]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame_incremental - fs/f2fs/f2fs.h
f2fs: add sysfs node to control ra_pages for fadvise seq file
[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 (WARN_ON(condition)) \
37 set_sbi_flag(sbi, SBI_NEED_FSCK); \
38 } while (0)
39#endif
40
41enum {
42 FAULT_KMALLOC,
43 FAULT_KVMALLOC,
44 FAULT_PAGE_ALLOC,
45 FAULT_PAGE_GET,
46 FAULT_ALLOC_NID,
47 FAULT_ORPHAN,
48 FAULT_BLOCK,
49 FAULT_DIR_DEPTH,
50 FAULT_EVICT_INODE,
51 FAULT_TRUNCATE,
52 FAULT_READ_IO,
53 FAULT_CHECKPOINT,
54 FAULT_DISCARD,
55 FAULT_WRITE_IO,
56 FAULT_MAX,
57};
58
59#ifdef CONFIG_F2FS_FAULT_INJECTION
60#define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
61
62struct f2fs_fault_info {
63 atomic_t inject_ops;
64 unsigned int inject_rate;
65 unsigned int inject_type;
66};
67
68extern const char *f2fs_fault_name[FAULT_MAX];
69#define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
70#endif
71
72/*
73 * For mount options
74 */
75#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
76#define F2FS_MOUNT_DISCARD 0x00000004
77#define F2FS_MOUNT_NOHEAP 0x00000008
78#define F2FS_MOUNT_XATTR_USER 0x00000010
79#define F2FS_MOUNT_POSIX_ACL 0x00000020
80#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
81#define F2FS_MOUNT_INLINE_XATTR 0x00000080
82#define F2FS_MOUNT_INLINE_DATA 0x00000100
83#define F2FS_MOUNT_INLINE_DENTRY 0x00000200
84#define F2FS_MOUNT_FLUSH_MERGE 0x00000400
85#define F2FS_MOUNT_NOBARRIER 0x00000800
86#define F2FS_MOUNT_FASTBOOT 0x00001000
87#define F2FS_MOUNT_EXTENT_CACHE 0x00002000
88#define F2FS_MOUNT_DATA_FLUSH 0x00008000
89#define F2FS_MOUNT_FAULT_INJECTION 0x00010000
90#define F2FS_MOUNT_USRQUOTA 0x00080000
91#define F2FS_MOUNT_GRPQUOTA 0x00100000
92#define F2FS_MOUNT_PRJQUOTA 0x00200000
93#define F2FS_MOUNT_QUOTA 0x00400000
94#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
95#define F2FS_MOUNT_RESERVE_ROOT 0x01000000
96#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
97#define F2FS_MOUNT_NORECOVERY 0x04000000
98#define F2FS_MOUNT_ATGC 0x08000000
99#define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000
100#define F2FS_MOUNT_GC_MERGE 0x20000000
101#define F2FS_MOUNT_COMPRESS_CACHE 0x40000000
102
103#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
104#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
105#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
106#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
107
108#define ver_after(a, b) (typecheck(unsigned long long, a) && \
109 typecheck(unsigned long long, b) && \
110 ((long long)((a) - (b)) > 0))
111
112typedef u32 block_t; /*
113 * should not change u32, since it is the on-disk block
114 * address format, __le32.
115 */
116typedef u32 nid_t;
117
118#define COMPRESS_EXT_NUM 16
119
120struct f2fs_mount_info {
121 unsigned int opt;
122 int write_io_size_bits; /* Write IO size bits */
123 block_t root_reserved_blocks; /* root reserved blocks */
124 kuid_t s_resuid; /* reserved blocks for uid */
125 kgid_t s_resgid; /* reserved blocks for gid */
126 int active_logs; /* # of active logs */
127 int inline_xattr_size; /* inline xattr size */
128#ifdef CONFIG_F2FS_FAULT_INJECTION
129 struct f2fs_fault_info fault_info; /* For fault injection */
130#endif
131#ifdef CONFIG_QUOTA
132 /* Names of quota files with journalled quota */
133 char *s_qf_names[MAXQUOTAS];
134 int s_jquota_fmt; /* Format of quota to use */
135#endif
136 /* For which write hints are passed down to block layer */
137 int whint_mode;
138 int alloc_mode; /* segment allocation policy */
139 int fsync_mode; /* fsync policy */
140 int fs_mode; /* fs mode: LFS or ADAPTIVE */
141 int bggc_mode; /* bggc mode: off, on or sync */
142 int discard_unit; /*
143 * discard command's offset/size should
144 * be aligned to this unit: block,
145 * segment or section
146 */
147 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
148 block_t unusable_cap_perc; /* percentage for cap */
149 block_t unusable_cap; /* Amount of space allowed to be
150 * unusable when disabling checkpoint
151 */
152
153 /* For compression */
154 unsigned char compress_algorithm; /* algorithm type */
155 unsigned char compress_log_size; /* cluster log size */
156 unsigned char compress_level; /* compress level */
157 bool compress_chksum; /* compressed data chksum */
158 unsigned char compress_ext_cnt; /* extension count */
159 unsigned char nocompress_ext_cnt; /* nocompress extension count */
160 int compress_mode; /* compression mode */
161 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
162 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
163};
164
165#define F2FS_FEATURE_ENCRYPT 0x0001
166#define F2FS_FEATURE_BLKZONED 0x0002
167#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
168#define F2FS_FEATURE_EXTRA_ATTR 0x0008
169#define F2FS_FEATURE_PRJQUOTA 0x0010
170#define F2FS_FEATURE_INODE_CHKSUM 0x0020
171#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
172#define F2FS_FEATURE_QUOTA_INO 0x0080
173#define F2FS_FEATURE_INODE_CRTIME 0x0100
174#define F2FS_FEATURE_LOST_FOUND 0x0200
175#define F2FS_FEATURE_VERITY 0x0400
176#define F2FS_FEATURE_SB_CHKSUM 0x0800
177#define F2FS_FEATURE_CASEFOLD 0x1000
178#define F2FS_FEATURE_COMPRESSION 0x2000
179#define F2FS_FEATURE_RO 0x4000
180
181#define __F2FS_HAS_FEATURE(raw_super, mask) \
182 ((raw_super->feature & cpu_to_le32(mask)) != 0)
183#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
184#define F2FS_SET_FEATURE(sbi, mask) \
185 (sbi->raw_super->feature |= cpu_to_le32(mask))
186#define F2FS_CLEAR_FEATURE(sbi, mask) \
187 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
188
189/*
190 * Default values for user and/or group using reserved blocks
191 */
192#define F2FS_DEF_RESUID 0
193#define F2FS_DEF_RESGID 0
194
195/*
196 * For checkpoint manager
197 */
198enum {
199 NAT_BITMAP,
200 SIT_BITMAP
201};
202
203#define CP_UMOUNT 0x00000001
204#define CP_FASTBOOT 0x00000002
205#define CP_SYNC 0x00000004
206#define CP_RECOVERY 0x00000008
207#define CP_DISCARD 0x00000010
208#define CP_TRIMMED 0x00000020
209#define CP_PAUSE 0x00000040
210#define CP_RESIZE 0x00000080
211
212#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
213#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
214#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
215#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
216#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
217#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
218#define DEF_CP_INTERVAL 60 /* 60 secs */
219#define DEF_IDLE_INTERVAL 5 /* 5 secs */
220#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
221#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
222#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
223
224struct cp_control {
225 int reason;
226 __u64 trim_start;
227 __u64 trim_end;
228 __u64 trim_minlen;
229};
230
231/*
232 * indicate meta/data type
233 */
234enum {
235 META_CP,
236 META_NAT,
237 META_SIT,
238 META_SSA,
239 META_MAX,
240 META_POR,
241 DATA_GENERIC, /* check range only */
242 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
243 DATA_GENERIC_ENHANCE_READ, /*
244 * strong check on range and segment
245 * bitmap but no warning due to race
246 * condition of read on truncated area
247 * by extent_cache
248 */
249 META_GENERIC,
250};
251
252/* for the list of ino */
253enum {
254 ORPHAN_INO, /* for orphan ino list */
255 APPEND_INO, /* for append ino list */
256 UPDATE_INO, /* for update ino list */
257 TRANS_DIR_INO, /* for trasactions dir ino list */
258 FLUSH_INO, /* for multiple device flushing */
259 MAX_INO_ENTRY, /* max. list */
260};
261
262struct ino_entry {
263 struct list_head list; /* list head */
264 nid_t ino; /* inode number */
265 unsigned int dirty_device; /* dirty device bitmap */
266};
267
268/* for the list of inodes to be GCed */
269struct inode_entry {
270 struct list_head list; /* list head */
271 struct inode *inode; /* vfs inode pointer */
272};
273
274struct fsync_node_entry {
275 struct list_head list; /* list head */
276 struct page *page; /* warm node page pointer */
277 unsigned int seq_id; /* sequence id */
278};
279
280struct ckpt_req {
281 struct completion wait; /* completion for checkpoint done */
282 struct llist_node llnode; /* llist_node to be linked in wait queue */
283 int ret; /* return code of checkpoint */
284 ktime_t queue_time; /* request queued time */
285};
286
287struct ckpt_req_control {
288 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
289 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
290 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
291 atomic_t issued_ckpt; /* # of actually issued ckpts */
292 atomic_t total_ckpt; /* # of total ckpts */
293 atomic_t queued_ckpt; /* # of queued ckpts */
294 struct llist_head issue_list; /* list for command issue */
295 spinlock_t stat_lock; /* lock for below checkpoint time stats */
296 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
297 unsigned int peak_time; /* peak wait time in msec until now */
298};
299
300/* for the bitmap indicate blocks to be discarded */
301struct discard_entry {
302 struct list_head list; /* list head */
303 block_t start_blkaddr; /* start blockaddr of current segment */
304 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
305};
306
307/* default discard granularity of inner discard thread, unit: block count */
308#define DEFAULT_DISCARD_GRANULARITY 16
309
310/* max discard pend list number */
311#define MAX_PLIST_NUM 512
312#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
313 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
314
315enum {
316 D_PREP, /* initial */
317 D_PARTIAL, /* partially submitted */
318 D_SUBMIT, /* all submitted */
319 D_DONE, /* finished */
320};
321
322struct discard_info {
323 block_t lstart; /* logical start address */
324 block_t len; /* length */
325 block_t start; /* actual start address in dev */
326};
327
328struct discard_cmd {
329 struct rb_node rb_node; /* rb node located in rb-tree */
330 union {
331 struct {
332 block_t lstart; /* logical start address */
333 block_t len; /* length */
334 block_t start; /* actual start address in dev */
335 };
336 struct discard_info di; /* discard info */
337
338 };
339 struct list_head list; /* command list */
340 struct completion wait; /* compleation */
341 struct block_device *bdev; /* bdev */
342 unsigned short ref; /* reference count */
343 unsigned char state; /* state */
344 unsigned char queued; /* queued discard */
345 int error; /* bio error */
346 spinlock_t lock; /* for state/bio_ref updating */
347 unsigned short bio_ref; /* bio reference count */
348};
349
350enum {
351 DPOLICY_BG,
352 DPOLICY_FORCE,
353 DPOLICY_FSTRIM,
354 DPOLICY_UMOUNT,
355 MAX_DPOLICY,
356};
357
358struct discard_policy {
359 int type; /* type of discard */
360 unsigned int min_interval; /* used for candidates exist */
361 unsigned int mid_interval; /* used for device busy */
362 unsigned int max_interval; /* used for candidates not exist */
363 unsigned int max_requests; /* # of discards issued per round */
364 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
365 bool io_aware; /* issue discard in idle time */
366 bool sync; /* submit discard with REQ_SYNC flag */
367 bool ordered; /* issue discard by lba order */
368 bool timeout; /* discard timeout for put_super */
369 unsigned int granularity; /* discard granularity */
370};
371
372struct discard_cmd_control {
373 struct task_struct *f2fs_issue_discard; /* discard thread */
374 struct list_head entry_list; /* 4KB discard entry list */
375 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
376 struct list_head wait_list; /* store on-flushing entries */
377 struct list_head fstrim_list; /* in-flight discard from fstrim */
378 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
379 unsigned int discard_wake; /* to wake up discard thread */
380 struct mutex cmd_lock;
381 unsigned int nr_discards; /* # of discards in the list */
382 unsigned int max_discards; /* max. discards to be issued */
383 unsigned int discard_granularity; /* discard granularity */
384 unsigned int undiscard_blks; /* # of undiscard blocks */
385 unsigned int next_pos; /* next discard position */
386 atomic_t issued_discard; /* # of issued discard */
387 atomic_t queued_discard; /* # of queued discard */
388 atomic_t discard_cmd_cnt; /* # of cached cmd count */
389 struct rb_root_cached root; /* root of discard rb-tree */
390 bool rbtree_check; /* config for consistence check */
391};
392
393/* for the list of fsync inodes, used only during recovery */
394struct fsync_inode_entry {
395 struct list_head list; /* list head */
396 struct inode *inode; /* vfs inode pointer */
397 block_t blkaddr; /* block address locating the last fsync */
398 block_t last_dentry; /* block address locating the last dentry */
399};
400
401#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
402#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
403
404#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
405#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
406#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
407#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
408
409#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
410#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
411
412static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
413{
414 int before = nats_in_cursum(journal);
415
416 journal->n_nats = cpu_to_le16(before + i);
417 return before;
418}
419
420static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
421{
422 int before = sits_in_cursum(journal);
423
424 journal->n_sits = cpu_to_le16(before + i);
425 return before;
426}
427
428static inline bool __has_cursum_space(struct f2fs_journal *journal,
429 int size, int type)
430{
431 if (type == NAT_JOURNAL)
432 return size <= MAX_NAT_JENTRIES(journal);
433 return size <= MAX_SIT_JENTRIES(journal);
434}
435
436/* for inline stuff */
437#define DEF_INLINE_RESERVED_SIZE 1
438static inline int get_extra_isize(struct inode *inode);
439static inline int get_inline_xattr_addrs(struct inode *inode);
440#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
441 (CUR_ADDRS_PER_INODE(inode) - \
442 get_inline_xattr_addrs(inode) - \
443 DEF_INLINE_RESERVED_SIZE))
444
445/* for inline dir */
446#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
447 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
448 BITS_PER_BYTE + 1))
449#define INLINE_DENTRY_BITMAP_SIZE(inode) \
450 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
451#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
452 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
453 NR_INLINE_DENTRY(inode) + \
454 INLINE_DENTRY_BITMAP_SIZE(inode)))
455
456/*
457 * For INODE and NODE manager
458 */
459/* for directory operations */
460
461struct f2fs_filename {
462 /*
463 * The filename the user specified. This is NULL for some
464 * filesystem-internal operations, e.g. converting an inline directory
465 * to a non-inline one, or roll-forward recovering an encrypted dentry.
466 */
467 const struct qstr *usr_fname;
468
469 /*
470 * The on-disk filename. For encrypted directories, this is encrypted.
471 * This may be NULL for lookups in an encrypted dir without the key.
472 */
473 struct fscrypt_str disk_name;
474
475 /* The dirhash of this filename */
476 f2fs_hash_t hash;
477
478#ifdef CONFIG_FS_ENCRYPTION
479 /*
480 * For lookups in encrypted directories: either the buffer backing
481 * disk_name, or a buffer that holds the decoded no-key name.
482 */
483 struct fscrypt_str crypto_buf;
484#endif
485#ifdef CONFIG_UNICODE
486 /*
487 * For casefolded directories: the casefolded name, but it's left NULL
488 * if the original name is not valid Unicode, if the directory is both
489 * casefolded and encrypted and its encryption key is unavailable, or if
490 * the filesystem is doing an internal operation where usr_fname is also
491 * NULL. In all these cases we fall back to treating the name as an
492 * opaque byte sequence.
