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; /* compleation */ 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 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 736 pgoff_t *m_next_extent; /* point to next possible extent */ 737 int m_seg_type; 738 bool m_may_create; /* indicate it is from write path */ 739 bool m_multidev_dio; /* indicate it allows multi-device dio */ 740 }; 741 742 /* for flag in get_data_block */ 743 enum { 744 F2FS_GET_BLOCK_DEFAULT, 745 F2FS_GET_BLOCK_FIEMAP, 746 F2FS_GET_BLOCK_BMAP, 747 F2FS_GET_BLOCK_DIO, 748 F2FS_GET_BLOCK_PRE_DIO, 749 F2FS_GET_BLOCK_PRE_AIO, 750 F2FS_GET_BLOCK_PRECACHE, 751 }; 752 753 /* 754 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 755 */ 756 #define FADVISE_COLD_BIT 0x01 757 #define FADVISE_LOST_PINO_BIT 0x02 758 #define FADVISE_ENCRYPT_BIT 0x04 759 #define FADVISE_ENC_NAME_BIT 0x08 760 #define FADVISE_KEEP_SIZE_BIT 0x10 761 #define FADVISE_HOT_BIT 0x20 762 #define FADVISE_VERITY_BIT 0x40 763 #define FADVISE_TRUNC_BIT 0x80 764 765 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 766 767 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 768 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 769 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 770 771 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 772 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 773 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 774 775 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 776 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 777 778 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 779 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 780 781 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 782 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 783 784 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 785 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 786 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 787 788 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 789 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 790 791 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 792 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 793 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 794 795 #define DEF_DIR_LEVEL 0 796 797 /* used for f2fs_inode_info->flags */ 798 enum { 799 FI_NEW_INODE, /* indicate newly allocated inode */ 800 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 801 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 802 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 803 FI_INC_LINK, /* need to increment i_nlink */ 804 FI_ACL_MODE, /* indicate acl mode */ 805 FI_NO_ALLOC, /* should not allocate any blocks */ 806 FI_FREE_NID, /* free allocated nide */ 807 FI_NO_EXTENT, /* not to use the extent cache */ 808 FI_INLINE_XATTR, /* used for inline xattr */ 809 FI_INLINE_DATA, /* used for inline data*/ 810 FI_INLINE_DENTRY, /* used for inline dentry */ 811 FI_APPEND_WRITE, /* inode has appended data */ 812 FI_UPDATE_WRITE, /* inode has in-place-update data */ 813 FI_NEED_IPU, /* used for ipu per file */ 814 FI_ATOMIC_FILE, /* indicate atomic file */ 815 FI_DATA_EXIST, /* indicate data exists */ 816 FI_SKIP_WRITES, /* should skip data page writeback */ 817 FI_OPU_WRITE, /* used for opu per file */ 818 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 819 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 820 FI_HOT_DATA, /* indicate file is hot */ 821 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 822 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 823 FI_PIN_FILE, /* indicate file should not be gced */ 824 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 825 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 826 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 827 FI_MMAP_FILE, /* indicate file was mmapped */ 828 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 829 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 830 FI_ALIGNED_WRITE, /* enable aligned write */ 831 FI_COW_FILE, /* indicate COW file */ 832 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 833 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 834 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 835 FI_OPENED_FILE, /* indicate file has been opened */ 836 FI_DONATE_FINISHED, /* indicate page donation of file has been finished */ 837 FI_MAX, /* max flag, never be used */ 838 }; 839 840 struct f2fs_inode_info { 841 struct inode vfs_inode; /* serve a vfs inode */ 842 unsigned long i_flags; /* keep an inode flags for ioctl */ 843 unsigned char i_advise; /* use to give file attribute hints */ 844 unsigned char i_dir_level; /* use for dentry level for large dir */ 845 union { 846 unsigned int i_current_depth; /* only for directory depth */ 847 unsigned short i_gc_failures; /* for gc failure statistic */ 848 }; 849 unsigned int i_pino; /* parent inode number */ 850 umode_t i_acl_mode; /* keep file acl mode temporarily */ 851 852 /* Use below internally in f2fs*/ 853 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 854 unsigned int ioprio_hint; /* hint for IO priority */ 855 struct f2fs_rwsem i_sem; /* protect fi info */ 856 atomic_t dirty_pages; /* # of dirty pages */ 857 f2fs_hash_t chash; /* hash value of given file name */ 858 unsigned int clevel; /* maximum level of given file name */ 859 struct task_struct *task; /* lookup and create consistency */ 860 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 861 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 862 nid_t i_xattr_nid; /* node id that contains xattrs */ 863 loff_t last_disk_size; /* lastly written file size */ 864 spinlock_t i_size_lock; /* protect last_disk_size */ 865 866 #ifdef CONFIG_QUOTA 867 struct dquot __rcu *i_dquot[MAXQUOTAS]; 868 869 /* quota space reservation, managed internally by quota code */ 870 qsize_t i_reserved_quota; 871 #endif 872 struct list_head dirty_list; /* dirty list for dirs and files */ 873 struct list_head gdirty_list; /* linked in global dirty list */ 874 875 /* linked in global inode list for cache donation */ 876 struct list_head gdonate_list; 877 pgoff_t donate_start, donate_end; /* inclusive */ 878 879 struct task_struct *atomic_write_task; /* store atomic write task */ 880 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 881 /* cached extent_tree entry */ 882 union { 883 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 884 struct inode *atomic_inode; 885 /* point to atomic_inode, available only for cow_inode */ 886 }; 887 888 /* avoid racing between foreground op and gc */ 889 struct f2fs_rwsem i_gc_rwsem[2]; 890 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 891 892 int i_extra_isize; /* size of extra space located in i_addr */ 893 kprojid_t i_projid; /* id for project quota */ 894 int i_inline_xattr_size; /* inline xattr size */ 895 struct timespec64 i_crtime; /* inode creation time */ 896 struct timespec64 i_disk_time[3];/* inode disk times */ 897 898 /* for file compress */ 899 atomic_t i_compr_blocks; /* # of compressed blocks */ 900 unsigned char i_compress_algorithm; /* algorithm type */ 901 unsigned char i_log_cluster_size; /* log of cluster size */ 902 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 903 unsigned char i_compress_flag; /* compress flag */ 904 unsigned int i_cluster_size; /* cluster size */ 905 906 unsigned int atomic_write_cnt; 907 loff_t original_i_size; /* original i_size before atomic write */ 908 }; 909 910 static inline void get_read_extent_info(struct extent_info *ext, 911 struct f2fs_extent *i_ext) 912 { 913 ext->fofs = le32_to_cpu(i_ext->fofs); 914 ext->blk = le32_to_cpu(i_ext->blk); 915 ext->len = le32_to_cpu(i_ext->len); 916 } 917 918 static inline void set_raw_read_extent(struct extent_info *ext, 919 struct f2fs_extent *i_ext) 920 { 921 i_ext->fofs = cpu_to_le32(ext->fofs); 922 i_ext->blk = cpu_to_le32(ext->blk); 923 i_ext->len = cpu_to_le32(ext->len); 924 } 925 926 static inline bool __is_discard_mergeable(struct discard_info *back, 927 struct discard_info *front, unsigned int max_len) 928 { 929 return (back->lstart + back->len == front->lstart) && 930 (back->len + front->len <= max_len); 931 } 932 933 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 934 struct discard_info *back, unsigned int max_len) 935 { 936 return __is_discard_mergeable(back, cur, max_len); 937 } 938 939 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 940 struct discard_info *front, unsigned int max_len) 941 { 942 return __is_discard_mergeable(cur, front, max_len); 943 } 944 945 /* 946 * For free nid management 947 */ 948 enum nid_state { 949 FREE_NID, /* newly added to free nid list */ 950 PREALLOC_NID, /* it is preallocated */ 951 MAX_NID_STATE, 952 }; 953 954 enum nat_state { 955 TOTAL_NAT, 956 DIRTY_NAT, 957 RECLAIMABLE_NAT, 958 MAX_NAT_STATE, 959 }; 960 961 struct f2fs_nm_info { 962 block_t nat_blkaddr; /* base disk address of NAT */ 963 nid_t max_nid; /* maximum possible node ids */ 964 nid_t available_nids; /* # of available node ids */ 965 nid_t next_scan_nid; /* the next nid to be scanned */ 966 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 967 unsigned int ram_thresh; /* control the memory footprint */ 968 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 969 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 970 971 /* NAT cache management */ 972 struct radix_tree_root nat_root;/* root of the nat entry cache */ 973 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 974 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 975 struct list_head nat_entries; /* cached nat entry list (clean) */ 976 spinlock_t nat_list_lock; /* protect clean nat entry list */ 977 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 978 unsigned int nat_blocks; /* # of nat blocks */ 979 980 /* free node ids management */ 981 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 982 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 983 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 984 spinlock_t nid_list_lock; /* protect nid lists ops */ 985 struct mutex build_lock; /* lock for build free nids */ 986 unsigned char **free_nid_bitmap; 987 unsigned char *nat_block_bitmap; 988 unsigned short *free_nid_count; /* free nid count of NAT block */ 989 990 /* for checkpoint */ 991 char *nat_bitmap; /* NAT bitmap pointer */ 992 993 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 994 unsigned char *nat_bits; /* NAT bits blocks */ 995 unsigned char *full_nat_bits; /* full NAT pages */ 996 unsigned char *empty_nat_bits; /* empty NAT pages */ 997 #ifdef CONFIG_F2FS_CHECK_FS 998 char *nat_bitmap_mir; /* NAT bitmap mirror */ 999 #endif 1000 int bitmap_size; /* bitmap size */ 1001 }; 1002 1003 /* 1004 * this structure is used as one of function parameters. 1005 * all the information are dedicated to a given direct node block determined 1006 * by the data offset in a file. 1007 */ 1008 struct dnode_of_data { 1009 struct inode *inode; /* vfs inode pointer */ 1010 struct folio *inode_folio; /* its inode folio, NULL is possible */ 1011 struct folio *node_folio; /* cached direct node folio */ 1012 nid_t nid; /* node id of the direct node block */ 1013 unsigned int ofs_in_node; /* data offset in the node page */ 1014 bool inode_folio_locked; /* inode folio is locked or not */ 1015 bool node_changed; /* is node block changed */ 1016 char cur_level; /* level of hole node page */ 1017 char max_level; /* level of current page located */ 1018 block_t data_blkaddr; /* block address of the node block */ 1019 }; 1020 1021 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 1022 struct folio *ifolio, struct folio *nfolio, nid_t nid) 1023 { 1024 memset(dn, 0, sizeof(*dn)); 1025 dn->inode = inode; 1026 dn->inode_folio = ifolio; 1027 dn->node_folio = nfolio; 1028 dn->nid = nid; 1029 } 1030 1031 /* 1032 * For SIT manager 1033 * 1034 * By default, there are 6 active log areas across the whole main area. 1035 * When considering hot and cold data separation to reduce cleaning overhead, 1036 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1037 * respectively. 1038 * In the current design, you should not change the numbers intentionally. 1039 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1040 * logs individually according to the underlying devices. (default: 6) 1041 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1042 * data and 8 for node logs. 1043 */ 1044 #define NR_CURSEG_DATA_TYPE (3) 1045 #define NR_CURSEG_NODE_TYPE (3) 1046 #define NR_CURSEG_INMEM_TYPE (2) 1047 #define NR_CURSEG_RO_TYPE (2) 1048 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1049 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1050 1051 enum log_type { 1052 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1053 CURSEG_WARM_DATA, /* data blocks */ 1054 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1055 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1056 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1057 CURSEG_COLD_NODE, /* indirect node blocks */ 1058 NR_PERSISTENT_LOG, /* number of persistent log */ 1059 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1060 /* pinned file that needs consecutive block address */ 1061 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1062 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1063 }; 1064 1065 struct flush_cmd { 1066 struct completion wait; 1067 struct llist_node llnode; 1068 nid_t ino; 1069 int ret; 1070 }; 1071 1072 struct flush_cmd_control { 1073 struct task_struct *f2fs_issue_flush; /* flush thread */ 1074 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1075 atomic_t issued_flush; /* # of issued flushes */ 1076 atomic_t queued_flush; /* # of queued flushes */ 1077 struct llist_head issue_list; /* list for command issue */ 1078 struct llist_node *dispatch_list; /* list for command dispatch */ 1079 }; 1080 1081 struct f2fs_sm_info { 1082 struct sit_info *sit_info; /* whole segment information */ 1083 struct free_segmap_info *free_info; /* free segment information */ 1084 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1085 struct curseg_info *curseg_array; /* active segment information */ 1086 1087 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1088 1089 block_t seg0_blkaddr; /* block address of 0'th segment */ 1090 block_t main_blkaddr; /* start block address of main area */ 1091 block_t ssa_blkaddr; /* start block address of SSA area */ 1092 1093 unsigned int segment_count; /* total # of segments */ 1094 unsigned int main_segments; /* # of segments in main area */ 1095 unsigned int reserved_segments; /* # of reserved segments */ 1096 unsigned int ovp_segments; /* # of overprovision segments */ 1097 1098 /* a threshold to reclaim prefree segments */ 1099 unsigned int rec_prefree_segments; 1100 1101 struct list_head sit_entry_set; /* sit entry set list */ 1102 1103 unsigned int ipu_policy; /* in-place-update policy */ 1104 unsigned int min_ipu_util; /* in-place-update threshold */ 1105 unsigned int min_fsync_blocks; /* threshold for fsync */ 1106 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1107 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1108 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1109 1110 /* for flush command control */ 1111 struct flush_cmd_control *fcc_info; 1112 1113 /* for discard command control */ 1114 struct discard_cmd_control *dcc_info; 1115 }; 1116 1117 /* 1118 * For superblock 1119 */ 1120 /* 1121 * COUNT_TYPE for monitoring 1122 * 1123 * f2fs monitors the number of several block types such as on-writeback, 1124 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1125 */ 1126 #define WB_DATA_TYPE(p, f) \ 1127 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1128 enum count_type { 1129 F2FS_DIRTY_DENTS, 1130 F2FS_DIRTY_DATA, 1131 F2FS_DIRTY_QDATA, 1132 F2FS_DIRTY_NODES, 1133 F2FS_DIRTY_META, 1134 F2FS_DIRTY_IMETA, 1135 F2FS_WB_CP_DATA, 1136 F2FS_WB_DATA, 1137 F2FS_RD_DATA, 1138 F2FS_RD_NODE, 1139 F2FS_RD_META, 1140 F2FS_DIO_WRITE, 1141 F2FS_DIO_READ, 1142 NR_COUNT_TYPE, 1143 }; 1144 1145 /* 1146 * The below are the page types of bios used in submit_bio(). 1147 * The available types are: 1148 * DATA User data pages. It operates as async mode. 1149 * NODE Node pages. It operates as async mode. 1150 * META FS metadata pages such as SIT, NAT, CP. 1151 * NR_PAGE_TYPE The number of page types. 