Loading...
Note: File does not exist in v3.5.6.
1// SPDX-License-Identifier: GPL-2.0
2/*
3 * fs/f2fs/f2fs.h
4 *
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
7 */
8#ifndef _LINUX_F2FS_H
9#define _LINUX_F2FS_H
10
11#include <linux/uio.h>
12#include <linux/types.h>
13#include <linux/page-flags.h>
14#include <linux/buffer_head.h>
15#include <linux/slab.h>
16#include <linux/crc32.h>
17#include <linux/magic.h>
18#include <linux/kobject.h>
19#include <linux/sched.h>
20#include <linux/cred.h>
21#include <linux/vmalloc.h>
22#include <linux/bio.h>
23#include <linux/blkdev.h>
24#include <linux/quotaops.h>
25#include <crypto/hash.h>
26
27#include <linux/fscrypt.h>
28#include <linux/fsverity.h>
29
30#ifdef CONFIG_F2FS_CHECK_FS
31#define f2fs_bug_on(sbi, condition) BUG_ON(condition)
32#else
33#define f2fs_bug_on(sbi, condition) \
34 do { \
35 if (unlikely(condition)) { \
36 WARN_ON(1); \
37 set_sbi_flag(sbi, SBI_NEED_FSCK); \
38 } \
39 } while (0)
40#endif
41
42enum {
43 FAULT_KMALLOC,
44 FAULT_KVMALLOC,
45 FAULT_PAGE_ALLOC,
46 FAULT_PAGE_GET,
47 FAULT_ALLOC_BIO,
48 FAULT_ALLOC_NID,
49 FAULT_ORPHAN,
50 FAULT_BLOCK,
51 FAULT_DIR_DEPTH,
52 FAULT_EVICT_INODE,
53 FAULT_TRUNCATE,
54 FAULT_READ_IO,
55 FAULT_CHECKPOINT,
56 FAULT_DISCARD,
57 FAULT_WRITE_IO,
58 FAULT_MAX,
59};
60
61#ifdef CONFIG_F2FS_FAULT_INJECTION
62#define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
63
64struct f2fs_fault_info {
65 atomic_t inject_ops;
66 unsigned int inject_rate;
67 unsigned int inject_type;
68};
69
70extern const char *f2fs_fault_name[FAULT_MAX];
71#define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
72#endif
73
74/*
75 * For mount options
76 */
77#define F2FS_MOUNT_BG_GC 0x00000001
78#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
79#define F2FS_MOUNT_DISCARD 0x00000004
80#define F2FS_MOUNT_NOHEAP 0x00000008
81#define F2FS_MOUNT_XATTR_USER 0x00000010
82#define F2FS_MOUNT_POSIX_ACL 0x00000020
83#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
84#define F2FS_MOUNT_INLINE_XATTR 0x00000080
85#define F2FS_MOUNT_INLINE_DATA 0x00000100
86#define F2FS_MOUNT_INLINE_DENTRY 0x00000200
87#define F2FS_MOUNT_FLUSH_MERGE 0x00000400
88#define F2FS_MOUNT_NOBARRIER 0x00000800
89#define F2FS_MOUNT_FASTBOOT 0x00001000
90#define F2FS_MOUNT_EXTENT_CACHE 0x00002000
91#define F2FS_MOUNT_FORCE_FG_GC 0x00004000
92#define F2FS_MOUNT_DATA_FLUSH 0x00008000
93#define F2FS_MOUNT_FAULT_INJECTION 0x00010000
94#define F2FS_MOUNT_ADAPTIVE 0x00020000
95#define F2FS_MOUNT_LFS 0x00040000
96#define F2FS_MOUNT_USRQUOTA 0x00080000
97#define F2FS_MOUNT_GRPQUOTA 0x00100000
98#define F2FS_MOUNT_PRJQUOTA 0x00200000
99#define F2FS_MOUNT_QUOTA 0x00400000
100#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
101#define F2FS_MOUNT_RESERVE_ROOT 0x01000000
102#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
103
104#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
105#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
106#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
107#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
108
109#define ver_after(a, b) (typecheck(unsigned long long, a) && \
110 typecheck(unsigned long long, b) && \
111 ((long long)((a) - (b)) > 0))
112
113typedef u32 block_t; /*
114 * should not change u32, since it is the on-disk block
115 * address format, __le32.
116 */
117typedef u32 nid_t;
118
119struct f2fs_mount_info {
120 unsigned int opt;
121 int write_io_size_bits; /* Write IO size bits */
122 block_t root_reserved_blocks; /* root reserved blocks */
123 kuid_t s_resuid; /* reserved blocks for uid */
124 kgid_t s_resgid; /* reserved blocks for gid */
125 int active_logs; /* # of active logs */
126 int inline_xattr_size; /* inline xattr size */
127#ifdef CONFIG_F2FS_FAULT_INJECTION
128 struct f2fs_fault_info fault_info; /* For fault injection */
129#endif
130#ifdef CONFIG_QUOTA
131 /* Names of quota files with journalled quota */
132 char *s_qf_names[MAXQUOTAS];
133 int s_jquota_fmt; /* Format of quota to use */
134#endif
135 /* For which write hints are passed down to block layer */
136 int whint_mode;
137 int alloc_mode; /* segment allocation policy */
138 int fsync_mode; /* fsync policy */
139 bool test_dummy_encryption; /* test dummy encryption */
140 block_t unusable_cap; /* Amount of space allowed to be
141 * unusable when disabling checkpoint
142 */
143};
144
145#define F2FS_FEATURE_ENCRYPT 0x0001
146#define F2FS_FEATURE_BLKZONED 0x0002
147#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
148#define F2FS_FEATURE_EXTRA_ATTR 0x0008
149#define F2FS_FEATURE_PRJQUOTA 0x0010
150#define F2FS_FEATURE_INODE_CHKSUM 0x0020
151#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
152#define F2FS_FEATURE_QUOTA_INO 0x0080
153#define F2FS_FEATURE_INODE_CRTIME 0x0100
154#define F2FS_FEATURE_LOST_FOUND 0x0200
155#define F2FS_FEATURE_VERITY 0x0400
156#define F2FS_FEATURE_SB_CHKSUM 0x0800
157#define F2FS_FEATURE_CASEFOLD 0x1000
158
159#define __F2FS_HAS_FEATURE(raw_super, mask) \
160 ((raw_super->feature & cpu_to_le32(mask)) != 0)
161#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
162#define F2FS_SET_FEATURE(sbi, mask) \
163 (sbi->raw_super->feature |= cpu_to_le32(mask))
164#define F2FS_CLEAR_FEATURE(sbi, mask) \
165 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
166
167/*
168 * Default values for user and/or group using reserved blocks
169 */
170#define F2FS_DEF_RESUID 0
171#define F2FS_DEF_RESGID 0
172
173/*
174 * For checkpoint manager
175 */
176enum {
177 NAT_BITMAP,
178 SIT_BITMAP
179};
180
181#define CP_UMOUNT 0x00000001
182#define CP_FASTBOOT 0x00000002
183#define CP_SYNC 0x00000004
184#define CP_RECOVERY 0x00000008
185#define CP_DISCARD 0x00000010
186#define CP_TRIMMED 0x00000020
187#define CP_PAUSE 0x00000040
188
189#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
190#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
191#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
192#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
193#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
194#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
195#define DEF_CP_INTERVAL 60 /* 60 secs */
196#define DEF_IDLE_INTERVAL 5 /* 5 secs */
197#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
198#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
199#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
200
201struct cp_control {
202 int reason;
203 __u64 trim_start;
204 __u64 trim_end;
205 __u64 trim_minlen;
206};
207
208/*
209 * indicate meta/data type
210 */
211enum {
212 META_CP,
213 META_NAT,
214 META_SIT,
215 META_SSA,
216 META_MAX,
217 META_POR,
218 DATA_GENERIC, /* check range only */
219 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
220 DATA_GENERIC_ENHANCE_READ, /*
221 * strong check on range and segment
222 * bitmap but no warning due to race
223 * condition of read on truncated area
224 * by extent_cache
225 */
226 META_GENERIC,
227};
228
229/* for the list of ino */
230enum {
231 ORPHAN_INO, /* for orphan ino list */
232 APPEND_INO, /* for append ino list */
233 UPDATE_INO, /* for update ino list */
234 TRANS_DIR_INO, /* for trasactions dir ino list */
235 FLUSH_INO, /* for multiple device flushing */
236 MAX_INO_ENTRY, /* max. list */
237};
238
239struct ino_entry {
240 struct list_head list; /* list head */
241 nid_t ino; /* inode number */
242 unsigned int dirty_device; /* dirty device bitmap */
243};
244
245/* for the list of inodes to be GCed */
246struct inode_entry {
247 struct list_head list; /* list head */
248 struct inode *inode; /* vfs inode pointer */
249};
250
251struct fsync_node_entry {
252 struct list_head list; /* list head */
253 struct page *page; /* warm node page pointer */
254 unsigned int seq_id; /* sequence id */
255};
256
257/* for the bitmap indicate blocks to be discarded */
258struct discard_entry {
259 struct list_head list; /* list head */
260 block_t start_blkaddr; /* start blockaddr of current segment */
261 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
262};
263
264/* default discard granularity of inner discard thread, unit: block count */
265#define DEFAULT_DISCARD_GRANULARITY 16
266
267/* max discard pend list number */
268#define MAX_PLIST_NUM 512
269#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
270 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
271
272enum {
273 D_PREP, /* initial */
274 D_PARTIAL, /* partially submitted */
275 D_SUBMIT, /* all submitted */
276 D_DONE, /* finished */
277};
278
279struct discard_info {
280 block_t lstart; /* logical start address */
281 block_t len; /* length */
282 block_t start; /* actual start address in dev */
283};
284
285struct discard_cmd {
286 struct rb_node rb_node; /* rb node located in rb-tree */
287 union {
288 struct {
289 block_t lstart; /* logical start address */
290 block_t len; /* length */
291 block_t start; /* actual start address in dev */
292 };
293 struct discard_info di; /* discard info */
294
295 };
296 struct list_head list; /* command list */
297 struct completion wait; /* compleation */
298 struct block_device *bdev; /* bdev */
299 unsigned short ref; /* reference count */
300 unsigned char state; /* state */
301 unsigned char queued; /* queued discard */
302 int error; /* bio error */
303 spinlock_t lock; /* for state/bio_ref updating */
304 unsigned short bio_ref; /* bio reference count */
305};
306
307enum {
308 DPOLICY_BG,
309 DPOLICY_FORCE,
310 DPOLICY_FSTRIM,
311 DPOLICY_UMOUNT,
312 MAX_DPOLICY,
313};
314
315struct discard_policy {
316 int type; /* type of discard */
317 unsigned int min_interval; /* used for candidates exist */
318 unsigned int mid_interval; /* used for device busy */
319 unsigned int max_interval; /* used for candidates not exist */
320 unsigned int max_requests; /* # of discards issued per round */
321 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
322 bool io_aware; /* issue discard in idle time */
323 bool sync; /* submit discard with REQ_SYNC flag */
324 bool ordered; /* issue discard by lba order */
325 unsigned int granularity; /* discard granularity */
326 int timeout; /* discard timeout for put_super */
327};
328
329struct discard_cmd_control {
330 struct task_struct *f2fs_issue_discard; /* discard thread */
331 struct list_head entry_list; /* 4KB discard entry list */
332 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
333 struct list_head wait_list; /* store on-flushing entries */
334 struct list_head fstrim_list; /* in-flight discard from fstrim */
335 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
336 unsigned int discard_wake; /* to wake up discard thread */
337 struct mutex cmd_lock;
338 unsigned int nr_discards; /* # of discards in the list */
339 unsigned int max_discards; /* max. discards to be issued */
340 unsigned int discard_granularity; /* discard granularity */
341 unsigned int undiscard_blks; /* # of undiscard blocks */
342 unsigned int next_pos; /* next discard position */
343 atomic_t issued_discard; /* # of issued discard */
344 atomic_t queued_discard; /* # of queued discard */
345 atomic_t discard_cmd_cnt; /* # of cached cmd count */
346 struct rb_root_cached root; /* root of discard rb-tree */
347 bool rbtree_check; /* config for consistence check */
348};
349
350/* for the list of fsync inodes, used only during recovery */
351struct fsync_inode_entry {
352 struct list_head list; /* list head */
353 struct inode *inode; /* vfs inode pointer */
354 block_t blkaddr; /* block address locating the last fsync */
355 block_t last_dentry; /* block address locating the last dentry */
356};
357
358#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
359#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
360
361#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
362#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
363#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
364#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
365
366#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
367#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
368
369static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
370{
371 int before = nats_in_cursum(journal);
372
373 journal->n_nats = cpu_to_le16(before + i);
374 return before;
375}
376
377static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
378{
379 int before = sits_in_cursum(journal);
380
381 journal->n_sits = cpu_to_le16(before + i);
382 return before;
383}
384
385static inline bool __has_cursum_space(struct f2fs_journal *journal,
386 int size, int type)
387{
388 if (type == NAT_JOURNAL)
389 return size <= MAX_NAT_JENTRIES(journal);
390 return size <= MAX_SIT_JENTRIES(journal);
391}
392
393/*
394 * ioctl commands
395 */
396#define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
397#define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
398#define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
399
400#define F2FS_IOCTL_MAGIC 0xf5
401#define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
402#define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
403#define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
404#define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
405#define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
406#define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
407#define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
408#define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
409 struct f2fs_defragment)
410#define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
411 struct f2fs_move_range)
412#define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
413 struct f2fs_flush_device)
414#define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
415 struct f2fs_gc_range)
416#define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
417#define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
418#define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
419#define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
420#define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64)
421
422#define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL
423#define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL
424
425#define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
426#define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
427#define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
428
429/*
430 * should be same as XFS_IOC_GOINGDOWN.
