]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame_incremental - fs/f2fs/f2fs.h
UBUNTU: Ubuntu-5.15.0-39.42
[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_BIO, /* it's obsolete due to bio_alloc() will never fail */
47 FAULT_ALLOC_NID,
48 FAULT_ORPHAN,
49 FAULT_BLOCK,
50 FAULT_DIR_DEPTH,
51 FAULT_EVICT_INODE,
52 FAULT_TRUNCATE,
53 FAULT_READ_IO,
54 FAULT_CHECKPOINT,
55 FAULT_DISCARD,
56 FAULT_WRITE_IO,
57 FAULT_SLAB_ALLOC,
58 FAULT_MAX,
59};
60
61#ifdef CONFIG_F2FS_FAULT_INJECTION
62#define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1)
63
64struct f2fs_fault_info {
65 atomic_t inject_ops;
66 unsigned int inject_rate;
67 unsigned int inject_type;
68};
69
70extern const char *f2fs_fault_name[FAULT_MAX];
71#define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type)))
72#endif
73
74/*
75 * For mount options
76 */
77#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002
78#define F2FS_MOUNT_DISCARD 0x00000004
79#define F2FS_MOUNT_NOHEAP 0x00000008
80#define F2FS_MOUNT_XATTR_USER 0x00000010
81#define F2FS_MOUNT_POSIX_ACL 0x00000020
82#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040
83#define F2FS_MOUNT_INLINE_XATTR 0x00000080
84#define F2FS_MOUNT_INLINE_DATA 0x00000100
85#define F2FS_MOUNT_INLINE_DENTRY 0x00000200
86#define F2FS_MOUNT_FLUSH_MERGE 0x00000400
87#define F2FS_MOUNT_NOBARRIER 0x00000800
88#define F2FS_MOUNT_FASTBOOT 0x00001000
89#define F2FS_MOUNT_EXTENT_CACHE 0x00002000
90#define F2FS_MOUNT_DATA_FLUSH 0x00008000
91#define F2FS_MOUNT_FAULT_INJECTION 0x00010000
92#define F2FS_MOUNT_USRQUOTA 0x00080000
93#define F2FS_MOUNT_GRPQUOTA 0x00100000
94#define F2FS_MOUNT_PRJQUOTA 0x00200000
95#define F2FS_MOUNT_QUOTA 0x00400000
96#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000
97#define F2FS_MOUNT_RESERVE_ROOT 0x01000000
98#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000
99#define F2FS_MOUNT_NORECOVERY 0x04000000
100#define F2FS_MOUNT_ATGC 0x08000000
101#define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000
102#define F2FS_MOUNT_GC_MERGE 0x20000000
103#define F2FS_MOUNT_COMPRESS_CACHE 0x40000000
104
105#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
106#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
107#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
108#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
109
110#define ver_after(a, b) (typecheck(unsigned long long, a) && \
111 typecheck(unsigned long long, b) && \
112 ((long long)((a) - (b)) > 0))
113
114typedef u32 block_t; /*
115 * should not change u32, since it is the on-disk block
116 * address format, __le32.
117 */
118typedef u32 nid_t;
119
120#define COMPRESS_EXT_NUM 16
121
122struct f2fs_mount_info {
123 unsigned int opt;
124 int write_io_size_bits; /* Write IO size bits */
125 block_t root_reserved_blocks; /* root reserved blocks */
126 kuid_t s_resuid; /* reserved blocks for uid */
127 kgid_t s_resgid; /* reserved blocks for gid */
128 int active_logs; /* # of active logs */
129 int inline_xattr_size; /* inline xattr size */
130#ifdef CONFIG_F2FS_FAULT_INJECTION
131 struct f2fs_fault_info fault_info; /* For fault injection */
132#endif
133#ifdef CONFIG_QUOTA
134 /* Names of quota files with journalled quota */
135 char *s_qf_names[MAXQUOTAS];
136 int s_jquota_fmt; /* Format of quota to use */
137#endif
138 /* For which write hints are passed down to block layer */
139 int whint_mode;
140 int alloc_mode; /* segment allocation policy */
141 int fsync_mode; /* fsync policy */
142 int fs_mode; /* fs mode: LFS or ADAPTIVE */
143 int bggc_mode; /* bggc mode: off, on or sync */
144 int discard_unit; /*
145 * discard command's offset/size should
146 * be aligned to this unit: block,
147 * segment or section
148 */
149 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
150 block_t unusable_cap_perc; /* percentage for cap */
151 block_t unusable_cap; /* Amount of space allowed to be
152 * unusable when disabling checkpoint
153 */
154
155 /* For compression */
156 unsigned char compress_algorithm; /* algorithm type */
157 unsigned char compress_log_size; /* cluster log size */
158 unsigned char compress_level; /* compress level */
159 bool compress_chksum; /* compressed data chksum */
160 unsigned char compress_ext_cnt; /* extension count */
161 unsigned char nocompress_ext_cnt; /* nocompress extension count */
162 int compress_mode; /* compression mode */
163 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
164 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
165};
166
167#define F2FS_FEATURE_ENCRYPT 0x0001
168#define F2FS_FEATURE_BLKZONED 0x0002
169#define F2FS_FEATURE_ATOMIC_WRITE 0x0004
170#define F2FS_FEATURE_EXTRA_ATTR 0x0008
171#define F2FS_FEATURE_PRJQUOTA 0x0010
172#define F2FS_FEATURE_INODE_CHKSUM 0x0020
173#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040
174#define F2FS_FEATURE_QUOTA_INO 0x0080
175#define F2FS_FEATURE_INODE_CRTIME 0x0100
176#define F2FS_FEATURE_LOST_FOUND 0x0200
177#define F2FS_FEATURE_VERITY 0x0400
178#define F2FS_FEATURE_SB_CHKSUM 0x0800
179#define F2FS_FEATURE_CASEFOLD 0x1000
180#define F2FS_FEATURE_COMPRESSION 0x2000
181#define F2FS_FEATURE_RO 0x4000
182
183#define __F2FS_HAS_FEATURE(raw_super, mask) \
184 ((raw_super->feature & cpu_to_le32(mask)) != 0)
185#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
186#define F2FS_SET_FEATURE(sbi, mask) \
187 (sbi->raw_super->feature |= cpu_to_le32(mask))
188#define F2FS_CLEAR_FEATURE(sbi, mask) \
189 (sbi->raw_super->feature &= ~cpu_to_le32(mask))
190
191/*
192 * Default values for user and/or group using reserved blocks
193 */
194#define F2FS_DEF_RESUID 0
195#define F2FS_DEF_RESGID 0
196
197/*
198 * For checkpoint manager
199 */
200enum {
201 NAT_BITMAP,
202 SIT_BITMAP
203};
204
205#define CP_UMOUNT 0x00000001
206#define CP_FASTBOOT 0x00000002
207#define CP_SYNC 0x00000004
208#define CP_RECOVERY 0x00000008
209#define CP_DISCARD 0x00000010
210#define CP_TRIMMED 0x00000020
211#define CP_PAUSE 0x00000040
212#define CP_RESIZE 0x00000080
213
214#define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi)
215#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
216#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
217#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
218#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
219#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
220#define DEF_CP_INTERVAL 60 /* 60 secs */
221#define DEF_IDLE_INTERVAL 5 /* 5 secs */
222#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
223#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
224#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
225
226struct cp_control {
227 int reason;
228 __u64 trim_start;
229 __u64 trim_end;
230 __u64 trim_minlen;
231};
232
233/*
234 * indicate meta/data type
235 */
236enum {
237 META_CP,
238 META_NAT,
239 META_SIT,
240 META_SSA,
241 META_MAX,
242 META_POR,
243 DATA_GENERIC, /* check range only */
244 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
245 DATA_GENERIC_ENHANCE_READ, /*
246 * strong check on range and segment
247 * bitmap but no warning due to race
248 * condition of read on truncated area
249 * by extent_cache
250 */
251 META_GENERIC,
252};
253
254/* for the list of ino */
255enum {
256 ORPHAN_INO, /* for orphan ino list */
257 APPEND_INO, /* for append ino list */
258 UPDATE_INO, /* for update ino list */
259 TRANS_DIR_INO, /* for trasactions dir ino list */
260 FLUSH_INO, /* for multiple device flushing */
261 MAX_INO_ENTRY, /* max. list */
262};
263
264struct ino_entry {
265 struct list_head list; /* list head */
266 nid_t ino; /* inode number */
267 unsigned int dirty_device; /* dirty device bitmap */
268};
269
270/* for the list of inodes to be GCed */
271struct inode_entry {
272 struct list_head list; /* list head */
273 struct inode *inode; /* vfs inode pointer */
274};
275
276struct fsync_node_entry {
277 struct list_head list; /* list head */
278 struct page *page; /* warm node page pointer */
279 unsigned int seq_id; /* sequence id */
280};
281
282struct ckpt_req {
283 struct completion wait; /* completion for checkpoint done */
284 struct llist_node llnode; /* llist_node to be linked in wait queue */
285 int ret; /* return code of checkpoint */
286 ktime_t queue_time; /* request queued time */
287};
288
289struct ckpt_req_control {
290 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
291 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
292 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
293 atomic_t issued_ckpt; /* # of actually issued ckpts */
294 atomic_t total_ckpt; /* # of total ckpts */
295 atomic_t queued_ckpt; /* # of queued ckpts */
296 struct llist_head issue_list; /* list for command issue */
297 spinlock_t stat_lock; /* lock for below checkpoint time stats */
298 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
299 unsigned int peak_time; /* peak wait time in msec until now */
300};
301
302/* for the bitmap indicate blocks to be discarded */
303struct discard_entry {
304 struct list_head list; /* list head */
305 block_t start_blkaddr; /* start blockaddr of current segment */
306 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
307};
308
309/* default discard granularity of inner discard thread, unit: block count */
310#define DEFAULT_DISCARD_GRANULARITY 16
311
312/* max discard pend list number */
313#define MAX_PLIST_NUM 512
314#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
315 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
316
317enum {
318 D_PREP, /* initial */
319 D_PARTIAL, /* partially submitted */
320 D_SUBMIT, /* all submitted */
321 D_DONE, /* finished */
322};
323
324struct discard_info {
325 block_t lstart; /* logical start address */
326 block_t len; /* length */
327 block_t start; /* actual start address in dev */
328};
329
330struct discard_cmd {
331 struct rb_node rb_node; /* rb node located in rb-tree */
332 union {
333 struct {
334 block_t lstart; /* logical start address */
335 block_t len; /* length */
336 block_t start; /* actual start address in dev */
337 };
338 struct discard_info di; /* discard info */
339
340 };
341 struct list_head list; /* command list */
342 struct completion wait; /* compleation */
343 struct block_device *bdev; /* bdev */
344 unsigned short ref; /* reference count */
345 unsigned char state; /* state */
346 unsigned char queued; /* queued discard */
347 int error; /* bio error */
348 spinlock_t lock; /* for state/bio_ref updating */
349 unsigned short bio_ref; /* bio reference count */
350};
351
352enum {
353 DPOLICY_BG,
354 DPOLICY_FORCE,
355 DPOLICY_FSTRIM,
356 DPOLICY_UMOUNT,
357 MAX_DPOLICY,
358};
359
360struct discard_policy {
361 int type; /* type of discard */
362 unsigned int min_interval; /* used for candidates exist */
363 unsigned int mid_interval; /* used for device busy */
364 unsigned int max_interval; /* used for candidates not exist */
365 unsigned int max_requests; /* # of discards issued per round */
366 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
367 bool io_aware; /* issue discard in idle time */
368 bool sync; /* submit discard with REQ_SYNC flag */
369 bool ordered; /* issue discard by lba order */
370 bool timeout; /* discard timeout for put_super */
371 unsigned int granularity; /* discard granularity */
372};
373
374struct discard_cmd_control {
375 struct task_struct *f2fs_issue_discard; /* discard thread */
376 struct list_head entry_list; /* 4KB discard entry list */
377 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
378 struct list_head wait_list; /* store on-flushing entries */
379 struct list_head fstrim_list; /* in-flight discard from fstrim */
380 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
381 unsigned int discard_wake; /* to wake up discard thread */
382 struct mutex cmd_lock;
383 unsigned int nr_discards; /* # of discards in the list */
384 unsigned int max_discards; /* max. discards to be issued */
385 unsigned int discard_granularity; /* discard granularity */
386 unsigned int undiscard_blks; /* # of undiscard blocks */
387 unsigned int next_pos; /* next discard position */
388 atomic_t issued_discard; /* # of issued discard */
389 atomic_t queued_discard; /* # of queued discard */
390 atomic_t discard_cmd_cnt; /* # of cached cmd count */
391 struct rb_root_cached root; /* root of discard rb-tree */
392 bool rbtree_check; /* config for consistence check */
393};
394
395/* for the list of fsync inodes, used only during recovery */
396struct fsync_inode_entry {
397 struct list_head list; /* list head */
398 struct inode *inode; /* vfs inode pointer */
399 block_t blkaddr; /* block address locating the last fsync */
400 block_t last_dentry; /* block address locating the last dentry */
401};
402
403#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
404#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
405
406#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
407#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
408#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
409#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
410
411#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
412#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
413
414static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
415{
416 int before = nats_in_cursum(journal);
417
418 journal->n_nats = cpu_to_le16(before + i);
419 return before;
420}
421
422static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
423{
424 int before = sits_in_cursum(journal);
425
426 journal->n_sits = cpu_to_le16(before + i);
427 return before;
428}
429
430static inline bool __has_cursum_space(struct f2fs_journal *journal,
431 int size, int type)
432{
433 if (type == NAT_JOURNAL)
434 return size <= MAX_NAT_JENTRIES(journal);
435 return size <= MAX_SIT_JENTRIES(journal);
436}
437
438/* for inline stuff */
439#define DEF_INLINE_RESERVED_SIZE 1
440static inline int get_extra_isize(struct inode *inode);
441static inline int get_inline_xattr_addrs(struct inode *inode);
442#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
443 (CUR_ADDRS_PER_INODE(inode) - \
444 get_inline_xattr_addrs(inode) - \
445 DEF_INLINE_RESERVED_SIZE))
446
447/* for inline dir */
448#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
449 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
450 BITS_PER_BYTE + 1))
451#define INLINE_DENTRY_BITMAP_SIZE(inode) \
452 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
453#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
454 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
455 NR_INLINE_DENTRY(inode) + \
456 INLINE_DENTRY_BITMAP_SIZE(inode)))
457
458/*
459 * For INODE and NODE manager
460 */
461/* for directory operations */
462
463struct f2fs_filename {
464 /*
465 * The filename the user specified. This is NULL for some
466 * filesystem-internal operations, e.g. converting an inline directory
467 * to a non-inline one, or roll-forward recovering an encrypted dentry.
468 */
469 const struct qstr *usr_fname;
470
471 /*
472 * The on-disk filename. For encrypted directories, this is encrypted.
473 * This may be NULL for lookups in an encrypted dir without the key.
474 */
475 struct fscrypt_str disk_name;
476
477 /* The dirhash of this filename */
478 f2fs_hash_t hash;
479
480#ifdef CONFIG_FS_ENCRYPTION
481 /*
482 * For lookups in encrypted directories: either the buffer backing
483 * disk_name, or a buffer that holds the decoded no-key name.
484 */
485 struct fscrypt_str crypto_buf;
486#endif
487#ifdef CONFIG_UNICODE
488 /*
489 * For casefolded directories: the casefolded name, but it's left NULL
490 * if the original name is not valid Unicode, if the directory is both
491 * casefolded and encrypted and its encryption key is unavailable, or if
492 * the filesystem is doing an internal operation where usr_fname is also
493 * NULL. In all these cases we fall back to treating the name as an
494 * opaque byte sequence.
495 */
496 struct fscrypt_str cf_name;
497#endif
498};
499
500struct f2fs_dentry_ptr {
501 struct inode *inode;
502 void *bitmap;
503 struct f2fs_dir_entry *dentry;
504 __u8 (*filename)[F2FS_SLOT_LEN];
505 int max;
506 int nr_bitmap;
507};
508
509static inline void make_dentry_ptr_block(struct inode *inode,
510 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
511{
512 d->inode = inode;
513 d->max = NR_DENTRY_IN_BLOCK;
514 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
515 d->bitmap = t->dentry_bitmap;
516 d->dentry = t->dentry;
517 d->filename = t->filename;
518}
519
520static inline void make_dentry_ptr_inline(struct inode *inode,
521 struct f2fs_dentry_ptr *d, void *t)
522{
523 int entry_cnt = NR_INLINE_DENTRY(inode);
524 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
525 int reserved_size = INLINE_RESERVED_SIZE(inode);
526
527 d->inode = inode;
528 d->max = entry_cnt;
529 d->nr_bitmap = bitmap_size;
530 d->bitmap = t;
531 d->dentry = t + bitmap_size + reserved_size;
532 d->filename = t + bitmap_size + reserved_size +
533 SIZE_OF_DIR_ENTRY * entry_cnt;
534}
535
536/*
537 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
538 * as its node offset to distinguish from index node blocks.
