]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/blob - fs/f2fs/f2fs.h
f2fs: add quick mode of checkpoint=disable for QA
[mirror_ubuntu-hirsute-kernel.git] / fs / f2fs / f2fs.h
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 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_F2FS_FS_ENCRYPTION)
28 #include <linux/fscrypt.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
42 enum {
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
64 struct f2fs_fault_info {
65 atomic_t inject_ops;
66 unsigned int inject_rate;
67 unsigned int inject_type;
68 };
69
70 extern 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
113 typedef u32 block_t; /*
114 * should not change u32, since it is the on-disk block
115 * address format, __le32.
116 */
117 typedef u32 nid_t;
118
119 struct 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 };
141
142 #define F2FS_FEATURE_ENCRYPT 0x0001
143 #define F2FS_FEATURE_BLKZONED 0x0002
144 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
145 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
146 #define F2FS_FEATURE_PRJQUOTA 0x0010
147 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
148 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
149 #define F2FS_FEATURE_QUOTA_INO 0x0080
150 #define F2FS_FEATURE_INODE_CRTIME 0x0100
151 #define F2FS_FEATURE_LOST_FOUND 0x0200
152 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */
153 #define F2FS_FEATURE_SB_CHKSUM 0x0800
154
155 #define __F2FS_HAS_FEATURE(raw_super, mask) \
156 ((raw_super->feature & cpu_to_le32(mask)) != 0)
157 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
158 #define F2FS_SET_FEATURE(sbi, mask) \
159 (sbi->raw_super->feature |= cpu_to_le32(mask))
160 #define F2FS_CLEAR_FEATURE(sbi, mask) \
161 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
162
163 /*
164 * Default values for user and/or group using reserved blocks
165 */
166 #define F2FS_DEF_RESUID 0
167 #define F2FS_DEF_RESGID 0
168
169 /*
170 * For checkpoint manager
171 */
172 enum {
173 NAT_BITMAP,
174 SIT_BITMAP
175 };
176
177 #define CP_UMOUNT 0x00000001
178 #define CP_FASTBOOT 0x00000002
179 #define CP_SYNC 0x00000004
180 #define CP_RECOVERY 0x00000008
181 #define CP_DISCARD 0x00000010
182 #define CP_TRIMMED 0x00000020
183 #define CP_PAUSE 0x00000040
184
185 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
186 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
187 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
188 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
189 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
190 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
191 #define DEF_CP_INTERVAL 60 /* 60 secs */
192 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
193 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
194 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
195 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
196
197 struct cp_control {
198 int reason;
199 __u64 trim_start;
200 __u64 trim_end;
201 __u64 trim_minlen;
202 };
203
204 /*
205 * indicate meta/data type
206 */
207 enum {
208 META_CP,
209 META_NAT,
210 META_SIT,
211 META_SSA,
212 META_MAX,
213 META_POR,
214 DATA_GENERIC,
215 META_GENERIC,
216 };
217
218 /* for the list of ino */
219 enum {
220 ORPHAN_INO, /* for orphan ino list */
221 APPEND_INO, /* for append ino list */
222 UPDATE_INO, /* for update ino list */
223 TRANS_DIR_INO, /* for trasactions dir ino list */
224 FLUSH_INO, /* for multiple device flushing */
225 MAX_INO_ENTRY, /* max. list */
226 };
227
228 struct ino_entry {
229 struct list_head list; /* list head */
230 nid_t ino; /* inode number */
231 unsigned int dirty_device; /* dirty device bitmap */
232 };
233
234 /* for the list of inodes to be GCed */
235 struct inode_entry {
236 struct list_head list; /* list head */
237 struct inode *inode; /* vfs inode pointer */
238 };
239
240 struct fsync_node_entry {
241 struct list_head list; /* list head */
242 struct page *page; /* warm node page pointer */
243 unsigned int seq_id; /* sequence id */
244 };
245
246 /* for the bitmap indicate blocks to be discarded */
247 struct discard_entry {
248 struct list_head list; /* list head */
249 block_t start_blkaddr; /* start blockaddr of current segment */
250 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
251 };
252
253 /* default discard granularity of inner discard thread, unit: block count */
254 #define DEFAULT_DISCARD_GRANULARITY 16
255
256 /* max discard pend list number */
257 #define MAX_PLIST_NUM 512
258 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
259 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
260
261 enum {
262 D_PREP, /* initial */
263 D_PARTIAL, /* partially submitted */
264 D_SUBMIT, /* all submitted */
265 D_DONE, /* finished */
266 };
267
268 struct discard_info {
269 block_t lstart; /* logical start address */
270 block_t len; /* length */
271 block_t start; /* actual start address in dev */
272 };
273
274 struct discard_cmd {
275 struct rb_node rb_node; /* rb node located in rb-tree */
276 union {
277 struct {
278 block_t lstart; /* logical start address */
279 block_t len; /* length */
280 block_t start; /* actual start address in dev */
281 };
282 struct discard_info di; /* discard info */
283
284 };
285 struct list_head list; /* command list */
286 struct completion wait; /* compleation */
287 struct block_device *bdev; /* bdev */
288 unsigned short ref; /* reference count */
289 unsigned char state; /* state */
290 unsigned char queued; /* queued discard */
291 int error; /* bio error */
292 spinlock_t lock; /* for state/bio_ref updating */
293 unsigned short bio_ref; /* bio reference count */
294 };
295
296 enum {
297 DPOLICY_BG,
298 DPOLICY_FORCE,
299 DPOLICY_FSTRIM,
300 DPOLICY_UMOUNT,
301 MAX_DPOLICY,
302 };
303
304 struct discard_policy {
305 int type; /* type of discard */
306 unsigned int min_interval; /* used for candidates exist */
307 unsigned int mid_interval; /* used for device busy */
308 unsigned int max_interval; /* used for candidates not exist */
309 unsigned int max_requests; /* # of discards issued per round */
310 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
311 bool io_aware; /* issue discard in idle time */
312 bool sync; /* submit discard with REQ_SYNC flag */
313 bool ordered; /* issue discard by lba order */
314 unsigned int granularity; /* discard granularity */
315 int timeout; /* discard timeout for put_super */
316 };
317
318 struct discard_cmd_control {
319 struct task_struct *f2fs_issue_discard; /* discard thread */
320 struct list_head entry_list; /* 4KB discard entry list */
321 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
322 struct list_head wait_list; /* store on-flushing entries */
323 struct list_head fstrim_list; /* in-flight discard from fstrim */
324 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
325 unsigned int discard_wake; /* to wake up discard thread */
326 struct mutex cmd_lock;
327 unsigned int nr_discards; /* # of discards in the list */
328 unsigned int max_discards; /* max. discards to be issued */
329 unsigned int discard_granularity; /* discard granularity */
330 unsigned int undiscard_blks; /* # of undiscard blocks */
331 unsigned int next_pos; /* next discard position */
332 atomic_t issued_discard; /* # of issued discard */
333 atomic_t queued_discard; /* # of queued discard */
334 atomic_t discard_cmd_cnt; /* # of cached cmd count */
335 struct rb_root_cached root; /* root of discard rb-tree */
336 bool rbtree_check; /* config for consistence check */
337 };
338
339 /* for the list of fsync inodes, used only during recovery */
340 struct fsync_inode_entry {
341 struct list_head list; /* list head */
342 struct inode *inode; /* vfs inode pointer */
343 block_t blkaddr; /* block address locating the last fsync */
344 block_t last_dentry; /* block address locating the last dentry */
345 };
346
347 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
348 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
349
350 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
351 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
352 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
353 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
354
355 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
356 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
357
358 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
359 {
360 int before = nats_in_cursum(journal);
361
362 journal->n_nats = cpu_to_le16(before + i);
363 return before;
364 }
365
366 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
367 {
368 int before = sits_in_cursum(journal);
369
370 journal->n_sits = cpu_to_le16(before + i);
371 return before;
372 }
373
374 static inline bool __has_cursum_space(struct f2fs_journal *journal,
375 int size, int type)
376 {
377 if (type == NAT_JOURNAL)
378 return size <= MAX_NAT_JENTRIES(journal);
379 return size <= MAX_SIT_JENTRIES(journal);
380 }
381
382 /*
383 * ioctl commands
384 */
385 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
386 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
387 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
388
389 #define F2FS_IOCTL_MAGIC 0xf5
390 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
391 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
392 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
393 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
394 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
395 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
396 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
397 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
398 struct f2fs_defragment)
399 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
400 struct f2fs_move_range)
401 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
402 struct f2fs_flush_device)
403 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
404 struct f2fs_gc_range)
405 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
406 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
407 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
408 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
409
410 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
411 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
412 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
413
414 /*
415 * should be same as XFS_IOC_GOINGDOWN.
416 * Flags for going down operation used by FS_IOC_GOINGDOWN
417 */
418 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
419 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
420 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
421 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
422 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
423 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
424
425 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
426 /*
427 * ioctl commands in 32 bit emulation
428 */
429 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
430 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
431 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
432 #endif
433
434 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
435 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
436
437 struct f2fs_gc_range {
438 u32 sync;
439 u64 start;
440 u64 len;
441 };
442
443 struct f2fs_defragment {
444 u64 start;
445 u64 len;
446 };
447
448 struct f2fs_move_range {
449 u32 dst_fd; /* destination fd */
450 u64 pos_in; /* start position in src_fd */
451 u64 pos_out; /* start position in dst_fd */
452 u64 len; /* size to move */
453 };
454
455 struct f2fs_flush_device {
456 u32 dev_num; /* device number to flush */
457 u32 segments; /* # of segments to flush */
458 };
459
460 /* for inline stuff */
461 #define DEF_INLINE_RESERVED_SIZE 1
462 #define DEF_MIN_INLINE_SIZE 1
463 static inline int get_extra_isize(struct inode *inode);
464 static inline int get_inline_xattr_addrs(struct inode *inode);
465 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
466 (CUR_ADDRS_PER_INODE(inode) - \
467 get_inline_xattr_addrs(inode) - \
468 DEF_INLINE_RESERVED_SIZE))
469
470 /* for inline dir */
471 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
472 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
473 BITS_PER_BYTE + 1))
474 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \
475 BITS_PER_BYTE - 1) / BITS_PER_BYTE)
476 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
477 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
478 NR_INLINE_DENTRY(inode) + \
479 INLINE_DENTRY_BITMAP_SIZE(inode)))
480
481 /*
482 * For INODE and NODE manager
483 */
484 /* for directory operations */
485 struct f2fs_dentry_ptr {
486 struct inode *inode;
487 void *bitmap;
488 struct f2fs_dir_entry *dentry;
489 __u8 (*filename)[F2FS_SLOT_LEN];
490 int max;
491 int nr_bitmap;
492 };
493
494 static inline void make_dentry_ptr_block(struct inode *inode,
495 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
496 {
497 d->inode = inode;
498 d->max = NR_DENTRY_IN_BLOCK;
499 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
500 d->bitmap = t->dentry_bitmap;
501 d->dentry = t->dentry;
502 d->filename = t->filename;
503 }
504
505 static inline void make_dentry_ptr_inline(struct inode *inode,
506 struct f2fs_dentry_ptr *d, void *t)
507 {
508 int entry_cnt = NR_INLINE_DENTRY(inode);
509 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
510 int reserved_size = INLINE_RESERVED_SIZE(inode);
511
512 d->inode = inode;
513 d->max = entry_cnt;
514 d->nr_bitmap = bitmap_size;
515 d->bitmap = t;
516 d->dentry = t + bitmap_size + reserved_size;
517 d->filename = t + bitmap_size + reserved_size +
518 SIZE_OF_DIR_ENTRY * entry_cnt;
519 }
520
521 /*
522 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
523 * as its node offset to distinguish from index node blocks.
524 * But some bits are used to mark the node block.
525 */
526 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
527 >> OFFSET_BIT_SHIFT)
528 enum {
529 ALLOC_NODE, /* allocate a new node page if needed */
530 LOOKUP_NODE, /* look up a node without readahead */
531 LOOKUP_NODE_RA, /*
532 * look up a node with readahead called
533 * by get_data_block.
