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