539 * But some bits are used to mark the node block.
540 */
541#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
542 >> OFFSET_BIT_SHIFT)
543enum {
544 ALLOC_NODE, /* allocate a new node page if needed */
545 LOOKUP_NODE, /* look up a node without readahead */
546 LOOKUP_NODE_RA, /*
547 * look up a node with readahead called
548 * by get_data_block.
549 */
550};
551
552#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
553
554/* congestion wait timeout value, default: 20ms */
555#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
556
557/* maximum retry quota flush count */
558#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
559
560#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
561
562#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
563
564/* for in-memory extent cache entry */
565#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
566
567/* number of extent info in extent cache we try to shrink */
568#define EXTENT_CACHE_SHRINK_NUMBER 128
569
570struct rb_entry {
571 struct rb_node rb_node; /* rb node located in rb-tree */
572 union {
573 struct {
574 unsigned int ofs; /* start offset of the entry */
575 unsigned int len; /* length of the entry */
576 };
577 unsigned long long key; /* 64-bits key */
578 } __packed;
579};
580
581struct extent_info {
582 unsigned int fofs; /* start offset in a file */
583 unsigned int len; /* length of the extent */
584 u32 blk; /* start block address of the extent */
585#ifdef CONFIG_F2FS_FS_COMPRESSION
586 unsigned int c_len; /* physical extent length of compressed blocks */
587#endif
588};
589
590struct extent_node {
591 struct rb_node rb_node; /* rb node located in rb-tree */
592 struct extent_info ei; /* extent info */
593 struct list_head list; /* node in global extent list of sbi */
594 struct extent_tree *et; /* extent tree pointer */
595};
596
597struct extent_tree {
598 nid_t ino; /* inode number */
599 struct rb_root_cached root; /* root of extent info rb-tree */
600 struct extent_node *cached_en; /* recently accessed extent node */
601 struct extent_info largest; /* largested extent info */
602 struct list_head list; /* to be used by sbi->zombie_list */
603 rwlock_t lock; /* protect extent info rb-tree */
604 atomic_t node_cnt; /* # of extent node in rb-tree*/
605 bool largest_updated; /* largest extent updated */
606};
607
608/*
609 * This structure is taken from ext4_map_blocks.
610 *
611 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks().
612 */
613#define F2FS_MAP_NEW (1 << BH_New)
614#define F2FS_MAP_MAPPED (1 << BH_Mapped)
615#define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten)
616#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
617 F2FS_MAP_UNWRITTEN)
618
619struct f2fs_map_blocks {
620 block_t m_pblk;
621 block_t m_lblk;
622 unsigned int m_len;
623 unsigned int m_flags;
624 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
625 pgoff_t *m_next_extent; /* point to next possible extent */
626 int m_seg_type;
627 bool m_may_create; /* indicate it is from write path */
628};
629
630/* for flag in get_data_block */
631enum {
632 F2FS_GET_BLOCK_DEFAULT,
633 F2FS_GET_BLOCK_FIEMAP,
634 F2FS_GET_BLOCK_BMAP,
635 F2FS_GET_BLOCK_DIO,
636 F2FS_GET_BLOCK_PRE_DIO,
637 F2FS_GET_BLOCK_PRE_AIO,
638 F2FS_GET_BLOCK_PRECACHE,
639};
640
641/*
642 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
643 */
644#define FADVISE_COLD_BIT 0x01
645#define FADVISE_LOST_PINO_BIT 0x02
646#define FADVISE_ENCRYPT_BIT 0x04
647#define FADVISE_ENC_NAME_BIT 0x08
648#define FADVISE_KEEP_SIZE_BIT 0x10
649#define FADVISE_HOT_BIT 0x20
650#define FADVISE_VERITY_BIT 0x40
651
652#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
653
654#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
655#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
656#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
657
658#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
659#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
660#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
661
662#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
663#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
664
665#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
666#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
667
668#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
669#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
670
671#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
672#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
673#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
674
675#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
676#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
677
678#define DEF_DIR_LEVEL 0
679
680enum {
681 GC_FAILURE_PIN,
682 GC_FAILURE_ATOMIC,
683 MAX_GC_FAILURE
684};
685
686/* used for f2fs_inode_info->flags */
687enum {
688 FI_NEW_INODE, /* indicate newly allocated inode */
689 FI_DIRTY_INODE, /* indicate inode is dirty or not */
690 FI_AUTO_RECOVER, /* indicate inode is recoverable */
691 FI_DIRTY_DIR, /* indicate directory has dirty pages */
692 FI_INC_LINK, /* need to increment i_nlink */
693 FI_ACL_MODE, /* indicate acl mode */
694 FI_NO_ALLOC, /* should not allocate any blocks */
695 FI_FREE_NID, /* free allocated nide */
696 FI_NO_EXTENT, /* not to use the extent cache */
697 FI_INLINE_XATTR, /* used for inline xattr */
698 FI_INLINE_DATA, /* used for inline data*/
699 FI_INLINE_DENTRY, /* used for inline dentry */
700 FI_APPEND_WRITE, /* inode has appended data */
701 FI_UPDATE_WRITE, /* inode has in-place-update data */
702 FI_NEED_IPU, /* used for ipu per file */
703 FI_ATOMIC_FILE, /* indicate atomic file */
704 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */
705 FI_VOLATILE_FILE, /* indicate volatile file */
706 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
707 FI_DROP_CACHE, /* drop dirty page cache */
708 FI_DATA_EXIST, /* indicate data exists */
709 FI_INLINE_DOTS, /* indicate inline dot dentries */
710 FI_DO_DEFRAG, /* indicate defragment is running */
711 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
712 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */
713 FI_HOT_DATA, /* indicate file is hot */
714 FI_EXTRA_ATTR, /* indicate file has extra attribute */
715 FI_PROJ_INHERIT, /* indicate file inherits projectid */
716 FI_PIN_FILE, /* indicate file should not be gced */
717 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */
718 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
719 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
720 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
721 FI_MMAP_FILE, /* indicate file was mmapped */
722 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
723 FI_COMPRESS_RELEASED, /* compressed blocks were released */
724 FI_ALIGNED_WRITE, /* enable aligned write */
725 FI_MAX, /* max flag, never be used */
726};
727
728struct f2fs_inode_info {
729 struct inode vfs_inode; /* serve a vfs inode */
730 unsigned long i_flags; /* keep an inode flags for ioctl */
731 unsigned char i_advise; /* use to give file attribute hints */
732 unsigned char i_dir_level; /* use for dentry level for large dir */
733 unsigned int i_current_depth; /* only for directory depth */
734 /* for gc failure statistic */
735 unsigned int i_gc_failures[MAX_GC_FAILURE];
736 unsigned int i_pino; /* parent inode number */
737 umode_t i_acl_mode; /* keep file acl mode temporarily */
738
739 /* Use below internally in f2fs*/
740 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
741 struct rw_semaphore i_sem; /* protect fi info */
742 atomic_t dirty_pages; /* # of dirty pages */
743 f2fs_hash_t chash; /* hash value of given file name */
744 unsigned int clevel; /* maximum level of given file name */
745 struct task_struct *task; /* lookup and create consistency */
746 struct task_struct *cp_task; /* separate cp/wb IO stats*/
747 nid_t i_xattr_nid; /* node id that contains xattrs */
748 loff_t last_disk_size; /* lastly written file size */
749 spinlock_t i_size_lock; /* protect last_disk_size */
750
751#ifdef CONFIG_QUOTA
752 struct dquot *i_dquot[MAXQUOTAS];
753
754 /* quota space reservation, managed internally by quota code */
755 qsize_t i_reserved_quota;
756#endif
757 struct list_head dirty_list; /* dirty list for dirs and files */
758 struct list_head gdirty_list; /* linked in global dirty list */
759 struct list_head inmem_ilist; /* list for inmem inodes */
760 struct list_head inmem_pages; /* inmemory pages managed by f2fs */
761 struct task_struct *inmem_task; /* store inmemory task */
762 struct mutex inmem_lock; /* lock for inmemory pages */
763 struct extent_tree *extent_tree; /* cached extent_tree entry */
764
765 /* avoid racing between foreground op and gc */
766 struct rw_semaphore i_gc_rwsem[2];
767 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */
768
769 int i_extra_isize; /* size of extra space located in i_addr */
770 kprojid_t i_projid; /* id for project quota */
771 int i_inline_xattr_size; /* inline xattr size */
772 struct timespec64 i_crtime; /* inode creation time */
773 struct timespec64 i_disk_time[4];/* inode disk times */
774
775 /* for file compress */
776 atomic_t i_compr_blocks; /* # of compressed blocks */
777 unsigned char i_compress_algorithm; /* algorithm type */
778 unsigned char i_log_cluster_size; /* log of cluster size */
779 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
780 unsigned short i_compress_flag; /* compress flag */
781 unsigned int i_cluster_size; /* cluster size */
782};
783
784static inline void get_extent_info(struct extent_info *ext,
785 struct f2fs_extent *i_ext)
786{
787 ext->fofs = le32_to_cpu(i_ext->fofs);
788 ext->blk = le32_to_cpu(i_ext->blk);
789 ext->len = le32_to_cpu(i_ext->len);
790}
791
792static inline void set_raw_extent(struct extent_info *ext,
793 struct f2fs_extent *i_ext)
794{
795 i_ext->fofs = cpu_to_le32(ext->fofs);
796 i_ext->blk = cpu_to_le32(ext->blk);
797 i_ext->len = cpu_to_le32(ext->len);
798}
799
800static inline void set_extent_info(struct extent_info *ei, unsigned int fofs,
801 u32 blk, unsigned int len)
802{
803 ei->fofs = fofs;
804 ei->blk = blk;
805 ei->len = len;
806#ifdef CONFIG_F2FS_FS_COMPRESSION
807 ei->c_len = 0;
808#endif
809}
810
811static inline bool __is_discard_mergeable(struct discard_info *back,
812 struct discard_info *front, unsigned int max_len)
813{
814 return (back->lstart + back->len == front->lstart) &&
815 (back->len + front->len <= max_len);
816}
817
818static inline bool __is_discard_back_mergeable(struct discard_info *cur,
819 struct discard_info *back, unsigned int max_len)
820{
821 return __is_discard_mergeable(back, cur, max_len);
822}
823
824static inline bool __is_discard_front_mergeable(struct discard_info *cur,
825 struct discard_info *front, unsigned int max_len)
826{
827 return __is_discard_mergeable(cur, front, max_len);
828}
829
830static inline bool __is_extent_mergeable(struct extent_info *back,
831 struct extent_info *front)
832{
833#ifdef CONFIG_F2FS_FS_COMPRESSION
834 if (back->c_len && back->len != back->c_len)
835 return false;
836 if (front->c_len && front->len != front->c_len)
837 return false;
838#endif
839 return (back->fofs + back->len == front->fofs &&
840 back->blk + back->len == front->blk);
841}
842
843static inline bool __is_back_mergeable(struct extent_info *cur,
844 struct extent_info *back)
845{
846 return __is_extent_mergeable(back, cur);
847}
848
849static inline bool __is_front_mergeable(struct extent_info *cur,
850 struct extent_info *front)
851{
852 return __is_extent_mergeable(cur, front);
853}
854
855extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
856static inline void __try_update_largest_extent(struct extent_tree *et,
857 struct extent_node *en)
858{
859 if (en->ei.len > et->largest.len) {
860 et->largest = en->ei;
861 et->largest_updated = true;
862 }
863}
864
865/*
866 * For free nid management
867 */
868enum nid_state {
869 FREE_NID, /* newly added to free nid list */
870 PREALLOC_NID, /* it is preallocated */
871 MAX_NID_STATE,
872};
873
874enum nat_state {
875 TOTAL_NAT,
876 DIRTY_NAT,
877 RECLAIMABLE_NAT,
878 MAX_NAT_STATE,
879};
880
881struct f2fs_nm_info {
882 block_t nat_blkaddr; /* base disk address of NAT */
883 nid_t max_nid; /* maximum possible node ids */
884 nid_t available_nids; /* # of available node ids */
885 nid_t next_scan_nid; /* the next nid to be scanned */
886 unsigned int ram_thresh; /* control the memory footprint */
887 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
888 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
889
890 /* NAT cache management */
891 struct radix_tree_root nat_root;/* root of the nat entry cache */
892 struct radix_tree_root nat_set_root;/* root of the nat set cache */
893 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */
894 struct list_head nat_entries; /* cached nat entry list (clean) */
895 spinlock_t nat_list_lock; /* protect clean nat entry list */
896 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
897 unsigned int nat_blocks; /* # of nat blocks */
898
899 /* free node ids management */
900 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
901 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
902 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
903 spinlock_t nid_list_lock; /* protect nid lists ops */
904 struct mutex build_lock; /* lock for build free nids */
905 unsigned char **free_nid_bitmap;
906 unsigned char *nat_block_bitmap;
907 unsigned short *free_nid_count; /* free nid count of NAT block */
908
909 /* for checkpoint */
910 char *nat_bitmap; /* NAT bitmap pointer */
911
912 unsigned int nat_bits_blocks; /* # of nat bits blocks */
913 unsigned char *nat_bits; /* NAT bits blocks */
914 unsigned char *full_nat_bits; /* full NAT pages */
915 unsigned char *empty_nat_bits; /* empty NAT pages */
916#ifdef CONFIG_F2FS_CHECK_FS
917 char *nat_bitmap_mir; /* NAT bitmap mirror */
918#endif
919 int bitmap_size; /* bitmap size */
920};
921
922/*
923 * this structure is used as one of function parameters.
924 * all the information are dedicated to a given direct node block determined
925 * by the data offset in a file.
926 */
927struct dnode_of_data {
928 struct inode *inode; /* vfs inode pointer */
929 struct page *inode_page; /* its inode page, NULL is possible */
930 struct page *node_page; /* cached direct node page */
931 nid_t nid; /* node id of the direct node block */
932 unsigned int ofs_in_node; /* data offset in the node page */
933 bool inode_page_locked; /* inode page is locked or not */
934 bool node_changed; /* is node block changed */
935 char cur_level; /* level of hole node page */
936 char max_level; /* level of current page located */
937 block_t data_blkaddr; /* block address of the node block */
938};
939
940static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
941 struct page *ipage, struct page *npage, nid_t nid)
942{
943 memset(dn, 0, sizeof(*dn));
944 dn->inode = inode;
945 dn->inode_page = ipage;
946 dn->node_page = npage;
947 dn->nid = nid;
948}
949
950/*
951 * For SIT manager
952 *
953 * By default, there are 6 active log areas across the whole main area.
954 * When considering hot and cold data separation to reduce cleaning overhead,
955 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
956 * respectively.
957 * In the current design, you should not change the numbers intentionally.
958 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
959 * logs individually according to the underlying devices. (default: 6)
960 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
961 * data and 8 for node logs.
