at for-next 158 kB view raw
1/* SPDX-License-Identifier: GPL-2.0 */ 2/* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8#ifndef _LINUX_F2FS_H 9#define _LINUX_F2FS_H 10 11#include <linux/uio.h> 12#include <linux/types.h> 13#include <linux/page-flags.h> 14#include <linux/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_FS 35#define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36#else 37#define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42#endif 43 44enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR_VALIDITY, 64 FAULT_BLKADDR_CONSISTENCE, 65 FAULT_NO_SEGMENT, 66 FAULT_MAX, 67}; 68 69#ifdef CONFIG_F2FS_FAULT_INJECTION 70#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 8 83#else 84#define DEFAULT_FAILURE_RETRY_COUNT 1 85#endif 86 87/* 88 * For mount options 89 */ 90#define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 91#define F2FS_MOUNT_DISCARD 0x00000002 92#define F2FS_MOUNT_NOHEAP 0x00000004 93#define F2FS_MOUNT_XATTR_USER 0x00000008 94#define F2FS_MOUNT_POSIX_ACL 0x00000010 95#define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 96#define F2FS_MOUNT_INLINE_XATTR 0x00000040 97#define F2FS_MOUNT_INLINE_DATA 0x00000080 98#define F2FS_MOUNT_INLINE_DENTRY 0x00000100 99#define F2FS_MOUNT_FLUSH_MERGE 0x00000200 100#define F2FS_MOUNT_NOBARRIER 0x00000400 101#define F2FS_MOUNT_FASTBOOT 0x00000800 102#define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 103#define F2FS_MOUNT_DATA_FLUSH 0x00002000 104#define F2FS_MOUNT_FAULT_INJECTION 0x00004000 105#define F2FS_MOUNT_USRQUOTA 0x00008000 106#define F2FS_MOUNT_GRPQUOTA 0x00010000 107#define F2FS_MOUNT_PRJQUOTA 0x00020000 108#define F2FS_MOUNT_QUOTA 0x00040000 109#define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 110#define F2FS_MOUNT_RESERVE_ROOT 0x00100000 111#define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 112#define F2FS_MOUNT_NORECOVERY 0x00400000 113#define F2FS_MOUNT_ATGC 0x00800000 114#define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 115#define F2FS_MOUNT_GC_MERGE 0x02000000 116#define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 117#define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 118 119#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 block 130 * address format, __le32. 131 */ 132typedef u32 nid_t; 133 134#define COMPRESS_EXT_NUM 16 135 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. This 144 * prevents priority inversion when a low-priority reader acquires the read lock 145 * while sleeping on the write lock but the write lock is needed by 146 * higher-priority clients. 147 */ 148 149struct f2fs_rwsem { 150 struct rw_semaphore internal_rwsem; 151#ifdef CONFIG_F2FS_UNFAIR_RWSEM 152 wait_queue_head_t read_waiters; 153#endif 154}; 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_INJECTION 164 struct f2fs_fault_info fault_info; /* For fault injection */ 165#endif 166#ifdef CONFIG_QUOTA 167 /* Names of quota files with journalled quota */ 168 char *s_qf_names[MAXQUOTAS]; 169 int s_jquota_fmt; /* Format of quota to use */ 170#endif 171 /* 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 should 180 * be aligned to this unit: block, 181 * segment or section 182 */ 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 be 186 * unusable when disabling checkpoint 187 */ 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 0x00000001 202#define F2FS_FEATURE_BLKZONED 0x00000002 203#define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 204#define F2FS_FEATURE_EXTRA_ATTR 0x00000008 205#define F2FS_FEATURE_PRJQUOTA 0x00000010 206#define F2FS_FEATURE_INODE_CHKSUM 0x00000020 207#define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 208#define F2FS_FEATURE_QUOTA_INO 0x00000080 209#define F2FS_FEATURE_INODE_CRTIME 0x00000100 210#define F2FS_FEATURE_LOST_FOUND 0x00000200 211#define F2FS_FEATURE_VERITY 0x00000400 212#define F2FS_FEATURE_SB_CHKSUM 0x00000800 213#define F2FS_FEATURE_CASEFOLD 0x00001000 214#define F2FS_FEATURE_COMPRESSION 0x00002000 215#define F2FS_FEATURE_RO 0x00004000 216#define F2FS_FEATURE_DEVICE_ALIAS 0x00008000 217 218#define __F2FS_HAS_FEATURE(raw_super, mask) \ 219 ((raw_super->feature & cpu_to_le32(mask)) != 0) 220#define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 221 222/* 223 * Default values for user and/or group using reserved blocks 224 */ 225#define F2FS_DEF_RESUID 0 226#define F2FS_DEF_RESGID 0 227 228/* 229 * For checkpoint manager 230 */ 231enum { 232 NAT_BITMAP, 233 SIT_BITMAP 234}; 235 236#define CP_UMOUNT 0x00000001 237#define CP_FASTBOOT 0x00000002 238#define CP_SYNC 0x00000004 239#define CP_RECOVERY 0x00000008 240#define CP_DISCARD 0x00000010 241#define CP_TRIMMED 0x00000020 242#define CP_PAUSE 0x00000040 243#define CP_RESIZE 0x00000080 244 245#define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 246#define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 247#define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 248#define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 249#define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 250#define DEF_CP_INTERVAL 60 /* 60 secs */ 251#define DEF_IDLE_INTERVAL 5 /* 5 secs */ 252#define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 253#define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 254#define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 255 256struct cp_control { 257 int reason; 258 __u64 trim_start; 259 __u64 trim_end; 260 __u64 trim_minlen; 261}; 262 263/* 264 * indicate meta/data type 265 */ 266enum { 267 META_CP, 268 META_NAT, 269 META_SIT, 270 META_SSA, 271 META_MAX, 272 META_POR, 273 DATA_GENERIC, /* check range only */ 274 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 275 DATA_GENERIC_ENHANCE_READ, /* 276 * strong check on range and segment 277 * bitmap but no warning due to race 278 * condition of read on truncated area 279 * by extent_cache 280 */ 281 DATA_GENERIC_ENHANCE_UPDATE, /* 282 * strong check on range and segment 283 * bitmap for update case 284 */ 285 META_GENERIC, 286}; 287 288/* for the list of ino */ 289enum { 290 ORPHAN_INO, /* for orphan ino list */ 291 APPEND_INO, /* for append ino list */ 292 UPDATE_INO, /* for update ino list */ 293 TRANS_DIR_INO, /* for transactions dir ino list */ 294 XATTR_DIR_INO, /* for xattr updated dir ino list */ 295 FLUSH_INO, /* for multiple device flushing */ 296 MAX_INO_ENTRY, /* max. list */ 297}; 298 299struct ino_entry { 300 struct list_head list; /* list head */ 301 nid_t ino; /* inode number */ 302 unsigned int dirty_device; /* dirty device bitmap */ 303}; 304 305/* for the list of inodes to be GCed */ 306struct inode_entry { 307 struct list_head list; /* list head */ 308 struct inode *inode; /* vfs inode pointer */ 309}; 310 311struct fsync_node_entry { 312 struct list_head list; /* list head */ 313 struct page *page; /* warm node page pointer */ 314 unsigned int seq_id; /* sequence id */ 315}; 316 317struct ckpt_req { 318 struct completion wait; /* completion for checkpoint done */ 319 struct llist_node llnode; /* llist_node to be linked in wait queue */ 320 int ret; /* return code of checkpoint */ 321 ktime_t queue_time; /* request queued time */ 322}; 323 324struct ckpt_req_control { 325 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 326 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 327 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 328 atomic_t issued_ckpt; /* # of actually issued ckpts */ 329 atomic_t total_ckpt; /* # of total ckpts */ 330 atomic_t queued_ckpt; /* # of queued ckpts */ 331 struct llist_head issue_list; /* list for command issue */ 332 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 333 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 334 unsigned int peak_time; /* peak wait time in msec until now */ 335}; 336 337/* for the bitmap indicate blocks to be discarded */ 338struct discard_entry { 339 struct list_head list; /* list head */ 340 block_t start_blkaddr; /* start blockaddr of current segment */ 341 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 342}; 343 344/* minimum discard granularity, unit: block count */ 345#define MIN_DISCARD_GRANULARITY 1 346/* default discard granularity of inner discard thread, unit: block count */ 347#define DEFAULT_DISCARD_GRANULARITY 16 348/* default maximum discard granularity of ordered discard, unit: block count */ 349#define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 350 351/* max discard pend list number */ 352#define MAX_PLIST_NUM 512 353#define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 354 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 355 356enum { 357 D_PREP, /* initial */ 358 D_PARTIAL, /* partially submitted */ 359 D_SUBMIT, /* all submitted */ 360 D_DONE, /* finished */ 361}; 362 363struct discard_info { 364 block_t lstart; /* logical start address */ 365 block_t len; /* length */ 366 block_t start; /* actual start address in dev */ 367}; 368 369struct discard_cmd { 370 struct rb_node rb_node; /* rb node located in rb-tree */ 371 struct discard_info di; /* discard info */ 372 struct list_head list; /* command list */ 373 struct completion wait; /* compleation */ 374 struct block_device *bdev; /* bdev */ 375 unsigned short ref; /* reference count */ 376 unsigned char state; /* state */ 377 unsigned char queued; /* queued discard */ 378 int error; /* bio error */ 379 spinlock_t lock; /* for state/bio_ref updating */ 380 unsigned short bio_ref; /* bio reference count */ 381}; 382 383enum { 384 DPOLICY_BG, 385 DPOLICY_FORCE, 386 DPOLICY_FSTRIM, 387 DPOLICY_UMOUNT, 388 MAX_DPOLICY, 389}; 390 391enum { 392 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 393 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 394 DPOLICY_IO_AWARE_MAX, 395}; 396 397struct discard_policy { 398 int type; /* type of discard */ 399 unsigned int min_interval; /* used for candidates exist */ 400 unsigned int mid_interval; /* used for device busy */ 401 unsigned int max_interval; /* used for candidates not exist */ 402 unsigned int max_requests; /* # of discards issued per round */ 403 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 404 bool io_aware; /* issue discard in idle time */ 405 bool sync; /* submit discard with REQ_SYNC flag */ 406 bool ordered; /* issue discard by lba order */ 407 bool timeout; /* discard timeout for put_super */ 408 unsigned int granularity; /* discard granularity */ 409}; 410 411struct discard_cmd_control { 412 struct task_struct *f2fs_issue_discard; /* discard thread */ 413 struct list_head entry_list; /* 4KB discard entry list */ 414 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 415 struct list_head wait_list; /* store on-flushing entries */ 416 struct list_head fstrim_list; /* in-flight discard from fstrim */ 417 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 418 struct mutex cmd_lock; 419 unsigned int nr_discards; /* # of discards in the list */ 420 unsigned int max_discards; /* max. discards to be issued */ 421 unsigned int max_discard_request; /* max. discard request per round */ 422 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 423 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 424 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 425 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 426 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 427 unsigned int discard_granularity; /* discard granularity */ 428 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 429 unsigned int discard_io_aware; /* io_aware policy */ 430 unsigned int undiscard_blks; /* # of undiscard blocks */ 431 unsigned int next_pos; /* next discard position */ 432 atomic_t issued_discard; /* # of issued discard */ 433 atomic_t queued_discard; /* # of queued discard */ 434 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 435 struct rb_root_cached root; /* root of discard rb-tree */ 436 bool rbtree_check; /* config for consistence check */ 437 bool discard_wake; /* to wake up discard thread */ 438}; 439 440/* for the list of fsync inodes, used only during recovery */ 441struct fsync_inode_entry { 442 struct list_head list; /* list head */ 443 struct inode *inode; /* vfs inode pointer */ 444 block_t blkaddr; /* block address locating the last fsync */ 445 block_t last_dentry; /* block address locating the last dentry */ 446}; 447 448#define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 449#define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 450 451#define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 452#define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 453#define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 454#define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 455 456#define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 457#define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 458 459static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 460{ 461 int before = nats_in_cursum(journal); 462 463 journal->n_nats = cpu_to_le16(before + i); 464 return before; 465} 466 467static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 468{ 469 int before = sits_in_cursum(journal); 470 471 journal->n_sits = cpu_to_le16(before + i); 472 return before; 473} 474 475static inline bool __has_cursum_space(struct f2fs_journal *journal, 476 int size, int type) 477{ 478 if (type == NAT_JOURNAL) 479 return size <= MAX_NAT_JENTRIES(journal); 480 return size <= MAX_SIT_JENTRIES(journal); 481} 482 483/* for inline stuff */ 484#define DEF_INLINE_RESERVED_SIZE 1 485static inline int get_extra_isize(struct inode *inode); 486static inline int get_inline_xattr_addrs(struct inode *inode); 487#define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 488 (CUR_ADDRS_PER_INODE(inode) - \ 489 get_inline_xattr_addrs(inode) - \ 490 DEF_INLINE_RESERVED_SIZE)) 491 492/* for inline dir */ 493#define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 494 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 495 BITS_PER_BYTE + 1)) 496#define INLINE_DENTRY_BITMAP_SIZE(inode) \ 497 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 498#define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 499 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 500 NR_INLINE_DENTRY(inode) + \ 501 INLINE_DENTRY_BITMAP_SIZE(inode))) 502 503/* 504 * For INODE and NODE manager 505 */ 506/* for directory operations */ 507 508struct f2fs_filename { 509 /* 510 * The filename the user specified. This is NULL for some 511 * filesystem-internal operations, e.g. converting an inline directory 512 * to a non-inline one, or roll-forward recovering an encrypted dentry. 513 */ 514 const struct qstr *usr_fname; 515 516 /* 517 * The on-disk filename. For encrypted directories, this is encrypted. 518 * This may be NULL for lookups in an encrypted dir without the key. 519 */ 520 struct fscrypt_str disk_name; 521 522 /* The dirhash of this filename */ 523 f2fs_hash_t hash; 524 525#ifdef CONFIG_FS_ENCRYPTION 526 /* 527 * For lookups in encrypted directories: either the buffer backing 528 * disk_name, or a buffer that holds the decoded no-key name. 529 */ 530 struct fscrypt_str crypto_buf; 531#endif 532#if IS_ENABLED(CONFIG_UNICODE) 533 /* 534 * For casefolded directories: the casefolded name, but it's left NULL 535 * if the original name is not valid Unicode, if the original name is 536 * "." or "..", if the directory is both casefolded and encrypted and 537 * its encryption key is unavailable, or if the filesystem is doing an 538 * internal operation where usr_fname is also NULL. In all these cases 539 * we fall back to treating the name as an opaque byte sequence. 540 */ 541 struct qstr cf_name; 542#endif 543}; 544 545struct f2fs_dentry_ptr { 546 struct inode *inode; 547 void *bitmap; 548 struct f2fs_dir_entry *dentry; 549 __u8 (*filename)[F2FS_SLOT_LEN]; 550 int max; 551 int nr_bitmap; 552}; 553 554static inline void make_dentry_ptr_block(struct inode *inode, 555 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 556{ 557 d->inode = inode; 558 d->max = NR_DENTRY_IN_BLOCK; 559 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 560 d->bitmap = t->dentry_bitmap; 561 d->dentry = t->dentry; 562 d->filename = t->filename; 563} 564 565static inline void make_dentry_ptr_inline(struct inode *inode, 566 struct f2fs_dentry_ptr *d, void *t) 567{ 568 int entry_cnt = NR_INLINE_DENTRY(inode); 569 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 570 int reserved_size = INLINE_RESERVED_SIZE(inode); 571 572 d->inode = inode; 573 d->max = entry_cnt; 574 d->nr_bitmap = bitmap_size; 575 d->bitmap = t; 576 d->dentry = t + bitmap_size + reserved_size; 577 d->filename = t + bitmap_size + reserved_size + 578 SIZE_OF_DIR_ENTRY * entry_cnt; 579} 580 581/* 582 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 583 * as its node offset to distinguish from index node blocks. 584 * But some bits are used to mark the node block. 585 */ 586#define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 587 >> OFFSET_BIT_SHIFT) 588enum { 589 ALLOC_NODE, /* allocate a new node page if needed */ 590 LOOKUP_NODE, /* look up a node without readahead */ 591 LOOKUP_NODE_RA, /* 592 * look up a node with readahead called 593 * by get_data_block. 594 */ 595}; 596 597#define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 598 599/* congestion wait timeout value, default: 20ms */ 600#define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 601 602/* maximum retry quota flush count */ 603#define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 604 605/* maximum retry of EIO'ed page */ 606#define MAX_RETRY_PAGE_EIO 100 607 608#define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 609 610#define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 611 612/* dirty segments threshold for triggering CP */ 613#define DEFAULT_DIRTY_THRESHOLD 4 614 615#define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 616#define RECOVERY_MIN_RA_BLOCKS 1 617 618#define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 619 620/* for in-memory extent cache entry */ 621#define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 622 623/* number of extent info in extent cache we try to shrink */ 624#define READ_EXTENT_CACHE_SHRINK_NUMBER 128 625 626/* number of age extent info in extent cache we try to shrink */ 627#define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 628#define LAST_AGE_WEIGHT 30 629#define SAME_AGE_REGION 1024 630 631/* 632 * Define data block with age less than 1GB as hot data 633 * define data block with age less than 10GB but more than 1GB as warm data 634 */ 635#define DEF_HOT_DATA_AGE_THRESHOLD 262144 636#define DEF_WARM_DATA_AGE_THRESHOLD 2621440 637 638/* default max read extent count per inode */ 639#define DEF_MAX_READ_EXTENT_COUNT 10240 640 641/* extent cache type */ 642enum extent_type { 643 EX_READ, 644 EX_BLOCK_AGE, 645 NR_EXTENT_CACHES, 646}; 647 648struct extent_info { 649 unsigned int fofs; /* start offset in a file */ 650 unsigned int len; /* length of the extent */ 651 union { 652 /* read extent_cache */ 653 struct { 654 /* start block address of the extent */ 655 block_t blk; 656#ifdef CONFIG_F2FS_FS_COMPRESSION 657 /* physical extent length of compressed blocks */ 658 unsigned int c_len; 659#endif 660 }; 661 /* block age extent_cache */ 662 struct { 663 /* block age of the extent */ 664 unsigned long long age; 665 /* last total blocks allocated */ 666 unsigned long long last_blocks; 667 }; 668 }; 669}; 670 671struct extent_node { 672 struct rb_node rb_node; /* rb node located in rb-tree */ 673 struct extent_info ei; /* extent info */ 674 struct list_head list; /* node in global extent list of sbi */ 675 struct extent_tree *et; /* extent tree pointer */ 676}; 677 678struct extent_tree { 679 nid_t ino; /* inode number */ 680 enum extent_type type; /* keep the extent tree type */ 681 struct rb_root_cached root; /* root of extent info rb-tree */ 682 struct extent_node *cached_en; /* recently accessed extent node */ 683 struct list_head list; /* to be used by sbi->zombie_list */ 684 rwlock_t lock; /* protect extent info rb-tree */ 685 atomic_t node_cnt; /* # of extent node in rb-tree*/ 686 bool largest_updated; /* largest extent updated */ 687 struct extent_info largest; /* largest cached extent for EX_READ */ 688}; 689 690struct extent_tree_info { 691 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 692 struct mutex extent_tree_lock; /* locking extent radix tree */ 693 struct list_head extent_list; /* lru list for shrinker */ 694 spinlock_t extent_lock; /* locking extent lru list */ 695 atomic_t total_ext_tree; /* extent tree count */ 696 struct list_head zombie_list; /* extent zombie tree list */ 697 atomic_t total_zombie_tree; /* extent zombie tree count */ 698 atomic_t total_ext_node; /* extent info count */ 699}; 700 701/* 702 * State of block returned by f2fs_map_blocks. 