fs_sm_info *sm_info; /* segment manager */
353 struct bio *bio[NR_PAGE_TYPE]; /* bios to merge */
354 sector_t last_block_in_bio[NR_PAGE_TYPE]; /* last block number */
355 struct rw_semaphore bio_sem; /* IO semaphore */
358 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
359 struct inode *meta_inode; /* cache meta blocks */
360 struct mutex cp_mutex; /* checkpoint procedure lock */
361 struct rw_semaphore cp_rwsem; /* blocking FS operations */
362 struct mutex node_write; /* locking node writes */
363 struct mutex writepages; /* mutex for writepages() */
364 bool por_doing; /* recovery is doing or not */
365 bool on_build_free_nids; /* build_free_nids is doing */
366 struct task_struct *cp_task; /* checkpoint task */
368 /* for orphan inode management */
369 struct list_head orphan_inode_list; /* orphan inode list */
370 struct mutex orphan_inode_mutex; /* for orphan inode list */
371 unsigned int n_orphans; /* # of orphan inodes */
373 /* for directory inode management */
374 struct list_head dir_inode_list; /* dir inode list */
375 spinlock_t dir_inode_lock; /* for dir inode list lock */
377 /* basic file system units */
378 unsigned int log_sectors_per_block; /* log2 sectors per block */
379 unsigned int log_blocksize; /* log2 block size */
380 unsigned int blocksize; /* block size */
381 unsigned int root_ino_num; /* root inode number*/
382 unsigned int node_ino_num; /* node inode number*/
383 unsigned int meta_ino_num; /* meta inode number*/
384 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
385 unsigned int blocks_per_seg; /* blocks per segment */
386 unsigned int segs_per_sec; /* segments per section */
387 unsigned int secs_per_zone; /* sections per zone */
388 unsigned int total_sections; /* total section count */
389 unsigned int total_node_count; /* total node block count */
390 unsigned int total_valid_node_count; /* valid node block count */
391 unsigned int total_valid_inode_count; /* valid inode count */
392 int active_logs; /* # of active logs */
394 block_t user_block_count; /* # of user blocks */
395 block_t total_valid_block_count; /* # of valid blocks */
396 block_t alloc_valid_block_count; /* # of allocated blocks */
397 block_t last_valid_block_count; /* for recovery */
398 u32 s_next_generation; /* for NFS support */
399 atomic_t nr_pages[NR_COUNT_TYPE]; /* # of pages, see count_type */
401 struct f2fs_mount_info mount_opt; /* mount options */
403 /* for cleaning operations */
404 struct mutex gc_mutex; /* mutex for GC */
405 struct f2fs_gc_kthread *gc_thread; /* GC thread */
406 unsigned int cur_victim_sec; /* current victim section num */
409 * for stat information.
410 * one is for the LFS mode, and the other is for the SSR mode.
412 #ifdef CONFIG_F2FS_STAT_FS
413 struct f2fs_stat_info *stat_info; /* FS status information */
414 unsigned int segment_count[2]; /* # of allocated segments */
415 unsigned int block_count[2]; /* # of allocated blocks */
416 int total_hit_ext, read_hit_ext; /* extent cache hit ratio */
417 int bg_gc; /* background gc calls */
418 unsigned int n_dirty_dirs; /* # of dir inodes */
420 unsigned int last_victim[2]; /* last victim segment # */
421 spinlock_t stat_lock; /* lock for stat operations */
423 /* For sysfs suppport */
424 struct kobject s_kobj;
425 struct completion s_kobj_unregister;
431 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
433 return container_of(inode, struct f2fs_inode_info, vfs_inode);
436 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
438 return sb->s_fs_info;
441 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
443 return (struct f2fs_super_block *)(sbi->raw_super);
446 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
448 return (struct f2fs_checkpoint *)(sbi->ckpt);
451 static inline struct f2fs_node *F2FS_NODE(struct page *page)
453 return (struct f2fs_node *)page_address(page);
456 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
458 return (struct f2fs_nm_info *)(sbi->nm_info);
461 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
463 return (struct f2fs_sm_info *)(sbi->sm_info);
466 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
468 return (struct sit_info *)(SM_I(sbi)->sit_info);
471 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
473 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
476 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
478 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
481 static inline void F2FS_SET_SB_DIRT(struct f2fs_sb_info *sbi)
486 static inline void F2FS_RESET_SB_DIRT(struct f2fs_sb_info *sbi)
491 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
493 return le64_to_cpu(cp->checkpoint_ver);
496 static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
498 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
499 return ckpt_flags & f;
502 static inline void set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
504 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
506 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
509 static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
511 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
513 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
516 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
518 down_read(&sbi->cp_rwsem);
521 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
523 up_read(&sbi->cp_rwsem);
526 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
528 down_write_nest_lock(&sbi->cp_rwsem, &sbi->cp_mutex);
531 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
533 up_write(&sbi->cp_rwsem);
537 * Check whether the given nid is within node id range.