431 * Flags for going down operation used by FS_IOC_GOINGDOWN
432 */
433#define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
434#define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
435#define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
436#define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
437#define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
438#define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
439
440#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
441/*
442 * ioctl commands in 32 bit emulation
443 */
444#define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
445#define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
446#define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
447#endif
448
449#define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
450#define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
451
452struct f2fs_gc_range {
453 u32 sync;
454 u64 start;
455 u64 len;
456};
457
458struct f2fs_defragment {
459 u64 start;
460 u64 len;
461};
462
463struct f2fs_move_range {
464 u32 dst_fd; /* destination fd */
465 u64 pos_in; /* start position in src_fd */
466 u64 pos_out; /* start position in dst_fd */
467 u64 len; /* size to move */
468};
469
470struct f2fs_flush_device {
471 u32 dev_num; /* device number to flush */
472 u32 segments; /* # of segments to flush */
473};
474
475/* for inline stuff */
476#define DEF_INLINE_RESERVED_SIZE 1
477static inline int get_extra_isize(struct inode *inode);
478static inline int get_inline_xattr_addrs(struct inode *inode);
479#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
480 (CUR_ADDRS_PER_INODE(inode) - \
481 get_inline_xattr_addrs(inode) - \
482 DEF_INLINE_RESERVED_SIZE))
483
484/* for inline dir */
485#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
486 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
487 BITS_PER_BYTE + 1))
488#define INLINE_DENTRY_BITMAP_SIZE(inode) \
489 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
490#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
491 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
492 NR_INLINE_DENTRY(inode) + \
493 INLINE_DENTRY_BITMAP_SIZE(inode)))
494
495/*
496 * For INODE and NODE manager
497 */
498/* for directory operations */
499struct f2fs_dentry_ptr {
500 struct inode *inode;
501 void *bitmap;
502 struct f2fs_dir_entry *dentry;
503 __u8 (*filename)[F2FS_SLOT_LEN];
504 int max;
505 int nr_bitmap;
506};
507
508static inline void make_dentry_ptr_block(struct inode *inode,
509 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
510{
511 d->inode = inode;
512 d->max = NR_DENTRY_IN_BLOCK;
513 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
514 d->bitmap = t->dentry_bitmap;
515 d->dentry = t->dentry;
516 d->filename = t->filename;
517}
518
519static inline void make_dentry_ptr_inline(struct inode *inode,
520 struct f2fs_dentry_ptr *d, void *t)
521{
522 int entry_cnt = NR_INLINE_DENTRY(inode);
523 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
524 int reserved_size = INLINE_RESERVED_SIZE(inode);
525
526 d->inode = inode;
527 d->max = entry_cnt;
528 d->nr_bitmap = bitmap_size;
529 d->bitmap = t;
530 d->dentry = t + bitmap_size + reserved_size;
531 d->filename = t + bitmap_size + reserved_size +
532 SIZE_OF_DIR_ENTRY * entry_cnt;
533}
534
535/*
536 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
537 * as its node offset to distinguish from index node blocks.
538 * But some bits are used to mark the node block.
539 */
540#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
541 >> OFFSET_BIT_SHIFT)
542enum {
543 ALLOC_NODE, /* allocate a new node page if needed */
544 LOOKUP_NODE, /* look up a node without readahead */
545 LOOKUP_NODE_RA, /*
546 * look up a node with readahead called
547 * by get_data_block.
548 */
549};
550
551#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
552
553/* maximum retry quota flush count */
554#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
555
556#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
557
558#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
559
560/* for in-memory extent cache entry */
561#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
562
563/* number of extent info in extent cache we try to shrink */
564#define EXTENT_CACHE_SHRINK_NUMBER 128
565
566struct rb_entry {
567 struct rb_node rb_node; /* rb node located in rb-tree */
568 unsigned int ofs; /* start offset of the entry */
569 unsigned int len; /* length of the entry */
570};
571
572struct extent_info {
573 unsigned int fofs; /* start offset in a file */
574 unsigned int len; /* length of the extent */
575 u32 blk; /* start block address of the extent */
576};
577
578struct extent_node {
579 struct rb_node rb_node; /* rb node located in rb-tree */
580 struct extent_info ei; /* extent info */
581 struct list_head list; /* node in global extent list of sbi */
582 struct extent_tree *et; /* extent tree pointer */
583};
584
585struct extent_tree {
586 nid_t ino; /* inode number */
587 struct rb_root_cached root; /* root of extent info rb-tree */
588 struct extent_node *cached_en; /* recently accessed extent node */
589 struct extent_info largest; /* largested extent info */
590 struct list_head list; /* to be used by sbi->zombie_list */
591 rwlock_t lock; /* protect extent info rb-tree */
592 atomic_t node_cnt; /* # of extent node in rb-tree*/
593 bool largest_updated; /* largest extent updated */
594};
595
596/*
597 * This structure is taken from ext4_map_blocks.
598 *
599 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
600 */
601#define F2FS_MAP_NEW (1 << BH_New)
602#define F2FS_MAP_MAPPED (1 << BH_Mapped)
603#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
604#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
605 F2FS_MAP_UNWRITTEN)
606
607struct f2fs_map_blocks {
608 block_t m_pblk;
609 block_t m_lblk;
610 unsigned int m_len;
611 unsigned int m_flags;
612 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
613 pgoff_t *m_next_extent; /* point to next possible extent */
614 int m_seg_type;
615 bool m_may_create; /* indicate it is from write path */
616};
617
618/* for flag in get_data_block */
619enum {
620 F2FS_GET_BLOCK_DEFAULT,
621 F2FS_GET_BLOCK_FIEMAP,
622 F2FS_GET_BLOCK_BMAP,
623 F2FS_GET_BLOCK_DIO,
624 F2FS_GET_BLOCK_PRE_DIO,
625 F2FS_GET_BLOCK_PRE_AIO,
626 F2FS_GET_BLOCK_PRECACHE,
627};
628
629/*
630 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
631 */
632#define FADVISE_COLD_BIT 0x01
633#define FADVISE_LOST_PINO_BIT 0x02
634#define FADVISE_ENCRYPT_BIT 0x04
635#define FADVISE_ENC_NAME_BIT 0x08
636#define FADVISE_KEEP_SIZE_BIT 0x10
637#define FADVISE_HOT_BIT 0x20
638#define FADVISE_VERITY_BIT 0x40
639
640#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
641
642#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
643#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
644#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
645#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
646#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
647#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
648#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
649#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
650#define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
651#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
652#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
653#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
654#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
655#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
656#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
657#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
658#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
659#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
660
661#define DEF_DIR_LEVEL 0
662
663enum {
664 GC_FAILURE_PIN,
665 GC_FAILURE_ATOMIC,
666 MAX_GC_FAILURE
667};
668
669struct f2fs_inode_info {
670 struct inode vfs_inode; /* serve a vfs inode */
671 unsigned long i_flags; /* keep an inode flags for ioctl */
672 unsigned char i_advise; /* use to give file attribute hints */
673 unsigned char i_dir_level; /* use for dentry level for large dir */
674 unsigned int i_current_depth; /* only for directory depth */
675 /* for gc failure statistic */
676 unsigned int i_gc_failures[MAX_GC_FAILURE];
677 unsigned int i_pino; /* parent inode number */
678 umode_t i_acl_mode; /* keep file acl mode temporarily */
679
680 /* Use below internally in f2fs*/
681 unsigned long flags; /* use to pass per-file flags */
682 struct rw_semaphore i_sem; /* protect fi info */
683 atomic_t dirty_pages; /* # of dirty pages */
684 f2fs_hash_t chash; /* hash value of given file name */
685 unsigned int clevel; /* maximum level of given file name */
686 struct task_struct *task; /* lookup and create consistency */
687 struct task_struct *cp_task; /* separate cp/wb IO stats*/
688 nid_t i_xattr_nid; /* node id that contains xattrs */
689 loff_t last_disk_size; /* lastly written file size */
690
691#ifdef CONFIG_QUOTA
692 struct dquot *i_dquot[MAXQUOTAS];
693
694 /* quota space reservation, managed internally by quota code */
695 qsize_t i_reserved_quota;
696#endif
697 struct list_head dirty_list; /* dirty list for dirs and files */
698 struct list_head gdirty_list; /* linked in global dirty list */
699 struct list_head inmem_ilist; /* list for inmem inodes */
700 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
701 struct task_struct *inmem_task; /* store inmemory task */
702 struct mutex inmem_lock; /* lock for inmemory pages */
703 struct extent_tree *extent_tree; /* cached extent_tree entry */
704
705 /* avoid racing between foreground op and gc */
706 struct rw_semaphore i_gc_rwsem[2];
707 struct rw_semaphore i_mmap_sem;
708 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
709
710 int i_extra_isize; /* size of extra space located in i_addr */
711 kprojid_t i_projid; /* id for project quota */
712 int i_inline_xattr_size; /* inline xattr size */
713 struct timespec64 i_crtime; /* inode creation time */
714 struct timespec64 i_disk_time[4];/* inode disk times */
715};
716
717static inline void get_extent_info(struct extent_info *ext,
718 struct f2fs_extent *i_ext)
719{
720 ext->fofs = le32_to_cpu(i_ext->fofs);
721 ext->blk = le32_to_cpu(i_ext->blk);
722 ext->len = le32_to_cpu(i_ext->len);
723}
724
725static inline void set_raw_extent(struct extent_info *ext,
726 struct f2fs_extent *i_ext)
727{
728 i_ext->fofs = cpu_to_le32(ext->fofs);
729 i_ext->blk = cpu_to_le32(ext->blk);
730 i_ext->len = cpu_to_le32(ext->len);
731}
732
733static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
734 u32 blk, unsigned int len)
735{
736 ei->fofs = fofs;
737 ei->blk = blk;
738 ei->len = len;
739}
740
741static inline bool __is_discard_mergeable(struct discard_info *back,
742 struct discard_info *front, unsigned int max_len)
743{
744 return (back->lstart + back->len == front->lstart) &&
745 (back->len + front->len <= max_len);
746}
747
748static inline bool __is_discard_back_mergeable(struct discard_info *cur,
749 struct discard_info *back, unsigned int max_len)
750{
751 return __is_discard_mergeable(back, cur, max_len);
752}
753
754static inline bool __is_discard_front_mergeable(struct discard_info *cur,
755 struct discard_info *front, unsigned int max_len)
756{
757 return __is_discard_mergeable(cur, front, max_len);
758}
759
760static inline bool __is_extent_mergeable(struct extent_info *back,
761 struct extent_info *front)
762{
763 return (back->fofs + back->len == front->fofs &&
764 back->blk + back->len == front->blk);
765}
766
767static inline bool __is_back_mergeable(struct extent_info *cur,
768 struct extent_info *back)
769{
770 return __is_extent_mergeable(back, cur);
771}
772
773static inline bool __is_front_mergeable(struct extent_info *cur,
774 struct extent_info *front)
775{
776 return __is_extent_mergeable(cur, front);
777}
778
779extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
780static inline void __try_update_largest_extent(struct extent_tree *et,
781 struct extent_node *en)
782{
783 if (en->ei.len > et->largest.len) {
784 et->largest = en->ei;
785 et->largest_updated = true;
786 }
787}
788
789/*
790 * For free nid management
791 */
792enum nid_state {
793 FREE_NID, /* newly added to free nid list */
794 PREALLOC_NID, /* it is preallocated */
795 MAX_NID_STATE,
796};
797
798struct f2fs_nm_info {
799 block_t nat_blkaddr; /* base disk address of NAT */
800 nid_t max_nid; /* maximum possible node ids */
801 nid_t available_nids; /* # of available node ids */
802 nid_t next_scan_nid; /* the next nid to be scanned */
803 unsigned int ram_thresh; /* control the memory footprint */
804 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
805 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
806
807 /* NAT cache management */
808 struct radix_tree_root nat_root;/* root of the nat entry cache */
809 struct radix_tree_root nat_set_root;/* root of the nat set cache */
810 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
811 struct list_head nat_entries; /* cached nat entry list (clean) */
812 spinlock_t nat_list_lock; /* protect clean nat entry list */
813 unsigned int nat_cnt; /* the # of cached nat entries */
814 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
815 unsigned int nat_blocks; /* # of nat blocks */
816
817 /* free node ids management */
818 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
819 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
820 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
821 spinlock_t nid_list_lock; /* protect nid lists ops */
822 struct mutex build_lock; /* lock for build free nids */
823 unsigned char **free_nid_bitmap;
824 unsigned char *nat_block_bitmap;
825 unsigned short *free_nid_count; /* free nid count of NAT block */
826
827 /* for checkpoint */
828 char *nat_bitmap; /* NAT bitmap pointer */
829
830 unsigned int nat_bits_blocks; /* # of nat bits blocks */
831 unsigned char *nat_bits; /* NAT bits blocks */
832 unsigned char *full_nat_bits; /* full NAT pages */
833 unsigned char *empty_nat_bits; /* empty NAT pages */
834#ifdef CONFIG_F2FS_CHECK_FS
835 char *nat_bitmap_mir; /* NAT bitmap mirror */
836#endif
837 int bitmap_size; /* bitmap size */
838};
839
840/*
841 * this structure is used as one of function parameters.
842 * all the information are dedicated to a given direct node block determined
843 * by the data offset in a file.
