]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame_incremental - fs/f2fs/f2fs.h
f2fs: compress: fix zstd data corruption
[mirror_ubuntu-jammy-kernel.git] / fs / f2fs / f2fs.h
... / ...
CommitLineData
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 bool test_dummy_encryption; /* test dummy encryption */
142 block_t unusable_cap; /* Amount of space allowed to be
143 * unusable when disabling checkpoint
144 */
145
146 /* For compression */
147 unsigned char compress_algorithm; /* algorithm type */
148 unsigned compress_log_size; /* cluster log size */
149 unsigned char compress_ext_cnt; /* extension count */
150 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
151};
152
153#define F2FS_FEATURE_ENCRYPT 0x0001
154#define F2FS_FEATURE_BLKZONED 0x0002
155#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
156#define F2FS_FEATURE_EXTRA_ATTR 0x0008
157#define F2FS_FEATURE_PRJQUOTA 0x0010
158#define F2FS_FEATURE_INODE_CHKSUM 0x0020
159#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
160#define F2FS_FEATURE_QUOTA_INO 0x0080
161#define F2FS_FEATURE_INODE_CRTIME 0x0100
162#define F2FS_FEATURE_LOST_FOUND 0x0200
163#define F2FS_FEATURE_VERITY 0x0400
164#define F2FS_FEATURE_SB_CHKSUM 0x0800
165#define F2FS_FEATURE_CASEFOLD 0x1000
166#define F2FS_FEATURE_COMPRESSION 0x2000
167
168#define __F2FS_HAS_FEATURE(raw_super, mask) \
169 ((raw_super->feature & cpu_to_le32(mask)) != 0)
170#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
171#define F2FS_SET_FEATURE(sbi, mask) \
172 (sbi->raw_super->feature |= cpu_to_le32(mask))
173#define F2FS_CLEAR_FEATURE(sbi, mask) \
174 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
175
176/*
177 * Default values for user and/or group using reserved blocks
178 */
179#define F2FS_DEF_RESUID 0
180#define F2FS_DEF_RESGID 0
181
182/*
183 * For checkpoint manager
184 */
185enum {
186 NAT_BITMAP,
187 SIT_BITMAP
188};
189
190#define CP_UMOUNT 0x00000001
191#define CP_FASTBOOT 0x00000002
192#define CP_SYNC 0x00000004
193#define CP_RECOVERY 0x00000008
194#define CP_DISCARD 0x00000010
195#define CP_TRIMMED 0x00000020
196#define CP_PAUSE 0x00000040
197#define CP_RESIZE 0x00000080
198
199#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
200#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
201#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
202#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
203#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
204#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
205#define DEF_CP_INTERVAL 60 /* 60 secs */
206#define DEF_IDLE_INTERVAL 5 /* 5 secs */
207#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
208#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
209#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
210
211struct cp_control {
212 int reason;
213 __u64 trim_start;
214 __u64 trim_end;
215 __u64 trim_minlen;
216};
217
218/*
219 * indicate meta/data type
220 */
221enum {
222 META_CP,
223 META_NAT,
224 META_SIT,
225 META_SSA,
226 META_MAX,
227 META_POR,
228 DATA_GENERIC, /* check range only */
229 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
230 DATA_GENERIC_ENHANCE_READ, /*
231 * strong check on range and segment
232 * bitmap but no warning due to race
233 * condition of read on truncated area
234 * by extent_cache
235 */
236 META_GENERIC,
237};
238
239/* for the list of ino */
240enum {
241 ORPHAN_INO, /* for orphan ino list */
242 APPEND_INO, /* for append ino list */
243 UPDATE_INO, /* for update ino list */
244 TRANS_DIR_INO, /* for trasactions dir ino list */
245 FLUSH_INO, /* for multiple device flushing */
246 MAX_INO_ENTRY, /* max. list */
247};
248
249struct ino_entry {
250 struct list_head list; /* list head */
251 nid_t ino; /* inode number */
252 unsigned int dirty_device; /* dirty device bitmap */
253};
254
255/* for the list of inodes to be GCed */
256struct inode_entry {
257 struct list_head list; /* list head */
258 struct inode *inode; /* vfs inode pointer */
259};
260
261struct fsync_node_entry {
262 struct list_head list; /* list head */
263 struct page *page; /* warm node page pointer */
264 unsigned int seq_id; /* sequence id */
265};
266
267/* for the bitmap indicate blocks to be discarded */
268struct discard_entry {
269 struct list_head list; /* list head */
270 block_t start_blkaddr; /* start blockaddr of current segment */
271 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
272};
273
274/* default discard granularity of inner discard thread, unit: block count */
275#define DEFAULT_DISCARD_GRANULARITY 16
276
277/* max discard pend list number */
278#define MAX_PLIST_NUM 512
279#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
280 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
281
282enum {
283 D_PREP, /* initial */
284 D_PARTIAL, /* partially submitted */
285 D_SUBMIT, /* all submitted */
286 D_DONE, /* finished */
287};
288
289struct discard_info {
290 block_t lstart; /* logical start address */
291 block_t len; /* length */
292 block_t start; /* actual start address in dev */
293};
294
295struct discard_cmd {
296 struct rb_node rb_node; /* rb node located in rb-tree */
297 union {
298 struct {
299 block_t lstart; /* logical start address */
300 block_t len; /* length */
301 block_t start; /* actual start address in dev */
302 };
303 struct discard_info di; /* discard info */
304
305 };
306 struct list_head list; /* command list */
307 struct completion wait; /* compleation */
308 struct block_device *bdev; /* bdev */
309 unsigned short ref; /* reference count */
310 unsigned char state; /* state */
311 unsigned char queued; /* queued discard */
312 int error; /* bio error */
313 spinlock_t lock; /* for state/bio_ref updating */
314 unsigned short bio_ref; /* bio reference count */
315};
316
317enum {
318 DPOLICY_BG,
319 DPOLICY_FORCE,
320 DPOLICY_FSTRIM,
321 DPOLICY_UMOUNT,
322 MAX_DPOLICY,
323};
324
325struct discard_policy {
326 int type; /* type of discard */
327 unsigned int min_interval; /* used for candidates exist */
328 unsigned int mid_interval; /* used for device busy */
329 unsigned int max_interval; /* used for candidates not exist */
330 unsigned int max_requests; /* # of discards issued per round */
331 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
332 bool io_aware; /* issue discard in idle time */
333 bool sync; /* submit discard with REQ_SYNC flag */
334 bool ordered; /* issue discard by lba order */
335 bool timeout; /* discard timeout for put_super */
336 unsigned int granularity; /* discard granularity */
337};
338
339struct discard_cmd_control {
340 struct task_struct *f2fs_issue_discard; /* discard thread */
341 struct list_head entry_list; /* 4KB discard entry list */
342 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
343 struct list_head wait_list; /* store on-flushing entries */
344 struct list_head fstrim_list; /* in-flight discard from fstrim */
345 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
346 unsigned int discard_wake; /* to wake up discard thread */
347 struct mutex cmd_lock;
348 unsigned int nr_discards; /* # of discards in the list */
349 unsigned int max_discards; /* max. discards to be issued */
350 unsigned int discard_granularity; /* discard granularity */
351 unsigned int undiscard_blks; /* # of undiscard blocks */
352 unsigned int next_pos; /* next discard position */
353 atomic_t issued_discard; /* # of issued discard */
354 atomic_t queued_discard; /* # of queued discard */
355 atomic_t discard_cmd_cnt; /* # of cached cmd count */
356 struct rb_root_cached root; /* root of discard rb-tree */
357 bool rbtree_check; /* config for consistence check */
358};
359
360/* for the list of fsync inodes, used only during recovery */
361struct fsync_inode_entry {
362 struct list_head list; /* list head */
363 struct inode *inode; /* vfs inode pointer */
364 block_t blkaddr; /* block address locating the last fsync */
365 block_t last_dentry; /* block address locating the last dentry */
366};
367
368#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
369#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
370
371#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
372#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
373#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
374#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
375
376#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
377#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
378
379static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
380{
381 int before = nats_in_cursum(journal);
382
383 journal->n_nats = cpu_to_le16(before + i);
384 return before;
385}
386
387static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
388{
389 int before = sits_in_cursum(journal);
390
391 journal->n_sits = cpu_to_le16(before + i);
392 return before;
393}
394
395static inline bool __has_cursum_space(struct f2fs_journal *journal,
396 int size, int type)
397{
398 if (type == NAT_JOURNAL)
399 return size <= MAX_NAT_JENTRIES(journal);
400 return size <= MAX_SIT_JENTRIES(journal);
401}
402
403/*
404 * ioctl commands
405 */
406#define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS
407#define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS
408#define F2FS_IOC_GETVERSION FS_IOC_GETVERSION
409
410#define F2FS_IOCTL_MAGIC 0xf5
411#define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1)
412#define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2)
413#define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3)
414#define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4)
415#define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5)
416#define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32)
417#define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7)
418#define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \
419 struct f2fs_defragment)
420#define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \
421 struct f2fs_move_range)
422#define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \
423 struct f2fs_flush_device)
424#define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \
425 struct f2fs_gc_range)
426#define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32)
427#define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32)
428#define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32)
429#define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15)
430#define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64)
431#define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64)
432#define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \
433 _IOR(F2FS_IOCTL_MAGIC, 18, __u64)
434#define F2FS_IOC_RESERVE_COMPRESS_BLOCKS \
435 _IOR(F2FS_IOCTL_MAGIC, 19, __u64)
436
437#define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL
438#define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL
439
440#define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
441#define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
442#define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
443
444/*
445 * should be same as XFS_IOC_GOINGDOWN.
446 * Flags for going down operation used by FS_IOC_GOINGDOWN
447 */
448#define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
449#define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */
450#define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */
451#define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
452#define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */
453#define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */
454
455#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
456/*
457 * ioctl commands in 32 bit emulation
458 */
459#define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
460#define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
461#define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION
462#endif
463
464#define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR
465#define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR
466
467struct f2fs_gc_range {
468 u32 sync;
469 u64 start;
470 u64 len;
471};
472
473struct f2fs_defragment {
474 u64 start;
475 u64 len;
476};
477
478struct f2fs_move_range {
479 u32 dst_fd; /* destination fd */
480 u64 pos_in; /* start position in src_fd */
481 u64 pos_out; /* start position in dst_fd */
482 u64 len; /* size to move */
483};
484
485struct f2fs_flush_device {
486 u32 dev_num; /* device number to flush */
487 u32 segments; /* # of segments to flush */
488};
489
490/* for inline stuff */
491#define DEF_INLINE_RESERVED_SIZE 1
492static inline int get_extra_isize(struct inode *inode);
493static inline int get_inline_xattr_addrs(struct inode *inode);
494#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
495 (CUR_ADDRS_PER_INODE(inode) - \
496 get_inline_xattr_addrs(inode) - \
497 DEF_INLINE_RESERVED_SIZE))
498
499/* for inline dir */
500#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
501 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
502 BITS_PER_BYTE + 1))
503#define INLINE_DENTRY_BITMAP_SIZE(inode) \
504 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
505#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
506 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
507 NR_INLINE_DENTRY(inode) + \
508 INLINE_DENTRY_BITMAP_SIZE(inode)))
509
510/*
511 * For INODE and NODE manager
512 */
513/* for directory operations */
514
515struct f2fs_filename {
516 /*
517 * The filename the user specified. This is NULL for some
518 * filesystem-internal operations, e.g. converting an inline directory
519 * to a non-inline one, or roll-forward recovering an encrypted dentry.
520 */
521 const struct qstr *usr_fname;
522
523 /*
524 * The on-disk filename. For encrypted directories, this is encrypted.
525 * This may be NULL for lookups in an encrypted dir without the key.
526 */
527 struct fscrypt_str disk_name;
528
529 /* The dirhash of this filename */
530 f2fs_hash_t hash;
531
532#ifdef CONFIG_FS_ENCRYPTION
533 /*
534 * For lookups in encrypted directories: either the buffer backing
535 * disk_name, or a buffer that holds the decoded no-key name.
536 */
537 struct fscrypt_str crypto_buf;
538#endif
539#ifdef CONFIG_UNICODE
540 /*
541 * For casefolded directories: the casefolded name, but it's left NULL
542 * if the original name is not valid Unicode or if the filesystem is
543 * doing an internal operation where usr_fname is also NULL. In these
544 * cases we fall back to treating the name as an opaque byte sequence.
545 */
546 struct fscrypt_str cf_name;
547#endif
548};
549
550struct f2fs_dentry_ptr {
551 struct inode *inode;
552 void *bitmap;
553 struct f2fs_dir_entry *dentry;
554 __u8 (*filename)[F2FS_SLOT_LEN];
555 int max;
556 int nr_bitmap;
557};
558
559static inline void make_dentry_ptr_block(struct inode *inode,
560 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
561{
562 d->inode = inode;
563 d->max = NR_DENTRY_IN_BLOCK;
564 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
565 d->bitmap = t->dentry_bitmap;
566 d->dentry = t->dentry;
567 d->filename = t->filename;
568}
569
570static inline void make_dentry_ptr_inline(struct inode *inode,
571 struct f2fs_dentry_ptr *d, void *t)
572{
573 int entry_cnt = NR_INLINE_DENTRY(inode);
574 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
575 int reserved_size = INLINE_RESERVED_SIZE(inode);
576
577 d->inode = inode;
578 d->max = entry_cnt;
579 d->nr_bitmap = bitmap_size;
580 d->bitmap = t;
581 d->dentry = t + bitmap_size + reserved_size;
582 d->filename = t + bitmap_size + reserved_size +
583 SIZE_OF_DIR_ENTRY * entry_cnt;
584}
585
586/*
587 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
588 * as its node offset to distinguish from index node blocks.
589 * But some bits are used to mark the node block.
590 */
591#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
592 >> OFFSET_BIT_SHIFT)
593enum {
594 ALLOC_NODE, /* allocate a new node page if needed */
595 LOOKUP_NODE, /* look up a node without readahead */
596 LOOKUP_NODE_RA, /*
597 * look up a node with readahead called
598 * by get_data_block.
599 */
600};
601
602#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
603
604/* congestion wait timeout value, default: 20ms */
605#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
606
607/* maximum retry quota flush count */
608#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
609
610#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
611
612#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
613
614/* for in-memory extent cache entry */
615#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
616
617/* number of extent info in extent cache we try to shrink */
618#define EXTENT_CACHE_SHRINK_NUMBER 128
619
620struct rb_entry {
621 struct rb_node rb_node; /* rb node located in rb-tree */
622 unsigned int ofs; /* start offset of the entry */
623 unsigned int len; /* length of the entry */
624};
625
626struct extent_info {
627 unsigned int fofs; /* start offset in a file */
628 unsigned int len; /* length of the extent */
629 u32 blk; /* start block address of the extent */
630};
631
632struct extent_node {
633 struct rb_node rb_node; /* rb node located in rb-tree */
634 struct extent_info ei; /* extent info */
635 struct list_head list; /* node in global extent list of sbi */
636 struct extent_tree *et; /* extent tree pointer */
637};
638
639struct extent_tree {
640 nid_t ino; /* inode number */
641 struct rb_root_cached root; /* root of extent info rb-tree */
642 struct extent_node *cached_en; /* recently accessed extent node */
643 struct extent_info largest; /* largested extent info */
644 struct list_head list; /* to be used by sbi->zombie_list */
645 rwlock_t lock; /* protect extent info rb-tree */
646 atomic_t node_cnt; /* # of extent node in rb-tree*/
647 bool largest_updated; /* largest extent updated */
648};
649
650/*
651 * This structure is taken from ext4_map_blocks.
