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 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 100 #define F2FS_MOUNT_DISCARD 0x00000002 101 #define F2FS_MOUNT_NOHEAP 0x00000004 102 #define F2FS_MOUNT_XATTR_USER 0x00000008 103 #define F2FS_MOUNT_POSIX_ACL 0x00000010 104 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 105 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 106 #define F2FS_MOUNT_INLINE_DATA 0x00000080 107 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 108 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 109 #define F2FS_MOUNT_NOBARRIER 0x00000400 110 #define F2FS_MOUNT_FASTBOOT 0x00000800 111 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 112 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 113 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 114 #define F2FS_MOUNT_USRQUOTA 0x00008000 115 #define F2FS_MOUNT_GRPQUOTA 0x00010000 116 #define F2FS_MOUNT_PRJQUOTA 0x00020000 117 #define F2FS_MOUNT_QUOTA 0x00040000 118 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 119 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 120 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 121 #define F2FS_MOUNT_NORECOVERY 0x00400000 122 #define F2FS_MOUNT_ATGC 0x00800000 123 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 124 #define F2FS_MOUNT_GC_MERGE 0x02000000 125 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 126 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 127 #define F2FS_MOUNT_NAT_BITS 0x10000000 128 #define F2FS_MOUNT_INLINECRYPT 0x20000000 129 /* 130 * Some f2fs environments expect to be able to pass the "lazytime" option 131 * string rather than using the MS_LAZYTIME flag, so this must remain. 132 */ 133 #define F2FS_MOUNT_LAZYTIME 0x40000000 134 #define F2FS_MOUNT_RESERVE_NODE 0x80000000 135 136 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 137 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 138 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 139 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 140 141 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 142 typecheck(unsigned long long, b) && \ 143 ((long long)((a) - (b)) > 0)) 144 145 typedef u32 block_t; /* 146 * should not change u32, since it is the on-disk block 147 * address format, __le32. 148 */ 149 typedef u32 nid_t; 150 151 #define COMPRESS_EXT_NUM 16 152 153 enum blkzone_allocation_policy { 154 BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */ 155 BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */ 156 BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */ 157 }; 158 159 enum bggc_io_aware_policy { 160 AWARE_ALL_IO, /* skip background GC if there is any kind of pending IO */ 161 AWARE_READ_IO, /* skip background GC if there is pending read IO */ 162 AWARE_NONE, /* don't aware IO for background GC */ 163 }; 164 165 enum device_allocation_policy { 166 ALLOCATE_FORWARD_NOHINT, 167 ALLOCATE_FORWARD_WITHIN_HINT, 168 ALLOCATE_FORWARD_FROM_HINT, 169 }; 170 171 /* 172 * An implementation of an rwsem that is explicitly unfair to readers. This 173 * prevents priority inversion when a low-priority reader acquires the read lock 174 * while sleeping on the write lock but the write lock is needed by 175 * higher-priority clients. 176 */ 177 178 struct f2fs_rwsem { 179 struct rw_semaphore internal_rwsem; 180 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 181 wait_queue_head_t read_waiters; 182 #endif 183 }; 184 185 struct f2fs_mount_info { 186 unsigned int opt; 187 block_t root_reserved_blocks; /* root reserved blocks */ 188 block_t root_reserved_nodes; /* root reserved nodes */ 189 kuid_t s_resuid; /* reserved blocks for uid */ 190 kgid_t s_resgid; /* reserved blocks for gid */ 191 int active_logs; /* # of active logs */ 192 int inline_xattr_size; /* inline xattr size */ 193 #ifdef CONFIG_F2FS_FAULT_INJECTION 194 struct f2fs_fault_info fault_info; /* For fault injection */ 195 #endif 196 #ifdef CONFIG_QUOTA 197 /* Names of quota files with journalled quota */ 198 char *s_qf_names[MAXQUOTAS]; 199 int s_jquota_fmt; /* Format of quota to use */ 200 #endif 201 /* For which write hints are passed down to block layer */ 202 int alloc_mode; /* segment allocation policy */ 203 int fsync_mode; /* fsync policy */ 204 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 205 int bggc_mode; /* bggc mode: off, on or sync */ 206 int memory_mode; /* memory mode */ 207 int errors; /* errors parameter */ 208 int discard_unit; /* 209 * discard command's offset/size should 210 * be aligned to this unit: block, 211 * segment or section 212 */ 213 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 214 block_t unusable_cap_perc; /* percentage for cap */ 215 block_t unusable_cap; /* Amount of space allowed to be 216 * unusable when disabling checkpoint 217 */ 218 219 /* For compression */ 220 unsigned char compress_algorithm; /* algorithm type */ 221 unsigned char compress_log_size; /* cluster log size */ 222 unsigned char compress_level; /* compress level */ 223 bool compress_chksum; /* compressed data chksum */ 224 unsigned char compress_ext_cnt; /* extension count */ 225 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 226 int compress_mode; /* compression mode */ 227 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 228 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 229 unsigned int lookup_mode; 230 }; 231 232 #define F2FS_FEATURE_ENCRYPT 0x00000001 233 #define F2FS_FEATURE_BLKZONED 0x00000002 234 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 235 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 236 #define F2FS_FEATURE_PRJQUOTA 0x00000010 237 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 238 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 239 #define F2FS_FEATURE_QUOTA_INO 0x00000080 240 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 241 #define F2FS_FEATURE_LOST_FOUND 0x00000200 242 #define F2FS_FEATURE_VERITY 0x00000400 243 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 244 #define F2FS_FEATURE_CASEFOLD 0x00001000 245 #define F2FS_FEATURE_COMPRESSION 0x00002000 246 #define F2FS_FEATURE_RO 0x00004000 247 #define F2FS_FEATURE_DEVICE_ALIAS 0x00008000 248 249 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 250 ((raw_super->feature & cpu_to_le32(mask)) != 0) 251 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 252 253 /* 254 * Default values for user and/or group using reserved blocks 255 */ 256 #define F2FS_DEF_RESUID 0 257 #define F2FS_DEF_RESGID 0 258 259 /* 260 * For checkpoint manager 261 */ 262 enum { 263 NAT_BITMAP, 264 SIT_BITMAP 265 }; 266 267 #define CP_UMOUNT 0x00000001 268 #define CP_FASTBOOT 0x00000002 269 #define CP_SYNC 0x00000004 270 #define CP_RECOVERY 0x00000008 271 #define CP_DISCARD 0x00000010 272 #define CP_TRIMMED 0x00000020 273 #define CP_PAUSE 0x00000040 274 #define CP_RESIZE 0x00000080 275 276 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 277 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 278 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 279 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 280 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 281 #define DEF_CP_INTERVAL 60 /* 60 secs */ 282 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 283 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 284 #define DEF_ENABLE_INTERVAL 16 /* 16 secs */ 285 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 286 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 287 288 enum cp_time { 289 CP_TIME_START, /* begin */ 290 CP_TIME_LOCK, /* after cp_global_sem */ 291 CP_TIME_OP_LOCK, /* after block_operation */ 292 CP_TIME_FLUSH_META, /* after flush sit/nat */ 293 CP_TIME_SYNC_META, /* after sync_meta_pages */ 294 CP_TIME_SYNC_CP_META, /* after sync cp meta pages */ 295 CP_TIME_WAIT_DIRTY_META,/* after wait on dirty meta */ 296 CP_TIME_WAIT_CP_DATA, /* after wait on cp data */ 297 CP_TIME_FLUSH_DEVICE, /* after flush device cache */ 298 CP_TIME_WAIT_LAST_CP, /* after wait on last cp pack */ 299 CP_TIME_END, /* after unblock_operation */ 300 CP_TIME_MAX, 301 }; 302 303 /* time cost stats of checkpoint */ 304 struct cp_stats { 305 ktime_t times[CP_TIME_MAX]; 306 }; 307 308 struct cp_control { 309 int reason; 310 __u64 trim_start; 311 __u64 trim_end; 312 __u64 trim_minlen; 313 struct cp_stats stats; 314 }; 315 316 /* 317 * indicate meta/data type 318 */ 319 enum { 320 META_CP, 321 META_NAT, 322 META_SIT, 323 META_SSA, 324 META_MAX, 325 META_POR, 326 DATA_GENERIC, /* check range only */ 327 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 328 DATA_GENERIC_ENHANCE_READ, /* 329 * strong check on range and segment 330 * bitmap but no warning due to race 331 * condition of read on truncated area 332 * by extent_cache 333 */ 334 DATA_GENERIC_ENHANCE_UPDATE, /* 335 * strong check on range and segment 336 * bitmap for update case 337 */ 338 META_GENERIC, 339 }; 340 341 /* for the list of ino */ 342 enum { 343 ORPHAN_INO, /* for orphan ino list */ 344 APPEND_INO, /* for append ino list */ 345 UPDATE_INO, /* for update ino list */ 346 TRANS_DIR_INO, /* for transactions dir ino list */ 347 XATTR_DIR_INO, /* for xattr updated dir ino list */ 348 FLUSH_INO, /* for multiple device flushing */ 349 MAX_INO_ENTRY, /* max. list */ 350 }; 351 352 struct ino_entry { 353 struct list_head list; /* list head */ 354 nid_t ino; /* inode number */ 355 unsigned int dirty_device; /* dirty device bitmap */ 356 }; 357 358 /* for the list of inodes to be GCed */ 359 struct inode_entry { 360 struct list_head list; /* list head */ 361 struct inode *inode; /* vfs inode pointer */ 362 }; 363 364 struct fsync_node_entry { 365 struct list_head list; /* list head */ 366 struct folio *folio; /* warm node folio pointer */ 367 unsigned int seq_id; /* sequence id */ 368 }; 369 370 struct ckpt_req { 371 struct completion wait; /* completion for checkpoint done */ 372 struct llist_node llnode; /* llist_node to be linked in wait queue */ 373 int ret; /* return code of checkpoint */ 374 union { 375 ktime_t queue_time; /* request queued time */ 376 ktime_t delta_time; /* time in queue */ 377 }; 378 }; 379 380 struct ckpt_req_control { 381 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 382 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 383 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 384 atomic_t issued_ckpt; /* # of actually issued ckpts */ 385 atomic_t total_ckpt; /* # of total ckpts */ 386 atomic_t queued_ckpt; /* # of queued ckpts */ 387 struct llist_head issue_list; /* list for command issue */ 388 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 389 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 390 unsigned int peak_time; /* peak wait time in msec until now */ 391 }; 392 393 /* a time threshold that checkpoint was blocked for, unit: ms */ 394 #define CP_LONG_LATENCY_THRESHOLD 5000 395 396 /* for the bitmap indicate blocks to be discarded */ 397 struct discard_entry { 398 struct list_head list; /* list head */ 399 block_t start_blkaddr; /* start blockaddr of current segment */ 400 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 401 }; 402 403 /* minimum discard granularity, unit: block count */ 404 #define MIN_DISCARD_GRANULARITY 1 405 /* default discard granularity of inner discard thread, unit: block count */ 406 #define DEFAULT_DISCARD_GRANULARITY 16 407 /* default maximum discard granularity of ordered discard, unit: block count */ 408 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 409 410 /* max discard pend list number */ 411 #define MAX_PLIST_NUM 512 412 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 413 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 414 415 enum { 416 D_PREP, /* initial */ 417 D_PARTIAL, /* partially submitted */ 418 D_SUBMIT, /* all submitted */ 419 D_DONE, /* finished */ 420 }; 421 422 struct discard_info { 423 block_t lstart; /* logical start address */ 424 block_t len; /* length */ 425 block_t start; /* actual start address in dev */ 426 }; 427 428 struct discard_cmd { 429 struct rb_node rb_node; /* rb node located in rb-tree */ 430 struct discard_info di; /* discard info */ 431 struct list_head list; /* command list */ 432 struct completion wait; /* completion */ 433 struct block_device *bdev; /* bdev */ 434 unsigned short ref; /* reference count */ 435 unsigned char state; /* state */ 436 unsigned char queued; /* queued discard */ 437 int error; /* bio error */ 438 spinlock_t lock; /* for state/bio_ref updating */ 439 unsigned short bio_ref; /* bio reference count */ 440 }; 441 442 enum { 443 DPOLICY_BG, 444 DPOLICY_FORCE, 445 DPOLICY_FSTRIM, 446 DPOLICY_UMOUNT, 447 MAX_DPOLICY, 448 }; 449 450 enum { 451 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 452 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 453 DPOLICY_IO_AWARE_MAX, 454 }; 455 456 struct discard_policy { 457 int type; /* type of discard */ 458 unsigned int min_interval; /* used for candidates exist */ 459 unsigned int mid_interval; /* used for device busy */ 460 unsigned int max_interval; /* used for candidates not exist */ 461 unsigned int max_requests; /* # of discards issued per round */ 462 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 463 bool io_aware; /* issue discard in idle time */ 464 bool sync; /* submit discard with REQ_SYNC flag */ 465 bool ordered; /* issue discard by lba order */ 466 bool timeout; /* discard timeout for put_super */ 467 unsigned int granularity; /* discard granularity */ 468 }; 469 470 struct discard_cmd_control { 471 struct task_struct *f2fs_issue_discard; /* discard thread */ 472 struct list_head entry_list; /* 4KB discard entry list */ 473 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 474 struct list_head wait_list; /* store on-flushing entries */ 475 struct list_head fstrim_list; /* in-flight discard from fstrim */ 476 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 477 struct mutex cmd_lock; 478 unsigned int nr_discards; /* # of discards in the list */ 479 unsigned int max_discards; /* max. discards to be issued */ 480 unsigned int max_discard_request; /* max. discard request per round */ 481 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 482 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 483 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 484 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 485 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 486 unsigned int discard_granularity; /* discard granularity */ 487 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 488 unsigned int discard_io_aware; /* io_aware policy */ 489 unsigned int undiscard_blks; /* # of undiscard blocks */ 490 unsigned int next_pos; /* next discard position */ 491 atomic_t issued_discard; /* # of issued discard */ 492 atomic_t queued_discard; /* # of queued discard */ 493 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 494 struct rb_root_cached root; /* root of discard rb-tree */ 495 bool rbtree_check; /* config for consistence check */ 496 bool discard_wake; /* to wake up discard thread */ 497 }; 498 499 /* for the list of fsync inodes, used only during recovery */ 500 struct fsync_inode_entry { 501 struct list_head list; /* list head */ 502 struct inode *inode; /* vfs inode pointer */ 503 block_t blkaddr; /* block address locating the last fsync */ 504 block_t last_dentry; /* block address locating the last dentry */ 505 }; 506 507 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 508 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 509 510 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 511 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 512 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 513 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 514 515 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 516 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 517 518 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 519 { 520 int before = nats_in_cursum(journal); 521 522 journal->n_nats = cpu_to_le16(before + i); 523 return before; 524 } 525 526 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 527 { 528 int before = sits_in_cursum(journal); 529 530 journal->n_sits = cpu_to_le16(before + i); 531 return before; 532 } 533 534 static inline bool __has_cursum_space(struct f2fs_journal *journal, 535 int size, int type) 536 { 537 if (type == NAT_JOURNAL) 538 return size <= MAX_NAT_JENTRIES(journal); 539 return size <= MAX_SIT_JENTRIES(journal); 540 } 541 542 /* for inline stuff */ 543 #define DEF_INLINE_RESERVED_SIZE 1 544 static inline int get_extra_isize(struct inode *inode); 545 static inline int get_inline_xattr_addrs(struct inode *inode); 546 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 547 (CUR_ADDRS_PER_INODE(inode) - \ 548 get_inline_xattr_addrs(inode) - \ 549 DEF_INLINE_RESERVED_SIZE)) 550 551 /* for inline dir */ 552 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 553 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 554 BITS_PER_BYTE + 1)) 555 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 556 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 557 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 558 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 559 NR_INLINE_DENTRY(inode) + \ 560 INLINE_DENTRY_BITMAP_SIZE(inode))) 561 562 /* 563 * For INODE and NODE manager 564 */ 565 /* for directory operations */ 566 567 struct f2fs_filename { 568 /* 569 * The filename the user specified. This is NULL for some 570 * filesystem-internal operations, e.g. converting an inline directory 571 * to a non-inline one, or roll-forward recovering an encrypted dentry. 572 */ 573 const struct qstr *usr_fname; 574 575 /* 576 * The on-disk filename. For encrypted directories, this is encrypted. 577 * This may be NULL for lookups in an encrypted dir without the key. 578 */ 579 struct fscrypt_str disk_name; 580 581 /* The dirhash of this filename */ 582 f2fs_hash_t hash; 583 584 #ifdef CONFIG_FS_ENCRYPTION 585 /* 586 * For lookups in encrypted directories: either the buffer backing 587 * disk_name, or a buffer that holds the decoded no-key name. 588 */ 589 struct fscrypt_str crypto_buf; 590 #endif 591 #if IS_ENABLED(CONFIG_UNICODE) 592 /* 593 * For casefolded directories: the casefolded name, but it's left NULL 594 * if the original name is not valid Unicode, if the original name is 595 * "." or "..", if the directory is both casefolded and encrypted and 596 * its encryption key is unavailable, or if the filesystem is doing an 597 * internal operation where usr_fname is also NULL. In all these cases 598 * we fall back to treating the name as an opaque byte sequence. 599 */ 600 struct qstr cf_name; 601 #endif 602 }; 603 604 struct f2fs_dentry_ptr { 605 struct inode *inode; 606 void *bitmap; 607 struct f2fs_dir_entry *dentry; 608 __u8 (*filename)[F2FS_SLOT_LEN]; 609 int max; 610 int nr_bitmap; 611 }; 612 613 static inline void make_dentry_ptr_block(struct inode *inode, 614 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 615 { 616 d->inode = inode; 617 d->max = NR_DENTRY_IN_BLOCK; 618 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 619 d->bitmap = t->dentry_bitmap; 620 d->dentry = t->dentry; 621 d->filename = t->filename; 622 } 623 624 static inline void make_dentry_ptr_inline(struct inode *inode, 625 struct f2fs_dentry_ptr *d, void *t) 626 { 627 int entry_cnt = NR_INLINE_DENTRY(inode); 628 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 629 int reserved_size = INLINE_RESERVED_SIZE(inode); 630 631 d->inode = inode; 632 d->max = entry_cnt; 633 d->nr_bitmap = bitmap_size; 634 d->bitmap = t; 635 d->dentry = t + bitmap_size + reserved_size; 636 d->filename = t + bitmap_size + reserved_size + 637 SIZE_OF_DIR_ENTRY * entry_cnt; 638 } 639 640 /* 641 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 642 * as its node offset to distinguish from index node blocks. 643 * But some bits are used to mark the node block. 644 */ 645 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 646 >> OFFSET_BIT_SHIFT) 647 enum { 648 ALLOC_NODE, /* allocate a new node page if needed */ 649 LOOKUP_NODE, /* look up a node without readahead */ 650 LOOKUP_NODE_RA, /* 651 * look up a node with readahead called 652 * by get_data_block. 