844 */
845struct dnode_of_data {
846 struct inode *inode; /* vfs inode pointer */
847 struct page *inode_page; /* its inode page, NULL is possible */
848 struct page *node_page; /* cached direct node page */
849 nid_t nid; /* node id of the direct node block */
850 unsigned int ofs_in_node; /* data offset in the node page */
851 bool inode_page_locked; /* inode page is locked or not */
852 bool node_changed; /* is node block changed */
853 char cur_level; /* level of hole node page */
854 char max_level; /* level of current page located */
855 block_t data_blkaddr; /* block address of the node block */
856};
857
858static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
859 struct page *ipage, struct page *npage, nid_t nid)
860{
861 memset(dn, 0, sizeof(*dn));
862 dn->inode = inode;
863 dn->inode_page = ipage;
864 dn->node_page = npage;
865 dn->nid = nid;
866}
867
868/*
869 * For SIT manager
870 *
871 * By default, there are 6 active log areas across the whole main area.
872 * When considering hot and cold data separation to reduce cleaning overhead,
873 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
874 * respectively.
875 * In the current design, you should not change the numbers intentionally.
876 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
877 * logs individually according to the underlying devices. (default: 6)
878 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
879 * data and 8 for node logs.
880 */
881#define NR_CURSEG_DATA_TYPE (3)
882#define NR_CURSEG_NODE_TYPE (3)
883#define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
884
885enum {
886 CURSEG_HOT_DATA = 0, /* directory entry blocks */
887 CURSEG_WARM_DATA, /* data blocks */
888 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
889 CURSEG_HOT_NODE, /* direct node blocks of directory files */
890 CURSEG_WARM_NODE, /* direct node blocks of normal files */
891 CURSEG_COLD_NODE, /* indirect node blocks */
892 NO_CHECK_TYPE,
893};
894
895struct flush_cmd {
896 struct completion wait;
897 struct llist_node llnode;
898 nid_t ino;
899 int ret;
900};
901
902struct flush_cmd_control {
903 struct task_struct *f2fs_issue_flush; /* flush thread */
904 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
905 atomic_t issued_flush; /* # of issued flushes */
906 atomic_t queued_flush; /* # of queued flushes */
907 struct llist_head issue_list; /* list for command issue */
908 struct llist_node *dispatch_list; /* list for command dispatch */
909};
910
911struct f2fs_sm_info {
912 struct sit_info *sit_info; /* whole segment information */
913 struct free_segmap_info *free_info; /* free segment information */
914 struct dirty_seglist_info *dirty_info; /* dirty segment information */
915 struct curseg_info *curseg_array; /* active segment information */
916
917 struct rw_semaphore curseg_lock; /* for preventing curseg change */
918
919 block_t seg0_blkaddr; /* block address of 0'th segment */
920 block_t main_blkaddr; /* start block address of main area */
921 block_t ssa_blkaddr; /* start block address of SSA area */
922
923 unsigned int segment_count; /* total # of segments */
924 unsigned int main_segments; /* # of segments in main area */
925 unsigned int reserved_segments; /* # of reserved segments */
926 unsigned int ovp_segments; /* # of overprovision segments */
927
928 /* a threshold to reclaim prefree segments */
929 unsigned int rec_prefree_segments;
930
931 /* for batched trimming */
932 unsigned int trim_sections; /* # of sections to trim */
933
934 struct list_head sit_entry_set; /* sit entry set list */
935
936 unsigned int ipu_policy; /* in-place-update policy */
937 unsigned int min_ipu_util; /* in-place-update threshold */
938 unsigned int min_fsync_blocks; /* threshold for fsync */
939 unsigned int min_seq_blocks; /* threshold for sequential blocks */
940 unsigned int min_hot_blocks; /* threshold for hot block allocation */
941 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
942
943 /* for flush command control */
944 struct flush_cmd_control *fcc_info;
945
946 /* for discard command control */
947 struct discard_cmd_control *dcc_info;
948};
949
950/*
951 * For superblock
952 */
953/*
954 * COUNT_TYPE for monitoring
955 *
956 * f2fs monitors the number of several block types such as on-writeback,
957 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
958 */
959#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
960enum count_type {
961 F2FS_DIRTY_DENTS,
962 F2FS_DIRTY_DATA,
963 F2FS_DIRTY_QDATA,
964 F2FS_DIRTY_NODES,
965 F2FS_DIRTY_META,
966 F2FS_INMEM_PAGES,
967 F2FS_DIRTY_IMETA,
968 F2FS_WB_CP_DATA,
969 F2FS_WB_DATA,
970 F2FS_RD_DATA,
971 F2FS_RD_NODE,
972 F2FS_RD_META,
973 F2FS_DIO_WRITE,
974 F2FS_DIO_READ,
975 NR_COUNT_TYPE,
976};
977
978/*
979 * The below are the page types of bios used in submit_bio().
980 * The available types are:
981 * DATA User data pages. It operates as async mode.
982 * NODE Node pages. It operates as async mode.
983 * META FS metadata pages such as SIT, NAT, CP.
984 * NR_PAGE_TYPE The number of page types.
985 * META_FLUSH Make sure the previous pages are written
986 * with waiting the bio's completion
987 * ... Only can be used with META.
988 */
989#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
990enum page_type {
991 DATA,
992 NODE,
993 META,
994 NR_PAGE_TYPE,
995 META_FLUSH,
996 INMEM, /* the below types are used by tracepoints only. */
997 INMEM_DROP,
998 INMEM_INVALIDATE,
999 INMEM_REVOKE,
1000 IPU,
1001 OPU,
1002};
1003
1004enum temp_type {
1005 HOT = 0, /* must be zero for meta bio */
1006 WARM,
1007 COLD,
1008 NR_TEMP_TYPE,
1009};
1010
1011enum need_lock_type {
1012 LOCK_REQ = 0,
1013 LOCK_DONE,
1014 LOCK_RETRY,
1015};
1016
1017enum cp_reason_type {
1018 CP_NO_NEEDED,
1019 CP_NON_REGULAR,
1020 CP_HARDLINK,
1021 CP_SB_NEED_CP,
1022 CP_WRONG_PINO,
1023 CP_NO_SPC_ROLL,
1024 CP_NODE_NEED_CP,
1025 CP_FASTBOOT_MODE,
1026 CP_SPEC_LOG_NUM,
1027 CP_RECOVER_DIR,
1028};
1029
1030enum iostat_type {
1031 APP_DIRECT_IO, /* app direct IOs */
1032 APP_BUFFERED_IO, /* app buffered IOs */
1033 APP_WRITE_IO, /* app write IOs */
1034 APP_MAPPED_IO, /* app mapped IOs */
1035 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1036 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1037 FS_META_IO, /* meta IOs from kworker/reclaimer */
1038 FS_GC_DATA_IO, /* data IOs from forground gc */
1039 FS_GC_NODE_IO, /* node IOs from forground gc */
1040 FS_CP_DATA_IO, /* data IOs from checkpoint */
1041 FS_CP_NODE_IO, /* node IOs from checkpoint */
1042 FS_CP_META_IO, /* meta IOs from checkpoint */
1043 FS_DISCARD, /* discard */
1044 NR_IO_TYPE,
1045};
1046
1047struct f2fs_io_info {
1048 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1049 nid_t ino; /* inode number */
1050 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1051 enum temp_type temp; /* contains HOT/WARM/COLD */
1052 int op; /* contains REQ_OP_ */
1053 int op_flags; /* req_flag_bits */
1054 block_t new_blkaddr; /* new block address to be written */
1055 block_t old_blkaddr; /* old block address before Cow */
1056 struct page *page; /* page to be written */
1057 struct page *encrypted_page; /* encrypted page */
1058 struct list_head list; /* serialize IOs */
1059 bool submitted; /* indicate IO submission */
1060 int need_lock; /* indicate we need to lock cp_rwsem */
1061 bool in_list; /* indicate fio is in io_list */
1062 bool is_por; /* indicate IO is from recovery or not */
1063 bool retry; /* need to reallocate block address */
1064 enum iostat_type io_type; /* io type */
1065 struct writeback_control *io_wbc; /* writeback control */
1066 struct bio **bio; /* bio for ipu */
1067 sector_t *last_block; /* last block number in bio */
1068 unsigned char version; /* version of the node */
1069};
1070
1071#define is_read_io(rw) ((rw) == READ)
1072struct f2fs_bio_info {
1073 struct f2fs_sb_info *sbi; /* f2fs superblock */
1074 struct bio *bio; /* bios to merge */
1075 sector_t last_block_in_bio; /* last block number */
1076 struct f2fs_io_info fio; /* store buffered io info. */
1077 struct rw_semaphore io_rwsem; /* blocking op for bio */
1078 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1079 struct list_head io_list; /* track fios */
1080};
1081
1082#define FDEV(i) (sbi->devs[i])
1083#define RDEV(i) (raw_super->devs[i])
1084struct f2fs_dev_info {
1085 struct block_device *bdev;
1086 char path[MAX_PATH_LEN];
1087 unsigned int total_segments;
1088 block_t start_blk;
1089 block_t end_blk;
1090#ifdef CONFIG_BLK_DEV_ZONED
1091 unsigned int nr_blkz; /* Total number of zones */
1092 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1093#endif
1094};
1095
1096enum inode_type {
1097 DIR_INODE, /* for dirty dir inode */
1098 FILE_INODE, /* for dirty regular/symlink inode */
1099 DIRTY_META, /* for all dirtied inode metadata */
1100 ATOMIC_FILE, /* for all atomic files */
1101 NR_INODE_TYPE,
1102};
1103
1104/* for inner inode cache management */
1105struct inode_management {
1106 struct radix_tree_root ino_root; /* ino entry array */
1107 spinlock_t ino_lock; /* for ino entry lock */
1108 struct list_head ino_list; /* inode list head */
1109 unsigned long ino_num; /* number of entries */
1110};
1111
1112/* For s_flag in struct f2fs_sb_info */
1113enum {
1114 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1115 SBI_IS_CLOSE, /* specify unmounting */
1116 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1117 SBI_POR_DOING, /* recovery is doing or not */
1118 SBI_NEED_SB_WRITE, /* need to recover superblock */
1119 SBI_NEED_CP, /* need to checkpoint */
1120 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1121 SBI_IS_RECOVERED, /* recovered orphan/data */
1122 SBI_CP_DISABLED, /* CP was disabled last mount */
1123 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1124 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1125 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1126 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1127 SBI_IS_RESIZEFS, /* resizefs is in process */
1128};
1129
1130enum {
1131 CP_TIME,
1132 REQ_TIME,
1133 DISCARD_TIME,
1134 GC_TIME,
1135 DISABLE_TIME,
1136 UMOUNT_DISCARD_TIMEOUT,
1137 MAX_TIME,
1138};
1139
1140enum {
1141 GC_NORMAL,
1142 GC_IDLE_CB,
1143 GC_IDLE_GREEDY,
1144 GC_URGENT,
1145};
1146
1147enum {
1148 WHINT_MODE_OFF, /* not pass down write hints */
1149 WHINT_MODE_USER, /* try to pass down hints given by users */
1150 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1151};
1152
1153enum {
1154 ALLOC_MODE_DEFAULT, /* stay default */
1155 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1156};
1157
1158enum fsync_mode {
1159 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1160 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1161 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1162};
1163
1164#ifdef CONFIG_FS_ENCRYPTION
1165#define DUMMY_ENCRYPTION_ENABLED(sbi) \
1166 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1167#else
1168#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1169#endif
1170
1171struct f2fs_sb_info {
1172 struct super_block *sb; /* pointer to VFS super block */
1173 struct proc_dir_entry *s_proc; /* proc entry */
1174 struct f2fs_super_block *raw_super; /* raw super block pointer */
1175 struct rw_semaphore sb_lock; /* lock for raw super block */
1176 int valid_super_block; /* valid super block no */
1177 unsigned long s_flag; /* flags for sbi */
1178 struct mutex writepages; /* mutex for writepages() */
1179#ifdef CONFIG_UNICODE
1180 struct unicode_map *s_encoding;
1181 __u16 s_encoding_flags;
1182#endif
1183
1184#ifdef CONFIG_BLK_DEV_ZONED
1185 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1186 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1187#endif
1188
1189 /* for node-related operations */
1190 struct f2fs_nm_info *nm_info; /* node manager */
1191 struct inode *node_inode; /* cache node blocks */
1192
1193 /* for segment-related operations */
1194 struct f2fs_sm_info *sm_info; /* segment manager */
1195
1196 /* for bio operations */
1197 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1198 /* keep migration IO order for LFS mode */
1199 struct rw_semaphore io_order_lock;
1200 mempool_t *write_io_dummy; /* Dummy pages */
1201
1202 /* for checkpoint */
1203 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1204 int cur_cp_pack; /* remain current cp pack */
1205 spinlock_t cp_lock; /* for flag in ckpt */
1206 struct inode *meta_inode; /* cache meta blocks */
1207 struct mutex cp_mutex; /* checkpoint procedure lock */
1208 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1209 struct rw_semaphore node_write; /* locking node writes */
1210 struct rw_semaphore node_change; /* locking node change */
1211 wait_queue_head_t cp_wait;
1212 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1213 long interval_time[MAX_TIME]; /* to store thresholds */
1214
1215 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1216
1217 spinlock_t fsync_node_lock; /* for node entry lock */
1218 struct list_head fsync_node_list; /* node list head */
1219 unsigned int fsync_seg_id; /* sequence id */
1220 unsigned int fsync_node_num; /* number of node entries */
1221
1222 /* for orphan inode, use 0'th array */
1223 unsigned int max_orphans; /* max orphan inodes */
1224
1225 /* for inode management */
1226 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1227 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1228 struct mutex flush_lock; /* for flush exclusion */
1229
1230 /* for extent tree cache */
1231 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1232 struct mutex extent_tree_lock; /* locking extent radix tree */
1233 struct list_head extent_list; /* lru list for shrinker */
1234 spinlock_t extent_lock; /* locking extent lru list */
1235 atomic_t total_ext_tree; /* extent tree count */
1236 struct list_head zombie_list; /* extent zombie tree list */
1237 atomic_t total_zombie_tree; /* extent zombie tree count */
1238 atomic_t total_ext_node; /* extent info count */
1239
1240 /* basic filesystem units */
1241 unsigned int log_sectors_per_block; /* log2 sectors per block */
1242 unsigned int log_blocksize; /* log2 block size */
1243 unsigned int blocksize; /* block size */
1244 unsigned int root_ino_num; /* root inode number*/
1245 unsigned int node_ino_num; /* node inode number*/
1246 unsigned int meta_ino_num; /* meta inode number*/
1247 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1248 unsigned int blocks_per_seg; /* blocks per segment */
1249 unsigned int segs_per_sec; /* segments per section */
1250 unsigned int secs_per_zone; /* sections per zone */
1251 unsigned int total_sections; /* total section count */
1252 struct mutex resize_mutex; /* for resize exclusion */
1253 unsigned int total_node_count; /* total node block count */
1254 unsigned int total_valid_node_count; /* valid node block count */
1255 loff_t max_file_blocks; /* max block index of file */
1256 int dir_level; /* directory level */
1257 int readdir_ra; /* readahead inode in readdir */
1258
1259 block_t user_block_count; /* # of user blocks */
1260 block_t total_valid_block_count; /* # of valid blocks */
1261 block_t discard_blks; /* discard command candidats */
1262 block_t last_valid_block_count; /* for recovery */
1263 block_t reserved_blocks; /* configurable reserved blocks */
1264 block_t current_reserved_blocks; /* current reserved blocks */
1265
1266 /* Additional tracking for no checkpoint mode */
1267 block_t unusable_block_count; /* # of blocks saved by last cp */
1268
1269 unsigned int nquota_files; /* # of quota sysfile */
1270 struct rw_semaphore quota_sem; /* blocking cp for flags */
1271
1272 /* # of pages, see count_type */
1273 atomic_t nr_pages[NR_COUNT_TYPE];
1274 /* # of allocated blocks */
1275 struct percpu_counter alloc_valid_block_count;
1276
1277 /* writeback control */
1278 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1279
1280 /* valid inode count */
1281 struct percpu_counter total_valid_inode_count;
1282
1283 struct f2fs_mount_info mount_opt; /* mount options */
1284
1285 /* for cleaning operations */
1286 struct mutex gc_mutex; /* mutex for GC */
1287 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1288 unsigned int cur_victim_sec; /* current victim section num */
1289 unsigned int gc_mode; /* current GC state */
1290 unsigned int next_victim_seg[2]; /* next segment in victim section */
1291 /* for skip statistic */
1292 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1293 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1294
1295 /* threshold for gc trials on pinned files */
1296 u64 gc_pin_file_threshold;
1297
1298 /* maximum # of trials to find a victim segment for SSR and GC */
1299 unsigned int max_victim_search;
1300 /* migration granularity of garbage collection, unit: segment */
1301 unsigned int migration_granularity;
1302
1303 /*
1304 * for stat information.