652 *
653 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
654 */
655#define F2FS_MAP_NEW (1 << BH_New)
656#define F2FS_MAP_MAPPED (1 << BH_Mapped)
657#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
658#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
659 F2FS_MAP_UNWRITTEN)
660
661struct f2fs_map_blocks {
662 block_t m_pblk;
663 block_t m_lblk;
664 unsigned int m_len;
665 unsigned int m_flags;
666 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
667 pgoff_t *m_next_extent; /* point to next possible extent */
668 int m_seg_type;
669 bool m_may_create; /* indicate it is from write path */
670};
671
672/* for flag in get_data_block */
673enum {
674 F2FS_GET_BLOCK_DEFAULT,
675 F2FS_GET_BLOCK_FIEMAP,
676 F2FS_GET_BLOCK_BMAP,
677 F2FS_GET_BLOCK_DIO,
678 F2FS_GET_BLOCK_PRE_DIO,
679 F2FS_GET_BLOCK_PRE_AIO,
680 F2FS_GET_BLOCK_PRECACHE,
681};
682
683/*
684 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
685 */
686#define FADVISE_COLD_BIT 0x01
687#define FADVISE_LOST_PINO_BIT 0x02
688#define FADVISE_ENCRYPT_BIT 0x04
689#define FADVISE_ENC_NAME_BIT 0x08
690#define FADVISE_KEEP_SIZE_BIT 0x10
691#define FADVISE_HOT_BIT 0x20
692#define FADVISE_VERITY_BIT 0x40
693
694#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
695
696#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
697#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
698#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
699#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
700#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
701#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
702#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
703#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
704#define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT)
705#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
706#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
707#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
708#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
709#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
710#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
711#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
712#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
713#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
714
715#define DEF_DIR_LEVEL 0
716
717enum {
718 GC_FAILURE_PIN,
719 GC_FAILURE_ATOMIC,
720 MAX_GC_FAILURE
721};
722
723/* used for f2fs_inode_info->flags */
724enum {
725 FI_NEW_INODE, /* indicate newly allocated inode */
726 FI_DIRTY_INODE, /* indicate inode is dirty or not */
727 FI_AUTO_RECOVER, /* indicate inode is recoverable */
728 FI_DIRTY_DIR, /* indicate directory has dirty pages */
729 FI_INC_LINK, /* need to increment i_nlink */
730 FI_ACL_MODE, /* indicate acl mode */
731 FI_NO_ALLOC, /* should not allocate any blocks */
732 FI_FREE_NID, /* free allocated nide */
733 FI_NO_EXTENT, /* not to use the extent cache */
734 FI_INLINE_XATTR, /* used for inline xattr */
735 FI_INLINE_DATA, /* used for inline data*/
736 FI_INLINE_DENTRY, /* used for inline dentry */
737 FI_APPEND_WRITE, /* inode has appended data */
738 FI_UPDATE_WRITE, /* inode has in-place-update data */
739 FI_NEED_IPU, /* used for ipu per file */
740 FI_ATOMIC_FILE, /* indicate atomic file */
741 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
742 FI_VOLATILE_FILE, /* indicate volatile file */
743 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
744 FI_DROP_CACHE, /* drop dirty page cache */
745 FI_DATA_EXIST, /* indicate data exists */
746 FI_INLINE_DOTS, /* indicate inline dot dentries */
747 FI_DO_DEFRAG, /* indicate defragment is running */
748 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
749 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
750 FI_HOT_DATA, /* indicate file is hot */
751 FI_EXTRA_ATTR, /* indicate file has extra attribute */
752 FI_PROJ_INHERIT, /* indicate file inherits projectid */
753 FI_PIN_FILE, /* indicate file should not be gced */
754 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
755 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
756 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
757 FI_MMAP_FILE, /* indicate file was mmapped */
758 FI_MAX, /* max flag, never be used */
759};
760
761struct f2fs_inode_info {
762 struct inode vfs_inode; /* serve a vfs inode */
763 unsigned long i_flags; /* keep an inode flags for ioctl */
764 unsigned char i_advise; /* use to give file attribute hints */
765 unsigned char i_dir_level; /* use for dentry level for large dir */
766 unsigned int i_current_depth; /* only for directory depth */
767 /* for gc failure statistic */
768 unsigned int i_gc_failures[MAX_GC_FAILURE];
769 unsigned int i_pino; /* parent inode number */
770 umode_t i_acl_mode; /* keep file acl mode temporarily */
771
772 /* Use below internally in f2fs*/
773 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
774 struct rw_semaphore i_sem; /* protect fi info */
775 atomic_t dirty_pages; /* # of dirty pages */
776 f2fs_hash_t chash; /* hash value of given file name */
777 unsigned int clevel; /* maximum level of given file name */
778 struct task_struct *task; /* lookup and create consistency */
779 struct task_struct *cp_task; /* separate cp/wb IO stats*/
780 nid_t i_xattr_nid; /* node id that contains xattrs */
781 loff_t last_disk_size; /* lastly written file size */
782 spinlock_t i_size_lock; /* protect last_disk_size */
783
784#ifdef CONFIG_QUOTA
785 struct dquot *i_dquot[MAXQUOTAS];
786
787 /* quota space reservation, managed internally by quota code */
788 qsize_t i_reserved_quota;
789#endif
790 struct list_head dirty_list; /* dirty list for dirs and files */
791 struct list_head gdirty_list; /* linked in global dirty list */
792 struct list_head inmem_ilist; /* list for inmem inodes */
793 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
794 struct task_struct *inmem_task; /* store inmemory task */
795 struct mutex inmem_lock; /* lock for inmemory pages */
796 struct extent_tree *extent_tree; /* cached extent_tree entry */
797
798 /* avoid racing between foreground op and gc */
799 struct rw_semaphore i_gc_rwsem[2];
800 struct rw_semaphore i_mmap_sem;
801 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
802
803 int i_extra_isize; /* size of extra space located in i_addr */
804 kprojid_t i_projid; /* id for project quota */
805 int i_inline_xattr_size; /* inline xattr size */
806 struct timespec64 i_crtime; /* inode creation time */
807 struct timespec64 i_disk_time[4];/* inode disk times */
808
809 /* for file compress */
810 u64 i_compr_blocks; /* # of compressed blocks */
811 unsigned char i_compress_algorithm; /* algorithm type */
812 unsigned char i_log_cluster_size; /* log of cluster size */
813 unsigned int i_cluster_size; /* cluster size */
814};
815
816static inline void get_extent_info(struct extent_info *ext,
817 struct f2fs_extent *i_ext)
818{
819 ext->fofs = le32_to_cpu(i_ext->fofs);
820 ext->blk = le32_to_cpu(i_ext->blk);
821 ext->len = le32_to_cpu(i_ext->len);
822}
823
824static inline void set_raw_extent(struct extent_info *ext,
825 struct f2fs_extent *i_ext)
826{
827 i_ext->fofs = cpu_to_le32(ext->fofs);
828 i_ext->blk = cpu_to_le32(ext->blk);
829 i_ext->len = cpu_to_le32(ext->len);
830}
831
832static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
833 u32 blk, unsigned int len)
834{
835 ei->fofs = fofs;
836 ei->blk = blk;
837 ei->len = len;
838}
839
840static inline bool __is_discard_mergeable(struct discard_info *back,
841 struct discard_info *front, unsigned int max_len)
842{
843 return (back->lstart + back->len == front->lstart) &&
844 (back->len + front->len <= max_len);
845}
846
847static inline bool __is_discard_back_mergeable(struct discard_info *cur,
848 struct discard_info *back, unsigned int max_len)
849{
850 return __is_discard_mergeable(back, cur, max_len);
851}
852
853static inline bool __is_discard_front_mergeable(struct discard_info *cur,
854 struct discard_info *front, unsigned int max_len)
855{
856 return __is_discard_mergeable(cur, front, max_len);
857}
858
859static inline bool __is_extent_mergeable(struct extent_info *back,
860 struct extent_info *front)
861{
862 return (back->fofs + back->len == front->fofs &&
863 back->blk + back->len == front->blk);
864}
865
866static inline bool __is_back_mergeable(struct extent_info *cur,
867 struct extent_info *back)
868{
869 return __is_extent_mergeable(back, cur);
870}
871
872static inline bool __is_front_mergeable(struct extent_info *cur,
873 struct extent_info *front)
874{
875 return __is_extent_mergeable(cur, front);
876}
877
878extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
879static inline void __try_update_largest_extent(struct extent_tree *et,
880 struct extent_node *en)
881{
882 if (en->ei.len > et->largest.len) {
883 et->largest = en->ei;
884 et->largest_updated = true;
885 }
886}
887
888/*
889 * For free nid management
890 */
891enum nid_state {
892 FREE_NID, /* newly added to free nid list */
893 PREALLOC_NID, /* it is preallocated */
894 MAX_NID_STATE,
895};
896
897struct f2fs_nm_info {
898 block_t nat_blkaddr; /* base disk address of NAT */
899 nid_t max_nid; /* maximum possible node ids */
900 nid_t available_nids; /* # of available node ids */
901 nid_t next_scan_nid; /* the next nid to be scanned */
902 unsigned int ram_thresh; /* control the memory footprint */
903 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
904 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
905
906 /* NAT cache management */
907 struct radix_tree_root nat_root;/* root of the nat entry cache */
908 struct radix_tree_root nat_set_root;/* root of the nat set cache */
909 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */
910 struct list_head nat_entries; /* cached nat entry list (clean) */
911 spinlock_t nat_list_lock; /* protect clean nat entry list */
912 unsigned int nat_cnt; /* the # of cached nat entries */
913 unsigned int dirty_nat_cnt; /* total num of nat entries in set */
914 unsigned int nat_blocks; /* # of nat blocks */
915
916 /* free node ids management */
917 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
918 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
919 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
920 spinlock_t nid_list_lock; /* protect nid lists ops */
921 struct mutex build_lock; /* lock for build free nids */
922 unsigned char **free_nid_bitmap;
923 unsigned char *nat_block_bitmap;
924 unsigned short *free_nid_count; /* free nid count of NAT block */
925
926 /* for checkpoint */
927 char *nat_bitmap; /* NAT bitmap pointer */
928
929 unsigned int nat_bits_blocks; /* # of nat bits blocks */
930 unsigned char *nat_bits; /* NAT bits blocks */
931 unsigned char *full_nat_bits; /* full NAT pages */
932 unsigned char *empty_nat_bits; /* empty NAT pages */
933#ifdef CONFIG_F2FS_CHECK_FS
934 char *nat_bitmap_mir; /* NAT bitmap mirror */
935#endif
936 int bitmap_size; /* bitmap size */
937};
938
939/*
940 * this structure is used as one of function parameters.
941 * all the information are dedicated to a given direct node block determined
942 * by the data offset in a file.
943 */
944struct dnode_of_data {
945 struct inode *inode; /* vfs inode pointer */
946 struct page *inode_page; /* its inode page, NULL is possible */
947 struct page *node_page; /* cached direct node page */
948 nid_t nid; /* node id of the direct node block */
949 unsigned int ofs_in_node; /* data offset in the node page */
950 bool inode_page_locked; /* inode page is locked or not */
951 bool node_changed; /* is node block changed */
952 char cur_level; /* level of hole node page */
953 char max_level; /* level of current page located */
954 block_t data_blkaddr; /* block address of the node block */
955};
956
957static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
958 struct page *ipage, struct page *npage, nid_t nid)
959{
960 memset(dn, 0, sizeof(*dn));
961 dn->inode = inode;
962 dn->inode_page = ipage;
963 dn->node_page = npage;
964 dn->nid = nid;
965}
966
967/*
968 * For SIT manager
969 *
970 * By default, there are 6 active log areas across the whole main area.
971 * When considering hot and cold data separation to reduce cleaning overhead,
972 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
973 * respectively.
974 * In the current design, you should not change the numbers intentionally.
975 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
976 * logs individually according to the underlying devices. (default: 6)
977 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
978 * data and 8 for node logs.
979 */
980#define NR_CURSEG_DATA_TYPE (3)
981#define NR_CURSEG_NODE_TYPE (3)
982#define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
983
984enum {
985 CURSEG_HOT_DATA = 0, /* directory entry blocks */
986 CURSEG_WARM_DATA, /* data blocks */
987 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
988 CURSEG_HOT_NODE, /* direct node blocks of directory files */
989 CURSEG_WARM_NODE, /* direct node blocks of normal files */
990 CURSEG_COLD_NODE, /* indirect node blocks */
991 NO_CHECK_TYPE,
992 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */
993};
994
995struct flush_cmd {
996 struct completion wait;
997 struct llist_node llnode;
998 nid_t ino;
999 int ret;
1000};
1001
1002struct flush_cmd_control {
1003 struct task_struct *f2fs_issue_flush; /* flush thread */
1004 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
1005 atomic_t issued_flush; /* # of issued flushes */
1006 atomic_t queued_flush; /* # of queued flushes */
1007 struct llist_head issue_list; /* list for command issue */
1008 struct llist_node *dispatch_list; /* list for command dispatch */
1009};
1010
1011struct f2fs_sm_info {
1012 struct sit_info *sit_info; /* whole segment information */
1013 struct free_segmap_info *free_info; /* free segment information */
1014 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1015 struct curseg_info *curseg_array; /* active segment information */
1016
1017 struct rw_semaphore curseg_lock; /* for preventing curseg change */
1018
1019 block_t seg0_blkaddr; /* block address of 0'th segment */
1020 block_t main_blkaddr; /* start block address of main area */
1021 block_t ssa_blkaddr; /* start block address of SSA area */
1022
1023 unsigned int segment_count; /* total # of segments */
1024 unsigned int main_segments; /* # of segments in main area */
1025 unsigned int reserved_segments; /* # of reserved segments */
1026 unsigned int ovp_segments; /* # of overprovision segments */
1027
1028 /* a threshold to reclaim prefree segments */
1029 unsigned int rec_prefree_segments;
1030
1031 /* for batched trimming */
1032 unsigned int trim_sections; /* # of sections to trim */
1033
1034 struct list_head sit_entry_set; /* sit entry set list */
1035
1036 unsigned int ipu_policy; /* in-place-update policy */
1037 unsigned int min_ipu_util; /* in-place-update threshold */
1038 unsigned int min_fsync_blocks; /* threshold for fsync */
1039 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1040 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1041 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1042
1043 /* for flush command control */
1044 struct flush_cmd_control *fcc_info;
1045
1046 /* for discard command control */
1047 struct discard_cmd_control *dcc_info;
1048};
1049
1050/*
1051 * For superblock
1052 */
1053/*
1054 * COUNT_TYPE for monitoring
1055 *
1056 * f2fs monitors the number of several block types such as on-writeback,
1057 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1058 */
1059#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1060enum count_type {
1061 F2FS_DIRTY_DENTS,
1062 F2FS_DIRTY_DATA,
1063 F2FS_DIRTY_QDATA,
1064 F2FS_DIRTY_NODES,
1065 F2FS_DIRTY_META,
1066 F2FS_INMEM_PAGES,
1067 F2FS_DIRTY_IMETA,
1068 F2FS_WB_CP_DATA,
1069 F2FS_WB_DATA,
1070 F2FS_RD_DATA,
1071 F2FS_RD_NODE,
1072 F2FS_RD_META,
1073 F2FS_DIO_WRITE,
1074 F2FS_DIO_READ,
1075 NR_COUNT_TYPE,
1076};
1077
1078/*
1079 * The below are the page types of bios used in submit_bio().