653 */ 654 }; 655 656 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 657 658 /* congestion wait timeout value, default: 20ms */ 659 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 660 661 /* timeout value injected, default: 1000ms */ 662 #define DEFAULT_FAULT_TIMEOUT (msecs_to_jiffies(1000)) 663 664 /* maximum retry quota flush count */ 665 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 666 667 /* maximum retry of EIO'ed page */ 668 #define MAX_RETRY_PAGE_EIO 100 669 670 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 671 672 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 673 674 /* dirty segments threshold for triggering CP */ 675 #define DEFAULT_DIRTY_THRESHOLD 4 676 677 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 678 #define RECOVERY_MIN_RA_BLOCKS 1 679 680 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 681 682 /* for in-memory extent cache entry */ 683 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 684 685 /* number of extent info in extent cache we try to shrink */ 686 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 687 688 /* number of age extent info in extent cache we try to shrink */ 689 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 690 #define LAST_AGE_WEIGHT 30 691 #define SAME_AGE_REGION 1024 692 693 /* 694 * Define data block with age less than 1GB as hot data 695 * define data block with age less than 10GB but more than 1GB as warm data 696 */ 697 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 698 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 699 700 /* default max read extent count per inode */ 701 #define DEF_MAX_READ_EXTENT_COUNT 10240 702 703 /* extent cache type */ 704 enum extent_type { 705 EX_READ, 706 EX_BLOCK_AGE, 707 NR_EXTENT_CACHES, 708 }; 709 710 struct extent_info { 711 unsigned int fofs; /* start offset in a file */ 712 unsigned int len; /* length of the extent */ 713 union { 714 /* read extent_cache */ 715 struct { 716 /* start block address of the extent */ 717 block_t blk; 718 #ifdef CONFIG_F2FS_FS_COMPRESSION 719 /* physical extent length of compressed blocks */ 720 unsigned int c_len; 721 #endif 722 }; 723 /* block age extent_cache */ 724 struct { 725 /* block age of the extent */ 726 unsigned long long age; 727 /* last total blocks allocated */ 728 unsigned long long last_blocks; 729 }; 730 }; 731 }; 732 733 struct extent_node { 734 struct rb_node rb_node; /* rb node located in rb-tree */ 735 struct extent_info ei; /* extent info */ 736 struct list_head list; /* node in global extent list of sbi */ 737 struct extent_tree *et; /* extent tree pointer */ 738 }; 739 740 struct extent_tree { 741 nid_t ino; /* inode number */ 742 enum extent_type type; /* keep the extent tree type */ 743 struct rb_root_cached root; /* root of extent info rb-tree */ 744 struct extent_node *cached_en; /* recently accessed extent node */ 745 struct list_head list; /* to be used by sbi->zombie_list */ 746 rwlock_t lock; /* protect extent info rb-tree */ 747 atomic_t node_cnt; /* # of extent node in rb-tree*/ 748 bool largest_updated; /* largest extent updated */ 749 struct extent_info largest; /* largest cached extent for EX_READ */ 750 }; 751 752 struct extent_tree_info { 753 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 754 struct mutex extent_tree_lock; /* locking extent radix tree */ 755 struct list_head extent_list; /* lru list for shrinker */ 756 spinlock_t extent_lock; /* locking extent lru list */ 757 atomic_t total_ext_tree; /* extent tree count */ 758 struct list_head zombie_list; /* extent zombie tree list */ 759 atomic_t total_zombie_tree; /* extent zombie tree count */ 760 atomic_t total_ext_node; /* extent info count */ 761 }; 762 763 /* 764 * State of block returned by f2fs_map_blocks. 765 */ 766 #define F2FS_MAP_NEW (1U << 0) 767 #define F2FS_MAP_MAPPED (1U << 1) 768 #define F2FS_MAP_DELALLOC (1U << 2) 769 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 770 F2FS_MAP_DELALLOC) 771 772 struct f2fs_map_blocks { 773 struct block_device *m_bdev; /* for multi-device dio */ 774 block_t m_pblk; 775 block_t m_lblk; 776 unsigned int m_len; 777 unsigned int m_flags; 778 unsigned long m_last_pblk; /* last allocated block, only used for DIO in LFS mode */ 779 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 780 pgoff_t *m_next_extent; /* point to next possible extent */ 781 int m_seg_type; 782 bool m_may_create; /* indicate it is from write path */ 783 bool m_multidev_dio; /* indicate it allows multi-device dio */ 784 }; 785 786 /* for flag in get_data_block */ 787 enum { 788 F2FS_GET_BLOCK_DEFAULT, 789 F2FS_GET_BLOCK_FIEMAP, 790 F2FS_GET_BLOCK_BMAP, 791 F2FS_GET_BLOCK_DIO, 792 F2FS_GET_BLOCK_PRE_DIO, 793 F2FS_GET_BLOCK_PRE_AIO, 794 F2FS_GET_BLOCK_PRECACHE, 795 }; 796 797 /* 798 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 799 */ 800 #define FADVISE_COLD_BIT 0x01 801 #define FADVISE_LOST_PINO_BIT 0x02 802 #define FADVISE_ENCRYPT_BIT 0x04 803 #define FADVISE_ENC_NAME_BIT 0x08 804 #define FADVISE_KEEP_SIZE_BIT 0x10 805 #define FADVISE_HOT_BIT 0x20 806 #define FADVISE_VERITY_BIT 0x40 807 #define FADVISE_TRUNC_BIT 0x80 808 809 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 810 811 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 812 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 813 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 814 815 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 816 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 817 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 818 819 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 820 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 821 822 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 823 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 824 825 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 826 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 827 828 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 829 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 830 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 831 832 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 833 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 834 835 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 836 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 837 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 838 839 #define DEF_DIR_LEVEL 0 840 841 /* used for f2fs_inode_info->flags */ 842 enum { 843 FI_NEW_INODE, /* indicate newly allocated inode */ 844 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 845 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 846 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 847 FI_INC_LINK, /* need to increment i_nlink */ 848 FI_ACL_MODE, /* indicate acl mode */ 849 FI_NO_ALLOC, /* should not allocate any blocks */ 850 FI_FREE_NID, /* free allocated nide */ 851 FI_NO_EXTENT, /* not to use the extent cache */ 852 FI_INLINE_XATTR, /* used for inline xattr */ 853 FI_INLINE_DATA, /* used for inline data*/ 854 FI_INLINE_DENTRY, /* used for inline dentry */ 855 FI_APPEND_WRITE, /* inode has appended data */ 856 FI_UPDATE_WRITE, /* inode has in-place-update data */ 857 FI_NEED_IPU, /* used for ipu per file */ 858 FI_ATOMIC_FILE, /* indicate atomic file */ 859 FI_DATA_EXIST, /* indicate data exists */ 860 FI_SKIP_WRITES, /* should skip data page writeback */ 861 FI_OPU_WRITE, /* used for opu per file */ 862 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 863 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 864 FI_HOT_DATA, /* indicate file is hot */ 865 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 866 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 867 FI_PIN_FILE, /* indicate file should not be gced */ 868 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 869 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 870 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 871 FI_MMAP_FILE, /* indicate file was mmapped */ 872 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 873 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 874 FI_ALIGNED_WRITE, /* enable aligned write */ 875 FI_COW_FILE, /* indicate COW file */ 876 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 877 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 878 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 879 FI_OPENED_FILE, /* indicate file has been opened */ 880 FI_DONATE_FINISHED, /* indicate page donation of file has been finished */ 881 FI_MAX, /* max flag, never be used */ 882 }; 883 884 struct f2fs_inode_info { 885 struct inode vfs_inode; /* serve a vfs inode */ 886 unsigned long i_flags; /* keep an inode flags for ioctl */ 887 unsigned char i_advise; /* use to give file attribute hints */ 888 unsigned char i_dir_level; /* use for dentry level for large dir */ 889 union { 890 unsigned int i_current_depth; /* only for directory depth */ 891 unsigned short i_gc_failures; /* for gc failure statistic */ 892 }; 893 unsigned int i_pino; /* parent inode number */ 894 umode_t i_acl_mode; /* keep file acl mode temporarily */ 895 896 /* Use below internally in f2fs*/ 897 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 898 unsigned int ioprio_hint; /* hint for IO priority */ 899 struct f2fs_rwsem i_sem; /* protect fi info */ 900 atomic_t dirty_pages; /* # of dirty pages */ 901 f2fs_hash_t chash; /* hash value of given file name */ 902 unsigned int clevel; /* maximum level of given file name */ 903 struct task_struct *task; /* lookup and create consistency */ 904 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 905 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 906 nid_t i_xattr_nid; /* node id that contains xattrs */ 907 loff_t last_disk_size; /* lastly written file size */ 908 spinlock_t i_size_lock; /* protect last_disk_size */ 909 910 #ifdef CONFIG_QUOTA 911 struct dquot __rcu *i_dquot[MAXQUOTAS]; 912 913 /* quota space reservation, managed internally by quota code */ 914 qsize_t i_reserved_quota; 915 #endif 916 struct list_head dirty_list; /* dirty list for dirs and files */ 917 struct list_head gdirty_list; /* linked in global dirty list */ 918 919 /* linked in global inode list for cache donation */ 920 struct list_head gdonate_list; 921 pgoff_t donate_start, donate_end; /* inclusive */ 922 atomic_t open_count; /* # of open files */ 923 924 struct task_struct *atomic_write_task; /* store atomic write task */ 925 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 926 /* cached extent_tree entry */ 927 union { 928 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 929 struct inode *atomic_inode; 930 /* point to atomic_inode, available only for cow_inode */ 931 }; 932 933 /* avoid racing between foreground op and gc */ 934 struct f2fs_rwsem i_gc_rwsem[2]; 935 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 936 937 int i_extra_isize; /* size of extra space located in i_addr */ 938 kprojid_t i_projid; /* id for project quota */ 939 int i_inline_xattr_size; /* inline xattr size */ 940 struct timespec64 i_crtime; /* inode creation time */ 941 struct timespec64 i_disk_time[3];/* inode disk times */ 942 943 /* for file compress */ 944 atomic_t i_compr_blocks; /* # of compressed blocks */ 945 unsigned char i_compress_algorithm; /* algorithm type */ 946 unsigned char i_log_cluster_size; /* log of cluster size */ 947 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 948 unsigned char i_compress_flag; /* compress flag */ 949 unsigned int i_cluster_size; /* cluster size */ 950 951 unsigned int atomic_write_cnt; 952 loff_t original_i_size; /* original i_size before atomic write */ 953 #ifdef CONFIG_FS_ENCRYPTION 954 struct fscrypt_inode_info *i_crypt_info; /* filesystem encryption info */ 955 #endif 956 #ifdef CONFIG_FS_VERITY 957 struct fsverity_info *i_verity_info; /* filesystem verity info */ 958 #endif 959 }; 960 961 static inline void get_read_extent_info(struct extent_info *ext, 962 struct f2fs_extent *i_ext) 963 { 964 ext->fofs = le32_to_cpu(i_ext->fofs); 965 ext->blk = le32_to_cpu(i_ext->blk); 966 ext->len = le32_to_cpu(i_ext->len); 967 } 968 969 static inline void set_raw_read_extent(struct extent_info *ext, 970 struct f2fs_extent *i_ext) 971 { 972 i_ext->fofs = cpu_to_le32(ext->fofs); 973 i_ext->blk = cpu_to_le32(ext->blk); 974 i_ext->len = cpu_to_le32(ext->len); 975 } 976 977 static inline bool __is_discard_mergeable(struct discard_info *back, 978 struct discard_info *front, unsigned int max_len) 979 { 980 return (back->lstart + back->len == front->lstart) && 981 (back->len + front->len <= max_len); 982 } 983 984 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 985 struct discard_info *back, unsigned int max_len) 986 { 987 return __is_discard_mergeable(back, cur, max_len); 988 } 989 990 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 991 struct discard_info *front, unsigned int max_len) 992 { 993 return __is_discard_mergeable(cur, front, max_len); 994 } 995 996 /* 997 * For free nid management 998 */ 999 enum nid_state { 1000 FREE_NID, /* newly added to free nid list */ 1001 PREALLOC_NID, /* it is preallocated */ 1002 MAX_NID_STATE, 1003 }; 1004 1005 enum nat_state { 1006 TOTAL_NAT, 1007 DIRTY_NAT, 1008 RECLAIMABLE_NAT, 1009 MAX_NAT_STATE, 1010 }; 1011 1012 struct f2fs_nm_info { 1013 block_t nat_blkaddr; /* base disk address of NAT */ 1014 nid_t max_nid; /* maximum possible node ids */ 1015 nid_t available_nids; /* # of available node ids */ 1016 nid_t next_scan_nid; /* the next nid to be scanned */ 1017 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 1018 unsigned int ram_thresh; /* control the memory footprint */ 1019 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 1020 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 1021 1022 /* NAT cache management */ 1023 struct radix_tree_root nat_root;/* root of the nat entry cache */ 1024 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 1025 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 1026 struct list_head nat_entries; /* cached nat entry list (clean) */ 1027 spinlock_t nat_list_lock; /* protect clean nat entry list */ 1028 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 1029 unsigned int nat_blocks; /* # of nat blocks */ 1030 1031 /* free node ids management */ 1032 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 1033 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 1034 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 1035 spinlock_t nid_list_lock; /* protect nid lists ops */ 1036 struct mutex build_lock; /* lock for build free nids */ 1037 unsigned char **free_nid_bitmap; 1038 unsigned char *nat_block_bitmap; 1039 unsigned short *free_nid_count; /* free nid count of NAT block */ 1040 1041 /* for checkpoint */ 1042 char *nat_bitmap; /* NAT bitmap pointer */ 1043 1044 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 1045 unsigned char *nat_bits; /* NAT bits blocks */ 1046 unsigned char *full_nat_bits; /* full NAT pages */ 1047 unsigned char *empty_nat_bits; /* empty NAT pages */ 1048 #ifdef CONFIG_F2FS_CHECK_FS 1049 char *nat_bitmap_mir; /* NAT bitmap mirror */ 1050 #endif 1051 int bitmap_size; /* bitmap size */ 1052 }; 1053 1054 /* 1055 * this structure is used as one of function parameters. 1056 * all the information are dedicated to a given direct node block determined 1057 * by the data offset in a file. 1058 */ 1059 struct dnode_of_data { 1060 struct inode *inode; /* vfs inode pointer */ 1061 struct folio *inode_folio; /* its inode folio, NULL is possible */ 1062 struct folio *node_folio; /* cached direct node folio */ 1063 nid_t nid; /* node id of the direct node block */ 1064 unsigned int ofs_in_node; /* data offset in the node page */ 1065 bool inode_folio_locked; /* inode folio is locked or not */ 1066 bool node_changed; /* is node block changed */ 1067 char cur_level; /* level of hole node page */ 1068 char max_level; /* level of current page located */ 1069 block_t data_blkaddr; /* block address of the node block */ 1070 }; 1071 1072 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 1073 struct folio *ifolio, struct folio *nfolio, nid_t nid) 1074 { 1075 memset(dn, 0, sizeof(*dn)); 1076 dn->inode = inode; 1077 dn->inode_folio = ifolio; 1078 dn->node_folio = nfolio; 1079 dn->nid = nid; 1080 } 1081 1082 /* 1083 * For SIT manager 1084 * 1085 * By default, there are 6 active log areas across the whole main area. 1086 * When considering hot and cold data separation to reduce cleaning overhead, 1087 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1088 * respectively. 1089 * In the current design, you should not change the numbers intentionally. 1090 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1091 * logs individually according to the underlying devices. (default: 6) 1092 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1093 * data and 8 for node logs. 1094 */ 1095 #define NR_CURSEG_DATA_TYPE (3) 1096 #define NR_CURSEG_NODE_TYPE (3) 1097 #define NR_CURSEG_INMEM_TYPE (2) 1098 #define NR_CURSEG_RO_TYPE (2) 1099 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1100 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1101 1102 enum log_type { 1103 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1104 CURSEG_WARM_DATA, /* data blocks */ 1105 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1106 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1107 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1108 CURSEG_COLD_NODE, /* indirect node blocks */ 1109 NR_PERSISTENT_LOG, /* number of persistent log */ 1110 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1111 /* pinned file that needs consecutive block address */ 1112 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1113 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1114 }; 1115 1116 struct flush_cmd { 1117 struct completion wait; 1118 struct llist_node llnode; 1119 nid_t ino; 1120 int ret; 1121 }; 1122 1123 struct flush_cmd_control { 1124 struct task_struct *f2fs_issue_flush; /* flush thread */ 1125 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1126 atomic_t issued_flush; /* # of issued flushes */ 1127 atomic_t queued_flush; /* # of queued flushes */ 1128 struct llist_head issue_list; /* list for command issue */ 1129 struct llist_node *dispatch_list; /* list for command dispatch */ 1130 }; 1131 1132 struct f2fs_sm_info { 1133 struct sit_info *sit_info; /* whole segment information */ 1134 struct free_segmap_info *free_info; /* free segment information */ 1135 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1136 struct curseg_info *curseg_array; /* active segment information */ 1137 1138 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1139 1140 block_t seg0_blkaddr; /* block address of 0'th segment */ 1141 block_t main_blkaddr; /* start block address of main area */ 1142 block_t ssa_blkaddr; /* start block address of SSA area */ 1143 1144 unsigned int segment_count; /* total # of segments */ 1145 unsigned int main_segments; /* # of segments in main area */ 1146 unsigned int reserved_segments; /* # of reserved segments */ 1147 unsigned int ovp_segments; /* # of overprovision segments */ 1148 1149 /* a threshold to reclaim prefree segments */ 1150 unsigned int rec_prefree_segments; 1151 1152 struct list_head sit_entry_set; /* sit entry set list */ 1153 1154 unsigned int ipu_policy; /* in-place-update policy */ 1155 unsigned int min_ipu_util; /* in-place-update threshold */ 1156 unsigned int min_fsync_blocks; /* threshold for fsync */ 1157 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1158 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1159 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1160 1161 /* for flush command control */ 1162 struct flush_cmd_control *fcc_info; 1163 1164 /* for discard command control */ 1165 struct discard_cmd_control *dcc_info; 1166 }; 1167 1168 /* 1169 * For superblock 1170 */ 1171 /* 1172 * COUNT_TYPE for monitoring 1173 * 1174 * f2fs monitors the number of several block types such as on-writeback, 1175 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1176 */ 1177 #define WB_DATA_TYPE(folio, f) \ 1178 (f || f2fs_is_cp_guaranteed(folio) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1179 enum count_type { 1180 F2FS_DIRTY_DENTS, 1181 F2FS_DIRTY_DATA, 1182 F2FS_DIRTY_QDATA, 1183 F2FS_DIRTY_NODES, 1184 F2FS_DIRTY_META, 1185 F2FS_DIRTY_IMETA, 1186 F2FS_WB_CP_DATA, 1187 F2FS_WB_DATA, 1188 F2FS_RD_DATA, 1189 F2FS_RD_NODE, 1190 F2FS_RD_META, 1191 F2FS_DIO_WRITE, 1192 F2FS_DIO_READ, 1193 NR_COUNT_TYPE, 1194 }; 1195 1196 /* 1197 * The below are the page types of bios used in submit_bio(). 1198 * The available types are: 1199 * DATA User data pages. It operates as async mode. 1200 * NODE Node pages. It operates as async mode. 1201 * META FS metadata pages such as SIT, NAT, CP. 1202 * NR_PAGE_TYPE The number of page types. 