Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
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 <linux/rw_hint.h>
28#include <crypto/hash.h>
29
30#include <linux/fscrypt.h>
31#include <linux/fsverity.h>
32
33struct pagevec;
34
35#ifdef CONFIG_F2FS_CHECK_FS
36#define f2fs_bug_on(sbi, condition) BUG_ON(condition)
37#else
38#define f2fs_bug_on(sbi, condition) \
39 do { \
40 if (WARN_ON(condition)) \
41 set_sbi_flag(sbi, SBI_NEED_FSCK); \
42 } while (0)
43#endif
44
45enum {
46 FAULT_KMALLOC,
47 FAULT_KVMALLOC,
48 FAULT_PAGE_ALLOC,
49 FAULT_PAGE_GET,
50 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */
51 FAULT_ALLOC_NID,
52 FAULT_ORPHAN,
53 FAULT_BLOCK,
54 FAULT_DIR_DEPTH,
55 FAULT_EVICT_INODE,
56 FAULT_TRUNCATE,
57 FAULT_READ_IO,
58 FAULT_CHECKPOINT,
59 FAULT_DISCARD,
60 FAULT_WRITE_IO,
61 FAULT_SLAB_ALLOC,
62 FAULT_DQUOT_INIT,
63 FAULT_LOCK_OP,
64 FAULT_BLKADDR_VALIDITY,
65 FAULT_BLKADDR_CONSISTENCE,
66 FAULT_NO_SEGMENT,
67 FAULT_MAX,
68};
69
70#ifdef CONFIG_F2FS_FAULT_INJECTION
71#define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0))
72
73struct f2fs_fault_info {
74 atomic_t inject_ops;
75 int inject_rate;
76 unsigned int inject_type;
77};
78
79extern const char *f2fs_fault_name[FAULT_MAX];
80#define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))
81
82/* maximum retry count for injected failure */
83#define DEFAULT_FAILURE_RETRY_COUNT 8
84#else
85#define DEFAULT_FAILURE_RETRY_COUNT 1
86#endif
87
88/*
89 * For mount options
90 */
91#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001
92#define F2FS_MOUNT_DISCARD 0x00000002
93#define F2FS_MOUNT_NOHEAP 0x00000004
94#define F2FS_MOUNT_XATTR_USER 0x00000008
95#define F2FS_MOUNT_POSIX_ACL 0x00000010
96#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020
97#define F2FS_MOUNT_INLINE_XATTR 0x00000040
98#define F2FS_MOUNT_INLINE_DATA 0x00000080
99#define F2FS_MOUNT_INLINE_DENTRY 0x00000100
100#define F2FS_MOUNT_FLUSH_MERGE 0x00000200
101#define F2FS_MOUNT_NOBARRIER 0x00000400
102#define F2FS_MOUNT_FASTBOOT 0x00000800
103#define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000
104#define F2FS_MOUNT_DATA_FLUSH 0x00002000
105#define F2FS_MOUNT_FAULT_INJECTION 0x00004000
106#define F2FS_MOUNT_USRQUOTA 0x00008000
107#define F2FS_MOUNT_GRPQUOTA 0x00010000
108#define F2FS_MOUNT_PRJQUOTA 0x00020000
109#define F2FS_MOUNT_QUOTA 0x00040000
110#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000
111#define F2FS_MOUNT_RESERVE_ROOT 0x00100000
112#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000
113#define F2FS_MOUNT_NORECOVERY 0x00400000
114#define F2FS_MOUNT_ATGC 0x00800000
115#define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000
116#define F2FS_MOUNT_GC_MERGE 0x02000000
117#define F2FS_MOUNT_COMPRESS_CACHE 0x04000000
118#define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000
119
120#define F2FS_OPTION(sbi) ((sbi)->mount_opt)
121#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)
122#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)
123#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)
124
125#define ver_after(a, b) (typecheck(unsigned long long, a) && \
126 typecheck(unsigned long long, b) && \
127 ((long long)((a) - (b)) > 0))
128
129typedef u32 block_t; /*
130 * should not change u32, since it is the on-disk block
131 * address format, __le32.
132 */
133typedef u32 nid_t;
134
135#define COMPRESS_EXT_NUM 16
136
137/*
138 * An implementation of an rwsem that is explicitly unfair to readers. This
139 * prevents priority inversion when a low-priority reader acquires the read lock
140 * while sleeping on the write lock but the write lock is needed by
141 * higher-priority clients.
142 */
143
144struct f2fs_rwsem {
145 struct rw_semaphore internal_rwsem;
146#ifdef CONFIG_F2FS_UNFAIR_RWSEM
147 wait_queue_head_t read_waiters;
148#endif
149};
150
151struct f2fs_mount_info {
152 unsigned int opt;
153 block_t root_reserved_blocks; /* root reserved blocks */
154 kuid_t s_resuid; /* reserved blocks for uid */
155 kgid_t s_resgid; /* reserved blocks for gid */
156 int active_logs; /* # of active logs */
157 int inline_xattr_size; /* inline xattr size */
158#ifdef CONFIG_F2FS_FAULT_INJECTION
159 struct f2fs_fault_info fault_info; /* For fault injection */
160#endif
161#ifdef CONFIG_QUOTA
162 /* Names of quota files with journalled quota */
163 char *s_qf_names[MAXQUOTAS];
164 int s_jquota_fmt; /* Format of quota to use */
165#endif
166 /* For which write hints are passed down to block layer */
167 int alloc_mode; /* segment allocation policy */
168 int fsync_mode; /* fsync policy */
169 int fs_mode; /* fs mode: LFS or ADAPTIVE */
170 int bggc_mode; /* bggc mode: off, on or sync */
171 int memory_mode; /* memory mode */
172 int errors; /* errors parameter */
173 int discard_unit; /*
174 * discard command's offset/size should
175 * be aligned to this unit: block,
176 * segment or section
177 */
178 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */
179 block_t unusable_cap_perc; /* percentage for cap */
180 block_t unusable_cap; /* Amount of space allowed to be
181 * unusable when disabling checkpoint
182 */
183
184 /* For compression */
185 unsigned char compress_algorithm; /* algorithm type */
186 unsigned char compress_log_size; /* cluster log size */
187 unsigned char compress_level; /* compress level */
188 bool compress_chksum; /* compressed data chksum */
189 unsigned char compress_ext_cnt; /* extension count */
190 unsigned char nocompress_ext_cnt; /* nocompress extension count */
191 int compress_mode; /* compression mode */
192 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
193 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */
194};
195
196#define F2FS_FEATURE_ENCRYPT 0x00000001
197#define F2FS_FEATURE_BLKZONED 0x00000002
198#define F2FS_FEATURE_ATOMIC_WRITE 0x00000004
199#define F2FS_FEATURE_EXTRA_ATTR 0x00000008
200#define F2FS_FEATURE_PRJQUOTA 0x00000010
201#define F2FS_FEATURE_INODE_CHKSUM 0x00000020
202#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040
203#define F2FS_FEATURE_QUOTA_INO 0x00000080
204#define F2FS_FEATURE_INODE_CRTIME 0x00000100
205#define F2FS_FEATURE_LOST_FOUND 0x00000200
206#define F2FS_FEATURE_VERITY 0x00000400
207#define F2FS_FEATURE_SB_CHKSUM 0x00000800
208#define F2FS_FEATURE_CASEFOLD 0x00001000
209#define F2FS_FEATURE_COMPRESSION 0x00002000
210#define F2FS_FEATURE_RO 0x00004000
211
212#define __F2FS_HAS_FEATURE(raw_super, mask) \
213 ((raw_super->feature & cpu_to_le32(mask)) != 0)
214#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
215
216/*
217 * Default values for user and/or group using reserved blocks
218 */
219#define F2FS_DEF_RESUID 0
220#define F2FS_DEF_RESGID 0
221
222/*
223 * For checkpoint manager
224 */
225enum {
226 NAT_BITMAP,
227 SIT_BITMAP
228};
229
230#define CP_UMOUNT 0x00000001
231#define CP_FASTBOOT 0x00000002
232#define CP_SYNC 0x00000004
233#define CP_RECOVERY 0x00000008
234#define CP_DISCARD 0x00000010
235#define CP_TRIMMED 0x00000020
236#define CP_PAUSE 0x00000040
237#define CP_RESIZE 0x00000080
238
239#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */
240#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */
241#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */
242#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */
243#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */
244#define DEF_CP_INTERVAL 60 /* 60 secs */
245#define DEF_IDLE_INTERVAL 5 /* 5 secs */
246#define DEF_DISABLE_INTERVAL 5 /* 5 secs */
247#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */
248#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */
249
250struct cp_control {
251 int reason;
252 __u64 trim_start;
253 __u64 trim_end;
254 __u64 trim_minlen;
255};
256
257/*
258 * indicate meta/data type
259 */
260enum {
261 META_CP,
262 META_NAT,
263 META_SIT,
264 META_SSA,
265 META_MAX,
266 META_POR,
267 DATA_GENERIC, /* check range only */
268 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */
269 DATA_GENERIC_ENHANCE_READ, /*
270 * strong check on range and segment
271 * bitmap but no warning due to race
272 * condition of read on truncated area
273 * by extent_cache
274 */
275 DATA_GENERIC_ENHANCE_UPDATE, /*
276 * strong check on range and segment
277 * bitmap for update case
278 */
279 META_GENERIC,
280};
281
282/* for the list of ino */
283enum {
284 ORPHAN_INO, /* for orphan ino list */
285 APPEND_INO, /* for append ino list */
286 UPDATE_INO, /* for update ino list */
287 TRANS_DIR_INO, /* for transactions dir ino list */
288 FLUSH_INO, /* for multiple device flushing */
289 MAX_INO_ENTRY, /* max. list */
290};
291
292struct ino_entry {
293 struct list_head list; /* list head */
294 nid_t ino; /* inode number */
295 unsigned int dirty_device; /* dirty device bitmap */
296};
297
298/* for the list of inodes to be GCed */
299struct inode_entry {
300 struct list_head list; /* list head */
301 struct inode *inode; /* vfs inode pointer */
302};
303
304struct fsync_node_entry {
305 struct list_head list; /* list head */
306 struct page *page; /* warm node page pointer */
307 unsigned int seq_id; /* sequence id */
308};
309
310struct ckpt_req {
311 struct completion wait; /* completion for checkpoint done */
312 struct llist_node llnode; /* llist_node to be linked in wait queue */
313 int ret; /* return code of checkpoint */
314 ktime_t queue_time; /* request queued time */
315};
316
317struct ckpt_req_control {
318 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */
319 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */
320 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */
321 atomic_t issued_ckpt; /* # of actually issued ckpts */
322 atomic_t total_ckpt; /* # of total ckpts */
323 atomic_t queued_ckpt; /* # of queued ckpts */
324 struct llist_head issue_list; /* list for command issue */
325 spinlock_t stat_lock; /* lock for below checkpoint time stats */
326 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */
327 unsigned int peak_time; /* peak wait time in msec until now */
328};
329
330/* for the bitmap indicate blocks to be discarded */
331struct discard_entry {
332 struct list_head list; /* list head */
333 block_t start_blkaddr; /* start blockaddr of current segment */
334 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */
335};
336
337/* minimum discard granularity, unit: block count */
338#define MIN_DISCARD_GRANULARITY 1
339/* default discard granularity of inner discard thread, unit: block count */
340#define DEFAULT_DISCARD_GRANULARITY 16
341/* default maximum discard granularity of ordered discard, unit: block count */
342#define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16
343
344/* max discard pend list number */
345#define MAX_PLIST_NUM 512
346#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \
347 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))
348
349enum {
350 D_PREP, /* initial */
351 D_PARTIAL, /* partially submitted */
352 D_SUBMIT, /* all submitted */
353 D_DONE, /* finished */
354};
355
356struct discard_info {
357 block_t lstart; /* logical start address */
358 block_t len; /* length */
359 block_t start; /* actual start address in dev */
360};
361
362struct discard_cmd {
363 struct rb_node rb_node; /* rb node located in rb-tree */
364 struct discard_info di; /* discard info */
365 struct list_head list; /* command list */
366 struct completion wait; /* compleation */
367 struct block_device *bdev; /* bdev */
368 unsigned short ref; /* reference count */
369 unsigned char state; /* state */
370 unsigned char queued; /* queued discard */
371 int error; /* bio error */
372 spinlock_t lock; /* for state/bio_ref updating */
373 unsigned short bio_ref; /* bio reference count */
374};
375
376enum {
377 DPOLICY_BG,
378 DPOLICY_FORCE,
379 DPOLICY_FSTRIM,
380 DPOLICY_UMOUNT,
381 MAX_DPOLICY,
382};
383
384enum {
385 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */
386 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */
387 DPOLICY_IO_AWARE_MAX,
388};
389
390struct discard_policy {
391 int type; /* type of discard */
392 unsigned int min_interval; /* used for candidates exist */
393 unsigned int mid_interval; /* used for device busy */
394 unsigned int max_interval; /* used for candidates not exist */
395 unsigned int max_requests; /* # of discards issued per round */
396 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */
397 bool io_aware; /* issue discard in idle time */
398 bool sync; /* submit discard with REQ_SYNC flag */
399 bool ordered; /* issue discard by lba order */
400 bool timeout; /* discard timeout for put_super */
401 unsigned int granularity; /* discard granularity */
402};
403
404struct discard_cmd_control {
405 struct task_struct *f2fs_issue_discard; /* discard thread */
406 struct list_head entry_list; /* 4KB discard entry list */
407 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */
408 struct list_head wait_list; /* store on-flushing entries */
409 struct list_head fstrim_list; /* in-flight discard from fstrim */
410 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */
411 struct mutex cmd_lock;
412 unsigned int nr_discards; /* # of discards in the list */
413 unsigned int max_discards; /* max. discards to be issued */
414 unsigned int max_discard_request; /* max. discard request per round */
415 unsigned int min_discard_issue_time; /* min. interval between discard issue */
416 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */
417 unsigned int max_discard_issue_time; /* max. interval between discard issue */
418 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */
419 unsigned int discard_urgent_util; /* utilization which issue discard proactively */
420 unsigned int discard_granularity; /* discard granularity */
421 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */
422 unsigned int discard_io_aware; /* io_aware policy */
423 unsigned int undiscard_blks; /* # of undiscard blocks */
424 unsigned int next_pos; /* next discard position */
425 atomic_t issued_discard; /* # of issued discard */
426 atomic_t queued_discard; /* # of queued discard */
427 atomic_t discard_cmd_cnt; /* # of cached cmd count */
428 struct rb_root_cached root; /* root of discard rb-tree */
429 bool rbtree_check; /* config for consistence check */
430 bool discard_wake; /* to wake up discard thread */
431};
432
433/* for the list of fsync inodes, used only during recovery */
434struct fsync_inode_entry {
435 struct list_head list; /* list head */
436 struct inode *inode; /* vfs inode pointer */
437 block_t blkaddr; /* block address locating the last fsync */
438 block_t last_dentry; /* block address locating the last dentry */
439};
440
441#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))
442#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))
443
444#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)
445#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)
446#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)
447#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)
448
449#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))
450#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))
451
452static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)
453{
454 int before = nats_in_cursum(journal);
455
456 journal->n_nats = cpu_to_le16(before + i);
457 return before;
458}
459
460static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)
461{
462 int before = sits_in_cursum(journal);
463
464 journal->n_sits = cpu_to_le16(before + i);
465 return before;
466}
467
468static inline bool __has_cursum_space(struct f2fs_journal *journal,
469 int size, int type)
470{
471 if (type == NAT_JOURNAL)
472 return size <= MAX_NAT_JENTRIES(journal);
473 return size <= MAX_SIT_JENTRIES(journal);
474}
475
476/* for inline stuff */
477#define DEF_INLINE_RESERVED_SIZE 1
478static inline int get_extra_isize(struct inode *inode);
479static inline int get_inline_xattr_addrs(struct inode *inode);
480#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \
481 (CUR_ADDRS_PER_INODE(inode) - \
482 get_inline_xattr_addrs(inode) - \
483 DEF_INLINE_RESERVED_SIZE))
484
485/* for inline dir */
486#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \
487 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
488 BITS_PER_BYTE + 1))
489#define INLINE_DENTRY_BITMAP_SIZE(inode) \
490 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)
491#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \
492 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \
493 NR_INLINE_DENTRY(inode) + \
494 INLINE_DENTRY_BITMAP_SIZE(inode)))
495
496/*
497 * For INODE and NODE manager
498 */
499/* for directory operations */
500
501struct f2fs_filename {
502 /*
503 * The filename the user specified. This is NULL for some
504 * filesystem-internal operations, e.g. converting an inline directory
505 * to a non-inline one, or roll-forward recovering an encrypted dentry.
506 */
507 const struct qstr *usr_fname;
508
509 /*
510 * The on-disk filename. For encrypted directories, this is encrypted.
511 * This may be NULL for lookups in an encrypted dir without the key.
512 */
513 struct fscrypt_str disk_name;
514
515 /* The dirhash of this filename */
516 f2fs_hash_t hash;
517
518#ifdef CONFIG_FS_ENCRYPTION
519 /*
520 * For lookups in encrypted directories: either the buffer backing
521 * disk_name, or a buffer that holds the decoded no-key name.
522 */
523 struct fscrypt_str crypto_buf;
524#endif
525#if IS_ENABLED(CONFIG_UNICODE)
526 /*
527 * For casefolded directories: the casefolded name, but it's left NULL
528 * if the original name is not valid Unicode, if the original name is
529 * "." or "..", if the directory is both casefolded and encrypted and
530 * its encryption key is unavailable, or if the filesystem is doing an
531 * internal operation where usr_fname is also NULL. In all these cases
532 * we fall back to treating the name as an opaque byte sequence.
533 */
534 struct fscrypt_str cf_name;
535#endif
536};
537
538struct f2fs_dentry_ptr {
539 struct inode *inode;
540 void *bitmap;
541 struct f2fs_dir_entry *dentry;
542 __u8 (*filename)[F2FS_SLOT_LEN];
543 int max;
544 int nr_bitmap;
545};
546
547static inline void make_dentry_ptr_block(struct inode *inode,
548 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)
549{
550 d->inode = inode;
551 d->max = NR_DENTRY_IN_BLOCK;
552 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;
553 d->bitmap = t->dentry_bitmap;
554 d->dentry = t->dentry;
555 d->filename = t->filename;
556}
557
558static inline void make_dentry_ptr_inline(struct inode *inode,
559 struct f2fs_dentry_ptr *d, void *t)
560{
561 int entry_cnt = NR_INLINE_DENTRY(inode);
562 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);
563 int reserved_size = INLINE_RESERVED_SIZE(inode);
564
565 d->inode = inode;
566 d->max = entry_cnt;
567 d->nr_bitmap = bitmap_size;
568 d->bitmap = t;
569 d->dentry = t + bitmap_size + reserved_size;
570 d->filename = t + bitmap_size + reserved_size +
571 SIZE_OF_DIR_ENTRY * entry_cnt;
572}
573
574/*
575 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1
576 * as its node offset to distinguish from index node blocks.
577 * But some bits are used to mark the node block.