1080 * The available types are:
1081 * DATA User data pages. It operates as async mode.
1082 * NODE Node pages. It operates as async mode.
1083 * META FS metadata pages such as SIT, NAT, CP.
1084 * NR_PAGE_TYPE The number of page types.
1085 * META_FLUSH Make sure the previous pages are written
1086 * with waiting the bio's completion
1087 * ... Only can be used with META.
1088 */
1089#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1090enum page_type {
1091 DATA,
1092 NODE,
1093 META,
1094 NR_PAGE_TYPE,
1095 META_FLUSH,
1096 INMEM, /* the below types are used by tracepoints only. */
1097 INMEM_DROP,
1098 INMEM_INVALIDATE,
1099 INMEM_REVOKE,
1100 IPU,
1101 OPU,
1102};
1103
1104enum temp_type {
1105 HOT = 0, /* must be zero for meta bio */
1106 WARM,
1107 COLD,
1108 NR_TEMP_TYPE,
1109};
1110
1111enum need_lock_type {
1112 LOCK_REQ = 0,
1113 LOCK_DONE,
1114 LOCK_RETRY,
1115};
1116
1117enum cp_reason_type {
1118 CP_NO_NEEDED,
1119 CP_NON_REGULAR,
1120 CP_COMPRESSED,
1121 CP_HARDLINK,
1122 CP_SB_NEED_CP,
1123 CP_WRONG_PINO,
1124 CP_NO_SPC_ROLL,
1125 CP_NODE_NEED_CP,
1126 CP_FASTBOOT_MODE,
1127 CP_SPEC_LOG_NUM,
1128 CP_RECOVER_DIR,
1129};
1130
1131enum iostat_type {
1132 /* WRITE IO */
1133 APP_DIRECT_IO, /* app direct write IOs */
1134 APP_BUFFERED_IO, /* app buffered write IOs */
1135 APP_WRITE_IO, /* app write IOs */
1136 APP_MAPPED_IO, /* app mapped IOs */
1137 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1138 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1139 FS_META_IO, /* meta IOs from kworker/reclaimer */
1140 FS_GC_DATA_IO, /* data IOs from forground gc */
1141 FS_GC_NODE_IO, /* node IOs from forground gc */
1142 FS_CP_DATA_IO, /* data IOs from checkpoint */
1143 FS_CP_NODE_IO, /* node IOs from checkpoint */
1144 FS_CP_META_IO, /* meta IOs from checkpoint */
1145
1146 /* READ IO */
1147 APP_DIRECT_READ_IO, /* app direct read IOs */
1148 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1149 APP_READ_IO, /* app read IOs */
1150 APP_MAPPED_READ_IO, /* app mapped read IOs */
1151 FS_DATA_READ_IO, /* data read IOs */
1152 FS_GDATA_READ_IO, /* data read IOs from background gc */
1153 FS_CDATA_READ_IO, /* compressed data read IOs */
1154 FS_NODE_READ_IO, /* node read IOs */
1155 FS_META_READ_IO, /* meta read IOs */
1156
1157 /* other */
1158 FS_DISCARD, /* discard */
1159 NR_IO_TYPE,
1160};
1161
1162struct f2fs_io_info {
1163 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1164 nid_t ino; /* inode number */
1165 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1166 enum temp_type temp; /* contains HOT/WARM/COLD */
1167 int op; /* contains REQ_OP_ */
1168 int op_flags; /* req_flag_bits */
1169 block_t new_blkaddr; /* new block address to be written */
1170 block_t old_blkaddr; /* old block address before Cow */
1171 struct page *page; /* page to be written */
1172 struct page *encrypted_page; /* encrypted page */
1173 struct page *compressed_page; /* compressed page */
1174 struct list_head list; /* serialize IOs */
1175 bool submitted; /* indicate IO submission */
1176 int need_lock; /* indicate we need to lock cp_rwsem */
1177 bool in_list; /* indicate fio is in io_list */
1178 bool is_por; /* indicate IO is from recovery or not */
1179 bool retry; /* need to reallocate block address */
1180 int compr_blocks; /* # of compressed block addresses */
1181 bool encrypted; /* indicate file is encrypted */
1182 enum iostat_type io_type; /* io type */
1183 struct writeback_control *io_wbc; /* writeback control */
1184 struct bio **bio; /* bio for ipu */
1185 sector_t *last_block; /* last block number in bio */
1186 unsigned char version; /* version of the node */
1187};
1188
1189struct bio_entry {
1190 struct bio *bio;
1191 struct list_head list;
1192};
1193
1194#define is_read_io(rw) ((rw) == READ)
1195struct f2fs_bio_info {
1196 struct f2fs_sb_info *sbi; /* f2fs superblock */
1197 struct bio *bio; /* bios to merge */
1198 sector_t last_block_in_bio; /* last block number */
1199 struct f2fs_io_info fio; /* store buffered io info. */
1200 struct rw_semaphore io_rwsem; /* blocking op for bio */
1201 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1202 struct list_head io_list; /* track fios */
1203 struct list_head bio_list; /* bio entry list head */
1204 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1205};
1206
1207#define FDEV(i) (sbi->devs[i])
1208#define RDEV(i) (raw_super->devs[i])
1209struct f2fs_dev_info {
1210 struct block_device *bdev;
1211 char path[MAX_PATH_LEN];
1212 unsigned int total_segments;
1213 block_t start_blk;
1214 block_t end_blk;
1215#ifdef CONFIG_BLK_DEV_ZONED
1216 unsigned int nr_blkz; /* Total number of zones */
1217 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1218#endif
1219};
1220
1221enum inode_type {
1222 DIR_INODE, /* for dirty dir inode */
1223 FILE_INODE, /* for dirty regular/symlink inode */
1224 DIRTY_META, /* for all dirtied inode metadata */
1225 ATOMIC_FILE, /* for all atomic files */
1226 NR_INODE_TYPE,
1227};
1228
1229/* for inner inode cache management */
1230struct inode_management {
1231 struct radix_tree_root ino_root; /* ino entry array */
1232 spinlock_t ino_lock; /* for ino entry lock */
1233 struct list_head ino_list; /* inode list head */
1234 unsigned long ino_num; /* number of entries */
1235};
1236
1237/* For s_flag in struct f2fs_sb_info */
1238enum {
1239 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1240 SBI_IS_CLOSE, /* specify unmounting */
1241 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1242 SBI_POR_DOING, /* recovery is doing or not */
1243 SBI_NEED_SB_WRITE, /* need to recover superblock */
1244 SBI_NEED_CP, /* need to checkpoint */
1245 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1246 SBI_IS_RECOVERED, /* recovered orphan/data */
1247 SBI_CP_DISABLED, /* CP was disabled last mount */
1248 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1249 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1250 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1251 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1252 SBI_IS_RESIZEFS, /* resizefs is in process */
1253};
1254
1255enum {
1256 CP_TIME,
1257 REQ_TIME,
1258 DISCARD_TIME,
1259 GC_TIME,
1260 DISABLE_TIME,
1261 UMOUNT_DISCARD_TIMEOUT,
1262 MAX_TIME,
1263};
1264
1265enum {
1266 GC_NORMAL,
1267 GC_IDLE_CB,
1268 GC_IDLE_GREEDY,
1269 GC_URGENT,
1270};
1271
1272enum {
1273 BGGC_MODE_ON, /* background gc is on */
1274 BGGC_MODE_OFF, /* background gc is off */
1275 BGGC_MODE_SYNC, /*
1276 * background gc is on, migrating blocks
1277 * like foreground gc
1278 */
1279};
1280
1281enum {
1282 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1283 FS_MODE_LFS, /* use lfs allocation only */
1284};
1285
1286enum {
1287 WHINT_MODE_OFF, /* not pass down write hints */
1288 WHINT_MODE_USER, /* try to pass down hints given by users */
1289 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1290};
1291
1292enum {
1293 ALLOC_MODE_DEFAULT, /* stay default */
1294 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1295};
1296
1297enum fsync_mode {
1298 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1299 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1300 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1301};
1302
1303/*
1304 * this value is set in page as a private data which indicate that
1305 * the page is atomically written, and it is in inmem_pages list.
1306 */
1307#define ATOMIC_WRITTEN_PAGE ((unsigned long)-1)
1308#define DUMMY_WRITTEN_PAGE ((unsigned long)-2)
1309
1310#define IS_ATOMIC_WRITTEN_PAGE(page) \
1311 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE)
1312#define IS_DUMMY_WRITTEN_PAGE(page) \
1313 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE)
1314
1315#ifdef CONFIG_FS_ENCRYPTION
1316#define DUMMY_ENCRYPTION_ENABLED(sbi) \
1317 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption))
1318#else
1319#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1320#endif
1321
1322/* For compression */
1323enum compress_algorithm_type {
1324 COMPRESS_LZO,
1325 COMPRESS_LZ4,
1326 COMPRESS_ZSTD,
1327 COMPRESS_LZORLE,
1328 COMPRESS_MAX,
1329};
1330
1331#define COMPRESS_DATA_RESERVED_SIZE 5
1332struct compress_data {
1333 __le32 clen; /* compressed data size */
1334 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1335 u8 cdata[]; /* compressed data */
1336};
1337
1338#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1339
1340#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1341
1342/* compress context */
1343struct compress_ctx {
1344 struct inode *inode; /* inode the context belong to */
1345 pgoff_t cluster_idx; /* cluster index number */
1346 unsigned int cluster_size; /* page count in cluster */
1347 unsigned int log_cluster_size; /* log of cluster size */
1348 struct page **rpages; /* pages store raw data in cluster */
1349 unsigned int nr_rpages; /* total page number in rpages */
1350 struct page **cpages; /* pages store compressed data in cluster */
1351 unsigned int nr_cpages; /* total page number in cpages */
1352 void *rbuf; /* virtual mapped address on rpages */
1353 struct compress_data *cbuf; /* virtual mapped address on cpages */
1354 size_t rlen; /* valid data length in rbuf */
1355 size_t clen; /* valid data length in cbuf */
1356 void *private; /* payload buffer for specified compression algorithm */
1357 void *private2; /* extra payload buffer */
1358};
1359
1360/* compress context for write IO path */
1361struct compress_io_ctx {
1362 u32 magic; /* magic number to indicate page is compressed */
1363 struct inode *inode; /* inode the context belong to */
1364 struct page **rpages; /* pages store raw data in cluster */
1365 unsigned int nr_rpages; /* total page number in rpages */
1366 refcount_t ref; /* referrence count of raw page */
1367};
1368
1369/* decompress io context for read IO path */
1370struct decompress_io_ctx {
1371 u32 magic; /* magic number to indicate page is compressed */
1372 struct inode *inode; /* inode the context belong to */
1373 pgoff_t cluster_idx; /* cluster index number */
1374 unsigned int cluster_size; /* page count in cluster */
1375 unsigned int log_cluster_size; /* log of cluster size */
1376 struct page **rpages; /* pages store raw data in cluster */
1377 unsigned int nr_rpages; /* total page number in rpages */
1378 struct page **cpages; /* pages store compressed data in cluster */
1379 unsigned int nr_cpages; /* total page number in cpages */
1380 struct page **tpages; /* temp pages to pad holes in cluster */
1381 void *rbuf; /* virtual mapped address on rpages */
1382 struct compress_data *cbuf; /* virtual mapped address on cpages */
1383 size_t rlen; /* valid data length in rbuf */
1384 size_t clen; /* valid data length in cbuf */
1385 refcount_t ref; /* referrence count of compressed page */
1386 bool failed; /* indicate IO error during decompression */
1387 void *private; /* payload buffer for specified decompression algorithm */
1388 void *private2; /* extra payload buffer */
1389};
1390
1391#define NULL_CLUSTER ((unsigned int)(~0))
1392#define MIN_COMPRESS_LOG_SIZE 2
1393#define MAX_COMPRESS_LOG_SIZE 8
1394#define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE)
1395
1396struct f2fs_sb_info {
1397 struct super_block *sb; /* pointer to VFS super block */
1398 struct proc_dir_entry *s_proc; /* proc entry */
1399 struct f2fs_super_block *raw_super; /* raw super block pointer */
1400 struct rw_semaphore sb_lock; /* lock for raw super block */
1401 int valid_super_block; /* valid super block no */
1402 unsigned long s_flag; /* flags for sbi */
1403 struct mutex writepages; /* mutex for writepages() */
1404#ifdef CONFIG_UNICODE
1405 struct unicode_map *s_encoding;
1406 __u16 s_encoding_flags;
1407#endif
1408
1409#ifdef CONFIG_BLK_DEV_ZONED
1410 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1411 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1412#endif
1413
1414 /* for node-related operations */
1415 struct f2fs_nm_info *nm_info; /* node manager */
1416 struct inode *node_inode; /* cache node blocks */
1417
1418 /* for segment-related operations */
1419 struct f2fs_sm_info *sm_info; /* segment manager */
1420
1421 /* for bio operations */
1422 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1423 /* keep migration IO order for LFS mode */
1424 struct rw_semaphore io_order_lock;
1425 mempool_t *write_io_dummy; /* Dummy pages */
1426
1427 /* for checkpoint */
1428 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1429 int cur_cp_pack; /* remain current cp pack */
1430 spinlock_t cp_lock; /* for flag in ckpt */
1431 struct inode *meta_inode; /* cache meta blocks */
1432 struct mutex cp_mutex; /* checkpoint procedure lock */
1433 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1434 struct rw_semaphore node_write; /* locking node writes */
1435 struct rw_semaphore node_change; /* locking node change */
1436 wait_queue_head_t cp_wait;
1437 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1438 long interval_time[MAX_TIME]; /* to store thresholds */
1439
1440 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1441
1442 spinlock_t fsync_node_lock; /* for node entry lock */
1443 struct list_head fsync_node_list; /* node list head */
1444 unsigned int fsync_seg_id; /* sequence id */
1445 unsigned int fsync_node_num; /* number of node entries */
1446
1447 /* for orphan inode, use 0'th array */
1448 unsigned int max_orphans; /* max orphan inodes */
1449
1450 /* for inode management */
1451 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1452 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1453 struct mutex flush_lock; /* for flush exclusion */
1454
1455 /* for extent tree cache */
1456 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1457 struct mutex extent_tree_lock; /* locking extent radix tree */
1458 struct list_head extent_list; /* lru list for shrinker */
1459 spinlock_t extent_lock; /* locking extent lru list */
1460 atomic_t total_ext_tree; /* extent tree count */
1461 struct list_head zombie_list; /* extent zombie tree list */
1462 atomic_t total_zombie_tree; /* extent zombie tree count */
1463 atomic_t total_ext_node; /* extent info count */
1464
1465 /* basic filesystem units */
1466 unsigned int log_sectors_per_block; /* log2 sectors per block */
1467 unsigned int log_blocksize; /* log2 block size */
1468 unsigned int blocksize; /* block size */
1469 unsigned int root_ino_num; /* root inode number*/
1470 unsigned int node_ino_num; /* node inode number*/
1471 unsigned int meta_ino_num; /* meta inode number*/
1472 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1473 unsigned int blocks_per_seg; /* blocks per segment */
1474 unsigned int segs_per_sec; /* segments per section */
1475 unsigned int secs_per_zone; /* sections per zone */
1476 unsigned int total_sections; /* total section count */
1477 unsigned int total_node_count; /* total node block count */
1478 unsigned int total_valid_node_count; /* valid node block count */
1479 loff_t max_file_blocks; /* max block index of file */
1480 int dir_level; /* directory level */
1481 int readdir_ra; /* readahead inode in readdir */
1482
1483 block_t user_block_count; /* # of user blocks */
1484 block_t total_valid_block_count; /* # of valid blocks */
1485 block_t discard_blks; /* discard command candidats */
1486 block_t last_valid_block_count; /* for recovery */
1487 block_t reserved_blocks; /* configurable reserved blocks */
1488 block_t current_reserved_blocks; /* current reserved blocks */
1489
1490 /* Additional tracking for no checkpoint mode */
1491 block_t unusable_block_count; /* # of blocks saved by last cp */
1492
1493 unsigned int nquota_files; /* # of quota sysfile */
1494 struct rw_semaphore quota_sem; /* blocking cp for flags */
1495
1496 /* # of pages, see count_type */
1497 atomic_t nr_pages[NR_COUNT_TYPE];
1498 /* # of allocated blocks */
1499 struct percpu_counter alloc_valid_block_count;
1500
1501 /* writeback control */
1502 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1503
1504 /* valid inode count */
1505 struct percpu_counter total_valid_inode_count;
1506
1507 struct f2fs_mount_info mount_opt; /* mount options */
1508
1509 /* for cleaning operations */
1510 struct rw_semaphore gc_lock; /*
1511 * semaphore for GC, avoid
1512 * race between GC and GC or CP
1513 */
1514 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1515 unsigned int cur_victim_sec; /* current victim section num */
1516 unsigned int gc_mode; /* current GC state */
1517 unsigned int next_victim_seg[2]; /* next segment in victim section */
1518 /* for skip statistic */
1519 unsigned int atomic_files; /* # of opened atomic file */
1520 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1521 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1522
1523 /* threshold for gc trials on pinned files */
1524 u64 gc_pin_file_threshold;
1525 struct rw_semaphore pin_sem;
1526
1527 /* maximum # of trials to find a victim segment for SSR and GC */
1528 unsigned int max_victim_search;
1529 /* migration granularity of garbage collection, unit: segment */
1530 unsigned int migration_granularity;
1531
1532 /*
1533 * for stat information.