703 */ 704#define F2FS_MAP_NEW (1U << 0) 705#define F2FS_MAP_MAPPED (1U << 1) 706#define F2FS_MAP_DELALLOC (1U << 2) 707#define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 708 F2FS_MAP_DELALLOC) 709 710struct f2fs_map_blocks { 711 struct block_device *m_bdev; /* for multi-device dio */ 712 block_t m_pblk; 713 block_t m_lblk; 714 unsigned int m_len; 715 unsigned int m_flags; 716 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 717 pgoff_t *m_next_extent; /* point to next possible extent */ 718 int m_seg_type; 719 bool m_may_create; /* indicate it is from write path */ 720 bool m_multidev_dio; /* indicate it allows multi-device dio */ 721}; 722 723/* for flag in get_data_block */ 724enum { 725 F2FS_GET_BLOCK_DEFAULT, 726 F2FS_GET_BLOCK_FIEMAP, 727 F2FS_GET_BLOCK_BMAP, 728 F2FS_GET_BLOCK_DIO, 729 F2FS_GET_BLOCK_PRE_DIO, 730 F2FS_GET_BLOCK_PRE_AIO, 731 F2FS_GET_BLOCK_PRECACHE, 732}; 733 734/* 735 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 736 */ 737#define FADVISE_COLD_BIT 0x01 738#define FADVISE_LOST_PINO_BIT 0x02 739#define FADVISE_ENCRYPT_BIT 0x04 740#define FADVISE_ENC_NAME_BIT 0x08 741#define FADVISE_KEEP_SIZE_BIT 0x10 742#define FADVISE_HOT_BIT 0x20 743#define FADVISE_VERITY_BIT 0x40 744#define FADVISE_TRUNC_BIT 0x80 745 746#define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 747 748#define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 749#define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 750#define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 751 752#define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 753#define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 754#define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 755 756#define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 757#define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 758 759#define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 760#define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 761 762#define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 763#define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 764 765#define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 766#define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 767#define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 768 769#define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 770#define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 771 772#define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 773#define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 774#define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 775 776#define DEF_DIR_LEVEL 0 777 778/* used for f2fs_inode_info->flags */ 779enum { 780 FI_NEW_INODE, /* indicate newly allocated inode */ 781 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 782 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 783 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 784 FI_INC_LINK, /* need to increment i_nlink */ 785 FI_ACL_MODE, /* indicate acl mode */ 786 FI_NO_ALLOC, /* should not allocate any blocks */ 787 FI_FREE_NID, /* free allocated nide */ 788 FI_NO_EXTENT, /* not to use the extent cache */ 789 FI_INLINE_XATTR, /* used for inline xattr */ 790 FI_INLINE_DATA, /* used for inline data*/ 791 FI_INLINE_DENTRY, /* used for inline dentry */ 792 FI_APPEND_WRITE, /* inode has appended data */ 793 FI_UPDATE_WRITE, /* inode has in-place-update data */ 794 FI_NEED_IPU, /* used for ipu per file */ 795 FI_ATOMIC_FILE, /* indicate atomic file */ 796 FI_DATA_EXIST, /* indicate data exists */ 797 FI_SKIP_WRITES, /* should skip data page writeback */ 798 FI_OPU_WRITE, /* used for opu per file */ 799 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 800 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 801 FI_HOT_DATA, /* indicate file is hot */ 802 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 803 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 804 FI_PIN_FILE, /* indicate file should not be gced */ 805 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 806 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 807 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 808 FI_MMAP_FILE, /* indicate file was mmapped */ 809 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 810 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 811 FI_ALIGNED_WRITE, /* enable aligned write */ 812 FI_COW_FILE, /* indicate COW file */ 813 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 814 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 815 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 816 FI_OPENED_FILE, /* indicate file has been opened */ 817 FI_MAX, /* max flag, never be used */ 818}; 819 820struct f2fs_inode_info { 821 struct inode vfs_inode; /* serve a vfs inode */ 822 unsigned long i_flags; /* keep an inode flags for ioctl */ 823 unsigned char i_advise; /* use to give file attribute hints */ 824 unsigned char i_dir_level; /* use for dentry level for large dir */ 825 union { 826 unsigned int i_current_depth; /* only for directory depth */ 827 unsigned short i_gc_failures; /* for gc failure statistic */ 828 }; 829 unsigned int i_pino; /* parent inode number */ 830 umode_t i_acl_mode; /* keep file acl mode temporarily */ 831 832 /* Use below internally in f2fs*/ 833 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 834 struct f2fs_rwsem i_sem; /* protect fi info */ 835 atomic_t dirty_pages; /* # of dirty pages */ 836 f2fs_hash_t chash; /* hash value of given file name */ 837 unsigned int clevel; /* maximum level of given file name */ 838 struct task_struct *task; /* lookup and create consistency */ 839 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 840 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 841 nid_t i_xattr_nid; /* node id that contains xattrs */ 842 loff_t last_disk_size; /* lastly written file size */ 843 spinlock_t i_size_lock; /* protect last_disk_size */ 844 845#ifdef CONFIG_QUOTA 846 struct dquot __rcu *i_dquot[MAXQUOTAS]; 847 848 /* quota space reservation, managed internally by quota code */ 849 qsize_t i_reserved_quota; 850#endif 851 struct list_head dirty_list; /* dirty list for dirs and files */ 852 struct list_head gdirty_list; /* linked in global dirty list */ 853 struct task_struct *atomic_write_task; /* store atomic write task */ 854 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 855 /* cached extent_tree entry */ 856 union { 857 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 858 struct inode *atomic_inode; 859 /* point to atomic_inode, available only for cow_inode */ 860 }; 861 862 /* avoid racing between foreground op and gc */ 863 struct f2fs_rwsem i_gc_rwsem[2]; 864 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 865 866 int i_extra_isize; /* size of extra space located in i_addr */ 867 kprojid_t i_projid; /* id for project quota */ 868 int i_inline_xattr_size; /* inline xattr size */ 869 struct timespec64 i_crtime; /* inode creation time */ 870 struct timespec64 i_disk_time[3];/* inode disk times */ 871 872 /* for file compress */ 873 atomic_t i_compr_blocks; /* # of compressed blocks */ 874 unsigned char i_compress_algorithm; /* algorithm type */ 875 unsigned char i_log_cluster_size; /* log of cluster size */ 876 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 877 unsigned char i_compress_flag; /* compress flag */ 878 unsigned int i_cluster_size; /* cluster size */ 879 880 unsigned int atomic_write_cnt; 881 loff_t original_i_size; /* original i_size before atomic write */ 882}; 883 884static inline void get_read_extent_info(struct extent_info *ext, 885 struct f2fs_extent *i_ext) 886{ 887 ext->fofs = le32_to_cpu(i_ext->fofs); 888 ext->blk = le32_to_cpu(i_ext->blk); 889 ext->len = le32_to_cpu(i_ext->len); 890} 891 892static inline void set_raw_read_extent(struct extent_info *ext, 893 struct f2fs_extent *i_ext) 894{ 895 i_ext->fofs = cpu_to_le32(ext->fofs); 896 i_ext->blk = cpu_to_le32(ext->blk); 897 i_ext->len = cpu_to_le32(ext->len); 898} 899 900static inline bool __is_discard_mergeable(struct discard_info *back, 901 struct discard_info *front, unsigned int max_len) 902{ 903 return (back->lstart + back->len == front->lstart) && 904 (back->len + front->len <= max_len); 905} 906 907static inline bool __is_discard_back_mergeable(struct discard_info *cur, 908 struct discard_info *back, unsigned int max_len) 909{ 910 return __is_discard_mergeable(back, cur, max_len); 911} 912 913static inline bool __is_discard_front_mergeable(struct discard_info *cur, 914 struct discard_info *front, unsigned int max_len) 915{ 916 return __is_discard_mergeable(cur, front, max_len); 917} 918 919/* 920 * For free nid management 921 */ 922enum nid_state { 923 FREE_NID, /* newly added to free nid list */ 924 PREALLOC_NID, /* it is preallocated */ 925 MAX_NID_STATE, 926}; 927 928enum nat_state { 929 TOTAL_NAT, 930 DIRTY_NAT, 931 RECLAIMABLE_NAT, 932 MAX_NAT_STATE, 933}; 934 935struct f2fs_nm_info { 936 block_t nat_blkaddr; /* base disk address of NAT */ 937 nid_t max_nid; /* maximum possible node ids */ 938 nid_t available_nids; /* # of available node ids */ 939 nid_t next_scan_nid; /* the next nid to be scanned */ 940 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 941 unsigned int ram_thresh; /* control the memory footprint */ 942 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 943 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 944 945 /* NAT cache management */ 946 struct radix_tree_root nat_root;/* root of the nat entry cache */ 947 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 948 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 949 struct list_head nat_entries; /* cached nat entry list (clean) */ 950 spinlock_t nat_list_lock; /* protect clean nat entry list */ 951 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 952 unsigned int nat_blocks; /* # of nat blocks */ 953 954 /* free node ids management */ 955 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 956 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 957 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 958 spinlock_t nid_list_lock; /* protect nid lists ops */ 959 struct mutex build_lock; /* lock for build free nids */ 960 unsigned char **free_nid_bitmap; 961 unsigned char *nat_block_bitmap; 962 unsigned short *free_nid_count; /* free nid count of NAT block */ 963 964 /* for checkpoint */ 965 char *nat_bitmap; /* NAT bitmap pointer */ 966 967 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 968 unsigned char *nat_bits; /* NAT bits blocks */ 969 unsigned char *full_nat_bits; /* full NAT pages */ 970 unsigned char *empty_nat_bits; /* empty NAT pages */ 971#ifdef CONFIG_F2FS_CHECK_FS 972 char *nat_bitmap_mir; /* NAT bitmap mirror */ 973#endif 974 int bitmap_size; /* bitmap size */ 975}; 976 977/* 978 * this structure is used as one of function parameters. 979 * all the information are dedicated to a given direct node block determined 980 * by the data offset in a file. 981 */ 982struct dnode_of_data { 983 struct inode *inode; /* vfs inode pointer */ 984 struct page *inode_page; /* its inode page, NULL is possible */ 985 struct page *node_page; /* cached direct node page */ 986 nid_t nid; /* node id of the direct node block */ 987 unsigned int ofs_in_node; /* data offset in the node page */ 988 bool inode_page_locked; /* inode page is locked or not */ 989 bool node_changed; /* is node block changed */ 990 char cur_level; /* level of hole node page */ 991 char max_level; /* level of current page located */ 992 block_t data_blkaddr; /* block address of the node block */ 993}; 994 995static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 996 struct page *ipage, struct page *npage, nid_t nid) 997{ 998 memset(dn, 0, sizeof(*dn)); 999 dn->inode = inode; 1000 dn->inode_page = ipage; 1001 dn->node_page = npage; 1002 dn->nid = nid; 1003} 1004 1005/* 1006 * For SIT manager 1007 * 1008 * By default, there are 6 active log areas across the whole main area. 1009 * When considering hot and cold data separation to reduce cleaning overhead, 1010 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1011 * respectively. 1012 * In the current design, you should not change the numbers intentionally. 1013 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1014 * logs individually according to the underlying devices. (default: 6) 1015 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1016 * data and 8 for node logs. 1017 */ 1018#define NR_CURSEG_DATA_TYPE (3) 1019#define NR_CURSEG_NODE_TYPE (3) 1020#define NR_CURSEG_INMEM_TYPE (2) 1021#define NR_CURSEG_RO_TYPE (2) 1022#define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1023#define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1024 1025enum log_type { 1026 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1027 CURSEG_WARM_DATA, /* data blocks */ 1028 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1029 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1030 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1031 CURSEG_COLD_NODE, /* indirect node blocks */ 1032 NR_PERSISTENT_LOG, /* number of persistent log */ 1033 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1034 /* pinned file that needs consecutive block address */ 1035 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1036 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1037}; 1038 1039struct flush_cmd { 1040 struct completion wait; 1041 struct llist_node llnode; 1042 nid_t ino; 1043 int ret; 1044}; 1045 1046struct flush_cmd_control { 1047 struct task_struct *f2fs_issue_flush; /* flush thread */ 1048 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1049 atomic_t issued_flush; /* # of issued flushes */ 1050 atomic_t queued_flush; /* # of queued flushes */ 1051 struct llist_head issue_list; /* list for command issue */ 1052 struct llist_node *dispatch_list; /* list for command dispatch */ 1053}; 1054 1055struct f2fs_sm_info { 1056 struct sit_info *sit_info; /* whole segment information */ 1057 struct free_segmap_info *free_info; /* free segment information */ 1058 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1059 struct curseg_info *curseg_array; /* active segment information */ 1060 1061 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1062 1063 block_t seg0_blkaddr; /* block address of 0'th segment */ 1064 block_t main_blkaddr; /* start block address of main area */ 1065 block_t ssa_blkaddr; /* start block address of SSA area */ 1066 1067 unsigned int segment_count; /* total # of segments */ 1068 unsigned int main_segments; /* # of segments in main area */ 1069 unsigned int reserved_segments; /* # of reserved segments */ 1070 unsigned int ovp_segments; /* # of overprovision segments */ 1071 1072 /* a threshold to reclaim prefree segments */ 1073 unsigned int rec_prefree_segments; 1074 1075 struct list_head sit_entry_set; /* sit entry set list */ 1076 1077 unsigned int ipu_policy; /* in-place-update policy */ 1078 unsigned int min_ipu_util; /* in-place-update threshold */ 1079 unsigned int min_fsync_blocks; /* threshold for fsync */ 1080 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1081 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1082 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1083 1084 /* for flush command control */ 1085 struct flush_cmd_control *fcc_info; 1086 1087 /* for discard command control */ 1088 struct discard_cmd_control *dcc_info; 1089}; 1090 1091/* 1092 * For superblock 1093 */ 1094/* 1095 * COUNT_TYPE for monitoring 1096 * 1097 * f2fs monitors the number of several block types such as on-writeback, 1098 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1099 */ 1100#define WB_DATA_TYPE(p, f) \ 1101 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1102enum count_type { 1103 F2FS_DIRTY_DENTS, 1104 F2FS_DIRTY_DATA, 1105 F2FS_DIRTY_QDATA, 1106 F2FS_DIRTY_NODES, 1107 F2FS_DIRTY_META, 1108 F2FS_DIRTY_IMETA, 1109 F2FS_WB_CP_DATA, 1110 F2FS_WB_DATA, 1111 F2FS_RD_DATA, 1112 F2FS_RD_NODE, 1113 F2FS_RD_META, 1114 F2FS_DIO_WRITE, 1115 F2FS_DIO_READ, 1116 NR_COUNT_TYPE, 1117}; 1118 1119/* 1120 * The below are the page types of bios used in submit_bio(). 1121 * The available types are: 1122 * DATA User data pages. It operates as async mode. 1123 * NODE Node pages. It operates as async mode. 1124 * META FS metadata pages such as SIT, NAT, CP. 1125 * NR_PAGE_TYPE The number of page types. 1126 * META_FLUSH Make sure the previous pages are written 1127 * with waiting the bio's completion 1128 * ... Only can be used with META. 1129 */ 1130#define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1131#define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1132enum page_type { 1133 DATA = 0, 1134 NODE = 1, /* should not change this */ 1135 META, 1136 NR_PAGE_TYPE, 1137 META_FLUSH, 1138 IPU, /* the below types are used by tracepoints only. */ 1139 OPU, 1140}; 1141 1142enum temp_type { 1143 HOT = 0, /* must be zero for meta bio */ 1144 WARM, 1145 COLD, 1146 NR_TEMP_TYPE, 1147}; 1148 1149enum need_lock_type { 1150 LOCK_REQ = 0, 1151 LOCK_DONE, 1152 LOCK_RETRY, 1153}; 1154 1155enum cp_reason_type { 1156 CP_NO_NEEDED, 1157 CP_NON_REGULAR, 1158 CP_COMPRESSED, 1159 CP_HARDLINK, 1160 CP_SB_NEED_CP, 1161 CP_WRONG_PINO, 1162 CP_NO_SPC_ROLL, 1163 CP_NODE_NEED_CP, 1164 CP_FASTBOOT_MODE, 1165 CP_SPEC_LOG_NUM, 1166 CP_RECOVER_DIR, 1167 CP_XATTR_DIR, 1168}; 1169 1170enum iostat_type { 1171 /* WRITE IO */ 1172 APP_DIRECT_IO, /* app direct write IOs */ 1173 APP_BUFFERED_IO, /* app buffered write IOs */ 1174 APP_WRITE_IO, /* app write IOs */ 1175 APP_MAPPED_IO, /* app mapped IOs */ 1176 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1177 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1178 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1179 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1180 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1181 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1182 FS_GC_DATA_IO, /* data IOs from forground gc */ 1183 FS_GC_NODE_IO, /* node IOs from forground gc */ 1184 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1185 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1186 FS_CP_META_IO, /* meta IOs from checkpoint */ 1187 1188 /* READ IO */ 1189 APP_DIRECT_READ_IO, /* app direct read IOs */ 1190 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1191 APP_READ_IO, /* app read IOs */ 1192 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1193 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1194 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1195 FS_DATA_READ_IO, /* data read IOs */ 1196 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1197 FS_CDATA_READ_IO, /* compressed data read IOs */ 1198 FS_NODE_READ_IO, /* node read IOs */ 1199 FS_META_READ_IO, /* meta read IOs */ 1200 1201 /* other */ 1202 FS_DISCARD_IO, /* discard */ 1203 FS_FLUSH_IO, /* flush */ 1204 FS_ZONE_RESET_IO, /* zone reset */ 1205 NR_IO_TYPE, 1206}; 1207 1208struct f2fs_io_info { 1209 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1210 nid_t ino; /* inode number */ 1211 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1212 enum temp_type temp; /* contains HOT/WARM/COLD */ 1213 enum req_op op; /* contains REQ_OP_ */ 1214 blk_opf_t op_flags; /* req_flag_bits */ 1215 block_t new_blkaddr; /* new block address to be written */ 1216 block_t old_blkaddr; /* old block address before Cow */ 1217 struct page *page; /* page to be written */ 1218 struct page *encrypted_page; /* encrypted page */ 1219 struct page *compressed_page; /* compressed page */ 1220 struct list_head list; /* serialize IOs */ 1221 unsigned int compr_blocks; /* # of compressed block addresses */ 1222 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1223 unsigned int version:8; /* version of the node */ 1224 unsigned int submitted:1; /* indicate IO submission */ 1225 unsigned int in_list:1; /* indicate fio is in io_list */ 1226 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1227 unsigned int encrypted:1; /* indicate file is encrypted */ 