1305 * one is for the LFS mode, and the other is for the SSR mode.
1306 */
1307#ifdef CONFIG_F2FS_STAT_FS
1308 struct f2fs_stat_info *stat_info; /* FS status information */
1309 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1310 unsigned int segment_count[2]; /* # of allocated segments */
1311 unsigned int block_count[2]; /* # of allocated blocks */
1312 atomic_t inplace_count; /* # of inplace update */
1313 atomic64_t total_hit_ext; /* # of lookup extent cache */
1314 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1315 atomic64_t read_hit_largest; /* # of hit largest extent node */
1316 atomic64_t read_hit_cached; /* # of hit cached extent node */
1317 atomic_t inline_xattr; /* # of inline_xattr inodes */
1318 atomic_t inline_inode; /* # of inline_data inodes */
1319 atomic_t inline_dir; /* # of inline_dentry inodes */
1320 atomic_t aw_cnt; /* # of atomic writes */
1321 atomic_t vw_cnt; /* # of volatile writes */
1322 atomic_t max_aw_cnt; /* max # of atomic writes */
1323 atomic_t max_vw_cnt; /* max # of volatile writes */
1324 int bg_gc; /* background gc calls */
1325 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1326 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1327 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1328#endif
1329 spinlock_t stat_lock; /* lock for stat operations */
1330
1331 /* For app/fs IO statistics */
1332 spinlock_t iostat_lock;
1333 unsigned long long write_iostat[NR_IO_TYPE];
1334 bool iostat_enable;
1335
1336 /* For sysfs suppport */
1337 struct kobject s_kobj;
1338 struct completion s_kobj_unregister;
1339
1340 /* For shrinker support */
1341 struct list_head s_list;
1342 int s_ndevs; /* number of devices */
1343 struct f2fs_dev_info *devs; /* for device list */
1344 unsigned int dirty_device; /* for checkpoint data flush */
1345 spinlock_t dev_lock; /* protect dirty_device */
1346 struct mutex umount_mutex;
1347 unsigned int shrinker_run_no;
1348
1349 /* For write statistics */
1350 u64 sectors_written_start;
1351 u64 kbytes_written;
1352
1353 /* Reference to checksum algorithm driver via cryptoapi */
1354 struct crypto_shash *s_chksum_driver;
1355
1356 /* Precomputed FS UUID checksum for seeding other checksums */
1357 __u32 s_chksum_seed;
1358};
1359
1360struct f2fs_private_dio {
1361 struct inode *inode;
1362 void *orig_private;
1363 bio_end_io_t *orig_end_io;
1364 bool write;
1365};
1366
1367#ifdef CONFIG_F2FS_FAULT_INJECTION
1368#define f2fs_show_injection_info(type) \
1369 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pS\n", \
1370 KERN_INFO, f2fs_fault_name[type], \
1371 __func__, __builtin_return_address(0))
1372static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1373{
1374 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1375
1376 if (!ffi->inject_rate)
1377 return false;
1378
1379 if (!IS_FAULT_SET(ffi, type))
1380 return false;
1381
1382 atomic_inc(&ffi->inject_ops);
1383 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1384 atomic_set(&ffi->inject_ops, 0);
1385 return true;
1386 }
1387 return false;
1388}
1389#else
1390#define f2fs_show_injection_info(type) do { } while (0)
1391static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1392{
1393 return false;
1394}
1395#endif
1396
1397/*
1398 * Test if the mounted volume is a multi-device volume.
1399 * - For a single regular disk volume, sbi->s_ndevs is 0.
1400 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1401 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1402 */
1403static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1404{
1405 return sbi->s_ndevs > 1;
1406}
1407
1408/* For write statistics. Suppose sector size is 512 bytes,
1409 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1410 */
1411#define BD_PART_WRITTEN(s) \
1412(((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1413 (s)->sectors_written_start) >> 1)
1414
1415static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1416{
1417 unsigned long now = jiffies;
1418
1419 sbi->last_time[type] = now;
1420
1421 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1422 if (type == REQ_TIME) {
1423 sbi->last_time[DISCARD_TIME] = now;
1424 sbi->last_time[GC_TIME] = now;
1425 }
1426}
1427
1428static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1429{
1430 unsigned long interval = sbi->interval_time[type] * HZ;
1431
1432 return time_after(jiffies, sbi->last_time[type] + interval);
1433}
1434
1435static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1436 int type)
1437{
1438 unsigned long interval = sbi->interval_time[type] * HZ;
1439 unsigned int wait_ms = 0;
1440 long delta;
1441
1442 delta = (sbi->last_time[type] + interval) - jiffies;
1443 if (delta > 0)
1444 wait_ms = jiffies_to_msecs(delta);
1445
1446 return wait_ms;
1447}
1448
1449/*
1450 * Inline functions
1451 */
1452static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1453 const void *address, unsigned int length)
1454{
1455 struct {
1456 struct shash_desc shash;
1457 char ctx[4];
1458 } desc;
1459 int err;
1460
1461 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1462
1463 desc.shash.tfm = sbi->s_chksum_driver;
1464 *(u32 *)desc.ctx = crc;
1465
1466 err = crypto_shash_update(&desc.shash, address, length);
1467 BUG_ON(err);
1468
1469 return *(u32 *)desc.ctx;
1470}
1471
1472static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1473 unsigned int length)
1474{
1475 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1476}
1477
1478static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1479 void *buf, size_t buf_size)
1480{
1481 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1482}
1483
1484static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1485 const void *address, unsigned int length)
1486{
1487 return __f2fs_crc32(sbi, crc, address, length);
1488}
1489
1490static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1491{
1492 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1493}
1494
1495static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1496{
1497 return sb->s_fs_info;
1498}
1499
1500static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1501{
1502 return F2FS_SB(inode->i_sb);
1503}
1504
1505static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1506{
1507 return F2FS_I_SB(mapping->host);
1508}
1509
1510static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1511{
1512 return F2FS_M_SB(page_file_mapping(page));
1513}
1514
1515static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1516{
1517 return (struct f2fs_super_block *)(sbi->raw_super);
1518}
1519
1520static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1521{
1522 return (struct f2fs_checkpoint *)(sbi->ckpt);
1523}
1524
1525static inline struct f2fs_node *F2FS_NODE(struct page *page)
1526{
1527 return (struct f2fs_node *)page_address(page);
1528}
1529
1530static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1531{
1532 return &((struct f2fs_node *)page_address(page))->i;
1533}
1534
1535static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1536{
1537 return (struct f2fs_nm_info *)(sbi->nm_info);
1538}
1539
1540static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1541{
1542 return (struct f2fs_sm_info *)(sbi->sm_info);
1543}
1544
1545static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1546{
1547 return (struct sit_info *)(SM_I(sbi)->sit_info);
1548}
1549
1550static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1551{
1552 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1553}
1554
1555static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1556{
1557 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1558}
1559
1560static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1561{
1562 return sbi->meta_inode->i_mapping;
1563}
1564
1565static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1566{
1567 return sbi->node_inode->i_mapping;
1568}
1569
1570static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1571{
1572 return test_bit(type, &sbi->s_flag);
1573}
1574
1575static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1576{
1577 set_bit(type, &sbi->s_flag);
1578}
1579
1580static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1581{
1582 clear_bit(type, &sbi->s_flag);
1583}
1584
1585static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1586{
1587 return le64_to_cpu(cp->checkpoint_ver);
1588}
1589
1590static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1591{
1592 if (type < F2FS_MAX_QUOTAS)
1593 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1594 return 0;
1595}
1596
1597static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1598{
1599 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1600 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1601}
1602
1603static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1604{
1605 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1606
1607 return ckpt_flags & f;
1608}
1609
1610static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1611{
1612 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1613}
1614
1615static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1616{
1617 unsigned int ckpt_flags;
1618
1619 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1620 ckpt_flags |= f;
1621 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1622}
1623
1624static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1625{
1626 unsigned long flags;
1627
1628 spin_lock_irqsave(&sbi->cp_lock, flags);
1629 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1630 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1631}
1632
1633static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1634{
1635 unsigned int ckpt_flags;
1636
1637 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1638 ckpt_flags &= (~f);
1639 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1640}
1641
1642static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1643{
1644 unsigned long flags;
1645
1646 spin_lock_irqsave(&sbi->cp_lock, flags);
1647 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1648 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1649}
1650
1651static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1652{
1653 unsigned long flags;
1654 unsigned char *nat_bits;
1655
1656 /*
1657 * In order to re-enable nat_bits we need to call fsck.f2fs by
1658 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1659 * so let's rely on regular fsck or unclean shutdown.