1152 * META_FLUSH Make sure the previous pages are written 1153 * with waiting the bio's completion 1154 * ... Only can be used with META. 1155 */ 1156 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1157 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1158 enum page_type { 1159 DATA = 0, 1160 NODE = 1, /* should not change this */ 1161 META, 1162 NR_PAGE_TYPE, 1163 META_FLUSH, 1164 IPU, /* the below types are used by tracepoints only. */ 1165 OPU, 1166 }; 1167 1168 enum temp_type { 1169 HOT = 0, /* must be zero for meta bio */ 1170 WARM, 1171 COLD, 1172 NR_TEMP_TYPE, 1173 }; 1174 1175 enum need_lock_type { 1176 LOCK_REQ = 0, 1177 LOCK_DONE, 1178 LOCK_RETRY, 1179 }; 1180 1181 enum cp_reason_type { 1182 CP_NO_NEEDED, 1183 CP_NON_REGULAR, 1184 CP_COMPRESSED, 1185 CP_HARDLINK, 1186 CP_SB_NEED_CP, 1187 CP_WRONG_PINO, 1188 CP_NO_SPC_ROLL, 1189 CP_NODE_NEED_CP, 1190 CP_FASTBOOT_MODE, 1191 CP_SPEC_LOG_NUM, 1192 CP_RECOVER_DIR, 1193 CP_XATTR_DIR, 1194 }; 1195 1196 enum iostat_type { 1197 /* WRITE IO */ 1198 APP_DIRECT_IO, /* app direct write IOs */ 1199 APP_BUFFERED_IO, /* app buffered write IOs */ 1200 APP_WRITE_IO, /* app write IOs */ 1201 APP_MAPPED_IO, /* app mapped IOs */ 1202 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1203 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1204 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1205 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1206 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1207 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1208 FS_GC_DATA_IO, /* data IOs from forground gc */ 1209 FS_GC_NODE_IO, /* node IOs from forground gc */ 1210 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1211 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1212 FS_CP_META_IO, /* meta IOs from checkpoint */ 1213 1214 /* READ IO */ 1215 APP_DIRECT_READ_IO, /* app direct read IOs */ 1216 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1217 APP_READ_IO, /* app read IOs */ 1218 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1219 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1220 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1221 FS_DATA_READ_IO, /* data read IOs */ 1222 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1223 FS_CDATA_READ_IO, /* compressed data read IOs */ 1224 FS_NODE_READ_IO, /* node read IOs */ 1225 FS_META_READ_IO, /* meta read IOs */ 1226 1227 /* other */ 1228 FS_DISCARD_IO, /* discard */ 1229 FS_FLUSH_IO, /* flush */ 1230 FS_ZONE_RESET_IO, /* zone reset */ 1231 NR_IO_TYPE, 1232 }; 1233 1234 struct f2fs_io_info { 1235 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1236 nid_t ino; /* inode number */ 1237 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1238 enum temp_type temp; /* contains HOT/WARM/COLD */ 1239 enum req_op op; /* contains REQ_OP_ */ 1240 blk_opf_t op_flags; /* req_flag_bits */ 1241 block_t new_blkaddr; /* new block address to be written */ 1242 block_t old_blkaddr; /* old block address before Cow */ 1243 struct page *page; /* page to be written */ 1244 struct page *encrypted_page; /* encrypted page */ 1245 struct page *compressed_page; /* compressed page */ 1246 struct list_head list; /* serialize IOs */ 1247 unsigned int compr_blocks; /* # of compressed block addresses */ 1248 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1249 unsigned int version:8; /* version of the node */ 1250 unsigned int submitted:1; /* indicate IO submission */ 1251 unsigned int in_list:1; /* indicate fio is in io_list */ 1252 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1253 unsigned int encrypted:1; /* indicate file is encrypted */ 1254 unsigned int meta_gc:1; /* require meta inode GC */ 1255 enum iostat_type io_type; /* io type */ 1256 struct writeback_control *io_wbc; /* writeback control */ 1257 struct bio **bio; /* bio for ipu */ 1258 sector_t *last_block; /* last block number in bio */ 1259 }; 1260 1261 struct bio_entry { 1262 struct bio *bio; 1263 struct list_head list; 1264 }; 1265 1266 #define is_read_io(rw) ((rw) == READ) 1267 struct f2fs_bio_info { 1268 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1269 struct bio *bio; /* bios to merge */ 1270 sector_t last_block_in_bio; /* last block number */ 1271 struct f2fs_io_info fio; /* store buffered io info. */ 1272 #ifdef CONFIG_BLK_DEV_ZONED 1273 struct completion zone_wait; /* condition value for the previous open zone to close */ 1274 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1275 void *bi_private; /* previous bi_private for pending bio */ 1276 #endif 1277 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1278 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1279 struct list_head io_list; /* track fios */ 1280 struct list_head bio_list; /* bio entry list head */ 1281 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1282 }; 1283 1284 #define FDEV(i) (sbi->devs[i]) 1285 #define RDEV(i) (raw_super->devs[i]) 1286 struct f2fs_dev_info { 1287 struct file *bdev_file; 1288 struct block_device *bdev; 1289 char path[MAX_PATH_LEN]; 1290 unsigned int total_segments; 1291 block_t start_blk; 1292 block_t end_blk; 1293 #ifdef CONFIG_BLK_DEV_ZONED 1294 unsigned int nr_blkz; /* Total number of zones */ 1295 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1296 #endif 1297 }; 1298 1299 enum inode_type { 1300 DIR_INODE, /* for dirty dir inode */ 1301 FILE_INODE, /* for dirty regular/symlink inode */ 1302 DIRTY_META, /* for all dirtied inode metadata */ 1303 DONATE_INODE, /* for all inode to donate pages */ 1304 NR_INODE_TYPE, 1305 }; 1306 1307 /* for inner inode cache management */ 1308 struct inode_management { 1309 struct radix_tree_root ino_root; /* ino entry array */ 1310 spinlock_t ino_lock; /* for ino entry lock */ 1311 struct list_head ino_list; /* inode list head */ 1312 unsigned long ino_num; /* number of entries */ 1313 }; 1314 1315 /* for GC_AT */ 1316 struct atgc_management { 1317 bool atgc_enabled; /* ATGC is enabled or not */ 1318 struct rb_root_cached root; /* root of victim rb-tree */ 1319 struct list_head victim_list; /* linked with all victim entries */ 1320 unsigned int victim_count; /* victim count in rb-tree */ 1321 unsigned int candidate_ratio; /* candidate ratio */ 1322 unsigned int max_candidate_count; /* max candidate count */ 1323 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1324 unsigned long long age_threshold; /* age threshold */ 1325 }; 1326 1327 struct f2fs_gc_control { 1328 unsigned int victim_segno; /* target victim segment number */ 1329 int init_gc_type; /* FG_GC or BG_GC */ 1330 bool no_bg_gc; /* check the space and stop bg_gc */ 1331 bool should_migrate_blocks; /* should migrate blocks */ 1332 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1333 bool one_time; /* require one time GC in one migration unit */ 1334 unsigned int nr_free_secs; /* # of free sections to do GC */ 1335 }; 1336 1337 /* 1338 * For s_flag in struct f2fs_sb_info 1339 * Modification on enum should be synchronized with s_flag array 1340 */ 1341 enum { 1342 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1343 SBI_IS_CLOSE, /* specify unmounting */ 1344 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1345 SBI_POR_DOING, /* recovery is doing or not */ 1346 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1347 SBI_NEED_CP, /* need to checkpoint */ 1348 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1349 SBI_IS_RECOVERED, /* recovered orphan/data */ 1350 SBI_CP_DISABLED, /* CP was disabled last mount */ 1351 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1352 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1353 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1354 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1355 SBI_IS_RESIZEFS, /* resizefs is in process */ 1356 SBI_IS_FREEZING, /* freezefs is in process */ 1357 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1358 MAX_SBI_FLAG, 1359 }; 1360 1361 enum { 1362 CP_TIME, 1363 REQ_TIME, 1364 DISCARD_TIME, 1365 GC_TIME, 1366 DISABLE_TIME, 1367 UMOUNT_DISCARD_TIMEOUT, 1368 MAX_TIME, 1369 }; 1370 1371 /* Note that you need to keep synchronization with this gc_mode_names array */ 1372 enum { 1373 GC_NORMAL, 1374 GC_IDLE_CB, 1375 GC_IDLE_GREEDY, 1376 GC_IDLE_AT, 1377 GC_URGENT_HIGH, 1378 GC_URGENT_LOW, 1379 GC_URGENT_MID, 1380 MAX_GC_MODE, 1381 }; 1382 1383 enum { 1384 BGGC_MODE_ON, /* background gc is on */ 1385 BGGC_MODE_OFF, /* background gc is off */ 1386 BGGC_MODE_SYNC, /* 1387 * background gc is on, migrating blocks 1388 * like foreground gc 1389 */ 1390 }; 1391 1392 enum { 1393 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1394 FS_MODE_LFS, /* use lfs allocation only */ 1395 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1396 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1397 }; 1398 1399 enum { 1400 ALLOC_MODE_DEFAULT, /* stay default */ 1401 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1402 }; 1403 1404 enum fsync_mode { 1405 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1406 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1407 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1408 }; 1409 1410 enum { 1411 COMPR_MODE_FS, /* 1412 * automatically compress compression 1413 * enabled files 1414 */ 1415 COMPR_MODE_USER, /* 1416 * automatical compression is disabled. 1417 * user can control the file compression 1418 * using ioctls 1419 */ 1420 }; 1421 1422 enum { 1423 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1424 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1425 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1426 }; 1427 1428 enum { 1429 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1430 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1431 }; 1432 1433 enum errors_option { 1434 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1435 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1436 MOUNT_ERRORS_PANIC, /* panic on errors */ 1437 }; 1438 1439 enum { 1440 BACKGROUND, 1441 FOREGROUND, 1442 MAX_CALL_TYPE, 1443 TOTAL_CALL = FOREGROUND, 1444 }; 1445 1446 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1447 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1448 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1449 1450 /* 1451 * Layout of f2fs page.private: 1452 * 1453 * Layout A: lowest bit should be 1 1454 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1455 * bit 0 PAGE_PRIVATE_NOT_POINTER 1456 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1457 * bit 2 PAGE_PRIVATE_INLINE_INODE 1458 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1459 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1460 * bit 5- f2fs private data 1461 * 1462 * Layout B: lowest bit should be 0 1463 * page.private is a wrapped pointer. 1464 */ 1465 enum { 1466 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1467 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1468 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1469 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1470 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1471 PAGE_PRIVATE_MAX 1472 }; 1473 1474 /* For compression */ 1475 enum compress_algorithm_type { 1476 COMPRESS_LZO, 1477 COMPRESS_LZ4, 1478 COMPRESS_ZSTD, 1479 COMPRESS_LZORLE, 1480 COMPRESS_MAX, 1481 }; 1482 1483 enum compress_flag { 1484 COMPRESS_CHKSUM, 1485 COMPRESS_MAX_FLAG, 1486 }; 1487 1488 #define COMPRESS_WATERMARK 20 1489 #define COMPRESS_PERCENT 20 1490 1491 #define COMPRESS_DATA_RESERVED_SIZE 4 1492 struct compress_data { 1493 __le32 clen; /* compressed data size */ 1494 __le32 chksum; /* compressed data chksum */ 1495 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1496 u8 cdata[]; /* compressed data */ 1497 }; 1498 1499 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1500 1501 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1502 1503 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1504 1505 #define COMPRESS_LEVEL_OFFSET 8 1506 1507 /* compress context */ 1508 struct compress_ctx { 1509 struct inode *inode; /* inode the context belong to */ 1510 pgoff_t cluster_idx; /* cluster index number */ 1511 unsigned int cluster_size; /* page count in cluster */ 1512 unsigned int log_cluster_size; /* log of cluster size */ 1513 struct page **rpages; /* pages store raw data in cluster */ 1514 unsigned int nr_rpages; /* total page number in rpages */ 1515 struct page **cpages; /* pages store compressed data in cluster */ 1516 unsigned int nr_cpages; /* total page number in cpages */ 1517 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1518 void *rbuf; /* virtual mapped address on rpages */ 1519 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1520 size_t rlen; /* valid data length in rbuf */ 1521 size_t clen; /* valid data length in cbuf */ 1522 void *private; /* payload buffer for specified compression algorithm */ 1523 void *private2; /* extra payload buffer */ 1524 }; 1525 1526 /* compress context for write IO path */ 1527 struct compress_io_ctx { 1528 u32 magic; /* magic number to indicate page is compressed */ 1529 struct inode *inode; /* inode the context belong to */ 1530 struct page **rpages; /* pages store raw data in cluster */ 1531 unsigned int nr_rpages; /* total page number in rpages */ 1532 atomic_t pending_pages; /* in-flight compressed page count */ 1533 }; 1534 1535 /* Context for decompressing one cluster on the read IO path */ 1536 struct decompress_io_ctx { 1537 u32 magic; /* magic number to indicate page is compressed */ 1538 struct inode *inode; /* inode the context belong to */ 1539 pgoff_t cluster_idx; /* cluster index number */ 1540 unsigned int cluster_size; /* page count in cluster */ 1541 unsigned int log_cluster_size; /* log of cluster size */ 1542 struct page **rpages; /* pages store raw data in cluster */ 1543 unsigned int nr_rpages; /* total page number in rpages */ 1544 struct page **cpages; /* pages store compressed data in cluster */ 1545 unsigned int nr_cpages; /* total page number in cpages */ 1546 struct page **tpages; /* temp pages to pad holes in cluster */ 1547 void *rbuf; /* virtual mapped address on rpages */ 1548 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1549 size_t rlen; /* valid data length in rbuf */ 1550 size_t clen; /* valid data length in cbuf */ 1551 1552 /* 1553 * The number of compressed pages remaining to be read in this cluster. 1554 * This is initially nr_cpages. It is decremented by 1 each time a page 1555 * has been read (or failed to be read). When it reaches 0, the cluster 1556 * is decompressed (or an error is reported). 1557 * 1558 * If an error occurs before all the pages have been submitted for I/O, 1559 * then this will never reach 0. In this case the I/O submitter is 1560 * responsible for calling f2fs_decompress_end_io() instead. 1561 */ 1562 atomic_t remaining_pages; 1563 1564 /* 1565 * Number of references to this decompress_io_ctx. 1566 * 1567 * One reference is held for I/O completion. This reference is dropped 1568 * after the pagecache pages are updated and unlocked -- either after 1569 * decompression (and verity if enabled), or after an error. 1570 * 1571 * In addition, each compressed page holds a reference while it is in a 1572 * bio. These references are necessary prevent compressed pages from 1573 * being freed while they are still in a bio. 1574 */ 1575 refcount_t refcnt; 1576 1577 bool failed; /* IO error occurred before decompression? */ 1578 bool need_verity; /* need fs-verity verification after decompression? */ 1579 void *private; /* payload buffer for specified decompression algorithm */ 1580 void *private2; /* extra payload buffer */ 1581 struct work_struct verity_work; /* work to verify the decompressed pages */ 1582 struct work_struct free_work; /* work for late free this structure itself */ 1583 }; 1584 1585 #define NULL_CLUSTER ((unsigned int)(~0)) 1586 #define MIN_COMPRESS_LOG_SIZE 2 1587 #define MAX_COMPRESS_LOG_SIZE 8 1588 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1589 1590 struct f2fs_sb_info { 1591 struct super_block *sb; /* pointer to VFS super block */ 1592 struct proc_dir_entry *s_proc; /* proc entry */ 1593 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1594 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1595 int valid_super_block; /* valid super block no */ 1596 unsigned long s_flag; /* flags for sbi */ 1597 struct mutex writepages; /* mutex for writepages() */ 1598 1599 #ifdef CONFIG_BLK_DEV_ZONED 1600 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1601 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1602 /* For adjust the priority writing position of data in zone UFS */ 1603 unsigned int blkzone_alloc_policy; 1604 #endif 1605 1606 /* for node-related operations */ 1607 struct f2fs_nm_info *nm_info; /* node manager */ 1608 struct inode *node_inode; /* cache node blocks */ 1609 1610 /* for segment-related operations */ 1611 struct f2fs_sm_info *sm_info; /* segment manager */ 1612 1613 /* for bio operations */ 1614 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1615 /* keep migration IO order for LFS mode */ 1616 struct f2fs_rwsem io_order_lock; 1617 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1618 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1619 1620 /* for checkpoint */ 1621 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1622 int cur_cp_pack; /* remain current cp pack */ 1623 spinlock_t cp_lock; /* for flag in ckpt */ 1624 struct inode *meta_inode; /* cache meta blocks */ 1625 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1626 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1627 struct f2fs_rwsem node_write; /* locking node writes */ 1628 struct f2fs_rwsem node_change; /* locking node change */ 1629 wait_queue_head_t cp_wait; 1630 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1631 long interval_time[MAX_TIME]; /* to store thresholds */ 1632 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1633 1634 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1635 1636 spinlock_t fsync_node_lock; /* for node entry lock */ 1637 struct list_head fsync_node_list; /* node list head */ 1638 unsigned int fsync_seg_id; /* sequence id */ 1639 unsigned int fsync_node_num; /* number of node entries */ 1640 1641 /* for orphan inode, use 0'th array */ 1642 unsigned int max_orphans; /* max orphan inodes */ 1643 1644 /* for inode management */ 1645 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1646 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1647 