4747 lines · c
1/* SPDX-License-Identifier: GPL-2.0 */2/*3 * fs/f2fs/f2fs.h4 *5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.6 * http://www.samsung.com/7 */8#ifndef _LINUX_F2FS_H9#define _LINUX_F2FS_H10 11#include <linux/uio.h>12#include <linux/types.h>13#include <linux/page-flags.h>14#include <linux/slab.h>15#include <linux/crc32.h>16#include <linux/magic.h>17#include <linux/kobject.h>18#include <linux/sched.h>19#include <linux/cred.h>20#include <linux/sched/mm.h>21#include <linux/vmalloc.h>22#include <linux/bio.h>23#include <linux/blkdev.h>24#include <linux/quotaops.h>25#include <linux/part_stat.h>26#include <linux/rw_hint.h>27#include <crypto/hash.h>28 29#include <linux/fscrypt.h>30#include <linux/fsverity.h>31 32struct pagevec;33 34#ifdef CONFIG_F2FS_CHECK_FS35#define f2fs_bug_on(sbi, condition) BUG_ON(condition)36#else37#define f2fs_bug_on(sbi, condition) \38 do { \39 if (WARN_ON(condition)) \40 set_sbi_flag(sbi, SBI_NEED_FSCK); \41 } while (0)42#endif43 44enum {45 FAULT_KMALLOC,46 FAULT_KVMALLOC,47 FAULT_PAGE_ALLOC,48 FAULT_PAGE_GET,49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */50 FAULT_ALLOC_NID,51 FAULT_ORPHAN,52 FAULT_BLOCK,53 FAULT_DIR_DEPTH,54 FAULT_EVICT_INODE,55 FAULT_TRUNCATE,56 FAULT_READ_IO,57 FAULT_CHECKPOINT,58 FAULT_DISCARD,59 FAULT_WRITE_IO,60 FAULT_SLAB_ALLOC,61 FAULT_DQUOT_INIT,62 FAULT_LOCK_OP,63 FAULT_BLKADDR_VALIDITY,64 FAULT_BLKADDR_CONSISTENCE,65 FAULT_NO_SEGMENT,66 FAULT_MAX,67};68 69#ifdef CONFIG_F2FS_FAULT_INJECTION70#define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0))71 72struct f2fs_fault_info {73 atomic_t inject_ops;74 int inject_rate;75 unsigned int inject_type;76};77 78extern const char *f2fs_fault_name[FAULT_MAX];79#define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))80 81/* maximum retry count for injected failure */82#define DEFAULT_FAILURE_RETRY_COUNT 883#else84#define DEFAULT_FAILURE_RETRY_COUNT 185#endif86 87/*88 * For mount options89 */90#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x0000000191#define F2FS_MOUNT_DISCARD 0x0000000292#define F2FS_MOUNT_NOHEAP 0x0000000493#define F2FS_MOUNT_XATTR_USER 0x0000000894#define F2FS_MOUNT_POSIX_ACL 0x0000001095#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x0000002096#define F2FS_MOUNT_INLINE_XATTR 0x0000004097#define F2FS_MOUNT_INLINE_DATA 0x0000008098#define F2FS_MOUNT_INLINE_DENTRY 0x0000010099#define F2FS_MOUNT_FLUSH_MERGE 0x00000200100#define F2FS_MOUNT_NOBARRIER 0x00000400101#define F2FS_MOUNT_FASTBOOT 0x00000800102#define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000103#define F2FS_MOUNT_DATA_FLUSH 0x00002000104#define F2FS_MOUNT_FAULT_INJECTION 0x00004000105#define F2FS_MOUNT_USRQUOTA 0x00008000106#define F2FS_MOUNT_GRPQUOTA 0x00010000107#define F2FS_MOUNT_PRJQUOTA 0x00020000108#define F2FS_MOUNT_QUOTA 0x00040000109#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000110#define F2FS_MOUNT_RESERVE_ROOT 0x00100000111#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000112#define F2FS_MOUNT_NORECOVERY 0x00400000113#define F2FS_MOUNT_ATGC 0x00800000114#define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000115#define F2FS_MOUNT_GC_MERGE 0x02000000116#define F2FS_MOUNT_COMPRESS_CACHE 0x04000000117#define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000118 119#define F2FS_OPTION(sbi) ((sbi)->mount_opt)120#define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option)121#define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option)122#define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option)123 124#define ver_after(a, b) (typecheck(unsigned long long, a) && \125 typecheck(unsigned long long, b) && \126 ((long long)((a) - (b)) > 0))127 128typedef u32 block_t; /*129 * should not change u32, since it is the on-disk block130 * address format, __le32.131 */132typedef u32 nid_t;133 134#define COMPRESS_EXT_NUM 16135 136enum blkzone_allocation_policy {137 BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */138 BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */139 BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */140};141 142/*143 * An implementation of an rwsem that is explicitly unfair to readers. This144 * prevents priority inversion when a low-priority reader acquires the read lock145 * while sleeping on the write lock but the write lock is needed by146 * higher-priority clients.147 */148 149struct f2fs_rwsem {150 struct rw_semaphore internal_rwsem;151#ifdef CONFIG_F2FS_UNFAIR_RWSEM152 wait_queue_head_t read_waiters;153#endif154};155 156struct f2fs_mount_info {157 unsigned int opt;158 block_t root_reserved_blocks; /* root reserved blocks */159 kuid_t s_resuid; /* reserved blocks for uid */160 kgid_t s_resgid; /* reserved blocks for gid */161 int active_logs; /* # of active logs */162 int inline_xattr_size; /* inline xattr size */163#ifdef CONFIG_F2FS_FAULT_INJECTION164 struct f2fs_fault_info fault_info; /* For fault injection */165#endif166#ifdef CONFIG_QUOTA167 /* Names of quota files with journalled quota */168 char *s_qf_names[MAXQUOTAS];169 int s_jquota_fmt; /* Format of quota to use */170#endif171 /* For which write hints are passed down to block layer */172 int alloc_mode; /* segment allocation policy */173 int fsync_mode; /* fsync policy */174 int fs_mode; /* fs mode: LFS or ADAPTIVE */175 int bggc_mode; /* bggc mode: off, on or sync */176 int memory_mode; /* memory mode */177 int errors; /* errors parameter */178 int discard_unit; /*179 * discard command's offset/size should180 * be aligned to this unit: block,181 * segment or section182 */183 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */184 block_t unusable_cap_perc; /* percentage for cap */185 block_t unusable_cap; /* Amount of space allowed to be186 * unusable when disabling checkpoint187 */188 189 /* For compression */190 unsigned char compress_algorithm; /* algorithm type */191 unsigned char compress_log_size; /* cluster log size */192 unsigned char compress_level; /* compress level */193 bool compress_chksum; /* compressed data chksum */194 unsigned char compress_ext_cnt; /* extension count */195 unsigned char nocompress_ext_cnt; /* nocompress extension count */196 int compress_mode; /* compression mode */197 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */198 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */199};200 201#define F2FS_FEATURE_ENCRYPT 0x00000001202#define F2FS_FEATURE_BLKZONED 0x00000002203#define F2FS_FEATURE_ATOMIC_WRITE 0x00000004204#define F2FS_FEATURE_EXTRA_ATTR 0x00000008205#define F2FS_FEATURE_PRJQUOTA 0x00000010206#define F2FS_FEATURE_INODE_CHKSUM 0x00000020207#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040208#define F2FS_FEATURE_QUOTA_INO 0x00000080209#define F2FS_FEATURE_INODE_CRTIME 0x00000100210#define F2FS_FEATURE_LOST_FOUND 0x00000200211#define F2FS_FEATURE_VERITY 0x00000400212#define F2FS_FEATURE_SB_CHKSUM 0x00000800213#define F2FS_FEATURE_CASEFOLD 0x00001000214#define F2FS_FEATURE_COMPRESSION 0x00002000215#define F2FS_FEATURE_RO 0x00004000216 217#define __F2FS_HAS_FEATURE(raw_super, mask) \218 ((raw_super->feature & cpu_to_le32(mask)) != 0)219#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)220 221/*222 * Default values for user and/or group using reserved blocks223 */224#define F2FS_DEF_RESUID 0225#define F2FS_DEF_RESGID 0226 227/*228 * For checkpoint manager229 */230enum {231 NAT_BITMAP,232 SIT_BITMAP233};234 235#define CP_UMOUNT 0x00000001236#define CP_FASTBOOT 0x00000002237#define CP_SYNC 0x00000004238#define CP_RECOVERY 0x00000008239#define CP_DISCARD 0x00000010240#define CP_TRIMMED 0x00000020241#define CP_PAUSE 0x00000040242#define CP_RESIZE 0x00000080243 244#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */245#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */246#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */247#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */248#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */249#define DEF_CP_INTERVAL 60 /* 60 secs */250#define DEF_IDLE_INTERVAL 5 /* 5 secs */251#define DEF_DISABLE_INTERVAL 5 /* 5 secs */252#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */253#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */254 255struct cp_control {256 int reason;257 __u64 trim_start;258 __u64 trim_end;259 __u64 trim_minlen;260};261 262/*263 * indicate meta/data type264 */265enum {266 META_CP,267 META_NAT,268 META_SIT,269 META_SSA,270 META_MAX,271 META_POR,272 DATA_GENERIC, /* check range only */273 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */274 DATA_GENERIC_ENHANCE_READ, /*275 * strong check on range and segment276 * bitmap but no warning due to race277 * condition of read on truncated area278 * by extent_cache279 */280 DATA_GENERIC_ENHANCE_UPDATE, /*281 * strong check on range and segment282 * bitmap for update case283 */284 META_GENERIC,285};286 287/* for the list of ino */288enum {289 ORPHAN_INO, /* for orphan ino list */290 APPEND_INO, /* for append ino list */291 UPDATE_INO, /* for update ino list */292 TRANS_DIR_INO, /* for transactions dir ino list */293 XATTR_DIR_INO, /* for xattr updated dir ino list */294 FLUSH_INO, /* for multiple device flushing */295 MAX_INO_ENTRY, /* max. list */296};297 298struct ino_entry {299 struct list_head list; /* list head */300 nid_t ino; /* inode number */301 unsigned int dirty_device; /* dirty device bitmap */302};303 304/* for the list of inodes to be GCed */305struct inode_entry {306 struct list_head list; /* list head */307 struct inode *inode; /* vfs inode pointer */308};309 310struct fsync_node_entry {311 struct list_head list; /* list head */312 struct page *page; /* warm node page pointer */313 unsigned int seq_id; /* sequence id */314};315 316struct ckpt_req {317 struct completion wait; /* completion for checkpoint done */318 struct llist_node llnode; /* llist_node to be linked in wait queue */319 int ret; /* return code of checkpoint */320 ktime_t queue_time; /* request queued time */321};322 323struct ckpt_req_control {324 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */325 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */326 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */327 atomic_t issued_ckpt; /* # of actually issued ckpts */328 atomic_t total_ckpt; /* # of total ckpts */329 atomic_t queued_ckpt; /* # of queued ckpts */330 struct llist_head issue_list; /* list for command issue */331 spinlock_t stat_lock; /* lock for below checkpoint time stats */332 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */333 unsigned int peak_time; /* peak wait time in msec until now */334};335 336/* for the bitmap indicate blocks to be discarded */337struct discard_entry {338 struct list_head list; /* list head */339 block_t start_blkaddr; /* start blockaddr of current segment */340 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */341};342 343/* minimum discard granularity, unit: block count */344#define MIN_DISCARD_GRANULARITY 1345/* default discard granularity of inner discard thread, unit: block count */346#define DEFAULT_DISCARD_GRANULARITY 16347/* default maximum discard granularity of ordered discard, unit: block count */348#define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16349 350/* max discard pend list number */351#define MAX_PLIST_NUM 512352#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \353 (MAX_PLIST_NUM - 1) : ((blk_num) - 1))354 355enum {356 D_PREP, /* initial */357 D_PARTIAL, /* partially submitted */358 D_SUBMIT, /* all submitted */359 D_DONE, /* finished */360};361 362struct discard_info {363 block_t lstart; /* logical start address */364 block_t len; /* length */365 block_t start; /* actual start address in dev */366};367 368struct discard_cmd {369 struct rb_node rb_node; /* rb node located in rb-tree */370 struct discard_info di; /* discard info */371 struct list_head list; /* command list */372 struct completion wait; /* compleation */373 struct block_device *bdev; /* bdev */374 unsigned short ref; /* reference count */375 unsigned char state; /* state */376 unsigned char queued; /* queued discard */377 int error; /* bio error */378 spinlock_t lock; /* for state/bio_ref updating */379 unsigned short bio_ref; /* bio reference count */380};381 382enum {383 DPOLICY_BG,384 DPOLICY_FORCE,385 DPOLICY_FSTRIM,386 DPOLICY_UMOUNT,387 MAX_DPOLICY,388};389 390enum {391 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */392 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */393 DPOLICY_IO_AWARE_MAX,394};395 396struct discard_policy {397 int type; /* type of discard */398 unsigned int min_interval; /* used for candidates exist */399 unsigned int mid_interval; /* used for device busy */400 unsigned int max_interval; /* used for candidates not exist */401 unsigned int max_requests; /* # of discards issued per round */402 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */403 bool io_aware; /* issue discard in idle time */404 bool sync; /* submit discard with REQ_SYNC flag */405 bool ordered; /* issue discard by lba order */406 bool timeout; /* discard timeout for put_super */407 unsigned int granularity; /* discard granularity */408};409 410struct discard_cmd_control {411 struct task_struct *f2fs_issue_discard; /* discard thread */412 struct list_head entry_list; /* 4KB discard entry list */413 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */414 struct list_head wait_list; /* store on-flushing entries */415 struct list_head fstrim_list; /* in-flight discard from fstrim */416 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */417 struct mutex cmd_lock;418 unsigned int nr_discards; /* # of discards in the list */419 unsigned int max_discards; /* max. discards to be issued */420 unsigned int max_discard_request; /* max. discard request per round */421 unsigned int min_discard_issue_time; /* min. interval between discard issue */422 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */423 unsigned int max_discard_issue_time; /* max. interval between discard issue */424 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */425 unsigned int discard_urgent_util; /* utilization which issue discard proactively */426 unsigned int discard_granularity; /* discard granularity */427 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */428 unsigned int discard_io_aware; /* io_aware policy */429 unsigned int undiscard_blks; /* # of undiscard blocks */430 unsigned int next_pos; /* next discard position */431 atomic_t issued_discard; /* # of issued discard */432 atomic_t queued_discard; /* # of queued discard */433 atomic_t discard_cmd_cnt; /* # of cached cmd count */434 struct rb_root_cached root; /* root of discard rb-tree */435 bool rbtree_check; /* config for consistence check */436 bool discard_wake; /* to wake up discard thread */437};438 439/* for the list of fsync inodes, used only during recovery */440struct fsync_inode_entry {441 struct list_head list; /* list head */442 struct inode *inode; /* vfs inode pointer */443 block_t blkaddr; /* block address locating the last fsync */444 block_t last_dentry; /* block address locating the last dentry */445};446 447#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats))448#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits))449 450#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne)451#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid)452#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se)453#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno)454 455#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl))456#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl))457 458static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i)459{460 int before = nats_in_cursum(journal);461 462 journal->n_nats = cpu_to_le16(before + i);463 return before;464}465 466static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i)467{468 int before = sits_in_cursum(journal);469 470 journal->n_sits = cpu_to_le16(before + i);471 return before;472}473 474static inline bool __has_cursum_space(struct f2fs_journal *journal,475 int size, int type)476{477 if (type == NAT_JOURNAL)478 return size <= MAX_NAT_JENTRIES(journal);479 return size <= MAX_SIT_JENTRIES(journal);480}481 482/* for inline stuff */483#define DEF_INLINE_RESERVED_SIZE 1484static inline int get_extra_isize(struct inode *inode);485static inline int get_inline_xattr_addrs(struct inode *inode);486#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \487 (CUR_ADDRS_PER_INODE(inode) - \488 get_inline_xattr_addrs(inode) - \489 DEF_INLINE_RESERVED_SIZE))490 491/* for inline dir */492#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \493 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \494 BITS_PER_BYTE + 1))495#define INLINE_DENTRY_BITMAP_SIZE(inode) \496 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE)497#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \498 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \499 NR_INLINE_DENTRY(inode) + \500 INLINE_DENTRY_BITMAP_SIZE(inode)))501 502/*503 * For INODE and NODE manager504 */505/* for directory operations */506 507struct f2fs_filename {508 /*509 * The filename the user specified. This is NULL for some510 * filesystem-internal operations, e.g. converting an inline directory511 * to a non-inline one, or roll-forward recovering an encrypted dentry.512 */513 const struct qstr *usr_fname;514 515 /*516 * The on-disk filename. For encrypted directories, this is encrypted.517 * This may be NULL for lookups in an encrypted dir without the key.518 */519 struct fscrypt_str disk_name;520 521 /* The dirhash of this filename */522 f2fs_hash_t hash;523 524#ifdef CONFIG_FS_ENCRYPTION525 /*526 * For lookups in encrypted directories: either the buffer backing527 * disk_name, or a buffer that holds the decoded no-key name.528 */529 struct fscrypt_str crypto_buf;530#endif531#if IS_ENABLED(CONFIG_UNICODE)532 /*533 * For casefolded directories: the casefolded name, but it's left NULL534 * if the original name is not valid Unicode, if the original name is535 * "." or "..", if the directory is both casefolded and encrypted and536 * its encryption key is unavailable, or if the filesystem is doing an537 * internal operation where usr_fname is also NULL. In all these cases538 * we fall back to treating the name as an opaque byte sequence.539 */540 struct qstr cf_name;541#endif542};543 544struct f2fs_dentry_ptr {545 struct inode *inode;546 void *bitmap;547 struct f2fs_dir_entry *dentry;548 __u8 (*filename)[F2FS_SLOT_LEN];549 int max;550 int nr_bitmap;551};552 553static inline void make_dentry_ptr_block(struct inode *inode,554 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t)555{556 d->inode = inode;557 d->max = NR_DENTRY_IN_BLOCK;558 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP;559 d->bitmap = t->dentry_bitmap;560 d->dentry = t->dentry;561 d->filename = t->filename;562}563 564static inline void make_dentry_ptr_inline(struct inode *inode,565 struct f2fs_dentry_ptr *d, void *t)566{567 int entry_cnt = NR_INLINE_DENTRY(inode);568 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode);569 int reserved_size = INLINE_RESERVED_SIZE(inode);570 571 d->inode = inode;572 d->max = entry_cnt;573 d->nr_bitmap = bitmap_size;574 d->bitmap = t;575 d->dentry = t + bitmap_size + reserved_size;576 d->filename = t + bitmap_size + reserved_size +577 SIZE_OF_DIR_ENTRY * entry_cnt;578}579 580/*581 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1582 * as its node offset to distinguish from index node blocks.583 * But some bits are used to mark the node block.584 */585#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \586 >> OFFSET_BIT_SHIFT)587enum {588 ALLOC_NODE, /* allocate a new node page if needed */589 LOOKUP_NODE, /* look up a node without readahead */590 LOOKUP_NODE_RA, /*591 * look up a node with readahead called592 * by get_data_block.593 */594};595 596#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */597 598/* congestion wait timeout value, default: 20ms */599#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20))600 601/* maximum retry quota flush count */602#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8603 604/* maximum retry of EIO'ed page */605#define MAX_RETRY_PAGE_EIO 100606 607#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */608 609#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */610 611/* dirty segments threshold for triggering CP */612#define DEFAULT_DIRTY_THRESHOLD 4613 614#define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS615#define RECOVERY_MIN_RA_BLOCKS 1616 617#define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */618 619/* for in-memory