Loading...
Note: File does not exist in v3.1.
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
43enum {
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
65struct f2fs_fault_info {
66 atomic_t inject_ops;
67 unsigned int inject_rate;
68 unsigned int inject_type;
69};
70
71extern 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
111typedef u32 block_t; /*
112 * should not change u32, since it is the on-disk block
113 * address format, __le32.
114 */
115typedef u32 nid_t;
116
117#define COMPRESS_EXT_NUM 16
118
119struct f2fs_mount_info {
120 unsigned int opt;
121 int write_io_size_bits; /* Write IO size bits */
122 block_t root_reserved_blocks; /* root reserved blocks */
123 kuid_t s_resuid; /* reserved blocks for uid */
124 kgid_t s_resgid; /* reserved blocks for gid */
125 int active_logs; /* # of active logs */
126 int inline_xattr_size; /* inline xattr size */
127#ifdef CONFIG_F2FS_FAULT_INJECTION
128 struct f2fs_fault_info fault_info; /* For fault injection */
129#endif
130#ifdef CONFIG_QUOTA
131 /* Names of quota files with journalled quota */
132 char *s_qf_names[MAXQUOTAS];
133 int s_jquota_fmt; /* Format of quota to use */
134#endif
135 /* For which write hints are passed down to block layer */
136 int whint_mode;
137 int alloc_mode; /* segment allocation policy */
138 int fsync_mode; /* fsync policy */
139 int fs_mode; /* fs mode: LFS or ADAPTIVE */
140 int bggc_mode; /* bggc mode: off, on or sync */
141 struct fscrypt_dummy_context dummy_enc_ctx; /* test dummy encryption */
142 block_t unusable_cap_perc; /* percentage for cap */
143 block_t unusable_cap; /* Amount of space allowed to be
144 * unusable when disabling checkpoint
145 */
146
147 /* For compression */
148 unsigned char compress_algorithm; /* algorithm type */
149 unsigned compress_log_size; /* cluster log size */
150 unsigned char compress_ext_cnt; /* extension count */
151 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
152};
153
154#define F2FS_FEATURE_ENCRYPT 0x0001
155#define F2FS_FEATURE_BLKZONED 0x0002
156#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
157#define F2FS_FEATURE_EXTRA_ATTR 0x0008
158#define F2FS_FEATURE_PRJQUOTA 0x0010
159#define F2FS_FEATURE_INODE_CHKSUM 0x0020
160#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
161#define F2FS_FEATURE_QUOTA_INO 0x0080
162#define F2FS_FEATURE_INODE_CRTIME 0x0100
163#define F2FS_FEATURE_LOST_FOUND 0x0200
164#define F2FS_FEATURE_VERITY 0x0400
165#define F2FS_FEATURE_SB_CHKSUM 0x0800
166#define F2FS_FEATURE_CASEFOLD 0x1000
167#define F2FS_FEATURE_COMPRESSION 0x2000
168
169#define __F2FS_HAS_FEATURE(raw_super, mask) \
170 ((raw_super->feature & cpu_to_le32(mask)) != 0)
171#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
172#define F2FS_SET_FEATURE(sbi, mask) \
173 (sbi->raw_super->feature |= cpu_to_le32(mask))
174#define F2FS_CLEAR_FEATURE(sbi, mask) \
175 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
176
177/*
178 * Default values for user and/or group using reserved blocks
179 */
180#define F2FS_DEF_RESUID 0
181#define F2FS_DEF_RESGID 0
182
183/*
184 * For checkpoint manager
185 */
186enum {
187 NAT_BITMAP,
188 SIT_BITMAP
189};
190
191#define CP_UMOUNT 0x00000001
192#define CP_FASTBOOT 0x00000002
193#define CP_SYNC 0x00000004
194#define CP_RECOVERY 0x00000008
195#define CP_DISCARD 0x00000010
196#define CP_TRIMMED 0x00000020
197#define CP_PAUSE 0x00000040
198#define CP_RESIZE 0x00000080
199
200#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
201#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
202#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
203#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
204#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
205#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
206#define DEF_CP_INTERVAL 60 /* 60 secs */
207#define DEF_IDLE_INTERVAL 5 /* 5 secs */
208#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
209#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
210#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
211
212struct cp_control {
213 int reason;
214 __u64 trim_start;
215 __u64 trim_end;
216 __u64 trim_minlen;
217};
218
219/*
220 * indicate meta/data type
221 */
222enum {
223 META_CP,
224 META_NAT,
225 META_SIT,
226 META_SSA,
227 META_MAX,
228 META_POR,
229 DATA_GENERIC, /* check range only */
230 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
231 DATA_GENERIC_ENHANCE_READ, /*
232 * strong check on range and segment
233 * bitmap but no warning due to race
234 * condition of read on truncated area
235 * by extent_cache
236 */
237 META_GENERIC,
238};
239
240/* for the list of ino */
241enum {
242 ORPHAN_INO, /* for orphan ino list */
243 APPEND_INO, /* for append ino list */
244 UPDATE_INO, /* for update ino list */
245 TRANS_DIR_INO, /* for trasactions dir ino list */
246 FLUSH_INO, /* for multiple device flushing */
247 MAX_INO_ENTRY, /* max. list */
248};
249
250struct ino_entry {
251 struct list_head list; /* list head */
252 nid_t ino; /* inode number */
253 unsigned int dirty_device; /* dirty device bitmap */
254};
255
256/* for the list of inodes to be GCed */
257struct inode_entry {
258 struct list_head list; /* list head */
259 struct inode *inode; /* vfs inode pointer */
260};
261
262struct fsync_node_entry {
263 struct list_head list; /* list head */
264 struct page *page; /* warm node page pointer */
265 unsigned int seq_id; /* sequence id */
266};
267
268/* for the bitmap indicate blocks to be discarded */
269struct discard_entry {
270 struct list_head list; /* list head */
271 block_t start_blkaddr; /* start blockaddr of current segment */
272 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
273};
274
275/* default discard granularity of inner discard thread, unit: block count */
276#define DEFAULT_DISCARD_GRANULARITY 16
277
278/* max discard pend list number */
279#define MAX_PLIST_NUM 512
280#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
281 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
282
283enum {
284 D_PREP, /* initial */
285 D_PARTIAL, /* partially submitted */
286 D_SUBMIT, /* all submitted */
287 D_DONE, /* finished */
288};
289
290struct discard_info {
291 block_t lstart; /* logical start address */
292 block_t len; /* length */
293 block_t start; /* actual start address in dev */
294};
295
296struct discard_cmd {
297 struct rb_node rb_node; /* rb node located in rb-tree */
298 union {
299 struct {
300 block_t lstart; /* logical start address */
301 block_t len; /* length */
302 block_t start; /* actual start address in dev */
303 };
304 struct discard_info di; /* discard info */
305
306 };
307 struct list_head list; /* command list */
308 struct completion wait; /* compleation */
309 struct block_device *bdev; /* bdev */
310 unsigned short ref; /* reference count */
311 unsigned char state; /* state */
312 unsigned char queued; /* queued discard */
313 int error; /* bio error */
314 spinlock_t lock; /* for state/bio_ref updating */
315 unsigned short bio_ref; /* bio reference count */
316};
317
318enum {
319 DPOLICY_BG,
320 DPOLICY_FORCE,
321 DPOLICY_FSTRIM,
322 DPOLICY_UMOUNT,
323 MAX_DPOLICY,
324};
325
326struct discard_policy {
327 int type; /* type of discard */
328 unsigned int min_interval; /* used for candidates exist */
329 unsigned int mid_interval; /* used for device busy */
330 unsigned int max_interval; /* used for candidates not exist */
331 unsigned int max_requests; /* # of discards issued per round */
332 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
333 bool io_aware; /* issue discard in idle time */
334 bool sync; /* submit discard with REQ_SYNC flag */
335 bool ordered; /* issue discard by lba order */
336 bool timeout; /* discard timeout for put_super */
337 unsigned int granularity; /* discard granularity */
338};
339
340struct discard_cmd_control {
341 struct task_struct *f2fs_issue_discard; /* discard thread */
342 struct list_head entry_list; /* 4KB discard entry list */
343 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
344 struct list_head wait_list; /* store on-flushing entries */
345 struct list_head fstrim_list; /* in-flight discard from fstrim */
346 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
347 unsigned int discard_wake; /* to wake up discard thread */
348 struct mutex cmd_lock;
349 unsigned int nr_discards; /* # of discards in the list */
350 unsigned int max_discards; /* max. discards to be issued */
351 unsigned int discard_granularity; /* discard granularity */
352 unsigned int undiscard_blks; /* # of undiscard blocks */
353 unsigned int next_pos; /* next discard position */
354 atomic_t issued_discard; /* # of issued discard */
355 atomic_t queued_discard; /* # of queued discard */
356 atomic_t discard_cmd_cnt; /* # of cached cmd count */
357 struct rb_root_cached root; /* root of discard rb-tree */
358 bool rbtree_check; /* config for consistence check */
359};
360
361/* for the list of fsync inodes, used only during recovery */
362struct fsync_inode_entry {
363 struct list_head list; /* list head */
364 struct inode *inode; /* vfs inode pointer */
365 block_t blkaddr; /* block address locating the last fsync */
366 block_t last_dentry; /* block address locating the last dentry */
367};
368
369#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
370#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
371
372#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
373#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
374#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
375#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
376
377#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
378#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
379
380static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
381{
382 int before = nats_in_cursum(journal);
383
384 journal->n_nats = cpu_to_le16(before + i);
385 return before;
386}
387
388static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
389{
390 int before = sits_in_cursum(journal);
391
392 journal->n_sits = cpu_to_le16(before + i);
393 return before;
394}
395
396static inline bool __has_cursum_space(struct f2fs_journal *journal,
397 int size, int type)
398{
399 if (type == NAT_JOURNAL)
400 return size <= MAX_NAT_JENTRIES(journal);
401 return size <= MAX_SIT_JENTRIES(journal);
402}
403
404/*
405 * f2fs-specific ioctl commands
406 */
407#define F2FS_IOCTL_MAGIC 0xf5
408#define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
409#define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
410#define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
411#define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
412#define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
413#define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
414#define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
415#define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
416 struct f2fs_defragment)
417#define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
418 struct f2fs_move_range)
419#define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
420 struct f2fs_flush_device)
421#define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
422 struct f2fs_gc_range)
423#define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
424#define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
425#define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
426#define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
427#define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64)
428#define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64)
429#define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \
430 _IOR(F2FS_IOCTL_MAGIC, 18, __u64)
431#define F2FS_IOC_RESERVE_COMPRESS_BLOCKS \
432 _IOR(F2FS_IOCTL_MAGIC, 19, __u64)
433#define F2FS_IOC_SEC_TRIM_FILE _IOW(F2FS_IOCTL_MAGIC, 20, \
434 struct f2fs_sectrim_range)
435
436/*
437 * should be same as XFS_IOC_GOINGDOWN.
438 * Flags for going down operation used by FS_IOC_GOINGDOWN
439 */
440#define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
441#define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
442#define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
443#define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
444#define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
445#define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
446
447/*
448 * Flags used by F2FS_IOC_SEC_TRIM_FILE
449 */
450#define F2FS_TRIM_FILE_DISCARD 0x1 /* send discard command */
451#define F2FS_TRIM_FILE_ZEROOUT 0x2 /* zero out */
452#define F2FS_TRIM_FILE_MASK 0x3
453
454struct f2fs_gc_range {
455 u32 sync;
456 u64 start;
457 u64 len;
458};
459
460struct f2fs_defragment {
461 u64 start;
462 u64 len;
463};
464
465struct f2fs_move_range {
466 u32 dst_fd; /* destination fd */
467 u64 pos_in; /* start position in src_fd */
468 u64 pos_out; /* start position in dst_fd */
469 u64 len; /* size to move */
470};
471
472struct f2fs_flush_device {
473 u32 dev_num; /* device number to flush */
474 u32 segments; /* # of segments to flush */
475};
476
477struct f2fs_sectrim_range {
478 u64 start;
479 u64 len;
480 u64 flags;
481};
482
483/* for inline stuff */
484#define DEF_INLINE_RESERVED_SIZE 1
485static inline int get_extra_isize(struct inode *inode);
486static inline int get_inline_xattr_addrs(struct inode *inode);
487#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
488 (CUR_ADDRS_PER_INODE(inode) - \
489 get_inline_xattr_addrs(inode) - \
490 DEF_INLINE_RESERVED_SIZE))
491
492/* for inline dir */
493#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
494 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
495 BITS_PER_BYTE + 1))
496#define INLINE_DENTRY_BITMAP_SIZE(inode) \
497 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
498#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
499 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
500 NR_INLINE_DENTRY(inode) + \
501 INLINE_DENTRY_BITMAP_SIZE(inode)))
502
503/*
504 * For INODE and NODE manager
505 */
506/* for directory operations */
507
508struct f2fs_filename {
509 /*
510 * The filename the user specified. This is NULL for some
511 * filesystem-internal operations, e.g. converting an inline directory
512 * to a non-inline one, or roll-forward recovering an encrypted dentry.
513 */
514 const struct qstr *usr_fname;
515
516 /*
517 * The on-disk filename. For encrypted directories, this is encrypted.
518 * This may be NULL for lookups in an encrypted dir without the key.
519 */
520 struct fscrypt_str disk_name;
521
522 /* The dirhash of this filename */
523 f2fs_hash_t hash;
524
525#ifdef CONFIG_FS_ENCRYPTION
526 /*
527 * For lookups in encrypted directories: either the buffer backing
528 * disk_name, or a buffer that holds the decoded no-key name.
529 */
530 struct fscrypt_str crypto_buf;
531#endif
532#ifdef CONFIG_UNICODE
533 /*
534 * For casefolded directories: the casefolded name, but it's left NULL
535 * if the original name is not valid Unicode or if the filesystem is
536 * doing an internal operation where usr_fname is also NULL. In these
537 * cases we fall back to treating the name as an opaque byte sequence.
538 */
539 struct fscrypt_str cf_name;
540#endif
541};
542
543struct f2fs_dentry_ptr {
544 struct inode *inode;
545 void *bitmap;
546 struct f2fs_dir_entry *dentry;
547 __u8 (*filename)[F2FS_SLOT_LEN];
548 int max;
549 int nr_bitmap;
550};
551
552static inline void make_dentry_ptr_block(struct inode *inode,
553 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
554{
555 d->inode = inode;
556 d->max = NR_DENTRY_IN_BLOCK;
557 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
558 d->bitmap = t->dentry_bitmap;
559 d->dentry = t->dentry;
560 d->filename = t->filename;
561}
562
563static inline void make_dentry_ptr_inline(struct inode *inode,
564 struct f2fs_dentry_ptr *d, void *t)
565{
566 int entry_cnt = NR_INLINE_DENTRY(inode);
567 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
568 int reserved_size = INLINE_RESERVED_SIZE(inode);
569
570 d->inode = inode;
571 d->max = entry_cnt;
572 d->nr_bitmap = bitmap_size;
573 d->bitmap = t;
574 d->dentry = t + bitmap_size + reserved_size;
575 d->filename = t + bitmap_size + reserved_size +
576 SIZE_OF_DIR_ENTRY * entry_cnt;
577}
578
579/*
580 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
581 * as its node offset to distinguish from index node blocks.
