]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - fs/f2fs/f2fs.h
f2fs: introduce mempool for {,de}compress intermediate page allocation
[mirror_ubuntu-jammy-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 <linux/part_stat.h>
26 #include <crypto/hash.h>
27
28 #include <linux/fscrypt.h>
29 #include <linux/fsverity.h>
30
31 #ifdef CONFIG_F2FS_CHECK_FS
32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition)
33 #else
34 #define f2fs_bug_on(sbi, condition) \
35 do { \
36 if (unlikely(condition)) { \
37 WARN_ON(1); \
38 set_sbi_flag(sbi, SBI_NEED_FSCK); \
39 } \
40 } while (0)
41 #endif
42
43 enum {
44 FAULT_KMALLOC,
45 FAULT_KVMALLOC,
46 FAULT_PAGE_ALLOC,
47 FAULT_PAGE_GET,
48 FAULT_ALLOC_BIO,
49 FAULT_ALLOC_NID,
50 FAULT_ORPHAN,
51 FAULT_BLOCK,
52 FAULT_DIR_DEPTH,
53 FAULT_EVICT_INODE,
54 FAULT_TRUNCATE,
55 FAULT_READ_IO,
56 FAULT_CHECKPOINT,
57 FAULT_DISCARD,
58 FAULT_WRITE_IO,
59 FAULT_MAX,
60 };
61
62 #ifdef CONFIG_F2FS_FAULT_INJECTION
63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
64
65 struct f2fs_fault_info {
66 atomic_t inject_ops;
67 unsigned int inject_rate;
68 unsigned int inject_type;
69 };
70
71 extern const char *f2fs_fault_name[FAULT_MAX];
72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
73 #endif
74
75 /*
76 * For mount options
77 */
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_DATA_FLUSH 0x00008000
92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000
93 #define F2FS_MOUNT_USRQUOTA 0x00080000
94 #define F2FS_MOUNT_GRPQUOTA 0x00100000
95 #define F2FS_MOUNT_PRJQUOTA 0x00200000
96 #define F2FS_MOUNT_QUOTA 0x00400000
97 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
98 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000
99 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
100 #define F2FS_MOUNT_NORECOVERY 0x04000000
101
102 #define F2FS_OPTION(sbi) ((sbi)->mount_opt)
103 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
104 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
105 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
106
107 #define ver_after(a, b) (typecheck(unsigned long long, a) && \
108 typecheck(unsigned long long, b) && \
109 ((long long)((a) - (b)) > 0))
110
111 typedef u32 block_t; /*
112 * should not change u32, since it is the on-disk block
113 * address format, __le32.
114 */
115 typedef u32 nid_t;
116
117 #define COMPRESS_EXT_NUM 16
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 int fs_mode; /* fs mode: LFS or ADAPTIVE */
140 int bggc_mode; /* bggc mode: off, on or sync */
141 bool test_dummy_encryption; /* test dummy encryption */
142 block_t unusable_cap; /* Amount of space allowed to be
143 * unusable when disabling checkpoint
144 */
145
146 /* For compression */
147 unsigned char compress_algorithm; /* algorithm type */
148 unsigned compress_log_size; /* cluster log size */
149 unsigned char compress_ext_cnt; /* extension count */
150 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
151 };
152
153 #define F2FS_FEATURE_ENCRYPT 0x0001
154 #define F2FS_FEATURE_BLKZONED 0x0002
155 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004
156 #define F2FS_FEATURE_EXTRA_ATTR 0x0008
157 #define F2FS_FEATURE_PRJQUOTA 0x0010
158 #define F2FS_FEATURE_INODE_CHKSUM 0x0020
159 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
160 #define F2FS_FEATURE_QUOTA_INO 0x0080
161 #define F2FS_FEATURE_INODE_CRTIME 0x0100
162 #define F2FS_FEATURE_LOST_FOUND 0x0200
163 #define F2FS_FEATURE_VERITY 0x0400
164 #define F2FS_FEATURE_SB_CHKSUM 0x0800
165 #define F2FS_FEATURE_CASEFOLD 0x1000
166 #define F2FS_FEATURE_COMPRESSION 0x2000
167
168 #define __F2FS_HAS_FEATURE(raw_super, mask) \
169 ((raw_super->feature & cpu_to_le32(mask)) != 0)
170 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
171 #define F2FS_SET_FEATURE(sbi, mask) \
172 (sbi->raw_super->feature |= cpu_to_le32(mask))
173 #define F2FS_CLEAR_FEATURE(sbi, mask) \
174 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
175
176 /*
177 * Default values for user and/or group using reserved blocks
178 */
179 #define F2FS_DEF_RESUID 0
180 #define F2FS_DEF_RESGID 0
181
182 /*
183 * For checkpoint manager
184 */
185 enum {
186 NAT_BITMAP,
187 SIT_BITMAP
188 };
189
190 #define CP_UMOUNT 0x00000001
191 #define CP_FASTBOOT 0x00000002
192 #define CP_SYNC 0x00000004
193 #define CP_RECOVERY 0x00000008
194 #define CP_DISCARD 0x00000010
195 #define CP_TRIMMED 0x00000020
196 #define CP_PAUSE 0x00000040
197
198 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
199 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
200 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
201 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
202 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
203 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
204 #define DEF_CP_INTERVAL 60 /* 60 secs */
205 #define DEF_IDLE_INTERVAL 5 /* 5 secs */
206 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */
207 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
208 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
209
210 struct cp_control {
211 int reason;
212 __u64 trim_start;
213 __u64 trim_end;
214 __u64 trim_minlen;
215 };
216
217 /*
218 * indicate meta/data type
219 */
220 enum {
221 META_CP,
222 META_NAT,
223 META_SIT,
224 META_SSA,
225 META_MAX,
226 META_POR,
227 DATA_GENERIC, /* check range only */
228 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
229 DATA_GENERIC_ENHANCE_READ, /*
230 * strong check on range and segment
231 * bitmap but no warning due to race
232 * condition of read on truncated area
233 * by extent_cache
234 */
235 META_GENERIC,
236 };
237
238 /* for the list of ino */
239 enum {
240 ORPHAN_INO, /* for orphan ino list */
241 APPEND_INO, /* for append ino list */
242 UPDATE_INO, /* for update ino list */
243 TRANS_DIR_INO, /* for trasactions dir ino list */
244 FLUSH_INO, /* for multiple device flushing */
245 MAX_INO_ENTRY, /* max. list */
246 };
247
248 struct ino_entry {
249 struct list_head list; /* list head */
250 nid_t ino; /* inode number */
251 unsigned int dirty_device; /* dirty device bitmap */
252 };
253
254 /* for the list of inodes to be GCed */
255 struct inode_entry {
256 struct list_head list; /* list head */
257 struct inode *inode; /* vfs inode pointer */
258 };
259
260 struct fsync_node_entry {
261 struct list_head list; /* list head */
262 struct page *page; /* warm node page pointer */
263 unsigned int seq_id; /* sequence id */
264 };
265
266 /* for the bitmap indicate blocks to be discarded */
267 struct discard_entry {
268 struct list_head list; /* list head */
269 block_t start_blkaddr; /* start blockaddr of current segment */
270 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
271 };
272
273 /* default discard granularity of inner discard thread, unit: block count */
274 #define DEFAULT_DISCARD_GRANULARITY 16
275
276 /* max discard pend list number */
277 #define MAX_PLIST_NUM 512
278 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
279 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
280
281 enum {
282 D_PREP, /* initial */
283 D_PARTIAL, /* partially submitted */
284 D_SUBMIT, /* all submitted */
285 D_DONE, /* finished */
286 };
287
288 struct discard_info {
289 block_t lstart; /* logical start address */
290 block_t len; /* length */
291 block_t start; /* actual start address in dev */
292 };
293
294 struct discard_cmd {
295 struct rb_node rb_node; /* rb node located in rb-tree */
296 union {
297 struct {
298 block_t lstart; /* logical start address */
299 block_t len; /* length */
300 block_t start; /* actual start address in dev */
301 };
302 struct discard_info di; /* discard info */
303
304 };
305 struct list_head list; /* command list */
306 struct completion wait; /* compleation */
307 struct block_device *bdev; /* bdev */
308 unsigned short ref; /* reference count */
309 unsigned char state; /* state */
310 unsigned char queued; /* queued discard */
311 int error; /* bio error */
312 spinlock_t lock; /* for state/bio_ref updating */
313 unsigned short bio_ref; /* bio reference count */
314 };
315
316 enum {
317 DPOLICY_BG,
318 DPOLICY_FORCE,
319 DPOLICY_FSTRIM,
320 DPOLICY_UMOUNT,
321 MAX_DPOLICY,
322 };
323
324 struct discard_policy {
325 int type; /* type of discard */
326 unsigned int min_interval; /* used for candidates exist */
327 unsigned int mid_interval; /* used for device busy */
328 unsigned int max_interval; /* used for candidates not exist */
329 unsigned int max_requests; /* # of discards issued per round */
330 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
331 bool io_aware; /* issue discard in idle time */
332 bool sync; /* submit discard with REQ_SYNC flag */
333 bool ordered; /* issue discard by lba order */
334 bool timeout; /* discard timeout for put_super */
335 unsigned int granularity; /* discard granularity */
336 };
337
338 struct discard_cmd_control {
339 struct task_struct *f2fs_issue_discard; /* discard thread */
340 struct list_head entry_list; /* 4KB discard entry list */
341 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
342 struct list_head wait_list; /* store on-flushing entries */
343 struct list_head fstrim_list; /* in-flight discard from fstrim */
344 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
345 unsigned int discard_wake; /* to wake up discard thread */
346 struct mutex cmd_lock;
347 unsigned int nr_discards; /* # of discards in the list */
348 unsigned int max_discards; /* max. discards to be issued */
349 unsigned int discard_granularity; /* discard granularity */
350 unsigned int undiscard_blks; /* # of undiscard blocks */
351 unsigned int next_pos; /* next discard position */
352 atomic_t issued_discard; /* # of issued discard */
353 atomic_t queued_discard; /* # of queued discard */
354 atomic_t discard_cmd_cnt; /* # of cached cmd count */
355 struct rb_root_cached root; /* root of discard rb-tree */
356 bool rbtree_check; /* config for consistence check */
357 };
358
359 /* for the list of fsync inodes, used only during recovery */
360 struct fsync_inode_entry {
361 struct list_head list; /* list head */
362 struct inode *inode; /* vfs inode pointer */
363 block_t blkaddr; /* block address locating the last fsync */
364 block_t last_dentry; /* block address locating the last dentry */
365 };
366
367 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
368 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
369
370 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
371 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
372 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
373 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
374
375 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
376 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
377
378 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
379 {
380 int before = nats_in_cursum(journal);
381
382 journal->n_nats = cpu_to_le16(before + i);
383 return before;
384 }
385
386 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
387 {
388 int before = sits_in_cursum(journal);
389
390 journal->n_sits = cpu_to_le16(before + i);
391 return before;
392 }
393
394 static inline bool __has_cursum_space(struct f2fs_journal *journal,
395 int size, int type)
396 {
397 if (type == NAT_JOURNAL)
398 return size <= MAX_NAT_JENTRIES(journal);
399 return size <= MAX_SIT_JENTRIES(journal);
400 }
401
402 /*
403 * ioctl commands
404 */
405 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
406 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
407 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
408
409 #define F2FS_IOCTL_MAGIC 0xf5
410 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
411 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
412 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
413 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
414 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
415 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
416 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
417 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
418 struct f2fs_defragment)
419 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
420 struct f2fs_move_range)
421 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
422 struct f2fs_flush_device)
423 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
424 struct f2fs_gc_range)
425 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
426 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
427 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
428 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
429 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64)
430 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64)
431
432 #define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL
433 #define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL
434
435 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
436 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
437 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
438
439 /*
440 * should be same as XFS_IOC_GOINGDOWN.
441 * Flags for going down operation used by FS_IOC_GOINGDOWN
442 */
443 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
444 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
445 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
446 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
447 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
448 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
449
450 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
451 /*
452 * ioctl commands in 32 bit emulation
453 */
454 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
455 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
456 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
457 #endif
458
459 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
460 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
461
462 struct f2fs_gc_range {
463 u32 sync;
464 u64 start;
465 u64 len;
466 };
467
468 struct f2fs_defragment {
469 u64 start;
470 u64 len;
471 };
472
473 struct f2fs_move_range {
474 u32 dst_fd; /* destination fd */
475 u64 pos_in; /* start position in src_fd */
476 u64 pos_out; /* start position in dst_fd */
477 u64 len; /* size to move */
478 };
479
480 struct f2fs_flush_device {
481 u32 dev_num; /* device number to flush */
482 u32 segments; /* # of segments to flush */
483 };
484
485 /* for inline stuff */
486 #define DEF_INLINE_RESERVED_SIZE 1
487 static inline int get_extra_isize(struct inode *inode);
488 static inline int get_inline_xattr_addrs(struct inode *inode);
489 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
490 (CUR_ADDRS_PER_INODE(inode) - \
491 get_inline_xattr_addrs(inode) - \
492 DEF_INLINE_RESERVED_SIZE))
493
494 /* for inline dir */
495 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
496 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
497 BITS_PER_BYTE + 1))
498 #define INLINE_DENTRY_BITMAP_SIZE(inode) \
499 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
500 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
501 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
502 NR_INLINE_DENTRY(inode) + \
503 INLINE_DENTRY_BITMAP_SIZE(inode)))
504
505 /*
506 * For INODE and NODE manager
507 */
508 /* for directory operations */
509 struct f2fs_dentry_ptr {
510 struct inode *inode;
511 void *bitmap;
512 struct f2fs_dir_entry *dentry;
513 __u8 (*filename)[F2FS_SLOT_LEN];
514 int max;
515 int nr_bitmap;
516 };
517
518 static inline void make_dentry_ptr_block(struct inode *inode,
519 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
520 {
521 d->inode = inode;
522 d->max = NR_DENTRY_IN_BLOCK;
523 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
524 d->bitmap = t->dentry_bitmap;
525 d->dentry = t->dentry;
526 d->filename = t->filename;
527 }
528
529 static inline void make_dentry_ptr_inline(struct inode *inode,
530 struct f2fs_dentry_ptr *d, void *t)
531 {
532 int entry_cnt = NR_INLINE_DENTRY(inode);
533 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
534 int reserved_size = INLINE_RESERVED_SIZE(inode);
535
536 d->inode = inode;
537 d->max = entry_cnt;
538 d->nr_bitmap = bitmap_size;
539 d->bitmap = t;
540 d->dentry = t + bitmap_size + reserved_size;
541 d->filename = t + bitmap_size + reserved_size +
542 SIZE_OF_DIR_ENTRY * entry_cnt;
543 }
544
545 /*
546 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
547 * as its node offset to distinguish from index node blocks.
548 * But some bits are used to mark the node block.
549 */
550 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
551 >> OFFSET_BIT_SHIFT)
552 enum {
553 ALLOC_NODE, /* allocate a new node page if needed */
554 LOOKUP_NODE, /* look up a node without readahead */
555 LOOKUP_NODE_RA, /*
556 * look up a node with readahead called
557 * by get_data_block.