578 */
579#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \
580 >> OFFSET_BIT_SHIFT)
581enum {
582 ALLOC_NODE, /* allocate a new node page if needed */
583 LOOKUP_NODE, /* look up a node without readahead */
584 LOOKUP_NODE_RA, /*
585 * look up a node with readahead called
586 * by get_data_block.
587 */
588};
589
590#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */
591
592/* congestion wait timeout value, default: 20ms */
593#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))
594
595/* maximum retry quota flush count */
596#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8
597
598/* maximum retry of EIO'ed page */
599#define MAX_RETRY_PAGE_EIO 100
600
601#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */
602
603#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */
604
605/* dirty segments threshold for triggering CP */
606#define DEFAULT_DIRTY_THRESHOLD 4
607
608#define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS
609#define RECOVERY_MIN_RA_BLOCKS 1
610
611#define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */
612
613/* for in-memory extent cache entry */
614#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */
615
616/* number of extent info in extent cache we try to shrink */
617#define READ_EXTENT_CACHE_SHRINK_NUMBER 128
618
619/* number of age extent info in extent cache we try to shrink */
620#define AGE_EXTENT_CACHE_SHRINK_NUMBER 128
621#define LAST_AGE_WEIGHT 30
622#define SAME_AGE_REGION 1024
623
624/*
625 * Define data block with age less than 1GB as hot data
626 * define data block with age less than 10GB but more than 1GB as warm data
627 */
628#define DEF_HOT_DATA_AGE_THRESHOLD 262144
629#define DEF_WARM_DATA_AGE_THRESHOLD 2621440
630
631/* extent cache type */
632enum extent_type {
633 EX_READ,
634 EX_BLOCK_AGE,
635 NR_EXTENT_CACHES,
636};
637
638struct extent_info {
639 unsigned int fofs; /* start offset in a file */
640 unsigned int len; /* length of the extent */
641 union {
642 /* read extent_cache */
643 struct {
644 /* start block address of the extent */
645 block_t blk;
646#ifdef CONFIG_F2FS_FS_COMPRESSION
647 /* physical extent length of compressed blocks */
648 unsigned int c_len;
649#endif
650 };
651 /* block age extent_cache */
652 struct {
653 /* block age of the extent */
654 unsigned long long age;
655 /* last total blocks allocated */
656 unsigned long long last_blocks;
657 };
658 };
659};
660
661struct extent_node {
662 struct rb_node rb_node; /* rb node located in rb-tree */
663 struct extent_info ei; /* extent info */
664 struct list_head list; /* node in global extent list of sbi */
665 struct extent_tree *et; /* extent tree pointer */
666};
667
668struct extent_tree {
669 nid_t ino; /* inode number */
670 enum extent_type type; /* keep the extent tree type */
671 struct rb_root_cached root; /* root of extent info rb-tree */
672 struct extent_node *cached_en; /* recently accessed extent node */
673 struct list_head list; /* to be used by sbi->zombie_list */
674 rwlock_t lock; /* protect extent info rb-tree */
675 atomic_t node_cnt; /* # of extent node in rb-tree*/
676 bool largest_updated; /* largest extent updated */
677 struct extent_info largest; /* largest cached extent for EX_READ */
678};
679
680struct extent_tree_info {
681 struct radix_tree_root extent_tree_root;/* cache extent cache entries */
682 struct mutex extent_tree_lock; /* locking extent radix tree */
683 struct list_head extent_list; /* lru list for shrinker */
684 spinlock_t extent_lock; /* locking extent lru list */
685 atomic_t total_ext_tree; /* extent tree count */
686 struct list_head zombie_list; /* extent zombie tree list */
687 atomic_t total_zombie_tree; /* extent zombie tree count */
688 atomic_t total_ext_node; /* extent info count */
689};
690
691/*
692 * State of block returned by f2fs_map_blocks.
693 */
694#define F2FS_MAP_NEW (1U << 0)
695#define F2FS_MAP_MAPPED (1U << 1)
696#define F2FS_MAP_DELALLOC (1U << 2)
697#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\
698 F2FS_MAP_DELALLOC)
699
700struct f2fs_map_blocks {
701 struct block_device *m_bdev; /* for multi-device dio */
702 block_t m_pblk;
703 block_t m_lblk;
704 unsigned int m_len;
705 unsigned int m_flags;
706 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */
707 pgoff_t *m_next_extent; /* point to next possible extent */
708 int m_seg_type;
709 bool m_may_create; /* indicate it is from write path */
710 bool m_multidev_dio; /* indicate it allows multi-device dio */
711};
712
713/* for flag in get_data_block */
714enum {
715 F2FS_GET_BLOCK_DEFAULT,
716 F2FS_GET_BLOCK_FIEMAP,
717 F2FS_GET_BLOCK_BMAP,
718 F2FS_GET_BLOCK_DIO,
719 F2FS_GET_BLOCK_PRE_DIO,
720 F2FS_GET_BLOCK_PRE_AIO,
721 F2FS_GET_BLOCK_PRECACHE,
722};
723
724/*
725 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.
726 */
727#define FADVISE_COLD_BIT 0x01
728#define FADVISE_LOST_PINO_BIT 0x02
729#define FADVISE_ENCRYPT_BIT 0x04
730#define FADVISE_ENC_NAME_BIT 0x08
731#define FADVISE_KEEP_SIZE_BIT 0x10
732#define FADVISE_HOT_BIT 0x20
733#define FADVISE_VERITY_BIT 0x40
734#define FADVISE_TRUNC_BIT 0x80
735
736#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)
737
738#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
739#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
740#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
741
742#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
743#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
744#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
745
746#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)
747#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)
748
749#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)
750#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)
751
752#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)
753#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)
754
755#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)
756#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)
757#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)
758
759#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)
760#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)
761
762#define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)
763#define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)
764#define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)
765
766#define DEF_DIR_LEVEL 0
767
768/* used for f2fs_inode_info->flags */
769enum {
770 FI_NEW_INODE, /* indicate newly allocated inode */
771 FI_DIRTY_INODE, /* indicate inode is dirty or not */
772 FI_AUTO_RECOVER, /* indicate inode is recoverable */
773 FI_DIRTY_DIR, /* indicate directory has dirty pages */
774 FI_INC_LINK, /* need to increment i_nlink */
775 FI_ACL_MODE, /* indicate acl mode */
776 FI_NO_ALLOC, /* should not allocate any blocks */
777 FI_FREE_NID, /* free allocated nide */
778 FI_NO_EXTENT, /* not to use the extent cache */
779 FI_INLINE_XATTR, /* used for inline xattr */
780 FI_INLINE_DATA, /* used for inline data*/
781 FI_INLINE_DENTRY, /* used for inline dentry */
782 FI_APPEND_WRITE, /* inode has appended data */
783 FI_UPDATE_WRITE, /* inode has in-place-update data */
784 FI_NEED_IPU, /* used for ipu per file */
785 FI_ATOMIC_FILE, /* indicate atomic file */
786 FI_DATA_EXIST, /* indicate data exists */
787 FI_INLINE_DOTS, /* indicate inline dot dentries */
788 FI_SKIP_WRITES, /* should skip data page writeback */
789 FI_OPU_WRITE, /* used for opu per file */
790 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */
791 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */
792 FI_HOT_DATA, /* indicate file is hot */
793 FI_EXTRA_ATTR, /* indicate file has extra attribute */
794 FI_PROJ_INHERIT, /* indicate file inherits projectid */
795 FI_PIN_FILE, /* indicate file should not be gced */
796 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
797 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */
798 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */
799 FI_MMAP_FILE, /* indicate file was mmapped */
800 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */
801 FI_COMPRESS_RELEASED, /* compressed blocks were released */
802 FI_ALIGNED_WRITE, /* enable aligned write */
803 FI_COW_FILE, /* indicate COW file */
804 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */
805 FI_ATOMIC_REPLACE, /* indicate atomic replace */
806 FI_MAX, /* max flag, never be used */
807};
808
809struct f2fs_inode_info {
810 struct inode vfs_inode; /* serve a vfs inode */
811 unsigned long i_flags; /* keep an inode flags for ioctl */
812 unsigned char i_advise; /* use to give file attribute hints */
813 unsigned char i_dir_level; /* use for dentry level for large dir */
814 union {
815 unsigned int i_current_depth; /* only for directory depth */
816 unsigned short i_gc_failures; /* for gc failure statistic */
817 };
818 unsigned int i_pino; /* parent inode number */
819 umode_t i_acl_mode; /* keep file acl mode temporarily */
820
821 /* Use below internally in f2fs*/
822 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */
823 struct f2fs_rwsem i_sem; /* protect fi info */
824 atomic_t dirty_pages; /* # of dirty pages */
825 f2fs_hash_t chash; /* hash value of given file name */
826 unsigned int clevel; /* maximum level of given file name */
827 struct task_struct *task; /* lookup and create consistency */
828 struct task_struct *cp_task; /* separate cp/wb IO stats*/
829 struct task_struct *wb_task; /* indicate inode is in context of writeback */
830 nid_t i_xattr_nid; /* node id that contains xattrs */
831 loff_t last_disk_size; /* lastly written file size */
832 spinlock_t i_size_lock; /* protect last_disk_size */
833
834#ifdef CONFIG_QUOTA
835 struct dquot __rcu *i_dquot[MAXQUOTAS];
836
837 /* quota space reservation, managed internally by quota code */
838 qsize_t i_reserved_quota;
839#endif
840 struct list_head dirty_list; /* dirty list for dirs and files */
841 struct list_head gdirty_list; /* linked in global dirty list */
842 struct task_struct *atomic_write_task; /* store atomic write task */
843 struct extent_tree *extent_tree[NR_EXTENT_CACHES];
844 /* cached extent_tree entry */
845 struct inode *cow_inode; /* copy-on-write inode for atomic write */
846
847 /* avoid racing between foreground op and gc */
848 struct f2fs_rwsem i_gc_rwsem[2];
849 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */
850
851 int i_extra_isize; /* size of extra space located in i_addr */
852 kprojid_t i_projid; /* id for project quota */
853 int i_inline_xattr_size; /* inline xattr size */
854 struct timespec64 i_crtime; /* inode creation time */
855 struct timespec64 i_disk_time[3];/* inode disk times */
856
857 /* for file compress */
858 atomic_t i_compr_blocks; /* # of compressed blocks */
859 unsigned char i_compress_algorithm; /* algorithm type */
860 unsigned char i_log_cluster_size; /* log of cluster size */
861 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */
862 unsigned char i_compress_flag; /* compress flag */
863 unsigned int i_cluster_size; /* cluster size */
864
865 unsigned int atomic_write_cnt;
866 loff_t original_i_size; /* original i_size before atomic write */
867};
868
869static inline void get_read_extent_info(struct extent_info *ext,
870 struct f2fs_extent *i_ext)
871{
872 ext->fofs = le32_to_cpu(i_ext->fofs);
873 ext->blk = le32_to_cpu(i_ext->blk);
874 ext->len = le32_to_cpu(i_ext->len);
875}
876
877static inline void set_raw_read_extent(struct extent_info *ext,
878 struct f2fs_extent *i_ext)
879{
880 i_ext->fofs = cpu_to_le32(ext->fofs);
881 i_ext->blk = cpu_to_le32(ext->blk);
882 i_ext->len = cpu_to_le32(ext->len);
883}
884
885static inline bool __is_discard_mergeable(struct discard_info *back,
886 struct discard_info *front, unsigned int max_len)
887{
888 return (back->lstart + back->len == front->lstart) &&
889 (back->len + front->len <= max_len);
890}
891
892static inline bool __is_discard_back_mergeable(struct discard_info *cur,
893 struct discard_info *back, unsigned int max_len)
894{
895 return __is_discard_mergeable(back, cur, max_len);
896}
897
898static inline bool __is_discard_front_mergeable(struct discard_info *cur,
899 struct discard_info *front, unsigned int max_len)
900{
901 return __is_discard_mergeable(cur, front, max_len);
902}
903
904/*
905 * For free nid management
906 */
907enum nid_state {
908 FREE_NID, /* newly added to free nid list */
909 PREALLOC_NID, /* it is preallocated */
910 MAX_NID_STATE,
911};
912
913enum nat_state {
914 TOTAL_NAT,
915 DIRTY_NAT,
916 RECLAIMABLE_NAT,
917 MAX_NAT_STATE,
918};
919
920struct f2fs_nm_info {
921 block_t nat_blkaddr; /* base disk address of NAT */
922 nid_t max_nid; /* maximum possible node ids */
923 nid_t available_nids; /* # of available node ids */
924 nid_t next_scan_nid; /* the next nid to be scanned */
925 nid_t max_rf_node_blocks; /* max # of nodes for recovery */
926 unsigned int ram_thresh; /* control the memory footprint */
927 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
928 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
929
930 /* NAT cache management */
931 struct radix_tree_root nat_root;/* root of the nat entry cache */
932 struct radix_tree_root nat_set_root;/* root of the nat set cache */
933 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */
934 struct list_head nat_entries; /* cached nat entry list (clean) */
935 spinlock_t nat_list_lock; /* protect clean nat entry list */
936 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */
937 unsigned int nat_blocks; /* # of nat blocks */
938
939 /* free node ids management */
940 struct radix_tree_root free_nid_root;/* root of the free_nid cache */
941 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */
942 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */
943 spinlock_t nid_list_lock; /* protect nid lists ops */
944 struct mutex build_lock; /* lock for build free nids */
945 unsigned char **free_nid_bitmap;
946 unsigned char *nat_block_bitmap;
947 unsigned short *free_nid_count; /* free nid count of NAT block */
948
949 /* for checkpoint */
950 char *nat_bitmap; /* NAT bitmap pointer */
951
952 unsigned int nat_bits_blocks; /* # of nat bits blocks */
953 unsigned char *nat_bits; /* NAT bits blocks */
954 unsigned char *full_nat_bits; /* full NAT pages */
955 unsigned char *empty_nat_bits; /* empty NAT pages */
956#ifdef CONFIG_F2FS_CHECK_FS
957 char *nat_bitmap_mir; /* NAT bitmap mirror */
958#endif
959 int bitmap_size; /* bitmap size */
960};
961
962/*
963 * this structure is used as one of function parameters.
964 * all the information are dedicated to a given direct node block determined
965 * by the data offset in a file.
966 */
967struct dnode_of_data {
968 struct inode *inode; /* vfs inode pointer */
969 struct page *inode_page; /* its inode page, NULL is possible */
970 struct page *node_page; /* cached direct node page */
971 nid_t nid; /* node id of the direct node block */
972 unsigned int ofs_in_node; /* data offset in the node page */
973 bool inode_page_locked; /* inode page is locked or not */
974 bool node_changed; /* is node block changed */
975 char cur_level; /* level of hole node page */
976 char max_level; /* level of current page located */
977 block_t data_blkaddr; /* block address of the node block */
978};
979
980static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,
981 struct page *ipage, struct page *npage, nid_t nid)
982{
983 memset(dn, 0, sizeof(*dn));
984 dn->inode = inode;
985 dn->inode_page = ipage;
986 dn->node_page = npage;
987 dn->nid = nid;
988}
989
990/*
991 * For SIT manager
992 *
993 * By default, there are 6 active log areas across the whole main area.
994 * When considering hot and cold data separation to reduce cleaning overhead,
995 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,
996 * respectively.
997 * In the current design, you should not change the numbers intentionally.
998 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6
999 * logs individually according to the underlying devices. (default: 6)
1000 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for
1001 * data and 8 for node logs.
1002 */
1003#define NR_CURSEG_DATA_TYPE (3)
1004#define NR_CURSEG_NODE_TYPE (3)
1005#define NR_CURSEG_INMEM_TYPE (2)
1006#define NR_CURSEG_RO_TYPE (2)
1007#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)
1008#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)
1009
1010enum {
1011 CURSEG_HOT_DATA = 0, /* directory entry blocks */
1012 CURSEG_WARM_DATA, /* data blocks */
1013 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */
1014 CURSEG_HOT_NODE, /* direct node blocks of directory files */
1015 CURSEG_WARM_NODE, /* direct node blocks of normal files */
1016 CURSEG_COLD_NODE, /* indirect node blocks */
1017 NR_PERSISTENT_LOG, /* number of persistent log */
1018 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,
1019 /* pinned file that needs consecutive block address */
1020 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */
1021 NO_CHECK_TYPE, /* number of persistent & inmem log */
1022};
1023
1024struct flush_cmd {
1025 struct completion wait;
1026 struct llist_node llnode;
1027 nid_t ino;
1028 int ret;
1029};
1030
1031struct flush_cmd_control {
1032 struct task_struct *f2fs_issue_flush; /* flush thread */
1033 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */
1034 atomic_t issued_flush; /* # of issued flushes */
1035 atomic_t queued_flush; /* # of queued flushes */
1036 struct llist_head issue_list; /* list for command issue */
1037 struct llist_node *dispatch_list; /* list for command dispatch */
1038};
1039
1040struct f2fs_sm_info {
1041 struct sit_info *sit_info; /* whole segment information */
1042 struct free_segmap_info *free_info; /* free segment information */
1043 struct dirty_seglist_info *dirty_info; /* dirty segment information */
1044 struct curseg_info *curseg_array; /* active segment information */
1045
1046 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */
1047
1048 block_t seg0_blkaddr; /* block address of 0'th segment */
1049 block_t main_blkaddr; /* start block address of main area */
1050 block_t ssa_blkaddr; /* start block address of SSA area */
1051
1052 unsigned int segment_count; /* total # of segments */
1053 unsigned int main_segments; /* # of segments in main area */
1054 unsigned int reserved_segments; /* # of reserved segments */
1055 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */
1056 unsigned int ovp_segments; /* # of overprovision segments */
1057
1058 /* a threshold to reclaim prefree segments */
1059 unsigned int rec_prefree_segments;
1060
1061 struct list_head sit_entry_set; /* sit entry set list */
1062
1063 unsigned int ipu_policy; /* in-place-update policy */
1064 unsigned int min_ipu_util; /* in-place-update threshold */
1065 unsigned int min_fsync_blocks; /* threshold for fsync */
1066 unsigned int min_seq_blocks; /* threshold for sequential blocks */
1067 unsigned int min_hot_blocks; /* threshold for hot block allocation */
1068 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */
1069
1070 /* for flush command control */
1071 struct flush_cmd_control *fcc_info;
1072
1073 /* for discard command control */
1074 struct discard_cmd_control *dcc_info;
1075};
1076
1077/*
1078 * For superblock
1079 */
1080/*
1081 * COUNT_TYPE for monitoring
1082 *
1083 * f2fs monitors the number of several block types such as on-writeback,
1084 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.
1085 */
1086#define WB_DATA_TYPE(p, f) \
1087 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)
1088enum count_type {
1089 F2FS_DIRTY_DENTS,
1090 F2FS_DIRTY_DATA,
1091 F2FS_DIRTY_QDATA,
1092 F2FS_DIRTY_NODES,
1093 F2FS_DIRTY_META,
1094 F2FS_DIRTY_IMETA,
1095 F2FS_WB_CP_DATA,
1096 F2FS_WB_DATA,
1097 F2FS_RD_DATA,
1098 F2FS_RD_NODE,
1099 F2FS_RD_META,
1100 F2FS_DIO_WRITE,
1101 F2FS_DIO_READ,
1102 NR_COUNT_TYPE,
1103};
1104
1105/*
1106 * The below are the page types of bios used in submit_bio().