1228 unsigned int meta_gc:1; /* require meta inode GC */ 1229 enum iostat_type io_type; /* io type */ 1230 struct writeback_control *io_wbc; /* writeback control */ 1231 struct bio **bio; /* bio for ipu */ 1232 sector_t *last_block; /* last block number in bio */ 1233}; 1234 1235struct bio_entry { 1236 struct bio *bio; 1237 struct list_head list; 1238}; 1239 1240#define is_read_io(rw) ((rw) == READ) 1241struct f2fs_bio_info { 1242 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1243 struct bio *bio; /* bios to merge */ 1244 sector_t last_block_in_bio; /* last block number */ 1245 struct f2fs_io_info fio; /* store buffered io info. */ 1246#ifdef CONFIG_BLK_DEV_ZONED 1247 struct completion zone_wait; /* condition value for the previous open zone to close */ 1248 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1249 void *bi_private; /* previous bi_private for pending bio */ 1250#endif 1251 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1252 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1253 struct list_head io_list; /* track fios */ 1254 struct list_head bio_list; /* bio entry list head */ 1255 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1256}; 1257 1258#define FDEV(i) (sbi->devs[i]) 1259#define RDEV(i) (raw_super->devs[i]) 1260struct f2fs_dev_info { 1261 struct file *bdev_file; 1262 struct block_device *bdev; 1263 char path[MAX_PATH_LEN]; 1264 unsigned int total_segments; 1265 block_t start_blk; 1266 block_t end_blk; 1267#ifdef CONFIG_BLK_DEV_ZONED 1268 unsigned int nr_blkz; /* Total number of zones */ 1269 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1270#endif 1271}; 1272 1273enum inode_type { 1274 DIR_INODE, /* for dirty dir inode */ 1275 FILE_INODE, /* for dirty regular/symlink inode */ 1276 DIRTY_META, /* for all dirtied inode metadata */ 1277 NR_INODE_TYPE, 1278}; 1279 1280/* for inner inode cache management */ 1281struct inode_management { 1282 struct radix_tree_root ino_root; /* ino entry array */ 1283 spinlock_t ino_lock; /* for ino entry lock */ 1284 struct list_head ino_list; /* inode list head */ 1285 unsigned long ino_num; /* number of entries */ 1286}; 1287 1288/* for GC_AT */ 1289struct atgc_management { 1290 bool atgc_enabled; /* ATGC is enabled or not */ 1291 struct rb_root_cached root; /* root of victim rb-tree */ 1292 struct list_head victim_list; /* linked with all victim entries */ 1293 unsigned int victim_count; /* victim count in rb-tree */ 1294 unsigned int candidate_ratio; /* candidate ratio */ 1295 unsigned int max_candidate_count; /* max candidate count */ 1296 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1297 unsigned long long age_threshold; /* age threshold */ 1298}; 1299 1300struct f2fs_gc_control { 1301 unsigned int victim_segno; /* target victim segment number */ 1302 int init_gc_type; /* FG_GC or BG_GC */ 1303 bool no_bg_gc; /* check the space and stop bg_gc */ 1304 bool should_migrate_blocks; /* should migrate blocks */ 1305 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1306 bool one_time; /* require one time GC in one migration unit */ 1307 unsigned int nr_free_secs; /* # of free sections to do GC */ 1308}; 1309 1310/* 1311 * For s_flag in struct f2fs_sb_info 1312 * Modification on enum should be synchronized with s_flag array 1313 */ 1314enum { 1315 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1316 SBI_IS_CLOSE, /* specify unmounting */ 1317 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1318 SBI_POR_DOING, /* recovery is doing or not */ 1319 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1320 SBI_NEED_CP, /* need to checkpoint */ 1321 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1322 SBI_IS_RECOVERED, /* recovered orphan/data */ 1323 SBI_CP_DISABLED, /* CP was disabled last mount */ 1324 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1325 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1326 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1327 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1328 SBI_IS_RESIZEFS, /* resizefs is in process */ 1329 SBI_IS_FREEZING, /* freezefs is in process */ 1330 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1331 MAX_SBI_FLAG, 1332}; 1333 1334enum { 1335 CP_TIME, 1336 REQ_TIME, 1337 DISCARD_TIME, 1338 GC_TIME, 1339 DISABLE_TIME, 1340 UMOUNT_DISCARD_TIMEOUT, 1341 MAX_TIME, 1342}; 1343 1344/* Note that you need to keep synchronization with this gc_mode_names array */ 1345enum { 1346 GC_NORMAL, 1347 GC_IDLE_CB, 1348 GC_IDLE_GREEDY, 1349 GC_IDLE_AT, 1350 GC_URGENT_HIGH, 1351 GC_URGENT_LOW, 1352 GC_URGENT_MID, 1353 MAX_GC_MODE, 1354}; 1355 1356enum { 1357 BGGC_MODE_ON, /* background gc is on */ 1358 BGGC_MODE_OFF, /* background gc is off */ 1359 BGGC_MODE_SYNC, /* 1360 * background gc is on, migrating blocks 1361 * like foreground gc 1362 */ 1363}; 1364 1365enum { 1366 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1367 FS_MODE_LFS, /* use lfs allocation only */ 1368 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1369 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1370}; 1371 1372enum { 1373 ALLOC_MODE_DEFAULT, /* stay default */ 1374 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1375}; 1376 1377enum fsync_mode { 1378 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1379 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1380 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1381}; 1382 1383enum { 1384 COMPR_MODE_FS, /* 1385 * automatically compress compression 1386 * enabled files 1387 */ 1388 COMPR_MODE_USER, /* 1389 * automatical compression is disabled. 1390 * user can control the file compression 1391 * using ioctls 1392 */ 1393}; 1394 1395enum { 1396 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1397 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1398 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1399}; 1400 1401enum { 1402 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1403 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1404}; 1405 1406enum errors_option { 1407 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1408 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1409 MOUNT_ERRORS_PANIC, /* panic on errors */ 1410}; 1411 1412enum { 1413 BACKGROUND, 1414 FOREGROUND, 1415 MAX_CALL_TYPE, 1416 TOTAL_CALL = FOREGROUND, 1417}; 1418 1419static inline int f2fs_test_bit(unsigned int nr, char *addr); 1420static inline void f2fs_set_bit(unsigned int nr, char *addr); 1421static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1422 1423/* 1424 * Layout of f2fs page.private: 1425 * 1426 * Layout A: lowest bit should be 1 1427 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1428 * bit 0 PAGE_PRIVATE_NOT_POINTER 1429 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1430 * bit 2 PAGE_PRIVATE_INLINE_INODE 1431 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1432 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1433 * bit 5- f2fs private data 1434 * 1435 * Layout B: lowest bit should be 0 1436 * page.private is a wrapped pointer. 1437 */ 1438enum { 1439 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1440 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1441 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1442 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1443 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1444 PAGE_PRIVATE_MAX 1445}; 1446 1447/* For compression */ 1448enum compress_algorithm_type { 1449 COMPRESS_LZO, 1450 COMPRESS_LZ4, 1451 COMPRESS_ZSTD, 1452 COMPRESS_LZORLE, 1453 COMPRESS_MAX, 1454}; 1455 1456enum compress_flag { 1457 COMPRESS_CHKSUM, 1458 COMPRESS_MAX_FLAG, 1459}; 1460 1461#define COMPRESS_WATERMARK 20 1462#define COMPRESS_PERCENT 20 1463 1464#define COMPRESS_DATA_RESERVED_SIZE 4 1465struct compress_data { 1466 __le32 clen; /* compressed data size */ 1467 __le32 chksum; /* compressed data chksum */ 1468 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1469 u8 cdata[]; /* compressed data */ 1470}; 1471 1472#define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1473 1474#define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1475 1476#define F2FS_ZSTD_DEFAULT_CLEVEL 1 1477 1478#define COMPRESS_LEVEL_OFFSET 8 1479 1480/* compress context */ 1481struct compress_ctx { 1482 struct inode *inode; /* inode the context belong to */ 1483 pgoff_t cluster_idx; /* cluster index number */ 1484 unsigned int cluster_size; /* page count in cluster */ 1485 unsigned int log_cluster_size; /* log of cluster size */ 1486 struct page **rpages; /* pages store raw data in cluster */ 1487 unsigned int nr_rpages; /* total page number in rpages */ 1488 struct page **cpages; /* pages store compressed data in cluster */ 1489 unsigned int nr_cpages; /* total page number in cpages */ 1490 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1491 void *rbuf; /* virtual mapped address on rpages */ 1492 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1493 size_t rlen; /* valid data length in rbuf */ 1494 size_t clen; /* valid data length in cbuf */ 1495 void *private; /* payload buffer for specified compression algorithm */ 1496 void *private2; /* extra payload buffer */ 1497}; 1498 1499/* compress context for write IO path */ 1500struct compress_io_ctx { 1501 u32 magic; /* magic number to indicate page is compressed */ 1502 struct inode *inode; /* inode the context belong to */ 1503 struct page **rpages; /* pages store raw data in cluster */ 1504 unsigned int nr_rpages; /* total page number in rpages */ 1505 atomic_t pending_pages; /* in-flight compressed page count */ 1506}; 1507 1508/* Context for decompressing one cluster on the read IO path */ 1509struct decompress_io_ctx { 1510 u32 magic; /* magic number to indicate page is compressed */ 1511 struct inode *inode; /* inode the context belong to */ 1512 pgoff_t cluster_idx; /* cluster index number */ 1513 unsigned int cluster_size; /* page count in cluster */ 1514 unsigned int log_cluster_size; /* log of cluster size */ 1515 struct page **rpages; /* pages store raw data in cluster */ 1516 unsigned int nr_rpages; /* total page number in rpages */ 1517 struct page **cpages; /* pages store compressed data in cluster */ 1518 unsigned int nr_cpages; /* total page number in cpages */ 1519 struct page **tpages; /* temp pages to pad holes in cluster */ 1520 void *rbuf; /* virtual mapped address on rpages */ 1521 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1522 size_t rlen; /* valid data length in rbuf */ 1523 size_t clen; /* valid data length in cbuf */ 1524 1525 /* 1526 * The number of compressed pages remaining to be read in this cluster. 1527 * This is initially nr_cpages. It is decremented by 1 each time a page 1528 * has been read (or failed to be read). When it reaches 0, the cluster 1529 * is decompressed (or an error is reported). 1530 * 1531 * If an error occurs before all the pages have been submitted for I/O, 1532 * then this will never reach 0. In this case the I/O submitter is 1533 * responsible for calling f2fs_decompress_end_io() instead. 1534 */ 1535 atomic_t remaining_pages; 1536 1537 /* 1538 * Number of references to this decompress_io_ctx. 1539 * 1540 * One reference is held for I/O completion. This reference is dropped 1541 * after the pagecache pages are updated and unlocked -- either after 1542 * decompression (and verity if enabled), or after an error. 1543 * 1544 * In addition, each compressed page holds a reference while it is in a 1545 * bio. These references are necessary prevent compressed pages from 1546 * being freed while they are still in a bio. 1547 */ 1548 refcount_t refcnt; 1549 1550 bool failed; /* IO error occurred before decompression? */ 1551 bool need_verity; /* need fs-verity verification after decompression? */ 1552 void *private; /* payload buffer for specified decompression algorithm */ 1553 void *private2; /* extra payload buffer */ 1554 struct work_struct verity_work; /* work to verify the decompressed pages */ 1555 struct work_struct free_work; /* work for late free this structure itself */ 1556}; 1557 1558#define NULL_CLUSTER ((unsigned int)(~0)) 1559#define MIN_COMPRESS_LOG_SIZE 2 1560#define MAX_COMPRESS_LOG_SIZE 8 1561#define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1562 1563struct f2fs_sb_info { 1564 struct super_block *sb; /* pointer to VFS super block */ 1565 struct proc_dir_entry *s_proc; /* proc entry */ 1566 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1567 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1568 int valid_super_block; /* valid super block no */ 1569 unsigned long s_flag; /* flags for sbi */ 1570 struct mutex writepages; /* mutex for writepages() */ 1571 1572#ifdef CONFIG_BLK_DEV_ZONED 1573 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1574 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1575 /* For adjust the priority writing position of data in zone UFS */ 1576 unsigned int blkzone_alloc_policy; 1577#endif 1578 1579 /* for node-related operations */ 1580 struct f2fs_nm_info *nm_info; /* node manager */ 1581 struct inode *node_inode; /* cache node blocks */ 1582 1583 /* for segment-related operations */ 1584 struct f2fs_sm_info *sm_info; /* segment manager */ 1585 1586 /* for bio operations */ 1587 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1588 /* keep migration IO order for LFS mode */ 1589 struct f2fs_rwsem io_order_lock; 1590 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1591 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1592 1593 /* for checkpoint */ 1594 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1595 int cur_cp_pack; /* remain current cp pack */ 1596 spinlock_t cp_lock; /* for flag in ckpt */ 1597 struct inode *meta_inode; /* cache meta blocks */ 1598 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1599 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1600 struct f2fs_rwsem node_write; /* locking node writes */ 1601 struct f2fs_rwsem node_change; /* locking node change */ 1602 wait_queue_head_t cp_wait; 1603 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1604 long interval_time[MAX_TIME]; /* to store thresholds */ 1605 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1606 1607 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1608 1609 spinlock_t fsync_node_lock; /* for node entry lock */ 1610 struct list_head fsync_node_list; /* node list head */ 1611 unsigned int fsync_seg_id; /* sequence id */ 1612 unsigned int fsync_node_num; /* number of node entries */ 1613 1614 /* for orphan inode, use 0'th array */ 1615 unsigned int max_orphans; /* max orphan inodes */ 1616 1617 /* for inode management */ 1618 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1619 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1620 struct mutex flush_lock; /* for flush exclusion */ 1621 1622 /* for extent tree cache */ 1623 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1624 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1625 unsigned int max_read_extent_count; /* max read extent count per inode */ 1626 1627 /* The threshold used for hot and warm data seperation*/ 1628 unsigned int hot_data_age_threshold; 1629 unsigned int warm_data_age_threshold; 1630 unsigned int last_age_weight; 1631 1632 /* basic filesystem units */ 1633 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1634 unsigned int log_blocksize; /* log2 block size */ 1635 unsigned int blocksize; /* block size */ 1636 unsigned int root_ino_num; /* root inode number*/ 1637 unsigned int node_ino_num; /* node inode number*/ 1638 unsigned int meta_ino_num; /* meta inode number*/ 1639 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1640 unsigned int blocks_per_seg; /* blocks per segment */ 1641 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1642 unsigned int segs_per_sec; /* segments per section */ 1643 unsigned int secs_per_zone; /* sections per zone */ 1644 unsigned int total_sections; /* total section count */ 1645 unsigned int total_node_count; /* total node block count */ 1646 unsigned int total_valid_node_count; /* valid node block count */ 1647 int dir_level; /* directory level */ 1648 bool readdir_ra; /* readahead inode in readdir */ 1649 u64 max_io_bytes; /* max io bytes to merge IOs */ 1650 1651 block_t user_block_count; /* # of user blocks */ 1652 block_t total_valid_block_count; /* # of valid blocks */ 1653 block_t discard_blks; /* discard command candidats */ 1654 block_t last_valid_block_count; /* for recovery */ 1655 block_t reserved_blocks; /* configurable reserved blocks */ 1656 block_t current_reserved_blocks; /* current reserved blocks */ 1657 1658 /* Additional tracking for no checkpoint mode */ 1659 block_t unusable_block_count; /* # of blocks saved by last cp */ 1660 1661 unsigned int nquota_files; /* # of quota sysfile */ 1662 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1663 1664 /* # of pages, see count_type */ 1665 atomic_t nr_pages[NR_COUNT_TYPE]; 1666 /* # of allocated blocks */ 1667 struct percpu_counter alloc_valid_block_count; 1668 /* # of node block writes as roll forward recovery */ 1669 struct percpu_counter rf_node_block_count; 1670 1671 /* writeback control */ 1672 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1673 1674 /* valid inode count */ 1675 struct percpu_counter total_valid_inode_count; 1676 1677 struct f2fs_mount_info mount_opt; /* mount options */ 1678 1679 /* for cleaning operations */ 1680 struct f2fs_rwsem gc_lock; /* 1681 * semaphore for GC, avoid 1682 * race between GC and GC or CP 1683 */ 1684 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1685 struct atgc_management am; /* atgc management */ 1686 unsigned int cur_victim_sec; /* current victim section num */ 1687 unsigned int gc_mode; /* current GC state */ 1688 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1689 spinlock_t gc_remaining_trials_lock; 1690 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1691 unsigned int gc_remaining_trials; 1692 1693 /* for skip statistic */ 1694 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1695 1696 /* threshold for gc trials on pinned files */ 1697 unsigned short gc_pin_file_threshold; 1698 struct f2fs_rwsem pin_sem; 1699 1700 /* maximum # of trials to find a victim segment for SSR and GC */ 1701 unsigned int max_victim_search; 1702 /* migration granularity of garbage collection, unit: segment */ 1703 unsigned int migration_granularity; 1704 /* migration window granularity of garbage collection, unit: segment */ 1705 unsigned int migration_window_granularity; 1706 1707 /* 1708 * for stat information. 