558 */
559 };
560
561 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
562
563 /* congestion wait timeout value, default: 20ms */
564 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
565
566 /* maximum retry quota flush count */
567 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
568
569 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
570
571 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
572
573 /* for in-memory extent cache entry */
574 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
575
576 /* number of extent info in extent cache we try to shrink */
577 #define EXTENT_CACHE_SHRINK_NUMBER 128
578
579 struct rb_entry {
580 struct rb_node rb_node; /* rb node located in rb-tree */
581 unsigned int ofs; /* start offset of the entry */
582 unsigned int len; /* length of the entry */
583 };
584
585 struct extent_info {
586 unsigned int fofs; /* start offset in a file */
587 unsigned int len; /* length of the extent */
588 u32 blk; /* start block address of the extent */
589 };
590
591 struct extent_node {
592 struct rb_node rb_node; /* rb node located in rb-tree */
593 struct extent_info ei; /* extent info */
594 struct list_head list; /* node in global extent list of sbi */
595 struct extent_tree *et; /* extent tree pointer */
596 };
597
598 struct extent_tree {
599 nid_t ino; /* inode number */
600 struct rb_root_cached root; /* root of extent info rb-tree */
601 struct extent_node *cached_en; /* recently accessed extent node */
602 struct extent_info largest; /* largested extent info */
603 struct list_head list; /* to be used by sbi->zombie_list */
604 rwlock_t lock; /* protect extent info rb-tree */
605 atomic_t node_cnt; /* # of extent node in rb-tree*/
606 bool largest_updated; /* largest extent updated */
607 };
608
609 /*
610 * This structure is taken from ext4_map_blocks.
611 *
612 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
613 */
614 #define F2FS_MAP_NEW (1 << BH_New)
615 #define F2FS_MAP_MAPPED (1 << BH_Mapped)
616 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
617 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
618 F2FS_MAP_UNWRITTEN)
619
620 struct f2fs_map_blocks {
621 block_t m_pblk;
622 block_t m_lblk;
623 unsigned int m_len;
624 unsigned int m_flags;
625 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
626 pgoff_t *m_next_extent; /* point to next possible extent */
627 int m_seg_type;
628 bool m_may_create; /* indicate it is from write path */
629 };
630
631 /* for flag in get_data_block */
632 enum {
633 F2FS_GET_BLOCK_DEFAULT,
634 F2FS_GET_BLOCK_FIEMAP,
635 F2FS_GET_BLOCK_BMAP,
636 F2FS_GET_BLOCK_DIO,
637 F2FS_GET_BLOCK_PRE_DIO,
638 F2FS_GET_BLOCK_PRE_AIO,
639 F2FS_GET_BLOCK_PRECACHE,
640 };
641
642 /*
643 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
644 */
645 #define FADVISE_COLD_BIT 0x01
646 #define FADVISE_LOST_PINO_BIT 0x02
647 #define FADVISE_ENCRYPT_BIT 0x04
648 #define FADVISE_ENC_NAME_BIT 0x08
649 #define FADVISE_KEEP_SIZE_BIT 0x10
650 #define FADVISE_HOT_BIT 0x20
651 #define FADVISE_VERITY_BIT 0x40
652
653 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
654
655 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
656 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
657 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
658 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
659 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
660 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
661 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
662 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
663 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
664 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
665 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
666 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
667 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
668 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
669 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
670 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
671 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
672 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
673
674 #define DEF_DIR_LEVEL 0
675
676 enum {
677 GC_FAILURE_PIN,
678 GC_FAILURE_ATOMIC,
679 MAX_GC_FAILURE
680 };
681
682 /* used for f2fs_inode_info->flags */
683 enum {
684 FI_NEW_INODE, /* indicate newly allocated inode */
685 FI_DIRTY_INODE, /* indicate inode is dirty or not */
686 FI_AUTO_RECOVER, /* indicate inode is recoverable */
687 FI_DIRTY_DIR, /* indicate directory has dirty pages */
688 FI_INC_LINK, /* need to increment i_nlink */
689 FI_ACL_MODE, /* indicate acl mode */
690 FI_NO_ALLOC, /* should not allocate any blocks */
691 FI_FREE_NID, /* free allocated nide */
692 FI_NO_EXTENT, /* not to use the extent cache */
693 FI_INLINE_XATTR, /* used for inline xattr */
694 FI_INLINE_DATA, /* used for inline data*/
695 FI_INLINE_DENTRY, /* used for inline dentry */
696 FI_APPEND_WRITE, /* inode has appended data */
697 FI_UPDATE_WRITE, /* inode has in-place-update data */
698 FI_NEED_IPU, /* used for ipu per file */
699 FI_ATOMIC_FILE, /* indicate atomic file */
700 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
701 FI_VOLATILE_FILE, /* indicate volatile file */
702 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
703 FI_DROP_CACHE, /* drop dirty page cache */
704 FI_DATA_EXIST, /* indicate data exists */
705 FI_INLINE_DOTS, /* indicate inline dot dentries */
706 FI_DO_DEFRAG, /* indicate defragment is running */
707 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
708 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
709 FI_HOT_DATA, /* indicate file is hot */
710 FI_EXTRA_ATTR, /* indicate file has extra attribute */
711 FI_PROJ_INHERIT, /* indicate file inherits projectid */
712 FI_PIN_FILE, /* indicate file should not be gced */
713 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
714 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
715 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
716 FI_MMAP_FILE, /* indicate file was mmapped */
717 FI_MAX, /* max flag, never be used */
718 };
719
720 struct f2fs_inode_info {
721 struct inode vfs_inode; /* serve a vfs inode */
722 unsigned long i_flags; /* keep an inode flags for ioctl */
723 unsigned char i_advise; /* use to give file attribute hints */
724 unsigned char i_dir_level; /* use for dentry level for large dir */
725 unsigned int i_current_depth; /* only for directory depth */
726 /* for gc failure statistic */
727 unsigned int i_gc_failures[MAX_GC_FAILURE];
728 unsigned int i_pino; /* parent inode number */
729 umode_t i_acl_mode; /* keep file acl mode temporarily */
730
731 /* Use below internally in f2fs*/
732 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
733 struct rw_semaphore i_sem; /* protect fi info */
734 atomic_t dirty_pages; /* # of dirty pages */
735 f2fs_hash_t chash; /* hash value of given file name */
736 unsigned int clevel; /* maximum level of given file name */
737 struct task_struct *task; /* lookup and create consistency */
738 struct task_struct *cp_task; /* separate cp/wb IO stats*/
739 nid_t i_xattr_nid; /* node id that contains xattrs */
740 loff_t last_disk_size; /* lastly written file size */
741 spinlock_t i_size_lock; /* protect last_disk_size */
742
743 #ifdef CONFIG_QUOTA
744 struct dquot *i_dquot[MAXQUOTAS];
745
746 /* quota space reservation, managed internally by quota code */
747 qsize_t i_reserved_quota;
748 #endif
749 struct list_head dirty_list; /* dirty list for dirs and files */
750 struct list_head gdirty_list; /* linked in global dirty list */
751 struct list_head inmem_ilist; /* list for inmem inodes */
752 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
753 struct task_struct *inmem_task; /* store inmemory task */
754 struct mutex inmem_lock; /* lock for inmemory pages */
755 struct extent_tree *extent_tree; /* cached extent_tree entry */
756
757 /* avoid racing between foreground op and gc */
758 struct rw_semaphore i_gc_rwsem[2];
759 struct rw_semaphore i_mmap_sem;
760 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
761
762 int i_extra_isize; /* size of extra space located in i_addr */
763 kprojid_t i_projid; /* id for project quota */
764 int i_inline_xattr_size; /* inline xattr size */
765 struct timespec64 i_crtime; /* inode creation time */
766 struct timespec64 i_disk_time[4];/* inode disk times */
767
768 /* for file compress */
769 u64 i_compr_blocks; /* # of compressed blocks */
770 unsigned char i_compress_algorithm; /* algorithm type */
771 unsigned char i_log_cluster_size; /* log of cluster size */
772 unsigned int i_cluster_size; /* cluster size */
773 };
774
775 static inline void get_extent_info(struct extent_info *ext,
776 struct f2fs_extent *i_ext)
777 {
778 ext->fofs = le32_to_cpu(i_ext->fofs);
779 ext->blk = le32_to_cpu(i_ext->blk);
780 ext->len = le32_to_cpu(i_ext->len);
781 }
782
783 static inline void set_raw_extent(struct extent_info *ext,
784 struct f2fs_extent *i_ext)
785 {
786 i_ext->fofs = cpu_to_le32(ext->fofs);
787 i_ext->blk = cpu_to_le32(ext->blk);
788 i_ext->len = cpu_to_le32(ext->len);
789 }
790
791 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
792 u32 blk, unsigned int len)
793 {
794 ei->fofs = fofs;
795 ei->blk = blk;
796 ei->len = len;
797 }
798
799 static inline bool __is_discard_mergeable(struct discard_info *back,
800 struct discard_info *front, unsigned int max_len)
801 {
802 return (back->lstart + back->len == front->lstart) &&
803 (back->len + front->len <= max_len);
804 }
805
806 static inline bool __is_discard_back_mergeable(struct discard_info *cur,
807 struct discard_info *back, unsigned int max_len)
808 {
809 return __is_discard_mergeable(back, cur, max_len);
810 }
811
812 static inline bool __is_discard_front_mergeable(struct discard_info *cur,
813 struct discard_info *front, unsigned int max_len)
814 {
815 return __is_discard_mergeable(cur, front, max_len);
816 }
817
818 static inline bool __is_extent_mergeable(struct extent_info *back,
819 struct extent_info *front)
820 {
821 return (back->fofs + back->len == front->fofs &&
822 back->blk + back->len == front->blk);
823 }
824
825 static inline bool __is_back_mergeable(struct extent_info *cur,
826 struct extent_info *back)
827 {
828 return __is_extent_mergeable(back, cur);
829 }
830
831 static inline bool __is_front_mergeable(struct extent_info *cur,
832 struct extent_info *front)
833 {
834 return __is_extent_mergeable(cur, front);
835 }
836
837 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
838 static inline void __try_update_largest_extent(struct extent_tree *et,
839 struct extent_node *en)
840 {
841 if (en->ei.len > et->largest.len) {
842 et->largest = en->ei;
843 et->largest_updated = true;
844 }
845 }
846
847 /*
848 * For free nid management
849 */
850 enum nid_state {
851 FREE_NID, /* newly added to free nid list */
852 PREALLOC_NID, /* it is preallocated */
853 MAX_NID_STATE,
854 };
855
856 struct f2fs_nm_info {
857 block_t nat_blkaddr; /* base disk address of NAT */
858 nid_t max_nid; /* maximum possible node ids */
859 nid_t available_nids; /* # of available node ids */
860 nid_t next_scan_nid; /* the next nid to be scanned */
861 unsigned int ram_thresh; /* control the memory footprint */
862 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
863 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
864
865 /* NAT cache management */
866 struct radix_tree_root nat_root;/* root of the nat entry cache */
867 struct radix_tree_root nat_set_root;/* root of the nat set cache */
868 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
869 struct list_head nat_entries; /* cached nat entry list (clean) */
870 spinlock_t nat_list_lock; /* protect clean nat entry list */
871 unsigned int nat_cnt; /* the # of cached nat entries */
872 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
873 unsigned int nat_blocks; /* # of nat blocks */
874
875 /* free node ids management */
876 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
877 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
878 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
879 spinlock_t nid_list_lock; /* protect nid lists ops */
880 struct mutex build_lock; /* lock for build free nids */
881 unsigned char **free_nid_bitmap;
882 unsigned char *nat_block_bitmap;
883 unsigned short *free_nid_count; /* free nid count of NAT block */
884
885 /* for checkpoint */
886 char *nat_bitmap; /* NAT bitmap pointer */
887
888 unsigned int nat_bits_blocks; /* # of nat bits blocks */
889 unsigned char *nat_bits; /* NAT bits blocks */
890 unsigned char *full_nat_bits; /* full NAT pages */
891 unsigned char *empty_nat_bits; /* empty NAT pages */
892 #ifdef CONFIG_F2FS_CHECK_FS
893 char *nat_bitmap_mir; /* NAT bitmap mirror */
894 #endif
895 int bitmap_size; /* bitmap size */
896 };
897
898 /*
899 * this structure is used as one of function parameters.
900 * all the information are dedicated to a given direct node block determined
901 * by the data offset in a file.
902 */
903 struct dnode_of_data {
904 struct inode *inode; /* vfs inode pointer */
905 struct page *inode_page; /* its inode page, NULL is possible */
906 struct page *node_page; /* cached direct node page */
907 nid_t nid; /* node id of the direct node block */
908 unsigned int ofs_in_node; /* data offset in the node page */
909 bool inode_page_locked; /* inode page is locked or not */
910 bool node_changed; /* is node block changed */
911 char cur_level; /* level of hole node page */
912 char max_level; /* level of current page located */
913 block_t data_blkaddr; /* block address of the node block */
914 };
915
916 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
917 struct page *ipage, struct page *npage, nid_t nid)
918 {
919 memset(dn, 0, sizeof(*dn));
920 dn->inode = inode;
921 dn->inode_page = ipage;
922 dn->node_page = npage;
923 dn->nid = nid;
924 }
925
926 /*
927 * For SIT manager
928 *
929 * By default, there are 6 active log areas across the whole main area.
930 * When considering hot and cold data separation to reduce cleaning overhead,
931 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
932 * respectively.
933 * In the current design, you should not change the numbers intentionally.
934 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
935 * logs individually according to the underlying devices. (default: 6)
936 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
937 * data and 8 for node logs.
938 */
939 #define NR_CURSEG_DATA_TYPE (3)
940 #define NR_CURSEG_NODE_TYPE (3)
941 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
942
943 enum {
944 CURSEG_HOT_DATA = 0, /* directory entry blocks */
945 CURSEG_WARM_DATA, /* data blocks */
946 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
947 CURSEG_HOT_NODE, /* direct node blocks of directory files */
948 CURSEG_WARM_NODE, /* direct node blocks of normal files */
949 CURSEG_COLD_NODE, /* indirect node blocks */
950 NO_CHECK_TYPE,
951 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */
952 };
953
954 struct flush_cmd {
955 struct completion wait;
956 struct llist_node llnode;
957 nid_t ino;
958 int ret;
959 };
960
961 struct flush_cmd_control {
962 struct task_struct *f2fs_issue_flush; /* flush thread */
963 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
964 atomic_t issued_flush; /* # of issued flushes */
965 atomic_t queued_flush; /* # of queued flushes */
966 struct llist_head issue_list; /* list for command issue */
967 struct llist_node *dispatch_list; /* list for command dispatch */
968 };
969
970 struct f2fs_sm_info {
971 struct sit_info *sit_info; /* whole segment information */
972 struct free_segmap_info *free_info; /* free segment information */
973 struct dirty_seglist_info *dirty_info; /* dirty segment information */
974 struct curseg_info *curseg_array; /* active segment information */
975
976 struct rw_semaphore curseg_lock; /* for preventing curseg change */
977
978 block_t seg0_blkaddr; /* block address of 0'th segment */
979 block_t main_blkaddr; /* start block address of main area */
980 block_t ssa_blkaddr; /* start block address of SSA area */
981
982 unsigned int segment_count; /* total # of segments */
983 unsigned int main_segments; /* # of segments in main area */
984 unsigned int reserved_segments; /* # of reserved segments */
985 unsigned int ovp_segments; /* # of overprovision segments */
986
987 /* a threshold to reclaim prefree segments */
988 unsigned int rec_prefree_segments;
989
990 /* for batched trimming */
991 unsigned int trim_sections; /* # of sections to trim */
992
993 struct list_head sit_entry_set; /* sit entry set list */
994
995 unsigned int ipu_policy; /* in-place-update policy */
996 unsigned int min_ipu_util; /* in-place-update threshold */
997 unsigned int min_fsync_blocks; /* threshold for fsync */
998 unsigned int min_seq_blocks; /* threshold for sequential blocks */
999 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1000 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1001
1002 /* for flush command control */
1003 struct flush_cmd_control *fcc_info;
1004
1005 /* for discard command control */
1006 struct discard_cmd_control *dcc_info;
1007 };
1008
1009 /*
1010 * For superblock
1011 */
1012 /*
1013 * COUNT_TYPE for monitoring
1014 *
1015 * f2fs monitors the number of several block types such as on-writeback,
1016 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1017 */
1018 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1019 enum count_type {
1020 F2FS_DIRTY_DENTS,
1021 F2FS_DIRTY_DATA,
1022 F2FS_DIRTY_QDATA,
1023 F2FS_DIRTY_NODES,
1024 F2FS_DIRTY_META,
1025 F2FS_INMEM_PAGES,
1026 F2FS_DIRTY_IMETA,
1027 F2FS_WB_CP_DATA,
1028 F2FS_WB_DATA,
1029 F2FS_RD_DATA,
1030 F2FS_RD_NODE,
1031 F2FS_RD_META,
1032 F2FS_DIO_WRITE,
1033 F2FS_DIO_READ,
1034 NR_COUNT_TYPE,
1035 };
1036
1037 /*
1038 * The below are the page types of bios used in submit_bio().