1107 * The available types are:
1108 * DATA User data pages. It operates as async mode.
1109 * NODE Node pages. It operates as async mode.
1110 * META FS metadata pages such as SIT, NAT, CP.
1111 * NR_PAGE_TYPE The number of page types.
1112 * META_FLUSH Make sure the previous pages are written
1113 * with waiting the bio's completion
1114 * ... Only can be used with META.
1115 */
1116#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))
1117#define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE)
1118enum page_type {
1119 DATA = 0,
1120 NODE = 1, /* should not change this */
1121 META,
1122 NR_PAGE_TYPE,
1123 META_FLUSH,
1124 IPU, /* the below types are used by tracepoints only. */
1125 OPU,
1126};
1127
1128enum temp_type {
1129 HOT = 0, /* must be zero for meta bio */
1130 WARM,
1131 COLD,
1132 NR_TEMP_TYPE,
1133};
1134
1135enum need_lock_type {
1136 LOCK_REQ = 0,
1137 LOCK_DONE,
1138 LOCK_RETRY,
1139};
1140
1141enum cp_reason_type {
1142 CP_NO_NEEDED,
1143 CP_NON_REGULAR,
1144 CP_COMPRESSED,
1145 CP_HARDLINK,
1146 CP_SB_NEED_CP,
1147 CP_WRONG_PINO,
1148 CP_NO_SPC_ROLL,
1149 CP_NODE_NEED_CP,
1150 CP_FASTBOOT_MODE,
1151 CP_SPEC_LOG_NUM,
1152 CP_RECOVER_DIR,
1153};
1154
1155enum iostat_type {
1156 /* WRITE IO */
1157 APP_DIRECT_IO, /* app direct write IOs */
1158 APP_BUFFERED_IO, /* app buffered write IOs */
1159 APP_WRITE_IO, /* app write IOs */
1160 APP_MAPPED_IO, /* app mapped IOs */
1161 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */
1162 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */
1163 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */
1164 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */
1165 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */
1166 FS_META_IO, /* meta IOs from kworker/reclaimer */
1167 FS_GC_DATA_IO, /* data IOs from forground gc */
1168 FS_GC_NODE_IO, /* node IOs from forground gc */
1169 FS_CP_DATA_IO, /* data IOs from checkpoint */
1170 FS_CP_NODE_IO, /* node IOs from checkpoint */
1171 FS_CP_META_IO, /* meta IOs from checkpoint */
1172
1173 /* READ IO */
1174 APP_DIRECT_READ_IO, /* app direct read IOs */
1175 APP_BUFFERED_READ_IO, /* app buffered read IOs */
1176 APP_READ_IO, /* app read IOs */
1177 APP_MAPPED_READ_IO, /* app mapped read IOs */
1178 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */
1179 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */
1180 FS_DATA_READ_IO, /* data read IOs */
1181 FS_GDATA_READ_IO, /* data read IOs from background gc */
1182 FS_CDATA_READ_IO, /* compressed data read IOs */
1183 FS_NODE_READ_IO, /* node read IOs */
1184 FS_META_READ_IO, /* meta read IOs */
1185
1186 /* other */
1187 FS_DISCARD_IO, /* discard */
1188 FS_FLUSH_IO, /* flush */
1189 FS_ZONE_RESET_IO, /* zone reset */
1190 NR_IO_TYPE,
1191};
1192
1193struct f2fs_io_info {
1194 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */
1195 nid_t ino; /* inode number */
1196 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */
1197 enum temp_type temp; /* contains HOT/WARM/COLD */
1198 enum req_op op; /* contains REQ_OP_ */
1199 blk_opf_t op_flags; /* req_flag_bits */
1200 block_t new_blkaddr; /* new block address to be written */
1201 block_t old_blkaddr; /* old block address before Cow */
1202 struct page *page; /* page to be written */
1203 struct page *encrypted_page; /* encrypted page */
1204 struct page *compressed_page; /* compressed page */
1205 struct list_head list; /* serialize IOs */
1206 unsigned int compr_blocks; /* # of compressed block addresses */
1207 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */
1208 unsigned int version:8; /* version of the node */
1209 unsigned int submitted:1; /* indicate IO submission */
1210 unsigned int in_list:1; /* indicate fio is in io_list */
1211 unsigned int is_por:1; /* indicate IO is from recovery or not */
1212 unsigned int encrypted:1; /* indicate file is encrypted */
1213 unsigned int post_read:1; /* require post read */
1214 enum iostat_type io_type; /* io type */
1215 struct writeback_control *io_wbc; /* writeback control */
1216 struct bio **bio; /* bio for ipu */
1217 sector_t *last_block; /* last block number in bio */
1218};
1219
1220struct bio_entry {
1221 struct bio *bio;
1222 struct list_head list;
1223};
1224
1225#define is_read_io(rw) ((rw) == READ)
1226struct f2fs_bio_info {
1227 struct f2fs_sb_info *sbi; /* f2fs superblock */
1228 struct bio *bio; /* bios to merge */
1229 sector_t last_block_in_bio; /* last block number */
1230 struct f2fs_io_info fio; /* store buffered io info. */
1231#ifdef CONFIG_BLK_DEV_ZONED
1232 struct completion zone_wait; /* condition value for the previous open zone to close */
1233 struct bio *zone_pending_bio; /* pending bio for the previous zone */
1234 void *bi_private; /* previous bi_private for pending bio */
1235#endif
1236 struct f2fs_rwsem io_rwsem; /* blocking op for bio */
1237 spinlock_t io_lock; /* serialize DATA/NODE IOs */
1238 struct list_head io_list; /* track fios */
1239 struct list_head bio_list; /* bio entry list head */
1240 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */
1241};
1242
1243#define FDEV(i) (sbi->devs[i])
1244#define RDEV(i) (raw_super->devs[i])
1245struct f2fs_dev_info {
1246 struct file *bdev_file;
1247 struct block_device *bdev;
1248 char path[MAX_PATH_LEN];
1249 unsigned int total_segments;
1250 block_t start_blk;
1251 block_t end_blk;
1252#ifdef CONFIG_BLK_DEV_ZONED
1253 unsigned int nr_blkz; /* Total number of zones */
1254 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */
1255#endif
1256};
1257
1258enum inode_type {
1259 DIR_INODE, /* for dirty dir inode */
1260 FILE_INODE, /* for dirty regular/symlink inode */
1261 DIRTY_META, /* for all dirtied inode metadata */
1262 NR_INODE_TYPE,
1263};
1264
1265/* for inner inode cache management */
1266struct inode_management {
1267 struct radix_tree_root ino_root; /* ino entry array */
1268 spinlock_t ino_lock; /* for ino entry lock */
1269 struct list_head ino_list; /* inode list head */
1270 unsigned long ino_num; /* number of entries */
1271};
1272
1273/* for GC_AT */
1274struct atgc_management {
1275 bool atgc_enabled; /* ATGC is enabled or not */
1276 struct rb_root_cached root; /* root of victim rb-tree */
1277 struct list_head victim_list; /* linked with all victim entries */
1278 unsigned int victim_count; /* victim count in rb-tree */
1279 unsigned int candidate_ratio; /* candidate ratio */
1280 unsigned int max_candidate_count; /* max candidate count */
1281 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */
1282 unsigned long long age_threshold; /* age threshold */
1283};
1284
1285struct f2fs_gc_control {
1286 unsigned int victim_segno; /* target victim segment number */
1287 int init_gc_type; /* FG_GC or BG_GC */
1288 bool no_bg_gc; /* check the space and stop bg_gc */
1289 bool should_migrate_blocks; /* should migrate blocks */
1290 bool err_gc_skipped; /* return EAGAIN if GC skipped */
1291 unsigned int nr_free_secs; /* # of free sections to do GC */
1292};
1293
1294/*
1295 * For s_flag in struct f2fs_sb_info
1296 * Modification on enum should be synchronized with s_flag array
1297 */
1298enum {
1299 SBI_IS_DIRTY, /* dirty flag for checkpoint */
1300 SBI_IS_CLOSE, /* specify unmounting */
1301 SBI_NEED_FSCK, /* need fsck.f2fs to fix */
1302 SBI_POR_DOING, /* recovery is doing or not */
1303 SBI_NEED_SB_WRITE, /* need to recover superblock */
1304 SBI_NEED_CP, /* need to checkpoint */
1305 SBI_IS_SHUTDOWN, /* shutdown by ioctl */
1306 SBI_IS_RECOVERED, /* recovered orphan/data */
1307 SBI_CP_DISABLED, /* CP was disabled last mount */
1308 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */
1309 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */
1310 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */
1311 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */
1312 SBI_IS_RESIZEFS, /* resizefs is in process */
1313 SBI_IS_FREEZING, /* freezefs is in process */
1314 SBI_IS_WRITABLE, /* remove ro mountoption transiently */
1315 MAX_SBI_FLAG,
1316};
1317
1318enum {
1319 CP_TIME,
1320 REQ_TIME,
1321 DISCARD_TIME,
1322 GC_TIME,
1323 DISABLE_TIME,
1324 UMOUNT_DISCARD_TIMEOUT,
1325 MAX_TIME,
1326};
1327
1328/* Note that you need to keep synchronization with this gc_mode_names array */
1329enum {
1330 GC_NORMAL,
1331 GC_IDLE_CB,
1332 GC_IDLE_GREEDY,
1333 GC_IDLE_AT,
1334 GC_URGENT_HIGH,
1335 GC_URGENT_LOW,
1336 GC_URGENT_MID,
1337 MAX_GC_MODE,
1338};
1339
1340enum {
1341 BGGC_MODE_ON, /* background gc is on */
1342 BGGC_MODE_OFF, /* background gc is off */
1343 BGGC_MODE_SYNC, /*
1344 * background gc is on, migrating blocks
1345 * like foreground gc
1346 */
1347};
1348
1349enum {
1350 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */
1351 FS_MODE_LFS, /* use lfs allocation only */
1352 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */
1353 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */
1354};
1355
1356enum {
1357 ALLOC_MODE_DEFAULT, /* stay default */
1358 ALLOC_MODE_REUSE, /* reuse segments as much as possible */
1359};
1360
1361enum fsync_mode {
1362 FSYNC_MODE_POSIX, /* fsync follows posix semantics */
1363 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */
1364 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */
1365};
1366
1367enum {
1368 COMPR_MODE_FS, /*
1369 * automatically compress compression
1370 * enabled files
1371 */
1372 COMPR_MODE_USER, /*
1373 * automatical compression is disabled.
1374 * user can control the file compression
1375 * using ioctls
1376 */
1377};
1378
1379enum {
1380 DISCARD_UNIT_BLOCK, /* basic discard unit is block */
1381 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */
1382 DISCARD_UNIT_SECTION, /* basic discard unit is section */
1383};
1384
1385enum {
1386 MEMORY_MODE_NORMAL, /* memory mode for normal devices */
1387 MEMORY_MODE_LOW, /* memory mode for low memry devices */
1388};
1389
1390enum errors_option {
1391 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */
1392 MOUNT_ERRORS_CONTINUE, /* continue on errors */
1393 MOUNT_ERRORS_PANIC, /* panic on errors */
1394};
1395
1396enum {
1397 BACKGROUND,
1398 FOREGROUND,
1399 MAX_CALL_TYPE,
1400 TOTAL_CALL = FOREGROUND,
1401};
1402
1403static inline int f2fs_test_bit(unsigned int nr, char *addr);
1404static inline void f2fs_set_bit(unsigned int nr, char *addr);
1405static inline void f2fs_clear_bit(unsigned int nr, char *addr);
1406
1407/*
1408 * Layout of f2fs page.private:
1409 *
1410 * Layout A: lowest bit should be 1
1411 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |
1412 * bit 0 PAGE_PRIVATE_NOT_POINTER
1413 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION
1414 * bit 2 PAGE_PRIVATE_INLINE_INODE
1415 * bit 3 PAGE_PRIVATE_REF_RESOURCE
1416 * bit 4- f2fs private data
1417 *
1418 * Layout B: lowest bit should be 0
1419 * page.private is a wrapped pointer.
1420 */
1421enum {
1422 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */
1423 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */
1424 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */
1425 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */
1426 PAGE_PRIVATE_MAX
1427};
1428
1429/* For compression */
1430enum compress_algorithm_type {
1431 COMPRESS_LZO,
1432 COMPRESS_LZ4,
1433 COMPRESS_ZSTD,
1434 COMPRESS_LZORLE,
1435 COMPRESS_MAX,
1436};
1437
1438enum compress_flag {
1439 COMPRESS_CHKSUM,
1440 COMPRESS_MAX_FLAG,
1441};
1442
1443#define COMPRESS_WATERMARK 20
1444#define COMPRESS_PERCENT 20
1445
1446#define COMPRESS_DATA_RESERVED_SIZE 4
1447struct compress_data {
1448 __le32 clen; /* compressed data size */
1449 __le32 chksum; /* compressed data chksum */
1450 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */
1451 u8 cdata[]; /* compressed data */
1452};
1453
1454#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))
1455
1456#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000
1457
1458#define F2FS_ZSTD_DEFAULT_CLEVEL 1
1459
1460#define COMPRESS_LEVEL_OFFSET 8
1461
1462/* compress context */
1463struct compress_ctx {
1464 struct inode *inode; /* inode the context belong to */
1465 pgoff_t cluster_idx; /* cluster index number */
1466 unsigned int cluster_size; /* page count in cluster */
1467 unsigned int log_cluster_size; /* log of cluster size */
1468 struct page **rpages; /* pages store raw data in cluster */
1469 unsigned int nr_rpages; /* total page number in rpages */
1470 struct page **cpages; /* pages store compressed data in cluster */
1471 unsigned int nr_cpages; /* total page number in cpages */
1472 unsigned int valid_nr_cpages; /* valid page number in cpages */
1473 void *rbuf; /* virtual mapped address on rpages */
1474 struct compress_data *cbuf; /* virtual mapped address on cpages */
1475 size_t rlen; /* valid data length in rbuf */
1476 size_t clen; /* valid data length in cbuf */
1477 void *private; /* payload buffer for specified compression algorithm */
1478 void *private2; /* extra payload buffer */
1479};
1480
1481/* compress context for write IO path */
1482struct compress_io_ctx {
1483 u32 magic; /* magic number to indicate page is compressed */
1484 struct inode *inode; /* inode the context belong to */
1485 struct page **rpages; /* pages store raw data in cluster */
1486 unsigned int nr_rpages; /* total page number in rpages */
1487 atomic_t pending_pages; /* in-flight compressed page count */
1488};
1489
1490/* Context for decompressing one cluster on the read IO path */
1491struct decompress_io_ctx {
1492 u32 magic; /* magic number to indicate page is compressed */
1493 struct inode *inode; /* inode the context belong to */
1494 pgoff_t cluster_idx; /* cluster index number */
1495 unsigned int cluster_size; /* page count in cluster */
1496 unsigned int log_cluster_size; /* log of cluster size */
1497 struct page **rpages; /* pages store raw data in cluster */
1498 unsigned int nr_rpages; /* total page number in rpages */
1499 struct page **cpages; /* pages store compressed data in cluster */
1500 unsigned int nr_cpages; /* total page number in cpages */
1501 struct page **tpages; /* temp pages to pad holes in cluster */
1502 void *rbuf; /* virtual mapped address on rpages */
1503 struct compress_data *cbuf; /* virtual mapped address on cpages */
1504 size_t rlen; /* valid data length in rbuf */
1505 size_t clen; /* valid data length in cbuf */
1506
1507 /*
1508 * The number of compressed pages remaining to be read in this cluster.
1509 * This is initially nr_cpages. It is decremented by 1 each time a page
1510 * has been read (or failed to be read). When it reaches 0, the cluster
1511 * is decompressed (or an error is reported).
1512 *
1513 * If an error occurs before all the pages have been submitted for I/O,
1514 * then this will never reach 0. In this case the I/O submitter is
1515 * responsible for calling f2fs_decompress_end_io() instead.
1516 */
1517 atomic_t remaining_pages;
1518
1519 /*
1520 * Number of references to this decompress_io_ctx.
1521 *
1522 * One reference is held for I/O completion. This reference is dropped
1523 * after the pagecache pages are updated and unlocked -- either after
1524 * decompression (and verity if enabled), or after an error.
1525 *
1526 * In addition, each compressed page holds a reference while it is in a
1527 * bio. These references are necessary prevent compressed pages from
1528 * being freed while they are still in a bio.