1709 * one is for the LFS mode, and the other is for the SSR mode. 1710 */ 1711#ifdef CONFIG_F2FS_STAT_FS 1712 struct f2fs_stat_info *stat_info; /* FS status information */ 1713 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1714 unsigned int segment_count[2]; /* # of allocated segments */ 1715 unsigned int block_count[2]; /* # of allocated blocks */ 1716 atomic_t inplace_count; /* # of inplace update */ 1717 /* # of lookup extent cache */ 1718 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1719 /* # of hit rbtree extent node */ 1720 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1721 /* # of hit cached extent node */ 1722 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1723 /* # of hit largest extent node in read extent cache */ 1724 atomic64_t read_hit_largest; 1725 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1726 atomic_t inline_inode; /* # of inline_data inodes */ 1727 atomic_t inline_dir; /* # of inline_dentry inodes */ 1728 atomic_t compr_inode; /* # of compressed inodes */ 1729 atomic64_t compr_blocks; /* # of compressed blocks */ 1730 atomic_t swapfile_inode; /* # of swapfile inodes */ 1731 atomic_t atomic_files; /* # of opened atomic file */ 1732 atomic_t max_aw_cnt; /* max # of atomic writes */ 1733 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1734 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1735 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1736 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1737#endif 1738 spinlock_t stat_lock; /* lock for stat operations */ 1739 1740 /* to attach REQ_META|REQ_FUA flags */ 1741 unsigned int data_io_flag; 1742 unsigned int node_io_flag; 1743 1744 /* For sysfs support */ 1745 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1746 struct completion s_kobj_unregister; 1747 1748 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1749 struct completion s_stat_kobj_unregister; 1750 1751 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1752 struct completion s_feature_list_kobj_unregister; 1753 1754 /* For shrinker support */ 1755 struct list_head s_list; 1756 struct mutex umount_mutex; 1757 unsigned int shrinker_run_no; 1758 1759 /* For multi devices */ 1760 int s_ndevs; /* number of devices */ 1761 struct f2fs_dev_info *devs; /* for device list */ 1762 unsigned int dirty_device; /* for checkpoint data flush */ 1763 spinlock_t dev_lock; /* protect dirty_device */ 1764 bool aligned_blksize; /* all devices has the same logical blksize */ 1765 unsigned int first_zoned_segno; /* first zoned segno */ 1766 1767 /* For write statistics */ 1768 u64 sectors_written_start; 1769 u64 kbytes_written; 1770 1771 /* Reference to checksum algorithm driver via cryptoapi */ 1772 struct crypto_shash *s_chksum_driver; 1773 1774 /* Precomputed FS UUID checksum for seeding other checksums */ 1775 __u32 s_chksum_seed; 1776 1777 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1778 1779 /* 1780 * If we are in irq context, let's update error information into 1781 * on-disk superblock in the work. 1782 */ 1783 struct work_struct s_error_work; 1784 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1785 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1786 spinlock_t error_lock; /* protect errors/stop_reason array */ 1787 bool error_dirty; /* errors of sb is dirty */ 1788 1789 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1790 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1791 1792 /* For reclaimed segs statistics per each GC mode */ 1793 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1794 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1795 1796 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1797 1798 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1799 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1800 1801 /* For atomic write statistics */ 1802 atomic64_t current_atomic_write; 1803 s64 peak_atomic_write; 1804 u64 committed_atomic_block; 1805 u64 revoked_atomic_block; 1806 1807#ifdef CONFIG_F2FS_FS_COMPRESSION 1808 struct kmem_cache *page_array_slab; /* page array entry */ 1809 unsigned int page_array_slab_size; /* default page array slab size */ 1810 1811 /* For runtime compression statistics */ 1812 u64 compr_written_block; 1813 u64 compr_saved_block; 1814 u32 compr_new_inode; 1815 1816 /* For compressed block cache */ 1817 struct inode *compress_inode; /* cache compressed blocks */ 1818 unsigned int compress_percent; /* cache page percentage */ 1819 unsigned int compress_watermark; /* cache page watermark */ 1820 atomic_t compress_page_hit; /* cache hit count */ 1821#endif 1822 1823#ifdef CONFIG_F2FS_IOSTAT 1824 /* For app/fs IO statistics */ 1825 spinlock_t iostat_lock; 1826 unsigned long long iostat_count[NR_IO_TYPE]; 1827 unsigned long long iostat_bytes[NR_IO_TYPE]; 1828 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1829 bool iostat_enable; 1830 unsigned long iostat_next_period; 1831 unsigned int iostat_period_ms; 1832 1833 /* For io latency related statistics info in one iostat period */ 1834 spinlock_t iostat_lat_lock; 1835 struct iostat_lat_info *iostat_io_lat; 1836#endif 1837}; 1838 1839/* Definitions to access f2fs_sb_info */ 1840#define SEGS_TO_BLKS(sbi, segs) \ 1841 ((segs) << (sbi)->log_blocks_per_seg) 1842#define BLKS_TO_SEGS(sbi, blks) \ 1843 ((blks) >> (sbi)->log_blocks_per_seg) 1844 1845#define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 1846#define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 1847#define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 1848 1849__printf(3, 4) 1850void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1851 1852#define f2fs_err(sbi, fmt, ...) \ 1853 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1854#define f2fs_warn(sbi, fmt, ...) \ 1855 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1856#define f2fs_notice(sbi, fmt, ...) \ 1857 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1858#define f2fs_info(sbi, fmt, ...) \ 1859 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1860#define f2fs_debug(sbi, fmt, ...) \ 1861 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1862 1863#define f2fs_err_ratelimited(sbi, fmt, ...) \ 1864 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1865#define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1866 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1867#define f2fs_info_ratelimited(sbi, fmt, ...) \ 1868 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1869 1870#ifdef CONFIG_F2FS_FAULT_INJECTION 1871#define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1872 __builtin_return_address(0)) 1873static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1874 const char *func, const char *parent_func) 1875{ 1876 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1877 1878 if (!ffi->inject_rate) 1879 return false; 1880 1881 if (!IS_FAULT_SET(ffi, type)) 1882 return false; 1883 1884 atomic_inc(&ffi->inject_ops); 1885 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1886 atomic_set(&ffi->inject_ops, 0); 1887 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1888 f2fs_fault_name[type], func, parent_func); 1889 return true; 1890 } 1891 return false; 1892} 1893#else 1894static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1895{ 1896 return false; 1897} 1898#endif 1899 1900/* 1901 * Test if the mounted volume is a multi-device volume. 1902 * - For a single regular disk volume, sbi->s_ndevs is 0. 1903 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1904 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1905 */ 1906static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1907{ 1908 return sbi->s_ndevs > 1; 1909} 1910 1911static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1912{ 1913 unsigned long now = jiffies; 1914 1915 sbi->last_time[type] = now; 1916 1917 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1918 if (type == REQ_TIME) { 1919 sbi->last_time[DISCARD_TIME] = now; 1920 sbi->last_time[GC_TIME] = now; 1921 } 1922} 1923 1924static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1925{ 1926 unsigned long interval = sbi->interval_time[type] * HZ; 1927 1928 return time_after(jiffies, sbi->last_time[type] + interval); 1929} 1930 1931static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1932 int type) 1933{ 1934 unsigned long interval = sbi->interval_time[type] * HZ; 1935 unsigned int wait_ms = 0; 1936 long delta; 1937 1938 delta = (sbi->last_time[type] + interval) - jiffies; 1939 if (delta > 0) 1940 wait_ms = jiffies_to_msecs(delta); 1941 1942 return wait_ms; 1943} 1944 1945/* 1946 * Inline functions 1947 */ 1948static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1949 const void *address, unsigned int length) 1950{ 1951 struct { 1952 struct shash_desc shash; 1953 char ctx[4]; 1954 } desc; 1955 int err; 1956 1957 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1958 1959 desc.shash.tfm = sbi->s_chksum_driver; 1960 *(u32 *)desc.ctx = crc; 1961 1962 err = crypto_shash_update(&desc.shash, address, length); 1963 BUG_ON(err); 1964 1965 return *(u32 *)desc.ctx; 1966} 1967 1968static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1969 unsigned int length) 1970{ 1971 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1972} 1973 1974static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1975 void *buf, size_t buf_size) 1976{ 1977 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1978} 1979 1980static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1981 const void *address, unsigned int length) 1982{ 1983 return __f2fs_crc32(sbi, crc, address, length); 1984} 1985 1986static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1987{ 1988 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1989} 1990 1991static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1992{ 1993 return sb->s_fs_info; 1994} 1995 1996static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1997{ 1998 return F2FS_SB(inode->i_sb); 1999} 2000 2001static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2002{ 2003 return F2FS_I_SB(mapping->host); 2004} 2005 2006static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2007{ 2008 return F2FS_M_SB(page_file_mapping(page)); 2009} 2010 2011static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2012{ 2013 return (struct f2fs_super_block *)(sbi->raw_super); 2014} 2015 2016static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2017 pgoff_t index) 2018{ 2019 pgoff_t idx_in_folio = index % (1 << folio_order(folio)); 2020 2021 return (struct f2fs_super_block *) 2022 (page_address(folio_page(folio, idx_in_folio)) + 2023 F2FS_SUPER_OFFSET); 2024} 2025 2026static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2027{ 2028 return (struct f2fs_checkpoint *)(sbi->ckpt); 2029} 2030 2031static inline struct f2fs_node *F2FS_NODE(struct page *page) 2032{ 2033 return (struct f2fs_node *)page_address(page); 2034} 2035 2036static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2037{ 2038 return &((struct f2fs_node *)page_address(page))->i; 2039} 2040 2041static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2042{ 2043 return (struct f2fs_nm_info *)(sbi->nm_info); 2044} 2045 2046static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2047{ 2048 return (struct f2fs_sm_info *)(sbi->sm_info); 2049} 2050 2051static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2052{ 2053 return (struct sit_info *)(SM_I(sbi)->sit_info); 2054} 2055 2056static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2057{ 2058 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2059} 2060 2061static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2062{ 2063 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2064} 2065 2066static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2067{ 2068 return sbi->meta_inode->i_mapping; 2069} 2070 2071static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2072{ 2073 return sbi->node_inode->i_mapping; 2074} 2075 2076static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2077{ 2078 return test_bit(type, &sbi->s_flag); 2079} 2080 2081static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2082{ 2083 set_bit(type, &sbi->s_flag); 2084} 2085 2086static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2087{ 2088 clear_bit(type, &sbi->s_flag); 2089} 2090 2091static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2092{ 2093 return le64_to_cpu(cp->checkpoint_ver); 2094} 2095 2096static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2097{ 2098 if (type < F2FS_MAX_QUOTAS) 2099 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2100 return 0; 2101} 2102 2103static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2104{ 2105 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2106 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2107} 2108 2109static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2110{ 2111 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2112 2113 return ckpt_flags & f; 2114} 2115 2116static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2117{ 2118 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2119} 2120 2121static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2122{ 2123 unsigned int ckpt_flags; 2124 2125 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2126 ckpt_flags |= f; 2127 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2128} 2129 2130static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2131{ 2132 unsigned long flags; 2133 2134 spin_lock_irqsave(&sbi->cp_lock, flags); 2135 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2136 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2137} 2138 2139static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2140{ 2141 unsigned int ckpt_flags; 2142 2143 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2144 ckpt_flags &= (~f); 2145 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2146} 2147 2148static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2149{ 2150 unsigned long flags; 2151 2152 spin_lock_irqsave(&sbi->cp_lock, flags); 2153 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2154 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2155} 2156 2157#define init_f2fs_rwsem(sem) \ 2158do { \ 2159 static struct lock_class_key __key; \ 2160 \ 2161 __init_f2fs_rwsem((sem), #sem, &__key); \ 2162} while (0) 2163 2164static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2165 const char *sem_name, struct lock_class_key *key) 2166{ 2167 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2168#ifdef CONFIG_F2FS_UNFAIR_RWSEM 2169 init_waitqueue_head(&sem->read_waiters); 2170#endif 2171} 2172 2173static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2174{ 2175 return rwsem_is_locked(&sem->internal_rwsem); 2176} 2177 2178static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2179{ 2180 return rwsem_is_contended(&sem->internal_rwsem); 2181} 2182 2183static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2184{ 2185#ifdef CONFIG_F2FS_UNFAIR_RWSEM 2186 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2187#else 2188 down_read(&sem->internal_rwsem); 2189#endif 2190} 2191 2192static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2193{ 2194 return down_read_trylock(&sem->internal_rwsem); 2195} 2196 2197static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2198{ 2199 up_read(&sem->internal_rwsem); 2200} 2201 2202static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2203{ 2204 down_write(&sem->internal_rwsem); 2205} 2206 2207#ifdef CONFIG_DEBUG_LOCK_ALLOC 2208static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2209{ 2210 down_read_nested(&sem->internal_rwsem, subclass); 2211} 2212 2213static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2214{ 2215 down_write_nested(&sem->internal_rwsem, subclass); 2216} 2217#else 2218#define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2219#define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2220#endif 2221 2222static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2223{ 2224 return down_write_trylock(&sem->internal_rwsem); 2225} 2226 2227static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2228{ 2229 up_write(&sem->internal_rwsem); 2230#ifdef CONFIG_F2FS_UNFAIR_RWSEM 2231 wake_up_all(&sem->read_waiters); 2232#endif 2233} 2234 2235static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2236{ 2237 f2fs_down_read(&sbi->cp_rwsem); 2238} 2239 2240static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2241{ 2242 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2243 return 0; 2244 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2245} 2246 2247static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2248{ 2249 f2fs_up_read(&sbi->cp_rwsem); 2250} 2251 2252static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2253{ 2254 f2fs_down_write(&sbi->cp_rwsem); 2255} 2256 2257static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2258{ 2259 f2fs_up_write(&sbi->cp_rwsem); 2260} 2261 2262static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2263{ 2264 int reason = CP_SYNC; 2265 2266 if (test_opt(sbi, FASTBOOT)) 2267 reason = CP_FASTBOOT; 2268 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2269 reason = CP_UMOUNT; 2270 return reason; 2271} 2272 2273static inline bool __remain_node_summaries(int reason) 2274{ 2275 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2276} 2277 2278static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2279{ 2280 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2281 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2282} 2283 2284/* 2285 * Check whether the inode has blocks or not 2286 */ 2287static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2288{ 2289 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2290 2291 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2292} 2293 2294static inline bool f2fs_has_xattr_block(unsigned int ofs) 2295{ 2296 return ofs == XATTR_NODE_OFFSET; 2297} 2298 2299static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2300 struct inode *inode, bool cap) 2301{ 2302 if (!inode) 2303 return true; 2304 if (!test_opt(sbi, RESERVE_ROOT)) 2305 return false; 2306 if (IS_NOQUOTA(inode)) 2307 return true; 2308 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2309 return true; 2310 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2311 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2312 return true; 2313 if (cap && capable(CAP_SYS_RESOURCE)) 2314 return true; 2315 return false; 2316} 2317 2318static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2319 struct inode *inode, bool cap) 2320{ 2321 block_t avail_user_block_count; 2322 2323 avail_user_block_count = sbi->user_block_count - 2324 sbi->current_reserved_blocks; 2325 2326 if (!__allow_reserved_blocks(sbi, inode, cap)) 2327 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2328 2329 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2330 if (avail_user_block_count > sbi->unusable_block_count) 2331 avail_user_block_count -= sbi->unusable_block_count; 2332 else 2333 avail_user_block_count = 0; 2334 } 2335 2336 return avail_user_block_count; 2337} 2338 2339static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2340static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2341 struct inode *inode, blkcnt_t *count, bool partial) 2342{ 2343 long long diff = 0, release = 0; 2344 block_t avail_user_block_count; 2345 int ret; 2346 2347 ret = dquot_reserve_block(inode, *count); 2348 if (ret) 2349 return ret; 2350 2351 if (time_to_inject(sbi, FAULT_BLOCK)) { 2352 release = *count; 2353 goto release_quota; 2354 } 2355 2356 /* 2357 * let's increase this in prior to actual block count change in order 2358 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2359 */ 2360 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2361 2362 spin_lock(&sbi->stat_lock); 2363 2364 avail_user_block_count = get_available_block_count(sbi, inode, true); 2365 diff = (long long)sbi->total_valid_block_count + *count - 2366 avail_user_block_count; 2367 if (unlikely(diff > 0)) { 2368 if (!partial) { 2369 spin_unlock(&sbi->stat_lock); 2370 release = *count; 2371 goto enospc; 2372 } 2373 if (diff > *count) 2374 diff = *count; 2375 *count -= diff; 2376 release = diff; 2377 if (!