extent cache entry */620#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */621 622/* number of extent info in extent cache we try to shrink */623#define READ_EXTENT_CACHE_SHRINK_NUMBER 128624 625/* number of age extent info in extent cache we try to shrink */626#define AGE_EXTENT_CACHE_SHRINK_NUMBER 128627#define LAST_AGE_WEIGHT 30628#define SAME_AGE_REGION 1024629 630/*631 * Define data block with age less than 1GB as hot data632 * define data block with age less than 10GB but more than 1GB as warm data633 */634#define DEF_HOT_DATA_AGE_THRESHOLD 262144635#define DEF_WARM_DATA_AGE_THRESHOLD 2621440636 637/* extent cache type */638enum extent_type {639 EX_READ,640 EX_BLOCK_AGE,641 NR_EXTENT_CACHES,642};643 644struct extent_info {645 unsigned int fofs; /* start offset in a file */646 unsigned int len; /* length of the extent */647 union {648 /* read extent_cache */649 struct {650 /* start block address of the extent */651 block_t blk;652#ifdef CONFIG_F2FS_FS_COMPRESSION653 /* physical extent length of compressed blocks */654 unsigned int c_len;655#endif656 };657 /* block age extent_cache */658 struct {659 /* block age of the extent */660 unsigned long long age;661 /* last total blocks allocated */662 unsigned long long last_blocks;663 };664 };665};666 667struct extent_node {668 struct rb_node rb_node; /* rb node located in rb-tree */669 struct extent_info ei; /* extent info */670 struct list_head list; /* node in global extent list of sbi */671 struct extent_tree *et; /* extent tree pointer */672};673 674struct extent_tree {675 nid_t ino; /* inode number */676 enum extent_type type; /* keep the extent tree type */677 struct rb_root_cached root; /* root of extent info rb-tree */678 struct extent_node *cached_en; /* recently accessed extent node */679 struct list_head list; /* to be used by sbi->zombie_list */680 rwlock_t lock; /* protect extent info rb-tree */681 atomic_t node_cnt; /* # of extent node in rb-tree*/682 bool largest_updated; /* largest extent updated */683 struct extent_info largest; /* largest cached extent for EX_READ */684};685 686struct extent_tree_info {687 struct radix_tree_root extent_tree_root;/* cache extent cache entries */688 struct mutex extent_tree_lock; /* locking extent radix tree */689 struct list_head extent_list; /* lru list for shrinker */690 spinlock_t extent_lock; /* locking extent lru list */691 atomic_t total_ext_tree; /* extent tree count */692 struct list_head zombie_list; /* extent zombie tree list */693 atomic_t total_zombie_tree; /* extent zombie tree count */694 atomic_t total_ext_node; /* extent info count */695};696 697/*698 * State of block returned by f2fs_map_blocks.699 */700#define F2FS_MAP_NEW (1U << 0)701#define F2FS_MAP_MAPPED (1U << 1)702#define F2FS_MAP_DELALLOC (1U << 2)703#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\704 F2FS_MAP_DELALLOC)705 706struct f2fs_map_blocks {707 struct block_device *m_bdev; /* for multi-device dio */708 block_t m_pblk;709 block_t m_lblk;710 unsigned int m_len;711 unsigned int m_flags;712 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */713 pgoff_t *m_next_extent; /* point to next possible extent */714 int m_seg_type;715 bool m_may_create; /* indicate it is from write path */716 bool m_multidev_dio; /* indicate it allows multi-device dio */717};718 719/* for flag in get_data_block */720enum {721 F2FS_GET_BLOCK_DEFAULT,722 F2FS_GET_BLOCK_FIEMAP,723 F2FS_GET_BLOCK_BMAP,724 F2FS_GET_BLOCK_DIO,725 F2FS_GET_BLOCK_PRE_DIO,726 F2FS_GET_BLOCK_PRE_AIO,727 F2FS_GET_BLOCK_PRECACHE,728};729 730/*731 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later.732 */733#define FADVISE_COLD_BIT 0x01734#define FADVISE_LOST_PINO_BIT 0x02735#define FADVISE_ENCRYPT_BIT 0x04736#define FADVISE_ENC_NAME_BIT 0x08737#define FADVISE_KEEP_SIZE_BIT 0x10738#define FADVISE_HOT_BIT 0x20739#define FADVISE_VERITY_BIT 0x40740#define FADVISE_TRUNC_BIT 0x80741 742#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT)743 744#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)745#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)746#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)747 748#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)749#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)750#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)751 752#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT)753#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT)754 755#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT)756#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT)757 758#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT)759#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT)760 761#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT)762#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT)763#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT)764 765#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT)766#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT)767 768#define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT)769#define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT)770#define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT)771 772#define DEF_DIR_LEVEL 0773 774/* used for f2fs_inode_info->flags */775enum {776 FI_NEW_INODE, /* indicate newly allocated inode */777 FI_DIRTY_INODE, /* indicate inode is dirty or not */778 FI_AUTO_RECOVER, /* indicate inode is recoverable */779 FI_DIRTY_DIR, /* indicate directory has dirty pages */780 FI_INC_LINK, /* need to increment i_nlink */781 FI_ACL_MODE, /* indicate acl mode */782 FI_NO_ALLOC, /* should not allocate any blocks */783 FI_FREE_NID, /* free allocated nide */784 FI_NO_EXTENT, /* not to use the extent cache */785 FI_INLINE_XATTR, /* used for inline xattr */786 FI_INLINE_DATA, /* used for inline data*/787 FI_INLINE_DENTRY, /* used for inline dentry */788 FI_APPEND_WRITE, /* inode has appended data */789 FI_UPDATE_WRITE, /* inode has in-place-update data */790 FI_NEED_IPU, /* used for ipu per file */791 FI_ATOMIC_FILE, /* indicate atomic file */792 FI_DATA_EXIST, /* indicate data exists */793 FI_SKIP_WRITES, /* should skip data page writeback */794 FI_OPU_WRITE, /* used for opu per file */795 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */796 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */797 FI_HOT_DATA, /* indicate file is hot */798 FI_EXTRA_ATTR, /* indicate file has extra attribute */799 FI_PROJ_INHERIT, /* indicate file inherits projectid */800 FI_PIN_FILE, /* indicate file should not be gced */801 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */802 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */803 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */804 FI_MMAP_FILE, /* indicate file was mmapped */805 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */806 FI_COMPRESS_RELEASED, /* compressed blocks were released */807 FI_ALIGNED_WRITE, /* enable aligned write */808 FI_COW_FILE, /* indicate COW file */809 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */810 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */811 FI_ATOMIC_REPLACE, /* indicate atomic replace */812 FI_OPENED_FILE, /* indicate file has been opened */813 FI_MAX, /* max flag, never be used */814};815 816struct f2fs_inode_info {817 struct inode vfs_inode; /* serve a vfs inode */818 unsigned long i_flags; /* keep an inode flags for ioctl */819 unsigned char i_advise; /* use to give file attribute hints */820 unsigned char i_dir_level; /* use for dentry level for large dir */821 union {822 unsigned int i_current_depth; /* only for directory depth */823 unsigned short i_gc_failures; /* for gc failure statistic */824 };825 unsigned int i_pino; /* parent inode number */826 umode_t i_acl_mode; /* keep file acl mode temporarily */827 828 /* Use below internally in f2fs*/829 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */830 struct f2fs_rwsem i_sem; /* protect fi info */831 atomic_t dirty_pages; /* # of dirty pages */832 f2fs_hash_t chash; /* hash value of given file name */833 unsigned int clevel; /* maximum level of given file name */834 struct task_struct *task; /* lookup and create consistency */835 struct task_struct *cp_task; /* separate cp/wb IO stats*/836 struct task_struct *wb_task; /* indicate inode is in context of writeback */837 nid_t i_xattr_nid; /* node id that contains xattrs */838 loff_t last_disk_size; /* lastly written file size */839 spinlock_t i_size_lock; /* protect last_disk_size */840 841#ifdef CONFIG_QUOTA842 struct dquot __rcu *i_dquot[MAXQUOTAS];843 844 /* quota space reservation, managed internally by quota code */845 qsize_t i_reserved_quota;846#endif847 struct list_head dirty_list; /* dirty list for dirs and files */848 struct list_head gdirty_list; /* linked in global dirty list */849 struct task_struct *atomic_write_task; /* store atomic write task */850 struct extent_tree *extent_tree[NR_EXTENT_CACHES];851 /* cached extent_tree entry */852 union {853 struct inode *cow_inode; /* copy-on-write inode for atomic write */854 struct inode *atomic_inode;855 /* point to atomic_inode, available only for cow_inode */856 };857 858 /* avoid racing between foreground op and gc */859 struct f2fs_rwsem i_gc_rwsem[2];860 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */861 862 int i_extra_isize; /* size of extra space located in i_addr */863 kprojid_t i_projid; /* id for project quota */864 int i_inline_xattr_size; /* inline xattr size */865 struct timespec64 i_crtime; /* inode creation time */866 struct timespec64 i_disk_time[3];/* inode disk times */867 868 /* for file compress */869 atomic_t i_compr_blocks; /* # of compressed blocks */870 unsigned char i_compress_algorithm; /* algorithm type */871 unsigned char i_log_cluster_size; /* log of cluster size */872 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */873 unsigned char i_compress_flag; /* compress flag */874 unsigned int i_cluster_size; /* cluster size */875 876 unsigned int atomic_write_cnt;877 loff_t original_i_size; /* original i_size before atomic write */878};879 880static inline void get_read_extent_info(struct extent_info *ext,881 struct f2fs_extent *i_ext)882{883 ext->fofs = le32_to_cpu(i_ext->fofs);884 ext->blk = le32_to_cpu(i_ext->blk);885 ext->len = le32_to_cpu(i_ext->len);886}887 888static inline void set_raw_read_extent(struct extent_info *ext,889 struct f2fs_extent *i_ext)890{891 i_ext->fofs = cpu_to_le32(ext->fofs);892 i_ext->blk = cpu_to_le32(ext->blk);893 i_ext->len = cpu_to_le32(ext->len);894}895 896static inline bool __is_discard_mergeable(struct discard_info *back,897 struct discard_info *front, unsigned int max_len)898{899 return (back->lstart + back->len == front->lstart) &&900 (back->len + front->len <= max_len);901}902 903static inline bool __is_discard_back_mergeable(struct discard_info *cur,904 struct discard_info *back, unsigned int max_len)905{906 return __is_discard_mergeable(back, cur, max_len);907}908 909static inline bool __is_discard_front_mergeable(struct discard_info *cur,910 struct discard_info *front, unsigned int max_len)911{912 return __is_discard_mergeable(cur, front, max_len);913}914 915/*916 * For free nid management917 */918enum nid_state {919 FREE_NID, /* newly added to free nid list */920 PREALLOC_NID, /* it is preallocated */921 MAX_NID_STATE,922};923 924enum nat_state {925 TOTAL_NAT,926 DIRTY_NAT,927 RECLAIMABLE_NAT,928 MAX_NAT_STATE,929};930 931struct f2fs_nm_info {932 block_t nat_blkaddr; /* base disk address of NAT */933 nid_t max_nid; /* maximum possible node ids */934 nid_t available_nids; /* # of available node ids */935 nid_t next_scan_nid; /* the next nid to be scanned */936 nid_t max_rf_node_blocks; /* max # of nodes for recovery */937 unsigned int ram_thresh; /* control the memory footprint */938 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */939 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */940 941 /* NAT cache management */942 struct radix_tree_root nat_root;/* root of the nat entry cache */943 struct radix_tree_root nat_set_root;/* root of the nat set cache */944 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */945 struct list_head nat_entries; /* cached nat entry list (clean) */946 spinlock_t nat_list_lock; /* protect clean nat entry list */947 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */948 unsigned int nat_blocks; /* # of nat blocks */949 950 /* free node ids management */951 struct radix_tree_root free_nid_root;/* root of the free_nid cache */952 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */953 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */954 spinlock_t nid_list_lock; /* protect nid lists ops */955 struct mutex build_lock; /* lock for build free nids */956 unsigned char **free_nid_bitmap;957 unsigned char *nat_block_bitmap;958 unsigned short *free_nid_count; /* free nid count of NAT block */959 960 /* for checkpoint */961 char *nat_bitmap; /* NAT bitmap pointer */962 963 unsigned int nat_bits_blocks; /* # of nat bits blocks */964 unsigned char *nat_bits; /* NAT bits blocks */965 unsigned char *full_nat_bits; /* full NAT pages */966 unsigned char *empty_nat_bits; /* empty NAT pages */967#ifdef CONFIG_F2FS_CHECK_FS968 char *nat_bitmap_mir; /* NAT bitmap mirror */969#endif970 int bitmap_size; /* bitmap size */971};972 973/*974 * this structure is used as one of function parameters.975 * all the information are dedicated to a given direct node block determined976 * by the data offset in a file.977 */978struct dnode_of_data {979 struct inode *inode; /* vfs inode pointer */980 struct page *inode_page; /* its inode page, NULL is possible */981 struct page *node_page; /* cached direct node page */982 nid_t nid; /* node id of the direct node block */983 unsigned int ofs_in_node; /* data offset in the node page */984 bool inode_page_locked; /* inode page is locked or not */985 bool node_changed; /* is node block changed */986 char cur_level; /* level of hole node page */987 char max_level; /* level of current page located */988 block_t data_blkaddr; /* block address of the node block */989};990 991static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode,992 struct page *ipage, struct page *npage, nid_t nid)993{994 memset(dn, 0, sizeof(*dn));995 dn->inode = inode;996 dn->inode_page = ipage;997 dn->node_page = npage;998 dn->nid = nid;999}1000 1001/*1002 * For SIT manager1003 *1004 * By default, there are 6 active log areas across the whole main area.1005 * When considering hot and cold data separation to reduce cleaning overhead,1006 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types,1007 * respectively.1008 * In the current design, you should not change the numbers intentionally.1009 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 61010 * logs individually according to the underlying devices. (default: 6)1011 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for1012 * data and 8 for node logs.1013 */1014#define NR_CURSEG_DATA_TYPE (3)1015#define NR_CURSEG_NODE_TYPE (3)1016#define NR_CURSEG_INMEM_TYPE (2)1017#define NR_CURSEG_RO_TYPE (2)1018#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE)1019#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE)1020 1021enum {1022 CURSEG_HOT_DATA = 0, /* directory entry blocks */1023 CURSEG_WARM_DATA, /* data blocks */1024 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */1025 CURSEG_HOT_NODE, /* direct node blocks of directory files */1026 CURSEG_WARM_NODE, /* direct node blocks of normal files */1027 CURSEG_COLD_NODE, /* indirect node blocks */1028 NR_PERSISTENT_LOG, /* number of persistent log */1029 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG,1030 /* pinned file that needs consecutive block address */1031 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */1032 NO_CHECK_TYPE, /* number of persistent & inmem log */1033};1034 1035struct flush_cmd {1036 struct completion wait;1037 struct llist_node llnode;1038 nid_t ino;1039 int ret;1040};1041 1042struct flush_cmd_control {1043 struct task_struct *f2fs_issue_flush; /* flush thread */1044 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */1045 atomic_t issued_flush; /* # of issued flushes */1046 atomic_t queued_flush; /* # of queued flushes */1047 struct llist_head issue_list; /* list for command issue */1048 struct llist_node *dispatch_list; /* list for command dispatch */1049};1050 1051struct f2fs_sm_info {1052 struct sit_info *sit_info; /* whole segment information */1053 struct free_segmap_info *free_info; /* free segment information */1054 struct dirty_seglist_info *dirty_info; /* dirty segment information */1055 struct curseg_info *curseg_array; /* active segment information */1056 1057 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */1058 1059 block_t seg0_blkaddr; /* block address of 0'th segment */1060 block_t main_blkaddr; /* start block address of main area */1061 block_t ssa_blkaddr; /* start block address of SSA area */1062 1063 unsigned int segment_count; /* total # of segments */1064 unsigned int main_segments; /* # of segments in main area */1065 unsigned int reserved_segments; /* # of reserved segments */1066 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */1067 unsigned int ovp_segments; /* # of overprovision segments */1068 1069 /* a threshold to reclaim prefree segments */1070 unsigned int rec_prefree_segments;1071 1072 struct list_head sit_entry_set; /* sit entry set list */1073 1074 unsigned int ipu_policy; /* in-place-update policy */1075 unsigned int min_ipu_util; /* in-place-update threshold */1076 unsigned int min_fsync_blocks; /* threshold for fsync */1077 unsigned int min_seq_blocks; /* threshold for sequential blocks */1078 unsigned int min_hot_blocks; /* threshold for hot block allocation */1079 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */1080 1081 /* for flush command control */1082 struct flush_cmd_control *fcc_info;1083 1084 /* for discard command control */1085 struct discard_cmd_control *dcc_info;1086};1087 1088/*1089 * For superblock1090 */1091/*1092 * COUNT_TYPE for monitoring1093 *1094 * f2fs monitors the number of several block types such as on-writeback,1095 * dirty dentry blocks, dirty node blocks, and dirty meta blocks.1096 */1097#define WB_DATA_TYPE(p, f) \1098 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA)1099enum count_type {1100 F2FS_DIRTY_DENTS,1101 F2FS_DIRTY_DATA,1102 F2FS_DIRTY_QDATA,1103 F2FS_DIRTY_NODES,1104 F2FS_DIRTY_META,1105 F2FS_DIRTY_IMETA,1106 F2FS_WB_CP_DATA,1107 F2FS_WB_DATA,1108 F2FS_RD_DATA,1109 F2FS_RD_NODE,1110 F2FS_RD_META,1111 F2FS_DIO_WRITE,1112 F2FS_DIO_READ,1113 NR_COUNT_TYPE,1114};1115 1116/*1117 * The below are the page types of bios used in submit_bio().1118 * The available types are:1119 * DATA User data pages. It operates as async mode.1120 * NODE Node pages. It operates as async mode.1121 * META FS metadata pages such as SIT, NAT, CP.1122 * NR_PAGE_TYPE The number of page types.1123 * META_FLUSH Make sure the previous pages are written1124 * with waiting the bio's completion1125 * ... Only can be used with META.1126 */1127#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type))1128#define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE)1129enum page_type {1130 DATA = 0,1131 NODE = 1, /* should not change this */1132 META,1133 NR_PAGE_TYPE,1134 META_FLUSH,1135 IPU, /* the below types are used by tracepoints only. */1136 OPU,1137};1138 1139enum temp_type {1140 HOT = 0, /* must be zero for meta bio */1141 WARM,1142 COLD,1143 NR_TEMP_TYPE,1144};1145 1146enum need_lock_type {1147 LOCK_REQ = 0,1148 LOCK_DONE,1149 LOCK_RETRY,1150};1151 1152enum cp_reason_type {1153 CP_NO_NEEDED,1154 CP_NON_REGULAR,1155 CP_COMPRESSED,1156 CP_HARDLINK,1157 CP_SB_NEED_CP,1158 CP_WRONG_PINO,1159 CP_NO_SPC_ROLL,1160 CP_NODE_NEED_CP,1161 CP_FASTBOOT_MODE,1162 CP_SPEC_LOG_NUM,1163 CP_RECOVER_DIR,1164 CP_XATTR_DIR,1165};1166 1167enum iostat_type {1168 /* WRITE IO */1169 APP_DIRECT_IO, /* app direct write IOs */1170 APP_BUFFERED_IO, /* app buffered write IOs */1171 APP_WRITE_IO, /* app write IOs */1172 APP_MAPPED_IO, /* app mapped IOs */1173 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */1174 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */1175 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */1176 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */1177 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */1178 FS_META_IO, /* meta IOs from kworker/reclaimer */1179 FS_GC_DATA_IO, /* data IOs from forground gc */1180 FS_GC_NODE_IO, /* node IOs from forground gc */1181 FS_CP_DATA_IO, /* data IOs from checkpoint */1182 FS_CP_NODE_IO, /* node