1039 * The available types are:
1040 * DATA User data pages. It operates as async mode.
1041 * NODE Node pages. It operates as async mode.
1042 * META FS metadata pages such as SIT, NAT, CP.
1043 * NR_PAGE_TYPE The number of page types.
1044 * META_FLUSH Make sure the previous pages are written
1045 * with waiting the bio's completion
1046 * ... Only can be used with META.
1047 */
1048 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1049 enum page_type {
1050 DATA,
1051 NODE,
1052 META,
1053 NR_PAGE_TYPE,
1054 META_FLUSH,
1055 INMEM, /* the below types are used by tracepoints only. */
1056 INMEM_DROP,
1057 INMEM_INVALIDATE,
1058 INMEM_REVOKE,
1059 IPU,
1060 OPU,
1061 };
1062
1063 enum temp_type {
1064 HOT = 0, /* must be zero for meta bio */
1065 WARM,
1066 COLD,
1067 NR_TEMP_TYPE,
1068 };
1069
1070 enum need_lock_type {
1071 LOCK_REQ = 0,
1072 LOCK_DONE,
1073 LOCK_RETRY,
1074 };
1075
1076 enum cp_reason_type {
1077 CP_NO_NEEDED,
1078 CP_NON_REGULAR,
1079 CP_COMPRESSED,
1080 CP_HARDLINK,
1081 CP_SB_NEED_CP,
1082 CP_WRONG_PINO,
1083 CP_NO_SPC_ROLL,
1084 CP_NODE_NEED_CP,
1085 CP_FASTBOOT_MODE,
1086 CP_SPEC_LOG_NUM,
1087 CP_RECOVER_DIR,
1088 };
1089
1090 enum iostat_type {
1091 /* WRITE IO */
1092 APP_DIRECT_IO, /* app direct write IOs */
1093 APP_BUFFERED_IO, /* app buffered write IOs */
1094 APP_WRITE_IO, /* app write IOs */
1095 APP_MAPPED_IO, /* app mapped IOs */
1096 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1097 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1098 FS_META_IO, /* meta IOs from kworker/reclaimer */
1099 FS_GC_DATA_IO, /* data IOs from forground gc */
1100 FS_GC_NODE_IO, /* node IOs from forground gc */
1101 FS_CP_DATA_IO, /* data IOs from checkpoint */
1102 FS_CP_NODE_IO, /* node IOs from checkpoint */
1103 FS_CP_META_IO, /* meta IOs from checkpoint */
1104
1105 /* READ IO */
1106 APP_DIRECT_READ_IO, /* app direct read IOs */
1107 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1108 APP_READ_IO, /* app read IOs */
1109 APP_MAPPED_READ_IO, /* app mapped read IOs */
1110 FS_DATA_READ_IO, /* data read IOs */
1111 FS_NODE_READ_IO, /* node read IOs */
1112 FS_META_READ_IO, /* meta read IOs */
1113
1114 /* other */
1115 FS_DISCARD, /* discard */
1116 NR_IO_TYPE,
1117 };
1118
1119 struct f2fs_io_info {
1120 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1121 nid_t ino; /* inode number */
1122 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1123 enum temp_type temp; /* contains HOT/WARM/COLD */
1124 int op; /* contains REQ_OP_ */
1125 int op_flags; /* req_flag_bits */
1126 block_t new_blkaddr; /* new block address to be written */
1127 block_t old_blkaddr; /* old block address before Cow */
1128 struct page *page; /* page to be written */
1129 struct page *encrypted_page; /* encrypted page */
1130 struct page *compressed_page; /* compressed page */
1131 struct list_head list; /* serialize IOs */
1132 bool submitted; /* indicate IO submission */
1133 int need_lock; /* indicate we need to lock cp_rwsem */
1134 bool in_list; /* indicate fio is in io_list */
1135 bool is_por; /* indicate IO is from recovery or not */
1136 bool retry; /* need to reallocate block address */
1137 int compr_blocks; /* # of compressed block addresses */
1138 bool encrypted; /* indicate file is encrypted */
1139 enum iostat_type io_type; /* io type */
1140 struct writeback_control *io_wbc; /* writeback control */
1141 struct bio **bio; /* bio for ipu */
1142 sector_t *last_block; /* last block number in bio */
1143 unsigned char version; /* version of the node */
1144 };
1145
1146 struct bio_entry {
1147 struct bio *bio;
1148 struct list_head list;
1149 };
1150
1151 #define is_read_io(rw) ((rw) == READ)
1152 struct f2fs_bio_info {
1153 struct f2fs_sb_info *sbi; /* f2fs superblock */
1154 struct bio *bio; /* bios to merge */
1155 sector_t last_block_in_bio; /* last block number */
1156 struct f2fs_io_info fio; /* store buffered io info. */
1157 struct rw_semaphore io_rwsem; /* blocking op for bio */
1158 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1159 struct list_head io_list; /* track fios */
1160 struct list_head bio_list; /* bio entry list head */
1161 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1162 };
1163
1164 #define FDEV(i) (sbi->devs[i])
1165 #define RDEV(i) (raw_super->devs[i])
1166 struct f2fs_dev_info {
1167 struct block_device *bdev;
1168 char path[MAX_PATH_LEN];
1169 unsigned int total_segments;
1170 block_t start_blk;
1171 block_t end_blk;
1172 #ifdef CONFIG_BLK_DEV_ZONED
1173 unsigned int nr_blkz; /* Total number of zones */
1174 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1175 #endif
1176 };
1177
1178 enum inode_type {
1179 DIR_INODE, /* for dirty dir inode */
1180 FILE_INODE, /* for dirty regular/symlink inode */
1181 DIRTY_META, /* for all dirtied inode metadata */
1182 ATOMIC_FILE, /* for all atomic files */
1183 NR_INODE_TYPE,
1184 };
1185
1186 /* for inner inode cache management */
1187 struct inode_management {
1188 struct radix_tree_root ino_root; /* ino entry array */
1189 spinlock_t ino_lock; /* for ino entry lock */
1190 struct list_head ino_list; /* inode list head */
1191 unsigned long ino_num; /* number of entries */
1192 };
1193
1194 /* For s_flag in struct f2fs_sb_info */
1195 enum {
1196 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1197 SBI_IS_CLOSE, /* specify unmounting */
1198 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1199 SBI_POR_DOING, /* recovery is doing or not */
1200 SBI_NEED_SB_WRITE, /* need to recover superblock */
1201 SBI_NEED_CP, /* need to checkpoint */
1202 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1203 SBI_IS_RECOVERED, /* recovered orphan/data */
1204 SBI_CP_DISABLED, /* CP was disabled last mount */
1205 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1206 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1207 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1208 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1209 SBI_IS_RESIZEFS, /* resizefs is in process */
1210 };
1211
1212 enum {
1213 CP_TIME,
1214 REQ_TIME,
1215 DISCARD_TIME,
1216 GC_TIME,
1217 DISABLE_TIME,
1218 UMOUNT_DISCARD_TIMEOUT,
1219 MAX_TIME,
1220 };
1221
1222 enum {
1223 GC_NORMAL,
1224 GC_IDLE_CB,
1225 GC_IDLE_GREEDY,
1226 GC_URGENT,
1227 };
1228
1229 enum {
1230 BGGC_MODE_ON, /* background gc is on */
1231 BGGC_MODE_OFF, /* background gc is off */
1232 BGGC_MODE_SYNC, /*
1233 * background gc is on, migrating blocks
1234 * like foreground gc
1235 */
1236 };
1237
1238 enum {
1239 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1240 FS_MODE_LFS, /* use lfs allocation only */
1241 };
1242
1243 enum {
1244 WHINT_MODE_OFF, /* not pass down write hints */
1245 WHINT_MODE_USER, /* try to pass down hints given by users */
1246 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1247 };
1248
1249 enum {
1250 ALLOC_MODE_DEFAULT, /* stay default */
1251 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1252 };
1253
1254 enum fsync_mode {
1255 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1256 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1257 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1258 };
1259
1260 /*
1261 * this value is set in page as a private data which indicate that
1262 * the page is atomically written, and it is in inmem_pages list.
1263 */
1264 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1265 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1266
1267 #define IS_ATOMIC_WRITTEN_PAGE(page) \
1268 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE)
1269 #define IS_DUMMY_WRITTEN_PAGE(page) \
1270 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE)
1271
1272 #ifdef CONFIG_FS_ENCRYPTION
1273 #define DUMMY_ENCRYPTION_ENABLED(sbi) \
1274 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1275 #else
1276 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1277 #endif
1278
1279 /* For compression */
1280 enum compress_algorithm_type {
1281 COMPRESS_LZO,
1282 COMPRESS_LZ4,
1283 COMPRESS_ZSTD,
1284 COMPRESS_MAX,
1285 };
1286
1287 #define COMPRESS_DATA_RESERVED_SIZE 5
1288 struct compress_data {
1289 __le32 clen; /* compressed data size */
1290 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1291 u8 cdata[]; /* compressed data */
1292 };
1293
1294 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1295
1296 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1297
1298 /* compress context */
1299 struct compress_ctx {
1300 struct inode *inode; /* inode the context belong to */
1301 pgoff_t cluster_idx; /* cluster index number */
1302 unsigned int cluster_size; /* page count in cluster */
1303 unsigned int log_cluster_size; /* log of cluster size */
1304 struct page **rpages; /* pages store raw data in cluster */
1305 unsigned int nr_rpages; /* total page number in rpages */
1306 struct page **cpages; /* pages store compressed data in cluster */
1307 unsigned int nr_cpages; /* total page number in cpages */
1308 void *rbuf; /* virtual mapped address on rpages */
1309 struct compress_data *cbuf; /* virtual mapped address on cpages */
1310 size_t rlen; /* valid data length in rbuf */
1311 size_t clen; /* valid data length in cbuf */
1312 void *private; /* payload buffer for specified compression algorithm */
1313 void *private2; /* extra payload buffer */
1314 };
1315
1316 /* compress context for write IO path */
1317 struct compress_io_ctx {
1318 u32 magic; /* magic number to indicate page is compressed */
1319 struct inode *inode; /* inode the context belong to */
1320 struct page **rpages; /* pages store raw data in cluster */
1321 unsigned int nr_rpages; /* total page number in rpages */
1322 refcount_t ref; /* referrence count of raw page */
1323 };
1324
1325 /* decompress io context for read IO path */
1326 struct decompress_io_ctx {
1327 u32 magic; /* magic number to indicate page is compressed */
1328 struct inode *inode; /* inode the context belong to */
1329 pgoff_t cluster_idx; /* cluster index number */
1330 unsigned int cluster_size; /* page count in cluster */
1331 unsigned int log_cluster_size; /* log of cluster size */
1332 struct page **rpages; /* pages store raw data in cluster */
1333 unsigned int nr_rpages; /* total page number in rpages */
1334 struct page **cpages; /* pages store compressed data in cluster */
1335 unsigned int nr_cpages; /* total page number in cpages */
1336 struct page **tpages; /* temp pages to pad holes in cluster */
1337 void *rbuf; /* virtual mapped address on rpages */
1338 struct compress_data *cbuf; /* virtual mapped address on cpages */
1339 size_t rlen; /* valid data length in rbuf */
1340 size_t clen; /* valid data length in cbuf */
1341 refcount_t ref; /* referrence count of compressed page */
1342 bool failed; /* indicate IO error during decompression */
1343 void *private; /* payload buffer for specified decompression algorithm */
1344 void *private2; /* extra payload buffer */
1345 };
1346
1347 #define NULL_CLUSTER ((unsigned int)(~0))
1348 #define MIN_COMPRESS_LOG_SIZE 2
1349 #define MAX_COMPRESS_LOG_SIZE 8
1350 #define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE)
1351
1352 struct f2fs_sb_info {
1353 struct super_block *sb; /* pointer to VFS super block */
1354 struct proc_dir_entry *s_proc; /* proc entry */
1355 struct f2fs_super_block *raw_super; /* raw super block pointer */
1356 struct rw_semaphore sb_lock; /* lock for raw super block */
1357 int valid_super_block; /* valid super block no */
1358 unsigned long s_flag; /* flags for sbi */
1359 struct mutex writepages; /* mutex for writepages() */
1360 #ifdef CONFIG_UNICODE
1361 struct unicode_map *s_encoding;
1362 __u16 s_encoding_flags;
1363 #endif
1364
1365 #ifdef CONFIG_BLK_DEV_ZONED
1366 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1367 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1368 #endif
1369
1370 /* for node-related operations */
1371 struct f2fs_nm_info *nm_info; /* node manager */
1372 struct inode *node_inode; /* cache node blocks */
1373
1374 /* for segment-related operations */
1375 struct f2fs_sm_info *sm_info; /* segment manager */
1376
1377 /* for bio operations */
1378 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1379 /* keep migration IO order for LFS mode */
1380 struct rw_semaphore io_order_lock;
1381 mempool_t *write_io_dummy; /* Dummy pages */
1382
1383 /* for checkpoint */
1384 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1385 int cur_cp_pack; /* remain current cp pack */
1386 spinlock_t cp_lock; /* for flag in ckpt */
1387 struct inode *meta_inode; /* cache meta blocks */
1388 struct mutex cp_mutex; /* checkpoint procedure lock */
1389 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1390 struct rw_semaphore node_write; /* locking node writes */
1391 struct rw_semaphore node_change; /* locking node change */
1392 wait_queue_head_t cp_wait;
1393 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1394 long interval_time[MAX_TIME]; /* to store thresholds */
1395
1396 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1397
1398 spinlock_t fsync_node_lock; /* for node entry lock */
1399 struct list_head fsync_node_list; /* node list head */
1400 unsigned int fsync_seg_id; /* sequence id */
1401 unsigned int fsync_node_num; /* number of node entries */
1402
1403 /* for orphan inode, use 0'th array */
1404 unsigned int max_orphans; /* max orphan inodes */
1405
1406 /* for inode management */
1407 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1408 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1409 struct mutex flush_lock; /* for flush exclusion */
1410
1411 /* for extent tree cache */
1412 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1413 struct mutex extent_tree_lock; /* locking extent radix tree */
1414 struct list_head extent_list; /* lru list for shrinker */
1415 spinlock_t extent_lock; /* locking extent lru list */
1416 atomic_t total_ext_tree; /* extent tree count */
1417 struct list_head zombie_list; /* extent zombie tree list */
1418 atomic_t total_zombie_tree; /* extent zombie tree count */
1419 atomic_t total_ext_node; /* extent info count */
1420
1421 /* basic filesystem units */
1422 unsigned int log_sectors_per_block; /* log2 sectors per block */
1423 unsigned int log_blocksize; /* log2 block size */
1424 unsigned int blocksize; /* block size */
1425 unsigned int root_ino_num; /* root inode number*/
1426 unsigned int node_ino_num; /* node inode number*/
1427 unsigned int meta_ino_num; /* meta inode number*/
1428 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1429 unsigned int blocks_per_seg; /* blocks per segment */
1430 unsigned int segs_per_sec; /* segments per section */
1431 unsigned int secs_per_zone; /* sections per zone */
1432 unsigned int total_sections; /* total section count */
1433 struct mutex resize_mutex; /* for resize exclusion */
1434 unsigned int total_node_count; /* total node block count */
1435 unsigned int total_valid_node_count; /* valid node block count */
1436 loff_t max_file_blocks; /* max block index of file */
1437 int dir_level; /* directory level */
1438 int readdir_ra; /* readahead inode in readdir */
1439
1440 block_t user_block_count; /* # of user blocks */
1441 block_t total_valid_block_count; /* # of valid blocks */
1442 block_t discard_blks; /* discard command candidats */
1443 block_t last_valid_block_count; /* for recovery */
1444 block_t reserved_blocks; /* configurable reserved blocks */
1445 block_t current_reserved_blocks; /* current reserved blocks */
1446
1447 /* Additional tracking for no checkpoint mode */
1448 block_t unusable_block_count; /* # of blocks saved by last cp */
1449
1450 unsigned int nquota_files; /* # of quota sysfile */
1451 struct rw_semaphore quota_sem; /* blocking cp for flags */
1452
1453 /* # of pages, see count_type */
1454 atomic_t nr_pages[NR_COUNT_TYPE];
1455 /* # of allocated blocks */
1456 struct percpu_counter alloc_valid_block_count;
1457
1458 /* writeback control */
1459 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1460
1461 /* valid inode count */
1462 struct percpu_counter total_valid_inode_count;
1463
1464 struct f2fs_mount_info mount_opt; /* mount options */
1465
1466 /* for cleaning operations */
1467 struct rw_semaphore gc_lock; /*
1468 * semaphore for GC, avoid
1469 * race between GC and GC or CP
1470 */
1471 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1472 unsigned int cur_victim_sec; /* current victim section num */
1473 unsigned int gc_mode; /* current GC state */
1474 unsigned int next_victim_seg[2]; /* next segment in victim section */
1475 /* for skip statistic */
1476 unsigned int atomic_files; /* # of opened atomic file */
1477 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1478 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1479
1480 /* threshold for gc trials on pinned files */
1481 u64 gc_pin_file_threshold;
1482 struct rw_semaphore pin_sem;
1483
1484 /* maximum # of trials to find a victim segment for SSR and GC */
1485 unsigned int max_victim_search;
1486 /* migration granularity of garbage collection, unit: segment */
1487 unsigned int migration_granularity;
1488
1489 /*
1490 * for stat information.