*count) { 2378 spin_unlock(&sbi->stat_lock); 2379 goto enospc; 2380 } 2381 } 2382 sbi->total_valid_block_count += (block_t)(*count); 2383 2384 spin_unlock(&sbi->stat_lock); 2385 2386 if (unlikely(release)) { 2387 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2388 dquot_release_reservation_block(inode, release); 2389 } 2390 f2fs_i_blocks_write(inode, *count, true, true); 2391 return 0; 2392 2393enospc: 2394 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2395release_quota: 2396 dquot_release_reservation_block(inode, release); 2397 return -ENOSPC; 2398} 2399 2400#define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2401static inline bool page_private_##name(struct page *page) \ 2402{ \ 2403 return PagePrivate(page) && \ 2404 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2405 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2406} 2407 2408#define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2409static inline void set_page_private_##name(struct page *page) \ 2410{ \ 2411 if (!PagePrivate(page)) \ 2412 attach_page_private(page, (void *)0); \ 2413 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2414 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2415} 2416 2417#define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2418static inline void clear_page_private_##name(struct page *page) \ 2419{ \ 2420 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2421 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2422 detach_page_private(page); \ 2423} 2424 2425PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2426PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2427PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2428PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2429 2430PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2431PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2432PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2433PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2434 2435PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2436PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2437PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2438PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2439 2440static inline unsigned long get_page_private_data(struct page *page) 2441{ 2442 unsigned long data = page_private(page); 2443 2444 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2445 return 0; 2446 return data >> PAGE_PRIVATE_MAX; 2447} 2448 2449static inline void set_page_private_data(struct page *page, unsigned long data) 2450{ 2451 if (!PagePrivate(page)) 2452 attach_page_private(page, (void *)0); 2453 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2454 page_private(page) |= data << PAGE_PRIVATE_MAX; 2455} 2456 2457static inline void clear_page_private_data(struct page *page) 2458{ 2459 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2460 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2461 detach_page_private(page); 2462} 2463 2464static inline void clear_page_private_all(struct page *page) 2465{ 2466 clear_page_private_data(page); 2467 clear_page_private_reference(page); 2468 clear_page_private_gcing(page); 2469 clear_page_private_inline(page); 2470 clear_page_private_atomic(page); 2471 2472 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2473} 2474 2475static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2476 struct inode *inode, 2477 block_t count) 2478{ 2479 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2480 2481 spin_lock(&sbi->stat_lock); 2482 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2483 sbi->total_valid_block_count -= (block_t)count; 2484 if (sbi->reserved_blocks && 2485 sbi->current_reserved_blocks < sbi->reserved_blocks) 2486 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2487 sbi->current_reserved_blocks + count); 2488 spin_unlock(&sbi->stat_lock); 2489 if (unlikely(inode->i_blocks < sectors)) { 2490 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2491 inode->i_ino, 2492 (unsigned long long)inode->i_blocks, 2493 (unsigned long long)sectors); 2494 set_sbi_flag(sbi, SBI_NEED_FSCK); 2495 return; 2496 } 2497 f2fs_i_blocks_write(inode, count, false, true); 2498} 2499 2500static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2501{ 2502 atomic_inc(&sbi->nr_pages[count_type]); 2503 2504 if (count_type == F2FS_DIRTY_DENTS || 2505 count_type == F2FS_DIRTY_NODES || 2506 count_type == F2FS_DIRTY_META || 2507 count_type == F2FS_DIRTY_QDATA || 2508 count_type == F2FS_DIRTY_IMETA) 2509 set_sbi_flag(sbi, SBI_IS_DIRTY); 2510} 2511 2512static inline void inode_inc_dirty_pages(struct inode *inode) 2513{ 2514 atomic_inc(&F2FS_I(inode)->dirty_pages); 2515 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2516 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2517 if (IS_NOQUOTA(inode)) 2518 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2519} 2520 2521static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2522{ 2523 atomic_dec(&sbi->nr_pages[count_type]); 2524} 2525 2526static inline void inode_dec_dirty_pages(struct inode *inode) 2527{ 2528 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2529 !S_ISLNK(inode->i_mode)) 2530 return; 2531 2532 atomic_dec(&F2FS_I(inode)->dirty_pages); 2533 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2534 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2535 if (IS_NOQUOTA(inode)) 2536 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2537} 2538 2539static inline void inc_atomic_write_cnt(struct inode *inode) 2540{ 2541 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2542 struct f2fs_inode_info *fi = F2FS_I(inode); 2543 u64 current_write; 2544 2545 fi->atomic_write_cnt++; 2546 atomic64_inc(&sbi->current_atomic_write); 2547 current_write = atomic64_read(&sbi->current_atomic_write); 2548 if (current_write > sbi->peak_atomic_write) 2549 sbi->peak_atomic_write = current_write; 2550} 2551 2552static inline void release_atomic_write_cnt(struct inode *inode) 2553{ 2554 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2555 struct f2fs_inode_info *fi = F2FS_I(inode); 2556 2557 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2558 fi->atomic_write_cnt = 0; 2559} 2560 2561static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2562{ 2563 return atomic_read(&sbi->nr_pages[count_type]); 2564} 2565 2566static inline int get_dirty_pages(struct inode *inode) 2567{ 2568 return atomic_read(&F2FS_I(inode)->dirty_pages); 2569} 2570 2571static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2572{ 2573 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2574 BLKS_PER_SEC(sbi)); 2575} 2576 2577static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2578{ 2579 return sbi->total_valid_block_count; 2580} 2581 2582static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2583{ 2584 return sbi->discard_blks; 2585} 2586 2587static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2588{ 2589 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2590 2591 /* return NAT or SIT bitmap */ 2592 if (flag == NAT_BITMAP) 2593 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2594 else if (flag == SIT_BITMAP) 2595 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2596 2597 return 0; 2598} 2599 2600static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2601{ 2602 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2603} 2604 2605static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2606{ 2607 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2608 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2609 int offset; 2610 2611 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2612 offset = (flag == SIT_BITMAP) ? 2613 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2614 /* 2615 * if large_nat_bitmap feature is enabled, leave checksum 2616 * protection for all nat/sit bitmaps. 2617 */ 2618 return tmp_ptr + offset + sizeof(__le32); 2619 } 2620 2621 if (__cp_payload(sbi) > 0) { 2622 if (flag == NAT_BITMAP) 2623 return tmp_ptr; 2624 else 2625 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2626 } else { 2627 offset = (flag == NAT_BITMAP) ? 2628 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2629 return tmp_ptr + offset; 2630 } 2631} 2632 2633static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2634{ 2635 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2636 2637 if (sbi->cur_cp_pack == 2) 2638 start_addr += BLKS_PER_SEG(sbi); 2639 return start_addr; 2640} 2641 2642static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2643{ 2644 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2645 2646 if (sbi->cur_cp_pack == 1) 2647 start_addr += BLKS_PER_SEG(sbi); 2648 return start_addr; 2649} 2650 2651static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2652{ 2653 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2654} 2655 2656static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2657{ 2658 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2659} 2660 2661extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2662static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2663 struct inode *inode, bool is_inode) 2664{ 2665 block_t valid_block_count; 2666 unsigned int valid_node_count; 2667 unsigned int avail_user_block_count; 2668 int err; 2669 2670 if (is_inode) { 2671 if (inode) { 2672 err = dquot_alloc_inode(inode); 2673 if (err) 2674 return err; 2675 } 2676 } else { 2677 err = dquot_reserve_block(inode, 1); 2678 if (err) 2679 return err; 2680 } 2681 2682 if (time_to_inject(sbi, FAULT_BLOCK)) 2683 goto enospc; 2684 2685 spin_lock(&sbi->stat_lock); 2686 2687 valid_block_count = sbi->total_valid_block_count + 1; 2688 avail_user_block_count = get_available_block_count(sbi, inode, false); 2689 2690 if (unlikely(valid_block_count > avail_user_block_count)) { 2691 spin_unlock(&sbi->stat_lock); 2692 goto enospc; 2693 } 2694 2695 valid_node_count = sbi->total_valid_node_count + 1; 2696 if (unlikely(valid_node_count > sbi->total_node_count)) { 2697 spin_unlock(&sbi->stat_lock); 2698 goto enospc; 2699 } 2700 2701 sbi->total_valid_node_count++; 2702 sbi->total_valid_block_count++; 2703 spin_unlock(&sbi->stat_lock); 2704 2705 if (inode) { 2706 if (is_inode) 2707 f2fs_mark_inode_dirty_sync(inode, true); 2708 else 2709 f2fs_i_blocks_write(inode, 1, true, true); 2710 } 2711 2712 percpu_counter_inc(&sbi->alloc_valid_block_count); 2713 return 0; 2714 2715enospc: 2716 if (is_inode) { 2717 if (inode) 2718 dquot_free_inode(inode); 2719 } else { 2720 dquot_release_reservation_block(inode, 1); 2721 } 2722 return -ENOSPC; 2723} 2724 2725static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2726 struct inode *inode, bool is_inode) 2727{ 2728 spin_lock(&sbi->stat_lock); 2729 2730 if (unlikely(!sbi->total_valid_block_count || 2731 !sbi->total_valid_node_count)) { 2732 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2733 sbi->total_valid_block_count, 2734 sbi->total_valid_node_count); 2735 set_sbi_flag(sbi, SBI_NEED_FSCK); 2736 } else { 2737 sbi->total_valid_block_count--; 2738 sbi->total_valid_node_count--; 2739 } 2740 2741 if (sbi->reserved_blocks && 2742 sbi->current_reserved_blocks < sbi->reserved_blocks) 2743 sbi->current_reserved_blocks++; 2744 2745 spin_unlock(&sbi->stat_lock); 2746 2747 if (is_inode) { 2748 dquot_free_inode(inode); 2749 } else { 2750 if (unlikely(inode->i_blocks == 0)) { 2751 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2752 inode->i_ino, 2753 (unsigned long long)inode->i_blocks); 2754 set_sbi_flag(sbi, SBI_NEED_FSCK); 2755 return; 2756 } 2757 f2fs_i_blocks_write(inode, 1, false, true); 2758 } 2759} 2760 2761static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2762{ 2763 return sbi->total_valid_node_count; 2764} 2765 2766static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2767{ 2768 percpu_counter_inc(&sbi->total_valid_inode_count); 2769} 2770 2771static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2772{ 2773 percpu_counter_dec(&sbi->total_valid_inode_count); 2774} 2775 2776static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2777{ 2778 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2779} 2780 2781static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2782 pgoff_t index, bool for_write) 2783{ 2784 struct page *page; 2785 unsigned int flags; 2786 2787 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2788 if (!for_write) 2789 page = find_get_page_flags(mapping, index, 2790 FGP_LOCK | FGP_ACCESSED); 2791 else 2792 page = find_lock_page(mapping, index); 2793 if (page) 2794 return page; 2795 2796 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2797 return NULL; 2798 } 2799 2800 if (!for_write) 2801 return grab_cache_page(mapping, index); 2802 2803 flags = memalloc_nofs_save(); 2804 page = grab_cache_page_write_begin(mapping, index); 2805 memalloc_nofs_restore(flags); 2806 2807 return page; 2808} 2809 2810static inline struct page *f2fs_pagecache_get_page( 2811 struct address_space *mapping, pgoff_t index, 2812 fgf_t fgp_flags, gfp_t gfp_mask) 2813{ 2814 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2815 return NULL; 2816 2817 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2818} 2819 2820static inline void f2fs_put_page(struct page *page, int unlock) 2821{ 2822 if (!page) 2823 return; 2824 2825 if (unlock) { 2826 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2827 unlock_page(page); 2828 } 2829 put_page(page); 2830} 2831 2832static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2833{ 2834 if (dn->node_page) 2835 f2fs_put_page(dn->node_page, 1); 2836 if (dn->inode_page && dn->node_page != dn->inode_page) 2837 f2fs_put_page(dn->inode_page, 0); 2838 dn->node_page = NULL; 2839 dn->inode_page = NULL; 2840} 2841 2842static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2843 size_t size) 2844{ 2845 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2846} 2847 2848static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2849 gfp_t flags) 2850{ 2851 void *entry; 2852 2853 entry = kmem_cache_alloc(cachep, flags); 2854 if (!entry) 2855 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2856 return entry; 2857} 2858 2859static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2860 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2861{ 2862 if (nofail) 2863 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2864 2865 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2866 return NULL; 2867 2868 return kmem_cache_alloc(cachep, flags); 2869} 2870 2871static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2872{ 2873 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2874 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2875 get_pages(sbi, F2FS_WB_CP_DATA) || 2876 get_pages(sbi, F2FS_DIO_READ) || 2877 get_pages(sbi, F2FS_DIO_WRITE)) 2878 return true; 2879 2880 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2881 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2882 return true; 2883 2884 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2885 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2886 return true; 2887 return false; 2888} 2889 2890static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 2891{ 2892 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ); 2893} 2894 2895static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2896{ 2897 bool zoned_gc = (type == GC_TIME && 2898 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 2899 2900 if (sbi->gc_mode == GC_URGENT_HIGH) 2901 return true; 2902 2903 if (zoned_gc) { 2904 if (is_inflight_read_io(sbi)) 2905 return false; 2906 } else { 2907 if (is_inflight_io(sbi, type)) 2908 return false; 2909 } 2910 2911 if (sbi->gc_mode == GC_URGENT_MID) 2912 return true; 2913 2914 if (sbi->gc_mode == GC_URGENT_LOW && 2915 (type == DISCARD_TIME || type == GC_TIME)) 2916 return true; 2917 2918 if (zoned_gc) 2919 return true; 2920 2921 return f2fs_time_over(sbi, type); 2922} 2923 2924static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2925 unsigned long index, void *item) 2926{ 2927 while (radix_tree_insert(root, index, item)) 2928 cond_resched(); 2929} 2930 2931#define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2932 2933static inline bool IS_INODE(struct page *page) 2934{ 2935 struct f2fs_node *p = F2FS_NODE(page); 2936 2937 return RAW_IS_INODE(p); 2938} 2939 2940static inline int offset_in_addr(struct f2fs_inode *i) 2941{ 2942 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2943 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2944} 2945 2946static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2947{ 2948 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2949} 2950 2951static inline int f2fs_has_extra_attr(struct inode *inode); 2952static inline unsigned int get_dnode_base(struct inode *inode, 2953 struct page *node_page) 2954{ 2955 if (!IS_INODE(node_page)) 2956 return 0; 2957 2958 return inode ? get_extra_isize(inode) : 2959 offset_in_addr(&F2FS_NODE(node_page)->i); 2960} 2961 2962static inline __le32 *get_dnode_addr(struct inode *inode, 2963 struct page *node_page) 2964{ 2965 return blkaddr_in_node(F2FS_NODE(node_page)) + 2966 get_dnode_base(inode, node_page); 2967} 2968 2969static inline block_t data_blkaddr(struct inode *inode, 2970 struct page *node_page, unsigned int offset) 2971{ 2972 return le32_to_cpu(*(get_dnode_addr(inode, node_page) + offset)); 2973} 2974 2975static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2976{ 2977 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2978} 2979 2980static inline int f2fs_test_bit(unsigned int nr, char *addr) 2981{ 2982 int mask; 2983 2984 addr += (nr >> 3); 2985 mask = BIT(7 - (nr & 0x07)); 2986 return mask & *addr; 2987} 2988 2989static inline void f2fs_set_bit(unsigned int nr, char *addr) 2990{ 2991 int mask; 2992 2993 addr += (nr >> 3); 2994 mask = BIT(7 - (nr & 0x07)); 2995 *addr |= mask; 2996} 2997 2998static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2999{ 3000 int mask; 3001 3002 addr += (nr >> 3); 3003 mask = BIT(7 - (nr & 0x07)); 3004 *addr &= ~mask; 3005} 3006 3007static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3008{ 3009 int mask; 3010 int ret; 3011 3012 addr += (nr >> 3); 3013 mask = BIT(7 - (nr & 0x07)); 3014 ret = mask & *addr; 3015 *addr |= mask; 3016 return ret; 3017} 3018 3019static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3020{ 3021 int mask; 3022 int ret; 3023 3024 addr += (nr >> 3); 3025 mask = BIT(7 - (nr & 0x07)); 3026 ret = mask & *addr; 3027 *addr &= ~mask; 3028 return ret; 3029} 3030 3031static inline void f2fs_change_bit(unsigned int nr, char *addr) 3032{ 3033 int mask; 3034 3035 addr += (nr >> 3); 3036 mask = BIT(7 - (nr & 0x07)); 3037 *addr ^= mask; 3038} 3039 3040/* 3041 * On-disk inode flags (f2fs_inode::i_flags) 3042 */ 3043#define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3044#define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3045#define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3046#define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3047#define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3048#define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3049#define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3050#define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3051#define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3052#define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3053#define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3054#define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3055 3056#define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3057 3058/* Flags that should be inherited by new inodes from their parent. */ 3059#define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3060 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3061 F2FS_CASEFOLD_FL) 3062 3063/* Flags that are appropriate for regular files (all but dir-specific ones). */ 3064#define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3065 F2FS_CASEFOLD_FL)) 3066 3067/* Flags that are appropriate for non-directories/regular files. */ 3068#define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3069 3070#define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3071 3072static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3073{ 3074 if (S_ISDIR(mode)) 3075 return flags; 3076 else if (S_ISREG(mode)) 3077 return flags & F2FS_REG_FLMASK; 3078 else 3079 return flags & F2FS_OTHER_FLMASK; 3080} 3081 3082static inline void __mark_inode_dirty_flag(struct inode *inode, 3083 int flag, bool set) 3084{ 3085 switch (flag) { 3086 case FI_INLINE_XATTR: 3087 case FI_INLINE_DATA: 3088 case FI_INLINE_DENTRY: 3089 case FI_NEW_INODE: 3090 if (set) 3091 return; 3092 fallthrough; 3093 case FI_DATA_EXIST: 3094 case FI_PIN_FILE: 3095 case FI_COMPRESS_RELEASED: 3096 f2fs_mark_inode_dirty_sync(inode, true); 3097 } 3098} 3099 3100static inline void set_inode_flag(struct inode *inode, int flag) 3101{ 3102 set_bit(flag, F2FS_I(inode)->flags); 3103 __mark_inode_dirty_flag(inode, flag, true); 3104} 3105 3106static inline int is_inode_flag_set(struct inode *inode, int flag) 3107{ 3108 return test_bit(flag, F2FS_I(inode)->flags); 3109} 3110 3111static inline void clear_inode_flag(struct inode *inode, int flag) 3112{ 3113 clear_bit(flag, F2FS_I(inode)->flags); 3114 __mark_inode_dirty_flag(inode, flag, false); 3115} 3116 3117static inline bool f2fs_verity_in_progress(struct inode *inode) 3118{ 3119 return IS_ENABLED(CONFIG_FS_VERITY) && 3120 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3121} 3122 3123static inline void set_acl_inode(struct inode *inode, umode_t mode) 3124{ 3125 F2FS_I(inode)->i_acl_mode = mode; 3126 set_inode_flag(inode, FI_ACL_MODE); 3127 f2fs_mark_inode_dirty_sync(inode, false); 3128} 3129 3130static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3131{ 3132 if (inc) 3133 inc_nlink(inode); 3134 else 3135 drop_nlink(inode); 3136 f2fs_mark_inode_dirty_sync(inode, true); 3137} 3138 3139static inline void f2fs_i_blocks_write(struct inode *inode, 3140 block_t diff, bool add, bool claim) 3141{ 3142 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3143 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3144 3145 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3146 if (add) { 3147 if (claim) 3148 dquot_claim_block(inode, diff); 3149 else 3150 dquot_alloc_block_nofail(inode, diff); 3151 } else { 3152 dquot_free_block(inode, diff); 3153 } 3154 3155 f2fs_mark_inode_dirty_sync(inode, true); 3156 if (clean || recover) 3157 set_inode_flag(inode, FI_AUTO_RECOVER); 3158} 3159 3160static inline bool f2fs_is_atomic_file(struct