493 */
494 struct fscrypt_str cf_name;
495#endif
496};
497
498struct f2fs_dentry_ptr {
499 struct inode *inode;
500 void *bitmap;
501 struct f2fs_dir_entry *dentry;
502 __u8 (*filename)[F2FS_SLOT_LEN];
503 int max;
504 int nr_bitmap;
505};
506
507static inline void make_dentry_ptr_block(struct inode *inode,
508 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
509{
510 d->inode = inode;
511 d->max = NR_DENTRY_IN_BLOCK;
512 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
513 d->bitmap = t->dentry_bitmap;
514 d->dentry = t->dentry;
515 d->filename = t->filename;
516}
517
518static inline void make_dentry_ptr_inline(struct inode *inode,
519 struct f2fs_dentry_ptr *d, void *t)
520{
521 int entry_cnt = NR_INLINE_DENTRY(inode);
522 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
523 int reserved_size = INLINE_RESERVED_SIZE(inode);
524
525 d->inode = inode;
526 d->max = entry_cnt;
527 d->nr_bitmap = bitmap_size;
528 d->bitmap = t;
529 d->dentry = t + bitmap_size + reserved_size;
530 d->filename = t + bitmap_size + reserved_size +
531 SIZE_OF_DIR_ENTRY * entry_cnt;
532}
533
534/*
535 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
536 * as its node offset to distinguish from index node blocks.
537 * But some bits are used to mark the node block.
538 */
539#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
540 >> OFFSET_BIT_SHIFT)
541enum {
542 ALLOC_NODE, /* allocate a new node page if needed */
543 LOOKUP_NODE, /* look up a node without readahead */
544 LOOKUP_NODE_RA, /*
545 * look up a node with readahead called
546 * by get_data_block.
547 */
548};
549
550#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */
551
552/* congestion wait timeout value, default: 20ms */
553#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
554
555/* maximum retry quota flush count */
556#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
557
558#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
559
560#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
561
562/* for in-memory extent cache entry */
563#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
564
565/* number of extent info in extent cache we try to shrink */
566#define EXTENT_CACHE_SHRINK_NUMBER 128
567
568struct rb_entry {
569 struct rb_node rb_node; /* rb node located in rb-tree */
570 union {
571 struct {
572 unsigned int ofs; /* start offset of the entry */
573 unsigned int len; /* length of the entry */
574 };
575 unsigned long long key; /* 64-bits key */
576 } __packed;
577};
578
579struct extent_info {
580 unsigned int fofs; /* start offset in a file */
581 unsigned int len; /* length of the extent */
582 u32 blk; /* start block address of the extent */
583};
584
585struct extent_node {
586 struct rb_node rb_node; /* rb node located in rb-tree */
587 struct extent_info ei; /* extent info */
588 struct list_head list; /* node in global extent list of sbi */
589 struct extent_tree *et; /* extent tree pointer */
590};
591
592struct extent_tree {
593 nid_t ino; /* inode number */
594 struct rb_root_cached root; /* root of extent info rb-tree */
595 struct extent_node *cached_en; /* recently accessed extent node */
596 struct extent_info largest; /* largested extent info */
597 struct list_head list; /* to be used by sbi->zombie_list */
598 rwlock_t lock; /* protect extent info rb-tree */
599 atomic_t node_cnt; /* # of extent node in rb-tree*/
600 bool largest_updated; /* largest extent updated */
601};
602
603/*
604 * This structure is taken from ext4_map_blocks.
605 *
606 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
607 */
608#define F2FS_MAP_NEW (1 << BH_New)
609#define F2FS_MAP_MAPPED (1 << BH_Mapped)
610#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
611#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
612 F2FS_MAP_UNWRITTEN)
613
614struct f2fs_map_blocks {
615 block_t m_pblk;
616 block_t m_lblk;
617 unsigned int m_len;
618 unsigned int m_flags;
619 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
620 pgoff_t *m_next_extent; /* point to next possible extent */
621 int m_seg_type;
622 bool m_may_create; /* indicate it is from write path */
623};
624
625/* for flag in get_data_block */
626enum {
627 F2FS_GET_BLOCK_DEFAULT,
628 F2FS_GET_BLOCK_FIEMAP,
629 F2FS_GET_BLOCK_BMAP,
630 F2FS_GET_BLOCK_DIO,
631 F2FS_GET_BLOCK_PRE_DIO,
632 F2FS_GET_BLOCK_PRE_AIO,
633 F2FS_GET_BLOCK_PRECACHE,
634};
635
636/*
637 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
638 */
639#define FADVISE_COLD_BIT 0x01
640#define FADVISE_LOST_PINO_BIT 0x02
641#define FADVISE_ENCRYPT_BIT 0x04
642#define FADVISE_ENC_NAME_BIT 0x08
643#define FADVISE_KEEP_SIZE_BIT 0x10
644#define FADVISE_HOT_BIT 0x20
645#define FADVISE_VERITY_BIT 0x40
646
647#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
648
649#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
650#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
651#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
652
653#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
654#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
655#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
656
657#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
658#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
659
660#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
661#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
662
663#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
664#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
665
666#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
667#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
668#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
669
670#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
671#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
672
673#define DEF_DIR_LEVEL 0
674
675enum {
676 GC_FAILURE_PIN,
677 GC_FAILURE_ATOMIC,
678 MAX_GC_FAILURE
679};
680
681/* used for f2fs_inode_info->flags */
682enum {
683 FI_NEW_INODE, /* indicate newly allocated inode */
684 FI_DIRTY_INODE, /* indicate inode is dirty or not */
685 FI_AUTO_RECOVER, /* indicate inode is recoverable */
686 FI_DIRTY_DIR, /* indicate directory has dirty pages */
687 FI_INC_LINK, /* need to increment i_nlink */
688 FI_ACL_MODE, /* indicate acl mode */
689 FI_NO_ALLOC, /* should not allocate any blocks */
690 FI_FREE_NID, /* free allocated nide */
691 FI_NO_EXTENT, /* not to use the extent cache */
692 FI_INLINE_XATTR, /* used for inline xattr */
693 FI_INLINE_DATA, /* used for inline data*/
694 FI_INLINE_DENTRY, /* used for inline dentry */
695 FI_APPEND_WRITE, /* inode has appended data */
696 FI_UPDATE_WRITE, /* inode has in-place-update data */
697 FI_NEED_IPU, /* used for ipu per file */
698 FI_ATOMIC_FILE, /* indicate atomic file */
699 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
700 FI_VOLATILE_FILE, /* indicate volatile file */
701 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
702 FI_DROP_CACHE, /* drop dirty page cache */
703 FI_DATA_EXIST, /* indicate data exists */
704 FI_INLINE_DOTS, /* indicate inline dot dentries */
705 FI_DO_DEFRAG, /* indicate defragment is running */
706 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
707 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
708 FI_HOT_DATA, /* indicate file is hot */
709 FI_EXTRA_ATTR, /* indicate file has extra attribute */
710 FI_PROJ_INHERIT, /* indicate file inherits projectid */
711 FI_PIN_FILE, /* indicate file should not be gced */
712 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
713 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
714 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
715 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
716 FI_MMAP_FILE, /* indicate file was mmapped */
717 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
718 FI_COMPRESS_RELEASED, /* compressed blocks were released */
719 FI_ALIGNED_WRITE, /* enable aligned write */
720 FI_MAX, /* max flag, never be used */
721};
722
723struct f2fs_inode_info {
724 struct inode vfs_inode; /* serve a vfs inode */
725 unsigned long i_flags; /* keep an inode flags for ioctl */
726 unsigned char i_advise; /* use to give file attribute hints */
727 unsigned char i_dir_level; /* use for dentry level for large dir */
728 unsigned int i_current_depth; /* only for directory depth */
729 /* for gc failure statistic */
730 unsigned int i_gc_failures[MAX_GC_FAILURE];
731 unsigned int i_pino; /* parent inode number */
732 umode_t i_acl_mode; /* keep file acl mode temporarily */
733
734 /* Use below internally in f2fs*/
735 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
736 struct rw_semaphore i_sem; /* protect fi info */
737 atomic_t dirty_pages; /* # of dirty pages */
738 f2fs_hash_t chash; /* hash value of given file name */
739 unsigned int clevel; /* maximum level of given file name */
740 struct task_struct *task; /* lookup and create consistency */
741 struct task_struct *cp_task; /* separate cp/wb IO stats*/
742 nid_t i_xattr_nid; /* node id that contains xattrs */
743 loff_t last_disk_size; /* lastly written file size */
744 spinlock_t i_size_lock; /* protect last_disk_size */
745
746#ifdef CONFIG_QUOTA
747 struct dquot *i_dquot[MAXQUOTAS];
748
749 /* quota space reservation, managed internally by quota code */
750 qsize_t i_reserved_quota;
751#endif
752 struct list_head dirty_list; /* dirty list for dirs and files */
753 struct list_head gdirty_list; /* linked in global dirty list */
754 struct list_head inmem_ilist; /* list for inmem inodes */
755 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
756 struct task_struct *inmem_task; /* store inmemory task */
757 struct mutex inmem_lock; /* lock for inmemory pages */
758 struct extent_tree *extent_tree; /* cached extent_tree entry */
759
760 /* avoid racing between foreground op and gc */
761 struct rw_semaphore i_gc_rwsem[2];
762 struct rw_semaphore i_mmap_sem;
763 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
764
765 int i_extra_isize; /* size of extra space located in i_addr */
766 kprojid_t i_projid; /* id for project quota */
767 int i_inline_xattr_size; /* inline xattr size */
768 struct timespec64 i_crtime; /* inode creation time */
769 struct timespec64 i_disk_time[4];/* inode disk times */
770
771 /* for file compress */
772 atomic_t i_compr_blocks; /* # of compressed blocks */
773 unsigned char i_compress_algorithm; /* algorithm type */
774 unsigned char i_log_cluster_size; /* log of cluster size */
775 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
776 unsigned short i_compress_flag; /* compress flag */
777 unsigned int i_cluster_size; /* cluster size */
778};
779
780static inline void get_extent_info(struct extent_info *ext,
781 struct f2fs_extent *i_ext)
782{
783 ext->fofs = le32_to_cpu(i_ext->fofs);
784 ext->blk = le32_to_cpu(i_ext->blk);
785 ext->len = le32_to_cpu(i_ext->len);
786}
787
788static inline void set_raw_extent(struct extent_info *ext,
789 struct f2fs_extent *i_ext)
790{
791 i_ext->fofs = cpu_to_le32(ext->fofs);
792 i_ext->blk = cpu_to_le32(ext->blk);
793 i_ext->len = cpu_to_le32(ext->len);
794}
795
796static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
797 u32 blk, unsigned int len)
798{
799 ei->fofs = fofs;
800 ei->blk = blk;
801 ei->len = len;
802}
803
804static inline bool __is_discard_mergeable(struct discard_info *back,
805 struct discard_info *front, unsigned int max_len)
806{
807 return (back->lstart + back->len == front->lstart) &&
808 (back->len + front->len <= max_len);
809}
810
811static inline bool __is_discard_back_mergeable(struct discard_info *cur,
812 struct discard_info *back, unsigned int max_len)
813{
814 return __is_discard_mergeable(back, cur, max_len);
815}
816
817static inline bool __is_discard_front_mergeable(struct discard_info *cur,
818 struct discard_info *front, unsigned int max_len)
819{
820 return __is_discard_mergeable(cur, front, max_len);
821}
822
823static inline bool __is_extent_mergeable(struct extent_info *back,
824 struct extent_info *front)
825{
826 return (back->fofs + back->len == front->fofs &&
827 back->blk + back->len == front->blk);
828}
829
830static inline bool __is_back_mergeable(struct extent_info *cur,
831 struct extent_info *back)
832{
833 return __is_extent_mergeable(back, cur);
834}
835
836static inline bool __is_front_mergeable(struct extent_info *cur,
837 struct extent_info *front)
838{
839 return __is_extent_mergeable(cur, front);
840}
841
842extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
843static inline void __try_update_largest_extent(struct extent_tree *et,
844 struct extent_node *en)
845{
846 if (en->ei.len > et->largest.len) {
847 et->largest = en->ei;
848 et->largest_updated = true;
849 }
850}
851
852/*
853 * For free nid management
854 */
855enum nid_state {
856 FREE_NID, /* newly added to free nid list */
857 PREALLOC_NID, /* it is preallocated */
858 MAX_NID_STATE,
859};
860
861enum nat_state {
862 TOTAL_NAT,
863 DIRTY_NAT,
864 RECLAIMABLE_NAT,
865 MAX_NAT_STATE,
866};
867
868struct f2fs_nm_info {
869 block_t nat_blkaddr; /* base disk address of NAT */
870 nid_t max_nid; /* maximum possible node ids */
871 nid_t available_nids; /* # of available node ids */
872 nid_t next_scan_nid; /* the next nid to be scanned */
873 unsigned int ram_thresh; /* control the memory footprint */
874 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
875 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
876
877 /* NAT cache management */
878 struct radix_tree_root nat_root;/* root of the nat entry cache */
879 struct radix_tree_root nat_set_root;/* root of the nat set cache */
880 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */
881 struct list_head nat_entries; /* cached nat entry list (clean) */
882 spinlock_t nat_list_lock; /* protect clean nat entry list */
883 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
884 unsigned int nat_blocks; /* # of nat blocks */
885
886 /* free node ids management */
887 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
888 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
889 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
890 spinlock_t nid_list_lock; /* protect nid lists ops */
891 struct mutex build_lock; /* lock for build free nids */
892 unsigned char **free_nid_bitmap;
893 unsigned char *nat_block_bitmap;
894 unsigned short *free_nid_count; /* free nid count of NAT block */
895
896 /* for checkpoint */
897 char *nat_bitmap; /* NAT bitmap pointer */
898
899 unsigned int nat_bits_blocks; /* # of nat bits blocks */
900 unsigned char *nat_bits; /* NAT bits blocks */
901 unsigned char *full_nat_bits; /* full NAT pages */
902 unsigned char *empty_nat_bits; /* empty NAT pages */
903#ifdef CONFIG_F2FS_CHECK_FS
904 char *nat_bitmap_mir; /* NAT bitmap mirror */
905#endif
906 int bitmap_size; /* bitmap size */
907};
908
909/*
910 * this structure is used as one of function parameters.
911 * all the information are dedicated to a given direct node block determined
912 * by the data offset in a file.
913 */
914struct dnode_of_data {
915 struct inode *inode; /* vfs inode pointer */
916 struct page *inode_page; /* its inode page, NULL is possible */
917 struct page *node_page; /* cached direct node page */
918 nid_t nid; /* node id of the direct node block */
919 unsigned int ofs_in_node; /* data offset in the node page */
920 bool inode_page_locked; /* inode page is locked or not */
921 bool node_changed; /* is node block changed */
922 char cur_level; /* level of hole node page */
923 char max_level; /* level of current page located */
924 block_t data_blkaddr; /* block address of the node block */
925};
926
927static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
928 struct page *ipage, struct page *npage, nid_t nid)
929{
930 memset(dn, 0, sizeof(*dn));
931 dn->inode = inode;
932 dn->inode_page = ipage;
933 dn->node_page = npage;
934 dn->nid = nid;
935}
936
937/*
938 * For SIT manager
939 *
940 * By default, there are 6 active log areas across the whole main area.
941 * When considering hot and cold data separation to reduce cleaning overhead,
942 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
943 * respectively.
944 * In the current design, you should not change the numbers intentionally.
945 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
946 * logs individually according to the underlying devices. (default: 6)
947 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
948 * data and 8 for node logs.