1491 * one is for the LFS mode, and the other is for the SSR mode.
1492 */
1493 #ifdef CONFIG_F2FS_STAT_FS
1494 struct f2fs_stat_info *stat_info; /* FS status information */
1495 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1496 unsigned int segment_count[2]; /* # of allocated segments */
1497 unsigned int block_count[2]; /* # of allocated blocks */
1498 atomic_t inplace_count; /* # of inplace update */
1499 atomic64_t total_hit_ext; /* # of lookup extent cache */
1500 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1501 atomic64_t read_hit_largest; /* # of hit largest extent node */
1502 atomic64_t read_hit_cached; /* # of hit cached extent node */
1503 atomic_t inline_xattr; /* # of inline_xattr inodes */
1504 atomic_t inline_inode; /* # of inline_data inodes */
1505 atomic_t inline_dir; /* # of inline_dentry inodes */
1506 atomic_t compr_inode; /* # of compressed inodes */
1507 atomic_t compr_blocks; /* # of compressed blocks */
1508 atomic_t vw_cnt; /* # of volatile writes */
1509 atomic_t max_aw_cnt; /* max # of atomic writes */
1510 atomic_t max_vw_cnt; /* max # of volatile writes */
1511 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1512 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1513 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1514 #endif
1515 spinlock_t stat_lock; /* lock for stat operations */
1516
1517 /* For app/fs IO statistics */
1518 spinlock_t iostat_lock;
1519 unsigned long long rw_iostat[NR_IO_TYPE];
1520 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1521 bool iostat_enable;
1522 unsigned long iostat_next_period;
1523 unsigned int iostat_period_ms;
1524
1525 /* to attach REQ_META|REQ_FUA flags */
1526 unsigned int data_io_flag;
1527
1528 /* For sysfs suppport */
1529 struct kobject s_kobj;
1530 struct completion s_kobj_unregister;
1531
1532 /* For shrinker support */
1533 struct list_head s_list;
1534 int s_ndevs; /* number of devices */
1535 struct f2fs_dev_info *devs; /* for device list */
1536 unsigned int dirty_device; /* for checkpoint data flush */
1537 spinlock_t dev_lock; /* protect dirty_device */
1538 struct mutex umount_mutex;
1539 unsigned int shrinker_run_no;
1540
1541 /* For write statistics */
1542 u64 sectors_written_start;
1543 u64 kbytes_written;
1544
1545 /* Reference to checksum algorithm driver via cryptoapi */
1546 struct crypto_shash *s_chksum_driver;
1547
1548 /* Precomputed FS UUID checksum for seeding other checksums */
1549 __u32 s_chksum_seed;
1550
1551 struct workqueue_struct *post_read_wq; /* post read workqueue */
1552
1553 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1554 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1555 };
1556
1557 struct f2fs_private_dio {
1558 struct inode *inode;
1559 void *orig_private;
1560 bio_end_io_t *orig_end_io;
1561 bool write;
1562 };
1563
1564 #ifdef CONFIG_F2FS_FAULT_INJECTION
1565 #define f2fs_show_injection_info(sbi, type) \
1566 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1567 KERN_INFO, sbi->sb->s_id, \
1568 f2fs_fault_name[type], \
1569 __func__, __builtin_return_address(0))
1570 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1571 {
1572 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1573
1574 if (!ffi->inject_rate)
1575 return false;
1576
1577 if (!IS_FAULT_SET(ffi, type))
1578 return false;
1579
1580 atomic_inc(&ffi->inject_ops);
1581 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1582 atomic_set(&ffi->inject_ops, 0);
1583 return true;
1584 }
1585 return false;
1586 }
1587 #else
1588 #define f2fs_show_injection_info(sbi, type) do { } while (0)
1589 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1590 {
1591 return false;
1592 }
1593 #endif
1594
1595 /*
1596 * Test if the mounted volume is a multi-device volume.
1597 * - For a single regular disk volume, sbi->s_ndevs is 0.
1598 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1599 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1600 */
1601 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1602 {
1603 return sbi->s_ndevs > 1;
1604 }
1605
1606 /* For write statistics. Suppose sector size is 512 bytes,
1607 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1608 */
1609 #define BD_PART_WRITTEN(s) \
1610 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1611 (s)->sectors_written_start) >> 1)
1612
1613 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1614 {
1615 unsigned long now = jiffies;
1616
1617 sbi->last_time[type] = now;
1618
1619 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1620 if (type == REQ_TIME) {
1621 sbi->last_time[DISCARD_TIME] = now;
1622 sbi->last_time[GC_TIME] = now;
1623 }
1624 }
1625
1626 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1627 {
1628 unsigned long interval = sbi->interval_time[type] * HZ;
1629
1630 return time_after(jiffies, sbi->last_time[type] + interval);
1631 }
1632
1633 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1634 int type)
1635 {
1636 unsigned long interval = sbi->interval_time[type] * HZ;
1637 unsigned int wait_ms = 0;
1638 long delta;
1639
1640 delta = (sbi->last_time[type] + interval) - jiffies;
1641 if (delta > 0)
1642 wait_ms = jiffies_to_msecs(delta);
1643
1644 return wait_ms;
1645 }
1646
1647 /*
1648 * Inline functions
1649 */
1650 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1651 const void *address, unsigned int length)
1652 {
1653 struct {
1654 struct shash_desc shash;
1655 char ctx[4];
1656 } desc;
1657 int err;
1658
1659 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1660
1661 desc.shash.tfm = sbi->s_chksum_driver;
1662 *(u32 *)desc.ctx = crc;
1663
1664 err = crypto_shash_update(&desc.shash, address, length);
1665 BUG_ON(err);
1666
1667 return *(u32 *)desc.ctx;
1668 }
1669
1670 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1671 unsigned int length)
1672 {
1673 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1674 }
1675
1676 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1677 void *buf, size_t buf_size)
1678 {
1679 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1680 }
1681
1682 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1683 const void *address, unsigned int length)
1684 {
1685 return __f2fs_crc32(sbi, crc, address, length);
1686 }
1687
1688 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1689 {
1690 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1691 }
1692
1693 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1694 {
1695 return sb->s_fs_info;
1696 }
1697
1698 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1699 {
1700 return F2FS_SB(inode->i_sb);
1701 }
1702
1703 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1704 {
1705 return F2FS_I_SB(mapping->host);
1706 }
1707
1708 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1709 {
1710 return F2FS_M_SB(page_file_mapping(page));
1711 }
1712
1713 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1714 {
1715 return (struct f2fs_super_block *)(sbi->raw_super);
1716 }
1717
1718 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1719 {
1720 return (struct f2fs_checkpoint *)(sbi->ckpt);
1721 }
1722
1723 static inline struct f2fs_node *F2FS_NODE(struct page *page)
1724 {
1725 return (struct f2fs_node *)page_address(page);
1726 }
1727
1728 static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1729 {
1730 return &((struct f2fs_node *)page_address(page))->i;
1731 }
1732
1733 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1734 {
1735 return (struct f2fs_nm_info *)(sbi->nm_info);
1736 }
1737
1738 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1739 {
1740 return (struct f2fs_sm_info *)(sbi->sm_info);
1741 }
1742
1743 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1744 {
1745 return (struct sit_info *)(SM_I(sbi)->sit_info);
1746 }
1747
1748 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1749 {
1750 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1751 }
1752
1753 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1754 {
1755 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1756 }
1757
1758 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1759 {
1760 return sbi->meta_inode->i_mapping;
1761 }
1762
1763 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1764 {
1765 return sbi->node_inode->i_mapping;
1766 }
1767
1768 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1769 {
1770 return test_bit(type, &sbi->s_flag);
1771 }
1772
1773 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1774 {
1775 set_bit(type, &sbi->s_flag);
1776 }
1777
1778 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1779 {
1780 clear_bit(type, &sbi->s_flag);
1781 }
1782
1783 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1784 {
1785 return le64_to_cpu(cp->checkpoint_ver);
1786 }
1787
1788 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1789 {
1790 if (type < F2FS_MAX_QUOTAS)
1791 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1792 return 0;
1793 }
1794
1795 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1796 {
1797 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1798 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1799 }
1800
1801 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1802 {
1803 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1804
1805 return ckpt_flags & f;
1806 }
1807
1808 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1809 {
1810 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1811 }
1812
1813 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1814 {
1815 unsigned int ckpt_flags;
1816
1817 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1818 ckpt_flags |= f;
1819 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1820 }
1821
1822 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1823 {
1824 unsigned long flags;
1825
1826 spin_lock_irqsave(&sbi->cp_lock, flags);
1827 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1828 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1829 }
1830
1831 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1832 {
1833 unsigned int ckpt_flags;
1834
1835 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1836 ckpt_flags &= (~f);
1837 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1838 }
1839
1840 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1841 {
1842 unsigned long flags;
1843
1844 spin_lock_irqsave(&sbi->cp_lock, flags);
1845 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1846 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1847 }
1848
1849 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1850 {
1851 unsigned long flags;
1852 unsigned char *nat_bits;
1853
1854 /*
1855 * In order to re-enable nat_bits we need to call fsck.f2fs by
1856 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1857 * so let's rely on regular fsck or unclean shutdown.
1858 */
1859
1860 if (lock)
1861 spin_lock_irqsave(&sbi->cp_lock, flags);
1862 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1863 nat_bits = NM_I(sbi)->nat_bits;
1864 NM_I(sbi)->nat_bits = NULL;
1865 if (lock)
1866 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1867
1868 kvfree(nat_bits);
1869 }
1870
1871 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1872 struct cp_control *cpc)
1873 {
1874 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1875
1876 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1877 }
1878
1879 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1880 {
1881 down_read(&sbi->cp_rwsem);
1882 }
1883
1884 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1885 {
1886 return down_read_trylock(&sbi->cp_rwsem);
1887 }
1888
1889 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1890 {
1891 up_read(&sbi->cp_rwsem);
1892 }
1893
1894 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1895 {
1896 down_write(&sbi->cp_rwsem);
1897 }
1898
1899 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1900 {
1901 up_write(&sbi->cp_rwsem);
1902 }
1903
1904 static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1905 {
1906 int reason = CP_SYNC;
1907
1908 if (test_opt(sbi, FASTBOOT))
1909 reason = CP_FASTBOOT;
1910 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1911 reason = CP_UMOUNT;
1912 return reason;
1913 }
1914
1915 static inline bool __remain_node_summaries(int reason)
1916 {
1917 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1918 }
1919
1920 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1921 {
1922 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1923 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1924 }
1925
1926 /*
1927 * Check whether the inode has blocks or not
1928 */
1929 static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1930 {
1931 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1932
1933 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1934 }
1935
1936 static inline bool f2fs_has_xattr_block(unsigned int ofs)
1937 {
1938 return ofs == XATTR_NODE_OFFSET;
1939 }
1940
1941 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1942 struct inode *inode, bool cap)
1943 {
1944 if (!inode)
1945 return true;
1946 if (!test_opt(sbi, RESERVE_ROOT))
1947 return false;
1948 if (IS_NOQUOTA(inode))
1949 return true;
1950 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1951 return true;
1952 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1953 in_group_p(F2FS_OPTION(sbi).s_resgid))
1954 return true;
1955 if (cap && capable(CAP_SYS_RESOURCE))
1956 return true;
1957 return false;
1958 }
1959
1960 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
1961 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
1962 struct inode *inode, blkcnt_t *count)
1963 {
1964 blkcnt_t diff = 0, release = 0;
1965 block_t avail_user_block_count;
1966 int ret;
1967
1968 ret = dquot_reserve_block(inode, *count);
1969 if (ret)
1970 return ret;
1971
1972 if (time_to_inject(sbi, FAULT_BLOCK)) {
1973 f2fs_show_injection_info(sbi, FAULT_BLOCK);
1974 release = *count;
1975 goto release_quota;
1976 }
1977
1978 /*
1979 * let's increase this in prior to actual block count change in order
1980 * for f2fs_sync_file to avoid data races when deciding checkpoint.