582 * But some bits are used to mark the node block.
583 */
584#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
585 >> OFFSET_BIT_SHIFT)
586enum {
587 ALLOC_NODE, /* allocate a new node page if needed */
588 LOOKUP_NODE, /* look up a node without readahead */
589 LOOKUP_NODE_RA, /*
590 * look up a node with readahead called
591 * by get_data_block.
592 */
593};
594
595#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
596
597/* congestion wait timeout value, default: 20ms */
598#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
599
600/* maximum retry quota flush count */
601#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
602
603#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
604
605#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
606
607/* for in-memory extent cache entry */
608#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
609
610/* number of extent info in extent cache we try to shrink */
611#define EXTENT_CACHE_SHRINK_NUMBER 128
612
613struct rb_entry {
614 struct rb_node rb_node; /* rb node located in rb-tree */
615 unsigned int ofs; /* start offset of the entry */
616 unsigned int len; /* length of the entry */
617};
618
619struct extent_info {
620 unsigned int fofs; /* start offset in a file */
621 unsigned int len; /* length of the extent */
622 u32 blk; /* start block address of the extent */
623};
624
625struct extent_node {
626 struct rb_node rb_node; /* rb node located in rb-tree */
627 struct extent_info ei; /* extent info */
628 struct list_head list; /* node in global extent list of sbi */
629 struct extent_tree *et; /* extent tree pointer */
630};
631
632struct extent_tree {
633 nid_t ino; /* inode number */
634 struct rb_root_cached root; /* root of extent info rb-tree */
635 struct extent_node *cached_en; /* recently accessed extent node */
636 struct extent_info largest; /* largested extent info */
637 struct list_head list; /* to be used by sbi->zombie_list */
638 rwlock_t lock; /* protect extent info rb-tree */
639 atomic_t node_cnt; /* # of extent node in rb-tree*/
640 bool largest_updated; /* largest extent updated */
641};
642
643/*
644 * This structure is taken from ext4_map_blocks.
645 *
646 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
647 */
648#define F2FS_MAP_NEW (1 << BH_New)
649#define F2FS_MAP_MAPPED (1 << BH_Mapped)
650#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
651#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
652 F2FS_MAP_UNWRITTEN)
653
654struct f2fs_map_blocks {
655 block_t m_pblk;
656 block_t m_lblk;
657 unsigned int m_len;
658 unsigned int m_flags;
659 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
660 pgoff_t *m_next_extent; /* point to next possible extent */
661 int m_seg_type;
662 bool m_may_create; /* indicate it is from write path */
663};
664
665/* for flag in get_data_block */
666enum {
667 F2FS_GET_BLOCK_DEFAULT,
668 F2FS_GET_BLOCK_FIEMAP,
669 F2FS_GET_BLOCK_BMAP,
670 F2FS_GET_BLOCK_DIO,
671 F2FS_GET_BLOCK_PRE_DIO,
672 F2FS_GET_BLOCK_PRE_AIO,
673 F2FS_GET_BLOCK_PRECACHE,
674};
675
676/*
677 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
678 */
679#define FADVISE_COLD_BIT 0x01
680#define FADVISE_LOST_PINO_BIT 0x02
681#define FADVISE_ENCRYPT_BIT 0x04
682#define FADVISE_ENC_NAME_BIT 0x08
683#define FADVISE_KEEP_SIZE_BIT 0x10
684#define FADVISE_HOT_BIT 0x20
685#define FADVISE_VERITY_BIT 0x40
686
687#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
688
689#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
690#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
691#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
692#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
693#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
694#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
695#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
696#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
697#define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
698#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
699#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
700#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
701#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
702#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
703#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
704#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
705#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
706#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
707
708#define DEF_DIR_LEVEL 0
709
710enum {
711 GC_FAILURE_PIN,
712 GC_FAILURE_ATOMIC,
713 MAX_GC_FAILURE
714};
715
716/* used for f2fs_inode_info->flags */
717enum {
718 FI_NEW_INODE, /* indicate newly allocated inode */
719 FI_DIRTY_INODE, /* indicate inode is dirty or not */
720 FI_AUTO_RECOVER, /* indicate inode is recoverable */
721 FI_DIRTY_DIR, /* indicate directory has dirty pages */
722 FI_INC_LINK, /* need to increment i_nlink */
723 FI_ACL_MODE, /* indicate acl mode */
724 FI_NO_ALLOC, /* should not allocate any blocks */
725 FI_FREE_NID, /* free allocated nide */
726 FI_NO_EXTENT, /* not to use the extent cache */
727 FI_INLINE_XATTR, /* used for inline xattr */
728 FI_INLINE_DATA, /* used for inline data*/
729 FI_INLINE_DENTRY, /* used for inline dentry */
730 FI_APPEND_WRITE, /* inode has appended data */
731 FI_UPDATE_WRITE, /* inode has in-place-update data */
732 FI_NEED_IPU, /* used for ipu per file */
733 FI_ATOMIC_FILE, /* indicate atomic file */
734 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
735 FI_VOLATILE_FILE, /* indicate volatile file */
736 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
737 FI_DROP_CACHE, /* drop dirty page cache */
738 FI_DATA_EXIST, /* indicate data exists */
739 FI_INLINE_DOTS, /* indicate inline dot dentries */
740 FI_DO_DEFRAG, /* indicate defragment is running */
741 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
742 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
743 FI_HOT_DATA, /* indicate file is hot */
744 FI_EXTRA_ATTR, /* indicate file has extra attribute */
745 FI_PROJ_INHERIT, /* indicate file inherits projectid */
746 FI_PIN_FILE, /* indicate file should not be gced */
747 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
748 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
749 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
750 FI_MMAP_FILE, /* indicate file was mmapped */
751 FI_MAX, /* max flag, never be used */
752};
753
754struct f2fs_inode_info {
755 struct inode vfs_inode; /* serve a vfs inode */
756 unsigned long i_flags; /* keep an inode flags for ioctl */
757 unsigned char i_advise; /* use to give file attribute hints */
758 unsigned char i_dir_level; /* use for dentry level for large dir */
759 unsigned int i_current_depth; /* only for directory depth */
760 /* for gc failure statistic */
761 unsigned int i_gc_failures[MAX_GC_FAILURE];
762 unsigned int i_pino; /* parent inode number */
763 umode_t i_acl_mode; /* keep file acl mode temporarily */
764
765 /* Use below internally in f2fs*/
766 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
767 struct rw_semaphore i_sem; /* protect fi info */
768 atomic_t dirty_pages; /* # of dirty pages */
769 f2fs_hash_t chash; /* hash value of given file name */
770 unsigned int clevel; /* maximum level of given file name */
771 struct task_struct *task; /* lookup and create consistency */
772 struct task_struct *cp_task; /* separate cp/wb IO stats*/
773 nid_t i_xattr_nid; /* node id that contains xattrs */
774 loff_t last_disk_size; /* lastly written file size */
775 spinlock_t i_size_lock; /* protect last_disk_size */
776
777#ifdef CONFIG_QUOTA
778 struct dquot *i_dquot[MAXQUOTAS];
779
780 /* quota space reservation, managed internally by quota code */
781 qsize_t i_reserved_quota;
782#endif
783 struct list_head dirty_list; /* dirty list for dirs and files */
784 struct list_head gdirty_list; /* linked in global dirty list */
785 struct list_head inmem_ilist; /* list for inmem inodes */
786 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
787 struct task_struct *inmem_task; /* store inmemory task */
788 struct mutex inmem_lock; /* lock for inmemory pages */
789 pgoff_t ra_offset; /* ongoing readahead offset */
790 struct extent_tree *extent_tree; /* cached extent_tree entry */
791
792 /* avoid racing between foreground op and gc */
793 struct rw_semaphore i_gc_rwsem[2];
794 struct rw_semaphore i_mmap_sem;
795 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
796
797 int i_extra_isize; /* size of extra space located in i_addr */
798 kprojid_t i_projid; /* id for project quota */
799 int i_inline_xattr_size; /* inline xattr size */
800 struct timespec64 i_crtime; /* inode creation time */
801 struct timespec64 i_disk_time[4];/* inode disk times */
802
803 /* for file compress */
804 u64 i_compr_blocks; /* # of compressed blocks */
805 unsigned char i_compress_algorithm; /* algorithm type */
806 unsigned char i_log_cluster_size; /* log of cluster size */
807 unsigned int i_cluster_size; /* cluster size */
808};
809
810static inline void get_extent_info(struct extent_info *ext,
811 struct f2fs_extent *i_ext)
812{
813 ext->fofs = le32_to_cpu(i_ext->fofs);
814 ext->blk = le32_to_cpu(i_ext->blk);
815 ext->len = le32_to_cpu(i_ext->len);
816}
817
818static inline void set_raw_extent(struct extent_info *ext,
819 struct f2fs_extent *i_ext)
820{
821 i_ext->fofs = cpu_to_le32(ext->fofs);
822 i_ext->blk = cpu_to_le32(ext->blk);
823 i_ext->len = cpu_to_le32(ext->len);
824}
825
826static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
827 u32 blk, unsigned int len)
828{
829 ei->fofs = fofs;
830 ei->blk = blk;
831 ei->len = len;
832}
833
834static inline bool __is_discard_mergeable(struct discard_info *back,
835 struct discard_info *front, unsigned int max_len)
836{
837 return (back->lstart + back->len == front->lstart) &&
838 (back->len + front->len <= max_len);
839}
840
841static inline bool __is_discard_back_mergeable(struct discard_info *cur,
842 struct discard_info *back, unsigned int max_len)
843{
844 return __is_discard_mergeable(back, cur, max_len);
845}
846
847static inline bool __is_discard_front_mergeable(struct discard_info *cur,
848 struct discard_info *front, unsigned int max_len)
849{
850 return __is_discard_mergeable(cur, front, max_len);
851}
852
853static inline bool __is_extent_mergeable(struct extent_info *back,
854 struct extent_info *front)
855{
856 return (back->fofs + back->len == front->fofs &&
857 back->blk + back->len == front->blk);
858}
859
860static inline bool __is_back_mergeable(struct extent_info *cur,
861 struct extent_info *back)
862{
863 return __is_extent_mergeable(back, cur);
864}
865
866static inline bool __is_front_mergeable(struct extent_info *cur,
867 struct extent_info *front)
868{
869 return __is_extent_mergeable(cur, front);
870}
871
872extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
873static inline void __try_update_largest_extent(struct extent_tree *et,
874 struct extent_node *en)
875{
876 if (en->ei.len > et->largest.len) {
877 et->largest = en->ei;
878 et->largest_updated = true;
879 }
880}
881
882/*
883 * For free nid management
884 */
885enum nid_state {
886 FREE_NID, /* newly added to free nid list */
887 PREALLOC_NID, /* it is preallocated */
888 MAX_NID_STATE,
889};
890
891struct f2fs_nm_info {
892 block_t nat_blkaddr; /* base disk address of NAT */
893 nid_t max_nid; /* maximum possible node ids */
894 nid_t available_nids; /* # of available node ids */
895 nid_t next_scan_nid; /* the next nid to be scanned */
896 unsigned int ram_thresh; /* control the memory footprint */
897 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
898 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
899
900 /* NAT cache management */
901 struct radix_tree_root nat_root;/* root of the nat entry cache */
902 struct radix_tree_root nat_set_root;/* root of the nat set cache */
903 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
904 struct list_head nat_entries; /* cached nat entry list (clean) */
905 spinlock_t nat_list_lock; /* protect clean nat entry list */
906 unsigned int nat_cnt; /* the # of cached nat entries */
907 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
908 unsigned int nat_blocks; /* # of nat blocks */
909
910 /* free node ids management */
911 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
912 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
913 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
914 spinlock_t nid_list_lock; /* protect nid lists ops */
915 struct mutex build_lock; /* lock for build free nids */
916 unsigned char **free_nid_bitmap;
917 unsigned char *nat_block_bitmap;
918 unsigned short *free_nid_count; /* free nid count of NAT block */
919
920 /* for checkpoint */
921 char *nat_bitmap; /* NAT bitmap pointer */
922
923 unsigned int nat_bits_blocks; /* # of nat bits blocks */
924 unsigned char *nat_bits; /* NAT bits blocks */
925 unsigned char *full_nat_bits; /* full NAT pages */
926 unsigned char *empty_nat_bits; /* empty NAT pages */
927#ifdef CONFIG_F2FS_CHECK_FS
928 char *nat_bitmap_mir; /* NAT bitmap mirror */
929#endif
930 int bitmap_size; /* bitmap size */
931};
932
933/*
934 * this structure is used as one of function parameters.
935 * all the information are dedicated to a given direct node block determined
936 * by the data offset in a file.
937 */
938struct dnode_of_data {
939 struct inode *inode; /* vfs inode pointer */
940 struct page *inode_page; /* its inode page, NULL is possible */
941 struct page *node_page; /* cached direct node page */
942 nid_t nid; /* node id of the direct node block */
943 unsigned int ofs_in_node; /* data offset in the node page */
944 bool inode_page_locked; /* inode page is locked or not */
945 bool node_changed; /* is node block changed */
946 char cur_level; /* level of hole node page */
947 char max_level; /* level of current page located */
948 block_t data_blkaddr; /* block address of the node block */
949};
950
951static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
952 struct page *ipage, struct page *npage, nid_t nid)
953{
954 memset(dn, 0, sizeof(*dn));
955 dn->inode = inode;
956 dn->inode_page = ipage;
957 dn->node_page = npage;
958 dn->nid = nid;
959}
960
961/*
962 * For SIT manager
963 *
964 * By default, there are 6 active log areas across the whole main area.
965 * When considering hot and cold data separation to reduce cleaning overhead,
966 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
967 * respectively.
968 * In the current design, you should not change the numbers intentionally.
969 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
970 * logs individually according to the underlying devices. (default: 6)
971 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
972 * data and 8 for node logs.