534 */
535 };
536
537 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
538
539 /* maximum retry quota flush count */
540 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
541
542 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
543
544 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
545
546 /* for in-memory extent cache entry */
547 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
548
549 /* number of extent info in extent cache we try to shrink */
550 #define EXTENT_CACHE_SHRINK_NUMBER 128
551
552 struct rb_entry {
553 struct rb_node rb_node; /* rb node located in rb-tree */
554 unsigned int ofs; /* start offset of the entry */
555 unsigned int len; /* length of the entry */
556 };
557
558 struct extent_info {
559 unsigned int fofs; /* start offset in a file */
560 unsigned int len; /* length of the extent */
561 u32 blk; /* start block address of the extent */
562 };
563
564 struct extent_node {
565 struct rb_node rb_node; /* rb node located in rb-tree */
566 struct extent_info ei; /* extent info */
567 struct list_head list; /* node in global extent list of sbi */
568 struct extent_tree *et; /* extent tree pointer */
569 };
570
571 struct extent_tree {
572 nid_t ino; /* inode number */
573 struct rb_root_cached root; /* root of extent info rb-tree */
574 struct extent_node *cached_en; /* recently accessed extent node */
575 struct extent_info largest; /* largested extent info */
576 struct list_head list; /* to be used by sbi->zombie_list */
577 rwlock_t lock; /* protect extent info rb-tree */
578 atomic_t node_cnt; /* # of extent node in rb-tree*/
579 bool largest_updated; /* largest extent updated */
580 };
581
582 /*
583 * This structure is taken from ext4_map_blocks.
584 *
585 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
586 */
587 #define F2FS_MAP_NEW (1 << BH_New)
588 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
589 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
590 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
591 F2FS_MAP_UNWRITTEN)
592
593 struct f2fs_map_blocks {
594 block_t m_pblk;
595 block_t m_lblk;
596 unsigned int m_len;
597 unsigned int m_flags;
598 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
599 pgoff_t *m_next_extent; /* point to next possible extent */
600 int m_seg_type;
601 bool m_may_create; /* indicate it is from write path */
602 };
603
604 /* for flag in get_data_block */
605 enum {
606 F2FS_GET_BLOCK_DEFAULT,
607 F2FS_GET_BLOCK_FIEMAP,
608 F2FS_GET_BLOCK_BMAP,
609 F2FS_GET_BLOCK_DIO,
610 F2FS_GET_BLOCK_PRE_DIO,
611 F2FS_GET_BLOCK_PRE_AIO,
612 F2FS_GET_BLOCK_PRECACHE,
613 };
614
615 /*
616 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
617 */
618 #define FADVISE_COLD_BIT 0x01
619 #define FADVISE_LOST_PINO_BIT 0x02
620 #define FADVISE_ENCRYPT_BIT 0x04
621 #define FADVISE_ENC_NAME_BIT 0x08
622 #define FADVISE_KEEP_SIZE_BIT 0x10
623 #define FADVISE_HOT_BIT 0x20
624 #define FADVISE_VERITY_BIT 0x40 /* reserved */
625
626 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
627
628 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
629 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
630 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
631 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
632 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
633 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
634 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
635 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
636 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
637 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
638 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
639 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
640 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
641 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
642 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
643 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
644
645 #define DEF_DIR_LEVEL 0
646
647 enum {
648 GC_FAILURE_PIN,
649 GC_FAILURE_ATOMIC,
650 MAX_GC_FAILURE
651 };
652
653 struct f2fs_inode_info {
654 struct inode vfs_inode; /* serve a vfs inode */
655 unsigned long i_flags; /* keep an inode flags for ioctl */
656 unsigned char i_advise; /* use to give file attribute hints */
657 unsigned char i_dir_level; /* use for dentry level for large dir */
658 unsigned int i_current_depth; /* only for directory depth */
659 /* for gc failure statistic */
660 unsigned int i_gc_failures[MAX_GC_FAILURE];
661 unsigned int i_pino; /* parent inode number */
662 umode_t i_acl_mode; /* keep file acl mode temporarily */
663
664 /* Use below internally in f2fs*/
665 unsigned long flags; /* use to pass per-file flags */
666 struct rw_semaphore i_sem; /* protect fi info */
667 atomic_t dirty_pages; /* # of dirty pages */
668 f2fs_hash_t chash; /* hash value of given file name */
669 unsigned int clevel; /* maximum level of given file name */
670 struct task_struct *task; /* lookup and create consistency */
671 struct task_struct *cp_task; /* separate cp/wb IO stats*/
672 nid_t i_xattr_nid; /* node id that contains xattrs */
673 loff_t last_disk_size; /* lastly written file size */
674
675 #ifdef CONFIG_QUOTA
676 struct dquot *i_dquot[MAXQUOTAS];
677
678 /* quota space reservation, managed internally by quota code */
679 qsize_t i_reserved_quota;
680 #endif
681 struct list_head dirty_list; /* dirty list for dirs and files */
682 struct list_head gdirty_list; /* linked in global dirty list */
683 struct list_head inmem_ilist; /* list for inmem inodes */
684 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
685 struct task_struct *inmem_task; /* store inmemory task */
686 struct mutex inmem_lock; /* lock for inmemory pages */
687 struct extent_tree *extent_tree; /* cached extent_tree entry */
688
689 /* avoid racing between foreground op and gc */
690 struct rw_semaphore i_gc_rwsem[2];
691 struct rw_semaphore i_mmap_sem;
692 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
693
694 int i_extra_isize; /* size of extra space located in i_addr */
695 kprojid_t i_projid; /* id for project quota */
696 int i_inline_xattr_size; /* inline xattr size */
697 struct timespec64 i_crtime; /* inode creation time */
698 struct timespec64 i_disk_time[4];/* inode disk times */
699 };
700
701 static inline void get_extent_info(struct extent_info *ext,
702 struct f2fs_extent *i_ext)
703 {
704 ext->fofs = le32_to_cpu(i_ext->fofs);
705 ext->blk = le32_to_cpu(i_ext->blk);
706 ext->len = le32_to_cpu(i_ext->len);
707 }
708
709 static inline void set_raw_extent(struct extent_info *ext,
710 struct f2fs_extent *i_ext)
711 {
712 i_ext->fofs = cpu_to_le32(ext->fofs);
713 i_ext->blk = cpu_to_le32(ext->blk);
714 i_ext->len = cpu_to_le32(ext->len);
715 }
716
717 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
718 u32 blk, unsigned int len)
719 {
720 ei->fofs = fofs;
721 ei->blk = blk;
722 ei->len = len;
723 }
724
725 static inline bool __is_discard_mergeable(struct discard_info *back,
726 struct discard_info *front, unsigned int max_len)
727 {
728 return (back->lstart + back->len == front->lstart) &&
729 (back->len + front->len <= max_len);
730 }
731
732 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
733 struct discard_info *back, unsigned int max_len)
734 {
735 return __is_discard_mergeable(back, cur, max_len);
736 }
737
738 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
739 struct discard_info *front, unsigned int max_len)
740 {
741 return __is_discard_mergeable(cur, front, max_len);
742 }
743
744 static inline bool __is_extent_mergeable(struct extent_info *back,
745 struct extent_info *front)
746 {
747 return (back->fofs + back->len == front->fofs &&
748 back->blk + back->len == front->blk);
749 }
750
751 static inline bool __is_back_mergeable(struct extent_info *cur,
752 struct extent_info *back)
753 {
754 return __is_extent_mergeable(back, cur);
755 }
756
757 static inline bool __is_front_mergeable(struct extent_info *cur,
758 struct extent_info *front)
759 {
760 return __is_extent_mergeable(cur, front);
761 }
762
763 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
764 static inline void __try_update_largest_extent(struct extent_tree *et,
765 struct extent_node *en)
766 {
767 if (en->ei.len > et->largest.len) {
768 et->largest = en->ei;
769 et->largest_updated = true;
770 }
771 }
772
773 /*
774 * For free nid management
775 */
776 enum nid_state {
777 FREE_NID, /* newly added to free nid list */
778 PREALLOC_NID, /* it is preallocated */
779 MAX_NID_STATE,
780 };
781
782 struct f2fs_nm_info {
783 block_t nat_blkaddr; /* base disk address of NAT */
784 nid_t max_nid; /* maximum possible node ids */
785 nid_t available_nids; /* # of available node ids */
786 nid_t next_scan_nid; /* the next nid to be scanned */
787 unsigned int ram_thresh; /* control the memory footprint */
788 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
789 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
790
791 /* NAT cache management */
792 struct radix_tree_root nat_root;/* root of the nat entry cache */
793 struct radix_tree_root nat_set_root;/* root of the nat set cache */
794 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
795 struct list_head nat_entries; /* cached nat entry list (clean) */
796 spinlock_t nat_list_lock; /* protect clean nat entry list */
797 unsigned int nat_cnt; /* the # of cached nat entries */
798 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
799 unsigned int nat_blocks; /* # of nat blocks */
800
801 /* free node ids management */
802 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
803 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
804 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
805 spinlock_t nid_list_lock; /* protect nid lists ops */
806 struct mutex build_lock; /* lock for build free nids */
807 unsigned char **free_nid_bitmap;
808 unsigned char *nat_block_bitmap;
809 unsigned short *free_nid_count; /* free nid count of NAT block */
810
811 /* for checkpoint */
812 char *nat_bitmap; /* NAT bitmap pointer */
813
814 unsigned int nat_bits_blocks; /* # of nat bits blocks */
815 unsigned char *nat_bits; /* NAT bits blocks */
816 unsigned char *full_nat_bits; /* full NAT pages */
817 unsigned char *empty_nat_bits; /* empty NAT pages */
818 #ifdef CONFIG_F2FS_CHECK_FS
819 char *nat_bitmap_mir; /* NAT bitmap mirror */
820 #endif
821 int bitmap_size; /* bitmap size */
822 };
823
824 /*
825 * this structure is used as one of function parameters.
826 * all the information are dedicated to a given direct node block determined
827 * by the data offset in a file.
828 */
829 struct dnode_of_data {
830 struct inode *inode; /* vfs inode pointer */
831 struct page *inode_page; /* its inode page, NULL is possible */
832 struct page *node_page; /* cached direct node page */
833 nid_t nid; /* node id of the direct node block */
834 unsigned int ofs_in_node; /* data offset in the node page */
835 bool inode_page_locked; /* inode page is locked or not */
836 bool node_changed; /* is node block changed */
837 char cur_level; /* level of hole node page */
838 char max_level; /* level of current page located */
839 block_t data_blkaddr; /* block address of the node block */
840 };
841
842 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
843 struct page *ipage, struct page *npage, nid_t nid)
844 {
845 memset(dn, 0, sizeof(*dn));
846 dn->inode = inode;
847 dn->inode_page = ipage;
848 dn->node_page = npage;
849 dn->nid = nid;
850 }
851
852 /*
853 * For SIT manager
854 *
855 * By default, there are 6 active log areas across the whole main area.
856 * When considering hot and cold data separation to reduce cleaning overhead,
857 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
858 * respectively.
859 * In the current design, you should not change the numbers intentionally.
860 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
861 * logs individually according to the underlying devices. (default: 6)
862 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
863 * data and 8 for node logs.
864 */
865 #define NR_CURSEG_DATA_TYPE (3)
866 #define NR_CURSEG_NODE_TYPE (3)
867 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
868
869 enum {
870 CURSEG_HOT_DATA = 0, /* directory entry blocks */
871 CURSEG_WARM_DATA, /* data blocks */
872 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
873 CURSEG_HOT_NODE, /* direct node blocks of directory files */
874 CURSEG_WARM_NODE, /* direct node blocks of normal files */
875 CURSEG_COLD_NODE, /* indirect node blocks */
876 NO_CHECK_TYPE,
877 };
878
879 struct flush_cmd {
880 struct completion wait;
881 struct llist_node llnode;
882 nid_t ino;
883 int ret;
884 };
885
886 struct flush_cmd_control {
887 struct task_struct *f2fs_issue_flush; /* flush thread */
888 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
889 atomic_t issued_flush; /* # of issued flushes */
890 atomic_t queued_flush; /* # of queued flushes */
891 struct llist_head issue_list; /* list for command issue */
892 struct llist_node *dispatch_list; /* list for command dispatch */
893 };
894
895 struct f2fs_sm_info {
896 struct sit_info *sit_info; /* whole segment information */
897 struct free_segmap_info *free_info; /* free segment information */
898 struct dirty_seglist_info *dirty_info; /* dirty segment information */
899 struct curseg_info *curseg_array; /* active segment information */
900
901 struct rw_semaphore curseg_lock; /* for preventing curseg change */
902
903 block_t seg0_blkaddr; /* block address of 0'th segment */
904 block_t main_blkaddr; /* start block address of main area */
905 block_t ssa_blkaddr; /* start block address of SSA area */
906
907 unsigned int segment_count; /* total # of segments */
908 unsigned int main_segments; /* # of segments in main area */
909 unsigned int reserved_segments; /* # of reserved segments */
910 unsigned int ovp_segments; /* # of overprovision segments */
911
912 /* a threshold to reclaim prefree segments */
913 unsigned int rec_prefree_segments;
914
915 /* for batched trimming */
916 unsigned int trim_sections; /* # of sections to trim */
917
918 struct list_head sit_entry_set; /* sit entry set list */
919
920 unsigned int ipu_policy; /* in-place-update policy */
921 unsigned int min_ipu_util; /* in-place-update threshold */
922 unsigned int min_fsync_blocks; /* threshold for fsync */
923 unsigned int min_seq_blocks; /* threshold for sequential blocks */
924 unsigned int min_hot_blocks; /* threshold for hot block allocation */
925 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
926
927 /* for flush command control */
928 struct flush_cmd_control *fcc_info;
929
930 /* for discard command control */
931 struct discard_cmd_control *dcc_info;
932 };
933
934 /*
935 * For superblock
936 */
937 /*
938 * COUNT_TYPE for monitoring
939 *
940 * f2fs monitors the number of several block types such as on-writeback,
941 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
942 */
943 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
944 enum count_type {
945 F2FS_DIRTY_DENTS,
946 F2FS_DIRTY_DATA,
947 F2FS_DIRTY_QDATA,
948 F2FS_DIRTY_NODES,
949 F2FS_DIRTY_META,
950 F2FS_INMEM_PAGES,
951 F2FS_DIRTY_IMETA,
952 F2FS_WB_CP_DATA,
953 F2FS_WB_DATA,
954 F2FS_RD_DATA,
955 F2FS_RD_NODE,
956 F2FS_RD_META,
957 F2FS_DIO_WRITE,
958 F2FS_DIO_READ,
959 NR_COUNT_TYPE,
960 };
961
962 /*
963 * The below are the page types of bios used in submit_bio().