1981 */
1982 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
1983
1984 spin_lock(&sbi->stat_lock);
1985 sbi->total_valid_block_count += (block_t)(*count);
1986 avail_user_block_count = sbi->user_block_count -
1987 sbi->current_reserved_blocks;
1988
1989 if (!__allow_reserved_blocks(sbi, inode, true))
1990 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
1991 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1992 if (avail_user_block_count > sbi->unusable_block_count)
1993 avail_user_block_count -= sbi->unusable_block_count;
1994 else
1995 avail_user_block_count = 0;
1996 }
1997 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
1998 diff = sbi->total_valid_block_count - avail_user_block_count;
1999 if (diff > *count)
2000 diff = *count;
2001 *count -= diff;
2002 release = diff;
2003 sbi->total_valid_block_count -= diff;
2004 if (!*count) {
2005 spin_unlock(&sbi->stat_lock);
2006 goto enospc;
2007 }
2008 }
2009 spin_unlock(&sbi->stat_lock);
2010
2011 if (unlikely(release)) {
2012 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2013 dquot_release_reservation_block(inode, release);
2014 }
2015 f2fs_i_blocks_write(inode, *count, true, true);
2016 return 0;
2017
2018 enospc:
2019 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2020 release_quota:
2021 dquot_release_reservation_block(inode, release);
2022 return -ENOSPC;
2023 }
2024
2025 __printf(2, 3)
2026 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2027
2028 #define f2fs_err(sbi, fmt, ...) \
2029 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2030 #define f2fs_warn(sbi, fmt, ...) \
2031 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2032 #define f2fs_notice(sbi, fmt, ...) \
2033 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2034 #define f2fs_info(sbi, fmt, ...) \
2035 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2036 #define f2fs_debug(sbi, fmt, ...) \
2037 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2038
2039 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2040 struct inode *inode,
2041 block_t count)
2042 {
2043 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2044
2045 spin_lock(&sbi->stat_lock);
2046 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2047 sbi->total_valid_block_count -= (block_t)count;
2048 if (sbi->reserved_blocks &&
2049 sbi->current_reserved_blocks < sbi->reserved_blocks)
2050 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2051 sbi->current_reserved_blocks + count);
2052 spin_unlock(&sbi->stat_lock);
2053 if (unlikely(inode->i_blocks < sectors)) {
2054 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2055 inode->i_ino,
2056 (unsigned long long)inode->i_blocks,
2057 (unsigned long long)sectors);
2058 set_sbi_flag(sbi, SBI_NEED_FSCK);
2059 return;
2060 }
2061 f2fs_i_blocks_write(inode, count, false, true);
2062 }
2063
2064 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2065 {
2066 atomic_inc(&sbi->nr_pages[count_type]);
2067
2068 if (count_type == F2FS_DIRTY_DENTS ||
2069 count_type == F2FS_DIRTY_NODES ||
2070 count_type == F2FS_DIRTY_META ||
2071 count_type == F2FS_DIRTY_QDATA ||
2072 count_type == F2FS_DIRTY_IMETA)
2073 set_sbi_flag(sbi, SBI_IS_DIRTY);
2074 }
2075
2076 static inline void inode_inc_dirty_pages(struct inode *inode)
2077 {
2078 atomic_inc(&F2FS_I(inode)->dirty_pages);
2079 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2080 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2081 if (IS_NOQUOTA(inode))
2082 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2083 }
2084
2085 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2086 {
2087 atomic_dec(&sbi->nr_pages[count_type]);
2088 }
2089
2090 static inline void inode_dec_dirty_pages(struct inode *inode)
2091 {
2092 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2093 !S_ISLNK(inode->i_mode))
2094 return;
2095
2096 atomic_dec(&F2FS_I(inode)->dirty_pages);
2097 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2098 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2099 if (IS_NOQUOTA(inode))
2100 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2101 }
2102
2103 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2104 {
2105 return atomic_read(&sbi->nr_pages[count_type]);
2106 }
2107
2108 static inline int get_dirty_pages(struct inode *inode)
2109 {
2110 return atomic_read(&F2FS_I(inode)->dirty_pages);
2111 }
2112
2113 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2114 {
2115 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2116 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2117 sbi->log_blocks_per_seg;
2118
2119 return segs / sbi->segs_per_sec;
2120 }
2121
2122 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2123 {
2124 return sbi->total_valid_block_count;
2125 }
2126
2127 static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2128 {
2129 return sbi->discard_blks;
2130 }
2131
2132 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2133 {
2134 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2135
2136 /* return NAT or SIT bitmap */
2137 if (flag == NAT_BITMAP)
2138 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2139 else if (flag == SIT_BITMAP)
2140 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2141
2142 return 0;
2143 }
2144
2145 static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2146 {
2147 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2148 }
2149
2150 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2151 {
2152 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2153 int offset;
2154
2155 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2156 offset = (flag == SIT_BITMAP) ?
2157 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2158 /*
2159 * if large_nat_bitmap feature is enabled, leave checksum
2160 * protection for all nat/sit bitmaps.
2161 */
2162 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2163 }
2164
2165 if (__cp_payload(sbi) > 0) {
2166 if (flag == NAT_BITMAP)
2167 return &ckpt->sit_nat_version_bitmap;
2168 else
2169 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2170 } else {
2171 offset = (flag == NAT_BITMAP) ?
2172 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2173 return &ckpt->sit_nat_version_bitmap + offset;
2174 }
2175 }
2176
2177 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2178 {
2179 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2180
2181 if (sbi->cur_cp_pack == 2)
2182 start_addr += sbi->blocks_per_seg;
2183 return start_addr;
2184 }
2185
2186 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2187 {
2188 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2189
2190 if (sbi->cur_cp_pack == 1)
2191 start_addr += sbi->blocks_per_seg;
2192 return start_addr;
2193 }
2194
2195 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2196 {
2197 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2198 }
2199
2200 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2201 {
2202 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2203 }
2204
2205 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2206 struct inode *inode, bool is_inode)
2207 {
2208 block_t valid_block_count;
2209 unsigned int valid_node_count, user_block_count;
2210 int err;
2211
2212 if (is_inode) {
2213 if (inode) {
2214 err = dquot_alloc_inode(inode);
2215 if (err)
2216 return err;
2217 }
2218 } else {
2219 err = dquot_reserve_block(inode, 1);
2220 if (err)
2221 return err;
2222 }
2223
2224 if (time_to_inject(sbi, FAULT_BLOCK)) {
2225 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2226 goto enospc;
2227 }
2228
2229 spin_lock(&sbi->stat_lock);
2230
2231 valid_block_count = sbi->total_valid_block_count +
2232 sbi->current_reserved_blocks + 1;
2233
2234 if (!__allow_reserved_blocks(sbi, inode, false))
2235 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2236 user_block_count = sbi->user_block_count;
2237 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2238 user_block_count -= sbi->unusable_block_count;
2239
2240 if (unlikely(valid_block_count > user_block_count)) {
2241 spin_unlock(&sbi->stat_lock);
2242 goto enospc;
2243 }
2244
2245 valid_node_count = sbi->total_valid_node_count + 1;
2246 if (unlikely(valid_node_count > sbi->total_node_count)) {
2247 spin_unlock(&sbi->stat_lock);
2248 goto enospc;
2249 }
2250
2251 sbi->total_valid_node_count++;
2252 sbi->total_valid_block_count++;
2253 spin_unlock(&sbi->stat_lock);
2254
2255 if (inode) {
2256 if (is_inode)
2257 f2fs_mark_inode_dirty_sync(inode, true);
2258 else
2259 f2fs_i_blocks_write(inode, 1, true, true);
2260 }
2261
2262 percpu_counter_inc(&sbi->alloc_valid_block_count);
2263 return 0;
2264
2265 enospc:
2266 if (is_inode) {
2267 if (inode)
2268 dquot_free_inode(inode);
2269 } else {
2270 dquot_release_reservation_block(inode, 1);
2271 }
2272 return -ENOSPC;
2273 }
2274
2275 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2276 struct inode *inode, bool is_inode)
2277 {
2278 spin_lock(&sbi->stat_lock);
2279
2280 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2281 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2282
2283 sbi->total_valid_node_count--;
2284 sbi->total_valid_block_count--;
2285 if (sbi->reserved_blocks &&
2286 sbi->current_reserved_blocks < sbi->reserved_blocks)
2287 sbi->current_reserved_blocks++;
2288
2289 spin_unlock(&sbi->stat_lock);
2290
2291 if (is_inode) {
2292 dquot_free_inode(inode);
2293 } else {
2294 if (unlikely(inode->i_blocks == 0)) {
2295 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2296 inode->i_ino,
2297 (unsigned long long)inode->i_blocks);
2298 set_sbi_flag(sbi, SBI_NEED_FSCK);
2299 return;
2300 }
2301 f2fs_i_blocks_write(inode, 1, false, true);
2302 }
2303 }
2304
2305 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2306 {
2307 return sbi->total_valid_node_count;
2308 }
2309
2310 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2311 {
2312 percpu_counter_inc(&sbi->total_valid_inode_count);
2313 }
2314
2315 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2316 {
2317 percpu_counter_dec(&sbi->total_valid_inode_count);
2318 }
2319
2320 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2321 {
2322 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2323 }
2324
2325 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2326 pgoff_t index, bool for_write)
2327 {
2328 struct page *page;
2329
2330 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2331 if (!for_write)
2332 page = find_get_page_flags(mapping, index,
2333 FGP_LOCK | FGP_ACCESSED);
2334 else
2335 page = find_lock_page(mapping, index);
2336 if (page)
2337 return page;
2338
2339 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2340 f2fs_show_injection_info(F2FS_M_SB(mapping),
2341 FAULT_PAGE_ALLOC);
2342 return NULL;
2343 }
2344 }
2345
2346 if (!for_write)
2347 return grab_cache_page(mapping, index);
2348 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2349 }
2350
2351 static inline struct page *f2fs_pagecache_get_page(
2352 struct address_space *mapping, pgoff_t index,
2353 int fgp_flags, gfp_t gfp_mask)
2354 {
2355 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2356 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2357 return NULL;
2358 }
2359
2360 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2361 }
2362
2363 static inline void f2fs_copy_page(struct page *src, struct page *dst)
2364 {
2365 char *src_kaddr = kmap(src);
2366 char *dst_kaddr = kmap(dst);
2367
2368 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2369 kunmap(dst);
2370 kunmap(src);
2371 }
2372
2373 static inline void f2fs_put_page(struct page *page, int unlock)
2374 {
2375 if (!page)
2376 return;
2377
2378 if (unlock) {
2379 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2380 unlock_page(page);
2381 }
2382 put_page(page);
2383 }
2384
2385 static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2386 {
2387 if (dn->node_page)
2388 f2fs_put_page(dn->node_page, 1);
2389 if (dn->inode_page && dn->node_page != dn->inode_page)
2390 f2fs_put_page(dn->inode_page, 0);
2391 dn->node_page = NULL;
2392 dn->inode_page = NULL;
2393 }
2394
2395 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2396 size_t size)
2397 {
2398 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2399 }
2400
2401 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2402 gfp_t flags)
2403 {
2404 void *entry;
2405
2406 entry = kmem_cache_alloc(cachep, flags);
2407 if (!entry)
2408 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2409 return entry;
2410 }
2411
2412 static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2413 {
2414 if (sbi->gc_mode == GC_URGENT)
2415 return true;
2416
2417 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2418 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2419 get_pages(sbi, F2FS_WB_CP_DATA) ||
2420 get_pages(sbi, F2FS_DIO_READ) ||
2421 get_pages(sbi, F2FS_DIO_WRITE))
2422 return false;
2423
2424 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2425 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2426 return false;
2427
2428 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2429 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2430 return false;
2431
2432 return f2fs_time_over(sbi, type);
2433 }
2434
2435 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2436 unsigned long index, void *item)
2437 {
2438 while (radix_tree_insert(root, index, item))
2439 cond_resched();
2440 }
2441
2442 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2443
2444 static inline bool IS_INODE(struct page *page)
2445 {
2446 struct f2fs_node *p = F2FS_NODE(page);
2447
2448 return RAW_IS_INODE(p);
2449 }
2450
2451 static inline int offset_in_addr(struct f2fs_inode *i)
2452 {
2453 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2454 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2455 }
2456
2457 static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2458 {
2459 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2460 }
2461
2462 static inline int f2fs_has_extra_attr(struct inode *inode);