962 */
963#define NR_CURSEG_DATA_TYPE (3)
964#define NR_CURSEG_NODE_TYPE (3)
965#define NR_CURSEG_INMEM_TYPE (2)
966#define NR_CURSEG_RO_TYPE (2)
967#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
968#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
969
970enum {
971 CURSEG_HOT_DATA = 0, /* directory entry blocks */
972 CURSEG_WARM_DATA, /* data blocks */
973 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
974 CURSEG_HOT_NODE, /* direct node blocks of directory files */
975 CURSEG_WARM_NODE, /* direct node blocks of normal files */
976 CURSEG_COLD_NODE, /* indirect node blocks */
977 NR_PERSISTENT_LOG, /* number of persistent log */
978 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
979 /* pinned file that needs consecutive block address */
980 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
981 NO_CHECK_TYPE, /* number of persistent & inmem log */
982};
983
984struct flush_cmd {
985 struct completion wait;
986 struct llist_node llnode;
987 nid_t ino;
988 int ret;
989};
990
991struct flush_cmd_control {
992 struct task_struct *f2fs_issue_flush; /* flush thread */
993 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
994 atomic_t issued_flush; /* # of issued flushes */
995 atomic_t queued_flush; /* # of queued flushes */
996 struct llist_head issue_list; /* list for command issue */
997 struct llist_node *dispatch_list; /* list for command dispatch */
998};
999
1000struct f2fs_sm_info {
1001 struct sit_info *sit_info; /* whole segment information */
1002 struct free_segmap_info *free_info; /* free segment information */
1003 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1004 struct curseg_info *curseg_array; /* active segment information */
1005
1006 struct rw_semaphore curseg_lock; /* for preventing curseg change */
1007
1008 block_t seg0_blkaddr; /* block address of 0'th segment */
1009 block_t main_blkaddr; /* start block address of main area */
1010 block_t ssa_blkaddr; /* start block address of SSA area */
1011
1012 unsigned int segment_count; /* total # of segments */
1013 unsigned int main_segments; /* # of segments in main area */
1014 unsigned int reserved_segments; /* # of reserved segments */
1015 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1016 unsigned int ovp_segments; /* # of overprovision segments */
1017
1018 /* a threshold to reclaim prefree segments */
1019 unsigned int rec_prefree_segments;
1020
1021 /* for batched trimming */
1022 unsigned int trim_sections; /* # of sections to trim */
1023
1024 struct list_head sit_entry_set; /* sit entry set list */
1025
1026 unsigned int ipu_policy; /* in-place-update policy */
1027 unsigned int min_ipu_util; /* in-place-update threshold */
1028 unsigned int min_fsync_blocks; /* threshold for fsync */
1029 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1030 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1031 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1032
1033 /* for flush command control */
1034 struct flush_cmd_control *fcc_info;
1035
1036 /* for discard command control */
1037 struct discard_cmd_control *dcc_info;
1038};
1039
1040/*
1041 * For superblock
1042 */
1043/*
1044 * COUNT_TYPE for monitoring
1045 *
1046 * f2fs monitors the number of several block types such as on-writeback,
1047 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1048 */
1049#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1050enum count_type {
1051 F2FS_DIRTY_DENTS,
1052 F2FS_DIRTY_DATA,
1053 F2FS_DIRTY_QDATA,
1054 F2FS_DIRTY_NODES,
1055 F2FS_DIRTY_META,
1056 F2FS_INMEM_PAGES,
1057 F2FS_DIRTY_IMETA,
1058 F2FS_WB_CP_DATA,
1059 F2FS_WB_DATA,
1060 F2FS_RD_DATA,
1061 F2FS_RD_NODE,
1062 F2FS_RD_META,
1063 F2FS_DIO_WRITE,
1064 F2FS_DIO_READ,
1065 NR_COUNT_TYPE,
1066};
1067
1068/*
1069 * The below are the page types of bios used in submit_bio().
1070 * The available types are:
1071 * DATA User data pages. It operates as async mode.
1072 * NODE Node pages. It operates as async mode.
1073 * META FS metadata pages such as SIT, NAT, CP.
1074 * NR_PAGE_TYPE The number of page types.
1075 * META_FLUSH Make sure the previous pages are written
1076 * with waiting the bio's completion
1077 * ... Only can be used with META.
1078 */
1079#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1080enum page_type {
1081 DATA,
1082 NODE,
1083 META,
1084 NR_PAGE_TYPE,
1085 META_FLUSH,
1086 INMEM, /* the below types are used by tracepoints only. */
1087 INMEM_DROP,
1088 INMEM_INVALIDATE,
1089 INMEM_REVOKE,
1090 IPU,
1091 OPU,
1092};
1093
1094enum temp_type {
1095 HOT = 0, /* must be zero for meta bio */
1096 WARM,
1097 COLD,
1098 NR_TEMP_TYPE,
1099};
1100
1101enum need_lock_type {
1102 LOCK_REQ = 0,
1103 LOCK_DONE,
1104 LOCK_RETRY,
1105};
1106
1107enum cp_reason_type {
1108 CP_NO_NEEDED,
1109 CP_NON_REGULAR,
1110 CP_COMPRESSED,
1111 CP_HARDLINK,
1112 CP_SB_NEED_CP,
1113 CP_WRONG_PINO,
1114 CP_NO_SPC_ROLL,
1115 CP_NODE_NEED_CP,
1116 CP_FASTBOOT_MODE,
1117 CP_SPEC_LOG_NUM,
1118 CP_RECOVER_DIR,
1119};
1120
1121enum iostat_type {
1122 /* WRITE IO */
1123 APP_DIRECT_IO, /* app direct write IOs */
1124 APP_BUFFERED_IO, /* app buffered write IOs */
1125 APP_WRITE_IO, /* app write IOs */
1126 APP_MAPPED_IO, /* app mapped IOs */
1127 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1128 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1129 FS_META_IO, /* meta IOs from kworker/reclaimer */
1130 FS_GC_DATA_IO, /* data IOs from forground gc */
1131 FS_GC_NODE_IO, /* node IOs from forground gc */
1132 FS_CP_DATA_IO, /* data IOs from checkpoint */
1133 FS_CP_NODE_IO, /* node IOs from checkpoint */
1134 FS_CP_META_IO, /* meta IOs from checkpoint */
1135
1136 /* READ IO */
1137 APP_DIRECT_READ_IO, /* app direct read IOs */
1138 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1139 APP_READ_IO, /* app read IOs */
1140 APP_MAPPED_READ_IO, /* app mapped read IOs */
1141 FS_DATA_READ_IO, /* data read IOs */
1142 FS_GDATA_READ_IO, /* data read IOs from background gc */
1143 FS_CDATA_READ_IO, /* compressed data read IOs */
1144 FS_NODE_READ_IO, /* node read IOs */
1145 FS_META_READ_IO, /* meta read IOs */
1146
1147 /* other */
1148 FS_DISCARD, /* discard */
1149 NR_IO_TYPE,
1150};
1151
1152struct f2fs_io_info {
1153 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1154 nid_t ino; /* inode number */
1155 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1156 enum temp_type temp; /* contains HOT/WARM/COLD */
1157 int op; /* contains REQ_OP_ */
1158 int op_flags; /* req_flag_bits */
1159 block_t new_blkaddr; /* new block address to be written */
1160 block_t old_blkaddr; /* old block address before Cow */
1161 struct page *page; /* page to be written */
1162 struct page *encrypted_page; /* encrypted page */
1163 struct page *compressed_page; /* compressed page */
1164 struct list_head list; /* serialize IOs */
1165 bool submitted; /* indicate IO submission */
1166 int need_lock; /* indicate we need to lock cp_rwsem */
1167 bool in_list; /* indicate fio is in io_list */
1168 bool is_por; /* indicate IO is from recovery or not */
1169 bool retry; /* need to reallocate block address */
1170 int compr_blocks; /* # of compressed block addresses */
1171 bool encrypted; /* indicate file is encrypted */
1172 enum iostat_type io_type; /* io type */
1173 struct writeback_control *io_wbc; /* writeback control */
1174 struct bio **bio; /* bio for ipu */
1175 sector_t *last_block; /* last block number in bio */
1176 unsigned char version; /* version of the node */
1177};
1178
1179struct bio_entry {
1180 struct bio *bio;
1181 struct list_head list;
1182};
1183
1184#define is_read_io(rw) ((rw) == READ)
1185struct f2fs_bio_info {
1186 struct f2fs_sb_info *sbi; /* f2fs superblock */
1187 struct bio *bio; /* bios to merge */
1188 sector_t last_block_in_bio; /* last block number */
1189 struct f2fs_io_info fio; /* store buffered io info. */
1190 struct rw_semaphore io_rwsem; /* blocking op for bio */
1191 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1192 struct list_head io_list; /* track fios */
1193 struct list_head bio_list; /* bio entry list head */
1194 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */
1195};
1196
1197#define FDEV(i) (sbi->devs[i])
1198#define RDEV(i) (raw_super->devs[i])
1199struct f2fs_dev_info {
1200 struct block_device *bdev;
1201 char path[MAX_PATH_LEN];
1202 unsigned int total_segments;
1203 block_t start_blk;
1204 block_t end_blk;
1205#ifdef CONFIG_BLK_DEV_ZONED
1206 unsigned int nr_blkz; /* Total number of zones */
1207 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1208 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */
1209#endif
1210};
1211
1212enum inode_type {
1213 DIR_INODE, /* for dirty dir inode */
1214 FILE_INODE, /* for dirty regular/symlink inode */
1215 DIRTY_META, /* for all dirtied inode metadata */
1216 ATOMIC_FILE, /* for all atomic files */
1217 NR_INODE_TYPE,
1218};
1219
1220/* for inner inode cache management */
1221struct inode_management {
1222 struct radix_tree_root ino_root; /* ino entry array */
1223 spinlock_t ino_lock; /* for ino entry lock */
1224 struct list_head ino_list; /* inode list head */
1225 unsigned long ino_num; /* number of entries */
1226};
1227
1228/* for GC_AT */
1229struct atgc_management {
1230 bool atgc_enabled; /* ATGC is enabled or not */
1231 struct rb_root_cached root; /* root of victim rb-tree */
1232 struct list_head victim_list; /* linked with all victim entries */
1233 unsigned int victim_count; /* victim count in rb-tree */
1234 unsigned int candidate_ratio; /* candidate ratio */
1235 unsigned int max_candidate_count; /* max candidate count */
1236 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1237 unsigned long long age_threshold; /* age threshold */
1238};
1239
1240/* For s_flag in struct f2fs_sb_info */
1241enum {
1242 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1243 SBI_IS_CLOSE, /* specify unmounting */
1244 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1245 SBI_POR_DOING, /* recovery is doing or not */
1246 SBI_NEED_SB_WRITE, /* need to recover superblock */
1247 SBI_NEED_CP, /* need to checkpoint */
1248 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1249 SBI_IS_RECOVERED, /* recovered orphan/data */
1250 SBI_CP_DISABLED, /* CP was disabled last mount */
1251 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1252 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1253 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1254 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1255 SBI_IS_RESIZEFS, /* resizefs is in process */
1256 SBI_IS_FREEZING, /* freezefs is in process */
1257};
1258
1259enum {
1260 CP_TIME,
1261 REQ_TIME,
1262 DISCARD_TIME,
1263 GC_TIME,
1264 DISABLE_TIME,
1265 UMOUNT_DISCARD_TIMEOUT,
1266 MAX_TIME,
1267};
1268
1269enum {
1270 GC_NORMAL,
1271 GC_IDLE_CB,
1272 GC_IDLE_GREEDY,
1273 GC_IDLE_AT,
1274 GC_URGENT_HIGH,
1275 GC_URGENT_LOW,
1276 MAX_GC_MODE,
1277};
1278
1279enum {
1280 BGGC_MODE_ON, /* background gc is on */
1281 BGGC_MODE_OFF, /* background gc is off */
1282 BGGC_MODE_SYNC, /*
1283 * background gc is on, migrating blocks
1284 * like foreground gc
1285 */
1286};
1287
1288enum {
1289 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1290 FS_MODE_LFS, /* use lfs allocation only */
1291};
1292
1293enum {
1294 WHINT_MODE_OFF, /* not pass down write hints */
1295 WHINT_MODE_USER, /* try to pass down hints given by users */
1296 WHINT_MODE_FS, /* pass down hints with F2FS policy */
1297};
1298
1299enum {
1300 ALLOC_MODE_DEFAULT, /* stay default */
1301 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1302};
1303
1304enum fsync_mode {
1305 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1306 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1307 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1308};
1309
1310enum {
1311 COMPR_MODE_FS, /*
1312 * automatically compress compression
1313 * enabled files
1314 */
1315 COMPR_MODE_USER, /*
1316 * automatical compression is disabled.
1317 * user can control the file compression
1318 * using ioctls
1319 */
1320};
1321
1322enum {
1323 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1324 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1325 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1326};
1327
1328static inline int f2fs_test_bit(unsigned int nr, char *addr);
1329static inline void f2fs_set_bit(unsigned int nr, char *addr);
1330static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1331
1332/*
1333 * Layout of f2fs page.private:
1334 *
1335 * Layout A: lowest bit should be 1
1336 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1337 * bit 0 PAGE_PRIVATE_NOT_POINTER
1338 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE
1339 * bit 2 PAGE_PRIVATE_DUMMY_WRITE
1340 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION
1341 * bit 4 PAGE_PRIVATE_INLINE_INODE
1342 * bit 5 PAGE_PRIVATE_REF_RESOURCE
1343 * bit 6- f2fs private data
1344 *
1345 * Layout B: lowest bit should be 0
1346 * page.private is a wrapped pointer.
1347 */
1348enum {
1349 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1350 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */
1351 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1352 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1353 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1354 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1355 PAGE_PRIVATE_MAX
1356};
1357
1358#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
1359static inline bool page_private_##name(struct page *page) \
1360{ \
1361 return PagePrivate(page) && \
1362 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
1363 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1364}
1365
1366#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
1367static inline void set_page_private_##name(struct page *page) \
1368{ \
1369 if (!PagePrivate(page)) { \
1370 get_page(page); \
1371 SetPagePrivate(page); \
1372 set_page_private(page, 0); \
1373 } \
1374 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
1375 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1376}
1377
1378#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
1379static inline void clear_page_private_##name(struct page *page) \
1380{ \
1381 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
1382 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \
1383 set_page_private(page, 0); \
1384 if (PagePrivate(page)) { \
1385 ClearPagePrivate(page); \
1386 put_page(page); \
1387 }\
1388 } \
1389}
1390
1391PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
1392PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE);
1393PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
1394PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
1395PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);
1396PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
1397
1398PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
1399PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
1400PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
1401PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);
1402PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
1403
1404PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
1405PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
1406PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
1407PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);
1408PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
1409
1410static inline unsigned long get_page_private_data(struct page *page)
1411{
1412 unsigned long data = page_private(page);
1413
1414 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
1415 return 0;
1416 return data >> PAGE_PRIVATE_MAX;
1417}
1418
1419static inline void set_page_private_data(struct page *page, unsigned long data)
1420{
1421 if (!PagePrivate(page)) {
1422 get_page(page);
1423 SetPagePrivate(page);
1424 set_page_private(page, 0);
1425 }
1426 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
1427 page_private(page) |= data << PAGE_PRIVATE_MAX;
1428}
1429
1430static inline void clear_page_private_data(struct page *page)
1431{
1432 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1;
1433 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) {
1434 set_page_private(page, 0);
1435 if (PagePrivate(page)) {
1436 ClearPagePrivate(page);
1437 put_page(page);
1438 }
1439 }
1440}
1441
1442/* For compression */
1443enum compress_algorithm_type {
1444 COMPRESS_LZO,
1445 COMPRESS_LZ4,
1446 COMPRESS_ZSTD,
1447 COMPRESS_LZORLE,
1448 COMPRESS_MAX,
1449};
1450
1451enum compress_flag {
1452 COMPRESS_CHKSUM,
1453 COMPRESS_MAX_FLAG,
1454};
1455
1456#define COMPRESS_WATERMARK 20
1457#define COMPRESS_PERCENT 20
1458
1459#define COMPRESS_DATA_RESERVED_SIZE 4
1460struct compress_data {
1461 __le32 clen; /* compressed data size */
1462 __le32 chksum; /* compressed data chksum */
1463 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1464 u8 cdata[]; /* compressed data */
1465};
1466
1467#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1468
1469#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1470
1471#define COMPRESS_LEVEL_OFFSET 8
1472
1473/* compress context */
1474struct compress_ctx {
1475 struct inode *inode; /* inode the context belong to */
1476 pgoff_t cluster_idx; /* cluster index number */
1477 unsigned int cluster_size; /* page count in cluster */
1478 unsigned int log_cluster_size; /* log of cluster size */
1479 struct page **rpages; /* pages store raw data in cluster */
1480 unsigned int nr_rpages; /* total page number in rpages */
1481 struct page **cpages; /* pages store compressed data in cluster */
1482 unsigned int nr_cpages; /* total page number in cpages */
1483 void *rbuf; /* virtual mapped address on rpages */
1484 struct compress_data *cbuf; /* virtual mapped address on cpages */
1485 size_t rlen; /* valid data length in rbuf */
1486 size_t clen; /* valid data length in cbuf */
1487 void *private; /* payload buffer for specified compression algorithm */
1488 void *private2; /* extra payload buffer */
1489};
1490
1491/* compress context for write IO path */
1492struct compress_io_ctx {
1493 u32 magic; /* magic number to indicate page is compressed */
1494 struct inode *inode; /* inode the context belong to */
1495 struct page **rpages; /* pages store raw data in cluster */
1496 unsigned int nr_rpages; /* total page number in rpages */
1497 atomic_t pending_pages; /* in-flight compressed page count */
1498};
1499
1500/* Context for decompressing one cluster on the read IO path */
1501struct decompress_io_ctx {
1502 u32 magic; /* magic number to indicate page is compressed */
1503 struct inode *inode; /* inode the context belong to */
1504 pgoff_t cluster_idx; /* cluster index number */
1505 unsigned int cluster_size; /* page count in cluster */
1506 unsigned int log_cluster_size; /* log of cluster size */
1507 struct page **rpages; /* pages store raw data in cluster */
1508 unsigned int nr_rpages; /* total page number in rpages */
1509 struct page **cpages; /* pages store compressed data in cluster */
1510 unsigned int nr_cpages; /* total page number in cpages */
1511 struct page **tpages; /* temp pages to pad holes in cluster */
1512 void *rbuf; /* virtual mapped address on rpages */
1513 struct compress_data *cbuf; /* virtual mapped address on cpages */
1514 size_t rlen; /* valid data length in rbuf */
1515 size_t clen; /* valid data length in cbuf */
1516
1517 /*
1518 * The number of compressed pages remaining to be read in this cluster.