IOs from checkpoint */1183 FS_CP_META_IO, /* meta IOs from checkpoint */1184 1185 /* READ IO */1186 APP_DIRECT_READ_IO, /* app direct read IOs */1187 APP_BUFFERED_READ_IO, /* app buffered read IOs */1188 APP_READ_IO, /* app read IOs */1189 APP_MAPPED_READ_IO, /* app mapped read IOs */1190 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */1191 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */1192 FS_DATA_READ_IO, /* data read IOs */1193 FS_GDATA_READ_IO, /* data read IOs from background gc */1194 FS_CDATA_READ_IO, /* compressed data read IOs */1195 FS_NODE_READ_IO, /* node read IOs */1196 FS_META_READ_IO, /* meta read IOs */1197 1198 /* other */1199 FS_DISCARD_IO, /* discard */1200 FS_FLUSH_IO, /* flush */1201 FS_ZONE_RESET_IO, /* zone reset */1202 NR_IO_TYPE,1203};1204 1205struct f2fs_io_info {1206 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */1207 nid_t ino; /* inode number */1208 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */1209 enum temp_type temp; /* contains HOT/WARM/COLD */1210 enum req_op op; /* contains REQ_OP_ */1211 blk_opf_t op_flags; /* req_flag_bits */1212 block_t new_blkaddr; /* new block address to be written */1213 block_t old_blkaddr; /* old block address before Cow */1214 struct page *page; /* page to be written */1215 struct page *encrypted_page; /* encrypted page */1216 struct page *compressed_page; /* compressed page */1217 struct list_head list; /* serialize IOs */1218 unsigned int compr_blocks; /* # of compressed block addresses */1219 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */1220 unsigned int version:8; /* version of the node */1221 unsigned int submitted:1; /* indicate IO submission */1222 unsigned int in_list:1; /* indicate fio is in io_list */1223 unsigned int is_por:1; /* indicate IO is from recovery or not */1224 unsigned int encrypted:1; /* indicate file is encrypted */1225 unsigned int meta_gc:1; /* require meta inode GC */1226 enum iostat_type io_type; /* io type */1227 struct writeback_control *io_wbc; /* writeback control */1228 struct bio **bio; /* bio for ipu */1229 sector_t *last_block; /* last block number in bio */1230};1231 1232struct bio_entry {1233 struct bio *bio;1234 struct list_head list;1235};1236 1237#define is_read_io(rw) ((rw) == READ)1238struct f2fs_bio_info {1239 struct f2fs_sb_info *sbi; /* f2fs superblock */1240 struct bio *bio; /* bios to merge */1241 sector_t last_block_in_bio; /* last block number */1242 struct f2fs_io_info fio; /* store buffered io info. */1243#ifdef CONFIG_BLK_DEV_ZONED1244 struct completion zone_wait; /* condition value for the previous open zone to close */1245 struct bio *zone_pending_bio; /* pending bio for the previous zone */1246 void *bi_private; /* previous bi_private for pending bio */1247#endif1248 struct f2fs_rwsem io_rwsem; /* blocking op for bio */1249 spinlock_t io_lock; /* serialize DATA/NODE IOs */1250 struct list_head io_list; /* track fios */1251 struct list_head bio_list; /* bio entry list head */1252 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */1253};1254 1255#define FDEV(i) (sbi->devs[i])1256#define RDEV(i) (raw_super->devs[i])1257struct f2fs_dev_info {1258 struct file *bdev_file;1259 struct block_device *bdev;1260 char path[MAX_PATH_LEN];1261 unsigned int total_segments;1262 block_t start_blk;1263 block_t end_blk;1264#ifdef CONFIG_BLK_DEV_ZONED1265 unsigned int nr_blkz; /* Total number of zones */1266 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */1267#endif1268};1269 1270enum inode_type {1271 DIR_INODE, /* for dirty dir inode */1272 FILE_INODE, /* for dirty regular/symlink inode */1273 DIRTY_META, /* for all dirtied inode metadata */1274 NR_INODE_TYPE,1275};1276 1277/* for inner inode cache management */1278struct inode_management {1279 struct radix_tree_root ino_root; /* ino entry array */1280 spinlock_t ino_lock; /* for ino entry lock */1281 struct list_head ino_list; /* inode list head */1282 unsigned long ino_num; /* number of entries */1283};1284 1285/* for GC_AT */1286struct atgc_management {1287 bool atgc_enabled; /* ATGC is enabled or not */1288 struct rb_root_cached root; /* root of victim rb-tree */1289 struct list_head victim_list; /* linked with all victim entries */1290 unsigned int victim_count; /* victim count in rb-tree */1291 unsigned int candidate_ratio; /* candidate ratio */1292 unsigned int max_candidate_count; /* max candidate count */1293 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */1294 unsigned long long age_threshold; /* age threshold */1295};1296 1297struct f2fs_gc_control {1298 unsigned int victim_segno; /* target victim segment number */1299 int init_gc_type; /* FG_GC or BG_GC */1300 bool no_bg_gc; /* check the space and stop bg_gc */1301 bool should_migrate_blocks; /* should migrate blocks */1302 bool err_gc_skipped; /* return EAGAIN if GC skipped */1303 bool one_time; /* require one time GC in one migration unit */1304 unsigned int nr_free_secs; /* # of free sections to do GC */1305};1306 1307/*1308 * For s_flag in struct f2fs_sb_info1309 * Modification on enum should be synchronized with s_flag array1310 */1311enum {1312 SBI_IS_DIRTY, /* dirty flag for checkpoint */1313 SBI_IS_CLOSE, /* specify unmounting */1314 SBI_NEED_FSCK, /* need fsck.f2fs to fix */1315 SBI_POR_DOING, /* recovery is doing or not */1316 SBI_NEED_SB_WRITE, /* need to recover superblock */1317 SBI_NEED_CP, /* need to checkpoint */1318 SBI_IS_SHUTDOWN, /* shutdown by ioctl */1319 SBI_IS_RECOVERED, /* recovered orphan/data */1320 SBI_CP_DISABLED, /* CP was disabled last mount */1321 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */1322 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */1323 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */1324 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */1325 SBI_IS_RESIZEFS, /* resizefs is in process */1326 SBI_IS_FREEZING, /* freezefs is in process */1327 SBI_IS_WRITABLE, /* remove ro mountoption transiently */1328 MAX_SBI_FLAG,1329};1330 1331enum {1332 CP_TIME,1333 REQ_TIME,1334 DISCARD_TIME,1335 GC_TIME,1336 DISABLE_TIME,1337 UMOUNT_DISCARD_TIMEOUT,1338 MAX_TIME,1339};1340 1341/* Note that you need to keep synchronization with this gc_mode_names array */1342enum {1343 GC_NORMAL,1344 GC_IDLE_CB,1345 GC_IDLE_GREEDY,1346 GC_IDLE_AT,1347 GC_URGENT_HIGH,1348 GC_URGENT_LOW,1349 GC_URGENT_MID,1350 MAX_GC_MODE,1351};1352 1353enum {1354 BGGC_MODE_ON, /* background gc is on */1355 BGGC_MODE_OFF, /* background gc is off */1356 BGGC_MODE_SYNC, /*1357 * background gc is on, migrating blocks1358 * like foreground gc1359 */1360};1361 1362enum {1363 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */1364 FS_MODE_LFS, /* use lfs allocation only */1365 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */1366 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */1367};1368 1369enum {1370 ALLOC_MODE_DEFAULT, /* stay default */1371 ALLOC_MODE_REUSE, /* reuse segments as much as possible */1372};1373 1374enum fsync_mode {1375 FSYNC_MODE_POSIX, /* fsync follows posix semantics */1376 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */1377 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */1378};1379 1380enum {1381 COMPR_MODE_FS, /*1382 * automatically compress compression1383 * enabled files1384 */1385 COMPR_MODE_USER, /*1386 * automatical compression is disabled.1387 * user can control the file compression1388 * using ioctls1389 */1390};1391 1392enum {1393 DISCARD_UNIT_BLOCK, /* basic discard unit is block */1394 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */1395 DISCARD_UNIT_SECTION, /* basic discard unit is section */1396};1397 1398enum {1399 MEMORY_MODE_NORMAL, /* memory mode for normal devices */1400 MEMORY_MODE_LOW, /* memory mode for low memry devices */1401};1402 1403enum errors_option {1404 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */1405 MOUNT_ERRORS_CONTINUE, /* continue on errors */1406 MOUNT_ERRORS_PANIC, /* panic on errors */1407};1408 1409enum {1410 BACKGROUND,1411 FOREGROUND,1412 MAX_CALL_TYPE,1413 TOTAL_CALL = FOREGROUND,1414};1415 1416static inline int f2fs_test_bit(unsigned int nr, char *addr);1417static inline void f2fs_set_bit(unsigned int nr, char *addr);1418static inline void f2fs_clear_bit(unsigned int nr, char *addr);1419 1420/*1421 * Layout of f2fs page.private:1422 *1423 * Layout A: lowest bit should be 11424 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... |1425 * bit 0 PAGE_PRIVATE_NOT_POINTER1426 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION1427 * bit 2 PAGE_PRIVATE_INLINE_INODE1428 * bit 3 PAGE_PRIVATE_REF_RESOURCE1429 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE1430 * bit 5- f2fs private data1431 *1432 * Layout B: lowest bit should be 01433 * page.private is a wrapped pointer.1434 */1435enum {1436 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */1437 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */1438 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */1439 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */1440 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */1441 PAGE_PRIVATE_MAX1442};1443 1444/* For compression */1445enum compress_algorithm_type {1446 COMPRESS_LZO,1447 COMPRESS_LZ4,1448 COMPRESS_ZSTD,1449 COMPRESS_LZORLE,1450 COMPRESS_MAX,1451};1452 1453enum compress_flag {1454 COMPRESS_CHKSUM,1455 COMPRESS_MAX_FLAG,1456};1457 1458#define COMPRESS_WATERMARK 201459#define COMPRESS_PERCENT 201460 1461#define COMPRESS_DATA_RESERVED_SIZE 41462struct compress_data {1463 __le32 clen; /* compressed data size */1464 __le32 chksum; /* compressed data chksum */1465 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */1466 u8 cdata[]; /* compressed data */1467};1468 1469#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data))1470 1471#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C0001472 1473#define F2FS_ZSTD_DEFAULT_CLEVEL 11474 1475#define COMPRESS_LEVEL_OFFSET 81476 1477/* compress context */1478struct compress_ctx {1479 struct inode *inode; /* inode the context belong to */1480 pgoff_t cluster_idx; /* cluster index number */1481 unsigned int cluster_size; /* page count in cluster */1482 unsigned int log_cluster_size; /* log of cluster size */1483 struct page **rpages; /* pages store raw data in cluster */1484 unsigned int nr_rpages; /* total page number in rpages */1485 struct page **cpages; /* pages store compressed data in cluster */1486 unsigned int nr_cpages; /* total page number in cpages */1487 unsigned int valid_nr_cpages; /* valid page number in cpages */1488 void *rbuf; /* virtual mapped address on rpages */1489 struct compress_data *cbuf; /* virtual mapped address on cpages */1490 size_t rlen; /* valid data length in rbuf */1491 size_t clen; /* valid data length in cbuf */1492 void *private; /* payload buffer for specified compression algorithm */1493 void *private2; /* extra payload buffer */1494};1495 1496/* compress context for write IO path */1497struct compress_io_ctx {1498 u32 magic; /* magic number to indicate page is compressed */1499 struct inode *inode; /* inode the context belong to */1500 struct page **rpages; /* pages store raw data in cluster */1501 unsigned int nr_rpages; /* total page number in rpages */1502 atomic_t pending_pages; /* in-flight compressed page count */1503};1504 1505/* Context for decompressing one cluster on the read IO path */1506struct decompress_io_ctx {1507 u32 magic; /* magic number to indicate page is compressed */1508 struct inode *inode; /* inode the context belong to */1509 pgoff_t cluster_idx; /* cluster index number */1510 unsigned int cluster_size; /* page count in cluster */1511 unsigned int log_cluster_size; /* log of cluster size */1512 struct page **rpages; /* pages store raw data in cluster */1513 unsigned int nr_rpages; /* total page number in rpages */1514 struct page **cpages; /* pages store compressed data in cluster */1515 unsigned int nr_cpages; /* total page number in cpages */1516 struct page **tpages; /* temp pages to pad holes in cluster */1517 void *rbuf; /* virtual mapped address on rpages */1518 struct compress_data *cbuf; /* virtual mapped address on cpages */1519 size_t rlen; /* valid data length in rbuf */1520 size_t clen; /* valid data length in cbuf */1521 1522 /*1523 * The number of compressed pages remaining to be read in this cluster.1524 * This is initially nr_cpages. It is decremented by 1 each time a page1525 * has been read (or failed to be read). When it reaches 0, the cluster1526 * is decompressed (or an error is reported).1527 *1528 * If an error occurs before all the pages have been submitted for I/O,1529 * then this will never reach 0. In this case the I/O submitter is1530 * responsible for calling f2fs_decompress_end_io() instead.1531 */1532 atomic_t remaining_pages;1533 1534 /*1535 * Number of references to this decompress_io_ctx.1536 *1537 * One reference is held for I/O completion. This reference is dropped1538 * after the pagecache pages are updated and unlocked -- either after1539 * decompression (and verity if enabled), or after an error.1540 *1541 * In addition, each compressed page holds a reference while it is in a1542 * bio. These references are necessary prevent compressed pages from1543 * being freed while they are still in a bio.1544 */1545 refcount_t refcnt;1546 1547 bool failed; /* IO error occurred before decompression? */1548 bool need_verity; /* need fs-verity verification after decompression? */1549 void *private; /* payload buffer for specified decompression algorithm */1550 void *private2; /* extra payload buffer */1551 struct work_struct verity_work; /* work to verify the decompressed pages */1552 struct work_struct free_work; /* work for late free this structure itself */1553};1554 1555#define NULL_CLUSTER ((unsigned int)(~0))1556#define MIN_COMPRESS_LOG_SIZE 21557#define MAX_COMPRESS_LOG_SIZE 81558#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size))1559 1560struct f2fs_sb_info {1561 struct super_block *sb; /* pointer to VFS super block */1562 struct proc_dir_entry *s_proc; /* proc entry */1563 struct f2fs_super_block *raw_super; /* raw super block pointer */1564 struct f2fs_rwsem sb_lock; /* lock for raw super block */1565 int valid_super_block; /* valid super block no */1566 unsigned long s_flag; /* flags for sbi */1567 struct mutex writepages; /* mutex for writepages() */1568 1569#ifdef CONFIG_BLK_DEV_ZONED1570 unsigned int blocks_per_blkz; /* F2FS blocks per zone */1571 unsigned int max_open_zones; /* max open zone resources of the zoned device */1572 /* For adjust the priority writing position of data in zone UFS */1573 unsigned int blkzone_alloc_policy;1574#endif1575 1576 /* for node-related operations */1577 struct f2fs_nm_info *nm_info; /* node manager */1578 struct inode *node_inode; /* cache node blocks */1579 1580 /* for segment-related operations */1581 struct f2fs_sm_info *sm_info; /* segment manager */1582 1583 /* for bio operations */1584 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */1585 /* keep migration IO order for LFS mode */1586 struct f2fs_rwsem io_order_lock;1587 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */1588 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */1589 1590 /* for checkpoint */1591 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */1592 int cur_cp_pack; /* remain current cp pack */1593 spinlock_t cp_lock; /* for flag in ckpt */1594 struct inode *meta_inode; /* cache meta blocks */1595 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */1596 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */1597 struct f2fs_rwsem node_write; /* locking node writes */1598 struct f2fs_rwsem node_change; /* locking node change */1599 wait_queue_head_t cp_wait;1600 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */1601 long interval_time[MAX_TIME]; /* to store thresholds */1602 struct ckpt_req_control cprc_info; /* for checkpoint request control */1603 1604 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */1605 1606 spinlock_t fsync_node_lock; /* for node entry lock */1607 struct list_head fsync_node_list; /* node list head */1608 unsigned int fsync_seg_id; /* sequence id */1609 unsigned int fsync_node_num; /* number of node entries */1610 1611 /* for orphan inode, use 0'th array */1612 unsigned int max_orphans; /* max orphan inodes */1613 1614 /* for inode management */1615 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */1616 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */1617 struct mutex flush_lock; /* for flush exclusion */1618 1619 /* for extent tree cache */1620 struct extent_tree_info extent_tree[NR_EXTENT_CACHES];1621 atomic64_t allocated_data_blocks; /* for block age extent_cache */1622 1623 /* The threshold used for hot and warm data seperation*/1624 unsigned int hot_data_age_threshold;1625 unsigned int warm_data_age_threshold;1626 unsigned int last_age_weight;1627 1628 /* basic filesystem units */1629 unsigned int log_sectors_per_block; /* log2 sectors per block */1630 unsigned int log_blocksize; /* log2 block size */1631 unsigned int blocksize; /* block size */1632 unsigned int root_ino_num; /* root inode number*/1633 unsigned int node_ino_num; /* node inode number*/1634 unsigned int meta_ino_num; /* meta inode number*/1635 unsigned int log_blocks_per_seg; /* log2 blocks per segment */1636 unsigned int blocks_per_seg; /* blocks per segment */1637 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */1638 unsigned int segs_per_sec; /* segments per section */1639 unsigned int secs_per_zone; /* sections per zone */1640 unsigned int total_sections; /* total section count */1641 unsigned int total_node_count; /* total node block count */1642 unsigned int total_valid_node_count; /* valid node block count */1643 int dir_level; /* directory level */1644 bool readdir_ra; /* readahead inode in readdir */1645 u64 max_io_bytes; /* max io bytes to merge IOs */1646 1647 block_t user_block_count; /* # of user blocks */1648 block_t total_valid_block_count; /* # of valid blocks */1649 block_t discard_blks; /* discard command candidats */1650 block_t last_valid_block_count; /* for recovery */1651 block_t reserved_blocks; /* configurable reserved blocks */1652 block_t current_reserved_blocks; /* current reserved blocks */1653 1654 /* Additional tracking for no checkpoint mode */1655 block_t unusable_block_count; /* # of blocks saved by last cp */1656 1657 unsigned int nquota_files; /* # of quota sysfile */1658 struct f2fs_rwsem quota_sem; /* blocking cp for flags */1659 1660 /* # of pages, see count_type */1661 atomic_t nr_pages[NR_COUNT_TYPE];1662 /* # of allocated blocks */1663 struct percpu_counter alloc_valid_block_count;1664 /* # of node block writes as roll forward recovery */1665 struct percpu_counter rf_node_block_count;1666 1667 /* writeback control */1668 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */1669 1670 /* valid inode count */1671 struct percpu_counter total_valid_inode_count;1672 1673 struct f2fs_mount_info mount_opt; /* mount options */1674 1675 /* for cleaning operations */1676 struct f2fs_rwsem gc_lock; /*1677 * semaphore for GC, avoid1678 * race between GC and GC or CP1679 */1680 struct f2fs_gc_kthread *gc_thread; /* GC thread */1681 struct atgc_management am; /* atgc management */1682 unsigned int cur_victim_sec; /* current victim section num */1683 unsigned int gc_mode; /* current GC state */1684 unsigned int next_victim_seg[2]; /* next segment in victim section */1685 spinlock_t gc_remaining_trials_lock;1686 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */1687 unsigned int gc_remaining_trials;1688 1689 /* for skip statistic */1690 unsigned long long skipped_gc_rwsem; /* FG_GC only */1691 1692 /* threshold for gc trials on pinned files */1693 unsigned short gc_pin_file_threshold;1694 struct f2fs_rwsem pin_sem;1695 1696 /* maximum # of trials to find a victim segment for SSR and GC */1697 unsigned int max_victim_search;1698 /* migration granularity of garbage collection, unit: segment */1699 unsigned int migration_granularity;1700 /* migration window granularity of garbage collection, unit: segment */1701 unsigned int migration_window_granularity;1702 1703 /*1704 * for stat information.1705 * one is for the LFS mode, and the other is for the SSR mode.1706 */1707#ifdef CONFIG_F2FS_STAT_FS1708 struct f2fs_stat_info *stat_info; /* FS status information */1709 atomic_t meta_count[META_MAX]; /* # of meta blocks */1710 unsigned int segment_count[2]; /* # of allocated segments */1711 unsigned int block_count[2]; /* # of allocated blocks */1712 atomic_t inplace_count; /* # of inplace update */1713 /* # of lookup