973 */
974#define NR_CURSEG_DATA_TYPE (3)
975#define NR_CURSEG_NODE_TYPE (3)
976#define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
977
978enum {
979 CURSEG_HOT_DATA = 0, /* directory entry blocks */
980 CURSEG_WARM_DATA, /* data blocks */
981 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
982 CURSEG_HOT_NODE, /* direct node blocks of directory files */
983 CURSEG_WARM_NODE, /* direct node blocks of normal files */
984 CURSEG_COLD_NODE, /* indirect node blocks */
985 NO_CHECK_TYPE,
986 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */
987};
988
989struct flush_cmd {
990 struct completion wait;
991 struct llist_node llnode;
992 nid_t ino;
993 int ret;
994};
995
996struct flush_cmd_control {
997 struct task_struct *f2fs_issue_flush; /* flush thread */
998 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
999 atomic_t issued_flush; /* # of issued flushes */
1000 atomic_t queued_flush; /* # of queued flushes */
1001 struct llist_head issue_list; /* list for command issue */
1002 struct llist_node *dispatch_list; /* list for command dispatch */
1003};
1004
1005struct f2fs_sm_info {
1006 struct sit_info *sit_info; /* whole segment information */
1007 struct free_segmap_info *free_info; /* free segment information */
1008 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1009 struct curseg_info *curseg_array; /* active segment information */
1010
1011 struct rw_semaphore curseg_lock; /* for preventing curseg change */
1012
1013 block_t seg0_blkaddr; /* block address of 0'th segment */
1014 block_t main_blkaddr; /* start block address of main area */
1015 block_t ssa_blkaddr; /* start block address of SSA area */
1016
1017 unsigned int segment_count; /* total # of segments */
1018 unsigned int main_segments; /* # of segments in main area */
1019 unsigned int reserved_segments; /* # of reserved segments */
1020 unsigned int ovp_segments; /* # of overprovision segments */
1021
1022 /* a threshold to reclaim prefree segments */
1023 unsigned int rec_prefree_segments;
1024
1025 /* for batched trimming */
1026 unsigned int trim_sections; /* # of sections to trim */
1027
1028 struct list_head sit_entry_set; /* sit entry set list */
1029
1030 unsigned int ipu_policy; /* in-place-update policy */
1031 unsigned int min_ipu_util; /* in-place-update threshold */
1032 unsigned int min_fsync_blocks; /* threshold for fsync */
1033 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1034 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1035 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1036
1037 /* for flush command control */
1038 struct flush_cmd_control *fcc_info;
1039
1040 /* for discard command control */
1041 struct discard_cmd_control *dcc_info;
1042};
1043
1044/*
1045 * For superblock
1046 */
1047/*
1048 * COUNT_TYPE for monitoring
1049 *
1050 * f2fs monitors the number of several block types such as on-writeback,
1051 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1052 */
1053#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1054enum count_type {
1055 F2FS_DIRTY_DENTS,
1056 F2FS_DIRTY_DATA,
1057 F2FS_DIRTY_QDATA,
1058 F2FS_DIRTY_NODES,
1059 F2FS_DIRTY_META,
1060 F2FS_INMEM_PAGES,
1061 F2FS_DIRTY_IMETA,
1062 F2FS_WB_CP_DATA,
1063 F2FS_WB_DATA,
1064 F2FS_RD_DATA,
1065 F2FS_RD_NODE,
1066 F2FS_RD_META,
1067 F2FS_DIO_WRITE,
1068 F2FS_DIO_READ,
1069 NR_COUNT_TYPE,
1070};
1071
1072/*
1073 * The below are the page types of bios used in submit_bio().
1074 * The available types are:
1075 * DATA User data pages. It operates as async mode.
1076 * NODE Node pages. It operates as async mode.
1077 * META FS metadata pages such as SIT, NAT, CP.
1078 * NR_PAGE_TYPE The number of page types.
1079 * META_FLUSH Make sure the previous pages are written
1080 * with waiting the bio's completion
1081 * ... Only can be used with META.
1082 */
1083#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1084enum page_type {
1085 DATA,
1086 NODE,
1087 META,
1088 NR_PAGE_TYPE,
1089 META_FLUSH,
1090 INMEM, /* the below types are used by tracepoints only. */
1091 INMEM_DROP,
1092 INMEM_INVALIDATE,
1093 INMEM_REVOKE,
1094 IPU,
1095 OPU,
1096};
1097
1098enum temp_type {
1099 HOT = 0, /* must be zero for meta bio */
1100 WARM,
1101 COLD,
1102 NR_TEMP_TYPE,
1103};
1104
1105enum need_lock_type {
1106 LOCK_REQ = 0,
1107 LOCK_DONE,
1108 LOCK_RETRY,
1109};
1110
1111enum cp_reason_type {
1112 CP_NO_NEEDED,
1113 CP_NON_REGULAR,
1114 CP_COMPRESSED,
1115 CP_HARDLINK,
1116 CP_SB_NEED_CP,
1117 CP_WRONG_PINO,
1118 CP_NO_SPC_ROLL,
1119 CP_NODE_NEED_CP,
1120 CP_FASTBOOT_MODE,
1121 CP_SPEC_LOG_NUM,
1122 CP_RECOVER_DIR,
1123};
1124
1125enum iostat_type {
1126 /* WRITE IO */
1127 APP_DIRECT_IO, /* app direct write IOs */
1128 APP_BUFFERED_IO, /* app buffered write IOs */
1129 APP_WRITE_IO, /* app write IOs */
1130 APP_MAPPED_IO, /* app mapped IOs */
1131 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1132 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1133 FS_META_IO, /* meta IOs from kworker/reclaimer */
1134 FS_GC_DATA_IO, /* data IOs from forground gc */
1135 FS_GC_NODE_IO, /* node IOs from forground gc */
1136 FS_CP_DATA_IO, /* data IOs from checkpoint */
1137 FS_CP_NODE_IO, /* node IOs from checkpoint */
1138 FS_CP_META_IO, /* meta IOs from checkpoint */
1139
1140 /* READ IO */
1141 APP_DIRECT_READ_IO, /* app direct read IOs */
1142 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1143 APP_READ_IO, /* app read IOs */
1144 APP_MAPPED_READ_IO, /* app mapped read IOs */
1145 FS_DATA_READ_IO, /* data read IOs */
1146 FS_GDATA_READ_IO, /* data read IOs from background gc */
1147 FS_CDATA_READ_IO, /* compressed data read IOs */
1148 FS_NODE_READ_IO, /* node read IOs */
1149 FS_META_READ_IO, /* meta read IOs */
1150
1151 /* other */
1152 FS_DISCARD, /* discard */
1153 NR_IO_TYPE,
1154};
1155
1156struct f2fs_io_info {
1157 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1158 nid_t ino; /* inode number */
1159 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1160 enum temp_type temp; /* contains HOT/WARM/COLD */
1161 int op; /* contains REQ_OP_ */
1162 int op_flags; /* req_flag_bits */
1163 block_t new_blkaddr; /* new block address to be written */
1164 block_t old_blkaddr; /* old block address before Cow */
1165 struct page *page; /* page to be written */
1166 struct page *encrypted_page; /* encrypted page */
1167 struct page *compressed_page; /* compressed page */
1168 struct list_head list; /* serialize IOs */
1169 bool submitted; /* indicate IO submission */
1170 int need_lock; /* indicate we need to lock cp_rwsem */
1171 bool in_list; /* indicate fio is in io_list */
1172 bool is_por; /* indicate IO is from recovery or not */
1173 bool retry; /* need to reallocate block address */
1174 int compr_blocks; /* # of compressed block addresses */
1175 bool encrypted; /* indicate file is encrypted */
1176 enum iostat_type io_type; /* io type */
1177 struct writeback_control *io_wbc; /* writeback control */
1178 struct bio **bio; /* bio for ipu */
1179 sector_t *last_block; /* last block number in bio */
1180 unsigned char version; /* version of the node */
1181};
1182
1183struct bio_entry {
1184 struct bio *bio;
1185 struct list_head list;
1186};
1187
1188#define is_read_io(rw) ((rw) == READ)
1189struct f2fs_bio_info {
1190 struct f2fs_sb_info *sbi; /* f2fs superblock */
1191 struct bio *bio; /* bios to merge */
1192 sector_t last_block_in_bio; /* last block number */
1193 struct f2fs_io_info fio; /* store buffered io info. */
1194 struct rw_semaphore io_rwsem; /* blocking op for bio */
1195 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1196 struct list_head io_list; /* track fios */
1197 struct list_head bio_list; /* bio entry list head */
1198 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1199};
1200
1201#define FDEV(i) (sbi->devs[i])
1202#define RDEV(i) (raw_super->devs[i])
1203struct f2fs_dev_info {
1204 struct block_device *bdev;
1205 char path[MAX_PATH_LEN];
1206 unsigned int total_segments;
1207 block_t start_blk;
1208 block_t end_blk;
1209#ifdef CONFIG_BLK_DEV_ZONED
1210 unsigned int nr_blkz; /* Total number of zones */
1211 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1212#endif
1213};
1214
1215enum inode_type {
1216 DIR_INODE, /* for dirty dir inode */
1217 FILE_INODE, /* for dirty regular/symlink inode */
1218 DIRTY_META, /* for all dirtied inode metadata */
1219 ATOMIC_FILE, /* for all atomic files */
1220 NR_INODE_TYPE,
1221};
1222
1223/* for inner inode cache management */
1224struct inode_management {
1225 struct radix_tree_root ino_root; /* ino entry array */
1226 spinlock_t ino_lock; /* for ino entry lock */
1227 struct list_head ino_list; /* inode list head */
1228 unsigned long ino_num; /* number of entries */
1229};
1230
1231/* For s_flag in struct f2fs_sb_info */
1232enum {
1233 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1234 SBI_IS_CLOSE, /* specify unmounting */
1235 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1236 SBI_POR_DOING, /* recovery is doing or not */
1237 SBI_NEED_SB_WRITE, /* need to recover superblock */
1238 SBI_NEED_CP, /* need to checkpoint */
1239 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1240 SBI_IS_RECOVERED, /* recovered orphan/data */
1241 SBI_CP_DISABLED, /* CP was disabled last mount */
1242 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1243 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1244 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1245 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1246 SBI_IS_RESIZEFS, /* resizefs is in process */
1247};
1248
1249enum {
1250 CP_TIME,
1251 REQ_TIME,
1252 DISCARD_TIME,
1253 GC_TIME,
1254 DISABLE_TIME,
1255 UMOUNT_DISCARD_TIMEOUT,
1256 MAX_TIME,
1257};
1258
1259enum {
1260 GC_NORMAL,
1261 GC_IDLE_CB,
1262 GC_IDLE_GREEDY,
1263 GC_URGENT_HIGH,
1264 GC_URGENT_LOW,
1265};
1266
1267enum {
1268 BGGC_MODE_ON, /* background gc is on */
1269 BGGC_MODE_OFF, /* background gc is off */
1270 BGGC_MODE_SYNC, /*
1271 * background gc is on, migrating blocks
1272 * like foreground gc
1273 */
1274};
1275
1276enum {
1277 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1278 FS_MODE_LFS, /* use lfs allocation only */
1279};
1280
1281enum {
1282 WHINT_MODE_OFF, /* not pass down write hints */
1283 WHINT_MODE_USER, /* try to pass down hints given by users */
1284 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1285};
1286
1287enum {
1288 ALLOC_MODE_DEFAULT, /* stay default */
1289 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1290};
1291
1292enum fsync_mode {
1293 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1294 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1295 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1296};
1297
1298/*
1299 * this value is set in page as a private data which indicate that
1300 * the page is atomically written, and it is in inmem_pages list.