1519 * This is initially nr_cpages. It is decremented by 1 each time a page
1520 * has been read (or failed to be read). When it reaches 0, the cluster
1521 * is decompressed (or an error is reported).
1522 *
1523 * If an error occurs before all the pages have been submitted for I/O,
1524 * then this will never reach 0. In this case the I/O submitter is
1525 * responsible for calling f2fs_decompress_end_io() instead.
1526 */
1527 atomic_t remaining_pages;
1528
1529 /*
1530 * Number of references to this decompress_io_ctx.
1531 *
1532 * One reference is held for I/O completion. This reference is dropped
1533 * after the pagecache pages are updated and unlocked -- either after
1534 * decompression (and verity if enabled), or after an error.
1535 *
1536 * In addition, each compressed page holds a reference while it is in a
1537 * bio. These references are necessary prevent compressed pages from
1538 * being freed while they are still in a bio.
1539 */
1540 refcount_t refcnt;
1541
1542 bool failed; /* IO error occurred before decompression? */
1543 bool need_verity; /* need fs-verity verification after decompression? */
1544 void *private; /* payload buffer for specified decompression algorithm */
1545 void *private2; /* extra payload buffer */
1546 struct work_struct verity_work; /* work to verify the decompressed pages */
1547};
1548
1549#define NULL_CLUSTER ((unsigned int)(~0))
1550#define MIN_COMPRESS_LOG_SIZE 2
1551#define MAX_COMPRESS_LOG_SIZE 8
1552#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1553
1554struct f2fs_sb_info {
1555 struct super_block *sb; /* pointer to VFS super block */
1556 struct proc_dir_entry *s_proc; /* proc entry */
1557 struct f2fs_super_block *raw_super; /* raw super block pointer */
1558 struct rw_semaphore sb_lock; /* lock for raw super block */
1559 int valid_super_block; /* valid super block no */
1560 unsigned long s_flag; /* flags for sbi */
1561 struct mutex writepages; /* mutex for writepages() */
1562
1563#ifdef CONFIG_BLK_DEV_ZONED
1564 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1565 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */
1566#endif
1567
1568 /* for node-related operations */
1569 struct f2fs_nm_info *nm_info; /* node manager */
1570 struct inode *node_inode; /* cache node blocks */
1571
1572 /* for segment-related operations */
1573 struct f2fs_sm_info *sm_info; /* segment manager */
1574
1575 /* for bio operations */
1576 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1577 /* keep migration IO order for LFS mode */
1578 struct rw_semaphore io_order_lock;
1579 mempool_t *write_io_dummy; /* Dummy pages */
1580
1581 /* for checkpoint */
1582 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1583 int cur_cp_pack; /* remain current cp pack */
1584 spinlock_t cp_lock; /* for flag in ckpt */
1585 struct inode *meta_inode; /* cache meta blocks */
1586 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */
1587 struct rw_semaphore cp_rwsem; /* blocking FS operations */
1588 struct rw_semaphore node_write; /* locking node writes */
1589 struct rw_semaphore node_change; /* locking node change */
1590 wait_queue_head_t cp_wait;
1591 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1592 long interval_time[MAX_TIME]; /* to store thresholds */
1593 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1594
1595 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1596
1597 spinlock_t fsync_node_lock; /* for node entry lock */
1598 struct list_head fsync_node_list; /* node list head */
1599 unsigned int fsync_seg_id; /* sequence id */
1600 unsigned int fsync_node_num; /* number of node entries */
1601
1602 /* for orphan inode, use 0'th array */
1603 unsigned int max_orphans; /* max orphan inodes */
1604
1605 /* for inode management */
1606 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1607 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1608 struct mutex flush_lock; /* for flush exclusion */
1609
1610 /* for extent tree cache */
1611 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
1612 struct mutex extent_tree_lock; /* locking extent radix tree */
1613 struct list_head extent_list; /* lru list for shrinker */
1614 spinlock_t extent_lock; /* locking extent lru list */
1615 atomic_t total_ext_tree; /* extent tree count */
1616 struct list_head zombie_list; /* extent zombie tree list */
1617 atomic_t total_zombie_tree; /* extent zombie tree count */
1618 atomic_t total_ext_node; /* extent info count */
1619
1620 /* basic filesystem units */
1621 unsigned int log_sectors_per_block; /* log2 sectors per block */
1622 unsigned int log_blocksize; /* log2 block size */
1623 unsigned int blocksize; /* block size */
1624 unsigned int root_ino_num; /* root inode number*/
1625 unsigned int node_ino_num; /* node inode number*/
1626 unsigned int meta_ino_num; /* meta inode number*/
1627 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1628 unsigned int blocks_per_seg; /* blocks per segment */
1629 unsigned int segs_per_sec; /* segments per section */
1630 unsigned int secs_per_zone; /* sections per zone */
1631 unsigned int total_sections; /* total section count */
1632 unsigned int total_node_count; /* total node block count */
1633 unsigned int total_valid_node_count; /* valid node block count */
1634 int dir_level; /* directory level */
1635 int readdir_ra; /* readahead inode in readdir */
1636 u64 max_io_bytes; /* max io bytes to merge IOs */
1637
1638 block_t user_block_count; /* # of user blocks */
1639 block_t total_valid_block_count; /* # of valid blocks */
1640 block_t discard_blks; /* discard command candidats */
1641 block_t last_valid_block_count; /* for recovery */
1642 block_t reserved_blocks; /* configurable reserved blocks */
1643 block_t current_reserved_blocks; /* current reserved blocks */
1644
1645 /* Additional tracking for no checkpoint mode */
1646 block_t unusable_block_count; /* # of blocks saved by last cp */
1647
1648 unsigned int nquota_files; /* # of quota sysfile */
1649 struct rw_semaphore quota_sem; /* blocking cp for flags */
1650
1651 /* # of pages, see count_type */
1652 atomic_t nr_pages[NR_COUNT_TYPE];
1653 /* # of allocated blocks */
1654 struct percpu_counter alloc_valid_block_count;
1655
1656 /* writeback control */
1657 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1658
1659 /* valid inode count */
1660 struct percpu_counter total_valid_inode_count;
1661
1662 struct f2fs_mount_info mount_opt; /* mount options */
1663
1664 /* for cleaning operations */
1665 struct rw_semaphore gc_lock; /*
1666 * semaphore for GC, avoid
1667 * race between GC and GC or CP
1668 */
1669 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1670 struct atgc_management am; /* atgc management */
1671 unsigned int cur_victim_sec; /* current victim section num */
1672 unsigned int gc_mode; /* current GC state */
1673 unsigned int next_victim_seg[2]; /* next segment in victim section */
1674
1675 /* for skip statistic */
1676 unsigned int atomic_files; /* # of opened atomic file */
1677 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */
1678 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1679
1680 /* threshold for gc trials on pinned files */
1681 u64 gc_pin_file_threshold;
1682 struct rw_semaphore pin_sem;
1683
1684 /* maximum # of trials to find a victim segment for SSR and GC */
1685 unsigned int max_victim_search;
1686 /* migration granularity of garbage collection, unit: segment */
1687 unsigned int migration_granularity;
1688
1689 /*
1690 * for stat information.
1691 * one is for the LFS mode, and the other is for the SSR mode.
1692 */
1693#ifdef CONFIG_F2FS_STAT_FS
1694 struct f2fs_stat_info *stat_info; /* FS status information */
1695 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1696 unsigned int segment_count[2]; /* # of allocated segments */
1697 unsigned int block_count[2]; /* # of allocated blocks */
1698 atomic_t inplace_count; /* # of inplace update */
1699 atomic64_t total_hit_ext; /* # of lookup extent cache */
1700 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */
1701 atomic64_t read_hit_largest; /* # of hit largest extent node */
1702 atomic64_t read_hit_cached; /* # of hit cached extent node */
1703 atomic_t inline_xattr; /* # of inline_xattr inodes */
1704 atomic_t inline_inode; /* # of inline_data inodes */
1705 atomic_t inline_dir; /* # of inline_dentry inodes */
1706 atomic_t compr_inode; /* # of compressed inodes */
1707 atomic64_t compr_blocks; /* # of compressed blocks */
1708 atomic_t vw_cnt; /* # of volatile writes */
1709 atomic_t max_aw_cnt; /* max # of atomic writes */
1710 atomic_t max_vw_cnt; /* max # of volatile writes */
1711 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1712 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1713 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1714#endif
1715 spinlock_t stat_lock; /* lock for stat operations */
1716
1717 /* to attach REQ_META|REQ_FUA flags */
1718 unsigned int data_io_flag;
1719 unsigned int node_io_flag;
1720
1721 /* For sysfs suppport */
1722 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1723 struct completion s_kobj_unregister;
1724
1725 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1726 struct completion s_stat_kobj_unregister;
1727
1728 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1729 struct completion s_feature_list_kobj_unregister;
1730
1731 /* For shrinker support */
1732 struct list_head s_list;
1733 int s_ndevs; /* number of devices */
1734 struct f2fs_dev_info *devs; /* for device list */
1735 unsigned int dirty_device; /* for checkpoint data flush */
1736 spinlock_t dev_lock; /* protect dirty_device */
1737 struct mutex umount_mutex;
1738 unsigned int shrinker_run_no;
1739
1740 /* For write statistics */
1741 u64 sectors_written_start;
1742 u64 kbytes_written;
1743
1744 /* Reference to checksum algorithm driver via cryptoapi */
1745 struct crypto_shash *s_chksum_driver;
1746
1747 /* Precomputed FS UUID checksum for seeding other checksums */
1748 __u32 s_chksum_seed;
1749
1750 struct workqueue_struct *post_read_wq; /* post read workqueue */
1751
1752 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1753 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1754
1755 /* For reclaimed segs statistics per each GC mode */
1756 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1757 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1758
1759 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1760
1761#ifdef CONFIG_F2FS_FS_COMPRESSION
1762 struct kmem_cache *page_array_slab; /* page array entry */
1763 unsigned int page_array_slab_size; /* default page array slab size */
1764
1765 /* For runtime compression statistics */
1766 u64 compr_written_block;
1767 u64 compr_saved_block;
1768 u32 compr_new_inode;
1769
1770 /* For compressed block cache */
1771 struct inode *compress_inode; /* cache compressed blocks */
1772 unsigned int compress_percent; /* cache page percentage */
1773 unsigned int compress_watermark; /* cache page watermark */
1774 atomic_t compress_page_hit; /* cache hit count */
1775#endif
1776
1777#ifdef CONFIG_F2FS_IOSTAT
1778 /* For app/fs IO statistics */
1779 spinlock_t iostat_lock;
1780 unsigned long long rw_iostat[NR_IO_TYPE];
1781 unsigned long long prev_rw_iostat[NR_IO_TYPE];
1782 bool iostat_enable;
1783 unsigned long iostat_next_period;
1784 unsigned int iostat_period_ms;
1785
1786 /* For io latency related statistics info in one iostat period */
1787 spinlock_t iostat_lat_lock;
1788 struct iostat_lat_info *iostat_io_lat;
1789#endif
1790};
1791
1792struct f2fs_private_dio {
1793 struct inode *inode;
1794 void *orig_private;
1795 bio_end_io_t *orig_end_io;
1796 bool write;
1797};
1798
1799#ifdef CONFIG_F2FS_FAULT_INJECTION
1800#define f2fs_show_injection_info(sbi, type) \
1801 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \
1802 KERN_INFO, sbi->sb->s_id, \
1803 f2fs_fault_name[type], \
1804 __func__, __builtin_return_address(0))
1805static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1806{
1807 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1808
1809 if (!ffi->inject_rate)
1810 return false;
1811
1812 if (!IS_FAULT_SET(ffi, type))
1813 return false;
1814
1815 atomic_inc(&ffi->inject_ops);
1816 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1817 atomic_set(&ffi->inject_ops, 0);
1818 return true;
1819 }
1820 return false;
1821}
1822#else
1823#define f2fs_show_injection_info(sbi, type) do { } while (0)
1824static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1825{
1826 return false;
1827}
1828#endif
1829
1830/*
1831 * Test if the mounted volume is a multi-device volume.
1832 * - For a single regular disk volume, sbi->s_ndevs is 0.