inode *inode); 3161 3162static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3163{ 3164 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3165 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3166 3167 if (i_size_read(inode) == i_size) 3168 return; 3169 3170 i_size_write(inode, i_size); 3171 3172 if (f2fs_is_atomic_file(inode)) 3173 return; 3174 3175 f2fs_mark_inode_dirty_sync(inode, true); 3176 if (clean || recover) 3177 set_inode_flag(inode, FI_AUTO_RECOVER); 3178} 3179 3180static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3181{ 3182 F2FS_I(inode)->i_current_depth = depth; 3183 f2fs_mark_inode_dirty_sync(inode, true); 3184} 3185 3186static inline void f2fs_i_gc_failures_write(struct inode *inode, 3187 unsigned int count) 3188{ 3189 F2FS_I(inode)->i_gc_failures = count; 3190 f2fs_mark_inode_dirty_sync(inode, true); 3191} 3192 3193static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3194{ 3195 F2FS_I(inode)->i_xattr_nid = xnid; 3196 f2fs_mark_inode_dirty_sync(inode, true); 3197} 3198 3199static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3200{ 3201 F2FS_I(inode)->i_pino = pino; 3202 f2fs_mark_inode_dirty_sync(inode, true); 3203} 3204 3205static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3206{ 3207 struct f2fs_inode_info *fi = F2FS_I(inode); 3208 3209 if (ri->i_inline & F2FS_INLINE_XATTR) 3210 set_bit(FI_INLINE_XATTR, fi->flags); 3211 if (ri->i_inline & F2FS_INLINE_DATA) 3212 set_bit(FI_INLINE_DATA, fi->flags); 3213 if (ri->i_inline & F2FS_INLINE_DENTRY) 3214 set_bit(FI_INLINE_DENTRY, fi->flags); 3215 if (ri->i_inline & F2FS_DATA_EXIST) 3216 set_bit(FI_DATA_EXIST, fi->flags); 3217 if (ri->i_inline & F2FS_EXTRA_ATTR) 3218 set_bit(FI_EXTRA_ATTR, fi->flags); 3219 if (ri->i_inline & F2FS_PIN_FILE) 3220 set_bit(FI_PIN_FILE, fi->flags); 3221 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3222 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3223} 3224 3225static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3226{ 3227 ri->i_inline = 0; 3228 3229 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3230 ri->i_inline |= F2FS_INLINE_XATTR; 3231 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3232 ri->i_inline |= F2FS_INLINE_DATA; 3233 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3234 ri->i_inline |= F2FS_INLINE_DENTRY; 3235 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3236 ri->i_inline |= F2FS_DATA_EXIST; 3237 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3238 ri->i_inline |= F2FS_EXTRA_ATTR; 3239 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3240 ri->i_inline |= F2FS_PIN_FILE; 3241 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3242 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3243} 3244 3245static inline int f2fs_has_extra_attr(struct inode *inode) 3246{ 3247 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3248} 3249 3250static inline int f2fs_has_inline_xattr(struct inode *inode) 3251{ 3252 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3253} 3254 3255static inline int f2fs_compressed_file(struct inode *inode) 3256{ 3257 return S_ISREG(inode->i_mode) && 3258 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3259} 3260 3261static inline bool f2fs_need_compress_data(struct inode *inode) 3262{ 3263 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3264 3265 if (!f2fs_compressed_file(inode)) 3266 return false; 3267 3268 if (compress_mode == COMPR_MODE_FS) 3269 return true; 3270 else if (compress_mode == COMPR_MODE_USER && 3271 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3272 return true; 3273 3274 return false; 3275} 3276 3277static inline unsigned int addrs_per_page(struct inode *inode, 3278 bool is_inode) 3279{ 3280 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) - 3281 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK; 3282 3283 if (f2fs_compressed_file(inode)) 3284 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3285 return addrs; 3286} 3287 3288static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3289{ 3290 struct f2fs_inode *ri = F2FS_INODE(page); 3291 3292 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3293 get_inline_xattr_addrs(inode)]); 3294} 3295 3296static inline int inline_xattr_size(struct inode *inode) 3297{ 3298 if (f2fs_has_inline_xattr(inode)) 3299 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3300 return 0; 3301} 3302 3303/* 3304 * Notice: check inline_data flag without inode page lock is unsafe. 3305 * It could change at any time by f2fs_convert_inline_page(). 3306 */ 3307static inline int f2fs_has_inline_data(struct inode *inode) 3308{ 3309 return is_inode_flag_set(inode, FI_INLINE_DATA); 3310} 3311 3312static inline int f2fs_exist_data(struct inode *inode) 3313{ 3314 return is_inode_flag_set(inode, FI_DATA_EXIST); 3315} 3316 3317static inline int f2fs_is_mmap_file(struct inode *inode) 3318{ 3319 return is_inode_flag_set(inode, FI_MMAP_FILE); 3320} 3321 3322static inline bool f2fs_is_pinned_file(struct inode *inode) 3323{ 3324 return is_inode_flag_set(inode, FI_PIN_FILE); 3325} 3326 3327static inline bool f2fs_is_atomic_file(struct inode *inode) 3328{ 3329 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3330} 3331 3332static inline bool f2fs_is_cow_file(struct inode *inode) 3333{ 3334 return is_inode_flag_set(inode, FI_COW_FILE); 3335} 3336 3337static inline void *inline_data_addr(struct inode *inode, struct page *page) 3338{ 3339 __le32 *addr = get_dnode_addr(inode, page); 3340 3341 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3342} 3343 3344static inline int f2fs_has_inline_dentry(struct inode *inode) 3345{ 3346 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3347} 3348 3349static inline int is_file(struct inode *inode, int type) 3350{ 3351 return F2FS_I(inode)->i_advise & type; 3352} 3353 3354static inline void set_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 void clear_file(struct inode *inode, int type) 3363{ 3364 if (!is_file(inode, type)) 3365 return; 3366 F2FS_I(inode)->i_advise &= ~type; 3367 f2fs_mark_inode_dirty_sync(inode, true); 3368} 3369 3370static inline bool f2fs_is_time_consistent(struct inode *inode) 3371{ 3372 struct timespec64 ts = inode_get_atime(inode); 3373 3374 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3375 return false; 3376 ts = inode_get_ctime(inode); 3377 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3378 return false; 3379 ts = inode_get_mtime(inode); 3380 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3381 return false; 3382 return true; 3383} 3384 3385static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3386{ 3387 bool ret; 3388 3389 if (dsync) { 3390 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3391 3392 spin_lock(&sbi->inode_lock[DIRTY_META]); 3393 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3394 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3395 return ret; 3396 } 3397 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3398 file_keep_isize(inode) || 3399 i_size_read(inode) & ~PAGE_MASK) 3400 return false; 3401 3402 if (!f2fs_is_time_consistent(inode)) 3403 return false; 3404 3405 spin_lock(&F2FS_I(inode)->i_size_lock); 3406 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3407 spin_unlock(&F2FS_I(inode)->i_size_lock); 3408 3409 return ret; 3410} 3411 3412static inline bool f2fs_readonly(struct super_block *sb) 3413{ 3414 return sb_rdonly(sb); 3415} 3416 3417static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3418{ 3419 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3420} 3421 3422static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3423 size_t size, gfp_t flags) 3424{ 3425 if (time_to_inject(sbi, FAULT_KMALLOC)) 3426 return NULL; 3427 3428 return kmalloc(size, flags); 3429} 3430 3431static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3432{ 3433 if (time_to_inject(sbi, FAULT_KMALLOC)) 3434 return NULL; 3435 3436 return __getname(); 3437} 3438 3439static inline void f2fs_putname(char *buf) 3440{ 3441 __putname(buf); 3442} 3443 3444static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3445 size_t size, gfp_t flags) 3446{ 3447 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3448} 3449 3450static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3451 size_t size, gfp_t flags) 3452{ 3453 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3454 return NULL; 3455 3456 return kvmalloc(size, flags); 3457} 3458 3459static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3460 size_t size, gfp_t flags) 3461{ 3462 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3463} 3464 3465static inline int get_extra_isize(struct inode *inode) 3466{ 3467 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3468} 3469 3470static inline int get_inline_xattr_addrs(struct inode *inode) 3471{ 3472 return F2FS_I(inode)->i_inline_xattr_size; 3473} 3474 3475#define f2fs_get_inode_mode(i) \ 3476 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3477 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3478 3479#define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3480 3481#define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3482 (offsetof(struct f2fs_inode, i_extra_end) - \ 3483 offsetof(struct f2fs_inode, i_extra_isize)) \ 3484 3485#define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3486#define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3487 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3488 sizeof((f2fs_inode)->field)) \ 3489 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3490 3491#define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3492 3493#define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3494 3495bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3496 block_t blkaddr, int type); 3497static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3498 block_t blkaddr, int type) 3499{ 3500 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3501 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3502 blkaddr, type); 3503} 3504 3505static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3506{ 3507 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3508 blkaddr == COMPRESS_ADDR) 3509 return false; 3510 return true; 3511} 3512 3513/* 3514 * file.c 3515 */ 3516int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3517int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3518int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3519int f2fs_truncate(struct inode *inode); 3520int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3521 struct kstat *stat, u32 request_mask, unsigned int flags); 3522int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3523 struct iattr *attr); 3524int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3525void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3526int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3527 bool readonly, bool need_lock); 3528int f2fs_precache_extents(struct inode *inode); 3529int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3530int f2fs_fileattr_set(struct mnt_idmap *idmap, 3531 struct dentry *dentry, struct fileattr *fa); 3532long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3533long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3534int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3535int f2fs_pin_file_control(struct inode *inode, bool inc); 3536 3537/* 3538 * inode.c 3539 */ 3540void f2fs_set_inode_flags(struct inode *inode); 3541bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3542void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3543struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3544struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3545int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3546void f2fs_update_inode(struct inode *inode, struct page *node_page); 3547void f2fs_update_inode_page(struct inode *inode); 3548int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3549void f2fs_evict_inode(struct inode *inode); 3550void f2fs_handle_failed_inode(struct inode *inode); 3551 3552/* 3553 * namei.c 3554 */ 3555int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3556 bool hot, bool set); 3557struct dentry *f2fs_get_parent(struct dentry *child); 3558int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3559 struct inode **new_inode); 3560 3561/* 3562 * dir.c 3563 */ 3564#if IS_ENABLED(CONFIG_UNICODE) 3565int f2fs_init_casefolded_name(const struct inode *dir, 3566 struct f2fs_filename *fname); 3567void f2fs_free_casefolded_name(struct f2fs_filename *fname); 3568#else 3569static inline int f2fs_init_casefolded_name(const struct inode *dir, 3570 struct f2fs_filename *fname) 3571{ 3572 return 0; 3573} 3574 3575static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 3576{ 3577} 3578#endif /* CONFIG_UNICODE */ 3579 3580int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3581 int lookup, struct f2fs_filename *fname); 3582int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3583 struct f2fs_filename *fname); 3584void f2fs_free_filename(struct f2fs_filename *fname); 3585struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3586 const struct f2fs_filename *fname, int *max_slots); 3587int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3588 unsigned int start_pos, struct fscrypt_str *fstr); 3589void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3590 struct f2fs_dentry_ptr *d); 3591struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3592 const struct f2fs_filename *fname, struct page *dpage); 3593void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3594 unsigned int current_depth); 3595int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3596void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3597struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3598 const struct f2fs_filename *fname, 3599 struct page **res_page); 3600struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3601 const struct qstr *child, struct page **res_page); 3602struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3603ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3604 struct page **page); 3605void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3606 struct page *page, struct inode *inode); 3607bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3608 const struct f2fs_filename *fname); 3609void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3610 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3611 unsigned int bit_pos); 3612int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3613 struct inode *inode, nid_t ino, umode_t mode); 3614int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3615 struct inode *inode, nid_t ino, umode_t mode); 3616int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3617 struct inode *inode, nid_t ino, umode_t mode); 3618void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3619 struct inode *dir, struct inode *inode); 3620int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3621 struct f2fs_filename *fname); 3622bool f2fs_empty_dir(struct inode *dir); 3623 3624static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3625{ 3626 if (fscrypt_is_nokey_name(dentry)) 3627 return -ENOKEY; 3628 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3629 inode, inode->i_ino, inode->i_mode); 3630} 3631 3632/* 3633 * super.c 3634 */ 3635int f2fs_inode_dirtied(struct inode *inode, bool sync); 3636void f2fs_inode_synced(struct inode *inode); 3637int f2fs_dquot_initialize(struct inode *inode); 3638int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3639int f2fs_quota_sync(struct super_block *sb, int type); 3640loff_t max_file_blocks(struct inode *inode); 3641void f2fs_quota_off_umount(struct super_block *sb); 3642void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3643void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3644void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3645void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3646int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3647int f2fs_sync_fs(struct super_block *sb, int sync); 3648int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3649 3650/* 3651 * hash.c 3652 */ 3653void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3654 3655/* 3656 * node.c 3657 */ 3658struct node_info; 3659 3660int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3661bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3662bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3663void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3664void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3665void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3666int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3667bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3668bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3669int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3670 struct node_info *ni, bool checkpoint_context); 3671pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3672int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3673int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3674int f2fs_truncate_xattr_node(struct inode *inode); 3675int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3676 unsigned int seq_id); 3677bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3678int f2fs_remove_inode_page(struct inode *inode); 3679struct page *f2fs_new_inode_page(struct inode *inode); 3680struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3681void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3682struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3683struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3684int f2fs_move_node_page(struct page *node_page, int gc_type); 3685void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3686int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3687 struct writeback_control *wbc, bool atomic, 3688 unsigned int *seq_id); 3689int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3690 struct writeback_control *wbc, 3691 bool do_balance, enum iostat_type io_type); 3692int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3693bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3694void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3695void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3696int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3697int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3698int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3699int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3700int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3701 unsigned int segno, struct f2fs_summary_block *sum); 3702void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3703int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3704int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3705void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3706int __init f2fs_create_node_manager_caches(void); 3707void f2fs_destroy_node_manager_caches(void); 3708 3709/* 3710 * segment.c 3711 */ 3712bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3713int f2fs_commit_atomic_write(struct inode *inode); 3714void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3715void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3716void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3717int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3718int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3719int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3720void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3721void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3722bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3723int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3724void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3725void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3726bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3727void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3728 struct cp_control *cpc); 3729void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3730block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3731int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3732void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3733int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3734bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3735int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3736int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 