1534 * one is for the LFS mode, and the other is for the SSR mode.
1535 */
1536#ifdef CONFIG_F2FS_STAT_FS
1537 struct f2fs_stat_info *stat_info; /* FS status information */
1538 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1539 unsigned int segment_count[2]; /* # of allocated segments */
1540 unsigned int block_count[2]; /* # of allocated blocks */
1541 atomic_t inplace_count; /* # of inplace update */
1542 atomic64_t total_hit_ext; /* # of lookup extent cache */
1543 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1544 atomic64_t read_hit_largest; /* # of hit largest extent node */
1545 atomic64_t read_hit_cached; /* # of hit cached extent node */
1546 atomic_t inline_xattr; /* # of inline_xattr inodes */
1547 atomic_t inline_inode; /* # of inline_data inodes */
1548 atomic_t inline_dir; /* # of inline_dentry inodes */
1549 atomic_t compr_inode; /* # of compressed inodes */
1550 atomic_t compr_blocks; /* # of compressed blocks */
1551 atomic_t vw_cnt; /* # of volatile writes */
1552 atomic_t max_aw_cnt; /* max # of atomic writes */
1553 atomic_t max_vw_cnt; /* max # of volatile writes */
1554 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1555 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1556 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1557#endif
1558 spinlock_t stat_lock; /* lock for stat operations */
1559
1560 /* For app/fs IO statistics */
1561 spinlock_t iostat_lock;
1562 unsigned long long rw_iostat[NR_IO_TYPE];
1563 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1564 bool iostat_enable;
1565 unsigned long iostat_next_period;
1566 unsigned int iostat_period_ms;
1567
1568 /* to attach REQ_META|REQ_FUA flags */
1569 unsigned int data_io_flag;
1570
1571 /* For sysfs suppport */
1572 struct kobject s_kobj;
1573 struct completion s_kobj_unregister;
1574
1575 /* For shrinker support */
1576 struct list_head s_list;
1577 int s_ndevs; /* number of devices */
1578 struct f2fs_dev_info *devs; /* for device list */
1579 unsigned int dirty_device; /* for checkpoint data flush */
1580 spinlock_t dev_lock; /* protect dirty_device */
1581 struct mutex umount_mutex;
1582 unsigned int shrinker_run_no;
1583
1584 /* For write statistics */
1585 u64 sectors_written_start;
1586 u64 kbytes_written;
1587
1588 /* Reference to checksum algorithm driver via cryptoapi */
1589 struct crypto_shash *s_chksum_driver;
1590
1591 /* Precomputed FS UUID checksum for seeding other checksums */
1592 __u32 s_chksum_seed;
1593
1594 struct workqueue_struct *post_read_wq; /* post read workqueue */
1595
1596 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1597 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1598};
1599
1600struct f2fs_private_dio {
1601 struct inode *inode;
1602 void *orig_private;
1603 bio_end_io_t *orig_end_io;
1604 bool write;
1605};
1606
1607#ifdef CONFIG_F2FS_FAULT_INJECTION
1608#define f2fs_show_injection_info(sbi, type) \
1609 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1610 KERN_INFO, sbi->sb->s_id, \
1611 f2fs_fault_name[type], \
1612 __func__, __builtin_return_address(0))
1613static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1614{
1615 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1616
1617 if (!ffi->inject_rate)
1618 return false;
1619
1620 if (!IS_FAULT_SET(ffi, type))
1621 return false;
1622
1623 atomic_inc(&ffi->inject_ops);
1624 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1625 atomic_set(&ffi->inject_ops, 0);
1626 return true;
1627 }
1628 return false;
1629}
1630#else
1631#define f2fs_show_injection_info(sbi, type) do { } while (0)
1632static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1633{
1634 return false;
1635}
1636#endif
1637
1638/*
1639 * Test if the mounted volume is a multi-device volume.
1640 * - For a single regular disk volume, sbi->s_ndevs is 0.
1641 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1642 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1643 */
1644static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1645{
1646 return sbi->s_ndevs > 1;
1647}
1648
1649/* For write statistics. Suppose sector size is 512 bytes,
1650 * and the return value is in kbytes. s is of struct f2fs_sb_info.
1651 */
1652#define BD_PART_WRITTEN(s) \
1653(((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \
1654 (s)->sectors_written_start) >> 1)
1655
1656static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1657{
1658 unsigned long now = jiffies;
1659
1660 sbi->last_time[type] = now;
1661
1662 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1663 if (type == REQ_TIME) {
1664 sbi->last_time[DISCARD_TIME] = now;
1665 sbi->last_time[GC_TIME] = now;
1666 }
1667}
1668
1669static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1670{
1671 unsigned long interval = sbi->interval_time[type] * HZ;
1672
1673 return time_after(jiffies, sbi->last_time[type] + interval);
1674}
1675
1676static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1677 int type)
1678{
1679 unsigned long interval = sbi->interval_time[type] * HZ;
1680 unsigned int wait_ms = 0;
1681 long delta;
1682
1683 delta = (sbi->last_time[type] + interval) - jiffies;
1684 if (delta > 0)
1685 wait_ms = jiffies_to_msecs(delta);
1686
1687 return wait_ms;
1688}
1689
1690/*
1691 * Inline functions
1692 */
1693static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1694 const void *address, unsigned int length)
1695{
1696 struct {
1697 struct shash_desc shash;
1698 char ctx[4];
1699 } desc;
1700 int err;
1701
1702 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1703
1704 desc.shash.tfm = sbi->s_chksum_driver;
1705 *(u32 *)desc.ctx = crc;
1706
1707 err = crypto_shash_update(&desc.shash, address, length);
1708 BUG_ON(err);
1709
1710 return *(u32 *)desc.ctx;
1711}
1712
1713static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1714 unsigned int length)
1715{
1716 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1717}
1718
1719static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1720 void *buf, size_t buf_size)
1721{
1722 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1723}
1724
1725static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1726 const void *address, unsigned int length)
1727{
1728 return __f2fs_crc32(sbi, crc, address, length);
1729}
1730
1731static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1732{
1733 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1734}
1735
1736static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1737{
1738 return sb->s_fs_info;
1739}
1740
1741static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1742{
1743 return F2FS_SB(inode->i_sb);
1744}
1745
1746static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1747{
1748 return F2FS_I_SB(mapping->host);
1749}
1750
1751static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1752{
1753 return F2FS_M_SB(page_file_mapping(page));
1754}
1755
1756static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1757{
1758 return (struct f2fs_super_block *)(sbi->raw_super);
1759}
1760
1761static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1762{
1763 return (struct f2fs_checkpoint *)(sbi->ckpt);
1764}
1765
1766static inline struct f2fs_node *F2FS_NODE(struct page *page)
1767{
1768 return (struct f2fs_node *)page_address(page);
1769}
1770
1771static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1772{
1773 return &((struct f2fs_node *)page_address(page))->i;
1774}
1775
1776static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1777{
1778 return (struct f2fs_nm_info *)(sbi->nm_info);
1779}
1780
1781static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1782{
1783 return (struct f2fs_sm_info *)(sbi->sm_info);
1784}
1785
1786static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1787{
1788 return (struct sit_info *)(SM_I(sbi)->sit_info);
1789}
1790
1791static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1792{
1793 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1794}
1795
1796static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1797{
1798 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1799}
1800
1801static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1802{
1803 return sbi->meta_inode->i_mapping;
1804}
1805
1806static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1807{
1808 return sbi->node_inode->i_mapping;
1809}
1810
1811static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1812{
1813 return test_bit(type, &sbi->s_flag);
1814}
1815
1816static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1817{
1818 set_bit(type, &sbi->s_flag);
1819}
1820
1821static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1822{
1823 clear_bit(type, &sbi->s_flag);
1824}
1825
1826static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1827{
1828 return le64_to_cpu(cp->checkpoint_ver);
1829}
1830
1831static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1832{
1833 if (type < F2FS_MAX_QUOTAS)
1834 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1835 return 0;
1836}
1837
1838static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
1839{
1840 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
1841 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
1842}
1843
1844static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1845{
1846 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1847
1848 return ckpt_flags & f;
1849}
1850
1851static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1852{
1853 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
1854}
1855
1856static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1857{
1858 unsigned int ckpt_flags;
1859
1860 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1861 ckpt_flags |= f;
1862 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1863}
1864
1865static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1866{
1867 unsigned long flags;
1868
1869 spin_lock_irqsave(&sbi->cp_lock, flags);
1870 __set_ckpt_flags(F2FS_CKPT(sbi), f);
1871 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1872}
1873
1874static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
1875{
1876 unsigned int ckpt_flags;
1877
1878 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
1879 ckpt_flags &= (~f);
1880 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
1881}
1882
1883static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
1884{
1885 unsigned long flags;
1886
1887 spin_lock_irqsave(&sbi->cp_lock, flags);
1888 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
1889 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1890}
1891
1892static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
1893{
1894 unsigned long flags;
1895 unsigned char *nat_bits;
1896
1897 /*
1898 * In order to re-enable nat_bits we need to call fsck.f2fs by
1899 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
1900 * so let's rely on regular fsck or unclean shutdown.
