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 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_BLKADDR_VALIDITY, 63 FAULT_BLKADDR_CONSISTENCE, 64 FAULT_NO_SEGMENT, 65 FAULT_INCONSISTENT_FOOTER, 66 FAULT_TIMEOUT, 67 FAULT_VMALLOC, 68 FAULT_MAX, 69 }; 70 71 /* indicate which option to update */ 72 enum fault_option { 73 FAULT_RATE = 1, /* only update fault rate */ 74 FAULT_TYPE = 2, /* only update fault type */ 75 FAULT_ALL = 4, /* reset all fault injection options/stats */ 76 }; 77 78 #ifdef CONFIG_F2FS_FAULT_INJECTION 79 struct f2fs_fault_info { 80 atomic_t inject_ops; 81 int inject_rate; 82 unsigned int inject_type; 83 /* Used to account total count of injection for each type */ 84 unsigned int inject_count[FAULT_MAX]; 85 }; 86 87 extern const char *f2fs_fault_name[FAULT_MAX]; 88 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 89 90 /* maximum retry count for injected failure */ 91 #define DEFAULT_FAILURE_RETRY_COUNT 8 92 #else 93 #define DEFAULT_FAILURE_RETRY_COUNT 1 94 #endif 95 96 /* 97 * For mount options 98 */ 99 enum f2fs_mount_opt { 100 F2FS_MOUNT_DISABLE_ROLL_FORWARD, 101 F2FS_MOUNT_DISCARD, 102 F2FS_MOUNT_NOHEAP, 103 F2FS_MOUNT_XATTR_USER, 104 F2FS_MOUNT_POSIX_ACL, 105 F2FS_MOUNT_DISABLE_EXT_IDENTIFY, 106 F2FS_MOUNT_INLINE_XATTR, 107 F2FS_MOUNT_INLINE_DATA, 108 F2FS_MOUNT_INLINE_DENTRY, 109 F2FS_MOUNT_FLUSH_MERGE, 110 F2FS_MOUNT_NOBARRIER, 111 F2FS_MOUNT_FASTBOOT, 112 F2FS_MOUNT_READ_EXTENT_CACHE, 113 F2FS_MOUNT_DATA_FLUSH, 114 F2FS_MOUNT_FAULT_INJECTION, 115 F2FS_MOUNT_USRQUOTA, 116 F2FS_MOUNT_GRPQUOTA, 117 F2FS_MOUNT_PRJQUOTA, 118 F2FS_MOUNT_QUOTA, 119 F2FS_MOUNT_INLINE_XATTR_SIZE, 120 F2FS_MOUNT_RESERVE_ROOT, 121 F2FS_MOUNT_DISABLE_CHECKPOINT, 122 F2FS_MOUNT_NORECOVERY, 123 F2FS_MOUNT_ATGC, 124 F2FS_MOUNT_MERGE_CHECKPOINT, 125 F2FS_MOUNT_GC_MERGE, 126 F2FS_MOUNT_COMPRESS_CACHE, 127 F2FS_MOUNT_AGE_EXTENT_CACHE, 128 F2FS_MOUNT_NAT_BITS, 129 F2FS_MOUNT_INLINECRYPT, 130 /* 131 * Some f2fs environments expect to be able to pass the "lazytime" option 132 * string rather than using the MS_LAZYTIME flag, so this must remain. 133 */ 134 F2FS_MOUNT_LAZYTIME, 135 F2FS_MOUNT_RESERVE_NODE, 136 }; 137 138 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 139 #define clear_opt(sbi, option) \ 140 (F2FS_OPTION(sbi).opt &= ~BIT(F2FS_MOUNT_##option)) 141 #define set_opt(sbi, option) \ 142 (F2FS_OPTION(sbi).opt |= BIT(F2FS_MOUNT_##option)) 143 #define test_opt(sbi, option) \ 144 (F2FS_OPTION(sbi).opt & BIT(F2FS_MOUNT_##option)) 145 146 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 147 typecheck(unsigned long long, b) && \ 148 ((long long)((a) - (b)) > 0)) 149 150 typedef u32 block_t; /* 151 * should not change u32, since it is the on-disk block 152 * address format, __le32. 153 */ 154 typedef u32 nid_t; 155 156 #define COMPRESS_EXT_NUM 16 157 158 enum blkzone_allocation_policy { 159 BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */ 160 BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */ 161 BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */ 162 }; 163 164 enum bggc_io_aware_policy { 165 AWARE_ALL_IO, /* skip background GC if there is any kind of pending IO */ 166 AWARE_READ_IO, /* skip background GC if there is pending read IO */ 167 AWARE_NONE, /* don't aware IO for background GC */ 168 }; 169 170 enum device_allocation_policy { 171 ALLOCATE_FORWARD_NOHINT, 172 ALLOCATE_FORWARD_WITHIN_HINT, 173 ALLOCATE_FORWARD_FROM_HINT, 174 }; 175 176 /* 177 * An implementation of an rwsem that is explicitly unfair to readers. This 178 * prevents priority inversion when a low-priority reader acquires the read lock 179 * while sleeping on the write lock but the write lock is needed by 180 * higher-priority clients. 181 */ 182 183 struct f2fs_rwsem { 184 struct rw_semaphore internal_rwsem; 185 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 186 wait_queue_head_t read_waiters; 187 #endif 188 }; 189 190 struct f2fs_mount_info { 191 unsigned long long opt; 192 block_t root_reserved_blocks; /* root reserved blocks */ 193 block_t root_reserved_nodes; /* root reserved nodes */ 194 kuid_t s_resuid; /* reserved blocks for uid */ 195 kgid_t s_resgid; /* reserved blocks for gid */ 196 int active_logs; /* # of active logs */ 197 int inline_xattr_size; /* inline xattr size */ 198 #ifdef CONFIG_F2FS_FAULT_INJECTION 199 struct f2fs_fault_info fault_info; /* For fault injection */ 200 #endif 201 #ifdef CONFIG_QUOTA 202 /* Names of quota files with journalled quota */ 203 char *s_qf_names[MAXQUOTAS]; 204 int s_jquota_fmt; /* Format of quota to use */ 205 #endif 206 /* For which write hints are passed down to block layer */ 207 int alloc_mode; /* segment allocation policy */ 208 int fsync_mode; /* fsync policy */ 209 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 210 int bggc_mode; /* bggc mode: off, on or sync */ 211 int memory_mode; /* memory mode */ 212 int errors; /* errors parameter */ 213 int discard_unit; /* 214 * discard command's offset/size should 215 * be aligned to this unit: block, 216 * segment or section 217 */ 218 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 219 block_t unusable_cap_perc; /* percentage for cap */ 220 block_t unusable_cap; /* Amount of space allowed to be 221 * unusable when disabling checkpoint 222 */ 223 224 /* For compression */ 225 unsigned char compress_algorithm; /* algorithm type */ 226 unsigned char compress_log_size; /* cluster log size */ 227 unsigned char compress_level; /* compress level */ 228 bool compress_chksum; /* compressed data chksum */ 229 unsigned char compress_ext_cnt; /* extension count */ 230 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 231 int compress_mode; /* compression mode */ 232 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 233 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 234 unsigned int lookup_mode; 235 }; 236 237 #define F2FS_FEATURE_ENCRYPT 0x00000001 238 #define F2FS_FEATURE_BLKZONED 0x00000002 239 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 240 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 241 #define F2FS_FEATURE_PRJQUOTA 0x00000010 242 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 243 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 244 #define F2FS_FEATURE_QUOTA_INO 0x00000080 245 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 246 #define F2FS_FEATURE_LOST_FOUND 0x00000200 247 #define F2FS_FEATURE_VERITY 0x00000400 248 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 249 #define F2FS_FEATURE_CASEFOLD 0x00001000 250 #define F2FS_FEATURE_COMPRESSION 0x00002000 251 #define F2FS_FEATURE_RO 0x00004000 252 #define F2FS_FEATURE_DEVICE_ALIAS 0x00008000 253 #define F2FS_FEATURE_PACKED_SSA 0x00010000 254 255 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 256 ((raw_super->feature & cpu_to_le32(mask)) != 0) 257 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 258 259 /* 260 * Default values for user and/or group using reserved blocks 261 */ 262 #define F2FS_DEF_RESUID 0 263 #define F2FS_DEF_RESGID 0 264 265 /* 266 * For checkpoint manager 267 */ 268 enum { 269 NAT_BITMAP, 270 SIT_BITMAP 271 }; 272 273 #define CP_UMOUNT 0x00000001 274 #define CP_FASTBOOT 0x00000002 275 #define CP_SYNC 0x00000004 276 #define CP_RECOVERY 0x00000008 277 #define CP_DISCARD 0x00000010 278 #define CP_TRIMMED 0x00000020 279 #define CP_PAUSE 0x00000040 280 #define CP_RESIZE 0x00000080 281 282 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 283 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 284 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 285 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 286 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 287 #define DEF_CP_INTERVAL 60 /* 60 secs */ 288 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 289 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 290 #define DEF_ENABLE_INTERVAL 5 /* 5 secs */ 291 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 292 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 293 294 enum cp_time { 295 CP_TIME_START, /* begin */ 296 CP_TIME_LOCK, /* after cp_global_sem */ 297 CP_TIME_OP_LOCK, /* after block_operation */ 298 CP_TIME_FLUSH_META, /* after flush sit/nat */ 299 CP_TIME_SYNC_META, /* after sync_meta_pages */ 300 CP_TIME_SYNC_CP_META, /* after sync cp meta pages */ 301 CP_TIME_WAIT_DIRTY_META,/* after wait on dirty meta */ 302 CP_TIME_WAIT_CP_DATA, /* after wait on cp data */ 303 CP_TIME_FLUSH_DEVICE, /* after flush device cache */ 304 CP_TIME_WAIT_LAST_CP, /* after wait on last cp pack */ 305 CP_TIME_END, /* after unblock_operation */ 306 CP_TIME_MAX, 307 }; 308 309 /* time cost stats of checkpoint */ 310 struct cp_stats { 311 ktime_t times[CP_TIME_MAX]; 312 }; 313 314 struct cp_control { 315 int reason; 316 __u64 trim_start; 317 __u64 trim_end; 318 __u64 trim_minlen; 319 struct cp_stats stats; 320 }; 321 322 enum f2fs_cp_phase { 323 CP_PHASE_START_BLOCK_OPS, 324 CP_PHASE_FINISH_BLOCK_OPS, 325 CP_PHASE_FINISH_CHECKPOINT, 326 }; 327 328 /* 329 * indicate meta/data type 330 */ 331 enum { 332 META_CP, 333 META_NAT, 334 META_SIT, 335 META_SSA, 336 META_MAX, 337 META_POR, 338 DATA_GENERIC, /* check range only */ 339 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 340 DATA_GENERIC_ENHANCE_READ, /* 341 * strong check on range and segment 342 * bitmap but no warning due to race 343 * condition of read on truncated area 344 * by extent_cache 345 */ 346 DATA_GENERIC_ENHANCE_UPDATE, /* 347 * strong check on range and segment 348 * bitmap for update case 349 */ 350 META_GENERIC, 351 }; 352 353 /* for the list of ino */ 354 enum { 355 ORPHAN_INO, /* for orphan ino list */ 356 APPEND_INO, /* for append ino list */ 357 UPDATE_INO, /* for update ino list */ 358 TRANS_DIR_INO, /* for transactions dir ino list */ 359 XATTR_DIR_INO, /* for xattr updated dir ino list */ 360 FLUSH_INO, /* for multiple device flushing */ 361 MAX_INO_ENTRY, /* max. list */ 362 }; 363 364 struct ino_entry { 365 struct list_head list; /* list head */ 366 nid_t ino; /* inode number */ 367 unsigned int dirty_device; /* dirty device bitmap */ 368 }; 369 370 /* for the list of inodes to be GCed */ 371 struct inode_entry { 372 struct list_head list; /* list head */ 373 struct inode *inode; /* vfs inode pointer */ 374 }; 375 376 struct fsync_node_entry { 377 struct list_head list; /* list head */ 378 struct folio *folio; /* warm node folio pointer */ 379 unsigned int seq_id; /* sequence id */ 380 }; 381 382 struct ckpt_req { 383 struct completion wait; /* completion for checkpoint done */ 384 struct llist_node llnode; /* llist_node to be linked in wait queue */ 385 int ret; /* return code of checkpoint */ 386 union { 387 ktime_t queue_time; /* request queued time */ 388 ktime_t delta_time; /* time in queue */ 389 }; 390 }; 391 392 struct ckpt_req_control { 393 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 394 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 395 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 396 atomic_t issued_ckpt; /* # of actually issued ckpts */ 397 atomic_t total_ckpt; /* # of total ckpts */ 398 atomic_t queued_ckpt; /* # of queued ckpts */ 399 struct llist_head issue_list; /* list for command issue */ 400 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 401 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 402 unsigned int peak_time; /* peak wait time in msec until now */ 403 }; 404 405 /* a time threshold that checkpoint was blocked for, unit: ms */ 406 #define CP_LONG_LATENCY_THRESHOLD 5000 407 408 /* for the bitmap indicate blocks to be discarded */ 409 struct discard_entry { 410 struct list_head list; /* list head */ 411 block_t start_blkaddr; /* start blockaddr of current segment */ 412 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 413 }; 414 415 /* minimum discard granularity, unit: block count */ 416 #define MIN_DISCARD_GRANULARITY 1 417 /* default discard granularity of inner discard thread, unit: block count */ 418 #define DEFAULT_DISCARD_GRANULARITY 16 419 /* default maximum discard granularity of ordered discard, unit: block count */ 420 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 421 /* default interval of periodical discard submission */ 422 #define DEFAULT_DISCARD_INTERVAL (msecs_to_jiffies(20)) 423 424 /* max discard pend list number */ 425 #define MAX_PLIST_NUM 512 426 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 427 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 428 429 enum { 430 D_PREP, /* initial */ 431 D_PARTIAL, /* partially submitted */ 432 D_SUBMIT, /* all submitted */ 433 D_DONE, /* finished */ 434 }; 435 436 struct discard_info { 437 block_t lstart; /* logical start address */ 438 block_t len; /* length */ 439 block_t start; /* actual start address in dev */ 440 }; 441 442 struct discard_cmd { 443 struct rb_node rb_node; /* rb node located in rb-tree */ 444 struct discard_info di; /* discard info */ 445 struct list_head list; /* command list */ 446 struct completion wait; /* completion */ 447 struct block_device *bdev; /* bdev */ 448 unsigned short ref; /* reference count */ 449 unsigned char state; /* state */ 450 unsigned char queued; /* queued discard */ 451 int error; /* bio error */ 452 spinlock_t lock; /* for state/bio_ref updating */ 453 unsigned short bio_ref; /* bio reference count */ 454 }; 455 456 enum { 457 DPOLICY_BG, 458 DPOLICY_FORCE, 459 DPOLICY_FSTRIM, 460 DPOLICY_UMOUNT, 461 MAX_DPOLICY, 462 }; 463 464 enum { 465 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 466 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 467 DPOLICY_IO_AWARE_MAX, 468 }; 469 470 struct discard_policy { 471 int type; /* type of discard */ 472 unsigned int min_interval; /* used for candidates exist */ 473 unsigned int mid_interval; /* used for device busy */ 474 unsigned int max_interval; /* used for candidates not exist */ 475 unsigned int max_requests; /* # of discards issued per round */ 476 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 477 bool io_aware; /* issue discard in idle time */ 478 bool sync; /* submit discard with REQ_SYNC flag */ 479 bool ordered; /* issue discard by lba order */ 480 bool timeout; /* discard timeout for put_super */ 481 unsigned int granularity; /* discard granularity */ 482 }; 483 484 struct discard_cmd_control { 485 struct task_struct *f2fs_issue_discard; /* discard thread */ 486 struct list_head entry_list; /* 4KB discard entry list */ 487 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 488 struct list_head wait_list; /* store on-flushing entries */ 489 struct list_head fstrim_list; /* in-flight discard from fstrim */ 490 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 491 struct mutex cmd_lock; 492 unsigned int nr_discards; /* # of discards in the list */ 493 unsigned int max_discards; /* max. discards to be issued */ 494 unsigned int max_discard_request; /* max. discard request per round */ 495 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 496 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 497 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 498 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 499 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 500 unsigned int discard_granularity; /* discard granularity */ 501 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 502 unsigned int discard_io_aware; /* io_aware policy */ 503 unsigned int undiscard_blks; /* # of undiscard blocks */ 504 unsigned int next_pos; /* next discard position */ 505 atomic_t issued_discard; /* # of issued discard */ 506 atomic_t queued_discard; /* # of queued discard */ 507 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 508 struct rb_root_cached root; /* root of discard rb-tree */ 509 bool rbtree_check; /* config for consistence check */ 510 bool discard_wake; /* to wake up discard thread */ 511 }; 512 513 /* for the list of fsync inodes, used only during recovery */ 514 struct fsync_inode_entry { 515 struct list_head list; /* list head */ 516 struct inode *inode; /* vfs inode pointer */ 517 block_t blkaddr; /* block address locating the last fsync */ 518 block_t last_dentry; /* block address locating the last dentry */ 519 }; 520 521 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 522 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 523 524 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 525 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 526 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 527 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 528 529 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 530 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 531 532 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 533 { 534 int before = nats_in_cursum(journal); 535 536 journal->n_nats = cpu_to_le16(before + i); 537 return before; 538 } 539 540 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 541 { 542 int before = sits_in_cursum(journal); 543 544 journal->n_sits = cpu_to_le16(before + i); 545 return before; 546 } 547 548 static inline bool __has_cursum_space(struct f2fs_journal *journal, 549 int size, int type) 550 { 551 if (type == NAT_JOURNAL) 552 return size <= MAX_NAT_JENTRIES(journal); 553 return size <= MAX_SIT_JENTRIES(journal); 554 } 555 556 /* for inline stuff */ 557 #define DEF_INLINE_RESERVED_SIZE 1 558 static inline int get_extra_isize(struct inode *inode); 559 static inline int get_inline_xattr_addrs(struct inode *inode); 560 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 561 (CUR_ADDRS_PER_INODE(inode) - \ 562 get_inline_xattr_addrs(inode) - \ 563 DEF_INLINE_RESERVED_SIZE)) 564 565 /* for inline dir */ 566 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 567 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 568 BITS_PER_BYTE + 1)) 569 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 570 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 571 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 572 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 573 NR_INLINE_DENTRY(inode) + \ 574 INLINE_DENTRY_BITMAP_SIZE(inode))) 575 576 /* 577 * For INODE and NODE manager 578 */ 579 /* for directory operations */ 580 581 struct f2fs_filename { 582 /* 583 * The filename the user specified. This is NULL for some 584 * filesystem-internal operations, e.g. converting an inline directory 585 * to a non-inline one, or roll-forward recovering an encrypted dentry. 586 */ 587 const struct qstr *usr_fname; 588 589 /* 590 * The on-disk filename. For encrypted directories, this is encrypted. 591 * This may be NULL for lookups in an encrypted dir without the key. 592 */ 593 struct fscrypt_str disk_name; 594 595 /* The dirhash of this filename */ 596 f2fs_hash_t hash; 597 598 #ifdef CONFIG_FS_ENCRYPTION 599 /* 600 * For lookups in encrypted directories: either the buffer backing 601 * disk_name, or a buffer that holds the decoded no-key name. 602 */ 603 struct fscrypt_str crypto_buf; 604 #endif 605 #if IS_ENABLED(CONFIG_UNICODE) 606 /* 607 * For casefolded directories: the casefolded name, but it's left NULL 608 * if the original name is not valid Unicode, if the original name is 609 * "." or "..", if the directory is both casefolded and encrypted and 610 * its encryption key is unavailable, or if the filesystem is doing an 611 * internal operation where usr_fname is also NULL. In all these cases 612 * we fall back to treating the name as an opaque byte sequence. 613 */ 614 struct qstr cf_name; 615 #endif 616 }; 617 618 struct f2fs_dentry_ptr { 619 struct inode *inode; 620 void *bitmap; 621 struct f2fs_dir_entry *dentry; 622 __u8 (*filename)[F2FS_SLOT_LEN]; 623 int max; 624 int nr_bitmap; 625 }; 626 627 static inline void make_dentry_ptr_block(struct inode *inode, 628 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 629 { 630 d->inode = inode; 631 d->max = NR_DENTRY_IN_BLOCK; 632 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 633 d->bitmap = t->dentry_bitmap; 634 d->dentry = t->dentry; 635 d->filename = t->filename; 636 } 637 638 static inline void make_dentry_ptr_inline(struct inode *inode, 639 struct f2fs_dentry_ptr *d, void *t) 640 { 641 int entry_cnt = NR_INLINE_DENTRY(inode); 642 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 643 int reserved_size = INLINE_RESERVED_SIZE(inode); 644 645 d->inode = inode; 646 d->max = entry_cnt; 647 d->nr_bitmap = bitmap_size; 648 d->bitmap = t; 649 d->dentry = t + bitmap_size + reserved_size; 650 d->filename = t + bitmap_size + reserved_size + 651 SIZE_OF_DIR_ENTRY * entry_cnt; 652 } 653 654 /* 655 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 656 * as its node offset to distinguish from index node blocks. 657 * But some bits are used to mark the node block. 658 */ 659 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 660 >> OFFSET_BIT_SHIFT) 661 enum { 662 ALLOC_NODE, /* allocate a new node page if needed */ 663 LOOKUP_NODE, /* look up a node without readahead */ 664 LOOKUP_NODE_RA, /* 665 * look up a node with readahead called 666 * by get_data_block. 