949 */
950#define NR_CURSEG_DATA_TYPE (3)
951#define NR_CURSEG_NODE_TYPE (3)
952#define NR_CURSEG_INMEM_TYPE (2)
953#define NR_CURSEG_RO_TYPE (2)
954#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
955#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
956
957enum {
958 CURSEG_HOT_DATA = 0, /* directory entry blocks */
959 CURSEG_WARM_DATA, /* data blocks */
960 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
961 CURSEG_HOT_NODE, /* direct node blocks of directory files */
962 CURSEG_WARM_NODE, /* direct node blocks of normal files */
963 CURSEG_COLD_NODE, /* indirect node blocks */
964 NR_PERSISTENT_LOG, /* number of persistent log */
965 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
966 /* pinned file that needs consecutive block address */
967 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
968 NO_CHECK_TYPE, /* number of persistent & inmem log */
969};
970
971struct flush_cmd {
972 struct completion wait;
973 struct llist_node llnode;
974 nid_t ino;
975 int ret;
976};
977
978struct flush_cmd_control {
979 struct task_struct *f2fs_issue_flush; /* flush thread */
980 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
981 atomic_t issued_flush; /* # of issued flushes */
982 atomic_t queued_flush; /* # of queued flushes */
983 struct llist_head issue_list; /* list for command issue */
984 struct llist_node *dispatch_list; /* list for command dispatch */
985};
986
987struct f2fs_sm_info {
988 struct sit_info *sit_info; /* whole segment information */
989 struct free_segmap_info *free_info; /* free segment information */
990 struct dirty_seglist_info *dirty_info; /* dirty segment information */
991 struct curseg_info *curseg_array; /* active segment information */
992
993 struct rw_semaphore curseg_lock; /* for preventing curseg change */
994
995 block_t seg0_blkaddr; /* block address of 0'th segment */
996 block_t main_blkaddr; /* start block address of main area */
997 block_t ssa_blkaddr; /* start block address of SSA area */
998
999 unsigned int segment_count; /* total # of segments */
1000 unsigned int main_segments; /* # of segments in main area */
1001 unsigned int reserved_segments; /* # of reserved segments */
1002 unsigned int ovp_segments; /* # of overprovision segments */
1003
1004 /* a threshold to reclaim prefree segments */
1005 unsigned int rec_prefree_segments;
1006
1007 /* for batched trimming */
1008 unsigned int trim_sections; /* # of sections to trim */
1009
1010 struct list_head sit_entry_set; /* sit entry set list */
1011
1012 unsigned int ipu_policy; /* in-place-update policy */
1013 unsigned int min_ipu_util; /* in-place-update threshold */
1014 unsigned int min_fsync_blocks; /* threshold for fsync */
1015 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1016 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1017 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1018
1019 /* for flush command control */
1020 struct flush_cmd_control *fcc_info;
1021
1022 /* for discard command control */
1023 struct discard_cmd_control *dcc_info;
1024};
1025
1026/*
1027 * For superblock
1028 */
1029/*
1030 * COUNT_TYPE for monitoring
1031 *
1032 * f2fs monitors the number of several block types such as on-writeback,
1033 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1034 */
1035#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1036enum count_type {
1037 F2FS_DIRTY_DENTS,
1038 F2FS_DIRTY_DATA,
1039 F2FS_DIRTY_QDATA,
1040 F2FS_DIRTY_NODES,
1041 F2FS_DIRTY_META,
1042 F2FS_INMEM_PAGES,
1043 F2FS_DIRTY_IMETA,
1044 F2FS_WB_CP_DATA,
1045 F2FS_WB_DATA,
1046 F2FS_RD_DATA,
1047 F2FS_RD_NODE,
1048 F2FS_RD_META,
1049 F2FS_DIO_WRITE,
1050 F2FS_DIO_READ,
1051 NR_COUNT_TYPE,
1052};
1053
1054/*
1055 * The below are the page types of bios used in submit_bio().
1056 * The available types are:
1057 * DATA User data pages. It operates as async mode.
1058 * NODE Node pages. It operates as async mode.
1059 * META FS metadata pages such as SIT, NAT, CP.
1060 * NR_PAGE_TYPE The number of page types.
1061 * META_FLUSH Make sure the previous pages are written
1062 * with waiting the bio's completion
1063 * ... Only can be used with META.
1064 */
1065#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1066enum page_type {
1067 DATA,
1068 NODE,
1069 META,
1070 NR_PAGE_TYPE,
1071 META_FLUSH,
1072 INMEM, /* the below types are used by tracepoints only. */
1073 INMEM_DROP,
1074 INMEM_INVALIDATE,
1075 INMEM_REVOKE,
1076 IPU,
1077 OPU,
1078};
1079
1080enum temp_type {
1081 HOT = 0, /* must be zero for meta bio */
1082 WARM,
1083 COLD,
1084 NR_TEMP_TYPE,
1085};
1086
1087enum need_lock_type {
1088 LOCK_REQ = 0,
1089 LOCK_DONE,
1090 LOCK_RETRY,
1091};
1092
1093enum cp_reason_type {
1094 CP_NO_NEEDED,
1095 CP_NON_REGULAR,
1096 CP_COMPRESSED,
1097 CP_HARDLINK,
1098 CP_SB_NEED_CP,
1099 CP_WRONG_PINO,
1100 CP_NO_SPC_ROLL,
1101 CP_NODE_NEED_CP,
1102 CP_FASTBOOT_MODE,
1103 CP_SPEC_LOG_NUM,
1104 CP_RECOVER_DIR,
1105};
1106
1107enum iostat_type {
1108 /* WRITE IO */
1109 APP_DIRECT_IO, /* app direct write IOs */
1110 APP_BUFFERED_IO, /* app buffered write IOs */
1111 APP_WRITE_IO, /* app write IOs */
1112 APP_MAPPED_IO, /* app mapped IOs */
1113 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1114 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1115 FS_META_IO, /* meta IOs from kworker/reclaimer */
1116 FS_GC_DATA_IO, /* data IOs from forground gc */
1117 FS_GC_NODE_IO, /* node IOs from forground gc */
1118 FS_CP_DATA_IO, /* data IOs from checkpoint */
1119 FS_CP_NODE_IO, /* node IOs from checkpoint */
1120 FS_CP_META_IO, /* meta IOs from checkpoint */
1121
1122 /* READ IO */
1123 APP_DIRECT_READ_IO, /* app direct read IOs */
1124 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1125 APP_READ_IO, /* app read IOs */
1126 APP_MAPPED_READ_IO, /* app mapped read IOs */
1127 FS_DATA_READ_IO, /* data read IOs */
1128 FS_GDATA_READ_IO, /* data read IOs from background gc */
1129 FS_CDATA_READ_IO, /* compressed data read IOs */
1130 FS_NODE_READ_IO, /* node read IOs */
1131 FS_META_READ_IO, /* meta read IOs */
1132
1133 /* other */
1134 FS_DISCARD, /* discard */
1135 NR_IO_TYPE,
1136};
1137
1138struct f2fs_io_info {
1139 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1140 nid_t ino; /* inode number */
1141 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1142 enum temp_type temp; /* contains HOT/WARM/COLD */
1143 int op; /* contains REQ_OP_ */
1144 int op_flags; /* req_flag_bits */
1145 block_t new_blkaddr; /* new block address to be written */
1146 block_t old_blkaddr; /* old block address before Cow */
1147 struct page *page; /* page to be written */
1148 struct page *encrypted_page; /* encrypted page */
1149 struct page *compressed_page; /* compressed page */
1150 struct list_head list; /* serialize IOs */
1151 bool submitted; /* indicate IO submission */
1152 int need_lock; /* indicate we need to lock cp_rwsem */
1153 bool in_list; /* indicate fio is in io_list */
1154 bool is_por; /* indicate IO is from recovery or not */
1155 bool retry; /* need to reallocate block address */
1156 int compr_blocks; /* # of compressed block addresses */
1157 bool encrypted; /* indicate file is encrypted */
1158 enum iostat_type io_type; /* io type */
1159 struct writeback_control *io_wbc; /* writeback control */
1160 struct bio **bio; /* bio for ipu */
1161 sector_t *last_block; /* last block number in bio */
1162 unsigned char version; /* version of the node */
1163};
1164
1165struct bio_entry {
1166 struct bio *bio;
1167 struct list_head list;
1168};
1169
1170#define is_read_io(rw) ((rw) == READ)
1171struct f2fs_bio_info {
1172 struct f2fs_sb_info *sbi; /* f2fs superblock */
1173 struct bio *bio; /* bios to merge */
1174 sector_t last_block_in_bio; /* last block number */
1175 struct f2fs_io_info fio; /* store buffered io info. */
1176 struct rw_semaphore io_rwsem; /* blocking op for bio */
1177 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1178 struct list_head io_list; /* track fios */
1179 struct list_head bio_list; /* bio entry list head */
1180 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1181};
1182
1183#define FDEV(i) (sbi->devs[i])
1184#define RDEV(i) (raw_super->devs[i])
1185struct f2fs_dev_info {
1186 struct block_device *bdev;
1187 char path[MAX_PATH_LEN];
1188 unsigned int total_segments;
1189 block_t start_blk;
1190 block_t end_blk;
1191#ifdef CONFIG_BLK_DEV_ZONED
1192 unsigned int nr_blkz; /* Total number of zones */
1193 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1194 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */
1195#endif
1196};
1197
1198enum inode_type {
1199 DIR_INODE, /* for dirty dir inode */
1200 FILE_INODE, /* for dirty regular/symlink inode */
1201 DIRTY_META, /* for all dirtied inode metadata */
1202 ATOMIC_FILE, /* for all atomic files */
1203 NR_INODE_TYPE,
1204};
1205
1206/* for inner inode cache management */
1207struct inode_management {
1208 struct radix_tree_root ino_root; /* ino entry array */
1209 spinlock_t ino_lock; /* for ino entry lock */
1210 struct list_head ino_list; /* inode list head */
1211 unsigned long ino_num; /* number of entries */
1212};
1213
1214/* for GC_AT */
1215struct atgc_management {
1216 bool atgc_enabled; /* ATGC is enabled or not */
1217 struct rb_root_cached root; /* root of victim rb-tree */
1218 struct list_head victim_list; /* linked with all victim entries */
1219 unsigned int victim_count; /* victim count in rb-tree */
1220 unsigned int candidate_ratio; /* candidate ratio */
1221 unsigned int max_candidate_count; /* max candidate count */
1222 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1223 unsigned long long age_threshold; /* age threshold */
1224};
1225
1226/* For s_flag in struct f2fs_sb_info */
1227enum {
1228 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1229 SBI_IS_CLOSE, /* specify unmounting */
1230 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1231 SBI_POR_DOING, /* recovery is doing or not */
1232 SBI_NEED_SB_WRITE, /* need to recover superblock */
1233 SBI_NEED_CP, /* need to checkpoint */
1234 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1235 SBI_IS_RECOVERED, /* recovered orphan/data */
1236 SBI_CP_DISABLED, /* CP was disabled last mount */
1237 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1238 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1239 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1240 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1241 SBI_IS_RESIZEFS, /* resizefs is in process */
1242};
1243
1244enum {
1245 CP_TIME,
1246 REQ_TIME,
1247 DISCARD_TIME,
1248 GC_TIME,
1249 DISABLE_TIME,
1250 UMOUNT_DISCARD_TIMEOUT,
1251 MAX_TIME,
1252};
1253
1254enum {
1255 GC_NORMAL,
1256 GC_IDLE_CB,
1257 GC_IDLE_GREEDY,
1258 GC_IDLE_AT,
1259 GC_URGENT_HIGH,
1260 GC_URGENT_LOW,
1261 MAX_GC_MODE,
1262};
1263
1264enum {
1265 BGGC_MODE_ON, /* background gc is on */
1266 BGGC_MODE_OFF, /* background gc is off */
1267 BGGC_MODE_SYNC, /*
1268 * background gc is on, migrating blocks
1269 * like foreground gc
1270 */
1271};
1272
1273enum {
1274 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1275 FS_MODE_LFS, /* use lfs allocation only */
1276};
1277
1278enum {
1279 WHINT_MODE_OFF, /* not pass down write hints */
1280 WHINT_MODE_USER, /* try to pass down hints given by users */
1281 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1282};
1283
1284enum {
1285 ALLOC_MODE_DEFAULT, /* stay default */
1286 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1287};
1288
1289enum fsync_mode {
1290 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1291 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1292 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1293};
1294
1295enum {
1296 COMPR_MODE_FS, /*
1297 * automatically compress compression
1298 * enabled files
1299 */
1300 COMPR_MODE_USER, /*
1301 * automatical compression is disabled.
1302 * user can control the file compression
1303 * using ioctls
1304 */
1305};
1306
1307enum {
1308 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1309 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1310 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1311};
1312
1313static inline int f2fs_test_bit(unsigned int nr, char *addr);
1314static inline void f2fs_set_bit(unsigned int nr, char *addr);
1315static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1316
1317/*
1318 * Layout of f2fs page.private:
1319 *
1320 * Layout A: lowest bit should be 1
1321 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1322 * bit 0 PAGE_PRIVATE_NOT_POINTER
1323 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE
1324 * bit 2 PAGE_PRIVATE_DUMMY_WRITE
1325 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION
1326 * bit 4 PAGE_PRIVATE_INLINE_INODE
1327 * bit 5 PAGE_PRIVATE_REF_RESOURCE
1328 * bit 6- f2fs private data
1329 *
1330 * Layout B: lowest bit should be 0
1331 * page.private is a wrapped pointer.
1332 */
1333enum {
1334 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1335 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */
1336 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1337 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1338 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1339 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1340 PAGE_PRIVATE_MAX
1341};
1342
1343#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
1344static inline bool page_private_##name(struct page *page) \
1345{ \
1346 return PagePrivate(page) && \
1347 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
1348 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1349}
1350
1351#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
1352static inline void set_page_private_##name(struct page *page) \
1353{ \
1354 if (!PagePrivate(page)) { \
1355 get_page(page); \
1356 SetPagePrivate(page); \
1357 set_page_private(page, 0); \
1358 } \
1359 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
1360 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1361}
1362
1363#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
1364static inline void clear_page_private_##name(struct page *page) \
1365{ \
1366 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1367 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \
1368 set_page_private(page, 0); \
1369 if (PagePrivate(page)) { \
1370 ClearPagePrivate(page); \
1371 put_page(page); \
1372 }\
1373 } \
1374}
1375
1376PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
1377PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE);
1378PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
1379PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
1380PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);
1381PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
1382
1383PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
1384PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
1385PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
1386PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);
1387PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
1388
1389PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
1390PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
1391PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
1392PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);
1393PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
1394
1395static inline unsigned long get_page_private_data(struct page *page)
1396{
1397 unsigned long data = page_private(page);
1398
1399 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
1400 return 0;
1401 return data >> PAGE_PRIVATE_MAX;
1402}
1403
1404static inline void set_page_private_data(struct page *page, unsigned long data)
1405{
1406 if (!PagePrivate(page)) {
1407 get_page(page);
1408 SetPagePrivate(page);
1409 set_page_private(page, 0);
1410 }
1411 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
1412 page_private(page) |= data << PAGE_PRIVATE_MAX;
1413}
1414
1415static inline void clear_page_private_data(struct page *page)
1416{
1417 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1;
1418 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) {
1419 set_page_private(page, 0);
1420 if (PagePrivate(page)) {
1421 ClearPagePrivate(page);
1422 put_page(page);
1423 }
1424 }
1425}
1426
1427/* For compression */
1428enum compress_algorithm_type {
1429 COMPRESS_LZO,
1430 COMPRESS_LZ4,
1431 COMPRESS_ZSTD,
1432 COMPRESS_LZORLE,
1433 COMPRESS_MAX,
1434};
1435
1436enum compress_flag {
1437 COMPRESS_CHKSUM,
1438 COMPRESS_MAX_FLAG,
1439};
1440
1441#define COMPRESS_WATERMARK 20
1442#define COMPRESS_PERCENT 20
1443
1444#define COMPRESS_DATA_RESERVED_SIZE 4
1445struct compress_data {
1446 __le32 clen; /* compressed data size */
1447 __le32 chksum; /* compressed data chksum */
1448 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1449 u8 cdata[]; /* compressed data */
1450};
1451
1452#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1453
1454#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1455
1456#define COMPRESS_LEVEL_OFFSET 8
1457
1458/* compress context */
1459struct compress_ctx {
1460 struct inode *inode; /* inode the context belong to */
1461 pgoff_t cluster_idx; /* cluster index number */
1462 unsigned int cluster_size; /* page count in cluster */
1463 unsigned int log_cluster_size; /* log of cluster size */
1464 struct page **rpages; /* pages store raw data in cluster */
1465 unsigned int nr_rpages; /* total page number in rpages */
1466 struct page **cpages; /* pages store compressed data in cluster */
1467 unsigned int nr_cpages; /* total page number in cpages */
1468 void *rbuf; /* virtual mapped address on rpages */
1469 struct compress_data *cbuf; /* virtual mapped address on cpages */
1470 size_t rlen; /* valid data length in rbuf */
1471 size_t clen; /* valid data length in cbuf */
1472 void *private; /* payload buffer for specified compression algorithm */
1473 void *private2; /* extra payload buffer */
1474};
1475
1476/* compress context for write IO path */
1477struct compress_io_ctx {
1478 u32 magic; /* magic number to indicate page is compressed */
1479 struct inode *inode; /* inode the context belong to */
1480 struct page **rpages; /* pages store raw data in cluster */
1481 unsigned int nr_rpages; /* total page number in rpages */
1482 atomic_t pending_pages; /* in-flight compressed page count */
1483};
1484
1485/* Context for decompressing one cluster on the read IO path */
1486struct decompress_io_ctx {
1487 u32 magic; /* magic number to indicate page is compressed */
1488 struct inode *inode; /* inode the context belong to */
1489 pgoff_t cluster_idx; /* cluster index number */
1490 unsigned int cluster_size; /* page count in cluster */
1491 unsigned int log_cluster_size; /* log of cluster size */
1492 struct page **rpages; /* pages store raw data in cluster */
1493 unsigned int nr_rpages; /* total page number in rpages */
1494 struct page **cpages; /* pages store compressed data in cluster */
1495 unsigned int nr_cpages; /* total page number in cpages */
1496 struct page **tpages; /* temp pages to pad holes in cluster */
1497 void *rbuf; /* virtual mapped address on rpages */
1498 struct compress_data *cbuf; /* virtual mapped address on cpages */
1499 size_t rlen; /* valid data length in rbuf */
1500 size_t clen; /* valid data length in cbuf */
1501
1502 /*
1503 * The number of compressed pages remaining to be read in this cluster.