1833 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1834 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1835 */
1836static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1837{
1838 return sbi->s_ndevs > 1;
1839}
1840
1841static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1842{
1843 unsigned long now = jiffies;
1844
1845 sbi->last_time[type] = now;
1846
1847 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1848 if (type == REQ_TIME) {
1849 sbi->last_time[DISCARD_TIME] = now;
1850 sbi->last_time[GC_TIME] = now;
1851 }
1852}
1853
1854static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1855{
1856 unsigned long interval = sbi->interval_time[type] * HZ;
1857
1858 return time_after(jiffies, sbi->last_time[type] + interval);
1859}
1860
1861static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1862 int type)
1863{
1864 unsigned long interval = sbi->interval_time[type] * HZ;
1865 unsigned int wait_ms = 0;
1866 long delta;
1867
1868 delta = (sbi->last_time[type] + interval) - jiffies;
1869 if (delta > 0)
1870 wait_ms = jiffies_to_msecs(delta);
1871
1872 return wait_ms;
1873}
1874
1875/*
1876 * Inline functions
1877 */
1878static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1879 const void *address, unsigned int length)
1880{
1881 struct {
1882 struct shash_desc shash;
1883 char ctx[4];
1884 } desc;
1885 int err;
1886
1887 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1888
1889 desc.shash.tfm = sbi->s_chksum_driver;
1890 *(u32 *)desc.ctx = crc;
1891
1892 err = crypto_shash_update(&desc.shash, address, length);
1893 BUG_ON(err);
1894
1895 return *(u32 *)desc.ctx;
1896}
1897
1898static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1899 unsigned int length)
1900{
1901 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1902}
1903
1904static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1905 void *buf, size_t buf_size)
1906{
1907 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1908}
1909
1910static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1911 const void *address, unsigned int length)
1912{
1913 return __f2fs_crc32(sbi, crc, address, length);
1914}
1915
1916static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1917{
1918 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1919}
1920
1921static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1922{
1923 return sb->s_fs_info;
1924}
1925
1926static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1927{
1928 return F2FS_SB(inode->i_sb);
1929}
1930
1931static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1932{
1933 return F2FS_I_SB(mapping->host);
1934}
1935
1936static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1937{
1938 return F2FS_M_SB(page_file_mapping(page));
1939}
1940
1941static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1942{
1943 return (struct f2fs_super_block *)(sbi->raw_super);
1944}
1945
1946static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1947{
1948 return (struct f2fs_checkpoint *)(sbi->ckpt);
1949}
1950
1951static inline struct f2fs_node *F2FS_NODE(struct page *page)
1952{
1953 return (struct f2fs_node *)page_address(page);
1954}
1955
1956static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1957{
1958 return &((struct f2fs_node *)page_address(page))->i;
1959}
1960
1961static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1962{
1963 return (struct f2fs_nm_info *)(sbi->nm_info);
1964}
1965
1966static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1967{
1968 return (struct f2fs_sm_info *)(sbi->sm_info);
1969}
1970
1971static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1972{
1973 return (struct sit_info *)(SM_I(sbi)->sit_info);
1974}
1975
1976static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1977{
1978 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1979}
1980
1981static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1982{
1983 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1984}
1985
1986static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1987{
1988 return sbi->meta_inode->i_mapping;
1989}
1990
1991static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
1992{
1993 return sbi->node_inode->i_mapping;
1994}
1995
1996static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
1997{
1998 return test_bit(type, &sbi->s_flag);
1999}
2000
2001static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2002{
2003 set_bit(type, &sbi->s_flag);
2004}
2005
2006static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2007{
2008 clear_bit(type, &sbi->s_flag);
2009}
2010
2011static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
2012{
2013 return le64_to_cpu(cp->checkpoint_ver);
2014}
2015
2016static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
2017{
2018 if (type < F2FS_MAX_QUOTAS)
2019 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
2020 return 0;
2021}
2022
2023static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2024{
2025 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2026 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2027}
2028
2029static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2030{
2031 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2032
2033 return ckpt_flags & f;
2034}
2035
2036static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2037{
2038 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2039}
2040
2041static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2042{
2043 unsigned int ckpt_flags;
2044
2045 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2046 ckpt_flags |= f;
2047 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2048}
2049
2050static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2051{
2052 unsigned long flags;
2053
2054 spin_lock_irqsave(&sbi->cp_lock, flags);
2055 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2056 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2057}
2058
2059static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2060{
2061 unsigned int ckpt_flags;
2062
2063 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2064 ckpt_flags &= (~f);
2065 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2066}
2067
2068static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2069{
2070 unsigned long flags;
2071
2072 spin_lock_irqsave(&sbi->cp_lock, flags);
2073 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2074 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2075}
2076
2077static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2078{
2079 down_read(&sbi->cp_rwsem);
2080}
2081
2082static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2083{
2084 return down_read_trylock(&sbi->cp_rwsem);
2085}
2086
2087static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2088{
2089 up_read(&sbi->cp_rwsem);
2090}
2091
2092static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2093{
2094 down_write(&sbi->cp_rwsem);
2095}
2096
2097static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2098{
2099 up_write(&sbi->cp_rwsem);
2100}
2101
2102static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2103{
2104 int reason = CP_SYNC;
2105
2106 if (test_opt(sbi, FASTBOOT))
2107 reason = CP_FASTBOOT;
2108 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2109 reason = CP_UMOUNT;
2110 return reason;
2111}
2112
2113static inline bool __remain_node_summaries(int reason)
2114{
2115 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2116}
2117
2118static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2119{
2120 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2121 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2122}
2123
2124/*
2125 * Check whether the inode has blocks or not
2126 */
2127static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2128{
2129 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2130
2131 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2132}
2133
2134static inline bool f2fs_has_xattr_block(unsigned int ofs)
2135{
2136 return ofs == XATTR_NODE_OFFSET;
2137}
2138
2139static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2140 struct inode *inode, bool cap)
2141{
2142 if (!inode)
2143 return true;
2144 if (!test_opt(sbi, RESERVE_ROOT))
2145 return false;
2146 if (IS_NOQUOTA(inode))
2147 return true;
2148 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2149 return true;
2150 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2151 in_group_p(F2FS_OPTION(sbi).s_resgid))
2152 return true;
2153 if (cap && capable(CAP_SYS_RESOURCE))
2154 return true;
2155 return false;
2156}
2157
2158static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2159static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2160 struct inode *inode, blkcnt_t *count)
2161{
2162 blkcnt_t diff = 0, release = 0;
2163 block_t avail_user_block_count;
2164 int ret;
2165
2166 ret = dquot_reserve_block(inode, *count);
2167 if (ret)
2168 return ret;
2169
2170 if (time_to_inject(sbi, FAULT_BLOCK)) {
2171 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2172 release = *count;
2173 goto release_quota;
2174 }
2175
2176 /*
2177 * let's increase this in prior to actual block count change in order
2178 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2179 */
2180 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2181
2182 spin_lock(&sbi->stat_lock);
2183 sbi->total_valid_block_count += (block_t)(*count);
2184 avail_user_block_count = sbi->user_block_count -
2185 sbi->current_reserved_blocks;
2186
2187 if (!__allow_reserved_blocks(sbi, inode, true))
2188 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2189
2190 if (F2FS_IO_ALIGNED(sbi))
2191 avail_user_block_count -= sbi->blocks_per_seg *
2192 SM_I(sbi)->additional_reserved_segments;
2193
2194 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2195 if (avail_user_block_count > sbi->unusable_block_count)
2196 avail_user_block_count -= sbi->unusable_block_count;
2197 else
2198 avail_user_block_count = 0;
2199 }
2200 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2201 diff = sbi->total_valid_block_count - avail_user_block_count;
2202 if (diff > *count)
2203 diff = *count;
2204 *count -= diff;
2205 release = diff;
2206 sbi->total_valid_block_count -= diff;
2207 if (!*count) {
2208 spin_unlock(&sbi->stat_lock);
2209 goto enospc;
2210 }
2211 }
2212 spin_unlock(&sbi->stat_lock);
2213
2214 if (unlikely(release)) {
2215 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2216 dquot_release_reservation_block(inode, release);
2217 }
2218 f2fs_i_blocks_write(inode, *count, true, true);
2219 return 0;
2220
2221enospc:
2222 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2223release_quota:
2224 dquot_release_reservation_block(inode, release);
2225 return -ENOSPC;
2226}
2227
2228__printf(2, 3)
2229void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2230
2231#define f2fs_err(sbi, fmt, ...) \
2232 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2233#define f2fs_warn(sbi, fmt, ...) \
2234 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2235#define f2fs_notice(sbi, fmt, ...) \
2236 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2237#define f2fs_info(sbi, fmt, ...) \
2238 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2239#define f2fs_debug(sbi, fmt, ...) \
2240 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2241
2242static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2243 struct inode *inode,
2244 block_t count)
2245{
2246 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2247
2248 spin_lock(&sbi->stat_lock);
2249 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2250 sbi->total_valid_block_count -= (block_t)count;
2251 if (sbi->reserved_blocks &&
2252 sbi->current_reserved_blocks < sbi->reserved_blocks)
2253 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2254 sbi->current_reserved_blocks + count);
2255 spin_unlock(&sbi->stat_lock);
2256 if (unlikely(inode->i_blocks < sectors)) {
2257 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2258 inode->i_ino,
2259 (unsigned long long)inode->i_blocks,
2260 (unsigned long long)sectors);
2261 set_sbi_flag(sbi, SBI_NEED_FSCK);
2262 return;
2263 }
2264 f2fs_i_blocks_write(inode, count, false, true);
2265}
2266
2267static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2268{
2269 atomic_inc(&sbi->nr_pages[count_type]);
2270
2271 if (count_type == F2FS_DIRTY_DENTS ||
2272 count_type == F2FS_DIRTY_NODES ||
2273 count_type == F2FS_DIRTY_META ||
2274 count_type == F2FS_DIRTY_QDATA ||
2275 count_type == F2FS_DIRTY_IMETA)
2276 set_sbi_flag(sbi, SBI_IS_DIRTY);
2277}
2278
2279static inline void inode_inc_dirty_pages(struct inode *inode)
2280{
2281 atomic_inc(&F2FS_I(inode)->dirty_pages);
2282 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2283 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2284 if (IS_NOQUOTA(inode))
2285 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2286}
2287
2288static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2289{
2290 atomic_dec(&sbi->nr_pages[count_type]);
2291}
2292
2293static inline void inode_dec_dirty_pages(struct inode *inode)
2294{
2295 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2296 !S_ISLNK(inode->i_mode))
2297 return;
2298
2299 atomic_dec(&F2FS_I(inode)->dirty_pages);
2300 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2301 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2302 if (IS_NOQUOTA(inode))
2303 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2304}
2305
2306static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2307{
2308 return atomic_read(&sbi->nr_pages[count_type]);
2309}
2310
2311static inline int get_dirty_pages(struct inode *inode)
2312{
2313 return atomic_read(&F2FS_I(inode)->dirty_pages);
2314}
2315
2316static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2317{
2318 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2319 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >>
2320 sbi->log_blocks_per_seg;
2321
2322 return segs / sbi->segs_per_sec;
2323}
2324
2325static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2326{
2327 return sbi->total_valid_block_count;
2328}
2329
2330static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2331{
2332 return sbi->discard_blks;
2333}
2334
2335static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2336{
2337 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2338
2339 /* return NAT or SIT bitmap */
2340 if (flag == NAT_BITMAP)
2341 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2342 else if (flag == SIT_BITMAP)
2343 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2344
2345 return 0;
2346}
2347
2348static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2349{
2350 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2351}
2352
2353static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2354{
2355 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2356 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2357 int offset;
2358
2359 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2360 offset = (flag == SIT_BITMAP) ?
2361 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2362 /*
2363 * if large_nat_bitmap feature is enabled, leave checksum
2364 * protection for all nat/sit bitmaps.
2365 */
2366 return tmp_ptr + offset + sizeof(__le32);
2367 }
2368
2369 if (__cp_payload(sbi) > 0) {
2370 if (flag == NAT_BITMAP)
2371 return &ckpt->sit_nat_version_bitmap;
2372 else
2373 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2374 } else {
2375 offset = (flag == NAT_BITMAP) ?
2376 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2377 return tmp_ptr + offset;
2378 }
2379}
2380
2381static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2382{
2383 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2384
2385 if (sbi->cur_cp_pack == 2)
2386 start_addr += sbi->blocks_per_seg;
2387 return start_addr;
2388}
2389
2390static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2391{
2392 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2393
2394 if (sbi->cur_cp_pack == 1)
2395 start_addr += sbi->blocks_per_seg;
2396 return start_addr;
2397}
2398
2399static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2400{
2401 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2402}
2403
2404static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2405{
2406 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2407}
2408
2409static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2410 struct inode *inode, bool is_inode)
2411{
2412 block_t valid_block_count;
2413 unsigned int valid_node_count, user_block_count;
2414 int err;
2415
2416 if (is_inode) {
2417 if (inode) {
2418 err = dquot_alloc_inode(inode);
2419 if (err)
2420 return err;
2421 }
2422 } else {
2423 err = dquot_reserve_block(inode, 1);
2424 if (err)
2425 return err;
2426 }
2427
2428 if (time_to_inject(sbi, FAULT_BLOCK)) {
2429 f2fs_show_injection_info(sbi, FAULT_BLOCK);
2430 goto enospc;
2431 }
2432
2433 spin_lock(&sbi->stat_lock);
2434
2435 valid_block_count = sbi->total_valid_block_count +
2436 sbi->current_reserved_blocks + 1;
2437
2438 if (!__allow_reserved_blocks(sbi, inode, false))
2439 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2440
2441 if (F2FS_IO_ALIGNED(sbi))
2442 valid_block_count += sbi->blocks_per_seg *
2443 SM_I(sbi)->additional_reserved_segments;
2444
2445 user_block_count = sbi->user_block_count;
2446 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2447 user_block_count -= sbi->unusable_block_count;
2448
2449 if (unlikely(valid_block_count > user_block_count)) {
2450 spin_unlock(&sbi->stat_lock);
2451 goto enospc;
2452 }
2453
2454 valid_node_count = sbi->total_valid_node_count + 1;
2455 if (unlikely(valid_node_count > sbi->total_node_count)) {
2456 spin_unlock(&sbi->stat_lock);
2457 goto enospc;
2458 }
2459
2460 sbi->total_valid_node_count++;
2461 sbi->total_valid_block_count++;
2462 spin_unlock(&sbi->stat_lock);
2463
2464 if (inode) {
2465 if (is_inode)
2466 f2fs_mark_inode_dirty_sync(inode, true);
2467 else
2468 f2fs_i_blocks_write(inode, 1, true, true);
2469 }
2470
2471 percpu_counter_inc(&sbi->alloc_valid_block_count);
2472 return 0;
2473
2474enospc:
2475 if (is_inode) {
2476 if (inode)
2477 dquot_free_inode(inode);
2478 } else {
2479 dquot_release_reservation_block(inode, 1);
2480 }
2481 return -ENOSPC;
2482}
2483
2484static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2485 struct inode *inode, bool is_inode)
2486{
2487 spin_lock(&sbi->stat_lock);
2488
2489 f2fs_bug_on(sbi, !sbi->total_valid_block_count);
2490 f2fs_bug_on(sbi, !sbi->total_valid_node_count);
2491
2492 sbi->total_valid_node_count--;
2493 sbi->total_valid_block_count--;
2494 if (sbi->reserved_blocks &&
2495 sbi->current_reserved_blocks < sbi->reserved_blocks)
2496 sbi->current_reserved_blocks++;
2497
2498 spin_unlock(&sbi->stat_lock);
2499
2500 if (is_inode) {
2501 dquot_free_inode(inode);
2502 } else {
2503 if (unlikely(inode->i_blocks == 0)) {
2504 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2505 inode->i_ino,
2506 (unsigned long long)inode->i_blocks);
2507 set_sbi_flag(sbi, SBI_NEED_FSCK);
2508 return;
2509 }
2510 f2fs_i_blocks_write(inode, 1, false, true);
2511 }
2512}