539 static inline int check_nid_range(struct f2fs_sb_info *sbi, nid_t nid)
541 WARN_ON((nid >= NM_I(sbi)->max_nid));
542 if (nid >= NM_I(sbi)->max_nid)
547 #define F2FS_DEFAULT_ALLOCATED_BLOCKS 1
550 * Check whether the inode has blocks or not
552 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
554 if (F2FS_I(inode)->i_xattr_nid)
555 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS + 1);
557 return (inode->i_blocks > F2FS_DEFAULT_ALLOCATED_BLOCKS);
560 static inline bool inc_valid_block_count(struct f2fs_sb_info *sbi,
561 struct inode *inode, blkcnt_t count)
563 block_t valid_block_count;
565 spin_lock(&sbi->stat_lock);
567 sbi->total_valid_block_count + (block_t)count;
568 if (valid_block_count > sbi->user_block_count) {
569 spin_unlock(&sbi->stat_lock);
572 inode->i_blocks += count;
573 sbi->total_valid_block_count = valid_block_count;
574 sbi->alloc_valid_block_count += (block_t)count;
575 spin_unlock(&sbi->stat_lock);
579 static inline int dec_valid_block_count(struct f2fs_sb_info *sbi,
583 spin_lock(&sbi->stat_lock);
584 BUG_ON(sbi->total_valid_block_count < (block_t) count);
585 BUG_ON(inode->i_blocks < count);
586 inode->i_blocks -= count;
587 sbi->total_valid_block_count -= (block_t)count;
588 spin_unlock(&sbi->stat_lock);
592 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
594 atomic_inc(&sbi->nr_pages[count_type]);
595 F2FS_SET_SB_DIRT(sbi);
598 static inline void inode_inc_dirty_dents(struct inode *inode)
600 atomic_inc(&F2FS_I(inode)->dirty_dents);
603 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
605 atomic_dec(&sbi->nr_pages[count_type]);
608 static inline void inode_dec_dirty_dents(struct inode *inode)
610 atomic_dec(&F2FS_I(inode)->dirty_dents);
613 static inline int get_pages(struct f2fs_sb_info *sbi, int count_type)
615 return atomic_read(&sbi->nr_pages[count_type]);
618 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
620 unsigned int pages_per_sec = sbi->segs_per_sec *
621 (1 << sbi->log_blocks_per_seg);
622 return ((get_pages(sbi, block_type) + pages_per_sec - 1)
623 >> sbi->log_blocks_per_seg) / sbi->segs_per_sec;
626 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
629 spin_lock(&sbi->stat_lock);
630 ret = sbi->total_valid_block_count;
631 spin_unlock(&sbi->stat_lock);
635 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
637 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
639 /* return NAT or SIT bitmap */
640 if (flag == NAT_BITMAP)
641 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
642 else if (flag == SIT_BITMAP)
643 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
648 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
650 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
651 int offset = (flag == NAT_BITMAP) ?