2463 static inline block_t data_blkaddr(struct inode *inode,
2464 struct page *node_page, unsigned int offset)
2465 {
2466 struct f2fs_node *raw_node;
2467 __le32 *addr_array;
2468 int base = 0;
2469 bool is_inode = IS_INODE(node_page);
2470
2471 raw_node = F2FS_NODE(node_page);
2472
2473 if (is_inode) {
2474 if (!inode)
2475 /* from GC path only */
2476 base = offset_in_addr(&raw_node->i);
2477 else if (f2fs_has_extra_attr(inode))
2478 base = get_extra_isize(inode);
2479 }
2480
2481 addr_array = blkaddr_in_node(raw_node);
2482 return le32_to_cpu(addr_array[base + offset]);
2483 }
2484
2485 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2486 {
2487 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2488 }
2489
2490 static inline int f2fs_test_bit(unsigned int nr, char *addr)
2491 {
2492 int mask;
2493
2494 addr += (nr >> 3);
2495 mask = 1 << (7 - (nr & 0x07));
2496 return mask & *addr;
2497 }
2498
2499 static inline void f2fs_set_bit(unsigned int nr, char *addr)
2500 {
2501 int mask;
2502
2503 addr += (nr >> 3);
2504 mask = 1 << (7 - (nr & 0x07));
2505 *addr |= mask;
2506 }
2507
2508 static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2509 {
2510 int mask;
2511
2512 addr += (nr >> 3);
2513 mask = 1 << (7 - (nr & 0x07));
2514 *addr &= ~mask;
2515 }
2516
2517 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2518 {
2519 int mask;
2520 int ret;
2521
2522 addr += (nr >> 3);
2523 mask = 1 << (7 - (nr & 0x07));
2524 ret = mask & *addr;
2525 *addr |= mask;
2526 return ret;
2527 }
2528
2529 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2530 {
2531 int mask;
2532 int ret;
2533
2534 addr += (nr >> 3);
2535 mask = 1 << (7 - (nr & 0x07));
2536 ret = mask & *addr;
2537 *addr &= ~mask;
2538 return ret;
2539 }
2540
2541 static inline void f2fs_change_bit(unsigned int nr, char *addr)
2542 {
2543 int mask;
2544
2545 addr += (nr >> 3);
2546 mask = 1 << (7 - (nr & 0x07));
2547 *addr ^= mask;
2548 }
2549
2550 /*
2551 * On-disk inode flags (f2fs_inode::i_flags)
2552 */
2553 #define F2FS_COMPR_FL 0x00000004 /* Compress file */
2554 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2555 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2556 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2557 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2558 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2559 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2560 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2561 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2562 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2563 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2564
2565 /* Flags that should be inherited by new inodes from their parent. */
2566 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2567 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2568 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2569
2570 /* Flags that are appropriate for regular files (all but dir-specific ones). */
2571 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2572 F2FS_CASEFOLD_FL))
2573
2574 /* Flags that are appropriate for non-directories/regular files. */
2575 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2576
2577 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2578 {
2579 if (S_ISDIR(mode))
2580 return flags;
2581 else if (S_ISREG(mode))
2582 return flags & F2FS_REG_FLMASK;
2583 else
2584 return flags & F2FS_OTHER_FLMASK;
2585 }
2586
2587 static inline void __mark_inode_dirty_flag(struct inode *inode,
2588 int flag, bool set)
2589 {
2590 switch (flag) {
2591 case FI_INLINE_XATTR:
2592 case FI_INLINE_DATA:
2593 case FI_INLINE_DENTRY:
2594 case FI_NEW_INODE:
2595 if (set)
2596 return;
2597 /* fall through */
2598 case FI_DATA_EXIST:
2599 case FI_INLINE_DOTS:
2600 case FI_PIN_FILE:
2601 f2fs_mark_inode_dirty_sync(inode, true);
2602 }
2603 }
2604
2605 static inline void set_inode_flag(struct inode *inode, int flag)
2606 {
2607 test_and_set_bit(flag, F2FS_I(inode)->flags);
2608 __mark_inode_dirty_flag(inode, flag, true);
2609 }
2610
2611 static inline int is_inode_flag_set(struct inode *inode, int flag)
2612 {
2613 return test_bit(flag, F2FS_I(inode)->flags);
2614 }
2615
2616 static inline void clear_inode_flag(struct inode *inode, int flag)
2617 {
2618 test_and_clear_bit(flag, F2FS_I(inode)->flags);
2619 __mark_inode_dirty_flag(inode, flag, false);
2620 }
2621
2622 static inline bool f2fs_verity_in_progress(struct inode *inode)
2623 {
2624 return IS_ENABLED(CONFIG_FS_VERITY) &&
2625 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2626 }
2627
2628 static inline void set_acl_inode(struct inode *inode, umode_t mode)
2629 {
2630 F2FS_I(inode)->i_acl_mode = mode;
2631 set_inode_flag(inode, FI_ACL_MODE);
2632 f2fs_mark_inode_dirty_sync(inode, false);
2633 }
2634
2635 static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2636 {
2637 if (inc)
2638 inc_nlink(inode);
2639 else
2640 drop_nlink(inode);
2641 f2fs_mark_inode_dirty_sync(inode, true);
2642 }
2643
2644 static inline void f2fs_i_blocks_write(struct inode *inode,
2645 block_t diff, bool add, bool claim)
2646 {
2647 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2648 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2649
2650 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2651 if (add) {
2652 if (claim)
2653 dquot_claim_block(inode, diff);
2654 else
2655 dquot_alloc_block_nofail(inode, diff);
2656 } else {
2657 dquot_free_block(inode, diff);
2658 }
2659
2660 f2fs_mark_inode_dirty_sync(inode, true);
2661 if (clean || recover)
2662 set_inode_flag(inode, FI_AUTO_RECOVER);
2663 }
2664
2665 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2666 {
2667 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2668 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2669
2670 if (i_size_read(inode) == i_size)
2671 return;
2672
2673 i_size_write(inode, i_size);
2674 f2fs_mark_inode_dirty_sync(inode, true);
2675 if (clean || recover)
2676 set_inode_flag(inode, FI_AUTO_RECOVER);
2677 }
2678
2679 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2680 {
2681 F2FS_I(inode)->i_current_depth = depth;
2682 f2fs_mark_inode_dirty_sync(inode, true);
2683 }
2684
2685 static inline void f2fs_i_gc_failures_write(struct inode *inode,
2686 unsigned int count)
2687 {
2688 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2689 f2fs_mark_inode_dirty_sync(inode, true);
2690 }
2691
2692 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2693 {
2694 F2FS_I(inode)->i_xattr_nid = xnid;
2695 f2fs_mark_inode_dirty_sync(inode, true);
2696 }
2697
2698 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2699 {
2700 F2FS_I(inode)->i_pino = pino;
2701 f2fs_mark_inode_dirty_sync(inode, true);
2702 }
2703
2704 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2705 {
2706 struct f2fs_inode_info *fi = F2FS_I(inode);
2707
2708 if (ri->i_inline & F2FS_INLINE_XATTR)
2709 set_bit(FI_INLINE_XATTR, fi->flags);
2710 if (ri->i_inline & F2FS_INLINE_DATA)
2711 set_bit(FI_INLINE_DATA, fi->flags);
2712 if (ri->i_inline & F2FS_INLINE_DENTRY)
2713 set_bit(FI_INLINE_DENTRY, fi->flags);
2714 if (ri->i_inline & F2FS_DATA_EXIST)
2715 set_bit(FI_DATA_EXIST, fi->flags);
2716 if (ri->i_inline & F2FS_INLINE_DOTS)
2717 set_bit(FI_INLINE_DOTS, fi->flags);
2718 if (ri->i_inline & F2FS_EXTRA_ATTR)
2719 set_bit(FI_EXTRA_ATTR, fi->flags);
2720 if (ri->i_inline & F2FS_PIN_FILE)
2721 set_bit(FI_PIN_FILE, fi->flags);
2722 }
2723
2724 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2725 {
2726 ri->i_inline = 0;
2727
2728 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2729 ri->i_inline |= F2FS_INLINE_XATTR;
2730 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2731 ri->i_inline |= F2FS_INLINE_DATA;
2732 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2733 ri->i_inline |= F2FS_INLINE_DENTRY;
2734 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2735 ri->i_inline |= F2FS_DATA_EXIST;
2736 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2737 ri->i_inline |= F2FS_INLINE_DOTS;
2738 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2739 ri->i_inline |= F2FS_EXTRA_ATTR;
2740 if (is_inode_flag_set(inode, FI_PIN_FILE))
2741 ri->i_inline |= F2FS_PIN_FILE;
2742 }
2743
2744 static inline int f2fs_has_extra_attr(struct inode *inode)
2745 {
2746 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2747 }
2748
2749 static inline int f2fs_has_inline_xattr(struct inode *inode)
2750 {
2751 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2752 }
2753
2754 static inline int f2fs_compressed_file(struct inode *inode)
2755 {
2756 return S_ISREG(inode->i_mode) &&
2757 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2758 }
2759
2760 static inline unsigned int addrs_per_inode(struct inode *inode)
2761 {
2762 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2763 get_inline_xattr_addrs(inode);
2764
2765 if (!f2fs_compressed_file(inode))
2766 return addrs;
2767 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2768 }
2769
2770 static inline unsigned int addrs_per_block(struct inode *inode)
2771 {
2772 if (!f2fs_compressed_file(inode))
2773 return DEF_ADDRS_PER_BLOCK;
2774 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2775 }
2776
2777 static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2778 {
2779 struct f2fs_inode *ri = F2FS_INODE(page);
2780
2781 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2782 get_inline_xattr_addrs(inode)]);
2783 }
2784
2785 static inline int inline_xattr_size(struct inode *inode)
2786 {
2787 if (f2fs_has_inline_xattr(inode))
2788 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2789 return 0;
2790 }
2791
2792 static inline int f2fs_has_inline_data(struct inode *inode)
2793 {
2794 return is_inode_flag_set(inode, FI_INLINE_DATA);
2795 }
2796
2797 static inline int f2fs_exist_data(struct inode *inode)
2798 {
2799 return is_inode_flag_set(inode, FI_DATA_EXIST);
2800 }
2801
2802 static inline int f2fs_has_inline_dots(struct inode *inode)
2803 {
2804 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2805 }
2806
2807 static inline int f2fs_is_mmap_file(struct inode *inode)
2808 {
2809 return is_inode_flag_set(inode, FI_MMAP_FILE);
2810 }
2811
2812 static inline bool f2fs_is_pinned_file(struct inode *inode)
2813 {
2814 return is_inode_flag_set(inode, FI_PIN_FILE);
2815 }
2816
2817 static inline bool f2fs_is_atomic_file(struct inode *inode)
2818 {
2819 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2820 }
2821
2822 static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2823 {
2824 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2825 }
2826
2827 static inline bool f2fs_is_volatile_file(struct inode *inode)
2828 {
2829 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2830 }
2831
2832 static inline bool f2fs_is_first_block_written(struct inode *inode)
2833 {
2834 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2835 }
2836
2837 static inline bool f2fs_is_drop_cache(struct inode *inode)
2838 {
2839 return is_inode_flag_set(inode, FI_DROP_CACHE);
2840 }
2841
2842 static inline void *inline_data_addr(struct inode *inode, struct page *page)
2843 {
2844 struct f2fs_inode *ri = F2FS_INODE(page);
2845 int extra_size = get_extra_isize(inode);
2846
2847 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2848 }
2849
2850 static inline int f2fs_has_inline_dentry(struct inode *inode)
2851 {
2852 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2853 }
2854
2855 static inline int is_file(struct inode *inode, int type)
2856 {
2857 return F2FS_I(inode)->i_advise & type;
2858 }
2859
2860 static inline void set_file(struct inode *inode, int type)
2861 {
2862 F2FS_I(inode)->i_advise |= type;
2863 f2fs_mark_inode_dirty_sync(inode, true);
2864 }
2865
2866 static inline void clear_file(struct inode *inode, int type)
2867 {
2868 F2FS_I(inode)->i_advise &= ~type;
2869 f2fs_mark_inode_dirty_sync(inode, true);
2870 }
2871
2872 static inline bool f2fs_is_time_consistent(struct inode *inode)
2873 {
2874 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2875 return false;
2876 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2877 return false;
2878 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2879 return false;
2880 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2881 &F2FS_I(inode)->i_crtime))
2882 return false;
2883 return true;
2884 }
2885
2886 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2887 {
2888 bool ret;
2889
2890 if (dsync) {
2891 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2892
2893 spin_lock(&sbi->inode_lock[DIRTY_META]);
2894 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2895 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2896 return ret;
2897 }
2898 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2899 file_keep_isize(inode) ||
2900 i_size_read(inode) & ~PAGE_MASK)
2901 return false;
2902
2903 if (!f2fs_is_time_consistent(inode))
2904 return false;
2905
2906 spin_lock(&F2FS_I(inode)->i_size_lock);
2907 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2908 spin_unlock(&F2FS_I(inode)->i_size_lock);
2909
2910 return ret;
2911 }
2912
2913 static inline bool f2fs_readonly(struct super_block *sb)
2914 {
2915 return sb_rdonly(sb);
2916 }
2917
2918 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2919 {
2920 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2921 }
2922
2923 static inline bool is_dot_dotdot(const struct qstr *str)
2924 {
2925 if (str->len == 1 && str->name[0] == '.')
2926 return true;
2927
2928 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.')
2929 return true;
2930
2931 return false;
2932 }
2933
2934 static inline bool f2fs_may_extent_tree(struct inode *inode)
2935 {
2936 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2937
2938 if (!test_opt(sbi, EXTENT_CACHE) ||
2939 is_inode_flag_set(inode, FI_NO_EXTENT) ||
2940 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
2941 return false;
2942
2943 /*
2944 * for recovered files during mount do not create extents
2945 * if shrinker is not registered.