667 */ 668 }; 669 670 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 671 672 /* IO/non-IO congestion wait timeout value, default: 1ms */ 673 #define DEFAULT_SCHEDULE_TIMEOUT (msecs_to_jiffies(1)) 674 675 /* timeout value injected, default: 1000ms */ 676 #define DEFAULT_FAULT_TIMEOUT (msecs_to_jiffies(1000)) 677 678 /* maximum retry quota flush count */ 679 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 680 681 /* maximum retry of EIO'ed page */ 682 #define MAX_RETRY_PAGE_EIO 100 683 684 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 685 686 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 687 688 /* dirty segments threshold for triggering CP */ 689 #define DEFAULT_DIRTY_THRESHOLD 4 690 691 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 692 #define RECOVERY_MIN_RA_BLOCKS 1 693 694 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 695 696 /* for in-memory extent cache entry */ 697 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 698 699 /* number of extent info in extent cache we try to shrink */ 700 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 701 702 /* number of age extent info in extent cache we try to shrink */ 703 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 704 #define LAST_AGE_WEIGHT 30 705 #define SAME_AGE_REGION 1024 706 707 /* 708 * Define data block with age less than 1GB as hot data 709 * define data block with age less than 10GB but more than 1GB as warm data 710 */ 711 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 712 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 713 714 /* default max read extent count per inode */ 715 #define DEF_MAX_READ_EXTENT_COUNT 10240 716 717 /* extent cache type */ 718 enum extent_type { 719 EX_READ, 720 EX_BLOCK_AGE, 721 NR_EXTENT_CACHES, 722 }; 723 724 /* 725 * Reserved value to mark invalid age extents, hence valid block range 726 * from 0 to ULLONG_MAX-1 727 */ 728 #define F2FS_EXTENT_AGE_INVALID ULLONG_MAX 729 730 struct extent_info { 731 unsigned int fofs; /* start offset in a file */ 732 unsigned int len; /* length of the extent */ 733 union { 734 /* read extent_cache */ 735 struct { 736 /* start block address of the extent */ 737 block_t blk; 738 #ifdef CONFIG_F2FS_FS_COMPRESSION 739 /* physical extent length of compressed blocks */ 740 unsigned int c_len; 741 #endif 742 }; 743 /* block age extent_cache */ 744 struct { 745 /* block age of the extent */ 746 unsigned long long age; 747 /* last total blocks allocated */ 748 unsigned long long last_blocks; 749 }; 750 }; 751 }; 752 753 struct extent_node { 754 struct rb_node rb_node; /* rb node located in rb-tree */ 755 struct extent_info ei; /* extent info */ 756 struct list_head list; /* node in global extent list of sbi */ 757 struct extent_tree *et; /* extent tree pointer */ 758 }; 759 760 struct extent_tree { 761 nid_t ino; /* inode number */ 762 enum extent_type type; /* keep the extent tree type */ 763 struct rb_root_cached root; /* root of extent info rb-tree */ 764 struct extent_node *cached_en; /* recently accessed extent node */ 765 struct list_head list; /* to be used by sbi->zombie_list */ 766 rwlock_t lock; /* protect extent info rb-tree */ 767 atomic_t node_cnt; /* # of extent node in rb-tree*/ 768 bool largest_updated; /* largest extent updated */ 769 struct extent_info largest; /* largest cached extent for EX_READ */ 770 }; 771 772 struct extent_tree_info { 773 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 774 struct mutex extent_tree_lock; /* locking extent radix tree */ 775 struct list_head extent_list; /* lru list for shrinker */ 776 spinlock_t extent_lock; /* locking extent lru list */ 777 atomic_t total_ext_tree; /* extent tree count */ 778 struct list_head zombie_list; /* extent zombie tree list */ 779 atomic_t total_zombie_tree; /* extent zombie tree count */ 780 atomic_t total_ext_node; /* extent info count */ 781 }; 782 783 /* 784 * State of block returned by f2fs_map_blocks. 785 */ 786 #define F2FS_MAP_NEW (1U << 0) 787 #define F2FS_MAP_MAPPED (1U << 1) 788 #define F2FS_MAP_DELALLOC (1U << 2) 789 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 790 F2FS_MAP_DELALLOC) 791 792 struct f2fs_map_blocks { 793 struct block_device *m_bdev; /* for multi-device dio */ 794 block_t m_pblk; 795 block_t m_lblk; 796 unsigned int m_len; 797 unsigned int m_flags; 798 unsigned long m_last_pblk; /* last allocated block, only used for DIO in LFS mode */ 799 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 800 pgoff_t *m_next_extent; /* point to next possible extent */ 801 int m_seg_type; 802 bool m_may_create; /* indicate it is from write path */ 803 bool m_multidev_dio; /* indicate it allows multi-device dio */ 804 }; 805 806 /* for flag in get_data_block */ 807 enum { 808 F2FS_GET_BLOCK_DEFAULT, 809 F2FS_GET_BLOCK_FIEMAP, 810 F2FS_GET_BLOCK_BMAP, 811 F2FS_GET_BLOCK_DIO, 812 F2FS_GET_BLOCK_PRE_DIO, 813 F2FS_GET_BLOCK_PRE_AIO, 814 F2FS_GET_BLOCK_PRECACHE, 815 }; 816 817 /* 818 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 819 */ 820 #define FADVISE_COLD_BIT 0x01 821 #define FADVISE_LOST_PINO_BIT 0x02 822 #define FADVISE_ENCRYPT_BIT 0x04 823 #define FADVISE_ENC_NAME_BIT 0x08 824 #define FADVISE_KEEP_SIZE_BIT 0x10 825 #define FADVISE_HOT_BIT 0x20 826 #define FADVISE_VERITY_BIT 0x40 827 #define FADVISE_TRUNC_BIT 0x80 828 829 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 830 831 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 832 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 833 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 834 835 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 836 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 837 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 838 839 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 840 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 841 842 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 843 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 844 845 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 846 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 847 848 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 849 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 850 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 851 852 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 853 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 854 855 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 856 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 857 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 858 859 #define DEF_DIR_LEVEL 0 860 861 /* used for f2fs_inode_info->flags */ 862 enum { 863 FI_NEW_INODE, /* indicate newly allocated inode */ 864 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 865 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 866 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 867 FI_INC_LINK, /* need to increment i_nlink */ 868 FI_ACL_MODE, /* indicate acl mode */ 869 FI_NO_ALLOC, /* should not allocate any blocks */ 870 FI_FREE_NID, /* free allocated nide */ 871 FI_NO_EXTENT, /* not to use the extent cache */ 872 FI_INLINE_XATTR, /* used for inline xattr */ 873 FI_INLINE_DATA, /* used for inline data*/ 874 FI_INLINE_DENTRY, /* used for inline dentry */ 875 FI_APPEND_WRITE, /* inode has appended data */ 876 FI_UPDATE_WRITE, /* inode has in-place-update data */ 877 FI_NEED_IPU, /* used for ipu per file */ 878 FI_ATOMIC_FILE, /* indicate atomic file */ 879 FI_DATA_EXIST, /* indicate data exists */ 880 FI_SKIP_WRITES, /* should skip data page writeback */ 881 FI_OPU_WRITE, /* used for opu per file */ 882 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 883 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 884 FI_HOT_DATA, /* indicate file is hot */ 885 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 886 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 887 FI_PIN_FILE, /* indicate file should not be gced */ 888 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 889 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 890 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 891 FI_MMAP_FILE, /* indicate file was mmapped */ 892 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 893 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 894 FI_ALIGNED_WRITE, /* enable aligned write */ 895 FI_COW_FILE, /* indicate COW file */ 896 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 897 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 898 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 899 FI_OPENED_FILE, /* indicate file has been opened */ 900 FI_DONATE_FINISHED, /* indicate page donation of file has been finished */ 901 FI_MAX, /* max flag, never be used */ 902 }; 903 904 struct f2fs_inode_info { 905 struct inode vfs_inode; /* serve a vfs inode */ 906 unsigned long i_flags; /* keep an inode flags for ioctl */ 907 unsigned char i_advise; /* use to give file attribute hints */ 908 unsigned char i_dir_level; /* use for dentry level for large dir */ 909 union { 910 unsigned int i_current_depth; /* only for directory depth */ 911 unsigned short i_gc_failures; /* for gc failure statistic */ 912 }; 913 unsigned int i_pino; /* parent inode number */ 914 umode_t i_acl_mode; /* keep file acl mode temporarily */ 915 916 /* Use below internally in f2fs*/ 917 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 918 unsigned int ioprio_hint; /* hint for IO priority */ 919 struct f2fs_rwsem i_sem; /* protect fi info */ 920 atomic_t dirty_pages; /* # of dirty pages */ 921 f2fs_hash_t chash; /* hash value of given file name */ 922 unsigned int clevel; /* maximum level of given file name */ 923 struct task_struct *task; /* lookup and create consistency */ 924 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 925 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 926 nid_t i_xattr_nid; /* node id that contains xattrs */ 927 loff_t last_disk_size; /* lastly written file size */ 928 spinlock_t i_size_lock; /* protect last_disk_size */ 929 930 #ifdef CONFIG_QUOTA 931 struct dquot __rcu *i_dquot[MAXQUOTAS]; 932 933 /* quota space reservation, managed internally by quota code */ 934 qsize_t i_reserved_quota; 935 #endif 936 struct list_head dirty_list; /* dirty list for dirs and files */ 937 struct list_head gdirty_list; /* linked in global dirty list */ 938 939 /* linked in global inode list for cache donation */ 940 struct list_head gdonate_list; 941 pgoff_t donate_start, donate_end; /* inclusive */ 942 atomic_t open_count; /* # of open files */ 943 944 struct task_struct *atomic_write_task; /* store atomic write task */ 945 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 946 /* cached extent_tree entry */ 947 union { 948 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 949 struct inode *atomic_inode; 950 /* point to atomic_inode, available only for cow_inode */ 951 }; 952 953 /* avoid racing between foreground op and gc */ 954 struct f2fs_rwsem i_gc_rwsem[2]; 955 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 956 957 int i_extra_isize; /* size of extra space located in i_addr */ 958 kprojid_t i_projid; /* id for project quota */ 959 int i_inline_xattr_size; /* inline xattr size */ 960 struct timespec64 i_crtime; /* inode creation time */ 961 struct timespec64 i_disk_time[3];/* inode disk times */ 962 963 /* for file compress */ 964 atomic_t i_compr_blocks; /* # of compressed blocks */ 965 unsigned char i_compress_algorithm; /* algorithm type */ 966 unsigned char i_log_cluster_size; /* log of cluster size */ 967 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 968 unsigned char i_compress_flag; /* compress flag */ 969 unsigned int i_cluster_size; /* cluster size */ 970 atomic_t writeback; /* count # of writeback thread */ 971 972 unsigned int atomic_write_cnt; 973 loff_t original_i_size; /* original i_size before atomic write */ 974 #ifdef CONFIG_FS_ENCRYPTION 975 struct fscrypt_inode_info *i_crypt_info; /* filesystem encryption info */ 976 #endif 977 }; 978 979 static inline void get_read_extent_info(struct extent_info *ext, 980 struct f2fs_extent *i_ext) 981 { 982 ext->fofs = le32_to_cpu(i_ext->fofs); 983 ext->blk = le32_to_cpu(i_ext->blk); 984 ext->len = le32_to_cpu(i_ext->len); 985 } 986 987 static inline void set_raw_read_extent(struct extent_info *ext, 988 struct f2fs_extent *i_ext) 989 { 990 i_ext->fofs = cpu_to_le32(ext->fofs); 991 i_ext->blk = cpu_to_le32(ext->blk); 992 i_ext->len = cpu_to_le32(ext->len); 993 } 994 995 static inline bool __is_discard_mergeable(struct discard_info *back, 996 struct discard_info *front, unsigned int max_len) 997 { 998 return (back->lstart + back->len == front->lstart) && 999 (back->len + front->len <= max_len); 1000 } 1001 1002 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 1003 struct discard_info *back, unsigned int max_len) 1004 { 1005 return __is_discard_mergeable(back, cur, max_len); 1006 } 1007 1008 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 1009 struct discard_info *front, unsigned int max_len) 1010 { 1011 return __is_discard_mergeable(cur, front, max_len); 1012 } 1013 1014 /* 1015 * For free nid management 1016 */ 1017 enum nid_state { 1018 FREE_NID, /* newly added to free nid list */ 1019 PREALLOC_NID, /* it is preallocated */ 1020 MAX_NID_STATE, 1021 }; 1022 1023 enum nat_state { 1024 TOTAL_NAT, 1025 DIRTY_NAT, 1026 RECLAIMABLE_NAT, 1027 MAX_NAT_STATE, 1028 }; 1029 1030 struct f2fs_nm_info { 1031 block_t nat_blkaddr; /* base disk address of NAT */ 1032 nid_t max_nid; /* maximum possible node ids */ 1033 nid_t available_nids; /* # of available node ids */ 1034 nid_t next_scan_nid; /* the next nid to be scanned */ 1035 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 1036 unsigned int ram_thresh; /* control the memory footprint */ 1037 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 1038 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 1039 1040 /* NAT cache management */ 1041 struct radix_tree_root nat_root;/* root of the nat entry cache */ 1042 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 1043 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 1044 struct list_head nat_entries; /* cached nat entry list (clean) */ 1045 spinlock_t nat_list_lock; /* protect clean nat entry list */ 1046 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 1047 unsigned int nat_blocks; /* # of nat blocks */ 1048 1049 /* free node ids management */ 1050 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 1051 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 1052 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 1053 spinlock_t nid_list_lock; /* protect nid lists ops */ 1054 struct mutex build_lock; /* lock for build free nids */ 1055 unsigned char **free_nid_bitmap; 1056 unsigned char *nat_block_bitmap; 1057 unsigned short *free_nid_count; /* free nid count of NAT block */ 1058 1059 /* for checkpoint */ 1060 char *nat_bitmap; /* NAT bitmap pointer */ 1061 1062 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 1063 unsigned char *nat_bits; /* NAT bits blocks */ 1064 unsigned char *full_nat_bits; /* full NAT pages */ 1065 unsigned char *empty_nat_bits; /* empty NAT pages */ 1066 #ifdef CONFIG_F2FS_CHECK_FS 1067 char *nat_bitmap_mir; /* NAT bitmap mirror */ 1068 #endif 1069 int bitmap_size; /* bitmap size */ 1070 }; 1071 1072 /* 1073 * this structure is used as one of function parameters. 1074 * all the information are dedicated to a given direct node block determined 1075 * by the data offset in a file. 1076 */ 1077 struct dnode_of_data { 1078 struct inode *inode; /* vfs inode pointer */ 1079 struct folio *inode_folio; /* its inode folio, NULL is possible */ 1080 struct folio *node_folio; /* cached direct node folio */ 1081 nid_t nid; /* node id of the direct node block */ 1082 unsigned int ofs_in_node; /* data offset in the node page */ 1083 bool inode_folio_locked; /* inode folio is locked or not */ 1084 bool node_changed; /* is node block changed */ 1085 char cur_level; /* level of hole node page */ 1086 char max_level; /* level of current page located */ 1087 block_t data_blkaddr; /* block address of the node block */ 1088 }; 1089 1090 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 1091 struct folio *ifolio, struct folio *nfolio, nid_t nid) 1092 { 1093 memset(dn, 0, sizeof(*dn)); 1094 dn->inode = inode; 1095 dn->inode_folio = ifolio; 1096 dn->node_folio = nfolio; 1097 dn->nid = nid; 1098 } 1099 1100 /* 1101 * For SIT manager 1102 * 1103 * By default, there are 6 active log areas across the whole main area. 1104 * When considering hot and cold data separation to reduce cleaning overhead, 1105 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1106 * respectively. 1107 * In the current design, you should not change the numbers intentionally. 1108 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1109 * logs individually according to the underlying devices. (default: 6) 1110 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1111 * data and 8 for node logs. 1112 */ 1113 #define NR_CURSEG_DATA_TYPE (3) 1114 #define NR_CURSEG_NODE_TYPE (3) 1115 #define NR_CURSEG_INMEM_TYPE (2) 1116 #define NR_CURSEG_RO_TYPE (2) 1117 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1118 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1119 1120 enum log_type { 1121 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1122 CURSEG_WARM_DATA, /* data blocks */ 1123 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1124 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1125 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1126 CURSEG_COLD_NODE, /* indirect node blocks */ 1127 NR_PERSISTENT_LOG, /* number of persistent log */ 1128 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1129 /* pinned file that needs consecutive block address */ 1130 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1131 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1132 }; 1133 1134 struct flush_cmd { 1135 struct completion wait; 1136 struct llist_node llnode; 1137 nid_t ino; 1138 int ret; 1139 }; 1140 1141 struct flush_cmd_control { 1142 struct task_struct *f2fs_issue_flush; /* flush thread */ 1143 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1144 atomic_t issued_flush; /* # of issued flushes */ 1145 atomic_t queued_flush; /* # of queued flushes */ 1146 struct llist_head issue_list; /* list for command issue */ 1147 struct llist_node *dispatch_list; /* list for command dispatch */ 1148 }; 1149 1150 struct f2fs_sm_info { 1151 struct sit_info *sit_info; /* whole segment information */ 1152 struct free_segmap_info *free_info; /* free segment information */ 1153 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1154 struct curseg_info *curseg_array; /* active segment information */ 1155 1156 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1157 1158 block_t seg0_blkaddr; /* block address of 0'th segment */ 1159 block_t main_blkaddr; /* start block address of main area */ 1160 block_t ssa_blkaddr; /* start block address of SSA area */ 1161 1162 unsigned int segment_count; /* total # of segments */ 1163 unsigned int main_segments; /* # of segments in main area */ 1164 unsigned int reserved_segments; /* # of reserved segments */ 1165 unsigned int ovp_segments; /* # of overprovision segments */ 1166 1167 /* a threshold to reclaim prefree segments */ 1168 unsigned int rec_prefree_segments; 1169 1170 struct list_head sit_entry_set; /* sit entry set list */ 1171 1172 unsigned int ipu_policy; /* in-place-update policy */ 1173 unsigned int min_ipu_util; /* in-place-update threshold */ 1174 unsigned int min_fsync_blocks; /* threshold for fsync */ 1175 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1176 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1177 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1178 1179 /* for flush command control */ 1180 struct flush_cmd_control *fcc_info; 1181 1182 /* for discard command control */ 1183 struct discard_cmd_control *dcc_info; 1184 }; 1185 1186 /* 1187 * For superblock 1188 */ 1189 /* 1190 * COUNT_TYPE for monitoring 1191 * 1192 * f2fs monitors the number of several block types such as on-writeback, 1193 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1194 */ 1195 #define WB_DATA_TYPE(folio, f) \ 1196 (f || f2fs_is_cp_guaranteed(folio) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1197 enum count_type { 1198 F2FS_DIRTY_DENTS, 1199 F2FS_DIRTY_DATA, 1200 F2FS_DIRTY_QDATA, 1201 F2FS_DIRTY_NODES, 1202 F2FS_DIRTY_META, 1203 F2FS_DIRTY_IMETA, 1204 F2FS_WB_CP_DATA, 1205 F2FS_WB_DATA, 1206 F2FS_RD_DATA, 1207 F2FS_RD_NODE, 1208 F2FS_RD_META, 1209 F2FS_DIO_WRITE, 1210 F2FS_DIO_READ, 1211 NR_COUNT_TYPE, 1212 }; 1213 1214 /* 1215 * The below are the page types of bios used in submit_bio(). 1216 * The available types are: 1217 * DATA User data pages. It operates as async mode. 1218 * NODE Node pages. It operates as async mode. 1219 * META FS metadata pages such as SIT, NAT, CP. 1220 * NR_PAGE_TYPE The number of page types. 