1901 */
1902
1903 if (lock)
1904 spin_lock_irqsave(&sbi->cp_lock, flags);
1905 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
1906 nat_bits = NM_I(sbi)->nat_bits;
1907 NM_I(sbi)->nat_bits = NULL;
1908 if (lock)
1909 spin_unlock_irqrestore(&sbi->cp_lock, flags);
1910
1911 kvfree(nat_bits);
1912}
1913
1914static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
1915 struct cp_control *cpc)
1916{
1917 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
1918
1919 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
1920}
1921
1922static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
1923{
1924 down_read(&sbi->cp_rwsem);
1925}
1926
1927static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
1928{
1929 return down_read_trylock(&sbi->cp_rwsem);
1930}
1931
1932static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
1933{
1934 up_read(&sbi->cp_rwsem);
1935}
1936
1937static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
1938{
1939 down_write(&sbi->cp_rwsem);
1940}
1941
1942static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
1943{
1944 up_write(&sbi->cp_rwsem);
1945}
1946
1947static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
1948{
1949 int reason = CP_SYNC;
1950
1951 if (test_opt(sbi, FASTBOOT))
1952 reason = CP_FASTBOOT;
1953 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
1954 reason = CP_UMOUNT;
1955 return reason;
1956}
1957
1958static inline bool __remain_node_summaries(int reason)
1959{
1960 return (reason & (CP_UMOUNT | CP_FASTBOOT));
1961}
1962
1963static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
1964{
1965 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
1966 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
1967}
1968
1969/*
1970 * Check whether the inode has blocks or not
1971 */
1972static inline int F2FS_HAS_BLOCKS(struct inode *inode)
1973{
1974 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
1975
1976 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
1977}
1978
1979static inline bool f2fs_has_xattr_block(unsigned int ofs)
1980{
1981 return ofs == XATTR_NODE_OFFSET;
1982}
1983
1984static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
1985 struct inode *inode, bool cap)
1986{
1987 if (!inode)
1988 return true;
1989 if (!test_opt(sbi, RESERVE_ROOT))
1990 return false;
1991 if (IS_NOQUOTA(inode))
1992 return true;
1993 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
1994 return true;
1995 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
1996 in_group_p(F2FS_OPTION(sbi).s_resgid))
1997 return true;
1998 if (cap && capable(CAP_SYS_RESOURCE))
1999 return true;
2000 return false;
2001}
2002
2003static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2004static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2005 struct inode *inode, blkcnt_t *count)
2006{
2007 blkcnt_t diff = 0, release = 0;
2008 block_t avail_user_block_count;
2009 int ret;
2010
2011 ret = dquot_reserve_block(inode, *count);
2012 if (ret)
2013 return ret;
2014
2015 if (time_to_inject(sbi, FAULT_BLOCK)) {
2016 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2017 release = *count;
2018 goto release_quota;
2019 }
2020
2021 /*
2022 * let's increase this in prior to actual block count change in order
2023 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2024 */
2025 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2026
2027 spin_lock(&sbi->stat_lock);
2028 sbi->total_valid_block_count += (block_t)(*count);
2029 avail_user_block_count = sbi->user_block_count -
2030 sbi->current_reserved_blocks;
2031
2032 if (!__allow_reserved_blocks(sbi, inode, true))
2033 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2034 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2035 if (avail_user_block_count > sbi->unusable_block_count)
2036 avail_user_block_count -= sbi->unusable_block_count;
2037 else
2038 avail_user_block_count = 0;
2039 }
2040 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2041 diff = sbi->total_valid_block_count - avail_user_block_count;
2042 if (diff > *count)
2043 diff = *count;
2044 *count -= diff;
2045 release = diff;
2046 sbi->total_valid_block_count -= diff;
2047 if (!*count) {
2048 spin_unlock(&sbi->stat_lock);
2049 goto enospc;
2050 }
2051 }
2052 spin_unlock(&sbi->stat_lock);
2053
2054 if (unlikely(release)) {
2055 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2056 dquot_release_reservation_block(inode, release);
2057 }
2058 f2fs_i_blocks_write(inode, *count, true, true);
2059 return 0;
2060
2061enospc:
2062 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2063release_quota:
2064 dquot_release_reservation_block(inode, release);
2065 return -ENOSPC;
2066}
2067
2068__printf(2, 3)
2069void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2070
2071#define f2fs_err(sbi, fmt, ...) \
2072 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2073#define f2fs_warn(sbi, fmt, ...) \
2074 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2075#define f2fs_notice(sbi, fmt, ...) \
2076 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2077#define f2fs_info(sbi, fmt, ...) \
2078 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2079#define f2fs_debug(sbi, fmt, ...) \
2080 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2081
2082static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2083 struct inode *inode,
2084 block_t count)
2085{
2086 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2087
2088 spin_lock(&sbi->stat_lock);
2089 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2090 sbi->total_valid_block_count -= (block_t)count;
2091 if (sbi->reserved_blocks &&
2092 sbi->current_reserved_blocks < sbi->reserved_blocks)
2093 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2094 sbi->current_reserved_blocks + count);
2095 spin_unlock(&sbi->stat_lock);
2096 if (unlikely(inode->i_blocks < sectors)) {
2097 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2098 inode->i_ino,
2099 (unsigned long long)inode->i_blocks,
2100 (unsigned long long)sectors);
2101 set_sbi_flag(sbi, SBI_NEED_FSCK);
2102 return;
2103 }
2104 f2fs_i_blocks_write(inode, count, false, true);
2105}
2106
2107static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2108{
2109 atomic_inc(&sbi->nr_pages[count_type]);
2110
2111 if (count_type == F2FS_DIRTY_DENTS ||
2112 count_type == F2FS_DIRTY_NODES ||
2113 count_type == F2FS_DIRTY_META ||
2114 count_type == F2FS_DIRTY_QDATA ||
2115 count_type == F2FS_DIRTY_IMETA)
2116 set_sbi_flag(sbi, SBI_IS_DIRTY);
2117}
2118
2119static inline void inode_inc_dirty_pages(struct inode *inode)
2120{
2121 atomic_inc(&F2FS_I(inode)->dirty_pages);
2122 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2123 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2124 if (IS_NOQUOTA(inode))
2125 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2126}
2127
2128static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2129{
2130 atomic_dec(&sbi->nr_pages[count_type]);
2131}
2132
2133static inline void inode_dec_dirty_pages(struct inode *inode)
2134{
2135 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2136 !S_ISLNK(inode->i_mode))
2137 return;
2138
2139 atomic_dec(&F2FS_I(inode)->dirty_pages);
2140 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2141 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2142 if (IS_NOQUOTA(inode))
2143 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2144}
2145
2146static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2147{
2148 return atomic_read(&sbi->nr_pages[count_type]);
2149}
2150
2151static inline int get_dirty_pages(struct inode *inode)
2152{
2153 return atomic_read(&F2FS_I(inode)->dirty_pages);
2154}
2155
2156static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2157{
2158 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2159 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2160 sbi->log_blocks_per_seg;
2161
2162 return segs / sbi->segs_per_sec;
2163}
2164
2165static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2166{
2167 return sbi->total_valid_block_count;
2168}
2169
2170static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2171{
2172 return sbi->discard_blks;
2173}
2174
2175static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2176{
2177 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2178
2179 /* return NAT or SIT bitmap */
2180 if (flag == NAT_BITMAP)
2181 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2182 else if (flag == SIT_BITMAP)
2183 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2184
2185 return 0;
2186}
2187
2188static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2189{
2190 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2191}
2192
2193static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2194{
2195 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2196 int offset;
2197
2198 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2199 offset = (flag == SIT_BITMAP) ?
2200 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2201 /*
2202 * if large_nat_bitmap feature is enabled, leave checksum
2203 * protection for all nat/sit bitmaps.
2204 */
2205 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32);
2206 }
2207
2208 if (__cp_payload(sbi) > 0) {
2209 if (flag == NAT_BITMAP)
2210 return &ckpt->sit_nat_version_bitmap;
2211 else
2212 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2213 } else {
2214 offset = (flag == NAT_BITMAP) ?
2215 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2216 return &ckpt->sit_nat_version_bitmap + offset;
2217 }
2218}
2219
2220static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2221{
2222 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2223
2224 if (sbi->cur_cp_pack == 2)
2225 start_addr += sbi->blocks_per_seg;
2226 return start_addr;
2227}
2228
2229static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2230{
2231 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2232
2233 if (sbi->cur_cp_pack == 1)
2234 start_addr += sbi->blocks_per_seg;
2235 return start_addr;
2236}
2237
2238static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2239{
2240 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2241}
2242
2243static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2244{
2245 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2246}
2247
2248static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2249 struct inode *inode, bool is_inode)
2250{
2251 block_t valid_block_count;
2252 unsigned int valid_node_count, user_block_count;
2253 int err;
2254
2255 if (is_inode) {
2256 if (inode) {
2257 err = dquot_alloc_inode(inode);
2258 if (err)
2259 return err;
2260 }
2261 } else {
2262 err = dquot_reserve_block(inode, 1);
2263 if (err)
2264 return err;
2265 }
2266
2267 if (time_to_inject(sbi, FAULT_BLOCK)) {
2268 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2269 goto enospc;
2270 }
2271
2272 spin_lock(&sbi->stat_lock);
2273
2274 valid_block_count = sbi->total_valid_block_count +
2275 sbi->current_reserved_blocks + 1;
2276
2277 if (!__allow_reserved_blocks(sbi, inode, false))
2278 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2279 user_block_count = sbi->user_block_count;
2280 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2281 user_block_count -= sbi->unusable_block_count;
2282
2283 if (unlikely(valid_block_count > user_block_count)) {
2284 spin_unlock(&sbi->stat_lock);
2285 goto enospc;
2286 }
2287
2288 valid_node_count = sbi->total_valid_node_count + 1;
2289 if (unlikely(valid_node_count > sbi->total_node_count)) {
2290 spin_unlock(&sbi->stat_lock);
2291 goto enospc;
2292 }
2293
2294 sbi->total_valid_node_count++;
2295 sbi->total_valid_block_count++;
2296 spin_unlock(&sbi->stat_lock);
2297
2298 if (inode) {
2299 if (is_inode)
2300 f2fs_mark_inode_dirty_sync(inode, true);
2301 else
2302 f2fs_i_blocks_write(inode, 1, true, true);
2303 }
2304
2305 percpu_counter_inc(&sbi->alloc_valid_block_count);
2306 return 0;
2307
2308enospc:
2309 if (is_inode) {
2310 if (inode)
2311 dquot_free_inode(inode);
2312 } else {
2313 dquot_release_reservation_block(inode, 1);
2314 }
2315 return -ENOSPC;
2316}
2317
2318static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2319 struct inode *inode, bool is_inode)
2320{
2321 spin_lock(&sbi->stat_lock);
2322
2323 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2324 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2325
2326 sbi->total_valid_node_count--;
2327 sbi->total_valid_block_count--;
2328 if (sbi->reserved_blocks &&
2329 sbi->current_reserved_blocks < sbi->reserved_blocks)
2330 sbi->current_reserved_blocks++;
2331
2332 spin_unlock(&sbi->stat_lock);
2333
2334 if (is_inode) {
2335 dquot_free_inode(inode);
2336 } else {
2337 if (unlikely(inode->i_blocks == 0)) {
2338 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2339 inode->i_ino,
2340 (unsigned long long)inode->i_blocks);
2341 set_sbi_flag(sbi, SBI_NEED_FSCK);
2342 return;
2343 }
2344 f2fs_i_blocks_write(inode, 1, false, true);
2345 }
2346}
2347
2348static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2349{
2350 return sbi->total_valid_node_count;
2351}
2352
2353static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2354{
2355 percpu_counter_inc(&sbi->total_valid_inode_count);
2356}
2357
2358static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2359{
2360 percpu_counter_dec(&sbi->total_valid_inode_count);
2361}
2362
2363static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2364{
2365 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2366}
2367
2368static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2369 pgoff_t index, bool for_write)
2370{
2371 struct page *page;
2372
2373 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2374 if (!for_write)
2375 page = find_get_page_flags(mapping, index,
2376 FGP_LOCK | FGP_ACCESSED);
2377 else
2378 page = find_lock_page(mapping, index);
2379 if (page)
2380 return page;
2381
2382 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2383 f2fs_show_injection_info(F2FS_M_SB(mapping),
2384 FAULT_PAGE_ALLOC);
2385 return NULL;
2386 }
2387 }
2388
2389 if (!for_write)
2390 return grab_cache_page(mapping, index);
2391 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2392}
2393
2394static inline struct page *f2fs_pagecache_get_page(
2395 struct address_space *mapping, pgoff_t index,
2396 int fgp_flags, gfp_t gfp_mask)
2397{
2398 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2399 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2400 return NULL;
2401 }
2402
2403 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2404}
2405
2406static inline void f2fs_copy_page(struct page *src, struct page *dst)
2407{
2408 char *src_kaddr = kmap(src);
2409 char *dst_kaddr = kmap(dst);
2410
2411 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2412 kunmap(dst);
2413 kunmap(src);
2414}
2415
2416static inline void f2fs_put_page(struct page *page, int unlock)
2417{
2418 if (!page)
2419 return;
2420
2421 if (unlock) {
2422 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2423 unlock_page(page);
2424 }
2425 put_page(page);
2426}
2427
2428static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2429{
2430 if (dn->node_page)
2431 f2fs_put_page(dn->node_page, 1);
2432 if (dn->inode_page && dn->node_page != dn->inode_page)
2433 f2fs_put_page(dn->inode_page, 0);
2434 dn->node_page = NULL;
2435 dn->inode_page = NULL;
2436}
2437
2438static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2439 size_t size)
2440{
2441 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2442}
2443