964 * The available types are:
965 * DATA User data pages. It operates as async mode.
966 * NODE Node pages. It operates as async mode.
967 * META FS metadata pages such as SIT, NAT, CP.
968 * NR_PAGE_TYPE The number of page types.
969 * META_FLUSH Make sure the previous pages are written
970 * with waiting the bio's completion
971 * ... Only can be used with META.
972 */
973 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
974 enum page_type {
975 DATA,
976 NODE,
977 META,
978 NR_PAGE_TYPE,
979 META_FLUSH,
980 INMEM, /* the below types are used by tracepoints only. */
981 INMEM_DROP,
982 INMEM_INVALIDATE,
983 INMEM_REVOKE,
984 IPU,
985 OPU,
986 };
987
988 enum temp_type {
989 HOT = 0, /* must be zero for meta bio */
990 WARM,
991 COLD,
992 NR_TEMP_TYPE,
993 };
994
995 enum need_lock_type {
996 LOCK_REQ = 0,
997 LOCK_DONE,
998 LOCK_RETRY,
999 };
1000
1001 enum cp_reason_type {
1002 CP_NO_NEEDED,
1003 CP_NON_REGULAR,
1004 CP_HARDLINK,
1005 CP_SB_NEED_CP,
1006 CP_WRONG_PINO,
1007 CP_NO_SPC_ROLL,
1008 CP_NODE_NEED_CP,
1009 CP_FASTBOOT_MODE,
1010 CP_SPEC_LOG_NUM,
1011 CP_RECOVER_DIR,
1012 };
1013
1014 enum iostat_type {
1015 APP_DIRECT_IO, /* app direct IOs */
1016 APP_BUFFERED_IO, /* app buffered IOs */
1017 APP_WRITE_IO, /* app write IOs */
1018 APP_MAPPED_IO, /* app mapped IOs */
1019 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1020 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1021 FS_META_IO, /* meta IOs from kworker/reclaimer */
1022 FS_GC_DATA_IO, /* data IOs from forground gc */
1023 FS_GC_NODE_IO, /* node IOs from forground gc */
1024 FS_CP_DATA_IO, /* data IOs from checkpoint */
1025 FS_CP_NODE_IO, /* node IOs from checkpoint */
1026 FS_CP_META_IO, /* meta IOs from checkpoint */
1027 FS_DISCARD, /* discard */
1028 NR_IO_TYPE,
1029 };
1030
1031 struct f2fs_io_info {
1032 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1033 nid_t ino; /* inode number */
1034 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1035 enum temp_type temp; /* contains HOT/WARM/COLD */
1036 int op; /* contains REQ_OP_ */
1037 int op_flags; /* req_flag_bits */
1038 block_t new_blkaddr; /* new block address to be written */
1039 block_t old_blkaddr; /* old block address before Cow */
1040 struct page *page; /* page to be written */
1041 struct page *encrypted_page; /* encrypted page */
1042 struct list_head list; /* serialize IOs */
1043 bool submitted; /* indicate IO submission */
1044 int need_lock; /* indicate we need to lock cp_rwsem */
1045 bool in_list; /* indicate fio is in io_list */
1046 bool is_meta; /* indicate borrow meta inode mapping or not */
1047 bool retry; /* need to reallocate block address */
1048 enum iostat_type io_type; /* io type */
1049 struct writeback_control *io_wbc; /* writeback control */
1050 unsigned char version; /* version of the node */
1051 };
1052
1053 #define is_read_io(rw) ((rw) == READ)
1054 struct f2fs_bio_info {
1055 struct f2fs_sb_info *sbi; /* f2fs superblock */
1056 struct bio *bio; /* bios to merge */
1057 sector_t last_block_in_bio; /* last block number */
1058 struct f2fs_io_info fio; /* store buffered io info. */
1059 struct rw_semaphore io_rwsem; /* blocking op for bio */
1060 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1061 struct list_head io_list; /* track fios */
1062 };
1063
1064 #define FDEV(i) (sbi->devs[i])
1065 #define RDEV(i) (raw_super->devs[i])
1066 struct f2fs_dev_info {
1067 struct block_device *bdev;
1068 char path[MAX_PATH_LEN];
1069 unsigned int total_segments;
1070 block_t start_blk;
1071 block_t end_blk;
1072 #ifdef CONFIG_BLK_DEV_ZONED
1073 unsigned int nr_blkz; /* Total number of zones */
1074 u8 *blkz_type; /* Array of zones type */
1075 #endif
1076 };
1077
1078 enum inode_type {
1079 DIR_INODE, /* for dirty dir inode */
1080 FILE_INODE, /* for dirty regular/symlink inode */
1081 DIRTY_META, /* for all dirtied inode metadata */
1082 ATOMIC_FILE, /* for all atomic files */
1083 NR_INODE_TYPE,
1084 };
1085
1086 /* for inner inode cache management */
1087 struct inode_management {
1088 struct radix_tree_root ino_root; /* ino entry array */
1089 spinlock_t ino_lock; /* for ino entry lock */
1090 struct list_head ino_list; /* inode list head */
1091 unsigned long ino_num; /* number of entries */
1092 };
1093
1094 /* For s_flag in struct f2fs_sb_info */
1095 enum {
1096 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1097 SBI_IS_CLOSE, /* specify unmounting */
1098 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1099 SBI_POR_DOING, /* recovery is doing or not */
1100 SBI_NEED_SB_WRITE, /* need to recover superblock */
1101 SBI_NEED_CP, /* need to checkpoint */
1102 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1103 SBI_IS_RECOVERED, /* recovered orphan/data */
1104 SBI_CP_DISABLED, /* CP was disabled last mount */
1105 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1106 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1107 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1108 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1109 };
1110
1111 enum {
1112 CP_TIME,
1113 REQ_TIME,
1114 DISCARD_TIME,
1115 GC_TIME,
1116 DISABLE_TIME,
1117 UMOUNT_DISCARD_TIMEOUT,
1118 MAX_TIME,
1119 };
1120
1121 enum {
1122 GC_NORMAL,
1123 GC_IDLE_CB,
1124 GC_IDLE_GREEDY,
1125 GC_URGENT,
1126 };
1127
1128 enum {
1129 WHINT_MODE_OFF, /* not pass down write hints */
1130 WHINT_MODE_USER, /* try to pass down hints given by users */
1131 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1132 };
1133
1134 enum {
1135 ALLOC_MODE_DEFAULT, /* stay default */
1136 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1137 };
1138
1139 enum fsync_mode {
1140 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1141 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1142 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1143 };
1144
1145 #ifdef CONFIG_F2FS_FS_ENCRYPTION
1146 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1147 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1148 #else
1149 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1150 #endif
1151
1152 struct f2fs_sb_info {
1153 struct super_block *sb; /* pointer to VFS super block */
1154 struct proc_dir_entry *s_proc; /* proc entry */
1155 struct f2fs_super_block *raw_super; /* raw super block pointer */
1156 struct rw_semaphore sb_lock; /* lock for raw super block */
1157 int valid_super_block; /* valid super block no */
1158 unsigned long s_flag; /* flags for sbi */
1159 struct mutex writepages; /* mutex for writepages() */
1160
1161 #ifdef CONFIG_BLK_DEV_ZONED
1162 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1163 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1164 #endif
1165
1166 /* for node-related operations */
1167 struct f2fs_nm_info *nm_info; /* node manager */
1168 struct inode *node_inode; /* cache node blocks */
1169
1170 /* for segment-related operations */
1171 struct f2fs_sm_info *sm_info; /* segment manager */
1172
1173 /* for bio operations */
1174 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1175 /* keep migration IO order for LFS mode */
1176 struct rw_semaphore io_order_lock;
1177 mempool_t *write_io_dummy; /* Dummy pages */
1178
1179 /* for checkpoint */
1180 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1181 int cur_cp_pack; /* remain current cp pack */
1182 spinlock_t cp_lock; /* for flag in ckpt */
1183 struct inode *meta_inode; /* cache meta blocks */
1184 struct mutex cp_mutex; /* checkpoint procedure lock */
1185 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1186 struct rw_semaphore node_write; /* locking node writes */
1187 struct rw_semaphore node_change; /* locking node change */
1188 wait_queue_head_t cp_wait;
1189 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1190 long interval_time[MAX_TIME]; /* to store thresholds */
1191
1192 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1193
1194 spinlock_t fsync_node_lock; /* for node entry lock */
1195 struct list_head fsync_node_list; /* node list head */
1196 unsigned int fsync_seg_id; /* sequence id */
1197 unsigned int fsync_node_num; /* number of node entries */
1198
1199 /* for orphan inode, use 0'th array */
1200 unsigned int max_orphans; /* max orphan inodes */
1201
1202 /* for inode management */
1203 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1204 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1205
1206 /* for extent tree cache */
1207 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1208 struct mutex extent_tree_lock; /* locking extent radix tree */
1209 struct list_head extent_list; /* lru list for shrinker */
1210 spinlock_t extent_lock; /* locking extent lru list */
1211 atomic_t total_ext_tree; /* extent tree count */
1212 struct list_head zombie_list; /* extent zombie tree list */
1213 atomic_t total_zombie_tree; /* extent zombie tree count */
1214 atomic_t total_ext_node; /* extent info count */
1215
1216 /* basic filesystem units */
1217 unsigned int log_sectors_per_block; /* log2 sectors per block */
1218 unsigned int log_blocksize; /* log2 block size */
1219 unsigned int blocksize; /* block size */
1220 unsigned int root_ino_num; /* root inode number*/
1221 unsigned int node_ino_num; /* node inode number*/
1222 unsigned int meta_ino_num; /* meta inode number*/
1223 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1224 unsigned int blocks_per_seg; /* blocks per segment */
1225 unsigned int segs_per_sec; /* segments per section */
1226 unsigned int secs_per_zone; /* sections per zone */
1227 unsigned int total_sections; /* total section count */
1228 unsigned int total_node_count; /* total node block count */
1229 unsigned int total_valid_node_count; /* valid node block count */
1230 loff_t max_file_blocks; /* max block index of file */
1231 int dir_level; /* directory level */
1232 int readdir_ra; /* readahead inode in readdir */
1233
1234 block_t user_block_count; /* # of user blocks */
1235 block_t total_valid_block_count; /* # of valid blocks */
1236 block_t discard_blks; /* discard command candidats */
1237 block_t last_valid_block_count; /* for recovery */
1238 block_t reserved_blocks; /* configurable reserved blocks */
1239 block_t current_reserved_blocks; /* current reserved blocks */
1240
1241 /* Additional tracking for no checkpoint mode */
1242 block_t unusable_block_count; /* # of blocks saved by last cp */
1243
1244 unsigned int nquota_files; /* # of quota sysfile */
1245
1246 u32 s_next_generation; /* for NFS support */
1247
1248 /* # of pages, see count_type */
1249 atomic_t nr_pages[NR_COUNT_TYPE];
1250 /* # of allocated blocks */
1251 struct percpu_counter alloc_valid_block_count;
1252
1253 /* writeback control */
1254 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1255
1256 /* valid inode count */
1257 struct percpu_counter total_valid_inode_count;
1258
1259 struct f2fs_mount_info mount_opt; /* mount options */
1260
1261 /* for cleaning operations */
1262 struct mutex gc_mutex; /* mutex for GC */
1263 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1264 unsigned int cur_victim_sec; /* current victim section num */
1265 unsigned int gc_mode; /* current GC state */
1266 unsigned int next_victim_seg[2]; /* next segment in victim section */
1267 /* for skip statistic */
1268 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1269 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1270
1271 /* threshold for gc trials on pinned files */
1272 u64 gc_pin_file_threshold;
1273
1274 /* maximum # of trials to find a victim segment for SSR and GC */
1275 unsigned int max_victim_search;
1276 /* migration granularity of garbage collection, unit: segment */
1277 unsigned int migration_granularity;
1278
1279 /*
1280 * for stat information.