1529 */
1530 refcount_t refcnt;
1531
1532 bool failed; /* IO error occurred before decompression? */
1533 bool need_verity; /* need fs-verity verification after decompression? */
1534 void *private; /* payload buffer for specified decompression algorithm */
1535 void *private2; /* extra payload buffer */
1536 struct work_struct verity_work; /* work to verify the decompressed pages */
1537 struct work_struct free_work; /* work for late free this structure itself */
1538};
1539
1540#define NULL_CLUSTER ((unsigned int)(~0))
1541#define MIN_COMPRESS_LOG_SIZE 2
1542#define MAX_COMPRESS_LOG_SIZE 8
1543#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))
1544
1545struct f2fs_sb_info {
1546 struct super_block *sb; /* pointer to VFS super block */
1547 struct proc_dir_entry *s_proc; /* proc entry */
1548 struct f2fs_super_block *raw_super; /* raw super block pointer */
1549 struct f2fs_rwsem sb_lock; /* lock for raw super block */
1550 int valid_super_block; /* valid super block no */
1551 unsigned long s_flag; /* flags for sbi */
1552 struct mutex writepages; /* mutex for writepages() */
1553
1554#ifdef CONFIG_BLK_DEV_ZONED
1555 unsigned int blocks_per_blkz; /* F2FS blocks per zone */
1556 unsigned int max_open_zones; /* max open zone resources of the zoned device */
1557#endif
1558
1559 /* for node-related operations */
1560 struct f2fs_nm_info *nm_info; /* node manager */
1561 struct inode *node_inode; /* cache node blocks */
1562
1563 /* for segment-related operations */
1564 struct f2fs_sm_info *sm_info; /* segment manager */
1565
1566 /* for bio operations */
1567 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */
1568 /* keep migration IO order for LFS mode */
1569 struct f2fs_rwsem io_order_lock;
1570 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */
1571 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */
1572
1573 /* for checkpoint */
1574 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */
1575 int cur_cp_pack; /* remain current cp pack */
1576 spinlock_t cp_lock; /* for flag in ckpt */
1577 struct inode *meta_inode; /* cache meta blocks */
1578 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */
1579 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */
1580 struct f2fs_rwsem node_write; /* locking node writes */
1581 struct f2fs_rwsem node_change; /* locking node change */
1582 wait_queue_head_t cp_wait;
1583 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */
1584 long interval_time[MAX_TIME]; /* to store thresholds */
1585 struct ckpt_req_control cprc_info; /* for checkpoint request control */
1586
1587 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */
1588
1589 spinlock_t fsync_node_lock; /* for node entry lock */
1590 struct list_head fsync_node_list; /* node list head */
1591 unsigned int fsync_seg_id; /* sequence id */
1592 unsigned int fsync_node_num; /* number of node entries */
1593
1594 /* for orphan inode, use 0'th array */
1595 unsigned int max_orphans; /* max orphan inodes */
1596
1597 /* for inode management */
1598 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */
1599 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */
1600 struct mutex flush_lock; /* for flush exclusion */
1601
1602 /* for extent tree cache */
1603 struct extent_tree_info extent_tree[NR_EXTENT_CACHES];
1604 atomic64_t allocated_data_blocks; /* for block age extent_cache */
1605
1606 /* The threshold used for hot and warm data seperation*/
1607 unsigned int hot_data_age_threshold;
1608 unsigned int warm_data_age_threshold;
1609 unsigned int last_age_weight;
1610
1611 /* basic filesystem units */
1612 unsigned int log_sectors_per_block; /* log2 sectors per block */
1613 unsigned int log_blocksize; /* log2 block size */
1614 unsigned int blocksize; /* block size */
1615 unsigned int root_ino_num; /* root inode number*/
1616 unsigned int node_ino_num; /* node inode number*/
1617 unsigned int meta_ino_num; /* meta inode number*/
1618 unsigned int log_blocks_per_seg; /* log2 blocks per segment */
1619 unsigned int blocks_per_seg; /* blocks per segment */
1620 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */
1621 unsigned int segs_per_sec; /* segments per section */
1622 unsigned int secs_per_zone; /* sections per zone */
1623 unsigned int total_sections; /* total section count */
1624 unsigned int total_node_count; /* total node block count */
1625 unsigned int total_valid_node_count; /* valid node block count */
1626 int dir_level; /* directory level */
1627 bool readdir_ra; /* readahead inode in readdir */
1628 u64 max_io_bytes; /* max io bytes to merge IOs */
1629
1630 block_t user_block_count; /* # of user blocks */
1631 block_t total_valid_block_count; /* # of valid blocks */
1632 block_t discard_blks; /* discard command candidats */
1633 block_t last_valid_block_count; /* for recovery */
1634 block_t reserved_blocks; /* configurable reserved blocks */
1635 block_t current_reserved_blocks; /* current reserved blocks */
1636
1637 /* Additional tracking for no checkpoint mode */
1638 block_t unusable_block_count; /* # of blocks saved by last cp */
1639
1640 unsigned int nquota_files; /* # of quota sysfile */
1641 struct f2fs_rwsem quota_sem; /* blocking cp for flags */
1642
1643 /* # of pages, see count_type */
1644 atomic_t nr_pages[NR_COUNT_TYPE];
1645 /* # of allocated blocks */
1646 struct percpu_counter alloc_valid_block_count;
1647 /* # of node block writes as roll forward recovery */
1648 struct percpu_counter rf_node_block_count;
1649
1650 /* writeback control */
1651 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */
1652
1653 /* valid inode count */
1654 struct percpu_counter total_valid_inode_count;
1655
1656 struct f2fs_mount_info mount_opt; /* mount options */
1657
1658 /* for cleaning operations */
1659 struct f2fs_rwsem gc_lock; /*
1660 * semaphore for GC, avoid
1661 * race between GC and GC or CP
1662 */
1663 struct f2fs_gc_kthread *gc_thread; /* GC thread */
1664 struct atgc_management am; /* atgc management */
1665 unsigned int cur_victim_sec; /* current victim section num */
1666 unsigned int gc_mode; /* current GC state */
1667 unsigned int next_victim_seg[2]; /* next segment in victim section */
1668 spinlock_t gc_remaining_trials_lock;
1669 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */
1670 unsigned int gc_remaining_trials;
1671
1672 /* for skip statistic */
1673 unsigned long long skipped_gc_rwsem; /* FG_GC only */
1674
1675 /* threshold for gc trials on pinned files */
1676 unsigned short gc_pin_file_threshold;
1677 struct f2fs_rwsem pin_sem;
1678
1679 /* maximum # of trials to find a victim segment for SSR and GC */
1680 unsigned int max_victim_search;
1681 /* migration granularity of garbage collection, unit: segment */
1682 unsigned int migration_granularity;
1683
1684 /*
1685 * for stat information.
1686 * one is for the LFS mode, and the other is for the SSR mode.
1687 */
1688#ifdef CONFIG_F2FS_STAT_FS
1689 struct f2fs_stat_info *stat_info; /* FS status information */
1690 atomic_t meta_count[META_MAX]; /* # of meta blocks */
1691 unsigned int segment_count[2]; /* # of allocated segments */
1692 unsigned int block_count[2]; /* # of allocated blocks */
1693 atomic_t inplace_count; /* # of inplace update */
1694 /* # of lookup extent cache */
1695 atomic64_t total_hit_ext[NR_EXTENT_CACHES];
1696 /* # of hit rbtree extent node */
1697 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES];
1698 /* # of hit cached extent node */
1699 atomic64_t read_hit_cached[NR_EXTENT_CACHES];
1700 /* # of hit largest extent node in read extent cache */
1701 atomic64_t read_hit_largest;
1702 atomic_t inline_xattr; /* # of inline_xattr inodes */
1703 atomic_t inline_inode; /* # of inline_data inodes */
1704 atomic_t inline_dir; /* # of inline_dentry inodes */
1705 atomic_t compr_inode; /* # of compressed inodes */
1706 atomic64_t compr_blocks; /* # of compressed blocks */
1707 atomic_t swapfile_inode; /* # of swapfile inodes */
1708 atomic_t atomic_files; /* # of opened atomic file */
1709 atomic_t max_aw_cnt; /* max # of atomic writes */
1710 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */
1711 unsigned int other_skip_bggc; /* skip background gc for other reasons */
1712 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */
1713 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */
1714#endif
1715 spinlock_t stat_lock; /* lock for stat operations */
1716
1717 /* to attach REQ_META|REQ_FUA flags */
1718 unsigned int data_io_flag;
1719 unsigned int node_io_flag;
1720
1721 /* For sysfs support */
1722 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */
1723 struct completion s_kobj_unregister;
1724
1725 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */
1726 struct completion s_stat_kobj_unregister;
1727
1728 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */
1729 struct completion s_feature_list_kobj_unregister;
1730
1731 /* For shrinker support */
1732 struct list_head s_list;
1733 struct mutex umount_mutex;
1734 unsigned int shrinker_run_no;
1735
1736 /* For multi devices */
1737 int s_ndevs; /* number of devices */
1738 struct f2fs_dev_info *devs; /* for device list */
1739 unsigned int dirty_device; /* for checkpoint data flush */
1740 spinlock_t dev_lock; /* protect dirty_device */
1741 bool aligned_blksize; /* all devices has the same logical blksize */
1742
1743 /* For write statistics */
1744 u64 sectors_written_start;
1745 u64 kbytes_written;
1746
1747 /* Reference to checksum algorithm driver via cryptoapi */
1748 struct crypto_shash *s_chksum_driver;
1749
1750 /* Precomputed FS UUID checksum for seeding other checksums */
1751 __u32 s_chksum_seed;
1752
1753 struct workqueue_struct *post_read_wq; /* post read workqueue */
1754
1755 /*
1756 * If we are in irq context, let's update error information into
1757 * on-disk superblock in the work.
1758 */
1759 struct work_struct s_error_work;
1760 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */
1761 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */
1762 spinlock_t error_lock; /* protect errors/stop_reason array */
1763 bool error_dirty; /* errors of sb is dirty */
1764
1765 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */
1766 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */
1767
1768 /* For reclaimed segs statistics per each GC mode */
1769 unsigned int gc_segment_mode; /* GC state for reclaimed segments */
1770 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */
1771
1772 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */
1773
1774 int max_fragment_chunk; /* max chunk size for block fragmentation mode */
1775 int max_fragment_hole; /* max hole size for block fragmentation mode */
1776
1777 /* For atomic write statistics */
1778 atomic64_t current_atomic_write;
1779 s64 peak_atomic_write;
1780 u64 committed_atomic_block;
1781 u64 revoked_atomic_block;
1782
1783#ifdef CONFIG_F2FS_FS_COMPRESSION
1784 struct kmem_cache *page_array_slab; /* page array entry */
1785 unsigned int page_array_slab_size; /* default page array slab size */
1786
1787 /* For runtime compression statistics */
1788 u64 compr_written_block;
1789 u64 compr_saved_block;
1790 u32 compr_new_inode;
1791
1792 /* For compressed block cache */
1793 struct inode *compress_inode; /* cache compressed blocks */
1794 unsigned int compress_percent; /* cache page percentage */
1795 unsigned int compress_watermark; /* cache page watermark */
1796 atomic_t compress_page_hit; /* cache hit count */
1797#endif
1798
1799#ifdef CONFIG_F2FS_IOSTAT
1800 /* For app/fs IO statistics */
1801 spinlock_t iostat_lock;
1802 unsigned long long iostat_count[NR_IO_TYPE];
1803 unsigned long long iostat_bytes[NR_IO_TYPE];
1804 unsigned long long prev_iostat_bytes[NR_IO_TYPE];
1805 bool iostat_enable;
1806 unsigned long iostat_next_period;
1807 unsigned int iostat_period_ms;
1808
1809 /* For io latency related statistics info in one iostat period */
1810 spinlock_t iostat_lat_lock;
1811 struct iostat_lat_info *iostat_io_lat;
1812#endif
1813};
1814
1815/* Definitions to access f2fs_sb_info */
1816#define SEGS_TO_BLKS(sbi, segs) \
1817 ((segs) << (sbi)->log_blocks_per_seg)
1818#define BLKS_TO_SEGS(sbi, blks) \
1819 ((blks) >> (sbi)->log_blocks_per_seg)
1820
1821#define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg)
1822#define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec))
1823#define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec)
1824
1825__printf(3, 4)
1826void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...);
1827
1828#define f2fs_err(sbi, fmt, ...) \
1829 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__)
1830#define f2fs_warn(sbi, fmt, ...) \
1831 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__)
1832#define f2fs_notice(sbi, fmt, ...) \
1833 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__)
1834#define f2fs_info(sbi, fmt, ...) \
1835 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__)
1836#define f2fs_debug(sbi, fmt, ...) \
1837 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__)
1838
1839#define f2fs_err_ratelimited(sbi, fmt, ...) \
1840 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__)
1841#define f2fs_warn_ratelimited(sbi, fmt, ...) \
1842 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__)
1843#define f2fs_info_ratelimited(sbi, fmt, ...) \
1844 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__)
1845
1846#ifdef CONFIG_F2FS_FAULT_INJECTION
1847#define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \
1848 __builtin_return_address(0))
1849static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type,
1850 const char *func, const char *parent_func)
1851{
1852 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1853
1854 if (!ffi->inject_rate)
1855 return false;
1856
1857 if (!IS_FAULT_SET(ffi, type))
1858 return false;
1859
1860 atomic_inc(&ffi->inject_ops);
1861 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {
1862 atomic_set(&ffi->inject_ops, 0);
1863 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS",
1864 f2fs_fault_name[type], func, parent_func);
1865 return true;
1866 }
1867 return false;
1868}
1869#else
1870static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)
1871{
1872 return false;
1873}
1874#endif
1875
1876/*
1877 * Test if the mounted volume is a multi-device volume.
1878 * - For a single regular disk volume, sbi->s_ndevs is 0.
1879 * - For a single zoned disk volume, sbi->s_ndevs is 1.
1880 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.
1881 */
1882static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)
1883{
1884 return sbi->s_ndevs > 1;
1885}
1886
1887static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)
1888{
1889 unsigned long now = jiffies;
1890
1891 sbi->last_time[type] = now;
1892
1893 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */
1894 if (type == REQ_TIME) {
1895 sbi->last_time[DISCARD_TIME] = now;
1896 sbi->last_time[GC_TIME] = now;
1897 }
1898}
1899
1900static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)
1901{
1902 unsigned long interval = sbi->interval_time[type] * HZ;
1903
1904 return time_after(jiffies, sbi->last_time[type] + interval);
1905}
1906
1907static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,
1908 int type)
1909{
1910 unsigned long interval = sbi->interval_time[type] * HZ;
1911 unsigned int wait_ms = 0;
1912 long delta;
1913
1914 delta = (sbi->last_time[type] + interval) - jiffies;
1915 if (delta > 0)
1916 wait_ms = jiffies_to_msecs(delta);
1917
1918 return wait_ms;
1919}
1920
1921/*
1922 * Inline functions
1923 */
1924static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,
1925 const void *address, unsigned int length)
1926{
1927 struct {
1928 struct shash_desc shash;
1929 char ctx[4];
1930 } desc;
1931 int err;
1932
1933 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));
1934
1935 desc.shash.tfm = sbi->s_chksum_driver;
1936 *(u32 *)desc.ctx = crc;
1937
1938 err = crypto_shash_update(&desc.shash, address, length);
1939 BUG_ON(err);
1940
1941 return *(u32 *)desc.ctx;
1942}
1943
1944static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,
1945 unsigned int length)
1946{
1947 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);
1948}
1949
1950static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,
1951 void *buf, size_t buf_size)
1952{
1953 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;
1954}
1955
1956static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,
1957 const void *address, unsigned int length)
1958{
1959 return __f2fs_crc32(sbi, crc, address, length);
1960}
1961
1962static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)
1963{
1964 return container_of(inode, struct f2fs_inode_info, vfs_inode);
1965}
1966
1967static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)
1968{
1969 return sb->s_fs_info;
1970}
1971
1972static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)
1973{
1974 return F2FS_SB(inode->i_sb);
1975}
1976
1977static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)
1978{
1979 return F2FS_I_SB(mapping->host);
1980}
1981
1982static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)
1983{
1984 return F2FS_M_SB(page_file_mapping(page));
1985}
1986
1987static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)
1988{
1989 return (struct f2fs_super_block *)(sbi->raw_super);
1990}
1991
1992static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)
1993{
1994 return (struct f2fs_checkpoint *)(sbi->ckpt);
1995}
1996
1997static inline struct f2fs_node *F2FS_NODE(struct page *page)
1998{
1999 return (struct f2fs_node *)page_address(page);
2000}
2001
2002static inline struct f2fs_inode *F2FS_INODE(struct page *page)
2003{
2004 return &((struct f2fs_node *)page_address(page))->i;
2005}
2006
2007static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)
2008{
2009 return (struct f2fs_nm_info *)(sbi->nm_info);
2010}
2011
2012static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)
2013{
2014 return (struct f2fs_sm_info *)(sbi->sm_info);
2015}
2016
2017static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)
2018{
2019 return (struct sit_info *)(SM_I(sbi)->sit_info);
2020}
2021
2022static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)
2023{
2024 return (struct free_segmap_info *)(SM_I(sbi)->free_info);
2025}
2026
2027static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)
2028{
2029 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);
2030}
2031
2032static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)
2033{
2034 return sbi->meta_inode->i_mapping;
2035}
2036
2037static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)
2038{
2039 return sbi->node_inode->i_mapping;
2040}
2041
2042static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)
2043{
2044 return test_bit(type, &sbi->s_flag);
2045}
2046
2047static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2048{
2049 set_bit(type, &sbi->s_flag);
2050}
2051
2052static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)
2053{
2054 clear_bit(type, &sbi->s_flag);
2055}
2056
2057static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)
2058{
2059 return le64_to_cpu(cp->checkpoint_ver);
2060}
2061
2062static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)
2063{
2064 if (type < F2FS_MAX_QUOTAS)
2065 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);
2066 return 0;
2067}
2068
2069static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)
2070{
2071 size_t crc_offset = le32_to_cpu(cp->checksum_offset);
2072 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));
2073}
2074
2075static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2076{
2077 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2078
2079 return ckpt_flags & f;
2080}
2081
2082static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2083{
2084 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);
2085}
2086
2087static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2088{
2089 unsigned int ckpt_flags;
2090
2091 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2092 ckpt_flags |= f;
2093 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2094}
2095
2096static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2097{
2098 unsigned long flags;
2099
2100 spin_lock_irqsave(&sbi->cp_lock, flags);
2101 __set_ckpt_flags(F2FS_CKPT(sbi), f);
2102 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2103}
2104
2105static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
2106{
2107 unsigned int ckpt_flags;
2108
2109 ckpt_flags = le32_to_cpu(cp->ckpt_flags);
2110 ckpt_flags &= (~f);
2111 cp->ckpt_flags = cpu_to_le32(ckpt_flags);
2112}
2113
2114static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)
2115{
2116 unsigned long flags;
2117
2118 spin_lock_irqsave(&sbi->cp_lock, flags);
2119 __clear_ckpt_flags(F2FS_CKPT(sbi), f);
2120 spin_unlock_irqrestore(&sbi->cp_lock, flags);
2121}
2122
2123#define init_f2fs_rwsem(sem) \
2124do { \
2125 static struct lock_class_key __key; \
2126 \
2127 __init_f2fs_rwsem((sem), #sem, &__key); \
2128} while (0)
2129
2130static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,
2131 const char *sem_name, struct lock_class_key *key)
2132{
2133 __init_rwsem(&sem->internal_rwsem, sem_name, key);
2134#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2135 init_waitqueue_head(&sem->read_waiters);
2136#endif
2137}
2138
2139static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)
2140{
2141 return rwsem_is_locked(&sem->internal_rwsem);
2142}
2143
2144static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)
2145{
2146 return rwsem_is_contended(&sem->internal_rwsem);
2147}
2148
2149static inline void f2fs_down_read(struct f2fs_rwsem *sem)
2150{
2151#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2152 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));
2153#else
2154 down_read(&sem->internal_rwsem);
2155#endif
2156}
2157
2158static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)
2159{
2160 return down_read_trylock(&sem->internal_rwsem);
2161}
2162
2163static inline void f2fs_up_read(struct f2fs_rwsem *sem)
2164{
2165 up_read(&sem->internal_rwsem);
2166}
2167
2168static inline void f2fs_down_write(struct f2fs_rwsem *sem)
2169{
2170 down_write(&sem->internal_rwsem);
2171}
2172
2173#ifdef CONFIG_DEBUG_LOCK_ALLOC
2174static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)
2175{
2176 down_read_nested(&sem->internal_rwsem, subclass);
2177}
2178
2179static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass)
2180{
2181 down_write_nested(&sem->internal_rwsem, subclass);
2182}
2183#else
2184#define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)
2185#define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem)
2186#endif
2187
2188static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)
2189{
2190 return down_write_trylock(&sem->internal_rwsem);
2191}
2192
2193static inline void f2fs_up_write(struct f2fs_rwsem *sem)
2194{
2195 up_write(&sem->internal_rwsem);
2196#ifdef CONFIG_F2FS_UNFAIR_RWSEM
2197 wake_up_all(&sem->read_waiters);
2198#endif
2199}
2200
2201static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)
2202{
2203 f2fs_down_read(&sbi->cp_rwsem);
2204}
2205
2206static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)
2207{
2208 if (time_to_inject(sbi, FAULT_LOCK_OP))
2209 return 0;
2210 return f2fs_down_read_trylock(&sbi->cp_rwsem);
2211}
2212
2213static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)
2214{
2215 f2fs_up_read(&sbi->cp_rwsem);
2216}
2217
2218static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)
2219{
2220 f2fs_down_write(&sbi->cp_rwsem);
2221}
2222
2223static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)
2224{
2225 f2fs_up_write(&sbi->cp_rwsem);
2226}
2227
2228static inline int __get_cp_reason(struct f2fs_sb_info *sbi)
2229{
2230 int reason = CP_SYNC;
2231
2232 if (test_opt(sbi, FASTBOOT))
2233 reason = CP_FASTBOOT;
2234 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))
2235 reason = CP_UMOUNT;
2236 return reason;
2237}
2238
2239static inline bool __remain_node_summaries(int reason)
2240{
2241 return (reason & (CP_UMOUNT | CP_FASTBOOT));
2242}
2243
2244static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)
2245{
2246 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||
2247 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));
2248}
2249
2250/*
2251 * Check whether the inode has blocks or not
2252 */
2253static inline int F2FS_HAS_BLOCKS(struct inode *inode)
2254{
2255 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;
2256
2257 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;
2258}
2259
2260static inline bool f2fs_has_xattr_block(unsigned int ofs)
2261{
2262 return ofs == XATTR_NODE_OFFSET;
2263}
2264
2265static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,
2266 struct inode *inode, bool cap)
2267{
2268 if (!inode)
2269 return true;
2270 if (!test_opt(sbi, RESERVE_ROOT))
2271 return false;
2272 if (IS_NOQUOTA(inode))
2273 return true;
2274 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))
2275 return true;
2276 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&
2277 in_group_p(F2FS_OPTION(sbi).s_resgid))
2278 return true;
2279 if (cap && capable(CAP_SYS_RESOURCE))
2280 return true;
2281 return false;
2282}
2283
2284static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi,
2285 struct inode *inode, bool cap)
2286{
2287 block_t avail_user_block_count;
2288
2289 avail_user_block_count = sbi->user_block_count -
2290 sbi->current_reserved_blocks;
2291
2292 if (!__allow_reserved_blocks(sbi, inode, cap))
2293 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;
2294
2295 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
2296 if (avail_user_block_count > sbi->unusable_block_count)
2297 avail_user_block_count -= sbi->unusable_block_count;
2298 else
2299 avail_user_block_count = 0;
2300 }
2301
2302 return avail_user_block_count;
2303}
2304
2305static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);
2306static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,
2307 struct inode *inode, blkcnt_t *count, bool partial)
2308{
2309 long long diff = 0, release = 0;
2310 block_t avail_user_block_count;
2311 int ret;
2312
2313 ret = dquot_reserve_block(inode, *count);
2314 if (ret)
2315 return ret;
2316
2317 if (time_to_inject(sbi, FAULT_BLOCK)) {
2318 release = *count;
2319 goto release_quota;
2320 }
2321
2322 /*
2323 * let's increase this in prior to actual block count change in order
2324 * for f2fs_sync_file to avoid data races when deciding checkpoint.