1660 */
1661
1662 if (lock)
1663 spin_lock_irqsave(&sbi->cp_lock, flags);
1664 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1665 nat_bits = NM_I(sbi)->nat_bits;
1666 NM_I(sbi)->nat_bits = NULL;
1667 if (lock)
1668 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1669
1670 kvfree(nat_bits);
1671}
1672
1673static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1674 struct cp_control *cpc)
1675{
1676 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1677
1678 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1679}
1680
1681static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1682{
1683 down_read(&sbi->cp_rwsem);
1684}
1685
1686static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1687{
1688 return down_read_trylock(&sbi->cp_rwsem);
1689}
1690
1691static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1692{
1693 up_read(&sbi->cp_rwsem);
1694}
1695
1696static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1697{
1698 down_write(&sbi->cp_rwsem);
1699}
1700
1701static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1702{
1703 up_write(&sbi->cp_rwsem);
1704}
1705
1706static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1707{
1708 int reason = CP_SYNC;
1709
1710 if (test_opt(sbi, FASTBOOT))
1711 reason = CP_FASTBOOT;
1712 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1713 reason = CP_UMOUNT;
1714 return reason;
1715}
1716
1717static inline bool __remain_node_summaries(int reason)
1718{
1719 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1720}
1721
1722static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1723{
1724 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1725 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1726}
1727
1728/*
1729 * Check whether the inode has blocks or not
1730 */
1731static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1732{
1733 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1734
1735 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1736}
1737
1738static inline bool f2fs_has_xattr_block(unsigned int ofs)
1739{
1740 return ofs == XATTR_NODE_OFFSET;
1741}
1742
1743static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1744 struct inode *inode, bool cap)
1745{
1746 if (!inode)
1747 return true;
1748 if (!test_opt(sbi, RESERVE_ROOT))
1749 return false;
1750 if (IS_NOQUOTA(inode))
1751 return true;
1752 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1753 return true;
1754 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1755 in_group_p(F2FS_OPTION(sbi).s_resgid))
1756 return true;
1757 if (cap && capable(CAP_SYS_RESOURCE))
1758 return true;
1759 return false;
1760}
1761
1762static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1763static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1764 struct inode *inode, blkcnt_t *count)
1765{
1766 blkcnt_t diff = 0, release = 0;
1767 block_t avail_user_block_count;
1768 int ret;
1769
1770 ret = dquot_reserve_block(inode, *count);
1771 if (ret)
1772 return ret;
1773
1774 if (time_to_inject(sbi, FAULT_BLOCK)) {
1775 f2fs_show_injection_info(FAULT_BLOCK);
1776 release = *count;
1777 goto release_quota;
1778 }
1779
1780 /*
1781 * let's increase this in prior to actual block count change in order
1782 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1783 */
1784 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1785
1786 spin_lock(&sbi->stat_lock);
1787 sbi->total_valid_block_count += (block_t)(*count);
1788 avail_user_block_count = sbi->user_block_count -
1789 sbi->current_reserved_blocks;
1790
1791 if (!__allow_reserved_blocks(sbi, inode, true))
1792 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1793 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1794 if (avail_user_block_count > sbi->unusable_block_count)
1795 avail_user_block_count -= sbi->unusable_block_count;
1796 else
1797 avail_user_block_count = 0;
1798 }
1799 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1800 diff = sbi->total_valid_block_count - avail_user_block_count;
1801 if (diff > *count)
1802 diff = *count;
1803 *count -= diff;
1804 release = diff;
1805 sbi->total_valid_block_count -= diff;
1806 if (!*count) {
1807 spin_unlock(&sbi->stat_lock);
1808 goto enospc;
1809 }
1810 }
1811 spin_unlock(&sbi->stat_lock);
1812
1813 if (unlikely(release)) {
1814 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1815 dquot_release_reservation_block(inode, release);
1816 }
1817 f2fs_i_blocks_write(inode, *count, true, true);
1818 return 0;
1819
1820enospc:
1821 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1822release_quota:
1823 dquot_release_reservation_block(inode, release);
1824 return -ENOSPC;
1825}
1826
1827__printf(2, 3)
1828void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
1829
1830#define f2fs_err(sbi, fmt, ...) \
1831 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
1832#define f2fs_warn(sbi, fmt, ...) \
1833 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
1834#define f2fs_notice(sbi, fmt, ...) \
1835 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
1836#define f2fs_info(sbi, fmt, ...) \
1837 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
1838#define f2fs_debug(sbi, fmt, ...) \
1839 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
1840
1841static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1842 struct inode *inode,
1843 block_t count)
1844{
1845 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1846
1847 spin_lock(&sbi->stat_lock);
1848 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1849 sbi->total_valid_block_count -= (block_t)count;
1850 if (sbi->reserved_blocks &&
1851 sbi->current_reserved_blocks < sbi->reserved_blocks)
1852 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1853 sbi->current_reserved_blocks + count);
1854 spin_unlock(&sbi->stat_lock);
1855 if (unlikely(inode->i_blocks < sectors)) {
1856 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
1857 inode->i_ino,
1858 (unsigned long long)inode->i_blocks,
1859 (unsigned long long)sectors);
1860 set_sbi_flag(sbi, SBI_NEED_FSCK);
1861 return;
1862 }
1863 f2fs_i_blocks_write(inode, count, false, true);
1864}
1865
1866static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1867{
1868 atomic_inc(&sbi->nr_pages[count_type]);
1869
1870 if (count_type == F2FS_DIRTY_DENTS ||
1871 count_type == F2FS_DIRTY_NODES ||
1872 count_type == F2FS_DIRTY_META ||
1873 count_type == F2FS_DIRTY_QDATA ||
1874 count_type == F2FS_DIRTY_IMETA)
1875 set_sbi_flag(sbi, SBI_IS_DIRTY);
1876}
1877
1878static inline void inode_inc_dirty_pages(struct inode *inode)
1879{
1880 atomic_inc(&F2FS_I(inode)->dirty_pages);
1881 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1882 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1883 if (IS_NOQUOTA(inode))
1884 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1885}
1886
1887static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1888{
1889 atomic_dec(&sbi->nr_pages[count_type]);
1890}
1891
1892static inline void inode_dec_dirty_pages(struct inode *inode)
1893{
1894 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1895 !S_ISLNK(inode->i_mode))
1896 return;
1897
1898 atomic_dec(&F2FS_I(inode)->dirty_pages);
1899 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1900 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1901 if (IS_NOQUOTA(inode))
1902 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1903}
1904
1905static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1906{
1907 return atomic_read(&sbi->nr_pages[count_type]);
1908}
1909
1910static inline int get_dirty_pages(struct inode *inode)
1911{
1912 return atomic_read(&F2FS_I(inode)->dirty_pages);
1913}
1914
1915static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1916{
1917 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1918 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1919 sbi->log_blocks_per_seg;
1920
1921 return segs / sbi->segs_per_sec;
1922}
1923
1924static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1925{
1926 return sbi->total_valid_block_count;
1927}
1928
1929static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1930{
1931 return sbi->discard_blks;
1932}
1933
1934static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1935{
1936 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1937
1938 /* return NAT or SIT bitmap */
1939 if (flag == NAT_BITMAP)
1940 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1941 else if (flag == SIT_BITMAP)
1942 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1943
1944 return 0;
1945}
1946
1947static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1948{
1949 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1950}
1951
1952static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1953{
1954 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1955 int offset;
1956
1957 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1958 offset = (flag == SIT_BITMAP) ?
1959 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1960 /*
1961 * if large_nat_bitmap feature is enabled, leave checksum
1962 * protection for all nat/sit bitmaps.
1963 */
1964 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
1965 }
1966
1967 if (__cp_payload(sbi) > 0) {
1968 if (flag == NAT_BITMAP)
1969 return &ckpt->sit_nat_version_bitmap;
1970 else
1971 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1972 } else {
1973 offset = (flag == NAT_BITMAP) ?
1974 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1975 return &ckpt->sit_nat_version_bitmap + offset;
1976 }
1977}
1978
1979static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1980{
1981 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1982
1983 if (sbi->cur_cp_pack == 2)
1984 start_addr += sbi->blocks_per_seg;
1985 return start_addr;
1986}
1987
1988static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1989{
1990 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1991
1992 if (sbi->cur_cp_pack == 1)
1993 start_addr += sbi->blocks_per_seg;
1994 return start_addr;
1995}
1996
1997static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1998{
1999 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2000}
2001
2002static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2003{
2004 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2005}
2006
2007static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2008 struct inode *inode, bool is_inode)
2009{
2010 block_t valid_block_count;
2011 unsigned int valid_node_count, user_block_count;
2012 int err;
2013
2014 if (is_inode) {
2015 if (inode) {
2016 err = dquot_alloc_inode(inode);
2017 if (err)
2018 return err;
2019 }
2020 } else {
2021 err = dquot_reserve_block(inode, 1);
2022 if (err)
2023 return err;
2024 }
2025
2026 if (time_to_inject(sbi, FAULT_BLOCK)) {
2027 f2fs_show_injection_info(FAULT_BLOCK);
2028 goto enospc;
2029 }
2030
2031 spin_lock(&sbi->stat_lock);
2032
2033 valid_block_count = sbi->total_valid_block_count +
2034 sbi->current_reserved_blocks + 1;
2035
2036 if (!__allow_reserved_blocks(sbi, inode, false))
2037 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2038 user_block_count = sbi->user_block_count;
2039 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2040 user_block_count -= sbi->unusable_block_count;
2041
2042 if (unlikely(valid_block_count > user_block_count)) {
2043 spin_unlock(&sbi->stat_lock);
2044 goto enospc;
2045 }
2046
2047 valid_node_count = sbi->total_valid_node_count + 1;
2048 if (unlikely(valid_node_count > sbi->total_node_count)) {
2049 spin_unlock(&sbi->stat_lock);
2050 goto enospc;
2051 }
2052
2053 sbi->total_valid_node_count++;
2054 sbi->total_valid_block_count++;
2055 spin_unlock(&sbi->stat_lock);
2056
2057 if (inode) {
2058 if (is_inode)
2059 f2fs_mark_inode_dirty_sync(inode, true);
2060 else
2061 f2fs_i_blocks_write(inode, 1, true, true);
2062 }
2063
2064 percpu_counter_inc(&sbi->alloc_valid_block_count);
2065 return 0;
2066
2067enospc:
2068 if (is_inode) {
2069 if (inode)
2070 dquot_free_inode(inode);
2071 } else {
2072 dquot_release_reservation_block(inode, 1);
2073 }
2074 return -ENOSPC;
2075}
2076
2077static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2078 struct inode *inode, bool is_inode)
2079{
2080 spin_lock(&sbi->stat_lock);
2081
2082 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2083 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2084
2085 sbi->total_valid_node_count--;
2086 sbi->total_valid_block_count--;
2087 if (sbi->reserved_blocks &&
2088 sbi->current_reserved_blocks < sbi->reserved_blocks)
2089 sbi->current_reserved_blocks++;
2090
2091 spin_unlock(&sbi->stat_lock);
2092
2093 if (is_inode) {
2094 dquot_free_inode(inode);
2095 } else {
2096 if (unlikely(inode->i_blocks == 0)) {
2097 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu",
2098 inode->i_ino,
2099 (unsigned long long)inode->i_blocks);
2100 set_sbi_flag(sbi, SBI_NEED_FSCK);
2101 return;
2102 }
2103 f2fs_i_blocks_write(inode, 1, false, true);
2104 }
2105}
2106
2107static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2108{
2109 return sbi->total_valid_node_count;
2110}
2111
2112static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2113{
2114 percpu_counter_inc(&sbi->total_valid_inode_count);
2115}
2116
2117static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2118{
2119 percpu_counter_dec(&sbi->total_valid_inode_count);
2120}
2121
2122static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2123{
2124 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2125}
2126
2127static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2128 pgoff_t index, bool for_write)
2129{
2130 struct page *page;
2131
2132 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2133 if (!for_write)
2134 page = find_get_page_flags(mapping, index,
2135 FGP_LOCK | FGP_ACCESSED);
2136 else
2137 page = find_lock_page(mapping, index);
2138 if (page)
2139 return page;
2140
2141 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2142 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2143 return NULL;
2144 }
2145 }
2146
2147 if (!for_write)
2148 return grab_cache_page(mapping, index);
2149 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2150}
2151
2152static inline struct page *f2fs_pagecache_get_page(
2153 struct address_space *mapping, pgoff_t index,
2154 int fgp_flags, gfp_t gfp_mask)
2155{
2156 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2157 f2fs_show_injection_info(FAULT_PAGE_GET);
2158 return NULL;
2159 }
2160
2161 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2162}
2163
2164static inline void f2fs_copy_page(struct page *src, struct page *dst)
2165{
2166 char *src_kaddr = kmap(src);
2167 char *dst_kaddr = kmap(dst);
2168
2169 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2170 kunmap(dst);
2171 kunmap(src);
2172}
2173
2174static inline void f2fs_put_page(struct page *page, int unlock)
2175{
2176 if (!page)
2177 return;
2178
2179 if (unlock) {
2180 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2181 unlock_page(page);
2182 }
2183 put_page(page);
2184}
2185
2186static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2187{
2188 if (dn->node_page)
2189 f2fs_put_page(dn->node_page, 1);
2190 if (dn->inode_page && dn->node_page != dn->inode_page)
2191 f2fs_put_page(dn->inode_page, 0);
2192 dn->node_page = NULL;
2193 dn->inode_page = NULL;
2194}
2195
2196static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2197 size_t size)
2198{
2199 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2200}
2201
2202static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2203 gfp_t flags)
2204{
2205 void *entry;
2206
2207 entry = kmem_cache_alloc(cachep, flags);
2208 if (!entry)
2209 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2210 return entry;
2211}
2212
2213static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2214 int npages, bool no_fail)
2215{
2216 struct bio *bio;
2217
2218 if (no_fail) {
2219 /* No failure on bio allocation */
2220 bio = bio_alloc(GFP_NOIO, npages);
2221 if (!bio)
2222 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2223 return bio;
2224 }
2225 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2226 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2227 return NULL;
2228 }
2229
2230 return bio_alloc(GFP_KERNEL, npages);
2231}
2232