2946 */
2947 if (list_empty(&sbi->s_list))
2948 return false;
2949
2950 return S_ISREG(inode->i_mode);
2951 }
2952
2953 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2954 size_t size, gfp_t flags)
2955 {
2956 void *ret;
2957
2958 if (time_to_inject(sbi, FAULT_KMALLOC)) {
2959 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
2960 return NULL;
2961 }
2962
2963 ret = kmalloc(size, flags);
2964 if (ret)
2965 return ret;
2966
2967 return kvmalloc(size, flags);
2968 }
2969
2970 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
2971 size_t size, gfp_t flags)
2972 {
2973 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
2974 }
2975
2976 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
2977 size_t size, gfp_t flags)
2978 {
2979 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
2980 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
2981 return NULL;
2982 }
2983
2984 return kvmalloc(size, flags);
2985 }
2986
2987 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
2988 size_t size, gfp_t flags)
2989 {
2990 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
2991 }
2992
2993 static inline int get_extra_isize(struct inode *inode)
2994 {
2995 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
2996 }
2997
2998 static inline int get_inline_xattr_addrs(struct inode *inode)
2999 {
3000 return F2FS_I(inode)->i_inline_xattr_size;
3001 }
3002
3003 #define f2fs_get_inode_mode(i) \
3004 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3005 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3006
3007 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3008 (offsetof(struct f2fs_inode, i_extra_end) - \
3009 offsetof(struct f2fs_inode, i_extra_isize)) \
3010
3011 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3012 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3013 ((offsetof(typeof(*(f2fs_inode)), field) + \
3014 sizeof((f2fs_inode)->field)) \
3015 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3016
3017 #define DEFAULT_IOSTAT_PERIOD_MS 3000
3018 #define MIN_IOSTAT_PERIOD_MS 100
3019 /* maximum period of iostat tracing is 1 day */
3020 #define MAX_IOSTAT_PERIOD_MS 8640000
3021
3022 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3023 {
3024 int i;
3025
3026 spin_lock(&sbi->iostat_lock);
3027 for (i = 0; i < NR_IO_TYPE; i++) {
3028 sbi->rw_iostat[i] = 0;
3029 sbi->prev_rw_iostat[i] = 0;
3030 }
3031 spin_unlock(&sbi->iostat_lock);
3032 }
3033
3034 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3035
3036 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3037 enum iostat_type type, unsigned long long io_bytes)
3038 {
3039 if (!sbi->iostat_enable)
3040 return;
3041 spin_lock(&sbi->iostat_lock);
3042 sbi->rw_iostat[type] += io_bytes;
3043
3044 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3045 sbi->rw_iostat[APP_BUFFERED_IO] =
3046 sbi->rw_iostat[APP_WRITE_IO] -
3047 sbi->rw_iostat[APP_DIRECT_IO];
3048
3049 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3050 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3051 sbi->rw_iostat[APP_READ_IO] -
3052 sbi->rw_iostat[APP_DIRECT_READ_IO];
3053 spin_unlock(&sbi->iostat_lock);
3054
3055 f2fs_record_iostat(sbi);
3056 }
3057
3058 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3059
3060 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3061
3062 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3063 block_t blkaddr, int type);
3064 static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3065 block_t blkaddr, int type)
3066 {
3067 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3068 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3069 blkaddr, type);
3070 f2fs_bug_on(sbi, 1);
3071 }
3072 }
3073
3074 static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3075 {
3076 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3077 blkaddr == COMPRESS_ADDR)
3078 return false;
3079 return true;
3080 }
3081
3082 static inline void f2fs_set_page_private(struct page *page,
3083 unsigned long data)
3084 {
3085 if (PagePrivate(page))
3086 return;
3087
3088 get_page(page);
3089 SetPagePrivate(page);
3090 set_page_private(page, data);
3091 }
3092
3093 static inline void f2fs_clear_page_private(struct page *page)
3094 {
3095 if (!PagePrivate(page))
3096 return;
3097
3098 set_page_private(page, 0);
3099 ClearPagePrivate(page);
3100 f2fs_put_page(page, 0);
3101 }
3102
3103 /*
3104 * file.c
3105 */
3106 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3107 void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3108 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3109 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3110 int f2fs_truncate(struct inode *inode);
3111 int f2fs_getattr(const struct path *path, struct kstat *stat,
3112 u32 request_mask, unsigned int flags);
3113 int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3114 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3115 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3116 int f2fs_precache_extents(struct inode *inode);
3117 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3118 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3119 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3120 int f2fs_pin_file_control(struct inode *inode, bool inc);
3121
3122 /*
3123 * inode.c
3124 */
3125 void f2fs_set_inode_flags(struct inode *inode);
3126 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3127 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3128 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3129 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3130 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3131 void f2fs_update_inode(struct inode *inode, struct page *node_page);
3132 void f2fs_update_inode_page(struct inode *inode);
3133 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3134 void f2fs_evict_inode(struct inode *inode);
3135 void f2fs_handle_failed_inode(struct inode *inode);
3136
3137 /*
3138 * namei.c
3139 */
3140 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3141 bool hot, bool set);
3142 struct dentry *f2fs_get_parent(struct dentry *child);
3143
3144 extern int f2fs_ci_compare(const struct inode *parent,
3145 const struct qstr *name,
3146 const struct qstr *entry,
3147 bool quick);
3148
3149 /*
3150 * dir.c
3151 */
3152 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3153 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname,
3154 f2fs_hash_t namehash, int *max_slots,
3155 struct f2fs_dentry_ptr *d);
3156 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3157 unsigned int start_pos, struct fscrypt_str *fstr);
3158 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3159 struct f2fs_dentry_ptr *d);
3160 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3161 const struct qstr *new_name,
3162 const struct qstr *orig_name, struct page *dpage);
3163 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3164 unsigned int current_depth);
3165 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3166 void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3167 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3168 struct fscrypt_name *fname, struct page **res_page);
3169 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3170 const struct qstr *child, struct page **res_page);
3171 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3172 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3173 struct page **page);
3174 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3175 struct page *page, struct inode *inode);
3176 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3177 struct fscrypt_name *fname);
3178 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3179 const struct qstr *name, f2fs_hash_t name_hash,
3180 unsigned int bit_pos);
3181 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
3182 const struct qstr *orig_name,
3183 struct inode *inode, nid_t ino, umode_t mode);
3184 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname,
3185 struct inode *inode, nid_t ino, umode_t mode);
3186 int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3187 struct inode *inode, nid_t ino, umode_t mode);
3188 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3189 struct inode *dir, struct inode *inode);
3190 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3191 bool f2fs_empty_dir(struct inode *dir);
3192
3193 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3194 {
3195 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3196 inode, inode->i_ino, inode->i_mode);
3197 }
3198
3199 /*
3200 * super.c
3201 */
3202 int f2fs_inode_dirtied(struct inode *inode, bool sync);
3203 void f2fs_inode_synced(struct inode *inode);
3204 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3205 int f2fs_quota_sync(struct super_block *sb, int type);
3206 void f2fs_quota_off_umount(struct super_block *sb);
3207 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3208 int f2fs_sync_fs(struct super_block *sb, int sync);
3209 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3210
3211 /*
3212 * hash.c
3213 */
3214 f2fs_hash_t f2fs_dentry_hash(const struct inode *dir,
3215 const struct qstr *name_info, struct fscrypt_name *fname);
3216
3217 /*
3218 * node.c
3219 */
3220 struct dnode_of_data;
3221 struct node_info;
3222
3223 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3224 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3225 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3226 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3227 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3228 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3229 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3230 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3231 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3232 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3233 struct node_info *ni);
3234 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3235 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3236 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3237 int f2fs_truncate_xattr_node(struct inode *inode);
3238 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3239 unsigned int seq_id);
3240 int f2fs_remove_inode_page(struct inode *inode);
3241 struct page *f2fs_new_inode_page(struct inode *inode);
3242 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3243 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3244 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3245 struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3246 int f2fs_move_node_page(struct page *node_page, int gc_type);
3247 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3248 struct writeback_control *wbc, bool atomic,
3249 unsigned int *seq_id);
3250 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3251 struct writeback_control *wbc,
3252 bool do_balance, enum iostat_type io_type);
3253 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3254 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3255 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3256 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3257 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3258 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3259 int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3260 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3261 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3262 unsigned int segno, struct f2fs_summary_block *sum);
3263 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3264 int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3265 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3266 int __init f2fs_create_node_manager_caches(void);
3267 void f2fs_destroy_node_manager_caches(void);
3268
3269 /*
3270 * segment.c
3271 */
3272 bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3273 void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3274 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3275 void f2fs_drop_inmem_pages(struct inode *inode);
3276 void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3277 int f2fs_commit_inmem_pages(struct inode *inode);
3278 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3279 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3280 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3281 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3282 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3283 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3284 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3285 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3286 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3287 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3288 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3289 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3290 struct cp_control *cpc);
3291 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3292 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3293 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3294 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3295 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3296 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3297 unsigned int start, unsigned int end);
3298 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type);
3299 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3300 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3301 struct cp_control *cpc);
3302 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3303 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3304 block_t blk_addr);
3305 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3306 enum iostat_type io_type);
3307 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3308 void f2fs_outplace_write_data(struct dnode_of_data *dn,
3309 struct f2fs_io_info *fio);
3310 int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3311 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3312 block_t old_blkaddr, block_t new_blkaddr,
3313 bool recover_curseg, bool recover_newaddr);
3314 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3315 block_t old_addr, block_t new_addr,
3316 unsigned char version, bool recover_curseg,
3317 bool recover_newaddr);
3318 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3319 block_t old_blkaddr, block_t *new_blkaddr,
3320 struct f2fs_summary *sum, int type,
3321 struct f2fs_io_info *fio, bool add_list);
3322 void f2fs_wait_on_page_writeback(struct page *page,
3323 enum page_type type, bool ordered, bool locked);
3324 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3325 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3326 block_t len);
3327 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3328 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3329 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3330 unsigned int val, int alloc);
3331 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3332 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3333 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3334 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3335 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3336 int __init f2fs_create_segment_manager_caches(void);
3337 void f2fs_destroy_segment_manager_caches(void);
3338 int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3339 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3340 enum page_type type, enum temp_type temp);
3341
3342 /*
3343 * checkpoint.c
3344 */
3345 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3346 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3347 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3348 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3349 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3350 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3351 block_t blkaddr, int type);
3352 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3353 int type, bool sync);
3354 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3355 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3356 long nr_to_write, enum iostat_type io_type);
3357 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3358 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3359 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3360 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3361 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3362 unsigned int devidx, int type);
3363 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3364 unsigned int devidx, int type);
3365 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3366 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3367 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3368 void f2fs_add_orphan_inode(struct inode *inode);
3369 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3370 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3371 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3372 void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3373 void f2fs_remove_dirty_inode(struct inode *inode);
3374 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3375 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3376 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3377 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3378 int __init f2fs_create_checkpoint_caches(void);
3379 void f2fs_destroy_checkpoint_caches(void);
3380
3381 /*
3382 * data.c
3383 */
3384 int __init f2fs_init_bioset(void);
3385 void f2fs_destroy_bioset(void);
3386 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio);
3387 int f2fs_init_bio_entry_cache(void);
3388 void f2fs_destroy_bio_entry_cache(void);
3389 void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3390 struct bio *bio, enum page_type type);
3391 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3392 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3393 struct inode *inode, struct page *page,
3394 nid_t ino, enum page_type type);
3395 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3396 struct bio **bio, struct page *page);
3397 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3398 int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3399 int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3400 void f2fs_submit_page_write(struct f2fs_io_info *fio);
3401 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3402 block_t blk_addr, struct bio *bio);
3403 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3404 void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3405 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3406 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3407 int f2fs_reserve_new_block(struct dnode_of_data *dn);
3408 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3409 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3410 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3411 int f2fs_mpage_readpages(struct address_space *mapping,
3412 struct list_head *pages, struct page *page,
3413 unsigned nr_pages, bool is_readahead);
3414 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3415 int op_flags, bool for_write);
3416 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3417 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3418 bool for_write);
3419 struct page *f2fs_get_new_data_page(struct inode *inode,
3420 struct page *ipage, pgoff_t index, bool new_i_size);
3421 int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3422 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3423 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3424 int create, int flag);
3425 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3426 u64 start, u64 len);
3427 int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3428 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3429 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3430 int f2fs_write_single_data_page(struct page *page, int *submitted,
3431 struct bio **bio, sector_t *last_block,
3432 struct writeback_control *wbc,
3433 enum iostat_type io_type,
3434 int compr_blocks);
3435 void f2fs_invalidate_page(struct page *page, unsigned int offset,
3436 unsigned int length);
3437 int f2fs_release_page(struct page *page, gfp_t wait);
3438 #ifdef CONFIG_MIGRATION
3439 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3440 struct page *page, enum migrate_mode mode);
3441 #endif
3442 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3443 void f2fs_clear_page_cache_dirty_tag(struct page *page);
3444 int f2fs_init_post_read_processing(void);
3445 void f2fs_destroy_post_read_processing(void);
3446 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3447 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3448
3449 /*
3450 * gc.c
3451 */
3452 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3453 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3454 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3455 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3456 unsigned int segno);
3457 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3458 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3459
3460 /*
3461 * recovery.c
3462 */
3463 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3464 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3465
3466 /*
3467 * debug.c
3468 */
3469 #ifdef CONFIG_F2FS_STAT_FS
3470 struct f2fs_stat_info {
3471 struct list_head stat_list;
3472 struct f2fs_sb_info *sbi;
3473 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3474 int main_area_segs, main_area_sections, main_area_zones;
3475 unsigned long long hit_largest, hit_cached, hit_rbtree;
3476 unsigned long long hit_total, total_ext;
3477 int ext_tree, zombie_tree, ext_node;
3478 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3479 int ndirty_data, ndirty_qdata;
3480 int inmem_pages;
3481 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3482 int nats, dirty_nats, sits, dirty_sits;
3483 int free_nids, avail_nids, alloc_nids;
3484 int total_count, utilization;
3485 int bg_gc, nr_wb_cp_data, nr_wb_data;
3486 int nr_rd_data, nr_rd_node, nr_rd_meta;
3487 int nr_dio_read, nr_dio_write;
3488 unsigned int io_skip_bggc, other_skip_bggc;
3489 int nr_flushing, nr_flushed, flush_list_empty;
3490 int nr_discarding, nr_discarded;
3491 int nr_discard_cmd;
3492 unsigned int undiscard_blks;
3493 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3494 int compr_inode, compr_blocks;
3495 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3496 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3497 unsigned int bimodal, avg_vblocks;
3498 int util_free, util_valid, util_invalid;
3499 int rsvd_segs, overp_segs;
3500 int dirty_count, node_pages, meta_pages;