2325 */
2326 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));
2327
2328 spin_lock(&sbi->stat_lock);
2329
2330 avail_user_block_count = get_available_block_count(sbi, inode, true);
2331 diff = (long long)sbi->total_valid_block_count + *count -
2332 avail_user_block_count;
2333 if (unlikely(diff > 0)) {
2334 if (!partial) {
2335 spin_unlock(&sbi->stat_lock);
2336 release = *count;
2337 goto enospc;
2338 }
2339 if (diff > *count)
2340 diff = *count;
2341 *count -= diff;
2342 release = diff;
2343 if (!*count) {
2344 spin_unlock(&sbi->stat_lock);
2345 goto enospc;
2346 }
2347 }
2348 sbi->total_valid_block_count += (block_t)(*count);
2349
2350 spin_unlock(&sbi->stat_lock);
2351
2352 if (unlikely(release)) {
2353 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2354 dquot_release_reservation_block(inode, release);
2355 }
2356 f2fs_i_blocks_write(inode, *count, true, true);
2357 return 0;
2358
2359enospc:
2360 percpu_counter_sub(&sbi->alloc_valid_block_count, release);
2361release_quota:
2362 dquot_release_reservation_block(inode, release);
2363 return -ENOSPC;
2364}
2365
2366#define PAGE_PRIVATE_GET_FUNC(name, flagname) \
2367static inline bool page_private_##name(struct page *page) \
2368{ \
2369 return PagePrivate(page) && \
2370 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \
2371 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2372}
2373
2374#define PAGE_PRIVATE_SET_FUNC(name, flagname) \
2375static inline void set_page_private_##name(struct page *page) \
2376{ \
2377 if (!PagePrivate(page)) \
2378 attach_page_private(page, (void *)0); \
2379 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \
2380 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2381}
2382
2383#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \
2384static inline void clear_page_private_##name(struct page *page) \
2385{ \
2386 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \
2387 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \
2388 detach_page_private(page); \
2389}
2390
2391PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);
2392PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);
2393PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);
2394
2395PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);
2396PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);
2397PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);
2398
2399PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);
2400PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);
2401PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);
2402
2403static inline unsigned long get_page_private_data(struct page *page)
2404{
2405 unsigned long data = page_private(page);
2406
2407 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))
2408 return 0;
2409 return data >> PAGE_PRIVATE_MAX;
2410}
2411
2412static inline void set_page_private_data(struct page *page, unsigned long data)
2413{
2414 if (!PagePrivate(page))
2415 attach_page_private(page, (void *)0);
2416 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));
2417 page_private(page) |= data << PAGE_PRIVATE_MAX;
2418}
2419
2420static inline void clear_page_private_data(struct page *page)
2421{
2422 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0);
2423 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER))
2424 detach_page_private(page);
2425}
2426
2427static inline void clear_page_private_all(struct page *page)
2428{
2429 clear_page_private_data(page);
2430 clear_page_private_reference(page);
2431 clear_page_private_gcing(page);
2432 clear_page_private_inline(page);
2433
2434 f2fs_bug_on(F2FS_P_SB(page), page_private(page));
2435}
2436
2437static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,
2438 struct inode *inode,
2439 block_t count)
2440{
2441 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;
2442
2443 spin_lock(&sbi->stat_lock);
2444 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);
2445 sbi->total_valid_block_count -= (block_t)count;
2446 if (sbi->reserved_blocks &&
2447 sbi->current_reserved_blocks < sbi->reserved_blocks)
2448 sbi->current_reserved_blocks = min(sbi->reserved_blocks,
2449 sbi->current_reserved_blocks + count);
2450 spin_unlock(&sbi->stat_lock);
2451 if (unlikely(inode->i_blocks < sectors)) {
2452 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",
2453 inode->i_ino,
2454 (unsigned long long)inode->i_blocks,
2455 (unsigned long long)sectors);
2456 set_sbi_flag(sbi, SBI_NEED_FSCK);
2457 return;
2458 }
2459 f2fs_i_blocks_write(inode, count, false, true);
2460}
2461
2462static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)
2463{
2464 atomic_inc(&sbi->nr_pages[count_type]);
2465
2466 if (count_type == F2FS_DIRTY_DENTS ||
2467 count_type == F2FS_DIRTY_NODES ||
2468 count_type == F2FS_DIRTY_META ||
2469 count_type == F2FS_DIRTY_QDATA ||
2470 count_type == F2FS_DIRTY_IMETA)
2471 set_sbi_flag(sbi, SBI_IS_DIRTY);
2472}
2473
2474static inline void inode_inc_dirty_pages(struct inode *inode)
2475{
2476 atomic_inc(&F2FS_I(inode)->dirty_pages);
2477 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2478 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2479 if (IS_NOQUOTA(inode))
2480 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2481}
2482
2483static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)
2484{
2485 atomic_dec(&sbi->nr_pages[count_type]);
2486}
2487
2488static inline void inode_dec_dirty_pages(struct inode *inode)
2489{
2490 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
2491 !S_ISLNK(inode->i_mode))
2492 return;
2493
2494 atomic_dec(&F2FS_I(inode)->dirty_pages);
2495 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?
2496 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);
2497 if (IS_NOQUOTA(inode))
2498 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);
2499}
2500
2501static inline void inc_atomic_write_cnt(struct inode *inode)
2502{
2503 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2504 struct f2fs_inode_info *fi = F2FS_I(inode);
2505 u64 current_write;
2506
2507 fi->atomic_write_cnt++;
2508 atomic64_inc(&sbi->current_atomic_write);
2509 current_write = atomic64_read(&sbi->current_atomic_write);
2510 if (current_write > sbi->peak_atomic_write)
2511 sbi->peak_atomic_write = current_write;
2512}
2513
2514static inline void release_atomic_write_cnt(struct inode *inode)
2515{
2516 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2517 struct f2fs_inode_info *fi = F2FS_I(inode);
2518
2519 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write);
2520 fi->atomic_write_cnt = 0;
2521}
2522
2523static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)
2524{
2525 return atomic_read(&sbi->nr_pages[count_type]);
2526}
2527
2528static inline int get_dirty_pages(struct inode *inode)
2529{
2530 return atomic_read(&F2FS_I(inode)->dirty_pages);
2531}
2532
2533static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)
2534{
2535 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1,
2536 BLKS_PER_SEC(sbi));
2537}
2538
2539static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)
2540{
2541 return sbi->total_valid_block_count;
2542}
2543
2544static inline block_t discard_blocks(struct f2fs_sb_info *sbi)
2545{
2546 return sbi->discard_blks;
2547}
2548
2549static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
2550{
2551 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2552
2553 /* return NAT or SIT bitmap */
2554 if (flag == NAT_BITMAP)
2555 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2556 else if (flag == SIT_BITMAP)
2557 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2558
2559 return 0;
2560}
2561
2562static inline block_t __cp_payload(struct f2fs_sb_info *sbi)
2563{
2564 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);
2565}
2566
2567static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
2568{
2569 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2570 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;
2571 int offset;
2572
2573 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {
2574 offset = (flag == SIT_BITMAP) ?
2575 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
2576 /*
2577 * if large_nat_bitmap feature is enabled, leave checksum
2578 * protection for all nat/sit bitmaps.
2579 */
2580 return tmp_ptr + offset + sizeof(__le32);
2581 }
2582
2583 if (__cp_payload(sbi) > 0) {
2584 if (flag == NAT_BITMAP)
2585 return tmp_ptr;
2586 else
2587 return (unsigned char *)ckpt + F2FS_BLKSIZE;
2588 } else {
2589 offset = (flag == NAT_BITMAP) ?
2590 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;
2591 return tmp_ptr + offset;
2592 }
2593}
2594
2595static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
2596{
2597 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2598
2599 if (sbi->cur_cp_pack == 2)
2600 start_addr += BLKS_PER_SEG(sbi);
2601 return start_addr;
2602}
2603
2604static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)
2605{
2606 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
2607
2608 if (sbi->cur_cp_pack == 1)
2609 start_addr += BLKS_PER_SEG(sbi);
2610 return start_addr;
2611}
2612
2613static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)
2614{
2615 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;
2616}
2617
2618static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)
2619{
2620 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);
2621}
2622
2623extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);
2624static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,
2625 struct inode *inode, bool is_inode)
2626{
2627 block_t valid_block_count;
2628 unsigned int valid_node_count;
2629 unsigned int avail_user_block_count;
2630 int err;
2631
2632 if (is_inode) {
2633 if (inode) {
2634 err = dquot_alloc_inode(inode);
2635 if (err)
2636 return err;
2637 }
2638 } else {
2639 err = dquot_reserve_block(inode, 1);
2640 if (err)
2641 return err;
2642 }
2643
2644 if (time_to_inject(sbi, FAULT_BLOCK))
2645 goto enospc;
2646
2647 spin_lock(&sbi->stat_lock);
2648
2649 valid_block_count = sbi->total_valid_block_count + 1;
2650 avail_user_block_count = get_available_block_count(sbi, inode, false);
2651
2652 if (unlikely(valid_block_count > avail_user_block_count)) {
2653 spin_unlock(&sbi->stat_lock);
2654 goto enospc;
2655 }
2656
2657 valid_node_count = sbi->total_valid_node_count + 1;
2658 if (unlikely(valid_node_count > sbi->total_node_count)) {
2659 spin_unlock(&sbi->stat_lock);
2660 goto enospc;
2661 }
2662
2663 sbi->total_valid_node_count++;
2664 sbi->total_valid_block_count++;
2665 spin_unlock(&sbi->stat_lock);
2666
2667 if (inode) {
2668 if (is_inode)
2669 f2fs_mark_inode_dirty_sync(inode, true);
2670 else
2671 f2fs_i_blocks_write(inode, 1, true, true);
2672 }
2673
2674 percpu_counter_inc(&sbi->alloc_valid_block_count);
2675 return 0;
2676
2677enospc:
2678 if (is_inode) {
2679 if (inode)
2680 dquot_free_inode(inode);
2681 } else {
2682 dquot_release_reservation_block(inode, 1);
2683 }
2684 return -ENOSPC;
2685}
2686
2687static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,
2688 struct inode *inode, bool is_inode)
2689{
2690 spin_lock(&sbi->stat_lock);
2691
2692 if (unlikely(!sbi->total_valid_block_count ||
2693 !sbi->total_valid_node_count)) {
2694 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",
2695 sbi->total_valid_block_count,
2696 sbi->total_valid_node_count);
2697 set_sbi_flag(sbi, SBI_NEED_FSCK);
2698 } else {
2699 sbi->total_valid_block_count--;
2700 sbi->total_valid_node_count--;
2701 }
2702
2703 if (sbi->reserved_blocks &&
2704 sbi->current_reserved_blocks < sbi->reserved_blocks)
2705 sbi->current_reserved_blocks++;
2706
2707 spin_unlock(&sbi->stat_lock);
2708
2709 if (is_inode) {
2710 dquot_free_inode(inode);
2711 } else {
2712 if (unlikely(inode->i_blocks == 0)) {
2713 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",
2714 inode->i_ino,
2715 (unsigned long long)inode->i_blocks);
2716 set_sbi_flag(sbi, SBI_NEED_FSCK);
2717 return;
2718 }
2719 f2fs_i_blocks_write(inode, 1, false, true);
2720 }
2721}
2722
2723static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)
2724{
2725 return sbi->total_valid_node_count;
2726}
2727
2728static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)
2729{
2730 percpu_counter_inc(&sbi->total_valid_inode_count);
2731}
2732
2733static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)
2734{
2735 percpu_counter_dec(&sbi->total_valid_inode_count);
2736}
2737
2738static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)
2739{
2740 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);
2741}
2742
2743static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,
2744 pgoff_t index, bool for_write)
2745{
2746 struct page *page;
2747 unsigned int flags;
2748
2749 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {
2750 if (!for_write)
2751 page = find_get_page_flags(mapping, index,
2752 FGP_LOCK | FGP_ACCESSED);
2753 else
2754 page = find_lock_page(mapping, index);
2755 if (page)
2756 return page;
2757
2758 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))
2759 return NULL;
2760 }
2761
2762 if (!for_write)
2763 return grab_cache_page(mapping, index);
2764
2765 flags = memalloc_nofs_save();
2766 page = grab_cache_page_write_begin(mapping, index);
2767 memalloc_nofs_restore(flags);
2768
2769 return page;
2770}
2771
2772static inline struct page *f2fs_pagecache_get_page(
2773 struct address_space *mapping, pgoff_t index,
2774 fgf_t fgp_flags, gfp_t gfp_mask)
2775{
2776 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET))
2777 return NULL;
2778
2779 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);
2780}
2781
2782static inline void f2fs_put_page(struct page *page, int unlock)
2783{
2784 if (!page)
2785 return;
2786
2787 if (unlock) {
2788 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));
2789 unlock_page(page);
2790 }
2791 put_page(page);
2792}
2793
2794static inline void f2fs_put_dnode(struct dnode_of_data *dn)
2795{
2796 if (dn->node_page)
2797 f2fs_put_page(dn->node_page, 1);
2798 if (dn->inode_page && dn->node_page != dn->inode_page)
2799 f2fs_put_page(dn->inode_page, 0);
2800 dn->node_page = NULL;
2801 dn->inode_page = NULL;
2802}
2803
2804static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,
2805 size_t size)
2806{
2807 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);
2808}
2809
2810static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,
2811 gfp_t flags)
2812{
2813 void *entry;
2814
2815 entry = kmem_cache_alloc(cachep, flags);
2816 if (!entry)
2817 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);
2818 return entry;
2819}
2820
2821static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,
2822 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)
2823{
2824 if (nofail)
2825 return f2fs_kmem_cache_alloc_nofail(cachep, flags);
2826
2827 if (time_to_inject(sbi, FAULT_SLAB_ALLOC))
2828 return NULL;
2829
2830 return kmem_cache_alloc(cachep, flags);
2831}
2832
2833static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)
2834{
2835 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||
2836 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||
2837 get_pages(sbi, F2FS_WB_CP_DATA) ||
2838 get_pages(sbi, F2FS_DIO_READ) ||
2839 get_pages(sbi, F2FS_DIO_WRITE))
2840 return true;
2841
2842 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&
2843 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))
2844 return true;
2845
2846 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&
2847 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))
2848 return true;
2849 return false;
2850}
2851
2852static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
2853{
2854 if (sbi->gc_mode == GC_URGENT_HIGH)
2855 return true;
2856
2857 if (is_inflight_io(sbi, type))
2858 return false;
2859
2860 if (sbi->gc_mode == GC_URGENT_MID)
2861 return true;
2862
2863 if (sbi->gc_mode == GC_URGENT_LOW &&
2864 (type == DISCARD_TIME || type == GC_TIME))
2865 return true;
2866
2867 return f2fs_time_over(sbi, type);
2868}
2869
2870static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,
2871 unsigned long index, void *item)
2872{
2873 while (radix_tree_insert(root, index, item))
2874 cond_resched();
2875}
2876
2877#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)
2878
2879static inline bool IS_INODE(struct page *page)
2880{
2881 struct f2fs_node *p = F2FS_NODE(page);
2882
2883 return RAW_IS_INODE(p);
2884}
2885
2886static inline int offset_in_addr(struct f2fs_inode *i)
2887{
2888 return (i->i_inline & F2FS_EXTRA_ATTR) ?