extent cache */1714 atomic64_t total_hit_ext[NR_EXTENT_CACHES];1715 /* # of hit rbtree extent node */1716 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES];1717 /* # of hit cached extent node */1718 atomic64_t read_hit_cached[NR_EXTENT_CACHES];1719 /* # of hit largest extent node in read extent cache */1720 atomic64_t read_hit_largest;1721 atomic_t inline_xattr; /* # of inline_xattr inodes */1722 atomic_t inline_inode; /* # of inline_data inodes */1723 atomic_t inline_dir; /* # of inline_dentry inodes */1724 atomic_t compr_inode; /* # of compressed inodes */1725 atomic64_t compr_blocks; /* # of compressed blocks */1726 atomic_t swapfile_inode; /* # of swapfile inodes */1727 atomic_t atomic_files; /* # of opened atomic file */1728 atomic_t max_aw_cnt; /* max # of atomic writes */1729 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */1730 unsigned int other_skip_bggc; /* skip background gc for other reasons */1731 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */1732 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */1733#endif1734 spinlock_t stat_lock; /* lock for stat operations */1735 1736 /* to attach REQ_META|REQ_FUA flags */1737 unsigned int data_io_flag;1738 unsigned int node_io_flag;1739 1740 /* For sysfs support */1741 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */1742 struct completion s_kobj_unregister;1743 1744 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */1745 struct completion s_stat_kobj_unregister;1746 1747 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */1748 struct completion s_feature_list_kobj_unregister;1749 1750 /* For shrinker support */1751 struct list_head s_list;1752 struct mutex umount_mutex;1753 unsigned int shrinker_run_no;1754 1755 /* For multi devices */1756 int s_ndevs; /* number of devices */1757 struct f2fs_dev_info *devs; /* for device list */1758 unsigned int dirty_device; /* for checkpoint data flush */1759 spinlock_t dev_lock; /* protect dirty_device */1760 bool aligned_blksize; /* all devices has the same logical blksize */1761 1762 /* For write statistics */1763 u64 sectors_written_start;1764 u64 kbytes_written;1765 1766 /* Reference to checksum algorithm driver via cryptoapi */1767 struct crypto_shash *s_chksum_driver;1768 1769 /* Precomputed FS UUID checksum for seeding other checksums */1770 __u32 s_chksum_seed;1771 1772 struct workqueue_struct *post_read_wq; /* post read workqueue */1773 1774 /*1775 * If we are in irq context, let's update error information into1776 * on-disk superblock in the work.1777 */1778 struct work_struct s_error_work;1779 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */1780 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */1781 spinlock_t error_lock; /* protect errors/stop_reason array */1782 bool error_dirty; /* errors of sb is dirty */1783 1784 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */1785 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */1786 1787 /* For reclaimed segs statistics per each GC mode */1788 unsigned int gc_segment_mode; /* GC state for reclaimed segments */1789 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */1790 1791 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */1792 1793 int max_fragment_chunk; /* max chunk size for block fragmentation mode */1794 int max_fragment_hole; /* max hole size for block fragmentation mode */1795 1796 /* For atomic write statistics */1797 atomic64_t current_atomic_write;1798 s64 peak_atomic_write;1799 u64 committed_atomic_block;1800 u64 revoked_atomic_block;1801 1802#ifdef CONFIG_F2FS_FS_COMPRESSION1803 struct kmem_cache *page_array_slab; /* page array entry */1804 unsigned int page_array_slab_size; /* default page array slab size */1805 1806 /* For runtime compression statistics */1807 u64 compr_written_block;1808 u64 compr_saved_block;1809 u32 compr_new_inode;1810 1811 /* For compressed block cache */1812 struct inode *compress_inode; /* cache compressed blocks */1813 unsigned int compress_percent; /* cache page percentage */1814 unsigned int compress_watermark; /* cache page watermark */1815 atomic_t compress_page_hit; /* cache hit count */1816#endif1817 1818#ifdef CONFIG_F2FS_IOSTAT1819 /* For app/fs IO statistics */1820 spinlock_t iostat_lock;1821 unsigned long long iostat_count[NR_IO_TYPE];1822 unsigned long long iostat_bytes[NR_IO_TYPE];1823 unsigned long long prev_iostat_bytes[NR_IO_TYPE];1824 bool iostat_enable;1825 unsigned long iostat_next_period;1826 unsigned int iostat_period_ms;1827 1828 /* For io latency related statistics info in one iostat period */1829 spinlock_t iostat_lat_lock;1830 struct iostat_lat_info *iostat_io_lat;1831#endif1832};1833 1834/* Definitions to access f2fs_sb_info */1835#define SEGS_TO_BLKS(sbi, segs) \1836 ((segs) << (sbi)->log_blocks_per_seg)1837#define BLKS_TO_SEGS(sbi, blks) \1838 ((blks) >> (sbi)->log_blocks_per_seg)1839 1840#define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg)1841#define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec))1842#define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec)1843 1844__printf(3, 4)1845void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...);1846 1847#define f2fs_err(sbi, fmt, ...) \1848 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__)1849#define f2fs_warn(sbi, fmt, ...) \1850 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__)1851#define f2fs_notice(sbi, fmt, ...) \1852 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__)1853#define f2fs_info(sbi, fmt, ...) \1854 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__)1855#define f2fs_debug(sbi, fmt, ...) \1856 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__)1857 1858#define f2fs_err_ratelimited(sbi, fmt, ...) \1859 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__)1860#define f2fs_warn_ratelimited(sbi, fmt, ...) \1861 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__)1862#define f2fs_info_ratelimited(sbi, fmt, ...) \1863 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__)1864 1865#ifdef CONFIG_F2FS_FAULT_INJECTION1866#define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \1867 __builtin_return_address(0))1868static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type,1869 const char *func, const char *parent_func)1870{1871 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;1872 1873 if (!ffi->inject_rate)1874 return false;1875 1876 if (!IS_FAULT_SET(ffi, type))1877 return false;1878 1879 atomic_inc(&ffi->inject_ops);1880 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) {1881 atomic_set(&ffi->inject_ops, 0);1882 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS",1883 f2fs_fault_name[type], func, parent_func);1884 return true;1885 }1886 return false;1887}1888#else1889static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type)1890{1891 return false;1892}1893#endif1894 1895/*1896 * Test if the mounted volume is a multi-device volume.1897 * - For a single regular disk volume, sbi->s_ndevs is 0.1898 * - For a single zoned disk volume, sbi->s_ndevs is 1.1899 * - For a multi-device volume, sbi->s_ndevs is always 2 or more.1900 */1901static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi)1902{1903 return sbi->s_ndevs > 1;1904}1905 1906static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type)1907{1908 unsigned long now = jiffies;1909 1910 sbi->last_time[type] = now;1911 1912 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */1913 if (type == REQ_TIME) {1914 sbi->last_time[DISCARD_TIME] = now;1915 sbi->last_time[GC_TIME] = now;1916 }1917}1918 1919static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type)1920{1921 unsigned long interval = sbi->interval_time[type] * HZ;1922 1923 return time_after(jiffies, sbi->last_time[type] + interval);1924}1925 1926static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi,1927 int type)1928{1929 unsigned long interval = sbi->interval_time[type] * HZ;1930 unsigned int wait_ms = 0;1931 long delta;1932 1933 delta = (sbi->last_time[type] + interval) - jiffies;1934 if (delta > 0)1935 wait_ms = jiffies_to_msecs(delta);1936 1937 return wait_ms;1938}1939 1940/*1941 * Inline functions1942 */1943static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc,1944 const void *address, unsigned int length)1945{1946 struct {1947 struct shash_desc shash;1948 char ctx[4];1949 } desc;1950 int err;1951 1952 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx));1953 1954 desc.shash.tfm = sbi->s_chksum_driver;1955 *(u32 *)desc.ctx = crc;1956 1957 err = crypto_shash_update(&desc.shash, address, length);1958 BUG_ON(err);1959 1960 return *(u32 *)desc.ctx;1961}1962 1963static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address,1964 unsigned int length)1965{1966 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length);1967}1968 1969static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc,1970 void *buf, size_t buf_size)1971{1972 return f2fs_crc32(sbi, buf, buf_size) == blk_crc;1973}1974 1975static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc,1976 const void *address, unsigned int length)1977{1978 return __f2fs_crc32(sbi, crc, address, length);1979}1980 1981static inline struct f2fs_inode_info *F2FS_I(struct inode *inode)1982{1983 return container_of(inode, struct f2fs_inode_info, vfs_inode);1984}1985 1986static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb)1987{1988 return sb->s_fs_info;1989}1990 1991static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode)1992{1993 return F2FS_SB(inode->i_sb);1994}1995 1996static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping)1997{1998 return F2FS_I_SB(mapping->host);1999}2000 2001static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page)2002{2003 return F2FS_M_SB(page_file_mapping(page));2004}2005 2006static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi)2007{2008 return (struct f2fs_super_block *)(sbi->raw_super);2009}2010 2011static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio,2012 pgoff_t index)2013{2014 pgoff_t idx_in_folio = index % (1 << folio_order(folio));2015 2016 return (struct f2fs_super_block *)2017 (page_address(folio_page(folio, idx_in_folio)) +2018 F2FS_SUPER_OFFSET);2019}2020 2021static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi)2022{2023 return (struct f2fs_checkpoint *)(sbi->ckpt);2024}2025 2026static inline struct f2fs_node *F2FS_NODE(struct page *page)2027{2028 return (struct f2fs_node *)page_address(page);2029}2030 2031static inline struct f2fs_inode *F2FS_INODE(struct page *page)2032{2033 return &((struct f2fs_node *)page_address(page))->i;2034}2035 2036static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi)2037{2038 return (struct f2fs_nm_info *)(sbi->nm_info);2039}2040 2041static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi)2042{2043 return (struct f2fs_sm_info *)(sbi->sm_info);2044}2045 2046static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi)2047{2048 return (struct sit_info *)(SM_I(sbi)->sit_info);2049}2050 2051static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi)2052{2053 return (struct free_segmap_info *)(SM_I(sbi)->free_info);2054}2055 2056static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi)2057{2058 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info);2059}2060 2061static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi)2062{2063 return sbi->meta_inode->i_mapping;2064}2065 2066static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi)2067{2068 return sbi->node_inode->i_mapping;2069}2070 2071static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type)2072{2073 return test_bit(type, &sbi->s_flag);2074}2075 2076static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)2077{2078 set_bit(type, &sbi->s_flag);2079}2080 2081static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type)2082{2083 clear_bit(type, &sbi->s_flag);2084}2085 2086static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp)2087{2088 return le64_to_cpu(cp->checkpoint_ver);2089}2090 2091static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type)2092{2093 if (type < F2FS_MAX_QUOTAS)2094 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]);2095 return 0;2096}2097 2098static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp)2099{2100 size_t crc_offset = le32_to_cpu(cp->checksum_offset);2101 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset)));2102}2103 2104static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)2105{2106 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);2107 2108 return ckpt_flags & f;2109}2110 2111static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)2112{2113 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f);2114}2115 2116static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)2117{2118 unsigned int ckpt_flags;2119 2120 ckpt_flags = le32_to_cpu(cp->ckpt_flags);2121 ckpt_flags |= f;2122 cp->ckpt_flags = cpu_to_le32(ckpt_flags);2123}2124 2125static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)2126{2127 unsigned long flags;2128 2129 spin_lock_irqsave(&sbi->cp_lock, flags);2130 __set_ckpt_flags(F2FS_CKPT(sbi), f);2131 spin_unlock_irqrestore(&sbi->cp_lock, flags);2132}2133 2134static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)2135{2136 unsigned int ckpt_flags;2137 2138 ckpt_flags = le32_to_cpu(cp->ckpt_flags);2139 ckpt_flags &= (~f);2140 cp->ckpt_flags = cpu_to_le32(ckpt_flags);2141}2142 2143static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f)2144{2145 unsigned long flags;2146 2147 spin_lock_irqsave(&sbi->cp_lock, flags);2148 __clear_ckpt_flags(F2FS_CKPT(sbi), f);2149 spin_unlock_irqrestore(&sbi->cp_lock, flags);2150}2151 2152#define init_f2fs_rwsem(sem) \2153do { \2154 static struct lock_class_key __key; \2155 \2156 __init_f2fs_rwsem((sem), #sem, &__key); \2157} while (0)2158 2159static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem,2160 const char *sem_name, struct lock_class_key *key)2161{2162 __init_rwsem(&sem->internal_rwsem, sem_name, key);2163#ifdef CONFIG_F2FS_UNFAIR_RWSEM2164 init_waitqueue_head(&sem->read_waiters);2165#endif2166}2167 2168static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem)2169{2170 return rwsem_is_locked(&sem->internal_rwsem);2171}2172 2173static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem)2174{2175 return rwsem_is_contended(&sem->internal_rwsem);2176}2177 2178static inline void f2fs_down_read(struct f2fs_rwsem *sem)2179{2180#ifdef CONFIG_F2FS_UNFAIR_RWSEM2181 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem));2182#else2183 down_read(&sem->internal_rwsem);2184#endif2185}2186 2187static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem)2188{2189 return down_read_trylock(&sem->internal_rwsem);2190}2191 2192static inline void f2fs_up_read(struct f2fs_rwsem *sem)2193{2194 up_read(&sem->internal_rwsem);2195}2196 2197static inline void f2fs_down_write(struct f2fs_rwsem *sem)2198{2199 down_write(&sem->internal_rwsem);2200}2201 2202#ifdef CONFIG_DEBUG_LOCK_ALLOC2203static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass)2204{2205 down_read_nested(&sem->internal_rwsem, subclass);2206}2207 2208static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass)2209{2210 down_write_nested(&sem->internal_rwsem, subclass);2211}2212#else2213#define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem)2214#define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem)2215#endif2216 2217static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem)2218{2219 return down_write_trylock(&sem->internal_rwsem);2220}2221 2222static inline void f2fs_up_write(struct f2fs_rwsem *sem)2223{2224 up_write(&sem->internal_rwsem);2225#ifdef CONFIG_F2FS_UNFAIR_RWSEM2226 wake_up_all(&sem->read_waiters);2227#endif2228}2229 2230static inline void f2fs_lock_op(struct f2fs_sb_info *sbi)2231{2232 f2fs_down_read(&sbi->cp_rwsem);2233}2234 2235static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi)2236{2237 if (time_to_inject(sbi, FAULT_LOCK_OP))2238 return 0;2239 return f2fs_down_read_trylock(&sbi->cp_rwsem);2240}2241 2242static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi)2243{2244 f2fs_up_read(&sbi->cp_rwsem);2245}2246 2247static inline void f2fs_lock_all(struct f2fs_sb_info *sbi)2248{2249 f2fs_down_write(&sbi->cp_rwsem);2250}2251 2252static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi)2253{2254 f2fs_up_write(&sbi->cp_rwsem);2255}2256 2257static inline int __get_cp_reason(struct f2fs_sb_info *sbi)2258{2259 int reason = CP_SYNC;2260 2261 if (test_opt(sbi, FASTBOOT))2262 reason = CP_FASTBOOT;2263 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE))2264 reason = CP_UMOUNT;2265 return reason;2266}2267 2268static inline bool __remain_node_summaries(int reason)2269{2270 return (reason & (CP_UMOUNT | CP_FASTBOOT));2271}2272 2273static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi)2274{2275 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) ||2276 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG));2277}2278 2279/*2280 * Check whether the inode has blocks or not2281 */2282static inline int F2FS_HAS_BLOCKS(struct inode *inode)2283{2284 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0;2285 2286 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block;2287}2288 2289static inline bool f2fs_has_xattr_block(unsigned int ofs)2290{2291 return ofs == XATTR_NODE_OFFSET;2292}2293 2294static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi,2295 struct inode *inode, bool cap)2296{2297 if (!inode)2298 return true;2299 if (!test_opt(sbi, RESERVE_ROOT))2300 return false;2301 if (IS_NOQUOTA(inode))2302 return true;2303 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid()))2304 return true;2305 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) &&2306 in_group_p(F2FS_OPTION(sbi).s_resgid))2307 return true;2308 if (cap && capable(CAP_SYS_RESOURCE))2309 return true;2310 return false;2311}2312 2313static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi,2314 struct inode *inode, bool cap)2315{2316 block_t avail_user_block_count;2317 2318 avail_user_block_count = sbi->user_block_count -2319 sbi->current_reserved_blocks;2320 2321 if (!__allow_reserved_blocks(sbi, inode, cap))2322 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks;2323 2324 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {2325 if (avail_user_block_count > sbi->unusable_block_count)2326 avail_user_block_count -= sbi->unusable_block_count;2327 else2328 avail_user_block_count = 0;2329 }2330 2331 return avail_user_block_count;2332}2333 2334static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool);2335static inline int inc_valid_block_count(struct f2fs_sb_info *sbi,2336 struct inode *inode, blkcnt_t *count, bool partial)2337{2338 long long diff = 0, release = 0;2339 block_t avail_user_block_count;2340 int ret;2341 2342 ret = dquot_reserve_block(inode, *count);2343 if (ret)2344 return ret;2345 2346 if (time_to_inject(sbi, FAULT_BLOCK)) {2347 release = *count;2348 goto release_quota;2349 }2350 2351 /*2352 * let's increase this in prior to actual block count change in order2353 * for f2fs_sync_file to avoid data races when deciding checkpoint.2354 */2355 percpu_counter_add(&sbi->alloc_valid_block_count, (*count));2356 2357 spin_lock(&sbi->stat_lock);2358 2359 avail_user_block_count = get_available_block_count(sbi, inode, true);2360 diff = (long long)sbi->total_valid_block_count + *count -2361 avail_user_block_count;2362 if (unlikely(diff > 0)) {2363 if (!partial) {2364 spin_unlock(&sbi->stat_lock);2365 release = *count;2366 goto enospc;2367 }2368 if (diff > *count)2369 diff = *count;2370 *count -= diff;2371 release = diff;2372 if (!