2233static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2234{
2235 if (sbi->gc_mode == GC_URGENT)
2236 return true;
2237
2238 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2239 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2240 get_pages(sbi, F2FS_WB_CP_DATA) ||
2241 get_pages(sbi, F2FS_DIO_READ) ||
2242 get_pages(sbi, F2FS_DIO_WRITE))
2243 return false;
2244
2245 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2246 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2247 return false;
2248
2249 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2250 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2251 return false;
2252
2253 return f2fs_time_over(sbi, type);
2254}
2255
2256static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2257 unsigned long index, void *item)
2258{
2259 while (radix_tree_insert(root, index, item))
2260 cond_resched();
2261}
2262
2263#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2264
2265static inline bool IS_INODE(struct page *page)
2266{
2267 struct f2fs_node *p = F2FS_NODE(page);
2268
2269 return RAW_IS_INODE(p);
2270}
2271
2272static inline int offset_in_addr(struct f2fs_inode *i)
2273{
2274 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2275 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2276}
2277
2278static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2279{
2280 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2281}
2282
2283static inline int f2fs_has_extra_attr(struct inode *inode);
2284static inline block_t datablock_addr(struct inode *inode,
2285 struct page *node_page, unsigned int offset)
2286{
2287 struct f2fs_node *raw_node;
2288 __le32 *addr_array;
2289 int base = 0;
2290 bool is_inode = IS_INODE(node_page);
2291
2292 raw_node = F2FS_NODE(node_page);
2293
2294 /* from GC path only */
2295 if (is_inode) {
2296 if (!inode)
2297 base = offset_in_addr(&raw_node->i);
2298 else if (f2fs_has_extra_attr(inode))
2299 base = get_extra_isize(inode);
2300 }
2301
2302 addr_array = blkaddr_in_node(raw_node);
2303 return le32_to_cpu(addr_array[base + offset]);
2304}
2305
2306static inline int f2fs_test_bit(unsigned int nr, char *addr)
2307{
2308 int mask;
2309
2310 addr += (nr >> 3);
2311 mask = 1 << (7 - (nr & 0x07));
2312 return mask & *addr;
2313}
2314
2315static inline void f2fs_set_bit(unsigned int nr, char *addr)
2316{
2317 int mask;
2318
2319 addr += (nr >> 3);
2320 mask = 1 << (7 - (nr & 0x07));
2321 *addr |= mask;
2322}
2323
2324static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2325{
2326 int mask;
2327
2328 addr += (nr >> 3);
2329 mask = 1 << (7 - (nr & 0x07));
2330 *addr &= ~mask;
2331}
2332
2333static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2334{
2335 int mask;
2336 int ret;
2337
2338 addr += (nr >> 3);
2339 mask = 1 << (7 - (nr & 0x07));
2340 ret = mask & *addr;
2341 *addr |= mask;
2342 return ret;
2343}
2344
2345static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2346{
2347 int mask;
2348 int ret;
2349
2350 addr += (nr >> 3);
2351 mask = 1 << (7 - (nr & 0x07));
2352 ret = mask & *addr;
2353 *addr &= ~mask;
2354 return ret;
2355}
2356
2357static inline void f2fs_change_bit(unsigned int nr, char *addr)
2358{
2359 int mask;
2360
2361 addr += (nr >> 3);
2362 mask = 1 << (7 - (nr & 0x07));
2363 *addr ^= mask;
2364}
2365
2366/*
2367 * On-disk inode flags (f2fs_inode::i_flags)
2368 */
2369#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2370#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2371#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2372#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2373#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2374#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2375#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2376#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2377#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2378
2379/* Flags that should be inherited by new inodes from their parent. */
2380#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2381 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2382 F2FS_CASEFOLD_FL)
2383
2384/* Flags that are appropriate for regular files (all but dir-specific ones). */
2385#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2386 F2FS_CASEFOLD_FL))
2387
2388/* Flags that are appropriate for non-directories/regular files. */
2389#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2390
2391static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2392{
2393 if (S_ISDIR(mode))
2394 return flags;
2395 else if (S_ISREG(mode))
2396 return flags & F2FS_REG_FLMASK;
2397 else
2398 return flags & F2FS_OTHER_FLMASK;
2399}
2400
2401/* used for f2fs_inode_info->flags */
2402enum {
2403 FI_NEW_INODE, /* indicate newly allocated inode */
2404 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2405 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2406 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2407 FI_INC_LINK, /* need to increment i_nlink */
2408 FI_ACL_MODE, /* indicate acl mode */
2409 FI_NO_ALLOC, /* should not allocate any blocks */
2410 FI_FREE_NID, /* free allocated nide */
2411 FI_NO_EXTENT, /* not to use the extent cache */
2412 FI_INLINE_XATTR, /* used for inline xattr */
2413 FI_INLINE_DATA, /* used for inline data*/
2414 FI_INLINE_DENTRY, /* used for inline dentry */
2415 FI_APPEND_WRITE, /* inode has appended data */
2416 FI_UPDATE_WRITE, /* inode has in-place-update data */
2417 FI_NEED_IPU, /* used for ipu per file */
2418 FI_ATOMIC_FILE, /* indicate atomic file */
2419 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2420 FI_VOLATILE_FILE, /* indicate volatile file */
2421 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2422 FI_DROP_CACHE, /* drop dirty page cache */
2423 FI_DATA_EXIST, /* indicate data exists */
2424 FI_INLINE_DOTS, /* indicate inline dot dentries */
2425 FI_DO_DEFRAG, /* indicate defragment is running */
2426 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2427 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2428 FI_HOT_DATA, /* indicate file is hot */
2429 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2430 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2431 FI_PIN_FILE, /* indicate file should not be gced */
2432 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2433 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
2434};
2435
2436static inline void __mark_inode_dirty_flag(struct inode *inode,
2437 int flag, bool set)
2438{
2439 switch (flag) {
2440 case FI_INLINE_XATTR:
2441 case FI_INLINE_DATA:
2442 case FI_INLINE_DENTRY:
2443 case FI_NEW_INODE:
2444 if (set)
2445 return;
2446 /* fall through */
2447 case FI_DATA_EXIST:
2448 case FI_INLINE_DOTS:
2449 case FI_PIN_FILE:
2450 f2fs_mark_inode_dirty_sync(inode, true);
2451 }
2452}
2453
2454static inline void set_inode_flag(struct inode *inode, int flag)
2455{
2456 if (!test_bit(flag, &F2FS_I(inode)->flags))
2457 set_bit(flag, &F2FS_I(inode)->flags);
2458 __mark_inode_dirty_flag(inode, flag, true);
2459}
2460
2461static inline int is_inode_flag_set(struct inode *inode, int flag)
2462{
2463 return test_bit(flag, &F2FS_I(inode)->flags);
2464}
2465
2466static inline void clear_inode_flag(struct inode *inode, int flag)
2467{
2468 if (test_bit(flag, &F2FS_I(inode)->flags))
2469 clear_bit(flag, &F2FS_I(inode)->flags);
2470 __mark_inode_dirty_flag(inode, flag, false);
2471}
2472
2473static inline bool f2fs_verity_in_progress(struct inode *inode)
2474{
2475 return IS_ENABLED(CONFIG_FS_VERITY) &&
2476 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2477}
2478
2479static inline void set_acl_inode(struct inode *inode, umode_t mode)
2480{
2481 F2FS_I(inode)->i_acl_mode = mode;
2482 set_inode_flag(inode, FI_ACL_MODE);
2483 f2fs_mark_inode_dirty_sync(inode, false);
2484}
2485
2486static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2487{
2488 if (inc)
2489 inc_nlink(inode);
2490 else
2491 drop_nlink(inode);
2492 f2fs_mark_inode_dirty_sync(inode, true);
2493}
2494
2495static inline void f2fs_i_blocks_write(struct inode *inode,
2496 block_t diff, bool add, bool claim)
2497{
2498 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2499 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2500
2501 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2502 if (add) {
2503 if (claim)
2504 dquot_claim_block(inode, diff);
2505 else
2506 dquot_alloc_block_nofail(inode, diff);
2507 } else {
2508 dquot_free_block(inode, diff);
2509 }
2510
2511 f2fs_mark_inode_dirty_sync(inode, true);
2512 if (clean || recover)
2513 set_inode_flag(inode, FI_AUTO_RECOVER);
2514}
2515
2516static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2517{
2518 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2519 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2520
2521 if (i_size_read(inode) == i_size)
2522 return;
2523
2524 i_size_write(inode, i_size);
2525 f2fs_mark_inode_dirty_sync(inode, true);
2526 if (clean || recover)
2527 set_inode_flag(inode, FI_AUTO_RECOVER);
2528}
2529
2530static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2531{
2532 F2FS_I(inode)->i_current_depth = depth;
2533 f2fs_mark_inode_dirty_sync(inode, true);
2534}
2535
2536static inline void f2fs_i_gc_failures_write(struct inode *inode,
2537 unsigned int count)
2538{
2539 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2540 f2fs_mark_inode_dirty_sync(inode, true);
2541}
2542
2543static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2544{
2545 F2FS_I(inode)->i_xattr_nid = xnid;
2546 f2fs_mark_inode_dirty_sync(inode, true);
2547}
2548
2549static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2550{
2551 F2FS_I(inode)->i_pino = pino;
2552 f2fs_mark_inode_dirty_sync(inode, true);
2553}
2554
2555static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2556{
2557 struct f2fs_inode_info *fi = F2FS_I(inode);
2558
2559 if (ri->i_inline & F2FS_INLINE_XATTR)
2560 set_bit(FI_INLINE_XATTR, &fi->flags);
2561 if (ri->i_inline & F2FS_INLINE_DATA)
2562 set_bit(FI_INLINE_DATA, &fi->flags);
2563 if (ri->i_inline & F2FS_INLINE_DENTRY)
2564 set_bit(FI_INLINE_DENTRY, &fi->flags);
2565 if (ri->i_inline & F2FS_DATA_EXIST)
2566 set_bit(FI_DATA_EXIST, &fi->flags);
2567 if (ri->i_inline & F2FS_INLINE_DOTS)
2568 set_bit(FI_INLINE_DOTS, &fi->flags);
2569 if (ri->i_inline & F2FS_EXTRA_ATTR)
2570 set_bit(FI_EXTRA_ATTR, &fi->flags);
2571 if (ri->i_inline & F2FS_PIN_FILE)
2572 set_bit(FI_PIN_FILE, &fi->flags);
2573}
2574
2575static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2576{
2577 ri->i_inline = 0;
2578
2579 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2580 ri->i_inline |= F2FS_INLINE_XATTR;
2581 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2582 ri->i_inline |= F2FS_INLINE_DATA;
2583 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2584 ri->i_inline |= F2FS_INLINE_DENTRY;
2585 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2586 ri->i_inline |= F2FS_DATA_EXIST;
2587 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2588 ri->i_inline |= F2FS_INLINE_DOTS;
2589 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2590 ri->i_inline |= F2FS_EXTRA_ATTR;
2591 if (is_inode_flag_set(inode, FI_PIN_FILE))
2592 ri->i_inline |= F2FS_PIN_FILE;
2593}
2594
2595static inline int f2fs_has_extra_attr(struct inode *inode)
2596{
2597 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2598}
2599
2600static inline int f2fs_has_inline_xattr(struct inode *inode)
2601{
2602 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2603}
2604
2605static inline unsigned int addrs_per_inode(struct inode *inode)
2606{
2607 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2608 get_inline_xattr_addrs(inode);
2609 return ALIGN_DOWN(addrs, 1);
2610}
2611
2612static inline unsigned int addrs_per_block(struct inode *inode)
2613{
2614 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, 1);
2615}
2616
2617static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2618{
2619 struct f2fs_inode *ri = F2FS_INODE(page);
2620
2621 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2622 get_inline_xattr_addrs(inode)]);
2623}
2624
2625static inline int inline_xattr_size(struct inode *inode)
2626{
2627 if (f2fs_has_inline_xattr(inode))
2628 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2629 return 0;
2630}
2631
2632static inline int f2fs_has_inline_data(struct inode *inode)
2633{
2634 return is_inode_flag_set(inode, FI_INLINE_DATA);
2635}
2636
2637static inline int f2fs_exist_data(struct inode *inode)
2638{
2639 return is_inode_flag_set(inode, FI_DATA_EXIST);
2640}
2641
2642static inline int f2fs_has_inline_dots(struct inode *inode)
2643{
2644 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2645}
2646
2647static inline bool f2fs_is_pinned_file(struct inode *inode)
2648{
2649 return is_inode_flag_set(inode, FI_PIN_FILE);
2650}
2651
2652static inline bool f2fs_is_atomic_file(struct inode *inode)
2653{
2654 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2655}
2656
2657static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2658{
2659 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2660}
2661
2662static inline bool f2fs_is_volatile_file(struct inode *inode)
2663{
2664 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2665}
2666
2667static inline bool f2fs_is_first_block_written(struct inode *inode)
2668{
2669 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2670}
2671
2672static inline bool f2fs_is_drop_cache(struct inode *inode)
2673{
2674 return is_inode_flag_set(inode, FI_DROP_CACHE);
2675}
2676
2677static inline void *inline_data_addr(struct inode *inode, struct page *page)
2678{
2679 struct f2fs_inode *ri = F2FS_INODE(page);
2680 int extra_size = get_extra_isize(inode);
2681
2682 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2683}
2684
2685static inline int f2fs_has_inline_dentry(struct inode *inode)
2686{
2687 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2688}
2689
2690static inline int is_file(struct inode *inode, int type)
2691{
2692 return F2FS_I(inode)->i_advise & type;
2693}
2694
2695static inline void set_file(struct inode *inode, int type)
2696{
2697 F2FS_I(inode)->i_advise |= type;
2698 f2fs_mark_inode_dirty_sync(inode, true);
2699}
2700
2701static inline void clear_file(struct inode *inode, int type)
2702{
2703 F2FS_I(inode)->i_advise &= ~type;
2704 f2fs_mark_inode_dirty_sync(inode, true);
2705}
2706
2707static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2708{
2709 bool ret;
2710
2711 if (dsync) {
2712 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2713
2714 spin_lock(&sbi->inode_lock[DIRTY_META]);
2715 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2716 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2717 return ret;
2718 }
2719 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2720 file_keep_isize(inode) ||
2721 i_size_read(inode) & ~PAGE_MASK)
2722 return false;
2723
2724 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2725 return false;
2726 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2727 return false;
2728 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2729 return false;
2730 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2731 &F2FS_I(inode)->i_crtime))
2732 return false;
2733
2734 down_read(&F2FS_I(inode)->i_sem);
2735 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2736 up_read(&F2FS_I(inode)->i_sem);
2737
2738 return ret;
2739}
2740
2741static inline bool f2fs_readonly(struct super_block *sb)
2742{
2743 return sb_rdonly(sb);
2744}
2745
2746static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2747{
2748 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2749}
2750
2751static inline bool is_dot_dotdot(const struct qstr *str)
2752{
2753 if (str->len == 1 && str->name[0] == '.')
2754 return true;
2755
2756 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2757 return true;
2758
2759 return false;
2760}
2761
2762static inline bool f2fs_may_extent_tree(struct inode *inode)
2763{
2764 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2765
2766 if (!test_opt(sbi, EXTENT_CACHE) ||
2767 is_inode_flag_set(inode, FI_NO_EXTENT))
2768 return false;
2769
2770 /*
2771 * for recovered files during mount do not create extents
2772 * if shrinker is not registered.
