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/sched/mm.h>
22#include <linux/vmalloc.h>
23#include <linux/bio.h>
24#include <linux/blkdev.h>
25#include <linux/quotaops.h>
26#include <linux/part_stat.h>
27#include <crypto/hash.h>
28
29#include <linux/fscrypt.h>
30#include <linux/fsverity.h>
31
32struct pagevec;
33
34#ifdef CONFIG_F2FS_CHECK_FS
35#define f2fs_bug_on(sbi, condition) BUG_ON(condition)
36#else
37#define f2fs_bug_on(sbi, condition) \
38 do { \
39 if (WARN_ON(condition)) \
40 set_sbi_flag(sbi, SBI_NEED_FSCK); \
41 } while (0)
42#endif
43
44enum {
45 FAULT_KMALLOC,
46 FAULT_KVMALLOC,
47 FAULT_PAGE_ALLOC,
48 FAULT_PAGE_GET,
49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */
50 FAULT_ALLOC_NID,
51 FAULT_ORPHAN,
52 FAULT_BLOCK,
53 FAULT_DIR_DEPTH,
54 FAULT_EVICT_INODE,
55 FAULT_TRUNCATE,
56 FAULT_READ_IO,
57 FAULT_CHECKPOINT,
58 FAULT_DISCARD,
59 FAULT_WRITE_IO,
60 FAULT_SLAB_ALLOC,
61 FAULT_DQUOT_INIT,
62 FAULT_LOCK_OP,
63 FAULT_BLKADDR,
64 FAULT_MAX,
65};
66
67#ifdef CONFIG_F2FS_FAULT_INJECTION
68#define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0))
69
70struct f2fs_fault_info {
71 atomic_t inject_ops;
72 unsigned int inject_rate;
73 unsigned int inject_type;
74};
75
76extern const char *f2fs_fault_name[FAULT_MAX];
77#define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))
78#endif
79
80/*
81 * For mount options
82 */
83#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001
84#define F2FS_MOUNT_DISCARD 0x00000002
85#define F2FS_MOUNT_NOHEAP 0x00000004
86#define F2FS_MOUNT_XATTR_USER 0x00000008
87#define F2FS_MOUNT_POSIX_ACL 0x00000010
88#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020
89#define F2FS_MOUNT_INLINE_XATTR 0x00000040
90#define F2FS_MOUNT_INLINE_DATA 0x00000080
91#define F2FS_MOUNT_INLINE_DENTRY 0x00000100
92#define F2FS_MOUNT_FLUSH_MERGE 0x00000200
93#define F2FS_MOUNT_NOBARRIER 0x00000400
94#define F2FS_MOUNT_FASTBOOT 0x00000800
95#define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000
96#define F2FS_MOUNT_DATA_FLUSH 0x00002000
97#define F2FS_MOUNT_FAULT_INJECTION 0x00004000
98#define F2FS_MOUNT_USRQUOTA 0x00008000
99#define F2FS_MOUNT_GRPQUOTA 0x00010000
100#define F2FS_MOUNT_PRJQUOTA 0x00020000
101#define F2FS_MOUNT_QUOTA 0x00040000
102#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000
103#define F2FS_MOUNT_RESERVE_ROOT 0x00100000
104#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000
105#define F2FS_MOUNT_NORECOVERY 0x00400000
106#define F2FS_MOUNT_ATGC 0x00800000
107#define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000
108#define F2FS_MOUNT_GC_MERGE 0x02000000
109#define F2FS_MOUNT_COMPRESS_CACHE 0x04000000
110#define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000
111
112#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
113#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
114#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
115#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
116
117#define ver_after(a, b) (typecheck(unsigned long long, a) && \
118 typecheck(unsigned long long, b) && \
119 ((long long)((a) - (b)) > 0))
120
121typedef u32 block_t; /*
122 * should not change u32, since it is the on-disk block
123 * address format, __le32.
124 */
125typedef u32 nid_t;
126
127#define COMPRESS_EXT_NUM 16
128
129/*
130 * An implementation of an rwsem that is explicitly unfair to readers. This
131 * prevents priority inversion when a low-priority reader acquires the read lock
132 * while sleeping on the write lock but the write lock is needed by
133 * higher-priority clients.
134 */
135
136struct f2fs_rwsem {
137 struct rw_semaphore internal_rwsem;
138#ifdef CONFIG_F2FS_UNFAIR_RWSEM
139 wait_queue_head_t read_waiters;
140#endif
141};
142
143struct f2fs_mount_info {
144 unsigned int opt;
145 int write_io_size_bits; /* Write IO size bits */
146 block_t root_reserved_blocks; /* root reserved blocks */
147 kuid_t s_resuid; /* reserved blocks for uid */
148 kgid_t s_resgid; /* reserved blocks for gid */
149 int active_logs; /* # of active logs */
150 int inline_xattr_size; /* inline xattr size */
151#ifdef CONFIG_F2FS_FAULT_INJECTION
152 struct f2fs_fault_info fault_info; /* For fault injection */
153#endif
154#ifdef CONFIG_QUOTA
155 /* Names of quota files with journalled quota */
156 char *s_qf_names[MAXQUOTAS];
157 int s_jquota_fmt; /* Format of quota to use */
158#endif
159 /* For which write hints are passed down to block layer */
160 int alloc_mode; /* segment allocation policy */
161 int fsync_mode; /* fsync policy */
162 int fs_mode; /* fs mode: LFS or ADAPTIVE */
163 int bggc_mode; /* bggc mode: off, on or sync */
164 int memory_mode; /* memory mode */
165 int errors; /* errors parameter */
166 int discard_unit; /*
167 * discard command's offset/size should
168 * be aligned to this unit: block,
169 * segment or section
170 */
171 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
172 block_t unusable_cap_perc; /* percentage for cap */
173 block_t unusable_cap; /* Amount of space allowed to be
174 * unusable when disabling checkpoint
175 */
176
177 /* For compression */
178 unsigned char compress_algorithm; /* algorithm type */
179 unsigned char compress_log_size; /* cluster log size */
180 unsigned char compress_level; /* compress level */
181 bool compress_chksum; /* compressed data chksum */
182 unsigned char compress_ext_cnt; /* extension count */
183 unsigned char nocompress_ext_cnt; /* nocompress extension count */
184 int compress_mode; /* compression mode */
185 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
186 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
187};
188
189#define F2FS_FEATURE_ENCRYPT 0x00000001
190#define F2FS_FEATURE_BLKZONED 0x00000002
191#define F2FS_FEATURE_ATOMIC_WRITE 0x00000004
192#define F2FS_FEATURE_EXTRA_ATTR 0x00000008
193#define F2FS_FEATURE_PRJQUOTA 0x00000010
194#define F2FS_FEATURE_INODE_CHKSUM 0x00000020
195#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040
196#define F2FS_FEATURE_QUOTA_INO 0x00000080
197#define F2FS_FEATURE_INODE_CRTIME 0x00000100
198#define F2FS_FEATURE_LOST_FOUND 0x00000200
199#define F2FS_FEATURE_VERITY 0x00000400
200#define F2FS_FEATURE_SB_CHKSUM 0x00000800
201#define F2FS_FEATURE_CASEFOLD 0x00001000
202#define F2FS_FEATURE_COMPRESSION 0x00002000
203#define F2FS_FEATURE_RO 0x00004000
204
205#define __F2FS_HAS_FEATURE(raw_super, mask) \
206 ((raw_super->feature & cpu_to_le32(mask)) != 0)
207#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
208
209/*
210 * Default values for user and/or group using reserved blocks
211 */
212#define F2FS_DEF_RESUID 0
213#define F2FS_DEF_RESGID 0
214
215/*
216 * For checkpoint manager
217 */
218enum {
219 NAT_BITMAP,
220 SIT_BITMAP
221};
222
223#define CP_UMOUNT 0x00000001
224#define CP_FASTBOOT 0x00000002
225#define CP_SYNC 0x00000004
226#define CP_RECOVERY 0x00000008
227#define CP_DISCARD 0x00000010
228#define CP_TRIMMED 0x00000020
229#define CP_PAUSE 0x00000040
230#define CP_RESIZE 0x00000080
231
232#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
233#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
234#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
235#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
236#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
237#define DEF_CP_INTERVAL 60 /* 60 secs */
238#define DEF_IDLE_INTERVAL 5 /* 5 secs */
239#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
240#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
241#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
242
243struct cp_control {
244 int reason;
245 __u64 trim_start;
246 __u64 trim_end;
247 __u64 trim_minlen;
248};
249
250/*
251 * indicate meta/data type
252 */
253enum {
254 META_CP,
255 META_NAT,
256 META_SIT,
257 META_SSA,
258 META_MAX,
259 META_POR,
260 DATA_GENERIC, /* check range only */
261 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
262 DATA_GENERIC_ENHANCE_READ, /*
263 * strong check on range and segment
264 * bitmap but no warning due to race
265 * condition of read on truncated area
266 * by extent_cache
267 */
268 DATA_GENERIC_ENHANCE_UPDATE, /*
269 * strong check on range and segment
270 * bitmap for update case
271 */
272 META_GENERIC,
273};
274
275/* for the list of ino */
276enum {
277 ORPHAN_INO, /* for orphan ino list */
278 APPEND_INO, /* for append ino list */
279 UPDATE_INO, /* for update ino list */
280 TRANS_DIR_INO, /* for transactions dir ino list */
281 FLUSH_INO, /* for multiple device flushing */
282 MAX_INO_ENTRY, /* max. list */
283};
284
285struct ino_entry {
286 struct list_head list; /* list head */
287 nid_t ino; /* inode number */
288 unsigned int dirty_device; /* dirty device bitmap */
289};
290
291/* for the list of inodes to be GCed */
292struct inode_entry {
293 struct list_head list; /* list head */
294 struct inode *inode; /* vfs inode pointer */
295};
296
297struct fsync_node_entry {
298 struct list_head list; /* list head */
299 struct page *page; /* warm node page pointer */
300 unsigned int seq_id; /* sequence id */
301};
302
303struct ckpt_req {
304 struct completion wait; /* completion for checkpoint done */
305 struct llist_node llnode; /* llist_node to be linked in wait queue */
306 int ret; /* return code of checkpoint */
307 ktime_t queue_time; /* request queued time */
308};
309
310struct ckpt_req_control {
311 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
312 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
313 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
314 atomic_t issued_ckpt; /* # of actually issued ckpts */
315 atomic_t total_ckpt; /* # of total ckpts */
316 atomic_t queued_ckpt; /* # of queued ckpts */
317 struct llist_head issue_list; /* list for command issue */
318 spinlock_t stat_lock; /* lock for below checkpoint time stats */
319 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
320 unsigned int peak_time; /* peak wait time in msec until now */
321};
322
323/* for the bitmap indicate blocks to be discarded */
324struct discard_entry {
325 struct list_head list; /* list head */
326 block_t start_blkaddr; /* start blockaddr of current segment */
327 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
328};
329
330/* minimum discard granularity, unit: block count */
331#define MIN_DISCARD_GRANULARITY 1
332/* default discard granularity of inner discard thread, unit: block count */
333#define DEFAULT_DISCARD_GRANULARITY 16
334/* default maximum discard granularity of ordered discard, unit: block count */
335#define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16
336
337/* max discard pend list number */
338#define MAX_PLIST_NUM 512
339#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
340 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
341
342enum {
343 D_PREP, /* initial */
344 D_PARTIAL, /* partially submitted */
345 D_SUBMIT, /* all submitted */
346 D_DONE, /* finished */
347};
348
349struct discard_info {
350 block_t lstart; /* logical start address */
351 block_t len; /* length */
352 block_t start; /* actual start address in dev */
353};
354
355struct discard_cmd {
356 struct rb_node rb_node; /* rb node located in rb-tree */
357 struct discard_info di; /* discard info */
358 struct list_head list; /* command list */
359 struct completion wait; /* compleation */
360 struct block_device *bdev; /* bdev */
361 unsigned short ref; /* reference count */
362 unsigned char state; /* state */
363 unsigned char queued; /* queued discard */
364 int error; /* bio error */
365 spinlock_t lock; /* for state/bio_ref updating */
366 unsigned short bio_ref; /* bio reference count */
367};
368
369enum {
370 DPOLICY_BG,
371 DPOLICY_FORCE,
372 DPOLICY_FSTRIM,
373 DPOLICY_UMOUNT,
374 MAX_DPOLICY,
375};
376
377struct discard_policy {
378 int type; /* type of discard */
379 unsigned int min_interval; /* used for candidates exist */
380 unsigned int mid_interval; /* used for device busy */
381 unsigned int max_interval; /* used for candidates not exist */
382 unsigned int max_requests; /* # of discards issued per round */
383 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
384 bool io_aware; /* issue discard in idle time */
385 bool sync; /* submit discard with REQ_SYNC flag */
386 bool ordered; /* issue discard by lba order */
387 bool timeout; /* discard timeout for put_super */
388 unsigned int granularity; /* discard granularity */
389};
390
391struct discard_cmd_control {
392 struct task_struct *f2fs_issue_discard; /* discard thread */
393 struct list_head entry_list; /* 4KB discard entry list */
394 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
395 struct list_head wait_list; /* store on-flushing entries */
396 struct list_head fstrim_list; /* in-flight discard from fstrim */
397 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
398 struct mutex cmd_lock;
399 unsigned int nr_discards; /* # of discards in the list */
400 unsigned int max_discards; /* max. discards to be issued */
401 unsigned int max_discard_request; /* max. discard request per round */
402 unsigned int min_discard_issue_time; /* min. interval between discard issue */
403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */
404 unsigned int max_discard_issue_time; /* max. interval between discard issue */
405 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */
406 unsigned int discard_urgent_util; /* utilization which issue discard proactively */
407 unsigned int discard_granularity; /* discard granularity */
408 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */
409 unsigned int undiscard_blks; /* # of undiscard blocks */
410 unsigned int next_pos; /* next discard position */
411 atomic_t issued_discard; /* # of issued discard */
412 atomic_t queued_discard; /* # of queued discard */
413 atomic_t discard_cmd_cnt; /* # of cached cmd count */
414 struct rb_root_cached root; /* root of discard rb-tree */
415 bool rbtree_check; /* config for consistence check */
416 bool discard_wake; /* to wake up discard thread */
417};
418
419/* for the list of fsync inodes, used only during recovery */
420struct fsync_inode_entry {
421 struct list_head list; /* list head */
422 struct inode *inode; /* vfs inode pointer */
423 block_t blkaddr; /* block address locating the last fsync */
424 block_t last_dentry; /* block address locating the last dentry */
425};
426
427#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
428#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
429
430#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
431#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
432#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
433#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
434
435#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
436#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
437
438static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
439{
440 int before = nats_in_cursum(journal);
441
442 journal->n_nats = cpu_to_le16(before + i);
443 return before;
444}
445
446static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
447{
448 int before = sits_in_cursum(journal);
449
450 journal->n_sits = cpu_to_le16(before + i);
451 return before;
452}
453
454static inline bool __has_cursum_space(struct f2fs_journal *journal,
455 int size, int type)
456{
457 if (type == NAT_JOURNAL)
458 return size <= MAX_NAT_JENTRIES(journal);
459 return size <= MAX_SIT_JENTRIES(journal);
460}
461
462/* for inline stuff */
463#define DEF_INLINE_RESERVED_SIZE 1
464static inline int get_extra_isize(struct inode *inode);
465static inline int get_inline_xattr_addrs(struct inode *inode);
466#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
467 (CUR_ADDRS_PER_INODE(inode) - \
468 get_inline_xattr_addrs(inode) - \
469 DEF_INLINE_RESERVED_SIZE))
470
471/* for inline dir */
472#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
473 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
474 BITS_PER_BYTE + 1))
475#define INLINE_DENTRY_BITMAP_SIZE(inode) \
476 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
477#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
478 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
479 NR_INLINE_DENTRY(inode) + \
480 INLINE_DENTRY_BITMAP_SIZE(inode)))
481
482/*
483 * For INODE and NODE manager
484 */
485/* for directory operations */
486
487struct f2fs_filename {
488 /*
489 * The filename the user specified. This is NULL for some
490 * filesystem-internal operations, e.g. converting an inline directory
491 * to a non-inline one, or roll-forward recovering an encrypted dentry.
492 */
493 const struct qstr *usr_fname;
494
495 /*
496 * The on-disk filename. For encrypted directories, this is encrypted.
497 * This may be NULL for lookups in an encrypted dir without the key.
498 */
499 struct fscrypt_str disk_name;
500
501 /* The dirhash of this filename */
502 f2fs_hash_t hash;
503
504#ifdef CONFIG_FS_ENCRYPTION
505 /*
506 * For lookups in encrypted directories: either the buffer backing
507 * disk_name, or a buffer that holds the decoded no-key name.
508 */
509 struct fscrypt_str crypto_buf;
510#endif
511#if IS_ENABLED(CONFIG_UNICODE)
512 /*
513 * For casefolded directories: the casefolded name, but it's left NULL
514 * if the original name is not valid Unicode, if the original name is
515 * "." or "..", if the directory is both casefolded and encrypted and
516 * its encryption key is unavailable, or if the filesystem is doing an
517 * internal operation where usr_fname is also NULL. In all these cases
518 * we fall back to treating the name as an opaque byte sequence.
519 */
520 struct fscrypt_str cf_name;
521#endif
522};
523
524struct f2fs_dentry_ptr {
525 struct inode *inode;
526 void *bitmap;
527 struct f2fs_dir_entry *dentry;
528 __u8 (*filename)[F2FS_SLOT_LEN];
529 int max;
530 int nr_bitmap;
531};
532
533static inline void make_dentry_ptr_block(struct inode *inode,
534 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
535{
536 d->inode = inode;
537 d->max = NR_DENTRY_IN_BLOCK;
538 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
539 d->bitmap = t->dentry_bitmap;
540 d->dentry = t->dentry;
541 d->filename = t->filename;
542}
543
544static inline void make_dentry_ptr_inline(struct inode *inode,
545 struct f2fs_dentry_ptr *d, void *t)
546{
547 int entry_cnt = NR_INLINE_DENTRY(inode);
548 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
549 int reserved_size = INLINE_RESERVED_SIZE(inode);
550
551 d->inode = inode;
552 d->max = entry_cnt;
553 d->nr_bitmap = bitmap_size;
554 d->bitmap = t;
555 d->dentry = t + bitmap_size + reserved_size;
556 d->filename = t + bitmap_size + reserved_size +
557 SIZE_OF_DIR_ENTRY * entry_cnt;
558}
559
560/*
561 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
562 * as its node offset to distinguish from index node blocks.