1301 */
1302#define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1303#define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1304
1305#define IS_ATOMIC_WRITTEN_PAGE(page) \
1306 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE)
1307#define IS_DUMMY_WRITTEN_PAGE(page) \
1308 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE)
1309
1310#ifdef CONFIG_F2FS_IO_TRACE
1311#define IS_IO_TRACED_PAGE(page) \
1312 (page_private(page) > 0 && \
1313 page_private(page) < (unsigned long)PID_MAX_LIMIT)
1314#else
1315#define IS_IO_TRACED_PAGE(page) (0)
1316#endif
1317
1318#ifdef CONFIG_FS_ENCRYPTION
1319#define DUMMY_ENCRYPTION_ENABLED(sbi) \
1320 (unlikely(F2FS_OPTION(sbi).dummy_enc_ctx.ctx != NULL))
1321#else
1322#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1323#endif
1324
1325/* For compression */
1326enum compress_algorithm_type {
1327 COMPRESS_LZO,
1328 COMPRESS_LZ4,
1329 COMPRESS_ZSTD,
1330 COMPRESS_LZORLE,
1331 COMPRESS_MAX,
1332};
1333
1334#define COMPRESS_DATA_RESERVED_SIZE 5
1335struct compress_data {
1336 __le32 clen; /* compressed data size */
1337 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1338 u8 cdata[]; /* compressed data */
1339};
1340
1341#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1342
1343#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1344
1345/* compress context */
1346struct compress_ctx {
1347 struct inode *inode; /* inode the context belong to */
1348 pgoff_t cluster_idx; /* cluster index number */
1349 unsigned int cluster_size; /* page count in cluster */
1350 unsigned int log_cluster_size; /* log of cluster size */
1351 struct page **rpages; /* pages store raw data in cluster */
1352 unsigned int nr_rpages; /* total page number in rpages */
1353 struct page **cpages; /* pages store compressed data in cluster */
1354 unsigned int nr_cpages; /* total page number in cpages */
1355 void *rbuf; /* virtual mapped address on rpages */
1356 struct compress_data *cbuf; /* virtual mapped address on cpages */
1357 size_t rlen; /* valid data length in rbuf */
1358 size_t clen; /* valid data length in cbuf */
1359 void *private; /* payload buffer for specified compression algorithm */
1360 void *private2; /* extra payload buffer */
1361};
1362
1363/* compress context for write IO path */
1364struct compress_io_ctx {
1365 u32 magic; /* magic number to indicate page is compressed */
1366 struct inode *inode; /* inode the context belong to */
1367 struct page **rpages; /* pages store raw data in cluster */
1368 unsigned int nr_rpages; /* total page number in rpages */
1369 refcount_t ref; /* referrence count of raw page */
1370};
1371
1372/* decompress io context for read IO path */
1373struct decompress_io_ctx {
1374 u32 magic; /* magic number to indicate page is compressed */
1375 struct inode *inode; /* inode the context belong to */
1376 pgoff_t cluster_idx; /* cluster index number */
1377 unsigned int cluster_size; /* page count in cluster */
1378 unsigned int log_cluster_size; /* log of cluster size */
1379 struct page **rpages; /* pages store raw data in cluster */
1380 unsigned int nr_rpages; /* total page number in rpages */
1381 struct page **cpages; /* pages store compressed data in cluster */
1382 unsigned int nr_cpages; /* total page number in cpages */
1383 struct page **tpages; /* temp pages to pad holes in cluster */
1384 void *rbuf; /* virtual mapped address on rpages */
1385 struct compress_data *cbuf; /* virtual mapped address on cpages */
1386 size_t rlen; /* valid data length in rbuf */
1387 size_t clen; /* valid data length in cbuf */
1388 refcount_t ref; /* referrence count of compressed page */
1389 bool failed; /* indicate IO error during decompression */
1390 void *private; /* payload buffer for specified decompression algorithm */
1391 void *private2; /* extra payload buffer */
1392};
1393
1394#define NULL_CLUSTER ((unsigned int)(~0))
1395#define MIN_COMPRESS_LOG_SIZE 2
1396#define MAX_COMPRESS_LOG_SIZE 8
1397#define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE)
1398
1399struct f2fs_sb_info {
1400 struct super_block *sb; /* pointer to VFS super block */
1401 struct proc_dir_entry *s_proc; /* proc entry */
1402 struct f2fs_super_block *raw_super; /* raw super block pointer */
1403 struct rw_semaphore sb_lock; /* lock for raw super block */
1404 int valid_super_block; /* valid super block no */
1405 unsigned long s_flag; /* flags for sbi */
1406 struct mutex writepages; /* mutex for writepages() */
1407#ifdef CONFIG_UNICODE
1408 struct unicode_map *s_encoding;
1409 __u16 s_encoding_flags;
1410#endif
1411
1412#ifdef CONFIG_BLK_DEV_ZONED
1413 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1414 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1415#endif
1416
1417 /* for node-related operations */
1418 struct f2fs_nm_info *nm_info; /* node manager */
1419 struct inode *node_inode; /* cache node blocks */
1420
1421 /* for segment-related operations */
1422 struct f2fs_sm_info *sm_info; /* segment manager */
1423
1424 /* for bio operations */
1425 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1426 /* keep migration IO order for LFS mode */
1427 struct rw_semaphore io_order_lock;
1428 mempool_t *write_io_dummy; /* Dummy pages */
1429
1430 /* for checkpoint */
1431 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1432 int cur_cp_pack; /* remain current cp pack */
1433 spinlock_t cp_lock; /* for flag in ckpt */
1434 struct inode *meta_inode; /* cache meta blocks */
1435 struct mutex cp_mutex; /* checkpoint procedure lock */
1436 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1437 struct rw_semaphore node_write; /* locking node writes */
1438 struct rw_semaphore node_change; /* locking node change */
1439 wait_queue_head_t cp_wait;
1440 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1441 long interval_time[MAX_TIME]; /* to store thresholds */
1442
1443 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1444
1445 spinlock_t fsync_node_lock; /* for node entry lock */
1446 struct list_head fsync_node_list; /* node list head */
1447 unsigned int fsync_seg_id; /* sequence id */
1448 unsigned int fsync_node_num; /* number of node entries */
1449
1450 /* for orphan inode, use 0'th array */
1451 unsigned int max_orphans; /* max orphan inodes */
1452
1453 /* for inode management */
1454 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1455 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1456 struct mutex flush_lock; /* for flush exclusion */
1457
1458 /* for extent tree cache */
1459 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1460 struct mutex extent_tree_lock; /* locking extent radix tree */
1461 struct list_head extent_list; /* lru list for shrinker */
1462 spinlock_t extent_lock; /* locking extent lru list */
1463 atomic_t total_ext_tree; /* extent tree count */
1464 struct list_head zombie_list; /* extent zombie tree list */
1465 atomic_t total_zombie_tree; /* extent zombie tree count */
1466 atomic_t total_ext_node; /* extent info count */
1467
1468 /* basic filesystem units */
1469 unsigned int log_sectors_per_block; /* log2 sectors per block */
1470 unsigned int log_blocksize; /* log2 block size */
1471 unsigned int blocksize; /* block size */
1472 unsigned int root_ino_num; /* root inode number*/
1473 unsigned int node_ino_num; /* node inode number*/
1474 unsigned int meta_ino_num; /* meta inode number*/
1475 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1476 unsigned int blocks_per_seg; /* blocks per segment */
1477 unsigned int segs_per_sec; /* segments per section */
1478 unsigned int secs_per_zone; /* sections per zone */
1479 unsigned int total_sections; /* total section count */
1480 unsigned int total_node_count; /* total node block count */
1481 unsigned int total_valid_node_count; /* valid node block count */
1482 loff_t max_file_blocks; /* max block index of file */
1483 int dir_level; /* directory level */
1484 int readdir_ra; /* readahead inode in readdir */
1485
1486 block_t user_block_count; /* # of user blocks */
1487 block_t total_valid_block_count; /* # of valid blocks */
1488 block_t discard_blks; /* discard command candidats */
1489 block_t last_valid_block_count; /* for recovery */
1490 block_t reserved_blocks; /* configurable reserved blocks */
1491 block_t current_reserved_blocks; /* current reserved blocks */
1492
1493 /* Additional tracking for no checkpoint mode */
1494 block_t unusable_block_count; /* # of blocks saved by last cp */
1495
1496 unsigned int nquota_files; /* # of quota sysfile */
1497 struct rw_semaphore quota_sem; /* blocking cp for flags */
1498
1499 /* # of pages, see count_type */
1500 atomic_t nr_pages[NR_COUNT_TYPE];
1501 /* # of allocated blocks */
1502 struct percpu_counter alloc_valid_block_count;
1503
1504 /* writeback control */
1505 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1506
1507 /* valid inode count */
1508 struct percpu_counter total_valid_inode_count;
1509
1510 struct f2fs_mount_info mount_opt; /* mount options */
1511
1512 /* for cleaning operations */
1513 struct rw_semaphore gc_lock; /*
1514 * semaphore for GC, avoid
1515 * race between GC and GC or CP
1516 */
1517 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1518 unsigned int cur_victim_sec; /* current victim section num */
1519 unsigned int gc_mode; /* current GC state */
1520 unsigned int next_victim_seg[2]; /* next segment in victim section */
1521
1522 /* for skip statistic */
1523 unsigned int atomic_files; /* # of opened atomic file */
1524 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1525 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1526
1527 /* threshold for gc trials on pinned files */
1528 u64 gc_pin_file_threshold;
1529 struct rw_semaphore pin_sem;
1530
1531 /* maximum # of trials to find a victim segment for SSR and GC */
1532 unsigned int max_victim_search;
1533 /* migration granularity of garbage collection, unit: segment */
1534 unsigned int migration_granularity;
1535
1536 /*
1537 * for stat information.
1538 * one is for the LFS mode, and the other is for the SSR mode.
1539 */
1540#ifdef CONFIG_F2FS_STAT_FS
1541 struct f2fs_stat_info *stat_info; /* FS status information */
1542 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1543 unsigned int segment_count[2]; /* # of allocated segments */
1544 unsigned int block_count[2]; /* # of allocated blocks */
1545 atomic_t inplace_count; /* # of inplace update */
1546 atomic64_t total_hit_ext; /* # of lookup extent cache */
1547 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1548 atomic64_t read_hit_largest; /* # of hit largest extent node */
1549 atomic64_t read_hit_cached; /* # of hit cached extent node */
1550 atomic_t inline_xattr; /* # of inline_xattr inodes */
1551 atomic_t inline_inode; /* # of inline_data inodes */
1552 atomic_t inline_dir; /* # of inline_dentry inodes */
1553 atomic_t compr_inode; /* # of compressed inodes */
1554 atomic_t compr_blocks; /* # of compressed blocks */
1555 atomic_t vw_cnt; /* # of volatile writes */
1556 atomic_t max_aw_cnt; /* max # of atomic writes */
1557 atomic_t max_vw_cnt; /* max # of volatile writes */
1558 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1559 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1560 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1561#endif
1562 spinlock_t stat_lock; /* lock for stat operations */
1563
1564 /* For app/fs IO statistics */
1565 spinlock_t iostat_lock;
1566 unsigned long long rw_iostat[NR_IO_TYPE];
1567 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1568 bool iostat_enable;
1569 unsigned long iostat_next_period;
1570 unsigned int iostat_period_ms;
1571
1572 /* to attach REQ_META|REQ_FUA flags */
1573 unsigned int data_io_flag;
1574 unsigned int node_io_flag;
1575
1576 /* For sysfs suppport */
1577 struct kobject s_kobj;
1578 struct completion s_kobj_unregister;
1579
1580 /* For shrinker support */
1581 struct list_head s_list;
1582 int s_ndevs; /* number of devices */
1583 struct f2fs_dev_info *devs; /* for device list */
1584 unsigned int dirty_device; /* for checkpoint data flush */
1585 spinlock_t dev_lock; /* protect dirty_device */
1586 struct mutex umount_mutex;
1587 unsigned int shrinker_run_no;
1588
1589 /* For write statistics */
1590 u64 sectors_written_start;
1591 u64 kbytes_written;
1592
1593 /* Reference to checksum algorithm driver via cryptoapi */
1594 struct crypto_shash *s_chksum_driver;
1595
1596 /* Precomputed FS UUID checksum for seeding other checksums */
1597 __u32 s_chksum_seed;
1598
1599 struct workqueue_struct *post_read_wq; /* post read workqueue */
1600
1601 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1602 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1603};
1604
1605struct f2fs_private_dio {
1606 struct inode *inode;
1607 void *orig_private;
1608 bio_end_io_t *orig_end_io;
1609 bool write;
1610};
1611
1612#ifdef CONFIG_F2FS_FAULT_INJECTION
1613#define f2fs_show_injection_info(sbi, type) \
1614 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1615 KERN_INFO, sbi->sb->s_id, \
1616 f2fs_fault_name[type], \
1617 __func__, __builtin_return_address(0))
1618static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1619{
1620 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1621
1622 if (!ffi->inject_rate)
1623 return false;
1624
1625 if (!IS_FAULT_SET(ffi, type))
1626 return false;
1627
1628 atomic_inc(&ffi->inject_ops);
1629 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1630 atomic_set(&ffi->inject_ops, 0);
1631 return true;
1632 }
1633 return false;
1634}
1635#else
1636#define f2fs_show_injection_info(sbi, type) do { } while (0)
1637static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1638{
1639 return false;
1640}
1641#endif
1642
1643/*
1644 * Test if the mounted volume is a multi-device volume.
1645 * - For a single regular disk volume, sbi->s_ndevs is 0.
1646 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1647 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1648 */
1649static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1650{
1651 return sbi->s_ndevs > 1;
1652}
1653
1654/* For write statistics. Suppose sector size is 512 bytes,
1655 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1656 */
1657#define BD_PART_WRITTEN(s) \
1658(((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1659 (s)->sectors_written_start) >> 1)
1660
1661static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1662{
1663 unsigned long now = jiffies;
1664
1665 sbi->last_time[type] = now;
1666
1667 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1668 if (type == REQ_TIME) {
1669 sbi->last_time[DISCARD_TIME] = now;
1670 sbi->last_time[GC_TIME] = now;
1671 }
1672}
1673
1674static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1675{
1676 unsigned long interval = sbi->interval_time[type] * HZ;
1677
1678 return time_after(jiffies, sbi->last_time[type] + interval);
1679}
1680
1681static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1682 int type)
1683{
1684 unsigned long interval = sbi->interval_time[type] * HZ;
1685 unsigned int wait_ms = 0;
1686 long delta;
1687
1688 delta = (sbi->last_time[type] + interval) - jiffies;
1689 if (delta > 0)
1690 wait_ms = jiffies_to_msecs(delta);
1691
1692 return wait_ms;
1693}
1694
1695/*
1696 * Inline functions
1697 */
1698static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1699 const void *address, unsigned int length)
1700{
1701 struct {
1702 struct shash_desc shash;
1703 char ctx[4];
1704 } desc;
1705 int err;
1706
1707 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1708
1709 desc.shash.tfm = sbi->s_chksum_driver;
1710 *(u32 *)desc.ctx = crc;
1711
1712 err = crypto_shash_update(&desc.shash, address, length);
1713 BUG_ON(err);
1714
1715 return *(u32 *)desc.ctx;
1716}
1717
1718static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1719 unsigned int length)
1720{
1721 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1722}
1723
1724static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1725 void *buf, size_t buf_size)
1726{
1727 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1728}
1729
1730static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1731 const void *address, unsigned int length)
1732{
1733 return __f2fs_crc32(sbi, crc, address, length);
1734}
1735
1736static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1737{
1738 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1739}
1740
1741static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1742{
1743 return sb->s_fs_info;
1744}
1745
1746static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1747{
1748 return F2FS_SB(inode->i_sb);
1749}
1750
1751static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1752{
1753 return F2FS_I_SB(mapping->host);
1754}
1755
1756static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1757{
1758 return F2FS_M_SB(page_file_mapping(page));
1759}
1760
1761static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1762{
1763 return (struct f2fs_super_block *)(sbi->raw_super);
1764}
1765
1766static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1767{
1768 return (struct f2fs_checkpoint *)(sbi->ckpt);
1769}
1770
1771static inline struct f2fs_node *F2FS_NODE(struct page *page)
1772{
1773 return (struct f2fs_node *)page_address(page);
1774}
1775
1776static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1777{
1778 return &((struct f2fs_node *)page_address(page))->i;
1779}
1780
1781static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1782{
1783 return (struct f2fs_nm_info *)(sbi->nm_info);
1784}
1785
1786static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1787{
1788 return (struct f2fs_sm_info *)(sbi->sm_info);
1789}
1790
1791static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1792{
1793 return (struct sit_info *)(SM_I(sbi)->sit_info);
1794}
1795
1796static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1797{
1798 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1799}
1800
1801static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1802{
1803 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1804}
1805
1806static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1807{
1808 return sbi->meta_inode->i_mapping;
1809}
1810
1811static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1812{
1813 return sbi->node_inode->i_mapping;
1814}
1815
1816static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1817{
1818 return test_bit(type, &sbi->s_flag);
1819}
1820
1821static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1822{
1823 set_bit(type, &sbi->s_flag);
1824}
1825
1826static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1827{
1828 clear_bit(type, &sbi->s_flag);
1829}
1830
1831static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1832{
1833 return le64_to_cpu(cp->checkpoint_ver);
1834}
1835
1836static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1837{
1838 if (type < F2FS_MAX_QUOTAS)
1839 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1840 return 0;
1841}
1842
1843static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1844{
1845 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1846 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1847}
1848
1849static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1850{
1851 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1852
1853 return ckpt_flags & f;
1854}
1855
1856static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1857{
1858 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1859}
1860
1861static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1862{
1863 unsigned int ckpt_flags;
1864
1865 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1866 ckpt_flags |= f;
1867 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1868}
1869
1870static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1871{
1872 unsigned long flags;
1873
1874 spin_lock_irqsave(&sbi->cp_lock, flags);
1875 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1876 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1877}
1878
1879static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1880{
1881 unsigned int ckpt_flags;
1882
1883 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1884 ckpt_flags &= (~f);
1885 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1886}
1887
1888static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1889{
1890 unsigned long flags;
1891
1892 spin_lock_irqsave(&sbi->cp_lock, flags);
1893 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1894 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1895}
1896
1897static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1898{
1899 unsigned long flags;
1900 unsigned char *nat_bits;
1901
1902 /*
1903 * In order to re-enable nat_bits we need to call fsck.f2fs by
1904 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1905 * so let's rely on regular fsck or unclean shutdown.