1221 * META_FLUSH Make sure the previous pages are written 1222 * with waiting the bio's completion 1223 * ... Only can be used with META. 1224 */ 1225 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1226 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1227 enum page_type { 1228 DATA = 0, 1229 NODE = 1, /* should not change this */ 1230 META, 1231 NR_PAGE_TYPE, 1232 META_FLUSH, 1233 IPU, /* the below types are used by tracepoints only. */ 1234 OPU, 1235 }; 1236 1237 enum temp_type { 1238 HOT = 0, /* must be zero for meta bio */ 1239 WARM, 1240 COLD, 1241 NR_TEMP_TYPE, 1242 }; 1243 1244 enum need_lock_type { 1245 LOCK_REQ = 0, 1246 LOCK_DONE, 1247 LOCK_RETRY, 1248 }; 1249 1250 enum cp_reason_type { 1251 CP_NO_NEEDED, 1252 CP_NON_REGULAR, 1253 CP_COMPRESSED, 1254 CP_HARDLINK, 1255 CP_SB_NEED_CP, 1256 CP_WRONG_PINO, 1257 CP_NO_SPC_ROLL, 1258 CP_NODE_NEED_CP, 1259 CP_FASTBOOT_MODE, 1260 CP_SPEC_LOG_NUM, 1261 CP_RECOVER_DIR, 1262 CP_XATTR_DIR, 1263 }; 1264 1265 enum iostat_type { 1266 /* WRITE IO */ 1267 APP_DIRECT_IO, /* app direct write IOs */ 1268 APP_BUFFERED_IO, /* app buffered write IOs */ 1269 APP_WRITE_IO, /* app write IOs */ 1270 APP_MAPPED_IO, /* app mapped IOs */ 1271 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1272 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1273 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1274 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1275 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1276 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1277 FS_GC_DATA_IO, /* data IOs from forground gc */ 1278 FS_GC_NODE_IO, /* node IOs from forground gc */ 1279 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1280 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1281 FS_CP_META_IO, /* meta IOs from checkpoint */ 1282 1283 /* READ IO */ 1284 APP_DIRECT_READ_IO, /* app direct read IOs */ 1285 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1286 APP_READ_IO, /* app read IOs */ 1287 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1288 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1289 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1290 FS_DATA_READ_IO, /* data read IOs */ 1291 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1292 FS_CDATA_READ_IO, /* compressed data read IOs */ 1293 FS_NODE_READ_IO, /* node read IOs */ 1294 FS_META_READ_IO, /* meta read IOs */ 1295 1296 /* other */ 1297 FS_DISCARD_IO, /* discard */ 1298 FS_FLUSH_IO, /* flush */ 1299 FS_ZONE_RESET_IO, /* zone reset */ 1300 NR_IO_TYPE, 1301 }; 1302 1303 struct f2fs_io_info { 1304 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1305 nid_t ino; /* inode number */ 1306 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1307 enum temp_type temp; /* contains HOT/WARM/COLD */ 1308 enum req_op op; /* contains REQ_OP_ */ 1309 blk_opf_t op_flags; /* req_flag_bits */ 1310 block_t new_blkaddr; /* new block address to be written */ 1311 block_t old_blkaddr; /* old block address before Cow */ 1312 union { 1313 struct page *page; /* page to be written */ 1314 struct folio *folio; 1315 }; 1316 struct page *encrypted_page; /* encrypted page */ 1317 struct page *compressed_page; /* compressed page */ 1318 struct list_head list; /* serialize IOs */ 1319 unsigned int compr_blocks; /* # of compressed block addresses */ 1320 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1321 unsigned int version:8; /* version of the node */ 1322 unsigned int submitted:1; /* indicate IO submission */ 1323 unsigned int in_list:1; /* indicate fio is in io_list */ 1324 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1325 unsigned int encrypted:1; /* indicate file is encrypted */ 1326 unsigned int meta_gc:1; /* require meta inode GC */ 1327 enum iostat_type io_type; /* io type */ 1328 struct writeback_control *io_wbc; /* writeback control */ 1329 struct bio **bio; /* bio for ipu */ 1330 sector_t *last_block; /* last block number in bio */ 1331 }; 1332 1333 struct bio_entry { 1334 struct bio *bio; 1335 struct list_head list; 1336 }; 1337 1338 #define is_read_io(rw) ((rw) == READ) 1339 struct f2fs_bio_info { 1340 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1341 struct bio *bio; /* bios to merge */ 1342 sector_t last_block_in_bio; /* last block number */ 1343 struct f2fs_io_info fio; /* store buffered io info. */ 1344 #ifdef CONFIG_BLK_DEV_ZONED 1345 struct completion zone_wait; /* condition value for the previous open zone to close */ 1346 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1347 void *bi_private; /* previous bi_private for pending bio */ 1348 #endif 1349 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1350 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1351 struct list_head io_list; /* track fios */ 1352 struct list_head bio_list; /* bio entry list head */ 1353 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1354 }; 1355 1356 #define FDEV(i) (sbi->devs[i]) 1357 #define RDEV(i) (raw_super->devs[i]) 1358 struct f2fs_dev_info { 1359 struct file *bdev_file; 1360 struct block_device *bdev; 1361 char path[MAX_PATH_LEN + 1]; 1362 unsigned int total_segments; 1363 block_t start_blk; 1364 block_t end_blk; 1365 #ifdef CONFIG_BLK_DEV_ZONED 1366 unsigned int nr_blkz; /* Total number of zones */ 1367 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1368 #endif 1369 }; 1370 1371 enum inode_type { 1372 DIR_INODE, /* for dirty dir inode */ 1373 FILE_INODE, /* for dirty regular/symlink inode */ 1374 DIRTY_META, /* for all dirtied inode metadata */ 1375 DONATE_INODE, /* for all inode to donate pages */ 1376 NR_INODE_TYPE, 1377 }; 1378 1379 /* for inner inode cache management */ 1380 struct inode_management { 1381 struct radix_tree_root ino_root; /* ino entry array */ 1382 spinlock_t ino_lock; /* for ino entry lock */ 1383 struct list_head ino_list; /* inode list head */ 1384 unsigned long ino_num; /* number of entries */ 1385 }; 1386 1387 /* for GC_AT */ 1388 struct atgc_management { 1389 bool atgc_enabled; /* ATGC is enabled or not */ 1390 struct rb_root_cached root; /* root of victim rb-tree */ 1391 struct list_head victim_list; /* linked with all victim entries */ 1392 unsigned int victim_count; /* victim count in rb-tree */ 1393 unsigned int candidate_ratio; /* candidate ratio */ 1394 unsigned int max_candidate_count; /* max candidate count */ 1395 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1396 unsigned long long age_threshold; /* age threshold */ 1397 }; 1398 1399 struct f2fs_gc_control { 1400 unsigned int victim_segno; /* target victim segment number */ 1401 int init_gc_type; /* FG_GC or BG_GC */ 1402 bool no_bg_gc; /* check the space and stop bg_gc */ 1403 bool should_migrate_blocks; /* should migrate blocks */ 1404 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1405 bool one_time; /* require one time GC in one migration unit */ 1406 unsigned int nr_free_secs; /* # of free sections to do GC */ 1407 }; 1408 1409 /* 1410 * For s_flag in struct f2fs_sb_info 1411 * Modification on enum should be synchronized with s_flag array 1412 */ 1413 enum { 1414 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1415 SBI_IS_CLOSE, /* specify unmounting */ 1416 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1417 SBI_POR_DOING, /* recovery is doing or not */ 1418 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1419 SBI_NEED_CP, /* need to checkpoint */ 1420 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1421 SBI_IS_RECOVERED, /* recovered orphan/data */ 1422 SBI_CP_DISABLED, /* CP was disabled last mount */ 1423 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1424 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1425 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1426 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1427 SBI_IS_RESIZEFS, /* resizefs is in process */ 1428 SBI_IS_FREEZING, /* freezefs is in process */ 1429 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1430 MAX_SBI_FLAG, 1431 }; 1432 1433 enum { 1434 CP_TIME, 1435 REQ_TIME, 1436 DISCARD_TIME, 1437 GC_TIME, 1438 DISABLE_TIME, 1439 ENABLE_TIME, 1440 UMOUNT_DISCARD_TIMEOUT, 1441 MAX_TIME, 1442 }; 1443 1444 /* Note that you need to keep synchronization with this gc_mode_names array */ 1445 enum { 1446 GC_NORMAL, 1447 GC_IDLE_CB, 1448 GC_IDLE_GREEDY, 1449 GC_IDLE_AT, 1450 GC_URGENT_HIGH, 1451 GC_URGENT_LOW, 1452 GC_URGENT_MID, 1453 MAX_GC_MODE, 1454 }; 1455 1456 enum { 1457 BGGC_MODE_ON, /* background gc is on */ 1458 BGGC_MODE_OFF, /* background gc is off */ 1459 BGGC_MODE_SYNC, /* 1460 * background gc is on, migrating blocks 1461 * like foreground gc 1462 */ 1463 }; 1464 1465 enum { 1466 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1467 FS_MODE_LFS, /* use lfs allocation only */ 1468 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1469 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1470 }; 1471 1472 enum { 1473 ALLOC_MODE_DEFAULT, /* stay default */ 1474 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1475 }; 1476 1477 enum fsync_mode { 1478 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1479 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1480 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1481 }; 1482 1483 enum { 1484 COMPR_MODE_FS, /* 1485 * automatically compress compression 1486 * enabled files 1487 */ 1488 COMPR_MODE_USER, /* 1489 * automatical compression is disabled. 1490 * user can control the file compression 1491 * using ioctls 1492 */ 1493 }; 1494 1495 enum { 1496 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1497 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1498 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1499 }; 1500 1501 enum { 1502 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1503 MEMORY_MODE_LOW, /* memory mode for low memory devices */ 1504 }; 1505 1506 enum errors_option { 1507 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1508 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1509 MOUNT_ERRORS_PANIC, /* panic on errors */ 1510 }; 1511 1512 enum { 1513 BACKGROUND, 1514 FOREGROUND, 1515 MAX_CALL_TYPE, 1516 TOTAL_CALL = FOREGROUND, 1517 }; 1518 1519 enum f2fs_lookup_mode { 1520 LOOKUP_PERF, 1521 LOOKUP_COMPAT, 1522 LOOKUP_AUTO, 1523 }; 1524 1525 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1526 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1527 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1528 1529 /* 1530 * Layout of f2fs page.private: 1531 * 1532 * Layout A: lowest bit should be 1 1533 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1534 * bit 0 PAGE_PRIVATE_NOT_POINTER 1535 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1536 * bit 2 PAGE_PRIVATE_INLINE_INODE 1537 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1538 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1539 * bit 5- f2fs private data 1540 * 1541 * Layout B: lowest bit should be 0 1542 * page.private is a wrapped pointer. 1543 */ 1544 enum { 1545 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1546 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1547 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1548 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1549 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1550 PAGE_PRIVATE_MAX 1551 }; 1552 1553 /* For compression */ 1554 enum compress_algorithm_type { 1555 COMPRESS_LZO, 1556 COMPRESS_LZ4, 1557 COMPRESS_ZSTD, 1558 COMPRESS_LZORLE, 1559 COMPRESS_MAX, 1560 }; 1561 1562 enum compress_flag { 1563 COMPRESS_CHKSUM, 1564 COMPRESS_MAX_FLAG, 1565 }; 1566 1567 #define COMPRESS_WATERMARK 20 1568 #define COMPRESS_PERCENT 20 1569 1570 #define COMPRESS_DATA_RESERVED_SIZE 4 1571 struct compress_data { 1572 __le32 clen; /* compressed data size */ 1573 __le32 chksum; /* compressed data checksum */ 1574 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1575 u8 cdata[]; /* compressed data */ 1576 }; 1577 1578 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1579 1580 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1581 1582 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1583 1584 #define COMPRESS_LEVEL_OFFSET 8 1585 1586 /* compress context */ 1587 struct compress_ctx { 1588 struct inode *inode; /* inode the context belong to */ 1589 pgoff_t cluster_idx; /* cluster index number */ 1590 unsigned int cluster_size; /* page count in cluster */ 1591 unsigned int log_cluster_size; /* log of cluster size */ 1592 struct page **rpages; /* pages store raw data in cluster */ 1593 unsigned int nr_rpages; /* total page number in rpages */ 1594 struct page **cpages; /* pages store compressed data in cluster */ 1595 unsigned int nr_cpages; /* total page number in cpages */ 1596 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1597 void *rbuf; /* virtual mapped address on rpages */ 1598 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1599 size_t rlen; /* valid data length in rbuf */ 1600 size_t clen; /* valid data length in cbuf */ 1601 void *private; /* payload buffer for specified compression algorithm */ 1602 void *private2; /* extra payload buffer */ 1603 struct fsverity_info *vi; /* verity info if needed */ 1604 }; 1605 1606 /* compress context for write IO path */ 1607 struct compress_io_ctx { 1608 u32 magic; /* magic number to indicate page is compressed */ 1609 struct inode *inode; /* inode the context belong to */ 1610 struct page **rpages; /* pages store raw data in cluster */ 1611 unsigned int nr_rpages; /* total page number in rpages */ 1612 atomic_t pending_pages; /* in-flight compressed page count */ 1613 }; 1614 1615 /* Context for decompressing one cluster on the read IO path */ 1616 struct decompress_io_ctx { 1617 u32 magic; /* magic number to indicate page is compressed */ 1618 struct inode *inode; /* inode the context belong to */ 1619 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1620 pgoff_t cluster_idx; /* cluster index number */ 1621 unsigned int cluster_size; /* page count in cluster */ 1622 unsigned int log_cluster_size; /* log of cluster size */ 1623 struct page **rpages; /* pages store raw data in cluster */ 1624 unsigned int nr_rpages; /* total page number in rpages */ 1625 struct page **cpages; /* pages store compressed data in cluster */ 1626 unsigned int nr_cpages; /* total page number in cpages */ 1627 struct page **tpages; /* temp pages to pad holes in cluster */ 1628 void *rbuf; /* virtual mapped address on rpages */ 1629 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1630 size_t rlen; /* valid data length in rbuf */ 1631 size_t clen; /* valid data length in cbuf */ 1632 1633 /* 1634 * The number of compressed pages remaining to be read in this cluster. 1635 * This is initially nr_cpages. It is decremented by 1 each time a page 1636 * has been read (or failed to be read). When it reaches 0, the cluster 1637 * is decompressed (or an error is reported). 1638 * 1639 * If an error occurs before all the pages have been submitted for I/O, 1640 * then this will never reach 0. In this case the I/O submitter is 1641 * responsible for calling f2fs_decompress_end_io() instead. 1642 */ 1643 atomic_t remaining_pages; 1644 1645 /* 1646 * Number of references to this decompress_io_ctx. 1647 * 1648 * One reference is held for I/O completion. This reference is dropped 1649 * after the pagecache pages are updated and unlocked -- either after 1650 * decompression (and verity if enabled), or after an error. 1651 * 1652 * In addition, each compressed page holds a reference while it is in a 1653 * bio. These references are necessary prevent compressed pages from 1654 * being freed while they are still in a bio. 1655 */ 1656 refcount_t refcnt; 1657 1658 bool failed; /* IO error occurred before decompression? */ 1659 struct fsverity_info *vi; /* fs-verity context if needed */ 1660 unsigned char compress_algorithm; /* backup algorithm type */ 1661 void *private; /* payload buffer for specified decompression algorithm */ 1662 void *private2; /* extra payload buffer */ 1663 struct work_struct verity_work; /* work to verify the decompressed pages */ 1664 struct work_struct free_work; /* work for late free this structure itself */ 1665 }; 1666 1667 #define NULL_CLUSTER ((unsigned int)(~0)) 1668 #define MIN_COMPRESS_LOG_SIZE 2 1669 #define MAX_COMPRESS_LOG_SIZE 8 1670 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1671 1672 struct f2fs_sb_info { 1673 struct super_block *sb; /* pointer to VFS super block */ 1674 struct proc_dir_entry *s_proc; /* proc entry */ 1675 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1676 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1677 int valid_super_block; /* valid super block no */ 1678 unsigned long s_flag; /* flags for sbi */ 1679 struct mutex writepages; /* mutex for writepages() */ 1680 1681 #ifdef CONFIG_BLK_DEV_ZONED 1682 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1683 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1684 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1685 /* For adjust the priority writing position of data in zone UFS */ 1686 unsigned int blkzone_alloc_policy; 1687 #endif 1688 1689 /* for node-related operations */ 1690 struct f2fs_nm_info *nm_info; /* node manager */ 1691 struct inode *node_inode; /* cache node blocks */ 1692 1693 /* for segment-related operations */ 1694 struct f2fs_sm_info *sm_info; /* segment manager */ 1695 1696 /* for bio operations */ 1697 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1698 /* keep migration IO order for LFS mode */ 1699 struct f2fs_rwsem io_order_lock; 1700 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1701 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1702 1703 /* for checkpoint */ 1704 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1705 int cur_cp_pack; /* remain current cp pack */ 1706 spinlock_t cp_lock; /* for flag in ckpt */ 1707 struct inode *meta_inode; /* cache meta blocks */ 1708 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1709 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1710 struct f2fs_rwsem node_write; /* locking node writes */ 1711 struct f2fs_rwsem node_change; /* locking node change */ 1712 wait_queue_head_t cp_wait; 1713 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1714 long interval_time[MAX_TIME]; /* to store thresholds */ 1715 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1716 struct cp_stats cp_stats; /* for time stat of checkpoint */ 1717 struct f2fs_rwsem cp_enable_rwsem; /* block cache/dio write */ 1718 1719 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1720 1721 spinlock_t fsync_node_lock; /* for node entry lock */ 1722 struct list_head fsync_node_list; /* node list head */ 1723 unsigned int fsync_seg_id; /* sequence id */ 1724 unsigned int fsync_node_num; /* number of node entries */ 1725 1726 /* for orphan inode, use 0'th array */ 1727 unsigned int max_orphans; /* max orphan inodes */ 1728 1729 /* for inode management */ 1730 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1731 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1732 struct mutex flush_lock; /* for flush exclusion */ 1733 1734 /* for extent tree cache */ 1735 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1736 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1737 unsigned int max_read_extent_count; /* max read extent count per inode */ 1738 1739 /* The threshold used for hot and warm data seperation*/ 1740 unsigned int hot_data_age_threshold; 1741 unsigned int warm_data_age_threshold; 1742 unsigned int last_age_weight; 1743 1744 /* control donate caches */ 1745 unsigned int donate_files; 1746 1747 /* basic filesystem units */ 1748 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1749 unsigned int log_blocksize; /* log2 block size */ 1750 unsigned int blocksize; /* block size */ 1751 unsigned int root_ino_num; /* root inode number*/ 1752 unsigned int node_ino_num; /* node inode number*/ 1753 unsigned int meta_ino_num; /* meta inode number*/ 1754 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1755 unsigned int blocks_per_seg; /* blocks per segment */ 1756 unsigned int segs_per_sec; /* segments per section */ 1757 unsigned int secs_per_zone; /* sections per zone */ 1758 unsigned int total_sections; /* total section count */ 1759 unsigned int total_node_count; /* total node block count */ 1760 unsigned int total_valid_node_count; /* valid node block count */ 1761 int dir_level; /* directory level */ 1762 bool readdir_ra; /* readahead inode in readdir */ 1763 u64 max_io_bytes; /* max io bytes to merge IOs */ 1764 1765 block_t user_block_count; /* # of user blocks */ 1766 block_t total_valid_block_count; /* # of valid blocks */ 1767 block_t discard_blks; /* discard command candidats */ 1768 block_t last_valid_block_count; /* for recovery */ 1769 block_t reserved_blocks; /* configurable reserved blocks */ 1770 block_t current_reserved_blocks; /* current reserved blocks */ 1771 1772 /* Additional tracking for no checkpoint mode */ 1773 block_t unusable_block_count; /* # of blocks saved by last cp */ 1774 1775 unsigned int nquota_files; /* # of quota sysfile */ 1776 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1777 struct task_struct *umount_lock_holder; /* s_umount lock holder */ 1778 1779 /* # of pages, see count_type */ 1780 atomic_t nr_pages[NR_COUNT_TYPE]; 1781 /* # of allocated blocks */ 1782 struct percpu_counter alloc_valid_block_count; 1783 /* # of node block writes as roll forward recovery */ 1784 struct percpu_counter rf_node_block_count; 1785 1786 /* writeback control */ 1787 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1788 1789 /* valid inode count */ 1790 struct percpu_counter total_valid_inode_count; 1791 1792 struct f2fs_mount_info mount_opt; /* mount options */ 1793 1794 /* for cleaning operations */ 1795 struct f2fs_rwsem gc_lock; /* 1796 * semaphore for GC, avoid 1797 * race between GC and GC or CP 1798 */ 1799 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1800 struct atgc_management am; /* atgc management */ 1801 unsigned int cur_victim_sec; /* current victim section num */ 1802 unsigned int gc_mode; /* current GC state */ 1803 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1804 spinlock_t gc_remaining_trials_lock; 1805 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1806 unsigned int gc_remaining_trials; 1807 1808 /* for skip statistic */ 1809 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1810 1811 /* free sections reserved for pinned file */ 1812 unsigned int reserved_pin_section; 1813 1814 /* threshold for gc trials on pinned files */ 1815 unsigned short gc_pin_file_threshold; 1816 struct f2fs_rwsem pin_sem; 1817 1818 /* maximum # of trials to find a victim segment for SSR and GC */ 1819 unsigned int max_victim_search; 1820 /* migration granularity of garbage collection, unit: segment */ 1821 unsigned int migration_granularity; 1822 /* migration window granularity of garbage collection, unit: segment */ 1823 unsigned int migration_window_granularity; 1824 1825 /* 1826 * for stat information. 