563 * But some bits are used to mark the node block.
564 */
565#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
566 >> OFFSET_BIT_SHIFT)
567enum {
568 ALLOC_NODE, /* allocate a new node page if needed */
569 LOOKUP_NODE, /* look up a node without readahead */
570 LOOKUP_NODE_RA, /*
571 * look up a node with readahead called
572 * by get_data_block.
573 */
574};
575
576#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
577
578/* congestion wait timeout value, default: 20ms */
579#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
580
581/* maximum retry quota flush count */
582#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
583
584/* maximum retry of EIO'ed page */
585#define MAX_RETRY_PAGE_EIO 100
586
587#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
588
589#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
590
591/* dirty segments threshold for triggering CP */
592#define DEFAULT_DIRTY_THRESHOLD 4
593
594#define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS
595#define RECOVERY_MIN_RA_BLOCKS 1
596
597#define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */
598
599/* for in-memory extent cache entry */
600#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
601
602/* number of extent info in extent cache we try to shrink */
603#define READ_EXTENT_CACHE_SHRINK_NUMBER 128
604
605/* number of age extent info in extent cache we try to shrink */
606#define AGE_EXTENT_CACHE_SHRINK_NUMBER 128
607#define LAST_AGE_WEIGHT 30
608#define SAME_AGE_REGION 1024
609
610/*
611 * Define data block with age less than 1GB as hot data
612 * define data block with age less than 10GB but more than 1GB as warm data
613 */
614#define DEF_HOT_DATA_AGE_THRESHOLD 262144
615#define DEF_WARM_DATA_AGE_THRESHOLD 2621440
616
617/* extent cache type */
618enum extent_type {
619 EX_READ,
620 EX_BLOCK_AGE,
621 NR_EXTENT_CACHES,
622};
623
624struct extent_info {
625 unsigned int fofs; /* start offset in a file */
626 unsigned int len; /* length of the extent */
627 union {
628 /* read extent_cache */
629 struct {
630 /* start block address of the extent */
631 block_t blk;
632#ifdef CONFIG_F2FS_FS_COMPRESSION
633 /* physical extent length of compressed blocks */
634 unsigned int c_len;
635#endif
636 };
637 /* block age extent_cache */
638 struct {
639 /* block age of the extent */
640 unsigned long long age;
641 /* last total blocks allocated */
642 unsigned long long last_blocks;
643 };
644 };
645};
646
647struct extent_node {
648 struct rb_node rb_node; /* rb node located in rb-tree */
649 struct extent_info ei; /* extent info */
650 struct list_head list; /* node in global extent list of sbi */
651 struct extent_tree *et; /* extent tree pointer */
652};
653
654struct extent_tree {
655 nid_t ino; /* inode number */
656 enum extent_type type; /* keep the extent tree type */
657 struct rb_root_cached root; /* root of extent info rb-tree */
658 struct extent_node *cached_en; /* recently accessed extent node */
659 struct list_head list; /* to be used by sbi->zombie_list */
660 rwlock_t lock; /* protect extent info rb-tree */
661 atomic_t node_cnt; /* # of extent node in rb-tree*/
662 bool largest_updated; /* largest extent updated */
663 struct extent_info largest; /* largest cached extent for EX_READ */
664};
665
666struct extent_tree_info {
667 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
668 struct mutex extent_tree_lock; /* locking extent radix tree */
669 struct list_head extent_list; /* lru list for shrinker */
670 spinlock_t extent_lock; /* locking extent lru list */
671 atomic_t total_ext_tree; /* extent tree count */
672 struct list_head zombie_list; /* extent zombie tree list */
673 atomic_t total_zombie_tree; /* extent zombie tree count */
674 atomic_t total_ext_node; /* extent info count */
675};
676
677/*
678 * State of block returned by f2fs_map_blocks.
679 */
680#define F2FS_MAP_NEW (1U << 0)
681#define F2FS_MAP_MAPPED (1U << 1)
682#define F2FS_MAP_DELALLOC (1U << 2)
683#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
684 F2FS_MAP_DELALLOC)
685
686struct f2fs_map_blocks {
687 struct block_device *m_bdev; /* for multi-device dio */
688 block_t m_pblk;
689 block_t m_lblk;
690 unsigned int m_len;
691 unsigned int m_flags;
692 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
693 pgoff_t *m_next_extent; /* point to next possible extent */
694 int m_seg_type;
695 bool m_may_create; /* indicate it is from write path */
696 bool m_multidev_dio; /* indicate it allows multi-device dio */
697};
698
699/* for flag in get_data_block */
700enum {
701 F2FS_GET_BLOCK_DEFAULT,
702 F2FS_GET_BLOCK_FIEMAP,
703 F2FS_GET_BLOCK_BMAP,
704 F2FS_GET_BLOCK_DIO,
705 F2FS_GET_BLOCK_PRE_DIO,
706 F2FS_GET_BLOCK_PRE_AIO,
707 F2FS_GET_BLOCK_PRECACHE,
708};
709
710/*
711 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
712 */
713#define FADVISE_COLD_BIT 0x01
714#define FADVISE_LOST_PINO_BIT 0x02
715#define FADVISE_ENCRYPT_BIT 0x04
716#define FADVISE_ENC_NAME_BIT 0x08
717#define FADVISE_KEEP_SIZE_BIT 0x10
718#define FADVISE_HOT_BIT 0x20
719#define FADVISE_VERITY_BIT 0x40
720#define FADVISE_TRUNC_BIT 0x80
721
722#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
723
724#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
725#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
726#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
727
728#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
729#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
730#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
731
732#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
733#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
734
735#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
736#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
737
738#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
739#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
740
741#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
742#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
743#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
744
745#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
746#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
747
748#define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)
749#define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)
750#define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)
751
752#define DEF_DIR_LEVEL 0
753
754enum {
755 GC_FAILURE_PIN,
756 MAX_GC_FAILURE
757};
758
759/* used for f2fs_inode_info->flags */
760enum {
761 FI_NEW_INODE, /* indicate newly allocated inode */
762 FI_DIRTY_INODE, /* indicate inode is dirty or not */
763 FI_AUTO_RECOVER, /* indicate inode is recoverable */
764 FI_DIRTY_DIR, /* indicate directory has dirty pages */
765 FI_INC_LINK, /* need to increment i_nlink */
766 FI_ACL_MODE, /* indicate acl mode */
767 FI_NO_ALLOC, /* should not allocate any blocks */
768 FI_FREE_NID, /* free allocated nide */
769 FI_NO_EXTENT, /* not to use the extent cache */
770 FI_INLINE_XATTR, /* used for inline xattr */
771 FI_INLINE_DATA, /* used for inline data*/
772 FI_INLINE_DENTRY, /* used for inline dentry */
773 FI_APPEND_WRITE, /* inode has appended data */
774 FI_UPDATE_WRITE, /* inode has in-place-update data */
775 FI_NEED_IPU, /* used for ipu per file */
776 FI_ATOMIC_FILE, /* indicate atomic file */
777 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */
778 FI_DROP_CACHE, /* drop dirty page cache */
779 FI_DATA_EXIST, /* indicate data exists */
780 FI_INLINE_DOTS, /* indicate inline dot dentries */
781 FI_SKIP_WRITES, /* should skip data page writeback */
782 FI_OPU_WRITE, /* used for opu per file */
783 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
784 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */
785 FI_HOT_DATA, /* indicate file is hot */
786 FI_EXTRA_ATTR, /* indicate file has extra attribute */
787 FI_PROJ_INHERIT, /* indicate file inherits projectid */
788 FI_PIN_FILE, /* indicate file should not be gced */
789 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
790 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
791 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
792 FI_MMAP_FILE, /* indicate file was mmapped */
793 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
794 FI_COMPRESS_RELEASED, /* compressed blocks were released */
795 FI_ALIGNED_WRITE, /* enable aligned write */
796 FI_COW_FILE, /* indicate COW file */
797 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */
798 FI_ATOMIC_REPLACE, /* indicate atomic replace */
799 FI_MAX, /* max flag, never be used */
800};
801
802struct f2fs_inode_info {
803 struct inode vfs_inode; /* serve a vfs inode */
804 unsigned long i_flags; /* keep an inode flags for ioctl */
805 unsigned char i_advise; /* use to give file attribute hints */
806 unsigned char i_dir_level; /* use for dentry level for large dir */
807 unsigned int i_current_depth; /* only for directory depth */
808 /* for gc failure statistic */
809 unsigned int i_gc_failures[MAX_GC_FAILURE];
810 unsigned int i_pino; /* parent inode number */
811 umode_t i_acl_mode; /* keep file acl mode temporarily */
812
813 /* Use below internally in f2fs*/
814 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
815 struct f2fs_rwsem i_sem; /* protect fi info */
816 atomic_t dirty_pages; /* # of dirty pages */
817 f2fs_hash_t chash; /* hash value of given file name */
818 unsigned int clevel; /* maximum level of given file name */
819 struct task_struct *task; /* lookup and create consistency */
820 struct task_struct *cp_task; /* separate cp/wb IO stats*/
821 struct task_struct *wb_task; /* indicate inode is in context of writeback */
822 nid_t i_xattr_nid; /* node id that contains xattrs */
823 loff_t last_disk_size; /* lastly written file size */
824 spinlock_t i_size_lock; /* protect last_disk_size */
825
826#ifdef CONFIG_QUOTA
827 struct dquot *i_dquot[MAXQUOTAS];
828
829 /* quota space reservation, managed internally by quota code */
830 qsize_t i_reserved_quota;
831#endif
832 struct list_head dirty_list; /* dirty list for dirs and files */
833 struct list_head gdirty_list; /* linked in global dirty list */
834 struct task_struct *atomic_write_task; /* store atomic write task */
835 struct extent_tree *extent_tree[NR_EXTENT_CACHES];
836 /* cached extent_tree entry */
837 struct inode *cow_inode; /* copy-on-write inode for atomic write */
838
839 /* avoid racing between foreground op and gc */
840 struct f2fs_rwsem i_gc_rwsem[2];
841 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */
842
843 int i_extra_isize; /* size of extra space located in i_addr */
844 kprojid_t i_projid; /* id for project quota */
845 int i_inline_xattr_size; /* inline xattr size */
846 struct timespec64 i_crtime; /* inode creation time */
847 struct timespec64 i_disk_time[3];/* inode disk times */
848
849 /* for file compress */
850 atomic_t i_compr_blocks; /* # of compressed blocks */
851 unsigned char i_compress_algorithm; /* algorithm type */
852 unsigned char i_log_cluster_size; /* log of cluster size */
853 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
854 unsigned char i_compress_flag; /* compress flag */
855 unsigned int i_cluster_size; /* cluster size */
856
857 unsigned int atomic_write_cnt;
858 loff_t original_i_size; /* original i_size before atomic write */
859};
860
861static inline void get_read_extent_info(struct extent_info *ext,
862 struct f2fs_extent *i_ext)
863{
864 ext->fofs = le32_to_cpu(i_ext->fofs);
865 ext->blk = le32_to_cpu(i_ext->blk);
866 ext->len = le32_to_cpu(i_ext->len);
867}
868
869static inline void set_raw_read_extent(struct extent_info *ext,
870 struct f2fs_extent *i_ext)
871{
872 i_ext->fofs = cpu_to_le32(ext->fofs);
873 i_ext->blk = cpu_to_le32(ext->blk);
874 i_ext->len = cpu_to_le32(ext->len);
875}
876
877static inline bool __is_discard_mergeable(struct discard_info *back,
878 struct discard_info *front, unsigned int max_len)
879{
880 return (back->lstart + back->len == front->lstart) &&
881 (back->len + front->len <= max_len);
882}
883
884static inline bool __is_discard_back_mergeable(struct discard_info *cur,
885 struct discard_info *back, unsigned int max_len)
886{
887 return __is_discard_mergeable(back, front: cur, max_len);
888}
889
890static inline bool __is_discard_front_mergeable(struct discard_info *cur,
891 struct discard_info *front, unsigned int max_len)
892{
893 return __is_discard_mergeable(back: cur, front, max_len);
894}
895
896/*
897 * For free nid management
898 */
899enum nid_state {
900 FREE_NID, /* newly added to free nid list */
901 PREALLOC_NID, /* it is preallocated */
902 MAX_NID_STATE,
903};
904
905enum nat_state {
906 TOTAL_NAT,
907 DIRTY_NAT,
908 RECLAIMABLE_NAT,
909 MAX_NAT_STATE,
910};
911
912struct f2fs_nm_info {
913 block_t nat_blkaddr; /* base disk address of NAT */
914 nid_t max_nid; /* maximum possible node ids */
915 nid_t available_nids; /* # of available node ids */
916 nid_t next_scan_nid; /* the next nid to be scanned */
917 nid_t max_rf_node_blocks; /* max # of nodes for recovery */
918 unsigned int ram_thresh; /* control the memory footprint */
919 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
920 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
921
922 /* NAT cache management */
923 struct radix_tree_root nat_root;/* root of the nat entry cache */
924 struct radix_tree_root nat_set_root;/* root of the nat set cache */
925 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */
926 struct list_head nat_entries; /* cached nat entry list (clean) */
927 spinlock_t nat_list_lock; /* protect clean nat entry list */
928 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
929 unsigned int nat_blocks; /* # of nat blocks */
930
931 /* free node ids management */
932 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
933 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
934 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
935 spinlock_t nid_list_lock; /* protect nid lists ops */
936 struct mutex build_lock; /* lock for build free nids */
937 unsigned char **free_nid_bitmap;
938 unsigned char *nat_block_bitmap;
939 unsigned short *free_nid_count; /* free nid count of NAT block */
940
941 /* for checkpoint */
942 char *nat_bitmap; /* NAT bitmap pointer */
943
944 unsigned int nat_bits_blocks; /* # of nat bits blocks */
945 unsigned char *nat_bits; /* NAT bits blocks */
946 unsigned char *full_nat_bits; /* full NAT pages */
947 unsigned char *empty_nat_bits; /* empty NAT pages */
948#ifdef CONFIG_F2FS_CHECK_FS
949 char *nat_bitmap_mir; /* NAT bitmap mirror */
950#endif
951 int bitmap_size; /* bitmap size */
952};
953
954/*
955 * this structure is used as one of function parameters.
956 * all the information are dedicated to a given direct node block determined
957 * by the data offset in a file.
958 */
959struct dnode_of_data {
960 struct inode *inode; /* vfs inode pointer */
961 struct page *inode_page; /* its inode page, NULL is possible */
962 struct page *node_page; /* cached direct node page */
963 nid_t nid; /* node id of the direct node block */
964 unsigned int ofs_in_node; /* data offset in the node page */
965 bool inode_page_locked; /* inode page is locked or not */
966 bool node_changed; /* is node block changed */
967 char cur_level; /* level of hole node page */
968 char max_level; /* level of current page located */
969 block_t data_blkaddr; /* block address of the node block */
970};
971
972static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
973 struct page *ipage, struct page *npage, nid_t nid)
974{
975 memset(dn, 0, sizeof(*dn));
976 dn->inode = inode;
977 dn->inode_page = ipage;
978 dn->node_page = npage;
979 dn->nid = nid;
980}
981
982/*
983 * For SIT manager
984 *
985 * By default, there are 6 active log areas across the whole main area.
986 * When considering hot and cold data separation to reduce cleaning overhead,
987 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
988 * respectively.
989 * In the current design, you should not change the numbers intentionally.
990 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
991 * logs individually according to the underlying devices. (default: 6)
992 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
993 * data and 8 for node logs.
994 */
995#define NR_CURSEG_DATA_TYPE (3)
996#define NR_CURSEG_NODE_TYPE (3)
997#define NR_CURSEG_INMEM_TYPE (2)
998#define NR_CURSEG_RO_TYPE (2)
999#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
1000#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
1001
1002enum {
1003 CURSEG_HOT_DATA = 0, /* directory entry blocks */
1004 CURSEG_WARM_DATA, /* data blocks */
1005 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
1006 CURSEG_HOT_NODE, /* direct node blocks of directory files */
1007 CURSEG_WARM_NODE, /* direct node blocks of normal files */
1008 CURSEG_COLD_NODE, /* indirect node blocks */
1009 NR_PERSISTENT_LOG, /* number of persistent log */
1010 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
1011 /* pinned file that needs consecutive block address */
1012 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
1013 NO_CHECK_TYPE, /* number of persistent & inmem log */
1014};
1015
1016struct flush_cmd {
1017 struct completion wait;
1018 struct llist_node llnode;
1019 nid_t ino;
1020 int ret;
1021};
1022
1023struct flush_cmd_control {
1024 struct task_struct *f2fs_issue_flush; /* flush thread */
1025 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
1026 atomic_t issued_flush; /* # of issued flushes */
1027 atomic_t queued_flush; /* # of queued flushes */
1028 struct llist_head issue_list; /* list for command issue */
1029 struct llist_node *dispatch_list; /* list for command dispatch */
1030};
1031
1032struct f2fs_sm_info {
1033 struct sit_info *sit_info; /* whole segment information */
1034 struct free_segmap_info *free_info; /* free segment information */
1035 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1036 struct curseg_info *curseg_array; /* active segment information */
1037
1038 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */
1039
1040 block_t seg0_blkaddr; /* block address of 0'th segment */
1041 block_t main_blkaddr; /* start block address of main area */
1042 block_t ssa_blkaddr; /* start block address of SSA area */
1043
1044 unsigned int segment_count; /* total # of segments */
1045 unsigned int main_segments; /* # of segments in main area */
1046 unsigned int reserved_segments; /* # of reserved segments */
1047 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1048 unsigned int ovp_segments; /* # of overprovision segments */
1049
1050 /* a threshold to reclaim prefree segments */
1051 unsigned int rec_prefree_segments;
1052
1053 struct list_head sit_entry_set; /* sit entry set list */
1054
1055 unsigned int ipu_policy; /* in-place-update policy */
1056 unsigned int min_ipu_util; /* in-place-update threshold */
1057 unsigned int min_fsync_blocks; /* threshold for fsync */
1058 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1059 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1060 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1061
1062 /* for flush command control */
1063 struct flush_cmd_control *fcc_info;
1064
1065 /* for discard command control */
1066 struct discard_cmd_control *dcc_info;
1067};
1068
1069/*
1070 * For superblock
1071 */
1072/*
1073 * COUNT_TYPE for monitoring
1074 *
1075 * f2fs monitors the number of several block types such as on-writeback,
1076 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1077 */
1078#define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1079enum count_type {
1080 F2FS_DIRTY_DENTS,
1081 F2FS_DIRTY_DATA,
1082 F2FS_DIRTY_QDATA,
1083 F2FS_DIRTY_NODES,
1084 F2FS_DIRTY_META,
1085 F2FS_DIRTY_IMETA,
1086 F2FS_WB_CP_DATA,
1087 F2FS_WB_DATA,
1088 F2FS_RD_DATA,
1089 F2FS_RD_NODE,
1090 F2FS_RD_META,
1091 F2FS_DIO_WRITE,
1092 F2FS_DIO_READ,
1093 NR_COUNT_TYPE,
1094};
1095
1096/*
1097 * The below are the page types of bios used in submit_bio().