1504 * This is initially nr_cpages. It is decremented by 1 each time a page
1505 * has been read (or failed to be read). When it reaches 0, the cluster
1506 * is decompressed (or an error is reported).
1507 *
1508 * If an error occurs before all the pages have been submitted for I/O,
1509 * then this will never reach 0. In this case the I/O submitter is
1510 * responsible for calling f2fs_decompress_end_io() instead.
1511 */
1512 atomic_t remaining_pages;
1513
1514 /*
1515 * Number of references to this decompress_io_ctx.
1516 *
1517 * One reference is held for I/O completion. This reference is dropped
1518 * after the pagecache pages are updated and unlocked -- either after
1519 * decompression (and verity if enabled), or after an error.
1520 *
1521 * In addition, each compressed page holds a reference while it is in a
1522 * bio. These references are necessary prevent compressed pages from
1523 * being freed while they are still in a bio.
1524 */
1525 refcount_t refcnt;
1526
1527 bool failed; /* IO error occurred before decompression? */
1528 bool need_verity; /* need fs-verity verification after decompression? */
1529 void *private; /* payload buffer for specified decompression algorithm */
1530 void *private2; /* extra payload buffer */
1531 struct work_struct verity_work; /* work to verify the decompressed pages */
1532};
1533
1534#define NULL_CLUSTER ((unsigned int)(~0))
1535#define MIN_COMPRESS_LOG_SIZE 2
1536#define MAX_COMPRESS_LOG_SIZE 8
1537#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1538
1539struct f2fs_sb_info {
1540 struct super_block *sb; /* pointer to VFS super block */
1541 struct proc_dir_entry *s_proc; /* proc entry */
1542 struct f2fs_super_block *raw_super; /* raw super block pointer */
1543 struct rw_semaphore sb_lock; /* lock for raw super block */
1544 int valid_super_block; /* valid super block no */
1545 unsigned long s_flag; /* flags for sbi */
1546 struct mutex writepages; /* mutex for writepages() */
1547
1548#ifdef CONFIG_BLK_DEV_ZONED
1549 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1550 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1551#endif
1552
1553 /* for node-related operations */
1554 struct f2fs_nm_info *nm_info; /* node manager */
1555 struct inode *node_inode; /* cache node blocks */
1556
1557 /* for segment-related operations */
1558 struct f2fs_sm_info *sm_info; /* segment manager */
1559
1560 /* for bio operations */
1561 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1562 /* keep migration IO order for LFS mode */
1563 struct rw_semaphore io_order_lock;
1564 mempool_t *write_io_dummy; /* Dummy pages */
1565
1566 /* for checkpoint */
1567 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1568 int cur_cp_pack; /* remain current cp pack */
1569 spinlock_t cp_lock; /* for flag in ckpt */
1570 struct inode *meta_inode; /* cache meta blocks */
1571 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */
1572 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1573 struct rw_semaphore node_write; /* locking node writes */
1574 struct rw_semaphore node_change; /* locking node change */
1575 wait_queue_head_t cp_wait;
1576 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1577 long interval_time[MAX_TIME]; /* to store thresholds */
1578 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1579
1580 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1581
1582 spinlock_t fsync_node_lock; /* for node entry lock */
1583 struct list_head fsync_node_list; /* node list head */
1584 unsigned int fsync_seg_id; /* sequence id */
1585 unsigned int fsync_node_num; /* number of node entries */
1586
1587 /* for orphan inode, use 0'th array */
1588 unsigned int max_orphans; /* max orphan inodes */
1589
1590 /* for inode management */
1591 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1592 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1593 struct mutex flush_lock; /* for flush exclusion */
1594
1595 /* for extent tree cache */
1596 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1597 struct mutex extent_tree_lock; /* locking extent radix tree */
1598 struct list_head extent_list; /* lru list for shrinker */
1599 spinlock_t extent_lock; /* locking extent lru list */
1600 atomic_t total_ext_tree; /* extent tree count */
1601 struct list_head zombie_list; /* extent zombie tree list */
1602 atomic_t total_zombie_tree; /* extent zombie tree count */
1603 atomic_t total_ext_node; /* extent info count */
1604
1605 /* basic filesystem units */
1606 unsigned int log_sectors_per_block; /* log2 sectors per block */
1607 unsigned int log_blocksize; /* log2 block size */
1608 unsigned int blocksize; /* block size */
1609 unsigned int root_ino_num; /* root inode number*/
1610 unsigned int node_ino_num; /* node inode number*/
1611 unsigned int meta_ino_num; /* meta inode number*/
1612 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1613 unsigned int blocks_per_seg; /* blocks per segment */
1614 unsigned int segs_per_sec; /* segments per section */
1615 unsigned int secs_per_zone; /* sections per zone */
1616 unsigned int total_sections; /* total section count */
1617 unsigned int total_node_count; /* total node block count */
1618 unsigned int total_valid_node_count; /* valid node block count */
1619 int dir_level; /* directory level */
1620 int readdir_ra; /* readahead inode in readdir */
1621 u64 max_io_bytes; /* max io bytes to merge IOs */
1622
1623 block_t user_block_count; /* # of user blocks */
1624 block_t total_valid_block_count; /* # of valid blocks */
1625 block_t discard_blks; /* discard command candidats */
1626 block_t last_valid_block_count; /* for recovery */
1627 block_t reserved_blocks; /* configurable reserved blocks */
1628 block_t current_reserved_blocks; /* current reserved blocks */
1629
1630 /* Additional tracking for no checkpoint mode */
1631 block_t unusable_block_count; /* # of blocks saved by last cp */
1632
1633 unsigned int nquota_files; /* # of quota sysfile */
1634 struct rw_semaphore quota_sem; /* blocking cp for flags */
1635
1636 /* # of pages, see count_type */
1637 atomic_t nr_pages[NR_COUNT_TYPE];
1638 /* # of allocated blocks */
1639 struct percpu_counter alloc_valid_block_count;
1640
1641 /* writeback control */
1642 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1643
1644 /* valid inode count */
1645 struct percpu_counter total_valid_inode_count;
1646
1647 struct f2fs_mount_info mount_opt; /* mount options */
1648
1649 /* for cleaning operations */
1650 struct rw_semaphore gc_lock; /*
1651 * semaphore for GC, avoid
1652 * race between GC and GC or CP
1653 */
1654 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1655 struct atgc_management am; /* atgc management */
1656 unsigned int cur_victim_sec; /* current victim section num */
1657 unsigned int gc_mode; /* current GC state */
1658 unsigned int next_victim_seg[2]; /* next segment in victim section */
1659
1660 /* for skip statistic */
1661 unsigned int atomic_files; /* # of opened atomic file */
1662 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1663 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1664
1665 /* threshold for gc trials on pinned files */
1666 u64 gc_pin_file_threshold;
1667 struct rw_semaphore pin_sem;
1668
1669 /* maximum # of trials to find a victim segment for SSR and GC */
1670 unsigned int max_victim_search;
1671 /* migration granularity of garbage collection, unit: segment */
1672 unsigned int migration_granularity;
1673
1674 /*
1675 * for stat information.
1676 * one is for the LFS mode, and the other is for the SSR mode.
1677 */
1678#ifdef CONFIG_F2FS_STAT_FS
1679 struct f2fs_stat_info *stat_info; /* FS status information */
1680 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1681 unsigned int segment_count[2]; /* # of allocated segments */
1682 unsigned int block_count[2]; /* # of allocated blocks */
1683 atomic_t inplace_count; /* # of inplace update */
1684 atomic64_t total_hit_ext; /* # of lookup extent cache */
1685 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1686 atomic64_t read_hit_largest; /* # of hit largest extent node */
1687 atomic64_t read_hit_cached; /* # of hit cached extent node */
1688 atomic_t inline_xattr; /* # of inline_xattr inodes */
1689 atomic_t inline_inode; /* # of inline_data inodes */
1690 atomic_t inline_dir; /* # of inline_dentry inodes */
1691 atomic_t compr_inode; /* # of compressed inodes */
1692 atomic64_t compr_blocks; /* # of compressed blocks */
1693 atomic_t vw_cnt; /* # of volatile writes */
1694 atomic_t max_aw_cnt; /* max # of atomic writes */
1695 atomic_t max_vw_cnt; /* max # of volatile writes */
1696 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1697 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1698 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1699#endif
1700 spinlock_t stat_lock; /* lock for stat operations */
1701
1702 /* For app/fs IO statistics */
1703 spinlock_t iostat_lock;
1704 unsigned long long rw_iostat[NR_IO_TYPE];
1705 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1706 bool iostat_enable;
1707 unsigned long iostat_next_period;
1708 unsigned int iostat_period_ms;
1709
1710 /* to attach REQ_META|REQ_FUA flags */
1711 unsigned int data_io_flag;
1712 unsigned int node_io_flag;
1713
1714 /* For sysfs suppport */
1715 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1716 struct completion s_kobj_unregister;
1717
1718 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1719 struct completion s_stat_kobj_unregister;
1720
1721 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1722 struct completion s_feature_list_kobj_unregister;
1723
1724 /* For shrinker support */
1725 struct list_head s_list;
1726 int s_ndevs; /* number of devices */
1727 struct f2fs_dev_info *devs; /* for device list */
1728 unsigned int dirty_device; /* for checkpoint data flush */
1729 spinlock_t dev_lock; /* protect dirty_device */
1730 struct mutex umount_mutex;
1731 unsigned int shrinker_run_no;
1732
1733 /* For write statistics */
1734 u64 sectors_written_start;
1735 u64 kbytes_written;
1736
1737 /* Reference to checksum algorithm driver via cryptoapi */
1738 struct crypto_shash *s_chksum_driver;
1739
1740 /* Precomputed FS UUID checksum for seeding other checksums */
1741 __u32 s_chksum_seed;
1742
1743 struct workqueue_struct *post_read_wq; /* post read workqueue */
1744
1745 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1746 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1747
1748 /* For reclaimed segs statistics per each GC mode */
1749 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1750 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1751
1752 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1753
1754#ifdef CONFIG_F2FS_FS_COMPRESSION
1755 struct kmem_cache *page_array_slab; /* page array entry */
1756 unsigned int page_array_slab_size; /* default page array slab size */
1757
1758 /* For runtime compression statistics */
1759 u64 compr_written_block;
1760 u64 compr_saved_block;
1761 u32 compr_new_inode;
1762
1763 /* For compressed block cache */
1764 struct inode *compress_inode; /* cache compressed blocks */
1765 unsigned int compress_percent; /* cache page percentage */
1766 unsigned int compress_watermark; /* cache page watermark */
1767 atomic_t compress_page_hit; /* cache hit count */
1768#endif
1769};
1770
1771struct f2fs_private_dio {
1772 struct inode *inode;
1773 void *orig_private;
1774 bio_end_io_t *orig_end_io;
1775 bool write;
1776};
1777
1778#ifdef CONFIG_F2FS_FAULT_INJECTION
1779#define f2fs_show_injection_info(sbi, type) \
1780 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1781 KERN_INFO, sbi->sb->s_id, \
1782 f2fs_fault_name[type], \
1783 __func__, __builtin_return_address(0))
1784static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1785{
1786 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1787
1788 if (!ffi->inject_rate)
1789 return false;
1790
1791 if (!IS_FAULT_SET(ffi, type))
1792 return false;
1793
1794 atomic_inc(&ffi->inject_ops);
1795 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1796 atomic_set(&ffi->inject_ops, 0);
1797 return true;
1798 }
1799 return false;
1800}
1801#else
1802#define f2fs_show_injection_info(sbi, type) do { } while (0)
1803static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1804{
1805 return false;
1806}
1807#endif
1808
1809/*
1810 * Test if the mounted volume is a multi-device volume.
1811 * - For a single regular disk volume, sbi->s_ndevs is 0.
1812 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1813 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1814 */
1815static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1816{
1817 return sbi->s_ndevs > 1;
1818}
1819
1820static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1821{
1822 unsigned long now = jiffies;
1823
1824 sbi->last_time[type] = now;
1825
1826 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1827 if (type == REQ_TIME) {
1828 sbi->last_time[DISCARD_TIME] = now;
1829 sbi->last_time[GC_TIME] = now;
1830 }
1831}
1832
1833static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1834{
1835 unsigned long interval = sbi->interval_time[type] * HZ;
1836
1837 return time_after(jiffies, sbi->last_time[type] + interval);
1838}
1839
1840static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1841 int type)
1842{
1843 unsigned long interval = sbi->interval_time[type] * HZ;
1844 unsigned int wait_ms = 0;
1845 long delta;
1846
1847 delta = (sbi->last_time[type] + interval) - jiffies;
1848 if (delta > 0)
1849 wait_ms = jiffies_to_msecs(delta);
1850
1851 return wait_ms;
1852}
1853
1854/*
1855 * Inline functions
1856 */
1857static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1858 const void *address, unsigned int length)
1859{
1860 struct {
1861 struct shash_desc shash;
1862 char ctx[4];
1863 } desc;
1864 int err;
1865
1866 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1867
1868 desc.shash.tfm = sbi->s_chksum_driver;
1869 *(u32 *)desc.ctx = crc;
1870
1871 err = crypto_shash_update(&desc.shash, address, length);
1872 BUG_ON(err);
1873
1874 return *(u32 *)desc.ctx;
1875}
1876
1877static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1878 unsigned int length)
1879{
1880 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1881}
1882
1883static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1884 void *buf, size_t buf_size)
1885{
1886 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1887}
1888
1889static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1890 const void *address, unsigned int length)
1891{
1892 return __f2fs_crc32(sbi, crc, address, length);
1893}
1894
1895static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1896{
1897 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1898}
1899
1900static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1901{
1902 return sb->s_fs_info;
1903}
1904
1905static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1906{
1907 return F2FS_SB(inode->i_sb);
1908}
1909
1910static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1911{
1912 return F2FS_I_SB(mapping->host);
1913}
1914
1915static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1916{
1917 return F2FS_M_SB(page_file_mapping(page));
1918}
1919
1920static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1921{
1922 return (struct f2fs_super_block *)(sbi->raw_super);
1923}
1924
1925static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1926{
1927 return (struct f2fs_checkpoint *)(sbi->ckpt);
1928}
1929
1930static inline struct f2fs_node *F2FS_NODE(struct page *page)
1931{
1932 return (struct f2fs_node *)page_address(page);
1933}
1934
1935static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1936{
1937 return &((struct f2fs_node *)page_address(page))->i;
1938}
1939
1940static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1941{
1942 return (struct f2fs_nm_info *)(sbi->nm_info);
1943}
1944
1945static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1946{
1947 return (struct f2fs_sm_info *)(sbi->sm_info);
1948}
1949
1950static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1951{
1952 return (struct sit_info *)(SM_I(sbi)->sit_info);
1953}
1954
1955static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1956{
1957 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1958}
1959
1960static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1961{
1962 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1963}
1964
1965static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1966{
1967 return sbi->meta_inode->i_mapping;
1968}
1969
1970static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1971{
1972 return sbi->node_inode->i_mapping;
1973}
1974
1975static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1976{
1977 return test_bit(type, &sbi->s_flag);
1978}
1979
1980static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1981{
1982 set_bit(type, &sbi->s_flag);
1983}
1984
1985static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
1986{
1987 clear_bit(type, &sbi->s_flag);
1988}
1989
1990static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
1991{
1992 return le64_to_cpu(cp->checkpoint_ver);
1993}
1994
1995static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
1996{
1997 if (type < F2FS_MAX_QUOTAS)
1998 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
1999 return 0;
2000}
2001
2002static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2003{
2004 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2005 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2006}
2007
2008static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2009{
2010 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2011
2012 return ckpt_flags & f;
2013}
2014
2015static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2016{
2017 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2018}
2019
2020static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2021{
2022 unsigned int ckpt_flags;
2023
2024 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2025 ckpt_flags |= f;
2026 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2027}
2028
2029static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2030{
2031 unsigned long flags;
2032
2033 spin_lock_irqsave(&sbi->cp_lock, flags);
2034 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2035 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2036}
2037
2038static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2039{
2040 unsigned int ckpt_flags;
2041
2042 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2043 ckpt_flags &= (~f);
2044 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2045}
2046
2047static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2048{
2049 unsigned long flags;
2050
2051 spin_lock_irqsave(&sbi->cp_lock, flags);
2052 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2053 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2054}
2055
2056static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock)
2057{
2058 unsigned long flags;
2059 unsigned char *nat_bits;
2060
2061 /*
2062 * In order to re-enable nat_bits we need to call fsck.f2fs by
2063 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost,
2064 * so let's rely on regular fsck or unclean shutdown.