1203 * META_FLUSH Make sure the previous pages are written 1204 * with waiting the bio's completion 1205 * ... Only can be used with META. 1206 */ 1207 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1208 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1209 enum page_type { 1210 DATA = 0, 1211 NODE = 1, /* should not change this */ 1212 META, 1213 NR_PAGE_TYPE, 1214 META_FLUSH, 1215 IPU, /* the below types are used by tracepoints only. */ 1216 OPU, 1217 }; 1218 1219 enum temp_type { 1220 HOT = 0, /* must be zero for meta bio */ 1221 WARM, 1222 COLD, 1223 NR_TEMP_TYPE, 1224 }; 1225 1226 enum need_lock_type { 1227 LOCK_REQ = 0, 1228 LOCK_DONE, 1229 LOCK_RETRY, 1230 }; 1231 1232 enum cp_reason_type { 1233 CP_NO_NEEDED, 1234 CP_NON_REGULAR, 1235 CP_COMPRESSED, 1236 CP_HARDLINK, 1237 CP_SB_NEED_CP, 1238 CP_WRONG_PINO, 1239 CP_NO_SPC_ROLL, 1240 CP_NODE_NEED_CP, 1241 CP_FASTBOOT_MODE, 1242 CP_SPEC_LOG_NUM, 1243 CP_RECOVER_DIR, 1244 CP_XATTR_DIR, 1245 }; 1246 1247 enum iostat_type { 1248 /* WRITE IO */ 1249 APP_DIRECT_IO, /* app direct write IOs */ 1250 APP_BUFFERED_IO, /* app buffered write IOs */ 1251 APP_WRITE_IO, /* app write IOs */ 1252 APP_MAPPED_IO, /* app mapped IOs */ 1253 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1254 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1255 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1256 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1257 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1258 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1259 FS_GC_DATA_IO, /* data IOs from forground gc */ 1260 FS_GC_NODE_IO, /* node IOs from forground gc */ 1261 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1262 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1263 FS_CP_META_IO, /* meta IOs from checkpoint */ 1264 1265 /* READ IO */ 1266 APP_DIRECT_READ_IO, /* app direct read IOs */ 1267 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1268 APP_READ_IO, /* app read IOs */ 1269 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1270 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1271 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1272 FS_DATA_READ_IO, /* data read IOs */ 1273 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1274 FS_CDATA_READ_IO, /* compressed data read IOs */ 1275 FS_NODE_READ_IO, /* node read IOs */ 1276 FS_META_READ_IO, /* meta read IOs */ 1277 1278 /* other */ 1279 FS_DISCARD_IO, /* discard */ 1280 FS_FLUSH_IO, /* flush */ 1281 FS_ZONE_RESET_IO, /* zone reset */ 1282 NR_IO_TYPE, 1283 }; 1284 1285 struct f2fs_io_info { 1286 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1287 nid_t ino; /* inode number */ 1288 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1289 enum temp_type temp; /* contains HOT/WARM/COLD */ 1290 enum req_op op; /* contains REQ_OP_ */ 1291 blk_opf_t op_flags; /* req_flag_bits */ 1292 block_t new_blkaddr; /* new block address to be written */ 1293 block_t old_blkaddr; /* old block address before Cow */ 1294 union { 1295 struct page *page; /* page to be written */ 1296 struct folio *folio; 1297 }; 1298 struct page *encrypted_page; /* encrypted page */ 1299 struct page *compressed_page; /* compressed page */ 1300 struct list_head list; /* serialize IOs */ 1301 unsigned int compr_blocks; /* # of compressed block addresses */ 1302 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1303 unsigned int version:8; /* version of the node */ 1304 unsigned int submitted:1; /* indicate IO submission */ 1305 unsigned int in_list:1; /* indicate fio is in io_list */ 1306 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1307 unsigned int encrypted:1; /* indicate file is encrypted */ 1308 unsigned int meta_gc:1; /* require meta inode GC */ 1309 enum iostat_type io_type; /* io type */ 1310 struct writeback_control *io_wbc; /* writeback control */ 1311 struct bio **bio; /* bio for ipu */ 1312 sector_t *last_block; /* last block number in bio */ 1313 }; 1314 1315 struct bio_entry { 1316 struct bio *bio; 1317 struct list_head list; 1318 }; 1319 1320 #define is_read_io(rw) ((rw) == READ) 1321 struct f2fs_bio_info { 1322 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1323 struct bio *bio; /* bios to merge */ 1324 sector_t last_block_in_bio; /* last block number */ 1325 struct f2fs_io_info fio; /* store buffered io info. */ 1326 #ifdef CONFIG_BLK_DEV_ZONED 1327 struct completion zone_wait; /* condition value for the previous open zone to close */ 1328 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1329 void *bi_private; /* previous bi_private for pending bio */ 1330 #endif 1331 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1332 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1333 struct list_head io_list; /* track fios */ 1334 struct list_head bio_list; /* bio entry list head */ 1335 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1336 }; 1337 1338 #define FDEV(i) (sbi->devs[i]) 1339 #define RDEV(i) (raw_super->devs[i]) 1340 struct f2fs_dev_info { 1341 struct file *bdev_file; 1342 struct block_device *bdev; 1343 char path[MAX_PATH_LEN + 1]; 1344 unsigned int total_segments; 1345 block_t start_blk; 1346 block_t end_blk; 1347 #ifdef CONFIG_BLK_DEV_ZONED 1348 unsigned int nr_blkz; /* Total number of zones */ 1349 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1350 #endif 1351 }; 1352 1353 enum inode_type { 1354 DIR_INODE, /* for dirty dir inode */ 1355 FILE_INODE, /* for dirty regular/symlink inode */ 1356 DIRTY_META, /* for all dirtied inode metadata */ 1357 DONATE_INODE, /* for all inode to donate pages */ 1358 NR_INODE_TYPE, 1359 }; 1360 1361 /* for inner inode cache management */ 1362 struct inode_management { 1363 struct radix_tree_root ino_root; /* ino entry array */ 1364 spinlock_t ino_lock; /* for ino entry lock */ 1365 struct list_head ino_list; /* inode list head */ 1366 unsigned long ino_num; /* number of entries */ 1367 }; 1368 1369 /* for GC_AT */ 1370 struct atgc_management { 1371 bool atgc_enabled; /* ATGC is enabled or not */ 1372 struct rb_root_cached root; /* root of victim rb-tree */ 1373 struct list_head victim_list; /* linked with all victim entries */ 1374 unsigned int victim_count; /* victim count in rb-tree */ 1375 unsigned int candidate_ratio; /* candidate ratio */ 1376 unsigned int max_candidate_count; /* max candidate count */ 1377 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1378 unsigned long long age_threshold; /* age threshold */ 1379 }; 1380 1381 struct f2fs_gc_control { 1382 unsigned int victim_segno; /* target victim segment number */ 1383 int init_gc_type; /* FG_GC or BG_GC */ 1384 bool no_bg_gc; /* check the space and stop bg_gc */ 1385 bool should_migrate_blocks; /* should migrate blocks */ 1386 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1387 bool one_time; /* require one time GC in one migration unit */ 1388 unsigned int nr_free_secs; /* # of free sections to do GC */ 1389 }; 1390 1391 /* 1392 * For s_flag in struct f2fs_sb_info 1393 * Modification on enum should be synchronized with s_flag array 1394 */ 1395 enum { 1396 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1397 SBI_IS_CLOSE, /* specify unmounting */ 1398 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1399 SBI_POR_DOING, /* recovery is doing or not */ 1400 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1401 SBI_NEED_CP, /* need to checkpoint */ 1402 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1403 SBI_IS_RECOVERED, /* recovered orphan/data */ 1404 SBI_CP_DISABLED, /* CP was disabled last mount */ 1405 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1406 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1407 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1408 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1409 SBI_IS_RESIZEFS, /* resizefs is in process */ 1410 SBI_IS_FREEZING, /* freezefs is in process */ 1411 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1412 MAX_SBI_FLAG, 1413 }; 1414 1415 enum { 1416 CP_TIME, 1417 REQ_TIME, 1418 DISCARD_TIME, 1419 GC_TIME, 1420 DISABLE_TIME, 1421 ENABLE_TIME, 1422 UMOUNT_DISCARD_TIMEOUT, 1423 MAX_TIME, 1424 }; 1425 1426 /* Note that you need to keep synchronization with this gc_mode_names array */ 1427 enum { 1428 GC_NORMAL, 1429 GC_IDLE_CB, 1430 GC_IDLE_GREEDY, 1431 GC_IDLE_AT, 1432 GC_URGENT_HIGH, 1433 GC_URGENT_LOW, 1434 GC_URGENT_MID, 1435 MAX_GC_MODE, 1436 }; 1437 1438 enum { 1439 BGGC_MODE_ON, /* background gc is on */ 1440 BGGC_MODE_OFF, /* background gc is off */ 1441 BGGC_MODE_SYNC, /* 1442 * background gc is on, migrating blocks 1443 * like foreground gc 1444 */ 1445 }; 1446 1447 enum { 1448 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1449 FS_MODE_LFS, /* use lfs allocation only */ 1450 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1451 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1452 }; 1453 1454 enum { 1455 ALLOC_MODE_DEFAULT, /* stay default */ 1456 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1457 }; 1458 1459 enum fsync_mode { 1460 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1461 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1462 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1463 }; 1464 1465 enum { 1466 COMPR_MODE_FS, /* 1467 * automatically compress compression 1468 * enabled files 1469 */ 1470 COMPR_MODE_USER, /* 1471 * automatical compression is disabled. 1472 * user can control the file compression 1473 * using ioctls 1474 */ 1475 }; 1476 1477 enum { 1478 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1479 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1480 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1481 }; 1482 1483 enum { 1484 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1485 MEMORY_MODE_LOW, /* memory mode for low memory devices */ 1486 }; 1487 1488 enum errors_option { 1489 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1490 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1491 MOUNT_ERRORS_PANIC, /* panic on errors */ 1492 }; 1493 1494 enum { 1495 BACKGROUND, 1496 FOREGROUND, 1497 MAX_CALL_TYPE, 1498 TOTAL_CALL = FOREGROUND, 1499 }; 1500 1501 enum f2fs_lookup_mode { 1502 LOOKUP_PERF, 1503 LOOKUP_COMPAT, 1504 LOOKUP_AUTO, 1505 }; 1506 1507 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1508 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1509 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1510 1511 /* 1512 * Layout of f2fs page.private: 1513 * 1514 * Layout A: lowest bit should be 1 1515 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1516 * bit 0 PAGE_PRIVATE_NOT_POINTER 1517 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1518 * bit 2 PAGE_PRIVATE_INLINE_INODE 1519 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1520 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1521 * bit 5- f2fs private data 1522 * 1523 * Layout B: lowest bit should be 0 1524 * page.private is a wrapped pointer. 1525 */ 1526 enum { 1527 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1528 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1529 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1530 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1531 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1532 PAGE_PRIVATE_MAX 1533 }; 1534 1535 /* For compression */ 1536 enum compress_algorithm_type { 1537 COMPRESS_LZO, 1538 COMPRESS_LZ4, 1539 COMPRESS_ZSTD, 1540 COMPRESS_LZORLE, 1541 COMPRESS_MAX, 1542 }; 1543 1544 enum compress_flag { 1545 COMPRESS_CHKSUM, 1546 COMPRESS_MAX_FLAG, 1547 }; 1548 1549 #define COMPRESS_WATERMARK 20 1550 #define COMPRESS_PERCENT 20 1551 1552 #define COMPRESS_DATA_RESERVED_SIZE 4 1553 struct compress_data { 1554 __le32 clen; /* compressed data size */ 1555 __le32 chksum; /* compressed data checksum */ 1556 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1557 u8 cdata[]; /* compressed data */ 1558 }; 1559 1560 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1561 1562 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1563 1564 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1565 1566 #define COMPRESS_LEVEL_OFFSET 8 1567 1568 /* compress context */ 1569 struct compress_ctx { 1570 struct inode *inode; /* inode the context belong to */ 1571 pgoff_t cluster_idx; /* cluster index number */ 1572 unsigned int cluster_size; /* page count in cluster */ 1573 unsigned int log_cluster_size; /* log of cluster size */ 1574 struct page **rpages; /* pages store raw data in cluster */ 1575 unsigned int nr_rpages; /* total page number in rpages */ 1576 struct page **cpages; /* pages store compressed data in cluster */ 1577 unsigned int nr_cpages; /* total page number in cpages */ 1578 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1579 void *rbuf; /* virtual mapped address on rpages */ 1580 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1581 size_t rlen; /* valid data length in rbuf */ 1582 size_t clen; /* valid data length in cbuf */ 1583 void *private; /* payload buffer for specified compression algorithm */ 1584 void *private2; /* extra payload buffer */ 1585 }; 1586 1587 /* compress context for write IO path */ 1588 struct compress_io_ctx { 1589 u32 magic; /* magic number to indicate page is compressed */ 1590 struct inode *inode; /* inode the context belong to */ 1591 struct page **rpages; /* pages store raw data in cluster */ 1592 unsigned int nr_rpages; /* total page number in rpages */ 1593 atomic_t pending_pages; /* in-flight compressed page count */ 1594 }; 1595 1596 /* Context for decompressing one cluster on the read IO path */ 1597 struct decompress_io_ctx { 1598 u32 magic; /* magic number to indicate page is compressed */ 1599 struct inode *inode; /* inode the context belong to */ 1600 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1601 pgoff_t cluster_idx; /* cluster index number */ 1602 unsigned int cluster_size; /* page count in cluster */ 1603 unsigned int log_cluster_size; /* log of cluster size */ 1604 struct page **rpages; /* pages store raw data in cluster */ 1605 unsigned int nr_rpages; /* total page number in rpages */ 1606 struct page **cpages; /* pages store compressed data in cluster */ 1607 unsigned int nr_cpages; /* total page number in cpages */ 1608 struct page **tpages; /* temp pages to pad holes in cluster */ 1609 void *rbuf; /* virtual mapped address on rpages */ 1610 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1611 size_t rlen; /* valid data length in rbuf */ 1612 size_t clen; /* valid data length in cbuf */ 1613 1614 /* 1615 * The number of compressed pages remaining to be read in this cluster. 1616 * This is initially nr_cpages. It is decremented by 1 each time a page 1617 * has been read (or failed to be read). When it reaches 0, the cluster 1618 * is decompressed (or an error is reported). 1619 * 1620 * If an error occurs before all the pages have been submitted for I/O, 1621 * then this will never reach 0. In this case the I/O submitter is 1622 * responsible for calling f2fs_decompress_end_io() instead. 1623 */ 1624 atomic_t remaining_pages; 1625 1626 /* 1627 * Number of references to this decompress_io_ctx. 1628 * 1629 * One reference is held for I/O completion. This reference is dropped 1630 * after the pagecache pages are updated and unlocked -- either after 1631 * decompression (and verity if enabled), or after an error. 1632 * 1633 * In addition, each compressed page holds a reference while it is in a 1634 * bio. These references are necessary prevent compressed pages from 1635 * being freed while they are still in a bio. 1636 */ 1637 refcount_t refcnt; 1638 1639 bool failed; /* IO error occurred before decompression? */ 1640 bool need_verity; /* need fs-verity verification after decompression? */ 1641 unsigned char compress_algorithm; /* backup algorithm type */ 1642 void *private; /* payload buffer for specified decompression algorithm */ 1643 void *private2; /* extra payload buffer */ 1644 struct work_struct verity_work; /* work to verify the decompressed pages */ 1645 struct work_struct free_work; /* work for late free this structure itself */ 1646 }; 1647 1648 #define NULL_CLUSTER ((unsigned int)(~0)) 1649 #define MIN_COMPRESS_LOG_SIZE 2 1650 #define MAX_COMPRESS_LOG_SIZE 8 1651 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1652 1653 struct f2fs_sb_info { 1654 struct super_block *sb; /* pointer to VFS super block */ 1655 struct proc_dir_entry *s_proc; /* proc entry */ 1656 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1657 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1658 int valid_super_block; /* valid super block no */ 1659 unsigned long s_flag; /* flags for sbi */ 1660 struct mutex writepages; /* mutex for writepages() */ 1661 1662 #ifdef CONFIG_BLK_DEV_ZONED 1663 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1664 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1665 /* For adjust the priority writing position of data in zone UFS */ 1666 unsigned int blkzone_alloc_policy; 1667 #endif 1668 1669 /* for node-related operations */ 1670 struct f2fs_nm_info *nm_info; /* node manager */ 1671 struct inode *node_inode; /* cache node blocks */ 1672 1673 /* for segment-related operations */ 1674 struct f2fs_sm_info *sm_info; /* segment manager */ 1675 1676 /* for bio operations */ 1677 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1678 /* keep migration IO order for LFS mode */ 1679 struct f2fs_rwsem io_order_lock; 1680 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1681 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1682 1683 /* for checkpoint */ 1684 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1685 int cur_cp_pack; /* remain current cp pack */ 1686 spinlock_t cp_lock; /* for flag in ckpt */ 1687 struct inode *meta_inode; /* cache meta blocks */ 1688 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1689 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1690 struct f2fs_rwsem node_write; /* locking node writes */ 1691 struct f2fs_rwsem node_change; /* locking node change */ 1692 wait_queue_head_t cp_wait; 1693 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1694 long interval_time[MAX_TIME]; /* to store thresholds */ 1695 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1696 struct cp_stats cp_stats; /* for time stat of checkpoint */ 1697 1698 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1699 1700 spinlock_t fsync_node_lock; /* for node entry lock */ 1701 struct list_head fsync_node_list; /* node list head */ 1702 unsigned int fsync_seg_id; /* sequence id */ 1703 unsigned int fsync_node_num; /* number of node entries */ 1704 1705 /* for orphan inode, use 0'th array */ 1706 unsigned int max_orphans; /* max orphan inodes */ 1707 1708 /* for inode management */ 1709 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1710 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1711 struct mutex flush_lock; /* for flush exclusion */ 1712 1713 /* for extent tree cache */ 1714 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1715 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1716 unsigned int max_read_extent_count; /* max read extent count per inode */ 1717 1718 /* The threshold used for hot and warm data seperation*/ 1719 unsigned int hot_data_age_threshold; 1720 unsigned int warm_data_age_threshold; 1721 unsigned int last_age_weight; 1722 1723 /* control donate caches */ 1724 unsigned int donate_files; 1725 1726 /* basic filesystem units */ 1727 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1728 unsigned int log_blocksize; /* log2 block size */ 1729 unsigned int blocksize; /* block size */ 1730 unsigned int root_ino_num; /* root inode number*/ 1731 unsigned int node_ino_num; /* node inode number*/ 1732 unsigned int meta_ino_num; /* meta inode number*/ 1733 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1734 unsigned int blocks_per_seg; /* blocks per segment */ 1735 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1736 unsigned int segs_per_sec; /* segments per section */ 1737 unsigned int secs_per_zone; /* sections per zone */ 1738 unsigned int total_sections; /* total section count */ 1739 unsigned int total_node_count; /* total node block count */ 1740 unsigned int total_valid_node_count; /* valid node block count */ 1741 int dir_level; /* directory level */ 1742 bool readdir_ra; /* readahead inode in readdir */ 1743 u64 max_io_bytes; /* max io bytes to merge IOs */ 1744 1745 block_t user_block_count; /* # of user blocks */ 1746 block_t total_valid_block_count; /* # of valid blocks */ 1747 block_t discard_blks; /* discard command candidats */ 1748 block_t last_valid_block_count; /* for recovery */ 1749 block_t reserved_blocks; /* configurable reserved blocks */ 1750 block_t current_reserved_blocks; /* current reserved blocks */ 1751 1752 /* Additional tracking for no checkpoint mode */ 1753 block_t unusable_block_count; /* # of blocks saved by last cp */ 1754 1755 unsigned int nquota_files; /* # of quota sysfile */ 1756 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1757 struct task_struct *umount_lock_holder; /* s_umount lock holder */ 1758 1759 /* # of pages, see count_type */ 1760 atomic_t nr_pages[NR_COUNT_TYPE]; 1761 /* # of allocated blocks */ 1762 struct percpu_counter alloc_valid_block_count; 1763 /* # of node block writes as roll forward recovery */ 1764 struct percpu_counter rf_node_block_count; 1765 1766 /* writeback control */ 1767 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1768 1769 /* valid inode count */ 1770 struct percpu_counter total_valid_inode_count; 1771 1772 struct f2fs_mount_info mount_opt; /* mount options */ 1773 1774 /* for cleaning operations */ 1775 struct f2fs_rwsem gc_lock; /* 1776 * semaphore for GC, avoid 1777 * race between GC and GC or CP 1778 */ 1779 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1780 struct atgc_management am; /* atgc management */ 1781 unsigned int cur_victim_sec; /* current victim section num */ 1782 unsigned int gc_mode; /* current GC state */ 1783 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1784 spinlock_t gc_remaining_trials_lock; 1785 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1786 unsigned int gc_remaining_trials; 1787 1788 /* for skip statistic */ 1789 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1790 1791 /* free sections reserved for pinned file */ 1792 unsigned int reserved_pin_section; 1793 1794 /* threshold for gc trials on pinned files */ 1795 unsigned short gc_pin_file_threshold; 1796 struct f2fs_rwsem pin_sem; 1797 1798 /* maximum # of trials to find a victim segment for SSR and GC */ 1799 unsigned int max_victim_search; 1800 /* migration granularity of garbage collection, unit: segment */ 1801 unsigned int migration_granularity; 1802 /* migration window granularity of garbage collection, unit: segment */ 1803 unsigned int migration_window_granularity; 1804 1805 /* 1806 * for stat information. 