652 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
653 return &ckpt->sit_nat_version_bitmap + offset;
656 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
659 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
660 unsigned long long ckpt_version = cur_cp_version(ckpt);
662 start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
665 * odd numbered checkpoint should at cp segment 0
666 * and even segent must be at cp segment 1
668 if (!(ckpt_version & 1))
669 start_addr += sbi->blocks_per_seg;
674 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
676 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
679 static inline bool inc_valid_node_count(struct f2fs_sb_info *sbi,
683 block_t valid_block_count;
684 unsigned int valid_node_count;
686 spin_lock(&sbi->stat_lock);
688 valid_block_count = sbi->total_valid_block_count + (block_t)count;
689 sbi->alloc_valid_block_count += (block_t)count;
690 valid_node_count = sbi->total_valid_node_count + count;
692 if (valid_block_count > sbi->user_block_count) {
693 spin_unlock(&sbi->stat_lock);
697 if (valid_node_count > sbi->total_node_count) {
698 spin_unlock(&sbi->stat_lock);
703 inode->i_blocks += count;
704 sbi->total_valid_node_count = valid_node_count;
705 sbi->total_valid_block_count = valid_block_count;
706 spin_unlock(&sbi->stat_lock);
711 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
715 spin_lock(&sbi->stat_lock);
717 BUG_ON(sbi->total_valid_block_count < count);
718 BUG_ON(sbi->total_valid_node_count < count);
719 BUG_ON(inode->i_blocks < count);
721 inode->i_blocks -= count;
722 sbi->total_valid_node_count -= count;
723 sbi->total_valid_block_count -= (block_t)count;
725 spin_unlock(&sbi->stat_lock);
728 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
731 spin_lock(&sbi->stat_lock);
732 ret = sbi->total_valid_node_count;
733 spin_unlock(&sbi->stat_lock);
737 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
739 spin_lock(&sbi->stat_lock);
740 BUG_ON(sbi->total_valid_inode_count == sbi->total_node_count);
741 sbi->total_valid_inode_count++;
742 spin_unlock(&sbi->stat_lock);
745 static inline int dec_valid_inode_count(struct f2fs_sb_info *sbi)
747 spin_lock(&sbi->stat_lock);
748 BUG_ON(!sbi->total_valid_inode_count);
749 sbi->total_valid_inode_count--;
750 spin_unlock(&sbi->stat_lock);
754 static inline unsigned int valid_inode_count(struct f2fs_sb_info *sbi)
757 spin_lock(&sbi->stat_lock);
758 ret = sbi->total_valid_inode_count;
759 spin_unlock(&sbi->stat_lock);
763 static inline void f2fs_put_page(struct page *page, int unlock)
765 if (!page || IS_ERR(page))
769 BUG_ON(!PageLocked(page));
772 page_cache_release(page);
775 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
778 f2fs_put_page(dn->node_page, 1);
779 if (dn->inode_page && dn->node_page != dn->inode_page)
780 f2fs_put_page(dn->inode_page, 0);
781 dn->node_page = NULL;
782 dn->inode_page = NULL;
785 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
786 size_t size, void (*ctor)(void *))
788 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, ctor);
791 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
796 entry = kmem_cache_alloc(cachep, flags);
805 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
807 static inline bool IS_INODE(struct page *page)
809 struct f2fs_node *p = F2FS_NODE(page);
810 return RAW_IS_INODE(p);
813 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
815 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
818 static inline block_t datablock_addr(struct page *node_page,
821 struct f2fs_node *raw_node;
823 raw_node = F2FS_NODE(node_page);
824 addr_array = blkaddr_in_node(raw_node);
825 return le32_to_cpu(addr_array[offset]);
828 static inline int f2fs_test_bit(unsigned int nr, char *addr)
833 mask = 1 << (7 - (nr & 0x07));
837 static inline int f2fs_set_bit(unsigned int nr, char *addr)
843 mask = 1 << (7 - (nr & 0x07));
849 static inline int f2fs_clear_bit(unsigned int nr, char *addr)
855 mask = 1 << (7 - (nr & 0x07));
861 /* used for f2fs_inode_info->flags */
863 FI_NEW_INODE, /* indicate newly allocated inode */
864 FI_DIRTY_INODE, /* indicate inode is dirty or not */