2889 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;
2890}
2891
2892static inline __le32 *blkaddr_in_node(struct f2fs_node *node)
2893{
2894 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;
2895}
2896
2897static inline int f2fs_has_extra_attr(struct inode *inode);
2898static inline block_t data_blkaddr(struct inode *inode,
2899 struct page *node_page, unsigned int offset)
2900{
2901 struct f2fs_node *raw_node;
2902 __le32 *addr_array;
2903 int base = 0;
2904 bool is_inode = IS_INODE(node_page);
2905
2906 raw_node = F2FS_NODE(node_page);
2907
2908 if (is_inode) {
2909 if (!inode)
2910 /* from GC path only */
2911 base = offset_in_addr(&raw_node->i);
2912 else if (f2fs_has_extra_attr(inode))
2913 base = get_extra_isize(inode);
2914 }
2915
2916 addr_array = blkaddr_in_node(raw_node);
2917 return le32_to_cpu(addr_array[base + offset]);
2918}
2919
2920static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)
2921{
2922 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);
2923}
2924
2925static inline int f2fs_test_bit(unsigned int nr, char *addr)
2926{
2927 int mask;
2928
2929 addr += (nr >> 3);
2930 mask = BIT(7 - (nr & 0x07));
2931 return mask & *addr;
2932}
2933
2934static inline void f2fs_set_bit(unsigned int nr, char *addr)
2935{
2936 int mask;
2937
2938 addr += (nr >> 3);
2939 mask = BIT(7 - (nr & 0x07));
2940 *addr |= mask;
2941}
2942
2943static inline void f2fs_clear_bit(unsigned int nr, char *addr)
2944{
2945 int mask;
2946
2947 addr += (nr >> 3);
2948 mask = BIT(7 - (nr & 0x07));
2949 *addr &= ~mask;
2950}
2951
2952static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)
2953{
2954 int mask;
2955 int ret;
2956
2957 addr += (nr >> 3);
2958 mask = BIT(7 - (nr & 0x07));
2959 ret = mask & *addr;
2960 *addr |= mask;
2961 return ret;
2962}
2963
2964static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)
2965{
2966 int mask;
2967 int ret;
2968
2969 addr += (nr >> 3);
2970 mask = BIT(7 - (nr & 0x07));
2971 ret = mask & *addr;
2972 *addr &= ~mask;
2973 return ret;
2974}
2975
2976static inline void f2fs_change_bit(unsigned int nr, char *addr)
2977{
2978 int mask;
2979
2980 addr += (nr >> 3);
2981 mask = BIT(7 - (nr & 0x07));
2982 *addr ^= mask;
2983}
2984
2985/*
2986 * On-disk inode flags (f2fs_inode::i_flags)
2987 */
2988#define F2FS_COMPR_FL 0x00000004 /* Compress file */
2989#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */
2990#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
2991#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */
2992#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */
2993#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */
2994#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */
2995#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */
2996#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
2997#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
2998#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */
2999
3000#define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL)
3001
3002/* Flags that should be inherited by new inodes from their parent. */
3003#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \
3004 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
3005 F2FS_CASEFOLD_FL)
3006
3007/* Flags that are appropriate for regular files (all but dir-specific ones). */
3008#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \
3009 F2FS_CASEFOLD_FL))
3010
3011/* Flags that are appropriate for non-directories/regular files. */
3012#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)
3013
3014static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)
3015{
3016 if (S_ISDIR(mode))
3017 return flags;
3018 else if (S_ISREG(mode))
3019 return flags & F2FS_REG_FLMASK;
3020 else
3021 return flags & F2FS_OTHER_FLMASK;
3022}
3023
3024static inline void __mark_inode_dirty_flag(struct inode *inode,
3025 int flag, bool set)
3026{
3027 switch (flag) {
3028 case FI_INLINE_XATTR:
3029 case FI_INLINE_DATA:
3030 case FI_INLINE_DENTRY:
3031 case FI_NEW_INODE:
3032 if (set)
3033 return;
3034 fallthrough;
3035 case FI_DATA_EXIST:
3036 case FI_INLINE_DOTS:
3037 case FI_PIN_FILE:
3038 case FI_COMPRESS_RELEASED:
3039 case FI_ATOMIC_COMMITTED:
3040 f2fs_mark_inode_dirty_sync(inode, true);
3041 }
3042}
3043
3044static inline void set_inode_flag(struct inode *inode, int flag)
3045{
3046 set_bit(flag, F2FS_I(inode)->flags);
3047 __mark_inode_dirty_flag(inode, flag, true);
3048}
3049
3050static inline int is_inode_flag_set(struct inode *inode, int flag)
3051{
3052 return test_bit(flag, F2FS_I(inode)->flags);
3053}
3054
3055static inline void clear_inode_flag(struct inode *inode, int flag)
3056{
3057 clear_bit(flag, F2FS_I(inode)->flags);
3058 __mark_inode_dirty_flag(inode, flag, false);
3059}
3060
3061static inline bool f2fs_verity_in_progress(struct inode *inode)
3062{
3063 return IS_ENABLED(CONFIG_FS_VERITY) &&
3064 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);
3065}
3066
3067static inline void set_acl_inode(struct inode *inode, umode_t mode)
3068{
3069 F2FS_I(inode)->i_acl_mode = mode;
3070 set_inode_flag(inode, FI_ACL_MODE);
3071 f2fs_mark_inode_dirty_sync(inode, false);
3072}
3073
3074static inline void f2fs_i_links_write(struct inode *inode, bool inc)
3075{
3076 if (inc)
3077 inc_nlink(inode);
3078 else
3079 drop_nlink(inode);
3080 f2fs_mark_inode_dirty_sync(inode, true);
3081}
3082
3083static inline void f2fs_i_blocks_write(struct inode *inode,
3084 block_t diff, bool add, bool claim)
3085{
3086 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3087 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3088
3089 /* add = 1, claim = 1 should be dquot_reserve_block in pair */
3090 if (add) {
3091 if (claim)
3092 dquot_claim_block(inode, diff);
3093 else
3094 dquot_alloc_block_nofail(inode, diff);
3095 } else {
3096 dquot_free_block(inode, diff);
3097 }
3098
3099 f2fs_mark_inode_dirty_sync(inode, true);
3100 if (clean || recover)
3101 set_inode_flag(inode, FI_AUTO_RECOVER);
3102}
3103
3104static inline bool f2fs_is_atomic_file(struct inode *inode);
3105
3106static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)
3107{
3108 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);
3109 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);
3110
3111 if (i_size_read(inode) == i_size)
3112 return;
3113
3114 i_size_write(inode, i_size);
3115
3116 if (f2fs_is_atomic_file(inode))
3117 return;
3118
3119 f2fs_mark_inode_dirty_sync(inode, true);
3120 if (clean || recover)
3121 set_inode_flag(inode, FI_AUTO_RECOVER);
3122}
3123
3124static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)
3125{
3126 F2FS_I(inode)->i_current_depth = depth;
3127 f2fs_mark_inode_dirty_sync(inode, true);
3128}
3129
3130static inline void f2fs_i_gc_failures_write(struct inode *inode,
3131 unsigned int count)
3132{
3133 F2FS_I(inode)->i_gc_failures = count;
3134 f2fs_mark_inode_dirty_sync(inode, true);
3135}
3136
3137static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)
3138{
3139 F2FS_I(inode)->i_xattr_nid = xnid;
3140 f2fs_mark_inode_dirty_sync(inode, true);
3141}
3142
3143static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)
3144{
3145 F2FS_I(inode)->i_pino = pino;
3146 f2fs_mark_inode_dirty_sync(inode, true);
3147}
3148
3149static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)
3150{
3151 struct f2fs_inode_info *fi = F2FS_I(inode);
3152
3153 if (ri->i_inline & F2FS_INLINE_XATTR)
3154 set_bit(FI_INLINE_XATTR, fi->flags);
3155 if (ri->i_inline & F2FS_INLINE_DATA)
3156 set_bit(FI_INLINE_DATA, fi->flags);
3157 if (ri->i_inline & F2FS_INLINE_DENTRY)
3158 set_bit(FI_INLINE_DENTRY, fi->flags);
3159 if (ri->i_inline & F2FS_DATA_EXIST)
3160 set_bit(FI_DATA_EXIST, fi->flags);
3161 if (ri->i_inline & F2FS_INLINE_DOTS)
3162 set_bit(FI_INLINE_DOTS, fi->flags);
3163 if (ri->i_inline & F2FS_EXTRA_ATTR)
3164 set_bit(FI_EXTRA_ATTR, fi->flags);
3165 if (ri->i_inline & F2FS_PIN_FILE)
3166 set_bit(FI_PIN_FILE, fi->flags);
3167 if (ri->i_inline & F2FS_COMPRESS_RELEASED)
3168 set_bit(FI_COMPRESS_RELEASED, fi->flags);
3169}
3170
3171static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)
3172{
3173 ri->i_inline = 0;
3174
3175 if (is_inode_flag_set(inode, FI_INLINE_XATTR))
3176 ri->i_inline |= F2FS_INLINE_XATTR;
3177 if (is_inode_flag_set(inode, FI_INLINE_DATA))
3178 ri->i_inline |= F2FS_INLINE_DATA;
3179 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))
3180 ri->i_inline |= F2FS_INLINE_DENTRY;
3181 if (is_inode_flag_set(inode, FI_DATA_EXIST))
3182 ri->i_inline |= F2FS_DATA_EXIST;
3183 if (is_inode_flag_set(inode, FI_INLINE_DOTS))
3184 ri->i_inline |= F2FS_INLINE_DOTS;
3185 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))
3186 ri->i_inline |= F2FS_EXTRA_ATTR;
3187 if (is_inode_flag_set(inode, FI_PIN_FILE))
3188 ri->i_inline |= F2FS_PIN_FILE;
3189 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))
3190 ri->i_inline |= F2FS_COMPRESS_RELEASED;
3191}
3192
3193static inline int f2fs_has_extra_attr(struct inode *inode)
3194{
3195 return is_inode_flag_set(inode, FI_EXTRA_ATTR);
3196}
3197
3198static inline int f2fs_has_inline_xattr(struct inode *inode)
3199{
3200 return is_inode_flag_set(inode, FI_INLINE_XATTR);
3201}
3202
3203static inline int f2fs_compressed_file(struct inode *inode)
3204{
3205 return S_ISREG(inode->i_mode) &&
3206 is_inode_flag_set(inode, FI_COMPRESSED_FILE);
3207}
3208
3209static inline bool f2fs_need_compress_data(struct inode *inode)
3210{
3211 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
3212
3213 if (!f2fs_compressed_file(inode))
3214 return false;
3215
3216 if (compress_mode == COMPR_MODE_FS)
3217 return true;
3218 else if (compress_mode == COMPR_MODE_USER &&
3219 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
3220 return true;
3221
3222 return false;
3223}
3224
3225static inline unsigned int addrs_per_inode(struct inode *inode)
3226{
3227 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) -
3228 get_inline_xattr_addrs(inode);
3229
3230 if (!f2fs_compressed_file(inode))
3231 return addrs;
3232 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);
3233}
3234
3235static inline unsigned int addrs_per_block(struct inode *inode)
3236{
3237 if (!f2fs_compressed_file(inode))
3238 return DEF_ADDRS_PER_BLOCK;
3239 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size);
3240}
3241
3242static inline void *inline_xattr_addr(struct inode *inode, struct page *page)
3243{
3244 struct f2fs_inode *ri = F2FS_INODE(page);
3245
3246 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -
3247 get_inline_xattr_addrs(inode)]);
3248}
3249
3250static inline int inline_xattr_size(struct inode *inode)
3251{
3252 if (f2fs_has_inline_xattr(inode))
3253 return get_inline_xattr_addrs(inode) * sizeof(__le32);
3254 return 0;
3255}
3256
3257/*
3258 * Notice: check inline_data flag without inode page lock is unsafe.
3259 * It could change at any time by f2fs_convert_inline_page().
3260 */
3261static inline int f2fs_has_inline_data(struct inode *inode)
3262{
3263 return is_inode_flag_set(inode, FI_INLINE_DATA);
3264}
3265
3266static inline int f2fs_exist_data(struct inode *inode)
3267{
3268 return is_inode_flag_set(inode, FI_DATA_EXIST);
3269}
3270
3271static inline int f2fs_has_inline_dots(struct inode *inode)
3272{
3273 return is_inode_flag_set(inode, FI_INLINE_DOTS);
3274}
3275
3276static inline int f2fs_is_mmap_file(struct inode *inode)
3277{
3278 return is_inode_flag_set(inode, FI_MMAP_FILE);
3279}
3280
3281static inline bool f2fs_is_pinned_file(struct inode *inode)
3282{
3283 return is_inode_flag_set(inode, FI_PIN_FILE);
3284}
3285
3286static inline bool f2fs_is_atomic_file(struct inode *inode)
3287{
3288 return is_inode_flag_set(inode, FI_ATOMIC_FILE);
3289}
3290
3291static inline bool f2fs_is_cow_file(struct inode *inode)
3292{
3293 return is_inode_flag_set(inode, FI_COW_FILE);
3294}
3295
3296static inline __le32 *get_dnode_addr(struct inode *inode,
3297 struct page *node_page);
3298static inline void *inline_data_addr(struct inode *inode, struct page *page)
3299{
3300 __le32 *addr = get_dnode_addr(inode, page);
3301
3302 return (void *)(addr + DEF_INLINE_RESERVED_SIZE);
3303}
3304
3305static inline int f2fs_has_inline_dentry(struct inode *inode)
3306{
3307 return is_inode_flag_set(inode, FI_INLINE_DENTRY);
3308}
3309
3310static inline int is_file(struct inode *inode, int type)
3311{
3312 return F2FS_I(inode)->i_advise & type;
3313}
3314
3315static inline void set_file(struct inode *inode, int type)
3316{
3317 if (is_file(inode, type))
3318 return;
3319 F2FS_I(inode)->i_advise |= type;
3320 f2fs_mark_inode_dirty_sync(inode, true);
3321}
3322
3323static inline void clear_file(struct inode *inode, int type)
3324{
3325 if (!is_file(inode, type))
3326 return;
3327 F2FS_I(inode)->i_advise &= ~type;
3328 f2fs_mark_inode_dirty_sync(inode, true);
3329}
3330
3331static inline bool f2fs_is_time_consistent(struct inode *inode)
3332{
3333 struct timespec64 ts = inode_get_atime(inode);
3334
3335 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts))
3336 return false;
3337 ts = inode_get_ctime(inode);
3338 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts))
3339 return false;
3340 ts = inode_get_mtime(inode);
3341 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts))
3342 return false;
3343 return true;
3344}
3345
3346static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
3347{
3348 bool ret;
3349
3350 if (dsync) {
3351 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3352
3353 spin_lock(&sbi->inode_lock[DIRTY_META]);
3354 ret = list_empty(&F2FS_I(inode)->gdirty_list);
3355 spin_unlock(&sbi->inode_lock[DIRTY_META]);
3356 return ret;
3357 }
3358 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||
3359 file_keep_isize(inode) ||
3360 i_size_read(inode) & ~PAGE_MASK)
3361 return false;
3362
3363 if (!f2fs_is_time_consistent(inode))
3364 return false;
3365
3366 spin_lock(&F2FS_I(inode)->i_size_lock);
3367 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
3368 spin_unlock(&F2FS_I(inode)->i_size_lock);
3369
3370 return ret;
3371}
3372
3373static inline bool f2fs_readonly(struct super_block *sb)
3374{
3375 return sb_rdonly(sb);
3376}
3377
3378static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)
3379{
3380 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);
3381}
3382
3383static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,
3384 size_t size, gfp_t flags)
3385{
3386 if (time_to_inject(sbi, FAULT_KMALLOC))
3387 return NULL;
3388
3389 return kmalloc(size, flags);
3390}
3391
3392static inline void *f2fs_getname(struct f2fs_sb_info *sbi)
3393{
3394 if (time_to_inject(sbi, FAULT_KMALLOC))
3395 return NULL;
3396
3397 return __getname();
3398}
3399
3400static inline void f2fs_putname(char *buf)
3401{
3402 __putname(buf);
3403}
3404
3405static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,
3406 size_t size, gfp_t flags)
3407{
3408 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);
3409}
3410
3411static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,
3412 size_t size, gfp_t flags)
3413{
3414 if (time_to_inject(sbi, FAULT_KVMALLOC))
3415 return NULL;
3416
3417 return kvmalloc(size, flags);
3418}
3419
3420static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,
3421 size_t size, gfp_t flags)
3422{
3423 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);
3424}
3425
3426static inline int get_extra_isize(struct inode *inode)
3427{
3428 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);
3429}
3430
3431static inline int get_inline_xattr_addrs(struct inode *inode)
3432{
3433 return F2FS_I(inode)->i_inline_xattr_size;
3434}
3435
3436static inline __le32 *get_dnode_addr(struct inode *inode,
3437 struct page *node_page)
3438{
3439 int base = 0;
3440
3441 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode))
3442 base = get_extra_isize(inode);
3443
3444 return blkaddr_in_node(F2FS_NODE(node_page)) + base;
3445}
3446
3447#define f2fs_get_inode_mode(i) \
3448 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \
3449 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))
3450
3451#define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32))
3452
3453#define F2FS_TOTAL_EXTRA_ATTR_SIZE \
3454 (offsetof(struct f2fs_inode, i_extra_end) - \
3455 offsetof(struct f2fs_inode, i_extra_isize)) \
3456
3457#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))
3458#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \
3459 ((offsetof(typeof(*(f2fs_inode)), field) + \
3460 sizeof((f2fs_inode)->field)) \
3461 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \
3462
3463#define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1)
3464
3465#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)
3466
3467bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3468 block_t blkaddr, int type);
3469static inline void verify_blkaddr(struct f2fs_sb_info *sbi,
3470 block_t blkaddr, int type)
3471{
3472 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type))
3473 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",
3474 blkaddr, type);
3475}
3476
3477static inline bool __is_valid_data_blkaddr(block_t blkaddr)
3478{
3479 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||
3480 blkaddr == COMPRESS_ADDR)
3481 return false;
3482 return true;
3483}
3484
3485/*
3486 * file.c
3487 */
3488int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);
3489int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);
3490int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);
3491int f2fs_truncate(struct inode *inode);
3492int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path,
3493 struct kstat *stat, u32 request_mask, unsigned int flags);
3494int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
3495 struct iattr *attr);
3496int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);
3497void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);
3498int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag,
3499 bool readonly);
3500int f2fs_precache_extents(struct inode *inode);
3501int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3502int f2fs_fileattr_set(struct mnt_idmap *idmap,
3503 struct dentry *dentry, struct fileattr *fa);
3504long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
3505long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
3506int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);
3507int f2fs_pin_file_control(struct inode *inode, bool inc);
3508
3509/*
3510 * inode.c
3511 */
3512void f2fs_set_inode_flags(struct inode *inode);
3513bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);
3514void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);
3515struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);
3516struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);
3517int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);
3518void f2fs_update_inode(struct inode *inode, struct page *node_page);
3519void f2fs_update_inode_page(struct inode *inode);
3520int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);
3521void f2fs_evict_inode(struct inode *inode);
3522void f2fs_handle_failed_inode(struct inode *inode);
3523
3524/*
3525 * namei.c
3526 */
3527int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,
3528 bool hot, bool set);
3529struct dentry *f2fs_get_parent(struct dentry *child);
3530int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
3531 struct inode **new_inode);
3532
3533/*
3534 * dir.c
3535 */
3536int f2fs_init_casefolded_name(const struct inode *dir,
3537 struct f2fs_filename *fname);
3538int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,
3539 int lookup, struct f2fs_filename *fname);
3540int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,
3541 struct f2fs_filename *fname);
3542void f2fs_free_filename(struct f2fs_filename *fname);
3543struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,
3544 const struct f2fs_filename *fname, int *max_slots);
3545int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
3546 unsigned int start_pos, struct fscrypt_str *fstr);
3547void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,
3548 struct f2fs_dentry_ptr *d);
3549struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,
3550 const struct f2fs_filename *fname, struct page *dpage);
3551void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,
3552 unsigned int current_depth);
3553int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);
3554void f2fs_drop_nlink(struct inode *dir, struct inode *inode);
3555struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
3556 const struct f2fs_filename *fname,
3557 struct page **res_page);
3558struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,
3559 const struct qstr *child, struct page **res_page);
3560struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);
3561ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,
3562 struct page **page);
3563void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,
3564 struct page *page, struct inode *inode);
3565bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,
3566 const struct f2fs_filename *fname);
3567void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
3568 const struct fscrypt_str *name, f2fs_hash_t name_hash,
3569 unsigned int bit_pos);
3570int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,
3571 struct inode *inode, nid_t ino, umode_t mode);
3572int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,
3573 struct inode *inode, nid_t ino, umode_t mode);
3574int f2fs_do_add_link(struct inode *dir, const struct qstr *name,
3575 struct inode *inode, nid_t ino, umode_t mode);
3576void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
3577 struct inode *dir, struct inode *inode);
3578int f2fs_do_tmpfile(struct inode *inode, struct inode *dir,
3579 struct f2fs_filename *fname);
3580bool f2fs_empty_dir(struct inode *dir);
3581
3582static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)
3583{
3584 if (fscrypt_is_nokey_name(dentry))
3585 return -ENOKEY;
3586 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,
3587 inode, inode->i_ino, inode->i_mode);
3588}
3589
3590/*
3591 * super.c
3592 */
3593int f2fs_inode_dirtied(struct inode *inode, bool sync);
3594void f2fs_inode_synced(struct inode *inode);
3595int f2fs_dquot_initialize(struct inode *inode);
3596int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);
3597int f2fs_quota_sync(struct super_block *sb, int type);
3598loff_t max_file_blocks(struct inode *inode);
3599void f2fs_quota_off_umount(struct super_block *sb);
3600void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag);
3601void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