3737void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3738void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3739int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3740 unsigned int start, unsigned int end); 3741int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3742int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3743int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3744int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3745bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3746 struct cp_control *cpc); 3747struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3748void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3749 block_t blk_addr); 3750void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio, 3751 enum iostat_type io_type); 3752void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3753void f2fs_outplace_write_data(struct dnode_of_data *dn, 3754 struct f2fs_io_info *fio); 3755int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3756void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3757 block_t old_blkaddr, block_t new_blkaddr, 3758 bool recover_curseg, bool recover_newaddr, 3759 bool from_gc); 3760void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3761 block_t old_addr, block_t new_addr, 3762 unsigned char version, bool recover_curseg, 3763 bool recover_newaddr); 3764enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 3765 enum log_type seg_type); 3766int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3767 block_t old_blkaddr, block_t *new_blkaddr, 3768 struct f2fs_summary *sum, int type, 3769 struct f2fs_io_info *fio); 3770void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3771 block_t blkaddr, unsigned int blkcnt); 3772void f2fs_wait_on_page_writeback(struct page *page, 3773 enum page_type type, bool ordered, bool locked); 3774void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3775void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3776 block_t len); 3777void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3778void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3779int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3780 unsigned int val, int alloc); 3781void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3782int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 3783int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3784void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3785int __init f2fs_create_segment_manager_caches(void); 3786void f2fs_destroy_segment_manager_caches(void); 3787int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 3788enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3789 enum page_type type, enum temp_type temp); 3790unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 3791unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3792 unsigned int segno); 3793unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 3794 unsigned int segno); 3795 3796#define DEF_FRAGMENT_SIZE 4 3797#define MIN_FRAGMENT_SIZE 1 3798#define MAX_FRAGMENT_SIZE 512 3799 3800static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3801{ 3802 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3803 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3804} 3805 3806/* 3807 * checkpoint.c 3808 */ 3809void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3810 unsigned char reason); 3811void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3812struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3813struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3814struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3815struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3816bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3817 block_t blkaddr, int type); 3818bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 3819 block_t blkaddr, int type); 3820int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3821 int type, bool sync); 3822void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3823 unsigned int ra_blocks); 3824long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3825 long nr_to_write, enum iostat_type io_type); 3826void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3827void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3828void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3829bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3830void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3831 unsigned int devidx, int type); 3832bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3833 unsigned int devidx, int type); 3834int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3835void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3836void f2fs_add_orphan_inode(struct inode *inode); 3837void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3838int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3839int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3840void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3841void f2fs_remove_dirty_inode(struct inode *inode); 3842int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3843 bool from_cp); 3844void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3845u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3846int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3847void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3848int __init f2fs_create_checkpoint_caches(void); 3849void f2fs_destroy_checkpoint_caches(void); 3850int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3851int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3852void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3853void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3854 3855/* 3856 * data.c 3857 */ 3858int __init f2fs_init_bioset(void); 3859void f2fs_destroy_bioset(void); 3860bool f2fs_is_cp_guaranteed(struct page *page); 3861int f2fs_init_bio_entry_cache(void); 3862void f2fs_destroy_bio_entry_cache(void); 3863void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3864 enum page_type type); 3865int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3866void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3867void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3868 struct inode *inode, struct page *page, 3869 nid_t ino, enum page_type type); 3870void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3871 struct bio **bio, struct page *page); 3872void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3873int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3874int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3875void f2fs_submit_page_write(struct f2fs_io_info *fio); 3876struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3877 block_t blk_addr, sector_t *sector); 3878int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3879void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3880void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3881int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3882int f2fs_reserve_new_block(struct dnode_of_data *dn); 3883int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3884int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3885struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3886 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3887struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3888 pgoff_t *next_pgofs); 3889struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3890 bool for_write); 3891struct page *f2fs_get_new_data_page(struct inode *inode, 3892 struct page *ipage, pgoff_t index, bool new_i_size); 3893int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3894int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3895int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3896 u64 start, u64 len); 3897int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3898bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3899bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3900int f2fs_write_single_data_page(struct folio *folio, int *submitted, 3901 struct bio **bio, sector_t *last_block, 3902 struct writeback_control *wbc, 3903 enum iostat_type io_type, 3904 int compr_blocks, bool allow_balance); 3905void f2fs_write_failed(struct inode *inode, loff_t to); 3906void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3907bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3908bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3909void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 3910int f2fs_init_post_read_processing(void); 3911void f2fs_destroy_post_read_processing(void); 3912int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3913void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3914extern const struct iomap_ops f2fs_iomap_ops; 3915 3916/* 3917 * gc.c 3918 */ 3919int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3920void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3921block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3922int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3923void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3924int f2fs_gc_range(struct f2fs_sb_info *sbi, 3925 unsigned int start_seg, unsigned int end_seg, 3926 bool dry_run, unsigned int dry_run_sections); 3927int f2fs_resize_fs(struct file *filp, __u64 block_count); 3928int __init f2fs_create_garbage_collection_cache(void); 3929void f2fs_destroy_garbage_collection_cache(void); 3930/* victim selection function for cleaning and SSR */ 3931int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3932 int gc_type, int type, char alloc_mode, 3933 unsigned long long age, bool one_time); 3934 3935/* 3936 * recovery.c 3937 */ 3938int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3939bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3940int __init f2fs_create_recovery_cache(void); 3941void f2fs_destroy_recovery_cache(void); 3942 3943/* 3944 * debug.c 3945 */ 3946#ifdef CONFIG_F2FS_STAT_FS 3947enum { 3948 DEVSTAT_INUSE, 3949 DEVSTAT_DIRTY, 3950 DEVSTAT_FULL, 3951 DEVSTAT_FREE, 3952 DEVSTAT_PREFREE, 3953 DEVSTAT_MAX, 3954}; 3955 3956struct f2fs_dev_stats { 3957 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 3958}; 3959 3960struct f2fs_stat_info { 3961 struct list_head stat_list; 3962 struct f2fs_sb_info *sbi; 3963 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3964 int main_area_segs, main_area_sections, main_area_zones; 3965 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3966 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3967 unsigned long long total_ext[NR_EXTENT_CACHES]; 3968 unsigned long long hit_total[NR_EXTENT_CACHES]; 3969 int ext_tree[NR_EXTENT_CACHES]; 3970 int zombie_tree[NR_EXTENT_CACHES]; 3971 int ext_node[NR_EXTENT_CACHES]; 3972 /* to count memory footprint */ 3973 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3974 /* for read extent cache */ 3975 unsigned long long hit_largest; 3976 /* for block age extent cache */ 3977 unsigned long long allocated_data_blocks; 3978 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3979 int ndirty_data, ndirty_qdata; 3980 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3981 int nats, dirty_nats, sits, dirty_sits; 3982 int free_nids, avail_nids, alloc_nids; 3983 int total_count, utilization; 3984 int nr_wb_cp_data, nr_wb_data; 3985 int nr_rd_data, nr_rd_node, nr_rd_meta; 3986 int nr_dio_read, nr_dio_write; 3987 unsigned int io_skip_bggc, other_skip_bggc; 3988 int nr_flushing, nr_flushed, flush_list_empty; 3989 int nr_discarding, nr_discarded; 3990 int nr_discard_cmd; 3991 unsigned int undiscard_blks; 3992 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3993 unsigned int cur_ckpt_time, peak_ckpt_time; 3994 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3995 int compr_inode, swapfile_inode; 3996 unsigned long long compr_blocks; 3997 int aw_cnt, max_aw_cnt; 3998 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3999 unsigned int bimodal, avg_vblocks; 4000 int util_free, util_valid, util_invalid; 4001 int rsvd_segs, overp_segs; 4002 int dirty_count, node_pages, meta_pages, compress_pages; 4003 int compress_page_hit; 4004 int prefree_count, free_segs, free_secs; 4005 int cp_call_count[MAX_CALL_TYPE], cp_count; 4006 int gc_call_count[MAX_CALL_TYPE]; 4007 int gc_segs[2][2]; 4008 int gc_secs[2][2]; 4009 int tot_blks, data_blks, node_blks; 4010 int bg_data_blks, bg_node_blks; 4011 int curseg[NR_CURSEG_TYPE]; 4012 int cursec[NR_CURSEG_TYPE]; 4013 int curzone[NR_CURSEG_TYPE]; 4014 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4015 unsigned int full_seg[NR_CURSEG_TYPE]; 4016 unsigned int valid_blks[NR_CURSEG_TYPE]; 4017 4018 unsigned int meta_count[META_MAX]; 4019 unsigned int segment_count[2]; 4020 unsigned int block_count[2]; 4021 unsigned int inplace_count; 4022 unsigned long long base_mem, cache_mem, page_mem; 4023 struct f2fs_dev_stats *dev_stats; 4024}; 4025 4026static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4027{ 4028 return (struct f2fs_stat_info *)sbi->stat_info; 4029} 4030 4031#define stat_inc_cp_call_count(sbi, foreground) \ 4032 atomic_inc(&sbi->cp_call_count[(foreground)]) 4033#define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4034#define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4035#define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4036#define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4037#define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4038#define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4039#define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4040#define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4041#define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4042#define stat_inc_inline_xattr(inode) \ 4043 do { \ 4044 if (f2fs_has_inline_xattr(inode)) \ 4045 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4046 } while (0) 4047#define stat_dec_inline_xattr(inode) \ 4048 do { \ 4049 if (f2fs_has_inline_xattr(inode)) \ 4050 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4051 } while (0) 4052#define stat_inc_inline_inode(inode) \ 4053 do { \ 4054 if (f2fs_has_inline_data(inode)) \ 4055 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4056 } while (0) 4057#define stat_dec_inline_inode(inode) \ 4058 do { \ 4059 if (f2fs_has_inline_data(inode)) \ 4060 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4061 } while (0) 4062#define stat_inc_inline_dir(inode) \ 4063 do { \ 4064 if (f2fs_has_inline_dentry(inode)) \ 4065 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4066 } while (0) 4067#define stat_dec_inline_dir(inode) \ 4068 do { \ 4069 if (f2fs_has_inline_dentry(inode)) \ 4070 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4071 } while (0) 4072#define stat_inc_compr_inode(inode) \ 4073 do { \ 4074 if (f2fs_compressed_file(inode)) \ 4075 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4076 } while (0) 4077#define stat_dec_compr_inode(inode) \ 4078 do { \ 4079 if (f2fs_compressed_file(inode)) \ 4080 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4081 } while (0) 4082#define stat_add_compr_blocks(inode, blocks) \ 4083 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4084#define stat_sub_compr_blocks(inode, blocks) \ 4085 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4086#define stat_inc_swapfile_inode(inode) \ 4087 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4088#define stat_dec_swapfile_inode(inode) \ 4089 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4090#define stat_inc_atomic_inode(inode) \ 4091 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4092#define stat_dec_atomic_inode(inode) \ 4093 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4094#define stat_inc_meta_count(sbi, blkaddr) \ 4095 do { \ 4096 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4097 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4098 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4099 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4100 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4101 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4102 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4103 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4104 } while (0) 4105#define stat_inc_seg_type(sbi, curseg) \ 4106 ((sbi)->segment_count[(curseg)->alloc_type]++) 4107#define stat_inc_block_count(sbi, curseg) \ 4108 ((sbi)->block_count[(curseg)->alloc_type]++) 4109#define stat_inc_inplace_blocks(sbi) \ 4110 (atomic_inc(&(sbi)->inplace_count)) 4111#define stat_update_max_atomic_write(inode) \ 4112 do { \ 4113 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4114 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4115 if (cur > max) \ 4116 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4117 } while (0) 4118#define stat_inc_gc_call_count(sbi, foreground) \ 4119 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4120#define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4121 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4122#define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4123 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4124 4125#define stat_inc_tot_blk_count(si, blks) \ 4126 ((si)->tot_blks += (blks)) 4127 4128#define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4129 do { \ 4130 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4131 stat_inc_tot_blk_count(si, blks); \ 4132 si->data_blks += (blks); \ 4133 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4134 } while (0) 4135 4136#define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4137 do { \ 4138 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4139 stat_inc_tot_blk_count(si, blks); \ 4140 si->node_blks += (blks); \ 4141 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4142 } while (0) 4143 4144int f2fs_build_stats(struct f2fs_sb_info *sbi); 4145void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4146void __init f2fs_create_root_stats(void); 4147void f2fs_destroy_root_stats(void); 4148void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4149#else 4150#define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4151#define stat_inc_cp_count(sbi) do { } while (0) 4152#define stat_io_skip_bggc_count(sbi) do { } while (0) 4153#define stat_other_skip_bggc_count(sbi) do { } while (0) 4154#define stat_inc_dirty_inode(sbi, type) do { } while (0) 4155#define stat_dec_dirty_inode(sbi, type) do { } while (0) 4156#define stat_inc_total_hit(sbi, type) do { } while (0) 4157#define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4158#define stat_inc_largest_node_hit(sbi) do { } while (0) 4159#define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4160#define stat_inc_inline_xattr(inode) do { } while (0) 4161#define stat_dec_inline_xattr(inode) do { } while (0) 4162#define stat_inc_inline_inode(inode) do { } while (0) 4163#define stat_dec_inline_inode(inode) do { } while (0) 4164#define stat_inc_inline_dir(inode) do { } while (0) 4165#define stat_dec_inline_dir(inode) do { } while (0) 4166#define stat_inc_compr_inode(inode) do { } while (0) 4167#define stat_dec_compr_inode(inode) do { } while (0) 4168#define stat_add_compr_blocks(inode, blocks) do { } while (0) 4169#define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4170#define stat_inc_swapfile_inode(inode) do { } while (0) 4171#define stat_dec_swapfile_inode(inode) do { } while (0) 4172#define stat_inc_atomic_inode(inode) do { } while (0) 4173#define stat_dec_atomic_inode(inode) do { } while (0) 4174#define stat_update_max_atomic_write(inode) do { } while (0) 4175#define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4176#define stat_inc_seg_type(sbi, curseg) do { } while (0) 4177#define stat_inc_block_count(sbi, curseg) do { } while (0) 4178#define stat_inc_inplace_blocks(sbi) do { } while (0) 4179#define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4180#define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4181#define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4182#define stat_inc_tot_blk_count(si, blks) do { } while (0) 4183#define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4184#define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4185 4186static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4187static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4188static inline void __init