2773 */
2774 if (list_empty(&sbi->s_list))
2775 return false;
2776
2777 return S_ISREG(inode->i_mode);
2778}
2779
2780static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2781 size_t size, gfp_t flags)
2782{
2783 void *ret;
2784
2785 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2786 f2fs_show_injection_info(FAULT_KMALLOC);
2787 return NULL;
2788 }
2789
2790 ret = kmalloc(size, flags);
2791 if (ret)
2792 return ret;
2793
2794 return kvmalloc(size, flags);
2795}
2796
2797static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2798 size_t size, gfp_t flags)
2799{
2800 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2801}
2802
2803static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2804 size_t size, gfp_t flags)
2805{
2806 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2807 f2fs_show_injection_info(FAULT_KVMALLOC);
2808 return NULL;
2809 }
2810
2811 return kvmalloc(size, flags);
2812}
2813
2814static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2815 size_t size, gfp_t flags)
2816{
2817 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2818}
2819
2820static inline int get_extra_isize(struct inode *inode)
2821{
2822 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2823}
2824
2825static inline int get_inline_xattr_addrs(struct inode *inode)
2826{
2827 return F2FS_I(inode)->i_inline_xattr_size;
2828}
2829
2830#define f2fs_get_inode_mode(i) \
2831 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2832 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2833
2834#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2835 (offsetof(struct f2fs_inode, i_extra_end) - \
2836 offsetof(struct f2fs_inode, i_extra_isize)) \
2837
2838#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2839#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2840 ((offsetof(typeof(*(f2fs_inode)), field) + \
2841 sizeof((f2fs_inode)->field)) \
2842 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
2843
2844static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2845{
2846 int i;
2847
2848 spin_lock(&sbi->iostat_lock);
2849 for (i = 0; i < NR_IO_TYPE; i++)
2850 sbi->write_iostat[i] = 0;
2851 spin_unlock(&sbi->iostat_lock);
2852}
2853
2854static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2855 enum iostat_type type, unsigned long long io_bytes)
2856{
2857 if (!sbi->iostat_enable)
2858 return;
2859 spin_lock(&sbi->iostat_lock);
2860 sbi->write_iostat[type] += io_bytes;
2861
2862 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2863 sbi->write_iostat[APP_BUFFERED_IO] =
2864 sbi->write_iostat[APP_WRITE_IO] -
2865 sbi->write_iostat[APP_DIRECT_IO];
2866 spin_unlock(&sbi->iostat_lock);
2867}
2868
2869#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2870
2871#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
2872
2873bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2874 block_t blkaddr, int type);
2875static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2876 block_t blkaddr, int type)
2877{
2878 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2879 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
2880 blkaddr, type);
2881 f2fs_bug_on(sbi, 1);
2882 }
2883}
2884
2885static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2886{
2887 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2888 return false;
2889 return true;
2890}
2891
2892static inline void f2fs_set_page_private(struct page *page,
2893 unsigned long data)
2894{
2895 if (PagePrivate(page))
2896 return;
2897
2898 get_page(page);
2899 SetPagePrivate(page);
2900 set_page_private(page, data);
2901}
2902
2903static inline void f2fs_clear_page_private(struct page *page)
2904{
2905 if (!PagePrivate(page))
2906 return;
2907
2908 set_page_private(page, 0);
2909 ClearPagePrivate(page);
2910 f2fs_put_page(page, 0);
2911}
2912
2913/*
2914 * file.c
2915 */
2916int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2917void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2918int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
2919int f2fs_truncate(struct inode *inode);
2920int f2fs_getattr(const struct path *path, struct kstat *stat,
2921 u32 request_mask, unsigned int flags);
2922int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2923int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2924void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2925int f2fs_precache_extents(struct inode *inode);
2926long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2927long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2928int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
2929int f2fs_pin_file_control(struct inode *inode, bool inc);
2930
2931/*
2932 * inode.c
2933 */
2934void f2fs_set_inode_flags(struct inode *inode);
2935bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2936void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2937struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2938struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2939int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2940void f2fs_update_inode(struct inode *inode, struct page *node_page);
2941void f2fs_update_inode_page(struct inode *inode);
2942int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2943void f2fs_evict_inode(struct inode *inode);
2944void f2fs_handle_failed_inode(struct inode *inode);
2945
2946/*
2947 * namei.c
2948 */
2949int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2950 bool hot, bool set);
2951struct dentry *f2fs_get_parent(struct dentry *child);
2952
2953extern int f2fs_ci_compare(const struct inode *parent,
2954 const struct qstr *name,
2955 const struct qstr *entry,
2956 bool quick);
2957
2958/*
2959 * dir.c
2960 */
2961unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2962struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2963 f2fs_hash_t namehash, int *max_slots,
2964 struct f2fs_dentry_ptr *d);
2965int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2966 unsigned int start_pos, struct fscrypt_str *fstr);
2967void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2968 struct f2fs_dentry_ptr *d);
2969struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2970 const struct qstr *new_name,
2971 const struct qstr *orig_name, struct page *dpage);
2972void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2973 unsigned int current_depth);
2974int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2975void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2976struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2977 struct fscrypt_name *fname, struct page **res_page);
2978struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2979 const struct qstr *child, struct page **res_page);
2980struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2981ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2982 struct page **page);
2983void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2984 struct page *page, struct inode *inode);
2985void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2986 const struct qstr *name, f2fs_hash_t name_hash,
2987 unsigned int bit_pos);
2988int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2989 const struct qstr *orig_name,
2990 struct inode *inode, nid_t ino, umode_t mode);
2991int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2992 struct inode *inode, nid_t ino, umode_t mode);
2993int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2994 struct inode *inode, nid_t ino, umode_t mode);
2995void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2996 struct inode *dir, struct inode *inode);
2997int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2998bool f2fs_empty_dir(struct inode *dir);
2999
3000static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3001{
3002 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3003 inode, inode->i_ino, inode->i_mode);
3004}
3005
3006/*
3007 * super.c
3008 */
3009int f2fs_inode_dirtied(struct inode *inode, bool sync);
3010void f2fs_inode_synced(struct inode *inode);
3011int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3012int f2fs_quota_sync(struct super_block *sb, int type);
3013void f2fs_quota_off_umount(struct super_block *sb);
3014int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3015int f2fs_sync_fs(struct super_block *sb, int sync);
3016int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3017
3018/*
3019 * hash.c
3020 */
3021f2fs_hash_t f2fs_dentry_hash(const struct inode *dir,
3022 const struct qstr *name_info, struct fscrypt_name *fname);
3023
3024/*
3025 * node.c
3026 */
3027struct dnode_of_data;
3028struct node_info;
3029
3030int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3031bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3032bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3033void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3034void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3035void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3036int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3037bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3038bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3039int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3040 struct node_info *ni);
3041pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3042int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3043int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3044int f2fs_truncate_xattr_node(struct inode *inode);
3045int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3046 unsigned int seq_id);
3047int f2fs_remove_inode_page(struct inode *inode);
3048struct page *f2fs_new_inode_page(struct inode *inode);
3049struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3050void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3051struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3052struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3053int f2fs_move_node_page(struct page *node_page, int gc_type);
3054int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3055 struct writeback_control *wbc, bool atomic,
3056 unsigned int *seq_id);
3057int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3058 struct writeback_control *wbc,
3059 bool do_balance, enum iostat_type io_type);
3060int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3061bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3062void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3063void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3064int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3065void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3066int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3067int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3068int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3069 unsigned int segno, struct f2fs_summary_block *sum);
3070int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3071int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3072void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3073int __init f2fs_create_node_manager_caches(void);
3074void f2fs_destroy_node_manager_caches(void);
3075
3076/*
3077 * segment.c
3078 */
3079bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3080void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3081void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3082void f2fs_drop_inmem_pages(struct inode *inode);
3083void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3084int f2fs_commit_inmem_pages(struct inode *inode);
3085void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3086void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
3087int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3088int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3089int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3090void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3091void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3092bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3093void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3094void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3095bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3096void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3097 struct cp_control *cpc);
3098void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3099block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3100int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3101void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3102int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3103void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3104 unsigned int start, unsigned int end);
3105void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3106int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3107bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3108 struct cp_control *cpc);
3109struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3110void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3111 block_t blk_addr);
3112void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3113 enum iostat_type io_type);
3114void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3115void f2fs_outplace_write_data(struct dnode_of_data *dn,
3116 struct f2fs_io_info *fio);
3117int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3118void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3119 block_t old_blkaddr, block_t new_blkaddr,
3120 bool recover_curseg, bool recover_newaddr);
3121void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3122 block_t old_addr, block_t new_addr,
3123 unsigned char version, bool recover_curseg,
3124 bool recover_newaddr);
3125void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3126 block_t old_blkaddr, block_t *new_blkaddr,
3127 struct f2fs_summary *sum, int type,
3128 struct f2fs_io_info *fio, bool add_list);
3129void f2fs_wait_on_page_writeback(struct page *page,
3130 enum page_type type, bool ordered, bool locked);
3131void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3132void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3133 block_t len);
3134void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3135void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3136int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3137 unsigned int val, int alloc);
3138void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3139int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3140void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3141int __init f2fs_create_segment_manager_caches(void);
3142void f2fs_destroy_segment_manager_caches(void);
3143int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3144enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3145 enum page_type type, enum temp_type temp);
3146
3147/*
3148 * checkpoint.c
3149 */
3150void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3151struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3152struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3153struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3154struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3155bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3156 block_t blkaddr, int type);
3157int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3158 int type, bool sync);
3159void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3160long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3161 long nr_to_write, enum iostat_type io_type);
3162void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3163void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3164void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3165bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3166void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3167 unsigned int devidx, int type);
3168bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3169 unsigned int devidx, int type);
3170int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3171int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3172void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3173void f2fs_add_orphan_inode(struct inode *inode);
3174void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3175int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3176int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3177void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3178void f2fs_remove_dirty_inode(struct inode *inode);
3179int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3180void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3181int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3182void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3183int __init f2fs_create_checkpoint_caches(void);
3184void f2fs_destroy_checkpoint_caches(void);
3185
3186/*
3187 * data.c
3188 */
3189int f2fs_init_post_read_processing(void);
3190void f2fs_destroy_post_read_processing(void);
3191void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3192void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3193 struct inode *inode, struct page *page,
3194 nid_t ino, enum page_type type);
3195void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3196int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3197int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3198void f2fs_submit_page_write(struct f2fs_io_info *fio);
3199struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3200 block_t blk_addr, struct bio *bio);
3201int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3202void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3203void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3204int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3205int f2fs_reserve_new_block(struct dnode_of_data *dn);
3206int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3207int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3208int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3209struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3210 int op_flags, bool for_write);
3211struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3212struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3213 bool for_write);
3214struct page *f2fs_get_new_data_page(struct inode *inode,
3215 struct page *ipage, pgoff_t index, bool new_i_size);
3216int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3217void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3218int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3219 int create, int flag);
3220int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3221 u64 start, u64 len);
3222bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3223bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3224void f2fs_invalidate_page(struct page *page, unsigned int offset,
3225 unsigned int length);
3226int f2fs_release_page(struct page *page, gfp_t wait);
3227#ifdef CONFIG_MIGRATION
3228int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3229 struct page *page, enum migrate_mode mode);
3230#endif
3231bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3232void f2fs_clear_page_cache_dirty_tag(struct page *page);
3233
3234/*
3235 * gc.c
3236 */
3237int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3238void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3239block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3240int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3241 unsigned int segno);