struct mutex flush_lock; /* for flush exclusion */ 1648 1649 /* for extent tree cache */ 1650 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1651 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1652 unsigned int max_read_extent_count; /* max read extent count per inode */ 1653 1654 /* The threshold used for hot and warm data seperation*/ 1655 unsigned int hot_data_age_threshold; 1656 unsigned int warm_data_age_threshold; 1657 unsigned int last_age_weight; 1658 1659 /* control donate caches */ 1660 unsigned int donate_files; 1661 1662 /* basic filesystem units */ 1663 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1664 unsigned int log_blocksize; /* log2 block size */ 1665 unsigned int blocksize; /* block size */ 1666 unsigned int root_ino_num; /* root inode number*/ 1667 unsigned int node_ino_num; /* node inode number*/ 1668 unsigned int meta_ino_num; /* meta inode number*/ 1669 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1670 unsigned int blocks_per_seg; /* blocks per segment */ 1671 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1672 unsigned int segs_per_sec; /* segments per section */ 1673 unsigned int secs_per_zone; /* sections per zone */ 1674 unsigned int total_sections; /* total section count */ 1675 unsigned int total_node_count; /* total node block count */ 1676 unsigned int total_valid_node_count; /* valid node block count */ 1677 int dir_level; /* directory level */ 1678 bool readdir_ra; /* readahead inode in readdir */ 1679 u64 max_io_bytes; /* max io bytes to merge IOs */ 1680 1681 block_t user_block_count; /* # of user blocks */ 1682 block_t total_valid_block_count; /* # of valid blocks */ 1683 block_t discard_blks; /* discard command candidats */ 1684 block_t last_valid_block_count; /* for recovery */ 1685 block_t reserved_blocks; /* configurable reserved blocks */ 1686 block_t current_reserved_blocks; /* current reserved blocks */ 1687 1688 /* Additional tracking for no checkpoint mode */ 1689 block_t unusable_block_count; /* # of blocks saved by last cp */ 1690 1691 unsigned int nquota_files; /* # of quota sysfile */ 1692 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1693 struct task_struct *umount_lock_holder; /* s_umount lock holder */ 1694 1695 /* # of pages, see count_type */ 1696 atomic_t nr_pages[NR_COUNT_TYPE]; 1697 /* # of allocated blocks */ 1698 struct percpu_counter alloc_valid_block_count; 1699 /* # of node block writes as roll forward recovery */ 1700 struct percpu_counter rf_node_block_count; 1701 1702 /* writeback control */ 1703 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1704 1705 /* valid inode count */ 1706 struct percpu_counter total_valid_inode_count; 1707 1708 struct f2fs_mount_info mount_opt; /* mount options */ 1709 1710 /* for cleaning operations */ 1711 struct f2fs_rwsem gc_lock; /* 1712 * semaphore for GC, avoid 1713 * race between GC and GC or CP 1714 */ 1715 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1716 struct atgc_management am; /* atgc management */ 1717 unsigned int cur_victim_sec; /* current victim section num */ 1718 unsigned int gc_mode; /* current GC state */ 1719 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1720 spinlock_t gc_remaining_trials_lock; 1721 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1722 unsigned int gc_remaining_trials; 1723 1724 /* for skip statistic */ 1725 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1726 1727 /* threshold for gc trials on pinned files */ 1728 unsigned short gc_pin_file_threshold; 1729 struct f2fs_rwsem pin_sem; 1730 1731 /* maximum # of trials to find a victim segment for SSR and GC */ 1732 unsigned int max_victim_search; 1733 /* migration granularity of garbage collection, unit: segment */ 1734 unsigned int migration_granularity; 1735 /* migration window granularity of garbage collection, unit: segment */ 1736 unsigned int migration_window_granularity; 1737 1738 /* 1739 * for stat information. 1740 * one is for the LFS mode, and the other is for the SSR mode. 1741 */ 1742 #ifdef CONFIG_F2FS_STAT_FS 1743 struct f2fs_stat_info *stat_info; /* FS status information */ 1744 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1745 unsigned int segment_count[2]; /* # of allocated segments */ 1746 unsigned int block_count[2]; /* # of allocated blocks */ 1747 atomic_t inplace_count; /* # of inplace update */ 1748 /* # of lookup extent cache */ 1749 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1750 /* # of hit rbtree extent node */ 1751 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1752 /* # of hit cached extent node */ 1753 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1754 /* # of hit largest extent node in read extent cache */ 1755 atomic64_t read_hit_largest; 1756 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1757 atomic_t inline_inode; /* # of inline_data inodes */ 1758 atomic_t inline_dir; /* # of inline_dentry inodes */ 1759 atomic_t compr_inode; /* # of compressed inodes */ 1760 atomic64_t compr_blocks; /* # of compressed blocks */ 1761 atomic_t swapfile_inode; /* # of swapfile inodes */ 1762 atomic_t atomic_files; /* # of opened atomic file */ 1763 atomic_t max_aw_cnt; /* max # of atomic writes */ 1764 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1765 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1766 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1767 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1768 #endif 1769 spinlock_t stat_lock; /* lock for stat operations */ 1770 1771 /* to attach REQ_META|REQ_FUA flags */ 1772 unsigned int data_io_flag; 1773 unsigned int node_io_flag; 1774 1775 /* For sysfs support */ 1776 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1777 struct completion s_kobj_unregister; 1778 1779 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1780 struct completion s_stat_kobj_unregister; 1781 1782 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1783 struct completion s_feature_list_kobj_unregister; 1784 1785 /* For shrinker support */ 1786 struct list_head s_list; 1787 struct mutex umount_mutex; 1788 unsigned int shrinker_run_no; 1789 1790 /* For multi devices */ 1791 int s_ndevs; /* number of devices */ 1792 struct f2fs_dev_info *devs; /* for device list */ 1793 unsigned int dirty_device; /* for checkpoint data flush */ 1794 spinlock_t dev_lock; /* protect dirty_device */ 1795 bool aligned_blksize; /* all devices has the same logical blksize */ 1796 unsigned int first_seq_zone_segno; /* first segno in sequential zone */ 1797 1798 /* For write statistics */ 1799 u64 sectors_written_start; 1800 u64 kbytes_written; 1801 1802 /* Precomputed FS UUID checksum for seeding other checksums */ 1803 __u32 s_chksum_seed; 1804 1805 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1806 1807 /* 1808 * If we are in irq context, let's update error information into 1809 * on-disk superblock in the work. 1810 */ 1811 struct work_struct s_error_work; 1812 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1813 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1814 spinlock_t error_lock; /* protect errors/stop_reason array */ 1815 bool error_dirty; /* errors of sb is dirty */ 1816 1817 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1818 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1819 1820 /* For reclaimed segs statistics per each GC mode */ 1821 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1822 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1823 1824 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1825 1826 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1827 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1828 1829 /* For atomic write statistics */ 1830 atomic64_t current_atomic_write; 1831 s64 peak_atomic_write; 1832 u64 committed_atomic_block; 1833 u64 revoked_atomic_block; 1834 1835 /* carve out reserved_blocks from total blocks */ 1836 bool carve_out; 1837 1838 #ifdef CONFIG_F2FS_FS_COMPRESSION 1839 struct kmem_cache *page_array_slab; /* page array entry */ 1840 unsigned int page_array_slab_size; /* default page array slab size */ 1841 1842 /* For runtime compression statistics */ 1843 u64 compr_written_block; 1844 u64 compr_saved_block; 1845 u32 compr_new_inode; 1846 1847 /* For compressed block cache */ 1848 struct inode *compress_inode; /* cache compressed blocks */ 1849 unsigned int compress_percent; /* cache page percentage */ 1850 unsigned int compress_watermark; /* cache page watermark */ 1851 atomic_t compress_page_hit; /* cache hit count */ 1852 #endif 1853 1854 #ifdef CONFIG_F2FS_IOSTAT 1855 /* For app/fs IO statistics */ 1856 spinlock_t iostat_lock; 1857 unsigned long long iostat_count[NR_IO_TYPE]; 1858 unsigned long long iostat_bytes[NR_IO_TYPE]; 1859 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1860 bool iostat_enable; 1861 unsigned long iostat_next_period; 1862 unsigned int iostat_period_ms; 1863 1864 /* For io latency related statistics info in one iostat period */ 1865 spinlock_t iostat_lat_lock; 1866 struct iostat_lat_info *iostat_io_lat; 1867 #endif 1868 }; 1869 1870 /* Definitions to access f2fs_sb_info */ 1871 #define SEGS_TO_BLKS(sbi, segs) \ 1872 ((segs) << (sbi)->log_blocks_per_seg) 1873 #define BLKS_TO_SEGS(sbi, blks) \ 1874 ((blks) >> (sbi)->log_blocks_per_seg) 1875 1876 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 1877 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 1878 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 1879 1880 __printf(3, 4) 1881 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1882 1883 #define f2fs_err(sbi, fmt, ...) \ 1884 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1885 #define f2fs_warn(sbi, fmt, ...) \ 1886 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1887 #define f2fs_notice(sbi, fmt, ...) \ 1888 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1889 #define f2fs_info(sbi, fmt, ...) \ 1890 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1891 #define f2fs_debug(sbi, fmt, ...) \ 1892 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1893 1894 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1895 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1896 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1897 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1898 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1899 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1900 1901 #ifdef CONFIG_F2FS_FAULT_INJECTION 1902 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1903 __builtin_return_address(0)) 1904 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1905 const char *func, const char *parent_func) 1906 { 1907 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1908 1909 if (!ffi->inject_rate) 1910 return false; 1911 1912 if (!IS_FAULT_SET(ffi, type)) 1913 return false; 1914 1915 atomic_inc(&ffi->inject_ops); 1916 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1917 atomic_set(&ffi->inject_ops, 0); 1918 ffi->inject_count[type]++; 1919 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1920 f2fs_fault_name[type], func, parent_func); 1921 return true; 1922 } 1923 return false; 1924 } 1925 #else 1926 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1927 { 1928 return false; 1929 } 1930 #endif 1931 1932 /* 1933 * Test if the mounted volume is a multi-device volume. 1934 * - For a single regular disk volume, sbi->s_ndevs is 0. 1935 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1936 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1937 */ 1938 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1939 { 1940 return sbi->s_ndevs > 1; 1941 } 1942 1943 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1944 { 1945 unsigned long now = jiffies; 1946 1947 sbi->last_time[type] = now; 1948 1949 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1950 if (type == REQ_TIME) { 1951 sbi->last_time[DISCARD_TIME] = now; 1952 sbi->last_time[GC_TIME] = now; 1953 } 1954 } 1955 1956 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1957 { 1958 unsigned long interval = sbi->interval_time[type] * HZ; 1959 1960 return time_after(jiffies, sbi->last_time[type] + interval); 1961 } 1962 1963 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1964 int type) 1965 { 1966 unsigned long interval = sbi->interval_time[type] * HZ; 1967 unsigned int wait_ms = 0; 1968 long delta; 1969 1970 delta = (sbi->last_time[type] + interval) - jiffies; 1971 if (delta > 0) 1972 wait_ms = jiffies_to_msecs(delta); 1973 1974 return wait_ms; 1975 } 1976 1977 /* 1978 * Inline functions 1979 */ 1980 static inline u32 __f2fs_crc32(u32 crc, const void *address, 1981 unsigned int length) 1982 { 1983 return crc32(crc, address, length); 1984 } 1985 1986 static inline u32 f2fs_crc32(const void *address, unsigned int length) 1987 { 1988 return __f2fs_crc32(F2FS_SUPER_MAGIC, address, length); 1989 } 1990 1991 static inline u32 f2fs_chksum(u32 crc, const void *address, unsigned int length) 1992 { 1993 return __f2fs_crc32(crc, address, length); 1994 } 1995 1996 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1997 { 1998 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1999 } 2000 2001 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2002 { 2003 return sb->s_fs_info; 2004 } 2005 2006 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2007 { 2008 return F2FS_SB(inode->i_sb); 2009 } 2010 2011 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2012 { 2013 return F2FS_I_SB(mapping->host); 2014 } 2015 2016 static inline struct f2fs_sb_info *F2FS_F_SB(struct folio *folio) 2017 { 2018 return F2FS_M_SB(folio->mapping); 2019 } 2020 2021 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2022 { 2023 return F2FS_F_SB(page_folio(page)); 2024 } 2025 2026 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2027 { 2028 return (struct f2fs_super_block *)(sbi->raw_super); 2029 } 2030 2031 static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2032 pgoff_t index) 2033 { 2034 pgoff_t idx_in_folio = index % (1 << folio_order(folio)); 2035 2036 return (struct f2fs_super_block *) 2037 (page_address(folio_page(folio, idx_in_folio)) + 2038 F2FS_SUPER_OFFSET); 2039 } 2040 2041 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2042 { 2043 return (struct f2fs_checkpoint *)(sbi->ckpt); 2044 } 2045 2046 static inline struct f2fs_node *F2FS_NODE(const struct page *page) 2047 { 2048 return (struct f2fs_node *)page_address(page); 2049 } 2050 2051 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2052 { 2053 return &((struct f2fs_node *)page_address(page))->i; 2054 } 2055 2056 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2057 { 2058 return (struct f2fs_nm_info *)(sbi->nm_info); 2059 } 2060 2061 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2062 { 2063 return (struct f2fs_sm_info *)(sbi->sm_info); 2064 } 2065 2066 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2067 { 2068 return (struct sit_info *)(SM_I(sbi)->sit_info); 2069 } 2070 2071 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2072 { 2073 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2074 } 2075 2076 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2077 { 2078 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2079 } 2080 2081 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2082 { 2083 return sbi->meta_inode->i_mapping; 2084 } 2085 2086 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2087 { 2088 return sbi->node_inode->i_mapping; 2089 } 2090 2091 static inline bool is_meta_folio(struct folio *folio) 2092 { 2093 return folio->mapping == META_MAPPING(F2FS_F_SB(folio)); 2094 } 2095 2096 static inline bool is_node_folio(struct folio *folio) 2097 { 2098 return folio->mapping == NODE_MAPPING(F2FS_F_SB(folio)); 2099 } 2100 2101 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2102 { 2103 return test_bit(type, &sbi->s_flag); 2104 } 2105 2106 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2107 { 2108 set_bit(type, &sbi->s_flag); 2109 } 2110 2111 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2112 { 2113 clear_bit(type, &sbi->s_flag); 2114 } 2115 2116 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2117 { 2118 return le64_to_cpu(cp->checkpoint_ver); 2119 } 2120 2121 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2122 { 2123 if (type < F2FS_MAX_QUOTAS) 2124 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2125 return 0; 2126 } 2127 2128 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2129 { 2130 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2131 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2132 } 2133 2134 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2135 { 2136 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2137 2138 return ckpt_flags & f; 2139 } 2140 2141 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2142 { 2143 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2144 } 2145 2146 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2147 { 2148 unsigned int ckpt_flags; 2149 2150 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2151 ckpt_flags |= f; 2152 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2153 } 2154 2155 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2156 { 2157 unsigned long flags; 2158 2159 spin_lock_irqsave(&sbi->cp_lock, flags); 2160 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2161 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2162 } 2163 2164 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2165 { 2166 unsigned int ckpt_flags; 2167 2168 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2169 ckpt_flags &= (~f); 2170 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2171 } 2172 2173 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2174 { 2175 unsigned long flags; 2176 2177 spin_lock_irqsave(&sbi->cp_lock, flags); 2178 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2179 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2180 } 2181 2182 #define init_f2fs_rwsem(sem) \ 2183 do { \ 2184 static struct lock_class_key __key; \ 2185 \ 2186 __init_f2fs_rwsem((sem), #sem, &__key); \ 2187 } while (0) 2188 2189 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2190 const char *sem_name, struct lock_class_key *key) 2191 { 2192 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2193 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2194 init_waitqueue_head(&sem->read_waiters); 2195 #endif 2196 } 2197 2198 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2199 { 2200 return rwsem_is_locked(&sem->internal_rwsem); 2201 } 2202 2203 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2204 { 2205 return rwsem_is_contended(&sem->internal_rwsem); 2206 } 2207 2208 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2209 { 2210 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2211 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2212 #else 2213 down_read(&sem->internal_rwsem); 2214 #endif 2215 } 2216 2217 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2218 { 2219 return down_read_trylock(&sem->internal_rwsem); 2220 } 2221 2222 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2223 { 2224 up_read(&sem->internal_rwsem); 2225 } 2226 2227 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2228 { 2229 down_write(&sem->internal_rwsem); 2230 } 2231 2232 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2233 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2234 { 2235 down_read_nested(&sem->internal_rwsem, subclass); 2236 } 2237 2238 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2239 { 2240 down_write_nested(&sem->internal_rwsem, subclass); 2241 } 2242 #else 2243 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2244 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2245 #endif 2246 2247 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2248 { 2249 return down_write_trylock(&sem->internal_rwsem); 2250 } 2251 2252 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2253 { 2254 up_write(&sem->internal_rwsem); 2255 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2256 wake_up_all(&sem->read_waiters); 2257 #endif 2258 } 2259 2260 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2261 { 2262 unsigned long flags; 2263 unsigned char *nat_bits; 2264 2265 /* 2266 * In order to re-enable nat_bits we need to call fsck.f2fs by 2267 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2268 * so let's rely on regular fsck or unclean shutdown. 2269 */ 2270 2271 if (lock) 2272 spin_lock_irqsave(&sbi->cp_lock, flags); 2273 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2274 nat_bits = NM_I(sbi)->nat_bits; 2275 NM_I(sbi)->nat_bits = NULL; 2276 if (lock) 2277 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2278 2279 kvfree(nat_bits); 2280 } 2281 2282 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2283 struct cp_control *cpc) 2284 { 2285 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2286 2287 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2288 } 2289 2290 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2291 { 2292 f2fs_down_read(&sbi->cp_rwsem); 2293 } 2294 2295 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2296 { 2297 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2298 return 0; 2299 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2300 } 2301 2302 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2303 { 2304 f2fs_up_read(&sbi->cp_rwsem); 2305 } 2306 2307 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2308 { 2309 f2fs_down_write(&sbi->cp_rwsem); 2310 } 2311 2312 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2313 { 2314 f2fs_up_write(&sbi->cp_rwsem); 2315 } 2316 2317 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2318 { 2319 int reason = CP_SYNC; 2320 2321 if (test_opt(sbi, FASTBOOT)) 2322 reason = CP_FASTBOOT; 2323 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2324 reason = CP_UMOUNT; 2325 return reason; 2326 } 2327 2328 static inline bool __remain_node_summaries(int reason) 2329 { 2330 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2331 } 2332 2333 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2334 { 2335 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2336 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2337 } 2338 2339 /* 2340 * Check whether the inode has blocks or not 2341 */ 2342 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2343 { 2344 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2345 2346 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2347 } 2348 2349 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2350 { 2351 return ofs == XATTR_NODE_OFFSET; 2352 } 2353 2354 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2355 struct inode *inode, bool cap) 2356 { 2357 if (!inode) 2358 return true; 2359 if (!test_opt(sbi, RESERVE_ROOT)) 2360 return false; 2361 if (IS_NOQUOTA(inode)) 2362 return true; 2363 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2364 return true; 2365 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2366 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2367 return true; 2368 if (cap && capable(CAP_SYS_RESOURCE)) 2369 return true; 2370 return false; 2371 } 2372 2373 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2374 struct inode *inode, bool cap) 2375 { 2376 block_t avail_user_block_count; 2377 2378 avail_user_block_count = sbi->user_block_count - 2379 sbi->current_reserved_blocks; 2380 2381 if (!__allow_reserved_blocks(sbi, inode, cap)) 2382 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2383 2384 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2385 if (avail_user_block_count > sbi->unusable_block_count) 2386 avail_user_block_count -= sbi->unusable_block_count; 2387 else 2388 avail_user_block_count = 0; 2389 } 2390 2391 return avail_user_block_count; 2392 } 2393 2394 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2395 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2396 struct inode *inode, blkcnt_t *count, bool partial) 2397 { 2398 long long diff = 0, release = 0; 2399 block_t avail_user_block_count; 2400 int ret; 2401 2402 ret = dquot_reserve_block(inode, *count); 2403 if (ret) 2404 return ret; 2405 2406 if (time_to_inject(sbi, FAULT_BLOCK)) { 2407 release = *count; 2408 goto release_quota; 2409 } 2410 2411 /* 2412 * let's increase this in prior to actual block count change in order 2413 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2414 */ 2415 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2416 2417 spin_lock(&sbi->stat_lock); 2418 2419 avail_user_block_count = get_available_block_count(sbi, inode, true); 2420 diff = (long long)sbi->total_valid_block_count + *count - 2421 avail_user_block_count; 2422 if (unlikely(diff > 0)) { 2423 if (!partial) { 2424 spin_unlock(&sbi->stat_lock); 2425 release = *count; 2426 goto enospc; 2427 } 2428 if (diff > *count) 2429 diff = *count; 2430 *count -= diff; 2431 release = diff; 2432 if (!*count) { 2433 spin_unlock(&sbi->stat_lock); 2434 goto enospc; 2435 } 2436 } 2437 sbi->total_valid_block_count += (block_t)(*count); 2438 2439 spin_unlock(&sbi->stat_lock); 2440 2441 if (unlikely(release)) { 2442 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2443 dquot_release_reservation_block(inode, release); 2444 } 2445 f2fs_i_blocks_write(inode, *count, true, true); 2446 return 0; 2447 2448 enospc: 2449 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2450 release_quota: 2451 dquot_release_reservation_block(inode, release); 2452 return -ENOSPC; 2453 } 2454 2455 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2456 static inline bool page_private_##name(struct page *page) \ 2457 { \ 2458 return PagePrivate(page) && \ 2459 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2460 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2461 } 2462 2463 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2464 static inline void set_page_private_##name(struct page *page) \ 2465 { \ 2466 if (!PagePrivate(page)) \ 2467 attach_page_private(page, (void *)0); \ 2468 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2469 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2470 } 2471 2472 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2473 static inline void clear_page_private_##name(struct page *page) \ 2474 { \ 2475 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2476 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2477 detach_page_private(page); \ 2478 } 2479 2480 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2481 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2482 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2483 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2484 2485 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2486 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2487 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2488 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2489 2490 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2491 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2492 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2493 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2494 2495 static inline unsigned long get_page_private_data(struct page *page) 2496 { 2497 unsigned long data = page_private(page); 2498 2499 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2500 return 0; 2501 return data >> PAGE_PRIVATE_MAX; 2502 } 2503 2504 static inline void set_page_private_data(struct page *page, unsigned long data) 2505 { 2506 if (!PagePrivate(page)) 2507 attach_page_private(page, (void *)0); 2508 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2509 page_private(page) |= data << PAGE_PRIVATE_MAX; 2510 } 2511 2512 static inline void clear_page_private_data(struct page *page) 2513 { 2514 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2515 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2516 detach_page_private(page); 2517 } 2518 2519 static inline void clear_page_private_all(struct page *page) 2520 { 2521 clear_page_private_data(page); 2522 clear_page_private_reference(page); 2523 clear_page_private_gcing(page); 2524 clear_page_private_inline(page); 2525 clear_page_private_atomic(page); 2526 2527 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2528 } 2529 2530 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2531 struct inode *inode, 2532 block_t count) 2533 { 2534 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2535 2536 spin_lock(&sbi->stat_lock); 2537 if (unlikely(sbi->total_valid_block_count < count)) { 2538 f2fs_warn(sbi, "Inconsistent total_valid_block_count:%u, ino:%lu, count:%u", 2539 sbi->total_valid_block_count, inode->i_ino, count); 2540 sbi->total_valid_block_count = 0; 2541 set_sbi_flag(sbi, SBI_NEED_FSCK); 2542 } else { 2543 sbi->total_valid_block_count -= count; 2544 } 2545 if (sbi->reserved_blocks && 2546 sbi->current_reserved_blocks < sbi->reserved_blocks) 2547 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2548 sbi->current_reserved_blocks + count); 2549 spin_unlock(&sbi->stat_lock); 2550 if (unlikely(inode->i_blocks < sectors)) { 2551 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2552 inode->i_ino, 2553 (unsigned long long)inode->i_blocks, 2554 (unsigned long long)sectors); 2555 set_sbi_flag(sbi, SBI_NEED_FSCK); 2556 return; 2557 } 2558 f2fs_i_blocks_write(inode, count, false, true); 2559 } 2560 2561 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2562 { 2563 atomic_inc(&sbi->nr_pages[count_type]); 2564 2565 if (count_type == F2FS_DIRTY_DENTS || 2566 count_type == F2FS_DIRTY_NODES || 2567 count_type == F2FS_DIRTY_META || 2568 count_type == F2FS_DIRTY_QDATA || 2569 count_type == F2FS_DIRTY_IMETA) 2570 set_sbi_flag(sbi, SBI_IS_DIRTY); 2571 } 2572 2573 static inline void inode_inc_dirty_pages(struct inode *inode) 2574 { 2575 atomic_inc(&F2FS_I(inode)->dirty_pages); 2576 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2577 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2578 if (IS_NOQUOTA(inode)) 2579 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2580 } 2581 2582 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2583 { 2584 atomic_dec(&sbi->nr_pages[count_type]); 2585 } 2586 2587 static inline void inode_dec_dirty_pages(struct inode *inode) 2588 { 2589 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2590 !S_ISLNK(inode->i_mode)) 2591 return; 2592 2593 atomic_dec(&F2FS_I(inode)->dirty_pages); 2594 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2595 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2596 if (IS_NOQUOTA(inode)) 2597 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2598 } 2599 2600 static inline void inc_atomic_write_cnt(struct inode *inode) 2601 { 2602 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2603 struct f2fs_inode_info *fi = F2FS_I(inode); 2604 u64 current_write; 2605 2606 fi->atomic_write_cnt++; 2607 atomic64_inc(&sbi->current_atomic_write); 2608 current_write = atomic64_read(&sbi->current_atomic_write); 2609 if (current_write > sbi->peak_atomic_write) 2610 sbi->peak_atomic_write = current_write; 2611 } 2612 2613 static inline void release_atomic_write_cnt(struct inode *inode) 2614 { 2615 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2616 struct f2fs_inode_info *fi = F2FS_I(inode); 2617 2618 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2619 fi->atomic_write_cnt = 0; 2620 } 2621 2622 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2623 { 2624 return atomic_read(&sbi->nr_pages[count_type]); 2625 } 2626 2627 static inline int get_dirty_pages(struct inode *inode) 2628 { 2629 return atomic_read(&F2FS_I(inode)->dirty_pages); 2630 } 2631 2632 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2633 { 2634 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2635 BLKS_PER_SEC(sbi)); 2636 } 2637 2638 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2639 { 2640 return sbi->total_valid_block_count; 2641 } 2642 2643 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2644 { 2645 return sbi->discard_blks; 2646 } 2647 2648 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2649 { 2650 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2651 2652 /* return NAT or SIT bitmap */ 2653 if (flag == NAT_BITMAP) 2654 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2655 else if (flag == SIT_BITMAP) 2656 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2657 2658 return 0; 2659 } 2660 2661 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2662 { 2663 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2664 } 2665 2666 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2667 { 2668 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2669 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2670 int offset; 2671 2672 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2673 offset = (flag == SIT_BITMAP) ? 2674 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2675 /* 2676 * if large_nat_bitmap feature is enabled, leave checksum 2677 * protection for all nat/sit bitmaps. 2678 */ 2679 return tmp_ptr + offset + sizeof(__le32); 2680 } 2681 2682 if (__cp_payload(sbi) > 0) { 2683 if (flag == NAT_BITMAP) 2684 return tmp_ptr; 2685 else 2686 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2687 } else { 2688 offset = (flag == NAT_BITMAP) ? 