*count) {2373 spin_unlock(&sbi->stat_lock);2374 goto enospc;2375 }2376 }2377 sbi->total_valid_block_count += (block_t)(*count);2378 2379 spin_unlock(&sbi->stat_lock);2380 2381 if (unlikely(release)) {2382 percpu_counter_sub(&sbi->alloc_valid_block_count, release);2383 dquot_release_reservation_block(inode, release);2384 }2385 f2fs_i_blocks_write(inode, *count, true, true);2386 return 0;2387 2388enospc:2389 percpu_counter_sub(&sbi->alloc_valid_block_count, release);2390release_quota:2391 dquot_release_reservation_block(inode, release);2392 return -ENOSPC;2393}2394 2395#define PAGE_PRIVATE_GET_FUNC(name, flagname) \2396static inline bool page_private_##name(struct page *page) \2397{ \2398 return PagePrivate(page) && \2399 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \2400 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \2401}2402 2403#define PAGE_PRIVATE_SET_FUNC(name, flagname) \2404static inline void set_page_private_##name(struct page *page) \2405{ \2406 if (!PagePrivate(page)) \2407 attach_page_private(page, (void *)0); \2408 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \2409 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \2410}2411 2412#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \2413static inline void clear_page_private_##name(struct page *page) \2414{ \2415 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \2416 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \2417 detach_page_private(page); \2418}2419 2420PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER);2421PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE);2422PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION);2423PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE);2424 2425PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE);2426PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE);2427PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION);2428PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE);2429 2430PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE);2431PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE);2432PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION);2433PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE);2434 2435static inline unsigned long get_page_private_data(struct page *page)2436{2437 unsigned long data = page_private(page);2438 2439 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data))2440 return 0;2441 return data >> PAGE_PRIVATE_MAX;2442}2443 2444static inline void set_page_private_data(struct page *page, unsigned long data)2445{2446 if (!PagePrivate(page))2447 attach_page_private(page, (void *)0);2448 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page));2449 page_private(page) |= data << PAGE_PRIVATE_MAX;2450}2451 2452static inline void clear_page_private_data(struct page *page)2453{2454 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0);2455 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER))2456 detach_page_private(page);2457}2458 2459static inline void clear_page_private_all(struct page *page)2460{2461 clear_page_private_data(page);2462 clear_page_private_reference(page);2463 clear_page_private_gcing(page);2464 clear_page_private_inline(page);2465 clear_page_private_atomic(page);2466 2467 f2fs_bug_on(F2FS_P_SB(page), page_private(page));2468}2469 2470static inline void dec_valid_block_count(struct f2fs_sb_info *sbi,2471 struct inode *inode,2472 block_t count)2473{2474 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK;2475 2476 spin_lock(&sbi->stat_lock);2477 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count);2478 sbi->total_valid_block_count -= (block_t)count;2479 if (sbi->reserved_blocks &&2480 sbi->current_reserved_blocks < sbi->reserved_blocks)2481 sbi->current_reserved_blocks = min(sbi->reserved_blocks,2482 sbi->current_reserved_blocks + count);2483 spin_unlock(&sbi->stat_lock);2484 if (unlikely(inode->i_blocks < sectors)) {2485 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu",2486 inode->i_ino,2487 (unsigned long long)inode->i_blocks,2488 (unsigned long long)sectors);2489 set_sbi_flag(sbi, SBI_NEED_FSCK);2490 return;2491 }2492 f2fs_i_blocks_write(inode, count, false, true);2493}2494 2495static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type)2496{2497 atomic_inc(&sbi->nr_pages[count_type]);2498 2499 if (count_type == F2FS_DIRTY_DENTS ||2500 count_type == F2FS_DIRTY_NODES ||2501 count_type == F2FS_DIRTY_META ||2502 count_type == F2FS_DIRTY_QDATA ||2503 count_type == F2FS_DIRTY_IMETA)2504 set_sbi_flag(sbi, SBI_IS_DIRTY);2505}2506 2507static inline void inode_inc_dirty_pages(struct inode *inode)2508{2509 atomic_inc(&F2FS_I(inode)->dirty_pages);2510 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?2511 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);2512 if (IS_NOQUOTA(inode))2513 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);2514}2515 2516static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type)2517{2518 atomic_dec(&sbi->nr_pages[count_type]);2519}2520 2521static inline void inode_dec_dirty_pages(struct inode *inode)2522{2523 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&2524 !S_ISLNK(inode->i_mode))2525 return;2526 2527 atomic_dec(&F2FS_I(inode)->dirty_pages);2528 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ?2529 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA);2530 if (IS_NOQUOTA(inode))2531 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA);2532}2533 2534static inline void inc_atomic_write_cnt(struct inode *inode)2535{2536 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);2537 struct f2fs_inode_info *fi = F2FS_I(inode);2538 u64 current_write;2539 2540 fi->atomic_write_cnt++;2541 atomic64_inc(&sbi->current_atomic_write);2542 current_write = atomic64_read(&sbi->current_atomic_write);2543 if (current_write > sbi->peak_atomic_write)2544 sbi->peak_atomic_write = current_write;2545}2546 2547static inline void release_atomic_write_cnt(struct inode *inode)2548{2549 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);2550 struct f2fs_inode_info *fi = F2FS_I(inode);2551 2552 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write);2553 fi->atomic_write_cnt = 0;2554}2555 2556static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type)2557{2558 return atomic_read(&sbi->nr_pages[count_type]);2559}2560 2561static inline int get_dirty_pages(struct inode *inode)2562{2563 return atomic_read(&F2FS_I(inode)->dirty_pages);2564}2565 2566static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type)2567{2568 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1,2569 BLKS_PER_SEC(sbi));2570}2571 2572static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi)2573{2574 return sbi->total_valid_block_count;2575}2576 2577static inline block_t discard_blocks(struct f2fs_sb_info *sbi)2578{2579 return sbi->discard_blks;2580}2581 2582static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)2583{2584 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);2585 2586 /* return NAT or SIT bitmap */2587 if (flag == NAT_BITMAP)2588 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);2589 else if (flag == SIT_BITMAP)2590 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);2591 2592 return 0;2593}2594 2595static inline block_t __cp_payload(struct f2fs_sb_info *sbi)2596{2597 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload);2598}2599 2600static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)2601{2602 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);2603 void *tmp_ptr = &ckpt->sit_nat_version_bitmap;2604 int offset;2605 2606 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) {2607 offset = (flag == SIT_BITMAP) ?2608 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;2609 /*2610 * if large_nat_bitmap feature is enabled, leave checksum2611 * protection for all nat/sit bitmaps.2612 */2613 return tmp_ptr + offset + sizeof(__le32);2614 }2615 2616 if (__cp_payload(sbi) > 0) {2617 if (flag == NAT_BITMAP)2618 return tmp_ptr;2619 else2620 return (unsigned char *)ckpt + F2FS_BLKSIZE;2621 } else {2622 offset = (flag == NAT_BITMAP) ?2623 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0;2624 return tmp_ptr + offset;2625 }2626}2627 2628static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)2629{2630 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);2631 2632 if (sbi->cur_cp_pack == 2)2633 start_addr += BLKS_PER_SEG(sbi);2634 return start_addr;2635}2636 2637static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi)2638{2639 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);2640 2641 if (sbi->cur_cp_pack == 1)2642 start_addr += BLKS_PER_SEG(sbi);2643 return start_addr;2644}2645 2646static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi)2647{2648 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1;2649}2650 2651static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi)2652{2653 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum);2654}2655 2656extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync);2657static inline int inc_valid_node_count(struct f2fs_sb_info *sbi,2658 struct inode *inode, bool is_inode)2659{2660 block_t valid_block_count;2661 unsigned int valid_node_count;2662 unsigned int avail_user_block_count;2663 int err;2664 2665 if (is_inode) {2666 if (inode) {2667 err = dquot_alloc_inode(inode);2668 if (err)2669 return err;2670 }2671 } else {2672 err = dquot_reserve_block(inode, 1);2673 if (err)2674 return err;2675 }2676 2677 if (time_to_inject(sbi, FAULT_BLOCK))2678 goto enospc;2679 2680 spin_lock(&sbi->stat_lock);2681 2682 valid_block_count = sbi->total_valid_block_count + 1;2683 avail_user_block_count = get_available_block_count(sbi, inode, false);2684 2685 if (unlikely(valid_block_count > avail_user_block_count)) {2686 spin_unlock(&sbi->stat_lock);2687 goto enospc;2688 }2689 2690 valid_node_count = sbi->total_valid_node_count + 1;2691 if (unlikely(valid_node_count > sbi->total_node_count)) {2692 spin_unlock(&sbi->stat_lock);2693 goto enospc;2694 }2695 2696 sbi->total_valid_node_count++;2697 sbi->total_valid_block_count++;2698 spin_unlock(&sbi->stat_lock);2699 2700 if (inode) {2701 if (is_inode)2702 f2fs_mark_inode_dirty_sync(inode, true);2703 else2704 f2fs_i_blocks_write(inode, 1, true, true);2705 }2706 2707 percpu_counter_inc(&sbi->alloc_valid_block_count);2708 return 0;2709 2710enospc:2711 if (is_inode) {2712 if (inode)2713 dquot_free_inode(inode);2714 } else {2715 dquot_release_reservation_block(inode, 1);2716 }2717 return -ENOSPC;2718}2719 2720static inline void dec_valid_node_count(struct f2fs_sb_info *sbi,2721 struct inode *inode, bool is_inode)2722{2723 spin_lock(&sbi->stat_lock);2724 2725 if (unlikely(!sbi->total_valid_block_count ||2726 !sbi->total_valid_node_count)) {2727 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u",2728 sbi->total_valid_block_count,2729 sbi->total_valid_node_count);2730 set_sbi_flag(sbi, SBI_NEED_FSCK);2731 } else {2732 sbi->total_valid_block_count--;2733 sbi->total_valid_node_count--;2734 }2735 2736 if (sbi->reserved_blocks &&2737 sbi->current_reserved_blocks < sbi->reserved_blocks)2738 sbi->current_reserved_blocks++;2739 2740 spin_unlock(&sbi->stat_lock);2741 2742 if (is_inode) {2743 dquot_free_inode(inode);2744 } else {2745 if (unlikely(inode->i_blocks == 0)) {2746 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu",2747 inode->i_ino,2748 (unsigned long long)inode->i_blocks);2749 set_sbi_flag(sbi, SBI_NEED_FSCK);2750 return;2751 }2752 f2fs_i_blocks_write(inode, 1, false, true);2753 }2754}2755 2756static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi)2757{2758 return sbi->total_valid_node_count;2759}2760 2761static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi)2762{2763 percpu_counter_inc(&sbi->total_valid_inode_count);2764}2765 2766static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi)2767{2768 percpu_counter_dec(&sbi->total_valid_inode_count);2769}2770 2771static inline s64 valid_inode_count(struct f2fs_sb_info *sbi)2772{2773 return percpu_counter_sum_positive(&sbi->total_valid_inode_count);2774}2775 2776static inline struct page *f2fs_grab_cache_page(struct address_space *mapping,2777 pgoff_t index, bool for_write)2778{2779 struct page *page;2780 unsigned int flags;2781 2782 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) {2783 if (!for_write)2784 page = find_get_page_flags(mapping, index,2785 FGP_LOCK | FGP_ACCESSED);2786 else2787 page = find_lock_page(mapping, index);2788 if (page)2789 return page;2790 2791 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC))2792 return NULL;2793 }2794 2795 if (!for_write)2796 return grab_cache_page(mapping, index);2797 2798 flags = memalloc_nofs_save();2799 page = grab_cache_page_write_begin(mapping, index);2800 memalloc_nofs_restore(flags);2801 2802 return page;2803}2804 2805static inline struct page *f2fs_pagecache_get_page(2806 struct address_space *mapping, pgoff_t index,2807 fgf_t fgp_flags, gfp_t gfp_mask)2808{2809 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET))2810 return NULL;2811 2812 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask);2813}2814 2815static inline void f2fs_put_page(struct page *page, int unlock)2816{2817 if (!page)2818 return;2819 2820 if (unlock) {2821 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page));2822 unlock_page(page);2823 }2824 put_page(page);2825}2826 2827static inline void f2fs_put_dnode(struct dnode_of_data *dn)2828{2829 if (dn->node_page)2830 f2fs_put_page(dn->node_page, 1);2831 if (dn->inode_page && dn->node_page != dn->inode_page)2832 f2fs_put_page(dn->inode_page, 0);2833 dn->node_page = NULL;2834 dn->inode_page = NULL;2835}2836 2837static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name,2838 size_t size)2839{2840 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL);2841}2842 2843static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep,2844 gfp_t flags)2845{2846 void *entry;2847 2848 entry = kmem_cache_alloc(cachep, flags);2849 if (!entry)2850 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL);2851 return entry;2852}2853 2854static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep,2855 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi)2856{2857 if (nofail)2858 return f2fs_kmem_cache_alloc_nofail(cachep, flags);2859 2860 if (time_to_inject(sbi, FAULT_SLAB_ALLOC))2861 return NULL;2862 2863 return kmem_cache_alloc(cachep, flags);2864}2865 2866static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type)2867{2868 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) ||2869 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) ||2870 get_pages(sbi, F2FS_WB_CP_DATA) ||2871 get_pages(sbi, F2FS_DIO_READ) ||2872 get_pages(sbi, F2FS_DIO_WRITE))2873 return true;2874 2875 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info &&2876 atomic_read(&SM_I(sbi)->dcc_info->queued_discard))2877 return true;2878 2879 if (SM_I(sbi) && SM_I(sbi)->fcc_info &&2880 atomic_read(&SM_I(sbi)->fcc_info->queued_flush))2881 return true;2882 return false;2883}2884 2885static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi)2886{2887 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ);2888}2889 2890static inline bool is_idle(struct f2fs_sb_info *sbi, int type)2891{2892 bool zoned_gc = (type == GC_TIME &&2893 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED));2894 2895 if (sbi->gc_mode == GC_URGENT_HIGH)2896 return true;2897 2898 if (zoned_gc) {2899 if (is_inflight_read_io(sbi))2900 return false;2901 } else {2902 if (is_inflight_io(sbi, type))2903 return false;2904 }2905 2906 if (sbi->gc_mode == GC_URGENT_MID)2907 return true;2908 2909 if (sbi->gc_mode == GC_URGENT_LOW &&2910 (type == DISCARD_TIME || type == GC_TIME))2911 return true;2912 2913 if (zoned_gc)2914 return true;2915 2916 return f2fs_time_over(sbi, type);2917}2918 2919static inline void f2fs_radix_tree_insert(struct radix_tree_root *root,2920 unsigned long index, void *item)2921{2922 while (radix_tree_insert(root, index, item))2923 cond_resched();2924}2925 2926#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino)2927 2928static inline bool IS_INODE(struct page *page)2929{2930 struct f2fs_node *p = F2FS_NODE(page);2931 2932 return RAW_IS_INODE(p);2933}2934 2935static inline int offset_in_addr(struct f2fs_inode *i)2936{2937 return (i->i_inline & F2FS_EXTRA_ATTR) ?2938 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0;2939}2940 2941static inline __le32 *blkaddr_in_node(struct f2fs_node *node)2942{2943 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr;2944}2945 2946static inline int f2fs_has_extra_attr(struct inode *inode);2947static inline unsigned int get_dnode_base(struct inode *inode,2948 struct page *node_page)2949{2950 if (!IS_INODE(node_page))2951 return 0;2952 2953 return inode ? get_extra_isize(inode) :2954 offset_in_addr(&F2FS_NODE(node_page)->i);2955}2956 2957static inline __le32 *get_dnode_addr(struct inode *inode,2958 struct page *node_page)2959{2960 return blkaddr_in_node(F2FS_NODE(node_page)) +2961 get_dnode_base(inode, node_page);2962}2963 2964static inline block_t data_blkaddr(struct inode *inode,2965 struct page *node_page, unsigned int offset)2966{2967 return le32_to_cpu(*(get_dnode_addr(inode, node_page) + offset));2968}2969 2970static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn)2971{2972 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node);2973}2974 2975static inline int f2fs_test_bit(unsigned int nr, char *addr)2976{2977 int mask;2978 2979 addr += (nr >> 3);2980 mask = BIT(7 - (nr & 0x07));2981 return mask & *addr;2982}2983 2984static inline void f2fs_set_bit(unsigned int nr, char *addr)2985{2986 int mask;2987 2988 addr += (nr >> 3);2989 mask = BIT(7 - (nr & 0x07));2990 *addr |= mask;2991}2992 2993static inline void f2fs_clear_bit(unsigned int nr, char *addr)2994{2995 int mask;2996 2997 addr += (nr >> 3);2998 mask = BIT(7 - (nr & 0x07));2999 *addr &= ~mask;3000}3001 3002static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr)3003{3004 int mask;3005 int ret;3006 3007 addr += (nr >> 3);3008 mask = BIT(7 - (nr & 0x07));3009 ret = mask & *addr;3010 *addr |= mask;3011 return ret;3012}3013 3014static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr)3015{3016 int mask;3017 int ret;3018 3019 addr += (nr >> 3);3020 mask = BIT(7 - (nr & 0x07));3021 ret = mask & *addr;3022 *addr &= ~mask;3023 return ret;3024}3025 3026static inline void f2fs_change_bit(unsigned int nr, char *addr)3027{3028 int mask;3029 3030 addr += (nr >> 3);3031 mask = BIT(7 - (nr & 0x07));3032 *addr ^= mask;3033}3034 3035/*3036 * On-disk inode flags (f2fs_inode::i_flags)3037 */3038#define F2FS_COMPR_FL 0x00000004 /* Compress file */3039#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */3040#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */3041#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */3042#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */3043#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */3044#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */3045#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */3046#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */3047#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */3048#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */3049 3050#define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL)3051 3052/* Flags that should be inherited by new inodes from their parent. */3053#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \3054 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \3055 F2FS_CASEFOLD_FL)3056 3057/* Flags that are appropriate for regular files (all but dir-specific ones). */3058#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \3059 F2FS_CASEFOLD_FL))3060 3061/* Flags that are appropriate for non-directories/regular files. */3062#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL)3063 3064static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags)3065{3066 if (S_ISDIR(mode))3067 return flags;3068 else if (S_ISREG(mode))3069 return flags & F2FS_REG_FLMASK;3070 else3071 return flags & F2FS_OTHER_FLMASK;3072}3073 3074static inline void __mark_inode_dirty_flag(struct inode *inode,3075 int flag, bool set)3076{3077 switch (flag) {3078 case FI_INLINE_XATTR:3079 case FI_INLINE_DATA:3080 case FI_INLINE_DENTRY:3081 case FI_NEW_INODE:3082 if (set)3083 return;3084 fallthrough;3085 case FI_DATA_EXIST:3086 case FI_PIN_FILE:3087 case FI_COMPRESS_RELEASED:3088 f2fs_mark_inode_dirty_sync(inode, true);3089 }3090}3091 3092static inline void set_inode_flag(struct inode *inode, int flag)3093{3094 set_bit(flag, F2FS_I(inode)->flags);3095 __mark_inode_dirty_flag(inode, flag, true);3096}3097 3098static inline int is_inode_flag_set(struct inode *inode, int flag)3099{3100 return test_bit(flag, F2FS_I(inode)->flags);3101}3102 3103static inline void clear_inode_flag(struct inode *inode, int flag)3104{3105 clear_bit(flag, F2FS_I(inode)->flags);3106 __mark_inode_dirty_flag(inode, flag, false);3107}3108 3109static inline bool f2fs_verity_in_progress(struct inode *inode)3110{3111 return IS_ENABLED(CONFIG_FS_VERITY) &&3112 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS);3113}3114 3115static inline void set_acl_inode(struct inode *inode, umode_t mode)3116{3117 F2FS_I(inode)->i_acl_mode = mode;3118 set_inode_flag(inode, FI_ACL_MODE);3119 f2fs_mark_inode_dirty_sync(inode, false);3120}3121 3122static inline void f2fs_i_links_write(struct inode *inode, bool inc)3123{3124 if (inc)3125 inc_nlink(inode);3126 else3127 drop_nlink(inode);3128 f2fs_mark_inode_dirty_sync(inode, true);3129}3130 3131static inline void f2fs_i_blocks_write(struct inode *inode,3132 block_t diff, bool add, bool claim)3133{3134 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);3135 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);3136 3137 /* add = 1, claim = 1 should be dquot_reserve_block in pair */3138 if (add) {3139 if (claim)3140 dquot_claim_block(inode, diff);3141 else3142 dquot_alloc_block_nofail(inode, diff);3143 } else {3144 dquot_free_block(inode, diff);3145 }3146 3147 f2fs_mark_inode_dirty_sync(inode, true);3148 if (clean || recover)3149 set_inode_flag(inode, FI_AUTO_RECOVER);3150}3151 3152static inline bool f2fs_is_atomic_file(struct inode *inode);3153 3154static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size)3155{3156 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE);3157 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER);3158 3159 if (i_size_read(inode) == i_size)3160 return;3161 3162 i_size_write(inode, i_size);3163 3164 if (f2fs_is_atomic_file(inode))3165 return;3166 3167 f2fs_mark_inode_dirty_sync(inode, true);3168 if (clean || recover)3169 set_inode_flag(inode, FI_AUTO_RECOVER);3170}3171 3172static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth)3173{3174 F2FS_I(inode)->i_current_depth = depth;3175 f2fs_mark_inode_dirty_sync(inode, true);3176}3177 3178static inline void f2fs_i_gc_failures_write(struct inode *inode,3179 unsigned int count)3180{3181 F2FS_I(inode)->i_gc_failures = count;3182 f2fs_mark_inode_dirty_sync(inode, true);3183}3184 3185static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid)3186{3187 F2FS_I(inode)->i_xattr_nid = xnid;3188 f2fs_mark_inode_dirty_sync(inode, true);3189}3190 3191static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino)3192{3193 F2FS_I(inode)->i_pino = pino;3194 f2fs_mark_inode_dirty_sync(inode, true);3195}3196 3197static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri)3198{3199 struct f2fs_inode_info *fi = F2FS_I(inode);3200 3201 if (ri->i_inline & F2FS_INLINE_XATTR)3202 set_bit(FI_INLINE_XATTR, fi->flags);3203 if (ri->i_inline & F2FS_INLINE_DATA)3204 set_bit(FI_INLINE_DATA, fi->flags);3205 if (ri->i_inline & F2FS_INLINE_DENTRY)3206 set_bit(FI_INLINE_DENTRY, fi->flags);3207 if (ri->i_inline & F2FS_DATA_EXIST)3208 set_bit(FI_DATA_EXIST, fi->flags);3209 if (ri->i_inline & F2FS_EXTRA_ATTR)3210 set_bit(FI_EXTRA_ATTR, fi->flags);3211 if (ri->i_inline & F2FS_PIN_FILE)3212 set_bit(FI_PIN_FILE, fi->flags);3213 if (ri->i_inline & F2FS_COMPRESS_RELEASED)3214 set_bit(FI_COMPRESS_RELEASED, fi->flags);3215}3216 3217static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri)3218{3219 ri->i_inline = 0;3220 3221 if (is_inode_flag_set(inode, FI_INLINE_XATTR))3222 ri->i_inline |= F2FS_INLINE_XATTR;3223 if (is_inode_flag_set(inode, FI_INLINE_DATA))3224 ri->i_inline |= F2FS_INLINE_DATA;3225 if (is_inode_flag_set(inode, FI_INLINE_DENTRY))3226 ri->i_inline |= F2FS_INLINE_DENTRY;3227 if (is_inode_flag_set(inode, FI_DATA_EXIST))3228 ri->i_inline |= F2FS_DATA_EXIST;3229 if (is_inode_flag_set(inode, FI_EXTRA_ATTR))3230 ri->i_inline |= F2FS_EXTRA_ATTR;3231 if (is_inode_flag_set(inode, FI_PIN_FILE))3232 ri->i_inline |= F2FS_PIN_FILE;3233 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED))3234 ri->i_inline |= F2FS_COMPRESS_RELEASED;3235}3236 3237static inline int f2fs_has_extra_attr(struct inode *inode)3238{3239 return is_inode_flag_set(inode, FI_EXTRA_ATTR);3240}3241 3242static inline int f2fs_has_inline_xattr(struct inode *inode)3243{3244 return is_inode_flag_set(inode, FI_INLINE_XATTR);3245}3246 3247static inline int f2fs_compressed_file(struct inode *inode)3248{3249 return S_ISREG(inode->i_mode) &&3250 is_inode_flag_set(inode, FI_COMPRESSED_FILE);3251}3252 3253static inline bool f2fs_need_compress_data(struct inode *inode)3254{3255 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;3256 3257 if (!f2fs_compressed_file(inode))3258 return false;3259 3260 if (compress_mode == COMPR_MODE_FS)3261 return true;3262 else if (compress_mode == COMPR_MODE_USER &&3263 is_inode_flag_set(inode, FI_ENABLE_COMPRESS))3264 return true;3265 3266 return false;3267}3268 3269static inline unsigned int addrs_per_page(struct inode *inode,3270 bool is_inode)3271{3272 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) -3273 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK;3274 3275 if (f2fs_compressed_file(inode))3276 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size);3277 return addrs;3278}3279 3280static inline void *inline_xattr_addr(struct inode *inode, struct page *page)3281{3282 struct f2fs_inode *ri = F2FS_INODE(page);3283 3284 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE -3285 get_inline_xattr_addrs(inode)]);3286}3287 3288static inline int inline_xattr_size(struct inode *inode)3289{3290 if (f2fs_has_inline_xattr(inode))3291 return get_inline_xattr_addrs(inode) * sizeof(__le32);3292 return 0;3293}3294 3295/*3296 * Notice: check inline_data flag without inode page lock is unsafe.3297 * It could change at any time by f2fs_convert_inline_page().3298 */3299static inline int f2fs_has_inline_data(struct inode *inode)3300{3301 return is_inode_flag_set(inode, FI_INLINE_DATA);3302}3303 3304static inline int f2fs_exist_data(struct inode *inode)3305{3306 return is_inode_flag_set(inode, FI_DATA_EXIST);3307}3308 3309static inline int f2fs_is_mmap_file(struct inode *inode)3310{3311 return is_inode_flag_set(inode, FI_MMAP_FILE);3312}3313 3314static inline bool f2fs_is_pinned_file(struct inode *inode)3315{3316 return is_inode_flag_set(inode, FI_PIN_FILE);3317}3318 3319static inline bool f2fs_is_atomic_file(struct inode *inode)3320{3321 return is_inode_flag_set(inode, FI_ATOMIC_FILE);3322}3323 3324static inline bool f2fs_is_cow_file(struct inode *inode)3325{3326 return is_inode_flag_set(inode, FI_COW_FILE);3327}3328 3329static inline void *inline_data_addr(struct inode *inode, struct page *page)3330{3331 __le32 *addr = get_dnode_addr(inode, page);3332 3333 return (void *)(addr + DEF_INLINE_RESERVED_SIZE);3334}3335 3336static inline int f2fs_has_inline_dentry(struct inode *inode)3337{3338 return is_inode_flag_set(inode, FI_INLINE_DENTRY);3339}3340 3341static inline int is_file(struct inode *inode, int type)3342{3343 return F2FS_I(inode)->i_advise & type;3344}3345 3346static inline void set_file(struct inode *inode, int type)3347{3348 if (is_file(inode, type))3349 return;3350 F2FS_I(inode)->i_advise |= type;3351 f2fs_mark_inode_dirty_sync(inode, true);3352}3353 3354static inline void clear_file(struct inode *inode, int type)3355{3356 if (!is_file(inode, type))3357 return;3358 F2FS_I(inode)->i_advise &= ~type;3359 f2fs_mark_inode_dirty_sync(inode, true);3360}3361 3362static inline bool f2fs_is_time_consistent(struct inode *inode)3363{3364 struct timespec64 ts = inode_get_atime(inode);3365 3366 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts))3367 return false;3368 ts = inode_get_ctime(inode);3369 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts))3370 return false;3371 ts = inode_get_mtime(inode);3372 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts))3373 return false;3374 return true;3375}3376 3377static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)3378{3379 bool ret;3380 3381 if (dsync) {3382 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);3383 3384 spin_lock(&sbi->inode_lock[DIRTY_META]);3385 ret = list_empty(&F2FS_I(inode)->gdirty_list);3386 spin_unlock(&sbi->inode_lock[DIRTY_META]);3387 return ret;3388 }3389 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) ||3390 file_keep_isize(inode) ||3391 i_size_read(inode) & ~PAGE_MASK)3392 return false;3393 3394 if (!f2fs_is_time_consistent(inode))3395 return false;3396 3397 spin_lock(&F2FS_I(inode)->i_size_lock);3398 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);3399 spin_unlock(&F2FS_I(inode)->i_size_lock);3400 3401 return ret;3402}3403 3404static inline bool f2fs_readonly(struct super_block *sb)3405{3406 return sb_rdonly(sb);3407}3408 3409static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi)3410{3411 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG);3412}3413 3414static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi,3415 size_t size, gfp_t flags)3416{3417 if (time_to_inject(sbi, FAULT_KMALLOC))3418 return NULL;3419 3420 return kmalloc(size, flags);3421}3422 3423static inline void *f2fs_getname(struct f2fs_sb_info *sbi)3424{3425 if (time_to_inject(sbi, FAULT_KMALLOC))3426 return NULL;3427 3428 return __getname();3429}3430 3431static inline void f2fs_putname(char *buf)3432{3433 __putname(buf);3434}3435 3436static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi,3437 size_t size, gfp_t flags)3438{3439 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO);3440}3441 3442static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi,3443 size_t size, gfp_t flags)3444{3445 if (time_to_inject(sbi, FAULT_KVMALLOC))3446 return NULL;3447 3448 return kvmalloc(size, flags);3449}3450 3451static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi,3452 size_t size, gfp_t flags)3453{3454 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO);3455}3456 3457static inline int get_extra_isize(struct inode *inode)3458{3459 return F2FS_I(inode)->i_extra_isize / sizeof(__le32);3460}3461 3462static inline int get_inline_xattr_addrs(struct inode *inode)3463{3464 return F2FS_I(inode)->i_inline_xattr_size;3465}3466 3467#define f2fs_get_inode_mode(i) \3468 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \3469 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode))3470 3471#define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32))3472 3473#define F2FS_TOTAL_EXTRA_ATTR_SIZE \3474 (offsetof(struct f2fs_inode, i_extra_end) - \3475 offsetof(struct f2fs_inode, i_extra_isize)) \3476 3477#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr))3478#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \3479 ((offsetof(typeof(*(f2fs_inode)), field) + \3480 sizeof((f2fs_inode)->field)) \3481 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \3482 3483#define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1)3484 3485#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META)3486 3487bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,3488 block_t blkaddr, int type);3489static inline void verify_blkaddr(struct f2fs_sb_info *sbi,3490 block_t blkaddr, int type)3491{3492 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type))3493 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.",3494 blkaddr, type);3495}3496 3497static inline bool __is_valid_data_blkaddr(block_t blkaddr)3498{3499 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR ||3500 blkaddr == COMPRESS_ADDR)3501 return false;3502 return true;3503}3504 3505/*3506 * file.c3507 */3508int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync);3509int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock);3510int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock);3511int f2fs_truncate(struct inode *inode);3512int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path,3513 struct kstat *stat, u32 request_mask, unsigned int flags);3514int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,3515 struct iattr *attr);3516int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end);3517void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count);3518int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag,3519 bool readonly, bool need_lock);3520int f2fs_precache_extents(struct inode *inode);3521int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa);3522int f2fs_fileattr_set(struct mnt_idmap *idmap,3523 struct dentry *dentry, struct fileattr *fa);3524long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);3525long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg);3526int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid);3527int f2fs_pin_file_control(struct inode *inode, bool inc);3528 3529/*3530 * inode.c3531 */3532void f2fs_set_inode_flags(struct inode *inode);3533bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page);3534void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page);3535struct inode *f2fs_iget(struct super_block *sb, unsigned long ino);3536struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino);3537int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink);3538void f2fs_update_inode(struct inode *inode, struct page *node_page);3539void f2fs_update_inode_page(struct inode *inode);3540int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc);3541void f2fs_evict_inode(struct inode *inode);3542void f2fs_handle_failed_inode(struct inode *inode);3543 3544/*3545 * namei.c3546 */3547int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name,3548 bool hot, bool set);3549struct dentry *f2fs_get_parent(struct dentry *child);3550int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir,3551 struct inode **new_inode);3552 3553/*3554 * dir.c3555 */3556#if IS_ENABLED(CONFIG_UNICODE)3557int f2fs_init_casefolded_name(const struct inode *dir,3558 struct f2fs_filename *fname);3559void f2fs_free_casefolded_name(struct f2fs_filename *fname);3560#else3561static inline int f2fs_init_casefolded_name(const struct inode *dir,3562 struct f2fs_filename *fname)3563{3564 return 0;3565}3566 3567static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname)3568{3569}3570#endif /* CONFIG_UNICODE */3571 3572int f2fs_setup_filename(struct inode *dir, const struct qstr *iname,3573 int lookup, struct f2fs_filename *fname);3574int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry,3575 struct f2fs_filename *fname);3576void f2fs_free_filename(struct f2fs_filename *fname);3577struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d,3578 const struct f2fs_filename *fname, int *max_slots);3579int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,3580 unsigned int start_pos, struct fscrypt_str *fstr);3581void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent,3582 struct f2fs_dentry_ptr *d);3583struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir,3584 const struct f2fs_filename *fname, struct page *dpage);3585void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode,3586 unsigned int current_depth);3587int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots);3588void f2fs_drop_nlink(struct inode *dir, struct inode *inode);3589struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,3590 const struct f2fs_filename *fname,3591 struct page **res_page);3592struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir,3593 const struct qstr *child, struct page **res_page);3594struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p);3595ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr,3596 struct page **page);3597void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de,3598 struct page *page, struct inode *inode);3599bool f2fs_has_enough_room(struct inode *dir, struct page *ipage,3600 const struct f2fs_filename *fname);3601void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,3602 const struct fscrypt_str *name, f2fs_hash_t name_hash,3603 unsigned int bit_pos);3604int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname,3605 struct inode *inode, nid_t ino, umode_t mode);3606int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname,3607 struct inode *inode, nid_t ino, umode_t mode);3608int f2fs_do_add_link(struct inode *dir, const struct qstr *name,3609 struct inode *inode, nid_t ino, umode_t mode);3610void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,3611 struct inode *dir, struct inode *inode);3612int f2fs_do_tmpfile(struct inode *inode, struct inode *dir,3613 struct f2fs_filename *fname);3614bool f2fs_empty_dir(struct inode *dir);3615 3616static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode)3617{3618 if (fscrypt_is_nokey_name(dentry))3619 return -ENOKEY;3620 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name,3621 inode, inode->i_ino, inode->i_mode);3622}3623 3624/*3625 * super.c3626 */3627int f2fs_inode_dirtied(struct inode *inode, bool sync);3628void f2fs_inode_synced(struct inode *inode);3629int f2fs_dquot_initialize(struct inode *inode);3630int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly);3631int f2fs_quota_sync(struct super_block *sb, int type);3632loff_t max_file_blocks(struct inode *inode);3633void f2fs_quota_off_umount(struct super_block *sb);3634void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag);3635void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,3636 bool irq_context);3637void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error);3638void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error);3639int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover);3640int f2fs_sync_fs(struct super_block *sb, int sync);3641int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi);3642 3643/*3644 * hash.c3645 */3646void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname);3647 3648/*3649 * node.c3650 */3651struct node_info;3652 3653int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid);3654bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type);3655bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page);3656void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi);3657void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page);3658void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi);3659int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid);3660bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid);3661bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino);3662int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid,3663 struct node_info *ni, bool checkpoint_context);3664pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs);3665int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode);3666int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from);3667int f2fs_truncate_xattr_node(struct inode *inode);3668int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi,3669 unsigned int seq_id);3670bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi);3671int f2fs_remove_inode_page(struct inode *inode);3672struct page *f2fs_new_inode_page(struct inode *inode);3673struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs);3674void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid);3675struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid);3676struct page *f2fs_get_node_page_ra(struct page *parent, int start);3677int f2fs_move_node_page(struct page *node_page, int gc_type);3678void f2fs_flush_inline_data(struct f2fs_sb_info *sbi);3679int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode,3680 struct writeback_control *wbc, bool atomic,3681 unsigned int *seq_id);3682int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,3683 struct writeback_control *wbc,3684 bool do_balance, enum iostat_type io_type);3685int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount);3686bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid);3687void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid);3688void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid);3689int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink);3690int f2fs_recover_inline_xattr(struct inode *inode, struct page *page);3691int f2fs_recover_xattr_data(struct inode *inode, struct page *page);3692int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page);3693int f2fs_restore_node_summary(struct f2fs_sb_info *sbi,3694 unsigned int segno, struct f2fs_summary_block *sum);3695void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi);3696int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);3697int f2fs_build_node_manager(struct f2fs_sb_info *sbi);3698void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi);3699int __init f2fs_create_node_manager_caches(void);3700void f2fs_destroy_node_manager_caches(void);3701 3702/*3703 * segment.c3704 */3705bool f2fs_need_SSR(struct f2fs_sb_info *sbi);3706int f2fs_commit_atomic_write(struct inode *inode);3707void f2fs_abort_atomic_write(struct inode *inode, bool clean);3708void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need);3709void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg);3710int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino);3711int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi);3712int f2fs_flush_device_cache(struct f2fs_sb_info *sbi);3713void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free);3714void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr);3715bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr);3716int f2fs_start_discard_thread(struct f2fs_sb_info *sbi);3717void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi);3718void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi);3719bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi);3720void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi,3721 struct cp_control *cpc);3722void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi);3723block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi);3724int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable);3725void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi);3726int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra);3727bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno);3728int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi);3729int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi);3730void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi);3731void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi);3732int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type,3733 unsigned int start, unsigned int end);3734int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force);3735int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi);3736int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi);3737int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range);3738bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi,3739 struct cp_control *cpc);3740struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno);3741void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src,3742 block_t blk_addr);3743void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio,3744 enum iostat_type io_type);3745void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio);3746void f2fs_outplace_write_data(struct dnode_of_data *dn,3747 struct f2fs_io_info *fio);3748int f2fs_inplace_write_data(struct f2fs_io_info *fio);3749void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,3750 block_t old_blkaddr, block_t new_blkaddr,3751 bool recover_curseg, bool recover_newaddr,3752 bool from_gc);3753void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn,3754 block_t old_addr, block_t new_addr,3755 unsigned char version, bool recover_curseg,3756 bool recover_newaddr);3757int f2fs_get_segment_temp(int seg_type);3758int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page,3759 block_t old_blkaddr, block_t *new_blkaddr,3760 struct f2fs_summary *sum, int type,3761 struct f2fs_io_info *fio);3762void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino,3763 block_t blkaddr, unsigned int blkcnt);3764void f2fs_wait_on_page_writeback(struct page *page,3765 enum page_type type, bool ordered, bool locked);3766void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr);3767void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr,3768 block_t len);3769void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk);3770void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk);3771int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type,3772 unsigned int val, int alloc);3773void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc);3774int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi);3775int f2fs_check_write_pointer(struct f2fs_sb_info *sbi);3776int f2fs_build_segment_manager(struct f2fs_sb_info *sbi);3777void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi);3778int __init f2fs_create_segment_manager_caches(void);3779void f2fs_destroy_segment_manager_caches(void);3780int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint);3781enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi,3782 enum page_type type, enum temp_type temp);3783unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi);3784unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi,3785 unsigned int segno);3786 3787#define DEF_FRAGMENT_SIZE 43788#define MIN_FRAGMENT_SIZE 13789#define MAX_FRAGMENT_SIZE 5123790 3791static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi)3792{3793 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG ||3794 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK;3795}3796 3797/*3798 * checkpoint.c3799 */3800void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io,3801 unsigned char reason);3802void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi);3803struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);3804struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index);3805struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index);3806struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index);3807bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,3808 block_t blkaddr, int type);3809bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi,3810 block_t blkaddr, int type);3811int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,3812 int type, bool sync);3813void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index,3814 unsigned int ra_blocks);3815long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,3816 long nr_to_write, enum iostat_type io_type);3817void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);3818void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type);3819void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all);3820bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode);3821void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,3822 unsigned int devidx, int type);3823bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino,3824 unsigned int devidx, int type);3825int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi);3826void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi);3827void f2fs_add_orphan_inode(struct inode *inode);3828void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino);3829int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi);3830int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi);3831void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio);3832void f2fs_remove_dirty_inode(struct inode *inode);3833int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type,3834 bool from_cp);3835void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type);3836u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi);3837int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc);3838void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi);3839int __init f2fs_create_checkpoint_caches(void);3840void f2fs_destroy_checkpoint_caches(void);3841int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi);3842int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi);3843void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi);3844void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi);3845 3846/*3847 * data.c3848 */3849int __init f2fs_init_bioset(void);3850void f2fs_destroy_bioset(void);3851bool f2fs_is_cp_guaranteed(struct page *page);3852int f2fs_init_bio_entry_cache(void);3853void f2fs_destroy_bio_entry_cache(void);3854void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio,3855 enum page_type type);3856int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi);3857void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type);3858void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,3859 struct inode *inode, struct page *page,3860 nid_t ino, enum page_type type);3861void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi,3862 struct bio **bio, struct page *page);3863void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi);3864int f2fs_submit_page_bio(struct f2fs_io_info *fio);3865int f2fs_merge_page_bio(struct f2fs_io_info *fio);3866void f2fs_submit_page_write(struct f2fs_io_info *fio);3867struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,3868 block_t blk_addr, sector_t *sector);3869int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr);3870void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);3871void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr);3872int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count);3873int f2fs_reserve_new_block(struct dnode_of_data *dn);3874int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index);3875int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index);3876struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index,3877 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs);3878struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index,3879 pgoff_t *next_pgofs);3880struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index,3881 bool for_write);3882struct page *f2fs_get_new_data_page(struct inode *inode,3883 struct page *ipage, pgoff_t index, bool new_i_size);3884int f2fs_do_write_data_page(struct f2fs_io_info *fio);3885int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag);3886int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,3887 u64 start, u64 len);3888int f2fs_encrypt_one_page(struct f2fs_io_info *fio);3889bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio);3890bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio);3891int f2fs_write_single_data_page(struct folio *folio, int *submitted,3892 struct bio **bio, sector_t *last_block,3893 struct writeback_control *wbc,3894 enum iostat_type io_type,3895 int compr_blocks, bool allow_balance);3896void f2fs_write_failed(struct inode *inode, loff_t to);3897void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length);3898bool f2fs_release_folio(struct folio *folio, gfp_t wait);3899bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len);3900void f2fs_clear_page_cache_dirty_tag(struct folio *folio);3901int f2fs_init_post_read_processing(void);3902void f2fs_destroy_post_read_processing(void);3903int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi);3904void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi);3905extern const struct iomap_ops f2fs_iomap_ops;3906 3907/*3908 * gc.c3909 */3910int f2fs_start_gc_thread(struct f2fs_sb_info *sbi);3911void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi);3912block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode);3913int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control);3914void f2fs_build_gc_manager(struct f2fs_sb_info *sbi);3915int f2fs_gc_range(struct f2fs_sb_info *sbi,3916 unsigned int start_seg, unsigned int end_seg,3917 bool dry_run, unsigned int dry_run_sections);3918int f2fs_resize_fs(struct file *filp, __u64 block_count);3919int __init f2fs_create_garbage_collection_cache(void);3920void f2fs_destroy_garbage_collection_cache(void);3921/* victim selection function for cleaning and SSR */3922int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result,3923 int gc_type, int type, char alloc_mode,3924 unsigned long long age, bool one_time);3925 3926/*3927 * recovery.c3928 */3929int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only);3930bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi);3931int __init f2fs_create_recovery_cache(void);3932void f2fs_destroy_recovery_cache(void);3933 3934/*3935 * debug.c3936 */3937#ifdef CONFIG_F2FS_STAT_FS3938struct f2fs_stat_info {3939 struct list_head stat_list;3940 struct f2fs_sb_info *sbi;3941 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs;3942 int main_area_segs, main_area_sections, main_area_zones;3943 unsigned long long hit_cached[NR_EXTENT_CACHES];3944 unsigned long long hit_rbtree[NR_EXTENT_CACHES];3945 unsigned long long total_ext[NR_EXTENT_CACHES];3946 unsigned long long hit_total[NR_EXTENT_CACHES];3947 int ext_tree[NR_EXTENT_CACHES];3948 int zombie_tree[NR_EXTENT_CACHES];3949 int ext_node[NR_EXTENT_CACHES];3950 /* to count memory footprint */3951 unsigned long long ext_mem[NR_EXTENT_CACHES];3952 /* for read extent cache */3953 unsigned long long hit_largest;3954 /* for block age extent cache */3955 unsigned long long allocated_data_blocks;3956 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta;3957 int ndirty_data, ndirty_qdata;3958 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all;3959 int nats, dirty_nats, sits, dirty_sits;3960 int free_nids, avail_nids, alloc_nids;3961 int total_count, utilization;3962 int nr_wb_cp_data, nr_wb_data;3963 int nr_rd_data, nr_rd_node, nr_rd_meta;3964 int nr_dio_read, nr_dio_write;3965 unsigned int io_skip_bggc, other_skip_bggc;3966 int nr_flushing, nr_flushed, flush_list_empty;3967 int nr_discarding, nr_discarded;3968 int nr_discard_cmd;3969 unsigned int undiscard_blks;3970 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt;3971 unsigned int cur_ckpt_time, peak_ckpt_time;3972 int inline_xattr, inline_inode, inline_dir, append, update, orphans;3973 int compr_inode, swapfile_inode;3974 unsigned long long compr_blocks;3975 int aw_cnt, max_aw_cnt;3976 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;3977 unsigned int bimodal, avg_vblocks;3978 int util_free, util_valid, util_invalid;3979 int rsvd_segs, overp_segs;3980 int dirty_count, node_pages, meta_pages, compress_pages;3981 int compress_page_hit;3982 int prefree_count, free_segs, free_secs;3983 int cp_call_count[MAX_CALL_TYPE], cp_count;3984 int gc_call_count[MAX_CALL_TYPE];3985 int gc_segs[2][2];3986 int gc_secs[2][2];3987 int tot_blks, data_blks, node_blks;3988 int bg_data_blks, bg_node_blks;3989 int curseg[NR_CURSEG_TYPE];3990 int cursec[NR_CURSEG_TYPE];3991 int curzone[NR_CURSEG_TYPE];3992 unsigned int dirty_seg[NR_CURSEG_TYPE];3993 unsigned int full_seg[NR_CURSEG_TYPE];3994 unsigned int valid_blks[NR_CURSEG_TYPE];3995 3996 unsigned int meta_count[META_MAX];3997 unsigned int segment_count[2];3998 unsigned int block_count[2];3999 unsigned int inplace_count;4000 unsigned long long base_mem, cache_mem, page_mem;4001};4002 4003static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)4004{4005 return (struct f2fs_stat_info *)sbi->stat_info;4006}4007 4008#define stat_inc_cp_call_count(sbi, foreground) \4009 atomic_inc(&sbi->cp_call_count[(foreground)])4010#define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++)4011#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++)4012#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++)4013#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++)4014#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--)4015#define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type]))4016#define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type]))4017#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest))4018#define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type]))4019#define stat_inc_inline_xattr(inode) \4020 do { \4021 if (f2fs_has_inline_xattr(inode)) \4022 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \4023 } while (0)4024#define stat_dec_inline_xattr(inode) \4025 do { \4026 if (f2fs_has_inline_xattr(inode)) \4027 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \4028 } while (0)4029#define stat_inc_inline_inode(inode) \4030 do { \4031 if (f2fs_has_inline_data(inode)) \4032 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \4033 } while (0)4034#define stat_dec_inline_inode(inode) \4035 do { \4036 if (f2fs_has_inline_data(inode)) \4037 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \4038 } while (0)4039#define stat_inc_inline_dir(inode) \4040 do { \4041 if (f2fs_has_inline_dentry(inode)) \4042 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \4043 } while (0)4044#define stat_dec_inline_dir(inode) \4045 do { \4046 if (f2fs_has_inline_dentry(inode)) \4047 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \4048 } while (0)4049#define stat_inc_compr_inode(inode) \4050 do { \4051 if (f2fs_compressed_file(inode)) \4052 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \4053 } while (0)4054#define stat_dec_compr_inode(inode) \4055 do { \4056 if (f2fs_compressed_file(inode)) \4057 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \4058 } while (0)4059#define stat_add_compr_blocks(inode, blocks) \4060 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))4061#define stat_sub_compr_blocks(inode, blocks) \4062 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))4063#define stat_inc_swapfile_inode(inode) \4064 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode))4065#define stat_dec_swapfile_inode(inode) \4066 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode))4067#define stat_inc_atomic_inode(inode) \4068 (atomic_inc(&F2FS_I_SB(inode)->atomic_files))4069#define stat_dec_atomic_inode(inode) \4070 (atomic_dec(&F2FS_I_SB(inode)->atomic_files))4071#define stat_inc_meta_count(sbi, blkaddr) \4072 do { \4073 if (blkaddr < SIT_I(sbi)->sit_base_addr) \4074 atomic_inc(&(sbi)->meta_count[META_CP]); \4075 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \4076 atomic_inc(&(sbi)->meta_count[META_SIT]); \4077 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \4078 atomic_inc(&(sbi)->meta_count[META_NAT]); \4079 else if (blkaddr < SM_I(sbi)->main_blkaddr) \4080 atomic_inc(&(sbi)->meta_count[META_SSA]); \4081 } while (0)4082#define stat_inc_seg_type(sbi, curseg) \4083 ((sbi)->segment_count[(curseg)->alloc_type]++)4084#define stat_inc_block_count(sbi, curseg) \4085 ((sbi)->block_count[(curseg)->alloc_type]++)4086#define stat_inc_inplace_blocks(sbi) \4087 (atomic_inc(&(sbi)->inplace_count))4088#define stat_update_max_atomic_write(inode) \4089 do { \4090 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \4091 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \4092 if (cur > max) \4093 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \4094 } while (0)4095#define stat_inc_gc_call_count(sbi, foreground) \4096 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++)4097#define stat_inc_gc_sec_count(sbi, type, gc_type) \4098 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++)4099#define stat_inc_gc_seg_count(sbi, type, gc_type) \4100 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++)4101 4102#define stat_inc_tot_blk_count(si, blks) \4103 ((si)->tot_blks += (blks))4104 4105#define stat_inc_data_blk_count(sbi, blks, gc_type) \4106 do { \4107 struct f2fs_stat_info *si = F2FS_STAT(sbi); \4108 stat_inc_tot_blk_count(si, blks); \4109 si->data_blks += (blks); \4110 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \4111 } while (0)4112 4113#define stat_inc_node_blk_count(sbi, blks, gc_type) \4114 do { \4115 struct f2fs_stat_info *si = F2FS_STAT(sbi); \4116 stat_inc_tot_blk_count(si, blks); \4117 si->node_blks += (blks); \4118 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \4119 } while (0)4120 4121int f2fs_build_stats(struct f2fs_sb_info *sbi);4122void f2fs_destroy_stats(struct f2fs_sb_info *sbi);4123void __init f2fs_create_root_stats(void);4124void f2fs_destroy_root_stats(void);4125void f2fs_update_sit_info(struct f2fs_sb_info *sbi);4126#else4127#define stat_inc_cp_call_count(sbi, foreground) do { } while (0)4128#define stat_inc_cp_count(sbi) do { } while (0)4129#define stat_io_skip_bggc_count(sbi) do { } while (0)4130#define stat_other_skip_bggc_count(sbi) do { } while (0)4131#define stat_inc_dirty_inode(sbi, type) do { } while (0)4132#define stat_dec_dirty_inode(sbi, type) do { } while (0)4133#define stat_inc_total_hit(sbi, type) do { } while (0)4134#define stat_inc_rbtree_node_hit(sbi, type) do { } while (0)4135#define stat_inc_largest_node_hit(sbi) do { } while (0)4136#define stat_inc_cached_node_hit(sbi, type) do { } while (0)4137#define stat_inc_inline_xattr(inode) do { } while (0)4138#define stat_dec_inline_xattr(inode) do { } while (0)4139#define stat_inc_inline_inode(inode) do { } while (0)4140#define stat_dec_inline_inode(inode) do { } while (0)4141#define stat_inc_inline_dir(inode) do { } while (0)4142#define stat_dec_inline_dir(inode) do { } while (0)4143#define stat_inc_compr_inode(inode) do { } while (0)4144#define stat_dec_compr_inode(inode) do { } while (0)4145#define stat_add_compr_blocks(inode, blocks) do { } while (0)4146#define stat_sub_compr_blocks(inode, blocks) do { } while (0)4147#define stat_inc_swapfile_inode(inode) do { } while (0)4148#define stat_dec_swapfile_inode(inode) do { } while (0)4149#define stat_inc_atomic_inode(inode) do { } while (0)4150#define stat_dec_atomic_inode(inode) do { } while (0)4151#define stat_update_max_atomic_write(inode) do { } while (0)4152#define stat_inc_meta_count(sbi, blkaddr) do { } while (0)4153#define stat_inc_seg_type(sbi, curseg) do { } while (0)4154#define stat_inc_block_count(sbi, curseg) do { } while (0)4155#define stat_inc_inplace_blocks(sbi) do { } while (0)4156#define stat_inc_gc_call_count(sbi, foreground) do { } while (0)4157#define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0)4158#define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0)4159#define stat_inc_tot_blk_count(si, blks) do { } while (0)4160#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0)4161#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0)4162 