2513
2514static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2515{
2516 return sbi->total_valid_node_count;
2517}
2518
2519static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2520{
2521 percpu_counter_inc(&sbi->total_valid_inode_count);
2522}
2523
2524static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2525{
2526 percpu_counter_dec(&sbi->total_valid_inode_count);
2527}
2528
2529static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2530{
2531 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2532}
2533
2534static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2535 pgoff_t index, bool for_write)
2536{
2537 struct page *page;
2538
2539 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2540 if (!for_write)
2541 page = find_get_page_flags(mapping, index,
2542 FGP_LOCK | FGP_ACCESSED);
2543 else
2544 page = find_lock_page(mapping, index);
2545 if (page)
2546 return page;
2547
2548 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) {
2549 f2fs_show_injection_info(F2FS_M_SB(mapping),
2550 FAULT_PAGE_ALLOC);
2551 return NULL;
2552 }
2553 }
2554
2555 if (!for_write)
2556 return grab_cache_page(mapping, index);
2557 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS);
2558}
2559
2560static inline struct page *f2fs_pagecache_get_page(
2561 struct address_space *mapping, pgoff_t index,
2562 int fgp_flags, gfp_t gfp_mask)
2563{
2564 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) {
2565 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET);
2566 return NULL;
2567 }
2568
2569 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2570}
2571
2572static inline void f2fs_copy_page(struct page *src, struct page *dst)
2573{
2574 char *src_kaddr = kmap(src);
2575 char *dst_kaddr = kmap(dst);
2576
2577 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE);
2578 kunmap(dst);
2579 kunmap(src);
2580}
2581
2582static inline void f2fs_put_page(struct page *page, int unlock)
2583{
2584 if (!page)
2585 return;
2586
2587 if (unlock) {
2588 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2589 unlock_page(page);
2590 }
2591 put_page(page);
2592}
2593
2594static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2595{
2596 if (dn->node_page)
2597 f2fs_put_page(dn->node_page, 1);
2598 if (dn->inode_page && dn->node_page != dn->inode_page)
2599 f2fs_put_page(dn->inode_page, 0);
2600 dn->node_page = NULL;
2601 dn->inode_page = NULL;
2602}
2603
2604static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2605 size_t size)
2606{
2607 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2608}
2609
2610static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
2611 gfp_t flags)
2612{
2613 void *entry;
2614
2615 entry = kmem_cache_alloc(cachep, flags);
2616 if (!entry)
2617 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2618 return entry;
2619}
2620
2621static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2622 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
2623{
2624 if (nofail)
2625 return f2fs_kmem_cache_alloc_nofail(cachep, flags);
2626
2627 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) {
2628 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC);
2629 return NULL;
2630 }
2631
2632 return kmem_cache_alloc(cachep, flags);
2633}
2634
2635static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2636{
2637 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2638 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2639 get_pages(sbi, F2FS_WB_CP_DATA) ||
2640 get_pages(sbi, F2FS_DIO_READ) ||
2641 get_pages(sbi, F2FS_DIO_WRITE))
2642 return true;
2643
2644 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2645 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2646 return true;
2647
2648 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2649 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2650 return true;
2651 return false;
2652}
2653
2654static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2655{
2656 if (sbi->gc_mode == GC_URGENT_HIGH)
2657 return true;
2658
2659 if (is_inflight_io(sbi, type))
2660 return false;
2661
2662 if (sbi->gc_mode == GC_URGENT_LOW &&
2663 (type == DISCARD_TIME || type == GC_TIME))
2664 return true;
2665
2666 return f2fs_time_over(sbi, type);
2667}
2668
2669static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2670 unsigned long index, void *item)
2671{
2672 while (radix_tree_insert(root, index, item))
2673 cond_resched();
2674}
2675
2676#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2677
2678static inline bool IS_INODE(struct page *page)
2679{
2680 struct f2fs_node *p = F2FS_NODE(page);
2681
2682 return RAW_IS_INODE(p);
2683}
2684
2685static inline int offset_in_addr(struct f2fs_inode *i)
2686{
2687 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2688 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2689}
2690
2691static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2692{
2693 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2694}
2695
2696static inline int f2fs_has_extra_attr(struct inode *inode);
2697static inline block_t data_blkaddr(struct inode *inode,
2698 struct page *node_page, unsigned int offset)
2699{
2700 struct f2fs_node *raw_node;
2701 __le32 *addr_array;
2702 int base = 0;
2703 bool is_inode = IS_INODE(node_page);
2704
2705 raw_node = F2FS_NODE(node_page);
2706
2707 if (is_inode) {
2708 if (!inode)
2709 /* from GC path only */
2710 base = offset_in_addr(&raw_node->i);
2711 else if (f2fs_has_extra_attr(inode))
2712 base = get_extra_isize(inode);
2713 }
2714
2715 addr_array = blkaddr_in_node(raw_node);
2716 return le32_to_cpu(addr_array[base + offset]);
2717}
2718
2719static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2720{
2721 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2722}
2723
2724static inline int f2fs_test_bit(unsigned int nr, char *addr)
2725{
2726 int mask;
2727
2728 addr += (nr >> 3);
2729 mask = 1 << (7 - (nr & 0x07));
2730 return mask & *addr;
2731}
2732
2733static inline void f2fs_set_bit(unsigned int nr, char *addr)
2734{
2735 int mask;
2736
2737 addr += (nr >> 3);
2738 mask = 1 << (7 - (nr & 0x07));
2739 *addr |= mask;
2740}
2741
2742static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2743{
2744 int mask;
2745
2746 addr += (nr >> 3);
2747 mask = 1 << (7 - (nr & 0x07));
2748 *addr &= ~mask;
2749}
2750
2751static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2752{
2753 int mask;
2754 int ret;
2755
2756 addr += (nr >> 3);
2757 mask = 1 << (7 - (nr & 0x07));
2758 ret = mask & *addr;
2759 *addr |= mask;
2760 return ret;
2761}
2762
2763static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2764{
2765 int mask;
2766 int ret;
2767
2768 addr += (nr >> 3);
2769 mask = 1 << (7 - (nr & 0x07));
2770 ret = mask & *addr;
2771 *addr &= ~mask;
2772 return ret;
2773}
2774
2775static inline void f2fs_change_bit(unsigned int nr, char *addr)
2776{
2777 int mask;
2778
2779 addr += (nr >> 3);
2780 mask = 1 << (7 - (nr & 0x07));
2781 *addr ^= mask;
2782}
2783
2784/*
2785 * On-disk inode flags (f2fs_inode::i_flags)
2786 */
2787#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2788#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2789#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2790#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2791#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2792#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2793#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2794#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2795#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2796#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2797#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2798
2799/* Flags that should be inherited by new inodes from their parent. */
2800#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2801 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2802 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL)
2803
2804/* Flags that are appropriate for regular files (all but dir-specific ones). */
2805#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2806 F2FS_CASEFOLD_FL))
2807
2808/* Flags that are appropriate for non-directories/regular files. */
2809#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2810
2811static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2812{
2813 if (S_ISDIR(mode))
2814 return flags;
2815 else if (S_ISREG(mode))
2816 return flags & F2FS_REG_FLMASK;
2817 else
2818 return flags & F2FS_OTHER_FLMASK;
2819}
2820
2821static inline void __mark_inode_dirty_flag(struct inode *inode,
2822 int flag, bool set)
2823{
2824 switch (flag) {
2825 case FI_INLINE_XATTR:
2826 case FI_INLINE_DATA:
2827 case FI_INLINE_DENTRY:
2828 case FI_NEW_INODE:
2829 if (set)
2830 return;
2831 fallthrough;
2832 case FI_DATA_EXIST:
2833 case FI_INLINE_DOTS:
2834 case FI_PIN_FILE:
2835 case FI_COMPRESS_RELEASED:
2836 f2fs_mark_inode_dirty_sync(inode, true);
2837 }
2838}
2839
2840static inline void set_inode_flag(struct inode *inode, int flag)
2841{
2842 set_bit(flag, F2FS_I(inode)->flags);
2843 __mark_inode_dirty_flag(inode, flag, true);
2844}
2845
2846static inline int is_inode_flag_set(struct inode *inode, int flag)
2847{
2848 return test_bit(flag, F2FS_I(inode)->flags);
2849}
2850
2851static inline void clear_inode_flag(struct inode *inode, int flag)
2852{
2853 clear_bit(flag, F2FS_I(inode)->flags);
2854 __mark_inode_dirty_flag(inode, flag, false);
2855}
2856
2857static inline bool f2fs_verity_in_progress(struct inode *inode)
2858{
2859 return IS_ENABLED(CONFIG_FS_VERITY) &&
2860 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
2861}
2862
2863static inline void set_acl_inode(struct inode *inode, umode_t mode)
2864{
2865 F2FS_I(inode)->i_acl_mode = mode;
2866 set_inode_flag(inode, FI_ACL_MODE);
2867 f2fs_mark_inode_dirty_sync(inode, false);
2868}
2869
2870static inline void f2fs_i_links_write(struct inode *inode, bool inc)
2871{
2872 if (inc)
2873 inc_nlink(inode);
2874 else
2875 drop_nlink(inode);
2876 f2fs_mark_inode_dirty_sync(inode, true);
2877}
2878
2879static inline void f2fs_i_blocks_write(struct inode *inode,
2880 block_t diff, bool add, bool claim)
2881{
2882 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2883 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2884
2885 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
2886 if (add) {
2887 if (claim)
2888 dquot_claim_block(inode, diff);
2889 else
2890 dquot_alloc_block_nofail(inode, diff);
2891 } else {
2892 dquot_free_block(inode, diff);
2893 }
2894
2895 f2fs_mark_inode_dirty_sync(inode, true);
2896 if (clean || recover)
2897 set_inode_flag(inode, FI_AUTO_RECOVER);
2898}
2899
2900static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
2901{
2902 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
2903 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
2904
2905 if (i_size_read(inode) == i_size)
2906 return;
2907
2908 i_size_write(inode, i_size);
2909 f2fs_mark_inode_dirty_sync(inode, true);
2910 if (clean || recover)
2911 set_inode_flag(inode, FI_AUTO_RECOVER);
2912}
2913
2914static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
2915{
2916 F2FS_I(inode)->i_current_depth = depth;
2917 f2fs_mark_inode_dirty_sync(inode, true);
2918}
2919
2920static inline void f2fs_i_gc_failures_write(struct inode *inode,
2921 unsigned int count)
2922{
2923 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
2924 f2fs_mark_inode_dirty_sync(inode, true);
2925}
2926
2927static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
2928{
2929 F2FS_I(inode)->i_xattr_nid = xnid;
2930 f2fs_mark_inode_dirty_sync(inode, true);
2931}
2932
2933static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
2934{
2935 F2FS_I(inode)->i_pino = pino;
2936 f2fs_mark_inode_dirty_sync(inode, true);
2937}
2938
2939static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
2940{
2941 struct f2fs_inode_info *fi = F2FS_I(inode);
2942
2943 if (ri->i_inline & F2FS_INLINE_XATTR)
2944 set_bit(FI_INLINE_XATTR, fi->flags);
2945 if (ri->i_inline & F2FS_INLINE_DATA)
2946 set_bit(FI_INLINE_DATA, fi->flags);
2947 if (ri->i_inline & F2FS_INLINE_DENTRY)
2948 set_bit(FI_INLINE_DENTRY, fi->flags);
2949 if (ri->i_inline & F2FS_DATA_EXIST)
2950 set_bit(FI_DATA_EXIST, fi->flags);
2951 if (ri->i_inline & F2FS_INLINE_DOTS)
2952 set_bit(FI_INLINE_DOTS, fi->flags);
2953 if (ri->i_inline & F2FS_EXTRA_ATTR)
2954 set_bit(FI_EXTRA_ATTR, fi->flags);
2955 if (ri->i_inline & F2FS_PIN_FILE)
2956 set_bit(FI_PIN_FILE, fi->flags);
2957 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
2958 set_bit(FI_COMPRESS_RELEASED, fi->flags);
2959}
2960
2961static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
2962{
2963 ri->i_inline = 0;
2964
2965 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
2966 ri->i_inline |= F2FS_INLINE_XATTR;
2967 if (is_inode_flag_set(inode, FI_INLINE_DATA))
2968 ri->i_inline |= F2FS_INLINE_DATA;
2969 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
2970 ri->i_inline |= F2FS_INLINE_DENTRY;
2971 if (is_inode_flag_set(inode, FI_DATA_EXIST))
2972 ri->i_inline |= F2FS_DATA_EXIST;
2973 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
2974 ri->i_inline |= F2FS_INLINE_DOTS;
2975 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
2976 ri->i_inline |= F2FS_EXTRA_ATTR;
2977 if (is_inode_flag_set(inode, FI_PIN_FILE))
2978 ri->i_inline |= F2FS_PIN_FILE;
2979 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
2980 ri->i_inline |= F2FS_COMPRESS_RELEASED;
2981}
2982
2983static inline int f2fs_has_extra_attr(struct inode *inode)
2984{
2985 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
2986}
2987
2988static inline int f2fs_has_inline_xattr(struct inode *inode)
2989{
2990 return is_inode_flag_set(inode, FI_INLINE_XATTR);
2991}
2992
2993static inline int f2fs_compressed_file(struct inode *inode)
2994{
2995 return S_ISREG(inode->i_mode) &&
2996 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
2997}
2998
2999static inline bool f2fs_need_compress_data(struct inode *inode)
3000{
3001 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
3002
3003 if (!f2fs_compressed_file(inode))
3004 return false;
3005
3006 if (compress_mode == COMPR_MODE_FS)
3007 return true;
3008 else if (compress_mode == COMPR_MODE_USER &&
3009 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
3010 return true;
3011
3012 return false;
3013}
3014
3015static inline unsigned int addrs_per_inode(struct inode *inode)
3016{
3017 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3018 get_inline_xattr_addrs(inode);
3019
3020 if (!f2fs_compressed_file(inode))
3021 return addrs;
3022 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3023}
3024
3025static inline unsigned int addrs_per_block(struct inode *inode)
3026{
3027 if (!f2fs_compressed_file(inode))
3028 return DEF_ADDRS_PER_BLOCK;
3029 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3030}
3031
3032static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3033{
3034 struct f2fs_inode *ri = F2FS_INODE(page);
3035
3036 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3037 get_inline_xattr_addrs(inode)]);
3038}
3039
3040static inline int inline_xattr_size(struct inode *inode)
3041{
3042 if (f2fs_has_inline_xattr(inode))
3043 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3044 return 0;
3045}
3046
3047static inline int f2fs_has_inline_data(struct inode *inode)
3048{
3049 return is_inode_flag_set(inode, FI_INLINE_DATA);
3050}
3051
3052static inline int f2fs_exist_data(struct inode *inode)
3053{
3054 return is_inode_flag_set(inode, FI_DATA_EXIST);
3055}
3056
3057static inline int f2fs_has_inline_dots(struct inode *inode)
3058{
3059 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3060}
3061
3062static inline int f2fs_is_mmap_file(struct inode *inode)
3063{
3064 return is_inode_flag_set(inode, FI_MMAP_FILE);
3065}
3066
3067static inline bool f2fs_is_pinned_file(struct inode *inode)
3068{
3069 return is_inode_flag_set(inode, FI_PIN_FILE);
3070}
3071
3072static inline bool f2fs_is_atomic_file(struct inode *inode)
3073{
3074 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3075}
3076
3077static inline bool f2fs_is_commit_atomic_write(struct inode *inode)
3078{
3079 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT);
3080}
3081
3082static inline bool f2fs_is_volatile_file(struct inode *inode)
3083{
3084 return is_inode_flag_set(inode, FI_VOLATILE_FILE);
3085}
3086
3087static inline bool f2fs_is_first_block_written(struct inode *inode)
3088{
3089 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
3090}
3091
3092static inline bool f2fs_is_drop_cache(struct inode *inode)
3093{
3094 return is_inode_flag_set(inode, FI_DROP_CACHE);
3095}
3096
3097static inline void *inline_data_addr(struct inode *inode, struct page *page)
3098{
3099 struct f2fs_inode *ri = F2FS_INODE(page);
3100 int extra_size = get_extra_isize(inode);
3101
3102 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
3103}
3104
3105static inline int f2fs_has_inline_dentry(struct inode *inode)
3106{
3107 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3108}
3109
3110static inline int is_file(struct inode *inode, int type)
3111{
3112 return F2FS_I(inode)->i_advise & type;
3113}
3114
3115static inline void set_file(struct inode *inode, int type)
3116{
3117 F2FS_I(inode)->i_advise |= type;
3118 f2fs_mark_inode_dirty_sync(inode, true);
3119}
3120
3121static inline void clear_file(struct inode *inode, int type)
3122{
3123 F2FS_I(inode)->i_advise &= ~type;
3124 f2fs_mark_inode_dirty_sync(inode, true);
3125}
3126
3127static inline bool f2fs_is_time_consistent(struct inode *inode)
3128{
3129 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
3130 return false;
3131 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
3132 return false;
3133 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
3134 return false;
3135 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3,
3136 &F2FS_I(inode)->i_crtime))
3137 return false;
3138 return true;
3139}
3140
3141static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3142{
3143 bool ret;
3144
3145 if (dsync) {
3146 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3147
3148 spin_lock(&sbi->inode_lock[DIRTY_META]);
3149 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3150 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3151 return ret;
3152 }
3153 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3154 file_keep_isize(inode) ||
3155 i_size_read(inode) & ~PAGE_MASK)
3156 return false;
3157
3158 if (!f2fs_is_time_consistent(inode))
3159 return false;
3160
3161 spin_lock(&F2FS_I(inode)->i_size_lock);
3162 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3163 spin_unlock(&F2FS_I(inode)->i_size_lock);
3164
3165 return ret;
3166}
3167
3168static inline bool f2fs_readonly(struct super_block *sb)
3169{
3170 return sb_rdonly(sb);
3171}
3172
3173static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3174{
3175 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3176}
3177
3178static inline bool is_dot_dotdot(const u8 *name, size_t len)
3179{
3180 if (len == 1 && name[0] == '.')
3181 return true;
3182
3183 if (len == 2 && name[0] == '.' && name[1] == '.')