2689 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2690 return tmp_ptr + offset; 2691 } 2692 } 2693 2694 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2695 { 2696 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2697 2698 if (sbi->cur_cp_pack == 2) 2699 start_addr += BLKS_PER_SEG(sbi); 2700 return start_addr; 2701 } 2702 2703 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2704 { 2705 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2706 2707 if (sbi->cur_cp_pack == 1) 2708 start_addr += BLKS_PER_SEG(sbi); 2709 return start_addr; 2710 } 2711 2712 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2713 { 2714 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2715 } 2716 2717 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2718 { 2719 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2720 } 2721 2722 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2723 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2724 struct inode *inode, bool is_inode) 2725 { 2726 block_t valid_block_count; 2727 unsigned int valid_node_count; 2728 unsigned int avail_user_block_count; 2729 int err; 2730 2731 if (is_inode) { 2732 if (inode) { 2733 err = dquot_alloc_inode(inode); 2734 if (err) 2735 return err; 2736 } 2737 } else { 2738 err = dquot_reserve_block(inode, 1); 2739 if (err) 2740 return err; 2741 } 2742 2743 if (time_to_inject(sbi, FAULT_BLOCK)) 2744 goto enospc; 2745 2746 spin_lock(&sbi->stat_lock); 2747 2748 valid_block_count = sbi->total_valid_block_count + 1; 2749 avail_user_block_count = get_available_block_count(sbi, inode, false); 2750 2751 if (unlikely(valid_block_count > avail_user_block_count)) { 2752 spin_unlock(&sbi->stat_lock); 2753 goto enospc; 2754 } 2755 2756 valid_node_count = sbi->total_valid_node_count + 1; 2757 if (unlikely(valid_node_count > sbi->total_node_count)) { 2758 spin_unlock(&sbi->stat_lock); 2759 goto enospc; 2760 } 2761 2762 sbi->total_valid_node_count++; 2763 sbi->total_valid_block_count++; 2764 spin_unlock(&sbi->stat_lock); 2765 2766 if (inode) { 2767 if (is_inode) 2768 f2fs_mark_inode_dirty_sync(inode, true); 2769 else 2770 f2fs_i_blocks_write(inode, 1, true, true); 2771 } 2772 2773 percpu_counter_inc(&sbi->alloc_valid_block_count); 2774 return 0; 2775 2776 enospc: 2777 if (is_inode) { 2778 if (inode) 2779 dquot_free_inode(inode); 2780 } else { 2781 dquot_release_reservation_block(inode, 1); 2782 } 2783 return -ENOSPC; 2784 } 2785 2786 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2787 struct inode *inode, bool is_inode) 2788 { 2789 spin_lock(&sbi->stat_lock); 2790 2791 if (unlikely(!sbi->total_valid_block_count || 2792 !sbi->total_valid_node_count)) { 2793 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2794 sbi->total_valid_block_count, 2795 sbi->total_valid_node_count); 2796 set_sbi_flag(sbi, SBI_NEED_FSCK); 2797 } else { 2798 sbi->total_valid_block_count--; 2799 sbi->total_valid_node_count--; 2800 } 2801 2802 if (sbi->reserved_blocks && 2803 sbi->current_reserved_blocks < sbi->reserved_blocks) 2804 sbi->current_reserved_blocks++; 2805 2806 spin_unlock(&sbi->stat_lock); 2807 2808 if (is_inode) { 2809 dquot_free_inode(inode); 2810 } else { 2811 if (unlikely(inode->i_blocks == 0)) { 2812 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2813 inode->i_ino, 2814 (unsigned long long)inode->i_blocks); 2815 set_sbi_flag(sbi, SBI_NEED_FSCK); 2816 return; 2817 } 2818 f2fs_i_blocks_write(inode, 1, false, true); 2819 } 2820 } 2821 2822 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2823 { 2824 return sbi->total_valid_node_count; 2825 } 2826 2827 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2828 { 2829 percpu_counter_inc(&sbi->total_valid_inode_count); 2830 } 2831 2832 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2833 { 2834 percpu_counter_dec(&sbi->total_valid_inode_count); 2835 } 2836 2837 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2838 { 2839 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2840 } 2841 2842 static inline struct folio *f2fs_grab_cache_folio(struct address_space *mapping, 2843 pgoff_t index, bool for_write) 2844 { 2845 struct folio *folio; 2846 unsigned int flags; 2847 2848 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2849 fgf_t fgf_flags; 2850 2851 if (!for_write) 2852 fgf_flags = FGP_LOCK | FGP_ACCESSED; 2853 else 2854 fgf_flags = FGP_LOCK; 2855 folio = __filemap_get_folio(mapping, index, fgf_flags, 0); 2856 if (!IS_ERR(folio)) 2857 return folio; 2858 2859 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2860 return ERR_PTR(-ENOMEM); 2861 } 2862 2863 if (!for_write) 2864 return filemap_grab_folio(mapping, index); 2865 2866 flags = memalloc_nofs_save(); 2867 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN, 2868 mapping_gfp_mask(mapping)); 2869 memalloc_nofs_restore(flags); 2870 2871 return folio; 2872 } 2873 2874 static inline struct folio *f2fs_filemap_get_folio( 2875 struct address_space *mapping, pgoff_t index, 2876 fgf_t fgp_flags, gfp_t gfp_mask) 2877 { 2878 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2879 return ERR_PTR(-ENOMEM); 2880 2881 return __filemap_get_folio(mapping, index, fgp_flags, gfp_mask); 2882 } 2883 2884 static inline struct page *f2fs_pagecache_get_page( 2885 struct address_space *mapping, pgoff_t index, 2886 fgf_t fgp_flags, gfp_t gfp_mask) 2887 { 2888 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2889 return NULL; 2890 2891 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2892 } 2893 2894 static inline void f2fs_folio_put(struct folio *folio, bool unlock) 2895 { 2896 if (IS_ERR_OR_NULL(folio)) 2897 return; 2898 2899 if (unlock) { 2900 f2fs_bug_on(F2FS_F_SB(folio), !folio_test_locked(folio)); 2901 folio_unlock(folio); 2902 } 2903 folio_put(folio); 2904 } 2905 2906 static inline void f2fs_put_page(struct page *page, int unlock) 2907 { 2908 if (!page) 2909 return; 2910 f2fs_folio_put(page_folio(page), unlock); 2911 } 2912 2913 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2914 { 2915 if (dn->node_folio) 2916 f2fs_folio_put(dn->node_folio, true); 2917 if (dn->inode_folio && dn->node_folio != dn->inode_folio) 2918 f2fs_folio_put(dn->inode_folio, false); 2919 dn->node_folio = NULL; 2920 dn->inode_folio = NULL; 2921 } 2922 2923 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2924 size_t size) 2925 { 2926 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2927 } 2928 2929 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2930 gfp_t flags) 2931 { 2932 void *entry; 2933 2934 entry = kmem_cache_alloc(cachep, flags); 2935 if (!entry) 2936 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2937 return entry; 2938 } 2939 2940 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2941 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2942 { 2943 if (nofail) 2944 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2945 2946 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2947 return NULL; 2948 2949 return kmem_cache_alloc(cachep, flags); 2950 } 2951 2952 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2953 { 2954 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2955 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2956 get_pages(sbi, F2FS_WB_CP_DATA) || 2957 get_pages(sbi, F2FS_DIO_READ) || 2958 get_pages(sbi, F2FS_DIO_WRITE)) 2959 return true; 2960 2961 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2962 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2963 return true; 2964 2965 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2966 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2967 return true; 2968 return false; 2969 } 2970 2971 static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 2972 { 2973 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ); 2974 } 2975 2976 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2977 { 2978 bool zoned_gc = (type == GC_TIME && 2979 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 2980 2981 if (sbi->gc_mode == GC_URGENT_HIGH) 2982 return true; 2983 2984 if (zoned_gc) { 2985 if (is_inflight_read_io(sbi)) 2986 return false; 2987 } else { 2988 if (is_inflight_io(sbi, type)) 2989 return false; 2990 } 2991 2992 if (sbi->gc_mode == GC_URGENT_MID) 2993 return true; 2994 2995 if (sbi->gc_mode == GC_URGENT_LOW && 2996 (type == DISCARD_TIME || type == GC_TIME)) 2997 return true; 2998 2999 if (zoned_gc) 3000 return true; 3001 3002 return f2fs_time_over(sbi, type); 3003 } 3004 3005 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 3006 unsigned long index, void *item) 3007 { 3008 while (radix_tree_insert(root, index, item)) 3009 cond_resched(); 3010 } 3011 3012 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 3013 3014 static inline bool IS_INODE(struct page *page) 3015 { 3016 struct f2fs_node *p = F2FS_NODE(page); 3017 3018 return RAW_IS_INODE(p); 3019 } 3020 3021 static inline int offset_in_addr(struct f2fs_inode *i) 3022 { 3023 return (i->i_inline & F2FS_EXTRA_ATTR) ? 3024 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 3025 } 3026 3027 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 3028 { 3029 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 3030 } 3031 3032 static inline int f2fs_has_extra_attr(struct inode *inode); 3033 static inline unsigned int get_dnode_base(struct inode *inode, 3034 struct page *node_page) 3035 { 3036 if (!IS_INODE(node_page)) 3037 return 0; 3038 3039 return inode ? get_extra_isize(inode) : 3040 offset_in_addr(&F2FS_NODE(node_page)->i); 3041 } 3042 3043 static inline __le32 *get_dnode_addr(struct inode *inode, 3044 struct folio *node_folio) 3045 { 3046 return blkaddr_in_node(F2FS_NODE(&node_folio->page)) + 3047 get_dnode_base(inode, &node_folio->page); 3048 } 3049 3050 static inline block_t data_blkaddr(struct inode *inode, 3051 struct folio *node_folio, unsigned int offset) 3052 { 3053 return le32_to_cpu(*(get_dnode_addr(inode, node_folio) + offset)); 3054 } 3055 3056 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 3057 { 3058 return data_blkaddr(dn->inode, dn->node_folio, dn->ofs_in_node); 3059 } 3060 3061 static inline int f2fs_test_bit(unsigned int nr, char *addr) 3062 { 3063 int mask; 3064 3065 addr += (nr >> 3); 3066 mask = BIT(7 - (nr & 0x07)); 3067 return mask & *addr; 3068 } 3069 3070 static inline void f2fs_set_bit(unsigned int nr, char *addr) 3071 { 3072 int mask; 3073 3074 addr += (nr >> 3); 3075 mask = BIT(7 - (nr & 0x07)); 3076 *addr |= mask; 3077 } 3078 3079 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 3080 { 3081 int mask; 3082 3083 addr += (nr >> 3); 3084 mask = BIT(7 - (nr & 0x07)); 3085 *addr &= ~mask; 3086 } 3087 3088 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3089 { 3090 int mask; 3091 int ret; 3092 3093 addr += (nr >> 3); 3094 mask = BIT(7 - (nr & 0x07)); 3095 ret = mask & *addr; 3096 *addr |= mask; 3097 return ret; 3098 } 3099 3100 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3101 { 3102 int mask; 3103 int ret; 3104 3105 addr += (nr >> 3); 3106 mask = BIT(7 - (nr & 0x07)); 3107 ret = mask & *addr; 3108 *addr &= ~mask; 3109 return ret; 3110 } 3111 3112 static inline void f2fs_change_bit(unsigned int nr, char *addr) 3113 { 3114 int mask; 3115 3116 addr += (nr >> 3); 3117 mask = BIT(7 - (nr & 0x07)); 3118 *addr ^= mask; 3119 } 3120 3121 /* 3122 * On-disk inode flags (f2fs_inode::i_flags) 3123 */ 3124 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3125 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3126 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3127 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3128 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3129 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3130 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3131 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3132 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3133 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3134 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3135 #define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3136 3137 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3138 3139 /* Flags that should be inherited by new inodes from their parent. */ 3140 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3141 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3142 F2FS_CASEFOLD_FL) 3143 3144 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3145 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3146 F2FS_CASEFOLD_FL)) 3147 3148 /* Flags that are appropriate for non-directories/regular files. */ 3149 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3150 3151 #define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3152 3153 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3154 { 3155 if (S_ISDIR(mode)) 3156 return flags; 3157 else if (S_ISREG(mode)) 3158 return flags & F2FS_REG_FLMASK; 3159 else 3160 return flags & F2FS_OTHER_FLMASK; 3161 } 3162 3163 static inline void __mark_inode_dirty_flag(struct inode *inode, 3164 int flag, bool set) 3165 { 3166 switch (flag) { 3167 case FI_INLINE_XATTR: 3168 case FI_INLINE_DATA: 3169 case FI_INLINE_DENTRY: 3170 case FI_NEW_INODE: 3171 if (set) 3172 return; 3173 fallthrough; 3174 case FI_DATA_EXIST: 3175 case FI_PIN_FILE: 3176 case FI_COMPRESS_RELEASED: 3177 f2fs_mark_inode_dirty_sync(inode, true); 3178 } 3179 } 3180 3181 static inline void set_inode_flag(struct inode *inode, int flag) 3182 { 3183 set_bit(flag, F2FS_I(inode)->flags); 3184 __mark_inode_dirty_flag(inode, flag, true); 3185 } 3186 3187 static inline int is_inode_flag_set(struct inode *inode, int flag) 3188 { 3189 return test_bit(flag, F2FS_I(inode)->flags); 3190 } 3191 3192 static inline void clear_inode_flag(struct inode *inode, int flag) 3193 { 3194 clear_bit(flag, F2FS_I(inode)->flags); 3195 __mark_inode_dirty_flag(inode, flag, false); 3196 } 3197 3198 static inline bool f2fs_verity_in_progress(struct inode *inode) 3199 { 3200 return IS_ENABLED(CONFIG_FS_VERITY) && 3201 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3202 } 3203 3204 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3205 { 3206 F2FS_I(inode)->i_acl_mode = mode; 3207 set_inode_flag(inode, FI_ACL_MODE); 3208 f2fs_mark_inode_dirty_sync(inode, false); 3209 } 3210 3211 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3212 { 3213 if (inc) 3214 inc_nlink(inode); 3215 else 3216 drop_nlink(inode); 3217 f2fs_mark_inode_dirty_sync(inode, true); 3218 } 3219 3220 static inline void f2fs_i_blocks_write(struct inode *inode, 3221 block_t diff, bool add, bool claim) 3222 { 3223 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3224 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3225 3226 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3227 if (add) { 3228 if (claim) 3229 dquot_claim_block(inode, diff); 3230 else 3231 dquot_alloc_block_nofail(inode, diff); 3232 } else { 3233 dquot_free_block(inode, diff); 3234 } 3235 3236 f2fs_mark_inode_dirty_sync(inode, true); 3237 if (clean || recover) 3238 set_inode_flag(inode, FI_AUTO_RECOVER); 3239 } 3240 3241 static inline bool f2fs_is_atomic_file(struct inode *inode); 3242 3243 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3244 { 3245 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3246 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3247 3248 if (i_size_read(inode) == i_size) 3249 return; 3250 3251 i_size_write(inode, i_size); 3252 3253 if (f2fs_is_atomic_file(inode)) 3254 return; 3255 3256 f2fs_mark_inode_dirty_sync(inode, true); 3257 if (clean || recover) 3258 set_inode_flag(inode, FI_AUTO_RECOVER); 3259 } 3260 3261 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3262 { 3263 F2FS_I(inode)->i_current_depth = depth; 3264 f2fs_mark_inode_dirty_sync(inode, true); 3265 } 3266 3267 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3268 unsigned int count) 3269 { 3270 F2FS_I(inode)->i_gc_failures = count; 3271 f2fs_mark_inode_dirty_sync(inode, true); 3272 } 3273 3274 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3275 { 3276 F2FS_I(inode)->i_xattr_nid = xnid; 3277 f2fs_mark_inode_dirty_sync(inode, true); 3278 } 3279 3280 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3281 { 3282 F2FS_I(inode)->i_pino = pino; 3283 f2fs_mark_inode_dirty_sync(inode, true); 3284 } 3285 3286 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3287 { 3288 struct f2fs_inode_info *fi = F2FS_I(inode); 3289 3290 if (ri->i_inline & F2FS_INLINE_XATTR) 3291 set_bit(FI_INLINE_XATTR, fi->flags); 3292 if (ri->i_inline & F2FS_INLINE_DATA) 3293 set_bit(FI_INLINE_DATA, fi->flags); 3294 if (ri->i_inline & F2FS_INLINE_DENTRY) 3295 set_bit(FI_INLINE_DENTRY, fi->flags); 3296 if (ri->i_inline & F2FS_DATA_EXIST) 3297 set_bit(FI_DATA_EXIST, fi->flags); 3298 if (ri->i_inline & F2FS_EXTRA_ATTR) 3299 set_bit(FI_EXTRA_ATTR, fi->flags); 3300 if (ri->i_inline & F2FS_PIN_FILE) 3301 set_bit(FI_PIN_FILE, fi->flags); 3302 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3303 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3304 } 3305 3306 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3307 { 3308 ri->i_inline = 0; 3309 3310 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3311 ri->i_inline |= F2FS_INLINE_XATTR; 3312 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3313 ri->i_inline |= F2FS_INLINE_DATA; 3314 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3315 ri->i_inline |= F2FS_INLINE_DENTRY; 3316 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3317 ri->i_inline |= F2FS_DATA_EXIST; 3318 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3319 ri->i_inline |= F2FS_EXTRA_ATTR; 3320 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3321 ri->i_inline |= F2FS_PIN_FILE; 3322 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3323 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3324 } 3325 3326 static inline int f2fs_has_extra_attr(struct inode *inode) 3327 { 3328 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3329 } 3330 3331 static inline int f2fs_has_inline_xattr(struct inode *inode) 3332 { 3333 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3334 } 3335 3336 static inline int f2fs_compressed_file(struct inode *inode) 3337 { 3338 return S_ISREG(inode->i_mode) && 3339 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3340 } 3341 3342 static inline bool f2fs_need_compress_data(struct inode *inode) 3343 { 3344 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3345 3346 if (!f2fs_compressed_file(inode)) 3347 return false; 3348 3349 if (compress_mode == COMPR_MODE_FS) 3350 return true; 3351 else if (compress_mode == COMPR_MODE_USER && 3352 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3353 return true; 3354 3355 return false; 3356 } 3357 3358 static inline unsigned int addrs_per_page(struct inode *inode, 3359 bool is_inode) 3360 { 3361 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) - 3362 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK; 3363 3364 if (f2fs_compressed_file(inode)) 3365 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3366 return addrs; 3367 } 3368 3369 static inline void *inline_xattr_addr(struct inode *inode, struct folio *folio) 3370 { 3371 struct f2fs_inode *ri = F2FS_INODE(&folio->page); 3372 3373 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3374 get_inline_xattr_addrs(inode)]); 3375 } 3376 3377 static inline int inline_xattr_size(struct inode *inode) 3378 { 3379 if (f2fs_has_inline_xattr(inode)) 3380 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3381 return 0; 3382 } 3383 3384 /* 3385 * Notice: check inline_data flag without inode page lock is unsafe. 