f2fs_create_root_stats(void) { } 4189static inline void f2fs_destroy_root_stats(void) { } 4190static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4191#endif 4192 4193extern const struct file_operations f2fs_dir_operations; 4194extern const struct file_operations f2fs_file_operations; 4195extern const struct inode_operations f2fs_file_inode_operations; 4196extern const struct address_space_operations f2fs_dblock_aops; 4197extern const struct address_space_operations f2fs_node_aops; 4198extern const struct address_space_operations f2fs_meta_aops; 4199extern const struct inode_operations f2fs_dir_inode_operations; 4200extern const struct inode_operations f2fs_symlink_inode_operations; 4201extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4202extern const struct inode_operations f2fs_special_inode_operations; 4203extern struct kmem_cache *f2fs_inode_entry_slab; 4204 4205/* 4206 * inline.c 4207 */ 4208bool f2fs_may_inline_data(struct inode *inode); 4209bool f2fs_sanity_check_inline_data(struct inode *inode, struct page *ipage); 4210bool f2fs_may_inline_dentry(struct inode *inode); 4211void f2fs_do_read_inline_data(struct folio *folio, struct page *ipage); 4212void f2fs_truncate_inline_inode(struct inode *inode, 4213 struct page *ipage, u64 from); 4214int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4215int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4216int f2fs_convert_inline_inode(struct inode *inode); 4217int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4218int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4219int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4220struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4221 const struct f2fs_filename *fname, 4222 struct page **res_page); 4223int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4224 struct page *ipage); 4225int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4226 struct inode *inode, nid_t ino, umode_t mode); 4227void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4228 struct page *page, struct inode *dir, 4229 struct inode *inode); 4230bool f2fs_empty_inline_dir(struct inode *dir); 4231int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4232 struct fscrypt_str *fstr); 4233int f2fs_inline_data_fiemap(struct inode *inode, 4234 struct fiemap_extent_info *fieinfo, 4235 __u64 start, __u64 len); 4236 4237/* 4238 * shrinker.c 4239 */ 4240unsigned long f2fs_shrink_count(struct shrinker *shrink, 4241 struct shrink_control *sc); 4242unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4243 struct shrink_control *sc); 4244void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4245void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4246 4247/* 4248 * extent_cache.c 4249 */ 4250bool sanity_check_extent_cache(struct inode *inode, struct page *ipage); 4251void f2fs_init_extent_tree(struct inode *inode); 4252void f2fs_drop_extent_tree(struct inode *inode); 4253void f2fs_destroy_extent_node(struct inode *inode); 4254void f2fs_destroy_extent_tree(struct inode *inode); 4255void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4256int __init f2fs_create_extent_cache(void); 4257void f2fs_destroy_extent_cache(void); 4258 4259/* read extent cache ops */ 4260void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4261bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4262 struct extent_info *ei); 4263bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4264 block_t *blkaddr); 4265void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4266void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4267 pgoff_t fofs, block_t blkaddr, unsigned int len); 4268unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4269 int nr_shrink); 4270 4271/* block age extent cache ops */ 4272void f2fs_init_age_extent_tree(struct inode *inode); 4273bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4274 struct extent_info *ei); 4275void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4276void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4277 pgoff_t fofs, unsigned int len); 4278unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4279 int nr_shrink); 4280 4281/* 4282 * sysfs.c 4283 */ 4284#define MIN_RA_MUL 2 4285#define MAX_RA_MUL 256 4286 4287int __init f2fs_init_sysfs(void); 4288void f2fs_exit_sysfs(void); 4289int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4290void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4291 4292/* verity.c */ 4293extern const struct fsverity_operations f2fs_verityops; 4294 4295/* 4296 * crypto support 4297 */ 4298static inline bool f2fs_encrypted_file(struct inode *inode) 4299{ 4300 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4301} 4302 4303static inline void f2fs_set_encrypted_inode(struct inode *inode) 4304{ 4305#ifdef CONFIG_FS_ENCRYPTION 4306 file_set_encrypt(inode); 4307 f2fs_set_inode_flags(inode); 4308#endif 4309} 4310 4311/* 4312 * Returns true if the reads of the inode's data need to undergo some 4313 * postprocessing step, like decryption or authenticity verification. 4314 */ 4315static inline bool f2fs_post_read_required(struct inode *inode) 4316{ 4317 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4318 f2fs_compressed_file(inode); 4319} 4320 4321static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4322{ 4323 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4324} 4325 4326static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4327{ 4328 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4329} 4330 4331/* 4332 * compress.c 4333 */ 4334#ifdef CONFIG_F2FS_FS_COMPRESSION 4335enum cluster_check_type { 4336 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4337 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4338 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4339}; 4340bool f2fs_is_compressed_page(struct page *page); 4341struct page *f2fs_compress_control_page(struct page *page); 4342int f2fs_prepare_compress_overwrite(struct inode *inode, 4343 struct page **pagep, pgoff_t index, void **fsdata); 4344bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4345 pgoff_t index, unsigned copied); 4346int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4347void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4348bool f2fs_is_compress_backend_ready(struct inode *inode); 4349bool f2fs_is_compress_level_valid(int alg, int lvl); 4350int __init f2fs_init_compress_mempool(void); 4351void f2fs_destroy_compress_mempool(void); 4352void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4353void f2fs_end_read_compressed_page(struct page *page, bool failed, 4354 block_t blkaddr, bool in_task); 4355bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4356bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4357bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4358 int index, int nr_pages, bool uptodate); 4359bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4360void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4361int f2fs_write_multi_pages(struct compress_ctx *cc, 4362 int *submitted, 4363 struct writeback_control *wbc, 4364 enum iostat_type io_type); 4365int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4366bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4367void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4368 pgoff_t fofs, block_t blkaddr, 4369 unsigned int llen, unsigned int c_len); 4370int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4371 unsigned nr_pages, sector_t *last_block_in_bio, 4372 struct readahead_control *rac, bool for_write); 4373struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4374void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4375 bool in_task); 4376void f2fs_put_page_dic(struct page *page, bool in_task); 4377unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4378 unsigned int ofs_in_node); 4379int f2fs_init_compress_ctx(struct compress_ctx *cc); 4380void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4381void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4382int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4383void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4384int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4385void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4386int __init f2fs_init_compress_cache(void); 4387void f2fs_destroy_compress_cache(void); 4388struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4389void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4390void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4391 nid_t ino, block_t blkaddr); 4392bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4393 block_t blkaddr); 4394void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4395#define inc_compr_inode_stat(inode) \ 4396 do { \ 4397 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4398 sbi->compr_new_inode++; \ 4399 } while (0) 4400#define add_compr_block_stat(inode, blocks) \ 4401 do { \ 4402 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4403 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4404 sbi->compr_written_block += blocks; \ 4405 sbi->compr_saved_block += diff; \ 4406 } while (0) 4407#else 4408static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4409static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4410{ 4411 if (!f2fs_compressed_file(inode)) 4412 return true; 4413 /* not support compression */ 4414 return false; 4415} 4416static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4417static inline struct page *f2fs_compress_control_page(struct page *page) 4418{ 4419 WARN_ON_ONCE(1); 4420 return ERR_PTR(-EINVAL); 4421} 4422static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4423static inline void f2fs_destroy_compress_mempool(void) { } 4424static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4425 bool in_task) { } 4426static inline void f2fs_end_read_compressed_page(struct page *page, 4427 bool failed, block_t blkaddr, bool in_task) 4428{ 4429 WARN_ON_ONCE(1); 4430} 4431static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4432{ 4433 WARN_ON_ONCE(1); 4434} 4435static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4436 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4437static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4438static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4439static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4440static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4441static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4442static inline int __init f2fs_init_compress_cache(void) { return 0; } 4443static inline void f2fs_destroy_compress_cache(void) { } 4444static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4445 block_t blkaddr) { } 4446static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4447 struct page *page, nid_t ino, block_t blkaddr) { } 4448static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4449 struct page *page, block_t blkaddr) { return false; } 4450static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4451 nid_t ino) { } 4452#define inc_compr_inode_stat(inode) do { } while (0) 4453static inline int f2fs_is_compressed_cluster( 4454 struct inode *inode, 4455 pgoff_t index) { return 0; } 4456static inline bool f2fs_is_sparse_cluster( 4457 struct inode *inode, 4458 pgoff_t index) { return true; } 4459static inline void f2fs_update_read_extent_tree_range_compressed( 4460 struct inode *inode, 4461 pgoff_t fofs, block_t blkaddr, 4462 unsigned int llen, unsigned int c_len) { } 4463#endif 4464 4465static inline int set_compress_context(struct inode *inode) 4466{ 4467#ifdef CONFIG_F2FS_FS_COMPRESSION 4468 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4469 struct f2fs_inode_info *fi = F2FS_I(inode); 4470 4471 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4472 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4473 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4474 BIT(COMPRESS_CHKSUM) : 0; 4475 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4476 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4477 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4478 F2FS_OPTION(sbi).compress_level) 4479 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4480 fi->i_flags |= F2FS_COMPR_FL; 4481 set_inode_flag(inode, FI_COMPRESSED_FILE); 4482 stat_inc_compr_inode(inode); 4483 inc_compr_inode_stat(inode); 4484 f2fs_mark_inode_dirty_sync(inode, true); 4485 return 0; 4486#else 4487 return -EOPNOTSUPP; 4488#endif 4489} 4490 4491static inline bool f2fs_disable_compressed_file(struct inode *inode) 4492{ 4493 struct f2fs_inode_info *fi = F2FS_I(inode); 4494 4495 f2fs_down_write(&fi->i_sem); 4496 4497 if (!f2fs_compressed_file(inode)) { 4498 f2fs_up_write(&fi->i_sem); 4499 return true; 4500 } 4501 if (f2fs_is_mmap_file(inode) || 4502 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4503 f2fs_up_write(&fi->i_sem); 4504 return false; 4505 } 4506 4507 fi->i_flags &= ~F2FS_COMPR_FL; 4508 stat_dec_compr_inode(inode); 4509 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4510 f2fs_mark_inode_dirty_sync(inode, true); 4511 4512 f2fs_up_write(&fi->i_sem); 4513 return true; 4514} 4515 4516#define F2FS_FEATURE_FUNCS(name, flagname) \ 4517static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4518{ \ 4519 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4520} 4521 4522F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4523F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4524F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4525F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4526F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4527F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4528F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4529F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4530F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4531F2FS_FEATURE_FUNCS(verity, VERITY); 4532F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4533F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4534F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4535F2FS_FEATURE_FUNCS(readonly, RO); 4536F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 4537 4538#ifdef CONFIG_BLK_DEV_ZONED 4539static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4540 block_t blkaddr) 4541{ 4542 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4543 4544 return test_bit(zno, FDEV(devi).blkz_seq); 4545} 4546#endif 4547 4548static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4549 struct block_device *bdev) 4550{ 4551 int i; 4552 4553 if (!f2fs_is_multi_device(sbi)) 4554 return 0; 4555 4556 for (i = 0; i < sbi->s_ndevs; i++) 4557 if (FDEV(i).bdev == bdev) 4558 return i; 4559 4560 WARN_ON(1); 4561 return -1; 4562} 4563 4564static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4565{ 4566 return f2fs_sb_has_blkzoned(sbi); 4567} 4568 4569static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4570{ 4571 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4572} 4573 4574static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4575{ 4576 int i; 4577 4578 if (!f2fs_is_multi_device(sbi)) 4579 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4580 4581 for (i = 0; i < sbi->s_ndevs; i++) 4582 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4583 return true; 4584 return false; 4585} 4586 4587static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4588{ 4589 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4590 f2fs_hw_should_discard(sbi); 4591} 4592 4593static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4594{ 4595 int i; 4596 4597 if (!f2fs_is_multi_device(sbi)) 4598 return bdev_read_only(sbi->sb->s_bdev); 4599 4600 for (i = 0; i < sbi->s_ndevs; i++) 4601 if (bdev_read_only(FDEV(i).bdev)) 4602 return true; 4603 return false; 4604} 4605 4606static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4607{ 4608 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4609} 4610 4611static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4612{ 4613 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4614} 4615 4616static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4617 block_t blkaddr) 4618{ 4619 if (f2fs_sb_has_blkzoned(sbi)) { 4620 int devi = f2fs_target_device_index(sbi, blkaddr); 4621 4622 return !bdev_is_zoned(FDEV(devi).bdev); 4623 } 4624 return true; 4625} 4626 4627static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4628{ 4629 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4630} 4631 4632static inline bool f2fs_may_compress(struct inode *inode) 4633{ 4634 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4635 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4636 f2fs_is_mmap_file(inode)) 4637 return false; 4638 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4639} 4640 4641static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4642 u64 blocks, bool add) 4643{ 4644 struct f2fs_inode_info *fi = F2FS_I(inode); 4645 int diff = fi->i_cluster_size - blocks; 4646 4647 /* don't update i_compr_blocks if saved blocks were released */ 4648 if (!add && !atomic_read(&fi->i_compr_blocks)) 4649 return; 4650 4651 if (add) { 4652 atomic_add(diff, &fi->i_compr_blocks); 4653 stat_add_compr_blocks(inode, diff); 4654 } else { 4655 atomic_sub(diff, &fi->i_compr_blocks); 4656 stat_sub_compr_blocks(inode, diff); 4657 } 4658 f2fs_mark_inode_dirty_sync(inode, true); 4659} 4660 4661static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4662 int flag) 4663{ 4664 if (!f2fs_is_multi_device(sbi)) 4665 return false; 4666 if (flag != F2FS_GET_BLOCK_DIO) 4667 return false; 4668 return sbi->aligned_blksize; 4669} 4670 4671static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4672{ 4673 return fsverity_active(inode) && 4674 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4675} 4676 4677#ifdef CONFIG_F2FS_FAULT_INJECTION 4678extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4679 unsigned long type); 4680#else 4681static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4682 unsigned long rate, unsigned long type) 4683{ 4684 return 0; 4685} 4686#endif 4687 4688static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4689{ 4690#ifdef CONFIG_QUOTA 4691 if (f2fs_sb_has_quota_ino(sbi)) 4692 return true; 4693 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4694 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4695 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4696 return true; 4697#endif 4698 return false; 4699} 4700 4701static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4702{ 4703 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4704} 4705 4706static inline void f2fs_io_schedule_timeout(long timeout) 4707{ 4708 set_current_state(TASK_UNINTERRUPTIBLE); 4709 io_schedule_timeout(timeout); 4710} 4711 4712static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 4713 struct folio *folio, enum page_type type) 4714{ 4715 pgoff_t ofs = folio->index; 4716 4717 if (unlikely(f2fs_cp_error(sbi))) 4718 return; 4719 4720 if (ofs == sbi->page_eio_ofs[type]) { 4721 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4722 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4723 } else { 4724 sbi->page_eio_ofs[type] = ofs; 4725 sbi->page_eio_cnt[type] = 0; 4726 } 4727} 4728 4729static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4730{ 4731 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4732} 4733 4734static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4735 block_t blkaddr, unsigned int cnt) 4736{ 4737 bool need_submit = false; 4738 int i = 0; 4739 4740 do { 4741 struct page *page; 4742 4743 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4744 if (page) { 4745 if (folio_test_writeback(page_folio(page))) 4746 need_submit = true; 4747 f2fs_put_page(page, 0); 4748 } 4749 } while (++i < cnt && !need_submit); 4750 4751 if (need_submit) 4752 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4753 NULL, 0, DATA); 4754 4755 truncate_inode_pages_range(META_MAPPING(sbi), 4756 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4757 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4758} 4759 4760static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4761 block_t blkaddr) 4762{ 4763 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4764 f2fs_invalidate_compress_page(sbi, blkaddr); 4765} 4766 4767#define EFSBADCRC EBADMSG /* Bad CRC detected */ 4768#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4769 4770#endif /* _LINUX_F2FS_H */