1098 * The available types are:
1099 * DATA User data pages. It operates as async mode.
1100 * NODE Node pages. It operates as async mode.
1101 * META FS metadata pages such as SIT, NAT, CP.
1102 * NR_PAGE_TYPE The number of page types.
1103 * META_FLUSH Make sure the previous pages are written
1104 * with waiting the bio's completion
1105 * ... Only can be used with META.
1106 */
1107#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1108enum page_type {
1109 DATA = 0,
1110 NODE = 1, /* should not change this */
1111 META,
1112 NR_PAGE_TYPE,
1113 META_FLUSH,
1114 IPU, /* the below types are used by tracepoints only. */
1115 OPU,
1116};
1117
1118enum temp_type {
1119 HOT = 0, /* must be zero for meta bio */
1120 WARM,
1121 COLD,
1122 NR_TEMP_TYPE,
1123};
1124
1125enum need_lock_type {
1126 LOCK_REQ = 0,
1127 LOCK_DONE,
1128 LOCK_RETRY,
1129};
1130
1131enum cp_reason_type {
1132 CP_NO_NEEDED,
1133 CP_NON_REGULAR,
1134 CP_COMPRESSED,
1135 CP_HARDLINK,
1136 CP_SB_NEED_CP,
1137 CP_WRONG_PINO,
1138 CP_NO_SPC_ROLL,
1139 CP_NODE_NEED_CP,
1140 CP_FASTBOOT_MODE,
1141 CP_SPEC_LOG_NUM,
1142 CP_RECOVER_DIR,
1143};
1144
1145enum iostat_type {
1146 /* WRITE IO */
1147 APP_DIRECT_IO, /* app direct write IOs */
1148 APP_BUFFERED_IO, /* app buffered write IOs */
1149 APP_WRITE_IO, /* app write IOs */
1150 APP_MAPPED_IO, /* app mapped IOs */
1151 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */
1152 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */
1153 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1154 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */
1155 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1156 FS_META_IO, /* meta IOs from kworker/reclaimer */
1157 FS_GC_DATA_IO, /* data IOs from forground gc */
1158 FS_GC_NODE_IO, /* node IOs from forground gc */
1159 FS_CP_DATA_IO, /* data IOs from checkpoint */
1160 FS_CP_NODE_IO, /* node IOs from checkpoint */
1161 FS_CP_META_IO, /* meta IOs from checkpoint */
1162
1163 /* READ IO */
1164 APP_DIRECT_READ_IO, /* app direct read IOs */
1165 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1166 APP_READ_IO, /* app read IOs */
1167 APP_MAPPED_READ_IO, /* app mapped read IOs */
1168 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */
1169 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */
1170 FS_DATA_READ_IO, /* data read IOs */
1171 FS_GDATA_READ_IO, /* data read IOs from background gc */
1172 FS_CDATA_READ_IO, /* compressed data read IOs */
1173 FS_NODE_READ_IO, /* node read IOs */
1174 FS_META_READ_IO, /* meta read IOs */
1175
1176 /* other */
1177 FS_DISCARD_IO, /* discard */
1178 FS_FLUSH_IO, /* flush */
1179 FS_ZONE_RESET_IO, /* zone reset */
1180 NR_IO_TYPE,
1181};
1182
1183struct f2fs_io_info {
1184 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1185 nid_t ino; /* inode number */
1186 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1187 enum temp_type temp; /* contains HOT/WARM/COLD */
1188 enum req_op op; /* contains REQ_OP_ */
1189 blk_opf_t op_flags; /* req_flag_bits */
1190 block_t new_blkaddr; /* new block address to be written */
1191 block_t old_blkaddr; /* old block address before Cow */
1192 struct page *page; /* page to be written */
1193 struct page *encrypted_page; /* encrypted page */
1194 struct page *compressed_page; /* compressed page */
1195 struct list_head list; /* serialize IOs */
1196 unsigned int compr_blocks; /* # of compressed block addresses */
1197 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */
1198 unsigned int version:8; /* version of the node */
1199 unsigned int submitted:1; /* indicate IO submission */
1200 unsigned int in_list:1; /* indicate fio is in io_list */
1201 unsigned int is_por:1; /* indicate IO is from recovery or not */
1202 unsigned int retry:1; /* need to reallocate block address */
1203 unsigned int encrypted:1; /* indicate file is encrypted */
1204 unsigned int post_read:1; /* require post read */
1205 enum iostat_type io_type; /* io type */
1206 struct writeback_control *io_wbc; /* writeback control */
1207 struct bio **bio; /* bio for ipu */
1208 sector_t *last_block; /* last block number in bio */
1209};
1210
1211struct bio_entry {
1212 struct bio *bio;
1213 struct list_head list;
1214};
1215
1216#define is_read_io(rw) ((rw) == READ)
1217struct f2fs_bio_info {
1218 struct f2fs_sb_info *sbi; /* f2fs superblock */
1219 struct bio *bio; /* bios to merge */
1220 sector_t last_block_in_bio; /* last block number */
1221 struct f2fs_io_info fio; /* store buffered io info. */
1222#ifdef CONFIG_BLK_DEV_ZONED
1223 struct completion zone_wait; /* condition value for the previous open zone to close */
1224 struct bio *zone_pending_bio; /* pending bio for the previous zone */
1225 void *bi_private; /* previous bi_private for pending bio */
1226#endif
1227 struct f2fs_rwsem io_rwsem; /* blocking op for bio */
1228 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1229 struct list_head io_list; /* track fios */
1230 struct list_head bio_list; /* bio entry list head */
1231 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */
1232};
1233
1234#define FDEV(i) (sbi->devs[i])
1235#define RDEV(i) (raw_super->devs[i])
1236struct f2fs_dev_info {
1237 struct bdev_handle *bdev_handle;
1238 struct block_device *bdev;
1239 char path[MAX_PATH_LEN];
1240 unsigned int total_segments;
1241 block_t start_blk;
1242 block_t end_blk;
1243#ifdef CONFIG_BLK_DEV_ZONED
1244 unsigned int nr_blkz; /* Total number of zones */
1245 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1246#endif
1247};
1248
1249enum inode_type {
1250 DIR_INODE, /* for dirty dir inode */
1251 FILE_INODE, /* for dirty regular/symlink inode */
1252 DIRTY_META, /* for all dirtied inode metadata */
1253 NR_INODE_TYPE,
1254};
1255
1256/* for inner inode cache management */
1257struct inode_management {
1258 struct radix_tree_root ino_root; /* ino entry array */
1259 spinlock_t ino_lock; /* for ino entry lock */
1260 struct list_head ino_list; /* inode list head */
1261 unsigned long ino_num; /* number of entries */
1262};
1263
1264/* for GC_AT */
1265struct atgc_management {
1266 bool atgc_enabled; /* ATGC is enabled or not */
1267 struct rb_root_cached root; /* root of victim rb-tree */
1268 struct list_head victim_list; /* linked with all victim entries */
1269 unsigned int victim_count; /* victim count in rb-tree */
1270 unsigned int candidate_ratio; /* candidate ratio */
1271 unsigned int max_candidate_count; /* max candidate count */
1272 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1273 unsigned long long age_threshold; /* age threshold */
1274};
1275
1276struct f2fs_gc_control {
1277 unsigned int victim_segno; /* target victim segment number */
1278 int init_gc_type; /* FG_GC or BG_GC */
1279 bool no_bg_gc; /* check the space and stop bg_gc */
1280 bool should_migrate_blocks; /* should migrate blocks */
1281 bool err_gc_skipped; /* return EAGAIN if GC skipped */
1282 unsigned int nr_free_secs; /* # of free sections to do GC */
1283};
1284
1285/*
1286 * For s_flag in struct f2fs_sb_info
1287 * Modification on enum should be synchronized with s_flag array
1288 */
1289enum {
1290 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1291 SBI_IS_CLOSE, /* specify unmounting */
1292 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1293 SBI_POR_DOING, /* recovery is doing or not */
1294 SBI_NEED_SB_WRITE, /* need to recover superblock */
1295 SBI_NEED_CP, /* need to checkpoint */
1296 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1297 SBI_IS_RECOVERED, /* recovered orphan/data */
1298 SBI_CP_DISABLED, /* CP was disabled last mount */
1299 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1300 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1301 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1302 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1303 SBI_IS_RESIZEFS, /* resizefs is in process */
1304 SBI_IS_FREEZING, /* freezefs is in process */
1305 SBI_IS_WRITABLE, /* remove ro mountoption transiently */
1306 MAX_SBI_FLAG,
1307};
1308
1309enum {
1310 CP_TIME,
1311 REQ_TIME,
1312 DISCARD_TIME,
1313 GC_TIME,
1314 DISABLE_TIME,
1315 UMOUNT_DISCARD_TIMEOUT,
1316 MAX_TIME,
1317};
1318
1319/* Note that you need to keep synchronization with this gc_mode_names array */
1320enum {
1321 GC_NORMAL,
1322 GC_IDLE_CB,
1323 GC_IDLE_GREEDY,
1324 GC_IDLE_AT,
1325 GC_URGENT_HIGH,
1326 GC_URGENT_LOW,
1327 GC_URGENT_MID,
1328 MAX_GC_MODE,
1329};
1330
1331enum {
1332 BGGC_MODE_ON, /* background gc is on */
1333 BGGC_MODE_OFF, /* background gc is off */
1334 BGGC_MODE_SYNC, /*
1335 * background gc is on, migrating blocks
1336 * like foreground gc
1337 */
1338};
1339
1340enum {
1341 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1342 FS_MODE_LFS, /* use lfs allocation only */
1343 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */
1344 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */
1345};
1346
1347enum {
1348 ALLOC_MODE_DEFAULT, /* stay default */
1349 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1350};
1351
1352enum fsync_mode {
1353 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1354 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1355 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1356};
1357
1358enum {
1359 COMPR_MODE_FS, /*
1360 * automatically compress compression
1361 * enabled files
1362 */
1363 COMPR_MODE_USER, /*
1364 * automatical compression is disabled.
1365 * user can control the file compression
1366 * using ioctls
1367 */
1368};
1369
1370enum {
1371 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1372 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1373 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1374};
1375
1376enum {
1377 MEMORY_MODE_NORMAL, /* memory mode for normal devices */
1378 MEMORY_MODE_LOW, /* memory mode for low memry devices */
1379};
1380
1381enum errors_option {
1382 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */
1383 MOUNT_ERRORS_CONTINUE, /* continue on errors */
1384 MOUNT_ERRORS_PANIC, /* panic on errors */
1385};
1386
1387enum {
1388 BACKGROUND,
1389 FOREGROUND,
1390 MAX_CALL_TYPE,
1391 TOTAL_CALL = FOREGROUND,
1392};
1393
1394static inline int f2fs_test_bit(unsigned int nr, char *addr);
1395static inline void f2fs_set_bit(unsigned int nr, char *addr);
1396static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1397
1398/*
1399 * Layout of f2fs page.private:
1400 *
1401 * Layout A: lowest bit should be 1
1402 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1403 * bit 0 PAGE_PRIVATE_NOT_POINTER
1404 * bit 1 PAGE_PRIVATE_DUMMY_WRITE
1405 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION
1406 * bit 3 PAGE_PRIVATE_INLINE_INODE
1407 * bit 4 PAGE_PRIVATE_REF_RESOURCE
1408 * bit 5- f2fs private data
1409 *
1410 * Layout B: lowest bit should be 0
1411 * page.private is a wrapped pointer.
1412 */
1413enum {
1414 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1415 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */
1416 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1417 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1418 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1419 PAGE_PRIVATE_MAX
1420};
1421
1422/* For compression */
1423enum compress_algorithm_type {
1424 COMPRESS_LZO,
1425 COMPRESS_LZ4,
1426 COMPRESS_ZSTD,
1427 COMPRESS_LZORLE,
1428 COMPRESS_MAX,
1429};
1430
1431enum compress_flag {
1432 COMPRESS_CHKSUM,
1433 COMPRESS_MAX_FLAG,
1434};
1435
1436#define COMPRESS_WATERMARK 20
1437#define COMPRESS_PERCENT 20
1438
1439#define COMPRESS_DATA_RESERVED_SIZE 4
1440struct compress_data {
1441 __le32 clen; /* compressed data size */
1442 __le32 chksum; /* compressed data chksum */
1443 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1444 u8 cdata[]; /* compressed data */
1445};
1446
1447#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1448
1449#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1450
1451#define F2FS_ZSTD_DEFAULT_CLEVEL 1
1452
1453#define COMPRESS_LEVEL_OFFSET 8
1454
1455/* compress context */
1456struct compress_ctx {
1457 struct inode *inode; /* inode the context belong to */
1458 pgoff_t cluster_idx; /* cluster index number */
1459 unsigned int cluster_size; /* page count in cluster */
1460 unsigned int log_cluster_size; /* log of cluster size */
1461 struct page **rpages; /* pages store raw data in cluster */
1462 unsigned int nr_rpages; /* total page number in rpages */
1463 struct page **cpages; /* pages store compressed data in cluster */
1464 unsigned int nr_cpages; /* total page number in cpages */
1465 unsigned int valid_nr_cpages; /* valid page number in cpages */
1466 void *rbuf; /* virtual mapped address on rpages */
1467 struct compress_data *cbuf; /* virtual mapped address on cpages */
1468 size_t rlen; /* valid data length in rbuf */
1469 size_t clen; /* valid data length in cbuf */
1470 void *private; /* payload buffer for specified compression algorithm */
1471 void *private2; /* extra payload buffer */
1472};
1473
1474/* compress context for write IO path */
1475struct compress_io_ctx {
1476 u32 magic; /* magic number to indicate page is compressed */
1477 struct inode *inode; /* inode the context belong to */
1478 struct page **rpages; /* pages store raw data in cluster */
1479 unsigned int nr_rpages; /* total page number in rpages */
1480 atomic_t pending_pages; /* in-flight compressed page count */
1481};
1482
1483/* Context for decompressing one cluster on the read IO path */
1484struct decompress_io_ctx {
1485 u32 magic; /* magic number to indicate page is compressed */
1486 struct inode *inode; /* inode the context belong to */
1487 pgoff_t cluster_idx; /* cluster index number */
1488 unsigned int cluster_size; /* page count in cluster */
1489 unsigned int log_cluster_size; /* log of cluster size */
1490 struct page **rpages; /* pages store raw data in cluster */
1491 unsigned int nr_rpages; /* total page number in rpages */
1492 struct page **cpages; /* pages store compressed data in cluster */
1493 unsigned int nr_cpages; /* total page number in cpages */
1494 struct page **tpages; /* temp pages to pad holes in cluster */
1495 void *rbuf; /* virtual mapped address on rpages */
1496 struct compress_data *cbuf; /* virtual mapped address on cpages */
1497 size_t rlen; /* valid data length in rbuf */
1498 size_t clen; /* valid data length in cbuf */
1499
1500 /*
1501 * The number of compressed pages remaining to be read in this cluster.
1502 * This is initially nr_cpages. It is decremented by 1 each time a page
1503 * has been read (or failed to be read). When it reaches 0, the cluster
1504 * is decompressed (or an error is reported).
1505 *
1506 * If an error occurs before all the pages have been submitted for I/O,
1507 * then this will never reach 0. In this case the I/O submitter is
1508 * responsible for calling f2fs_decompress_end_io() instead.
1509 */
1510 atomic_t remaining_pages;
1511
1512 /*
1513 * Number of references to this decompress_io_ctx.
1514 *
1515 * One reference is held for I/O completion. This reference is dropped
1516 * after the pagecache pages are updated and unlocked -- either after
1517 * decompression (and verity if enabled), or after an error.
1518 *
1519 * In addition, each compressed page holds a reference while it is in a
1520 * bio. These references are necessary prevent compressed pages from
1521 * being freed while they are still in a bio.