1906 */
1907
1908 if (lock)
1909 spin_lock_irqsave(&sbi->cp_lock, flags);
1910 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1911 nat_bits = NM_I(sbi)->nat_bits;
1912 NM_I(sbi)->nat_bits = NULL;
1913 if (lock)
1914 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1915
1916 kvfree(nat_bits);
1917}
1918
1919static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1920 struct cp_control *cpc)
1921{
1922 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1923
1924 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1925}
1926
1927static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1928{
1929 down_read(&sbi->cp_rwsem);
1930}
1931
1932static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1933{
1934 return down_read_trylock(&sbi->cp_rwsem);
1935}
1936
1937static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1938{
1939 up_read(&sbi->cp_rwsem);
1940}
1941
1942static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1943{
1944 down_write(&sbi->cp_rwsem);
1945}
1946
1947static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1948{
1949 up_write(&sbi->cp_rwsem);
1950}
1951
1952static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1953{
1954 int reason = CP_SYNC;
1955
1956 if (test_opt(sbi, FASTBOOT))
1957 reason = CP_FASTBOOT;
1958 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1959 reason = CP_UMOUNT;
1960 return reason;
1961}
1962
1963static inline bool __remain_node_summaries(int reason)
1964{
1965 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1966}
1967
1968static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1969{
1970 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1971 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1972}
1973
1974/*
1975 * Check whether the inode has blocks or not
1976 */
1977static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1978{
1979 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1980
1981 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1982}
1983
1984static inline bool f2fs_has_xattr_block(unsigned int ofs)
1985{
1986 return ofs == XATTR_NODE_OFFSET;
1987}
1988
1989static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1990 struct inode *inode, bool cap)
1991{
1992 if (!inode)
1993 return true;
1994 if (!test_opt(sbi, RESERVE_ROOT))
1995 return false;
1996 if (IS_NOQUOTA(inode))
1997 return true;
1998 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1999 return true;
2000 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2001 in_group_p(F2FS_OPTION(sbi).s_resgid))
2002 return true;
2003 if (cap && capable(CAP_SYS_RESOURCE))
2004 return true;
2005 return false;
2006}
2007
2008static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2009static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2010 struct inode *inode, blkcnt_t *count)
2011{
2012 blkcnt_t diff = 0, release = 0;
2013 block_t avail_user_block_count;
2014 int ret;
2015
2016 ret = dquot_reserve_block(inode, *count);
2017 if (ret)
2018 return ret;
2019
2020 if (time_to_inject(sbi, FAULT_BLOCK)) {
2021 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2022 release = *count;
2023 goto release_quota;
2024 }
2025
2026 /*
2027 * let's increase this in prior to actual block count change in order
2028 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2029 */
2030 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2031
2032 spin_lock(&sbi->stat_lock);
2033 sbi->total_valid_block_count += (block_t)(*count);
2034 avail_user_block_count = sbi->user_block_count -
2035 sbi->current_reserved_blocks;
2036
2037 if (!__allow_reserved_blocks(sbi, inode, true))
2038 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2039 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2040 if (avail_user_block_count > sbi->unusable_block_count)
2041 avail_user_block_count -= sbi->unusable_block_count;
2042 else
2043 avail_user_block_count = 0;
2044 }
2045 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2046 diff = sbi->total_valid_block_count - avail_user_block_count;
2047 if (diff > *count)
2048 diff = *count;
2049 *count -= diff;
2050 release = diff;
2051 sbi->total_valid_block_count -= diff;
2052 if (!*count) {
2053 spin_unlock(&sbi->stat_lock);
2054 goto enospc;
2055 }
2056 }
2057 spin_unlock(&sbi->stat_lock);
2058
2059 if (unlikely(release)) {
2060 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2061 dquot_release_reservation_block(inode, release);
2062 }
2063 f2fs_i_blocks_write(inode, *count, true, true);
2064 return 0;
2065
2066enospc:
2067 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2068release_quota:
2069 dquot_release_reservation_block(inode, release);
2070 return -ENOSPC;
2071}
2072
2073__printf(2, 3)
2074void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2075
2076#define f2fs_err(sbi, fmt, ...) \
2077 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2078#define f2fs_warn(sbi, fmt, ...) \
2079 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2080#define f2fs_notice(sbi, fmt, ...) \
2081 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2082#define f2fs_info(sbi, fmt, ...) \
2083 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2084#define f2fs_debug(sbi, fmt, ...) \
2085 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2086
2087static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2088 struct inode *inode,
2089 block_t count)
2090{
2091 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2092
2093 spin_lock(&sbi->stat_lock);
2094 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2095 sbi->total_valid_block_count -= (block_t)count;
2096 if (sbi->reserved_blocks &&
2097 sbi->current_reserved_blocks < sbi->reserved_blocks)
2098 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2099 sbi->current_reserved_blocks + count);
2100 spin_unlock(&sbi->stat_lock);
2101 if (unlikely(inode->i_blocks < sectors)) {
2102 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2103 inode->i_ino,
2104 (unsigned long long)inode->i_blocks,
2105 (unsigned long long)sectors);
2106 set_sbi_flag(sbi, SBI_NEED_FSCK);
2107 return;
2108 }
2109 f2fs_i_blocks_write(inode, count, false, true);
2110}
2111
2112static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2113{
2114 atomic_inc(&sbi->nr_pages[count_type]);
2115
2116 if (count_type == F2FS_DIRTY_DENTS ||
2117 count_type == F2FS_DIRTY_NODES ||
2118 count_type == F2FS_DIRTY_META ||
2119 count_type == F2FS_DIRTY_QDATA ||
2120 count_type == F2FS_DIRTY_IMETA)
2121 set_sbi_flag(sbi, SBI_IS_DIRTY);
2122}
2123
2124static inline void inode_inc_dirty_pages(struct inode *inode)
2125{
2126 atomic_inc(&F2FS_I(inode)->dirty_pages);
2127 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2128 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2129 if (IS_NOQUOTA(inode))
2130 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2131}
2132
2133static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2134{
2135 atomic_dec(&sbi->nr_pages[count_type]);
2136}
2137
2138static inline void inode_dec_dirty_pages(struct inode *inode)
2139{
2140 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2141 !S_ISLNK(inode->i_mode))
2142 return;
2143
2144 atomic_dec(&F2FS_I(inode)->dirty_pages);
2145 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2146 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2147 if (IS_NOQUOTA(inode))
2148 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2149}
2150
2151static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2152{
2153 return atomic_read(&sbi->nr_pages[count_type]);
2154}
2155
2156static inline int get_dirty_pages(struct inode *inode)
2157{
2158 return atomic_read(&F2FS_I(inode)->dirty_pages);
2159}
2160
2161static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2162{
2163 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2164 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2165 sbi->log_blocks_per_seg;
2166
2167 return segs / sbi->segs_per_sec;
2168}
2169
2170static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2171{
2172 return sbi->total_valid_block_count;
2173}
2174
2175static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2176{
2177 return sbi->discard_blks;
2178}
2179
2180static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2181{
2182 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2183
2184 /* return NAT or SIT bitmap */
2185 if (flag == NAT_BITMAP)
2186 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2187 else if (flag == SIT_BITMAP)
2188 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2189
2190 return 0;
2191}
2192
2193static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2194{
2195 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2196}
2197
2198static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2199{
2200 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2201 int offset;
2202
2203 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2204 offset = (flag == SIT_BITMAP) ?
2205 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2206 /*
2207 * if large_nat_bitmap feature is enabled, leave checksum
2208 * protection for all nat/sit bitmaps.
2209 */
2210 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2211 }
2212
2213 if (__cp_payload(sbi) > 0) {
2214 if (flag == NAT_BITMAP)
2215 return &ckpt->sit_nat_version_bitmap;
2216 else
2217 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2218 } else {
2219 offset = (flag == NAT_BITMAP) ?
2220 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2221 return &ckpt->sit_nat_version_bitmap + offset;
2222 }
2223}
2224
2225static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2226{
2227 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2228
2229 if (sbi->cur_cp_pack == 2)
2230 start_addr += sbi->blocks_per_seg;
2231 return start_addr;
2232}
2233
2234static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2235{
2236 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2237
2238 if (sbi->cur_cp_pack == 1)
2239 start_addr += sbi->blocks_per_seg;
2240 return start_addr;
2241}
2242
2243static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2244{
2245 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2246}
2247
2248static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2249{
2250 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2251}
2252
2253static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2254 struct inode *inode, bool is_inode)
2255{
2256 block_t valid_block_count;
2257 unsigned int valid_node_count, user_block_count;
2258 int err;
2259
2260 if (is_inode) {
2261 if (inode) {
2262 err = dquot_alloc_inode(inode);
2263 if (err)
2264 return err;
2265 }
2266 } else {
2267 err = dquot_reserve_block(inode, 1);
2268 if (err)
2269 return err;
2270 }
2271
2272 if (time_to_inject(sbi, FAULT_BLOCK)) {
2273 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2274 goto enospc;
2275 }
2276
2277 spin_lock(&sbi->stat_lock);
2278
2279 valid_block_count = sbi->total_valid_block_count +
2280 sbi->current_reserved_blocks + 1;
2281
2282 if (!__allow_reserved_blocks(sbi, inode, false))
2283 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2284 user_block_count = sbi->user_block_count;
2285 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2286 user_block_count -= sbi->unusable_block_count;
2287
2288 if (unlikely(valid_block_count > user_block_count)) {
2289 spin_unlock(&sbi->stat_lock);
2290 goto enospc;
2291 }
2292
2293 valid_node_count = sbi->total_valid_node_count + 1;
2294 if (unlikely(valid_node_count > sbi->total_node_count)) {
2295 spin_unlock(&sbi->stat_lock);
2296 goto enospc;
2297 }
2298
2299 sbi->total_valid_node_count++;
2300 sbi->total_valid_block_count++;
2301 spin_unlock(&sbi->stat_lock);
2302
2303 if (inode) {
2304 if (is_inode)
2305 f2fs_mark_inode_dirty_sync(inode, true);
2306 else
2307 f2fs_i_blocks_write(inode, 1, true, true);
2308 }
2309
2310 percpu_counter_inc(&sbi->alloc_valid_block_count);
2311 return 0;
2312
2313enospc:
2314 if (is_inode) {
2315 if (inode)
2316 dquot_free_inode(inode);
2317 } else {
2318 dquot_release_reservation_block(inode, 1);
2319 }
2320 return -ENOSPC;
2321}
2322
2323static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2324 struct inode *inode, bool is_inode)
2325{
2326 spin_lock(&sbi->stat_lock);
2327
2328 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2329 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2330
2331 sbi->total_valid_node_count--;
2332 sbi->total_valid_block_count--;
2333 if (sbi->reserved_blocks &&
2334 sbi->current_reserved_blocks < sbi->reserved_blocks)
2335 sbi->current_reserved_blocks++;
2336
2337 spin_unlock(&sbi->stat_lock);
2338
2339 if (is_inode) {
2340 dquot_free_inode(inode);
2341 } else {
2342 if (unlikely(inode->i_blocks == 0)) {
2343 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2344 inode->i_ino,
2345 (unsigned long long)inode->i_blocks);
2346 set_sbi_flag(sbi, SBI_NEED_FSCK);
2347 return;
2348 }
2349 f2fs_i_blocks_write(inode, 1, false, true);
2350 }
2351}
2352
2353static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2354{
2355 return sbi->total_valid_node_count;
2356}
2357
2358static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2359{
2360 percpu_counter_inc(&sbi->total_valid_inode_count);
2361}
2362
2363static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2364{
2365 percpu_counter_dec(&sbi->total_valid_inode_count);
2366}
2367
2368static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2369{
2370 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2371}
2372
2373static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2374 pgoff_t index, bool for_write)
2375{
2376 struct page *page;
2377
2378 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2379 if (!for_write)
2380 page = find_get_page_flags(mapping, index,
2381 FGP_LOCK | FGP_ACCESSED);
2382 else
2383 page = find_lock_page(mapping, index);
2384 if (page)
2385 return page;
2386
2387 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2388 f2fs_show_injection_info(F2FS_M_SB(mapping),
2389 FAULT_PAGE_ALLOC);
2390 return NULL;
2391 }
2392 }
2393
2394 if (!for_write)
2395 return grab_cache_page(mapping, index);
2396 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2397}
2398
2399static inline struct page *f2fs_pagecache_get_page(
2400 struct address_space *mapping, pgoff_t index,
2401 int fgp_flags, gfp_t gfp_mask)
2402{
2403 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2404 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2405 return NULL;
2406 }
2407
2408 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2409}
2410
2411static inline void f2fs_copy_page(struct page *src, struct page *dst)
2412{
2413 char *src_kaddr = kmap(src);
2414 char *dst_kaddr = kmap(dst);
2415
2416 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2417 kunmap(dst);
2418 kunmap(src);
2419}
2420
2421static inline void f2fs_put_page(struct page *page, int unlock)
2422{
2423 if (!page)
2424 return;
2425
2426 if (unlock) {
2427 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2428 unlock_page(page);
2429 }
2430 put_page(page);
2431}
2432
2433static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2434{
2435 if (dn->node_page)
2436 f2fs_put_page(dn->node_page, 1);
2437 if (dn->inode_page && dn->node_page != dn->inode_page)
2438 f2fs_put_page(dn->inode_page, 0);
2439 dn->node_page = NULL;
2440 dn->inode_page = NULL;
2441}
2442
2443static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2444 size_t size)
2445{
2446 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2447}
2448
2449static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2450 gfp_t flags)
2451{
2452 void *entry;
2453
2454 entry = kmem_cache_alloc(cachep, flags);