865 FI_INC_LINK, /* need to increment i_nlink */
866 FI_ACL_MODE, /* indicate acl mode */
867 FI_NO_ALLOC, /* should not allocate any blocks */
868 FI_UPDATE_DIR, /* should update inode block for consistency */
869 FI_DELAY_IPUT, /* used for the recovery */
870 FI_INLINE_XATTR, /* used for inline xattr */
873 static inline void set_inode_flag(struct f2fs_inode_info *fi, int flag)
875 set_bit(flag, &fi->flags);
878 static inline int is_inode_flag_set(struct f2fs_inode_info *fi, int flag)
880 return test_bit(flag, &fi->flags);
883 static inline void clear_inode_flag(struct f2fs_inode_info *fi, int flag)
885 clear_bit(flag, &fi->flags);
888 static inline void set_acl_inode(struct f2fs_inode_info *fi, umode_t mode)
890 fi->i_acl_mode = mode;
891 set_inode_flag(fi, FI_ACL_MODE);
894 static inline int cond_clear_inode_flag(struct f2fs_inode_info *fi, int flag)
896 if (is_inode_flag_set(fi, FI_ACL_MODE)) {
897 clear_inode_flag(fi, FI_ACL_MODE);
903 static inline void get_inline_info(struct f2fs_inode_info *fi,
904 struct f2fs_inode *ri)
906 if (ri->i_inline & F2FS_INLINE_XATTR)
907 set_inode_flag(fi, FI_INLINE_XATTR);
910 static inline void set_raw_inline(struct f2fs_inode_info *fi,
911 struct f2fs_inode *ri)
915 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
916 ri->i_inline |= F2FS_INLINE_XATTR;
919 static inline unsigned int addrs_per_inode(struct f2fs_inode_info *fi)
921 if (is_inode_flag_set(fi, FI_INLINE_XATTR))
922 return DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS;
923 return DEF_ADDRS_PER_INODE;
926 static inline void *inline_xattr_addr(struct page *page)
928 struct f2fs_inode *ri;
929 ri = (struct f2fs_inode *)page_address(page);
930 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
931 F2FS_INLINE_XATTR_ADDRS]);
934 static inline int inline_xattr_size(struct inode *inode)
936 if (is_inode_flag_set(F2FS_I(inode), FI_INLINE_XATTR))
937 return F2FS_INLINE_XATTR_ADDRS << 2;
942 static inline int f2fs_readonly(struct super_block *sb)
944 return sb->s_flags & MS_RDONLY;
950 int f2fs_sync_file(struct file *, loff_t, loff_t, int);
951 void truncate_data_blocks(struct dnode_of_data *);
952 void f2fs_truncate(struct inode *);
953 int f2fs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
954 int f2fs_setattr(struct dentry *, struct iattr *);
955 int truncate_hole(struct inode *, pgoff_t, pgoff_t);
956 int truncate_data_blocks_range(struct dnode_of_data *, int);
957 long f2fs_ioctl(struct file *, unsigned int, unsigned long);
958 long f2fs_compat_ioctl(struct file *, unsigned int, unsigned long);
963 void f2fs_set_inode_flags(struct inode *);
964 struct inode *f2fs_iget(struct super_block *, unsigned long);
965 void update_inode(struct inode *, struct page *);
966 int update_inode_page(struct inode *);
967 int f2fs_write_inode(struct inode *, struct writeback_control *);
968 void f2fs_evict_inode(struct inode *);
973 struct dentry *f2fs_get_parent(struct dentry *child);
978 struct f2fs_dir_entry *f2fs_find_entry(struct inode *, struct qstr *,
980 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *, struct page **);
981 ino_t f2fs_inode_by_name(struct inode *, struct qstr *);
982 void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
983 struct page *, struct inode *);
984 int update_dent_inode(struct inode *, const struct qstr *);
985 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *);
986 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *);
987 int f2fs_make_empty(struct inode *, struct inode *);
988 bool f2fs_empty_dir(struct inode *);
990 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
992 return __f2fs_add_link(dentry->d_parent->d_inode, &dentry->d_name,
999 int f2fs_sync_fs(struct super_block *, int);
1000 extern __printf(3, 4)
1001 void f2fs_msg(struct super_block *, const char *, const char *, ...);
1006 f2fs_hash_t f2fs_dentry_hash(const char *, size_t);
1011 struct dnode_of_data;
1014 int is_checkpointed_node(struct f2fs_sb_info *, nid_t);
1015 void get_node_info(struct f2fs_sb_info *, nid_t, struct node_info *);
1016 int get_dnode_of_data(struct dnode_of_data *, pgoff_t, int);
1017 int truncate_inode_blocks(struct inode *, pgoff_t);