1522 */
1523 refcount_t refcnt;
1524
1525 bool failed; /* IO error occurred before decompression? */
1526 bool need_verity; /* need fs-verity verification after decompression? */
1527 void *private; /* payload buffer for specified decompression algorithm */
1528 void *private2; /* extra payload buffer */
1529 struct work_struct verity_work; /* work to verify the decompressed pages */
1530 struct work_struct free_work; /* work for late free this structure itself */
1531};
1532
1533#define NULL_CLUSTER ((unsigned int)(~0))
1534#define MIN_COMPRESS_LOG_SIZE 2
1535#define MAX_COMPRESS_LOG_SIZE 8
1536#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1537
1538struct f2fs_sb_info {
1539 struct super_block *sb; /* pointer to VFS super block */
1540 struct proc_dir_entry *s_proc; /* proc entry */
1541 struct f2fs_super_block *raw_super; /* raw super block pointer */
1542 struct f2fs_rwsem sb_lock; /* lock for raw super block */
1543 int valid_super_block; /* valid super block no */
1544 unsigned long s_flag; /* flags for sbi */
1545 struct mutex writepages; /* mutex for writepages() */
1546
1547#ifdef CONFIG_BLK_DEV_ZONED
1548 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1549#endif
1550
1551 /* for node-related operations */
1552 struct f2fs_nm_info *nm_info; /* node manager */
1553 struct inode *node_inode; /* cache node blocks */
1554
1555 /* for segment-related operations */
1556 struct f2fs_sm_info *sm_info; /* segment manager */
1557
1558 /* for bio operations */
1559 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1560 /* keep migration IO order for LFS mode */
1561 struct f2fs_rwsem io_order_lock;
1562 mempool_t *write_io_dummy; /* Dummy pages */
1563 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */
1564 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */
1565
1566 /* for checkpoint */
1567 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1568 int cur_cp_pack; /* remain current cp pack */
1569 spinlock_t cp_lock; /* for flag in ckpt */
1570 struct inode *meta_inode; /* cache meta blocks */
1571 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */
1572 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */
1573 struct f2fs_rwsem node_write; /* locking node writes */
1574 struct f2fs_rwsem node_change; /* locking node change */
1575 wait_queue_head_t cp_wait;
1576 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1577 long interval_time[MAX_TIME]; /* to store thresholds */
1578 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1579
1580 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1581
1582 spinlock_t fsync_node_lock; /* for node entry lock */
1583 struct list_head fsync_node_list; /* node list head */
1584 unsigned int fsync_seg_id; /* sequence id */
1585 unsigned int fsync_node_num; /* number of node entries */
1586
1587 /* for orphan inode, use 0'th array */
1588 unsigned int max_orphans; /* max orphan inodes */
1589
1590 /* for inode management */
1591 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1592 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1593 struct mutex flush_lock; /* for flush exclusion */
1594
1595 /* for extent tree cache */
1596 struct extent_tree_info extent_tree[NR_EXTENT_CACHES];
1597 atomic64_t allocated_data_blocks; /* for block age extent_cache */
1598
1599 /* The threshold used for hot and warm data seperation*/
1600 unsigned int hot_data_age_threshold;
1601 unsigned int warm_data_age_threshold;
1602 unsigned int last_age_weight;
1603
1604 /* basic filesystem units */
1605 unsigned int log_sectors_per_block; /* log2 sectors per block */
1606 unsigned int log_blocksize; /* log2 block size */
1607 unsigned int blocksize; /* block size */
1608 unsigned int root_ino_num; /* root inode number*/
1609 unsigned int node_ino_num; /* node inode number*/
1610 unsigned int meta_ino_num; /* meta inode number*/
1611 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1612 unsigned int blocks_per_seg; /* blocks per segment */
1613 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1614 unsigned int segs_per_sec; /* segments per section */
1615 unsigned int secs_per_zone; /* sections per zone */
1616 unsigned int total_sections; /* total section count */
1617 unsigned int total_node_count; /* total node block count */
1618 unsigned int total_valid_node_count; /* valid node block count */
1619 int dir_level; /* directory level */
1620 bool readdir_ra; /* readahead inode in readdir */
1621 u64 max_io_bytes; /* max io bytes to merge IOs */
1622
1623 block_t user_block_count; /* # of user blocks */
1624 block_t total_valid_block_count; /* # of valid blocks */
1625 block_t discard_blks; /* discard command candidats */
1626 block_t last_valid_block_count; /* for recovery */
1627 block_t reserved_blocks; /* configurable reserved blocks */
1628 block_t current_reserved_blocks; /* current reserved blocks */
1629
1630 /* Additional tracking for no checkpoint mode */
1631 block_t unusable_block_count; /* # of blocks saved by last cp */
1632
1633 unsigned int nquota_files; /* # of quota sysfile */
1634 struct f2fs_rwsem quota_sem; /* blocking cp for flags */
1635
1636 /* # of pages, see count_type */
1637 atomic_t nr_pages[NR_COUNT_TYPE];
1638 /* # of allocated blocks */
1639 struct percpu_counter alloc_valid_block_count;
1640 /* # of node block writes as roll forward recovery */
1641 struct percpu_counter rf_node_block_count;
1642
1643 /* writeback control */
1644 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1645
1646 /* valid inode count */
1647 struct percpu_counter total_valid_inode_count;
1648
1649 struct f2fs_mount_info mount_opt; /* mount options */
1650
1651 /* for cleaning operations */
1652 struct f2fs_rwsem gc_lock; /*
1653 * semaphore for GC, avoid
1654 * race between GC and GC or CP
1655 */
1656 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1657 struct atgc_management am; /* atgc management */
1658 unsigned int cur_victim_sec; /* current victim section num */
1659 unsigned int gc_mode; /* current GC state */
1660 unsigned int next_victim_seg[2]; /* next segment in victim section */
1661 spinlock_t gc_remaining_trials_lock;
1662 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */
1663 unsigned int gc_remaining_trials;
1664
1665 /* for skip statistic */
1666 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1667
1668 /* threshold for gc trials on pinned files */
1669 u64 gc_pin_file_threshold;
1670 struct f2fs_rwsem pin_sem;
1671
1672 /* maximum # of trials to find a victim segment for SSR and GC */
1673 unsigned int max_victim_search;
1674 /* migration granularity of garbage collection, unit: segment */
1675 unsigned int migration_granularity;
1676
1677 /*
1678 * for stat information.
1679 * one is for the LFS mode, and the other is for the SSR mode.
1680 */
1681#ifdef CONFIG_F2FS_STAT_FS
1682 struct f2fs_stat_info *stat_info; /* FS status information */
1683 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1684 unsigned int segment_count[2]; /* # of allocated segments */
1685 unsigned int block_count[2]; /* # of allocated blocks */
1686 atomic_t inplace_count; /* # of inplace update */
1687 /* # of lookup extent cache */
1688 atomic64_t total_hit_ext[NR_EXTENT_CACHES];
1689 /* # of hit rbtree extent node */
1690 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES];
1691 /* # of hit cached extent node */
1692 atomic64_t read_hit_cached[NR_EXTENT_CACHES];
1693 /* # of hit largest extent node in read extent cache */
1694 atomic64_t read_hit_largest;
1695 atomic_t inline_xattr; /* # of inline_xattr inodes */
1696 atomic_t inline_inode; /* # of inline_data inodes */
1697 atomic_t inline_dir; /* # of inline_dentry inodes */
1698 atomic_t compr_inode; /* # of compressed inodes */
1699 atomic64_t compr_blocks; /* # of compressed blocks */
1700 atomic_t swapfile_inode; /* # of swapfile inodes */
1701 atomic_t atomic_files; /* # of opened atomic file */
1702 atomic_t max_aw_cnt; /* max # of atomic writes */
1703 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1704 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1705 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1706 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */
1707#endif
1708 spinlock_t stat_lock; /* lock for stat operations */
1709
1710 /* to attach REQ_META|REQ_FUA flags */
1711 unsigned int data_io_flag;
1712 unsigned int node_io_flag;
1713
1714 /* For sysfs support */
1715 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1716 struct completion s_kobj_unregister;
1717
1718 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1719 struct completion s_stat_kobj_unregister;
1720
1721 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1722 struct completion s_feature_list_kobj_unregister;
1723
1724 /* For shrinker support */
1725 struct list_head s_list;
1726 struct mutex umount_mutex;
1727 unsigned int shrinker_run_no;
1728
1729 /* For multi devices */
1730 int s_ndevs; /* number of devices */
1731 struct f2fs_dev_info *devs; /* for device list */
1732 unsigned int dirty_device; /* for checkpoint data flush */
1733 spinlock_t dev_lock; /* protect dirty_device */
1734 bool aligned_blksize; /* all devices has the same logical blksize */
1735
1736 /* For write statistics */
1737 u64 sectors_written_start;
1738 u64 kbytes_written;
1739
1740 /* Reference to checksum algorithm driver via cryptoapi */
1741 struct crypto_shash *s_chksum_driver;
1742
1743 /* Precomputed FS UUID checksum for seeding other checksums */
1744 __u32 s_chksum_seed;
1745
1746 struct workqueue_struct *post_read_wq; /* post read workqueue */
1747
1748 /*
1749 * If we are in irq context, let's update error information into
1750 * on-disk superblock in the work.
1751 */
1752 struct work_struct s_error_work;
1753 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
1754 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */
1755 spinlock_t error_lock; /* protect errors/stop_reason array */
1756 bool error_dirty; /* errors of sb is dirty */
1757
1758 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1759 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1760
1761 /* For reclaimed segs statistics per each GC mode */
1762 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1763 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1764
1765 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1766
1767 int max_fragment_chunk; /* max chunk size for block fragmentation mode */
1768 int max_fragment_hole; /* max hole size for block fragmentation mode */
1769
1770 /* For atomic write statistics */
1771 atomic64_t current_atomic_write;
1772 s64 peak_atomic_write;
1773 u64 committed_atomic_block;
1774 u64 revoked_atomic_block;
1775
1776#ifdef CONFIG_F2FS_FS_COMPRESSION
1777 struct kmem_cache *page_array_slab; /* page array entry */
1778 unsigned int page_array_slab_size; /* default page array slab size */
1779
1780 /* For runtime compression statistics */
1781 u64 compr_written_block;
1782 u64 compr_saved_block;
1783 u32 compr_new_inode;
1784
1785 /* For compressed block cache */
1786 struct inode *compress_inode; /* cache compressed blocks */
1787 unsigned int compress_percent; /* cache page percentage */
1788 unsigned int compress_watermark; /* cache page watermark */
1789 atomic_t compress_page_hit; /* cache hit count */
1790#endif
1791
1792#ifdef CONFIG_F2FS_IOSTAT
1793 /* For app/fs IO statistics */
1794 spinlock_t iostat_lock;
1795 unsigned long long iostat_count[NR_IO_TYPE];
1796 unsigned long long iostat_bytes[NR_IO_TYPE];
1797 unsigned long long prev_iostat_bytes[NR_IO_TYPE];
1798 bool iostat_enable;
1799 unsigned long iostat_next_period;
1800 unsigned int iostat_period_ms;
1801
1802 /* For io latency related statistics info in one iostat period */
1803 spinlock_t iostat_lat_lock;
1804 struct iostat_lat_info *iostat_io_lat;
1805#endif
1806};
1807
1808#ifdef CONFIG_F2FS_FAULT_INJECTION
1809#define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \
1810 __builtin_return_address(0))
1811static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type,
1812 const char *func, const char *parent_func)
1813{
1814 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1815
1816 if (!ffi->inject_rate)
1817 return false;
1818
1819 if (!IS_FAULT_SET(ffi, type))
1820 return false;
1821
1822 atomic_inc(v: &ffi->inject_ops);
1823 if (atomic_read(v: &ffi->inject_ops) >= ffi->inject_rate) {
1824 atomic_set(v: &ffi->inject_ops, i: 0);
1825 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n",
1826 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type],
1827 func, parent_func);
1828 return true;
1829 }
1830 return false;
1831}
1832#else
1833static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1834{
1835 return false;
1836}
1837#endif
1838
1839/*
1840 * Test if the mounted volume is a multi-device volume.
1841 * - For a single regular disk volume, sbi->s_ndevs is 0.
1842 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1843 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1844 */
1845static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1846{
1847 return sbi->s_ndevs > 1;
1848}
1849
1850static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1851{
1852 unsigned long now = jiffies;
1853
1854 sbi->last_time[type] = now;
1855
1856 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1857 if (type == REQ_TIME) {
1858 sbi->last_time[DISCARD_TIME] = now;
1859 sbi->last_time[GC_TIME] = now;
1860 }
1861}
1862
1863static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1864{
1865 unsigned long interval = sbi->interval_time[type] * HZ;
1866
1867 return time_after(jiffies, sbi->last_time[type] + interval);
1868}
1869
1870static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1871 int type)
1872{
1873 unsigned long interval = sbi->interval_time[type] * HZ;
1874 unsigned int wait_ms = 0;
1875 long delta;
1876
1877 delta = (sbi->last_time[type] + interval) - jiffies;
1878 if (delta > 0)
1879 wait_ms = jiffies_to_msecs(j: delta);
1880
1881 return wait_ms;
1882}
1883
1884/*
1885 * Inline functions
1886 */
1887static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1888 const void *address, unsigned int length)
1889{
1890 struct {
1891 struct shash_desc shash;
1892 char ctx[4];
1893 } desc;
1894 int err;
1895
1896 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1897
1898 desc.shash.tfm = sbi->s_chksum_driver;
1899 *(u32 *)desc.ctx = crc;
1900
1901 err = crypto_shash_update(desc: &desc.shash, data: address, len: length);
1902 BUG_ON(err);
1903
1904 return *(u32 *)desc.ctx;
1905}
1906
1907static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1908 unsigned int length)
1909{
1910 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1911}
1912
1913static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1914 void *buf, size_t buf_size)
1915{
1916 return f2fs_crc32(sbi, address: buf, length: buf_size) == blk_crc;
1917}
1918
1919static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1920 const void *address, unsigned int length)
1921{
1922 return __f2fs_crc32(sbi, crc, address, length);
1923}
1924
1925static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1926{
1927 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1928}
1929
1930static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1931{
1932 return sb->s_fs_info;
1933}
1934
1935static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1936{
1937 return F2FS_SB(sb: inode->i_sb);
1938}
1939
1940static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1941{
1942 return F2FS_I_SB(inode: mapping->host);
1943}
1944
1945static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1946{
1947 return F2FS_M_SB(mapping: page_file_mapping(page));
1948}
1949
1950static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1951{
1952 return (struct f2fs_super_block *)(sbi->raw_super);
1953}
1954
1955static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1956{
1957 return (struct f2fs_checkpoint *)(sbi->ckpt);
1958}
1959
1960static inline struct f2fs_node *F2FS_NODE(struct page *page)
1961{
1962 return (struct f2fs_node *)page_address(page);
1963}
1964
1965static inline struct f2fs_inode *F2FS_INODE(struct page *page)
1966{
1967 return &((struct f2fs_node *)page_address(page))->i;
1968}
1969
1970static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
1971{
1972 return (struct f2fs_nm_info *)(sbi->nm_info);
1973}
1974
1975static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
1976{
1977 return (struct f2fs_sm_info *)(sbi->sm_info);
1978}
1979
1980static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
1981{
1982 return (struct sit_info *)(SM_I(sbi)->sit_info);
1983}
1984
1985static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
1986{
1987 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
1988}
1989
1990static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
1991{
1992 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
1993}
1994
1995static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
1996{
1997 return sbi->meta_inode->i_mapping;
1998}
1999
2000static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
2001{
2002 return sbi->node_inode->i_mapping;
2003}
2004
2005static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
2006{
2007 return test_bit(type, &sbi->s_flag);
2008}
2009
2010static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2011{
2012 set_bit(nr: type, addr: &sbi->s_flag);
2013}
2014
2015static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2016{
2017 clear_bit(nr: type, addr: &sbi->s_flag);
2018}
2019
2020static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
2021{
2022 return le64_to_cpu(cp->checkpoint_ver);
2023}
2024
2025static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
2026{
2027 if (type < F2FS_MAX_QUOTAS)
2028 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
2029 return 0;
2030}
2031
2032static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2033{
2034 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2035 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2036}
2037
2038static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2039{
2040 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2041
2042 return ckpt_flags & f;
2043}
2044
2045static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2046{
2047 return __is_set_ckpt_flags(cp: F2FS_CKPT(sbi), f);
2048}
2049
2050static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2051{
2052 unsigned int ckpt_flags;
2053
2054 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2055 ckpt_flags |= f;
2056 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2057}
2058
2059static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2060{
2061 unsigned long flags;
2062
2063 spin_lock_irqsave(&sbi->cp_lock, flags);
2064 __set_ckpt_flags(cp: F2FS_CKPT(sbi), f);
2065 spin_unlock_irqrestore(lock: &sbi->cp_lock, flags);
2066}
2067
2068static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2069{
2070 unsigned int ckpt_flags;
2071
2072 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2073 ckpt_flags &= (~f);
2074 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2075}
2076
2077static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2078{
2079 unsigned long flags;
2080
2081 spin_lock_irqsave(&sbi->cp_lock, flags);
2082 __clear_ckpt_flags(cp: F2FS_CKPT(sbi), f);
2083 spin_unlock_irqrestore(lock: &sbi->cp_lock, flags);