3242void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3243int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3244
3245/*
3246 * recovery.c
3247 */
3248int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3249bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3250
3251/*
3252 * debug.c
3253 */
3254#ifdef CONFIG_F2FS_STAT_FS
3255struct f2fs_stat_info {
3256 struct list_head stat_list;
3257 struct f2fs_sb_info *sbi;
3258 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3259 int main_area_segs, main_area_sections, main_area_zones;
3260 unsigned long long hit_largest, hit_cached, hit_rbtree;
3261 unsigned long long hit_total, total_ext;
3262 int ext_tree, zombie_tree, ext_node;
3263 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3264 int ndirty_data, ndirty_qdata;
3265 int inmem_pages;
3266 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3267 int nats, dirty_nats, sits, dirty_sits;
3268 int free_nids, avail_nids, alloc_nids;
3269 int total_count, utilization;
3270 int bg_gc, nr_wb_cp_data, nr_wb_data;
3271 int nr_rd_data, nr_rd_node, nr_rd_meta;
3272 int nr_dio_read, nr_dio_write;
3273 unsigned int io_skip_bggc, other_skip_bggc;
3274 int nr_flushing, nr_flushed, flush_list_empty;
3275 int nr_discarding, nr_discarded;
3276 int nr_discard_cmd;
3277 unsigned int undiscard_blks;
3278 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3279 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3280 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3281 unsigned int bimodal, avg_vblocks;
3282 int util_free, util_valid, util_invalid;
3283 int rsvd_segs, overp_segs;
3284 int dirty_count, node_pages, meta_pages;
3285 int prefree_count, call_count, cp_count, bg_cp_count;
3286 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3287 int bg_node_segs, bg_data_segs;
3288 int tot_blks, data_blks, node_blks;
3289 int bg_data_blks, bg_node_blks;
3290 unsigned long long skipped_atomic_files[2];
3291 int curseg[NR_CURSEG_TYPE];
3292 int cursec[NR_CURSEG_TYPE];
3293 int curzone[NR_CURSEG_TYPE];
3294
3295 unsigned int meta_count[META_MAX];
3296 unsigned int segment_count[2];
3297 unsigned int block_count[2];
3298 unsigned int inplace_count;
3299 unsigned long long base_mem, cache_mem, page_mem;
3300};
3301
3302static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3303{
3304 return (struct f2fs_stat_info *)sbi->stat_info;
3305}
3306
3307#define stat_inc_cp_count(si) ((si)->cp_count++)
3308#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3309#define stat_inc_call_count(si) ((si)->call_count++)
3310#define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3311#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3312#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3313#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3314#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3315#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3316#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3317#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3318#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3319#define stat_inc_inline_xattr(inode) \
3320 do { \
3321 if (f2fs_has_inline_xattr(inode)) \
3322 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3323 } while (0)
3324#define stat_dec_inline_xattr(inode) \
3325 do { \
3326 if (f2fs_has_inline_xattr(inode)) \
3327 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3328 } while (0)
3329#define stat_inc_inline_inode(inode) \
3330 do { \
3331 if (f2fs_has_inline_data(inode)) \
3332 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3333 } while (0)
3334#define stat_dec_inline_inode(inode) \
3335 do { \
3336 if (f2fs_has_inline_data(inode)) \
3337 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3338 } while (0)
3339#define stat_inc_inline_dir(inode) \
3340 do { \
3341 if (f2fs_has_inline_dentry(inode)) \
3342 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3343 } while (0)
3344#define stat_dec_inline_dir(inode) \
3345 do { \
3346 if (f2fs_has_inline_dentry(inode)) \
3347 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3348 } while (0)
3349#define stat_inc_meta_count(sbi, blkaddr) \
3350 do { \
3351 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3352 atomic_inc(&(sbi)->meta_count[META_CP]); \
3353 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3354 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3355 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3356 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3357 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3358 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3359 } while (0)
3360#define stat_inc_seg_type(sbi, curseg) \
3361 ((sbi)->segment_count[(curseg)->alloc_type]++)
3362#define stat_inc_block_count(sbi, curseg) \
3363 ((sbi)->block_count[(curseg)->alloc_type]++)
3364#define stat_inc_inplace_blocks(sbi) \
3365 (atomic_inc(&(sbi)->inplace_count))
3366#define stat_inc_atomic_write(inode) \
3367 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3368#define stat_dec_atomic_write(inode) \
3369 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3370#define stat_update_max_atomic_write(inode) \
3371 do { \
3372 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3373 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3374 if (cur > max) \
3375 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3376 } while (0)
3377#define stat_inc_volatile_write(inode) \
3378 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3379#define stat_dec_volatile_write(inode) \
3380 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3381#define stat_update_max_volatile_write(inode) \
3382 do { \
3383 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3384 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3385 if (cur > max) \
3386 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3387 } while (0)
3388#define stat_inc_seg_count(sbi, type, gc_type) \
3389 do { \
3390 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3391 si->tot_segs++; \
3392 if ((type) == SUM_TYPE_DATA) { \
3393 si->data_segs++; \
3394 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3395 } else { \
3396 si->node_segs++; \
3397 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3398 } \
3399 } while (0)
3400
3401#define stat_inc_tot_blk_count(si, blks) \
3402 ((si)->tot_blks += (blks))
3403
3404#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3405 do { \
3406 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3407 stat_inc_tot_blk_count(si, blks); \
3408 si->data_blks += (blks); \
3409 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3410 } while (0)
3411
3412#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3413 do { \
3414 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3415 stat_inc_tot_blk_count(si, blks); \
3416 si->node_blks += (blks); \
3417 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3418 } while (0)
3419
3420int f2fs_build_stats(struct f2fs_sb_info *sbi);
3421void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3422void __init f2fs_create_root_stats(void);
3423void f2fs_destroy_root_stats(void);
3424#else
3425#define stat_inc_cp_count(si) do { } while (0)
3426#define stat_inc_bg_cp_count(si) do { } while (0)
3427#define stat_inc_call_count(si) do { } while (0)
3428#define stat_inc_bggc_count(si) do { } while (0)
3429#define stat_io_skip_bggc_count(sbi) do { } while (0)
3430#define stat_other_skip_bggc_count(sbi) do { } while (0)
3431#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3432#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3433#define stat_inc_total_hit(sb) do { } while (0)
3434#define stat_inc_rbtree_node_hit(sb) do { } while (0)
3435#define stat_inc_largest_node_hit(sbi) do { } while (0)
3436#define stat_inc_cached_node_hit(sbi) do { } while (0)
3437#define stat_inc_inline_xattr(inode) do { } while (0)
3438#define stat_dec_inline_xattr(inode) do { } while (0)
3439#define stat_inc_inline_inode(inode) do { } while (0)
3440#define stat_dec_inline_inode(inode) do { } while (0)
3441#define stat_inc_inline_dir(inode) do { } while (0)
3442#define stat_dec_inline_dir(inode) do { } while (0)
3443#define stat_inc_atomic_write(inode) do { } while (0)
3444#define stat_dec_atomic_write(inode) do { } while (0)
3445#define stat_update_max_atomic_write(inode) do { } while (0)
3446#define stat_inc_volatile_write(inode) do { } while (0)
3447#define stat_dec_volatile_write(inode) do { } while (0)
3448#define stat_update_max_volatile_write(inode) do { } while (0)
3449#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3450#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3451#define stat_inc_block_count(sbi, curseg) do { } while (0)
3452#define stat_inc_inplace_blocks(sbi) do { } while (0)
3453#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3454#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3455#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3456#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3457
3458static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3459static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3460static inline void __init f2fs_create_root_stats(void) { }
3461static inline void f2fs_destroy_root_stats(void) { }
3462#endif
3463
3464extern const struct file_operations f2fs_dir_operations;
3465#ifdef CONFIG_UNICODE
3466extern const struct dentry_operations f2fs_dentry_ops;
3467#endif
3468extern const struct file_operations f2fs_file_operations;
3469extern const struct inode_operations f2fs_file_inode_operations;
3470extern const struct address_space_operations f2fs_dblock_aops;
3471extern const struct address_space_operations f2fs_node_aops;
3472extern const struct address_space_operations f2fs_meta_aops;
3473extern const struct inode_operations f2fs_dir_inode_operations;
3474extern const struct inode_operations f2fs_symlink_inode_operations;
3475extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3476extern const struct inode_operations f2fs_special_inode_operations;
3477extern struct kmem_cache *f2fs_inode_entry_slab;
3478
3479/*
3480 * inline.c
3481 */
3482bool f2fs_may_inline_data(struct inode *inode);
3483bool f2fs_may_inline_dentry(struct inode *inode);
3484void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3485void f2fs_truncate_inline_inode(struct inode *inode,
3486 struct page *ipage, u64 from);
3487int f2fs_read_inline_data(struct inode *inode, struct page *page);
3488int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3489int f2fs_convert_inline_inode(struct inode *inode);
3490int f2fs_write_inline_data(struct inode *inode, struct page *page);
3491bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3492struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3493 struct fscrypt_name *fname, struct page **res_page);
3494int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3495 struct page *ipage);
3496int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3497 const struct qstr *orig_name,
3498 struct inode *inode, nid_t ino, umode_t mode);
3499void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3500 struct page *page, struct inode *dir,
3501 struct inode *inode);
3502bool f2fs_empty_inline_dir(struct inode *dir);
3503int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3504 struct fscrypt_str *fstr);
3505int f2fs_inline_data_fiemap(struct inode *inode,
3506 struct fiemap_extent_info *fieinfo,
3507 __u64 start, __u64 len);
3508
3509/*
3510 * shrinker.c
3511 */
3512unsigned long f2fs_shrink_count(struct shrinker *shrink,
3513 struct shrink_control *sc);
3514unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3515 struct shrink_control *sc);
3516void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3517void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3518
3519/*
3520 * extent_cache.c
3521 */
3522struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3523 struct rb_entry *cached_re, unsigned int ofs);
3524struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3525 struct rb_root_cached *root,
3526 struct rb_node **parent,
3527 unsigned int ofs, bool *leftmost);
3528struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3529 struct rb_entry *cached_re, unsigned int ofs,
3530 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3531 struct rb_node ***insert_p, struct rb_node **insert_parent,
3532 bool force, bool *leftmost);
3533bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3534 struct rb_root_cached *root);
3535unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3536bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3537void f2fs_drop_extent_tree(struct inode *inode);
3538unsigned int f2fs_destroy_extent_node(struct inode *inode);
3539void f2fs_destroy_extent_tree(struct inode *inode);
3540bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3541 struct extent_info *ei);
3542void f2fs_update_extent_cache(struct dnode_of_data *dn);
3543void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3544 pgoff_t fofs, block_t blkaddr, unsigned int len);
3545void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3546int __init f2fs_create_extent_cache(void);
3547void f2fs_destroy_extent_cache(void);
3548
3549/*
3550 * sysfs.c
3551 */
3552int __init f2fs_init_sysfs(void);
3553void f2fs_exit_sysfs(void);
3554int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3555void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3556
3557/* verity.c */
3558extern const struct fsverity_operations f2fs_verityops;
3559
3560/*
3561 * crypto support
3562 */
3563static inline bool f2fs_encrypted_file(struct inode *inode)
3564{
3565 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3566}
3567
3568static inline void f2fs_set_encrypted_inode(struct inode *inode)
3569{
3570#ifdef CONFIG_FS_ENCRYPTION
3571 file_set_encrypt(inode);
3572 f2fs_set_inode_flags(inode);
3573#endif
3574}
3575
3576/*
3577 * Returns true if the reads of the inode's data need to undergo some
3578 * postprocessing step, like decryption or authenticity verification.
3579 */
3580static inline bool f2fs_post_read_required(struct inode *inode)
3581{
3582 return f2fs_encrypted_file(inode) || fsverity_active(inode);
3583}
3584
3585#define F2FS_FEATURE_FUNCS(name, flagname) \
3586static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3587{ \
3588 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3589}
3590
3591F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3592F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3593F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3594F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3595F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3596F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3597F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3598F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3599F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3600F2FS_FEATURE_FUNCS(verity, VERITY);
3601F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3602F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
3603
3604#ifdef CONFIG_BLK_DEV_ZONED
3605static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
3606 block_t blkaddr)
3607{
3608 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3609
3610 return test_bit(zno, FDEV(devi).blkz_seq);
3611}
3612#endif
3613
3614static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3615{
3616 return f2fs_sb_has_blkzoned(sbi);
3617}
3618
3619static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
3620{
3621 return blk_queue_discard(bdev_get_queue(bdev)) ||
3622 bdev_is_zoned(bdev);
3623}
3624
3625static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3626{
3627 int i;
3628
3629 if (!f2fs_is_multi_device(sbi))
3630 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
3631
3632 for (i = 0; i < sbi->s_ndevs; i++)
3633 if (f2fs_bdev_support_discard(FDEV(i).bdev))
3634 return true;
3635 return false;
3636}
3637
3638static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3639{
3640 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3641 f2fs_hw_should_discard(sbi);
3642}
3643
3644static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
3645{
3646 int i;
3647
3648 if (!f2fs_is_multi_device(sbi))
3649 return bdev_read_only(sbi->sb->s_bdev);
3650
3651 for (i = 0; i < sbi->s_ndevs; i++)
3652 if (bdev_read_only(FDEV(i).bdev))
3653 return true;
3654 return false;
3655}
3656
3657
3658static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3659{
3660 clear_opt(sbi, ADAPTIVE);
3661 clear_opt(sbi, LFS);
3662
3663 switch (mt) {
3664 case F2FS_MOUNT_ADAPTIVE:
3665 set_opt(sbi, ADAPTIVE);
3666 break;
3667 case F2FS_MOUNT_LFS:
3668 set_opt(sbi, LFS);
3669 break;
3670 }
3671}
3672
3673static inline bool f2fs_may_encrypt(struct inode *inode)
3674{
3675#ifdef CONFIG_FS_ENCRYPTION
3676 umode_t mode = inode->i_mode;
3677
3678 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3679#else
3680 return false;
3681#endif
3682}
3683
3684static inline int block_unaligned_IO(struct inode *inode,
3685 struct kiocb *iocb, struct iov_iter *iter)
3686{
3687 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
3688 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
3689 loff_t offset = iocb->ki_pos;
3690 unsigned long align = offset | iov_iter_alignment(iter);
3691
3692 return align & blocksize_mask;
3693}
3694
3695static inline int allow_outplace_dio(struct inode *inode,
3696 struct kiocb *iocb, struct iov_iter *iter)
3697{
3698 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3699 int rw = iov_iter_rw(iter);
3700
3701 return (test_opt(sbi, LFS) && (rw == WRITE) &&
3702 !block_unaligned_IO(inode, iocb, iter));
3703}
3704
3705static inline bool f2fs_force_buffered_io(struct inode *inode,
3706 struct kiocb *iocb, struct iov_iter *iter)
3707{
3708 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3709 int rw = iov_iter_rw(iter);
3710
3711 if (f2fs_post_read_required(inode))
3712 return true;
3713 if (f2fs_is_multi_device(sbi))
3714 return true;
3715 /*
3716 * for blkzoned device, fallback direct IO to buffered IO, so
3717 * all IOs can be serialized by log-structured write.
3718 */
3719 if (f2fs_sb_has_blkzoned(sbi))
3720 return true;
3721 if (test_opt(sbi, LFS) && (rw == WRITE)) {
3722 if (block_unaligned_IO(inode, iocb, iter))
3723 return true;
3724 if (F2FS_IO_ALIGNED(sbi))
3725 return true;
3726 }
3727 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
3728 !IS_SWAPFILE(inode))
3729 return true;
3730
3731 return false;
3732}
3733
3734#ifdef CONFIG_F2FS_FAULT_INJECTION
3735extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3736 unsigned int type);
3737#else
3738#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3739#endif
3740
3741static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
3742{
3743#ifdef CONFIG_QUOTA
3744 if (f2fs_sb_has_quota_ino(sbi))
3745 return true;
3746 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
3747 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
3748 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
3749 return true;
3750#endif
3751 return false;
3752}
3753
3754#define EFSBADCRC EBADMSG /* Bad CRC detected */
3755#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3756
3757#endif /* _LINUX_F2FS_H */