2065 */
2066
2067 if (lock)
2068 spin_lock_irqsave(&sbi->cp_lock, flags);
2069 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG);
2070 nat_bits = NM_I(sbi)->nat_bits;
2071 NM_I(sbi)->nat_bits = NULL;
2072 if (lock)
2073 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2074
2075 kvfree(nat_bits);
2076}
2077
2078static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi,
2079 struct cp_control *cpc)
2080{
2081 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG);
2082
2083 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set;
2084}
2085
2086static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2087{
2088 down_read(&sbi->cp_rwsem);
2089}
2090
2091static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2092{
2093 return down_read_trylock(&sbi->cp_rwsem);
2094}
2095
2096static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2097{
2098 up_read(&sbi->cp_rwsem);
2099}
2100
2101static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2102{
2103 down_write(&sbi->cp_rwsem);
2104}
2105
2106static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2107{
2108 up_write(&sbi->cp_rwsem);
2109}
2110
2111static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2112{
2113 int reason = CP_SYNC;
2114
2115 if (test_opt(sbi, FASTBOOT))
2116 reason = CP_FASTBOOT;
2117 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2118 reason = CP_UMOUNT;
2119 return reason;
2120}
2121
2122static inline bool __remain_node_summaries(int reason)
2123{
2124 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2125}
2126
2127static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2128{
2129 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2130 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2131}
2132
2133/*
2134 * Check whether the inode has blocks or not
2135 */
2136static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2137{
2138 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2139
2140 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2141}
2142
2143static inline bool f2fs_has_xattr_block(unsigned int ofs)
2144{
2145 return ofs == XATTR_NODE_OFFSET;
2146}
2147
2148static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2149 struct inode *inode, bool cap)
2150{
2151 if (!inode)
2152 return true;
2153 if (!test_opt(sbi, RESERVE_ROOT))
2154 return false;
2155 if (IS_NOQUOTA(inode))
2156 return true;
2157 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2158 return true;
2159 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2160 in_group_p(F2FS_OPTION(sbi).s_resgid))
2161 return true;
2162 if (cap && capable(CAP_SYS_RESOURCE))
2163 return true;
2164 return false;
2165}
2166
2167static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2168static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2169 struct inode *inode, blkcnt_t *count)
2170{
2171 blkcnt_t diff = 0, release = 0;
2172 block_t avail_user_block_count;
2173 int ret;
2174
2175 ret = dquot_reserve_block(inode, *count);
2176 if (ret)
2177 return ret;
2178
2179 if (time_to_inject(sbi, FAULT_BLOCK)) {
2180 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2181 release = *count;
2182 goto release_quota;
2183 }
2184
2185 /*
2186 * let's increase this in prior to actual block count change in order
2187 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2188 */
2189 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2190
2191 spin_lock(&sbi->stat_lock);
2192 sbi->total_valid_block_count += (block_t)(*count);
2193 avail_user_block_count = sbi->user_block_count -
2194 sbi->current_reserved_blocks;
2195
2196 if (!__allow_reserved_blocks(sbi, inode, true))
2197 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2198 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2199 if (avail_user_block_count > sbi->unusable_block_count)
2200 avail_user_block_count -= sbi->unusable_block_count;
2201 else
2202 avail_user_block_count = 0;
2203 }
2204 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2205 diff = sbi->total_valid_block_count - avail_user_block_count;
2206 if (diff > *count)
2207 diff = *count;
2208 *count -= diff;
2209 release = diff;
2210 sbi->total_valid_block_count -= diff;
2211 if (!*count) {
2212 spin_unlock(&sbi->stat_lock);
2213 goto enospc;
2214 }
2215 }
2216 spin_unlock(&sbi->stat_lock);
2217
2218 if (unlikely(release)) {
2219 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2220 dquot_release_reservation_block(inode, release);
2221 }
2222 f2fs_i_blocks_write(inode, *count, true, true);
2223 return 0;
2224
2225enospc:
2226 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2227release_quota:
2228 dquot_release_reservation_block(inode, release);
2229 return -ENOSPC;
2230}
2231
2232__printf(2, 3)
2233void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2234
2235#define f2fs_err(sbi, fmt, ...) \
2236 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2237#define f2fs_warn(sbi, fmt, ...) \
2238 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2239#define f2fs_notice(sbi, fmt, ...) \
2240 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2241#define f2fs_info(sbi, fmt, ...) \
2242 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2243#define f2fs_debug(sbi, fmt, ...) \
2244 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2245
2246static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2247 struct inode *inode,
2248 block_t count)
2249{
2250 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2251
2252 spin_lock(&sbi->stat_lock);
2253 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2254 sbi->total_valid_block_count -= (block_t)count;
2255 if (sbi->reserved_blocks &&
2256 sbi->current_reserved_blocks < sbi->reserved_blocks)
2257 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2258 sbi->current_reserved_blocks + count);
2259 spin_unlock(&sbi->stat_lock);
2260 if (unlikely(inode->i_blocks < sectors)) {
2261 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2262 inode->i_ino,
2263 (unsigned long long)inode->i_blocks,
2264 (unsigned long long)sectors);
2265 set_sbi_flag(sbi, SBI_NEED_FSCK);
2266 return;
2267 }
2268 f2fs_i_blocks_write(inode, count, false, true);
2269}
2270
2271static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2272{
2273 atomic_inc(&sbi->nr_pages[count_type]);
2274
2275 if (count_type == F2FS_DIRTY_DENTS ||
2276 count_type == F2FS_DIRTY_NODES ||
2277 count_type == F2FS_DIRTY_META ||
2278 count_type == F2FS_DIRTY_QDATA ||
2279 count_type == F2FS_DIRTY_IMETA)
2280 set_sbi_flag(sbi, SBI_IS_DIRTY);
2281}
2282
2283static inline void inode_inc_dirty_pages(struct inode *inode)
2284{
2285 atomic_inc(&F2FS_I(inode)->dirty_pages);
2286 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2287 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2288 if (IS_NOQUOTA(inode))
2289 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2290}
2291
2292static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2293{
2294 atomic_dec(&sbi->nr_pages[count_type]);
2295}
2296
2297static inline void inode_dec_dirty_pages(struct inode *inode)
2298{
2299 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2300 !S_ISLNK(inode->i_mode))
2301 return;
2302
2303 atomic_dec(&F2FS_I(inode)->dirty_pages);
2304 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2305 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2306 if (IS_NOQUOTA(inode))
2307 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2308}
2309
2310static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2311{
2312 return atomic_read(&sbi->nr_pages[count_type]);
2313}
2314
2315static inline int get_dirty_pages(struct inode *inode)
2316{
2317 return atomic_read(&F2FS_I(inode)->dirty_pages);
2318}
2319
2320static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2321{
2322 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2323 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2324 sbi->log_blocks_per_seg;
2325
2326 return segs / sbi->segs_per_sec;
2327}
2328
2329static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2330{
2331 return sbi->total_valid_block_count;
2332}
2333
2334static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2335{
2336 return sbi->discard_blks;
2337}
2338
2339static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2340{
2341 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2342
2343 /* return NAT or SIT bitmap */
2344 if (flag == NAT_BITMAP)
2345 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2346 else if (flag == SIT_BITMAP)
2347 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2348
2349 return 0;
2350}
2351
2352static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2353{
2354 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2355}
2356
2357static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2358{
2359 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2360 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2361 int offset;
2362
2363 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2364 offset = (flag == SIT_BITMAP) ?
2365 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2366 /*
2367 * if large_nat_bitmap feature is enabled, leave checksum
2368 * protection for all nat/sit bitmaps.
2369 */
2370 return tmp_ptr + offset + sizeof(__le32);
2371 }
2372
2373 if (__cp_payload(sbi) > 0) {
2374 if (flag == NAT_BITMAP)
2375 return &ckpt->sit_nat_version_bitmap;
2376 else
2377 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2378 } else {
2379 offset = (flag == NAT_BITMAP) ?
2380 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2381 return tmp_ptr + offset;
2382 }
2383}
2384
2385static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2386{
2387 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2388
2389 if (sbi->cur_cp_pack == 2)
2390 start_addr += sbi->blocks_per_seg;
2391 return start_addr;
2392}
2393
2394static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2395{
2396 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2397
2398 if (sbi->cur_cp_pack == 1)
2399 start_addr += sbi->blocks_per_seg;
2400 return start_addr;
2401}
2402
2403static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2404{
2405 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2406}
2407
2408static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2409{
2410 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2411}
2412
2413static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2414 struct inode *inode, bool is_inode)
2415{
2416 block_t valid_block_count;
2417 unsigned int valid_node_count, user_block_count;
2418 int err;
2419
2420 if (is_inode) {
2421 if (inode) {
2422 err = dquot_alloc_inode(inode);
2423 if (err)
2424 return err;
2425 }
2426 } else {
2427 err = dquot_reserve_block(inode, 1);
2428 if (err)
2429 return err;
2430 }
2431
2432 if (time_to_inject(sbi, FAULT_BLOCK)) {
2433 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2434 goto enospc;
2435 }
2436
2437 spin_lock(&sbi->stat_lock);
2438
2439 valid_block_count = sbi->total_valid_block_count +
2440 sbi->current_reserved_blocks + 1;
2441
2442 if (!__allow_reserved_blocks(sbi, inode, false))
2443 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2444 user_block_count = sbi->user_block_count;
2445 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2446 user_block_count -= sbi->unusable_block_count;
2447
2448 if (unlikely(valid_block_count > user_block_count)) {
2449 spin_unlock(&sbi->stat_lock);
2450 goto enospc;
2451 }
2452
2453 valid_node_count = sbi->total_valid_node_count + 1;
2454 if (unlikely(valid_node_count > sbi->total_node_count)) {
2455 spin_unlock(&sbi->stat_lock);
2456 goto enospc;
2457 }
2458
2459 sbi->total_valid_node_count++;
2460 sbi->total_valid_block_count++;
2461 spin_unlock(&sbi->stat_lock);
2462
2463 if (inode) {
2464 if (is_inode)
2465 f2fs_mark_inode_dirty_sync(inode, true);
2466 else
2467 f2fs_i_blocks_write(inode, 1, true, true);
2468 }
2469
2470 percpu_counter_inc(&sbi->alloc_valid_block_count);
2471 return 0;
2472
2473enospc:
2474 if (is_inode) {
2475 if (inode)
2476 dquot_free_inode(inode);
2477 } else {
2478 dquot_release_reservation_block(inode, 1);
2479 }
2480 return -ENOSPC;
2481}
2482
2483static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2484 struct inode *inode, bool is_inode)
2485{
2486 spin_lock(&sbi->stat_lock);
2487
2488 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2489 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2490
2491 sbi->total_valid_node_count--;
2492 sbi->total_valid_block_count--;
2493 if (sbi->reserved_blocks &&
2494 sbi->current_reserved_blocks < sbi->reserved_blocks)
2495 sbi->current_reserved_blocks++;
2496
2497 spin_unlock(&sbi->stat_lock);
2498
2499 if (is_inode) {
2500 dquot_free_inode(inode);
2501 } else {
2502 if (unlikely(inode->i_blocks == 0)) {
2503 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2504 inode->i_ino,
2505 (unsigned long long)inode->i_blocks);
2506 set_sbi_flag(sbi, SBI_NEED_FSCK);
2507 return;
2508 }
2509 f2fs_i_blocks_write(inode, 1, false, true);
2510 }
2511}
2512
2513static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2514{
2515 return sbi->total_valid_node_count;
2516}
2517
2518static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2519{
2520 percpu_counter_inc(&sbi->total_valid_inode_count);
2521}
2522
2523static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2524{
2525 percpu_counter_dec(&sbi->total_valid_inode_count);
2526}
2527
2528static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2529{
2530 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2531}
2532
2533static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2534 pgoff_t index, bool for_write)
2535{
2536 struct page *page;
2537
2538 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2539 if (!for_write)
2540 page = find_get_page_flags(mapping, index,
2541 FGP_LOCK | FGP_ACCESSED);
2542 else
2543 page = find_lock_page(mapping, index);
2544 if (page)
2545 return page;
2546
2547 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2548 f2fs_show_injection_info(F2FS_M_SB(mapping),
2549 FAULT_PAGE_ALLOC);
2550 return NULL;
2551 }
2552 }
2553
2554 if (!for_write)
2555 return grab_cache_page(mapping, index);
2556 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2557}
2558
2559static inline struct page *f2fs_pagecache_get_page(
2560 struct address_space *mapping, pgoff_t index,
2561 int fgp_flags, gfp_t gfp_mask)
2562{
2563 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2564 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2565 return NULL;
2566 }
2567
2568 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2569}
2570
2571static inline void f2fs_copy_page(struct page *src, struct page *dst)
2572{
2573 char *src_kaddr = kmap(src);
2574 char *dst_kaddr = kmap(dst);
2575
2576 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2577 kunmap(dst);
2578 kunmap(src);
2579}
2580
2581static inline void f2fs_put_page(struct page *page, int unlock)
2582{
2583 if (!page)
2584 return;
2585
2586 if (unlock) {
2587 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2588 unlock_page(page);
2589 }
2590 put_page(page);
2591}
2592
2593static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2594{
2595 if (dn->node_page)
2596 f2fs_put_page(dn->node_page, 1);
2597 if (dn->inode_page && dn->node_page != dn->inode_page)
2598 f2fs_put_page(dn->inode_page, 0);
2599 dn->node_page = NULL;
2600 dn->inode_page = NULL;
2601}
2602
2603static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2604 size_t size)
2605{
2606 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2607}
2608
2609static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2610 gfp_t flags)
2611{
2612 void *entry;
2613
2614 entry = kmem_cache_alloc(cachep, flags);
2615 if (!entry)
2616 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2617 return entry;
2618}
2619