1827 * one is for the LFS mode, and the other is for the SSR mode. 1828 */ 1829 #ifdef CONFIG_F2FS_STAT_FS 1830 struct f2fs_stat_info *stat_info; /* FS status information */ 1831 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1832 unsigned int segment_count[2]; /* # of allocated segments */ 1833 unsigned int block_count[2]; /* # of allocated blocks */ 1834 atomic_t inplace_count; /* # of inplace update */ 1835 /* # of lookup extent cache */ 1836 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1837 /* # of hit rbtree extent node */ 1838 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1839 /* # of hit cached extent node */ 1840 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1841 /* # of hit largest extent node in read extent cache */ 1842 atomic64_t read_hit_largest; 1843 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1844 atomic_t inline_inode; /* # of inline_data inodes */ 1845 atomic_t inline_dir; /* # of inline_dentry inodes */ 1846 atomic_t compr_inode; /* # of compressed inodes */ 1847 atomic64_t compr_blocks; /* # of compressed blocks */ 1848 atomic_t swapfile_inode; /* # of swapfile inodes */ 1849 atomic_t atomic_files; /* # of opened atomic file */ 1850 atomic_t max_aw_cnt; /* max # of atomic writes */ 1851 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1852 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1853 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1854 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1855 #endif 1856 spinlock_t stat_lock; /* lock for stat operations */ 1857 1858 /* to attach REQ_META|REQ_FUA flags */ 1859 unsigned int data_io_flag; 1860 unsigned int node_io_flag; 1861 1862 /* For sysfs support */ 1863 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1864 struct completion s_kobj_unregister; 1865 1866 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1867 struct completion s_stat_kobj_unregister; 1868 1869 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1870 struct completion s_feature_list_kobj_unregister; 1871 1872 /* For shrinker support */ 1873 struct list_head s_list; 1874 struct mutex umount_mutex; 1875 unsigned int shrinker_run_no; 1876 1877 /* For multi devices */ 1878 int s_ndevs; /* number of devices */ 1879 struct f2fs_dev_info *devs; /* for device list */ 1880 unsigned int dirty_device; /* for checkpoint data flush */ 1881 spinlock_t dev_lock; /* protect dirty_device */ 1882 bool aligned_blksize; /* all devices has the same logical blksize */ 1883 unsigned int first_seq_zone_segno; /* first segno in sequential zone */ 1884 unsigned int bggc_io_aware; /* For adjust the BG_GC priority when pending IO */ 1885 unsigned int allocate_section_hint; /* the boundary position between devices */ 1886 unsigned int allocate_section_policy; /* determine the section writing priority */ 1887 1888 /* For write statistics */ 1889 u64 sectors_written_start; 1890 u64 kbytes_written; 1891 1892 /* Precomputed FS UUID checksum for seeding other checksums */ 1893 __u32 s_chksum_seed; 1894 1895 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1896 1897 /* 1898 * If we are in irq context, let's update error information into 1899 * on-disk superblock in the work. 1900 */ 1901 struct work_struct s_error_work; 1902 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1903 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1904 spinlock_t error_lock; /* protect errors/stop_reason array */ 1905 bool error_dirty; /* errors of sb is dirty */ 1906 1907 /* For reclaimed segs statistics per each GC mode */ 1908 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1909 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1910 1911 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1912 1913 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1914 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1915 1916 /* For atomic write statistics */ 1917 atomic64_t current_atomic_write; 1918 s64 peak_atomic_write; 1919 u64 committed_atomic_block; 1920 u64 revoked_atomic_block; 1921 1922 /* carve out reserved_blocks from total blocks */ 1923 bool carve_out; 1924 1925 #ifdef CONFIG_F2FS_FS_COMPRESSION 1926 struct kmem_cache *page_array_slab; /* page array entry */ 1927 unsigned int page_array_slab_size; /* default page array slab size */ 1928 1929 /* For runtime compression statistics */ 1930 u64 compr_written_block; 1931 u64 compr_saved_block; 1932 u32 compr_new_inode; 1933 1934 /* For compressed block cache */ 1935 struct inode *compress_inode; /* cache compressed blocks */ 1936 unsigned int compress_percent; /* cache page percentage */ 1937 unsigned int compress_watermark; /* cache page watermark */ 1938 atomic_t compress_page_hit; /* cache hit count */ 1939 #endif 1940 1941 #ifdef CONFIG_F2FS_IOSTAT 1942 /* For app/fs IO statistics */ 1943 spinlock_t iostat_lock; 1944 unsigned long long iostat_count[NR_IO_TYPE]; 1945 unsigned long long iostat_bytes[NR_IO_TYPE]; 1946 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1947 bool iostat_enable; 1948 unsigned long iostat_next_period; 1949 unsigned int iostat_period_ms; 1950 1951 /* For io latency related statistics info in one iostat period */ 1952 spinlock_t iostat_lat_lock; 1953 struct iostat_lat_info *iostat_io_lat; 1954 #endif 1955 }; 1956 1957 /* Definitions to access f2fs_sb_info */ 1958 #define SEGS_TO_BLKS(sbi, segs) \ 1959 ((segs) << (sbi)->log_blocks_per_seg) 1960 #define BLKS_TO_SEGS(sbi, blks) \ 1961 ((blks) >> (sbi)->log_blocks_per_seg) 1962 1963 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 1964 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 1965 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 1966 1967 __printf(3, 4) 1968 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1969 1970 #define f2fs_err(sbi, fmt, ...) \ 1971 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1972 #define f2fs_warn(sbi, fmt, ...) \ 1973 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1974 #define f2fs_notice(sbi, fmt, ...) \ 1975 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1976 #define f2fs_info(sbi, fmt, ...) \ 1977 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1978 #define f2fs_debug(sbi, fmt, ...) \ 1979 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1980 1981 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1982 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1983 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1984 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1985 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1986 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1987 1988 #ifdef CONFIG_F2FS_FAULT_INJECTION 1989 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1990 __builtin_return_address(0)) 1991 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1992 const char *func, const char *parent_func) 1993 { 1994 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1995 1996 if (!ffi->inject_rate) 1997 return false; 1998 1999 if (!IS_FAULT_SET(ffi, type)) 2000 return false; 2001 2002 atomic_inc(&ffi->inject_ops); 2003 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 2004 atomic_set(&ffi->inject_ops, 0); 2005 ffi->inject_count[type]++; 2006 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 2007 f2fs_fault_name[type], func, parent_func); 2008 return true; 2009 } 2010 return false; 2011 } 2012 #else 2013 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 2014 { 2015 return false; 2016 } 2017 #endif 2018 2019 /* 2020 * Test if the mounted volume is a multi-device volume. 2021 * - For a single regular disk volume, sbi->s_ndevs is 0. 2022 * - For a single zoned disk volume, sbi->s_ndevs is 1. 2023 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 2024 */ 2025 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 2026 { 2027 return sbi->s_ndevs > 1; 2028 } 2029 2030 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 2031 { 2032 unsigned long now = jiffies; 2033 2034 sbi->last_time[type] = now; 2035 2036 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 2037 if (type == REQ_TIME) { 2038 sbi->last_time[DISCARD_TIME] = now; 2039 sbi->last_time[GC_TIME] = now; 2040 } 2041 } 2042 2043 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 2044 { 2045 unsigned long interval = sbi->interval_time[type] * HZ; 2046 2047 return time_after(jiffies, sbi->last_time[type] + interval); 2048 } 2049 2050 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 2051 int type) 2052 { 2053 unsigned long interval = sbi->interval_time[type] * HZ; 2054 unsigned int wait_ms = 0; 2055 long delta; 2056 2057 delta = (sbi->last_time[type] + interval) - jiffies; 2058 if (delta > 0) 2059 wait_ms = jiffies_to_msecs(delta); 2060 2061 return wait_ms; 2062 } 2063 2064 /* 2065 * Inline functions 2066 */ 2067 static inline u32 __f2fs_crc32(u32 crc, const void *address, 2068 unsigned int length) 2069 { 2070 return crc32(crc, address, length); 2071 } 2072 2073 static inline u32 f2fs_crc32(const void *address, unsigned int length) 2074 { 2075 return __f2fs_crc32(F2FS_SUPER_MAGIC, address, length); 2076 } 2077 2078 static inline u32 f2fs_chksum(u32 crc, const void *address, unsigned int length) 2079 { 2080 return __f2fs_crc32(crc, address, length); 2081 } 2082 2083 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 2084 { 2085 return container_of(inode, struct f2fs_inode_info, vfs_inode); 2086 } 2087 2088 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2089 { 2090 return sb->s_fs_info; 2091 } 2092 2093 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2094 { 2095 return F2FS_SB(inode->i_sb); 2096 } 2097 2098 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2099 { 2100 return F2FS_I_SB(mapping->host); 2101 } 2102 2103 static inline struct f2fs_sb_info *F2FS_F_SB(const struct folio *folio) 2104 { 2105 return F2FS_M_SB(folio->mapping); 2106 } 2107 2108 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2109 { 2110 return (struct f2fs_super_block *)(sbi->raw_super); 2111 } 2112 2113 static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2114 pgoff_t index) 2115 { 2116 pgoff_t idx_in_folio = index % folio_nr_pages(folio); 2117 2118 return (struct f2fs_super_block *) 2119 (page_address(folio_page(folio, idx_in_folio)) + 2120 F2FS_SUPER_OFFSET); 2121 } 2122 2123 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2124 { 2125 return (struct f2fs_checkpoint *)(sbi->ckpt); 2126 } 2127 2128 static inline struct f2fs_node *F2FS_NODE(const struct folio *folio) 2129 { 2130 return (struct f2fs_node *)folio_address(folio); 2131 } 2132 2133 static inline struct f2fs_inode *F2FS_INODE(const struct folio *folio) 2134 { 2135 return &((struct f2fs_node *)folio_address(folio))->i; 2136 } 2137 2138 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2139 { 2140 return (struct f2fs_nm_info *)(sbi->nm_info); 2141 } 2142 2143 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2144 { 2145 return (struct f2fs_sm_info *)(sbi->sm_info); 2146 } 2147 2148 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2149 { 2150 return (struct sit_info *)(SM_I(sbi)->sit_info); 2151 } 2152 2153 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2154 { 2155 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2156 } 2157 2158 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2159 { 2160 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2161 } 2162 2163 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2164 { 2165 return sbi->meta_inode->i_mapping; 2166 } 2167 2168 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2169 { 2170 return sbi->node_inode->i_mapping; 2171 } 2172 2173 static inline bool is_meta_folio(struct folio *folio) 2174 { 2175 return folio->mapping == META_MAPPING(F2FS_F_SB(folio)); 2176 } 2177 2178 static inline bool is_node_folio(struct folio *folio) 2179 { 2180 return folio->mapping == NODE_MAPPING(F2FS_F_SB(folio)); 2181 } 2182 2183 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2184 { 2185 return test_bit(type, &sbi->s_flag); 2186 } 2187 2188 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2189 { 2190 set_bit(type, &sbi->s_flag); 2191 } 2192 2193 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2194 { 2195 clear_bit(type, &sbi->s_flag); 2196 } 2197 2198 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2199 { 2200 return le64_to_cpu(cp->checkpoint_ver); 2201 } 2202 2203 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2204 { 2205 if (type < F2FS_MAX_QUOTAS) 2206 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2207 return 0; 2208 } 2209 2210 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2211 { 2212 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2213 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2214 } 2215 2216 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2217 { 2218 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2219 2220 return ckpt_flags & f; 2221 } 2222 2223 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2224 { 2225 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2226 } 2227 2228 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2229 { 2230 unsigned int ckpt_flags; 2231 2232 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2233 ckpt_flags |= f; 2234 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2235 } 2236 2237 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2238 { 2239 unsigned long flags; 2240 2241 spin_lock_irqsave(&sbi->cp_lock, flags); 2242 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2243 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2244 } 2245 2246 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2247 { 2248 unsigned int ckpt_flags; 2249 2250 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2251 ckpt_flags &= (~f); 2252 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2253 } 2254 2255 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2256 { 2257 unsigned long flags; 2258 2259 spin_lock_irqsave(&sbi->cp_lock, flags); 2260 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2261 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2262 } 2263 2264 #define init_f2fs_rwsem(sem) \ 2265 do { \ 2266 static struct lock_class_key __key; \ 2267 \ 2268 __init_f2fs_rwsem((sem), #sem, &__key); \ 2269 } while (0) 2270 2271 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2272 const char *sem_name, struct lock_class_key *key) 2273 { 2274 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2275 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2276 init_waitqueue_head(&sem->read_waiters); 2277 #endif 2278 } 2279 2280 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2281 { 2282 return rwsem_is_locked(&sem->internal_rwsem); 2283 } 2284 2285 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2286 { 2287 return rwsem_is_contended(&sem->internal_rwsem); 2288 } 2289 2290 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2291 { 2292 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2293 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2294 #else 2295 down_read(&sem->internal_rwsem); 2296 #endif 2297 } 2298 2299 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2300 { 2301 return down_read_trylock(&sem->internal_rwsem); 2302 } 2303 2304 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2305 { 2306 up_read(&sem->internal_rwsem); 2307 } 2308 2309 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2310 { 2311 down_write(&sem->internal_rwsem); 2312 } 2313 2314 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2315 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2316 { 2317 down_read_nested(&sem->internal_rwsem, subclass); 2318 } 2319 2320 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2321 { 2322 down_write_nested(&sem->internal_rwsem, subclass); 2323 } 2324 #else 2325 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2326 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2327 #endif 2328 2329 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2330 { 2331 return down_write_trylock(&sem->internal_rwsem); 2332 } 2333 2334 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2335 { 2336 up_write(&sem->internal_rwsem); 2337 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2338 wake_up_all(&sem->read_waiters); 2339 #endif 2340 } 2341 2342 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2343 { 2344 unsigned long flags; 2345 unsigned char *nat_bits; 2346 2347 /* 2348 * In order to re-enable nat_bits we need to call fsck.f2fs by 2349 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2350 * so let's rely on regular fsck or unclean shutdown. 2351 */ 2352 2353 if (lock) 2354 spin_lock_irqsave(&sbi->cp_lock, flags); 2355 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2356 nat_bits = NM_I(sbi)->nat_bits; 2357 NM_I(sbi)->nat_bits = NULL; 2358 if (lock) 2359 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2360 2361 kvfree(nat_bits); 2362 } 2363 2364 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2365 struct cp_control *cpc) 2366 { 2367 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2368 2369 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2370 } 2371 2372 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2373 { 2374 f2fs_down_read(&sbi->cp_rwsem); 2375 } 2376 2377 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2378 { 2379 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2380 return 0; 2381 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2382 } 2383 2384 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2385 { 2386 f2fs_up_read(&sbi->cp_rwsem); 2387 } 2388 2389 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2390 { 2391 f2fs_down_write(&sbi->cp_rwsem); 2392 } 2393 2394 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2395 { 2396 f2fs_up_write(&sbi->cp_rwsem); 2397 } 2398 2399 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2400 { 2401 int reason = CP_SYNC; 2402 2403 if (test_opt(sbi, FASTBOOT)) 2404 reason = CP_FASTBOOT; 2405 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2406 reason = CP_UMOUNT; 2407 return reason; 2408 } 2409 2410 static inline bool __remain_node_summaries(int reason) 2411 { 2412 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2413 } 2414 2415 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2416 { 2417 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2418 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2419 } 2420 2421 /* 2422 * Check whether the inode has blocks or not 2423 */ 2424 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2425 { 2426 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2427 2428 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2429 } 2430 2431 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2432 { 2433 return ofs == XATTR_NODE_OFFSET; 2434 } 2435 2436 static inline bool __allow_reserved_root(struct f2fs_sb_info *sbi, 2437 struct inode *inode, bool cap) 2438 { 2439 if (!inode) 2440 return true; 2441 if (IS_NOQUOTA(inode)) 2442 return true; 2443 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2444 return true; 2445 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2446 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2447 return true; 2448 if (cap && capable(CAP_SYS_RESOURCE)) 2449 return true; 2450 return false; 2451 } 2452 2453 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2454 struct inode *inode, bool cap) 2455 { 2456 block_t avail_user_block_count; 2457 2458 avail_user_block_count = sbi->user_block_count - 2459 sbi->current_reserved_blocks; 2460 2461 if (test_opt(sbi, RESERVE_ROOT) && !__allow_reserved_root(sbi, inode, cap)) 2462 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2463 2464 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2465 if (avail_user_block_count > sbi->unusable_block_count) 2466 avail_user_block_count -= sbi->unusable_block_count; 2467 else 2468 avail_user_block_count = 0; 2469 } 2470 2471 return avail_user_block_count; 2472 } 2473 2474 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2475 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2476 struct inode *inode, blkcnt_t *count, bool partial) 2477 { 2478 long long diff = 0, release = 0; 2479 block_t avail_user_block_count; 2480 int ret; 2481 2482 ret = dquot_reserve_block(inode, *count); 2483 if (ret) 2484 return ret; 2485 2486 if (time_to_inject(sbi, FAULT_BLOCK)) { 2487 release = *count; 2488 goto release_quota; 2489 } 2490 2491 /* 2492 * let's increase this in prior to actual block count change in order 2493 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2494 */ 2495 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2496 2497 spin_lock(&sbi->stat_lock); 2498 2499 avail_user_block_count = get_available_block_count(sbi, inode, true); 2500 diff = (long long)sbi->total_valid_block_count + *count - 2501 avail_user_block_count; 2502 if (unlikely(diff > 0)) { 2503 if (!partial) { 2504 spin_unlock(&sbi->stat_lock); 2505 release = *count; 2506 goto enospc; 2507 } 2508 if (diff > *count) 2509 diff = *count; 2510 *count -= diff; 2511 release = diff; 2512 if (!