2084}
2085
2086#define init_f2fs_rwsem(sem) \
2087do { \
2088 static struct lock_class_key __key; \
2089 \
2090 __init_f2fs_rwsem((sem), #sem, &__key); \
2091} while (0)
2092
2093static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,
2094 const char *sem_name, struct lock_class_key *key)
2095{
2096 __init_rwsem(sem: &sem->internal_rwsem, name: sem_name, key);
2097#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2098 init_waitqueue_head(&sem->read_waiters);
2099#endif
2100}
2101
2102static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)
2103{
2104 return rwsem_is_locked(sem: &sem->internal_rwsem);
2105}
2106
2107static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)
2108{
2109 return rwsem_is_contended(sem: &sem->internal_rwsem);
2110}
2111
2112static inline void f2fs_down_read(struct f2fs_rwsem *sem)
2113{
2114#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2115 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));
2116#else
2117 down_read(&sem->internal_rwsem);
2118#endif
2119}
2120
2121static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)
2122{
2123 return down_read_trylock(sem: &sem->internal_rwsem);
2124}
2125
2126static inline void f2fs_up_read(struct f2fs_rwsem *sem)
2127{
2128 up_read(sem: &sem->internal_rwsem);
2129}
2130
2131static inline void f2fs_down_write(struct f2fs_rwsem *sem)
2132{
2133 down_write(sem: &sem->internal_rwsem);
2134}
2135
2136#ifdef CONFIG_DEBUG_LOCK_ALLOC
2137static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)
2138{
2139 down_read_nested(sem: &sem->internal_rwsem, subclass);
2140}
2141
2142static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass)
2143{
2144 down_write_nested(sem: &sem->internal_rwsem, subclass);
2145}
2146#else
2147#define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)
2148#define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem)
2149#endif
2150
2151static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)
2152{
2153 return down_write_trylock(sem: &sem->internal_rwsem);
2154}
2155
2156static inline void f2fs_up_write(struct f2fs_rwsem *sem)
2157{
2158 up_write(sem: &sem->internal_rwsem);
2159#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2160 wake_up_all(&sem->read_waiters);
2161#endif
2162}
2163
2164static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2165{
2166 f2fs_down_read(sem: &sbi->cp_rwsem);
2167}
2168
2169static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2170{
2171 if (time_to_inject(sbi, FAULT_LOCK_OP))
2172 return 0;
2173 return f2fs_down_read_trylock(sem: &sbi->cp_rwsem);
2174}
2175
2176static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2177{
2178 f2fs_up_read(sem: &sbi->cp_rwsem);
2179}
2180
2181static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2182{
2183 f2fs_down_write(sem: &sbi->cp_rwsem);
2184}
2185
2186static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2187{
2188 f2fs_up_write(sem: &sbi->cp_rwsem);
2189}
2190
2191static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2192{
2193 int reason = CP_SYNC;
2194
2195 if (test_opt(sbi, FASTBOOT))
2196 reason = CP_FASTBOOT;
2197 if (is_sbi_flag_set(sbi, type: SBI_IS_CLOSE))
2198 reason = CP_UMOUNT;
2199 return reason;
2200}
2201
2202static inline bool __remain_node_summaries(int reason)
2203{
2204 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2205}
2206
2207static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2208{
2209 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2210 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2211}
2212
2213/*
2214 * Check whether the inode has blocks or not
2215 */
2216static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2217{
2218 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2219
2220 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2221}
2222
2223static inline bool f2fs_has_xattr_block(unsigned int ofs)
2224{
2225 return ofs == XATTR_NODE_OFFSET;
2226}
2227
2228static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2229 struct inode *inode, bool cap)
2230{
2231 if (!inode)
2232 return true;
2233 if (!test_opt(sbi, RESERVE_ROOT))
2234 return false;
2235 if (IS_NOQUOTA(inode))
2236 return true;
2237 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2238 return true;
2239 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2240 in_group_p(F2FS_OPTION(sbi).s_resgid))
2241 return true;
2242 if (cap && capable(CAP_SYS_RESOURCE))
2243 return true;
2244 return false;
2245}
2246
2247static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2248static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2249 struct inode *inode, blkcnt_t *count)
2250{
2251 blkcnt_t diff = 0, release = 0;
2252 block_t avail_user_block_count;
2253 int ret;
2254
2255 ret = dquot_reserve_block(inode, nr: *count);
2256 if (ret)
2257 return ret;
2258
2259 if (time_to_inject(sbi, FAULT_BLOCK)) {
2260 release = *count;
2261 goto release_quota;
2262 }
2263
2264 /*
2265 * let's increase this in prior to actual block count change in order
2266 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2267 */
2268 percpu_counter_add(fbc: &sbi->alloc_valid_block_count, amount: (*count));
2269
2270 spin_lock(lock: &sbi->stat_lock);
2271 sbi->total_valid_block_count += (block_t)(*count);
2272 avail_user_block_count = sbi->user_block_count -
2273 sbi->current_reserved_blocks;
2274
2275 if (!__allow_reserved_blocks(sbi, inode, cap: true))
2276 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2277
2278 if (F2FS_IO_ALIGNED(sbi))
2279 avail_user_block_count -= sbi->blocks_per_seg *
2280 SM_I(sbi)->additional_reserved_segments;
2281
2282 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2283 if (avail_user_block_count > sbi->unusable_block_count)
2284 avail_user_block_count -= sbi->unusable_block_count;
2285 else
2286 avail_user_block_count = 0;
2287 }
2288 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) {
2289 diff = sbi->total_valid_block_count - avail_user_block_count;
2290 if (diff > *count)
2291 diff = *count;
2292 *count -= diff;
2293 release = diff;
2294 sbi->total_valid_block_count -= diff;
2295 if (!*count) {
2296 spin_unlock(lock: &sbi->stat_lock);
2297 goto enospc;
2298 }
2299 }
2300 spin_unlock(lock: &sbi->stat_lock);
2301
2302 if (unlikely(release)) {
2303 percpu_counter_sub(fbc: &sbi->alloc_valid_block_count, amount: release);
2304 dquot_release_reservation_block(inode, nr: release);
2305 }
2306 f2fs_i_blocks_write(inode, *count, true, true);
2307 return 0;
2308
2309enospc:
2310 percpu_counter_sub(fbc: &sbi->alloc_valid_block_count, amount: release);
2311release_quota:
2312 dquot_release_reservation_block(inode, nr: release);
2313 return -ENOSPC;
2314}
2315
2316__printf(2, 3)
2317void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...);
2318
2319#define f2fs_err(sbi, fmt, ...) \
2320 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__)
2321#define f2fs_warn(sbi, fmt, ...) \
2322 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__)
2323#define f2fs_notice(sbi, fmt, ...) \
2324 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__)
2325#define f2fs_info(sbi, fmt, ...) \
2326 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__)
2327#define f2fs_debug(sbi, fmt, ...) \
2328 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__)
2329
2330#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
2331static inline bool page_private_##name(struct page *page) \
2332{ \
2333 return PagePrivate(page) && \
2334 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
2335 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2336}
2337
2338#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
2339static inline void set_page_private_##name(struct page *page) \
2340{ \
2341 if (!PagePrivate(page)) \
2342 attach_page_private(page, (void *)0); \
2343 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
2344 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2345}
2346
2347#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
2348static inline void clear_page_private_##name(struct page *page) \
2349{ \
2350 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2351 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \
2352 detach_page_private(page); \
2353}
2354
2355PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
2356PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
2357PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
2358PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE);
2359
2360PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
2361PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
2362PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
2363PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE);
2364
2365PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
2366PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
2367PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
2368PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE);
2369
2370static inline unsigned long get_page_private_data(struct page *page)
2371{
2372 unsigned long data = page_private(page);
2373
2374 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
2375 return 0;
2376 return data >> PAGE_PRIVATE_MAX;
2377}
2378
2379static inline void set_page_private_data(struct page *page, unsigned long data)
2380{
2381 if (!PagePrivate(page))
2382 attach_page_private(page, data: (void *)0);
2383 set_bit(nr: PAGE_PRIVATE_NOT_POINTER, addr: &page_private(page));
2384 page_private(page) |= data << PAGE_PRIVATE_MAX;
2385}
2386
2387static inline void clear_page_private_data(struct page *page)
2388{
2389 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0);
2390 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER))
2391 detach_page_private(page);
2392}
2393
2394static inline void clear_page_private_all(struct page *page)
2395{
2396 clear_page_private_data(page);
2397 clear_page_private_reference(page);
2398 clear_page_private_gcing(page);
2399 clear_page_private_inline(page);
2400
2401 f2fs_bug_on(F2FS_P_SB(page), page_private(page));
2402}
2403
2404static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2405 struct inode *inode,
2406 block_t count)
2407{
2408 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2409
2410 spin_lock(lock: &sbi->stat_lock);
2411 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2412 sbi->total_valid_block_count -= (block_t)count;
2413 if (sbi->reserved_blocks &&
2414 sbi->current_reserved_blocks < sbi->reserved_blocks)
2415 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2416 sbi->current_reserved_blocks + count);
2417 spin_unlock(lock: &sbi->stat_lock);
2418 if (unlikely(inode->i_blocks < sectors)) {
2419 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2420 inode->i_ino,
2421 (unsigned long long)inode->i_blocks,
2422 (unsigned long long)sectors);
2423 set_sbi_flag(sbi, type: SBI_NEED_FSCK);
2424 return;
2425 }
2426 f2fs_i_blocks_write(inode, count, false, true);
2427}
2428
2429static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2430{
2431 atomic_inc(v: &sbi->nr_pages[count_type]);
2432
2433 if (count_type == F2FS_DIRTY_DENTS ||
2434 count_type == F2FS_DIRTY_NODES ||
2435 count_type == F2FS_DIRTY_META ||
2436 count_type == F2FS_DIRTY_QDATA ||
2437 count_type == F2FS_DIRTY_IMETA)
2438 set_sbi_flag(sbi, type: SBI_IS_DIRTY);
2439}
2440
2441static inline void inode_inc_dirty_pages(struct inode *inode)
2442{
2443 atomic_inc(v: &F2FS_I(inode)->dirty_pages);
2444 inc_page_count(sbi: F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2445 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2446 if (IS_NOQUOTA(inode))
2447 inc_page_count(sbi: F2FS_I_SB(inode), count_type: F2FS_DIRTY_QDATA);
2448}
2449
2450static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2451{
2452 atomic_dec(v: &sbi->nr_pages[count_type]);
2453}
2454
2455static inline void inode_dec_dirty_pages(struct inode *inode)
2456{
2457 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2458 !S_ISLNK(inode->i_mode))
2459 return;
2460
2461 atomic_dec(v: &F2FS_I(inode)->dirty_pages);
2462 dec_page_count(sbi: F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2463 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2464 if (IS_NOQUOTA(inode))
2465 dec_page_count(sbi: F2FS_I_SB(inode), count_type: F2FS_DIRTY_QDATA);
2466}
2467
2468static inline void inc_atomic_write_cnt(struct inode *inode)
2469{
2470 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2471 struct f2fs_inode_info *fi = F2FS_I(inode);
2472 u64 current_write;
2473
2474 fi->atomic_write_cnt++;
2475 atomic64_inc(v: &sbi->current_atomic_write);
2476 current_write = atomic64_read(v: &sbi->current_atomic_write);
2477 if (current_write > sbi->peak_atomic_write)
2478 sbi->peak_atomic_write = current_write;
2479}
2480
2481static inline void release_atomic_write_cnt(struct inode *inode)
2482{
2483 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2484 struct f2fs_inode_info *fi = F2FS_I(inode);
2485
2486 atomic64_sub(i: fi->atomic_write_cnt, v: &sbi->current_atomic_write);
2487 fi->atomic_write_cnt = 0;
2488}
2489
2490static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2491{
2492 return atomic_read(v: &sbi->nr_pages[count_type]);
2493}
2494
2495static inline int get_dirty_pages(struct inode *inode)
2496{
2497 return atomic_read(v: &F2FS_I(inode)->dirty_pages);
2498}
2499
2500static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2501{
2502 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg;
2503 unsigned int segs = (get_pages(sbi, count_type: block_type) + pages_per_sec - 1) >>
2504 sbi->log_blocks_per_seg;
2505
2506 return segs / sbi->segs_per_sec;
2507}
2508
2509static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2510{
2511 return sbi->total_valid_block_count;
2512}
2513
2514static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2515{
2516 return sbi->discard_blks;
2517}
2518
2519static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2520{
2521 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2522
2523 /* return NAT or SIT bitmap */
2524 if (flag == NAT_BITMAP)
2525 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2526 else if (flag == SIT_BITMAP)
2527 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2528
2529 return 0;
2530}
2531
2532static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2533{
2534 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2535}
2536
2537static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2538{
2539 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2540 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2541 int offset;
2542
2543 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2544 offset = (flag == SIT_BITMAP) ?
2545 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2546 /*
2547 * if large_nat_bitmap feature is enabled, leave checksum
2548 * protection for all nat/sit bitmaps.
2549 */
2550 return tmp_ptr + offset + sizeof(__le32);
2551 }
2552
2553 if (__cp_payload(sbi) > 0) {
2554 if (flag == NAT_BITMAP)
2555 return tmp_ptr;
2556 else
2557 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2558 } else {
2559 offset = (flag == NAT_BITMAP) ?
2560 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2561 return tmp_ptr + offset;
2562 }
2563}
2564
2565static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2566{
2567 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2568
2569 if (sbi->cur_cp_pack == 2)
2570 start_addr += sbi->blocks_per_seg;
2571 return start_addr;
2572}
2573
2574static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2575{
2576 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2577
2578 if (sbi->cur_cp_pack == 1)
2579 start_addr += sbi->blocks_per_seg;
2580 return start_addr;
2581}
2582
2583static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2584{
2585 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2586}
2587
2588static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2589{
2590 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2591}
2592
2593extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
2594static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2595 struct inode *inode, bool is_inode)
2596{
2597 block_t valid_block_count;
2598 unsigned int valid_node_count, user_block_count;
2599 int err;
2600
2601 if (is_inode) {
2602 if (inode) {
2603 err = dquot_alloc_inode(inode);
2604 if (err)
2605 return err;
2606 }
2607 } else {
2608 err = dquot_reserve_block(inode, nr: 1);
2609 if (err)
2610 return err;
2611 }
2612
2613 if (time_to_inject(sbi, FAULT_BLOCK))
2614 goto enospc;
2615
2616 spin_lock(lock: &sbi->stat_lock);
2617
2618 valid_block_count = sbi->total_valid_block_count +
2619 sbi->current_reserved_blocks + 1;
2620
2621 if (!__allow_reserved_blocks(sbi, inode, cap: false))
2622 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks;
2623
2624 if (F2FS_IO_ALIGNED(sbi))
2625 valid_block_count += sbi->blocks_per_seg *
2626 SM_I(sbi)->additional_reserved_segments;
2627
2628 user_block_count = sbi->user_block_count;
2629 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2630 user_block_count -= sbi->unusable_block_count;
2631
2632 if (unlikely(valid_block_count > user_block_count)) {
2633 spin_unlock(lock: &sbi->stat_lock);
2634 goto enospc;
2635 }
2636
2637 valid_node_count = sbi->total_valid_node_count + 1;
2638 if (unlikely(valid_node_count > sbi->total_node_count)) {
2639 spin_unlock(lock: &sbi->stat_lock);
2640 goto enospc;
2641 }
2642
2643 sbi->total_valid_node_count++;
2644 sbi->total_valid_block_count++;
2645 spin_unlock(lock: &sbi->stat_lock);
2646
2647 if (inode) {
2648 if (is_inode)
2649 f2fs_mark_inode_dirty_sync(inode, sync: true);
2650 else
2651 f2fs_i_blocks_write(inode, 1, true, true);
2652 }
2653
2654 percpu_counter_inc(fbc: &sbi->alloc_valid_block_count);
2655 return 0;
2656
2657enospc:
2658 if (is_inode) {
2659 if (inode)
2660 dquot_free_inode(inode);
2661 } else {
2662 dquot_release_reservation_block(inode, nr: 1);
2663 }
2664 return -ENOSPC;
2665}
2666
2667static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2668 struct inode *inode, bool is_inode)
2669{
2670 spin_lock(lock: &sbi->stat_lock);
2671
2672 if (unlikely(!sbi->total_valid_block_count ||
2673 !sbi->total_valid_node_count)) {
2674 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2675 sbi->total_valid_block_count,
2676 sbi->total_valid_node_count);
2677 set_sbi_flag(sbi, type: SBI_NEED_FSCK);
2678 } else {
2679 sbi->total_valid_block_count--;
2680 sbi->total_valid_node_count--;
2681 }
2682
2683 if (sbi->reserved_blocks &&
2684 sbi->current_reserved_blocks < sbi->reserved_blocks)
2685 sbi->current_reserved_blocks++;
2686
2687 spin_unlock(lock: &sbi->stat_lock);
2688
2689 if (is_inode) {
2690 dquot_free_inode(inode);
2691 } else {
2692 if (unlikely(inode->i_blocks == 0)) {
2693 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2694 inode->i_ino,
2695 (unsigned long long)inode->i_blocks);
2696 set_sbi_flag(sbi, type: SBI_NEED_FSCK);
2697 return;
2698 }
2699 f2fs_i_blocks_write(inode, 1, false, true);
2700 }
2701}
2702
2703static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2704{
2705 return sbi->total_valid_node_count;
2706}
2707
2708static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2709{
2710 percpu_counter_inc(fbc: &sbi->total_valid_inode_count);
2711}
2712
2713static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2714{