2444static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2445 gfp_t flags)
2446{
2447 void *entry;
2448
2449 entry = kmem_cache_alloc(cachep, flags);
2450 if (!entry)
2451 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2452 return entry;
2453}
2454
2455static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2456{
2457 if (sbi->gc_mode == GC_URGENT)
2458 return true;
2459
2460 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2461 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2462 get_pages(sbi, F2FS_WB_CP_DATA) ||
2463 get_pages(sbi, F2FS_DIO_READ) ||
2464 get_pages(sbi, F2FS_DIO_WRITE))
2465 return false;
2466
2467 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2468 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2469 return false;
2470
2471 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2472 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2473 return false;
2474
2475 return f2fs_time_over(sbi, type);
2476}
2477
2478static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2479 unsigned long index, void *item)
2480{
2481 while (radix_tree_insert(root, index, item))
2482 cond_resched();
2483}
2484
2485#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2486
2487static inline bool IS_INODE(struct page *page)
2488{
2489 struct f2fs_node *p = F2FS_NODE(page);
2490
2491 return RAW_IS_INODE(p);
2492}
2493
2494static inline int offset_in_addr(struct f2fs_inode *i)
2495{
2496 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2497 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2498}
2499
2500static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2501{
2502 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2503}
2504
2505static inline int f2fs_has_extra_attr(struct inode *inode);
2506static inline block_t data_blkaddr(struct inode *inode,
2507 struct page *node_page, unsigned int offset)
2508{
2509 struct f2fs_node *raw_node;
2510 __le32 *addr_array;
2511 int base = 0;
2512 bool is_inode = IS_INODE(node_page);
2513
2514 raw_node = F2FS_NODE(node_page);
2515
2516 if (is_inode) {
2517 if (!inode)
2518 /* from GC path only */
2519 base = offset_in_addr(&raw_node->i);
2520 else if (f2fs_has_extra_attr(inode))
2521 base = get_extra_isize(inode);
2522 }
2523
2524 addr_array = blkaddr_in_node(raw_node);
2525 return le32_to_cpu(addr_array[base + offset]);
2526}
2527
2528static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2529{
2530 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2531}
2532
2533static inline int f2fs_test_bit(unsigned int nr, char *addr)
2534{
2535 int mask;
2536
2537 addr += (nr >> 3);
2538 mask = 1 << (7 - (nr & 0x07));
2539 return mask & *addr;
2540}
2541
2542static inline void f2fs_set_bit(unsigned int nr, char *addr)
2543{
2544 int mask;
2545
2546 addr += (nr >> 3);
2547 mask = 1 << (7 - (nr & 0x07));
2548 *addr |= mask;
2549}
2550
2551static inline void f2fs_clear_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 int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2561{
2562 int mask;
2563 int ret;
2564
2565 addr += (nr >> 3);
2566 mask = 1 << (7 - (nr & 0x07));
2567 ret = mask & *addr;
2568 *addr |= mask;
2569 return ret;
2570}
2571
2572static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2573{
2574 int mask;
2575 int ret;
2576
2577 addr += (nr >> 3);
2578 mask = 1 << (7 - (nr & 0x07));
2579 ret = mask & *addr;
2580 *addr &= ~mask;
2581 return ret;
2582}
2583
2584static inline void f2fs_change_bit(unsigned int nr, char *addr)
2585{
2586 int mask;
2587
2588 addr += (nr >> 3);
2589 mask = 1 << (7 - (nr & 0x07));
2590 *addr ^= mask;
2591}
2592
2593/*
2594 * On-disk inode flags (f2fs_inode::i_flags)
2595 */
2596#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2597#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2598#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2599#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2600#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2601#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2602#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2603#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2604#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2605#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2606#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2607
2608/* Flags that should be inherited by new inodes from their parent. */
2609#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2610 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2611 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2612
2613/* Flags that are appropriate for regular files (all but dir-specific ones). */
2614#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2615 F2FS_CASEFOLD_FL))
2616
2617/* Flags that are appropriate for non-directories/regular files. */
2618#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2619
2620static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2621{
2622 if (S_ISDIR(mode))
2623 return flags;
2624 else if (S_ISREG(mode))
2625 return flags & F2FS_REG_FLMASK;
2626 else
2627 return flags & F2FS_OTHER_FLMASK;
2628}
2629
2630static inline void __mark_inode_dirty_flag(struct inode *inode,
2631 int flag, bool set)
2632{
2633 switch (flag) {
2634 case FI_INLINE_XATTR:
2635 case FI_INLINE_DATA:
2636 case FI_INLINE_DENTRY:
2637 case FI_NEW_INODE:
2638 if (set)
2639 return;
2640 /* fall through */
2641 case FI_DATA_EXIST:
2642 case FI_INLINE_DOTS:
2643 case FI_PIN_FILE:
2644 f2fs_mark_inode_dirty_sync(inode, true);
2645 }
2646}
2647
2648static inline void set_inode_flag(struct inode *inode, int flag)
2649{
2650 test_and_set_bit(flag, F2FS_I(inode)->flags);
2651 __mark_inode_dirty_flag(inode, flag, true);
2652}
2653
2654static inline int is_inode_flag_set(struct inode *inode, int flag)
2655{
2656 return test_bit(flag, F2FS_I(inode)->flags);
2657}
2658
2659static inline void clear_inode_flag(struct inode *inode, int flag)
2660{
2661 test_and_clear_bit(flag, F2FS_I(inode)->flags);
2662 __mark_inode_dirty_flag(inode, flag, false);
2663}
2664
2665static inline bool f2fs_verity_in_progress(struct inode *inode)
2666{
2667 return IS_ENABLED(CONFIG_FS_VERITY) &&
2668 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2669}
2670
2671static inline void set_acl_inode(struct inode *inode, umode_t mode)
2672{
2673 F2FS_I(inode)->i_acl_mode = mode;
2674 set_inode_flag(inode, FI_ACL_MODE);
2675 f2fs_mark_inode_dirty_sync(inode, false);
2676}
2677
2678static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2679{
2680 if (inc)
2681 inc_nlink(inode);
2682 else
2683 drop_nlink(inode);
2684 f2fs_mark_inode_dirty_sync(inode, true);
2685}
2686
2687static inline void f2fs_i_blocks_write(struct inode *inode,
2688 block_t diff, bool add, bool claim)
2689{
2690 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2691 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2692
2693 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2694 if (add) {
2695 if (claim)
2696 dquot_claim_block(inode, diff);
2697 else
2698 dquot_alloc_block_nofail(inode, diff);
2699 } else {
2700 dquot_free_block(inode, diff);
2701 }
2702
2703 f2fs_mark_inode_dirty_sync(inode, true);
2704 if (clean || recover)
2705 set_inode_flag(inode, FI_AUTO_RECOVER);
2706}
2707
2708static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2709{
2710 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2711 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2712
2713 if (i_size_read(inode) == i_size)
2714 return;
2715
2716 i_size_write(inode, i_size);
2717 f2fs_mark_inode_dirty_sync(inode, true);
2718 if (clean || recover)
2719 set_inode_flag(inode, FI_AUTO_RECOVER);
2720}
2721
2722static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2723{
2724 F2FS_I(inode)->i_current_depth = depth;
2725 f2fs_mark_inode_dirty_sync(inode, true);
2726}
2727
2728static inline void f2fs_i_gc_failures_write(struct inode *inode,
2729 unsigned int count)
2730{
2731 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2732 f2fs_mark_inode_dirty_sync(inode, true);
2733}
2734
2735static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2736{
2737 F2FS_I(inode)->i_xattr_nid = xnid;
2738 f2fs_mark_inode_dirty_sync(inode, true);
2739}
2740
2741static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2742{
2743 F2FS_I(inode)->i_pino = pino;
2744 f2fs_mark_inode_dirty_sync(inode, true);
2745}
2746
2747static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2748{
2749 struct f2fs_inode_info *fi = F2FS_I(inode);
2750
2751 if (ri->i_inline & F2FS_INLINE_XATTR)
2752 set_bit(FI_INLINE_XATTR, fi->flags);
2753 if (ri->i_inline & F2FS_INLINE_DATA)
2754 set_bit(FI_INLINE_DATA, fi->flags);
2755 if (ri->i_inline & F2FS_INLINE_DENTRY)
2756 set_bit(FI_INLINE_DENTRY, fi->flags);
2757 if (ri->i_inline & F2FS_DATA_EXIST)
2758 set_bit(FI_DATA_EXIST, fi->flags);
2759 if (ri->i_inline & F2FS_INLINE_DOTS)
2760 set_bit(FI_INLINE_DOTS, fi->flags);
2761 if (ri->i_inline & F2FS_EXTRA_ATTR)
2762 set_bit(FI_EXTRA_ATTR, fi->flags);
2763 if (ri->i_inline & F2FS_PIN_FILE)
2764 set_bit(FI_PIN_FILE, fi->flags);
2765}
2766
2767static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2768{
2769 ri->i_inline = 0;
2770
2771 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2772 ri->i_inline |= F2FS_INLINE_XATTR;
2773 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2774 ri->i_inline |= F2FS_INLINE_DATA;
2775 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2776 ri->i_inline |= F2FS_INLINE_DENTRY;
2777 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2778 ri->i_inline |= F2FS_DATA_EXIST;
2779 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2780 ri->i_inline |= F2FS_INLINE_DOTS;
2781 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2782 ri->i_inline |= F2FS_EXTRA_ATTR;
2783 if (is_inode_flag_set(inode, FI_PIN_FILE))
2784 ri->i_inline |= F2FS_PIN_FILE;
2785}
2786
2787static inline int f2fs_has_extra_attr(struct inode *inode)
2788{
2789 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2790}
2791
2792static inline int f2fs_has_inline_xattr(struct inode *inode)
2793{
2794 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2795}
2796
2797static inline int f2fs_compressed_file(struct inode *inode)
2798{
2799 return S_ISREG(inode->i_mode) &&
2800 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2801}
2802
2803static inline unsigned int addrs_per_inode(struct inode *inode)
2804{
2805 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
2806 get_inline_xattr_addrs(inode);
2807
2808 if (!f2fs_compressed_file(inode))
2809 return addrs;
2810 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
2811}
2812
2813static inline unsigned int addrs_per_block(struct inode *inode)
2814{
2815 if (!f2fs_compressed_file(inode))
2816 return DEF_ADDRS_PER_BLOCK;
2817 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
2818}
2819
2820static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
2821{
2822 struct f2fs_inode *ri = F2FS_INODE(page);
2823
2824 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
2825 get_inline_xattr_addrs(inode)]);
2826}
2827
2828static inline int inline_xattr_size(struct inode *inode)
2829{
2830 if (f2fs_has_inline_xattr(inode))
2831 return get_inline_xattr_addrs(inode) * sizeof(__le32);
2832 return 0;
2833}
2834
2835static inline int f2fs_has_inline_data(struct inode *inode)
2836{
2837 return is_inode_flag_set(inode, FI_INLINE_DATA);
2838}
2839
2840static inline int f2fs_exist_data(struct inode *inode)
2841{
2842 return is_inode_flag_set(inode, FI_DATA_EXIST);
2843}
2844
2845static inline int f2fs_has_inline_dots(struct inode *inode)
2846{
2847 return is_inode_flag_set(inode, FI_INLINE_DOTS);
2848}
2849
2850static inline int f2fs_is_mmap_file(struct inode *inode)
2851{
2852 return is_inode_flag_set(inode, FI_MMAP_FILE);
2853}
2854
2855static inline bool f2fs_is_pinned_file(struct inode *inode)
2856{
2857 return is_inode_flag_set(inode, FI_PIN_FILE);
2858}
2859
2860static inline bool f2fs_is_atomic_file(struct inode *inode)
2861{
2862 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
2863}
2864
2865static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
2866{
2867 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
2868}
2869
2870static inline bool f2fs_is_volatile_file(struct inode *inode)
2871{
2872 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
2873}
2874
2875static inline bool f2fs_is_first_block_written(struct inode *inode)
2876{
2877 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
2878}
2879
2880static inline bool f2fs_is_drop_cache(struct inode *inode)
2881{
2882 return is_inode_flag_set(inode, FI_DROP_CACHE);
2883}
2884
2885static inline void *inline_data_addr(struct inode *inode, struct page *page)
2886{
2887 struct f2fs_inode *ri = F2FS_INODE(page);
2888 int extra_size = get_extra_isize(inode);
2889
2890 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
2891}
2892
2893static inline int f2fs_has_inline_dentry(struct inode *inode)
2894{
2895 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
2896}
2897
2898static inline int is_file(struct inode *inode, int type)
2899{
2900 return F2FS_I(inode)->i_advise & type;
2901}
2902
2903static inline void set_file(struct inode *inode, int type)
2904{
2905 F2FS_I(inode)->i_advise |= type;
2906 f2fs_mark_inode_dirty_sync(inode, true);
2907}
2908
2909static inline void clear_file(struct inode *inode, int type)
2910{
2911 F2FS_I(inode)->i_advise &= ~type;
2912 f2fs_mark_inode_dirty_sync(inode, true);
2913}
2914
2915static inline bool f2fs_is_time_consistent(struct inode *inode)
2916{
2917 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
2918 return false;
2919 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
2920 return false;
2921 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
2922 return false;
2923 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
2924 &F2FS_I(inode)->i_crtime))
2925 return false;
2926 return true;
2927}
2928
2929static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
2930{
2931 bool ret;
2932
2933 if (dsync) {
2934 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2935
2936 spin_lock(&sbi->inode_lock[DIRTY_META]);
2937 ret = list_empty(&F2FS_I(inode)->gdirty_list);
2938 spin_unlock(&sbi->inode_lock[DIRTY_META]);
2939 return ret;
2940 }
2941 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
2942 file_keep_isize(inode) ||
2943 i_size_read(inode) & ~PAGE_MASK)
2944 return false;
2945
2946 if (!f2fs_is_time_consistent(inode))
2947 return false;
2948
2949 spin_lock(&F2FS_I(inode)->i_size_lock);
2950 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
2951 spin_unlock(&F2FS_I(inode)->i_size_lock);
2952
2953 return ret;
2954}
2955
2956static inline bool f2fs_readonly(struct super_block *sb)
2957{
2958 return sb_rdonly(sb);
2959}
2960
2961static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
2962{
2963 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
2964}
2965
2966static inline bool is_dot_dotdot(const u8 *name, size_t len)
2967{
2968 if (len == 1 && name[0] == '.')
2969 return true;
2970
2971 if (len == 2 && name[0] == '.' && name[1] == '.')
2972 return true;
2973
2974 return false;
2975}
2976
2977static inline bool f2fs_may_extent_tree(struct inode *inode)
2978{
2979 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2980
2981 if (!test_opt(sbi, EXTENT_CACHE) ||
2982 is_inode_flag_set(inode, FI_NO_EXTENT) ||
2983 is_inode_flag_set(inode, FI_COMPRESSED_FILE))
2984 return false;
2985
2986 /*
2987 * for recovered files during mount do not create extents
2988 * if shrinker is not registered.