*count) { 2513 spin_unlock(&sbi->stat_lock); 2514 goto enospc; 2515 } 2516 } 2517 sbi->total_valid_block_count += (block_t)(*count); 2518 2519 spin_unlock(&sbi->stat_lock); 2520 2521 if (unlikely(release)) { 2522 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2523 dquot_release_reservation_block(inode, release); 2524 } 2525 f2fs_i_blocks_write(inode, *count, true, true); 2526 return 0; 2527 2528 enospc: 2529 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2530 release_quota: 2531 dquot_release_reservation_block(inode, release); 2532 return -ENOSPC; 2533 } 2534 2535 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2536 static inline bool folio_test_f2fs_##name(const struct folio *folio) \ 2537 { \ 2538 unsigned long priv = (unsigned long)folio->private; \ 2539 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2540 (1UL << PAGE_PRIVATE_##flagname); \ 2541 return (priv & v) == v; \ 2542 } \ 2543 static inline bool page_private_##name(struct page *page) \ 2544 { \ 2545 return PagePrivate(page) && \ 2546 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2547 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2548 } 2549 2550 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2551 static inline void folio_set_f2fs_##name(struct folio *folio) \ 2552 { \ 2553 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2554 (1UL << PAGE_PRIVATE_##flagname); \ 2555 if (!folio->private) \ 2556 folio_attach_private(folio, (void *)v); \ 2557 else { \ 2558 v |= (unsigned long)folio->private; \ 2559 folio->private = (void *)v; \ 2560 } \ 2561 } \ 2562 static inline void set_page_private_##name(struct page *page) \ 2563 { \ 2564 if (!PagePrivate(page)) \ 2565 attach_page_private(page, (void *)0); \ 2566 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2567 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2568 } 2569 2570 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2571 static inline void folio_clear_f2fs_##name(struct folio *folio) \ 2572 { \ 2573 unsigned long v = (unsigned long)folio->private; \ 2574 \ 2575 v &= ~(1UL << PAGE_PRIVATE_##flagname); \ 2576 if (v == (1UL << PAGE_PRIVATE_NOT_POINTER)) \ 2577 folio_detach_private(folio); \ 2578 else \ 2579 folio->private = (void *)v; \ 2580 } \ 2581 static inline void clear_page_private_##name(struct page *page) \ 2582 { \ 2583 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2584 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2585 detach_page_private(page); \ 2586 } 2587 2588 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2589 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2590 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2591 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2592 2593 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2594 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2595 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2596 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2597 2598 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2599 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2600 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2601 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2602 2603 static inline unsigned long folio_get_f2fs_data(struct folio *folio) 2604 { 2605 unsigned long data = (unsigned long)folio->private; 2606 2607 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2608 return 0; 2609 return data >> PAGE_PRIVATE_MAX; 2610 } 2611 2612 static inline void folio_set_f2fs_data(struct folio *folio, unsigned long data) 2613 { 2614 data = (1UL << PAGE_PRIVATE_NOT_POINTER) | (data << PAGE_PRIVATE_MAX); 2615 2616 if (!folio_test_private(folio)) 2617 folio_attach_private(folio, (void *)data); 2618 else 2619 folio->private = (void *)((unsigned long)folio->private | data); 2620 } 2621 2622 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2623 struct inode *inode, 2624 block_t count) 2625 { 2626 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2627 2628 spin_lock(&sbi->stat_lock); 2629 if (unlikely(sbi->total_valid_block_count < count)) { 2630 f2fs_warn(sbi, "Inconsistent total_valid_block_count:%u, ino:%lu, count:%u", 2631 sbi->total_valid_block_count, inode->i_ino, count); 2632 sbi->total_valid_block_count = 0; 2633 set_sbi_flag(sbi, SBI_NEED_FSCK); 2634 } else { 2635 sbi->total_valid_block_count -= count; 2636 } 2637 if (sbi->reserved_blocks && 2638 sbi->current_reserved_blocks < sbi->reserved_blocks) 2639 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2640 sbi->current_reserved_blocks + count); 2641 spin_unlock(&sbi->stat_lock); 2642 if (unlikely(inode->i_blocks < sectors)) { 2643 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2644 inode->i_ino, 2645 (unsigned long long)inode->i_blocks, 2646 (unsigned long long)sectors); 2647 set_sbi_flag(sbi, SBI_NEED_FSCK); 2648 return; 2649 } 2650 f2fs_i_blocks_write(inode, count, false, true); 2651 } 2652 2653 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2654 { 2655 atomic_inc(&sbi->nr_pages[count_type]); 2656 2657 if (count_type == F2FS_DIRTY_DENTS || 2658 count_type == F2FS_DIRTY_NODES || 2659 count_type == F2FS_DIRTY_META || 2660 count_type == F2FS_DIRTY_QDATA || 2661 count_type == F2FS_DIRTY_IMETA) 2662 set_sbi_flag(sbi, SBI_IS_DIRTY); 2663 } 2664 2665 static inline void inode_inc_dirty_pages(struct inode *inode) 2666 { 2667 atomic_inc(&F2FS_I(inode)->dirty_pages); 2668 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2669 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2670 if (IS_NOQUOTA(inode)) 2671 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2672 } 2673 2674 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2675 { 2676 atomic_dec(&sbi->nr_pages[count_type]); 2677 } 2678 2679 static inline void inode_dec_dirty_pages(struct inode *inode) 2680 { 2681 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2682 !S_ISLNK(inode->i_mode)) 2683 return; 2684 2685 atomic_dec(&F2FS_I(inode)->dirty_pages); 2686 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2687 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2688 if (IS_NOQUOTA(inode)) 2689 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2690 } 2691 2692 static inline void inc_atomic_write_cnt(struct inode *inode) 2693 { 2694 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2695 struct f2fs_inode_info *fi = F2FS_I(inode); 2696 u64 current_write; 2697 2698 fi->atomic_write_cnt++; 2699 atomic64_inc(&sbi->current_atomic_write); 2700 current_write = atomic64_read(&sbi->current_atomic_write); 2701 if (current_write > sbi->peak_atomic_write) 2702 sbi->peak_atomic_write = current_write; 2703 } 2704 2705 static inline void release_atomic_write_cnt(struct inode *inode) 2706 { 2707 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2708 struct f2fs_inode_info *fi = F2FS_I(inode); 2709 2710 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2711 fi->atomic_write_cnt = 0; 2712 } 2713 2714 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2715 { 2716 return atomic_read(&sbi->nr_pages[count_type]); 2717 } 2718 2719 static inline int get_dirty_pages(struct inode *inode) 2720 { 2721 return atomic_read(&F2FS_I(inode)->dirty_pages); 2722 } 2723 2724 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2725 { 2726 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2727 BLKS_PER_SEC(sbi)); 2728 } 2729 2730 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2731 { 2732 return sbi->total_valid_block_count; 2733 } 2734 2735 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2736 { 2737 return sbi->discard_blks; 2738 } 2739 2740 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2741 { 2742 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2743 2744 /* return NAT or SIT bitmap */ 2745 if (flag == NAT_BITMAP) 2746 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2747 else if (flag == SIT_BITMAP) 2748 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2749 2750 return 0; 2751 } 2752 2753 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2754 { 2755 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2756 } 2757 2758 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2759 { 2760 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2761 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2762 int offset; 2763 2764 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2765 offset = (flag == SIT_BITMAP) ? 2766 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2767 /* 2768 * if large_nat_bitmap feature is enabled, leave checksum 2769 * protection for all nat/sit bitmaps. 2770 */ 2771 return tmp_ptr + offset + sizeof(__le32); 2772 } 2773 2774 if (__cp_payload(sbi) > 0) { 2775 if (flag == NAT_BITMAP) 2776 return tmp_ptr; 2777 else 2778 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2779 } else { 2780 offset = (flag == NAT_BITMAP) ? 2781 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2782 return tmp_ptr + offset; 2783 } 2784 } 2785 2786 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2787 { 2788 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2789 2790 if (sbi->cur_cp_pack == 2) 2791 start_addr += BLKS_PER_SEG(sbi); 2792 return start_addr; 2793 } 2794 2795 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2796 { 2797 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2798 2799 if (sbi->cur_cp_pack == 1) 2800 start_addr += BLKS_PER_SEG(sbi); 2801 return start_addr; 2802 } 2803 2804 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2805 { 2806 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2807 } 2808 2809 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2810 { 2811 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2812 } 2813 2814 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2815 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2816 struct inode *inode, bool is_inode) 2817 { 2818 block_t valid_block_count; 2819 unsigned int valid_node_count, avail_user_node_count; 2820 unsigned int avail_user_block_count; 2821 int err; 2822 2823 if (is_inode) { 2824 if (inode) { 2825 err = dquot_alloc_inode(inode); 2826 if (err) 2827 return err; 2828 } 2829 } else { 2830 err = dquot_reserve_block(inode, 1); 2831 if (err) 2832 return err; 2833 } 2834 2835 if (time_to_inject(sbi, FAULT_BLOCK)) 2836 goto enospc; 2837 2838 spin_lock(&sbi->stat_lock); 2839 2840 valid_block_count = sbi->total_valid_block_count + 1; 2841 avail_user_block_count = get_available_block_count(sbi, inode, 2842 test_opt(sbi, RESERVE_NODE)); 2843 2844 if (unlikely(valid_block_count > avail_user_block_count)) { 2845 spin_unlock(&sbi->stat_lock); 2846 goto enospc; 2847 } 2848 2849 avail_user_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM; 2850 if (test_opt(sbi, RESERVE_NODE) && 2851 !__allow_reserved_root(sbi, inode, true)) 2852 avail_user_node_count -= F2FS_OPTION(sbi).root_reserved_nodes; 2853 valid_node_count = sbi->total_valid_node_count + 1; 2854 if (unlikely(valid_node_count > avail_user_node_count)) { 2855 spin_unlock(&sbi->stat_lock); 2856 goto enospc; 2857 } 2858 2859 sbi->total_valid_node_count++; 2860 sbi->total_valid_block_count++; 2861 spin_unlock(&sbi->stat_lock); 2862 2863 if (inode) { 2864 if (is_inode) 2865 f2fs_mark_inode_dirty_sync(inode, true); 2866 else 2867 f2fs_i_blocks_write(inode, 1, true, true); 2868 } 2869 2870 percpu_counter_inc(&sbi->alloc_valid_block_count); 2871 return 0; 2872 2873 enospc: 2874 if (is_inode) { 2875 if (inode) 2876 dquot_free_inode(inode); 2877 } else { 2878 dquot_release_reservation_block(inode, 1); 2879 } 2880 return -ENOSPC; 2881 } 2882 2883 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2884 struct inode *inode, bool is_inode) 2885 { 2886 spin_lock(&sbi->stat_lock); 2887 2888 if (unlikely(!sbi->total_valid_block_count || 2889 !sbi->total_valid_node_count)) { 2890 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2891 sbi->total_valid_block_count, 2892 sbi->total_valid_node_count); 2893 set_sbi_flag(sbi, SBI_NEED_FSCK); 2894 } else { 2895 sbi->total_valid_block_count--; 2896 sbi->total_valid_node_count--; 2897 } 2898 2899 if (sbi->reserved_blocks && 2900 sbi->current_reserved_blocks < sbi->reserved_blocks) 2901 sbi->current_reserved_blocks++; 2902 2903 spin_unlock(&sbi->stat_lock); 2904 2905 if (is_inode) { 2906 dquot_free_inode(inode); 2907 } else { 2908 if (unlikely(inode->i_blocks == 0)) { 2909 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2910 inode->i_ino, 2911 (unsigned long long)inode->i_blocks); 2912 set_sbi_flag(sbi, SBI_NEED_FSCK); 2913 return; 2914 } 2915 f2fs_i_blocks_write(inode, 1, false, true); 2916 } 2917 } 2918 2919 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2920 { 2921 return sbi->total_valid_node_count; 2922 } 2923 2924 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2925 { 2926 percpu_counter_inc(&sbi->total_valid_inode_count); 2927 } 2928 2929 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2930 { 2931 percpu_counter_dec(&sbi->total_valid_inode_count); 2932 } 2933 2934 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2935 { 2936 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2937 } 2938 2939 static inline struct folio *f2fs_grab_cache_folio(struct address_space *mapping, 2940 pgoff_t index, bool for_write) 2941 { 2942 struct folio *folio; 2943 unsigned int flags; 2944 2945 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2946 fgf_t fgf_flags; 2947 2948 if (!for_write) 2949 fgf_flags = FGP_LOCK | FGP_ACCESSED; 2950 else 2951 fgf_flags = FGP_LOCK; 2952 folio = __filemap_get_folio(mapping, index, fgf_flags, 0); 2953 if (!IS_ERR(folio)) 2954 return folio; 2955 2956 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2957 return ERR_PTR(-ENOMEM); 2958 } 2959 2960 if (!for_write) 2961 return filemap_grab_folio(mapping, index); 2962 2963 flags = memalloc_nofs_save(); 2964 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN, 2965 mapping_gfp_mask(mapping)); 2966 memalloc_nofs_restore(flags); 2967 2968 return folio; 2969 } 2970 2971 static inline struct folio *f2fs_filemap_get_folio( 2972 struct address_space *mapping, pgoff_t index, 2973 fgf_t fgp_flags, gfp_t gfp_mask) 2974 { 2975 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2976 return ERR_PTR(-ENOMEM); 2977 2978 return __filemap_get_folio(mapping, index, fgp_flags, gfp_mask); 2979 } 2980 2981 static inline void f2fs_folio_put(struct folio *folio, bool unlock) 2982 { 2983 if (IS_ERR_OR_NULL(folio)) 2984 return; 2985 2986 if (unlock) { 2987 f2fs_bug_on(F2FS_F_SB(folio), !folio_test_locked(folio)); 2988 folio_unlock(folio); 2989 } 2990 folio_put(folio); 2991 } 2992 2993 static inline void f2fs_put_page(struct page *page, bool unlock) 2994 { 2995 if (!page) 2996 return; 2997 f2fs_folio_put(page_folio(page), unlock); 2998 } 2999 3000 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 3001 { 3002 if (dn->node_folio) 3003 f2fs_folio_put(dn->node_folio, true); 3004 if (dn->inode_folio && dn->node_folio != dn->inode_folio) 3005 f2fs_folio_put(dn->inode_folio, false); 3006 dn->node_folio = NULL; 3007 dn->inode_folio = NULL; 3008 } 3009 3010 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 3011 size_t size) 3012 { 3013 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 3014 } 3015 3016 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 3017 gfp_t flags) 3018 { 3019 void *entry; 3020 3021 entry = kmem_cache_alloc(cachep, flags); 3022 if (!entry) 3023 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 3024 return entry; 3025 } 3026 3027 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 3028 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 3029 { 3030 if (nofail) 3031 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 3032 3033 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 3034 return NULL; 3035 3036 return kmem_cache_alloc(cachep, flags); 3037 } 3038 3039 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 3040 { 3041 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 3042 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 3043 get_pages(sbi, F2FS_WB_CP_DATA) || 3044 get_pages(sbi, F2FS_DIO_READ) || 3045 get_pages(sbi, F2FS_DIO_WRITE)) 3046 return true; 3047 3048 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 3049 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 3050 return true; 3051 3052 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 3053 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 3054 return true; 3055 return false; 3056 } 3057 3058 static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 3059 { 3060 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ); 3061 } 3062 3063 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 3064 { 3065 bool zoned_gc = (type == GC_TIME && 3066 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 3067 3068 if (sbi->gc_mode == GC_URGENT_HIGH) 3069 return true; 3070 3071 if (sbi->bggc_io_aware == AWARE_READ_IO && is_inflight_read_io(sbi)) 3072 return false; 3073 if (sbi->bggc_io_aware == AWARE_ALL_IO && is_inflight_io(sbi, type)) 3074 return false; 3075 3076 if (sbi->gc_mode == GC_URGENT_MID) 3077 return true; 3078 3079 if (sbi->gc_mode == GC_URGENT_LOW && 3080 (type == DISCARD_TIME || type == GC_TIME)) 3081 return true; 3082 3083 if (zoned_gc) 3084 return true; 3085 3086 return f2fs_time_over(sbi, type); 3087 } 3088 3089 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 3090 unsigned long index, void *item) 3091 { 3092 while (radix_tree_insert(root, index, item)) 3093 cond_resched(); 3094 } 3095 3096 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 3097 3098 static inline bool IS_INODE(const struct folio *folio) 3099 { 3100 struct f2fs_node *p = F2FS_NODE(folio); 3101 3102 return RAW_IS_INODE(p); 3103 } 3104 3105 static inline int offset_in_addr(struct f2fs_inode *i) 3106 { 3107 return (i->i_inline & F2FS_EXTRA_ATTR) ? 3108 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 3109 } 3110 3111 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 3112 { 3113 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 3114 } 3115 3116 static inline int f2fs_has_extra_attr(struct inode *inode); 3117 static inline unsigned int get_dnode_base(struct inode *inode, 3118 struct folio *node_folio) 3119 { 3120 if (!IS_INODE(node_folio)) 3121 return 0; 3122 3123 return inode ? get_extra_isize(inode) : 3124 offset_in_addr(&F2FS_NODE(node_folio)->i); 3125 } 3126 3127 static inline __le32 *get_dnode_addr(struct inode *inode, 3128 struct folio *node_folio) 3129 { 3130 return blkaddr_in_node(F2FS_NODE(node_folio)) + 3131 get_dnode_base(inode, node_folio); 3132 } 3133 3134 static inline block_t data_blkaddr(struct inode *inode, 3135 struct folio *node_folio, unsigned int offset) 3136 { 3137 return le32_to_cpu(*(get_dnode_addr(inode, node_folio) + offset)); 3138 } 3139 3140 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 3141 { 3142 return data_blkaddr(dn->inode, dn->node_folio, dn->ofs_in_node); 3143 } 3144 3145 static inline int f2fs_test_bit(unsigned int nr, char *addr) 3146 { 3147 int mask; 3148 3149 addr += (nr >> 3); 3150 mask = BIT(7 - (nr & 0x07)); 3151 return mask & *addr; 3152 } 3153 3154 static inline void f2fs_set_bit(unsigned int nr, char *addr) 3155 { 3156 int mask; 3157 3158 addr += (nr >> 3); 3159 mask = BIT(7 - (nr & 0x07)); 3160 *addr |= mask; 3161 } 3162 3163 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 3164 { 3165 int mask; 3166 3167 addr += (nr >> 3); 3168 mask = BIT(7 - (nr & 0x07)); 3169 *addr &= ~mask; 3170 } 3171 3172 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3173 { 3174 int mask; 3175 int ret; 3176 3177 addr += (nr >> 3); 3178 mask = BIT(7 - (nr & 0x07)); 3179 ret = mask & *addr; 3180 *addr |= mask; 3181 return ret; 3182 } 3183 3184 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3185 { 3186 int mask; 3187 int ret; 3188 3189 addr += (nr >> 3); 3190 mask = BIT(7 - (nr & 0x07)); 3191 ret = mask & *addr; 3192 *addr &= ~mask; 3193 return ret; 3194 } 3195 3196 static inline void f2fs_change_bit(unsigned int nr, char *addr) 3197 { 3198 int mask; 3199 3200 addr += (nr >> 3); 3201 mask = BIT(7 - (nr & 0x07)); 3202 *addr ^= mask; 3203 } 3204 3205 /* 3206 * On-disk inode flags (f2fs_inode::i_flags) 3207 */ 3208 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3209 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3210 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3211 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3212 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3213 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3214 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3215 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3216 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3217 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3218 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3219 #define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3220 3221 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3222 3223 /* Flags that should be inherited by new inodes from their parent. */ 3224 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3225 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3226 F2FS_CASEFOLD_FL) 3227 3228 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3229 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3230 F2FS_CASEFOLD_FL)) 3231 3232 /* Flags that are appropriate for non-directories/regular files. */ 3233 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3234 3235 #define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3236 3237 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3238 { 3239 if (S_ISDIR(mode)) 3240 return flags; 3241 else if (S_ISREG(mode)) 3242 return flags & F2FS_REG_FLMASK; 3243 else 3244 return flags & F2FS_OTHER_FLMASK; 3245 } 3246 3247 static inline void __mark_inode_dirty_flag(struct inode *inode, 3248 int flag, bool set) 3249 { 3250 switch (flag) { 3251 case FI_INLINE_XATTR: 3252 case FI_INLINE_DATA: 3253 case FI_INLINE_DENTRY: 3254 case FI_NEW_INODE: 3255 if (set) 3256 return; 3257 fallthrough; 3258 case FI_DATA_EXIST: 3259 case FI_PIN_FILE: 3260 case FI_COMPRESS_RELEASED: 3261 f2fs_mark_inode_dirty_sync(inode, true); 3262 } 3263 } 3264 3265 static inline void set_inode_flag(struct inode *inode, int flag) 3266 { 3267 set_bit(flag, F2FS_I(inode)->flags); 3268 __mark_inode_dirty_flag(inode, flag, true); 3269 } 3270 3271 static inline int is_inode_flag_set(struct inode *inode, int flag) 3272 { 3273 return test_bit(flag, F2FS_I(inode)->flags); 3274 } 3275 3276 static inline void clear_inode_flag(struct inode *inode, int