1018 int truncate_xattr_node(struct inode *, struct page *);
1019 int remove_inode_page(struct inode *);
1020 struct page *new_inode_page(struct inode *, const struct qstr *);
1021 struct page *new_node_page(struct dnode_of_data *, unsigned int, struct page *);
1022 void ra_node_page(struct f2fs_sb_info *, nid_t);
1023 struct page *get_node_page(struct f2fs_sb_info *, pgoff_t);
1024 struct page *get_node_page_ra(struct page *, int);
1025 void sync_inode_page(struct dnode_of_data *);
1026 int sync_node_pages(struct f2fs_sb_info *, nid_t, struct writeback_control *);
1027 bool alloc_nid(struct f2fs_sb_info *, nid_t *);
1028 void alloc_nid_done(struct f2fs_sb_info *, nid_t);
1029 void alloc_nid_failed(struct f2fs_sb_info *, nid_t);
1030 void recover_node_page(struct f2fs_sb_info *, struct page *,
1031 struct f2fs_summary *, struct node_info *, block_t);
1032 int recover_inode_page(struct f2fs_sb_info *, struct page *);
1033 int restore_node_summary(struct f2fs_sb_info *, unsigned int,
1034 struct f2fs_summary_block *);
1035 void flush_nat_entries(struct f2fs_sb_info *);
1036 int build_node_manager(struct f2fs_sb_info *);
1037 void destroy_node_manager(struct f2fs_sb_info *);
1038 int __init create_node_manager_caches(void);
1039 void destroy_node_manager_caches(void);
1044 void f2fs_balance_fs(struct f2fs_sb_info *);
1045 void invalidate_blocks(struct f2fs_sb_info *, block_t);
1046 void clear_prefree_segments(struct f2fs_sb_info *);
1047 int npages_for_summary_flush(struct f2fs_sb_info *);
1048 void allocate_new_segments(struct f2fs_sb_info *);
1049 struct page *get_sum_page(struct f2fs_sb_info *, unsigned int);
1050 struct bio *f2fs_bio_alloc(struct block_device *, int);
1051 void f2fs_submit_bio(struct f2fs_sb_info *, enum page_type, bool);
1052 void f2fs_wait_on_page_writeback(struct page *, enum page_type, bool);
1053 void write_meta_page(struct f2fs_sb_info *, struct page *);
1054 void write_node_page(struct f2fs_sb_info *, struct page *, unsigned int,
1055 block_t, block_t *);
1056 void write_data_page(struct inode *, struct page *, struct dnode_of_data*,
1057 block_t, block_t *);
1058 void rewrite_data_page(struct f2fs_sb_info *, struct page *, block_t);
1059 void recover_data_page(struct f2fs_sb_info *, struct page *,
1060 struct f2fs_summary *, block_t, block_t);
1061 void rewrite_node_page(struct f2fs_sb_info *, struct page *,
1062 struct f2fs_summary *, block_t, block_t);
1063 void write_data_summaries(struct f2fs_sb_info *, block_t);
1064 void write_node_summaries(struct f2fs_sb_info *, block_t);
1065 int lookup_journal_in_cursum(struct f2fs_summary_block *,
1066 int, unsigned int, int);
1067 void flush_sit_entries(struct f2fs_sb_info *);
1068 int build_segment_manager(struct f2fs_sb_info *);
1069 void destroy_segment_manager(struct f2fs_sb_info *);
1074 struct page *grab_meta_page(struct f2fs_sb_info *, pgoff_t);
1075 struct page *get_meta_page(struct f2fs_sb_info *, pgoff_t);
1076 long sync_meta_pages(struct f2fs_sb_info *, enum page_type, long);
1077 int acquire_orphan_inode(struct f2fs_sb_info *);
1078 void release_orphan_inode(struct f2fs_sb_info *);
1079 void add_orphan_inode(struct f2fs_sb_info *, nid_t);
1080 void remove_orphan_inode(struct f2fs_sb_info *, nid_t);
1081 int recover_orphan_inodes(struct f2fs_sb_info *);
1082 int get_valid_checkpoint(struct f2fs_sb_info *);
1083 void set_dirty_dir_page(struct inode *, struct page *);
1084 void add_dirty_dir_inode(struct inode *);
1085 void remove_dirty_dir_inode(struct inode *);
1086 struct inode *check_dirty_dir_inode(struct f2fs_sb_info *, nid_t);
1087 void sync_dirty_dir_inodes(struct f2fs_sb_info *);
1088 void write_checkpoint(struct f2fs_sb_info *, bool);
1089 void init_orphan_info(struct f2fs_sb_info *);
1090 int __init create_checkpoint_caches(void);
1091 void destroy_checkpoint_caches(void);
1096 int reserve_new_block(struct dnode_of_data *);
1097 void update_extent_cache(block_t, struct dnode_of_data *);
1098 struct page *find_data_page(struct inode *, pgoff_t, bool);
1099 struct page *get_lock_data_page(struct inode *, pgoff_t);
1100 struct page *get_new_data_page(struct inode *, struct page *, pgoff_t, bool);