2715 percpu_counter_dec(fbc: &sbi->total_valid_inode_count);
2716}
2717
2718static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2719{
2720 return percpu_counter_sum_positive(fbc: &sbi->total_valid_inode_count);
2721}
2722
2723static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2724 pgoff_t index, bool for_write)
2725{
2726 struct page *page;
2727 unsigned int flags;
2728
2729 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2730 if (!for_write)
2731 page = find_get_page_flags(mapping, offset: index,
2732 FGP_LOCK | FGP_ACCESSED);
2733 else
2734 page = find_lock_page(mapping, index);
2735 if (page)
2736 return page;
2737
2738 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))
2739 return NULL;
2740 }
2741
2742 if (!for_write)
2743 return grab_cache_page(mapping, index);
2744
2745 flags = memalloc_nofs_save();
2746 page = grab_cache_page_write_begin(mapping, index);
2747 memalloc_nofs_restore(flags);
2748
2749 return page;
2750}
2751
2752static inline struct page *f2fs_pagecache_get_page(
2753 struct address_space *mapping, pgoff_t index,
2754 fgf_t fgp_flags, gfp_t gfp_mask)
2755{
2756 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET))
2757 return NULL;
2758
2759 return pagecache_get_page(mapping, index, fgp_flags, gfp: gfp_mask);
2760}
2761
2762static inline void f2fs_put_page(struct page *page, int unlock)
2763{
2764 if (!page)
2765 return;
2766
2767 if (unlock) {
2768 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2769 unlock_page(page);
2770 }
2771 put_page(page);
2772}
2773
2774static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2775{
2776 if (dn->node_page)
2777 f2fs_put_page(page: dn->node_page, unlock: 1);
2778 if (dn->inode_page && dn->node_page != dn->inode_page)
2779 f2fs_put_page(page: dn->inode_page, unlock: 0);
2780 dn->node_page = NULL;
2781 dn->inode_page = NULL;
2782}
2783
2784static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2785 size_t size)
2786{
2787 return kmem_cache_create(name, size, align: 0, SLAB_RECLAIM_ACCOUNT, NULL);
2788}
2789
2790static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
2791 gfp_t flags)
2792{
2793 void *entry;
2794
2795 entry = kmem_cache_alloc(cachep, flags);
2796 if (!entry)
2797 entry = kmem_cache_alloc(cachep, flags: flags | __GFP_NOFAIL);
2798 return entry;
2799}
2800
2801static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2802 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
2803{
2804 if (nofail)
2805 return f2fs_kmem_cache_alloc_nofail(cachep, flags);
2806
2807 if (time_to_inject(sbi, FAULT_SLAB_ALLOC))
2808 return NULL;
2809
2810 return kmem_cache_alloc(cachep, flags);
2811}
2812
2813static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2814{
2815 if (get_pages(sbi, count_type: F2FS_RD_DATA) || get_pages(sbi, count_type: F2FS_RD_NODE) ||
2816 get_pages(sbi, count_type: F2FS_RD_META) || get_pages(sbi, count_type: F2FS_WB_DATA) ||
2817 get_pages(sbi, count_type: F2FS_WB_CP_DATA) ||
2818 get_pages(sbi, count_type: F2FS_DIO_READ) ||
2819 get_pages(sbi, count_type: F2FS_DIO_WRITE))
2820 return true;
2821
2822 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2823 atomic_read(v: &SM_I(sbi)->dcc_info->queued_discard))
2824 return true;
2825
2826 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2827 atomic_read(v: &SM_I(sbi)->fcc_info->queued_flush))
2828 return true;
2829 return false;
2830}
2831
2832static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2833{
2834 if (sbi->gc_mode == GC_URGENT_HIGH)
2835 return true;
2836
2837 if (is_inflight_io(sbi, type))
2838 return false;
2839
2840 if (sbi->gc_mode == GC_URGENT_MID)
2841 return true;
2842
2843 if (sbi->gc_mode == GC_URGENT_LOW &&
2844 (type == DISCARD_TIME || type == GC_TIME))
2845 return true;
2846
2847 return f2fs_time_over(sbi, type);
2848}
2849
2850static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2851 unsigned long index, void *item)
2852{
2853 while (radix_tree_insert(root, index, item))
2854 cond_resched();
2855}
2856
2857#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2858
2859static inline bool IS_INODE(struct page *page)
2860{
2861 struct f2fs_node *p = F2FS_NODE(page);
2862
2863 return RAW_IS_INODE(p);
2864}
2865
2866static inline int offset_in_addr(struct f2fs_inode *i)
2867{
2868 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2869 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2870}
2871
2872static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2873{
2874 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2875}
2876
2877static inline int f2fs_has_extra_attr(struct inode *inode);
2878static inline block_t data_blkaddr(struct inode *inode,
2879 struct page *node_page, unsigned int offset)
2880{
2881 struct f2fs_node *raw_node;
2882 __le32 *addr_array;
2883 int base = 0;
2884 bool is_inode = IS_INODE(page: node_page);
2885
2886 raw_node = F2FS_NODE(page: node_page);
2887
2888 if (is_inode) {
2889 if (!inode)
2890 /* from GC path only */
2891 base = offset_in_addr(i: &raw_node->i);
2892 else if (f2fs_has_extra_attr(inode))
2893 base = get_extra_isize(inode);
2894 }
2895
2896 addr_array = blkaddr_in_node(node: raw_node);
2897 return le32_to_cpu(addr_array[base + offset]);
2898}
2899
2900static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2901{
2902 return data_blkaddr(inode: dn->inode, node_page: dn->node_page, offset: dn->ofs_in_node);
2903}
2904
2905static inline int f2fs_test_bit(unsigned int nr, char *addr)
2906{
2907 int mask;
2908
2909 addr += (nr >> 3);
2910 mask = BIT(7 - (nr & 0x07));
2911 return mask & *addr;
2912}
2913
2914static inline void f2fs_set_bit(unsigned int nr, char *addr)
2915{
2916 int mask;
2917
2918 addr += (nr >> 3);
2919 mask = BIT(7 - (nr & 0x07));
2920 *addr |= mask;
2921}
2922
2923static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2924{
2925 int mask;
2926
2927 addr += (nr >> 3);
2928 mask = BIT(7 - (nr & 0x07));
2929 *addr &= ~mask;
2930}
2931
2932static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2933{
2934 int mask;
2935 int ret;
2936
2937 addr += (nr >> 3);
2938 mask = BIT(7 - (nr & 0x07));
2939 ret = mask & *addr;
2940 *addr |= mask;
2941 return ret;
2942}
2943
2944static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2945{
2946 int mask;
2947 int ret;
2948
2949 addr += (nr >> 3);
2950 mask = BIT(7 - (nr & 0x07));
2951 ret = mask & *addr;
2952 *addr &= ~mask;
2953 return ret;
2954}
2955
2956static inline void f2fs_change_bit(unsigned int nr, char *addr)
2957{
2958 int mask;
2959
2960 addr += (nr >> 3);
2961 mask = BIT(7 - (nr & 0x07));
2962 *addr ^= mask;
2963}
2964
2965/*
2966 * On-disk inode flags (f2fs_inode::i_flags)
2967 */
2968#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2969#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2970#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2971#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2972#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2973#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2974#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2975#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2976#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2977#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2978#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2979
2980#define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL)
2981
2982/* Flags that should be inherited by new inodes from their parent. */
2983#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
2984 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2985 F2FS_CASEFOLD_FL)
2986
2987/* Flags that are appropriate for regular files (all but dir-specific ones). */
2988#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
2989 F2FS_CASEFOLD_FL))
2990
2991/* Flags that are appropriate for non-directories/regular files. */
2992#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
2993
2994static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
2995{
2996 if (S_ISDIR(mode))
2997 return flags;
2998 else if (S_ISREG(mode))
2999 return flags & F2FS_REG_FLMASK;
3000 else
3001 return flags & F2FS_OTHER_FLMASK;
3002}
3003
3004static inline void __mark_inode_dirty_flag(struct inode *inode,
3005 int flag, bool set)
3006{
3007 switch (flag) {
3008 case FI_INLINE_XATTR:
3009 case FI_INLINE_DATA:
3010 case FI_INLINE_DENTRY:
3011 case FI_NEW_INODE:
3012 if (set)
3013 return;
3014 fallthrough;
3015 case FI_DATA_EXIST:
3016 case FI_INLINE_DOTS:
3017 case FI_PIN_FILE:
3018 case FI_COMPRESS_RELEASED:
3019 f2fs_mark_inode_dirty_sync(inode, sync: true);
3020 }
3021}
3022
3023static inline void set_inode_flag(struct inode *inode, int flag)
3024{
3025 set_bit(nr: flag, addr: F2FS_I(inode)->flags);
3026 __mark_inode_dirty_flag(inode, flag, set: true);
3027}
3028
3029static inline int is_inode_flag_set(struct inode *inode, int flag)
3030{
3031 return test_bit(flag, F2FS_I(inode)->flags);
3032}
3033
3034static inline void clear_inode_flag(struct inode *inode, int flag)
3035{
3036 clear_bit(nr: flag, addr: F2FS_I(inode)->flags);
3037 __mark_inode_dirty_flag(inode, flag, set: false);
3038}
3039
3040static inline bool f2fs_verity_in_progress(struct inode *inode)
3041{
3042 return IS_ENABLED(CONFIG_FS_VERITY) &&
3043 is_inode_flag_set(inode, flag: FI_VERITY_IN_PROGRESS);
3044}
3045
3046static inline void set_acl_inode(struct inode *inode, umode_t mode)
3047{
3048 F2FS_I(inode)->i_acl_mode = mode;
3049 set_inode_flag(inode, flag: FI_ACL_MODE);
3050 f2fs_mark_inode_dirty_sync(inode, sync: false);
3051}
3052
3053static inline void f2fs_i_links_write(struct inode *inode, bool inc)
3054{
3055 if (inc)
3056 inc_nlink(inode);
3057 else
3058 drop_nlink(inode);
3059 f2fs_mark_inode_dirty_sync(inode, sync: true);
3060}
3061
3062static inline void f2fs_i_blocks_write(struct inode *inode,
3063 block_t diff, bool add, bool claim)
3064{
3065 bool clean = !is_inode_flag_set(inode, flag: FI_DIRTY_INODE);
3066 bool recover = is_inode_flag_set(inode, flag: FI_AUTO_RECOVER);
3067
3068 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
3069 if (add) {
3070 if (claim)
3071 dquot_claim_block(inode, nr: diff);
3072 else
3073 dquot_alloc_block_nofail(inode, nr: diff);
3074 } else {
3075 dquot_free_block(inode, nr: diff);
3076 }
3077
3078 f2fs_mark_inode_dirty_sync(inode, sync: true);
3079 if (clean || recover)
3080 set_inode_flag(inode, flag: FI_AUTO_RECOVER);
3081}
3082
3083static inline bool f2fs_is_atomic_file(struct inode *inode);
3084
3085static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
3086{
3087 bool clean = !is_inode_flag_set(inode, flag: FI_DIRTY_INODE);
3088 bool recover = is_inode_flag_set(inode, flag: FI_AUTO_RECOVER);
3089
3090 if (i_size_read(inode) == i_size)
3091 return;
3092
3093 i_size_write(inode, i_size);
3094
3095 if (f2fs_is_atomic_file(inode))
3096 return;
3097
3098 f2fs_mark_inode_dirty_sync(inode, sync: true);
3099 if (clean || recover)
3100 set_inode_flag(inode, flag: FI_AUTO_RECOVER);
3101}
3102
3103static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
3104{
3105 F2FS_I(inode)->i_current_depth = depth;
3106 f2fs_mark_inode_dirty_sync(inode, sync: true);
3107}
3108
3109static inline void f2fs_i_gc_failures_write(struct inode *inode,
3110 unsigned int count)
3111{
3112 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count;
3113 f2fs_mark_inode_dirty_sync(inode, sync: true);
3114}
3115
3116static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
3117{
3118 F2FS_I(inode)->i_xattr_nid = xnid;
3119 f2fs_mark_inode_dirty_sync(inode, sync: true);
3120}
3121
3122static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
3123{
3124 F2FS_I(inode)->i_pino = pino;
3125 f2fs_mark_inode_dirty_sync(inode, sync: true);
3126}
3127
3128static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
3129{
3130 struct f2fs_inode_info *fi = F2FS_I(inode);
3131
3132 if (ri->i_inline & F2FS_INLINE_XATTR)
3133 set_bit(nr: FI_INLINE_XATTR, addr: fi->flags);
3134 if (ri->i_inline & F2FS_INLINE_DATA)
3135 set_bit(nr: FI_INLINE_DATA, addr: fi->flags);
3136 if (ri->i_inline & F2FS_INLINE_DENTRY)
3137 set_bit(nr: FI_INLINE_DENTRY, addr: fi->flags);
3138 if (ri->i_inline & F2FS_DATA_EXIST)
3139 set_bit(nr: FI_DATA_EXIST, addr: fi->flags);
3140 if (ri->i_inline & F2FS_INLINE_DOTS)
3141 set_bit(nr: FI_INLINE_DOTS, addr: fi->flags);
3142 if (ri->i_inline & F2FS_EXTRA_ATTR)
3143 set_bit(nr: FI_EXTRA_ATTR, addr: fi->flags);
3144 if (ri->i_inline & F2FS_PIN_FILE)
3145 set_bit(nr: FI_PIN_FILE, addr: fi->flags);
3146 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
3147 set_bit(nr: FI_COMPRESS_RELEASED, addr: fi->flags);
3148}
3149
3150static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
3151{
3152 ri->i_inline = 0;
3153
3154 if (is_inode_flag_set(inode, flag: FI_INLINE_XATTR))
3155 ri->i_inline |= F2FS_INLINE_XATTR;
3156 if (is_inode_flag_set(inode, flag: FI_INLINE_DATA))
3157 ri->i_inline |= F2FS_INLINE_DATA;
3158 if (is_inode_flag_set(inode, flag: FI_INLINE_DENTRY))
3159 ri->i_inline |= F2FS_INLINE_DENTRY;
3160 if (is_inode_flag_set(inode, flag: FI_DATA_EXIST))
3161 ri->i_inline |= F2FS_DATA_EXIST;
3162 if (is_inode_flag_set(inode, flag: FI_INLINE_DOTS))
3163 ri->i_inline |= F2FS_INLINE_DOTS;
3164 if (is_inode_flag_set(inode, flag: FI_EXTRA_ATTR))
3165 ri->i_inline |= F2FS_EXTRA_ATTR;
3166 if (is_inode_flag_set(inode, flag: FI_PIN_FILE))
3167 ri->i_inline |= F2FS_PIN_FILE;
3168 if (is_inode_flag_set(inode, flag: FI_COMPRESS_RELEASED))
3169 ri->i_inline |= F2FS_COMPRESS_RELEASED;
3170}
3171
3172static inline int f2fs_has_extra_attr(struct inode *inode)
3173{
3174 return is_inode_flag_set(inode, flag: FI_EXTRA_ATTR);
3175}
3176
3177static inline int f2fs_has_inline_xattr(struct inode *inode)
3178{
3179 return is_inode_flag_set(inode, flag: FI_INLINE_XATTR);
3180}
3181
3182static inline int f2fs_compressed_file(struct inode *inode)
3183{
3184 return S_ISREG(inode->i_mode) &&
3185 is_inode_flag_set(inode, flag: FI_COMPRESSED_FILE);
3186}
3187
3188static inline bool f2fs_need_compress_data(struct inode *inode)
3189{
3190 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
3191
3192 if (!f2fs_compressed_file(inode))
3193 return false;
3194
3195 if (compress_mode == COMPR_MODE_FS)
3196 return true;
3197 else if (compress_mode == COMPR_MODE_USER &&
3198 is_inode_flag_set(inode, flag: FI_ENABLE_COMPRESS))
3199 return true;
3200
3201 return false;
3202}
3203
3204static inline unsigned int addrs_per_inode(struct inode *inode)
3205{
3206 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3207 get_inline_xattr_addrs(inode);
3208
3209 if (!f2fs_compressed_file(inode))
3210 return addrs;
3211 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3212}
3213
3214static inline unsigned int addrs_per_block(struct inode *inode)
3215{
3216 if (!f2fs_compressed_file(inode))
3217 return DEF_ADDRS_PER_BLOCK;
3218 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3219}
3220
3221static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3222{
3223 struct f2fs_inode *ri = F2FS_INODE(page);
3224
3225 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3226 get_inline_xattr_addrs(inode)]);
3227}
3228
3229static inline int inline_xattr_size(struct inode *inode)
3230{
3231 if (f2fs_has_inline_xattr(inode))
3232 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3233 return 0;
3234}
3235
3236/*
3237 * Notice: check inline_data flag without inode page lock is unsafe.
3238 * It could change at any time by f2fs_convert_inline_page().
3239 */
3240static inline int f2fs_has_inline_data(struct inode *inode)
3241{
3242 return is_inode_flag_set(inode, flag: FI_INLINE_DATA);
3243}
3244
3245static inline int f2fs_exist_data(struct inode *inode)
3246{
3247 return is_inode_flag_set(inode, flag: FI_DATA_EXIST);
3248}
3249
3250static inline int f2fs_has_inline_dots(struct inode *inode)
3251{
3252 return is_inode_flag_set(inode, flag: FI_INLINE_DOTS);
3253}
3254
3255static inline int f2fs_is_mmap_file(struct inode *inode)
3256{
3257 return is_inode_flag_set(inode, flag: FI_MMAP_FILE);
3258}
3259
3260static inline bool f2fs_is_pinned_file(struct inode *inode)
3261{
3262 return is_inode_flag_set(inode, flag: FI_PIN_FILE);
3263}
3264
3265static inline bool f2fs_is_atomic_file(struct inode *inode)
3266{
3267 return is_inode_flag_set(inode, flag: FI_ATOMIC_FILE);
3268}
3269
3270static inline bool f2fs_is_cow_file(struct inode *inode)
3271{
3272 return is_inode_flag_set(inode, flag: FI_COW_FILE);
3273}
3274
3275static inline bool f2fs_is_first_block_written(struct inode *inode)
3276{
3277 return is_inode_flag_set(inode, flag: FI_FIRST_BLOCK_WRITTEN);
3278}
3279
3280static inline bool f2fs_is_drop_cache(struct inode *inode)
3281{
3282 return is_inode_flag_set(inode, flag: FI_DROP_CACHE);
3283}
3284
3285static inline void *inline_data_addr(struct inode *inode, struct page *page)
3286{
3287 struct f2fs_inode *ri = F2FS_INODE(page);
3288 int extra_size = get_extra_isize(inode);
3289
3290 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]);
3291}
3292
3293static inline int f2fs_has_inline_dentry(struct inode *inode)
3294{
3295 return is_inode_flag_set(inode, flag: FI_INLINE_DENTRY);
3296}
3297
3298static inline int is_file(struct inode *inode, int type)
3299{
3300 return F2FS_I(inode)->i_advise & type;
3301}
3302
3303static inline void set_file(struct inode *inode, int type)
3304{
3305 if (is_file(inode, type))
3306 return;
3307 F2FS_I(inode)->i_advise |= type;
3308 f2fs_mark_inode_dirty_sync(inode, sync: true);
3309}
3310
3311static inline void clear_file(struct inode *inode, int type)
3312{
3313 if (!is_file(inode, type))
3314 return;
3315 F2FS_I(inode)->i_advise &= ~type;
3316 f2fs_mark_inode_dirty_sync(inode, sync: true);
3317}
3318
3319static inline bool f2fs_is_time_consistent(struct inode *inode)
3320{
3321 struct timespec64 ts = inode_get_atime(inode);
3322
3323 if (!timespec64_equal(a: F2FS_I(inode)->i_disk_time, b: &ts))
3324 return false;
3325 ts = inode_get_ctime(inode);
3326 if (!timespec64_equal(a: F2FS_I(inode)->i_disk_time + 1, b: &ts))
3327 return false;
3328 ts = inode_get_mtime(inode);
3329 if (!timespec64_equal(a: F2FS_I(inode)->i_disk_time + 2, b: &ts))
3330 return false;
3331 return true;
3332}
3333
3334static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3335{
3336 bool ret;
3337
3338 if (dsync) {
3339 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3340
3341 spin_lock(lock: &sbi->inode_lock[DIRTY_META]);
3342 ret = list_empty(head: &F2FS_I(inode)->gdirty_list);
3343 spin_unlock(lock: &sbi->inode_lock[DIRTY_META]);
3344 return ret;
3345 }
3346 if (!is_inode_flag_set(inode, flag: FI_AUTO_RECOVER) ||
3347 file_keep_isize(inode) ||
3348 i_size_read(inode) & ~PAGE_MASK)
3349 return false;
3350
3351 if (!f2fs_is_time_consistent(inode))
3352 return false;
3353
3354 spin_lock(lock: &F2FS_I(inode)->i_size_lock);
3355 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3356 spin_unlock(lock: &F2FS_I(inode)->i_size_lock);
3357
3358 return ret;
3359}
3360
3361static inline bool f2fs_readonly(struct super_block *sb)
3362{
3363 return sb_rdonly(sb);
3364}
3365
3366static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3367{
3368 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3369}
3370
3371static inline bool is_dot_dotdot(const u8 *name, size_t len)
3372{
3373 if (len == 1 && name[0] == '.')
3374 return true;
3375
3376 if (len == 2 && name[0] == '.' && name[1] == '.')