1807 * one is for the LFS mode, and the other is for the SSR mode. 1808 */ 1809 #ifdef CONFIG_F2FS_STAT_FS 1810 struct f2fs_stat_info *stat_info; /* FS status information */ 1811 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1812 unsigned int segment_count[2]; /* # of allocated segments */ 1813 unsigned int block_count[2]; /* # of allocated blocks */ 1814 atomic_t inplace_count; /* # of inplace update */ 1815 /* # of lookup extent cache */ 1816 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1817 /* # of hit rbtree extent node */ 1818 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1819 /* # of hit cached extent node */ 1820 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1821 /* # of hit largest extent node in read extent cache */ 1822 atomic64_t read_hit_largest; 1823 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1824 atomic_t inline_inode; /* # of inline_data inodes */ 1825 atomic_t inline_dir; /* # of inline_dentry inodes */ 1826 atomic_t compr_inode; /* # of compressed inodes */ 1827 atomic64_t compr_blocks; /* # of compressed blocks */ 1828 atomic_t swapfile_inode; /* # of swapfile inodes */ 1829 atomic_t atomic_files; /* # of opened atomic file */ 1830 atomic_t max_aw_cnt; /* max # of atomic writes */ 1831 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1832 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1833 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1834 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1835 #endif 1836 spinlock_t stat_lock; /* lock for stat operations */ 1837 1838 /* to attach REQ_META|REQ_FUA flags */ 1839 unsigned int data_io_flag; 1840 unsigned int node_io_flag; 1841 1842 /* For sysfs support */ 1843 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1844 struct completion s_kobj_unregister; 1845 1846 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1847 struct completion s_stat_kobj_unregister; 1848 1849 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1850 struct completion s_feature_list_kobj_unregister; 1851 1852 /* For shrinker support */ 1853 struct list_head s_list; 1854 struct mutex umount_mutex; 1855 unsigned int shrinker_run_no; 1856 1857 /* For multi devices */ 1858 int s_ndevs; /* number of devices */ 1859 struct f2fs_dev_info *devs; /* for device list */ 1860 unsigned int dirty_device; /* for checkpoint data flush */ 1861 spinlock_t dev_lock; /* protect dirty_device */ 1862 bool aligned_blksize; /* all devices has the same logical blksize */ 1863 unsigned int first_seq_zone_segno; /* first segno in sequential zone */ 1864 unsigned int bggc_io_aware; /* For adjust the BG_GC priority when pending IO */ 1865 unsigned int allocate_section_hint; /* the boundary position between devices */ 1866 unsigned int allocate_section_policy; /* determine the section writing priority */ 1867 1868 /* For write statistics */ 1869 u64 sectors_written_start; 1870 u64 kbytes_written; 1871 1872 /* Precomputed FS UUID checksum for seeding other checksums */ 1873 __u32 s_chksum_seed; 1874 1875 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1876 1877 /* 1878 * If we are in irq context, let's update error information into 1879 * on-disk superblock in the work. 1880 */ 1881 struct work_struct s_error_work; 1882 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1883 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1884 spinlock_t error_lock; /* protect errors/stop_reason array */ 1885 bool error_dirty; /* errors of sb is dirty */ 1886 1887 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1888 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1889 1890 /* For reclaimed segs statistics per each GC mode */ 1891 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1892 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1893 1894 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1895 1896 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1897 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1898 1899 /* For atomic write statistics */ 1900 atomic64_t current_atomic_write; 1901 s64 peak_atomic_write; 1902 u64 committed_atomic_block; 1903 u64 revoked_atomic_block; 1904 1905 /* carve out reserved_blocks from total blocks */ 1906 bool carve_out; 1907 1908 #ifdef CONFIG_F2FS_FS_COMPRESSION 1909 struct kmem_cache *page_array_slab; /* page array entry */ 1910 unsigned int page_array_slab_size; /* default page array slab size */ 1911 1912 /* For runtime compression statistics */ 1913 u64 compr_written_block; 1914 u64 compr_saved_block; 1915 u32 compr_new_inode; 1916 1917 /* For compressed block cache */ 1918 struct inode *compress_inode; /* cache compressed blocks */ 1919 unsigned int compress_percent; /* cache page percentage */ 1920 unsigned int compress_watermark; /* cache page watermark */ 1921 atomic_t compress_page_hit; /* cache hit count */ 1922 #endif 1923 1924 #ifdef CONFIG_F2FS_IOSTAT 1925 /* For app/fs IO statistics */ 1926 spinlock_t iostat_lock; 1927 unsigned long long iostat_count[NR_IO_TYPE]; 1928 unsigned long long iostat_bytes[NR_IO_TYPE]; 1929 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1930 bool iostat_enable; 1931 unsigned long iostat_next_period; 1932 unsigned int iostat_period_ms; 1933 1934 /* For io latency related statistics info in one iostat period */ 1935 spinlock_t iostat_lat_lock; 1936 struct iostat_lat_info *iostat_io_lat; 1937 #endif 1938 }; 1939 1940 /* Definitions to access f2fs_sb_info */ 1941 #define SEGS_TO_BLKS(sbi, segs) \ 1942 ((segs) << (sbi)->log_blocks_per_seg) 1943 #define BLKS_TO_SEGS(sbi, blks) \ 1944 ((blks) >> (sbi)->log_blocks_per_seg) 1945 1946 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 1947 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 1948 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 1949 1950 __printf(3, 4) 1951 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1952 1953 #define f2fs_err(sbi, fmt, ...) \ 1954 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1955 #define f2fs_warn(sbi, fmt, ...) \ 1956 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1957 #define f2fs_notice(sbi, fmt, ...) \ 1958 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1959 #define f2fs_info(sbi, fmt, ...) \ 1960 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1961 #define f2fs_debug(sbi, fmt, ...) \ 1962 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1963 1964 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1965 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1966 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1967 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1968 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1969 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1970 1971 #ifdef CONFIG_F2FS_FAULT_INJECTION 1972 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1973 __builtin_return_address(0)) 1974 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1975 const char *func, const char *parent_func) 1976 { 1977 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1978 1979 if (!ffi->inject_rate) 1980 return false; 1981 1982 if (!IS_FAULT_SET(ffi, type)) 1983 return false; 1984 1985 atomic_inc(&ffi->inject_ops); 1986 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1987 atomic_set(&ffi->inject_ops, 0); 1988 ffi->inject_count[type]++; 1989 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1990 f2fs_fault_name[type], func, parent_func); 1991 return true; 1992 } 1993 return false; 1994 } 1995 #else 1996 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1997 { 1998 return false; 1999 } 2000 #endif 2001 2002 /* 2003 * Test if the mounted volume is a multi-device volume. 2004 * - For a single regular disk volume, sbi->s_ndevs is 0. 2005 * - For a single zoned disk volume, sbi->s_ndevs is 1. 2006 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 2007 */ 2008 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 2009 { 2010 return sbi->s_ndevs > 1; 2011 } 2012 2013 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 2014 { 2015 unsigned long now = jiffies; 2016 2017 sbi->last_time[type] = now; 2018 2019 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 2020 if (type == REQ_TIME) { 2021 sbi->last_time[DISCARD_TIME] = now; 2022 sbi->last_time[GC_TIME] = now; 2023 } 2024 } 2025 2026 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 2027 { 2028 unsigned long interval = sbi->interval_time[type] * HZ; 2029 2030 return time_after(jiffies, sbi->last_time[type] + interval); 2031 } 2032 2033 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 2034 int type) 2035 { 2036 unsigned long interval = sbi->interval_time[type] * HZ; 2037 unsigned int wait_ms = 0; 2038 long delta; 2039 2040 delta = (sbi->last_time[type] + interval) - jiffies; 2041 if (delta > 0) 2042 wait_ms = jiffies_to_msecs(delta); 2043 2044 return wait_ms; 2045 } 2046 2047 /* 2048 * Inline functions 2049 */ 2050 static inline u32 __f2fs_crc32(u32 crc, const void *address, 2051 unsigned int length) 2052 { 2053 return crc32(crc, address, length); 2054 } 2055 2056 static inline u32 f2fs_crc32(const void *address, unsigned int length) 2057 { 2058 return __f2fs_crc32(F2FS_SUPER_MAGIC, address, length); 2059 } 2060 2061 static inline u32 f2fs_chksum(u32 crc, const void *address, unsigned int length) 2062 { 2063 return __f2fs_crc32(crc, address, length); 2064 } 2065 2066 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 2067 { 2068 return container_of(inode, struct f2fs_inode_info, vfs_inode); 2069 } 2070 2071 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2072 { 2073 return sb->s_fs_info; 2074 } 2075 2076 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2077 { 2078 return F2FS_SB(inode->i_sb); 2079 } 2080 2081 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2082 { 2083 return F2FS_I_SB(mapping->host); 2084 } 2085 2086 static inline struct f2fs_sb_info *F2FS_F_SB(const struct folio *folio) 2087 { 2088 return F2FS_M_SB(folio->mapping); 2089 } 2090 2091 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2092 { 2093 return (struct f2fs_super_block *)(sbi->raw_super); 2094 } 2095 2096 static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2097 pgoff_t index) 2098 { 2099 pgoff_t idx_in_folio = index % (1 << folio_order(folio)); 2100 2101 return (struct f2fs_super_block *) 2102 (page_address(folio_page(folio, idx_in_folio)) + 2103 F2FS_SUPER_OFFSET); 2104 } 2105 2106 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2107 { 2108 return (struct f2fs_checkpoint *)(sbi->ckpt); 2109 } 2110 2111 static inline struct f2fs_node *F2FS_NODE(const struct folio *folio) 2112 { 2113 return (struct f2fs_node *)folio_address(folio); 2114 } 2115 2116 static inline struct f2fs_inode *F2FS_INODE(const struct folio *folio) 2117 { 2118 return &((struct f2fs_node *)folio_address(folio))->i; 2119 } 2120 2121 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2122 { 2123 return (struct f2fs_nm_info *)(sbi->nm_info); 2124 } 2125 2126 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2127 { 2128 return (struct f2fs_sm_info *)(sbi->sm_info); 2129 } 2130 2131 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2132 { 2133 return (struct sit_info *)(SM_I(sbi)->sit_info); 2134 } 2135 2136 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2137 { 2138 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2139 } 2140 2141 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2142 { 2143 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2144 } 2145 2146 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2147 { 2148 return sbi->meta_inode->i_mapping; 2149 } 2150 2151 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2152 { 2153 return sbi->node_inode->i_mapping; 2154 } 2155 2156 static inline bool is_meta_folio(struct folio *folio) 2157 { 2158 return folio->mapping == META_MAPPING(F2FS_F_SB(folio)); 2159 } 2160 2161 static inline bool is_node_folio(struct folio *folio) 2162 { 2163 return folio->mapping == NODE_MAPPING(F2FS_F_SB(folio)); 2164 } 2165 2166 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2167 { 2168 return test_bit(type, &sbi->s_flag); 2169 } 2170 2171 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2172 { 2173 set_bit(type, &sbi->s_flag); 2174 } 2175 2176 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2177 { 2178 clear_bit(type, &sbi->s_flag); 2179 } 2180 2181 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2182 { 2183 return le64_to_cpu(cp->checkpoint_ver); 2184 } 2185 2186 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2187 { 2188 if (type < F2FS_MAX_QUOTAS) 2189 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2190 return 0; 2191 } 2192 2193 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2194 { 2195 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2196 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2197 } 2198 2199 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2200 { 2201 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2202 2203 return ckpt_flags & f; 2204 } 2205 2206 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2207 { 2208 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2209 } 2210 2211 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2212 { 2213 unsigned int ckpt_flags; 2214 2215 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2216 ckpt_flags |= f; 2217 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2218 } 2219 2220 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2221 { 2222 unsigned long flags; 2223 2224 spin_lock_irqsave(&sbi->cp_lock, flags); 2225 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2226 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2227 } 2228 2229 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2230 { 2231 unsigned int ckpt_flags; 2232 2233 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2234 ckpt_flags &= (~f); 2235 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2236 } 2237 2238 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2239 { 2240 unsigned long flags; 2241 2242 spin_lock_irqsave(&sbi->cp_lock, flags); 2243 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2244 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2245 } 2246 2247 #define init_f2fs_rwsem(sem) \ 2248 do { \ 2249 static struct lock_class_key __key; \ 2250 \ 2251 __init_f2fs_rwsem((sem), #sem, &__key); \ 2252 } while (0) 2253 2254 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2255 const char *sem_name, struct lock_class_key *key) 2256 { 2257 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2258 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2259 init_waitqueue_head(&sem->read_waiters); 2260 #endif 2261 } 2262 2263 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2264 { 2265 return rwsem_is_locked(&sem->internal_rwsem); 2266 } 2267 2268 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2269 { 2270 return rwsem_is_contended(&sem->internal_rwsem); 2271 } 2272 2273 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2274 { 2275 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2276 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2277 #else 2278 down_read(&sem->internal_rwsem); 2279 #endif 2280 } 2281 2282 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2283 { 2284 return down_read_trylock(&sem->internal_rwsem); 2285 } 2286 2287 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2288 { 2289 up_read(&sem->internal_rwsem); 2290 } 2291 2292 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2293 { 2294 down_write(&sem->internal_rwsem); 2295 } 2296 2297 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2298 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2299 { 2300 down_read_nested(&sem->internal_rwsem, subclass); 2301 } 2302 2303 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2304 { 2305 down_write_nested(&sem->internal_rwsem, subclass); 2306 } 2307 #else 2308 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2309 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2310 #endif 2311 2312 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2313 { 2314 return down_write_trylock(&sem->internal_rwsem); 2315 } 2316 2317 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2318 { 2319 up_write(&sem->internal_rwsem); 2320 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2321 wake_up_all(&sem->read_waiters); 2322 #endif 2323 } 2324 2325 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2326 { 2327 unsigned long flags; 2328 unsigned char *nat_bits; 2329 2330 /* 2331 * In order to re-enable nat_bits we need to call fsck.f2fs by 2332 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2333 * so let's rely on regular fsck or unclean shutdown. 2334 */ 2335 2336 if (lock) 2337 spin_lock_irqsave(&sbi->cp_lock, flags); 2338 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2339 nat_bits = NM_I(sbi)->nat_bits; 2340 NM_I(sbi)->nat_bits = NULL; 2341 if (lock) 2342 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2343 2344 kvfree(nat_bits); 2345 } 2346 2347 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2348 struct cp_control *cpc) 2349 { 2350 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2351 2352 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2353 } 2354 2355 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2356 { 2357 f2fs_down_read(&sbi->cp_rwsem); 2358 } 2359 2360 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2361 { 2362 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2363 return 0; 2364 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2365 } 2366 2367 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2368 { 2369 f2fs_up_read(&sbi->cp_rwsem); 2370 } 2371 2372 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2373 { 2374 f2fs_down_write(&sbi->cp_rwsem); 2375 } 2376 2377 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2378 { 2379 f2fs_up_write(&sbi->cp_rwsem); 2380 } 2381 2382 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2383 { 2384 int reason = CP_SYNC; 2385 2386 if (test_opt(sbi, FASTBOOT)) 2387 reason = CP_FASTBOOT; 2388 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2389 reason = CP_UMOUNT; 2390 return reason; 2391 } 2392 2393 static inline bool __remain_node_summaries(int reason) 2394 { 2395 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2396 } 2397 2398 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2399 { 2400 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2401 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2402 } 2403 2404 /* 2405 * Check whether the inode has blocks or not 2406 */ 2407 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2408 { 2409 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2410 2411 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2412 } 2413 2414 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2415 { 2416 return ofs == XATTR_NODE_OFFSET; 2417 } 2418 2419 static inline bool __allow_reserved_root(struct f2fs_sb_info *sbi, 2420 struct inode *inode, bool cap) 2421 { 2422 if (!inode) 2423 return true; 2424 if (IS_NOQUOTA(inode)) 2425 return true; 2426 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2427 return true; 2428 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2429 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2430 return true; 2431 if (cap && capable(CAP_SYS_RESOURCE)) 2432 return true; 2433 return false; 2434 } 2435 2436 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2437 struct inode *inode, bool cap) 2438 { 2439 block_t avail_user_block_count; 2440 2441 avail_user_block_count = sbi->user_block_count - 2442 sbi->current_reserved_blocks; 2443 2444 if (test_opt(sbi, RESERVE_ROOT) && !__allow_reserved_root(sbi, inode, cap)) 2445 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2446 2447 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2448 if (avail_user_block_count > sbi->unusable_block_count) 2449 avail_user_block_count -= sbi->unusable_block_count; 2450 else 2451 avail_user_block_count = 0; 2452 } 2453 2454 return avail_user_block_count; 2455 } 2456 2457 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2458 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2459 struct inode *inode, blkcnt_t *count, bool partial) 2460 { 2461 long long diff = 0, release = 0; 2462 block_t avail_user_block_count; 2463 int ret; 2464 2465 ret = dquot_reserve_block(inode, *count); 2466 if (ret) 2467 return ret; 2468 2469 if (time_to_inject(sbi, FAULT_BLOCK)) { 2470 release = *count; 2471 goto release_quota; 2472 } 2473 2474 /* 2475 * let's increase this in prior to actual block count change in order 2476 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2477 */ 2478 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2479 2480 spin_lock(&sbi->stat_lock); 2481 2482 avail_user_block_count = get_available_block_count(sbi, inode, true); 2483 diff = (long long)sbi->total_valid_block_count + *count - 2484 avail_user_block_count; 2485 if (unlikely(diff > 0)) { 2486 if (!partial) { 2487 spin_unlock(&sbi->stat_lock); 2488 release = *count; 2489 goto enospc; 2490 } 2491 if (diff > *count) 2492 diff = *count; 2493 *count -= diff; 2494 release = diff; 2495 if (!