4163static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; }4164static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { }4165static inline void __init f2fs_create_root_stats(void) { }4166static inline void f2fs_destroy_root_stats(void) { }4167static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {}4168#endif4169 4170extern const struct file_operations f2fs_dir_operations;4171extern const struct file_operations f2fs_file_operations;4172extern const struct inode_operations f2fs_file_inode_operations;4173extern const struct address_space_operations f2fs_dblock_aops;4174extern const struct address_space_operations f2fs_node_aops;4175extern const struct address_space_operations f2fs_meta_aops;4176extern const struct inode_operations f2fs_dir_inode_operations;4177extern const struct inode_operations f2fs_symlink_inode_operations;4178extern const struct inode_operations f2fs_encrypted_symlink_inode_operations;4179extern const struct inode_operations f2fs_special_inode_operations;4180extern struct kmem_cache *f2fs_inode_entry_slab;4181 4182/*4183 * inline.c4184 */4185bool f2fs_may_inline_data(struct inode *inode);4186bool f2fs_sanity_check_inline_data(struct inode *inode, struct page *ipage);4187bool f2fs_may_inline_dentry(struct inode *inode);4188void f2fs_do_read_inline_data(struct folio *folio, struct page *ipage);4189void f2fs_truncate_inline_inode(struct inode *inode,4190 struct page *ipage, u64 from);4191int f2fs_read_inline_data(struct inode *inode, struct folio *folio);4192int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page);4193int f2fs_convert_inline_inode(struct inode *inode);4194int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry);4195int f2fs_write_inline_data(struct inode *inode, struct folio *folio);4196int f2fs_recover_inline_data(struct inode *inode, struct page *npage);4197struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir,4198 const struct f2fs_filename *fname,4199 struct page **res_page);4200int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent,4201 struct page *ipage);4202int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname,4203 struct inode *inode, nid_t ino, umode_t mode);4204void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry,4205 struct page *page, struct inode *dir,4206 struct inode *inode);4207bool f2fs_empty_inline_dir(struct inode *dir);4208int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx,4209 struct fscrypt_str *fstr);4210int f2fs_inline_data_fiemap(struct inode *inode,4211 struct fiemap_extent_info *fieinfo,4212 __u64 start, __u64 len);4213 4214/*4215 * shrinker.c4216 */4217unsigned long f2fs_shrink_count(struct shrinker *shrink,4218 struct shrink_control *sc);4219unsigned long f2fs_shrink_scan(struct shrinker *shrink,4220 struct shrink_control *sc);4221void f2fs_join_shrinker(struct f2fs_sb_info *sbi);4222void f2fs_leave_shrinker(struct f2fs_sb_info *sbi);4223 4224/*4225 * extent_cache.c4226 */4227bool sanity_check_extent_cache(struct inode *inode, struct page *ipage);4228void f2fs_init_extent_tree(struct inode *inode);4229void f2fs_drop_extent_tree(struct inode *inode);4230void f2fs_destroy_extent_node(struct inode *inode);4231void f2fs_destroy_extent_tree(struct inode *inode);4232void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi);4233int __init f2fs_create_extent_cache(void);4234void f2fs_destroy_extent_cache(void);4235 4236/* read extent cache ops */4237void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage);4238bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs,4239 struct extent_info *ei);4240bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index,4241 block_t *blkaddr);4242void f2fs_update_read_extent_cache(struct dnode_of_data *dn);4243void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn,4244 pgoff_t fofs, block_t blkaddr, unsigned int len);4245unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi,4246 int nr_shrink);4247 4248/* block age extent cache ops */4249void f2fs_init_age_extent_tree(struct inode *inode);4250bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs,4251 struct extent_info *ei);4252void f2fs_update_age_extent_cache(struct dnode_of_data *dn);4253void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn,4254 pgoff_t fofs, unsigned int len);4255unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi,4256 int nr_shrink);4257 4258/*4259 * sysfs.c4260 */4261#define MIN_RA_MUL 24262#define MAX_RA_MUL 2564263 4264int __init f2fs_init_sysfs(void);4265void f2fs_exit_sysfs(void);4266int f2fs_register_sysfs(struct f2fs_sb_info *sbi);4267void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi);4268 4269/* verity.c */4270extern const struct fsverity_operations f2fs_verityops;4271 4272/*4273 * crypto support4274 */4275static inline bool f2fs_encrypted_file(struct inode *inode)4276{4277 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode);4278}4279 4280static inline void f2fs_set_encrypted_inode(struct inode *inode)4281{4282#ifdef CONFIG_FS_ENCRYPTION4283 file_set_encrypt(inode);4284 f2fs_set_inode_flags(inode);4285#endif4286}4287 4288/*4289 * Returns true if the reads of the inode's data need to undergo some4290 * postprocessing step, like decryption or authenticity verification.4291 */4292static inline bool f2fs_post_read_required(struct inode *inode)4293{4294 return f2fs_encrypted_file(inode) || fsverity_active(inode) ||4295 f2fs_compressed_file(inode);4296}4297 4298static inline bool f2fs_used_in_atomic_write(struct inode *inode)4299{4300 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode);4301}4302 4303static inline bool f2fs_meta_inode_gc_required(struct inode *inode)4304{4305 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode);4306}4307 4308/*4309 * compress.c4310 */4311#ifdef CONFIG_F2FS_FS_COMPRESSION4312enum cluster_check_type {4313 CLUSTER_IS_COMPR, /* check only if compressed cluster */4314 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */4315 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */4316};4317bool f2fs_is_compressed_page(struct page *page);4318struct page *f2fs_compress_control_page(struct page *page);4319int f2fs_prepare_compress_overwrite(struct inode *inode,4320 struct page **pagep, pgoff_t index, void **fsdata);4321bool f2fs_compress_write_end(struct inode *inode, void *fsdata,4322 pgoff_t index, unsigned copied);4323int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock);4324void f2fs_compress_write_end_io(struct bio *bio, struct page *page);4325bool f2fs_is_compress_backend_ready(struct inode *inode);4326bool f2fs_is_compress_level_valid(int alg, int lvl);4327int __init f2fs_init_compress_mempool(void);4328void f2fs_destroy_compress_mempool(void);4329void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task);4330void f2fs_end_read_compressed_page(struct page *page, bool failed,4331 block_t blkaddr, bool in_task);4332bool f2fs_cluster_is_empty(struct compress_ctx *cc);4333bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index);4334bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages,4335 int index, int nr_pages, bool uptodate);4336bool f2fs_sanity_check_cluster(struct dnode_of_data *dn);4337void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio);4338int f2fs_write_multi_pages(struct compress_ctx *cc,4339 int *submitted,4340 struct writeback_control *wbc,4341 enum iostat_type io_type);4342int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index);4343bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index);4344void f2fs_update_read_extent_tree_range_compressed(struct inode *inode,4345 pgoff_t fofs, block_t blkaddr,4346 unsigned int llen, unsigned int c_len);4347int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,4348 unsigned nr_pages, sector_t *last_block_in_bio,4349 struct readahead_control *rac, bool for_write);4350struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc);4351void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed,4352 bool in_task);4353void f2fs_put_page_dic(struct page *page, bool in_task);4354unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn,4355 unsigned int ofs_in_node);4356int f2fs_init_compress_ctx(struct compress_ctx *cc);4357void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse);4358void f2fs_init_compress_info(struct f2fs_sb_info *sbi);4359int f2fs_init_compress_inode(struct f2fs_sb_info *sbi);4360void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi);4361int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi);4362void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi);4363int __init f2fs_init_compress_cache(void);4364void f2fs_destroy_compress_cache(void);4365struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi);4366void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr);4367void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page,4368 nid_t ino, block_t blkaddr);4369bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page,4370 block_t blkaddr);4371void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino);4372#define inc_compr_inode_stat(inode) \4373 do { \4374 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \4375 sbi->compr_new_inode++; \4376 } while (0)4377#define add_compr_block_stat(inode, blocks) \4378 do { \4379 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \4380 int diff = F2FS_I(inode)->i_cluster_size - blocks; \4381 sbi->compr_written_block += blocks; \4382 sbi->compr_saved_block += diff; \4383 } while (0)4384#else4385static inline bool f2fs_is_compressed_page(struct page *page) { return false; }4386static inline bool f2fs_is_compress_backend_ready(struct inode *inode)4387{4388 if (!f2fs_compressed_file(inode))4389 return true;4390 /* not support compression */4391 return false;4392}4393static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; }4394static inline struct page *f2fs_compress_control_page(struct page *page)4395{4396 WARN_ON_ONCE(1);4397 return ERR_PTR(-EINVAL);4398}4399static inline int __init f2fs_init_compress_mempool(void) { return 0; }4400static inline void f2fs_destroy_compress_mempool(void) { }4401static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic,4402 bool in_task) { }4403static inline void f2fs_end_read_compressed_page(struct page *page,4404 bool failed, block_t blkaddr, bool in_task)4405{4406 WARN_ON_ONCE(1);4407}4408static inline void f2fs_put_page_dic(struct page *page, bool in_task)4409{4410 WARN_ON_ONCE(1);4411}4412static inline unsigned int f2fs_cluster_blocks_are_contiguous(4413 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; }4414static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; }4415static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; }4416static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { }4417static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; }4418static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { }4419static inline int __init f2fs_init_compress_cache(void) { return 0; }4420static inline void f2fs_destroy_compress_cache(void) { }4421static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi,4422 block_t blkaddr) { }4423static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi,4424 struct page *page, nid_t ino, block_t blkaddr) { }4425static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi,4426 struct page *page, block_t blkaddr) { return false; }4427static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi,4428 nid_t ino) { }4429#define inc_compr_inode_stat(inode) do { } while (0)4430static inline int f2fs_is_compressed_cluster(4431 struct inode *inode,4432 pgoff_t index) { return 0; }4433static inline bool f2fs_is_sparse_cluster(4434 struct inode *inode,4435 pgoff_t index) { return true; }4436static inline void f2fs_update_read_extent_tree_range_compressed(4437 struct inode *inode,4438 pgoff_t fofs, block_t blkaddr,4439 unsigned int llen, unsigned int c_len) { }4440#endif4441 4442static inline int set_compress_context(struct inode *inode)4443{4444#ifdef CONFIG_F2FS_FS_COMPRESSION4445 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);4446 struct f2fs_inode_info *fi = F2FS_I(inode);4447 4448 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm;4449 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size;4450 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ?4451 BIT(COMPRESS_CHKSUM) : 0;4452 fi->i_cluster_size = BIT(fi->i_log_cluster_size);4453 if ((fi->i_compress_algorithm == COMPRESS_LZ4 ||4454 fi->i_compress_algorithm == COMPRESS_ZSTD) &&4455 F2FS_OPTION(sbi).compress_level)4456 fi->i_compress_level = F2FS_OPTION(sbi).compress_level;4457 fi->i_flags |= F2FS_COMPR_FL;4458 set_inode_flag(inode, FI_COMPRESSED_FILE);4459 stat_inc_compr_inode(inode);4460 inc_compr_inode_stat(inode);4461 f2fs_mark_inode_dirty_sync(inode, true);4462 return 0;4463#else4464 return -EOPNOTSUPP;4465#endif4466}4467 4468static inline bool f2fs_disable_compressed_file(struct inode *inode)4469{4470 struct f2fs_inode_info *fi = F2FS_I(inode);4471 4472 f2fs_down_write(&fi->i_sem);4473 4474 if (!f2fs_compressed_file(inode)) {4475 f2fs_up_write(&fi->i_sem);4476 return true;4477 }4478 if (f2fs_is_mmap_file(inode) ||4479 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) {4480 f2fs_up_write(&fi->i_sem);4481 return false;4482 }4483 4484 fi->i_flags &= ~F2FS_COMPR_FL;4485 stat_dec_compr_inode(inode);4486 clear_inode_flag(inode, FI_COMPRESSED_FILE);4487 f2fs_mark_inode_dirty_sync(inode, true);4488 4489 f2fs_up_write(&fi->i_sem);4490 return true;4491}4492 4493#define F2FS_FEATURE_FUNCS(name, flagname) \4494static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \4495{ \4496 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \4497}4498 4499F2FS_FEATURE_FUNCS(encrypt, ENCRYPT);4500F2FS_FEATURE_FUNCS(blkzoned, BLKZONED);4501F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR);4502F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA);4503F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM);4504F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR);4505F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO);4506F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME);4507F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND);4508F2FS_FEATURE_FUNCS(verity, VERITY);4509F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM);4510F2FS_FEATURE_FUNCS(casefold, CASEFOLD);4511F2FS_FEATURE_FUNCS(compression, COMPRESSION);4512F2FS_FEATURE_FUNCS(readonly, RO);4513 4514#ifdef CONFIG_BLK_DEV_ZONED4515static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi,4516 block_t blkaddr)4517{4518 unsigned int zno = blkaddr / sbi->blocks_per_blkz;4519 4520 return test_bit(zno, FDEV(devi).blkz_seq);4521}4522#endif4523 4524static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi,4525 struct block_device *bdev)4526{4527 int i;4528 4529 if (!f2fs_is_multi_device(sbi))4530 return 0;4531 4532 for (i = 0; i < sbi->s_ndevs; i++)4533 if (FDEV(i).bdev == bdev)4534 return i;4535 4536 WARN_ON(1);4537 return -1;4538}4539 4540static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi)4541{4542 return f2fs_sb_has_blkzoned(sbi);4543}4544 4545static inline bool f2fs_bdev_support_discard(struct block_device *bdev)4546{4547 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev);4548}4549 4550static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi)4551{4552 int i;4553 4554 if (!f2fs_is_multi_device(sbi))4555 return f2fs_bdev_support_discard(sbi->sb->s_bdev);4556 4557 for (i = 0; i < sbi->s_ndevs; i++)4558 if (f2fs_bdev_support_discard(FDEV(i).bdev))4559 return true;4560 return false;4561}4562 4563static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi)4564{4565 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) ||4566 f2fs_hw_should_discard(sbi);4567}4568 4569static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi)4570{4571 int i;4572 4573 if (!f2fs_is_multi_device(sbi))4574 return bdev_read_only(sbi->sb->s_bdev);4575 4576 for (i = 0; i < sbi->s_ndevs; i++)4577 if (bdev_read_only(FDEV(i).bdev))4578 return true;4579 return false;4580}4581 4582static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi)4583{4584 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi);4585}4586 4587static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi)4588{4589 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS;4590}4591 4592static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi,4593 block_t blkaddr)4594{4595 if (f2fs_sb_has_blkzoned(sbi)) {4596 int devi = f2fs_target_device_index(sbi, blkaddr);4597 4598 return !bdev_is_zoned(FDEV(devi).bdev);4599 }4600 return true;4601}4602 4603static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi)4604{4605 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW;4606}4607 4608static inline bool f2fs_may_compress(struct inode *inode)4609{4610 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) ||4611 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) ||4612 f2fs_is_mmap_file(inode))4613 return false;4614 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode);4615}4616 4617static inline void f2fs_i_compr_blocks_update(struct inode *inode,4618 u64 blocks, bool add)4619{4620 struct f2fs_inode_info *fi = F2FS_I(inode);4621 int diff = fi->i_cluster_size - blocks;4622 4623 /* don't update i_compr_blocks if saved blocks were released */4624 if (!add && !atomic_read(&fi->i_compr_blocks))4625 return;4626 4627 if (add) {4628 atomic_add(diff, &fi->i_compr_blocks);4629 stat_add_compr_blocks(inode, diff);4630 } else {4631 atomic_sub(diff, &fi->i_compr_blocks);4632 stat_sub_compr_blocks(inode, diff);4633 }4634 f2fs_mark_inode_dirty_sync(inode, true);4635}4636 4637static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi,4638 int flag)4639{4640 if (!f2fs_is_multi_device(sbi))4641 return false;4642 if (flag != F2FS_GET_BLOCK_DIO)4643 return false;4644 return sbi->aligned_blksize;4645}4646 4647static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx)4648{4649 return fsverity_active(inode) &&4650 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE);4651}4652 4653#ifdef CONFIG_F2FS_FAULT_INJECTION4654extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,4655 unsigned long type);4656#else4657static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi,4658 unsigned long rate, unsigned long type)4659{4660 return 0;4661}4662#endif4663 4664static inline bool is_journalled_quota(struct f2fs_sb_info *sbi)4665{4666#ifdef CONFIG_QUOTA4667 if (f2fs_sb_has_quota_ino(sbi))4668 return true;4669 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||4670 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||4671 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])4672 return true;4673#endif4674 return false;4675}4676 4677static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi)4678{4679 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK;4680}4681 4682static inline void f2fs_io_schedule_timeout(long timeout)4683{4684 set_current_state(TASK_UNINTERRUPTIBLE);4685 io_schedule_timeout(timeout);4686}4687 4688static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi,4689 struct folio *folio, enum page_type type)4690{4691 pgoff_t ofs = folio->index;4692 4693 if (unlikely(f2fs_cp_error(sbi)))4694 return;4695 4696 if (ofs == sbi->page_eio_ofs[type]) {4697 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO)4698 set_ckpt_flags(sbi, CP_ERROR_FLAG);4699 } else {4700 sbi->page_eio_ofs[type] = ofs;4701 sbi->page_eio_cnt[type] = 0;4702 }4703}4704 4705static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi)4706{4707 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb);4708}4709 4710static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi,4711 block_t blkaddr, unsigned int cnt)4712{4713 bool need_submit = false;4714 int i = 0;4715 4716 do {4717 struct page *page;4718 4719 page = find_get_page(META_MAPPING(sbi), blkaddr + i);4720 if (page) {4721 if (folio_test_writeback(page_folio(page)))4722 need_submit = true;4723 f2fs_put_page(page, 0);4724 }4725 } while (++i < cnt && !need_submit);4726 4727 if (need_submit)4728 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode,4729 NULL, 0, DATA);4730 4731 truncate_inode_pages_range(META_MAPPING(sbi),4732 F2FS_BLK_TO_BYTES((loff_t)blkaddr),4733 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1)));4734}4735 4736static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi,4737 block_t blkaddr)4738{4739 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1);4740 f2fs_invalidate_compress_page(sbi, blkaddr);4741}4742 4743#define EFSBADCRC EBADMSG /* Bad CRC detected */4744#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */4745 4746#endif /* _LINUX_F2FS_H */4747