1281 * one is for the LFS mode, and the other is for the SSR mode.
1282 */
1283 #ifdef CONFIG_F2FS_STAT_FS
1284 struct f2fs_stat_info *stat_info; /* FS status information */
1285 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1286 unsigned int segment_count[2]; /* # of allocated segments */
1287 unsigned int block_count[2]; /* # of allocated blocks */
1288 atomic_t inplace_count; /* # of inplace update */
1289 atomic64_t total_hit_ext; /* # of lookup extent cache */
1290 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1291 atomic64_t read_hit_largest; /* # of hit largest extent node */
1292 atomic64_t read_hit_cached; /* # of hit cached extent node */
1293 atomic_t inline_xattr; /* # of inline_xattr inodes */
1294 atomic_t inline_inode; /* # of inline_data inodes */
1295 atomic_t inline_dir; /* # of inline_dentry inodes */
1296 atomic_t aw_cnt; /* # of atomic writes */
1297 atomic_t vw_cnt; /* # of volatile writes */
1298 atomic_t max_aw_cnt; /* max # of atomic writes */
1299 atomic_t max_vw_cnt; /* max # of volatile writes */
1300 int bg_gc; /* background gc calls */
1301 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1302 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1303 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1304 #endif
1305 spinlock_t stat_lock; /* lock for stat operations */
1306
1307 /* For app/fs IO statistics */
1308 spinlock_t iostat_lock;
1309 unsigned long long write_iostat[NR_IO_TYPE];
1310 bool iostat_enable;
1311
1312 /* For sysfs suppport */
1313 struct kobject s_kobj;
1314 struct completion s_kobj_unregister;
1315
1316 /* For shrinker support */
1317 struct list_head s_list;
1318 int s_ndevs; /* number of devices */
1319 struct f2fs_dev_info *devs; /* for device list */
1320 unsigned int dirty_device; /* for checkpoint data flush */
1321 spinlock_t dev_lock; /* protect dirty_device */
1322 struct mutex umount_mutex;
1323 unsigned int shrinker_run_no;
1324
1325 /* For write statistics */
1326 u64 sectors_written_start;
1327 u64 kbytes_written;
1328
1329 /* Reference to checksum algorithm driver via cryptoapi */
1330 struct crypto_shash *s_chksum_driver;
1331
1332 /* Precomputed FS UUID checksum for seeding other checksums */
1333 __u32 s_chksum_seed;
1334 };
1335
1336 struct f2fs_private_dio {
1337 struct inode *inode;
1338 void *orig_private;
1339 bio_end_io_t *orig_end_io;
1340 bool write;
1341 };
1342
1343 #ifdef CONFIG_F2FS_FAULT_INJECTION
1344 #define f2fs_show_injection_info(type) \
1345 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pF\n", \
1346 KERN_INFO, f2fs_fault_name[type], \
1347 __func__, __builtin_return_address(0))
1348 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1349 {
1350 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1351
1352 if (!ffi->inject_rate)
1353 return false;
1354
1355 if (!IS_FAULT_SET(ffi, type))
1356 return false;
1357
1358 atomic_inc(&ffi->inject_ops);
1359 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1360 atomic_set(&ffi->inject_ops, 0);
1361 return true;
1362 }
1363 return false;
1364 }
1365 #else
1366 #define f2fs_show_injection_info(type) do { } while (0)
1367 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1368 {
1369 return false;
1370 }
1371 #endif
1372
1373 /* For write statistics. Suppose sector size is 512 bytes,
1374 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1375 */
1376 #define BD_PART_WRITTEN(s) \
1377 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1378 (s)->sectors_written_start) >> 1)
1379
1380 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1381 {
1382 unsigned long now = jiffies;
1383
1384 sbi->last_time[type] = now;
1385
1386 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1387 if (type == REQ_TIME) {
1388 sbi->last_time[DISCARD_TIME] = now;
1389 sbi->last_time[GC_TIME] = now;
1390 }
1391 }
1392
1393 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1394 {
1395 unsigned long interval = sbi->interval_time[type] * HZ;
1396
1397 return time_after(jiffies, sbi->last_time[type] + interval);
1398 }
1399
1400 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1401 int type)
1402 {
1403 unsigned long interval = sbi->interval_time[type] * HZ;
1404 unsigned int wait_ms = 0;
1405 long delta;
1406
1407 delta = (sbi->last_time[type] + interval) - jiffies;
1408 if (delta > 0)
1409 wait_ms = jiffies_to_msecs(delta);
1410
1411 return wait_ms;
1412 }
1413
1414 /*
1415 * Inline functions
1416 */
1417 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1418 const void *address, unsigned int length)
1419 {
1420 struct {
1421 struct shash_desc shash;
1422 char ctx[4];
1423 } desc;
1424 int err;
1425
1426 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1427
1428 desc.shash.tfm = sbi->s_chksum_driver;
1429 desc.shash.flags = 0;
1430 *(u32 *)desc.ctx = crc;
1431
1432 err = crypto_shash_update(&desc.shash, address, length);
1433 BUG_ON(err);
1434
1435 return *(u32 *)desc.ctx;
1436 }
1437
1438 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1439 unsigned int length)
1440 {
1441 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1442 }
1443
1444 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1445 void *buf, size_t buf_size)
1446 {
1447 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1448 }
1449
1450 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1451 const void *address, unsigned int length)
1452 {
1453 return __f2fs_crc32(sbi, crc, address, length);
1454 }
1455
1456 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1457 {
1458 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1459 }
1460
1461 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1462 {
1463 return sb->s_fs_info;
1464 }
1465
1466 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1467 {
1468 return F2FS_SB(inode->i_sb);
1469 }
1470
1471 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1472 {
1473 return F2FS_I_SB(mapping->host);
1474 }
1475
1476 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1477 {
1478 return F2FS_M_SB(page->mapping);
1479 }
1480
1481 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1482 {
1483 return (struct f2fs_super_block *)(sbi->raw_super);
1484 }
1485
1486 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1487 {
1488 return (struct f2fs_checkpoint *)(sbi->ckpt);
1489 }
1490
1491 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1492 {
1493 return (struct f2fs_node *)page_address(page);
1494 }
1495
1496 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1497 {
1498 return &((struct f2fs_node *)page_address(page))->i;
1499 }
1500
1501 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1502 {
1503 return (struct f2fs_nm_info *)(sbi->nm_info);
1504 }
1505
1506 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1507 {
1508 return (struct f2fs_sm_info *)(sbi->sm_info);
1509 }
1510
1511 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1512 {
1513 return (struct sit_info *)(SM_I(sbi)->sit_info);
1514 }
1515
1516 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1517 {
1518 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1519 }
1520
1521 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1522 {
1523 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1524 }
1525
1526 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1527 {
1528 return sbi->meta_inode->i_mapping;
1529 }
1530
1531 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1532 {
1533 return sbi->node_inode->i_mapping;
1534 }
1535
1536 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1537 {
1538 return test_bit(type, &sbi->s_flag);
1539 }
1540
1541 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1542 {
1543 set_bit(type, &sbi->s_flag);
1544 }
1545
1546 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1547 {
1548 clear_bit(type, &sbi->s_flag);
1549 }
1550
1551 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1552 {
1553 return le64_to_cpu(cp->checkpoint_ver);
1554 }
1555
1556 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1557 {
1558 if (type < F2FS_MAX_QUOTAS)
1559 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1560 return 0;
1561 }
1562
1563 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1564 {
1565 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1566 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1567 }
1568
1569 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1570 {
1571 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1572
1573 return ckpt_flags & f;
1574 }
1575
1576 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1577 {
1578 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1579 }
1580
1581 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1582 {
1583 unsigned int ckpt_flags;
1584
1585 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1586 ckpt_flags |= f;
1587 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1588 }
1589
1590 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1591 {
1592 unsigned long flags;
1593
1594 spin_lock_irqsave(&sbi->cp_lock, flags);
1595 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1596 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1597 }
1598
1599 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1600 {
1601 unsigned int ckpt_flags;
1602
1603 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1604 ckpt_flags &= (~f);
1605 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1606 }
1607
1608 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1609 {
1610 unsigned long flags;
1611
1612 spin_lock_irqsave(&sbi->cp_lock, flags);
1613 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1614 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1615 }
1616
1617 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1618 {
1619 unsigned long flags;
1620
1621 /*
1622 * In order to re-enable nat_bits we need to call fsck.f2fs by
1623 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1624 * so let's rely on regular fsck or unclean shutdown.
1625 */
1626
1627 if (lock)
1628 spin_lock_irqsave(&sbi->cp_lock, flags);
1629 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1630 kvfree(NM_I(sbi)->nat_bits);
1631 NM_I(sbi)->nat_bits = NULL;
1632 if (lock)
1633 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1634 }
1635
1636 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1637 struct cp_control *cpc)
1638 {
1639 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1640
1641 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1642 }
1643
1644 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1645 {
1646 down_read(&sbi->cp_rwsem);
1647 }
1648
1649 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1650 {
1651 return down_read_trylock(&sbi->cp_rwsem);
1652 }
1653
1654 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1655 {
1656 up_read(&sbi->cp_rwsem);
1657 }
1658
1659 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1660 {
1661 down_write(&sbi->cp_rwsem);
1662 }
1663
1664 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1665 {
1666 up_write(&sbi->cp_rwsem);
1667 }
1668
1669 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1670 {
1671 int reason = CP_SYNC;
1672
1673 if (test_opt(sbi, FASTBOOT))
1674 reason = CP_FASTBOOT;
1675 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1676 reason = CP_UMOUNT;
1677 return reason;
1678 }
1679
1680 static inline bool __remain_node_summaries(int reason)
1681 {
1682 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1683 }
1684
1685 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1686 {
1687 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1688 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1689 }
1690
1691 /*
1692 * Check whether the inode has blocks or not
1693 */
1694 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1695 {
1696 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1697
1698 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1699 }
1700
1701 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1702 {
1703 return ofs == XATTR_NODE_OFFSET;
1704 }
1705
1706 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1707 struct inode *inode, bool cap)
1708 {
1709 if (!inode)
1710 return true;
1711 if (!test_opt(sbi, RESERVE_ROOT))
1712 return false;
1713 if (IS_NOQUOTA(inode))
1714 return true;
1715 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1716 return true;
1717 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1718 in_group_p(F2FS_OPTION(sbi).s_resgid))
1719 return true;
1720 if (cap && capable(CAP_SYS_RESOURCE))
1721 return true;
1722 return false;
1723 }
1724
1725 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1726 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1727 struct inode *inode, blkcnt_t *count)
1728 {
1729 blkcnt_t diff = 0, release = 0;
1730 block_t avail_user_block_count;
1731 int ret;
1732
1733 ret = dquot_reserve_block(inode, *count);
1734 if (ret)
1735 return ret;
1736
1737 if (time_to_inject(sbi, FAULT_BLOCK)) {
1738 f2fs_show_injection_info(FAULT_BLOCK);
1739 release = *count;
1740 goto enospc;
1741 }
1742
1743 /*
1744 * let's increase this in prior to actual block count change in order
1745 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1746 */
1747 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1748
1749 spin_lock(&sbi->stat_lock);
1750 sbi->total_valid_block_count += (block_t)(*count);
1751 avail_user_block_count = sbi->user_block_count -
1752 sbi->current_reserved_blocks;
1753
1754 if (!__allow_reserved_blocks(sbi, inode, true))
1755 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1756 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1757 avail_user_block_count -= sbi->unusable_block_count;
1758 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1759 diff = sbi->total_valid_block_count - avail_user_block_count;
1760 if (diff > *count)
1761 diff = *count;
1762 *count -= diff;
1763 release = diff;
1764 sbi->total_valid_block_count -= diff;
1765 if (!*count) {
1766 spin_unlock(&sbi->stat_lock);
1767 goto enospc;
1768 }
1769 }
1770 spin_unlock(&sbi->stat_lock);
1771
1772 if (unlikely(release)) {
1773 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1774 dquot_release_reservation_block(inode, release);
1775 }
1776 f2fs_i_blocks_write(inode, *count, true, true);
1777 return 0;
1778
1779 enospc:
1780 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
1781 dquot_release_reservation_block(inode, release);
1782 return -ENOSPC;
1783 }
1784
1785 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
1786 struct inode *inode,
1787 block_t count)
1788 {
1789 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
1790
1791 spin_lock(&sbi->stat_lock);
1792 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
1793 f2fs_bug_on(sbi, inode->i_blocks < sectors);
1794 sbi->total_valid_block_count -= (block_t)count;
1795 if (sbi->reserved_blocks &&
1796 sbi->current_reserved_blocks < sbi->reserved_blocks)
1797 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
1798 sbi->current_reserved_blocks + count);
1799 spin_unlock(&sbi->stat_lock);
1800 f2fs_i_blocks_write(inode, count, false, true);
1801 }
1802
1803 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
1804 {
1805 atomic_inc(&sbi->nr_pages[count_type]);
1806
1807 if (count_type == F2FS_DIRTY_DENTS ||
1808 count_type == F2FS_DIRTY_NODES ||
1809 count_type == F2FS_DIRTY_META ||
1810 count_type == F2FS_DIRTY_QDATA ||
1811 count_type == F2FS_DIRTY_IMETA)
1812 set_sbi_flag(sbi, SBI_IS_DIRTY);
1813 }
1814
1815 static inline void inode_inc_dirty_pages(struct inode *inode)
1816 {
1817 atomic_inc(&F2FS_I(inode)->dirty_pages);
1818 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1819 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1820 if (IS_NOQUOTA(inode))
1821 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1822 }
1823
1824 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
1825 {
1826 atomic_dec(&sbi->nr_pages[count_type]);
1827 }
1828
1829 static inline void inode_dec_dirty_pages(struct inode *inode)
1830 {
1831 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
1832 !S_ISLNK(inode->i_mode))
1833 return;
1834
1835 atomic_dec(&F2FS_I(inode)->dirty_pages);
1836 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
1837 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
1838 if (IS_NOQUOTA(inode))
1839 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
1840 }
1841
1842 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
1843 {
1844 return atomic_read(&sbi->nr_pages[count_type]);
1845 }
1846
1847 static inline int get_dirty_pages(struct inode *inode)
1848 {
1849 return atomic_read(&F2FS_I(inode)->dirty_pages);
1850 }
1851
1852 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
1853 {
1854 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
1855 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
1856 sbi->log_blocks_per_seg;
1857
1858 return segs / sbi->segs_per_sec;
1859 }
1860
1861 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
1862 {
1863 return sbi->total_valid_block_count;
1864 }
1865
1866 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
1867 {
1868 return sbi->discard_blks;
1869 }
1870
1871 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
1872 {
1873 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1874
1875 /* return NAT or SIT bitmap */
1876 if (flag == NAT_BITMAP)
1877 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
1878 else if (flag == SIT_BITMAP)
1879 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
1880
1881 return 0;
1882 }
1883
1884 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
1885 {
1886 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
1887 }
1888
1889 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
1890 {
1891 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1892 int offset;
1893
1894 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
1895 offset = (flag == SIT_BITMAP) ?