*count) { 2496 spin_unlock(&sbi->stat_lock); 2497 goto enospc; 2498 } 2499 } 2500 sbi->total_valid_block_count += (block_t)(*count); 2501 2502 spin_unlock(&sbi->stat_lock); 2503 2504 if (unlikely(release)) { 2505 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2506 dquot_release_reservation_block(inode, release); 2507 } 2508 f2fs_i_blocks_write(inode, *count, true, true); 2509 return 0; 2510 2511 enospc: 2512 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2513 release_quota: 2514 dquot_release_reservation_block(inode, release); 2515 return -ENOSPC; 2516 } 2517 2518 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2519 static inline bool folio_test_f2fs_##name(const struct folio *folio) \ 2520 { \ 2521 unsigned long priv = (unsigned long)folio->private; \ 2522 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2523 (1UL << PAGE_PRIVATE_##flagname); \ 2524 return (priv & v) == v; \ 2525 } \ 2526 static inline bool page_private_##name(struct page *page) \ 2527 { \ 2528 return PagePrivate(page) && \ 2529 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2530 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2531 } 2532 2533 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2534 static inline void folio_set_f2fs_##name(struct folio *folio) \ 2535 { \ 2536 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2537 (1UL << PAGE_PRIVATE_##flagname); \ 2538 if (!folio->private) \ 2539 folio_attach_private(folio, (void *)v); \ 2540 else { \ 2541 v |= (unsigned long)folio->private; \ 2542 folio->private = (void *)v; \ 2543 } \ 2544 } \ 2545 static inline void set_page_private_##name(struct page *page) \ 2546 { \ 2547 if (!PagePrivate(page)) \ 2548 attach_page_private(page, (void *)0); \ 2549 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2550 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2551 } 2552 2553 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2554 static inline void folio_clear_f2fs_##name(struct folio *folio) \ 2555 { \ 2556 unsigned long v = (unsigned long)folio->private; \ 2557 \ 2558 v &= ~(1UL << PAGE_PRIVATE_##flagname); \ 2559 if (v == (1UL << PAGE_PRIVATE_NOT_POINTER)) \ 2560 folio_detach_private(folio); \ 2561 else \ 2562 folio->private = (void *)v; \ 2563 } \ 2564 static inline void clear_page_private_##name(struct page *page) \ 2565 { \ 2566 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2567 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2568 detach_page_private(page); \ 2569 } 2570 2571 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2572 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2573 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2574 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2575 2576 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2577 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2578 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2579 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2580 2581 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2582 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2583 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2584 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2585 2586 static inline unsigned long folio_get_f2fs_data(struct folio *folio) 2587 { 2588 unsigned long data = (unsigned long)folio->private; 2589 2590 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2591 return 0; 2592 return data >> PAGE_PRIVATE_MAX; 2593 } 2594 2595 static inline void folio_set_f2fs_data(struct folio *folio, unsigned long data) 2596 { 2597 data = (1UL << PAGE_PRIVATE_NOT_POINTER) | (data << PAGE_PRIVATE_MAX); 2598 2599 if (!folio_test_private(folio)) 2600 folio_attach_private(folio, (void *)data); 2601 else 2602 folio->private = (void *)((unsigned long)folio->private | data); 2603 } 2604 2605 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2606 struct inode *inode, 2607 block_t count) 2608 { 2609 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2610 2611 spin_lock(&sbi->stat_lock); 2612 if (unlikely(sbi->total_valid_block_count < count)) { 2613 f2fs_warn(sbi, "Inconsistent total_valid_block_count:%u, ino:%lu, count:%u", 2614 sbi->total_valid_block_count, inode->i_ino, count); 2615 sbi->total_valid_block_count = 0; 2616 set_sbi_flag(sbi, SBI_NEED_FSCK); 2617 } else { 2618 sbi->total_valid_block_count -= count; 2619 } 2620 if (sbi->reserved_blocks && 2621 sbi->current_reserved_blocks < sbi->reserved_blocks) 2622 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2623 sbi->current_reserved_blocks + count); 2624 spin_unlock(&sbi->stat_lock); 2625 if (unlikely(inode->i_blocks < sectors)) { 2626 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2627 inode->i_ino, 2628 (unsigned long long)inode->i_blocks, 2629 (unsigned long long)sectors); 2630 set_sbi_flag(sbi, SBI_NEED_FSCK); 2631 return; 2632 } 2633 f2fs_i_blocks_write(inode, count, false, true); 2634 } 2635 2636 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2637 { 2638 atomic_inc(&sbi->nr_pages[count_type]); 2639 2640 if (count_type == F2FS_DIRTY_DENTS || 2641 count_type == F2FS_DIRTY_NODES || 2642 count_type == F2FS_DIRTY_META || 2643 count_type == F2FS_DIRTY_QDATA || 2644 count_type == F2FS_DIRTY_IMETA) 2645 set_sbi_flag(sbi, SBI_IS_DIRTY); 2646 } 2647 2648 static inline void inode_inc_dirty_pages(struct inode *inode) 2649 { 2650 atomic_inc(&F2FS_I(inode)->dirty_pages); 2651 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2652 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2653 if (IS_NOQUOTA(inode)) 2654 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2655 } 2656 2657 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2658 { 2659 atomic_dec(&sbi->nr_pages[count_type]); 2660 } 2661 2662 static inline void inode_dec_dirty_pages(struct inode *inode) 2663 { 2664 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2665 !S_ISLNK(inode->i_mode)) 2666 return; 2667 2668 atomic_dec(&F2FS_I(inode)->dirty_pages); 2669 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2670 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2671 if (IS_NOQUOTA(inode)) 2672 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2673 } 2674 2675 static inline void inc_atomic_write_cnt(struct inode *inode) 2676 { 2677 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2678 struct f2fs_inode_info *fi = F2FS_I(inode); 2679 u64 current_write; 2680 2681 fi->atomic_write_cnt++; 2682 atomic64_inc(&sbi->current_atomic_write); 2683 current_write = atomic64_read(&sbi->current_atomic_write); 2684 if (current_write > sbi->peak_atomic_write) 2685 sbi->peak_atomic_write = current_write; 2686 } 2687 2688 static inline void release_atomic_write_cnt(struct inode *inode) 2689 { 2690 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2691 struct f2fs_inode_info *fi = F2FS_I(inode); 2692 2693 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2694 fi->atomic_write_cnt = 0; 2695 } 2696 2697 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2698 { 2699 return atomic_read(&sbi->nr_pages[count_type]); 2700 } 2701 2702 static inline int get_dirty_pages(struct inode *inode) 2703 { 2704 return atomic_read(&F2FS_I(inode)->dirty_pages); 2705 } 2706 2707 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2708 { 2709 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2710 BLKS_PER_SEC(sbi)); 2711 } 2712 2713 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2714 { 2715 return sbi->total_valid_block_count; 2716 } 2717 2718 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2719 { 2720 return sbi->discard_blks; 2721 } 2722 2723 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2724 { 2725 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2726 2727 /* return NAT or SIT bitmap */ 2728 if (flag == NAT_BITMAP) 2729 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2730 else if (flag == SIT_BITMAP) 2731 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2732 2733 return 0; 2734 } 2735 2736 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2737 { 2738 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2739 } 2740 2741 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2742 { 2743 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2744 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2745 int offset; 2746 2747 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2748 offset = (flag == SIT_BITMAP) ? 2749 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2750 /* 2751 * if large_nat_bitmap feature is enabled, leave checksum 2752 * protection for all nat/sit bitmaps. 2753 */ 2754 return tmp_ptr + offset + sizeof(__le32); 2755 } 2756 2757 if (__cp_payload(sbi) > 0) { 2758 if (flag == NAT_BITMAP) 2759 return tmp_ptr; 2760 else 2761 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2762 } else { 2763 offset = (flag == NAT_BITMAP) ? 2764 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2765 return tmp_ptr + offset; 2766 } 2767 } 2768 2769 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2770 { 2771 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2772 2773 if (sbi->cur_cp_pack == 2) 2774 start_addr += BLKS_PER_SEG(sbi); 2775 return start_addr; 2776 } 2777 2778 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2779 { 2780 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2781 2782 if (sbi->cur_cp_pack == 1) 2783 start_addr += BLKS_PER_SEG(sbi); 2784 return start_addr; 2785 } 2786 2787 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2788 { 2789 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2790 } 2791 2792 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2793 { 2794 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2795 } 2796 2797 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2798 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2799 struct inode *inode, bool is_inode) 2800 { 2801 block_t valid_block_count; 2802 unsigned int valid_node_count, avail_user_node_count; 2803 unsigned int avail_user_block_count; 2804 int err; 2805 2806 if (is_inode) { 2807 if (inode) { 2808 err = dquot_alloc_inode(inode); 2809 if (err) 2810 return err; 2811 } 2812 } else { 2813 err = dquot_reserve_block(inode, 1); 2814 if (err) 2815 return err; 2816 } 2817 2818 if (time_to_inject(sbi, FAULT_BLOCK)) 2819 goto enospc; 2820 2821 spin_lock(&sbi->stat_lock); 2822 2823 valid_block_count = sbi->total_valid_block_count + 1; 2824 avail_user_block_count = get_available_block_count(sbi, inode, 2825 test_opt(sbi, RESERVE_NODE)); 2826 2827 if (unlikely(valid_block_count > avail_user_block_count)) { 2828 spin_unlock(&sbi->stat_lock); 2829 goto enospc; 2830 } 2831 2832 avail_user_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM; 2833 if (test_opt(sbi, RESERVE_NODE) && 2834 !__allow_reserved_root(sbi, inode, true)) 2835 avail_user_node_count -= F2FS_OPTION(sbi).root_reserved_nodes; 2836 valid_node_count = sbi->total_valid_node_count + 1; 2837 if (unlikely(valid_node_count > avail_user_node_count)) { 2838 spin_unlock(&sbi->stat_lock); 2839 goto enospc; 2840 } 2841 2842 sbi->total_valid_node_count++; 2843 sbi->total_valid_block_count++; 2844 spin_unlock(&sbi->stat_lock); 2845 2846 if (inode) { 2847 if (is_inode) 2848 f2fs_mark_inode_dirty_sync(inode, true); 2849 else 2850 f2fs_i_blocks_write(inode, 1, true, true); 2851 } 2852 2853 percpu_counter_inc(&sbi->alloc_valid_block_count); 2854 return 0; 2855 2856 enospc: 2857 if (is_inode) { 2858 if (inode) 2859 dquot_free_inode(inode); 2860 } else { 2861 dquot_release_reservation_block(inode, 1); 2862 } 2863 return -ENOSPC; 2864 } 2865 2866 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2867 struct inode *inode, bool is_inode) 2868 { 2869 spin_lock(&sbi->stat_lock); 2870 2871 if (unlikely(!sbi->total_valid_block_count || 2872 !sbi->total_valid_node_count)) { 2873 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2874 sbi->total_valid_block_count, 2875 sbi->total_valid_node_count); 2876 set_sbi_flag(sbi, SBI_NEED_FSCK); 2877 } else { 2878 sbi->total_valid_block_count--; 2879 sbi->total_valid_node_count--; 2880 } 2881 2882 if (sbi->reserved_blocks && 2883 sbi->current_reserved_blocks < sbi->reserved_blocks) 2884 sbi->current_reserved_blocks++; 2885 2886 spin_unlock(&sbi->stat_lock); 2887 2888 if (is_inode) { 2889 dquot_free_inode(inode); 2890 } else { 2891 if (unlikely(inode->i_blocks == 0)) { 2892 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2893 inode->i_ino, 2894 (unsigned long long)inode->i_blocks); 2895 set_sbi_flag(sbi, SBI_NEED_FSCK); 2896 return; 2897 } 2898 f2fs_i_blocks_write(inode, 1, false, true); 2899 } 2900 } 2901 2902 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2903 { 2904 return sbi->total_valid_node_count; 2905 } 2906 2907 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2908 { 2909 percpu_counter_inc(&sbi->total_valid_inode_count); 2910 } 2911 2912 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2913 { 2914 percpu_counter_dec(&sbi->total_valid_inode_count); 2915 } 2916 2917 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2918 { 2919 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2920 } 2921 2922 static inline struct folio *f2fs_grab_cache_folio(struct address_space *mapping, 2923 pgoff_t index, bool for_write) 2924 { 2925 struct folio *folio; 2926 unsigned int flags; 2927 2928 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2929 fgf_t fgf_flags; 2930 2931 if (!for_write) 2932 fgf_flags = FGP_LOCK | FGP_ACCESSED; 2933 else 2934 fgf_flags = FGP_LOCK; 2935 folio = __filemap_get_folio(mapping, index, fgf_flags, 0); 2936 if (!IS_ERR(folio)) 2937 return folio; 2938 2939 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2940 return ERR_PTR(-ENOMEM); 2941 } 2942 2943 if (!for_write) 2944 return filemap_grab_folio(mapping, index); 2945 2946 flags = memalloc_nofs_save(); 2947 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN, 2948 mapping_gfp_mask(mapping)); 2949 memalloc_nofs_restore(flags); 2950 2951 return folio; 2952 } 2953 2954 static inline struct folio *f2fs_filemap_get_folio( 2955 struct address_space *mapping, pgoff_t index, 2956 fgf_t fgp_flags, gfp_t gfp_mask) 2957 { 2958 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2959 return ERR_PTR(-ENOMEM); 2960 2961 return __filemap_get_folio(mapping, index, fgp_flags, gfp_mask); 2962 } 2963 2964 static inline struct page *f2fs_pagecache_get_page( 2965 struct address_space *mapping, pgoff_t index, 2966 fgf_t fgp_flags, gfp_t gfp_mask) 2967 { 2968 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2969 return NULL; 2970 2971 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2972 } 2973 2974 static inline void f2fs_folio_put(struct folio *folio, bool unlock) 2975 { 2976 if (IS_ERR_OR_NULL(folio)) 2977 return; 2978 2979 if (unlock) { 2980 f2fs_bug_on(F2FS_F_SB(folio), !folio_test_locked(folio)); 2981 folio_unlock(folio); 2982 } 2983 folio_put(folio); 2984 } 2985 2986 static inline void f2fs_put_page(struct page *page, int unlock) 2987 { 2988 if (!page) 2989 return; 2990 f2fs_folio_put(page_folio(page), unlock); 2991 } 2992 2993 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2994 { 2995 if (dn->node_folio) 2996 f2fs_folio_put(dn->node_folio, true); 2997 if (dn->inode_folio && dn->node_folio != dn->inode_folio) 2998 f2fs_folio_put(dn->inode_folio, false); 2999 dn->node_folio = NULL; 3000 dn->inode_folio = NULL; 3001 } 3002 3003 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 3004 size_t size) 3005 { 3006 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 3007 } 3008 3009 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 3010 gfp_t flags) 3011 { 3012 void *entry; 3013 3014 entry = kmem_cache_alloc(cachep, flags); 3015 if (!entry) 3016 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 3017 return entry; 3018 } 3019 3020 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 3021 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 3022 { 3023 if (nofail) 3024 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 3025 3026 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 3027 return NULL; 3028 3029 return kmem_cache_alloc(cachep, flags); 3030 } 3031 3032 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 3033 { 3034 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 3035 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 3036 get_pages(sbi, F2FS_WB_CP_DATA) || 3037 get_pages(sbi, F2FS_DIO_READ) || 3038 get_pages(sbi, F2FS_DIO_WRITE)) 3039 return true; 3040 3041 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 3042 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 3043 return true; 3044 3045 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 3046 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 3047 return true; 3048 return false; 3049 } 3050 3051 static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 3052 { 3053 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ); 3054 } 3055 3056 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 3057 { 3058 bool zoned_gc = (type == GC_TIME && 3059 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 3060 3061 if (sbi->gc_mode == GC_URGENT_HIGH) 3062 return true; 3063 3064 if (sbi->bggc_io_aware == AWARE_READ_IO && is_inflight_read_io(sbi)) 3065 return false; 3066 if (sbi->bggc_io_aware == AWARE_ALL_IO && is_inflight_io(sbi, type)) 3067 return false; 3068 3069 if (sbi->gc_mode == GC_URGENT_MID) 3070 return true; 3071 3072 if (sbi->gc_mode == GC_URGENT_LOW && 3073 (type == DISCARD_TIME || type == GC_TIME)) 3074 return true; 3075 3076 if (zoned_gc) 3077 return true; 3078 3079 return f2fs_time_over(sbi, type); 3080 } 3081 3082 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 3083 unsigned long index, void *item) 3084 { 3085 while (radix_tree_insert(root, index, item)) 3086 cond_resched(); 3087 } 3088 3089 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 3090 3091 static inline bool IS_INODE(const struct folio *folio) 3092 { 3093 struct f2fs_node *p = F2FS_NODE(folio); 3094 3095 return RAW_IS_INODE(p); 3096 } 3097 3098 static inline int offset_in_addr(struct f2fs_inode *i) 3099 { 3100 return (i->i_inline & F2FS_EXTRA_ATTR) ? 3101 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 3102 } 3103 3104 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 3105 { 3106 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 3107 } 3108 3109 static inline int f2fs_has_extra_attr(struct inode *inode); 3110 static inline unsigned int get_dnode_base(struct inode *inode, 3111 struct folio *node_folio) 3112 { 3113 if (!IS_INODE(node_folio)) 3114 return 0; 3115 3116 return inode ? get_extra_isize(inode) : 3117 offset_in_addr(&F2FS_NODE(node_folio)->i); 3118 } 3119 3120 static inline __le32 *get_dnode_addr(struct inode *inode, 3121 struct folio *node_folio) 3122 { 3123 return blkaddr_in_node(F2FS_NODE(node_folio)) + 3124 get_dnode_base(inode, node_folio); 3125 } 3126 3127 static inline block_t data_blkaddr(struct inode *inode, 3128 struct folio *node_folio, unsigned int offset) 3129 { 3130 return le32_to_cpu(*(get_dnode_addr(inode, node_folio) + offset)); 3131 } 3132 3133 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 3134 { 3135 return data_blkaddr(dn->inode, dn->node_folio, dn->ofs_in_node); 3136 } 3137 3138 static inline int f2fs_test_bit(unsigned int nr, char *addr) 3139 { 3140 int mask; 3141 3142 addr += (nr >> 3); 3143 mask = BIT(7 - (nr & 0x07)); 3144 return mask & *addr; 3145 } 3146 3147 static inline void f2fs_set_bit(unsigned int nr, char *addr) 3148 { 3149 int mask; 3150 3151 addr += (nr >> 3); 3152 mask = BIT(7 - (nr & 0x07)); 3153 *addr |= mask; 3154 } 3155 3156 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 3157 { 3158 int mask; 3159 3160 addr += (nr >> 3); 3161 mask = BIT(7 - (nr & 0x07)); 3162 *addr &= ~mask; 3163 } 3164 3165 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3166 { 3167 int mask; 3168 int ret; 3169 3170 addr += (nr >> 3); 3171 mask = BIT(7 - (nr & 0x07)); 3172 ret = mask & *addr; 3173 *addr |= mask; 3174 return ret; 3175 } 3176 3177 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3178 { 3179 int mask; 3180 int ret; 3181 3182 addr += (nr >> 3); 3183 mask = BIT(7 - (nr & 0x07)); 3184 ret = mask & *addr; 3185 *addr &= ~mask; 3186 return ret; 3187 } 3188 3189 static inline void f2fs_change_bit(unsigned int nr, char *addr) 3190 { 3191 int mask; 3192 3193 addr += (nr >> 3); 3194 mask = BIT(7 - (nr & 0x07)); 3195 *addr ^= mask; 3196 } 3197 3198 /* 3199 * On-disk inode flags (f2fs_inode::i_flags) 3200 */ 3201 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3202 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3203 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3204 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3205 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3206 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3207 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3208 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3209 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3210 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3211 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3212 #define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3213 3214 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3215 3216 /* Flags that should be inherited by new inodes from their parent. */ 3217 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3218 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3219 F2FS_CASEFOLD_FL) 3220 3221 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3222 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3223 F2FS_CASEFOLD_FL)) 3224 3225 /* Flags that are appropriate for non-directories/regular files. */ 3226 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3227 3228 #define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3229 3230 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3231 { 3232 if (S_ISDIR(mode)) 3233 return flags; 3234 else if (S_ISREG(mode)) 3235 return flags & F2FS_REG_FLMASK; 3236 else 3237 return flags & F2FS_OTHER_FLMASK; 3238 } 3239 3240 static inline void __mark_inode_dirty_flag(struct inode *inode, 3241 int flag, bool set) 3242 { 3243 switch (flag) { 3244 case FI_INLINE_XATTR: 3245 case FI_INLINE_DATA: 3246 case FI_INLINE_DENTRY: 3247 case FI_NEW_INODE: 3248 if (set) 3249 return; 3250 fallthrough; 3251 case FI_DATA_EXIST: 3252 case FI_PIN_FILE: 3253 case FI_COMPRESS_RELEASED: 3254 f2fs_mark_inode_dirty_sync(inode, true); 3255 } 3256 } 3257 3258 static inline void set_inode_flag(struct inode *inode, int flag) 3259 { 3260 set_bit(flag, F2FS_I(inode)->flags); 3261 __mark_inode_dirty_flag(inode, flag, true); 3262 } 3263 3264 static inline int is_inode_flag_set(struct inode *inode, int flag) 3265 { 3266 return test_bit(flag, F2FS_I(inode)->flags); 