3377 return true;
3378
3379 return false;
3380}
3381
3382static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3383 size_t size, gfp_t flags)
3384{
3385 if (time_to_inject(sbi, FAULT_KMALLOC))
3386 return NULL;
3387
3388 return kmalloc(size, flags);
3389}
3390
3391static inline void *f2fs_getname(struct f2fs_sb_info *sbi)
3392{
3393 if (time_to_inject(sbi, FAULT_KMALLOC))
3394 return NULL;
3395
3396 return __getname();
3397}
3398
3399static inline void f2fs_putname(char *buf)
3400{
3401 __putname(buf);
3402}
3403
3404static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3405 size_t size, gfp_t flags)
3406{
3407 return f2fs_kmalloc(sbi, size, flags: flags | __GFP_ZERO);
3408}
3409
3410static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3411 size_t size, gfp_t flags)
3412{
3413 if (time_to_inject(sbi, FAULT_KVMALLOC))
3414 return NULL;
3415
3416 return kvmalloc(size, flags);
3417}
3418
3419static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3420 size_t size, gfp_t flags)
3421{
3422 return f2fs_kvmalloc(sbi, size, flags: flags | __GFP_ZERO);
3423}
3424
3425static inline int get_extra_isize(struct inode *inode)
3426{
3427 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3428}
3429
3430static inline int get_inline_xattr_addrs(struct inode *inode)
3431{
3432 return F2FS_I(inode)->i_inline_xattr_size;
3433}
3434
3435#define f2fs_get_inode_mode(i) \
3436 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3437 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3438
3439#define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32))
3440
3441#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3442 (offsetof(struct f2fs_inode, i_extra_end) - \
3443 offsetof(struct f2fs_inode, i_extra_isize)) \
3444
3445#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3446#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3447 ((offsetof(typeof(*(f2fs_inode)), field) + \
3448 sizeof((f2fs_inode)->field)) \
3449 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3450
3451#define __is_large_section(sbi) ((sbi)->segs_per_sec > 1)
3452
3453#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3454
3455bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3456 block_t blkaddr, int type);
3457static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3458 block_t blkaddr, int type)
3459{
3460 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) {
3461 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3462 blkaddr, type);
3463 f2fs_bug_on(sbi, 1);
3464 }
3465}
3466
3467static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3468{
3469 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3470 blkaddr == COMPRESS_ADDR)
3471 return false;
3472 return true;
3473}
3474
3475/*
3476 * file.c
3477 */
3478int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3479int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3480int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3481int f2fs_truncate(struct inode *inode);
3482int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path,
3483 struct kstat *stat, u32 request_mask, unsigned int flags);
3484int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
3485 struct iattr *attr);
3486int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3487void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3488int f2fs_precache_extents(struct inode *inode);
3489int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3490int f2fs_fileattr_set(struct mnt_idmap *idmap,
3491 struct dentry *dentry, struct fileattr *fa);
3492long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3493long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3494int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3495int f2fs_pin_file_control(struct inode *inode, bool inc);
3496
3497/*
3498 * inode.c
3499 */
3500void f2fs_set_inode_flags(struct inode *inode);
3501bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3502void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3503struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3504struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3505int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3506void f2fs_update_inode(struct inode *inode, struct page *node_page);
3507void f2fs_update_inode_page(struct inode *inode);
3508int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3509void f2fs_evict_inode(struct inode *inode);
3510void f2fs_handle_failed_inode(struct inode *inode);
3511
3512/*
3513 * namei.c
3514 */
3515int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3516 bool hot, bool set);
3517struct dentry *f2fs_get_parent(struct dentry *child);
3518int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
3519 struct inode **new_inode);
3520
3521/*
3522 * dir.c
3523 */
3524int f2fs_init_casefolded_name(const struct inode *dir,
3525 struct f2fs_filename *fname);
3526int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3527 int lookup, struct f2fs_filename *fname);
3528int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3529 struct f2fs_filename *fname);
3530void f2fs_free_filename(struct f2fs_filename *fname);
3531struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3532 const struct f2fs_filename *fname, int *max_slots);
3533int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3534 unsigned int start_pos, struct fscrypt_str *fstr);
3535void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3536 struct f2fs_dentry_ptr *d);
3537struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3538 const struct f2fs_filename *fname, struct page *dpage);
3539void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3540 unsigned int current_depth);
3541int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3542void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3543struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3544 const struct f2fs_filename *fname,
3545 struct page **res_page);
3546struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3547 const struct qstr *child, struct page **res_page);
3548struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3549ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3550 struct page **page);
3551void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3552 struct page *page, struct inode *inode);
3553bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3554 const struct f2fs_filename *fname);
3555void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3556 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3557 unsigned int bit_pos);
3558int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3559 struct inode *inode, nid_t ino, umode_t mode);
3560int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3561 struct inode *inode, nid_t ino, umode_t mode);
3562int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3563 struct inode *inode, nid_t ino, umode_t mode);
3564void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3565 struct inode *dir, struct inode *inode);
3566int f2fs_do_tmpfile(struct inode *inode, struct inode *dir);
3567bool f2fs_empty_dir(struct inode *dir);
3568
3569static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3570{
3571 if (fscrypt_is_nokey_name(dentry))
3572 return -ENOKEY;
3573 return f2fs_do_add_link(dir: d_inode(dentry: dentry->d_parent), name: &dentry->d_name,
3574 inode, ino: inode->i_ino, mode: inode->i_mode);
3575}
3576
3577/*
3578 * super.c
3579 */
3580int f2fs_inode_dirtied(struct inode *inode, bool sync);
3581void f2fs_inode_synced(struct inode *inode);
3582int f2fs_dquot_initialize(struct inode *inode);
3583int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3584int f2fs_quota_sync(struct super_block *sb, int type);
3585loff_t max_file_blocks(struct inode *inode);
3586void f2fs_quota_off_umount(struct super_block *sb);
3587void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag);
3588void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
3589 bool irq_context);
3590void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
3591void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error);
3592int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3593int f2fs_sync_fs(struct super_block *sb, int sync);
3594int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3595
3596/*
3597 * hash.c
3598 */
3599void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3600
3601/*
3602 * node.c
3603 */
3604struct node_info;
3605
3606int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3607bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3608bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3609void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3610void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3611void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3612int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3613bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3614bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3615int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3616 struct node_info *ni, bool checkpoint_context);
3617pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3618int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3619int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3620int f2fs_truncate_xattr_node(struct inode *inode);
3621int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3622 unsigned int seq_id);
3623bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
3624int f2fs_remove_inode_page(struct inode *inode);
3625struct page *f2fs_new_inode_page(struct inode *inode);
3626struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3627void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3628struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3629struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3630int f2fs_move_node_page(struct page *node_page, int gc_type);
3631void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3632int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3633 struct writeback_control *wbc, bool atomic,
3634 unsigned int *seq_id);
3635int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3636 struct writeback_control *wbc,
3637 bool do_balance, enum iostat_type io_type);
3638int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3639bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3640void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3641void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3642int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3643int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3644int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3645int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3646int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3647 unsigned int segno, struct f2fs_summary_block *sum);
3648void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
3649int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3650int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3651void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3652int __init f2fs_create_node_manager_caches(void);
3653void f2fs_destroy_node_manager_caches(void);
3654
3655/*
3656 * segment.c
3657 */
3658bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3659int f2fs_commit_atomic_write(struct inode *inode);
3660void f2fs_abort_atomic_write(struct inode *inode, bool clean);
3661void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3662void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3663int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3664int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3665int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3666void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3667void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3668bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3669int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
3670void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3671void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3672bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3673void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3674 struct cp_control *cpc);
3675void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3676block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3677int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3678void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3679int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3680bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3681void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3682void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3683void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3684void f2fs_get_new_segment(struct f2fs_sb_info *sbi,
3685 unsigned int *newseg, bool new_sec, int dir);
3686void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3687 unsigned int start, unsigned int end);
3688void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3689void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3690int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3691bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3692 struct cp_control *cpc);
3693struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3694void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3695 block_t blk_addr);
3696void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3697 enum iostat_type io_type);
3698void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3699void f2fs_outplace_write_data(struct dnode_of_data *dn,
3700 struct f2fs_io_info *fio);
3701int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3702void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3703 block_t old_blkaddr, block_t new_blkaddr,
3704 bool recover_curseg, bool recover_newaddr,
3705 bool from_gc);
3706void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3707 block_t old_addr, block_t new_addr,
3708 unsigned char version, bool recover_curseg,
3709 bool recover_newaddr);
3710void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3711 block_t old_blkaddr, block_t *new_blkaddr,
3712 struct f2fs_summary *sum, int type,
3713 struct f2fs_io_info *fio);
3714void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,
3715 block_t blkaddr, unsigned int blkcnt);
3716void f2fs_wait_on_page_writeback(struct page *page,
3717 enum page_type type, bool ordered, bool locked);
3718void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3719void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3720 block_t len);
3721void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3722void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3723int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3724 unsigned int val, int alloc);
3725void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3726int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3727int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3728int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3729void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3730int __init f2fs_create_segment_manager_caches(void);
3731void f2fs_destroy_segment_manager_caches(void);
3732int f2fs_rw_hint_to_seg_type(enum rw_hint hint);
3733unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3734 unsigned int segno);
3735unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3736 unsigned int segno);
3737
3738#define DEF_FRAGMENT_SIZE 4
3739#define MIN_FRAGMENT_SIZE 1
3740#define MAX_FRAGMENT_SIZE 512
3741
3742static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)
3743{
3744 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||
3745 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;
3746}
3747
3748/*
3749 * checkpoint.c
3750 */
3751void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
3752 unsigned char reason);
3753void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);
3754struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3755struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3756struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3757struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3758bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3759 block_t blkaddr, int type);
3760int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3761 int type, bool sync);
3762void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,
3763 unsigned int ra_blocks);
3764long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3765 long nr_to_write, enum iostat_type io_type);
3766void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3767void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3768void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3769bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3770void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3771 unsigned int devidx, int type);
3772bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3773 unsigned int devidx, int type);
3774int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3775void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3776void f2fs_add_orphan_inode(struct inode *inode);
3777void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3778int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3779int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3780void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);
3781void f2fs_remove_dirty_inode(struct inode *inode);
3782int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3783 bool from_cp);
3784void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3785u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3786int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3787void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3788int __init f2fs_create_checkpoint_caches(void);
3789void f2fs_destroy_checkpoint_caches(void);
3790int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3791int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3792void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3793void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3794
3795/*
3796 * data.c
3797 */
3798int __init f2fs_init_bioset(void);
3799void f2fs_destroy_bioset(void);
3800int f2fs_init_bio_entry_cache(void);
3801void f2fs_destroy_bio_entry_cache(void);
3802void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio,
3803 enum page_type type);
3804int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);
3805void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3806void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3807 struct inode *inode, struct page *page,
3808 nid_t ino, enum page_type type);
3809void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3810 struct bio **bio, struct page *page);
3811void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3812int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3813int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3814void f2fs_submit_page_write(struct f2fs_io_info *fio);
3815struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3816 block_t blk_addr, sector_t *sector);
3817int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3818void f2fs_set_data_blkaddr(struct dnode_of_data *dn);
3819void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3820int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3821int f2fs_reserve_new_block(struct dnode_of_data *dn);
3822int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index);
3823int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3824struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3825 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs);
3826struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index,
3827 pgoff_t *next_pgofs);
3828struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3829 bool for_write);
3830struct page *f2fs_get_new_data_page(struct inode *inode,
3831 struct page *ipage, pgoff_t index, bool new_i_size);
3832int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3833int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag);
3834int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3835 u64 start, u64 len);
3836int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3837bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3838bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3839int f2fs_write_single_data_page(struct page *page, int *submitted,
3840 struct bio **bio, sector_t *last_block,
3841 struct writeback_control *wbc,
3842 enum iostat_type io_type,
3843 int compr_blocks, bool allow_balance);
3844void f2fs_write_failed(struct inode *inode, loff_t to);
3845void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);
3846bool f2fs_release_folio(struct folio *folio, gfp_t wait);
3847bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3848void f2fs_clear_page_cache_dirty_tag(struct page *page);
3849int f2fs_init_post_read_processing(void);
3850void f2fs_destroy_post_read_processing(void);
3851int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3852void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3853extern const struct iomap_ops f2fs_iomap_ops;
3854
3855/*
3856 * gc.c
3857 */
3858int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3859void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3860block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3861int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);
3862void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3863int f2fs_resize_fs(struct file *filp, __u64 block_count);
3864int __init f2fs_create_garbage_collection_cache(void);
3865void f2fs_destroy_garbage_collection_cache(void);
3866/* victim selection function for cleaning and SSR */
3867int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result,
3868 int gc_type, int type, char alloc_mode,
3869 unsigned long long age);
3870
3871/*
3872 * recovery.c
3873 */
3874int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3875bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3876int __init f2fs_create_recovery_cache(void);
3877void f2fs_destroy_recovery_cache(void);
3878
3879/*
3880 * debug.c
3881 */
3882#ifdef CONFIG_F2FS_STAT_FS
3883struct f2fs_stat_info {
3884 struct list_head stat_list;
3885 struct f2fs_sb_info *sbi;
3886 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3887 int main_area_segs, main_area_sections, main_area_zones;
3888 unsigned long long hit_cached[NR_EXTENT_CACHES];
3889 unsigned long long hit_rbtree[NR_EXTENT_CACHES];
3890 unsigned long long total_ext[NR_EXTENT_CACHES];
3891 unsigned long long hit_total[NR_EXTENT_CACHES];
3892 int ext_tree[NR_EXTENT_CACHES];
3893 int zombie_tree[NR_EXTENT_CACHES];
3894 int ext_node[NR_EXTENT_CACHES];
3895 /* to count memory footprint */
3896 unsigned long long ext_mem[NR_EXTENT_CACHES];
3897 /* for read extent cache */
3898 unsigned long long hit_largest;
3899 /* for block age extent cache */
3900 unsigned long long allocated_data_blocks;
3901 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3902 int ndirty_data, ndirty_qdata;
3903 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3904 int nats, dirty_nats, sits, dirty_sits;
3905 int free_nids, avail_nids, alloc_nids;
3906 int total_count, utilization;
3907 int nr_wb_cp_data, nr_wb_data;
3908 int nr_rd_data, nr_rd_node, nr_rd_meta;
3909 int nr_dio_read, nr_dio_write;
3910 unsigned int io_skip_bggc, other_skip_bggc;
3911 int nr_flushing, nr_flushed, flush_list_empty;
3912 int nr_discarding, nr_discarded;
3913 int nr_discard_cmd;
3914 unsigned int undiscard_blks;
3915 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3916 unsigned int cur_ckpt_time, peak_ckpt_time;
3917 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3918 int compr_inode, swapfile_inode;
3919 unsigned long long compr_blocks;
3920 int aw_cnt, max_aw_cnt;
3921 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3922 unsigned int bimodal, avg_vblocks;
3923 int util_free, util_valid, util_invalid;
3924 int rsvd_segs, overp_segs;
3925 int dirty_count, node_pages, meta_pages, compress_pages;
3926 int compress_page_hit;
3927 int prefree_count, free_segs, free_secs;
3928 int cp_call_count[MAX_CALL_TYPE], cp_count;
3929 int gc_call_count[MAX_CALL_TYPE];
3930 int gc_segs[2][2];
3931 int gc_secs[2][2];
3932 int tot_blks, data_blks, node_blks;
3933 int bg_data_blks, bg_node_blks;
3934 int curseg[NR_CURSEG_TYPE];
3935 int cursec[NR_CURSEG_TYPE];
3936 int curzone[NR_CURSEG_TYPE];
3937 unsigned int dirty_seg[NR_CURSEG_TYPE];
3938 unsigned int full_seg[NR_CURSEG_TYPE];
3939 unsigned int valid_blks[NR_CURSEG_TYPE];
3940
3941 unsigned int meta_count[META_MAX];
3942 unsigned int segment_count[2];
3943 unsigned int block_count[2];
3944 unsigned int inplace_count;
3945 unsigned long long base_mem, cache_mem, page_mem;
3946};
3947
3948static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3949{
3950 return (struct f2fs_stat_info *)sbi->stat_info;
3951}
3952
3953#define stat_inc_cp_call_count(sbi, foreground) \
3954 atomic_inc(&sbi->cp_call_count[(foreground)])
3955#define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++)
3956#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3957#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3958#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3959#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3960#define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type]))