2620static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2621{
2622 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2623 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2624 get_pages(sbi, F2FS_WB_CP_DATA) ||
2625 get_pages(sbi, F2FS_DIO_READ) ||
2626 get_pages(sbi, F2FS_DIO_WRITE))
2627 return true;
2628
2629 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2630 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2631 return true;
2632
2633 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2634 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2635 return true;
2636 return false;
2637}
2638
2639static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2640{
2641 if (sbi->gc_mode == GC_URGENT_HIGH)
2642 return true;
2643
2644 if (is_inflight_io(sbi, type))
2645 return false;
2646
2647 if (sbi->gc_mode == GC_URGENT_LOW &&
2648 (type == DISCARD_TIME || type == GC_TIME))
2649 return true;
2650
2651 return f2fs_time_over(sbi, type);
2652}
2653
2654static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2655 unsigned long index, void *item)
2656{
2657 while (radix_tree_insert(root, index, item))
2658 cond_resched();
2659}
2660
2661#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2662
2663static inline bool IS_INODE(struct page *page)
2664{
2665 struct f2fs_node *p = F2FS_NODE(page);
2666
2667 return RAW_IS_INODE(p);
2668}
2669
2670static inline int offset_in_addr(struct f2fs_inode *i)
2671{
2672 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2673 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2674}
2675
2676static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2677{
2678 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2679}
2680
2681static inline int f2fs_has_extra_attr(struct inode *inode);
2682static inline block_t data_blkaddr(struct inode *inode,
2683 struct page *node_page, unsigned int offset)
2684{
2685 struct f2fs_node *raw_node;
2686 __le32 *addr_array;
2687 int base = 0;
2688 bool is_inode = IS_INODE(node_page);
2689
2690 raw_node = F2FS_NODE(node_page);
2691
2692 if (is_inode) {
2693 if (!inode)
2694 /* from GC path only */
2695 base = offset_in_addr(&raw_node->i);
2696 else if (f2fs_has_extra_attr(inode))
2697 base = get_extra_isize(inode);
2698 }
2699
2700 addr_array = blkaddr_in_node(raw_node);
2701 return le32_to_cpu(addr_array[base + offset]);
2702}
2703
2704static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2705{
2706 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2707}
2708
2709static inline int f2fs_test_bit(unsigned int nr, char *addr)
2710{
2711 int mask;
2712
2713 addr += (nr >> 3);
2714 mask = 1 << (7 - (nr & 0x07));
2715 return mask & *addr;
2716}
2717
2718static inline void f2fs_set_bit(unsigned int nr, char *addr)
2719{
2720 int mask;
2721
2722 addr += (nr >> 3);
2723 mask = 1 << (7 - (nr & 0x07));
2724 *addr |= mask;
2725}
2726
2727static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2728{
2729 int mask;
2730
2731 addr += (nr >> 3);
2732 mask = 1 << (7 - (nr & 0x07));
2733 *addr &= ~mask;
2734}
2735
2736static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2737{
2738 int mask;
2739 int ret;
2740
2741 addr += (nr >> 3);
2742 mask = 1 << (7 - (nr & 0x07));
2743 ret = mask & *addr;
2744 *addr |= mask;
2745 return ret;
2746}
2747
2748static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2749{
2750 int mask;
2751 int ret;
2752
2753 addr += (nr >> 3);
2754 mask = 1 << (7 - (nr & 0x07));
2755 ret = mask & *addr;
2756 *addr &= ~mask;
2757 return ret;
2758}
2759
2760static inline void f2fs_change_bit(unsigned int nr, char *addr)
2761{
2762 int mask;
2763
2764 addr += (nr >> 3);
2765 mask = 1 << (7 - (nr & 0x07));
2766 *addr ^= mask;
2767}
2768
2769/*
2770 * On-disk inode flags (f2fs_inode::i_flags)
2771 */
2772#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2773#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2774#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2775#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2776#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2777#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2778#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2779#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2780#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2781#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2782#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2783
2784/* Flags that should be inherited by new inodes from their parent. */
2785#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2786 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2787 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2788
2789/* Flags that are appropriate for regular files (all but dir-specific ones). */
2790#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2791 F2FS_CASEFOLD_FL))
2792
2793/* Flags that are appropriate for non-directories/regular files. */
2794#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2795
2796static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2797{
2798 if (S_ISDIR(mode))
2799 return flags;
2800 else if (S_ISREG(mode))
2801 return flags & F2FS_REG_FLMASK;
2802 else
2803 return flags & F2FS_OTHER_FLMASK;
2804}
2805
2806static inline void __mark_inode_dirty_flag(struct inode *inode,
2807 int flag, bool set)
2808{
2809 switch (flag) {
2810 case FI_INLINE_XATTR:
2811 case FI_INLINE_DATA:
2812 case FI_INLINE_DENTRY:
2813 case FI_NEW_INODE:
2814 if (set)
2815 return;
2816 fallthrough;
2817 case FI_DATA_EXIST:
2818 case FI_INLINE_DOTS:
2819 case FI_PIN_FILE:
2820 case FI_COMPRESS_RELEASED:
2821 f2fs_mark_inode_dirty_sync(inode, true);
2822 }
2823}
2824
2825static inline void set_inode_flag(struct inode *inode, int flag)
2826{
2827 set_bit(flag, F2FS_I(inode)->flags);
2828 __mark_inode_dirty_flag(inode, flag, true);
2829}
2830
2831static inline int is_inode_flag_set(struct inode *inode, int flag)
2832{
2833 return test_bit(flag, F2FS_I(inode)->flags);
2834}
2835
2836static inline void clear_inode_flag(struct inode *inode, int flag)
2837{
2838 clear_bit(flag, F2FS_I(inode)->flags);
2839 __mark_inode_dirty_flag(inode, flag, false);
2840}
2841
2842static inline bool f2fs_verity_in_progress(struct inode *inode)
2843{
2844 return IS_ENABLED(CONFIG_FS_VERITY) &&
2845 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2846}
2847
2848static inline void set_acl_inode(struct inode *inode, umode_t mode)
2849{
2850 F2FS_I(inode)->i_acl_mode = mode;
2851 set_inode_flag(inode, FI_ACL_MODE);
2852 f2fs_mark_inode_dirty_sync(inode, false);
2853}
2854
2855static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2856{
2857 if (inc)
2858 inc_nlink(inode);
2859 else
2860 drop_nlink(inode);
2861 f2fs_mark_inode_dirty_sync(inode, true);
2862}
2863
2864static inline void f2fs_i_blocks_write(struct inode *inode,
2865 block_t diff, bool add, bool claim)
2866{
2867 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2868 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2869
2870 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2871 if (add) {
2872 if (claim)
2873 dquot_claim_block(inode, diff);
2874 else
2875 dquot_alloc_block_nofail(inode, diff);
2876 } else {
2877 dquot_free_block(inode, diff);
2878 }
2879
2880 f2fs_mark_inode_dirty_sync(inode, true);
2881 if (clean || recover)
2882 set_inode_flag(inode, FI_AUTO_RECOVER);
2883}
2884
2885static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2886{
2887 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2888 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2889
2890 if (i_size_read(inode) == i_size)
2891 return;
2892
2893 i_size_write(inode, i_size);
2894 f2fs_mark_inode_dirty_sync(inode, true);
2895 if (clean || recover)
2896 set_inode_flag(inode, FI_AUTO_RECOVER);
2897}
2898
2899static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2900{
2901 F2FS_I(inode)->i_current_depth = depth;
2902 f2fs_mark_inode_dirty_sync(inode, true);
2903}
2904
2905static inline void f2fs_i_gc_failures_write(struct inode *inode,
2906 unsigned int count)
2907{
2908 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2909 f2fs_mark_inode_dirty_sync(inode, true);
2910}
2911
2912static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2913{
2914 F2FS_I(inode)->i_xattr_nid = xnid;
2915 f2fs_mark_inode_dirty_sync(inode, true);
2916}
2917
2918static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2919{
2920 F2FS_I(inode)->i_pino = pino;
2921 f2fs_mark_inode_dirty_sync(inode, true);
2922}
2923
2924static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2925{
2926 struct f2fs_inode_info *fi = F2FS_I(inode);
2927
2928 if (ri->i_inline & F2FS_INLINE_XATTR)
2929 set_bit(FI_INLINE_XATTR, fi->flags);
2930 if (ri->i_inline & F2FS_INLINE_DATA)
2931 set_bit(FI_INLINE_DATA, fi->flags);
2932 if (ri->i_inline & F2FS_INLINE_DENTRY)
2933 set_bit(FI_INLINE_DENTRY, fi->flags);
2934 if (ri->i_inline & F2FS_DATA_EXIST)
2935 set_bit(FI_DATA_EXIST, fi->flags);
2936 if (ri->i_inline & F2FS_INLINE_DOTS)
2937 set_bit(FI_INLINE_DOTS, fi->flags);
2938 if (ri->i_inline & F2FS_EXTRA_ATTR)
2939 set_bit(FI_EXTRA_ATTR, fi->flags);
2940 if (ri->i_inline & F2FS_PIN_FILE)
2941 set_bit(FI_PIN_FILE, fi->flags);
2942 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
2943 set_bit(FI_COMPRESS_RELEASED, fi->flags);
2944}
2945
2946static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2947{
2948 ri->i_inline = 0;
2949
2950 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2951 ri->i_inline |= F2FS_INLINE_XATTR;
2952 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2953 ri->i_inline |= F2FS_INLINE_DATA;
2954 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2955 ri->i_inline |= F2FS_INLINE_DENTRY;
2956 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2957 ri->i_inline |= F2FS_DATA_EXIST;
2958 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2959 ri->i_inline |= F2FS_INLINE_DOTS;
2960 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2961 ri->i_inline |= F2FS_EXTRA_ATTR;
2962 if (is_inode_flag_set(inode, FI_PIN_FILE))
2963 ri->i_inline |= F2FS_PIN_FILE;
2964 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
2965 ri->i_inline |= F2FS_COMPRESS_RELEASED;
2966}
2967
2968static inline int f2fs_has_extra_attr(struct inode *inode)
2969{
2970 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2971}
2972
2973static inline int f2fs_has_inline_xattr(struct inode *inode)
2974{
2975 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2976}
2977
2978static inline int f2fs_compressed_file(struct inode *inode)
2979{
2980 return S_ISREG(inode->i_mode) &&
2981 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2982}
2983
2984static inline bool f2fs_need_compress_data(struct inode *inode)
2985{
2986 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
2987
2988 if (!f2fs_compressed_file(inode))
2989 return false;
2990
2991 if (compress_mode == COMPR_MODE_FS)
2992 return true;
2993 else if (compress_mode == COMPR_MODE_USER &&
2994 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
2995 return true;
2996
2997 return false;
2998}
2999
3000static inline unsigned int addrs_per_inode(struct inode *inode)
3001{
3002 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3003 get_inline_xattr_addrs(inode);
3004
3005 if (!f2fs_compressed_file(inode))
3006 return addrs;
3007 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3008}
3009
3010static inline unsigned int addrs_per_block(struct inode *inode)
3011{
3012 if (!f2fs_compressed_file(inode))
3013 return DEF_ADDRS_PER_BLOCK;
3014 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3015}
3016
3017static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3018{
3019 struct f2fs_inode *ri = F2FS_INODE(page);
3020
3021 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3022 get_inline_xattr_addrs(inode)]);
3023}
3024
3025static inline int inline_xattr_size(struct inode *inode)
3026{
3027 if (f2fs_has_inline_xattr(inode))
3028 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3029 return 0;
3030}
3031
3032static inline int f2fs_has_inline_data(struct inode *inode)
3033{
3034 return is_inode_flag_set(inode, FI_INLINE_DATA);
3035}
3036
3037static inline int f2fs_exist_data(struct inode *inode)
3038{
3039 return is_inode_flag_set(inode, FI_DATA_EXIST);
3040}
3041
3042static inline int f2fs_has_inline_dots(struct inode *inode)
3043{
3044 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3045}
3046
3047static inline int f2fs_is_mmap_file(struct inode *inode)
3048{
3049 return is_inode_flag_set(inode, FI_MMAP_FILE);
3050}
3051
3052static inline bool f2fs_is_pinned_file(struct inode *inode)
3053{
3054 return is_inode_flag_set(inode, FI_PIN_FILE);
3055}
3056
3057static inline bool f2fs_is_atomic_file(struct inode *inode)
3058{
3059 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3060}
3061
3062static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
3063{
3064 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
3065}
3066
3067static inline bool f2fs_is_volatile_file(struct inode *inode)
3068{
3069 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
3070}
3071
3072static inline bool f2fs_is_first_block_written(struct inode *inode)
3073{
3074 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
3075}
3076
3077static inline bool f2fs_is_drop_cache(struct inode *inode)
3078{
3079 return is_inode_flag_set(inode, FI_DROP_CACHE);
3080}
3081
3082static inline void *inline_data_addr(struct inode *inode, struct page *page)
3083{
3084 struct f2fs_inode *ri = F2FS_INODE(page);
3085 int extra_size = get_extra_isize(inode);
3086
3087 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
3088}
3089
3090static inline int f2fs_has_inline_dentry(struct inode *inode)
3091{
3092 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3093}
3094
3095static inline int is_file(struct inode *inode, int type)
3096{
3097 return F2FS_I(inode)->i_advise & type;
3098}
3099
3100static inline void set_file(struct inode *inode, int type)
3101{
3102 F2FS_I(inode)->i_advise |= type;
3103 f2fs_mark_inode_dirty_sync(inode, true);
3104}
3105
3106static inline void clear_file(struct inode *inode, int type)
3107{
3108 F2FS_I(inode)->i_advise &= ~type;
3109 f2fs_mark_inode_dirty_sync(inode, true);
3110}
3111
3112static inline bool f2fs_is_time_consistent(struct inode *inode)
3113{
3114 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
3115 return false;
3116 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
3117 return false;
3118 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
3119 return false;
3120 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
3121 &F2FS_I(inode)->i_crtime))
3122 return false;
3123 return true;
3124}
3125
3126static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3127{
3128 bool ret;
3129
3130 if (dsync) {
3131 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3132
3133 spin_lock(&sbi->inode_lock[DIRTY_META]);
3134 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3135 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3136 return ret;
3137 }
3138 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3139 file_keep_isize(inode) ||
3140 i_size_read(inode) & ~PAGE_MASK)
3141 return false;
3142
3143 if (!f2fs_is_time_consistent(inode))
3144 return false;
3145
3146 spin_lock(&F2FS_I(inode)->i_size_lock);
3147 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3148 spin_unlock(&F2FS_I(inode)->i_size_lock);
3149
3150 return ret;
3151}
3152
3153static inline bool f2fs_readonly(struct super_block *sb)
3154{
3155 return sb_rdonly(sb);
3156}
3157
3158static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3159{
3160 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3161}
3162
3163static inline bool is_dot_dotdot(const u8 *name, size_t len)
3164{
3165 if (len == 1 && name[0] == '.')
3166 return true;
3167
3168 if (len == 2 && name[0] == '.' && name[1] == '.')