3267 } 3268 3269 static inline void clear_inode_flag(struct inode *inode, int flag) 3270 { 3271 clear_bit(flag, F2FS_I(inode)->flags); 3272 __mark_inode_dirty_flag(inode, flag, false); 3273 } 3274 3275 static inline bool f2fs_verity_in_progress(struct inode *inode) 3276 { 3277 return IS_ENABLED(CONFIG_FS_VERITY) && 3278 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3279 } 3280 3281 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3282 { 3283 F2FS_I(inode)->i_acl_mode = mode; 3284 set_inode_flag(inode, FI_ACL_MODE); 3285 f2fs_mark_inode_dirty_sync(inode, false); 3286 } 3287 3288 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3289 { 3290 if (inc) 3291 inc_nlink(inode); 3292 else 3293 drop_nlink(inode); 3294 f2fs_mark_inode_dirty_sync(inode, true); 3295 } 3296 3297 static inline void f2fs_i_blocks_write(struct inode *inode, 3298 block_t diff, bool add, bool claim) 3299 { 3300 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3301 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3302 3303 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3304 if (add) { 3305 if (claim) 3306 dquot_claim_block(inode, diff); 3307 else 3308 dquot_alloc_block_nofail(inode, diff); 3309 } else { 3310 dquot_free_block(inode, diff); 3311 } 3312 3313 f2fs_mark_inode_dirty_sync(inode, true); 3314 if (clean || recover) 3315 set_inode_flag(inode, FI_AUTO_RECOVER); 3316 } 3317 3318 static inline bool f2fs_is_atomic_file(struct inode *inode); 3319 3320 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3321 { 3322 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3323 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3324 3325 if (i_size_read(inode) == i_size) 3326 return; 3327 3328 i_size_write(inode, i_size); 3329 3330 if (f2fs_is_atomic_file(inode)) 3331 return; 3332 3333 f2fs_mark_inode_dirty_sync(inode, true); 3334 if (clean || recover) 3335 set_inode_flag(inode, FI_AUTO_RECOVER); 3336 } 3337 3338 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3339 { 3340 F2FS_I(inode)->i_current_depth = depth; 3341 f2fs_mark_inode_dirty_sync(inode, true); 3342 } 3343 3344 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3345 unsigned int count) 3346 { 3347 F2FS_I(inode)->i_gc_failures = count; 3348 f2fs_mark_inode_dirty_sync(inode, true); 3349 } 3350 3351 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3352 { 3353 F2FS_I(inode)->i_xattr_nid = xnid; 3354 f2fs_mark_inode_dirty_sync(inode, true); 3355 } 3356 3357 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3358 { 3359 F2FS_I(inode)->i_pino = pino; 3360 f2fs_mark_inode_dirty_sync(inode, true); 3361 } 3362 3363 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3364 { 3365 struct f2fs_inode_info *fi = F2FS_I(inode); 3366 3367 if (ri->i_inline & F2FS_INLINE_XATTR) 3368 set_bit(FI_INLINE_XATTR, fi->flags); 3369 if (ri->i_inline & F2FS_INLINE_DATA) 3370 set_bit(FI_INLINE_DATA, fi->flags); 3371 if (ri->i_inline & F2FS_INLINE_DENTRY) 3372 set_bit(FI_INLINE_DENTRY, fi->flags); 3373 if (ri->i_inline & F2FS_DATA_EXIST) 3374 set_bit(FI_DATA_EXIST, fi->flags); 3375 if (ri->i_inline & F2FS_EXTRA_ATTR) 3376 set_bit(FI_EXTRA_ATTR, fi->flags); 3377 if (ri->i_inline & F2FS_PIN_FILE) 3378 set_bit(FI_PIN_FILE, fi->flags); 3379 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3380 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3381 } 3382 3383 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3384 { 3385 ri->i_inline = 0; 3386 3387 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3388 ri->i_inline |= F2FS_INLINE_XATTR; 3389 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3390 ri->i_inline |= F2FS_INLINE_DATA; 3391 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3392 ri->i_inline |= F2FS_INLINE_DENTRY; 3393 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3394 ri->i_inline |= F2FS_DATA_EXIST; 3395 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3396 ri->i_inline |= F2FS_EXTRA_ATTR; 3397 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3398 ri->i_inline |= F2FS_PIN_FILE; 3399 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3400 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3401 } 3402 3403 static inline int f2fs_has_extra_attr(struct inode *inode) 3404 { 3405 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3406 } 3407 3408 static inline int f2fs_has_inline_xattr(struct inode *inode) 3409 { 3410 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3411 } 3412 3413 static inline int f2fs_compressed_file(struct inode *inode) 3414 { 3415 return S_ISREG(inode->i_mode) && 3416 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3417 } 3418 3419 static inline bool f2fs_need_compress_data(struct inode *inode) 3420 { 3421 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3422 3423 if (!f2fs_compressed_file(inode)) 3424 return false; 3425 3426 if (compress_mode == COMPR_MODE_FS) 3427 return true; 3428 else if (compress_mode == COMPR_MODE_USER && 3429 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3430 return true; 3431 3432 return false; 3433 } 3434 3435 static inline unsigned int addrs_per_page(struct inode *inode, 3436 bool is_inode) 3437 { 3438 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) - 3439 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK; 3440 3441 if (f2fs_compressed_file(inode)) 3442 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3443 return addrs; 3444 } 3445 3446 static inline 3447 void *inline_xattr_addr(struct inode *inode, const struct folio *folio) 3448 { 3449 struct f2fs_inode *ri = F2FS_INODE(folio); 3450 3451 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3452 get_inline_xattr_addrs(inode)]); 3453 } 3454 3455 static inline int inline_xattr_size(struct inode *inode) 3456 { 3457 if (f2fs_has_inline_xattr(inode)) 3458 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3459 return 0; 3460 } 3461 3462 /* 3463 * Notice: check inline_data flag without inode page lock is unsafe. 3464 * It could change at any time by f2fs_convert_inline_folio(). 3465 */ 3466 static inline int f2fs_has_inline_data(struct inode *inode) 3467 { 3468 return is_inode_flag_set(inode, FI_INLINE_DATA); 3469 } 3470 3471 static inline int f2fs_exist_data(struct inode *inode) 3472 { 3473 return is_inode_flag_set(inode, FI_DATA_EXIST); 3474 } 3475 3476 static inline int f2fs_is_mmap_file(struct inode *inode) 3477 { 3478 return is_inode_flag_set(inode, FI_MMAP_FILE); 3479 } 3480 3481 static inline bool f2fs_is_pinned_file(struct inode *inode) 3482 { 3483 return is_inode_flag_set(inode, FI_PIN_FILE); 3484 } 3485 3486 static inline bool f2fs_is_atomic_file(struct inode *inode) 3487 { 3488 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3489 } 3490 3491 static inline bool f2fs_is_cow_file(struct inode *inode) 3492 { 3493 return is_inode_flag_set(inode, FI_COW_FILE); 3494 } 3495 3496 static inline void *inline_data_addr(struct inode *inode, struct folio *folio) 3497 { 3498 __le32 *addr = get_dnode_addr(inode, folio); 3499 3500 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3501 } 3502 3503 static inline int f2fs_has_inline_dentry(struct inode *inode) 3504 { 3505 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3506 } 3507 3508 static inline int is_file(struct inode *inode, int type) 3509 { 3510 return F2FS_I(inode)->i_advise & type; 3511 } 3512 3513 static inline void set_file(struct inode *inode, int type) 3514 { 3515 if (is_file(inode, type)) 3516 return; 3517 F2FS_I(inode)->i_advise |= type; 3518 f2fs_mark_inode_dirty_sync(inode, true); 3519 } 3520 3521 static inline void clear_file(struct inode *inode, int type) 3522 { 3523 if (!is_file(inode, type)) 3524 return; 3525 F2FS_I(inode)->i_advise &= ~type; 3526 f2fs_mark_inode_dirty_sync(inode, true); 3527 } 3528 3529 static inline bool f2fs_is_time_consistent(struct inode *inode) 3530 { 3531 struct timespec64 ts = inode_get_atime(inode); 3532 3533 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3534 return false; 3535 ts = inode_get_ctime(inode); 3536 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3537 return false; 3538 ts = inode_get_mtime(inode); 3539 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3540 return false; 3541 return true; 3542 } 3543 3544 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3545 { 3546 bool ret; 3547 3548 if (dsync) { 3549 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3550 3551 spin_lock(&sbi->inode_lock[DIRTY_META]); 3552 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3553 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3554 return ret; 3555 } 3556 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3557 file_keep_isize(inode) || 3558 i_size_read(inode) & ~PAGE_MASK) 3559 return false; 3560 3561 if (!f2fs_is_time_consistent(inode)) 3562 return false; 3563 3564 spin_lock(&F2FS_I(inode)->i_size_lock); 3565 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3566 spin_unlock(&F2FS_I(inode)->i_size_lock); 3567 3568 return ret; 3569 } 3570 3571 static inline bool f2fs_readonly(struct super_block *sb) 3572 { 3573 return sb_rdonly(sb); 3574 } 3575 3576 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3577 { 3578 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3579 } 3580 3581 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3582 size_t size, gfp_t flags) 3583 { 3584 if (time_to_inject(sbi, FAULT_KMALLOC)) 3585 return NULL; 3586 3587 return kmalloc(size, flags); 3588 } 3589 3590 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3591 { 3592 if (time_to_inject(sbi, FAULT_KMALLOC)) 3593 return NULL; 3594 3595 return __getname(); 3596 } 3597 3598 static inline void f2fs_putname(char *buf) 3599 { 3600 __putname(buf); 3601 } 3602 3603 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3604 size_t size, gfp_t flags) 3605 { 3606 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3607 } 3608 3609 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3610 size_t size, gfp_t flags) 3611 { 3612 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3613 return NULL; 3614 3615 return kvmalloc(size, flags); 3616 } 3617 3618 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3619 size_t size, gfp_t flags) 3620 { 3621 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3622 } 3623 3624 static inline void *f2fs_vmalloc(struct f2fs_sb_info *sbi, size_t size) 3625 { 3626 if (time_to_inject(sbi, FAULT_VMALLOC)) 3627 return NULL; 3628 3629 return vmalloc(size); 3630 } 3631 3632 static inline int get_extra_isize(struct inode *inode) 3633 { 3634 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3635 } 3636 3637 static inline int get_inline_xattr_addrs(struct inode *inode) 3638 { 3639 return F2FS_I(inode)->i_inline_xattr_size; 3640 } 3641 3642 #define f2fs_get_inode_mode(i) \ 3643 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3644 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3645 3646 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3647 3648 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3649 (offsetof(struct f2fs_inode, i_extra_end) - \ 3650 offsetof(struct f2fs_inode, i_extra_isize)) \ 3651 3652 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3653 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3654 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3655 sizeof((f2fs_inode)->field)) \ 3656 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3657 3658 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3659 3660 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3661 3662 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3663 block_t blkaddr, int type); 3664 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3665 block_t blkaddr, int type) 3666 { 3667 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3668 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3669 blkaddr, type); 3670 } 3671 3672 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3673 { 3674 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3675 blkaddr == COMPRESS_ADDR) 3676 return false; 3677 return true; 3678 } 3679 3680 /* 3681 * file.c 3682 */ 3683 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3684 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3685 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3686 int f2fs_truncate(struct inode *inode); 3687 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3688 struct kstat *stat, u32 request_mask, unsigned int flags); 3689 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3690 struct iattr *attr); 3691 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3692 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3693 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3694 bool readonly, bool need_lock); 3695 int f2fs_precache_extents(struct inode *inode); 3696 int f2fs_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3697 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3698 struct dentry *dentry, struct file_kattr *fa); 3699 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3700 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3701 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3702 int f2fs_pin_file_control(struct inode *inode, bool inc); 3703 3704 /* 3705 * inode.c 3706 */ 3707 void f2fs_set_inode_flags(struct inode *inode); 3708 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct folio *folio); 3709 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct folio *folio); 3710 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3711 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3712 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3713 void f2fs_update_inode(struct inode *inode, struct folio *node_folio); 3714 void f2fs_update_inode_page(struct inode *inode); 3715 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3716 void f2fs_remove_donate_inode(struct inode *inode); 3717 void f2fs_evict_inode(struct inode *inode); 3718 void f2fs_handle_failed_inode(struct inode *inode); 3719 3720 /* 3721 * namei.c 3722 */ 3723 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3724 bool hot, bool set); 3725 struct dentry *f2fs_get_parent(struct dentry *child); 3726 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3727 struct inode **new_inode); 3728 3729 /* 3730 * dir.c 3731 */ 3732 #if IS_ENABLED(CONFIG_UNICODE) 3733 int f2fs_init_casefolded_name(const struct inode *dir, 3734 struct f2fs_filename *fname); 3735 void f2fs_free_casefolded_name(struct f2fs_filename *fname); 3736 #else 3737 static inline int f2fs_init_casefolded_name(const struct inode *dir, 3738 struct f2fs_filename *fname) 3739 { 3740 return 0; 3741 } 3742 3743 static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 3744 { 3745 } 3746 #endif /* CONFIG_UNICODE */ 3747 3748 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3749 int lookup, struct f2fs_filename *fname); 3750 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3751 struct f2fs_filename *fname); 3752 void f2fs_free_filename(struct f2fs_filename *fname); 3753 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3754 const struct f2fs_filename *fname, int *max_slots, 3755 bool use_hash); 3756 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3757 unsigned int start_pos, struct fscrypt_str *fstr); 3758 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3759 struct f2fs_dentry_ptr *d); 3760 struct folio *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3761 const struct f2fs_filename *fname, struct folio *dfolio); 3762 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3763 unsigned int current_depth); 3764 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3765 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3766 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3767 const struct f2fs_filename *fname, struct folio **res_folio); 3768 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3769 const struct qstr *child, struct folio **res_folio); 3770 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct folio **f); 3771 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3772 struct folio **folio); 3773 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3774 struct folio *folio, struct inode *inode); 3775 bool f2fs_has_enough_room(struct inode *dir, struct folio *ifolio, 3776 const struct f2fs_filename *fname); 3777 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3778 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3779 unsigned int bit_pos); 3780 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3781 struct inode *inode, nid_t ino, umode_t mode); 3782 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3783 struct inode *inode, nid_t ino, umode_t mode); 3784 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3785 struct inode *inode, nid_t ino, umode_t mode); 3786 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct folio *folio, 3787 struct inode *dir, struct inode *inode); 3788 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3789 struct f2fs_filename *fname); 3790 bool f2fs_empty_dir(struct inode *dir); 3791 3792 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3793 { 3794 if (fscrypt_is_nokey_name(dentry)) 3795 return -ENOKEY; 3796 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3797 inode, inode->i_ino, inode->i_mode); 3798 } 3799 3800 /* 3801 * super.c 3802 */ 3803 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3804 void f2fs_inode_synced(struct inode *inode); 3805 int f2fs_dquot_initialize(struct inode *inode); 3806 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3807 int f2fs_do_quota_sync(struct super_block *sb, int type); 3808 loff_t max_file_blocks(struct inode *inode); 3809 void f2fs_quota_off_umount(struct super_block *sb); 3810 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3811 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3812 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3813 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3814 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3815 int f2fs_sync_fs(struct super_block *sb, int sync); 3816 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3817 3818 /* 3819 * hash.c 3820 */ 3821 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3822 3823 /* 3824 * node.c 3825 */ 3826 struct node_info; 3827 enum node_type; 3828 3829 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3830 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3831 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct folio *folio); 3832 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3833 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct folio *folio); 3834 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3835 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3836 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3837 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3838 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3839 struct node_info *ni, bool checkpoint_context); 3840 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3841 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3842 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3843 int f2fs_truncate_xattr_node(struct inode *inode); 3844 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3845 unsigned int seq_id); 3846 int f2fs_remove_inode_page(struct inode *inode); 3847 struct folio *f2fs_new_inode_folio(struct inode *inode); 3848 struct folio *f2fs_new_node_folio(struct dnode_of_data *dn, unsigned int ofs); 3849 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3850 struct folio *f2fs_get_node_folio(struct f2fs_sb_info *sbi, pgoff_t nid, 3851 enum node_type node_type); 3852 struct folio *f2fs_get_inode_folio(struct f2fs_sb_info *sbi, pgoff_t ino); 3853 struct folio *f2fs_get_xnode_folio(struct f2fs_sb_info *sbi, pgoff_t xnid); 3854 int f2fs_move_node_folio(struct folio *node_folio, int gc_type); 3855 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3856 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3857 struct writeback_control *wbc, bool atomic, 3858 unsigned int *seq_id); 3859 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3860 struct writeback_control *wbc, 3861 bool do_balance, enum iostat_type io_type); 3862 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3863 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3864 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3865 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3866 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3867 int f2fs_recover_inline_xattr(struct inode *inode, struct folio *folio); 3868 int f2fs_recover_xattr_data(struct inode *inode, struct folio *folio); 3869 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct folio *folio); 3870 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3871 unsigned int segno, struct f2fs_summary_block *sum); 3872 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3873 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3874 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3875 int __init f2fs_create_node_manager_caches(void); 3876 void f2fs_destroy_node_manager_caches(void); 3877 3878 /* 3879 * segment.c 3880 */ 3881 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3882 int f2fs_commit_atomic_write(struct inode *inode); 3883 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3884 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3885 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3886 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3887 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3888 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3889 