3184 return true;
3185
3186 return false;
3187}
3188
3189static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3190 size_t size, gfp_t flags)
3191{
3192 if (time_to_inject(sbi, FAULT_KMALLOC)) {
3193 f2fs_show_injection_info(sbi, FAULT_KMALLOC);
3194 return NULL;
3195 }
3196
3197 return kmalloc(size, flags);
3198}
3199
3200static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3201 size_t size, gfp_t flags)
3202{
3203 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3204}
3205
3206static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3207 size_t size, gfp_t flags)
3208{
3209 if (time_to_inject(sbi, FAULT_KVMALLOC)) {
3210 f2fs_show_injection_info(sbi, FAULT_KVMALLOC);
3211 return NULL;
3212 }
3213
3214 return kvmalloc(size, flags);
3215}
3216
3217static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3218 size_t size, gfp_t flags)
3219{
3220 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3221}
3222
3223static inline int get_extra_isize(struct inode *inode)
3224{
3225 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3226}
3227
3228static inline int get_inline_xattr_addrs(struct inode *inode)
3229{
3230 return F2FS_I(inode)->i_inline_xattr_size;
3231}
3232
3233#define f2fs_get_inode_mode(i) \
3234 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3235 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3236
3237#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3238 (offsetof(struct f2fs_inode, i_extra_end) - \
3239 offsetof(struct f2fs_inode, i_extra_isize)) \
3240
3241#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3242#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3243 ((offsetof(typeof(*(f2fs_inode)), field) + \
3244 sizeof((f2fs_inode)->field)) \
3245 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3246
3247#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3248
3249#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3250
3251bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3252 block_t blkaddr, int type);
3253static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3254 block_t blkaddr, int type)
3255{
3256 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3257 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3258 blkaddr, type);
3259 f2fs_bug_on(sbi, 1);
3260 }
3261}
3262
3263static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3264{
3265 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3266 blkaddr == COMPRESS_ADDR)
3267 return false;
3268 return true;
3269}
3270
3271/*
3272 * file.c
3273 */
3274int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3275void f2fs_truncate_data_blocks(struct dnode_of_data *dn);
3276int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3277int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3278int f2fs_truncate(struct inode *inode);
3279int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path,
3280 struct kstat *stat, u32 request_mask, unsigned int flags);
3281int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
3282 struct iattr *attr);
3283int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3284void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3285int f2fs_precache_extents(struct inode *inode);
3286int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3287int f2fs_fileattr_set(struct user_namespace *mnt_userns,
3288 struct dentry *dentry, struct fileattr *fa);
3289long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3290long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3291int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3292int f2fs_pin_file_control(struct inode *inode, bool inc);
3293
3294/*
3295 * inode.c
3296 */
3297void f2fs_set_inode_flags(struct inode *inode);
3298bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3299void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3300struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3301struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3302int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3303void f2fs_update_inode(struct inode *inode, struct page *node_page);
3304void f2fs_update_inode_page(struct inode *inode);
3305int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3306void f2fs_evict_inode(struct inode *inode);
3307void f2fs_handle_failed_inode(struct inode *inode);
3308
3309/*
3310 * namei.c
3311 */
3312int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3313 bool hot, bool set);
3314struct dentry *f2fs_get_parent(struct dentry *child);
3315
3316/*
3317 * dir.c
3318 */
3319unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de);
3320int f2fs_init_casefolded_name(const struct inode *dir,
3321 struct f2fs_filename *fname);
3322int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3323 int lookup, struct f2fs_filename *fname);
3324int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3325 struct f2fs_filename *fname);
3326void f2fs_free_filename(struct f2fs_filename *fname);
3327struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3328 const struct f2fs_filename *fname, int *max_slots);
3329int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3330 unsigned int start_pos, struct fscrypt_str *fstr);
3331void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3332 struct f2fs_dentry_ptr *d);
3333struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3334 const struct f2fs_filename *fname, struct page *dpage);
3335void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3336 unsigned int current_depth);
3337int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3338void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3339struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3340 const struct f2fs_filename *fname,
3341 struct page **res_page);
3342struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3343 const struct qstr *child, struct page **res_page);
3344struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3345ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3346 struct page **page);
3347void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3348 struct page *page, struct inode *inode);
3349bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3350 const struct f2fs_filename *fname);
3351void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3352 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3353 unsigned int bit_pos);
3354int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3355 struct inode *inode, nid_t ino, umode_t mode);
3356int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3357 struct inode *inode, nid_t ino, umode_t mode);
3358int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3359 struct inode *inode, nid_t ino, umode_t mode);
3360void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3361 struct inode *dir, struct inode *inode);
3362int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3363bool f2fs_empty_dir(struct inode *dir);
3364
3365static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3366{
3367 if (fscrypt_is_nokey_name(dentry))
3368 return -ENOKEY;
3369 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3370 inode, inode->i_ino, inode->i_mode);
3371}
3372
3373/*
3374 * super.c
3375 */
3376int f2fs_inode_dirtied(struct inode *inode, bool sync);
3377void f2fs_inode_synced(struct inode *inode);
3378int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3379int f2fs_quota_sync(struct super_block *sb, int type);
3380loff_t max_file_blocks(struct inode *inode);
3381void f2fs_quota_off_umount(struct super_block *sb);
3382int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3383int f2fs_sync_fs(struct super_block *sb, int sync);
3384int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3385
3386/*
3387 * hash.c
3388 */
3389void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3390
3391/*
3392 * node.c
3393 */
3394struct node_info;
3395
3396int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3397bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3398bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3399void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3400void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3401void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3402int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3403bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3404bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3405int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3406 struct node_info *ni);
3407pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3408int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3409int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3410int f2fs_truncate_xattr_node(struct inode *inode);
3411int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3412 unsigned int seq_id);
3413bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
3414int f2fs_remove_inode_page(struct inode *inode);
3415struct page *f2fs_new_inode_page(struct inode *inode);
3416struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3417void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3418struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3419struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3420int f2fs_move_node_page(struct page *node_page, int gc_type);
3421void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3422int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3423 struct writeback_control *wbc, bool atomic,
3424 unsigned int *seq_id);
3425int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3426 struct writeback_control *wbc,
3427 bool do_balance, enum iostat_type io_type);
3428int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3429bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3430void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3431void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3432int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3433int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3434int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3435int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3436int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3437 unsigned int segno, struct f2fs_summary_block *sum);
3438void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
3439int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3440int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3441void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3442int __init f2fs_create_node_manager_caches(void);
3443void f2fs_destroy_node_manager_caches(void);
3444
3445/*
3446 * segment.c
3447 */
3448bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3449void f2fs_register_inmem_page(struct inode *inode, struct page *page);
3450void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure);
3451void f2fs_drop_inmem_pages(struct inode *inode);
3452void f2fs_drop_inmem_page(struct inode *inode, struct page *page);
3453int f2fs_commit_inmem_pages(struct inode *inode);
3454void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3455void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3456int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3457int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3458int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3459void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3460void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3461bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3462int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
3463void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3464void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3465bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3466void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3467 struct cp_control *cpc);
3468void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3469block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3470int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3471void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3472int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3473bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3474void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3475void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3476void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3477void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3478 unsigned int *newseg, bool new_sec, int dir);
3479void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3480 unsigned int start, unsigned int end);
3481void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3482void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3483int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3484bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3485 struct cp_control *cpc);
3486struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3487void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3488 block_t blk_addr);
3489void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3490 enum iostat_type io_type);
3491void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3492void f2fs_outplace_write_data(struct dnode_of_data *dn,
3493 struct f2fs_io_info *fio);
3494int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3495void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3496 block_t old_blkaddr, block_t new_blkaddr,
3497 bool recover_curseg, bool recover_newaddr,
3498 bool from_gc);
3499void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3500 block_t old_addr, block_t new_addr,
3501 unsigned char version, bool recover_curseg,
3502 bool recover_newaddr);
3503void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3504 block_t old_blkaddr, block_t *new_blkaddr,
3505 struct f2fs_summary *sum, int type,
3506 struct f2fs_io_info *fio);
3507void f2fs_wait_on_page_writeback(struct page *page,
3508 enum page_type type, bool ordered, bool locked);
3509void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3510void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3511 block_t len);
3512void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3513void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3514int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3515 unsigned int val, int alloc);
3516void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3517int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3518int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3519int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3520void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3521int __init f2fs_create_segment_manager_caches(void);
3522void f2fs_destroy_segment_manager_caches(void);
3523int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3524enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3525 enum page_type type, enum temp_type temp);
3526unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3527 unsigned int segno);
3528unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3529 unsigned int segno);
3530
3531/*
3532 * checkpoint.c
3533 */
3534void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io);
3535struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3536struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3537struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3538struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3539bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3540 block_t blkaddr, int type);
3541int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3542 int type, bool sync);
3543void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index);
3544long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3545 long nr_to_write, enum iostat_type io_type);
3546void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3547void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3548void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3549bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3550void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3551 unsigned int devidx, int type);
3552bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3553 unsigned int devidx, int type);
3554int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi);
3555int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3556void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3557void f2fs_add_orphan_inode(struct inode *inode);
3558void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3559int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3560int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3561void f2fs_update_dirty_page(struct inode *inode, struct page *page);
3562void f2fs_remove_dirty_inode(struct inode *inode);
3563int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type);
3564void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3565u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3566int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3567void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3568int __init f2fs_create_checkpoint_caches(void);
3569void f2fs_destroy_checkpoint_caches(void);
3570int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3571int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3572void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3573void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3574
3575/*
3576 * data.c
3577 */
3578int __init f2fs_init_bioset(void);
3579void f2fs_destroy_bioset(void);
3580int f2fs_init_bio_entry_cache(void);
3581void f2fs_destroy_bio_entry_cache(void);
3582void f2fs_submit_bio(struct f2fs_sb_info *sbi,
3583 struct bio *bio, enum page_type type);
3584void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3585void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3586 struct inode *inode, struct page *page,
3587 nid_t ino, enum page_type type);
3588void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3589 struct bio **bio, struct page *page);
3590void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3591int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3592int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3593void f2fs_submit_page_write(struct f2fs_io_info *fio);
3594struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3595 block_t blk_addr, struct bio *bio);
3596int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3597void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3598void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3599int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3600int f2fs_reserve_new_block(struct dnode_of_data *dn);
3601int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index);
3602int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from);
3603int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3604struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3605 int op_flags, bool for_write);
3606struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index);
3607struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3608 bool for_write);
3609struct page *f2fs_get_new_data_page(struct inode *inode,
3610 struct page *ipage, pgoff_t index, bool new_i_size);
3611int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3612void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock);
3613int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
3614 int create, int flag);
3615int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3616 u64 start, u64 len);
3617int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3618bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3619bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3620int f2fs_write_single_data_page(struct page *page, int *submitted,
3621 struct bio **bio, sector_t *last_block,
3622 struct writeback_control *wbc,
3623 enum iostat_type io_type,
3624 int compr_blocks, bool allow_balance);
3625void f2fs_invalidate_page(struct page *page, unsigned int offset,
3626 unsigned int length);
3627int f2fs_release_page(struct page *page, gfp_t wait);
3628#ifdef CONFIG_MIGRATION
3629int f2fs_migrate_page(struct address_space *mapping, struct page *newpage,
3630 struct page *page, enum migrate_mode mode);
3631#endif
3632bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3633void f2fs_clear_page_cache_dirty_tag(struct page *page);
3634int f2fs_init_post_read_processing(void);
3635void f2fs_destroy_post_read_processing(void);
3636int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3637void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3638
3639/*
3640 * gc.c
3641 */
3642int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3643void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3644block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3645int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force,
3646 unsigned int segno);
3647void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3648int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count);
3649int __init f2fs_create_garbage_collection_cache(void);
3650void f2fs_destroy_garbage_collection_cache(void);
3651
3652/*
3653 * recovery.c
3654 */
3655int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3656bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3657int __init f2fs_create_recovery_cache(void);
3658void f2fs_destroy_recovery_cache(void);
3659
3660/*
3661 * debug.c
3662 */
3663#ifdef CONFIG_F2FS_STAT_FS
3664struct f2fs_stat_info {
3665 struct list_head stat_list;
3666 struct f2fs_sb_info *sbi;
3667 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3668 int main_area_segs, main_area_sections, main_area_zones;
3669 unsigned long long hit_largest, hit_cached, hit_rbtree;
3670 unsigned long long hit_total, total_ext;
3671 int ext_tree, zombie_tree, ext_node;
3672 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3673 int ndirty_data, ndirty_qdata;
3674 int inmem_pages;
3675 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3676 int nats, dirty_nats, sits, dirty_sits;
3677 int free_nids, avail_nids, alloc_nids;
3678 int total_count, utilization;
3679 int bg_gc, nr_wb_cp_data, nr_wb_data;
3680 int nr_rd_data, nr_rd_node, nr_rd_meta;
3681 int nr_dio_read, nr_dio_write;
3682 unsigned int io_skip_bggc, other_skip_bggc;
3683 int nr_flushing, nr_flushed, flush_list_empty;
3684 int nr_discarding, nr_discarded;
3685 int nr_discard_cmd;
3686 unsigned int undiscard_blks;
3687 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3688 unsigned int cur_ckpt_time, peak_ckpt_time;
3689 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3690 int compr_inode;
3691 unsigned long long compr_blocks;
3692 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
3693 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3694 unsigned int bimodal, avg_vblocks;
3695 int util_free, util_valid, util_invalid;
3696 int rsvd_segs, overp_segs;
3697 int dirty_count, node_pages, meta_pages, compress_pages;
3698 int compress_page_hit;
3699 int prefree_count, call_count, cp_count, bg_cp_count;
3700 int tot_segs, node_segs, data_segs, free_segs, free_secs;
3701 int bg_node_segs, bg_data_segs;
3702 int tot_blks, data_blks, node_blks;
3703 int bg_data_blks, bg_node_blks;
3704 unsigned long long skipped_atomic_files[2];
3705 int curseg[NR_CURSEG_TYPE];
3706 int cursec[NR_CURSEG_TYPE];
3707 int curzone[NR_CURSEG_TYPE];
3708 unsigned int dirty_seg[NR_CURSEG_TYPE];
3709 unsigned int full_seg[NR_CURSEG_TYPE];
3710 unsigned int valid_blks[NR_CURSEG_TYPE];
3711
3712 unsigned int meta_count[META_MAX];
3713 unsigned int segment_count[2];
3714 unsigned int block_count[2];
3715 unsigned int inplace_count;
3716 unsigned long long base_mem, cache_mem, page_mem;
3717};
3718
3719static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3720{
3721 return (struct f2fs_stat_info *)sbi->stat_info;
3722}
3723
3724#define stat_inc_cp_count(si) ((si)->cp_count++)
3725#define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++)
3726#define stat_inc_call_count(si) ((si)->call_count++)
3727#define stat_inc_bggc_count(si) ((si)->bg_gc++)
3728#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3729#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3730#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3731#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3732#define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext))
3733#define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree))
3734#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3735#define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached))
3736#define stat_inc_inline_xattr(inode) \
3737 do { \
3738 if (f2fs_has_inline_xattr(inode)) \