3386 * It could change at any time by f2fs_convert_inline_folio(). 3387 */ 3388 static inline int f2fs_has_inline_data(struct inode *inode) 3389 { 3390 return is_inode_flag_set(inode, FI_INLINE_DATA); 3391 } 3392 3393 static inline int f2fs_exist_data(struct inode *inode) 3394 { 3395 return is_inode_flag_set(inode, FI_DATA_EXIST); 3396 } 3397 3398 static inline int f2fs_is_mmap_file(struct inode *inode) 3399 { 3400 return is_inode_flag_set(inode, FI_MMAP_FILE); 3401 } 3402 3403 static inline bool f2fs_is_pinned_file(struct inode *inode) 3404 { 3405 return is_inode_flag_set(inode, FI_PIN_FILE); 3406 } 3407 3408 static inline bool f2fs_is_atomic_file(struct inode *inode) 3409 { 3410 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3411 } 3412 3413 static inline bool f2fs_is_cow_file(struct inode *inode) 3414 { 3415 return is_inode_flag_set(inode, FI_COW_FILE); 3416 } 3417 3418 static inline void *inline_data_addr(struct inode *inode, struct folio *folio) 3419 { 3420 __le32 *addr = get_dnode_addr(inode, folio); 3421 3422 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3423 } 3424 3425 static inline int f2fs_has_inline_dentry(struct inode *inode) 3426 { 3427 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3428 } 3429 3430 static inline int is_file(struct inode *inode, int type) 3431 { 3432 return F2FS_I(inode)->i_advise & type; 3433 } 3434 3435 static inline void set_file(struct inode *inode, int type) 3436 { 3437 if (is_file(inode, type)) 3438 return; 3439 F2FS_I(inode)->i_advise |= type; 3440 f2fs_mark_inode_dirty_sync(inode, true); 3441 } 3442 3443 static inline void clear_file(struct inode *inode, int type) 3444 { 3445 if (!is_file(inode, type)) 3446 return; 3447 F2FS_I(inode)->i_advise &= ~type; 3448 f2fs_mark_inode_dirty_sync(inode, true); 3449 } 3450 3451 static inline bool f2fs_is_time_consistent(struct inode *inode) 3452 { 3453 struct timespec64 ts = inode_get_atime(inode); 3454 3455 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3456 return false; 3457 ts = inode_get_ctime(inode); 3458 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3459 return false; 3460 ts = inode_get_mtime(inode); 3461 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3462 return false; 3463 return true; 3464 } 3465 3466 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3467 { 3468 bool ret; 3469 3470 if (dsync) { 3471 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3472 3473 spin_lock(&sbi->inode_lock[DIRTY_META]); 3474 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3475 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3476 return ret; 3477 } 3478 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3479 file_keep_isize(inode) || 3480 i_size_read(inode) & ~PAGE_MASK) 3481 return false; 3482 3483 if (!f2fs_is_time_consistent(inode)) 3484 return false; 3485 3486 spin_lock(&F2FS_I(inode)->i_size_lock); 3487 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3488 spin_unlock(&F2FS_I(inode)->i_size_lock); 3489 3490 return ret; 3491 } 3492 3493 static inline bool f2fs_readonly(struct super_block *sb) 3494 { 3495 return sb_rdonly(sb); 3496 } 3497 3498 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3499 { 3500 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3501 } 3502 3503 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3504 size_t size, gfp_t flags) 3505 { 3506 if (time_to_inject(sbi, FAULT_KMALLOC)) 3507 return NULL; 3508 3509 return kmalloc(size, flags); 3510 } 3511 3512 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3513 { 3514 if (time_to_inject(sbi, FAULT_KMALLOC)) 3515 return NULL; 3516 3517 return __getname(); 3518 } 3519 3520 static inline void f2fs_putname(char *buf) 3521 { 3522 __putname(buf); 3523 } 3524 3525 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3526 size_t size, gfp_t flags) 3527 { 3528 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3529 } 3530 3531 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3532 size_t size, gfp_t flags) 3533 { 3534 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3535 return NULL; 3536 3537 return kvmalloc(size, flags); 3538 } 3539 3540 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3541 size_t size, gfp_t flags) 3542 { 3543 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3544 } 3545 3546 static inline void *f2fs_vmalloc(struct f2fs_sb_info *sbi, size_t size) 3547 { 3548 if (time_to_inject(sbi, FAULT_VMALLOC)) 3549 return NULL; 3550 3551 return vmalloc(size); 3552 } 3553 3554 static inline int get_extra_isize(struct inode *inode) 3555 { 3556 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3557 } 3558 3559 static inline int get_inline_xattr_addrs(struct inode *inode) 3560 { 3561 return F2FS_I(inode)->i_inline_xattr_size; 3562 } 3563 3564 #define f2fs_get_inode_mode(i) \ 3565 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3566 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3567 3568 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3569 3570 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3571 (offsetof(struct f2fs_inode, i_extra_end) - \ 3572 offsetof(struct f2fs_inode, i_extra_isize)) \ 3573 3574 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3575 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3576 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3577 sizeof((f2fs_inode)->field)) \ 3578 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3579 3580 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3581 3582 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3583 3584 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3585 block_t blkaddr, int type); 3586 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3587 block_t blkaddr, int type) 3588 { 3589 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3590 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3591 blkaddr, type); 3592 } 3593 3594 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3595 { 3596 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3597 blkaddr == COMPRESS_ADDR) 3598 return false; 3599 return true; 3600 } 3601 3602 /* 3603 * file.c 3604 */ 3605 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3606 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3607 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3608 int f2fs_truncate(struct inode *inode); 3609 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3610 struct kstat *stat, u32 request_mask, unsigned int flags); 3611 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3612 struct iattr *attr); 3613 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3614 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3615 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3616 bool readonly, bool need_lock); 3617 int f2fs_precache_extents(struct inode *inode); 3618 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3619 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3620 struct dentry *dentry, struct fileattr *fa); 3621 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3622 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3623 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3624 int f2fs_pin_file_control(struct inode *inode, bool inc); 3625 3626 /* 3627 * inode.c 3628 */ 3629 void f2fs_set_inode_flags(struct inode *inode); 3630 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct folio *folio); 3631 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3632 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3633 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3634 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3635 void f2fs_update_inode(struct inode *inode, struct folio *node_folio); 3636 void f2fs_update_inode_page(struct inode *inode); 3637 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3638 void f2fs_evict_inode(struct inode *inode); 3639 void f2fs_handle_failed_inode(struct inode *inode); 3640 3641 /* 3642 * namei.c 3643 */ 3644 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3645 bool hot, bool set); 3646 struct dentry *f2fs_get_parent(struct dentry *child); 3647 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3648 struct inode **new_inode); 3649 3650 /* 3651 * dir.c 3652 */ 3653 #if IS_ENABLED(CONFIG_UNICODE) 3654 int f2fs_init_casefolded_name(const struct inode *dir, 3655 struct f2fs_filename *fname); 3656 void f2fs_free_casefolded_name(struct f2fs_filename *fname); 3657 #else 3658 static inline int f2fs_init_casefolded_name(const struct inode *dir, 3659 struct f2fs_filename *fname) 3660 { 3661 return 0; 3662 } 3663 3664 static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 3665 { 3666 } 3667 #endif /* CONFIG_UNICODE */ 3668 3669 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3670 int lookup, struct f2fs_filename *fname); 3671 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3672 struct f2fs_filename *fname); 3673 void f2fs_free_filename(struct f2fs_filename *fname); 3674 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3675 const struct f2fs_filename *fname, int *max_slots, 3676 bool use_hash); 3677 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3678 unsigned int start_pos, struct fscrypt_str *fstr); 3679 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3680 struct f2fs_dentry_ptr *d); 3681 struct folio *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3682 const struct f2fs_filename *fname, struct folio *dfolio); 3683 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3684 unsigned int current_depth); 3685 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3686 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3687 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3688 const struct f2fs_filename *fname, struct folio **res_folio); 3689 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3690 const struct qstr *child, struct folio **res_folio); 3691 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct folio **f); 3692 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3693 struct folio **folio); 3694 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3695 struct folio *folio, struct inode *inode); 3696 bool f2fs_has_enough_room(struct inode *dir, struct folio *ifolio, 3697 const struct f2fs_filename *fname); 3698 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3699 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3700 unsigned int bit_pos); 3701 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3702 struct inode *inode, nid_t ino, umode_t mode); 3703 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3704 struct inode *inode, nid_t ino, umode_t mode); 3705 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3706 struct inode *inode, nid_t ino, umode_t mode); 3707 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct folio *folio, 3708 struct inode *dir, struct inode *inode); 3709 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3710 struct f2fs_filename *fname); 3711 bool f2fs_empty_dir(struct inode *dir); 3712 3713 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3714 { 3715 if (fscrypt_is_nokey_name(dentry)) 3716 return -ENOKEY; 3717 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3718 inode, inode->i_ino, inode->i_mode); 3719 } 3720 3721 /* 3722 * super.c 3723 */ 3724 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3725 void f2fs_inode_synced(struct inode *inode); 3726 int f2fs_dquot_initialize(struct inode *inode); 3727 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3728 int f2fs_do_quota_sync(struct super_block *sb, int type); 3729 loff_t max_file_blocks(struct inode *inode); 3730 void f2fs_quota_off_umount(struct super_block *sb); 3731 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3732 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3733 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3734 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3735 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3736 int f2fs_sync_fs(struct super_block *sb, int sync); 3737 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3738 3739 /* 3740 * hash.c 3741 */ 3742 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3743 3744 /* 3745 * node.c 3746 */ 3747 struct node_info; 3748 3749 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3750 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3751 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct folio *folio); 3752 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3753 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct folio *folio); 3754 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3755 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3756 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3757 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3758 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3759 struct node_info *ni, bool checkpoint_context); 3760 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3761 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3762 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3763 int f2fs_truncate_xattr_node(struct inode *inode); 3764 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3765 unsigned int seq_id); 3766 int f2fs_remove_inode_page(struct inode *inode); 3767 struct folio *f2fs_new_inode_folio(struct inode *inode); 3768 struct folio *f2fs_new_node_folio(struct dnode_of_data *dn, unsigned int ofs); 3769 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3770 struct folio *f2fs_get_node_folio(struct f2fs_sb_info *sbi, pgoff_t nid); 3771 struct folio *f2fs_get_inode_folio(struct f2fs_sb_info *sbi, pgoff_t ino); 3772 struct folio *f2fs_get_xnode_folio(struct f2fs_sb_info *sbi, pgoff_t xnid); 3773 int f2fs_move_node_folio(struct folio *node_folio, int gc_type); 3774 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3775 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3776 struct writeback_control *wbc, bool atomic, 3777 unsigned int *seq_id); 3778 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3779 struct writeback_control *wbc, 3780 bool do_balance, enum iostat_type io_type); 3781 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3782 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3783 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3784 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3785 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3786 int f2fs_recover_inline_xattr(struct inode *inode, struct folio *folio); 3787 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3788 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3789 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3790 unsigned int segno, struct f2fs_summary_block *sum); 3791 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3792 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3793 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3794 int __init f2fs_create_node_manager_caches(void); 3795 void f2fs_destroy_node_manager_caches(void); 3796 3797 /* 3798 * segment.c 3799 */ 3800 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3801 int f2fs_commit_atomic_write(struct inode *inode); 3802 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3803 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3804 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3805 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3806 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3807 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3808 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3809 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr, 3810 unsigned int len); 3811 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3812 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3813 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3814 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3815 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3816 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3817 struct cp_control *cpc); 3818 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3819 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3820 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3821 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3822 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3823 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3824 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3825 int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 3826 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3827 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3828 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3829 unsigned int start, unsigned int end); 3830 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3831 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3832 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3833 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3834 