2455 if (!entry)
2456 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2457 return entry;
2458}
2459
2460static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2461{
2462 if (sbi->gc_mode == GC_URGENT_HIGH)
2463 return true;
2464
2465 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2466 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2467 get_pages(sbi, F2FS_WB_CP_DATA) ||
2468 get_pages(sbi, F2FS_DIO_READ) ||
2469 get_pages(sbi, F2FS_DIO_WRITE))
2470 return false;
2471
2472 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2473 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2474 return false;
2475
2476 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2477 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2478 return false;
2479
2480 if (sbi->gc_mode == GC_URGENT_LOW &&
2481 (type == DISCARD_TIME || type == GC_TIME))
2482 return true;
2483
2484 return f2fs_time_over(sbi, type);
2485}
2486
2487static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2488 unsigned long index, void *item)
2489{
2490 while (radix_tree_insert(root, index, item))
2491 cond_resched();
2492}
2493
2494#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2495
2496static inline bool IS_INODE(struct page *page)
2497{
2498 struct f2fs_node *p = F2FS_NODE(page);
2499
2500 return RAW_IS_INODE(p);
2501}
2502
2503static inline int offset_in_addr(struct f2fs_inode *i)
2504{
2505 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2506 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2507}
2508
2509static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2510{
2511 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2512}
2513
2514static inline int f2fs_has_extra_attr(struct inode *inode);
2515static inline block_t data_blkaddr(struct inode *inode,
2516 struct page *node_page, unsigned int offset)
2517{
2518 struct f2fs_node *raw_node;
2519 __le32 *addr_array;
2520 int base = 0;
2521 bool is_inode = IS_INODE(node_page);
2522
2523 raw_node = F2FS_NODE(node_page);
2524
2525 if (is_inode) {
2526 if (!inode)
2527 /* from GC path only */
2528 base = offset_in_addr(&raw_node->i);
2529 else if (f2fs_has_extra_attr(inode))
2530 base = get_extra_isize(inode);
2531 }
2532
2533 addr_array = blkaddr_in_node(raw_node);
2534 return le32_to_cpu(addr_array[base + offset]);
2535}
2536
2537static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2538{
2539 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2540}
2541
2542static inline int f2fs_test_bit(unsigned int nr, char *addr)
2543{
2544 int mask;
2545
2546 addr += (nr >> 3);
2547 mask = 1 << (7 - (nr & 0x07));
2548 return mask & *addr;
2549}
2550
2551static inline void f2fs_set_bit(unsigned int nr, char *addr)
2552{
2553 int mask;
2554
2555 addr += (nr >> 3);
2556 mask = 1 << (7 - (nr & 0x07));
2557 *addr |= mask;
2558}
2559
2560static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2561{
2562 int mask;
2563
2564 addr += (nr >> 3);
2565 mask = 1 << (7 - (nr & 0x07));
2566 *addr &= ~mask;
2567}
2568
2569static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2570{
2571 int mask;
2572 int ret;
2573
2574 addr += (nr >> 3);
2575 mask = 1 << (7 - (nr & 0x07));
2576 ret = mask & *addr;
2577 *addr |= mask;
2578 return ret;
2579}
2580
2581static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2582{
2583 int mask;
2584 int ret;
2585
2586 addr += (nr >> 3);
2587 mask = 1 << (7 - (nr & 0x07));
2588 ret = mask & *addr;
2589 *addr &= ~mask;
2590 return ret;
2591}
2592
2593static inline void f2fs_change_bit(unsigned int nr, char *addr)
2594{
2595 int mask;
2596
2597 addr += (nr >> 3);
2598 mask = 1 << (7 - (nr & 0x07));
2599 *addr ^= mask;
2600}
2601
2602/*
2603 * On-disk inode flags (f2fs_inode::i_flags)
2604 */
2605#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2606#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2607#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2608#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2609#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2610#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2611#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2612#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2613#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2614#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2615#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2616
2617/* Flags that should be inherited by new inodes from their parent. */
2618#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2619 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2620 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2621
2622/* Flags that are appropriate for regular files (all but dir-specific ones). */
2623#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2624 F2FS_CASEFOLD_FL))
2625
2626/* Flags that are appropriate for non-directories/regular files. */
2627#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2628
2629static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2630{
2631 if (S_ISDIR(mode))
2632 return flags;
2633 else if (S_ISREG(mode))
2634 return flags & F2FS_REG_FLMASK;
2635 else
2636 return flags & F2FS_OTHER_FLMASK;
2637}
2638
2639static inline void __mark_inode_dirty_flag(struct inode *inode,
2640 int flag, bool set)
2641{
2642 switch (flag) {
2643 case FI_INLINE_XATTR:
2644 case FI_INLINE_DATA:
2645 case FI_INLINE_DENTRY:
2646 case FI_NEW_INODE:
2647 if (set)
2648 return;
2649 fallthrough;
2650 case FI_DATA_EXIST:
2651 case FI_INLINE_DOTS:
2652 case FI_PIN_FILE:
2653 f2fs_mark_inode_dirty_sync(inode, true);
2654 }
2655}
2656
2657static inline void set_inode_flag(struct inode *inode, int flag)
2658{
2659 set_bit(flag, F2FS_I(inode)->flags);
2660 __mark_inode_dirty_flag(inode, flag, true);
2661}
2662
2663static inline int is_inode_flag_set(struct inode *inode, int flag)
2664{
2665 return test_bit(flag, F2FS_I(inode)->flags);
2666}
2667
2668static inline void clear_inode_flag(struct inode *inode, int flag)
2669{
2670 clear_bit(flag, F2FS_I(inode)->flags);
2671 __mark_inode_dirty_flag(inode, flag, false);
2672}
2673
2674static inline bool f2fs_verity_in_progress(struct inode *inode)
2675{
2676 return IS_ENABLED(CONFIG_FS_VERITY) &&
2677 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2678}
2679
2680static inline void set_acl_inode(struct inode *inode, umode_t mode)
2681{
2682 F2FS_I(inode)->i_acl_mode = mode;
2683 set_inode_flag(inode, FI_ACL_MODE);
2684 f2fs_mark_inode_dirty_sync(inode, false);
2685}
2686
2687static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2688{
2689 if (inc)
2690 inc_nlink(inode);
2691 else
2692 drop_nlink(inode);
2693 f2fs_mark_inode_dirty_sync(inode, true);
2694}
2695
2696static inline void f2fs_i_blocks_write(struct inode *inode,
2697 block_t diff, bool add, bool claim)
2698{
2699 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2700 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2701
2702 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2703 if (add) {
2704 if (claim)
2705 dquot_claim_block(inode, diff);
2706 else
2707 dquot_alloc_block_nofail(inode, diff);
2708 } else {
2709 dquot_free_block(inode, diff);
2710 }
2711
2712 f2fs_mark_inode_dirty_sync(inode, true);
2713 if (clean || recover)
2714 set_inode_flag(inode, FI_AUTO_RECOVER);
2715}
2716
2717static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2718{
2719 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2720 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2721
2722 if (i_size_read(inode) == i_size)
2723 return;
2724
2725 i_size_write(inode, i_size);
2726 f2fs_mark_inode_dirty_sync(inode, true);
2727 if (clean || recover)
2728 set_inode_flag(inode, FI_AUTO_RECOVER);
2729}
2730
2731static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2732{
2733 F2FS_I(inode)->i_current_depth = depth;
2734 f2fs_mark_inode_dirty_sync(inode, true);
2735}
2736
2737static inline void f2fs_i_gc_failures_write(struct inode *inode,
2738 unsigned int count)
2739{
2740 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2741 f2fs_mark_inode_dirty_sync(inode, true);
2742}
2743
2744static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2745{
2746 F2FS_I(inode)->i_xattr_nid = xnid;
2747 f2fs_mark_inode_dirty_sync(inode, true);
2748}
2749
2750static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2751{
2752 F2FS_I(inode)->i_pino = pino;
2753 f2fs_mark_inode_dirty_sync(inode, true);
2754}
2755
2756static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2757{
2758 struct f2fs_inode_info *fi = F2FS_I(inode);
2759
2760 if (ri->i_inline & F2FS_INLINE_XATTR)
2761 set_bit(FI_INLINE_XATTR, fi->flags);
2762 if (ri->i_inline & F2FS_INLINE_DATA)
2763 set_bit(FI_INLINE_DATA, fi->flags);
2764 if (ri->i_inline & F2FS_INLINE_DENTRY)
2765 set_bit(FI_INLINE_DENTRY, fi->flags);
2766 if (ri->i_inline & F2FS_DATA_EXIST)
2767 set_bit(FI_DATA_EXIST, fi->flags);
2768 if (ri->i_inline & F2FS_INLINE_DOTS)
2769 set_bit(FI_INLINE_DOTS, fi->flags);
2770 if (ri->i_inline & F2FS_EXTRA_ATTR)
2771 set_bit(FI_EXTRA_ATTR, fi->flags);
2772 if (ri->i_inline & F2FS_PIN_FILE)
2773 set_bit(FI_PIN_FILE, fi->flags);
2774}
2775
2776static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2777{
2778 ri->i_inline = 0;
2779
2780 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2781 ri->i_inline |= F2FS_INLINE_XATTR;
2782 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2783 ri->i_inline |= F2FS_INLINE_DATA;
2784 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2785 ri->i_inline |= F2FS_INLINE_DENTRY;
2786 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2787 ri->i_inline |= F2FS_DATA_EXIST;
2788 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2789 ri->i_inline |= F2FS_INLINE_DOTS;
2790 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2791 ri->i_inline |= F2FS_EXTRA_ATTR;
2792 if (is_inode_flag_set(inode, FI_PIN_FILE))
2793 ri->i_inline |= F2FS_PIN_FILE;
2794}
2795
2796static inline int f2fs_has_extra_attr(struct inode *inode)
2797{
2798 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2799}
2800
2801static inline int f2fs_has_inline_xattr(struct inode *inode)
2802{
2803 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2804}
2805
2806static inline int f2fs_compressed_file(struct inode *inode)
2807{
2808 return S_ISREG(inode->i_mode) &&
2809 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2810}
2811
2812static inline unsigned int addrs_per_inode(struct inode *inode)
2813{
2814 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2815 get_inline_xattr_addrs(inode);
2816
2817 if (!f2fs_compressed_file(inode))
2818 return addrs;
2819 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2820}
2821
2822static inline unsigned int addrs_per_block(struct inode *inode)
2823{
2824 if (!f2fs_compressed_file(inode))
2825 return DEF_ADDRS_PER_BLOCK;
2826 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2827}
2828
2829static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2830{
2831 struct f2fs_inode *ri = F2FS_INODE(page);
2832
2833 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2834 get_inline_xattr_addrs(inode)]);
2835}
2836
2837static inline int inline_xattr_size(struct inode *inode)
2838{
2839 if (f2fs_has_inline_xattr(inode))
2840 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2841 return 0;
2842}
2843
2844static inline int f2fs_has_inline_data(struct inode *inode)
2845{
2846 return is_inode_flag_set(inode, FI_INLINE_DATA);
2847}
2848
2849static inline int f2fs_exist_data(struct inode *inode)
2850{
2851 return is_inode_flag_set(inode, FI_DATA_EXIST);
2852}
2853
2854static inline int f2fs_has_inline_dots(struct inode *inode)
2855{
2856 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2857}
2858
2859static inline int f2fs_is_mmap_file(struct inode *inode)
2860{
2861 return is_inode_flag_set(inode, FI_MMAP_FILE);
2862}
2863
2864static inline bool f2fs_is_pinned_file(struct inode *inode)
2865{
2866 return is_inode_flag_set(inode, FI_PIN_FILE);
2867}
2868
2869static inline bool f2fs_is_atomic_file(struct inode *inode)
2870{
2871 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2872}
2873
2874static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2875{
2876 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2877}
2878
2879static inline bool f2fs_is_volatile_file(struct inode *inode)
2880{
2881 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2882}
2883
2884static inline bool f2fs_is_first_block_written(struct inode *inode)
2885{
2886 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2887}
2888
2889static inline bool f2fs_is_drop_cache(struct inode *inode)
2890{
2891 return is_inode_flag_set(inode, FI_DROP_CACHE);
2892}
2893
2894static inline void *inline_data_addr(struct inode *inode, struct page *page)
2895{
2896 struct f2fs_inode *ri = F2FS_INODE(page);
2897 int extra_size = get_extra_isize(inode);
2898
2899 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2900}
2901
2902static inline int f2fs_has_inline_dentry(struct inode *inode)
2903{
2904 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2905}
2906
2907static inline int is_file(struct inode *inode, int type)
2908{
2909 return F2FS_I(inode)->i_advise & type;
2910}
2911
2912static inline void set_file(struct inode *inode, int type)
2913{
2914 F2FS_I(inode)->i_advise |= type;
2915 f2fs_mark_inode_dirty_sync(inode, true);
2916}
2917
2918static inline void clear_file(struct inode *inode, int type)
2919{
2920 F2FS_I(inode)->i_advise &= ~type;
2921 f2fs_mark_inode_dirty_sync(inode, true);
2922}
2923
2924static inline bool f2fs_is_time_consistent(struct inode *inode)
2925{
2926 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2927 return false;
2928 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2929 return false;
2930 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2931 return false;
2932 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2933 &F2FS_I(inode)->i_crtime))
2934 return false;
2935 return true;
2936}
2937
2938static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2939{
2940 bool ret;
2941
2942 if (dsync) {
2943 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2944
2945 spin_lock(&sbi->inode_lock[DIRTY_META]);
2946 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2947 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2948 return ret;
2949 }
2950 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2951 file_keep_isize(inode) ||
2952 i_size_read(inode) & ~PAGE_MASK)
2953 return false;
2954
2955 if (!f2fs_is_time_consistent(inode))
2956 return false;
2957
2958 spin_lock(&F2FS_I(inode)->i_size_lock);
2959 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2960 spin_unlock(&F2FS_I(inode)->i_size_lock);
2961
2962 return ret;
2963}
2964
2965static inline bool f2fs_readonly(struct super_block *sb)
2966{
2967 return sb_rdonly(sb);
2968}
2969
2970static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2971{
2972 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2973}
2974
2975static inline bool is_dot_dotdot(const u8 *name, size_t len)
2976{
2977 if (len == 1 && name[0] == '.')
2978 return true;
2979
2980 if (len == 2 && name[0] == '.' && name[1] == '.')
2981 return true;
2982
2983 return false;
2984}
2985
2986static inline bool f2fs_may_extent_tree(struct inode *inode)
2987{
2988 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2989
2990 if (!test_opt(sbi, EXTENT_CACHE) ||
2991 is_inode_flag_set(inode, FI_NO_EXTENT) ||
2992 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
2993 return false;
2994
2995 /*
2996 * for recovered files during mount do not create extents
2997 * if shrinker is not registered.