3169 return true;
3170
3171 return false;
3172}
3173
3174static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3175 size_t size, gfp_t flags)
3176{
3177 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3178 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3179 return NULL;
3180 }
3181
3182 return kmalloc(size, flags);
3183}
3184
3185static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3186 size_t size, gfp_t flags)
3187{
3188 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3189}
3190
3191static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3192 size_t size, gfp_t flags)
3193{
3194 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3195 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3196 return NULL;
3197 }
3198
3199 return kvmalloc(size, flags);
3200}
3201
3202static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3203 size_t size, gfp_t flags)
3204{
3205 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3206}
3207
3208static inline int get_extra_isize(struct inode *inode)
3209{
3210 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3211}
3212
3213static inline int get_inline_xattr_addrs(struct inode *inode)
3214{
3215 return F2FS_I(inode)->i_inline_xattr_size;
3216}
3217
3218#define f2fs_get_inode_mode(i) \
3219 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3220 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3221
3222#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3223 (offsetof(struct f2fs_inode, i_extra_end) - \
3224 offsetof(struct f2fs_inode, i_extra_isize)) \
3225
3226#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3227#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3228 ((offsetof(typeof(*(f2fs_inode)), field) + \
3229 sizeof((f2fs_inode)->field)) \
3230 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3231
3232#define DEFAULT_IOSTAT_PERIOD_MS 3000
3233#define MIN_IOSTAT_PERIOD_MS 100
3234/* maximum period of iostat tracing is 1 day */
3235#define MAX_IOSTAT_PERIOD_MS 8640000
3236
3237static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
3238{
3239 int i;
3240
3241 spin_lock(&sbi->iostat_lock);
3242 for (i = 0; i < NR_IO_TYPE; i++) {
3243 sbi->rw_iostat[i] = 0;
3244 sbi->prev_rw_iostat[i] = 0;
3245 }
3246 spin_unlock(&sbi->iostat_lock);
3247}
3248
3249extern void f2fs_record_iostat(struct f2fs_sb_info *sbi);
3250
3251static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi,
3252 enum iostat_type type, unsigned long long io_bytes)
3253{
3254 if (!sbi->iostat_enable)
3255 return;
3256 spin_lock(&sbi->iostat_lock);
3257 sbi->rw_iostat[type] += io_bytes;
3258
3259 if (type == APP_WRITE_IO || type == APP_DIRECT_IO)
3260 sbi->rw_iostat[APP_BUFFERED_IO] =
3261 sbi->rw_iostat[APP_WRITE_IO] -
3262 sbi->rw_iostat[APP_DIRECT_IO];
3263
3264 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO)
3265 sbi->rw_iostat[APP_BUFFERED_READ_IO] =
3266 sbi->rw_iostat[APP_READ_IO] -
3267 sbi->rw_iostat[APP_DIRECT_READ_IO];
3268 spin_unlock(&sbi->iostat_lock);
3269
3270 f2fs_record_iostat(sbi);
3271}
3272
3273#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3274
3275#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3276
3277bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3278 block_t blkaddr, int type);
3279static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3280 block_t blkaddr, int type)
3281{
3282 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3283 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3284 blkaddr, type);
3285 f2fs_bug_on(sbi, 1);
3286 }
3287}
3288
3289static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3290{
3291 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3292 blkaddr == COMPRESS_ADDR)
3293 return false;
3294 return true;
3295}
3296
3297/*
3298 * file.c
3299 */
3300int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3301void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3302int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3303int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3304int f2fs_truncate(struct inode *inode);
3305int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path,
3306 struct kstat *stat, u32 request_mask, unsigned int flags);
3307int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
3308 struct iattr *attr);
3309int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3310void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3311int f2fs_precache_extents(struct inode *inode);
3312int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3313int f2fs_fileattr_set(struct user_namespace *mnt_userns,
3314 struct dentry *dentry, struct fileattr *fa);
3315long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3316long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3317int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3318int f2fs_pin_file_control(struct inode *inode, bool inc);
3319
3320/*
3321 * inode.c
3322 */
3323void f2fs_set_inode_flags(struct inode *inode);
3324bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3325void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3326struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3327struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3328int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3329void f2fs_update_inode(struct inode *inode, struct page *node_page);
3330void f2fs_update_inode_page(struct inode *inode);
3331int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3332void f2fs_evict_inode(struct inode *inode);
3333void f2fs_handle_failed_inode(struct inode *inode);
3334
3335/*
3336 * namei.c
3337 */
3338int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3339 bool hot, bool set);
3340struct dentry *f2fs_get_parent(struct dentry *child);
3341
3342/*
3343 * dir.c
3344 */
3345unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3346int f2fs_init_casefolded_name(const struct inode *dir,
3347 struct f2fs_filename *fname);
3348int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3349 int lookup, struct f2fs_filename *fname);
3350int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3351 struct f2fs_filename *fname);
3352void f2fs_free_filename(struct f2fs_filename *fname);
3353struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3354 const struct f2fs_filename *fname, int *max_slots);
3355int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3356 unsigned int start_pos, struct fscrypt_str *fstr);
3357void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3358 struct f2fs_dentry_ptr *d);
3359struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3360 const struct f2fs_filename *fname, struct page *dpage);
3361void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3362 unsigned int current_depth);
3363int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3364void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3365struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3366 const struct f2fs_filename *fname,
3367 struct page **res_page);
3368struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3369 const struct qstr *child, struct page **res_page);
3370struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3371ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3372 struct page **page);
3373void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3374 struct page *page, struct inode *inode);
3375bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3376 const struct f2fs_filename *fname);
3377void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3378 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3379 unsigned int bit_pos);
3380int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3381 struct inode *inode, nid_t ino, umode_t mode);
3382int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3383 struct inode *inode, nid_t ino, umode_t mode);
3384int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3385 struct inode *inode, nid_t ino, umode_t mode);
3386void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3387 struct inode *dir, struct inode *inode);
3388int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3389bool f2fs_empty_dir(struct inode *dir);
3390
3391static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3392{
3393 if (fscrypt_is_nokey_name(dentry))
3394 return -ENOKEY;
3395 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3396 inode, inode->i_ino, inode->i_mode);
3397}
3398
3399/*
3400 * super.c
3401 */
3402int f2fs_inode_dirtied(struct inode *inode, bool sync);
3403void f2fs_inode_synced(struct inode *inode);
3404int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3405int f2fs_quota_sync(struct super_block *sb, int type);
3406loff_t max_file_blocks(struct inode *inode);
3407void f2fs_quota_off_umount(struct super_block *sb);
3408int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3409int f2fs_sync_fs(struct super_block *sb, int sync);
3410int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3411
3412/*
3413 * hash.c
3414 */
3415void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3416
3417/*
3418 * node.c
3419 */
3420struct node_info;
3421
3422int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3423bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3424bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3425void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3426void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3427void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3428int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3429bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3430bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3431int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3432 struct node_info *ni);
3433pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3434int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3435int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3436int f2fs_truncate_xattr_node(struct inode *inode);
3437int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3438 unsigned int seq_id);
3439int f2fs_remove_inode_page(struct inode *inode);
3440struct page *f2fs_new_inode_page(struct inode *inode);
3441struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3442void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3443struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3444struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3445int f2fs_move_node_page(struct page *node_page, int gc_type);
3446void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3447int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3448 struct writeback_control *wbc, bool atomic,
3449 unsigned int *seq_id);
3450int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3451 struct writeback_control *wbc,
3452 bool do_balance, enum iostat_type io_type);
3453int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3454bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3455void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3456void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3457int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3458int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3459int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3460int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3461int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3462 unsigned int segno, struct f2fs_summary_block *sum);
3463int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3464int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3465void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3466int __init f2fs_create_node_manager_caches(void);
3467void f2fs_destroy_node_manager_caches(void);
3468
3469/*
3470 * segment.c
3471 */
3472bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3473void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3474void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3475void f2fs_drop_inmem_pages(struct inode *inode);
3476void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3477int f2fs_commit_inmem_pages(struct inode *inode);
3478void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3479void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3480int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3481int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3482int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3483void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3484void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3485bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3486void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3487void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3488bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3489void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3490 struct cp_control *cpc);
3491void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3492block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3493int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3494void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3495int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3496bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3497void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3498void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3499void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3500void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3501 unsigned int *newseg, bool new_sec, int dir);
3502void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3503 unsigned int start, unsigned int end);
3504void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3505void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3506int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3507bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3508 struct cp_control *cpc);
3509struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3510void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3511 block_t blk_addr);
3512void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3513 enum iostat_type io_type);
3514void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3515void f2fs_outplace_write_data(struct dnode_of_data *dn,
3516 struct f2fs_io_info *fio);
3517int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3518void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3519 block_t old_blkaddr, block_t new_blkaddr,
3520 bool recover_curseg, bool recover_newaddr,
3521 bool from_gc);
3522void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3523 block_t old_addr, block_t new_addr,
3524 unsigned char version, bool recover_curseg,
3525 bool recover_newaddr);
3526void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3527 block_t old_blkaddr, block_t *new_blkaddr,
3528 struct f2fs_summary *sum, int type,
3529 struct f2fs_io_info *fio);
3530void f2fs_wait_on_page_writeback(struct page *page,
3531 enum page_type type, bool ordered, bool locked);
3532void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3533void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3534 block_t len);
3535void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3536void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3537int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3538 unsigned int val, int alloc);
3539void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3540int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3541int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3542int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3543void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3544int __init f2fs_create_segment_manager_caches(void);
3545void f2fs_destroy_segment_manager_caches(void);
3546int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3547enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3548 enum page_type type, enum temp_type temp);
3549unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3550 unsigned int segno);
3551unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3552 unsigned int segno);
3553
3554/*
3555 * checkpoint.c
3556 */
3557void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3558struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3559struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3560struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3561struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3562bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3563 block_t blkaddr, int type);
3564int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3565 int type, bool sync);
3566void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3567long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3568 long nr_to_write, enum iostat_type io_type);
3569void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3570void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3571void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3572bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3573void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3574 unsigned int devidx, int type);
3575bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3576 unsigned int devidx, int type);
3577int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3578int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3579void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3580void f2fs_add_orphan_inode(struct inode *inode);
3581void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3582int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3583int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3584void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3585void f2fs_remove_dirty_inode(struct inode *inode);
3586int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3587void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3588u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3589int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3590void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3591int __init f2fs_create_checkpoint_caches(void);
3592void f2fs_destroy_checkpoint_caches(void);
3593int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3594int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3595void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3596void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3597
3598/*
3599 * data.c
3600 */
3601int __init f2fs_init_bioset(void);
3602void f2fs_destroy_bioset(void);
3603int f2fs_init_bio_entry_cache(void);
3604void f2fs_destroy_bio_entry_cache(void);
3605void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3606 struct bio *bio, enum page_type type);
3607void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3608void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3609 struct inode *inode, struct page *page,
3610 nid_t ino, enum page_type type);
3611void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3612 struct bio **bio, struct page *page);
3613void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3614int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3615int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3616void f2fs_submit_page_write(struct f2fs_io_info *fio);
3617struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3618 block_t blk_addr, struct bio *bio);
3619int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3620void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3621void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3622int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3623int f2fs_reserve_new_block(struct dnode_of_data *dn);
3624int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3625int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3626int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3627struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3628 int op_flags, bool for_write);
3629struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3630struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3631 bool for_write);
3632struct page *f2fs_get_new_data_page(struct inode *inode,
3633 struct page *ipage, pgoff_t index, bool new_i_size);
3634int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3635void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3636int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3637 int create, int flag);
3638int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3639 u64 start, u64 len);
3640int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3641bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3642bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3643int f2fs_write_single_data_page(struct page *page, int *submitted,
3644 struct bio **bio, sector_t *last_block,
3645 struct writeback_control *wbc,
3646 enum iostat_type io_type,
3647 int compr_blocks, bool allow_balance);
3648void f2fs_invalidate_page(struct page *page, unsigned int offset,
3649 unsigned int length);
3650int f2fs_release_page(struct page *page, gfp_t wait);
3651#ifdef CONFIG_MIGRATION
3652int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3653 struct page *page, enum migrate_mode mode);
3654#endif
3655bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3656void f2fs_clear_page_cache_dirty_tag(struct page *page);
3657int f2fs_init_post_read_processing(void);
3658void f2fs_destroy_post_read_processing(void);
3659int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3660void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3661
3662/*
3663 * gc.c
3664 */
3665int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3666void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3667block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3668int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force,
3669 unsigned int segno);
3670void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3671int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3672int __init f2fs_create_garbage_collection_cache(void);
3673void f2fs_destroy_garbage_collection_cache(void);
3674
3675/*
3676 * recovery.c
3677 */
3678int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3679bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3680int __init f2fs_create_recovery_cache(void);
3681void f2fs_destroy_recovery_cache(void);
3682
3683/*
3684 * debug.c
3685 */
3686#ifdef CONFIG_F2FS_STAT_FS
3687struct f2fs_stat_info {
3688 struct list_head stat_list;
3689 struct f2fs_sb_info *sbi;
3690 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3691 int main_area_segs, main_area_sections, main_area_zones;
3692 unsigned long long hit_largest, hit_cached, hit_rbtree;
3693 unsigned long long hit_total, total_ext;
3694 int ext_tree, zombie_tree, ext_node;
3695 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3696 int ndirty_data, ndirty_qdata;
3697 int inmem_pages;
3698 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3699 int nats, dirty_nats, sits, dirty_sits;
3700 int free_nids, avail_nids, alloc_nids;
3701 int total_count, utilization;
3702 int bg_gc, nr_wb_cp_data, nr_wb_data;
3703 int nr_rd_data, nr_rd_node, nr_rd_meta;
3704 int nr_dio_read, nr_dio_write;
3705 unsigned int io_skip_bggc, other_skip_bggc;
3706 int nr_flushing, nr_flushed, flush_list_empty;
3707 int nr_discarding, nr_discarded;
3708 int nr_discard_cmd;
3709 unsigned int undiscard_blks;
3710 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3711 unsigned int cur_ckpt_time, peak_ckpt_time;
3712 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3713 int compr_inode;
3714 unsigned long long compr_blocks;
3715 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3716 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3717 unsigned int bimodal, avg_vblocks;
3718 int util_free, util_valid, util_invalid;
3719 int rsvd_segs, overp_segs;
3720 int dirty_count, node_pages, meta_pages, compress_pages;
3721 int compress_page_hit;
3722 int prefree_count, call_count, cp_count, bg_cp_count;
3723 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3724 int bg_node_segs, bg_data_segs;
3725 int tot_blks, data_blks, node_blks;
3726 int bg_data_blks, bg_node_blks;
3727 unsigned long long skipped_atomic_files[2];
3728 int curseg[NR_CURSEG_TYPE];
3729 int cursec[NR_CURSEG_TYPE];
3730 int curzone[NR_CURSEG_TYPE];
3731 unsigned int dirty_seg[NR_CURSEG_TYPE];
3732 unsigned int full_seg[NR_CURSEG_TYPE];
3733 unsigned int valid_blks[NR_CURSEG_TYPE];
3734
3735 unsigned int meta_count[META_MAX];
3736 unsigned int segment_count[2];
3737 unsigned int block_count[2];
3738 unsigned int inplace_count;
3739 unsigned long long base_mem, cache_mem, page_mem;
3740};
3741
3742static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3743{
3744 return (struct f2fs_stat_info *)sbi->stat_info;
3745}
3746
3747#define stat_inc_cp_count(si) ((si)->cp_count++)