2989 */
2990 if (list_empty(&sbi->s_list))
2991 return false;
2992
2993 return S_ISREG(inode->i_mode);
2994}
2995
2996static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
2997 size_t size, gfp_t flags)
2998{
2999 void *ret;
3000
3001 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3002 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3003 return NULL;
3004 }
3005
3006 ret = kmalloc(size, flags);
3007 if (ret)
3008 return ret;
3009
3010 return kvmalloc(size, flags);
3011}
3012
3013static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3014 size_t size, gfp_t flags)
3015{
3016 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3017}
3018
3019static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3020 size_t size, gfp_t flags)
3021{
3022 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3023 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3024 return NULL;
3025 }
3026
3027 return kvmalloc(size, flags);
3028}
3029
3030static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3031 size_t size, gfp_t flags)
3032{
3033 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3034}
3035
3036static inline int get_extra_isize(struct inode *inode)
3037{
3038 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3039}
3040
3041static inline int get_inline_xattr_addrs(struct inode *inode)
3042{
3043 return F2FS_I(inode)->i_inline_xattr_size;
3044}
3045
3046#define f2fs_get_inode_mode(i) \
3047 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3048 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3049
3050#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3051 (offsetof(struct f2fs_inode, i_extra_end) - \
3052 offsetof(struct f2fs_inode, i_extra_isize)) \
3053
3054#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3055#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3056 ((offsetof(typeof(*(f2fs_inode)), field) + \
3057 sizeof((f2fs_inode)->field)) \
3058 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3059
3060#define DEFAULT_IOSTAT_PERIOD_MS 3000
3061#define MIN_IOSTAT_PERIOD_MS 100
3062/* maximum period of iostat tracing is 1 day */
3063#define MAX_IOSTAT_PERIOD_MS 8640000
3064
3065static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3066{
3067 int i;
3068
3069 spin_lock(&sbi->iostat_lock);
3070 for (i = 0; i < NR_IO_TYPE; i++) {
3071 sbi->rw_iostat[i] = 0;
3072 sbi->prev_rw_iostat[i] = 0;
3073 }
3074 spin_unlock(&sbi->iostat_lock);
3075}
3076
3077extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3078
3079static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3080 enum iostat_type type, unsigned long long io_bytes)
3081{
3082 if (!sbi->iostat_enable)
3083 return;
3084 spin_lock(&sbi->iostat_lock);
3085 sbi->rw_iostat[type] += io_bytes;
3086
3087 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3088 sbi->rw_iostat[APP_BUFFERED_IO] =
3089 sbi->rw_iostat[APP_WRITE_IO] -
3090 sbi->rw_iostat[APP_DIRECT_IO];
3091
3092 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3093 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3094 sbi->rw_iostat[APP_READ_IO] -
3095 sbi->rw_iostat[APP_DIRECT_READ_IO];
3096 spin_unlock(&sbi->iostat_lock);
3097
3098 f2fs_record_iostat(sbi);
3099}
3100
3101#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3102
3103#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3104
3105bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3106 block_t blkaddr, int type);
3107static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3108 block_t blkaddr, int type)
3109{
3110 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3111 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3112 blkaddr, type);
3113 f2fs_bug_on(sbi, 1);
3114 }
3115}
3116
3117static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3118{
3119 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3120 blkaddr == COMPRESS_ADDR)
3121 return false;
3122 return true;
3123}
3124
3125static inline void f2fs_set_page_private(struct page *page,
3126 unsigned long data)
3127{
3128 if (PagePrivate(page))
3129 return;
3130
3131 get_page(page);
3132 SetPagePrivate(page);
3133 set_page_private(page, data);
3134}
3135
3136static inline void f2fs_clear_page_private(struct page *page)
3137{
3138 if (!PagePrivate(page))
3139 return;
3140
3141 set_page_private(page, 0);
3142 ClearPagePrivate(page);
3143 f2fs_put_page(page, 0);
3144}
3145
3146/*
3147 * file.c
3148 */
3149int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3150void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3151int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3152int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3153int f2fs_truncate(struct inode *inode);
3154int f2fs_getattr(const struct path *path, struct kstat *stat,
3155 u32 request_mask, unsigned int flags);
3156int f2fs_setattr(struct dentry *dentry, struct iattr *attr);
3157int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3158void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3159int f2fs_precache_extents(struct inode *inode);
3160long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3161long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3162int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3163int f2fs_pin_file_control(struct inode *inode, bool inc);
3164
3165/*
3166 * inode.c
3167 */
3168void f2fs_set_inode_flags(struct inode *inode);
3169bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3170void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3171struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3172struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3173int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3174void f2fs_update_inode(struct inode *inode, struct page *node_page);
3175void f2fs_update_inode_page(struct inode *inode);
3176int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3177void f2fs_evict_inode(struct inode *inode);
3178void f2fs_handle_failed_inode(struct inode *inode);
3179
3180/*
3181 * namei.c
3182 */
3183int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3184 bool hot, bool set);
3185struct dentry *f2fs_get_parent(struct dentry *child);
3186
3187/*
3188 * dir.c
3189 */
3190unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3191int f2fs_init_casefolded_name(const struct inode *dir,
3192 struct f2fs_filename *fname);
3193int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3194 int lookup, struct f2fs_filename *fname);
3195int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3196 struct f2fs_filename *fname);
3197void f2fs_free_filename(struct f2fs_filename *fname);
3198struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3199 const struct f2fs_filename *fname, int *max_slots);
3200int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3201 unsigned int start_pos, struct fscrypt_str *fstr);
3202void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3203 struct f2fs_dentry_ptr *d);
3204struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3205 const struct f2fs_filename *fname, struct page *dpage);
3206void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3207 unsigned int current_depth);
3208int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3209void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3210struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3211 const struct f2fs_filename *fname,
3212 struct page **res_page);
3213struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3214 const struct qstr *child, struct page **res_page);
3215struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3216ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3217 struct page **page);
3218void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3219 struct page *page, struct inode *inode);
3220bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3221 const struct f2fs_filename *fname);
3222void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3223 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3224 unsigned int bit_pos);
3225int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3226 struct inode *inode, nid_t ino, umode_t mode);
3227int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3228 struct inode *inode, nid_t ino, umode_t mode);
3229int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3230 struct inode *inode, nid_t ino, umode_t mode);
3231void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3232 struct inode *dir, struct inode *inode);
3233int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3234bool f2fs_empty_dir(struct inode *dir);
3235
3236static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3237{
3238 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3239 inode, inode->i_ino, inode->i_mode);
3240}
3241
3242/*
3243 * super.c
3244 */
3245int f2fs_inode_dirtied(struct inode *inode, bool sync);
3246void f2fs_inode_synced(struct inode *inode);
3247int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3248int f2fs_quota_sync(struct super_block *sb, int type);
3249void f2fs_quota_off_umount(struct super_block *sb);
3250int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3251int f2fs_sync_fs(struct super_block *sb, int sync);
3252int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3253
3254/*
3255 * hash.c
3256 */
3257void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3258
3259/*
3260 * node.c
3261 */
3262struct dnode_of_data;
3263struct node_info;
3264
3265int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3266bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3267bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3268void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3269void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3270void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3271int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3272bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3273bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3274int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3275 struct node_info *ni);
3276pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3277int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3278int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3279int f2fs_truncate_xattr_node(struct inode *inode);
3280int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3281 unsigned int seq_id);
3282int f2fs_remove_inode_page(struct inode *inode);
3283struct page *f2fs_new_inode_page(struct inode *inode);
3284struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3285void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3286struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3287struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3288int f2fs_move_node_page(struct page *node_page, int gc_type);
3289int f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3290int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3291 struct writeback_control *wbc, bool atomic,
3292 unsigned int *seq_id);
3293int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3294 struct writeback_control *wbc,
3295 bool do_balance, enum iostat_type io_type);
3296int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3297bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3298void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3299void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3300int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3301void f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3302int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3303int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3304int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3305 unsigned int segno, struct f2fs_summary_block *sum);
3306int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3307int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3308void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3309int __init f2fs_create_node_manager_caches(void);
3310void f2fs_destroy_node_manager_caches(void);
3311
3312/*
3313 * segment.c
3314 */
3315bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3316void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3317void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3318void f2fs_drop_inmem_pages(struct inode *inode);
3319void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3320int f2fs_commit_inmem_pages(struct inode *inode);
3321void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3322void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3323int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3324int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3325int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3326void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3327void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3328bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3329void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3330void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3331bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3332void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3333 struct cp_control *cpc);
3334void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3335block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3336int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3337void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3338int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3339void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3340 unsigned int start, unsigned int end);
3341void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type);
3342int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3343bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3344 struct cp_control *cpc);
3345struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3346void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3347 block_t blk_addr);
3348void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3349 enum iostat_type io_type);
3350void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3351void f2fs_outplace_write_data(struct dnode_of_data *dn,
3352 struct f2fs_io_info *fio);
3353int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3354void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3355 block_t old_blkaddr, block_t new_blkaddr,
3356 bool recover_curseg, bool recover_newaddr);
3357void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3358 block_t old_addr, block_t new_addr,
3359 unsigned char version, bool recover_curseg,
3360 bool recover_newaddr);
3361void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3362 block_t old_blkaddr, block_t *new_blkaddr,
3363 struct f2fs_summary *sum, int type,
3364 struct f2fs_io_info *fio, bool add_list);
3365void f2fs_wait_on_page_writeback(struct page *page,
3366 enum page_type type, bool ordered, bool locked);
3367void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3368void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3369 block_t len);
3370void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3371void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3372int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3373 unsigned int val, int alloc);
3374void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3375int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3376int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3377int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3378void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3379int __init f2fs_create_segment_manager_caches(void);
3380void f2fs_destroy_segment_manager_caches(void);
3381int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3382enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3383 enum page_type type, enum temp_type temp);
3384
3385/*
3386 * checkpoint.c
3387 */
3388void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3389struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3390struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3391struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index);
3392struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3393bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3394 block_t blkaddr, int type);
3395int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3396 int type, bool sync);
3397void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3398long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3399 long nr_to_write, enum iostat_type io_type);
3400void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3401void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3402void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3403bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3404void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3405 unsigned int devidx, int type);
3406bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3407 unsigned int devidx, int type);
3408int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3409int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3410void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3411void f2fs_add_orphan_inode(struct inode *inode);
3412void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3413int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3414int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3415void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3416void f2fs_remove_dirty_inode(struct inode *inode);
3417int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3418void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3419int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3420void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3421int __init f2fs_create_checkpoint_caches(void);
3422void f2fs_destroy_checkpoint_caches(void);
3423
3424/*
3425 * data.c
3426 */
3427int __init f2fs_init_bioset(void);
3428void f2fs_destroy_bioset(void);
3429struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio);
3430int f2fs_init_bio_entry_cache(void);
3431void f2fs_destroy_bio_entry_cache(void);
3432void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3433 struct bio *bio, enum page_type type);
3434void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3435void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3436 struct inode *inode, struct page *page,
3437 nid_t ino, enum page_type type);
3438void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3439 struct bio **bio, struct page *page);
3440void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3441int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3442int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3443void f2fs_submit_page_write(struct f2fs_io_info *fio);
3444struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3445 block_t blk_addr, struct bio *bio);
3446int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3447void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3448void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3449int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3450int f2fs_reserve_new_block(struct dnode_of_data *dn);
3451int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3452int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3453int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3454int f2fs_mpage_readpages(struct address_space *mapping,
3455 struct list_head *pages, struct page *page,
3456 unsigned nr_pages, bool is_readahead);
3457struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3458 int op_flags, bool for_write);
3459struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3460struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3461 bool for_write);
3462struct page *f2fs_get_new_data_page(struct inode *inode,
3463 struct page *ipage, pgoff_t index, bool new_i_size);
3464int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3465void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3466int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3467 int create, int flag);
3468int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3469 u64 start, u64 len);
3470int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3471bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3472bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3473int f2fs_write_single_data_page(struct page *page, int *submitted,
3474 struct bio **bio, sector_t *last_block,
3475 struct writeback_control *wbc,
3476 enum iostat_type io_type,
3477 int compr_blocks);
3478void f2fs_invalidate_page(struct page *page, unsigned int offset,
3479 unsigned int length);
3480int f2fs_release_page(struct page *page, gfp_t wait);
3481#ifdef CONFIG_MIGRATION
3482int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3483 struct page *page, enum migrate_mode mode);
3484#endif
3485bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3486void f2fs_clear_page_cache_dirty_tag(struct page *page);
3487int f2fs_init_post_read_processing(void);
3488void f2fs_destroy_post_read_processing(void);
3489int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3490void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3491
3492/*
3493 * gc.c
3494 */
3495int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3496void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3497block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3498int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background,
3499 unsigned int segno);
3500void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3501int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3502
3503/*
3504 * recovery.c
3505 */
3506int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3507bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3508
3509/*
3510 * debug.c
3511 */
3512#ifdef CONFIG_F2FS_STAT_FS
3513struct f2fs_stat_info {
3514 struct list_head stat_list;
3515 struct f2fs_sb_info *sbi;
3516 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3517 int main_area_segs, main_area_sections, main_area_zones;
3518 unsigned long long hit_largest, hit_cached, hit_rbtree;
3519 unsigned long long hit_total, total_ext;
3520 int ext_tree, zombie_tree, ext_node;
3521 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3522 int ndirty_data, ndirty_qdata;
3523 int inmem_pages;
3524 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3525 int nats, dirty_nats, sits, dirty_sits;
3526 int free_nids, avail_nids, alloc_nids;
3527 int total_count, utilization;
3528 int bg_gc, nr_wb_cp_data, nr_wb_data;
3529 int nr_rd_data, nr_rd_node, nr_rd_meta;
3530 int nr_dio_read, nr_dio_write;
3531 unsigned int io_skip_bggc, other_skip_bggc;
3532 int nr_flushing, nr_flushed, flush_list_empty;
3533 int nr_discarding, nr_discarded;
3534 int nr_discard_cmd;
3535 unsigned int undiscard_blks;
3536 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3537 int compr_inode, compr_blocks;