3602 bool irq_context);
3603void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);
3604void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error);
3605int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);
3606int f2fs_sync_fs(struct super_block *sb, int sync);
3607int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);
3608
3609/*
3610 * hash.c
3611 */
3612void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);
3613
3614/*
3615 * node.c
3616 */
3617struct node_info;
3618
3619int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);
3620bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);
3621bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);
3622void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);
3623void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);
3624void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);
3625int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);
3626bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);
3627bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);
3628int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,
3629 struct node_info *ni, bool checkpoint_context);
3630pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);
3631int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);
3632int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);
3633int f2fs_truncate_xattr_node(struct inode *inode);
3634int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,
3635 unsigned int seq_id);
3636bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);
3637int f2fs_remove_inode_page(struct inode *inode);
3638struct page *f2fs_new_inode_page(struct inode *inode);
3639struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);
3640void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);
3641struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);
3642struct page *f2fs_get_node_page_ra(struct page *parent, int start);
3643int f2fs_move_node_page(struct page *node_page, int gc_type);
3644void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);
3645int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,
3646 struct writeback_control *wbc, bool atomic,
3647 unsigned int *seq_id);
3648int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
3649 struct writeback_control *wbc,
3650 bool do_balance, enum iostat_type io_type);
3651int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);
3652bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);
3653void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);
3654void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);
3655int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);
3656int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);
3657int f2fs_recover_xattr_data(struct inode *inode, struct page *page);
3658int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);
3659int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,
3660 unsigned int segno, struct f2fs_summary_block *sum);
3661void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);
3662int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3663int f2fs_build_node_manager(struct f2fs_sb_info *sbi);
3664void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);
3665int __init f2fs_create_node_manager_caches(void);
3666void f2fs_destroy_node_manager_caches(void);
3667
3668/*
3669 * segment.c
3670 */
3671bool f2fs_need_SSR(struct f2fs_sb_info *sbi);
3672int f2fs_commit_atomic_write(struct inode *inode);
3673void f2fs_abort_atomic_write(struct inode *inode, bool clean);
3674void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);
3675void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);
3676int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);
3677int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);
3678int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);
3679void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);
3680void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);
3681bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);
3682int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);
3683void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);
3684void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);
3685bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);
3686void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,
3687 struct cp_control *cpc);
3688void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);
3689block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);
3690int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);
3691void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);
3692int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);
3693bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);
3694int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);
3695void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);
3696void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);
3697int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,
3698 unsigned int start, unsigned int end);
3699int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);
3700int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi);
3701int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);
3702int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);
3703bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,
3704 struct cp_control *cpc);
3705struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);
3706void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,
3707 block_t blk_addr);
3708void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page,
3709 enum iostat_type io_type);
3710void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);
3711void f2fs_outplace_write_data(struct dnode_of_data *dn,
3712 struct f2fs_io_info *fio);
3713int f2fs_inplace_write_data(struct f2fs_io_info *fio);
3714void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
3715 block_t old_blkaddr, block_t new_blkaddr,
3716 bool recover_curseg, bool recover_newaddr,
3717 bool from_gc);
3718void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,
3719 block_t old_addr, block_t new_addr,
3720 unsigned char version, bool recover_curseg,
3721 bool recover_newaddr);
3722int f2fs_get_segment_temp(int seg_type);
3723int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,
3724 block_t old_blkaddr, block_t *new_blkaddr,
3725 struct f2fs_summary *sum, int type,
3726 struct f2fs_io_info *fio);
3727void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,
3728 block_t blkaddr, unsigned int blkcnt);
3729void f2fs_wait_on_page_writeback(struct page *page,
3730 enum page_type type, bool ordered, bool locked);
3731void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);
3732void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,
3733 block_t len);
3734void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3735void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);
3736int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,
3737 unsigned int val, int alloc);
3738void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3739int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);
3740int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);
3741int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);
3742void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);
3743int __init f2fs_create_segment_manager_caches(void);
3744void f2fs_destroy_segment_manager_caches(void);
3745int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint);
3746enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,
3747 enum page_type type, enum temp_type temp);
3748unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi,
3749 unsigned int segno);
3750unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,
3751 unsigned int segno);
3752
3753#define DEF_FRAGMENT_SIZE 4
3754#define MIN_FRAGMENT_SIZE 1
3755#define MAX_FRAGMENT_SIZE 512
3756
3757static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)
3758{
3759 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||
3760 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;
3761}
3762
3763/*
3764 * checkpoint.c
3765 */
3766void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,
3767 unsigned char reason);
3768void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);
3769struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3770struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);
3771struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);
3772struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);
3773bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
3774 block_t blkaddr, int type);
3775bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi,
3776 block_t blkaddr, int type);
3777int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
3778 int type, bool sync);
3779void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,
3780 unsigned int ra_blocks);
3781long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
3782 long nr_to_write, enum iostat_type io_type);
3783void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3784void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);
3785void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);
3786bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);
3787void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3788 unsigned int devidx, int type);
3789bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,
3790 unsigned int devidx, int type);
3791int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);
3792void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);
3793void f2fs_add_orphan_inode(struct inode *inode);
3794void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);
3795int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);
3796int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);
3797void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);
3798void f2fs_remove_dirty_inode(struct inode *inode);
3799int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,
3800 bool from_cp);
3801void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);
3802u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);
3803int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);
3804void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);
3805int __init f2fs_create_checkpoint_caches(void);
3806void f2fs_destroy_checkpoint_caches(void);
3807int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);
3808int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);
3809void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);
3810void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);
3811
3812/*
3813 * data.c
3814 */
3815int __init f2fs_init_bioset(void);
3816void f2fs_destroy_bioset(void);
3817bool f2fs_is_cp_guaranteed(struct page *page);
3818int f2fs_init_bio_entry_cache(void);
3819void f2fs_destroy_bio_entry_cache(void);
3820void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio,
3821 enum page_type type);
3822int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);
3823void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);
3824void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
3825 struct inode *inode, struct page *page,
3826 nid_t ino, enum page_type type);
3827void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,
3828 struct bio **bio, struct page *page);
3829void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);
3830int f2fs_submit_page_bio(struct f2fs_io_info *fio);
3831int f2fs_merge_page_bio(struct f2fs_io_info *fio);
3832void f2fs_submit_page_write(struct f2fs_io_info *fio);
3833struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
3834 block_t blk_addr, sector_t *sector);
3835int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);
3836void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3837void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);
3838int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);
3839int f2fs_reserve_new_block(struct dnode_of_data *dn);
3840int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index);
3841int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);
3842struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,
3843 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs);
3844struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index,
3845 pgoff_t *next_pgofs);
3846struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,
3847 bool for_write);
3848struct page *f2fs_get_new_data_page(struct inode *inode,
3849 struct page *ipage, pgoff_t index, bool new_i_size);
3850int f2fs_do_write_data_page(struct f2fs_io_info *fio);
3851int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag);
3852int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3853 u64 start, u64 len);
3854int f2fs_encrypt_one_page(struct f2fs_io_info *fio);
3855bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);
3856bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);
3857int f2fs_write_single_data_page(struct page *page, int *submitted,
3858 struct bio **bio, sector_t *last_block,
3859 struct writeback_control *wbc,
3860 enum iostat_type io_type,
3861 int compr_blocks, bool allow_balance);
3862void f2fs_write_failed(struct inode *inode, loff_t to);
3863void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);
3864bool f2fs_release_folio(struct folio *folio, gfp_t wait);
3865bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);
3866void f2fs_clear_page_cache_dirty_tag(struct page *page);
3867int f2fs_init_post_read_processing(void);
3868void f2fs_destroy_post_read_processing(void);
3869int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);
3870void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);
3871extern const struct iomap_ops f2fs_iomap_ops;
3872
3873/*
3874 * gc.c
3875 */
3876int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);
3877void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);
3878block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);
3879int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);
3880void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);
3881int f2fs_gc_range(struct f2fs_sb_info *sbi,
3882 unsigned int start_seg, unsigned int end_seg,
3883 bool dry_run, unsigned int dry_run_sections);
3884int f2fs_resize_fs(struct file *filp, __u64 block_count);
3885int __init f2fs_create_garbage_collection_cache(void);
3886void f2fs_destroy_garbage_collection_cache(void);
3887/* victim selection function for cleaning and SSR */
3888int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result,
3889 int gc_type, int type, char alloc_mode,
3890 unsigned long long age);
3891
3892/*
3893 * recovery.c
3894 */
3895int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);
3896bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);
3897int __init f2fs_create_recovery_cache(void);
3898void f2fs_destroy_recovery_cache(void);
3899
3900/*
3901 * debug.c
3902 */
3903#ifdef CONFIG_F2FS_STAT_FS
3904struct f2fs_stat_info {
3905 struct list_head stat_list;
3906 struct f2fs_sb_info *sbi;
3907 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;
3908 int main_area_segs, main_area_sections, main_area_zones;
3909 unsigned long long hit_cached[NR_EXTENT_CACHES];
3910 unsigned long long hit_rbtree[NR_EXTENT_CACHES];
3911 unsigned long long total_ext[NR_EXTENT_CACHES];
3912 unsigned long long hit_total[NR_EXTENT_CACHES];
3913 int ext_tree[NR_EXTENT_CACHES];
3914 int zombie_tree[NR_EXTENT_CACHES];
3915 int ext_node[NR_EXTENT_CACHES];
3916 /* to count memory footprint */
3917 unsigned long long ext_mem[NR_EXTENT_CACHES];
3918 /* for read extent cache */
3919 unsigned long long hit_largest;
3920 /* for block age extent cache */
3921 unsigned long long allocated_data_blocks;
3922 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;
3923 int ndirty_data, ndirty_qdata;
3924 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;
3925 int nats, dirty_nats, sits, dirty_sits;
3926 int free_nids, avail_nids, alloc_nids;
3927 int total_count, utilization;
3928 int nr_wb_cp_data, nr_wb_data;
3929 int nr_rd_data, nr_rd_node, nr_rd_meta;
3930 int nr_dio_read, nr_dio_write;
3931 unsigned int io_skip_bggc, other_skip_bggc;
3932 int nr_flushing, nr_flushed, flush_list_empty;
3933 int nr_discarding, nr_discarded;
3934 int nr_discard_cmd;
3935 unsigned int undiscard_blks;
3936 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;
3937 unsigned int cur_ckpt_time, peak_ckpt_time;
3938 int inline_xattr, inline_inode, inline_dir, append, update, orphans;
3939 int compr_inode, swapfile_inode;
3940 unsigned long long compr_blocks;
3941 int aw_cnt, max_aw_cnt;
3942 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
3943 unsigned int bimodal, avg_vblocks;
3944 int util_free, util_valid, util_invalid;
3945 int rsvd_segs, overp_segs;
3946 int dirty_count, node_pages, meta_pages, compress_pages;
3947 int compress_page_hit;
3948 int prefree_count, free_segs, free_secs;
3949 int cp_call_count[MAX_CALL_TYPE], cp_count;
3950 int gc_call_count[MAX_CALL_TYPE];
3951 int gc_segs[2][2];
3952 int gc_secs[2][2];
3953 int tot_blks, data_blks, node_blks;
3954 int bg_data_blks, bg_node_blks;
3955 int curseg[NR_CURSEG_TYPE];
3956 int cursec[NR_CURSEG_TYPE];
3957 int curzone[NR_CURSEG_TYPE];
3958 unsigned int dirty_seg[NR_CURSEG_TYPE];
3959 unsigned int full_seg[NR_CURSEG_TYPE];
3960 unsigned int valid_blks[NR_CURSEG_TYPE];
3961
3962 unsigned int meta_count[META_MAX];
3963 unsigned int segment_count[2];
3964 unsigned int block_count[2];
3965 unsigned int inplace_count;
3966 unsigned long long base_mem, cache_mem, page_mem;
3967};
3968
3969static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
3970{
3971 return (struct f2fs_stat_info *)sbi->stat_info;
3972}
3973
3974#define stat_inc_cp_call_count(sbi, foreground) \
3975 atomic_inc(&sbi->cp_call_count[(foreground)])
3976#define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++)
3977#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)
3978#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)
3979#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)
3980#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)
3981#define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type]))
3982#define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type]))
3983#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))
3984#define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type]))
3985#define stat_inc_inline_xattr(inode) \
3986 do { \
3987 if (f2fs_has_inline_xattr(inode)) \
3988 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \
3989 } while (0)
3990#define stat_dec_inline_xattr(inode) \
3991 do { \
3992 if (f2fs_has_inline_xattr(inode)) \
3993 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \
3994 } while (0)
3995#define stat_inc_inline_inode(inode) \
3996 do { \
3997 if (f2fs_has_inline_data(inode)) \
3998 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \
3999 } while (0)
4000#define stat_dec_inline_inode(inode) \
4001 do { \
4002 if (f2fs_has_inline_data(inode)) \
4003 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \
4004 } while (0)
4005#define stat_inc_inline_dir(inode) \
4006 do { \
4007 if (f2fs_has_inline_dentry(inode)) \
4008 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \
4009 } while (0)
4010#define stat_dec_inline_dir(inode) \
4011 do { \
4012 if (f2fs_has_inline_dentry(inode)) \
4013 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \
4014 } while (0)
4015#define stat_inc_compr_inode(inode) \
4016 do { \
4017 if (f2fs_compressed_file(inode)) \
4018 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \
4019 } while (0)
4020#define stat_dec_compr_inode(inode) \
4021 do { \
4022 if (f2fs_compressed_file(inode)) \
4023 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
4024 } while (0)
4025#define stat_add_compr_blocks(inode, blocks) \
4026 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
4027#define stat_sub_compr_blocks(inode, blocks) \
4028 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
4029#define stat_inc_swapfile_inode(inode) \
4030 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))
4031#define stat_dec_swapfile_inode(inode) \
4032 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))
4033#define stat_inc_atomic_inode(inode) \
4034 (atomic_inc(&F2FS_I_SB(inode)->atomic_files))
4035#define stat_dec_atomic_inode(inode) \
4036 (atomic_dec(&F2FS_I_SB(inode)->atomic_files))
4037#define stat_inc_meta_count(sbi, blkaddr) \
4038 do { \
4039 if (blkaddr < SIT_I(sbi)->sit_base_addr) \
4040 atomic_inc(&(sbi)->meta_count[META_CP]); \
4041 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \
4042 atomic_inc(&(sbi)->meta_count[META_SIT]); \
4043 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \
4044 atomic_inc(&(sbi)->meta_count[META_NAT]); \
4045 else if (blkaddr < SM_I(sbi)->main_blkaddr) \
4046 atomic_inc(&(sbi)->meta_count[META_SSA]); \
4047 } while (0)
4048#define stat_inc_seg_type(sbi, curseg) \
4049 ((sbi)->segment_count[(curseg)->alloc_type]++)
4050#define stat_inc_block_count(sbi, curseg) \
4051 ((sbi)->block_count[(curseg)->alloc_type]++)
4052#define stat_inc_inplace_blocks(sbi) \
4053 (atomic_inc(&(sbi)->inplace_count))
4054#define stat_update_max_atomic_write(inode) \
4055 do { \
4056 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \
4057 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \
4058 if (cur > max) \
4059 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \
4060 } while (0)
4061#define stat_inc_gc_call_count(sbi, foreground) \
4062 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++)
4063#define stat_inc_gc_sec_count(sbi, type, gc_type) \
4064 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++)
4065#define stat_inc_gc_seg_count(sbi, type, gc_type) \
4066 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++)
4067
4068#define stat_inc_tot_blk_count(si, blks) \
4069 ((si)->tot_blks += (blks))
4070
4071#define stat_inc_data_blk_count(sbi, blks, gc_type) \
4072 do { \
4073 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4074 stat_inc_tot_blk_count(si, blks); \
4075 si->data_blks += (blks); \
4076 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4077 } while (0)
4078
4079#define stat_inc_node_blk_count(sbi, blks, gc_type) \
4080 do { \
4081 struct f2fs_stat_info *si = F2FS_STAT(sbi); \
4082 stat_inc_tot_blk_count(si, blks); \
4083 si->node_blks += (blks); \
4084 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \
4085 } while (0)
4086
4087int f2fs_build_stats(struct f2fs_sb_info *sbi);