flag) 3277 { 3278 clear_bit(flag, F2FS_I(inode)->flags); 3279 __mark_inode_dirty_flag(inode, flag, false); 3280 } 3281 3282 static inline bool f2fs_verity_in_progress(struct inode *inode) 3283 { 3284 return IS_ENABLED(CONFIG_FS_VERITY) && 3285 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3286 } 3287 3288 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3289 { 3290 F2FS_I(inode)->i_acl_mode = mode; 3291 set_inode_flag(inode, FI_ACL_MODE); 3292 f2fs_mark_inode_dirty_sync(inode, false); 3293 } 3294 3295 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3296 { 3297 if (inc) 3298 inc_nlink(inode); 3299 else 3300 drop_nlink(inode); 3301 f2fs_mark_inode_dirty_sync(inode, true); 3302 } 3303 3304 static inline void f2fs_i_blocks_write(struct inode *inode, 3305 block_t diff, bool add, bool claim) 3306 { 3307 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3308 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3309 3310 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3311 if (add) { 3312 if (claim) 3313 dquot_claim_block(inode, diff); 3314 else 3315 dquot_alloc_block_nofail(inode, diff); 3316 } else { 3317 dquot_free_block(inode, diff); 3318 } 3319 3320 f2fs_mark_inode_dirty_sync(inode, true); 3321 if (clean || recover) 3322 set_inode_flag(inode, FI_AUTO_RECOVER); 3323 } 3324 3325 static inline bool f2fs_is_atomic_file(struct inode *inode); 3326 3327 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3328 { 3329 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3330 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3331 3332 if (i_size_read(inode) == i_size) 3333 return; 3334 3335 i_size_write(inode, i_size); 3336 3337 if (f2fs_is_atomic_file(inode)) 3338 return; 3339 3340 f2fs_mark_inode_dirty_sync(inode, true); 3341 if (clean || recover) 3342 set_inode_flag(inode, FI_AUTO_RECOVER); 3343 } 3344 3345 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3346 { 3347 F2FS_I(inode)->i_current_depth = depth; 3348 f2fs_mark_inode_dirty_sync(inode, true); 3349 } 3350 3351 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3352 unsigned int count) 3353 { 3354 F2FS_I(inode)->i_gc_failures = count; 3355 f2fs_mark_inode_dirty_sync(inode, true); 3356 } 3357 3358 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3359 { 3360 F2FS_I(inode)->i_xattr_nid = xnid; 3361 f2fs_mark_inode_dirty_sync(inode, true); 3362 } 3363 3364 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3365 { 3366 F2FS_I(inode)->i_pino = pino; 3367 f2fs_mark_inode_dirty_sync(inode, true); 3368 } 3369 3370 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3371 { 3372 struct f2fs_inode_info *fi = F2FS_I(inode); 3373 3374 if (ri->i_inline & F2FS_INLINE_XATTR) 3375 set_bit(FI_INLINE_XATTR, fi->flags); 3376 if (ri->i_inline & F2FS_INLINE_DATA) 3377 set_bit(FI_INLINE_DATA, fi->flags); 3378 if (ri->i_inline & F2FS_INLINE_DENTRY) 3379 set_bit(FI_INLINE_DENTRY, fi->flags); 3380 if (ri->i_inline & F2FS_DATA_EXIST) 3381 set_bit(FI_DATA_EXIST, fi->flags); 3382 if (ri->i_inline & F2FS_EXTRA_ATTR) 3383 set_bit(FI_EXTRA_ATTR, fi->flags); 3384 if (ri->i_inline & F2FS_PIN_FILE) 3385 set_bit(FI_PIN_FILE, fi->flags); 3386 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3387 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3388 } 3389 3390 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3391 { 3392 ri->i_inline = 0; 3393 3394 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3395 ri->i_inline |= F2FS_INLINE_XATTR; 3396 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3397 ri->i_inline |= F2FS_INLINE_DATA; 3398 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3399 ri->i_inline |= F2FS_INLINE_DENTRY; 3400 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3401 ri->i_inline |= F2FS_DATA_EXIST; 3402 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3403 ri->i_inline |= F2FS_EXTRA_ATTR; 3404 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3405 ri->i_inline |= F2FS_PIN_FILE; 3406 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3407 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3408 } 3409 3410 static inline int f2fs_has_extra_attr(struct inode *inode) 3411 { 3412 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3413 } 3414 3415 static inline int f2fs_has_inline_xattr(struct inode *inode) 3416 { 3417 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3418 } 3419 3420 static inline int f2fs_compressed_file(struct inode *inode) 3421 { 3422 return S_ISREG(inode->i_mode) && 3423 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3424 } 3425 3426 static inline bool f2fs_need_compress_data(struct inode *inode) 3427 { 3428 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3429 3430 if (!f2fs_compressed_file(inode)) 3431 return false; 3432 3433 if (compress_mode == COMPR_MODE_FS) 3434 return true; 3435 else if (compress_mode == COMPR_MODE_USER && 3436 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3437 return true; 3438 3439 return false; 3440 } 3441 3442 static inline unsigned int addrs_per_page(struct inode *inode, 3443 bool is_inode) 3444 { 3445 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) - 3446 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK; 3447 3448 if (f2fs_compressed_file(inode)) 3449 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3450 return addrs; 3451 } 3452 3453 static inline 3454 void *inline_xattr_addr(struct inode *inode, const struct folio *folio) 3455 { 3456 struct f2fs_inode *ri = F2FS_INODE(folio); 3457 3458 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3459 get_inline_xattr_addrs(inode)]); 3460 } 3461 3462 static inline int inline_xattr_size(struct inode *inode) 3463 { 3464 if (f2fs_has_inline_xattr(inode)) 3465 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3466 return 0; 3467 } 3468 3469 /* 3470 * Notice: check inline_data flag without inode page lock is unsafe. 3471 * It could change at any time by f2fs_convert_inline_folio(). 3472 */ 3473 static inline int f2fs_has_inline_data(struct inode *inode) 3474 { 3475 return is_inode_flag_set(inode, FI_INLINE_DATA); 3476 } 3477 3478 static inline int f2fs_exist_data(struct inode *inode) 3479 { 3480 return is_inode_flag_set(inode, FI_DATA_EXIST); 3481 } 3482 3483 static inline int f2fs_is_mmap_file(struct inode *inode) 3484 { 3485 return is_inode_flag_set(inode, FI_MMAP_FILE); 3486 } 3487 3488 static inline bool f2fs_is_pinned_file(struct inode *inode) 3489 { 3490 return is_inode_flag_set(inode, FI_PIN_FILE); 3491 } 3492 3493 static inline bool f2fs_is_atomic_file(struct inode *inode) 3494 { 3495 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3496 } 3497 3498 static inline bool f2fs_is_cow_file(struct inode *inode) 3499 { 3500 return is_inode_flag_set(inode, FI_COW_FILE); 3501 } 3502 3503 static inline void *inline_data_addr(struct inode *inode, struct folio *folio) 3504 { 3505 __le32 *addr = get_dnode_addr(inode, folio); 3506 3507 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3508 } 3509 3510 static inline int f2fs_has_inline_dentry(struct inode *inode) 3511 { 3512 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3513 } 3514 3515 static inline int is_file(struct inode *inode, int type) 3516 { 3517 return F2FS_I(inode)->i_advise & type; 3518 } 3519 3520 static inline void set_file(struct inode *inode, int type) 3521 { 3522 if (is_file(inode, type)) 3523 return; 3524 F2FS_I(inode)->i_advise |= type; 3525 f2fs_mark_inode_dirty_sync(inode, true); 3526 } 3527 3528 static inline void clear_file(struct inode *inode, int type) 3529 { 3530 if (!is_file(inode, type)) 3531 return; 3532 F2FS_I(inode)->i_advise &= ~type; 3533 f2fs_mark_inode_dirty_sync(inode, true); 3534 } 3535 3536 static inline bool f2fs_is_time_consistent(struct inode *inode) 3537 { 3538 struct timespec64 ts = inode_get_atime(inode); 3539 3540 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3541 return false; 3542 ts = inode_get_ctime(inode); 3543 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3544 return false; 3545 ts = inode_get_mtime(inode); 3546 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3547 return false; 3548 return true; 3549 } 3550 3551 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3552 { 3553 bool ret; 3554 3555 if (dsync) { 3556 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3557 3558 spin_lock(&sbi->inode_lock[DIRTY_META]); 3559 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3560 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3561 return ret; 3562 } 3563 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3564 file_keep_isize(inode) || 3565 i_size_read(inode) & ~PAGE_MASK) 3566 return false; 3567 3568 if (!f2fs_is_time_consistent(inode)) 3569 return false; 3570 3571 spin_lock(&F2FS_I(inode)->i_size_lock); 3572 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3573 spin_unlock(&F2FS_I(inode)->i_size_lock); 3574 3575 return ret; 3576 } 3577 3578 static inline bool f2fs_readonly(struct super_block *sb) 3579 { 3580 return sb_rdonly(sb); 3581 } 3582 3583 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3584 { 3585 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3586 } 3587 3588 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3589 size_t size, gfp_t flags) 3590 { 3591 if (time_to_inject(sbi, FAULT_KMALLOC)) 3592 return NULL; 3593 3594 return kmalloc(size, flags); 3595 } 3596 3597 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3598 { 3599 if (time_to_inject(sbi, FAULT_KMALLOC)) 3600 return NULL; 3601 3602 return __getname(); 3603 } 3604 3605 static inline void f2fs_putname(char *buf) 3606 { 3607 __putname(buf); 3608 } 3609 3610 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3611 size_t size, gfp_t flags) 3612 { 3613 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3614 } 3615 3616 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3617 size_t size, gfp_t flags) 3618 { 3619 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3620 return NULL; 3621 3622 return kvmalloc(size, flags); 3623 } 3624 3625 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3626 size_t size, gfp_t flags) 3627 { 3628 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3629 } 3630 3631 static inline void *f2fs_vmalloc(struct f2fs_sb_info *sbi, size_t size) 3632 { 3633 if (time_to_inject(sbi, FAULT_VMALLOC)) 3634 return NULL; 3635 3636 return vmalloc(size); 3637 } 3638 3639 static inline int get_extra_isize(struct inode *inode) 3640 { 3641 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3642 } 3643 3644 static inline int get_inline_xattr_addrs(struct inode *inode) 3645 { 3646 return F2FS_I(inode)->i_inline_xattr_size; 3647 } 3648 3649 #define f2fs_get_inode_mode(i) \ 3650 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3651 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3652 3653 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3654 3655 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3656 (offsetof(struct f2fs_inode, i_extra_end) - \ 3657 offsetof(struct f2fs_inode, i_extra_isize)) \ 3658 3659 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3660 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3661 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3662 sizeof((f2fs_inode)->field)) \ 3663 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3664 3665 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3666 3667 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3668 3669 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3670 block_t blkaddr, int type); 3671 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3672 block_t blkaddr, int type) 3673 { 3674 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3675 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3676 blkaddr, type); 3677 } 3678 3679 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3680 { 3681 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3682 blkaddr == COMPRESS_ADDR) 3683 return false; 3684 return true; 3685 } 3686 3687 /* 3688 * file.c 3689 */ 3690 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3691 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3692 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3693 int f2fs_truncate(struct inode *inode); 3694 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3695 struct kstat *stat, u32 request_mask, unsigned int flags); 3696 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3697 struct iattr *attr); 3698 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3699 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3700 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3701 bool readonly, bool need_lock); 3702 int f2fs_precache_extents(struct inode *inode); 3703 int f2fs_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3704 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3705 struct dentry *dentry, struct file_kattr *fa); 3706 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3707 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3708 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3709 int f2fs_pin_file_control(struct inode *inode, bool inc); 3710 3711 /* 3712 * inode.c 3713 */ 3714 void f2fs_set_inode_flags(struct inode *inode); 3715 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct folio *folio); 3716 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct folio *folio); 3717 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3718 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3719 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3720 void f2fs_update_inode(struct inode *inode, struct folio *node_folio); 3721 void f2fs_update_inode_page(struct inode *inode); 3722 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3723 void f2fs_remove_donate_inode(struct inode *inode); 3724 void f2fs_evict_inode(struct inode *inode); 3725 void f2fs_handle_failed_inode(struct inode *inode); 3726 3727 /* 3728 * namei.c 3729 */ 3730 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3731 bool hot, bool set); 3732 struct dentry *f2fs_get_parent(struct dentry *child); 3733 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3734 struct inode **new_inode); 3735 3736 /* 3737 * dir.c 3738 */ 3739 #if IS_ENABLED(CONFIG_UNICODE) 3740 int f2fs_init_casefolded_name(const struct inode *dir, 3741 struct f2fs_filename *fname); 3742 void f2fs_free_casefolded_name(struct f2fs_filename *fname); 3743 #else 3744 static inline int f2fs_init_casefolded_name(const struct inode *dir, 3745 struct f2fs_filename *fname) 3746 { 3747 return 0; 3748 } 3749 3750 static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 3751 { 3752 } 3753 #endif /* CONFIG_UNICODE */ 3754 3755 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3756 int lookup, struct f2fs_filename *fname); 3757 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3758 struct f2fs_filename *fname); 3759 void f2fs_free_filename(struct f2fs_filename *fname); 3760 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3761 const struct f2fs_filename *fname, int *max_slots, 3762 bool use_hash); 3763 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3764 unsigned int start_pos, struct fscrypt_str *fstr); 3765 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3766 struct f2fs_dentry_ptr *d); 3767 struct folio *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3768 const struct f2fs_filename *fname, struct folio *dfolio); 3769 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3770 unsigned int current_depth); 3771 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3772 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3773 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3774 const struct f2fs_filename *fname, struct folio **res_folio); 3775 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3776 const struct qstr *child, struct folio **res_folio); 3777 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct folio **f); 3778 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3779 struct folio **folio); 3780 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3781 struct folio *folio, struct inode *inode); 3782 bool f2fs_has_enough_room(struct inode *dir, struct folio *ifolio, 3783 const struct f2fs_filename *fname); 3784 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3785 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3786 unsigned int bit_pos); 3787 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3788 struct inode *inode, nid_t ino, umode_t mode); 3789 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3790 struct inode *inode, nid_t ino, umode_t mode); 3791 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3792 struct inode *inode, nid_t ino, umode_t mode); 3793 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct folio *folio, 3794 struct inode *dir, struct inode *inode); 3795 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3796 struct f2fs_filename *fname); 3797 bool f2fs_empty_dir(struct inode *dir); 3798 3799 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3800 { 3801 if (fscrypt_is_nokey_name(dentry)) 3802 return -ENOKEY; 3803 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3804 inode, inode->i_ino, inode->i_mode); 3805 } 3806 3807 /* 3808 * super.c 3809 */ 3810 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3811 void f2fs_inode_synced(struct inode *inode); 3812 int f2fs_dquot_initialize(struct inode *inode); 3813 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3814 int f2fs_do_quota_sync(struct super_block *sb, int type); 3815 loff_t max_file_blocks(struct inode *inode); 3816 void f2fs_quota_off_umount(struct super_block *sb); 3817 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3818 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3819 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3820 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3821 int f2fs_sync_fs(struct super_block *sb, int sync); 3822 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3823 3824 /* 3825 * hash.c 3826 */ 3827 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3828 3829 /* 3830 * node.c 3831 */ 3832 struct node_info; 3833 enum node_type; 3834 3835 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3836 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3837 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct folio *folio); 3838 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3839 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct folio *folio); 3840 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3841 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3842 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3843 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3844 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3845 struct node_info *ni, bool checkpoint_context); 3846 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3847 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3848 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3849 int f2fs_truncate_xattr_node(struct inode *inode); 3850 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3851 unsigned int seq_id); 3852 int f2fs_remove_inode_page(struct inode *inode); 3853 struct folio *f2fs_new_inode_folio(struct inode *inode); 3854 struct folio *f2fs_new_node_folio(struct dnode_of_data *dn, unsigned int ofs); 3855 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3856 struct folio *f2fs_get_node_folio(struct f2fs_sb_info *sbi, pgoff_t nid, 3857 enum node_type node_type); 3858 struct folio *f2fs_get_inode_folio(struct f2fs_sb_info *sbi, pgoff_t ino); 3859 struct folio *f2fs_get_xnode_folio(struct f2fs_sb_info *sbi, pgoff_t xnid); 3860 int f2fs_move_node_folio(struct folio *node_folio, int gc_type); 3861 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3862 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3863 struct writeback_control *wbc, bool atomic, 3864 unsigned int *seq_id); 3865 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3866 struct writeback_control *wbc, 3867 bool do_balance, enum iostat_type io_type); 3868 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3869 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3870 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3871 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3872 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3873 int f2fs_recover_inline_xattr(struct inode *inode, struct folio *folio); 3874 int f2fs_recover_xattr_data(struct inode *inode, struct folio *folio); 3875 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct folio *folio); 3876 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3877 unsigned int segno, struct f2fs_summary_block *sum); 3878 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3879 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3880 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3881 int __init f2fs_create_node_manager_caches(void); 3882 void f2fs_destroy_node_manager_caches(void); 3883 3884 /* 3885 * segment.c 3886 */ 3887 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3888 int f2fs_commit_atomic_write(struct inode *inode); 3889 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3890 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3891 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3892 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3893 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3894 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3895 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3896 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr, 3897 unsigned int len); 3898 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3899 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3900 