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3890 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr, 3891 unsigned int len); 3892 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3893 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3894 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3895 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3896 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3897 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3898 struct cp_control *cpc); 3899 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3900 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3901 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3902 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3903 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3904 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3905 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3906 int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 3907 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3908 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3909 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3910 unsigned int start, unsigned int end); 3911 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3912 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3913 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3914 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3915 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3916 struct cp_control *cpc); 3917 struct folio *f2fs_get_sum_folio(struct f2fs_sb_info *sbi, unsigned int segno); 3918 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3919 block_t blk_addr); 3920 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio, 3921 enum iostat_type io_type); 3922 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3923 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3924 struct f2fs_io_info *fio); 3925 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3926 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3927 block_t old_blkaddr, block_t new_blkaddr, 3928 bool recover_curseg, bool recover_newaddr, 3929 bool from_gc); 3930 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3931 block_t old_addr, block_t new_addr, 3932 unsigned char version, bool recover_curseg, 3933 bool recover_newaddr); 3934 enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 3935 enum log_type seg_type); 3936 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct folio *folio, 3937 block_t old_blkaddr, block_t *new_blkaddr, 3938 struct f2fs_summary *sum, int type, 3939 struct f2fs_io_info *fio); 3940 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3941 block_t blkaddr, unsigned int blkcnt); 3942 void f2fs_folio_wait_writeback(struct folio *folio, enum page_type type, 3943 bool ordered, bool locked); 3944 #define f2fs_wait_on_page_writeback(page, type, ordered, locked) \ 3945 f2fs_folio_wait_writeback(page_folio(page), type, ordered, locked) 3946 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3947 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3948 block_t len); 3949 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3950 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3951 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3952 unsigned int val, int alloc); 3953 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3954 int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 3955 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3956 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3957 int __init f2fs_create_segment_manager_caches(void); 3958 void f2fs_destroy_segment_manager_caches(void); 3959 int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 3960 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3961 enum page_type type, enum temp_type temp); 3962 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 3963 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3964 unsigned int segno); 3965 unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 3966 unsigned int segno); 3967 3968 static inline struct inode *fio_inode(struct f2fs_io_info *fio) 3969 { 3970 return fio->folio->mapping->host; 3971 } 3972 3973 #define DEF_FRAGMENT_SIZE 4 3974 #define MIN_FRAGMENT_SIZE 1 3975 #define MAX_FRAGMENT_SIZE 512 3976 3977 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3978 { 3979 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3980 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3981 } 3982 3983 /* 3984 * checkpoint.c 3985 */ 3986 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3987 unsigned char reason); 3988 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3989 struct folio *f2fs_grab_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3990 struct folio *f2fs_get_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3991 struct folio *f2fs_get_meta_folio_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3992 struct folio *f2fs_get_tmp_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3993 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3994 block_t blkaddr, int type); 3995 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 3996 block_t blkaddr, int type); 3997 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3998 int type, bool sync); 3999 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 4000 unsigned int ra_blocks); 4001 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 4002 long nr_to_write, enum iostat_type io_type); 4003 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4004 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4005 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 4006 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 4007 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4008 unsigned int devidx, int type); 4009 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4010 unsigned int devidx, int type); 4011 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 4012 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 4013 void f2fs_add_orphan_inode(struct inode *inode); 4014 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 4015 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 4016 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 4017 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 4018 void f2fs_remove_dirty_inode(struct inode *inode); 4019 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 4020 bool from_cp); 4021 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 4022 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 4023 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4024 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 4025 int __init f2fs_create_checkpoint_caches(void); 4026 void f2fs_destroy_checkpoint_caches(void); 4027 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 4028 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 4029 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 4030 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 4031 4032 /* 4033 * data.c 4034 */ 4035 int __init f2fs_init_bioset(void); 4036 void f2fs_destroy_bioset(void); 4037 bool f2fs_is_cp_guaranteed(const struct folio *folio); 4038 int f2fs_init_bio_entry_cache(void); 4039 void f2fs_destroy_bio_entry_cache(void); 4040 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 4041 enum page_type type); 4042 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 4043 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 4044 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 4045 struct inode *inode, struct folio *folio, 4046 nid_t ino, enum page_type type); 4047 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 4048 struct bio **bio, struct folio *folio); 4049 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 4050 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 4051 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 4052 void f2fs_submit_page_write(struct f2fs_io_info *fio); 4053 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 4054 block_t blk_addr, sector_t *sector); 4055 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 4056 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4057 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4058 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 4059 int f2fs_reserve_new_block(struct dnode_of_data *dn); 4060 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 4061 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 4062 struct folio *f2fs_get_read_data_folio(struct inode *inode, pgoff_t index, 4063 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 4064 struct folio *f2fs_find_data_folio(struct inode *inode, pgoff_t index, 4065 pgoff_t *next_pgofs); 4066 struct folio *f2fs_get_lock_data_folio(struct inode *inode, pgoff_t index, 4067 bool for_write); 4068 struct folio *f2fs_get_new_data_folio(struct inode *inode, 4069 struct folio *ifolio, pgoff_t index, bool new_i_size); 4070 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 4071 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 4072 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 4073 u64 start, u64 len); 4074 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 4075 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 4076 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 4077 int f2fs_write_single_data_page(struct folio *folio, int *submitted, 4078 struct bio **bio, sector_t *last_block, 4079 struct writeback_control *wbc, 4080 enum iostat_type io_type, 4081 int compr_blocks, bool allow_balance); 4082 void f2fs_write_failed(struct inode *inode, loff_t to); 4083 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 4084 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 4085 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 4086 void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 4087 int f2fs_init_post_read_processing(void); 4088 void f2fs_destroy_post_read_processing(void); 4089 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 4090 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 4091 extern const struct iomap_ops f2fs_iomap_ops; 4092 4093 /* 4094 * gc.c 4095 */ 4096 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 4097 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 4098 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 4099 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 4100 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 4101 int f2fs_gc_range(struct f2fs_sb_info *sbi, 4102 unsigned int start_seg, unsigned int end_seg, 4103 bool dry_run, unsigned int dry_run_sections); 4104 int f2fs_resize_fs(struct file *filp, __u64 block_count); 4105 int __init f2fs_create_garbage_collection_cache(void); 4106 void f2fs_destroy_garbage_collection_cache(void); 4107 /* victim selection function for cleaning and SSR */ 4108 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 4109 int gc_type, int type, char alloc_mode, 4110 unsigned long long age, bool one_time); 4111 4112 /* 4113 * recovery.c 4114 */ 4115 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 4116 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 4117 int __init f2fs_create_recovery_cache(void); 4118 void f2fs_destroy_recovery_cache(void); 4119 4120 /* 4121 * debug.c 4122 */ 4123 #ifdef CONFIG_F2FS_STAT_FS 4124 enum { 4125 DEVSTAT_INUSE, 4126 DEVSTAT_DIRTY, 4127 DEVSTAT_FULL, 4128 DEVSTAT_FREE, 4129 DEVSTAT_PREFREE, 4130 DEVSTAT_MAX, 4131 }; 4132 4133 struct f2fs_dev_stats { 4134 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 4135 }; 4136 4137 struct f2fs_stat_info { 4138 struct list_head stat_list; 4139 struct f2fs_sb_info *sbi; 4140 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 4141 int main_area_segs, main_area_sections, main_area_zones; 4142 unsigned long long hit_cached[NR_EXTENT_CACHES]; 4143 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 4144 unsigned long long total_ext[NR_EXTENT_CACHES]; 4145 unsigned long long hit_total[NR_EXTENT_CACHES]; 4146 int ext_tree[NR_EXTENT_CACHES]; 4147 int zombie_tree[NR_EXTENT_CACHES]; 4148 int ext_node[NR_EXTENT_CACHES]; 4149 /* to count memory footprint */ 4150 unsigned long long ext_mem[NR_EXTENT_CACHES]; 4151 /* for read extent cache */ 4152 unsigned long long hit_largest; 4153 /* for block age extent cache */ 4154 unsigned long long allocated_data_blocks; 4155 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 4156 int ndirty_data, ndirty_qdata; 4157 unsigned int ndirty_dirs, ndirty_files, ndirty_all; 4158 unsigned int nquota_files, ndonate_files; 4159 int nats, dirty_nats, sits, dirty_sits; 4160 int free_nids, avail_nids, alloc_nids; 4161 int total_count, utilization; 4162 int nr_wb_cp_data, nr_wb_data; 4163 int nr_rd_data, nr_rd_node, nr_rd_meta; 4164 int nr_dio_read, nr_dio_write; 4165 unsigned int io_skip_bggc, other_skip_bggc; 4166 int nr_flushing, nr_flushed, flush_list_empty; 4167 int nr_discarding, nr_discarded; 4168 int nr_discard_cmd; 4169 unsigned int undiscard_blks; 4170 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 4171 unsigned int cur_ckpt_time, peak_ckpt_time; 4172 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 4173 int compr_inode, swapfile_inode; 4174 unsigned long long compr_blocks; 4175 int aw_cnt, max_aw_cnt; 4176 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 4177 unsigned int bimodal, avg_vblocks; 4178 int util_free, util_valid, util_invalid; 4179 int rsvd_segs, overp_segs; 4180 int dirty_count, node_pages, meta_pages, compress_pages; 4181 int compress_page_hit; 4182 int prefree_count, free_segs, free_secs; 4183 int cp_call_count[MAX_CALL_TYPE], cp_count; 4184 int gc_call_count[MAX_CALL_TYPE]; 4185 int gc_segs[2][2]; 4186 int gc_secs[2][2]; 4187 int tot_blks, data_blks, node_blks; 4188 int bg_data_blks, bg_node_blks; 4189 int curseg[NR_CURSEG_TYPE]; 4190 int cursec[NR_CURSEG_TYPE]; 4191 int curzone[NR_CURSEG_TYPE]; 4192 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4193 unsigned int full_seg[NR_CURSEG_TYPE]; 4194 unsigned int valid_blks[NR_CURSEG_TYPE]; 4195 4196 unsigned int meta_count[META_MAX]; 4197 unsigned int segment_count[2]; 4198 unsigned int block_count[2]; 4199 unsigned int inplace_count; 4200 unsigned long long base_mem, cache_mem, page_mem; 4201 struct f2fs_dev_stats *dev_stats; 4202 }; 4203 4204 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4205 { 4206 return (struct f2fs_stat_info *)sbi->stat_info; 4207 } 4208 4209 #define stat_inc_cp_call_count(sbi, foreground) \ 4210 atomic_inc(&sbi->cp_call_count[(foreground)]) 4211 #define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4212 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4213 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4214 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4215 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4216 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4217 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4218 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4219 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4220 #define stat_inc_inline_xattr(inode) \ 4221 do { \ 4222 if (f2fs_has_inline_xattr(inode)) \ 4223 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4224 } while (0) 4225 #define stat_dec_inline_xattr(inode) \ 4226 do { \ 4227 if (f2fs_has_inline_xattr(inode)) \ 4228 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4229 } while (0) 4230 #define stat_inc_inline_inode(inode) \ 4231 do { \ 4232 if (f2fs_has_inline_data(inode)) \ 4233 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4234 } while (0) 4235 #define stat_dec_inline_inode(inode) \ 4236 do { \ 4237 if (f2fs_has_inline_data(inode)) \ 4238 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4239 } while (0) 4240 #define stat_inc_inline_dir(inode) \ 4241 do { \ 4242 if (f2fs_has_inline_dentry(inode)) \ 4243 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4244 } while (0) 4245 #define stat_dec_inline_dir(inode) \ 4246 do { \ 4247 if (f2fs_has_inline_dentry(inode)) \ 4248 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4249 } while (0) 4250 #define stat_inc_compr_inode(inode) \ 4251 do { \ 4252 if (f2fs_compressed_file(inode)) \ 4253 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4254 } while (0) 4255 #define stat_dec_compr_inode(inode) \ 4256 do { \ 4257 if (f2fs_compressed_file(inode)) \ 4258 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4259 } while (0) 4260 #define stat_add_compr_blocks(inode, blocks) \ 4261 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4262 #define stat_sub_compr_blocks(inode, blocks) \ 4263 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4264 #define stat_inc_swapfile_inode(inode) \ 4265 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4266 #define stat_dec_swapfile_inode(inode) \ 4267 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4268 #define stat_inc_atomic_inode(inode) \ 4269 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4270 #define stat_dec_atomic_inode(inode) \ 4271 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4272 #define stat_inc_meta_count(sbi, blkaddr) \ 4273 do { \ 4274 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4275 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4276 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4277 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4278 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4279 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4280 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4281 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4282 } while (0) 4283 #define stat_inc_seg_type(sbi, curseg) \ 4284 ((sbi)->segment_count[(curseg)->alloc_type]++) 4285 #define stat_inc_block_count(sbi, curseg) \ 4286 ((sbi)->block_count[(curseg)->alloc_type]++) 4287 #define stat_inc_inplace_blocks(sbi) \ 4288 (atomic_inc(&(sbi)->inplace_count)) 4289 #define stat_update_max_atomic_write(inode) \ 4290 do { \ 4291 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4292 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4293 if (cur > max) \ 4294 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4295 } while (0) 4296 #define stat_inc_gc_call_count(sbi, foreground) \ 4297 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4298 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4299 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4300 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4301 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4302 4303 #define stat_inc_tot_blk_count(si, blks) \ 4304 ((si)->tot_blks += (blks)) 4305 4306 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4307 do { \ 4308 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4309 stat_inc_tot_blk_count(si, blks); \ 4310 si->data_blks += (blks); \ 4311 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4312 } while (0) 4313 4314 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4315 do { \ 4316 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4317 stat_inc_tot_blk_count(si, blks); \ 4318 si->node_blks += (blks); \ 4319 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4320 } while (0) 4321 4322 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4323 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4324 void __init f2fs_create_root_stats(void); 4325 void f2fs_destroy_root_stats(void); 4326 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4327 #else 4328 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4329 #define stat_inc_cp_count(sbi) do { } while (0) 4330 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4331 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4332 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4333 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4334 #define stat_inc_total_hit(sbi, type) do { } while (0) 4335 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4336 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4337 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4338 #define stat_inc_inline_xattr(inode) do { } while (0) 4339 #define stat_dec_inline_xattr(inode) do { } while (0) 4340 #define stat_inc_inline_inode(inode) do { } while (0) 4341 #define stat_dec_inline_inode(inode) do { } while (0) 4342 #define stat_inc_inline_dir(inode) do { } while (0) 4343 #define stat_dec_inline_dir(inode) do { } while (0) 4344 #define stat_inc_compr_inode(inode) do { } while (0) 4345 #define stat_dec_compr_inode(inode) do { } while (0) 4346 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4347 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4348 #define stat_inc_swapfile_inode(inode) do { } while (0) 4349 #define stat_dec_swapfile_inode(inode) do { } while (0) 4350 #define stat_inc_atomic_inode(inode) do { } while (0) 4351 #define stat_dec_atomic_inode(inode) do { } while (0) 4352 #define stat_update_max_atomic_write(inode) do { } while (0) 4353 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4354 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4355 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4356 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4357 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4358 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4359 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4360 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4361 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4362 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4363 4364 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4365 