3501 int prefree_count, call_count, cp_count, bg_cp_count;
3502 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3503 int bg_node_segs, bg_data_segs;
3504 int tot_blks, data_blks, node_blks;
3505 int bg_data_blks, bg_node_blks;
3506 unsigned long long skipped_atomic_files[2];
3507 int curseg[NR_CURSEG_TYPE];
3508 int cursec[NR_CURSEG_TYPE];
3509 int curzone[NR_CURSEG_TYPE];
3510
3511 unsigned int meta_count[META_MAX];
3512 unsigned int segment_count[2];
3513 unsigned int block_count[2];
3514 unsigned int inplace_count;
3515 unsigned long long base_mem, cache_mem, page_mem;
3516 };
3517
3518 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3519 {
3520 return (struct f2fs_stat_info *)sbi->stat_info;
3521 }
3522
3523 #define stat_inc_cp_count(si) ((si)->cp_count++)
3524 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3525 #define stat_inc_call_count(si) ((si)->call_count++)
3526 #define stat_inc_bggc_count(si) ((si)->bg_gc++)
3527 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3528 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3529 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3530 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3531 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3532 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3533 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3534 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3535 #define stat_inc_inline_xattr(inode) \
3536 do { \
3537 if (f2fs_has_inline_xattr(inode)) \
3538 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3539 } while (0)
3540 #define stat_dec_inline_xattr(inode) \
3541 do { \
3542 if (f2fs_has_inline_xattr(inode)) \
3543 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3544 } while (0)
3545 #define stat_inc_inline_inode(inode) \
3546 do { \
3547 if (f2fs_has_inline_data(inode)) \
3548 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3549 } while (0)
3550 #define stat_dec_inline_inode(inode) \
3551 do { \
3552 if (f2fs_has_inline_data(inode)) \
3553 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3554 } while (0)
3555 #define stat_inc_inline_dir(inode) \
3556 do { \
3557 if (f2fs_has_inline_dentry(inode)) \
3558 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3559 } while (0)
3560 #define stat_dec_inline_dir(inode) \
3561 do { \
3562 if (f2fs_has_inline_dentry(inode)) \
3563 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3564 } while (0)
3565 #define stat_inc_compr_inode(inode) \
3566 do { \
3567 if (f2fs_compressed_file(inode)) \
3568 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3569 } while (0)
3570 #define stat_dec_compr_inode(inode) \
3571 do { \
3572 if (f2fs_compressed_file(inode)) \
3573 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3574 } while (0)
3575 #define stat_add_compr_blocks(inode, blocks) \
3576 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3577 #define stat_sub_compr_blocks(inode, blocks) \
3578 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3579 #define stat_inc_meta_count(sbi, blkaddr) \
3580 do { \
3581 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3582 atomic_inc(&(sbi)->meta_count[META_CP]); \
3583 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3584 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3585 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3586 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3587 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3588 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3589 } while (0)
3590 #define stat_inc_seg_type(sbi, curseg) \
3591 ((sbi)->segment_count[(curseg)->alloc_type]++)
3592 #define stat_inc_block_count(sbi, curseg) \
3593 ((sbi)->block_count[(curseg)->alloc_type]++)
3594 #define stat_inc_inplace_blocks(sbi) \
3595 (atomic_inc(&(sbi)->inplace_count))
3596 #define stat_update_max_atomic_write(inode) \
3597 do { \
3598 int cur = F2FS_I_SB(inode)->atomic_files; \
3599 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3600 if (cur > max) \
3601 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3602 } while (0)
3603 #define stat_inc_volatile_write(inode) \
3604 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3605 #define stat_dec_volatile_write(inode) \
3606 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3607 #define stat_update_max_volatile_write(inode) \
3608 do { \
3609 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3610 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3611 if (cur > max) \
3612 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3613 } while (0)
3614 #define stat_inc_seg_count(sbi, type, gc_type) \
3615 do { \
3616 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3617 si->tot_segs++; \
3618 if ((type) == SUM_TYPE_DATA) { \
3619 si->data_segs++; \
3620 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3621 } else { \
3622 si->node_segs++; \
3623 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3624 } \
3625 } while (0)
3626
3627 #define stat_inc_tot_blk_count(si, blks) \
3628 ((si)->tot_blks += (blks))
3629
3630 #define stat_inc_data_blk_count(sbi, blks, gc_type) \
3631 do { \
3632 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3633 stat_inc_tot_blk_count(si, blks); \
3634 si->data_blks += (blks); \
3635 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3636 } while (0)
3637
3638 #define stat_inc_node_blk_count(sbi, blks, gc_type) \
3639 do { \
3640 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3641 stat_inc_tot_blk_count(si, blks); \
3642 si->node_blks += (blks); \
3643 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3644 } while (0)
3645
3646 int f2fs_build_stats(struct f2fs_sb_info *sbi);
3647 void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3648 void __init f2fs_create_root_stats(void);
3649 void f2fs_destroy_root_stats(void);
3650 void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3651 #else
3652 #define stat_inc_cp_count(si) do { } while (0)
3653 #define stat_inc_bg_cp_count(si) do { } while (0)
3654 #define stat_inc_call_count(si) do { } while (0)
3655 #define stat_inc_bggc_count(si) do { } while (0)
3656 #define stat_io_skip_bggc_count(sbi) do { } while (0)
3657 #define stat_other_skip_bggc_count(sbi) do { } while (0)
3658 #define stat_inc_dirty_inode(sbi, type) do { } while (0)
3659 #define stat_dec_dirty_inode(sbi, type) do { } while (0)
3660 #define stat_inc_total_hit(sbi) do { } while (0)
3661 #define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3662 #define stat_inc_largest_node_hit(sbi) do { } while (0)
3663 #define stat_inc_cached_node_hit(sbi) do { } while (0)
3664 #define stat_inc_inline_xattr(inode) do { } while (0)
3665 #define stat_dec_inline_xattr(inode) do { } while (0)
3666 #define stat_inc_inline_inode(inode) do { } while (0)
3667 #define stat_dec_inline_inode(inode) do { } while (0)
3668 #define stat_inc_inline_dir(inode) do { } while (0)
3669 #define stat_dec_inline_dir(inode) do { } while (0)
3670 #define stat_inc_compr_inode(inode) do { } while (0)
3671 #define stat_dec_compr_inode(inode) do { } while (0)
3672 #define stat_add_compr_blocks(inode, blocks) do { } while (0)
3673 #define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3674 #define stat_inc_atomic_write(inode) do { } while (0)
3675 #define stat_dec_atomic_write(inode) do { } while (0)
3676 #define stat_update_max_atomic_write(inode) do { } while (0)
3677 #define stat_inc_volatile_write(inode) do { } while (0)
3678 #define stat_dec_volatile_write(inode) do { } while (0)
3679 #define stat_update_max_volatile_write(inode) do { } while (0)
3680 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3681 #define stat_inc_seg_type(sbi, curseg) do { } while (0)
3682 #define stat_inc_block_count(sbi, curseg) do { } while (0)
3683 #define stat_inc_inplace_blocks(sbi) do { } while (0)
3684 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3685 #define stat_inc_tot_blk_count(si, blks) do { } while (0)
3686 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3687 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3688
3689 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3690 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3691 static inline void __init f2fs_create_root_stats(void) { }
3692 static inline void f2fs_destroy_root_stats(void) { }
3693 static inline void update_sit_info(struct f2fs_sb_info *sbi) {}
3694 #endif
3695
3696 extern const struct file_operations f2fs_dir_operations;
3697 #ifdef CONFIG_UNICODE
3698 extern const struct dentry_operations f2fs_dentry_ops;
3699 #endif
3700 extern const struct file_operations f2fs_file_operations;
3701 extern const struct inode_operations f2fs_file_inode_operations;
3702 extern const struct address_space_operations f2fs_dblock_aops;
3703 extern const struct address_space_operations f2fs_node_aops;
3704 extern const struct address_space_operations f2fs_meta_aops;
3705 extern const struct inode_operations f2fs_dir_inode_operations;
3706 extern const struct inode_operations f2fs_symlink_inode_operations;
3707 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3708 extern const struct inode_operations f2fs_special_inode_operations;
3709 extern struct kmem_cache *f2fs_inode_entry_slab;
3710
3711 /*
3712 * inline.c
3713 */
3714 bool f2fs_may_inline_data(struct inode *inode);
3715 bool f2fs_may_inline_dentry(struct inode *inode);
3716 void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3717 void f2fs_truncate_inline_inode(struct inode *inode,
3718 struct page *ipage, u64 from);
3719 int f2fs_read_inline_data(struct inode *inode, struct page *page);
3720 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3721 int f2fs_convert_inline_inode(struct inode *inode);
3722 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3723 int f2fs_write_inline_data(struct inode *inode, struct page *page);
3724 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3725 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3726 struct fscrypt_name *fname, struct page **res_page);
3727 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3728 struct page *ipage);
3729 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name,
3730 const struct qstr *orig_name,
3731 struct inode *inode, nid_t ino, umode_t mode);
3732 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3733 struct page *page, struct inode *dir,
3734 struct inode *inode);
3735 bool f2fs_empty_inline_dir(struct inode *dir);
3736 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3737 struct fscrypt_str *fstr);
3738 int f2fs_inline_data_fiemap(struct inode *inode,
3739 struct fiemap_extent_info *fieinfo,
3740 __u64 start, __u64 len);
3741
3742 /*
3743 * shrinker.c
3744 */
3745 unsigned long f2fs_shrink_count(struct shrinker *shrink,
3746 struct shrink_control *sc);
3747 unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3748 struct shrink_control *sc);
3749 void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3750 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3751
3752 /*
3753 * extent_cache.c
3754 */
3755 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3756 struct rb_entry *cached_re, unsigned int ofs);
3757 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3758 struct rb_root_cached *root,
3759 struct rb_node **parent,
3760 unsigned int ofs, bool *leftmost);
3761 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3762 struct rb_entry *cached_re, unsigned int ofs,
3763 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3764 struct rb_node ***insert_p, struct rb_node **insert_parent,
3765 bool force, bool *leftmost);
3766 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3767 struct rb_root_cached *root);
3768 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3769 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3770 void f2fs_drop_extent_tree(struct inode *inode);
3771 unsigned int f2fs_destroy_extent_node(struct inode *inode);
3772 void f2fs_destroy_extent_tree(struct inode *inode);
3773 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3774 struct extent_info *ei);
3775 void f2fs_update_extent_cache(struct dnode_of_data *dn);
3776 void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3777 pgoff_t fofs, block_t blkaddr, unsigned int len);
3778 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3779 int __init f2fs_create_extent_cache(void);
3780 void f2fs_destroy_extent_cache(void);
3781
3782 /*
3783 * sysfs.c
3784 */
3785 int __init f2fs_init_sysfs(void);
3786 void f2fs_exit_sysfs(void);
3787 int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3788 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3789
3790 /* verity.c */
3791 extern const struct fsverity_operations f2fs_verityops;
3792
3793 /*
3794 * crypto support
3795 */
3796 static inline bool f2fs_encrypted_file(struct inode *inode)
3797 {
3798 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3799 }
3800
3801 static inline void f2fs_set_encrypted_inode(struct inode *inode)
3802 {
3803 #ifdef CONFIG_FS_ENCRYPTION
3804 file_set_encrypt(inode);
3805 f2fs_set_inode_flags(inode);
3806 #endif
3807 }
3808
3809 /*
3810 * Returns true if the reads of the inode's data need to undergo some
3811 * postprocessing step, like decryption or authenticity verification.
3812 */
3813 static inline bool f2fs_post_read_required(struct inode *inode)
3814 {
3815 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3816 f2fs_compressed_file(inode);
3817 }
3818
3819 /*
3820 * compress.c
3821 */
3822 #ifdef CONFIG_F2FS_FS_COMPRESSION
3823 bool f2fs_is_compressed_page(struct page *page);
3824 struct page *f2fs_compress_control_page(struct page *page);
3825 int f2fs_prepare_compress_overwrite(struct inode *inode,
3826 struct page **pagep, pgoff_t index, void **fsdata);
3827 bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3828 pgoff_t index, unsigned copied);
3829 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3830 void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3831 bool f2fs_is_compress_backend_ready(struct inode *inode);
3832 int f2fs_init_compress_mempool(void);
3833 void f2fs_destroy_compress_mempool(void);
3834 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity);
3835 bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3836 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3837 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3838 int f2fs_write_multi_pages(struct compress_ctx *cc,
3839 int *submitted,
3840 struct writeback_control *wbc,
3841 enum iostat_type io_type);
3842 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3843 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3844 unsigned nr_pages, sector_t *last_block_in_bio,
3845 bool is_readahead, bool for_write);
3846 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3847 void f2fs_free_dic(struct decompress_io_ctx *dic);
3848 void f2fs_decompress_end_io(struct page **rpages,
3849 unsigned int cluster_size, bool err, bool verity);
3850 int f2fs_init_compress_ctx(struct compress_ctx *cc);
3851 void f2fs_destroy_compress_ctx(struct compress_ctx *cc);
3852 void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3853 #else
3854 static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3855 static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3856 {
3857 if (!f2fs_compressed_file(inode))
3858 return true;
3859 /* not support compression */
3860 return false;
3861 }
3862 static inline struct page *f2fs_compress_control_page(struct page *page)
3863 {
3864 WARN_ON_ONCE(1);
3865 return ERR_PTR(-EINVAL);
3866 }
3867 static inline int f2fs_init_compress_mempool(void) { return 0; }
3868 static inline void f2fs_destroy_compress_mempool(void) { }
3869 #endif
3870
3871 static inline void set_compress_context(struct inode *inode)
3872 {
3873 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3874
3875 F2FS_I(inode)->i_compress_algorithm =
3876 F2FS_OPTION(sbi).compress_algorithm;
3877 F2FS_I(inode)->i_log_cluster_size =
3878 F2FS_OPTION(sbi).compress_log_size;
3879 F2FS_I(inode)->i_cluster_size =
3880 1 << F2FS_I(inode)->i_log_cluster_size;
3881 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
3882 set_inode_flag(inode, FI_COMPRESSED_FILE);
3883 stat_inc_compr_inode(inode);
3884 f2fs_mark_inode_dirty_sync(inode, true);
3885 }
3886
3887 static inline u64 f2fs_disable_compressed_file(struct inode *inode)
3888 {
3889 struct f2fs_inode_info *fi = F2FS_I(inode);
3890
3891 if (!f2fs_compressed_file(inode))
3892 return 0;
3893 if (S_ISREG(inode->i_mode)) {
3894 if (get_dirty_pages(inode))
3895 return 1;
3896 if (fi->i_compr_blocks)
3897 return fi->i_compr_blocks;
3898 }
3899
3900 fi->i_flags &= ~F2FS_COMPR_FL;
3901 stat_dec_compr_inode(inode);
3902 clear_inode_flag(inode, FI_COMPRESSED_FILE);
3903 f2fs_mark_inode_dirty_sync(inode, true);
3904 return 0;
3905 }
3906
3907 #define F2FS_FEATURE_FUNCS(name, flagname) \
3908 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3909 { \
3910 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3911 }
3912
3913 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3914 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3915 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3916 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3917 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3918 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3919 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3920 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3921 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3922 F2FS_FEATURE_FUNCS(verity, VERITY);
3923 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3924 F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
3925 F2FS_FEATURE_FUNCS(compression, COMPRESSION);
3926
3927 #ifdef CONFIG_BLK_DEV_ZONED
3928 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
3929 block_t blkaddr)
3930 {
3931 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3932
3933 return test_bit(zno, FDEV(devi).blkz_seq);
3934 }
3935 #endif
3936
3937 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3938 {
3939 return f2fs_sb_has_blkzoned(sbi);
3940 }
3941
3942 static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
3943 {
3944 return blk_queue_discard(bdev_get_queue(bdev)) ||
3945 bdev_is_zoned(bdev);
3946 }
3947
3948 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3949 {
3950 int i;
3951
3952 if (!f2fs_is_multi_device(sbi))
3953 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
3954
3955 for (i = 0; i < sbi->s_ndevs; i++)
3956 if (f2fs_bdev_support_discard(FDEV(i).bdev))
3957 return true;
3958 return false;
3959 }
3960
3961 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
3962 {
3963 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
3964 f2fs_hw_should_discard(sbi);
3965 }
3966
3967 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
3968 {
3969 int i;
3970
3971 if (!f2fs_is_multi_device(sbi))
3972 return bdev_read_only(sbi->sb->s_bdev);
3973
3974 for (i = 0; i < sbi->s_ndevs; i++)
3975 if (bdev_read_only(FDEV(i).bdev))
3976 return true;
3977 return false;
3978 }
3979
3980 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
3981 {
3982 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
3983 }
3984
3985 static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode)
3986 {
3987 #ifdef CONFIG_FS_ENCRYPTION
3988 struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
3989 umode_t mode = inode->i_mode;
3990
3991 /*
3992 * If the directory encrypted or dummy encryption enabled,
3993 * then we should encrypt the inode.
3994 */
3995 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi))
3996 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
3997 #endif
3998 return false;
3999 }
4000
4001 static inline bool f2fs_may_compress(struct inode *inode)
4002 {
4003 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4004 f2fs_is_atomic_file(inode) ||
4005 f2fs_is_volatile_file(inode))
4006 return false;
4007 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4008 }
4009
4010 static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4011 u64 blocks, bool add)
4012 {
4013 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4014
4015 if (add) {
4016 F2FS_I(inode)->i_compr_blocks += diff;
4017 stat_add_compr_blocks(inode, diff);
4018 } else {
4019 F2FS_I(inode)->i_compr_blocks -= diff;
4020 stat_sub_compr_blocks(inode, diff);
4021 }
4022 f2fs_mark_inode_dirty_sync(inode, true);
4023 }
4024
4025 static inline int block_unaligned_IO(struct inode *inode,
4026 struct kiocb *iocb, struct iov_iter *iter)
4027 {
4028 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4029 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4030 loff_t offset = iocb->ki_pos;
4031 unsigned long align = offset | iov_iter_alignment(iter);
4032
4033 return align & blocksize_mask;
4034 }
4035
4036 static inline int allow_outplace_dio(struct inode *inode,
4037 struct kiocb *iocb, struct iov_iter *iter)
4038 {
4039 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4040 int rw = iov_iter_rw(iter);
4041
4042 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4043 !block_unaligned_IO(inode, iocb, iter));
4044 }
4045
4046 static inline bool f2fs_force_buffered_io(struct inode *inode,
4047 struct kiocb *iocb, struct iov_iter *iter)
4048 {
4049 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4050 int rw = iov_iter_rw(iter);
4051
4052 if (f2fs_post_read_required(inode))
4053 return true;
4054 if (f2fs_is_multi_device(sbi))
4055 return true;
4056 /*
4057 * for blkzoned device, fallback direct IO to buffered IO, so
4058 * all IOs can be serialized by log-structured write.
4059 */
4060 if (f2fs_sb_has_blkzoned(sbi))
4061 return true;
4062 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4063 if (block_unaligned_IO(inode, iocb, iter))
4064 return true;
4065 if (F2FS_IO_ALIGNED(sbi))
4066 return true;
4067 }
4068 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4069 !IS_SWAPFILE(inode))
4070 return true;
4071
4072 return false;
4073 }
4074
4075 #ifdef CONFIG_F2FS_FAULT_INJECTION
4076 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4077 unsigned int type);
4078 #else
4079 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4080 #endif
4081
4082 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4083 {
4084 #ifdef CONFIG_QUOTA
4085 if (f2fs_sb_has_quota_ino(sbi))
4086 return true;
4087 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4088 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4089 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4090 return true;
4091 #endif
4092 return false;
4093 }
4094
4095 #define EFSBADCRC EBADMSG /* Bad CRC detected */
4096 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4097
4098 #endif /* _LINUX_F2FS_H */