3538 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3539 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3540 unsigned int bimodal, avg_vblocks;
3541 int util_free, util_valid, util_invalid;
3542 int rsvd_segs, overp_segs;
3543 int dirty_count, node_pages, meta_pages;
3544 int prefree_count, call_count, cp_count, bg_cp_count;
3545 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3546 int bg_node_segs, bg_data_segs;
3547 int tot_blks, data_blks, node_blks;
3548 int bg_data_blks, bg_node_blks;
3549 unsigned long long skipped_atomic_files[2];
3550 int curseg[NR_CURSEG_TYPE];
3551 int cursec[NR_CURSEG_TYPE];
3552 int curzone[NR_CURSEG_TYPE];
3553
3554 unsigned int meta_count[META_MAX];
3555 unsigned int segment_count[2];
3556 unsigned int block_count[2];
3557 unsigned int inplace_count;
3558 unsigned long long base_mem, cache_mem, page_mem;
3559};
3560
3561static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3562{
3563 return (struct f2fs_stat_info *)sbi->stat_info;
3564}
3565
3566#define stat_inc_cp_count(si) ((si)->cp_count++)
3567#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3568#define stat_inc_call_count(si) ((si)->call_count++)
3569#define stat_inc_bggc_count(si) ((si)->bg_gc++)
3570#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3571#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3572#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3573#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3574#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3575#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3576#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3577#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3578#define stat_inc_inline_xattr(inode) \
3579 do { \
3580 if (f2fs_has_inline_xattr(inode)) \
3581 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3582 } while (0)
3583#define stat_dec_inline_xattr(inode) \
3584 do { \
3585 if (f2fs_has_inline_xattr(inode)) \
3586 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3587 } while (0)
3588#define stat_inc_inline_inode(inode) \
3589 do { \
3590 if (f2fs_has_inline_data(inode)) \
3591 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3592 } while (0)
3593#define stat_dec_inline_inode(inode) \
3594 do { \
3595 if (f2fs_has_inline_data(inode)) \
3596 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3597 } while (0)
3598#define stat_inc_inline_dir(inode) \
3599 do { \
3600 if (f2fs_has_inline_dentry(inode)) \
3601 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3602 } while (0)
3603#define stat_dec_inline_dir(inode) \
3604 do { \
3605 if (f2fs_has_inline_dentry(inode)) \
3606 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3607 } while (0)
3608#define stat_inc_compr_inode(inode) \
3609 do { \
3610 if (f2fs_compressed_file(inode)) \
3611 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3612 } while (0)
3613#define stat_dec_compr_inode(inode) \
3614 do { \
3615 if (f2fs_compressed_file(inode)) \
3616 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3617 } while (0)
3618#define stat_add_compr_blocks(inode, blocks) \
3619 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3620#define stat_sub_compr_blocks(inode, blocks) \
3621 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3622#define stat_inc_meta_count(sbi, blkaddr) \
3623 do { \
3624 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3625 atomic_inc(&(sbi)->meta_count[META_CP]); \
3626 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3627 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3628 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3629 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3630 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3631 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3632 } while (0)
3633#define stat_inc_seg_type(sbi, curseg) \
3634 ((sbi)->segment_count[(curseg)->alloc_type]++)
3635#define stat_inc_block_count(sbi, curseg) \
3636 ((sbi)->block_count[(curseg)->alloc_type]++)
3637#define stat_inc_inplace_blocks(sbi) \
3638 (atomic_inc(&(sbi)->inplace_count))
3639#define stat_update_max_atomic_write(inode) \
3640 do { \
3641 int cur = F2FS_I_SB(inode)->atomic_files; \
3642 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3643 if (cur > max) \
3644 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3645 } while (0)
3646#define stat_inc_volatile_write(inode) \
3647 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3648#define stat_dec_volatile_write(inode) \
3649 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3650#define stat_update_max_volatile_write(inode) \
3651 do { \
3652 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3653 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3654 if (cur > max) \
3655 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3656 } while (0)
3657#define stat_inc_seg_count(sbi, type, gc_type) \
3658 do { \
3659 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3660 si->tot_segs++; \
3661 if ((type) == SUM_TYPE_DATA) { \
3662 si->data_segs++; \
3663 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3664 } else { \
3665 si->node_segs++; \
3666 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3667 } \
3668 } while (0)
3669
3670#define stat_inc_tot_blk_count(si, blks) \
3671 ((si)->tot_blks += (blks))
3672
3673#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3674 do { \
3675 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3676 stat_inc_tot_blk_count(si, blks); \
3677 si->data_blks += (blks); \
3678 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3679 } while (0)
3680
3681#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3682 do { \
3683 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3684 stat_inc_tot_blk_count(si, blks); \
3685 si->node_blks += (blks); \
3686 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3687 } while (0)
3688
3689int f2fs_build_stats(struct f2fs_sb_info *sbi);
3690void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3691void __init f2fs_create_root_stats(void);
3692void f2fs_destroy_root_stats(void);
3693void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3694#else
3695#define stat_inc_cp_count(si) do { } while (0)
3696#define stat_inc_bg_cp_count(si) do { } while (0)
3697#define stat_inc_call_count(si) do { } while (0)
3698#define stat_inc_bggc_count(si) do { } while (0)
3699#define stat_io_skip_bggc_count(sbi) do { } while (0)
3700#define stat_other_skip_bggc_count(sbi) do { } while (0)
3701#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3702#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3703#define stat_inc_total_hit(sbi) do { } while (0)
3704#define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3705#define stat_inc_largest_node_hit(sbi) do { } while (0)
3706#define stat_inc_cached_node_hit(sbi) do { } while (0)
3707#define stat_inc_inline_xattr(inode) do { } while (0)
3708#define stat_dec_inline_xattr(inode) do { } while (0)
3709#define stat_inc_inline_inode(inode) do { } while (0)
3710#define stat_dec_inline_inode(inode) do { } while (0)
3711#define stat_inc_inline_dir(inode) do { } while (0)
3712#define stat_dec_inline_dir(inode) do { } while (0)
3713#define stat_inc_compr_inode(inode) do { } while (0)
3714#define stat_dec_compr_inode(inode) do { } while (0)
3715#define stat_add_compr_blocks(inode, blocks) do { } while (0)
3716#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3717#define stat_inc_atomic_write(inode) do { } while (0)
3718#define stat_dec_atomic_write(inode) do { } while (0)
3719#define stat_update_max_atomic_write(inode) do { } while (0)
3720#define stat_inc_volatile_write(inode) do { } while (0)
3721#define stat_dec_volatile_write(inode) do { } while (0)
3722#define stat_update_max_volatile_write(inode) do { } while (0)
3723#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3724#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3725#define stat_inc_block_count(sbi, curseg) do { } while (0)
3726#define stat_inc_inplace_blocks(sbi) do { } while (0)
3727#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3728#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3729#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3730#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3731
3732static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3733static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3734static inline void __init f2fs_create_root_stats(void) { }
3735static inline void f2fs_destroy_root_stats(void) { }
3736static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3737#endif
3738
3739extern const struct file_operations f2fs_dir_operations;
3740#ifdef CONFIG_UNICODE
3741extern const struct dentry_operations f2fs_dentry_ops;
3742#endif
3743extern const struct file_operations f2fs_file_operations;
3744extern const struct inode_operations f2fs_file_inode_operations;
3745extern const struct address_space_operations f2fs_dblock_aops;
3746extern const struct address_space_operations f2fs_node_aops;
3747extern const struct address_space_operations f2fs_meta_aops;
3748extern const struct inode_operations f2fs_dir_inode_operations;
3749extern const struct inode_operations f2fs_symlink_inode_operations;
3750extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3751extern const struct inode_operations f2fs_special_inode_operations;
3752extern struct kmem_cache *f2fs_inode_entry_slab;
3753
3754/*
3755 * inline.c
3756 */
3757bool f2fs_may_inline_data(struct inode *inode);
3758bool f2fs_may_inline_dentry(struct inode *inode);
3759void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3760void f2fs_truncate_inline_inode(struct inode *inode,
3761 struct page *ipage, u64 from);
3762int f2fs_read_inline_data(struct inode *inode, struct page *page);
3763int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3764int f2fs_convert_inline_inode(struct inode *inode);
3765int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3766int f2fs_write_inline_data(struct inode *inode, struct page *page);
3767bool f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3768struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3769 const struct f2fs_filename *fname,
3770 struct page **res_page);
3771int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3772 struct page *ipage);
3773int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3774 struct inode *inode, nid_t ino, umode_t mode);
3775void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3776 struct page *page, struct inode *dir,
3777 struct inode *inode);
3778bool f2fs_empty_inline_dir(struct inode *dir);
3779int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3780 struct fscrypt_str *fstr);
3781int f2fs_inline_data_fiemap(struct inode *inode,
3782 struct fiemap_extent_info *fieinfo,
3783 __u64 start, __u64 len);
3784
3785/*
3786 * shrinker.c
3787 */
3788unsigned long f2fs_shrink_count(struct shrinker *shrink,
3789 struct shrink_control *sc);
3790unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3791 struct shrink_control *sc);
3792void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3793void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3794
3795/*
3796 * extent_cache.c
3797 */
3798struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3799 struct rb_entry *cached_re, unsigned int ofs);
3800struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3801 struct rb_root_cached *root,
3802 struct rb_node **parent,
3803 unsigned int ofs, bool *leftmost);
3804struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3805 struct rb_entry *cached_re, unsigned int ofs,
3806 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3807 struct rb_node ***insert_p, struct rb_node **insert_parent,
3808 bool force, bool *leftmost);
3809bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3810 struct rb_root_cached *root);
3811unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3812bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext);
3813void f2fs_drop_extent_tree(struct inode *inode);
3814unsigned int f2fs_destroy_extent_node(struct inode *inode);
3815void f2fs_destroy_extent_tree(struct inode *inode);
3816bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3817 struct extent_info *ei);
3818void f2fs_update_extent_cache(struct dnode_of_data *dn);
3819void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3820 pgoff_t fofs, block_t blkaddr, unsigned int len);
3821void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3822int __init f2fs_create_extent_cache(void);
3823void f2fs_destroy_extent_cache(void);
3824
3825/*
3826 * sysfs.c
3827 */
3828int __init f2fs_init_sysfs(void);
3829void f2fs_exit_sysfs(void);
3830int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3831void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3832
3833/* verity.c */
3834extern const struct fsverity_operations f2fs_verityops;
3835
3836/*
3837 * crypto support
3838 */
3839static inline bool f2fs_encrypted_file(struct inode *inode)
3840{
3841 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
3842}
3843
3844static inline void f2fs_set_encrypted_inode(struct inode *inode)
3845{
3846#ifdef CONFIG_FS_ENCRYPTION
3847 file_set_encrypt(inode);
3848 f2fs_set_inode_flags(inode);
3849#endif
3850}
3851
3852/*
3853 * Returns true if the reads of the inode's data need to undergo some
3854 * postprocessing step, like decryption or authenticity verification.
3855 */
3856static inline bool f2fs_post_read_required(struct inode *inode)
3857{
3858 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
3859 f2fs_compressed_file(inode);
3860}
3861
3862/*
3863 * compress.c
3864 */
3865#ifdef CONFIG_F2FS_FS_COMPRESSION
3866bool f2fs_is_compressed_page(struct page *page);
3867struct page *f2fs_compress_control_page(struct page *page);
3868int f2fs_prepare_compress_overwrite(struct inode *inode,
3869 struct page **pagep, pgoff_t index, void **fsdata);
3870bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
3871 pgoff_t index, unsigned copied);
3872int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
3873void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
3874bool f2fs_is_compress_backend_ready(struct inode *inode);
3875int f2fs_init_compress_mempool(void);
3876void f2fs_destroy_compress_mempool(void);
3877void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity);
3878bool f2fs_cluster_is_empty(struct compress_ctx *cc);
3879bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
3880void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
3881int f2fs_write_multi_pages(struct compress_ctx *cc,
3882 int *submitted,
3883 struct writeback_control *wbc,
3884 enum iostat_type io_type);
3885int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
3886int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
3887 unsigned nr_pages, sector_t *last_block_in_bio,
3888 bool is_readahead, bool for_write);
3889struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
3890void f2fs_free_dic(struct decompress_io_ctx *dic);
3891void f2fs_decompress_end_io(struct page **rpages,
3892 unsigned int cluster_size, bool err, bool verity);
3893int f2fs_init_compress_ctx(struct compress_ctx *cc);
3894void f2fs_destroy_compress_ctx(struct compress_ctx *cc);
3895void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
3896#else
3897static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
3898static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
3899{
3900 if (!f2fs_compressed_file(inode))
3901 return true;
3902 /* not support compression */
3903 return false;
3904}
3905static inline struct page *f2fs_compress_control_page(struct page *page)
3906{
3907 WARN_ON_ONCE(1);
3908 return ERR_PTR(-EINVAL);
3909}
3910static inline int f2fs_init_compress_mempool(void) { return 0; }
3911static inline void f2fs_destroy_compress_mempool(void) { }
3912#endif
3913
3914static inline void set_compress_context(struct inode *inode)
3915{
3916 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3917
3918 F2FS_I(inode)->i_compress_algorithm =
3919 F2FS_OPTION(sbi).compress_algorithm;
3920 F2FS_I(inode)->i_log_cluster_size =
3921 F2FS_OPTION(sbi).compress_log_size;
3922 F2FS_I(inode)->i_cluster_size =
3923 1 << F2FS_I(inode)->i_log_cluster_size;
3924 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
3925 set_inode_flag(inode, FI_COMPRESSED_FILE);
3926 stat_inc_compr_inode(inode);
3927 f2fs_mark_inode_dirty_sync(inode, true);
3928}
3929
3930static inline u64 f2fs_disable_compressed_file(struct inode *inode)
3931{
3932 struct f2fs_inode_info *fi = F2FS_I(inode);
3933
3934 if (!f2fs_compressed_file(inode))
3935 return 0;
3936 if (S_ISREG(inode->i_mode)) {
3937 if (get_dirty_pages(inode))
3938 return 1;
3939 if (fi->i_compr_blocks)
3940 return fi->i_compr_blocks;
3941 }
3942
3943 fi->i_flags &= ~F2FS_COMPR_FL;
3944 stat_dec_compr_inode(inode);
3945 clear_inode_flag(inode, FI_COMPRESSED_FILE);
3946 f2fs_mark_inode_dirty_sync(inode, true);
3947 return 0;
3948}
3949
3950#define F2FS_FEATURE_FUNCS(name, flagname) \
3951static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
3952{ \
3953 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
3954}
3955
3956F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
3957F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
3958F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
3959F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
3960F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
3961F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
3962F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
3963F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
3964F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
3965F2FS_FEATURE_FUNCS(verity, VERITY);
3966F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
3967F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
3968F2FS_FEATURE_FUNCS(compression, COMPRESSION);
3969
3970#ifdef CONFIG_BLK_DEV_ZONED
3971static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
3972 block_t blkaddr)
3973{
3974 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
3975
3976 return test_bit(zno, FDEV(devi).blkz_seq);
3977}
3978#endif
3979
3980static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
3981{
3982 return f2fs_sb_has_blkzoned(sbi);
3983}
3984
3985static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
3986{
3987 return blk_queue_discard(bdev_get_queue(bdev)) ||
3988 bdev_is_zoned(bdev);
3989}
3990
3991static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
3992{
3993 int i;
3994
3995 if (!f2fs_is_multi_device(sbi))
3996 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
3997
3998 for (i = 0; i < sbi->s_ndevs; i++)
3999 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4000 return true;
4001 return false;
4002}
4003
4004static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4005{
4006 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4007 f2fs_hw_should_discard(sbi);
4008}
4009
4010static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4011{
4012 int i;
4013
4014 if (!f2fs_is_multi_device(sbi))
4015 return bdev_read_only(sbi->sb->s_bdev);
4016
4017 for (i = 0; i < sbi->s_ndevs; i++)
4018 if (bdev_read_only(FDEV(i).bdev))
4019 return true;
4020 return false;
4021}
4022
4023static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4024{
4025 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4026}
4027
4028static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode)
4029{
4030#ifdef CONFIG_FS_ENCRYPTION
4031 struct f2fs_sb_info *sbi = F2FS_I_SB(dir);
4032 umode_t mode = inode->i_mode;
4033
4034 /*
4035 * If the directory encrypted or dummy encryption enabled,
4036 * then we should encrypt the inode.
4037 */
4038 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi))
4039 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode));
4040#endif
4041 return false;
4042}
4043
4044static inline bool f2fs_may_compress(struct inode *inode)
4045{
4046 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4047 f2fs_is_atomic_file(inode) ||
4048 f2fs_is_volatile_file(inode))
4049 return false;
4050 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4051}
4052
4053static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4054 u64 blocks, bool add)
4055{
4056 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4057
4058 /* don't update i_compr_blocks if saved blocks were released */
4059 if (!add && !F2FS_I(inode)->i_compr_blocks)
4060 return;
4061
4062 if (add) {
4063 F2FS_I(inode)->i_compr_blocks += diff;
4064 stat_add_compr_blocks(inode, diff);
4065 } else {
4066 F2FS_I(inode)->i_compr_blocks -= diff;
4067 stat_sub_compr_blocks(inode, diff);
4068 }
4069 f2fs_mark_inode_dirty_sync(inode, true);
4070}
4071
4072static inline int block_unaligned_IO(struct inode *inode,
4073 struct kiocb *iocb, struct iov_iter *iter)
4074{
4075 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4076 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4077 loff_t offset = iocb->ki_pos;
4078 unsigned long align = offset | iov_iter_alignment(iter);
4079
4080 return align & blocksize_mask;
4081}
4082
4083static inline int allow_outplace_dio(struct inode *inode,
4084 struct kiocb *iocb, struct iov_iter *iter)
4085{
4086 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4087 int rw = iov_iter_rw(iter);
4088
4089 return (f2fs_lfs_mode(sbi) && (rw == WRITE) &&
4090 !block_unaligned_IO(inode, iocb, iter));
4091}
4092
4093static inline bool f2fs_force_buffered_io(struct inode *inode,
4094 struct kiocb *iocb, struct iov_iter *iter)
4095{
4096 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4097 int rw = iov_iter_rw(iter);
4098
4099 if (f2fs_post_read_required(inode))
4100 return true;
4101 if (f2fs_is_multi_device(sbi))
4102 return true;
4103 /*
4104 * for blkzoned device, fallback direct IO to buffered IO, so
4105 * all IOs can be serialized by log-structured write.
4106 */
4107 if (f2fs_sb_has_blkzoned(sbi))
4108 return true;
4109 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4110 if (block_unaligned_IO(inode, iocb, iter))
4111 return true;
4112 if (F2FS_IO_ALIGNED(sbi))
4113 return true;
4114 }
4115 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) &&
4116 !IS_SWAPFILE(inode))
4117 return true;
4118
4119 return false;
4120}
4121
4122#ifdef CONFIG_F2FS_FAULT_INJECTION
4123extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4124 unsigned int type);
4125#else
4126#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4127#endif
4128
4129static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4130{
4131#ifdef CONFIG_QUOTA
4132 if (f2fs_sb_has_quota_ino(sbi))
4133 return true;
4134 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4135 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4136 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4137 return true;
4138#endif
4139 return false;
4140}
4141
4142#define EFSBADCRC EBADMSG /* Bad CRC detected */
4143#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4144
4145#endif /* _LINUX_F2FS_H */