1896 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
1897 return &ckpt->sit_nat_version_bitmap + offset;
1898 }
1899
1900 if (__cp_payload(sbi) > 0) {
1901 if (flag == NAT_BITMAP)
1902 return &ckpt->sit_nat_version_bitmap;
1903 else
1904 return (unsigned char *)ckpt + F2FS_BLKSIZE;
1905 } else {
1906 offset = (flag == NAT_BITMAP) ?
1907 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
1908 return &ckpt->sit_nat_version_bitmap + offset;
1909 }
1910 }
1911
1912 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
1913 {
1914 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1915
1916 if (sbi->cur_cp_pack == 2)
1917 start_addr += sbi->blocks_per_seg;
1918 return start_addr;
1919 }
1920
1921 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
1922 {
1923 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
1924
1925 if (sbi->cur_cp_pack == 1)
1926 start_addr += sbi->blocks_per_seg;
1927 return start_addr;
1928 }
1929
1930 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
1931 {
1932 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
1933 }
1934
1935 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
1936 {
1937 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
1938 }
1939
1940 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
1941 struct inode *inode, bool is_inode)
1942 {
1943 block_t valid_block_count;
1944 unsigned int valid_node_count;
1945 int err;
1946
1947 if (is_inode) {
1948 if (inode) {
1949 err = dquot_alloc_inode(inode);
1950 if (err)
1951 return err;
1952 }
1953 } else {
1954 err = dquot_reserve_block(inode, 1);
1955 if (err)
1956 return err;
1957 }
1958
1959 if (time_to_inject(sbi, FAULT_BLOCK)) {
1960 f2fs_show_injection_info(FAULT_BLOCK);
1961 goto enospc;
1962 }
1963
1964 spin_lock(&sbi->stat_lock);
1965
1966 valid_block_count = sbi->total_valid_block_count +
1967 sbi->current_reserved_blocks + 1;
1968
1969 if (!__allow_reserved_blocks(sbi, inode, false))
1970 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
1971 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1972 valid_block_count += sbi->unusable_block_count;
1973
1974 if (unlikely(valid_block_count > sbi->user_block_count)) {
1975 spin_unlock(&sbi->stat_lock);
1976 goto enospc;
1977 }
1978
1979 valid_node_count = sbi->total_valid_node_count + 1;
1980 if (unlikely(valid_node_count > sbi->total_node_count)) {
1981 spin_unlock(&sbi->stat_lock);
1982 goto enospc;
1983 }
1984
1985 sbi->total_valid_node_count++;
1986 sbi->total_valid_block_count++;
1987 spin_unlock(&sbi->stat_lock);
1988
1989 if (inode) {
1990 if (is_inode)
1991 f2fs_mark_inode_dirty_sync(inode, true);
1992 else
1993 f2fs_i_blocks_write(inode, 1, true, true);
1994 }
1995
1996 percpu_counter_inc(&sbi->alloc_valid_block_count);
1997 return 0;
1998
1999 enospc:
2000 if (is_inode) {
2001 if (inode)
2002 dquot_free_inode(inode);
2003 } else {
2004 dquot_release_reservation_block(inode, 1);
2005 }
2006 return -ENOSPC;
2007 }
2008
2009 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2010 struct inode *inode, bool is_inode)
2011 {
2012 spin_lock(&sbi->stat_lock);
2013
2014 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2015 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2016 f2fs_bug_on(sbi, !is_inode && !inode->i_blocks);
2017
2018 sbi->total_valid_node_count--;
2019 sbi->total_valid_block_count--;
2020 if (sbi->reserved_blocks &&
2021 sbi->current_reserved_blocks < sbi->reserved_blocks)
2022 sbi->current_reserved_blocks++;
2023
2024 spin_unlock(&sbi->stat_lock);
2025
2026 if (is_inode)
2027 dquot_free_inode(inode);
2028 else
2029 f2fs_i_blocks_write(inode, 1, false, true);
2030 }
2031
2032 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2033 {
2034 return sbi->total_valid_node_count;
2035 }
2036
2037 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2038 {
2039 percpu_counter_inc(&sbi->total_valid_inode_count);
2040 }
2041
2042 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2043 {
2044 percpu_counter_dec(&sbi->total_valid_inode_count);
2045 }
2046
2047 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2048 {
2049 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2050 }
2051
2052 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2053 pgoff_t index, bool for_write)
2054 {
2055 struct page *page;
2056
2057 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2058 if (!for_write)
2059 page = find_get_page_flags(mapping, index,
2060 FGP_LOCK | FGP_ACCESSED);
2061 else
2062 page = find_lock_page(mapping, index);
2063 if (page)
2064 return page;
2065
2066 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2067 f2fs_show_injection_info(FAULT_PAGE_ALLOC);
2068 return NULL;
2069 }
2070 }
2071
2072 if (!for_write)
2073 return grab_cache_page(mapping, index);
2074 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2075 }
2076
2077 static inline struct page *f2fs_pagecache_get_page(
2078 struct address_space *mapping, pgoff_t index,
2079 int fgp_flags, gfp_t gfp_mask)
2080 {
2081 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2082 f2fs_show_injection_info(FAULT_PAGE_GET);
2083 return NULL;
2084 }
2085
2086 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2087 }
2088
2089 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2090 {
2091 char *src_kaddr = kmap(src);
2092 char *dst_kaddr = kmap(dst);
2093
2094 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2095 kunmap(dst);
2096 kunmap(src);
2097 }
2098
2099 static inline void f2fs_put_page(struct page *page, int unlock)
2100 {
2101 if (!page)
2102 return;
2103
2104 if (unlock) {
2105 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2106 unlock_page(page);
2107 }
2108 put_page(page);
2109 }
2110
2111 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2112 {
2113 if (dn->node_page)
2114 f2fs_put_page(dn->node_page, 1);
2115 if (dn->inode_page && dn->node_page != dn->inode_page)
2116 f2fs_put_page(dn->inode_page, 0);
2117 dn->node_page = NULL;
2118 dn->inode_page = NULL;
2119 }
2120
2121 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2122 size_t size)
2123 {
2124 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2125 }
2126
2127 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2128 gfp_t flags)
2129 {
2130 void *entry;
2131
2132 entry = kmem_cache_alloc(cachep, flags);
2133 if (!entry)
2134 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2135 return entry;
2136 }
2137
2138 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi,
2139 int npages, bool no_fail)
2140 {
2141 struct bio *bio;
2142
2143 if (no_fail) {
2144 /* No failure on bio allocation */
2145 bio = bio_alloc(GFP_NOIO, npages);
2146 if (!bio)
2147 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages);
2148 return bio;
2149 }
2150 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) {
2151 f2fs_show_injection_info(FAULT_ALLOC_BIO);
2152 return NULL;
2153 }
2154
2155 return bio_alloc(GFP_KERNEL, npages);
2156 }
2157
2158 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2159 {
2160 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2161 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2162 get_pages(sbi, F2FS_WB_CP_DATA) ||
2163 get_pages(sbi, F2FS_DIO_READ) ||
2164 get_pages(sbi, F2FS_DIO_WRITE) ||
2165 atomic_read(&SM_I(sbi)->dcc_info->queued_discard) ||
2166 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2167 return false;
2168 return f2fs_time_over(sbi, type);
2169 }
2170
2171 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2172 unsigned long index, void *item)
2173 {
2174 while (radix_tree_insert(root, index, item))
2175 cond_resched();
2176 }
2177
2178 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2179
2180 static inline bool IS_INODE(struct page *page)
2181 {
2182 struct f2fs_node *p = F2FS_NODE(page);
2183
2184 return RAW_IS_INODE(p);
2185 }
2186
2187 static inline int offset_in_addr(struct f2fs_inode *i)
2188 {
2189 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2190 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2191 }
2192
2193 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2194 {
2195 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2196 }
2197
2198 static inline int f2fs_has_extra_attr(struct inode *inode);
2199 static inline block_t datablock_addr(struct inode *inode,
2200 struct page *node_page, unsigned int offset)
2201 {
2202 struct f2fs_node *raw_node;
2203 __le32 *addr_array;
2204 int base = 0;
2205 bool is_inode = IS_INODE(node_page);
2206
2207 raw_node = F2FS_NODE(node_page);
2208
2209 /* from GC path only */
2210 if (is_inode) {
2211 if (!inode)
2212 base = offset_in_addr(&raw_node->i);
2213 else if (f2fs_has_extra_attr(inode))
2214 base = get_extra_isize(inode);
2215 }
2216
2217 addr_array = blkaddr_in_node(raw_node);
2218 return le32_to_cpu(addr_array[base + offset]);
2219 }
2220
2221 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2222 {
2223 int mask;
2224
2225 addr += (nr >> 3);
2226 mask = 1 << (7 - (nr & 0x07));
2227 return mask & *addr;
2228 }
2229
2230 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2231 {
2232 int mask;
2233
2234 addr += (nr >> 3);
2235 mask = 1 << (7 - (nr & 0x07));
2236 *addr |= mask;
2237 }
2238
2239 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2240 {
2241 int mask;
2242
2243 addr += (nr >> 3);
2244 mask = 1 << (7 - (nr & 0x07));
2245 *addr &= ~mask;
2246 }
2247
2248 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2249 {
2250 int mask;
2251 int ret;
2252
2253 addr += (nr >> 3);
2254 mask = 1 << (7 - (nr & 0x07));
2255 ret = mask & *addr;
2256 *addr |= mask;
2257 return ret;
2258 }
2259
2260 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2261 {
2262 int mask;
2263 int ret;
2264
2265 addr += (nr >> 3);
2266 mask = 1 << (7 - (nr & 0x07));
2267 ret = mask & *addr;
2268 *addr &= ~mask;
2269 return ret;
2270 }
2271
2272 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2273 {
2274 int mask;
2275
2276 addr += (nr >> 3);
2277 mask = 1 << (7 - (nr & 0x07));
2278 *addr ^= mask;
2279 }
2280
2281 /*
2282 * Inode flags
2283 */
2284 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */
2285 #define F2FS_UNRM_FL 0x00000002 /* Undelete */
2286 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2287 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2288 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2289 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2290 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2291 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2292 /* Reserved for compression usage... */
2293 #define F2FS_DIRTY_FL 0x00000100
2294 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
2295 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */
2296 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */
2297 /* End compression flags --- maybe not all used */
2298 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2299 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */
2300 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
2301 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
2302 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2303 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
2304 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
2305 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */
2306 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */
2307 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
2308 #define F2FS_NOCOW_FL 0x00800000 /* Do not cow file */
2309 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
2310 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2311 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
2312
2313 #define F2FS_FL_USER_VISIBLE 0x30CBDFFF /* User visible flags */
2314 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
2315
2316 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */
2317 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \
2318 F2FS_IMMUTABLE_FL | \
2319 F2FS_APPEND_FL | \
2320 F2FS_NODUMP_FL | \