4088void f2fs_destroy_stats(struct f2fs_sb_info *sbi);
4089void __init f2fs_create_root_stats(void);
4090void f2fs_destroy_root_stats(void);
4091void f2fs_update_sit_info(struct f2fs_sb_info *sbi);
4092#else
4093#define stat_inc_cp_call_count(sbi, foreground) do { } while (0)
4094#define stat_inc_cp_count(sbi) do { } while (0)
4095#define stat_io_skip_bggc_count(sbi) do { } while (0)
4096#define stat_other_skip_bggc_count(sbi) do { } while (0)
4097#define stat_inc_dirty_inode(sbi, type) do { } while (0)
4098#define stat_dec_dirty_inode(sbi, type) do { } while (0)
4099#define stat_inc_total_hit(sbi, type) do { } while (0)
4100#define stat_inc_rbtree_node_hit(sbi, type) do { } while (0)
4101#define stat_inc_largest_node_hit(sbi) do { } while (0)
4102#define stat_inc_cached_node_hit(sbi, type) do { } while (0)
4103#define stat_inc_inline_xattr(inode) do { } while (0)
4104#define stat_dec_inline_xattr(inode) do { } while (0)
4105#define stat_inc_inline_inode(inode) do { } while (0)
4106#define stat_dec_inline_inode(inode) do { } while (0)
4107#define stat_inc_inline_dir(inode) do { } while (0)
4108#define stat_dec_inline_dir(inode) do { } while (0)
4109#define stat_inc_compr_inode(inode) do { } while (0)
4110#define stat_dec_compr_inode(inode) do { } while (0)
4111#define stat_add_compr_blocks(inode, blocks) do { } while (0)
4112#define stat_sub_compr_blocks(inode, blocks) do { } while (0)
4113#define stat_inc_swapfile_inode(inode) do { } while (0)
4114#define stat_dec_swapfile_inode(inode) do { } while (0)
4115#define stat_inc_atomic_inode(inode) do { } while (0)
4116#define stat_dec_atomic_inode(inode) do { } while (0)
4117#define stat_update_max_atomic_write(inode) do { } while (0)
4118#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)
4119#define stat_inc_seg_type(sbi, curseg) do { } while (0)
4120#define stat_inc_block_count(sbi, curseg) do { } while (0)
4121#define stat_inc_inplace_blocks(sbi) do { } while (0)
4122#define stat_inc_gc_call_count(sbi, foreground) do { } while (0)
4123#define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0)
4124#define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0)
4125#define stat_inc_tot_blk_count(si, blks) do { } while (0)
4126#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)
4127#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)
4128
4129static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }
4130static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }
4131static inline void __init f2fs_create_root_stats(void) { }
4132static inline void f2fs_destroy_root_stats(void) { }
4133static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}
4134#endif
4135
4136extern const struct file_operations f2fs_dir_operations;
4137extern const struct file_operations f2fs_file_operations;
4138extern const struct inode_operations f2fs_file_inode_operations;
4139extern const struct address_space_operations f2fs_dblock_aops;
4140extern const struct address_space_operations f2fs_node_aops;
4141extern const struct address_space_operations f2fs_meta_aops;
4142extern const struct inode_operations f2fs_dir_inode_operations;
4143extern const struct inode_operations f2fs_symlink_inode_operations;
4144extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;
4145extern const struct inode_operations f2fs_special_inode_operations;
4146extern struct kmem_cache *f2fs_inode_entry_slab;
4147
4148/*
4149 * inline.c
4150 */
4151bool f2fs_may_inline_data(struct inode *inode);
4152bool f2fs_sanity_check_inline_data(struct inode *inode);
4153bool f2fs_may_inline_dentry(struct inode *inode);
4154void f2fs_do_read_inline_data(struct folio *folio, struct page *ipage);
4155void f2fs_truncate_inline_inode(struct inode *inode,
4156 struct page *ipage, u64 from);
4157int f2fs_read_inline_data(struct inode *inode, struct folio *folio);
4158int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);
4159int f2fs_convert_inline_inode(struct inode *inode);
4160int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);
4161int f2fs_write_inline_data(struct inode *inode, struct page *page);
4162int f2fs_recover_inline_data(struct inode *inode, struct page *npage);
4163struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,
4164 const struct f2fs_filename *fname,
4165 struct page **res_page);
4166int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,
4167 struct page *ipage);
4168int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,
4169 struct inode *inode, nid_t ino, umode_t mode);
4170void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,
4171 struct page *page, struct inode *dir,
4172 struct inode *inode);
4173bool f2fs_empty_inline_dir(struct inode *dir);
4174int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,
4175 struct fscrypt_str *fstr);
4176int f2fs_inline_data_fiemap(struct inode *inode,
4177 struct fiemap_extent_info *fieinfo,
4178 __u64 start, __u64 len);
4179
4180/*
4181 * shrinker.c
4182 */
4183unsigned long f2fs_shrink_count(struct shrinker *shrink,
4184 struct shrink_control *sc);
4185unsigned long f2fs_shrink_scan(struct shrinker *shrink,
4186 struct shrink_control *sc);
4187void f2fs_join_shrinker(struct f2fs_sb_info *sbi);
4188void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);
4189
4190/*
4191 * extent_cache.c
4192 */
4193bool sanity_check_extent_cache(struct inode *inode);
4194void f2fs_init_extent_tree(struct inode *inode);
4195void f2fs_drop_extent_tree(struct inode *inode);
4196void f2fs_destroy_extent_node(struct inode *inode);
4197void f2fs_destroy_extent_tree(struct inode *inode);
4198void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);
4199int __init f2fs_create_extent_cache(void);
4200void f2fs_destroy_extent_cache(void);
4201
4202/* read extent cache ops */
4203void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage);
4204bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,
4205 struct extent_info *ei);
4206bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index,
4207 block_t *blkaddr);
4208void f2fs_update_read_extent_cache(struct dnode_of_data *dn);
4209void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,
4210 pgoff_t fofs, block_t blkaddr, unsigned int len);
4211unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi,
4212 int nr_shrink);
4213
4214/* block age extent cache ops */
4215void f2fs_init_age_extent_tree(struct inode *inode);
4216bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,
4217 struct extent_info *ei);
4218void f2fs_update_age_extent_cache(struct dnode_of_data *dn);
4219void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,
4220 pgoff_t fofs, unsigned int len);
4221unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi,
4222 int nr_shrink);
4223
4224/*
4225 * sysfs.c
4226 */
4227#define MIN_RA_MUL 2
4228#define MAX_RA_MUL 256
4229
4230int __init f2fs_init_sysfs(void);
4231void f2fs_exit_sysfs(void);
4232int f2fs_register_sysfs(struct f2fs_sb_info *sbi);
4233void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);
4234
4235/* verity.c */
4236extern const struct fsverity_operations f2fs_verityops;
4237
4238/*
4239 * crypto support
4240 */
4241static inline bool f2fs_encrypted_file(struct inode *inode)
4242{
4243 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);
4244}
4245
4246static inline void f2fs_set_encrypted_inode(struct inode *inode)
4247{
4248#ifdef CONFIG_FS_ENCRYPTION
4249 file_set_encrypt(inode);
4250 f2fs_set_inode_flags(inode);
4251#endif
4252}
4253
4254/*
4255 * Returns true if the reads of the inode's data need to undergo some
4256 * postprocessing step, like decryption or authenticity verification.
4257 */
4258static inline bool f2fs_post_read_required(struct inode *inode)
4259{
4260 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
4261 f2fs_compressed_file(inode);
4262}
4263
4264/*
4265 * compress.c
4266 */
4267#ifdef CONFIG_F2FS_FS_COMPRESSION
4268bool f2fs_is_compressed_page(struct page *page);
4269struct page *f2fs_compress_control_page(struct page *page);
4270int f2fs_prepare_compress_overwrite(struct inode *inode,
4271 struct page **pagep, pgoff_t index, void **fsdata);
4272bool f2fs_compress_write_end(struct inode *inode, void *fsdata,
4273 pgoff_t index, unsigned copied);
4274int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);
4275void f2fs_compress_write_end_io(struct bio *bio, struct page *page);
4276bool f2fs_is_compress_backend_ready(struct inode *inode);
4277bool f2fs_is_compress_level_valid(int alg, int lvl);
4278int __init f2fs_init_compress_mempool(void);
4279void f2fs_destroy_compress_mempool(void);
4280void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);
4281void f2fs_end_read_compressed_page(struct page *page, bool failed,
4282 block_t blkaddr, bool in_task);
4283bool f2fs_cluster_is_empty(struct compress_ctx *cc);
4284bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);
4285bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,
4286 int index, int nr_pages, bool uptodate);
4287bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);
4288void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page);
4289int f2fs_write_multi_pages(struct compress_ctx *cc,
4290 int *submitted,
4291 struct writeback_control *wbc,
4292 enum iostat_type io_type);
4293int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);
4294void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,
4295 pgoff_t fofs, block_t blkaddr,
4296 unsigned int llen, unsigned int c_len);
4297int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
4298 unsigned nr_pages, sector_t *last_block_in_bio,
4299 bool is_readahead, bool for_write);
4300struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);
4301void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,
4302 bool in_task);
4303void f2fs_put_page_dic(struct page *page, bool in_task);
4304unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn,
4305 unsigned int ofs_in_node);
4306int f2fs_init_compress_ctx(struct compress_ctx *cc);
4307void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);
4308void f2fs_init_compress_info(struct f2fs_sb_info *sbi);
4309int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);
4310void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);
4311int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);
4312void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);
4313int __init f2fs_init_compress_cache(void);
4314void f2fs_destroy_compress_cache(void);
4315struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);
4316void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);
4317void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4318 nid_t ino, block_t blkaddr);
4319bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,
4320 block_t blkaddr);
4321void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);
4322#define inc_compr_inode_stat(inode) \
4323 do { \
4324 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4325 sbi->compr_new_inode++; \
4326 } while (0)
4327#define add_compr_block_stat(inode, blocks) \
4328 do { \
4329 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \
4330 int diff = F2FS_I(inode)->i_cluster_size - blocks; \
4331 sbi->compr_written_block += blocks; \
4332 sbi->compr_saved_block += diff; \
4333 } while (0)
4334#else
4335static inline bool f2fs_is_compressed_page(struct page *page) { return false; }
4336static inline bool f2fs_is_compress_backend_ready(struct inode *inode)
4337{
4338 if (!f2fs_compressed_file(inode))
4339 return true;
4340 /* not support compression */
4341 return false;
4342}
4343static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; }
4344static inline struct page *f2fs_compress_control_page(struct page *page)
4345{
4346 WARN_ON_ONCE(1);
4347 return ERR_PTR(-EINVAL);
4348}
4349static inline int __init f2fs_init_compress_mempool(void) { return 0; }
4350static inline void f2fs_destroy_compress_mempool(void) { }
4351static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,
4352 bool in_task) { }
4353static inline void f2fs_end_read_compressed_page(struct page *page,
4354 bool failed, block_t blkaddr, bool in_task)
4355{
4356 WARN_ON_ONCE(1);
4357}
4358static inline void f2fs_put_page_dic(struct page *page, bool in_task)
4359{
4360 WARN_ON_ONCE(1);
4361}
4362static inline unsigned int f2fs_cluster_blocks_are_contiguous(
4363 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; }
4364static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }
4365static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }
4366static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }
4367static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }
4368static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }
4369static inline int __init f2fs_init_compress_cache(void) { return 0; }
4370static inline void f2fs_destroy_compress_cache(void) { }
4371static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,
4372 block_t blkaddr) { }
4373static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,
4374 struct page *page, nid_t ino, block_t blkaddr) { }
4375static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,
4376 struct page *page, block_t blkaddr) { return false; }
4377static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,
4378 nid_t ino) { }
4379#define inc_compr_inode_stat(inode) do { } while (0)
4380static inline void f2fs_update_read_extent_tree_range_compressed(
4381 struct inode *inode,
4382 pgoff_t fofs, block_t blkaddr,
4383 unsigned int llen, unsigned int c_len) { }
4384#endif
4385
4386static inline int set_compress_context(struct inode *inode)
4387{
4388#ifdef CONFIG_F2FS_FS_COMPRESSION
4389 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
4390
4391 F2FS_I(inode)->i_compress_algorithm =
4392 F2FS_OPTION(sbi).compress_algorithm;
4393 F2FS_I(inode)->i_log_cluster_size =
4394 F2FS_OPTION(sbi).compress_log_size;
4395 F2FS_I(inode)->i_compress_flag =
4396 F2FS_OPTION(sbi).compress_chksum ?
4397 BIT(COMPRESS_CHKSUM) : 0;
4398 F2FS_I(inode)->i_cluster_size =
4399 BIT(F2FS_I(inode)->i_log_cluster_size);
4400 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 ||
4401 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) &&
4402 F2FS_OPTION(sbi).compress_level)
4403 F2FS_I(inode)->i_compress_level =
4404 F2FS_OPTION(sbi).compress_level;
4405 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL;
4406 set_inode_flag(inode, FI_COMPRESSED_FILE);
4407 stat_inc_compr_inode(inode);
4408 inc_compr_inode_stat(inode);
4409 f2fs_mark_inode_dirty_sync(inode, true);
4410 return 0;
4411#else
4412 return -EOPNOTSUPP;
4413#endif
4414}
4415
4416static inline bool f2fs_disable_compressed_file(struct inode *inode)
4417{
4418 struct f2fs_inode_info *fi = F2FS_I(inode);
4419
4420 f2fs_down_write(&F2FS_I(inode)->i_sem);
4421
4422 if (!f2fs_compressed_file(inode)) {
4423 f2fs_up_write(&F2FS_I(inode)->i_sem);
4424 return true;
4425 }
4426 if (f2fs_is_mmap_file(inode) ||
4427 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) {
4428 f2fs_up_write(&F2FS_I(inode)->i_sem);
4429 return false;
4430 }
4431
4432 fi->i_flags &= ~F2FS_COMPR_FL;
4433 stat_dec_compr_inode(inode);
4434 clear_inode_flag(inode, FI_COMPRESSED_FILE);
4435 f2fs_mark_inode_dirty_sync(inode, true);
4436
4437 f2fs_up_write(&F2FS_I(inode)->i_sem);
4438 return true;
4439}
4440
4441#define F2FS_FEATURE_FUNCS(name, flagname) \
4442static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \
4443{ \
4444 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \
4445}
4446
4447F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);
4448F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);
4449F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);
4450F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);
4451F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);
4452F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);
4453F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);
4454F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);
4455F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);
4456F2FS_FEATURE_FUNCS(verity, VERITY);
4457F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);
4458F2FS_FEATURE_FUNCS(casefold, CASEFOLD);
4459F2FS_FEATURE_FUNCS(compression, COMPRESSION);
4460F2FS_FEATURE_FUNCS(readonly, RO);
4461
4462#ifdef CONFIG_BLK_DEV_ZONED
4463static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,
4464 block_t blkaddr)
4465{
4466 unsigned int zno = blkaddr / sbi->blocks_per_blkz;
4467
4468 return test_bit(zno, FDEV(devi).blkz_seq);
4469}
4470#endif
4471
4472static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi,
4473 struct block_device *bdev)
4474{
4475 int i;
4476
4477 if (!f2fs_is_multi_device(sbi))
4478 return 0;
4479
4480 for (i = 0; i < sbi->s_ndevs; i++)
4481 if (FDEV(i).bdev == bdev)
4482 return i;
4483
4484 WARN_ON(1);
4485 return -1;
4486}
4487
4488static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)
4489{
4490 return f2fs_sb_has_blkzoned(sbi);
4491}
4492
4493static inline bool f2fs_bdev_support_discard(struct block_device *bdev)
4494{
4495 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);
4496}
4497
4498static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)
4499{
4500 int i;
4501
4502 if (!f2fs_is_multi_device(sbi))
4503 return f2fs_bdev_support_discard(sbi->sb->s_bdev);
4504
4505 for (i = 0; i < sbi->s_ndevs; i++)
4506 if (f2fs_bdev_support_discard(FDEV(i).bdev))
4507 return true;
4508 return false;
4509}
4510
4511static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)
4512{
4513 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||
4514 f2fs_hw_should_discard(sbi);
4515}
4516
4517static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)
4518{
4519 int i;
4520
4521 if (!f2fs_is_multi_device(sbi))
4522 return bdev_read_only(sbi->sb->s_bdev);
4523
4524 for (i = 0; i < sbi->s_ndevs; i++)
4525 if (bdev_read_only(FDEV(i).bdev))
4526 return true;
4527 return false;
4528}
4529
4530static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi)
4531{
4532 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi);
4533}
4534
4535static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)
4536{
4537 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;
4538}
4539
4540static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi,
4541 block_t blkaddr)
4542{
4543 if (f2fs_sb_has_blkzoned(sbi)) {
4544 int devi = f2fs_target_device_index(sbi, blkaddr);
4545
4546 return !bdev_is_zoned(FDEV(devi).bdev);
4547 }
4548 return true;
4549}
4550
4551static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)
4552{
4553 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;
4554}
4555
4556static inline bool f2fs_may_compress(struct inode *inode)
4557{
4558 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||
4559 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) ||
4560 f2fs_is_mmap_file(inode))
4561 return false;
4562 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);
4563}
4564
4565static inline void f2fs_i_compr_blocks_update(struct inode *inode,
4566 u64 blocks, bool add)
4567{
4568 struct f2fs_inode_info *fi = F2FS_I(inode);
4569 int diff = fi->i_cluster_size - blocks;
4570
4571 /* don't update i_compr_blocks if saved blocks were released */
4572 if (!add && !atomic_read(&fi->i_compr_blocks))
4573 return;
4574
4575 if (add) {
4576 atomic_add(diff, &fi->i_compr_blocks);
4577 stat_add_compr_blocks(inode, diff);
4578 } else {
4579 atomic_sub(diff, &fi->i_compr_blocks);
4580 stat_sub_compr_blocks(inode, diff);
4581 }
4582 f2fs_mark_inode_dirty_sync(inode, true);
4583}
4584
4585static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,
4586 int flag)
4587{
4588 if (!f2fs_is_multi_device(sbi))
4589 return false;
4590 if (flag != F2FS_GET_BLOCK_DIO)
4591 return false;
4592 return sbi->aligned_blksize;
4593}
4594
4595static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)
4596{
4597 return fsverity_active(inode) &&
4598 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
4599}
4600
4601#ifdef CONFIG_F2FS_FAULT_INJECTION
4602extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,
4603 unsigned long type);
4604#else
4605static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi,
4606 unsigned long rate, unsigned long type)
4607{
4608 return 0;
4609}
4610#endif
4611
4612static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)
4613{
4614#ifdef CONFIG_QUOTA
4615 if (f2fs_sb_has_quota_ino(sbi))
4616 return true;
4617 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
4618 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
4619 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4620 return true;
4621#endif
4622 return false;
4623}
4624
4625static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)
4626{
4627 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;
4628}
4629
4630static inline void f2fs_io_schedule_timeout(long timeout)
4631{
4632 set_current_state(TASK_UNINTERRUPTIBLE);
4633 io_schedule_timeout(timeout);
4634}
4635
4636static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs,
4637 enum page_type type)
4638{
4639 if (unlikely(f2fs_cp_error(sbi)))
4640 return;
4641
4642 if (ofs == sbi->page_eio_ofs[type]) {
4643 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)
4644 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4645 } else {
4646 sbi->page_eio_ofs[type] = ofs;
4647 sbi->page_eio_cnt[type] = 0;
4648 }
4649}
4650
4651static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi)
4652{
4653 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb);
4654}
4655
4656static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi,
4657 block_t blkaddr, unsigned int cnt)
4658{
4659 bool need_submit = false;
4660 int i = 0;
4661
4662 do {
4663 struct page *page;
4664
4665 page = find_get_page(META_MAPPING(sbi), blkaddr + i);
4666 if (page) {
4667 if (folio_test_writeback(page_folio(page)))
4668 need_submit = true;
4669 f2fs_put_page(page, 0);
4670 }
4671 } while (++i < cnt && !need_submit);
4672
4673 if (need_submit)
4674 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode,
4675 NULL, 0, DATA);
4676
4677 truncate_inode_pages_range(META_MAPPING(sbi),
4678 F2FS_BLK_TO_BYTES((loff_t)blkaddr),
4679 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1)));
4680}
4681
4682static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi,
4683 block_t blkaddr)
4684{
4685 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1);
4686 f2fs_invalidate_compress_page(sbi, blkaddr);
4687}
4688
4689#define EFSBADCRC EBADMSG /* Bad CRC detected */
4690#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
4691
4692#endif /* _LINUX_F2FS_H */