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3901 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3902 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3903 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3904 struct cp_control *cpc); 3905 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3906 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3907 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3908 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3909 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3910 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3911 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3912 int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 3913 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3914 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3915 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3916 unsigned int start, unsigned int end); 3917 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3918 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3919 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3920 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3921 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3922 struct cp_control *cpc); 3923 struct folio *f2fs_get_sum_folio(struct f2fs_sb_info *sbi, unsigned int segno); 3924 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3925 block_t blk_addr); 3926 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio, 3927 enum iostat_type io_type); 3928 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3929 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3930 struct f2fs_io_info *fio); 3931 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3932 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3933 block_t old_blkaddr, block_t new_blkaddr, 3934 bool recover_curseg, bool recover_newaddr, 3935 bool from_gc); 3936 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3937 block_t old_addr, block_t new_addr, 3938 unsigned char version, bool recover_curseg, 3939 bool recover_newaddr); 3940 enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 3941 enum log_type seg_type); 3942 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct folio *folio, 3943 block_t old_blkaddr, block_t *new_blkaddr, 3944 struct f2fs_summary *sum, int type, 3945 struct f2fs_io_info *fio); 3946 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3947 block_t blkaddr, unsigned int blkcnt); 3948 void f2fs_folio_wait_writeback(struct folio *folio, enum page_type type, 3949 bool ordered, bool locked); 3950 #define f2fs_wait_on_page_writeback(page, type, ordered, locked) \ 3951 f2fs_folio_wait_writeback(page_folio(page), type, ordered, locked) 3952 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3953 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3954 block_t len); 3955 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3956 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3957 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3958 unsigned int val, int alloc); 3959 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3960 int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 3961 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3962 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3963 int __init f2fs_create_segment_manager_caches(void); 3964 void f2fs_destroy_segment_manager_caches(void); 3965 int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 3966 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3967 enum page_type type, enum temp_type temp); 3968 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 3969 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3970 unsigned int segno); 3971 unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 3972 unsigned int segno); 3973 3974 static inline struct inode *fio_inode(struct f2fs_io_info *fio) 3975 { 3976 return fio->folio->mapping->host; 3977 } 3978 3979 #define DEF_FRAGMENT_SIZE 4 3980 #define MIN_FRAGMENT_SIZE 1 3981 #define MAX_FRAGMENT_SIZE 512 3982 3983 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3984 { 3985 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3986 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3987 } 3988 3989 /* 3990 * checkpoint.c 3991 */ 3992 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3993 unsigned char reason); 3994 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3995 struct folio *f2fs_grab_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3996 struct folio *f2fs_get_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3997 struct folio *f2fs_get_meta_folio_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3998 struct folio *f2fs_get_tmp_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3999 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 4000 block_t blkaddr, int type); 4001 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 4002 block_t blkaddr, int type); 4003 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 4004 int type, bool sync); 4005 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 4006 unsigned int ra_blocks); 4007 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 4008 long nr_to_write, enum iostat_type io_type); 4009 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4010 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4011 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 4012 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 4013 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4014 unsigned int devidx, int type); 4015 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4016 unsigned int devidx, int type); 4017 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 4018 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 4019 void f2fs_add_orphan_inode(struct inode *inode); 4020 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 4021 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 4022 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 4023 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 4024 void f2fs_remove_dirty_inode(struct inode *inode); 4025 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 4026 bool from_cp); 4027 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 4028 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 4029 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4030 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 4031 int __init f2fs_create_checkpoint_caches(void); 4032 void f2fs_destroy_checkpoint_caches(void); 4033 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 4034 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 4035 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 4036 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 4037 4038 /* 4039 * data.c 4040 */ 4041 int __init f2fs_init_bioset(void); 4042 void f2fs_destroy_bioset(void); 4043 bool f2fs_is_cp_guaranteed(const struct folio *folio); 4044 int f2fs_init_bio_entry_cache(void); 4045 void f2fs_destroy_bio_entry_cache(void); 4046 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 4047 enum page_type type); 4048 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 4049 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 4050 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 4051 struct inode *inode, struct folio *folio, 4052 nid_t ino, enum page_type type); 4053 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 4054 struct bio **bio, struct folio *folio); 4055 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 4056 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 4057 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 4058 void f2fs_submit_page_write(struct f2fs_io_info *fio); 4059 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 4060 block_t blk_addr, sector_t *sector); 4061 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 4062 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4063 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4064 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 4065 int f2fs_reserve_new_block(struct dnode_of_data *dn); 4066 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 4067 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 4068 struct folio *f2fs_get_read_data_folio(struct inode *inode, pgoff_t index, 4069 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 4070 struct folio *f2fs_find_data_folio(struct inode *inode, pgoff_t index, 4071 pgoff_t *next_pgofs); 4072 struct folio *f2fs_get_lock_data_folio(struct inode *inode, pgoff_t index, 4073 bool for_write); 4074 struct folio *f2fs_get_new_data_folio(struct inode *inode, 4075 struct folio *ifolio, pgoff_t index, bool new_i_size); 4076 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 4077 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 4078 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 4079 u64 start, u64 len); 4080 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 4081 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 4082 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 4083 int f2fs_write_single_data_page(struct folio *folio, int *submitted, 4084 struct bio **bio, sector_t *last_block, 4085 struct writeback_control *wbc, 4086 enum iostat_type io_type, 4087 int compr_blocks, bool allow_balance); 4088 void f2fs_write_failed(struct inode *inode, loff_t to); 4089 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 4090 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 4091 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 4092 void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 4093 int f2fs_init_post_read_processing(void); 4094 void f2fs_destroy_post_read_processing(void); 4095 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 4096 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 4097 extern const struct iomap_ops f2fs_iomap_ops; 4098 4099 /* 4100 * gc.c 4101 */ 4102 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 4103 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 4104 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 4105 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 4106 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 4107 int f2fs_gc_range(struct f2fs_sb_info *sbi, 4108 unsigned int start_seg, unsigned int end_seg, 4109 bool dry_run, unsigned int dry_run_sections); 4110 int f2fs_resize_fs(struct file *filp, __u64 block_count); 4111 int __init f2fs_create_garbage_collection_cache(void); 4112 void f2fs_destroy_garbage_collection_cache(void); 4113 /* victim selection function for cleaning and SSR */ 4114 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 4115 int gc_type, int type, char alloc_mode, 4116 unsigned long long age, bool one_time); 4117 4118 /* 4119 * recovery.c 4120 */ 4121 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 4122 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 4123 int __init f2fs_create_recovery_cache(void); 4124 void f2fs_destroy_recovery_cache(void); 4125 4126 /* 4127 * debug.c 4128 */ 4129 #ifdef CONFIG_F2FS_STAT_FS 4130 enum { 4131 DEVSTAT_INUSE, 4132 DEVSTAT_DIRTY, 4133 DEVSTAT_FULL, 4134 DEVSTAT_FREE, 4135 DEVSTAT_PREFREE, 4136 DEVSTAT_MAX, 4137 }; 4138 4139 struct f2fs_dev_stats { 4140 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 4141 }; 4142 4143 struct f2fs_stat_info { 4144 struct list_head stat_list; 4145 struct f2fs_sb_info *sbi; 4146 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 4147 int main_area_segs, main_area_sections, main_area_zones; 4148 unsigned long long hit_cached[NR_EXTENT_CACHES]; 4149 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 4150 unsigned long long total_ext[NR_EXTENT_CACHES]; 4151 unsigned long long hit_total[NR_EXTENT_CACHES]; 4152 int ext_tree[NR_EXTENT_CACHES]; 4153 int zombie_tree[NR_EXTENT_CACHES]; 4154 int ext_node[NR_EXTENT_CACHES]; 4155 /* to count memory footprint */ 4156 unsigned long long ext_mem[NR_EXTENT_CACHES]; 4157 /* for read extent cache */ 4158 unsigned long long hit_largest; 4159 /* for block age extent cache */ 4160 unsigned long long allocated_data_blocks; 4161 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 4162 int ndirty_data, ndirty_qdata; 4163 unsigned int ndirty_dirs, ndirty_files, ndirty_all; 4164 unsigned int nquota_files, ndonate_files; 4165 int nats, dirty_nats, sits, dirty_sits; 4166 int free_nids, avail_nids, alloc_nids; 4167 int total_count, utilization; 4168 int nr_wb_cp_data, nr_wb_data; 4169 int nr_rd_data, nr_rd_node, nr_rd_meta; 4170 int nr_dio_read, nr_dio_write; 4171 unsigned int io_skip_bggc, other_skip_bggc; 4172 int nr_flushing, nr_flushed, flush_list_empty; 4173 int nr_discarding, nr_discarded; 4174 int nr_discard_cmd; 4175 unsigned int undiscard_blks; 4176 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 4177 unsigned int cur_ckpt_time, peak_ckpt_time; 4178 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 4179 int compr_inode, swapfile_inode; 4180 unsigned long long compr_blocks; 4181 int aw_cnt, max_aw_cnt; 4182 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 4183 unsigned int bimodal, avg_vblocks; 4184 int util_free, util_valid, util_invalid; 4185 int rsvd_segs, overp_segs; 4186 int dirty_count, node_pages, meta_pages, compress_pages; 4187 int compress_page_hit; 4188 int prefree_count, free_segs, free_secs; 4189 int cp_call_count[MAX_CALL_TYPE], cp_count; 4190 int gc_call_count[MAX_CALL_TYPE]; 4191 int gc_segs[2][2]; 4192 int gc_secs[2][2]; 4193 int tot_blks, data_blks, node_blks; 4194 int bg_data_blks, bg_node_blks; 4195 int blkoff[NR_CURSEG_TYPE]; 4196 int curseg[NR_CURSEG_TYPE]; 4197 int cursec[NR_CURSEG_TYPE]; 4198 int curzone[NR_CURSEG_TYPE]; 4199 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4200 unsigned int full_seg[NR_CURSEG_TYPE]; 4201 unsigned int valid_blks[NR_CURSEG_TYPE]; 4202 4203 unsigned int meta_count[META_MAX]; 4204 unsigned int segment_count[2]; 4205 unsigned int block_count[2]; 4206 unsigned int inplace_count; 4207 unsigned long long base_mem, cache_mem, page_mem; 4208 struct f2fs_dev_stats *dev_stats; 4209 }; 4210 4211 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4212 { 4213 return (struct f2fs_stat_info *)sbi->stat_info; 4214 } 4215 4216 #define stat_inc_cp_call_count(sbi, foreground) \ 4217 atomic_inc(&sbi->cp_call_count[(foreground)]) 4218 #define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4219 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4220 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4221 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4222 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4223 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4224 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4225 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4226 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4227 #define stat_inc_inline_xattr(inode) \ 4228 do { \ 4229 if (f2fs_has_inline_xattr(inode)) \ 4230 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4231 } while (0) 4232 #define stat_dec_inline_xattr(inode) \ 4233 do { \ 4234 if (f2fs_has_inline_xattr(inode)) \ 4235 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4236 } while (0) 4237 #define stat_inc_inline_inode(inode) \ 4238 do { \ 4239 if (f2fs_has_inline_data(inode)) \ 4240 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4241 } while (0) 4242 #define stat_dec_inline_inode(inode) \ 4243 do { \ 4244 if (f2fs_has_inline_data(inode)) \ 4245 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4246 } while (0) 4247 #define stat_inc_inline_dir(inode) \ 4248 do { \ 4249 if (f2fs_has_inline_dentry(inode)) \ 4250 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4251 } while (0) 4252 #define stat_dec_inline_dir(inode) \ 4253 do { \ 4254 if (f2fs_has_inline_dentry(inode)) \ 4255 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4256 } while (0) 4257 #define stat_inc_compr_inode(inode) \ 4258 do { \ 4259 if (f2fs_compressed_file(inode)) \ 4260 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4261 } while (0) 4262 #define stat_dec_compr_inode(inode) \ 4263 do { \ 4264 if (f2fs_compressed_file(inode)) \ 4265 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4266 } while (0) 4267 #define stat_add_compr_blocks(inode, blocks) \ 4268 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4269 #define stat_sub_compr_blocks(inode, blocks) \ 4270 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4271 #define stat_inc_swapfile_inode(inode) \ 4272 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4273 #define stat_dec_swapfile_inode(inode) \ 4274 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4275 #define stat_inc_atomic_inode(inode) \ 4276 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4277 #define stat_dec_atomic_inode(inode) \ 4278 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4279 #define stat_inc_meta_count(sbi, blkaddr) \ 4280 do { \ 4281 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4282 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4283 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4284 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4285 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4286 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4287 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4288 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4289 } while (0) 4290 #define stat_inc_seg_type(sbi, curseg) \ 4291 ((sbi)->segment_count[(curseg)->alloc_type]++) 4292 #define stat_inc_block_count(sbi, curseg) \ 4293 ((sbi)->block_count[(curseg)->alloc_type]++) 4294 #define stat_inc_inplace_blocks(sbi) \ 4295 (atomic_inc(&(sbi)->inplace_count)) 4296 #define stat_update_max_atomic_write(inode) \ 4297 do { \ 4298 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4299 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4300 if (cur > max) \ 4301 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4302 } while (0) 4303 #define stat_inc_gc_call_count(sbi, foreground) \ 4304 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4305 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4306 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4307 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4308 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4309 4310 #define stat_inc_tot_blk_count(si, blks) \ 4311 ((si)->tot_blks += (blks)) 4312 4313 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4314 do { \ 4315 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4316 stat_inc_tot_blk_count(si, blks); \ 4317 si->data_blks += (blks); \ 4318 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4319 } while (0) 4320 4321 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4322 do { \ 4323 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4324 stat_inc_tot_blk_count(si, blks); \ 4325 si->node_blks += (blks); \ 4326 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4327 } while (0) 4328 4329 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4330 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4331 void __init f2fs_create_root_stats(void); 4332 void f2fs_destroy_root_stats(void); 4333 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4334 #else 4335 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4336 #define stat_inc_cp_count(sbi) do { } while (0) 4337 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4338 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4339 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4340 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4341 #define stat_inc_total_hit(sbi, type) do { } while (0) 4342 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4343 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4344 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4345 #define stat_inc_inline_xattr(inode) do { } while (0) 4346 #define stat_dec_inline_xattr(inode) do { } while (0) 4347 #define stat_inc_inline_inode(inode) do { } while (0) 4348 #define stat_dec_inline_inode(inode) do { } while (0) 4349 #define stat_inc_inline_dir(inode) do { } while (0) 4350 #define stat_dec_inline_dir(inode) do { } while (0) 4351 #define stat_inc_compr_inode(inode) do { } while (0) 4352 #define stat_dec_compr_inode(inode) do { } while (0) 4353 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4354 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4355 #define stat_inc_swapfile_inode(inode) do { } while (0) 4356 #define stat_dec_swapfile_inode(inode) do { } while (0) 4357 #define stat_inc_atomic_inode(inode) do { } while (0) 4358 #define stat_dec_atomic_inode(inode) do { } while (0) 4359 #define stat_update_max_atomic_write(inode) do { } while (0) 4360 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4361 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4362 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4363 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4364 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4365 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4366 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4367 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4368 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4369 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4370 4371 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4372 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4373 static inline void __init f2fs_create_root_stats(void) { } 4374 static inline void f2fs_destroy_root_stats(void) { } 4375 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4376 #endif 4377 4378 extern const struct file_operations f2fs_dir_operations; 4379 extern const struct file_operations f2fs_file_operations; 4380 