2321 F2FS_NOATIME_FL | \
2322 F2FS_PROJINHERIT_FL)
2323
2324 /* Flags that should be inherited by new inodes from their parent. */
2325 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\
2326 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\
2327 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\
2328 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\
2329 F2FS_PROJINHERIT_FL)
2330
2331 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2332 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL))
2333
2334 /* Flags that are appropriate for non-directories/regular files. */
2335 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2336
2337 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2338 {
2339 if (S_ISDIR(mode))
2340 return flags;
2341 else if (S_ISREG(mode))
2342 return flags & F2FS_REG_FLMASK;
2343 else
2344 return flags & F2FS_OTHER_FLMASK;
2345 }
2346
2347 /* used for f2fs_inode_info->flags */
2348 enum {
2349 FI_NEW_INODE, /* indicate newly allocated inode */
2350 FI_DIRTY_INODE, /* indicate inode is dirty or not */
2351 FI_AUTO_RECOVER, /* indicate inode is recoverable */
2352 FI_DIRTY_DIR, /* indicate directory has dirty pages */
2353 FI_INC_LINK, /* need to increment i_nlink */
2354 FI_ACL_MODE, /* indicate acl mode */
2355 FI_NO_ALLOC, /* should not allocate any blocks */
2356 FI_FREE_NID, /* free allocated nide */
2357 FI_NO_EXTENT, /* not to use the extent cache */
2358 FI_INLINE_XATTR, /* used for inline xattr */
2359 FI_INLINE_DATA, /* used for inline data*/
2360 FI_INLINE_DENTRY, /* used for inline dentry */
2361 FI_APPEND_WRITE, /* inode has appended data */
2362 FI_UPDATE_WRITE, /* inode has in-place-update data */
2363 FI_NEED_IPU, /* used for ipu per file */
2364 FI_ATOMIC_FILE, /* indicate atomic file */
2365 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
2366 FI_VOLATILE_FILE, /* indicate volatile file */
2367 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
2368 FI_DROP_CACHE, /* drop dirty page cache */
2369 FI_DATA_EXIST, /* indicate data exists */
2370 FI_INLINE_DOTS, /* indicate inline dot dentries */
2371 FI_DO_DEFRAG, /* indicate defragment is running */
2372 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
2373 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
2374 FI_HOT_DATA, /* indicate file is hot */
2375 FI_EXTRA_ATTR, /* indicate file has extra attribute */
2376 FI_PROJ_INHERIT, /* indicate file inherits projectid */
2377 FI_PIN_FILE, /* indicate file should not be gced */
2378 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
2379 };
2380
2381 static inline void __mark_inode_dirty_flag(struct inode *inode,
2382 int flag, bool set)
2383 {
2384 switch (flag) {
2385 case FI_INLINE_XATTR:
2386 case FI_INLINE_DATA:
2387 case FI_INLINE_DENTRY:
2388 case FI_NEW_INODE:
2389 if (set)
2390 return;
2391 /* fall through */
2392 case FI_DATA_EXIST:
2393 case FI_INLINE_DOTS:
2394 case FI_PIN_FILE:
2395 f2fs_mark_inode_dirty_sync(inode, true);
2396 }
2397 }
2398
2399 static inline void set_inode_flag(struct inode *inode, int flag)
2400 {
2401 if (!test_bit(flag, &F2FS_I(inode)->flags))
2402 set_bit(flag, &F2FS_I(inode)->flags);
2403 __mark_inode_dirty_flag(inode, flag, true);
2404 }
2405
2406 static inline int is_inode_flag_set(struct inode *inode, int flag)
2407 {
2408 return test_bit(flag, &F2FS_I(inode)->flags);
2409 }
2410
2411 static inline void clear_inode_flag(struct inode *inode, int flag)
2412 {
2413 if (test_bit(flag, &F2FS_I(inode)->flags))
2414 clear_bit(flag, &F2FS_I(inode)->flags);
2415 __mark_inode_dirty_flag(inode, flag, false);
2416 }
2417
2418 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2419 {
2420 F2FS_I(inode)->i_acl_mode = mode;
2421 set_inode_flag(inode, FI_ACL_MODE);
2422 f2fs_mark_inode_dirty_sync(inode, false);
2423 }
2424
2425 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2426 {
2427 if (inc)
2428 inc_nlink(inode);
2429 else
2430 drop_nlink(inode);
2431 f2fs_mark_inode_dirty_sync(inode, true);
2432 }
2433
2434 static inline void f2fs_i_blocks_write(struct inode *inode,
2435 block_t diff, bool add, bool claim)
2436 {
2437 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2438 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2439
2440 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2441 if (add) {
2442 if (claim)
2443 dquot_claim_block(inode, diff);
2444 else
2445 dquot_alloc_block_nofail(inode, diff);
2446 } else {
2447 dquot_free_block(inode, diff);
2448 }
2449
2450 f2fs_mark_inode_dirty_sync(inode, true);
2451 if (clean || recover)
2452 set_inode_flag(inode, FI_AUTO_RECOVER);
2453 }
2454
2455 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2456 {
2457 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2458 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2459
2460 if (i_size_read(inode) == i_size)
2461 return;
2462
2463 i_size_write(inode, i_size);
2464 f2fs_mark_inode_dirty_sync(inode, true);
2465 if (clean || recover)
2466 set_inode_flag(inode, FI_AUTO_RECOVER);
2467 }
2468
2469 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2470 {
2471 F2FS_I(inode)->i_current_depth = depth;
2472 f2fs_mark_inode_dirty_sync(inode, true);
2473 }
2474
2475 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2476 unsigned int count)
2477 {
2478 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2479 f2fs_mark_inode_dirty_sync(inode, true);
2480 }
2481
2482 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2483 {
2484 F2FS_I(inode)->i_xattr_nid = xnid;
2485 f2fs_mark_inode_dirty_sync(inode, true);
2486 }
2487
2488 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2489 {
2490 F2FS_I(inode)->i_pino = pino;
2491 f2fs_mark_inode_dirty_sync(inode, true);
2492 }
2493
2494 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2495 {
2496 struct f2fs_inode_info *fi = F2FS_I(inode);
2497
2498 if (ri->i_inline & F2FS_INLINE_XATTR)
2499 set_bit(FI_INLINE_XATTR, &fi->flags);
2500 if (ri->i_inline & F2FS_INLINE_DATA)
2501 set_bit(FI_INLINE_DATA, &fi->flags);
2502 if (ri->i_inline & F2FS_INLINE_DENTRY)
2503 set_bit(FI_INLINE_DENTRY, &fi->flags);
2504 if (ri->i_inline & F2FS_DATA_EXIST)
2505 set_bit(FI_DATA_EXIST, &fi->flags);
2506 if (ri->i_inline & F2FS_INLINE_DOTS)
2507 set_bit(FI_INLINE_DOTS, &fi->flags);
2508 if (ri->i_inline & F2FS_EXTRA_ATTR)
2509 set_bit(FI_EXTRA_ATTR, &fi->flags);
2510 if (ri->i_inline & F2FS_PIN_FILE)
2511 set_bit(FI_PIN_FILE, &fi->flags);
2512 }
2513
2514 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2515 {
2516 ri->i_inline = 0;
2517
2518 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2519 ri->i_inline |= F2FS_INLINE_XATTR;
2520 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2521 ri->i_inline |= F2FS_INLINE_DATA;
2522 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2523 ri->i_inline |= F2FS_INLINE_DENTRY;
2524 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2525 ri->i_inline |= F2FS_DATA_EXIST;
2526 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2527 ri->i_inline |= F2FS_INLINE_DOTS;
2528 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2529 ri->i_inline |= F2FS_EXTRA_ATTR;
2530 if (is_inode_flag_set(inode, FI_PIN_FILE))
2531 ri->i_inline |= F2FS_PIN_FILE;
2532 }
2533
2534 static inline int f2fs_has_extra_attr(struct inode *inode)
2535 {
2536 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2537 }
2538
2539 static inline int f2fs_has_inline_xattr(struct inode *inode)
2540 {
2541 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2542 }
2543
2544 static inline unsigned int addrs_per_inode(struct inode *inode)
2545 {
2546 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode);
2547 }
2548
2549 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2550 {
2551 struct f2fs_inode *ri = F2FS_INODE(page);
2552
2553 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2554 get_inline_xattr_addrs(inode)]);
2555 }
2556
2557 static inline int inline_xattr_size(struct inode *inode)
2558 {
2559 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2560 }
2561
2562 static inline int f2fs_has_inline_data(struct inode *inode)
2563 {
2564 return is_inode_flag_set(inode, FI_INLINE_DATA);
2565 }
2566
2567 static inline int f2fs_exist_data(struct inode *inode)
2568 {
2569 return is_inode_flag_set(inode, FI_DATA_EXIST);
2570 }
2571
2572 static inline int f2fs_has_inline_dots(struct inode *inode)
2573 {
2574 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2575 }
2576
2577 static inline bool f2fs_is_pinned_file(struct inode *inode)
2578 {
2579 return is_inode_flag_set(inode, FI_PIN_FILE);
2580 }
2581
2582 static inline bool f2fs_is_atomic_file(struct inode *inode)
2583 {
2584 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2585 }
2586
2587 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2588 {
2589 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2590 }
2591
2592 static inline bool f2fs_is_volatile_file(struct inode *inode)
2593 {
2594 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2595 }
2596
2597 static inline bool f2fs_is_first_block_written(struct inode *inode)
2598 {
2599 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2600 }
2601
2602 static inline bool f2fs_is_drop_cache(struct inode *inode)
2603 {
2604 return is_inode_flag_set(inode, FI_DROP_CACHE);
2605 }
2606
2607 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2608 {
2609 struct f2fs_inode *ri = F2FS_INODE(page);
2610 int extra_size = get_extra_isize(inode);
2611
2612 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2613 }
2614
2615 static inline int f2fs_has_inline_dentry(struct inode *inode)
2616 {
2617 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2618 }
2619
2620 static inline int is_file(struct inode *inode, int type)
2621 {
2622 return F2FS_I(inode)->i_advise & type;
2623 }
2624
2625 static inline void set_file(struct inode *inode, int type)
2626 {
2627 F2FS_I(inode)->i_advise |= type;
2628 f2fs_mark_inode_dirty_sync(inode, true);
2629 }
2630
2631 static inline void clear_file(struct inode *inode, int type)
2632 {
2633 F2FS_I(inode)->i_advise &= ~type;
2634 f2fs_mark_inode_dirty_sync(inode, true);
2635 }
2636
2637 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2638 {
2639 bool ret;
2640
2641 if (dsync) {
2642 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2643
2644 spin_lock(&sbi->inode_lock[DIRTY_META]);
2645 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2646 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2647 return ret;
2648 }
2649 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2650 file_keep_isize(inode) ||
2651 i_size_read(inode) & ~PAGE_MASK)
2652 return false;
2653
2654 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2655 return false;
2656 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2657 return false;
2658 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2659 return false;
2660 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2661 &F2FS_I(inode)->i_crtime))
2662 return false;
2663
2664 down_read(&F2FS_I(inode)->i_sem);
2665 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2666 up_read(&F2FS_I(inode)->i_sem);
2667
2668 return ret;
2669 }
2670
2671 static inline bool f2fs_readonly(struct super_block *sb)
2672 {
2673 return sb_rdonly(sb);
2674 }
2675
2676 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2677 {
2678 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2679 }
2680
2681 static inline bool is_dot_dotdot(const struct qstr *str)
2682 {
2683 if (str->len == 1 && str->name[0] == '.')
2684 return true;
2685
2686 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2687 return true;
2688
2689 return false;
2690 }
2691
2692 static inline bool f2fs_may_extent_tree(struct inode *inode)
2693 {
2694 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2695
2696 if (!test_opt(sbi, EXTENT_CACHE) ||
2697 is_inode_flag_set(inode, FI_NO_EXTENT))
2698 return false;
2699
2700 /*
2701 * for recovered files during mount do not create extents
2702 * if shrinker is not registered.