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4366 static inline void __init f2fs_create_root_stats(void) { } 4367 static inline void f2fs_destroy_root_stats(void) { } 4368 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4369 #endif 4370 4371 extern const struct file_operations f2fs_dir_operations; 4372 extern const struct file_operations f2fs_file_operations; 4373 extern const struct inode_operations f2fs_file_inode_operations; 4374 extern const struct address_space_operations f2fs_dblock_aops; 4375 extern const struct address_space_operations f2fs_node_aops; 4376 extern const struct address_space_operations f2fs_meta_aops; 4377 extern const struct inode_operations f2fs_dir_inode_operations; 4378 extern const struct inode_operations f2fs_symlink_inode_operations; 4379 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4380 extern const struct inode_operations f2fs_special_inode_operations; 4381 extern struct kmem_cache *f2fs_inode_entry_slab; 4382 4383 /* 4384 * inline.c 4385 */ 4386 bool f2fs_may_inline_data(struct inode *inode); 4387 bool f2fs_sanity_check_inline_data(struct inode *inode, struct folio *ifolio); 4388 bool f2fs_may_inline_dentry(struct inode *inode); 4389 void f2fs_do_read_inline_data(struct folio *folio, struct folio *ifolio); 4390 void f2fs_truncate_inline_inode(struct inode *inode, struct folio *ifolio, 4391 u64 from); 4392 int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4393 int f2fs_convert_inline_folio(struct dnode_of_data *dn, struct folio *folio); 4394 int f2fs_convert_inline_inode(struct inode *inode); 4395 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4396 int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4397 int f2fs_recover_inline_data(struct inode *inode, struct folio *nfolio); 4398 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4399 const struct f2fs_filename *fname, struct folio **res_folio, 4400 bool use_hash); 4401 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4402 struct folio *ifolio); 4403 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4404 struct inode *inode, nid_t ino, umode_t mode); 4405 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4406 struct folio *folio, struct inode *dir, struct inode *inode); 4407 bool f2fs_empty_inline_dir(struct inode *dir); 4408 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4409 struct fscrypt_str *fstr); 4410 int f2fs_inline_data_fiemap(struct inode *inode, 4411 struct fiemap_extent_info *fieinfo, 4412 __u64 start, __u64 len); 4413 4414 /* 4415 * shrinker.c 4416 */ 4417 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4418 struct shrink_control *sc); 4419 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4420 struct shrink_control *sc); 4421 unsigned int f2fs_donate_files(void); 4422 void f2fs_reclaim_caches(unsigned int reclaim_caches_kb); 4423 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4424 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4425 4426 /* 4427 * extent_cache.c 4428 */ 4429 bool sanity_check_extent_cache(struct inode *inode, struct folio *ifolio); 4430 void f2fs_init_extent_tree(struct inode *inode); 4431 void f2fs_drop_extent_tree(struct inode *inode); 4432 void f2fs_destroy_extent_node(struct inode *inode); 4433 void f2fs_destroy_extent_tree(struct inode *inode); 4434 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4435 int __init f2fs_create_extent_cache(void); 4436 void f2fs_destroy_extent_cache(void); 4437 4438 /* read extent cache ops */ 4439 void f2fs_init_read_extent_tree(struct inode *inode, struct folio *ifolio); 4440 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4441 struct extent_info *ei); 4442 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4443 block_t *blkaddr); 4444 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4445 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4446 pgoff_t fofs, block_t blkaddr, unsigned int len); 4447 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4448 int nr_shrink); 4449 4450 /* block age extent cache ops */ 4451 void f2fs_init_age_extent_tree(struct inode *inode); 4452 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4453 struct extent_info *ei); 4454 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4455 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4456 pgoff_t fofs, unsigned int len); 4457 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4458 int nr_shrink); 4459 4460 /* 4461 * sysfs.c 4462 */ 4463 #define MIN_RA_MUL 2 4464 #define MAX_RA_MUL 256 4465 4466 int __init f2fs_init_sysfs(void); 4467 void f2fs_exit_sysfs(void); 4468 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4469 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4470 4471 /* verity.c */ 4472 extern const struct fsverity_operations f2fs_verityops; 4473 4474 /* 4475 * crypto support 4476 */ 4477 static inline bool f2fs_encrypted_file(struct inode *inode) 4478 { 4479 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4480 } 4481 4482 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4483 { 4484 #ifdef CONFIG_FS_ENCRYPTION 4485 file_set_encrypt(inode); 4486 f2fs_set_inode_flags(inode); 4487 #endif 4488 } 4489 4490 /* 4491 * Returns true if the reads of the inode's data need to undergo some 4492 * postprocessing step, like decryption or authenticity verification. 4493 */ 4494 static inline bool f2fs_post_read_required(struct inode *inode) 4495 { 4496 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4497 f2fs_compressed_file(inode); 4498 } 4499 4500 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4501 { 4502 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4503 } 4504 4505 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4506 { 4507 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4508 } 4509 4510 /* 4511 * compress.c 4512 */ 4513 #ifdef CONFIG_F2FS_FS_COMPRESSION 4514 enum cluster_check_type { 4515 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4516 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4517 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4518 }; 4519 bool f2fs_is_compressed_page(struct folio *folio); 4520 struct folio *f2fs_compress_control_folio(struct folio *folio); 4521 int f2fs_prepare_compress_overwrite(struct inode *inode, 4522 struct page **pagep, pgoff_t index, void **fsdata); 4523 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4524 pgoff_t index, unsigned copied); 4525 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4526 void f2fs_compress_write_end_io(struct bio *bio, struct folio *folio); 4527 bool f2fs_is_compress_backend_ready(struct inode *inode); 4528 bool f2fs_is_compress_level_valid(int alg, int lvl); 4529 int __init f2fs_init_compress_mempool(void); 4530 void f2fs_destroy_compress_mempool(void); 4531 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4532 void f2fs_end_read_compressed_page(struct folio *folio, bool failed, 4533 block_t blkaddr, bool in_task); 4534 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4535 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4536 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4537 int index, int nr_pages, bool uptodate); 4538 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4539 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4540 int f2fs_write_multi_pages(struct compress_ctx *cc, 4541 int *submitted, 4542 struct writeback_control *wbc, 4543 enum iostat_type io_type); 4544 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4545 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4546 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4547 pgoff_t fofs, block_t blkaddr, 4548 unsigned int llen, unsigned int c_len); 4549 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4550 unsigned nr_pages, sector_t *last_block_in_bio, 4551 struct readahead_control *rac, bool for_write); 4552 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4553 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4554 bool in_task); 4555 void f2fs_put_folio_dic(struct folio *folio, bool in_task); 4556 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4557 unsigned int ofs_in_node); 4558 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4559 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4560 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4561 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4562 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4563 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4564 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4565 int __init f2fs_init_compress_cache(void); 4566 void f2fs_destroy_compress_cache(void); 4567 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4568 void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4569 block_t blkaddr, unsigned int len); 4570 bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, struct folio *folio, 4571 block_t blkaddr); 4572 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4573 #define inc_compr_inode_stat(inode) \ 4574 do { \ 4575 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4576 sbi->compr_new_inode++; \ 4577 } while (0) 4578 #define add_compr_block_stat(inode, blocks) \ 4579 do { \ 4580 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4581 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4582 sbi->compr_written_block += blocks; \ 4583 sbi->compr_saved_block += diff; \ 4584 } while (0) 4585 #else 4586 static inline bool f2fs_is_compressed_page(struct folio *folio) { return false; } 4587 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4588 { 4589 if (!f2fs_compressed_file(inode)) 4590 return true; 4591 /* not support compression */ 4592 return false; 4593 } 4594 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4595 static inline struct folio *f2fs_compress_control_folio(struct folio *folio) 4596 { 4597 WARN_ON_ONCE(1); 4598 return ERR_PTR(-EINVAL); 4599 } 4600 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4601 static inline void f2fs_destroy_compress_mempool(void) { } 4602 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4603 bool in_task) { } 4604 static inline void f2fs_end_read_compressed_page(struct folio *folio, 4605 bool failed, block_t blkaddr, bool in_task) 4606 { 4607 WARN_ON_ONCE(1); 4608 } 4609 static inline void f2fs_put_folio_dic(struct folio *folio, bool in_task) 4610 { 4611 WARN_ON_ONCE(1); 4612 } 4613 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4614 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4615 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4616 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4617 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4618 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4619 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4620 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4621 static inline void f2fs_destroy_compress_cache(void) { } 4622 static inline void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4623 block_t blkaddr, unsigned int len) { } 4624 static inline bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, 4625 struct folio *folio, block_t blkaddr) { return false; } 4626 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4627 nid_t ino) { } 4628 #define inc_compr_inode_stat(inode) do { } while (0) 4629 static inline int f2fs_is_compressed_cluster( 4630 struct inode *inode, 4631 pgoff_t index) { return 0; } 4632 static inline bool f2fs_is_sparse_cluster( 4633 struct inode *inode, 4634 pgoff_t index) { return true; } 4635 static inline void f2fs_update_read_extent_tree_range_compressed( 4636 struct inode *inode, 4637 pgoff_t fofs, block_t blkaddr, 4638 unsigned int llen, unsigned int c_len) { } 4639 #endif 4640 4641 static inline int set_compress_context(struct inode *inode) 4642 { 4643 #ifdef CONFIG_F2FS_FS_COMPRESSION 4644 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4645 struct f2fs_inode_info *fi = F2FS_I(inode); 4646 4647 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4648 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4649 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4650 BIT(COMPRESS_CHKSUM) : 0; 4651 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4652 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4653 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4654 F2FS_OPTION(sbi).compress_level) 4655 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4656 fi->i_flags |= F2FS_COMPR_FL; 4657 set_inode_flag(inode, FI_COMPRESSED_FILE); 4658 stat_inc_compr_inode(inode); 4659 inc_compr_inode_stat(inode); 4660 f2fs_mark_inode_dirty_sync(inode, true); 4661 return 0; 4662 #else 4663 return -EOPNOTSUPP; 4664 #endif 4665 } 4666 4667 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4668 { 4669 struct f2fs_inode_info *fi = F2FS_I(inode); 4670 4671 f2fs_down_write(&fi->i_sem); 4672 4673 if (!f2fs_compressed_file(inode)) { 4674 f2fs_up_write(&fi->i_sem); 4675 return true; 4676 } 4677 if (f2fs_is_mmap_file(inode) || 4678 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4679 f2fs_up_write(&fi->i_sem); 4680 return false; 4681 } 4682 4683 fi->i_flags &= ~F2FS_COMPR_FL; 4684 stat_dec_compr_inode(inode); 4685 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4686 f2fs_mark_inode_dirty_sync(inode, true); 4687 4688 f2fs_up_write(&fi->i_sem); 4689 return true; 4690 } 4691 4692 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4693 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4694 { \ 4695 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4696 } 4697 4698 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4699 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4700 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4701 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4702 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4703 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4704 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4705 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4706 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4707 F2FS_FEATURE_FUNCS(verity, VERITY); 4708 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4709 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4710 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4711 F2FS_FEATURE_FUNCS(readonly, RO); 4712 F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 4713 4714 #ifdef CONFIG_BLK_DEV_ZONED 4715 static inline bool f2fs_zone_is_seq(struct f2fs_sb_info *sbi, int devi, 4716 unsigned int zone) 4717 { 4718 return test_bit(zone, FDEV(devi).blkz_seq); 4719 } 4720 4721 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4722 block_t blkaddr) 4723 { 4724 return f2fs_zone_is_seq(sbi, devi, blkaddr / sbi->blocks_per_blkz); 4725 } 4726 #endif 4727 4728 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4729 struct block_device *bdev) 4730 { 4731 int i; 4732 4733 if (!f2fs_is_multi_device(sbi)) 4734 return 0; 4735 4736 for (i = 0; i < sbi->s_ndevs; i++) 4737 if (FDEV(i).bdev == bdev) 4738 return i; 4739 4740 WARN_ON(1); 4741 return -1; 4742 } 4743 4744 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4745 { 4746 return f2fs_sb_has_blkzoned(sbi); 4747 } 4748 4749 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4750 { 4751 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4752 } 4753 4754 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4755 { 4756 int i; 4757 4758 if (!f2fs_is_multi_device(sbi)) 4759 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4760 4761 for (i = 0; i < sbi->s_ndevs; i++) 4762 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4763 return true; 4764 return false; 4765 } 4766 4767 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4768 { 4769 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4770 f2fs_hw_should_discard(sbi); 4771 } 4772 4773 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4774 { 4775 int i; 4776 4777 if (!f2fs_is_multi_device(sbi)) 4778 return bdev_read_only(sbi->sb->s_bdev); 4779 4780 for (i = 0; i < sbi->s_ndevs; i++) 4781 if (bdev_read_only(FDEV(i).bdev)) 4782 return true; 4783 return false; 4784 } 4785 4786 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4787 { 4788 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4789 } 4790 4791 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4792 { 4793 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4794 } 4795 4796 static inline bool f2fs_is_sequential_zone_area(struct f2fs_sb_info *sbi, 4797 block_t blkaddr) 4798 { 4799 if (f2fs_sb_has_blkzoned(sbi)) { 4800 #ifdef CONFIG_BLK_DEV_ZONED 4801 int devi = f2fs_target_device_index(sbi, blkaddr); 4802 4803 if (!bdev_is_zoned(FDEV(devi).bdev)) 4804 return false; 4805 4806 if (f2fs_is_multi_device(sbi)) { 4807 if (blkaddr < FDEV(devi).start_blk || 4808 blkaddr > FDEV(devi).end_blk) { 4809 f2fs_err(sbi, "Invalid block %x", blkaddr); 4810 return false; 4811 } 4812 blkaddr -= FDEV(devi).start_blk; 4813 } 4814 4815 return f2fs_blkz_is_seq(sbi, devi, blkaddr); 4816 #else 4817 return false; 4818 #endif 4819 } 4820 return false; 4821 } 4822 4823 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4824 { 4825 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4826 } 4827 4828 static inline bool f2fs_may_compress(struct inode *inode) 4829 { 4830 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4831 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4832 f2fs_is_mmap_file(inode)) 4833 return false; 4834 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4835 } 4836 4837 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4838 u64 blocks, bool add) 4839 { 4840 struct f2fs_inode_info *fi = F2FS_I(inode); 4841 int diff = fi->i_cluster_size - blocks; 4842 4843 /* don't update i_compr_blocks if saved blocks were released */ 4844 if (!add && !atomic_read(&fi->i_compr_blocks)) 4845 return; 4846 4847 if (add) { 4848 atomic_add(diff, &fi->i_compr_blocks); 4849 stat_add_compr_blocks(inode, diff); 4850 } else { 4851 atomic_sub(diff, &fi->i_compr_blocks); 4852 stat_sub_compr_blocks(inode, diff); 4853 } 4854 f2fs_mark_inode_dirty_sync(inode, true); 4855 } 4856 4857 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4858 int flag) 4859 { 4860 if (!f2fs_is_multi_device(sbi)) 4861 return false; 4862 if (flag != F2FS_GET_BLOCK_DIO) 4863 return false; 4864 return sbi->aligned_blksize; 4865 } 4866 4867 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4868 { 4869 return fsverity_active(inode) && 4870 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4871 } 4872 4873 #ifdef CONFIG_F2FS_FAULT_INJECTION 4874 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4875 unsigned long type, enum fault_option fo); 4876 #else 4877 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4878 unsigned long rate, unsigned long type, 4879 enum fault_option fo) 4880 { 4881 return 0; 4882 } 4883 #endif 4884 4885 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4886 { 4887 #ifdef CONFIG_QUOTA 4888 if (f2fs_sb_has_quota_ino(sbi)) 4889 return true; 4890 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4891 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4892 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4893 return true; 4894 #endif 4895 return false; 4896 } 4897 4898 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4899 { 4900 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4901 } 4902 4903 static inline void f2fs_io_schedule_timeout(long timeout) 4904 { 4905 set_current_state(TASK_UNINTERRUPTIBLE); 4906 io_schedule_timeout(timeout); 4907 } 4908 4909 static inline void f2fs_io_schedule_timeout_killable(long timeout) 4910 { 4911 while (timeout) { 4912 if (fatal_signal_pending(current)) 4913 return; 4914 set_current_state(TASK_UNINTERRUPTIBLE); 4915 io_schedule_timeout(DEFAULT_IO_TIMEOUT); 4916 if (timeout <= DEFAULT_IO_TIMEOUT) 4917 return; 4918 timeout -= DEFAULT_IO_TIMEOUT; 4919 } 4920 } 4921 4922 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 4923 struct folio *folio, enum page_type type) 4924 { 4925 pgoff_t ofs = folio->index; 4926 4927 if (unlikely(f2fs_cp_error(sbi))) 4928 return; 4929 4930 if (ofs == sbi->page_eio_ofs[type]) { 4931 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4932 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4933 } else { 4934 sbi->page_eio_ofs[type] = ofs; 4935 sbi->page_eio_cnt[type] = 0; 4936 } 4937 } 4938 4939 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4940 { 4941 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4942 } 4943 4944 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4945 block_t blkaddr, unsigned int cnt) 4946 { 4947 bool need_submit = false; 4948 int i = 0; 4949 4950 do { 4951 struct folio *folio; 4952 4953 folio = filemap_get_folio(META_MAPPING(sbi), blkaddr + i); 4954 if (!IS_ERR(folio)) { 4955 if (folio_test_writeback(folio)) 4956 need_submit = true; 4957 f2fs_folio_put(folio, false); 4958 } 4959 } while (++i < cnt && !need_submit); 4960 4961 if (need_submit) 4962 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4963 NULL, 0, DATA); 4964 4965 truncate_inode_pages_range(META_MAPPING(sbi), 4966 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4967 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4968 } 4969 4970 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4971 block_t blkaddr, unsigned int len) 4972 { 4973 f2fs_truncate_meta_inode_pages(sbi, blkaddr, len); 4974 f2fs_invalidate_compress_pages_range(sbi, blkaddr, len); 4975 } 4976 4977 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4978 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4979 4980 #endif /* _LINUX_F2FS_H */ 4981