2998 */
2999 if (list_empty(&sbi->s_list))
3000 return false;
3001
3002 return S_ISREG(inode->i_mode);
3003}
3004
3005static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3006 size_t size, gfp_t flags)
3007{
3008 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3009 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3010 return NULL;
3011 }
3012
3013 return kmalloc(size, flags);
3014}
3015
3016static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3017 size_t size, gfp_t flags)
3018{
3019 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3020}
3021
3022static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3023 size_t size, gfp_t flags)
3024{
3025 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3026 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3027 return NULL;
3028 }
3029
3030 return kvmalloc(size, flags);
3031}
3032
3033static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3034 size_t size, gfp_t flags)
3035{
3036 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3037}
3038
3039static inline int get_extra_isize(struct inode *inode)
3040{
3041 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3042}
3043
3044static inline int get_inline_xattr_addrs(struct inode *inode)
3045{
3046 return F2FS_I(inode)->i_inline_xattr_size;
3047}
3048
3049#define f2fs_get_inode_mode(i) \
3050 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3051 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3052
3053#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3054 (offsetof(struct f2fs_inode, i_extra_end) - \
3055 offsetof(struct f2fs_inode, i_extra_isize)) \
3056
3057#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3058#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3059 ((offsetof(typeof(*(f2fs_inode)), field) + \
3060 sizeof((f2fs_inode)->field)) \
3061 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3062
3063#define DEFAULT_IOSTAT_PERIOD_MS 3000
3064#define MIN_IOSTAT_PERIOD_MS 100
3065/* maximum period of iostat tracing is 1 day */
3066#define MAX_IOSTAT_PERIOD_MS 8640000
3067
3068static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3069{
3070 int i;
3071
3072 spin_lock(&sbi->iostat_lock);
3073 for (i = 0; i < NR_IO_TYPE; i++) {
3074 sbi->rw_iostat[i] = 0;
3075 sbi->prev_rw_iostat[i] = 0;
3076 }
3077 spin_unlock(&sbi->iostat_lock);
3078}
3079
3080extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3081
3082static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3083 enum iostat_type type, unsigned long long io_bytes)
3084{
3085 if (!sbi->iostat_enable)
3086 return;
3087 spin_lock(&sbi->iostat_lock);
3088 sbi->rw_iostat[type] += io_bytes;
3089
3090 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3091 sbi->rw_iostat[APP_BUFFERED_IO] =
3092 sbi->rw_iostat[APP_WRITE_IO] -
3093 sbi->rw_iostat[APP_DIRECT_IO];
3094
3095 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3096 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3097 sbi->rw_iostat[APP_READ_IO] -
3098 sbi->rw_iostat[APP_DIRECT_READ_IO];
3099 spin_unlock(&sbi->iostat_lock);
3100
3101 f2fs_record_iostat(sbi);
3102}
3103
3104#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3105
3106#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3107
3108bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3109 block_t blkaddr, int type);
3110static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3111 block_t blkaddr, int type)
3112{
3113 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3114 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3115 blkaddr, type);
3116 f2fs_bug_on(sbi, 1);
3117 }
3118}
3119
3120static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3121{
3122 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3123 blkaddr == COMPRESS_ADDR)
3124 return false;
3125 return true;
3126}
3127
3128static inline void f2fs_set_page_private(struct page *page,
3129 unsigned long data)
3130{
3131 if (PagePrivate(page))
3132 return;
3133
3134 attach_page_private(page, (void *)data);
3135}
3136
3137static inline void f2fs_clear_page_private(struct page *page)
3138{
3139 detach_page_private(page);
3140}
3141
3142/*
3143 * file.c
3144 */
3145int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3146void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3147int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3148int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3149int f2fs_truncate(struct inode *inode);
3150int f2fs_getattr(const struct path *path, struct kstat *stat,
3151 u32 request_mask, unsigned int flags);
3152int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3153int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3154void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3155int f2fs_precache_extents(struct inode *inode);
3156long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3157long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3158int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3159int f2fs_pin_file_control(struct inode *inode, bool inc);
3160
3161/*
3162 * inode.c
3163 */
3164void f2fs_set_inode_flags(struct inode *inode);
3165bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3166void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3167struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3168struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3169int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3170void f2fs_update_inode(struct inode *inode, struct page *node_page);
3171void f2fs_update_inode_page(struct inode *inode);
3172int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3173void f2fs_evict_inode(struct inode *inode);
3174void f2fs_handle_failed_inode(struct inode *inode);
3175
3176/*
3177 * namei.c
3178 */
3179int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3180 bool hot, bool set);
3181struct dentry *f2fs_get_parent(struct dentry *child);
3182
3183/*
3184 * dir.c
3185 */
3186unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3187int f2fs_init_casefolded_name(const struct inode *dir,
3188 struct f2fs_filename *fname);
3189int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3190 int lookup, struct f2fs_filename *fname);
3191int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3192 struct f2fs_filename *fname);
3193void f2fs_free_filename(struct f2fs_filename *fname);
3194struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3195 const struct f2fs_filename *fname, int *max_slots);
3196int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3197 unsigned int start_pos, struct fscrypt_str *fstr);
3198void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3199 struct f2fs_dentry_ptr *d);
3200struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3201 const struct f2fs_filename *fname, struct page *dpage);
3202void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3203 unsigned int current_depth);
3204int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3205void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3206struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3207 const struct f2fs_filename *fname,
3208 struct page **res_page);
3209struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3210 const struct qstr *child, struct page **res_page);
3211struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3212ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3213 struct page **page);
3214void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3215 struct page *page, struct inode *inode);
3216bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3217 const struct f2fs_filename *fname);
3218void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3219 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3220 unsigned int bit_pos);
3221int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3222 struct inode *inode, nid_t ino, umode_t mode);
3223int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3224 struct inode *inode, nid_t ino, umode_t mode);
3225int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3226 struct inode *inode, nid_t ino, umode_t mode);
3227void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3228 struct inode *dir, struct inode *inode);
3229int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3230bool f2fs_empty_dir(struct inode *dir);
3231
3232static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3233{
3234 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3235 inode, inode->i_ino, inode->i_mode);
3236}
3237
3238/*
3239 * super.c
3240 */
3241int f2fs_inode_dirtied(struct inode *inode, bool sync);
3242void f2fs_inode_synced(struct inode *inode);
3243int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3244int f2fs_quota_sync(struct super_block *sb, int type);
3245void f2fs_quota_off_umount(struct super_block *sb);
3246int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3247int f2fs_sync_fs(struct super_block *sb, int sync);
3248int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3249
3250/*
3251 * hash.c
3252 */
3253void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3254
3255/*
3256 * node.c
3257 */
3258struct dnode_of_data;
3259struct node_info;
3260
3261int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3262bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3263bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3264void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3265void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3266void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3267int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3268bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3269bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3270int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3271 struct node_info *ni);
3272pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3273int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3274int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3275int f2fs_truncate_xattr_node(struct inode *inode);
3276int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3277 unsigned int seq_id);
3278int f2fs_remove_inode_page(struct inode *inode);
3279struct page *f2fs_new_inode_page(struct inode *inode);
3280struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3281void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3282struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3283struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3284int f2fs_move_node_page(struct page *node_page, int gc_type);
3285void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3286int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3287 struct writeback_control *wbc, bool atomic,
3288 unsigned int *seq_id);
3289int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3290 struct writeback_control *wbc,
3291 bool do_balance, enum iostat_type io_type);
3292int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3293bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3294void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3295void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3296int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3297int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3298int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3299int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3300int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3301 unsigned int segno, struct f2fs_summary_block *sum);
3302int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3303int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3304void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3305int __init f2fs_create_node_manager_caches(void);
3306void f2fs_destroy_node_manager_caches(void);
3307
3308/*
3309 * segment.c
3310 */
3311bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3312void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3313void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3314void f2fs_drop_inmem_pages(struct inode *inode);
3315void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3316int f2fs_commit_inmem_pages(struct inode *inode);
3317void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3318void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3319int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3320int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3321int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3322void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3323void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3324bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3325void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3326void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3327bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3328void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3329 struct cp_control *cpc);
3330void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3331block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3332int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3333void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3334int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3335void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3336 unsigned int start, unsigned int end);
3337void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type);
3338void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3339int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3340bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3341 struct cp_control *cpc);
3342struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3343void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3344 block_t blk_addr);
3345void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3346 enum iostat_type io_type);
3347void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3348void f2fs_outplace_write_data(struct dnode_of_data *dn,
3349 struct f2fs_io_info *fio);
3350int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3351void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3352 block_t old_blkaddr, block_t new_blkaddr,
3353 bool recover_curseg, bool recover_newaddr);
3354void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3355 block_t old_addr, block_t new_addr,
3356 unsigned char version, bool recover_curseg,
3357 bool recover_newaddr);
3358void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3359 block_t old_blkaddr, block_t *new_blkaddr,
3360 struct f2fs_summary *sum, int type,
3361 struct f2fs_io_info *fio);
3362void f2fs_wait_on_page_writeback(struct page *page,
3363 enum page_type type, bool ordered, bool locked);
3364void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3365void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3366 block_t len);
3367void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3368void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3369int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3370 unsigned int val, int alloc);
3371void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3372int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3373int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3374int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3375void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3376int __init f2fs_create_segment_manager_caches(void);
3377void f2fs_destroy_segment_manager_caches(void);
3378int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3379enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3380 enum page_type type, enum temp_type temp);
3381
3382/*
3383 * checkpoint.c
3384 */
3385void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3386struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3387struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3388struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3389struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3390bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3391 block_t blkaddr, int type);
3392int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3393 int type, bool sync);
3394void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3395long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3396 long nr_to_write, enum iostat_type io_type);
3397void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3398void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3399void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3400bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3401void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3402 unsigned int devidx, int type);
3403bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3404 unsigned int devidx, int type);
3405int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3406int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3407void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3408void f2fs_add_orphan_inode(struct inode *inode);
3409void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3410int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3411int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3412void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3413void f2fs_remove_dirty_inode(struct inode *inode);
3414int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3415void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3416int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3417void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3418int __init f2fs_create_checkpoint_caches(void);
3419void f2fs_destroy_checkpoint_caches(void);
3420
3421/*
3422 * data.c
3423 */
3424int __init f2fs_init_bioset(void);
3425void f2fs_destroy_bioset(void);
3426struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio);
3427int f2fs_init_bio_entry_cache(void);
3428void f2fs_destroy_bio_entry_cache(void);
3429void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3430 struct bio *bio, enum page_type type);
3431void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3432void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3433 struct inode *inode, struct page *page,
3434 nid_t ino, enum page_type type);
3435void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3436 struct bio **bio, struct page *page);
3437void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3438int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3439int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3440void f2fs_submit_page_write(struct f2fs_io_info *fio);
3441struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3442 block_t blk_addr, struct bio *bio);
3443int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3444void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3445void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3446int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3447int f2fs_reserve_new_block(struct dnode_of_data *dn);
3448int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3449int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3450int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3451struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3452 int op_flags, bool for_write);
3453struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3454struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3455 bool for_write);
3456struct page *f2fs_get_new_data_page(struct inode *inode,
3457 struct page *ipage, pgoff_t index, bool new_i_size);
3458int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3459void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3460int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3461 int create, int flag);
3462int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3463 u64 start, u64 len);
3464int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3465bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3466bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3467int f2fs_write_single_data_page(struct page *page, int *submitted,
3468 struct bio **bio, sector_t *last_block,
3469 struct writeback_control *wbc,
3470 enum iostat_type io_type,
3471 int compr_blocks);
3472void f2fs_invalidate_page(struct page *page, unsigned int offset,
3473 unsigned int length);
3474int f2fs_release_page(struct page *page, gfp_t wait);
3475#ifdef CONFIG_MIGRATION
3476int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3477 struct page *page, enum migrate_mode mode);
3478#endif
3479bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3480void f2fs_clear_page_cache_dirty_tag(struct page *page);
3481int f2fs_init_post_read_processing(void);
3482void f2fs_destroy_post_read_processing(void);
3483int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3484void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3485
3486/*
3487 * gc.c
3488 */
3489int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3490void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3491block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3492int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3493 unsigned int segno);
3494void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3495int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3496
3497/*
3498 * recovery.c
3499 */
3500int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3501bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3502
3503/*
3504 * debug.c
3505 */
3506#ifdef CONFIG_F2FS_STAT_FS
3507struct f2fs_stat_info {
3508 struct list_head stat_list;
3509 struct f2fs_sb_info *sbi;
3510 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3511 int main_area_segs, main_area_sections, main_area_zones;
3512 unsigned long long hit_largest, hit_cached, hit_rbtree;
3513 unsigned long long hit_total, total_ext;
3514 int ext_tree, zombie_tree, ext_node;
3515 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3516 int ndirty_data, ndirty_qdata;
3517 int inmem_pages;
3518 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3519 int nats, dirty_nats, sits, dirty_sits;
3520 int free_nids, avail_nids, alloc_nids;
3521 int total_count, utilization;
3522 int bg_gc, nr_wb_cp_data, nr_wb_data;
3523 int nr_rd_data, nr_rd_node, nr_rd_meta;
3524 int nr_dio_read, nr_dio_write;
3525 unsigned int io_skip_bggc, other_skip_bggc;
3526 int nr_flushing, nr_flushed, flush_list_empty;
3527 int nr_discarding, nr_discarded;
3528 int nr_discard_cmd;
3529 unsigned int undiscard_blks;
3530 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3531 int compr_inode, compr_blocks;
3532 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3533 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3534 unsigned int bimodal, avg_vblocks;
3535 int util_free, util_valid, util_invalid;
3536 int rsvd_segs, overp_segs;
3537 int dirty_count, node_pages, meta_pages;