2703 */
2704 if (list_empty(&sbi->s_list))
2705 return false;
2706
2707 return S_ISREG(inode->i_mode);
2708 }
2709
2710 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2711 size_t size, gfp_t flags)
2712 {
2713 void *ret;
2714
2715 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2716 f2fs_show_injection_info(FAULT_KMALLOC);
2717 return NULL;
2718 }
2719
2720 ret = kmalloc(size, flags);
2721 if (ret)
2722 return ret;
2723
2724 return kvmalloc(size, flags);
2725 }
2726
2727 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2728 size_t size, gfp_t flags)
2729 {
2730 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2731 }
2732
2733 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2734 size_t size, gfp_t flags)
2735 {
2736 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2737 f2fs_show_injection_info(FAULT_KVMALLOC);
2738 return NULL;
2739 }
2740
2741 return kvmalloc(size, flags);
2742 }
2743
2744 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2745 size_t size, gfp_t flags)
2746 {
2747 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2748 }
2749
2750 static inline int get_extra_isize(struct inode *inode)
2751 {
2752 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2753 }
2754
2755 static inline int get_inline_xattr_addrs(struct inode *inode)
2756 {
2757 return F2FS_I(inode)->i_inline_xattr_size;
2758 }
2759
2760 #define f2fs_get_inode_mode(i) \
2761 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
2762 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
2763
2764 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
2765 (offsetof(struct f2fs_inode, i_extra_end) - \
2766 offsetof(struct f2fs_inode, i_extra_isize)) \
2767
2768 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
2769 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
2770 ((offsetof(typeof(*(f2fs_inode)), field) + \
2771 sizeof((f2fs_inode)->field)) \
2772 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
2773
2774 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
2775 {
2776 int i;
2777
2778 spin_lock(&sbi->iostat_lock);
2779 for (i = 0; i < NR_IO_TYPE; i++)
2780 sbi->write_iostat[i] = 0;
2781 spin_unlock(&sbi->iostat_lock);
2782 }
2783
2784 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
2785 enum iostat_type type, unsigned long long io_bytes)
2786 {
2787 if (!sbi->iostat_enable)
2788 return;
2789 spin_lock(&sbi->iostat_lock);
2790 sbi->write_iostat[type] += io_bytes;
2791
2792 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
2793 sbi->write_iostat[APP_BUFFERED_IO] =
2794 sbi->write_iostat[APP_WRITE_IO] -
2795 sbi->write_iostat[APP_DIRECT_IO];
2796 spin_unlock(&sbi->iostat_lock);
2797 }
2798
2799 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
2800
2801 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META && \
2802 (!is_read_io((fio)->op) || (fio)->is_meta))
2803
2804 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
2805 block_t blkaddr, int type);
2806 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2807 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
2808 block_t blkaddr, int type)
2809 {
2810 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
2811 f2fs_msg(sbi->sb, KERN_ERR,
2812 "invalid blkaddr: %u, type: %d, run fsck to fix.",
2813 blkaddr, type);
2814 f2fs_bug_on(sbi, 1);
2815 }
2816 }
2817
2818 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
2819 {
2820 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR)
2821 return false;
2822 return true;
2823 }
2824
2825 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi,
2826 block_t blkaddr)
2827 {
2828 if (!__is_valid_data_blkaddr(blkaddr))
2829 return false;
2830 verify_blkaddr(sbi, blkaddr, DATA_GENERIC);
2831 return true;
2832 }
2833
2834 /*
2835 * file.c
2836 */
2837 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
2838 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
2839 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock,
2840 bool buf_write);
2841 int f2fs_truncate(struct inode *inode);
2842 int f2fs_getattr(const struct path *path, struct kstat *stat,
2843 u32 request_mask, unsigned int flags);
2844 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
2845 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
2846 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
2847 int f2fs_precache_extents(struct inode *inode);
2848 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2849 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2850 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
2851 int f2fs_pin_file_control(struct inode *inode, bool inc);
2852
2853 /*
2854 * inode.c
2855 */
2856 void f2fs_set_inode_flags(struct inode *inode);
2857 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
2858 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
2859 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
2860 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
2861 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
2862 void f2fs_update_inode(struct inode *inode, struct page *node_page);
2863 void f2fs_update_inode_page(struct inode *inode);
2864 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
2865 void f2fs_evict_inode(struct inode *inode);
2866 void f2fs_handle_failed_inode(struct inode *inode);
2867
2868 /*
2869 * namei.c
2870 */
2871 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
2872 bool hot, bool set);
2873 struct dentry *f2fs_get_parent(struct dentry *child);
2874
2875 /*
2876 * dir.c
2877 */
2878 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
2879 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
2880 f2fs_hash_t namehash, int *max_slots,
2881 struct f2fs_dentry_ptr *d);
2882 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
2883 unsigned int start_pos, struct fscrypt_str *fstr);
2884 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
2885 struct f2fs_dentry_ptr *d);
2886 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
2887 const struct qstr *new_name,
2888 const struct qstr *orig_name, struct page *dpage);
2889 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
2890 unsigned int current_depth);
2891 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
2892 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
2893 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
2894 struct fscrypt_name *fname, struct page **res_page);
2895 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
2896 const struct qstr *child, struct page **res_page);
2897 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
2898 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
2899 struct page **page);
2900 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
2901 struct page *page, struct inode *inode);
2902 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
2903 const struct qstr *name, f2fs_hash_t name_hash,
2904 unsigned int bit_pos);
2905 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
2906 const struct qstr *orig_name,
2907 struct inode *inode, nid_t ino, umode_t mode);
2908 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
2909 struct inode *inode, nid_t ino, umode_t mode);
2910 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
2911 struct inode *inode, nid_t ino, umode_t mode);
2912 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
2913 struct inode *dir, struct inode *inode);
2914 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
2915 bool f2fs_empty_dir(struct inode *dir);
2916
2917 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
2918 {
2919 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
2920 inode, inode->i_ino, inode->i_mode);
2921 }
2922
2923 /*
2924 * super.c
2925 */
2926 int f2fs_inode_dirtied(struct inode *inode, bool sync);
2927 void f2fs_inode_synced(struct inode *inode);
2928 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
2929 int f2fs_quota_sync(struct super_block *sb, int type);
2930 void f2fs_quota_off_umount(struct super_block *sb);
2931 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
2932 int f2fs_sync_fs(struct super_block *sb, int sync);
2933 extern __printf(3, 4)
2934 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...);
2935 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
2936
2937 /*
2938 * hash.c
2939 */
2940 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info,
2941 struct fscrypt_name *fname);
2942
2943 /*
2944 * node.c
2945 */
2946 struct dnode_of_data;
2947 struct node_info;
2948
2949 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
2950 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
2951 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
2952 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
2953 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
2954 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
2955 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
2956 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
2957 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
2958 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
2959 struct node_info *ni);
2960 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
2961 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
2962 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
2963 int f2fs_truncate_xattr_node(struct inode *inode);
2964 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
2965 unsigned int seq_id);
2966 int f2fs_remove_inode_page(struct inode *inode);
2967 struct page *f2fs_new_inode_page(struct inode *inode);
2968 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
2969 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
2970 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
2971 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
2972 int f2fs_move_node_page(struct page *node_page, int gc_type);
2973 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
2974 struct writeback_control *wbc, bool atomic,
2975 unsigned int *seq_id);
2976 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
2977 struct writeback_control *wbc,
2978 bool do_balance, enum iostat_type io_type);
2979 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
2980 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
2981 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
2982 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
2983 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
2984 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
2985 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
2986 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
2987 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
2988 unsigned int segno, struct f2fs_summary_block *sum);
2989 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
2990 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
2991 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
2992 int __init f2fs_create_node_manager_caches(void);
2993 void f2fs_destroy_node_manager_caches(void);
2994
2995 /*
2996 * segment.c
2997 */
2998 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
2999 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3000 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3001 void f2fs_drop_inmem_pages(struct inode *inode);
3002 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3003 int f2fs_commit_inmem_pages(struct inode *inode);
3004 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3005 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi);
3006 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3007 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3008 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3009 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3010 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3011 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3012 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3013 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3014 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3015 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3016 struct cp_control *cpc);
3017 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3018 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi);
3019 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3020 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3021 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3022 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3023 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3024 struct cp_control *cpc);
3025 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3026 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3027 block_t blk_addr);
3028 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3029 enum iostat_type io_type);
3030 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3031 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3032 struct f2fs_io_info *fio);
3033 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3034 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3035 block_t old_blkaddr, block_t new_blkaddr,
3036 bool recover_curseg, bool recover_newaddr);
3037 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3038 block_t old_addr, block_t new_addr,
3039 unsigned char version, bool recover_curseg,
3040 bool recover_newaddr);
3041 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3042 block_t old_blkaddr, block_t *new_blkaddr,
3043 struct f2fs_summary *sum, int type,
3044 struct f2fs_io_info *fio, bool add_list);
3045 void f2fs_wait_on_page_writeback(struct page *page,
3046 enum page_type type, bool ordered, bool locked);
3047 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3048 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3049 block_t len);
3050 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3051 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3052 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3053 unsigned int val, int alloc);
3054 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3055 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3056 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3057 int __init f2fs_create_segment_manager_caches(void);
3058 void f2fs_destroy_segment_manager_caches(void);
3059 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3060 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3061 enum page_type type, enum temp_type temp);
3062
3063 /*
3064 * checkpoint.c
3065 */
3066 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3067 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3068 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3069 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3070 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3071 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3072 block_t blkaddr, int type);
3073 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3074 int type, bool sync);
3075 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3076 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3077 long nr_to_write, enum iostat_type io_type);
3078 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3079 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3080 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3081 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3082 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3083 unsigned int devidx, int type);
3084 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3085 unsigned int devidx, int type);
3086 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3087 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3088 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3089 void f2fs_add_orphan_inode(struct inode *inode);
3090 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3091 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3092 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3093 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3094 void f2fs_remove_dirty_inode(struct inode *inode);
3095 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3096 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi);
3097 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3098 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3099 int __init f2fs_create_checkpoint_caches(void);
3100 void f2fs_destroy_checkpoint_caches(void);
3101
3102 /*
3103 * data.c
3104 */
3105 int f2fs_init_post_read_processing(void);
3106 void f2fs_destroy_post_read_processing(void);
3107 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3108 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3109 struct inode *inode, struct page *page,
3110 nid_t ino, enum page_type type);
3111 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3112 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3113 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3114 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3115 block_t blk_addr, struct bio *bio);
3116 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3117 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3118 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3119 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3120 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3121 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3122 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3123 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3124 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3125 int op_flags, bool for_write);
3126 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3127 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3128 bool for_write);
3129 struct page *f2fs_get_new_data_page(struct inode *inode,
3130 struct page *ipage, pgoff_t index, bool new_i_size);
3131 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3132 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3133 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3134 int create, int flag);
3135 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3136 u64 start, u64 len);
3137 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3138 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3139 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3140 unsigned int length);
3141 int f2fs_release_page(struct page *page, gfp_t wait);
3142 #ifdef CONFIG_MIGRATION
3143 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3144 struct page *page, enum migrate_mode mode);
3145 #endif
3146 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3147 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3148