extern const struct inode_operations f2fs_file_inode_operations; 4381 extern const struct address_space_operations f2fs_dblock_aops; 4382 extern const struct address_space_operations f2fs_node_aops; 4383 extern const struct address_space_operations f2fs_meta_aops; 4384 extern const struct inode_operations f2fs_dir_inode_operations; 4385 extern const struct inode_operations f2fs_symlink_inode_operations; 4386 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4387 extern const struct inode_operations f2fs_special_inode_operations; 4388 extern struct kmem_cache *f2fs_inode_entry_slab; 4389 4390 /* 4391 * inline.c 4392 */ 4393 bool f2fs_may_inline_data(struct inode *inode); 4394 bool f2fs_sanity_check_inline_data(struct inode *inode, struct folio *ifolio); 4395 bool f2fs_may_inline_dentry(struct inode *inode); 4396 void f2fs_do_read_inline_data(struct folio *folio, struct folio *ifolio); 4397 void f2fs_truncate_inline_inode(struct inode *inode, struct folio *ifolio, 4398 u64 from); 4399 int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4400 int f2fs_convert_inline_folio(struct dnode_of_data *dn, struct folio *folio); 4401 int f2fs_convert_inline_inode(struct inode *inode); 4402 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4403 int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4404 int f2fs_recover_inline_data(struct inode *inode, struct folio *nfolio); 4405 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4406 const struct f2fs_filename *fname, struct folio **res_folio, 4407 bool use_hash); 4408 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4409 struct folio *ifolio); 4410 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4411 struct inode *inode, nid_t ino, umode_t mode); 4412 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4413 struct folio *folio, struct inode *dir, struct inode *inode); 4414 bool f2fs_empty_inline_dir(struct inode *dir); 4415 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4416 struct fscrypt_str *fstr); 4417 int f2fs_inline_data_fiemap(struct inode *inode, 4418 struct fiemap_extent_info *fieinfo, 4419 __u64 start, __u64 len); 4420 4421 /* 4422 * shrinker.c 4423 */ 4424 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4425 struct shrink_control *sc); 4426 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4427 struct shrink_control *sc); 4428 unsigned int f2fs_donate_files(void); 4429 void f2fs_reclaim_caches(unsigned int reclaim_caches_kb); 4430 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4431 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4432 4433 /* 4434 * extent_cache.c 4435 */ 4436 bool sanity_check_extent_cache(struct inode *inode, struct folio *ifolio); 4437 void f2fs_init_extent_tree(struct inode *inode); 4438 void f2fs_drop_extent_tree(struct inode *inode); 4439 void f2fs_destroy_extent_node(struct inode *inode); 4440 void f2fs_destroy_extent_tree(struct inode *inode); 4441 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4442 int __init f2fs_create_extent_cache(void); 4443 void f2fs_destroy_extent_cache(void); 4444 4445 /* read extent cache ops */ 4446 void f2fs_init_read_extent_tree(struct inode *inode, struct folio *ifolio); 4447 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4448 struct extent_info *ei); 4449 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4450 block_t *blkaddr); 4451 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4452 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4453 pgoff_t fofs, block_t blkaddr, unsigned int len); 4454 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4455 int nr_shrink); 4456 4457 /* block age extent cache ops */ 4458 void f2fs_init_age_extent_tree(struct inode *inode); 4459 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4460 struct extent_info *ei); 4461 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4462 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4463 pgoff_t fofs, unsigned int len); 4464 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4465 int nr_shrink); 4466 4467 /* 4468 * sysfs.c 4469 */ 4470 #define MIN_RA_MUL 2 4471 #define MAX_RA_MUL 256 4472 4473 int __init f2fs_init_sysfs(void); 4474 void f2fs_exit_sysfs(void); 4475 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4476 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4477 4478 /* verity.c */ 4479 extern const struct fsverity_operations f2fs_verityops; 4480 4481 /* 4482 * crypto support 4483 */ 4484 static inline bool f2fs_encrypted_file(struct inode *inode) 4485 { 4486 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4487 } 4488 4489 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4490 { 4491 #ifdef CONFIG_FS_ENCRYPTION 4492 file_set_encrypt(inode); 4493 f2fs_set_inode_flags(inode); 4494 #endif 4495 } 4496 4497 /* 4498 * Returns true if the reads of the inode's data need to undergo some 4499 * postprocessing step, like decryption or authenticity verification. 4500 */ 4501 static inline bool f2fs_post_read_required(struct inode *inode) 4502 { 4503 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4504 f2fs_compressed_file(inode); 4505 } 4506 4507 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4508 { 4509 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4510 } 4511 4512 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4513 { 4514 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4515 } 4516 4517 /* 4518 * compress.c 4519 */ 4520 #ifdef CONFIG_F2FS_FS_COMPRESSION 4521 enum cluster_check_type { 4522 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4523 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4524 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4525 }; 4526 bool f2fs_is_compressed_page(struct folio *folio); 4527 struct folio *f2fs_compress_control_folio(struct folio *folio); 4528 int f2fs_prepare_compress_overwrite(struct inode *inode, 4529 struct page **pagep, pgoff_t index, void **fsdata); 4530 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4531 pgoff_t index, unsigned copied); 4532 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4533 void f2fs_compress_write_end_io(struct bio *bio, struct folio *folio); 4534 bool f2fs_is_compress_backend_ready(struct inode *inode); 4535 bool f2fs_is_compress_level_valid(int alg, int lvl); 4536 int __init f2fs_init_compress_mempool(void); 4537 void f2fs_destroy_compress_mempool(void); 4538 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4539 void f2fs_end_read_compressed_page(struct folio *folio, bool failed, 4540 block_t blkaddr, bool in_task); 4541 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4542 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4543 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4544 int index, int nr_pages, bool uptodate); 4545 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4546 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4547 int f2fs_write_multi_pages(struct compress_ctx *cc, 4548 int *submitted, 4549 struct writeback_control *wbc, 4550 enum iostat_type io_type); 4551 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4552 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4553 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4554 pgoff_t fofs, block_t blkaddr, 4555 unsigned int llen, unsigned int c_len); 4556 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4557 unsigned nr_pages, sector_t *last_block_in_bio, 4558 struct readahead_control *rac, bool for_write); 4559 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4560 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4561 bool in_task); 4562 void f2fs_put_folio_dic(struct folio *folio, bool in_task); 4563 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4564 unsigned int ofs_in_node); 4565 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4566 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4567 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4568 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4569 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4570 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4571 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4572 int __init f2fs_init_compress_cache(void); 4573 void f2fs_destroy_compress_cache(void); 4574 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4575 void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4576 block_t blkaddr, unsigned int len); 4577 bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, struct folio *folio, 4578 block_t blkaddr); 4579 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4580 #define inc_compr_inode_stat(inode) \ 4581 do { \ 4582 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4583 sbi->compr_new_inode++; \ 4584 } while (0) 4585 #define add_compr_block_stat(inode, blocks) \ 4586 do { \ 4587 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4588 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4589 sbi->compr_written_block += blocks; \ 4590 sbi->compr_saved_block += diff; \ 4591 } while (0) 4592 #else 4593 static inline bool f2fs_is_compressed_page(struct folio *folio) { return false; } 4594 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4595 { 4596 if (!f2fs_compressed_file(inode)) 4597 return true; 4598 /* not support compression */ 4599 return false; 4600 } 4601 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4602 static inline struct folio *f2fs_compress_control_folio(struct folio *folio) 4603 { 4604 WARN_ON_ONCE(1); 4605 return ERR_PTR(-EINVAL); 4606 } 4607 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4608 static inline void f2fs_destroy_compress_mempool(void) { } 4609 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4610 bool in_task) { } 4611 static inline void f2fs_end_read_compressed_page(struct folio *folio, 4612 bool failed, block_t blkaddr, bool in_task) 4613 { 4614 WARN_ON_ONCE(1); 4615 } 4616 static inline void f2fs_put_folio_dic(struct folio *folio, bool in_task) 4617 { 4618 WARN_ON_ONCE(1); 4619 } 4620 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4621 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4622 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4623 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4624 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4625 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4626 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4627 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4628 static inline void f2fs_destroy_compress_cache(void) { } 4629 static inline void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4630 block_t blkaddr, unsigned int len) { } 4631 static inline bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, 4632 struct folio *folio, block_t blkaddr) { return false; } 4633 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4634 nid_t ino) { } 4635 #define inc_compr_inode_stat(inode) do { } while (0) 4636 static inline int f2fs_is_compressed_cluster( 4637 struct inode *inode, 4638 pgoff_t index) { return 0; } 4639 static inline bool f2fs_is_sparse_cluster( 4640 struct inode *inode, 4641 pgoff_t index) { return true; } 4642 static inline void f2fs_update_read_extent_tree_range_compressed( 4643 struct inode *inode, 4644 pgoff_t fofs, block_t blkaddr, 4645 unsigned int llen, unsigned int c_len) { } 4646 #endif 4647 4648 static inline int set_compress_context(struct inode *inode) 4649 { 4650 #ifdef CONFIG_F2FS_FS_COMPRESSION 4651 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4652 struct f2fs_inode_info *fi = F2FS_I(inode); 4653 4654 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4655 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4656 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4657 BIT(COMPRESS_CHKSUM) : 0; 4658 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4659 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4660 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4661 F2FS_OPTION(sbi).compress_level) 4662 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4663 fi->i_flags |= F2FS_COMPR_FL; 4664 set_inode_flag(inode, FI_COMPRESSED_FILE); 4665 stat_inc_compr_inode(inode); 4666 inc_compr_inode_stat(inode); 4667 f2fs_mark_inode_dirty_sync(inode, true); 4668 return 0; 4669 #else 4670 return -EOPNOTSUPP; 4671 #endif 4672 } 4673 4674 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4675 { 4676 struct f2fs_inode_info *fi = F2FS_I(inode); 4677 4678 f2fs_down_write(&fi->i_sem); 4679 4680 if (!f2fs_compressed_file(inode)) { 4681 f2fs_up_write(&fi->i_sem); 4682 return true; 4683 } 4684 if (f2fs_is_mmap_file(inode) || atomic_read(&fi->writeback) || 4685 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4686 f2fs_up_write(&fi->i_sem); 4687 return false; 4688 } 4689 4690 fi->i_flags &= ~F2FS_COMPR_FL; 4691 stat_dec_compr_inode(inode); 4692 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4693 f2fs_mark_inode_dirty_sync(inode, true); 4694 4695 f2fs_up_write(&fi->i_sem); 4696 return true; 4697 } 4698 4699 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4700 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4701 { \ 4702 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4703 } 4704 4705 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4706 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4707 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4708 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4709 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4710 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4711 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4712 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4713 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4714 F2FS_FEATURE_FUNCS(verity, VERITY); 4715 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4716 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4717 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4718 F2FS_FEATURE_FUNCS(readonly, RO); 4719 F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 4720 F2FS_FEATURE_FUNCS(packed_ssa, PACKED_SSA); 4721 4722 #ifdef CONFIG_BLK_DEV_ZONED 4723 static inline bool f2fs_zone_is_seq(struct f2fs_sb_info *sbi, int devi, 4724 unsigned int zone) 4725 { 4726 return test_bit(zone, FDEV(devi).blkz_seq); 4727 } 4728 4729 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4730 block_t blkaddr) 4731 { 4732 return f2fs_zone_is_seq(sbi, devi, blkaddr / sbi->blocks_per_blkz); 4733 } 4734 #endif 4735 4736 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4737 struct block_device *bdev) 4738 { 4739 int i; 4740 4741 if (!f2fs_is_multi_device(sbi)) 4742 return 0; 4743 4744 for (i = 0; i < sbi->s_ndevs; i++) 4745 if (FDEV(i).bdev == bdev) 4746 return i; 4747 4748 WARN_ON(1); 4749 return -1; 4750 } 4751 4752 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4753 { 4754 return f2fs_sb_has_blkzoned(sbi); 4755 } 4756 4757 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4758 { 4759 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4760 } 4761 4762 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4763 { 4764 int i; 4765 4766 if (!f2fs_is_multi_device(sbi)) 4767 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4768 4769 for (i = 0; i < sbi->s_ndevs; i++) 4770 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4771 return true; 4772 return false; 4773 } 4774 4775 static inline unsigned int f2fs_hw_discard_granularity(struct f2fs_sb_info *sbi) 4776 { 4777 int i = 1; 4778 unsigned int discard_granularity = bdev_discard_granularity(sbi->sb->s_bdev); 4779 4780 if (f2fs_is_multi_device(sbi)) 4781 for (; i < sbi->s_ndevs && !bdev_is_zoned(FDEV(i).bdev); i++) 4782 discard_granularity = max_t(unsigned int, discard_granularity, 4783 bdev_discard_granularity(FDEV(i).bdev)); 4784 return discard_granularity; 4785 } 4786 4787 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4788 { 4789 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4790 f2fs_hw_should_discard(sbi); 4791 } 4792 4793 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4794 { 4795 int i; 4796 4797 if (!f2fs_is_multi_device(sbi)) 4798 return bdev_read_only(sbi->sb->s_bdev); 4799 4800 for (i = 0; i < sbi->s_ndevs; i++) 4801 if (bdev_read_only(FDEV(i).bdev)) 4802 return true; 4803 return false; 4804 } 4805 4806 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4807 { 4808 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4809 } 4810 4811 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4812 { 4813 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4814 } 4815 4816 static inline bool f2fs_is_sequential_zone_area(struct f2fs_sb_info *sbi, 4817 block_t blkaddr) 4818 { 4819 if (f2fs_sb_has_blkzoned(sbi)) { 4820 #ifdef CONFIG_BLK_DEV_ZONED 4821 int devi = f2fs_target_device_index(sbi, blkaddr); 4822 4823 if (!bdev_is_zoned(FDEV(devi).bdev)) 4824 return false; 4825 4826 if (f2fs_is_multi_device(sbi)) { 4827 if (blkaddr < FDEV(devi).start_blk || 4828 blkaddr > FDEV(devi).end_blk) { 4829 f2fs_err(sbi, "Invalid block %x", blkaddr); 4830 return false; 4831 } 4832 blkaddr -= FDEV(devi).start_blk; 4833 } 4834 4835 return f2fs_blkz_is_seq(sbi, devi, blkaddr); 4836 #else 4837 return false; 4838 #endif 4839 } 4840 return false; 4841 } 4842 4843 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4844 { 4845 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4846 } 4847 4848 static inline bool f2fs_may_compress(struct inode *inode) 4849 { 4850 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4851 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4852 f2fs_is_mmap_file(inode)) 4853 return false; 4854 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4855 } 4856 4857 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4858 u64 blocks, bool add) 4859 { 4860 struct f2fs_inode_info *fi = F2FS_I(inode); 4861 int diff = fi->i_cluster_size - blocks; 4862 4863 /* don't update i_compr_blocks if saved blocks were released */ 4864 if (!add && !atomic_read(&fi->i_compr_blocks)) 4865 return; 4866 4867 if (add) { 4868 atomic_add(diff, &fi->i_compr_blocks); 4869 stat_add_compr_blocks(inode, diff); 4870 } else { 4871 atomic_sub(diff, &fi->i_compr_blocks); 4872 stat_sub_compr_blocks(inode, diff); 4873 } 4874 f2fs_mark_inode_dirty_sync(inode, true); 4875 } 4876 4877 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4878 int flag) 4879 { 4880 if (!f2fs_is_multi_device(sbi)) 4881 return false; 4882 if (flag != F2FS_GET_BLOCK_DIO) 4883 return false; 4884 return sbi->aligned_blksize; 4885 } 4886 4887 #ifdef CONFIG_F2FS_FAULT_INJECTION 4888 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4889 unsigned long type, enum fault_option fo); 4890 #else 4891 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4892 unsigned long rate, unsigned long type, 4893 enum fault_option fo) 4894 { 4895 return 0; 4896 } 4897 #endif 4898 4899 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4900 { 4901 #ifdef CONFIG_QUOTA 4902 if (f2fs_sb_has_quota_ino(sbi)) 4903 return true; 4904 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4905 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4906 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4907 return true; 4908 #endif 4909 return false; 4910 } 4911 4912 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4913 { 4914 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4915 } 4916 4917 static inline void __f2fs_schedule_timeout(long timeout, bool io) 4918 { 4919 set_current_state(TASK_UNINTERRUPTIBLE); 4920 if (io) 4921 io_schedule_timeout(timeout); 4922 else 4923 schedule_timeout(timeout); 4924 } 4925 4926 #define f2fs_io_schedule_timeout(timeout) \ 4927 __f2fs_schedule_timeout(timeout, true) 4928 #define f2fs_schedule_timeout(timeout) \ 4929 __f2fs_schedule_timeout(timeout, false) 4930 4931 static inline void f2fs_io_schedule_timeout_killable(long timeout) 4932 { 4933 while (timeout) { 4934 if (fatal_signal_pending(current)) 4935 return; 4936 set_current_state(TASK_UNINTERRUPTIBLE); 4937 io_schedule_timeout(DEFAULT_SCHEDULE_TIMEOUT); 4938 if (timeout <= DEFAULT_SCHEDULE_TIMEOUT) 4939 return; 4940 timeout -= DEFAULT_SCHEDULE_TIMEOUT; 4941 } 4942 } 4943 4944 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 4945 struct folio *folio, enum page_type type) 4946 { 4947 pgoff_t ofs = folio->index; 4948 4949 if (unlikely(f2fs_cp_error(sbi))) 4950 return; 4951 4952 if (ofs == sbi->page_eio_ofs[type]) { 4953 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4954 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4955 } else { 4956 sbi->page_eio_ofs[type] = ofs; 4957 sbi->page_eio_cnt[type] = 0; 4958 } 4959 } 4960 4961 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4962 { 4963 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4964 } 4965 4966 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4967 block_t blkaddr, unsigned int cnt) 4968 { 4969 bool need_submit = false; 4970 int i = 0; 4971 4972 do { 4973 struct folio *folio; 4974 4975 folio = filemap_get_folio(META_MAPPING(sbi), blkaddr + i); 4976 if (!IS_ERR(folio)) { 4977 if (folio_test_writeback(folio)) 4978 need_submit = true; 4979 f2fs_folio_put(folio, false); 4980 } 4981 } while (++i < cnt && !need_submit); 4982 4983 if (need_submit) 4984 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4985 NULL, 0, DATA); 4986 4987 truncate_inode_pages_range(META_MAPPING(sbi), 4988 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4989 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4990 } 4991 4992 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4993 block_t blkaddr, unsigned int len) 4994 { 4995 f2fs_truncate_meta_inode_pages(sbi, blkaddr, len); 4996 f2fs_invalidate_compress_pages_range(sbi, blkaddr, len); 4997 } 4998 4999 #endif /* _LINUX_F2FS_H */ 5000