3961#define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type]))
3962#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3963#define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type]))
3964#define stat_inc_inline_xattr(inode) \
3965 do { \
3966 if (f2fs_has_inline_xattr(inode)) \
3967 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3968 } while (0)
3969#define stat_dec_inline_xattr(inode) \
3970 do { \
3971 if (f2fs_has_inline_xattr(inode)) \
3972 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3973 } while (0)
3974#define stat_inc_inline_inode(inode) \
3975 do { \
3976 if (f2fs_has_inline_data(inode)) \
3977 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3978 } while (0)
3979#define stat_dec_inline_inode(inode) \
3980 do { \
3981 if (f2fs_has_inline_data(inode)) \
3982 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
3983 } while (0)
3984#define stat_inc_inline_dir(inode) \
3985 do { \
3986 if (f2fs_has_inline_dentry(inode)) \
3987 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
3988 } while (0)
3989#define stat_dec_inline_dir(inode) \
3990 do { \
3991 if (f2fs_has_inline_dentry(inode)) \
3992 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
3993 } while (0)
3994#define stat_inc_compr_inode(inode) \
3995 do { \
3996 if (f2fs_compressed_file(inode)) \
3997 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
3998 } while (0)
3999#define stat_dec_compr_inode(inode) \
4000 do { \
4001 if (f2fs_compressed_file(inode)) \
4002 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
4003 } while (0)
4004#define stat_add_compr_blocks(inode, blocks) \
4005 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
4006#define stat_sub_compr_blocks(inode, blocks) \
4007 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
4008#define stat_inc_swapfile_inode(inode) \
4009 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))
4010#define stat_dec_swapfile_inode(inode) \
4011 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))
4012#define stat_inc_atomic_inode(inode) \
4013 (atomic_inc(&F2FS_I_SB(inode)->atomic_files))
4014#define stat_dec_atomic_inode(inode) \
4015 (atomic_dec(&F2FS_I_SB(inode)->atomic_files))
4016#define stat_inc_meta_count(sbi, blkaddr) \
4017 do { \
4018 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
4019 atomic_inc(&(sbi)->meta_count[META_CP]); \
4020 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
4021 atomic_inc(&(sbi)->meta_count[META_SIT]); \
4022 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
4023 atomic_inc(&(sbi)->meta_count[META_NAT]); \
4024 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
4025 atomic_inc(&(sbi)->meta_count[META_SSA]); \
4026 } while (0)
4027#define stat_inc_seg_type(sbi, curseg) \
4028 ((sbi)->segment_count[(curseg)->alloc_type]++)
4029#define stat_inc_block_count(sbi, curseg) \
4030 ((sbi)->block_count[(curseg)->alloc_type]++)
4031#define stat_inc_inplace_blocks(sbi) \
4032 (atomic_inc(&(sbi)->inplace_count))
4033#define stat_update_max_atomic_write(inode) \
4034 do { \
4035 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \
4036 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
4037 if (cur > max) \
4038 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
4039 } while (0)
4040#define stat_inc_gc_call_count(sbi, foreground) \
4041 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++)
4042#define stat_inc_gc_sec_count(sbi, type, gc_type) \
4043 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++)
4044#define stat_inc_gc_seg_count(sbi, type, gc_type) \
4045 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++)
4046
4047#define stat_inc_tot_blk_count(si, blks) \
4048 ((si)->tot_blks += (blks))
4049
4050#define stat_inc_data_blk_count(sbi, blks, gc_type) \
4051 do { \
4052 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4053 stat_inc_tot_blk_count(si, blks); \
4054 si->data_blks += (blks); \
4055 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4056 } while (0)
4057
4058#define stat_inc_node_blk_count(sbi, blks, gc_type) \
4059 do { \
4060 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4061 stat_inc_tot_blk_count(si, blks); \
4062 si->node_blks += (blks); \
4063 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4064 } while (0)
4065
4066int f2fs_build_stats(struct f2fs_sb_info *sbi);
4067void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
4068void __init f2fs_create_root_stats(void);
4069void f2fs_destroy_root_stats(void);
4070void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
4071#else
4072#define stat_inc_cp_call_count(sbi, foreground) do { } while (0)
4073#define stat_inc_cp_count(sbi) do { } while (0)
4074#define stat_io_skip_bggc_count(sbi) do { } while (0)
4075#define stat_other_skip_bggc_count(sbi) do { } while (0)
4076#define stat_inc_dirty_inode(sbi, type) do { } while (0)
4077#define stat_dec_dirty_inode(sbi, type) do { } while (0)
4078#define stat_inc_total_hit(sbi, type) do { } while (0)
4079#define stat_inc_rbtree_node_hit(sbi, type) do { } while (0)
4080#define stat_inc_largest_node_hit(sbi) do { } while (0)
4081#define stat_inc_cached_node_hit(sbi, type) do { } while (0)
4082#define stat_inc_inline_xattr(inode) do { } while (0)
4083#define stat_dec_inline_xattr(inode) do { } while (0)
4084#define stat_inc_inline_inode(inode) do { } while (0)
4085#define stat_dec_inline_inode(inode) do { } while (0)
4086#define stat_inc_inline_dir(inode) do { } while (0)
4087#define stat_dec_inline_dir(inode) do { } while (0)
4088#define stat_inc_compr_inode(inode) do { } while (0)
4089#define stat_dec_compr_inode(inode) do { } while (0)
4090#define stat_add_compr_blocks(inode, blocks) do { } while (0)
4091#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
4092#define stat_inc_swapfile_inode(inode) do { } while (0)
4093#define stat_dec_swapfile_inode(inode) do { } while (0)
4094#define stat_inc_atomic_inode(inode) do { } while (0)
4095#define stat_dec_atomic_inode(inode) do { } while (0)
4096#define stat_update_max_atomic_write(inode) do { } while (0)
4097#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
4098#define stat_inc_seg_type(sbi, curseg) do { } while (0)
4099#define stat_inc_block_count(sbi, curseg) do { } while (0)
4100#define stat_inc_inplace_blocks(sbi) do { } while (0)
4101#define stat_inc_gc_call_count(sbi, foreground) do { } while (0)
4102#define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0)
4103#define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0)
4104#define stat_inc_tot_blk_count(si, blks) do { } while (0)
4105#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
4106#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
4107
4108static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
4109static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
4110static inline void __init f2fs_create_root_stats(void) { }
4111static inline void f2fs_destroy_root_stats(void) { }
4112static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
4113#endif
4114
4115extern const struct file_operations f2fs_dir_operations;
4116extern const struct file_operations f2fs_file_operations;
4117extern const struct inode_operations f2fs_file_inode_operations;
4118extern const struct address_space_operations f2fs_dblock_aops;
4119extern const struct address_space_operations f2fs_node_aops;
4120extern const struct address_space_operations f2fs_meta_aops;
4121extern const struct inode_operations f2fs_dir_inode_operations;
4122extern const struct inode_operations f2fs_symlink_inode_operations;
4123extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
4124extern const struct inode_operations f2fs_special_inode_operations;
4125extern struct kmem_cache *f2fs_inode_entry_slab;
4126
4127/*
4128 * inline.c
4129 */
4130bool f2fs_may_inline_data(struct inode *inode);
4131bool f2fs_sanity_check_inline_data(struct inode *inode);
4132bool f2fs_may_inline_dentry(struct inode *inode);
4133void f2fs_do_read_inline_data(struct page *page, struct page *ipage);
4134void f2fs_truncate_inline_inode(struct inode *inode,
4135 struct page *ipage, u64 from);
4136int f2fs_read_inline_data(struct inode *inode, struct page *page);
4137int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
4138int f2fs_convert_inline_inode(struct inode *inode);
4139int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
4140int f2fs_write_inline_data(struct inode *inode, struct page *page);
4141int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
4142struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
4143 const struct f2fs_filename *fname,
4144 struct page **res_page);
4145int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
4146 struct page *ipage);
4147int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
4148 struct inode *inode, nid_t ino, umode_t mode);
4149void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
4150 struct page *page, struct inode *dir,
4151 struct inode *inode);
4152bool f2fs_empty_inline_dir(struct inode *dir);
4153int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
4154 struct fscrypt_str *fstr);
4155int f2fs_inline_data_fiemap(struct inode *inode,
4156 struct fiemap_extent_info *fieinfo,
4157 __u64 start, __u64 len);
4158
4159/*
4160 * shrinker.c
4161 */
4162unsigned long f2fs_shrink_count(struct shrinker *shrink,
4163 struct shrink_control *sc);
4164unsigned long f2fs_shrink_scan(struct shrinker *shrink,
4165 struct shrink_control *sc);
4166void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
4167void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
4168
4169/*
4170 * extent_cache.c
4171 */
4172bool sanity_check_extent_cache(struct inode *inode);
4173void f2fs_init_extent_tree(struct inode *inode);
4174void f2fs_drop_extent_tree(struct inode *inode);
4175void f2fs_destroy_extent_node(struct inode *inode);
4176void f2fs_destroy_extent_tree(struct inode *inode);
4177void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4178int __init f2fs_create_extent_cache(void);
4179void f2fs_destroy_extent_cache(void);
4180
4181/* read extent cache ops */
4182void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage);
4183bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,
4184 struct extent_info *ei);
4185bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index,
4186 block_t *blkaddr);
4187void f2fs_update_read_extent_cache(struct dnode_of_data *dn);
4188void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,
4189 pgoff_t fofs, block_t blkaddr, unsigned int len);
4190unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi,
4191 int nr_shrink);
4192
4193/* block age extent cache ops */
4194void f2fs_init_age_extent_tree(struct inode *inode);
4195bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,
4196 struct extent_info *ei);
4197void f2fs_update_age_extent_cache(struct dnode_of_data *dn);
4198void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,
4199 pgoff_t fofs, unsigned int len);
4200unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi,
4201 int nr_shrink);
4202
4203/*
4204 * sysfs.c
4205 */
4206#define MIN_RA_MUL 2
4207#define MAX_RA_MUL 256
4208
4209int __init f2fs_init_sysfs(void);
4210void f2fs_exit_sysfs(void);
4211int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4212void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4213
4214/* verity.c */
4215extern const struct fsverity_operations f2fs_verityops;
4216
4217/*
4218 * crypto support
4219 */
4220static inline bool f2fs_encrypted_file(struct inode *inode)
4221{
4222 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4223}
4224
4225static inline void f2fs_set_encrypted_inode(struct inode *inode)
4226{
4227#ifdef CONFIG_FS_ENCRYPTION
4228 file_set_encrypt(inode);
4229 f2fs_set_inode_flags(inode);
4230#endif
4231}
4232
4233/*
4234 * Returns true if the reads of the inode's data need to undergo some
4235 * postprocessing step, like decryption or authenticity verification.
4236 */
4237static inline bool f2fs_post_read_required(struct inode *inode)
4238{
4239 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4240 f2fs_compressed_file(inode);
4241}
4242
4243/*
4244 * compress.c
4245 */
4246#ifdef CONFIG_F2FS_FS_COMPRESSION
4247bool f2fs_is_compressed_page(struct page *page);
4248struct page *f2fs_compress_control_page(struct page *page);
4249int f2fs_prepare_compress_overwrite(struct inode *inode,
4250 struct page **pagep, pgoff_t index, void **fsdata);
4251bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4252 pgoff_t index, unsigned copied);
4253int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4254void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4255bool f2fs_is_compress_backend_ready(struct inode *inode);
4256bool f2fs_is_compress_level_valid(int alg, int lvl);
4257int __init f2fs_init_compress_mempool(void);
4258void f2fs_destroy_compress_mempool(void);
4259void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);
4260void f2fs_end_read_compressed_page(struct page *page, bool failed,
4261 block_t blkaddr, bool in_task);
4262bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4263bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4264bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,
4265 int index, int nr_pages, bool uptodate);
4266bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
4267void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4268int f2fs_write_multi_pages(struct compress_ctx *cc,
4269 int *submitted,
4270 struct writeback_control *wbc,
4271 enum iostat_type io_type);
4272int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4273void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,
4274 pgoff_t fofs, block_t blkaddr,
4275 unsigned int llen, unsigned int c_len);
4276int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4277 unsigned nr_pages, sector_t *last_block_in_bio,
4278 bool is_readahead, bool for_write);
4279struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4280void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
4281 bool in_task);
4282void f2fs_put_page_dic(struct page *page, bool in_task);
4283unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn);
4284int f2fs_init_compress_ctx(struct compress_ctx *cc);
4285void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4286void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4287int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4288void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4289int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4290void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4291int __init f2fs_init_compress_cache(void);
4292void f2fs_destroy_compress_cache(void);
4293struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4294void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4295void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4296 nid_t ino, block_t blkaddr);
4297bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4298 block_t blkaddr);
4299void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4300#define inc_compr_inode_stat(inode) \
4301 do { \
4302 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4303 sbi->compr_new_inode++; \
4304 } while (0)
4305#define add_compr_block_stat(inode, blocks) \
4306 do { \
4307 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4308 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4309 sbi->compr_written_block += blocks; \
4310 sbi->compr_saved_block += diff; \
4311 } while (0)
4312#else
4313static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4314static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4315{
4316 if (!f2fs_compressed_file(inode))
4317 return true;
4318 /* not support compression */
4319 return false;
4320}
4321static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; }
4322static inline struct page *f2fs_compress_control_page(struct page *page)
4323{
4324 WARN_ON_ONCE(1);
4325 return ERR_PTR(-EINVAL);
4326}
4327static inline int __init f2fs_init_compress_mempool(void) { return 0; }
4328static inline void f2fs_destroy_compress_mempool(void) { }
4329static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,
4330 bool in_task) { }
4331static inline void f2fs_end_read_compressed_page(struct page *page,
4332 bool failed, block_t blkaddr, bool in_task)
4333{
4334 WARN_ON_ONCE(1);
4335}
4336static inline void f2fs_put_page_dic(struct page *page, bool in_task)
4337{
4338 WARN_ON_ONCE(1);
4339}
4340static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; }
4341static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
4342static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4343static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4344static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4345static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4346static inline int __init f2fs_init_compress_cache(void) { return 0; }
4347static inline void f2fs_destroy_compress_cache(void) { }
4348static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4349 block_t blkaddr) { }
4350static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4351 struct page *page, nid_t ino, block_t blkaddr) { }
4352static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4353 struct page *page, block_t blkaddr) { return false; }
4354static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4355 nid_t ino) { }
4356#define inc_compr_inode_stat(inode) do { } while (0)
4357static inline void f2fs_update_read_extent_tree_range_compressed(
4358 struct inode *inode,
4359 pgoff_t fofs, block_t blkaddr,
4360 unsigned int llen, unsigned int c_len) { }
4361#endif
4362
4363static inline int set_compress_context(struct inode *inode)
4364{
4365#ifdef CONFIG_F2FS_FS_COMPRESSION
4366 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4367
4368 F2FS_I(inode)->i_compress_algorithm =
4369 F2FS_OPTION(sbi).compress_algorithm;
4370 F2FS_I(inode)->i_log_cluster_size =
4371 F2FS_OPTION(sbi).compress_log_size;
4372 F2FS_I(inode)->i_compress_flag =
4373 F2FS_OPTION(sbi).compress_chksum ?
4374 BIT(COMPRESS_CHKSUM) : 0;
4375 F2FS_I(inode)->i_cluster_size =
4376 BIT(F2FS_I(inode)->i_log_cluster_size);
4377 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4378 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4379 F2FS_OPTION(sbi).compress_level)
4380 F2FS_I(inode)->i_compress_level =
4381 F2FS_OPTION(sbi).compress_level;
4382 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4383 set_inode_flag(inode, flag: FI_COMPRESSED_FILE);
4384 stat_inc_compr_inode(inode);
4385 inc_compr_inode_stat(inode);
4386 f2fs_mark_inode_dirty_sync(inode, sync: true);
4387 return 0;
4388#else
4389 return -EOPNOTSUPP;
4390#endif
4391}
4392
4393static inline bool f2fs_disable_compressed_file(struct inode *inode)
4394{
4395 struct f2fs_inode_info *fi = F2FS_I(inode);
4396
4397 if (!f2fs_compressed_file(inode))
4398 return true;
4399 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))
4400 return false;
4401
4402 fi->i_flags &= ~F2FS_COMPR_FL;
4403 stat_dec_compr_inode(inode);
4404 clear_inode_flag(inode, flag: FI_COMPRESSED_FILE);
4405 f2fs_mark_inode_dirty_sync(inode, sync: true);
4406 return true;
4407}
4408
4409#define F2FS_FEATURE_FUNCS(name, flagname) \
4410static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4411{ \
4412 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4413}
4414
4415F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4416F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4417F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4418F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4419F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4420F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4421F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4422F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4423F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4424F2FS_FEATURE_FUNCS(verity, VERITY);
4425F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4426F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4427F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4428F2FS_FEATURE_FUNCS(readonly, RO);
4429
4430#ifdef CONFIG_BLK_DEV_ZONED
4431static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4432 block_t blkaddr)
4433{
4434 unsigned int zno = blkaddr / sbi->blocks_per_blkz;
4435
4436 return test_bit(zno, FDEV(devi).blkz_seq);
4437}
4438#endif
4439
4440static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi,
4441 struct block_device *bdev)
4442{
4443 int i;
4444
4445 if (!f2fs_is_multi_device(sbi))
4446 return 0;
4447
4448 for (i = 0; i < sbi->s_ndevs; i++)
4449 if (FDEV(i).bdev == bdev)
4450 return i;
4451
4452 WARN_ON(1);
4453 return -1;
4454}
4455
4456static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4457{
4458 return f2fs_sb_has_blkzoned(sbi);
4459}
4460
4461static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4462{
4463 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);
4464}
4465
4466static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4467{
4468 int i;
4469
4470 if (!f2fs_is_multi_device(sbi))
4471 return f2fs_bdev_support_discard(bdev: sbi->sb->s_bdev);
4472
4473 for (i = 0; i < sbi->s_ndevs; i++)
4474 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4475 return true;
4476 return false;
4477}
4478
4479static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4480{
4481 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4482 f2fs_hw_should_discard(sbi);
4483}
4484
4485static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4486{
4487 int i;
4488
4489 if (!f2fs_is_multi_device(sbi))
4490 return bdev_read_only(bdev: sbi->sb->s_bdev);
4491
4492 for (i = 0; i < sbi->s_ndevs; i++)
4493 if (bdev_read_only(FDEV(i).bdev))
4494 return true;
4495 return false;
4496}
4497
4498static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi)
4499{
4500 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi);
4501}
4502
4503static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4504{
4505 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4506}
4507
4508static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)
4509{
4510 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;
4511}
4512
4513static inline bool f2fs_may_compress(struct inode *inode)
4514{
4515 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4516 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) ||
4517 f2fs_is_mmap_file(inode))
4518 return false;
4519 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4520}
4521
4522static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4523 u64 blocks, bool add)
4524{
4525 struct f2fs_inode_info *fi = F2FS_I(inode);
4526 int diff = fi->i_cluster_size - blocks;
4527
4528 /* don't update i_compr_blocks if saved blocks were released */
4529 if (!add && !atomic_read(v: &fi->i_compr_blocks))
4530 return;
4531
4532 if (add) {
4533 atomic_add(i: diff, v: &fi->i_compr_blocks);
4534 stat_add_compr_blocks(inode, diff);
4535 } else {
4536 atomic_sub(i: diff, v: &fi->i_compr_blocks);
4537 stat_sub_compr_blocks(inode, diff);
4538 }
4539 f2fs_mark_inode_dirty_sync(inode, sync: true);
4540}
4541
4542static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,
4543 int flag)
4544{
4545 if (!f2fs_is_multi_device(sbi))
4546 return false;
4547 if (flag != F2FS_GET_BLOCK_DIO)
4548 return false;
4549 return sbi->aligned_blksize;
4550}
4551
4552static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4553{
4554 return fsverity_active(inode) &&
4555 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4556}
4557
4558#ifdef CONFIG_F2FS_FAULT_INJECTION
4559extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
4560 unsigned int type);
4561#else
4562#define f2fs_build_fault_attr(sbi, rate, type) do { } while (0)
4563#endif
4564
4565static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4566{
4567#ifdef CONFIG_QUOTA
4568 if (f2fs_sb_has_quota_ino(sbi))
4569 return true;
4570 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4571 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4572 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4573 return true;
4574#endif
4575 return false;
4576}
4577
4578static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4579{
4580 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4581}
4582
4583static inline void f2fs_io_schedule_timeout(long timeout)
4584{
4585 set_current_state(TASK_UNINTERRUPTIBLE);
4586 io_schedule_timeout(timeout);
4587}
4588
4589static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs,
4590 enum page_type type)
4591{
4592 if (unlikely(f2fs_cp_error(sbi)))
4593 return;
4594
4595 if (ofs == sbi->page_eio_ofs[type]) {
4596 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)
4597 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4598 } else {
4599 sbi->page_eio_ofs[type] = ofs;
4600 sbi->page_eio_cnt[type] = 0;
4601 }
4602}
4603
4604static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi)
4605{
4606 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sb: sbi->sb);
4607}
4608
4609#define EFSBADCRC EBADMSG /* Bad CRC detected */
4610#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4611
4612#endif /* _LINUX_F2FS_H */
4613

source code of linux/fs/f2fs/f2fs.h