1101 int f2fs_readpage(struct f2fs_sb_info *, struct page *, block_t, int);
1102 int do_write_data_page(struct page *);
1107 int start_gc_thread(struct f2fs_sb_info *);
1108 void stop_gc_thread(struct f2fs_sb_info *);
1109 block_t start_bidx_of_node(unsigned int, struct f2fs_inode_info *);
1110 int f2fs_gc(struct f2fs_sb_info *);
1111 void build_gc_manager(struct f2fs_sb_info *);
1112 int __init create_gc_caches(void);
1113 void destroy_gc_caches(void);
1118 int recover_fsync_data(struct f2fs_sb_info *);
1119 bool space_for_roll_forward(struct f2fs_sb_info *);
1124 #ifdef CONFIG_F2FS_STAT_FS
1125 struct f2fs_stat_info {
1126 struct list_head stat_list;
1127 struct f2fs_sb_info *sbi;
1128 struct mutex stat_lock;
1129 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
1130 int main_area_segs, main_area_sections, main_area_zones;
1131 int hit_ext, total_ext;
1132 int ndirty_node, ndirty_dent, ndirty_dirs, ndirty_meta;
1133 int nats, sits, fnids;
1134 int total_count, utilization;
1136 unsigned int valid_count, valid_node_count, valid_inode_count;
1137 unsigned int bimodal, avg_vblocks;
1138 int util_free, util_valid, util_invalid;
1139 int rsvd_segs, overp_segs;
1140 int dirty_count, node_pages, meta_pages;
1141 int prefree_count, call_count;
1142 int tot_segs, node_segs, data_segs, free_segs, free_secs;
1143 int tot_blks, data_blks, node_blks;
1144 int curseg[NR_CURSEG_TYPE];
1145 int cursec[NR_CURSEG_TYPE];
1146 int curzone[NR_CURSEG_TYPE];
1148 unsigned int segment_count[2];
1149 unsigned int block_count[2];
1150 unsigned base_mem, cache_mem;
1153 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
1155 return (struct f2fs_stat_info*)sbi->stat_info;
1158 #define stat_inc_call_count(si) ((si)->call_count++)
1159 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
1160 #define stat_inc_dirty_dir(sbi) ((sbi)->n_dirty_dirs++)
1161 #define stat_dec_dirty_dir(sbi) ((sbi)->n_dirty_dirs--)
1162 #define stat_inc_hit_ext(sb) ((F2FS_SB(sb))->total_hit_ext++)
1163 #define stat_inc_alloc_type(sbi, curseg) \
1164 ((sbi)->segment_count[(curseg)->alloc_type]++)
1166 #define stat_inc_seg_count(sbi, type) \
1168 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1170 if (type == SUM_TYPE_DATA) \
1176 #define stat_inc_tot_blk_count(si, blks) \
1177 (si->tot_blks += (blks))
1179 #define stat_inc_data_blk_count(sbi, blks) \
1181 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1182 stat_inc_tot_blk_count(si, blks); \
1183 si->data_blks += (blks); \
1186 #define stat_inc_node_blk_count(sbi, blks) \
1188 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
1189 stat_inc_tot_blk_count(si, blks); \
1190 si->node_blks += (blks); \
1194 int f2fs_build_stats(struct f2fs_sb_info *);
1195 void f2fs_destroy_stats(struct f2fs_sb_info *);
1196 void __init f2fs_create_root_stats(void);
1197 void f2fs_destroy_root_stats(void);
1199 #define stat_inc_call_count(si)
1200 #define stat_inc_bggc_count(si)
1201 #define stat_inc_dirty_dir(sbi)
1202 #define stat_dec_dirty_dir(sbi)
1203 #define stat_inc_hit_ext(sb)
1204 #define stat_inc_alloc_type(sbi, curseg)
1205 #define stat_inc_seg_count(si, type)
1206 #define stat_inc_tot_blk_count(si, blks)
1207 #define stat_inc_data_blk_count(si, blks)
1208 #define stat_inc_node_blk_count(sbi, blks)
1210 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
1211 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
1212 static inline void __init f2fs_create_root_stats(void) { }
1213 static inline void f2fs_destroy_root_stats(void) { }
1216 extern const struct file_operations f2fs_dir_operations;
1217 extern const struct file_operations f2fs_file_operations;
1218 extern const struct inode_operations f2fs_file_inode_operations;
1219 extern const struct address_space_operations f2fs_dblock_aops;
1220 extern const struct address_space_operations f2fs_node_aops;
1221 extern const struct address_space_operations f2fs_meta_aops;
1222 extern const struct inode_operations f2fs_dir_inode_operations;
1223 extern const struct inode_operations f2fs_symlink_inode_operations;
1224 extern const struct inode_operations f2fs_special_inode_operations;