3149 /*
3150 * gc.c
3151 */
3152 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3153 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3154 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3155 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3156 unsigned int segno);
3157 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3158
3159 /*
3160 * recovery.c
3161 */
3162 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3163 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3164
3165 /*
3166 * debug.c
3167 */
3168 #ifdef CONFIG_F2FS_STAT_FS
3169 struct f2fs_stat_info {
3170 struct list_head stat_list;
3171 struct f2fs_sb_info *sbi;
3172 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3173 int main_area_segs, main_area_sections, main_area_zones;
3174 unsigned long long hit_largest, hit_cached, hit_rbtree;
3175 unsigned long long hit_total, total_ext;
3176 int ext_tree, zombie_tree, ext_node;
3177 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3178 int ndirty_data, ndirty_qdata;
3179 int inmem_pages;
3180 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3181 int nats, dirty_nats, sits, dirty_sits;
3182 int free_nids, avail_nids, alloc_nids;
3183 int total_count, utilization;
3184 int bg_gc, nr_wb_cp_data, nr_wb_data;
3185 int nr_rd_data, nr_rd_node, nr_rd_meta;
3186 int nr_dio_read, nr_dio_write;
3187 unsigned int io_skip_bggc, other_skip_bggc;
3188 int nr_flushing, nr_flushed, flush_list_empty;
3189 int nr_discarding, nr_discarded;
3190 int nr_discard_cmd;
3191 unsigned int undiscard_blks;
3192 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3193 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3194 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3195 unsigned int bimodal, avg_vblocks;
3196 int util_free, util_valid, util_invalid;
3197 int rsvd_segs, overp_segs;
3198 int dirty_count, node_pages, meta_pages;
3199 int prefree_count, call_count, cp_count, bg_cp_count;
3200 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3201 int bg_node_segs, bg_data_segs;
3202 int tot_blks, data_blks, node_blks;
3203 int bg_data_blks, bg_node_blks;
3204 unsigned long long skipped_atomic_files[2];
3205 int curseg[NR_CURSEG_TYPE];
3206 int cursec[NR_CURSEG_TYPE];
3207 int curzone[NR_CURSEG_TYPE];
3208
3209 unsigned int meta_count[META_MAX];
3210 unsigned int segment_count[2];
3211 unsigned int block_count[2];
3212 unsigned int inplace_count;
3213 unsigned long long base_mem, cache_mem, page_mem;
3214 };
3215
3216 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3217 {
3218 return (struct f2fs_stat_info *)sbi->stat_info;
3219 }
3220
3221 #define stat_inc_cp_count(si) ((si)->cp_count++)
3222 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3223 #define stat_inc_call_count(si) ((si)->call_count++)
3224 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++)
3225 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3226 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3227 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3228 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3229 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3230 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3231 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3232 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3233 #define stat_inc_inline_xattr(inode) \
3234 do { \
3235 if (f2fs_has_inline_xattr(inode)) \
3236 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3237 } while (0)
3238 #define stat_dec_inline_xattr(inode) \
3239 do { \
3240 if (f2fs_has_inline_xattr(inode)) \
3241 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3242 } while (0)
3243 #define stat_inc_inline_inode(inode) \
3244 do { \
3245 if (f2fs_has_inline_data(inode)) \
3246 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3247 } while (0)
3248 #define stat_dec_inline_inode(inode) \
3249 do { \
3250 if (f2fs_has_inline_data(inode)) \
3251 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3252 } while (0)
3253 #define stat_inc_inline_dir(inode) \
3254 do { \
3255 if (f2fs_has_inline_dentry(inode)) \
3256 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3257 } while (0)
3258 #define stat_dec_inline_dir(inode) \
3259 do { \
3260 if (f2fs_has_inline_dentry(inode)) \
3261 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3262 } while (0)
3263 #define stat_inc_meta_count(sbi, blkaddr) \
3264 do { \
3265 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3266 atomic_inc(&(sbi)->meta_count[META_CP]); \
3267 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3268 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3269 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3270 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3271 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3272 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3273 } while (0)
3274 #define stat_inc_seg_type(sbi, curseg) \
3275 ((sbi)->segment_count[(curseg)->alloc_type]++)
3276 #define stat_inc_block_count(sbi, curseg) \
3277 ((sbi)->block_count[(curseg)->alloc_type]++)
3278 #define stat_inc_inplace_blocks(sbi) \
3279 (atomic_inc(&(sbi)->inplace_count))
3280 #define stat_inc_atomic_write(inode) \
3281 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt))
3282 #define stat_dec_atomic_write(inode) \
3283 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt))
3284 #define stat_update_max_atomic_write(inode) \
3285 do { \
3286 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \
3287 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3288 if (cur > max) \
3289 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3290 } while (0)
3291 #define stat_inc_volatile_write(inode) \
3292 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3293 #define stat_dec_volatile_write(inode) \
3294 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3295 #define stat_update_max_volatile_write(inode) \
3296 do { \
3297 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3298 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3299 if (cur > max) \
3300 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3301 } while (0)
3302 #define stat_inc_seg_count(sbi, type, gc_type) \
3303 do { \
3304 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3305 si->tot_segs++; \
3306 if ((type) == SUM_TYPE_DATA) { \
3307 si->data_segs++; \
3308 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3309 } else { \
3310 si->node_segs++; \
3311 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3312 } \
3313 } while (0)
3314
3315 #define stat_inc_tot_blk_count(si, blks) \
3316 ((si)->tot_blks += (blks))
3317
3318 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3319 do { \
3320 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3321 stat_inc_tot_blk_count(si, blks); \
3322 si->data_blks += (blks); \
3323 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3324 } while (0)
3325
3326 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3327 do { \
3328 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3329 stat_inc_tot_blk_count(si, blks); \
3330 si->node_blks += (blks); \
3331 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3332 } while (0)
3333
3334 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3335 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3336 void __init f2fs_create_root_stats(void);
3337 void f2fs_destroy_root_stats(void);
3338 #else
3339 #define stat_inc_cp_count(si) do { } while (0)
3340 #define stat_inc_bg_cp_count(si) do { } while (0)
3341 #define stat_inc_call_count(si) do { } while (0)
3342 #define stat_inc_bggc_count(si) do { } while (0)
3343 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3344 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3345 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3346 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3347 #define stat_inc_total_hit(sb) do { } while (0)
3348 #define stat_inc_rbtree_node_hit(sb) do { } while (0)
3349 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3350 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3351 #define stat_inc_inline_xattr(inode) do { } while (0)
3352 #define stat_dec_inline_xattr(inode) do { } while (0)
3353 #define stat_inc_inline_inode(inode) do { } while (0)
3354 #define stat_dec_inline_inode(inode) do { } while (0)
3355 #define stat_inc_inline_dir(inode) do { } while (0)
3356 #define stat_dec_inline_dir(inode) do { } while (0)
3357 #define stat_inc_atomic_write(inode) do { } while (0)
3358 #define stat_dec_atomic_write(inode) do { } while (0)
3359 #define stat_update_max_atomic_write(inode) do { } while (0)
3360 #define stat_inc_volatile_write(inode) do { } while (0)
3361 #define stat_dec_volatile_write(inode) do { } while (0)
3362 #define stat_update_max_volatile_write(inode) do { } while (0)
3363 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3364 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3365 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3366 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3367 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3368 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3369 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3370 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3371
3372 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3373 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3374 static inline void __init f2fs_create_root_stats(void) { }
3375 static inline void f2fs_destroy_root_stats(void) { }
3376 #endif
3377
3378 extern const struct file_operations f2fs_dir_operations;
3379 extern const struct file_operations f2fs_file_operations;
3380 extern const struct inode_operations f2fs_file_inode_operations;
3381 extern const struct address_space_operations f2fs_dblock_aops;
3382 extern const struct address_space_operations f2fs_node_aops;
3383 extern const struct address_space_operations f2fs_meta_aops;
3384 extern const struct inode_operations f2fs_dir_inode_operations;
3385 extern const struct inode_operations f2fs_symlink_inode_operations;
3386 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3387 extern const struct inode_operations f2fs_special_inode_operations;
3388 extern struct kmem_cache *f2fs_inode_entry_slab;
3389
3390 /*
3391 * inline.c
3392 */
3393 bool f2fs_may_inline_data(struct inode *inode);
3394 bool f2fs_may_inline_dentry(struct inode *inode);
3395 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3396 void f2fs_truncate_inline_inode(struct inode *inode,
3397 struct page *ipage, u64 from);
3398 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3399 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3400 int f2fs_convert_inline_inode(struct inode *inode);
3401 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3402 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3403 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3404 struct fscrypt_name *fname, struct page **res_page);
3405 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3406 struct page *ipage);
3407 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3408 const struct qstr *orig_name,
3409 struct inode *inode, nid_t ino, umode_t mode);
3410 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3411 struct page *page, struct inode *dir,
3412 struct inode *inode);
3413 bool f2fs_empty_inline_dir(struct inode *dir);
3414 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3415 struct fscrypt_str *fstr);
3416 int f2fs_inline_data_fiemap(struct inode *inode,
3417 struct fiemap_extent_info *fieinfo,
3418 __u64 start, __u64 len);
3419
3420 /*
3421 * shrinker.c
3422 */
3423 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3424 struct shrink_control *sc);
3425 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3426 struct shrink_control *sc);
3427 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3428 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3429
3430 /*
3431 * extent_cache.c
3432 */
3433 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3434 struct rb_entry *cached_re, unsigned int ofs);
3435 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3436 struct rb_root_cached *root,
3437 struct rb_node **parent,
3438 unsigned int ofs, bool *leftmost);
3439 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3440 struct rb_entry *cached_re, unsigned int ofs,
3441 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3442 struct rb_node ***insert_p, struct rb_node **insert_parent,
3443 bool force, bool *leftmost);
3444 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3445 struct rb_root_cached *root);
3446 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3447 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3448 void f2fs_drop_extent_tree(struct inode *inode);
3449 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3450 void f2fs_destroy_extent_tree(struct inode *inode);
3451 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3452 struct extent_info *ei);
3453 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3454 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3455 pgoff_t fofs, block_t blkaddr, unsigned int len);
3456 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3457 int __init f2fs_create_extent_cache(void);
3458 void f2fs_destroy_extent_cache(void);
3459
3460 /*
3461 * sysfs.c
3462 */
3463 int __init f2fs_init_sysfs(void);
3464 void f2fs_exit_sysfs(void);
3465 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3466 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3467
3468 /*
3469 * crypto support
3470 */
3471 static inline bool f2fs_encrypted_inode(struct inode *inode)
3472 {
3473 return file_is_encrypt(inode);
3474 }
3475
3476 static inline bool f2fs_encrypted_file(struct inode *inode)
3477 {
3478 return f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode);
3479 }
3480
3481 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3482 {
3483 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3484 file_set_encrypt(inode);
3485 f2fs_set_inode_flags(inode);
3486 #endif
3487 }
3488
3489 /*
3490 * Returns true if the reads of the inode's data need to undergo some
3491 * postprocessing step, like decryption or authenticity verification.
3492 */
3493 static inline bool f2fs_post_read_required(struct inode *inode)
3494 {
3495 return f2fs_encrypted_file(inode);
3496 }
3497
3498 #define F2FS_FEATURE_FUNCS(name, flagname) \
3499 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3500 { \
3501 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3502 }
3503
3504 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3505 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3506 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3507 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3508 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3509 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3510 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3511 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3512 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3513 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3514
3515 #ifdef CONFIG_BLK_DEV_ZONED
3516 static inline int get_blkz_type(struct f2fs_sb_info *sbi,
3517 struct block_device *bdev, block_t blkaddr)
3518 {
3519 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3520 int i;
3521
3522 for (i = 0; i < sbi->s_ndevs; i++)
3523 if (FDEV(i).bdev == bdev)
3524 return FDEV(i).blkz_type[zno];
3525 return -EINVAL;
3526 }
3527 #endif
3528
3529 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3530 {
3531 return f2fs_sb_has_blkzoned(sbi);
3532 }
3533
3534 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3535 {
3536 return blk_queue_discard(bdev_get_queue(sbi->sb->s_bdev));
3537 }
3538
3539 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3540 {
3541 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3542 f2fs_hw_should_discard(sbi);
3543 }
3544
3545 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt)
3546 {
3547 clear_opt(sbi, ADAPTIVE);
3548 clear_opt(sbi, LFS);
3549
3550 switch (mt) {
3551 case F2FS_MOUNT_ADAPTIVE:
3552 set_opt(sbi, ADAPTIVE);
3553 break;
3554 case F2FS_MOUNT_LFS:
3555 set_opt(sbi, LFS);
3556 break;
3557 }
3558 }
3559
3560 static inline bool f2fs_may_encrypt(struct inode *inode)
3561 {
3562 #ifdef CONFIG_F2FS_FS_ENCRYPTION
3563 umode_t mode = inode->i_mode;
3564
3565 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3566 #else
3567 return false;
3568 #endif
3569 }
3570
3571 static inline int block_unaligned_IO(struct inode *inode,
3572 struct kiocb *iocb, struct iov_iter *iter)
3573 {
3574 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
3575 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
3576 loff_t offset = iocb->ki_pos;
3577 unsigned long align = offset | iov_iter_alignment(iter);
3578
3579 return align & blocksize_mask;
3580 }
3581
3582 static inline int allow_outplace_dio(struct inode *inode,
3583 struct kiocb *iocb, struct iov_iter *iter)
3584 {
3585 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3586 int rw = iov_iter_rw(iter);
3587
3588 return (test_opt(sbi, LFS) && (rw == WRITE) &&
3589 !block_unaligned_IO(inode, iocb, iter));
3590 }
3591
3592 static inline bool f2fs_force_buffered_io(struct inode *inode,
3593 struct kiocb *iocb, struct iov_iter *iter)
3594 {
3595 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3596 int rw = iov_iter_rw(iter);
3597
3598 if (f2fs_post_read_required(inode))
3599 return true;
3600 if (sbi->s_ndevs)
3601 return true;
3602 /*
3603 * for blkzoned device, fallback direct IO to buffered IO, so
3604 * all IOs can be serialized by log-structured write.
3605 */
3606 if (f2fs_sb_has_blkzoned(sbi))
3607 return true;
3608 if (test_opt(sbi, LFS) && (rw == WRITE) &&
3609 block_unaligned_IO(inode, iocb, iter))
3610 return true;
3611 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
3612 return true;
3613
3614 return false;
3615 }
3616
3617 #ifdef CONFIG_F2FS_FAULT_INJECTION
3618 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
3619 unsigned int type);
3620 #else
3621 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
3622 #endif
3623
3624 #endif
3625
3626 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
3627 {
3628 #ifdef CONFIG_QUOTA
3629 if (f2fs_sb_has_quota_ino(sbi))
3630 return true;
3631 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
3632 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
3633 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
3634 return true;
3635 #endif
3636 return false;
3637 }