3748#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3749#define stat_inc_call_count(si) ((si)->call_count++)
3750#define stat_inc_bggc_count(si) ((si)->bg_gc++)
3751#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3752#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3753#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3754#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3755#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3756#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3757#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3758#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3759#define stat_inc_inline_xattr(inode) \
3760 do { \
3761 if (f2fs_has_inline_xattr(inode)) \
3762 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3763 } while (0)
3764#define stat_dec_inline_xattr(inode) \
3765 do { \
3766 if (f2fs_has_inline_xattr(inode)) \
3767 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3768 } while (0)
3769#define stat_inc_inline_inode(inode) \
3770 do { \
3771 if (f2fs_has_inline_data(inode)) \
3772 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3773 } while (0)
3774#define stat_dec_inline_inode(inode) \
3775 do { \
3776 if (f2fs_has_inline_data(inode)) \
3777 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3778 } while (0)
3779#define stat_inc_inline_dir(inode) \
3780 do { \
3781 if (f2fs_has_inline_dentry(inode)) \
3782 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3783 } while (0)
3784#define stat_dec_inline_dir(inode) \
3785 do { \
3786 if (f2fs_has_inline_dentry(inode)) \
3787 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3788 } while (0)
3789#define stat_inc_compr_inode(inode) \
3790 do { \
3791 if (f2fs_compressed_file(inode)) \
3792 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3793 } while (0)
3794#define stat_dec_compr_inode(inode) \
3795 do { \
3796 if (f2fs_compressed_file(inode)) \
3797 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3798 } while (0)
3799#define stat_add_compr_blocks(inode, blocks) \
3800 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3801#define stat_sub_compr_blocks(inode, blocks) \
3802 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3803#define stat_inc_meta_count(sbi, blkaddr) \
3804 do { \
3805 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3806 atomic_inc(&(sbi)->meta_count[META_CP]); \
3807 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3808 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3809 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3810 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3811 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3812 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3813 } while (0)
3814#define stat_inc_seg_type(sbi, curseg) \
3815 ((sbi)->segment_count[(curseg)->alloc_type]++)
3816#define stat_inc_block_count(sbi, curseg) \
3817 ((sbi)->block_count[(curseg)->alloc_type]++)
3818#define stat_inc_inplace_blocks(sbi) \
3819 (atomic_inc(&(sbi)->inplace_count))
3820#define stat_update_max_atomic_write(inode) \
3821 do { \
3822 int cur = F2FS_I_SB(inode)->atomic_files; \
3823 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3824 if (cur > max) \
3825 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3826 } while (0)
3827#define stat_inc_volatile_write(inode) \
3828 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3829#define stat_dec_volatile_write(inode) \
3830 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3831#define stat_update_max_volatile_write(inode) \
3832 do { \
3833 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3834 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3835 if (cur > max) \
3836 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3837 } while (0)
3838#define stat_inc_seg_count(sbi, type, gc_type) \
3839 do { \
3840 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3841 si->tot_segs++; \
3842 if ((type) == SUM_TYPE_DATA) { \
3843 si->data_segs++; \
3844 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3845 } else { \
3846 si->node_segs++; \
3847 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3848 } \
3849 } while (0)
3850
3851#define stat_inc_tot_blk_count(si, blks) \
3852 ((si)->tot_blks += (blks))
3853
3854#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3855 do { \
3856 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3857 stat_inc_tot_blk_count(si, blks); \
3858 si->data_blks += (blks); \
3859 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3860 } while (0)
3861
3862#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3863 do { \
3864 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3865 stat_inc_tot_blk_count(si, blks); \
3866 si->node_blks += (blks); \
3867 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3868 } while (0)
3869
3870int f2fs_build_stats(struct f2fs_sb_info *sbi);
3871void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3872void __init f2fs_create_root_stats(void);
3873void f2fs_destroy_root_stats(void);
3874void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3875#else
3876#define stat_inc_cp_count(si) do { } while (0)
3877#define stat_inc_bg_cp_count(si) do { } while (0)
3878#define stat_inc_call_count(si) do { } while (0)
3879#define stat_inc_bggc_count(si) do { } while (0)
3880#define stat_io_skip_bggc_count(sbi) do { } while (0)
3881#define stat_other_skip_bggc_count(sbi) do { } while (0)
3882#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3883#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3884#define stat_inc_total_hit(sbi) do { } while (0)
3885#define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3886#define stat_inc_largest_node_hit(sbi) do { } while (0)
3887#define stat_inc_cached_node_hit(sbi) do { } while (0)
3888#define stat_inc_inline_xattr(inode) do { } while (0)
3889#define stat_dec_inline_xattr(inode) do { } while (0)
3890#define stat_inc_inline_inode(inode) do { } while (0)
3891#define stat_dec_inline_inode(inode) do { } while (0)
3892#define stat_inc_inline_dir(inode) do { } while (0)
3893#define stat_dec_inline_dir(inode) do { } while (0)
3894#define stat_inc_compr_inode(inode) do { } while (0)
3895#define stat_dec_compr_inode(inode) do { } while (0)
3896#define stat_add_compr_blocks(inode, blocks) do { } while (0)
3897#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3898#define stat_update_max_atomic_write(inode) do { } while (0)
3899#define stat_inc_volatile_write(inode) do { } while (0)
3900#define stat_dec_volatile_write(inode) do { } while (0)
3901#define stat_update_max_volatile_write(inode) do { } while (0)
3902#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3903#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3904#define stat_inc_block_count(sbi, curseg) do { } while (0)
3905#define stat_inc_inplace_blocks(sbi) do { } while (0)
3906#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3907#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3908#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3909#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3910
3911static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3912static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3913static inline void __init f2fs_create_root_stats(void) { }
3914static inline void f2fs_destroy_root_stats(void) { }
3915static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3916#endif
3917
3918extern const struct file_operations f2fs_dir_operations;
3919extern const struct file_operations f2fs_file_operations;
3920extern const struct inode_operations f2fs_file_inode_operations;
3921extern const struct address_space_operations f2fs_dblock_aops;
3922extern const struct address_space_operations f2fs_node_aops;
3923extern const struct address_space_operations f2fs_meta_aops;
3924extern const struct inode_operations f2fs_dir_inode_operations;
3925extern const struct inode_operations f2fs_symlink_inode_operations;
3926extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3927extern const struct inode_operations f2fs_special_inode_operations;
3928extern struct kmem_cache *f2fs_inode_entry_slab;
3929
3930/*
3931 * inline.c
3932 */
3933bool f2fs_may_inline_data(struct inode *inode);
3934bool f2fs_may_inline_dentry(struct inode *inode);
3935void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3936void f2fs_truncate_inline_inode(struct inode *inode,
3937 struct page *ipage, u64 from);
3938int f2fs_read_inline_data(struct inode *inode, struct page *page);
3939int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3940int f2fs_convert_inline_inode(struct inode *inode);
3941int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3942int f2fs_write_inline_data(struct inode *inode, struct page *page);
3943int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3944struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3945 const struct f2fs_filename *fname,
3946 struct page **res_page);
3947int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3948 struct page *ipage);
3949int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3950 struct inode *inode, nid_t ino, umode_t mode);
3951void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3952 struct page *page, struct inode *dir,
3953 struct inode *inode);
3954bool f2fs_empty_inline_dir(struct inode *dir);
3955int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3956 struct fscrypt_str *fstr);
3957int f2fs_inline_data_fiemap(struct inode *inode,
3958 struct fiemap_extent_info *fieinfo,
3959 __u64 start, __u64 len);
3960
3961/*
3962 * shrinker.c
3963 */
3964unsigned long f2fs_shrink_count(struct shrinker *shrink,
3965 struct shrink_control *sc);
3966unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3967 struct shrink_control *sc);
3968void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3969void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3970
3971/*
3972 * extent_cache.c
3973 */
3974struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3975 struct rb_entry *cached_re, unsigned int ofs);
3976struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
3977 struct rb_root_cached *root,
3978 struct rb_node **parent,
3979 unsigned long long key, bool *left_most);
3980struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3981 struct rb_root_cached *root,
3982 struct rb_node **parent,
3983 unsigned int ofs, bool *leftmost);
3984struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3985 struct rb_entry *cached_re, unsigned int ofs,
3986 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3987 struct rb_node ***insert_p, struct rb_node **insert_parent,
3988 bool force, bool *leftmost);
3989bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3990 struct rb_root_cached *root, bool check_key);
3991unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3992void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3993void f2fs_drop_extent_tree(struct inode *inode);
3994unsigned int f2fs_destroy_extent_node(struct inode *inode);
3995void f2fs_destroy_extent_tree(struct inode *inode);
3996bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3997 struct extent_info *ei);
3998void f2fs_update_extent_cache(struct dnode_of_data *dn);
3999void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
4000 pgoff_t fofs, block_t blkaddr, unsigned int len);
4001void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4002int __init f2fs_create_extent_cache(void);
4003void f2fs_destroy_extent_cache(void);
4004
4005/*
4006 * sysfs.c
4007 */
4008#define MIN_RA_MUL 2
4009#define MAX_RA_MUL 256
4010
4011int __init f2fs_init_sysfs(void);
4012void f2fs_exit_sysfs(void);
4013int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4014void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4015
4016/* verity.c */
4017extern const struct fsverity_operations f2fs_verityops;
4018
4019/*
4020 * crypto support
4021 */
4022static inline bool f2fs_encrypted_file(struct inode *inode)
4023{
4024 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4025}
4026
4027static inline void f2fs_set_encrypted_inode(struct inode *inode)
4028{
4029#ifdef CONFIG_FS_ENCRYPTION
4030 file_set_encrypt(inode);
4031 f2fs_set_inode_flags(inode);
4032#endif
4033}
4034
4035/*
4036 * Returns true if the reads of the inode's data need to undergo some
4037 * postprocessing step, like decryption or authenticity verification.
4038 */
4039static inline bool f2fs_post_read_required(struct inode *inode)
4040{
4041 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4042 f2fs_compressed_file(inode);
4043}
4044
4045/*
4046 * compress.c
4047 */
4048#ifdef CONFIG_F2FS_FS_COMPRESSION
4049bool f2fs_is_compressed_page(struct page *page);
4050struct page *f2fs_compress_control_page(struct page *page);
4051int f2fs_prepare_compress_overwrite(struct inode *inode,
4052 struct page **pagep, pgoff_t index, void **fsdata);
4053bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4054 pgoff_t index, unsigned copied);
4055int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4056void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4057bool f2fs_is_compress_backend_ready(struct inode *inode);
4058int f2fs_init_compress_mempool(void);
4059void f2fs_destroy_compress_mempool(void);
4060void f2fs_decompress_cluster(struct decompress_io_ctx *dic);
4061void f2fs_end_read_compressed_page(struct page *page, bool failed,
4062 block_t blkaddr);
4063bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4064bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4065void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4066int f2fs_write_multi_pages(struct compress_ctx *cc,
4067 int *submitted,
4068 struct writeback_control *wbc,
4069 enum iostat_type io_type);
4070int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4071int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4072 unsigned nr_pages, sector_t *last_block_in_bio,
4073 bool is_readahead, bool for_write);
4074struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4075void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed);
4076void f2fs_put_page_dic(struct page *page);
4077int f2fs_init_compress_ctx(struct compress_ctx *cc);
4078void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4079void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4080int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4081void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4082int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4083void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4084int __init f2fs_init_compress_cache(void);
4085void f2fs_destroy_compress_cache(void);
4086struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4087void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4088void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4089 nid_t ino, block_t blkaddr);
4090bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4091 block_t blkaddr);
4092void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4093#define inc_compr_inode_stat(inode) \
4094 do { \
4095 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4096 sbi->compr_new_inode++; \
4097 } while (0)
4098#define add_compr_block_stat(inode, blocks) \
4099 do { \
4100 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4101 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4102 sbi->compr_written_block += blocks; \
4103 sbi->compr_saved_block += diff; \
4104 } while (0)
4105#else
4106static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4107static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4108{
4109 if (!f2fs_compressed_file(inode))
4110 return true;
4111 /* not support compression */
4112 return false;
4113}
4114static inline struct page *f2fs_compress_control_page(struct page *page)
4115{
4116 WARN_ON_ONCE(1);
4117 return ERR_PTR(-EINVAL);
4118}
4119static inline int f2fs_init_compress_mempool(void) { return 0; }
4120static inline void f2fs_destroy_compress_mempool(void) { }
4121static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { }
4122static inline void f2fs_end_read_compressed_page(struct page *page,
4123 bool failed, block_t blkaddr)
4124{
4125 WARN_ON_ONCE(1);
4126}
4127static inline void f2fs_put_page_dic(struct page *page)
4128{
4129 WARN_ON_ONCE(1);
4130}
4131static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4132static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4133static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4134static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4135static inline int __init f2fs_init_compress_cache(void) { return 0; }
4136static inline void f2fs_destroy_compress_cache(void) { }
4137static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4138 block_t blkaddr) { }
4139static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4140 struct page *page, nid_t ino, block_t blkaddr) { }
4141static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4142 struct page *page, block_t blkaddr) { return false; }
4143static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4144 nid_t ino) { }
4145#define inc_compr_inode_stat(inode) do { } while (0)
4146#endif
4147
4148static inline void set_compress_context(struct inode *inode)
4149{
4150 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4151
4152 F2FS_I(inode)->i_compress_algorithm =
4153 F2FS_OPTION(sbi).compress_algorithm;
4154 F2FS_I(inode)->i_log_cluster_size =
4155 F2FS_OPTION(sbi).compress_log_size;
4156 F2FS_I(inode)->i_compress_flag =
4157 F2FS_OPTION(sbi).compress_chksum ?
4158 1 << COMPRESS_CHKSUM : 0;
4159 F2FS_I(inode)->i_cluster_size =
4160 1 << F2FS_I(inode)->i_log_cluster_size;
4161 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4162 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4163 F2FS_OPTION(sbi).compress_level)
4164 F2FS_I(inode)->i_compress_flag |=
4165 F2FS_OPTION(sbi).compress_level <<
4166 COMPRESS_LEVEL_OFFSET;
4167 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4168 set_inode_flag(inode, FI_COMPRESSED_FILE);
4169 stat_inc_compr_inode(inode);
4170 inc_compr_inode_stat(inode);
4171 f2fs_mark_inode_dirty_sync(inode, true);
4172}
4173
4174static inline bool f2fs_disable_compressed_file(struct inode *inode)
4175{
4176 struct f2fs_inode_info *fi = F2FS_I(inode);
4177
4178 if (!f2fs_compressed_file(inode))
4179 return true;
4180 if (S_ISREG(inode->i_mode) &&
4181 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks)))
4182 return false;
4183
4184 fi->i_flags &= ~F2FS_COMPR_FL;
4185 stat_dec_compr_inode(inode);
4186 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4187 f2fs_mark_inode_dirty_sync(inode, true);
4188 return true;
4189}
4190
4191#define F2FS_FEATURE_FUNCS(name, flagname) \
4192static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4193{ \
4194 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4195}
4196
4197F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4198F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4199F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4200F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4201F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4202F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4203F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4204F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4205F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4206F2FS_FEATURE_FUNCS(verity, VERITY);
4207F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4208F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4209F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4210F2FS_FEATURE_FUNCS(readonly, RO);
4211
4212static inline bool f2fs_may_extent_tree(struct inode *inode)
4213{
4214 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4215
4216 if (!test_opt(sbi, EXTENT_CACHE) ||
4217 is_inode_flag_set(inode, FI_NO_EXTENT) ||
4218 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) &&
4219 !f2fs_sb_has_readonly(sbi)))
4220 return false;
4221
4222 /*
4223 * for recovered files during mount do not create extents
4224 * if shrinker is not registered.
4225 */
4226 if (list_empty(&sbi->s_list))
4227 return false;
4228
4229 return S_ISREG(inode->i_mode);
4230}
4231
4232#ifdef CONFIG_BLK_DEV_ZONED
4233static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4234 block_t blkaddr)
4235{
4236 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4237
4238 return test_bit(zno, FDEV(devi).blkz_seq);
4239}
4240#endif
4241
4242static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4243{
4244 return f2fs_sb_has_blkzoned(sbi);
4245}
4246
4247static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4248{
4249 return blk_queue_discard(bdev_get_queue(bdev)) ||
4250 bdev_is_zoned(bdev);
4251}
4252
4253static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4254{
4255 int i;
4256
4257 if (!f2fs_is_multi_device(sbi))
4258 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4259
4260 for (i = 0; i < sbi->s_ndevs; i++)
4261 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4262 return true;
4263 return false;
4264}
4265
4266static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4267{
4268 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4269 f2fs_hw_should_discard(sbi);
4270}
4271
4272static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4273{
4274 int i;
4275
4276 if (!f2fs_is_multi_device(sbi))
4277 return bdev_read_only(sbi->sb->s_bdev);
4278
4279 for (i = 0; i < sbi->s_ndevs; i++)
4280 if (bdev_read_only(FDEV(i).bdev))
4281 return true;
4282 return false;
4283}
4284
4285static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4286{
4287 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4288}
4289
4290static inline bool f2fs_may_compress(struct inode *inode)
4291{
4292 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4293 f2fs_is_atomic_file(inode) ||
4294 f2fs_is_volatile_file(inode))
4295 return false;
4296 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4297}
4298
4299static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4300 u64 blocks, bool add)
4301{
4302 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4303 struct f2fs_inode_info *fi = F2FS_I(inode);
4304
4305 /* don't update i_compr_blocks if saved blocks were released */
4306 if (!add && !atomic_read(&fi->i_compr_blocks))
4307 return;
4308
4309 if (add) {
4310 atomic_add(diff, &fi->i_compr_blocks);
4311 stat_add_compr_blocks(inode, diff);
4312 } else {
4313 atomic_sub(diff, &fi->i_compr_blocks);
4314 stat_sub_compr_blocks(inode, diff);
4315 }
4316 f2fs_mark_inode_dirty_sync(inode, true);
4317}
4318
4319static inline int block_unaligned_IO(struct inode *inode,
4320 struct kiocb *iocb, struct iov_iter *iter)
4321{
4322 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4323 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4324 loff_t offset = iocb->ki_pos;
4325 unsigned long align = offset | iov_iter_alignment(iter);
4326
4327 return align & blocksize_mask;
4328}
4329
4330static inline bool f2fs_force_buffered_io(struct inode *inode,
4331 struct kiocb *iocb, struct iov_iter *iter)
4332{
4333 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4334 int rw = iov_iter_rw(iter);
4335
4336 if (f2fs_post_read_required(inode))
4337 return true;
4338 if (f2fs_is_multi_device(sbi))
4339 return true;
4340 /*
4341 * for blkzoned device, fallback direct IO to buffered IO, so
4342 * all IOs can be serialized by log-structured write.
4343 */
4344 if (f2fs_sb_has_blkzoned(sbi))
4345 return true;
4346 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4347 if (block_unaligned_IO(inode, iocb, iter))
4348 return true;
4349 if (F2FS_IO_ALIGNED(sbi))
4350 return true;
4351 }
4352 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
4353 return true;
4354
4355 return false;
4356}
4357
4358static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4359{
4360 return fsverity_active(inode) &&
4361 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4362}
4363
4364#ifdef CONFIG_F2FS_FAULT_INJECTION
4365extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4366 unsigned int type);
4367#else
4368#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4369#endif
4370
4371static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4372{
4373#ifdef CONFIG_QUOTA
4374 if (f2fs_sb_has_quota_ino(sbi))
4375 return true;
4376 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4377 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4378 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4379 return true;
4380#endif
4381 return false;
4382}
4383
4384static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4385{
4386 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4387}
4388
4389#define EFSBADCRC EBADMSG /* Bad CRC detected */
4390#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4391
4392#endif /* _LINUX_F2FS_H */