3538 int prefree_count, call_count, cp_count, bg_cp_count;
3539 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3540 int bg_node_segs, bg_data_segs;
3541 int tot_blks, data_blks, node_blks;
3542 int bg_data_blks, bg_node_blks;
3543 unsigned long long skipped_atomic_files[2];
3544 int curseg[NR_CURSEG_TYPE];
3545 int cursec[NR_CURSEG_TYPE];
3546 int curzone[NR_CURSEG_TYPE];
3547 unsigned int dirty_seg[NR_CURSEG_TYPE];
3548 unsigned int full_seg[NR_CURSEG_TYPE];
3549 unsigned int valid_blks[NR_CURSEG_TYPE];
3550
3551 unsigned int meta_count[META_MAX];
3552 unsigned int segment_count[2];
3553 unsigned int block_count[2];
3554 unsigned int inplace_count;
3555 unsigned long long base_mem, cache_mem, page_mem;
3556};
3557
3558static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3559{
3560 return (struct f2fs_stat_info *)sbi->stat_info;
3561}
3562
3563#define stat_inc_cp_count(si) ((si)->cp_count++)
3564#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3565#define stat_inc_call_count(si) ((si)->call_count++)
3566#define stat_inc_bggc_count(si) ((si)->bg_gc++)
3567#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3568#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3569#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3570#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3571#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3572#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3573#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3574#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3575#define stat_inc_inline_xattr(inode) \
3576 do { \
3577 if (f2fs_has_inline_xattr(inode)) \
3578 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3579 } while (0)
3580#define stat_dec_inline_xattr(inode) \
3581 do { \
3582 if (f2fs_has_inline_xattr(inode)) \
3583 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3584 } while (0)
3585#define stat_inc_inline_inode(inode) \
3586 do { \
3587 if (f2fs_has_inline_data(inode)) \
3588 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3589 } while (0)
3590#define stat_dec_inline_inode(inode) \
3591 do { \
3592 if (f2fs_has_inline_data(inode)) \
3593 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3594 } while (0)
3595#define stat_inc_inline_dir(inode) \
3596 do { \
3597 if (f2fs_has_inline_dentry(inode)) \
3598 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3599 } while (0)
3600#define stat_dec_inline_dir(inode) \
3601 do { \
3602 if (f2fs_has_inline_dentry(inode)) \
3603 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3604 } while (0)
3605#define stat_inc_compr_inode(inode) \
3606 do { \
3607 if (f2fs_compressed_file(inode)) \
3608 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3609 } while (0)
3610#define stat_dec_compr_inode(inode) \
3611 do { \
3612 if (f2fs_compressed_file(inode)) \
3613 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3614 } while (0)
3615#define stat_add_compr_blocks(inode, blocks) \
3616 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3617#define stat_sub_compr_blocks(inode, blocks) \
3618 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3619#define stat_inc_meta_count(sbi, blkaddr) \
3620 do { \
3621 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3622 atomic_inc(&(sbi)->meta_count[META_CP]); \
3623 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3624 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3625 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3626 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3627 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3628 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3629 } while (0)
3630#define stat_inc_seg_type(sbi, curseg) \
3631 ((sbi)->segment_count[(curseg)->alloc_type]++)
3632#define stat_inc_block_count(sbi, curseg) \
3633 ((sbi)->block_count[(curseg)->alloc_type]++)
3634#define stat_inc_inplace_blocks(sbi) \
3635 (atomic_inc(&(sbi)->inplace_count))
3636#define stat_update_max_atomic_write(inode) \
3637 do { \
3638 int cur = F2FS_I_SB(inode)->atomic_files; \
3639 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3640 if (cur > max) \
3641 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3642 } while (0)
3643#define stat_inc_volatile_write(inode) \
3644 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3645#define stat_dec_volatile_write(inode) \
3646 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3647#define stat_update_max_volatile_write(inode) \
3648 do { \
3649 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3650 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3651 if (cur > max) \
3652 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3653 } while (0)
3654#define stat_inc_seg_count(sbi, type, gc_type) \
3655 do { \
3656 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3657 si->tot_segs++; \
3658 if ((type) == SUM_TYPE_DATA) { \
3659 si->data_segs++; \
3660 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3661 } else { \
3662 si->node_segs++; \
3663 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3664 } \
3665 } while (0)
3666
3667#define stat_inc_tot_blk_count(si, blks) \
3668 ((si)->tot_blks += (blks))
3669
3670#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3671 do { \
3672 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3673 stat_inc_tot_blk_count(si, blks); \
3674 si->data_blks += (blks); \
3675 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3676 } while (0)
3677
3678#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3679 do { \
3680 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3681 stat_inc_tot_blk_count(si, blks); \
3682 si->node_blks += (blks); \
3683 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3684 } while (0)
3685
3686int f2fs_build_stats(struct f2fs_sb_info *sbi);
3687void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3688void __init f2fs_create_root_stats(void);
3689void f2fs_destroy_root_stats(void);
3690void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3691#else
3692#define stat_inc_cp_count(si) do { } while (0)
3693#define stat_inc_bg_cp_count(si) do { } while (0)
3694#define stat_inc_call_count(si) do { } while (0)
3695#define stat_inc_bggc_count(si) do { } while (0)
3696#define stat_io_skip_bggc_count(sbi) do { } while (0)
3697#define stat_other_skip_bggc_count(sbi) do { } while (0)
3698#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3699#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3700#define stat_inc_total_hit(sbi) do { } while (0)
3701#define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3702#define stat_inc_largest_node_hit(sbi) do { } while (0)
3703#define stat_inc_cached_node_hit(sbi) do { } while (0)
3704#define stat_inc_inline_xattr(inode) do { } while (0)
3705#define stat_dec_inline_xattr(inode) do { } while (0)
3706#define stat_inc_inline_inode(inode) do { } while (0)
3707#define stat_dec_inline_inode(inode) do { } while (0)
3708#define stat_inc_inline_dir(inode) do { } while (0)
3709#define stat_dec_inline_dir(inode) do { } while (0)
3710#define stat_inc_compr_inode(inode) do { } while (0)
3711#define stat_dec_compr_inode(inode) do { } while (0)
3712#define stat_add_compr_blocks(inode, blocks) do { } while (0)
3713#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3714#define stat_inc_atomic_write(inode) do { } while (0)
3715#define stat_dec_atomic_write(inode) do { } while (0)
3716#define stat_update_max_atomic_write(inode) do { } while (0)
3717#define stat_inc_volatile_write(inode) do { } while (0)
3718#define stat_dec_volatile_write(inode) do { } while (0)
3719#define stat_update_max_volatile_write(inode) do { } while (0)
3720#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3721#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3722#define stat_inc_block_count(sbi, curseg) do { } while (0)
3723#define stat_inc_inplace_blocks(sbi) do { } while (0)
3724#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3725#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3726#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3727#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3728
3729static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3730static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3731static inline void __init f2fs_create_root_stats(void) { }
3732static inline void f2fs_destroy_root_stats(void) { }
3733static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3734#endif
3735
3736extern const struct file_operations f2fs_dir_operations;
3737#ifdef CONFIG_UNICODE
3738extern const struct dentry_operations f2fs_dentry_ops;
3739#endif
3740extern const struct file_operations f2fs_file_operations;
3741extern const struct inode_operations f2fs_file_inode_operations;
3742extern const struct address_space_operations f2fs_dblock_aops;
3743extern const struct address_space_operations f2fs_node_aops;
3744extern const struct address_space_operations f2fs_meta_aops;
3745extern const struct inode_operations f2fs_dir_inode_operations;
3746extern const struct inode_operations f2fs_symlink_inode_operations;
3747extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3748extern const struct inode_operations f2fs_special_inode_operations;
3749extern struct kmem_cache *f2fs_inode_entry_slab;
3750
3751/*
3752 * inline.c
3753 */
3754bool f2fs_may_inline_data(struct inode *inode);
3755bool f2fs_may_inline_dentry(struct inode *inode);
3756void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3757void f2fs_truncate_inline_inode(struct inode *inode,
3758 struct page *ipage, u64 from);
3759int f2fs_read_inline_data(struct inode *inode, struct page *page);
3760int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3761int f2fs_convert_inline_inode(struct inode *inode);
3762int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3763int f2fs_write_inline_data(struct inode *inode, struct page *page);
3764int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3765struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3766 const struct f2fs_filename *fname,
3767 struct page **res_page);
3768int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3769 struct page *ipage);
3770int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3771 struct inode *inode, nid_t ino, umode_t mode);
3772void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3773 struct page *page, struct inode *dir,
3774 struct inode *inode);
3775bool f2fs_empty_inline_dir(struct inode *dir);
3776int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3777 struct fscrypt_str *fstr);
3778int f2fs_inline_data_fiemap(struct inode *inode,
3779 struct fiemap_extent_info *fieinfo,
3780 __u64 start, __u64 len);
3781
3782/*
3783 * shrinker.c
3784 */
3785unsigned long f2fs_shrink_count(struct shrinker *shrink,
3786 struct shrink_control *sc);
3787unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3788 struct shrink_control *sc);
3789void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3790void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3791
3792/*
3793 * extent_cache.c
3794 */
3795struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3796 struct rb_entry *cached_re, unsigned int ofs);
3797struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3798 struct rb_root_cached *root,
3799 struct rb_node **parent,
3800 unsigned int ofs, bool *leftmost);
3801struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3802 struct rb_entry *cached_re, unsigned int ofs,
3803 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3804 struct rb_node ***insert_p, struct rb_node **insert_parent,
3805 bool force, bool *leftmost);
3806bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3807 struct rb_root_cached *root);
3808unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3809void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3810void f2fs_drop_extent_tree(struct inode *inode);
3811unsigned int f2fs_destroy_extent_node(struct inode *inode);
3812void f2fs_destroy_extent_tree(struct inode *inode);
3813bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3814 struct extent_info *ei);
3815void f2fs_update_extent_cache(struct dnode_of_data *dn);
3816void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3817 pgoff_t fofs, block_t blkaddr, unsigned int len);
3818void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3819int __init f2fs_create_extent_cache(void);
3820void f2fs_destroy_extent_cache(void);
3821
3822/*
3823 * sysfs.c
3824 */
3825int __init f2fs_init_sysfs(void);
3826void f2fs_exit_sysfs(void);
3827int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3828void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3829
3830/* verity.c */
3831extern const struct fsverity_operations f2fs_verityops;
3832
3833/*
3834 * crypto support
3835 */
3836static inline bool f2fs_encrypted_file(struct inode *inode)
3837{
3838 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3839}
3840
3841static inline void f2fs_set_encrypted_inode(struct inode *inode)
3842{
3843#ifdef CONFIG_FS_ENCRYPTION
3844 file_set_encrypt(inode);
3845 f2fs_set_inode_flags(inode);
3846#endif
3847}
3848
3849/*
3850 * Returns true if the reads of the inode's data need to undergo some
3851 * postprocessing step, like decryption or authenticity verification.
3852 */
3853static inline bool f2fs_post_read_required(struct inode *inode)
3854{
3855 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3856 f2fs_compressed_file(inode);
3857}
3858
3859/*
3860 * compress.c
3861 */
3862#ifdef CONFIG_F2FS_FS_COMPRESSION
3863bool f2fs_is_compressed_page(struct page *page);
3864struct page *f2fs_compress_control_page(struct page *page);
3865int f2fs_prepare_compress_overwrite(struct inode *inode,
3866 struct page **pagep, pgoff_t index, void **fsdata);
3867bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3868 pgoff_t index, unsigned copied);
3869int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3870void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3871bool f2fs_is_compress_backend_ready(struct inode *inode);
3872int f2fs_init_compress_mempool(void);
3873void f2fs_destroy_compress_mempool(void);
3874void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity);
3875bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3876bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3877void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3878int f2fs_write_multi_pages(struct compress_ctx *cc,
3879 int *submitted,
3880 struct writeback_control *wbc,
3881 enum iostat_type io_type);
3882int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3883int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3884 unsigned nr_pages, sector_t *last_block_in_bio,
3885 bool is_readahead, bool for_write);
3886struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3887void f2fs_free_dic(struct decompress_io_ctx *dic);
3888void f2fs_decompress_end_io(struct page **rpages,
3889 unsigned int cluster_size, bool err, bool verity);
3890int f2fs_init_compress_ctx(struct compress_ctx *cc);
3891void f2fs_destroy_compress_ctx(struct compress_ctx *cc);
3892void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3893#else
3894static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3895static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3896{
3897 if (!f2fs_compressed_file(inode))
3898 return true;
3899 /* not support compression */
3900 return false;
3901}
3902static inline struct page *f2fs_compress_control_page(struct page *page)
3903{
3904 WARN_ON_ONCE(1);
3905 return ERR_PTR(-EINVAL);
3906}
3907static inline int f2fs_init_compress_mempool(void) { return 0; }
3908static inline void f2fs_destroy_compress_mempool(void) { }
3909#endif
3910
3911static inline void set_compress_context(struct inode *inode)
3912{
3913 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3914
3915 F2FS_I(inode)->i_compress_algorithm =
3916 F2FS_OPTION(sbi).compress_algorithm;
3917 F2FS_I(inode)->i_log_cluster_size =
3918 F2FS_OPTION(sbi).compress_log_size;
3919 F2FS_I(inode)->i_cluster_size =
3920 1 << F2FS_I(inode)->i_log_cluster_size;
3921 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
3922 set_inode_flag(inode, FI_COMPRESSED_FILE);
3923 stat_inc_compr_inode(inode);
3924 f2fs_mark_inode_dirty_sync(inode, true);
3925}
3926
3927static inline u64 f2fs_disable_compressed_file(struct inode *inode)
3928{
3929 struct f2fs_inode_info *fi = F2FS_I(inode);
3930
3931 if (!f2fs_compressed_file(inode))
3932 return 0;
3933 if (S_ISREG(inode->i_mode)) {
3934 if (get_dirty_pages(inode))
3935 return 1;
3936 if (fi->i_compr_blocks)
3937 return fi->i_compr_blocks;
3938 }
3939
3940 fi->i_flags &= ~F2FS_COMPR_FL;
3941 stat_dec_compr_inode(inode);
3942 clear_inode_flag(inode, FI_COMPRESSED_FILE);
3943 f2fs_mark_inode_dirty_sync(inode, true);
3944 return 0;
3945}
3946
3947#define F2FS_FEATURE_FUNCS(name, flagname) \
3948static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3949{ \
3950 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3951}
3952
3953F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3954F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3955F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3956F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3957F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3958F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3959F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3960F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3961F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3962F2FS_FEATURE_FUNCS(verity, VERITY);
3963F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3964F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
3965F2FS_FEATURE_FUNCS(compression, COMPRESSION);
3966
3967#ifdef CONFIG_BLK_DEV_ZONED
3968static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
3969 block_t blkaddr)
3970{
3971 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3972
3973 return test_bit(zno, FDEV(devi).blkz_seq);
3974}
3975#endif
3976
3977static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3978{
3979 return f2fs_sb_has_blkzoned(sbi);
3980}
3981
3982static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
3983{
3984 return blk_queue_discard(bdev_get_queue(bdev)) ||
3985 bdev_is_zoned(bdev);
3986}
3987
3988static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3989{
3990 int i;
3991
3992 if (!f2fs_is_multi_device(sbi))
3993 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
3994
3995 for (i = 0; i < sbi->s_ndevs; i++)
3996 if (f2fs_bdev_support_discard(FDEV(i).bdev))
3997 return true;
3998 return false;
3999}
4000
4001static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4002{
4003 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4004 f2fs_hw_should_discard(sbi);
4005}
4006
4007static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4008{
4009 int i;
4010
4011 if (!f2fs_is_multi_device(sbi))
4012 return bdev_read_only(sbi->sb->s_bdev);
4013
4014 for (i = 0; i < sbi->s_ndevs; i++)
4015 if (bdev_read_only(FDEV(i).bdev))
4016 return true;
4017 return false;
4018}
4019
4020static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4021{
4022 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4023}
4024
4025static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode)
4026{
4027#ifdef CONFIG_FS_ENCRYPTION
4028 struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
4029 umode_t mode = inode->i_mode;
4030
4031 /*
4032 * If the directory encrypted or dummy encryption enabled,
4033 * then we should encrypt the inode.
4034 */
4035 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi))
4036 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
4037#endif
4038 return false;
4039}
4040
4041static inline bool f2fs_may_compress(struct inode *inode)
4042{
4043 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4044 f2fs_is_atomic_file(inode) ||
4045 f2fs_is_volatile_file(inode))
4046 return false;
4047 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4048}
4049
4050static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4051 u64 blocks, bool add)
4052{
4053 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4054
4055 /* don't update i_compr_blocks if saved blocks were released */
4056 if (!add && !F2FS_I(inode)->i_compr_blocks)
4057 return;
4058
4059 if (add) {
4060 F2FS_I(inode)->i_compr_blocks += diff;
4061 stat_add_compr_blocks(inode, diff);
4062 } else {
4063 F2FS_I(inode)->i_compr_blocks -= diff;
4064 stat_sub_compr_blocks(inode, diff);
4065 }
4066 f2fs_mark_inode_dirty_sync(inode, true);
4067}
4068
4069static inline int block_unaligned_IO(struct inode *inode,
4070 struct kiocb *iocb, struct iov_iter *iter)
4071{
4072 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4073 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4074 loff_t offset = iocb->ki_pos;
4075 unsigned long align = offset | iov_iter_alignment(iter);
4076
4077 return align & blocksize_mask;
4078}
4079
4080static inline int allow_outplace_dio(struct inode *inode,
4081 struct kiocb *iocb, struct iov_iter *iter)
4082{
4083 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4084 int rw = iov_iter_rw(iter);
4085
4086 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4087 !block_unaligned_IO(inode, iocb, iter));
4088}
4089
4090static inline bool f2fs_force_buffered_io(struct inode *inode,
4091 struct kiocb *iocb, struct iov_iter *iter)
4092{
4093 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4094 int rw = iov_iter_rw(iter);
4095
4096 if (f2fs_post_read_required(inode))
4097 return true;
4098 if (f2fs_is_multi_device(sbi))
4099 return true;
4100 /*
4101 * for blkzoned device, fallback direct IO to buffered IO, so
4102 * all IOs can be serialized by log-structured write.
4103 */
4104 if (f2fs_sb_has_blkzoned(sbi))
4105 return true;
4106 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4107 if (block_unaligned_IO(inode, iocb, iter))
4108 return true;
4109 if (F2FS_IO_ALIGNED(sbi))
4110 return true;
4111 }
4112 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4113 !IS_SWAPFILE(inode))
4114 return true;
4115
4116 return false;
4117}
4118
4119#ifdef CONFIG_F2FS_FAULT_INJECTION
4120extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4121 unsigned int type);
4122#else
4123#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4124#endif
4125
4126static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4127{
4128#ifdef CONFIG_QUOTA
4129 if (f2fs_sb_has_quota_ino(sbi))
4130 return true;
4131 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4132 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4133 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4134 return true;
4135#endif
4136 return false;
4137}
4138
4139#define EFSBADCRC EBADMSG /* Bad CRC detected */
4140#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4141
4142#endif /* _LINUX_F2FS_H */