3739 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3740 } while (0)
3741#define stat_dec_inline_xattr(inode) \
3742 do { \
3743 if (f2fs_has_inline_xattr(inode)) \
3744 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3745 } while (0)
3746#define stat_inc_inline_inode(inode) \
3747 do { \
3748 if (f2fs_has_inline_data(inode)) \
3749 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3750 } while (0)
3751#define stat_dec_inline_inode(inode) \
3752 do { \
3753 if (f2fs_has_inline_data(inode)) \
3754 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3755 } while (0)
3756#define stat_inc_inline_dir(inode) \
3757 do { \
3758 if (f2fs_has_inline_dentry(inode)) \
3759 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3760 } while (0)
3761#define stat_dec_inline_dir(inode) \
3762 do { \
3763 if (f2fs_has_inline_dentry(inode)) \
3764 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3765 } while (0)
3766#define stat_inc_compr_inode(inode) \
3767 do { \
3768 if (f2fs_compressed_file(inode)) \
3769 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3770 } while (0)
3771#define stat_dec_compr_inode(inode) \
3772 do { \
3773 if (f2fs_compressed_file(inode)) \
3774 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
3775 } while (0)
3776#define stat_add_compr_blocks(inode, blocks) \
3777 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
3778#define stat_sub_compr_blocks(inode, blocks) \
3779 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
3780#define stat_inc_meta_count(sbi, blkaddr) \
3781 do { \
3782 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
3783 atomic_inc(&(sbi)->meta_count[META_CP]); \
3784 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
3785 atomic_inc(&(sbi)->meta_count[META_SIT]); \
3786 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
3787 atomic_inc(&(sbi)->meta_count[META_NAT]); \
3788 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
3789 atomic_inc(&(sbi)->meta_count[META_SSA]); \
3790 } while (0)
3791#define stat_inc_seg_type(sbi, curseg) \
3792 ((sbi)->segment_count[(curseg)->alloc_type]++)
3793#define stat_inc_block_count(sbi, curseg) \
3794 ((sbi)->block_count[(curseg)->alloc_type]++)
3795#define stat_inc_inplace_blocks(sbi) \
3796 (atomic_inc(&(sbi)->inplace_count))
3797#define stat_update_max_atomic_write(inode) \
3798 do { \
3799 int cur = F2FS_I_SB(inode)->atomic_files; \
3800 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
3801 if (cur > max) \
3802 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
3803 } while (0)
3804#define stat_inc_volatile_write(inode) \
3805 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt))
3806#define stat_dec_volatile_write(inode) \
3807 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt))
3808#define stat_update_max_volatile_write(inode) \
3809 do { \
3810 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \
3811 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \
3812 if (cur > max) \
3813 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \
3814 } while (0)
3815#define stat_inc_seg_count(sbi, type, gc_type) \
3816 do { \
3817 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3818 si->tot_segs++; \
3819 if ((type) == SUM_TYPE_DATA) { \
3820 si->data_segs++; \
3821 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \
3822 } else { \
3823 si->node_segs++; \
3824 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \
3825 } \
3826 } while (0)
3827
3828#define stat_inc_tot_blk_count(si, blks) \
3829 ((si)->tot_blks += (blks))
3830
3831#define stat_inc_data_blk_count(sbi, blks, gc_type) \
3832 do { \
3833 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3834 stat_inc_tot_blk_count(si, blks); \
3835 si->data_blks += (blks); \
3836 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3837 } while (0)
3838
3839#define stat_inc_node_blk_count(sbi, blks, gc_type) \
3840 do { \
3841 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
3842 stat_inc_tot_blk_count(si, blks); \
3843 si->node_blks += (blks); \
3844 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
3845 } while (0)
3846
3847int f2fs_build_stats(struct f2fs_sb_info *sbi);
3848void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
3849void __init f2fs_create_root_stats(void);
3850void f2fs_destroy_root_stats(void);
3851void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
3852#else
3853#define stat_inc_cp_count(si) do { } while (0)
3854#define stat_inc_bg_cp_count(si) do { } while (0)
3855#define stat_inc_call_count(si) do { } while (0)
3856#define stat_inc_bggc_count(si) do { } while (0)
3857#define stat_io_skip_bggc_count(sbi) do { } while (0)
3858#define stat_other_skip_bggc_count(sbi) do { } while (0)
3859#define stat_inc_dirty_inode(sbi, type) do { } while (0)
3860#define stat_dec_dirty_inode(sbi, type) do { } while (0)
3861#define stat_inc_total_hit(sbi) do { } while (0)
3862#define stat_inc_rbtree_node_hit(sbi) do { } while (0)
3863#define stat_inc_largest_node_hit(sbi) do { } while (0)
3864#define stat_inc_cached_node_hit(sbi) do { } while (0)
3865#define stat_inc_inline_xattr(inode) do { } while (0)
3866#define stat_dec_inline_xattr(inode) do { } while (0)
3867#define stat_inc_inline_inode(inode) do { } while (0)
3868#define stat_dec_inline_inode(inode) do { } while (0)
3869#define stat_inc_inline_dir(inode) do { } while (0)
3870#define stat_dec_inline_dir(inode) do { } while (0)
3871#define stat_inc_compr_inode(inode) do { } while (0)
3872#define stat_dec_compr_inode(inode) do { } while (0)
3873#define stat_add_compr_blocks(inode, blocks) do { } while (0)
3874#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
3875#define stat_update_max_atomic_write(inode) do { } while (0)
3876#define stat_inc_volatile_write(inode) do { } while (0)
3877#define stat_dec_volatile_write(inode) do { } while (0)
3878#define stat_update_max_volatile_write(inode) do { } while (0)
3879#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
3880#define stat_inc_seg_type(sbi, curseg) do { } while (0)
3881#define stat_inc_block_count(sbi, curseg) do { } while (0)
3882#define stat_inc_inplace_blocks(sbi) do { } while (0)
3883#define stat_inc_seg_count(sbi, type, gc_type) do { } while (0)
3884#define stat_inc_tot_blk_count(si, blks) do { } while (0)
3885#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
3886#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
3887
3888static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
3889static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
3890static inline void __init f2fs_create_root_stats(void) { }
3891static inline void f2fs_destroy_root_stats(void) { }
3892static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
3893#endif
3894
3895extern const struct file_operations f2fs_dir_operations;
3896extern const struct file_operations f2fs_file_operations;
3897extern const struct inode_operations f2fs_file_inode_operations;
3898extern const struct address_space_operations f2fs_dblock_aops;
3899extern const struct address_space_operations f2fs_node_aops;
3900extern const struct address_space_operations f2fs_meta_aops;
3901extern const struct inode_operations f2fs_dir_inode_operations;
3902extern const struct inode_operations f2fs_symlink_inode_operations;
3903extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
3904extern const struct inode_operations f2fs_special_inode_operations;
3905extern struct kmem_cache *f2fs_inode_entry_slab;
3906
3907/*
3908 * inline.c
3909 */
3910bool f2fs_may_inline_data(struct inode *inode);
3911bool f2fs_may_inline_dentry(struct inode *inode);
3912void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
3913void f2fs_truncate_inline_inode(struct inode *inode,
3914 struct page *ipage, u64 from);
3915int f2fs_read_inline_data(struct inode *inode, struct page *page);
3916int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
3917int f2fs_convert_inline_inode(struct inode *inode);
3918int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
3919int f2fs_write_inline_data(struct inode *inode, struct page *page);
3920int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
3921struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
3922 const struct f2fs_filename *fname,
3923 struct page **res_page);
3924int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
3925 struct page *ipage);
3926int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
3927 struct inode *inode, nid_t ino, umode_t mode);
3928void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
3929 struct page *page, struct inode *dir,
3930 struct inode *inode);
3931bool f2fs_empty_inline_dir(struct inode *dir);
3932int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
3933 struct fscrypt_str *fstr);
3934int f2fs_inline_data_fiemap(struct inode *inode,
3935 struct fiemap_extent_info *fieinfo,
3936 __u64 start, __u64 len);
3937
3938/*
3939 * shrinker.c
3940 */
3941unsigned long f2fs_shrink_count(struct shrinker *shrink,
3942 struct shrink_control *sc);
3943unsigned long f2fs_shrink_scan(struct shrinker *shrink,
3944 struct shrink_control *sc);
3945void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
3946void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
3947
3948/*
3949 * extent_cache.c
3950 */
3951struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root,
3952 struct rb_entry *cached_re, unsigned int ofs);
3953struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi,
3954 struct rb_root_cached *root,
3955 struct rb_node **parent,
3956 unsigned long long key, bool *left_most);
3957struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi,
3958 struct rb_root_cached *root,
3959 struct rb_node **parent,
3960 unsigned int ofs, bool *leftmost);
3961struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root,
3962 struct rb_entry *cached_re, unsigned int ofs,
3963 struct rb_entry **prev_entry, struct rb_entry **next_entry,
3964 struct rb_node ***insert_p, struct rb_node **insert_parent,
3965 bool force, bool *leftmost);
3966bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi,
3967 struct rb_root_cached *root, bool check_key);
3968unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink);
3969void f2fs_init_extent_tree(struct inode *inode, struct page *ipage);
3970void f2fs_drop_extent_tree(struct inode *inode);
3971unsigned int f2fs_destroy_extent_node(struct inode *inode);
3972void f2fs_destroy_extent_tree(struct inode *inode);
3973bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs,
3974 struct extent_info *ei);
3975void f2fs_update_extent_cache(struct dnode_of_data *dn);
3976void f2fs_update_extent_cache_range(struct dnode_of_data *dn,
3977 pgoff_t fofs, block_t blkaddr, unsigned int len);
3978void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
3979int __init f2fs_create_extent_cache(void);
3980void f2fs_destroy_extent_cache(void);
3981
3982/*
3983 * sysfs.c
3984 */
3985#define MIN_RA_MUL 2
3986#define MAX_RA_MUL 256
3987
3988int __init f2fs_init_sysfs(void);
3989void f2fs_exit_sysfs(void);
3990int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
3991void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
3992
3993/* verity.c */
3994extern const struct fsverity_operations f2fs_verityops;
3995
3996/*
3997 * crypto support
3998 */
3999static inline bool f2fs_encrypted_file(struct inode *inode)
4000{
4001 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4002}
4003
4004static inline void f2fs_set_encrypted_inode(struct inode *inode)
4005{
4006#ifdef CONFIG_FS_ENCRYPTION
4007 file_set_encrypt(inode);
4008 f2fs_set_inode_flags(inode);
4009#endif
4010}
4011
4012/*
4013 * Returns true if the reads of the inode's data need to undergo some
4014 * postprocessing step, like decryption or authenticity verification.
4015 */
4016static inline bool f2fs_post_read_required(struct inode *inode)
4017{
4018 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4019 f2fs_compressed_file(inode);
4020}
4021
4022/*
4023 * compress.c
4024 */
4025#ifdef CONFIG_F2FS_FS_COMPRESSION
4026bool f2fs_is_compressed_page(struct page *page);
4027struct page *f2fs_compress_control_page(struct page *page);
4028int f2fs_prepare_compress_overwrite(struct inode *inode,
4029 struct page **pagep, pgoff_t index, void **fsdata);
4030bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4031 pgoff_t index, unsigned copied);
4032int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4033void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4034bool f2fs_is_compress_backend_ready(struct inode *inode);
4035int f2fs_init_compress_mempool(void);
4036void f2fs_destroy_compress_mempool(void);
4037void f2fs_decompress_cluster(struct decompress_io_ctx *dic);
4038void f2fs_end_read_compressed_page(struct page *page, bool failed,
4039 block_t blkaddr);
4040bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4041bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4042bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
4043void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4044int f2fs_write_multi_pages(struct compress_ctx *cc,
4045 int *submitted,
4046 struct writeback_control *wbc,
4047 enum iostat_type io_type);
4048int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4049void f2fs_update_extent_tree_range_compressed(struct inode *inode,
4050 pgoff_t fofs, block_t blkaddr, unsigned int llen,
4051 unsigned int c_len);
4052int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4053 unsigned nr_pages, sector_t *last_block_in_bio,
4054 bool is_readahead, bool for_write);
4055struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4056void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed);
4057void f2fs_put_page_dic(struct page *page);
4058unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn);
4059int f2fs_init_compress_ctx(struct compress_ctx *cc);
4060void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4061void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4062int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4063void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4064int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4065void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4066int __init f2fs_init_compress_cache(void);
4067void f2fs_destroy_compress_cache(void);
4068struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4069void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4070void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4071 nid_t ino, block_t blkaddr);
4072bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4073 block_t blkaddr);
4074void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4075#define inc_compr_inode_stat(inode) \
4076 do { \
4077 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4078 sbi->compr_new_inode++; \
4079 } while (0)
4080#define add_compr_block_stat(inode, blocks) \
4081 do { \
4082 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4083 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4084 sbi->compr_written_block += blocks; \
4085 sbi->compr_saved_block += diff; \
4086 } while (0)
4087#else
4088static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4089static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4090{
4091 if (!f2fs_compressed_file(inode))
4092 return true;
4093 /* not support compression */
4094 return false;
4095}
4096static inline struct page *f2fs_compress_control_page(struct page *page)
4097{
4098 WARN_ON_ONCE(1);
4099 return ERR_PTR(-EINVAL);
4100}
4101static inline int f2fs_init_compress_mempool(void) { return 0; }
4102static inline void f2fs_destroy_compress_mempool(void) { }
4103static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { }
4104static inline void f2fs_end_read_compressed_page(struct page *page,
4105 bool failed, block_t blkaddr)
4106{
4107 WARN_ON_ONCE(1);
4108}
4109static inline void f2fs_put_page_dic(struct page *page)
4110{
4111 WARN_ON_ONCE(1);
4112}
4113static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; }
4114static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
4115static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4116static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4117static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4118static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4119static inline int __init f2fs_init_compress_cache(void) { return 0; }
4120static inline void f2fs_destroy_compress_cache(void) { }
4121static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4122 block_t blkaddr) { }
4123static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4124 struct page *page, nid_t ino, block_t blkaddr) { }
4125static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4126 struct page *page, block_t blkaddr) { return false; }
4127static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4128 nid_t ino) { }
4129#define inc_compr_inode_stat(inode) do { } while (0)
4130static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode,
4131 pgoff_t fofs, block_t blkaddr, unsigned int llen,
4132 unsigned int c_len) { }
4133#endif
4134
4135static inline void set_compress_context(struct inode *inode)
4136{
4137 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4138
4139 F2FS_I(inode)->i_compress_algorithm =
4140 F2FS_OPTION(sbi).compress_algorithm;
4141 F2FS_I(inode)->i_log_cluster_size =
4142 F2FS_OPTION(sbi).compress_log_size;
4143 F2FS_I(inode)->i_compress_flag =
4144 F2FS_OPTION(sbi).compress_chksum ?
4145 1 << COMPRESS_CHKSUM : 0;
4146 F2FS_I(inode)->i_cluster_size =
4147 1 << F2FS_I(inode)->i_log_cluster_size;
4148 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4149 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4150 F2FS_OPTION(sbi).compress_level)
4151 F2FS_I(inode)->i_compress_flag |=
4152 F2FS_OPTION(sbi).compress_level <<
4153 COMPRESS_LEVEL_OFFSET;
4154 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4155 set_inode_flag(inode, FI_COMPRESSED_FILE);
4156 stat_inc_compr_inode(inode);
4157 inc_compr_inode_stat(inode);
4158 f2fs_mark_inode_dirty_sync(inode, true);
4159}
4160
4161static inline bool f2fs_disable_compressed_file(struct inode *inode)
4162{
4163 struct f2fs_inode_info *fi = F2FS_I(inode);
4164
4165 if (!f2fs_compressed_file(inode))
4166 return true;
4167 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))
4168 return false;
4169
4170 fi->i_flags &= ~F2FS_COMPR_FL;
4171 stat_dec_compr_inode(inode);
4172 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4173 f2fs_mark_inode_dirty_sync(inode, true);
4174 return true;
4175}
4176
4177#define F2FS_FEATURE_FUNCS(name, flagname) \
4178static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4179{ \
4180 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4181}
4182
4183F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4184F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4185F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4186F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4187F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4188F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4189F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4190F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4191F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4192F2FS_FEATURE_FUNCS(verity, VERITY);
4193F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4194F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4195F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4196F2FS_FEATURE_FUNCS(readonly, RO);
4197
4198static inline bool f2fs_may_extent_tree(struct inode *inode)
4199{
4200 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4201
4202 if (!test_opt(sbi, EXTENT_CACHE) ||
4203 is_inode_flag_set(inode, FI_NO_EXTENT) ||
4204 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) &&
4205 !f2fs_sb_has_readonly(sbi)))
4206 return false;
4207
4208 /*
4209 * for recovered files during mount do not create extents
4210 * if shrinker is not registered.
4211 */
4212 if (list_empty(&sbi->s_list))
4213 return false;
4214
4215 return S_ISREG(inode->i_mode);
4216}
4217
4218#ifdef CONFIG_BLK_DEV_ZONED
4219static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4220 block_t blkaddr)
4221{
4222 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz;
4223
4224 return test_bit(zno, FDEV(devi).blkz_seq);
4225}
4226#endif
4227
4228static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4229{
4230 return f2fs_sb_has_blkzoned(sbi);
4231}
4232
4233static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4234{
4235 return blk_queue_discard(bdev_get_queue(bdev)) ||
4236 bdev_is_zoned(bdev);
4237}
4238
4239static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4240{
4241 int i;
4242
4243 if (!f2fs_is_multi_device(sbi))
4244 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4245
4246 for (i = 0; i < sbi->s_ndevs; i++)
4247 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4248 return true;
4249 return false;
4250}
4251
4252static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4253{
4254 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4255 f2fs_hw_should_discard(sbi);
4256}
4257
4258static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4259{
4260 int i;
4261
4262 if (!f2fs_is_multi_device(sbi))
4263 return bdev_read_only(sbi->sb->s_bdev);
4264
4265 for (i = 0; i < sbi->s_ndevs; i++)
4266 if (bdev_read_only(FDEV(i).bdev))
4267 return true;
4268 return false;
4269}
4270
4271static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4272{
4273 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4274}
4275
4276static inline bool f2fs_may_compress(struct inode *inode)
4277{
4278 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4279 f2fs_is_atomic_file(inode) ||
4280 f2fs_is_volatile_file(inode))
4281 return false;
4282 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4283}
4284
4285static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4286 u64 blocks, bool add)
4287{
4288 int diff = F2FS_I(inode)->i_cluster_size - blocks;
4289 struct f2fs_inode_info *fi = F2FS_I(inode);
4290
4291 /* don't update i_compr_blocks if saved blocks were released */
4292 if (!add && !atomic_read(&fi->i_compr_blocks))
4293 return;
4294
4295 if (add) {
4296 atomic_add(diff, &fi->i_compr_blocks);
4297 stat_add_compr_blocks(inode, diff);
4298 } else {
4299 atomic_sub(diff, &fi->i_compr_blocks);
4300 stat_sub_compr_blocks(inode, diff);
4301 }
4302 f2fs_mark_inode_dirty_sync(inode, true);
4303}
4304
4305static inline int block_unaligned_IO(struct inode *inode,
4306 struct kiocb *iocb, struct iov_iter *iter)
4307{
4308 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits);
4309 unsigned int blocksize_mask = (1 << i_blkbits) - 1;
4310 loff_t offset = iocb->ki_pos;
4311 unsigned long align = offset | iov_iter_alignment(iter);
4312
4313 return align & blocksize_mask;
4314}
4315
4316static inline bool f2fs_force_buffered_io(struct inode *inode,
4317 struct kiocb *iocb, struct iov_iter *iter)
4318{
4319 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4320 int rw = iov_iter_rw(iter);
4321
4322 if (f2fs_post_read_required(inode))
4323 return true;
4324 if (f2fs_is_multi_device(sbi))
4325 return true;
4326 /*
4327 * for blkzoned device, fallback direct IO to buffered IO, so
4328 * all IOs can be serialized by log-structured write.
4329 */
4330 if (f2fs_sb_has_blkzoned(sbi))
4331 return true;
4332 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) {
4333 if (block_unaligned_IO(inode, iocb, iter))
4334 return true;
4335 if (F2FS_IO_ALIGNED(sbi))
4336 return true;
4337 }
4338 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED))
4339 return true;
4340
4341 return false;
4342}
4343
4344static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4345{
4346 return fsverity_active(inode) &&
4347 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4348}
4349
4350#ifdef CONFIG_F2FS_FAULT_INJECTION
4351extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4352 unsigned int type);
4353#else
4354#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4355#endif
4356
4357static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4358{
4359#ifdef CONFIG_QUOTA
4360 if (f2fs_sb_has_quota_ino(sbi))
4361 return true;
4362 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4363 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4364 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4365 return true;
4366#endif
4367 return false;
4368}
4369
4370static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4371{
4372 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4373}
4374
4375#define EFSBADCRC EBADMSG /* Bad CRC detected */
4376#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4377
4378#endif /* _LINUX_F2FS_H */