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