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3835 struct cp_control *cpc); 3836 struct folio *f2fs_get_sum_folio(struct f2fs_sb_info *sbi, unsigned int segno); 3837 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3838 block_t blk_addr); 3839 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio, 3840 enum iostat_type io_type); 3841 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3842 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3843 struct f2fs_io_info *fio); 3844 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3845 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3846 block_t old_blkaddr, block_t new_blkaddr, 3847 bool recover_curseg, bool recover_newaddr, 3848 bool from_gc); 3849 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3850 block_t old_addr, block_t new_addr, 3851 unsigned char version, bool recover_curseg, 3852 bool recover_newaddr); 3853 enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 3854 enum log_type seg_type); 3855 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3856 block_t old_blkaddr, block_t *new_blkaddr, 3857 struct f2fs_summary *sum, int type, 3858 struct f2fs_io_info *fio); 3859 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3860 block_t blkaddr, unsigned int blkcnt); 3861 void f2fs_folio_wait_writeback(struct folio *folio, enum page_type type, 3862 bool ordered, bool locked); 3863 #define f2fs_wait_on_page_writeback(page, type, ordered, locked) \ 3864 f2fs_folio_wait_writeback(page_folio(page), type, ordered, locked) 3865 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3866 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3867 block_t len); 3868 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3869 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3870 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3871 unsigned int val, int alloc); 3872 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3873 int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 3874 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3875 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3876 int __init f2fs_create_segment_manager_caches(void); 3877 void f2fs_destroy_segment_manager_caches(void); 3878 int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 3879 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3880 enum page_type type, enum temp_type temp); 3881 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 3882 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3883 unsigned int segno); 3884 unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 3885 unsigned int segno); 3886 3887 static inline struct inode *fio_inode(struct f2fs_io_info *fio) 3888 { 3889 return page_folio(fio->page)->mapping->host; 3890 } 3891 3892 #define DEF_FRAGMENT_SIZE 4 3893 #define MIN_FRAGMENT_SIZE 1 3894 #define MAX_FRAGMENT_SIZE 512 3895 3896 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3897 { 3898 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3899 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3900 } 3901 3902 /* 3903 * checkpoint.c 3904 */ 3905 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3906 unsigned char reason); 3907 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3908 struct folio *f2fs_grab_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3909 struct folio *f2fs_get_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3910 struct folio *f2fs_get_meta_folio_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3911 struct folio *f2fs_get_tmp_folio(struct f2fs_sb_info *sbi, pgoff_t index); 3912 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3913 block_t blkaddr, int type); 3914 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 3915 block_t blkaddr, int type); 3916 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3917 int type, bool sync); 3918 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3919 unsigned int ra_blocks); 3920 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3921 long nr_to_write, enum iostat_type io_type); 3922 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3923 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3924 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3925 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3926 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3927 unsigned int devidx, int type); 3928 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3929 unsigned int devidx, int type); 3930 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3931 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3932 void f2fs_add_orphan_inode(struct inode *inode); 3933 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3934 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3935 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3936 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3937 void f2fs_remove_dirty_inode(struct inode *inode); 3938 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3939 bool from_cp); 3940 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3941 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3942 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3943 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3944 int __init f2fs_create_checkpoint_caches(void); 3945 void f2fs_destroy_checkpoint_caches(void); 3946 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3947 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3948 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3949 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3950 3951 /* 3952 * data.c 3953 */ 3954 int __init f2fs_init_bioset(void); 3955 void f2fs_destroy_bioset(void); 3956 bool f2fs_is_cp_guaranteed(struct page *page); 3957 int f2fs_init_bio_entry_cache(void); 3958 void f2fs_destroy_bio_entry_cache(void); 3959 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3960 enum page_type type); 3961 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3962 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3963 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3964 struct inode *inode, struct page *page, 3965 nid_t ino, enum page_type type); 3966 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3967 struct bio **bio, struct folio *folio); 3968 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3969 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3970 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3971 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3972 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3973 block_t blk_addr, sector_t *sector); 3974 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3975 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3976 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3977 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3978 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3979 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3980 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3981 struct folio *f2fs_get_read_data_folio(struct inode *inode, pgoff_t index, 3982 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3983 struct folio *f2fs_find_data_folio(struct inode *inode, pgoff_t index, 3984 pgoff_t *next_pgofs); 3985 struct folio *f2fs_get_lock_data_folio(struct inode *inode, pgoff_t index, 3986 bool for_write); 3987 struct folio *f2fs_get_new_data_folio(struct inode *inode, 3988 struct folio *ifolio, pgoff_t index, bool new_i_size); 3989 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3990 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3991 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3992 u64 start, u64 len); 3993 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3994 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3995 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3996 int f2fs_write_single_data_page(struct folio *folio, int *submitted, 3997 struct bio **bio, sector_t *last_block, 3998 struct writeback_control *wbc, 3999 enum iostat_type io_type, 4000 int compr_blocks, bool allow_balance); 4001 void f2fs_write_failed(struct inode *inode, loff_t to); 4002 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 4003 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 4004 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 4005 void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 4006 int f2fs_init_post_read_processing(void); 4007 void f2fs_destroy_post_read_processing(void); 4008 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 4009 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 4010 extern const struct iomap_ops f2fs_iomap_ops; 4011 4012 /* 4013 * gc.c 4014 */ 4015 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 4016 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 4017 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 4018 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 4019 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 4020 int f2fs_gc_range(struct f2fs_sb_info *sbi, 4021 unsigned int start_seg, unsigned int end_seg, 4022 bool dry_run, unsigned int dry_run_sections); 4023 int f2fs_resize_fs(struct file *filp, __u64 block_count); 4024 int __init f2fs_create_garbage_collection_cache(void); 4025 void f2fs_destroy_garbage_collection_cache(void); 4026 /* victim selection function for cleaning and SSR */ 4027 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 4028 int gc_type, int type, char alloc_mode, 4029 unsigned long long age, bool one_time); 4030 4031 /* 4032 * recovery.c 4033 */ 4034 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 4035 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 4036 int __init f2fs_create_recovery_cache(void); 4037 void f2fs_destroy_recovery_cache(void); 4038 4039 /* 4040 * debug.c 4041 */ 4042 #ifdef CONFIG_F2FS_STAT_FS 4043 enum { 4044 DEVSTAT_INUSE, 4045 DEVSTAT_DIRTY, 4046 DEVSTAT_FULL, 4047 DEVSTAT_FREE, 4048 DEVSTAT_PREFREE, 4049 DEVSTAT_MAX, 4050 }; 4051 4052 struct f2fs_dev_stats { 4053 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 4054 }; 4055 4056 struct f2fs_stat_info { 4057 struct list_head stat_list; 4058 struct f2fs_sb_info *sbi; 4059 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 4060 int main_area_segs, main_area_sections, main_area_zones; 4061 unsigned long long hit_cached[NR_EXTENT_CACHES]; 4062 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 4063 unsigned long long total_ext[NR_EXTENT_CACHES]; 4064 unsigned long long hit_total[NR_EXTENT_CACHES]; 4065 int ext_tree[NR_EXTENT_CACHES]; 4066 int zombie_tree[NR_EXTENT_CACHES]; 4067 int ext_node[NR_EXTENT_CACHES]; 4068 /* to count memory footprint */ 4069 unsigned long long ext_mem[NR_EXTENT_CACHES]; 4070 /* for read extent cache */ 4071 unsigned long long hit_largest; 4072 /* for block age extent cache */ 4073 unsigned long long allocated_data_blocks; 4074 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 4075 int ndirty_data, ndirty_qdata; 4076 unsigned int ndirty_dirs, ndirty_files, ndirty_all; 4077 unsigned int nquota_files, ndonate_files; 4078 int nats, dirty_nats, sits, dirty_sits; 4079 int free_nids, avail_nids, alloc_nids; 4080 int total_count, utilization; 4081 int nr_wb_cp_data, nr_wb_data; 4082 int nr_rd_data, nr_rd_node, nr_rd_meta; 4083 int nr_dio_read, nr_dio_write; 4084 unsigned int io_skip_bggc, other_skip_bggc; 4085 int nr_flushing, nr_flushed, flush_list_empty; 4086 int nr_discarding, nr_discarded; 4087 int nr_discard_cmd; 4088 unsigned int undiscard_blks; 4089 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 4090 unsigned int cur_ckpt_time, peak_ckpt_time; 4091 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 4092 int compr_inode, swapfile_inode; 4093 unsigned long long compr_blocks; 4094 int aw_cnt, max_aw_cnt; 4095 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 4096 unsigned int bimodal, avg_vblocks; 4097 int util_free, util_valid, util_invalid; 4098 int rsvd_segs, overp_segs; 4099 int dirty_count, node_pages, meta_pages, compress_pages; 4100 int compress_page_hit; 4101 int prefree_count, free_segs, free_secs; 4102 int cp_call_count[MAX_CALL_TYPE], cp_count; 4103 int gc_call_count[MAX_CALL_TYPE]; 4104 int gc_segs[2][2]; 4105 int gc_secs[2][2]; 4106 int tot_blks, data_blks, node_blks; 4107 int bg_data_blks, bg_node_blks; 4108 int curseg[NR_CURSEG_TYPE]; 4109 int cursec[NR_CURSEG_TYPE]; 4110 int curzone[NR_CURSEG_TYPE]; 4111 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4112 unsigned int full_seg[NR_CURSEG_TYPE]; 4113 unsigned int valid_blks[NR_CURSEG_TYPE]; 4114 4115 unsigned int meta_count[META_MAX]; 4116 unsigned int segment_count[2]; 4117 unsigned int block_count[2]; 4118 unsigned int inplace_count; 4119 unsigned long long base_mem, cache_mem, page_mem; 4120 struct f2fs_dev_stats *dev_stats; 4121 }; 4122 4123 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4124 { 4125 return (struct f2fs_stat_info *)sbi->stat_info; 4126 } 4127 4128 #define stat_inc_cp_call_count(sbi, foreground) \ 4129 atomic_inc(&sbi->cp_call_count[(foreground)]) 4130 #define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4131 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4132 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4133 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4134 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4135 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4136 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4137 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4138 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4139 #define stat_inc_inline_xattr(inode) \ 4140 do { \ 4141 if (f2fs_has_inline_xattr(inode)) \ 4142 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4143 } while (0) 4144 #define stat_dec_inline_xattr(inode) \ 4145 do { \ 4146 if (f2fs_has_inline_xattr(inode)) \ 4147 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4148 } while (0) 4149 #define stat_inc_inline_inode(inode) \ 4150 do { \ 4151 if (f2fs_has_inline_data(inode)) \ 4152 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4153 } while (0) 4154 #define stat_dec_inline_inode(inode) \ 4155 do { \ 4156 if (f2fs_has_inline_data(inode)) \ 4157 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4158 } while (0) 4159 #define stat_inc_inline_dir(inode) \ 4160 do { \ 4161 if (f2fs_has_inline_dentry(inode)) \ 4162 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4163 } while (0) 4164 #define stat_dec_inline_dir(inode) \ 4165 do { \ 4166 if (f2fs_has_inline_dentry(inode)) \ 4167 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4168 } while (0) 4169 #define stat_inc_compr_inode(inode) \ 4170 do { \ 4171 if (f2fs_compressed_file(inode)) \ 4172 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4173 } while (0) 4174 #define stat_dec_compr_inode(inode) \ 4175 do { \ 4176 if (f2fs_compressed_file(inode)) \ 4177 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4178 } while (0) 4179 #define stat_add_compr_blocks(inode, blocks) \ 4180 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4181 #define stat_sub_compr_blocks(inode, blocks) \ 4182 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4183 #define stat_inc_swapfile_inode(inode) \ 4184 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4185 #define stat_dec_swapfile_inode(inode) \ 4186 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4187 #define stat_inc_atomic_inode(inode) \ 4188 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4189 #define stat_dec_atomic_inode(inode) \ 4190 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4191 #define stat_inc_meta_count(sbi, blkaddr) \ 4192 do { \ 4193 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4194 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4195 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4196 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4197 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4198 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4199 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4200 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4201 } while (0) 4202 #define stat_inc_seg_type(sbi, curseg) \ 4203 ((sbi)->segment_count[(curseg)->alloc_type]++) 4204 #define stat_inc_block_count(sbi, curseg) \ 4205 ((sbi)->block_count[(curseg)->alloc_type]++) 4206 #define stat_inc_inplace_blocks(sbi) \ 4207 (atomic_inc(&(sbi)->inplace_count)) 4208 #define stat_update_max_atomic_write(inode) \ 4209 do { \ 4210 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4211 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4212 if (cur > max) \ 4213 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4214 } while (0) 4215 #define stat_inc_gc_call_count(sbi, foreground) \ 4216 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4217 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4218 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4219 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4220 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4221 4222 #define stat_inc_tot_blk_count(si, blks) \ 4223 ((si)->tot_blks += (blks)) 4224 4225 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4226 do { \ 4227 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4228 stat_inc_tot_blk_count(si, blks); \ 4229 si->data_blks += (blks); \ 4230 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4231 } while (0) 4232 4233 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4234 do { \ 4235 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4236 stat_inc_tot_blk_count(si, blks); \ 4237 si->node_blks += (blks); \ 4238 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4239 } while (0) 4240 4241 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4242 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4243 void __init f2fs_create_root_stats(void); 4244 void f2fs_destroy_root_stats(void); 4245 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4246 #else 4247 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4248 #define stat_inc_cp_count(sbi) do { } while (0) 4249 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4250 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4251 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4252 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4253 #define stat_inc_total_hit(sbi, type) do { } while (0) 4254 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4255 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4256 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4257 #define stat_inc_inline_xattr(inode) do { } while (0) 4258 #define stat_dec_inline_xattr(inode) do { } while (0) 4259 #define stat_inc_inline_inode(inode) do { } while (0) 4260 #define stat_dec_inline_inode(inode) do { } while (0) 4261 #define stat_inc_inline_dir(inode) do { } while (0) 4262 #define stat_dec_inline_dir(inode) do { } while (0) 4263 #define stat_inc_compr_inode(inode) do { } while (0) 4264 #define stat_dec_compr_inode(inode) do { } while (0) 4265 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4266 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4267 #define stat_inc_swapfile_inode(inode) do { } while (0) 4268 #define stat_dec_swapfile_inode(inode) do { } while (0) 4269 #define stat_inc_atomic_inode(inode) do { } while (0) 4270 #define stat_dec_atomic_inode(inode) do { } while (0) 4271 #define stat_update_max_atomic_write(inode) do { } while (0) 4272 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4273 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4274 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4275 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4276 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4277 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4278 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4279 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4280 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4281 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4282 4283 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4284 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4285 static inline void __init f2fs_create_root_stats(void) { } 4286 static inline void f2fs_destroy_root_stats(void) { } 4287 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4288 #endif 4289 4290 extern const struct file_operations f2fs_dir_operations; 4291 extern const struct file_operations f2fs_file_operations; 4292 extern const struct inode_operations f2fs_file_inode_operations; 4293 extern const struct address_space_operations f2fs_dblock_aops; 4294 extern const struct address_space_operations f2fs_node_aops; 4295 extern const struct address_space_operations f2fs_meta_aops; 4296 extern const struct inode_operations f2fs_dir_inode_operations; 4297 extern const struct inode_operations f2fs_symlink_inode_operations; 4298 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4299 extern const struct inode_operations f2fs_special_inode_operations; 4300 extern struct kmem_cache *f2fs_inode_entry_slab; 4301 4302 /* 4303 * inline.c 4304 */ 4305 bool f2fs_may_inline_data(struct inode *inode); 4306 bool f2fs_sanity_check_inline_data(struct inode *inode, struct page *ipage); 4307 bool f2fs_may_inline_dentry(struct inode *inode); 4308 void f2fs_do_read_inline_data(struct folio *folio, struct folio *ifolio); 4309 void f2fs_truncate_inline_inode(struct inode *inode, struct folio *ifolio, 4310 u64 from); 4311 int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4312 int f2fs_convert_inline_folio(struct dnode_of_data *dn, struct folio *folio); 4313 int f2fs_convert_inline_inode(struct inode *inode); 4314 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4315 int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4316 int f2fs_recover_inline_data(struct inode *inode, struct folio *nfolio); 4317 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4318 const struct f2fs_filename *fname, struct folio **res_folio, 4319 bool use_hash); 4320 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4321 struct folio *ifolio); 4322 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4323 struct inode *inode, nid_t ino, umode_t mode); 4324 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4325 struct folio *folio, struct inode *dir, struct inode *inode); 4326 bool f2fs_empty_inline_dir(struct inode *dir); 4327 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4328 struct fscrypt_str *fstr); 4329 int f2fs_inline_data_fiemap(struct inode *inode, 4330 struct fiemap_extent_info *fieinfo, 4331 __u64 start, __u64 len); 4332 4333 /* 4334 * shrinker.c 4335 */ 4336 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4337 struct shrink_control *sc); 4338 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4339 struct shrink_control *sc); 4340 unsigned int f2fs_donate_files(void); 4341 void f2fs_reclaim_caches(unsigned int reclaim_caches_kb); 4342 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4343 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4344 4345 /* 4346 * extent_cache.c 4347 */ 4348 bool sanity_check_extent_cache(struct inode *inode, struct page *ipage); 4349 void f2fs_init_extent_tree(struct inode *inode); 4350 void f2fs_drop_extent_tree(struct inode *inode); 4351 void f2fs_destroy_extent_node(struct inode *inode); 4352 void f2fs_destroy_extent_tree(struct inode *inode); 4353 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4354 int __init f2fs_create_extent_cache(void); 4355 void f2fs_destroy_extent_cache(void); 4356 4357 /* read extent cache ops */ 4358 void f2fs_init_read_extent_tree(struct inode *inode, struct folio *ifolio); 4359 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4360 struct extent_info *ei); 4361 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4362 block_t *blkaddr); 4363 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4364 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4365 pgoff_t fofs, block_t blkaddr, unsigned int len); 4366 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4367 int nr_shrink); 4368 4369 /* block age extent cache ops */ 4370 void f2fs_init_age_extent_tree(struct inode *inode); 4371 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4372 struct extent_info *ei); 4373 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4374 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4375 pgoff_t fofs, unsigned int len); 4376 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4377 int nr_shrink); 4378 4379 /* 4380 * sysfs.c 4381 */ 4382 #define MIN_RA_MUL 2 4383 #define MAX_RA_MUL 256 4384 4385 int __init f2fs_init_sysfs(void); 4386 void f2fs_exit_sysfs(void); 4387 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4388 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4389 4390 /* verity.c */ 4391 extern const struct fsverity_operations f2fs_verityops; 4392 4393 /* 4394 * crypto support 4395 */ 4396 static inline bool f2fs_encrypted_file(struct inode *inode) 4397 { 4398 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4399 } 4400 4401 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4402 { 4403 #ifdef CONFIG_FS_ENCRYPTION 4404 file_set_encrypt(inode); 4405 f2fs_set_inode_flags(inode); 4406 #endif 4407 } 4408 4409 /* 4410 * Returns true if the reads of the inode's data need to undergo some 4411 * postprocessing step, like decryption or authenticity verification. 4412 */ 4413 static inline bool f2fs_post_read_required(struct inode *inode) 4414 { 4415 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4416 f2fs_compressed_file(inode); 4417 } 4418 4419 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4420 { 4421 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4422 } 4423 4424 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4425 { 4426 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4427 } 4428 4429 /* 4430 * compress.c 4431 */ 4432 #ifdef CONFIG_F2FS_FS_COMPRESSION 4433 enum cluster_check_type { 4434 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4435 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4436 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4437 }; 4438 bool f2fs_is_compressed_page(struct page *page); 4439 struct folio *f2fs_compress_control_folio(struct folio *folio); 4440 int f2fs_prepare_compress_overwrite(struct inode *inode, 4441 struct page **pagep, pgoff_t index, void **fsdata); 4442 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4443 pgoff_t index, unsigned copied); 4444 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4445 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4446 bool f2fs_is_compress_backend_ready(struct inode *inode); 4447 bool f2fs_is_compress_level_valid(int alg, int lvl); 4448 int __init f2fs_init_compress_mempool(void); 4449 void f2fs_destroy_compress_mempool(void); 4450 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4451 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4452 block_t blkaddr, bool in_task); 4453 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4454 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4455 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4456 int index, int nr_pages, bool uptodate); 4457 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4458 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4459 int f2fs_write_multi_pages(struct compress_ctx *cc, 4460 int *submitted, 4461 struct writeback_control *wbc, 4462 enum iostat_type io_type); 4463 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4464 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4465 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4466 pgoff_t fofs, block_t blkaddr, 4467 unsigned int llen, unsigned int c_len); 4468 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4469 unsigned nr_pages, sector_t *last_block_in_bio, 4470 struct readahead_control *rac, bool for_write); 4471 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4472 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4473 bool in_task); 4474 void f2fs_put_folio_dic(struct folio *folio, bool in_task); 4475 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4476 unsigned int ofs_in_node); 4477 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4478 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4479 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4480 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4481 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4482 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4483 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4484 int __init f2fs_init_compress_cache(void); 4485 void f2fs_destroy_compress_cache(void); 4486 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4487 void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4488 block_t blkaddr, unsigned int len); 4489 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4490 nid_t ino, block_t blkaddr); 4491 bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, struct folio *folio, 4492 block_t blkaddr); 4493 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4494 #define inc_compr_inode_stat(inode) \ 4495 do { \ 4496 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4497 sbi->compr_new_inode++; \ 4498 } while (0) 4499 #define add_compr_block_stat(inode, blocks) \ 4500 do { \ 4501 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4502 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4503 sbi->compr_written_block += blocks; \ 4504 sbi->compr_saved_block += diff; \ 4505 } while (0) 4506 #else 4507 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4508 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4509 { 4510 if (!f2fs_compressed_file(inode)) 4511 return true; 4512 /* not support compression */ 4513 return false; 4514 } 4515 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4516 static inline struct folio *f2fs_compress_control_folio(struct folio *folio) 4517 { 4518 WARN_ON_ONCE(1); 4519 return ERR_PTR(-EINVAL); 4520 } 4521 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4522 static inline void f2fs_destroy_compress_mempool(void) { } 4523 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4524 bool in_task) { } 4525 static inline void f2fs_end_read_compressed_page(struct page *page, 4526 bool failed, block_t blkaddr, bool in_task) 4527 { 4528 WARN_ON_ONCE(1); 4529 } 4530 static inline void f2fs_put_folio_dic(struct folio *folio, bool in_task) 4531 { 4532 WARN_ON_ONCE(1); 4533 } 4534 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4535 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4536 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4537 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4538 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4539 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4540 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4541 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4542 static inline void f2fs_destroy_compress_cache(void) { } 4543 static inline void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4544 block_t blkaddr, unsigned int len) { } 4545 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4546 struct page *page, nid_t ino, block_t blkaddr) { } 4547 static inline bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, 4548 struct folio *folio, block_t blkaddr) { return false; } 4549 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4550 nid_t ino) { } 4551 #define inc_compr_inode_stat(inode) do { } while (0) 4552 static inline int f2fs_is_compressed_cluster( 4553 struct inode *inode, 4554 pgoff_t index) { return 0; } 4555 static inline bool f2fs_is_sparse_cluster( 4556 struct inode *inode, 4557 pgoff_t index) { return true; } 4558 static inline void f2fs_update_read_extent_tree_range_compressed( 4559 struct inode *inode, 4560 pgoff_t fofs, block_t blkaddr, 4561 unsigned int llen, unsigned int c_len) { } 4562 #endif 4563 4564 static inline int set_compress_context(struct inode *inode) 4565 { 4566 #ifdef CONFIG_F2FS_FS_COMPRESSION 4567 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4568 struct f2fs_inode_info *fi = F2FS_I(inode); 4569 4570 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4571 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4572 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4573 BIT(COMPRESS_CHKSUM) : 0; 4574 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4575 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4576 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4577 F2FS_OPTION(sbi).compress_level) 4578 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4579 fi->i_flags |= F2FS_COMPR_FL; 4580 set_inode_flag(inode, FI_COMPRESSED_FILE); 4581 stat_inc_compr_inode(inode); 4582 inc_compr_inode_stat(inode); 4583 f2fs_mark_inode_dirty_sync(inode, true); 4584 return 0; 4585 #else 4586 return -EOPNOTSUPP; 4587 #endif 4588 } 4589 4590 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4591 { 4592 struct f2fs_inode_info *fi = F2FS_I(inode); 4593 4594 f2fs_down_write(&fi->i_sem); 4595 4596 if (!f2fs_compressed_file(inode)) { 4597 f2fs_up_write(&fi->i_sem); 4598 return true; 4599 } 4600 if (f2fs_is_mmap_file(inode) || 4601 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4602 f2fs_up_write(&fi->i_sem); 4603 return false; 4604 } 4605 4606 fi->i_flags &= ~F2FS_COMPR_FL; 4607 stat_dec_compr_inode(inode); 4608 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4609 f2fs_mark_inode_dirty_sync(inode, true); 4610 4611 f2fs_up_write(&fi->i_sem); 4612 return true; 4613 } 4614 4615 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4616 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4617 { \ 4618 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4619 } 4620 4621 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4622 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4623 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4624 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4625 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4626 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4627 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4628 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4629 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4630 F2FS_FEATURE_FUNCS(verity, VERITY); 4631 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4632 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4633 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4634 F2FS_FEATURE_FUNCS(readonly, RO); 4635 F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 4636 4637 #ifdef CONFIG_BLK_DEV_ZONED 4638 static inline bool f2fs_zone_is_seq(struct f2fs_sb_info *sbi, int devi, 4639 unsigned int zone) 4640 { 4641 return test_bit(zone, FDEV(devi).blkz_seq); 4642 } 4643 4644 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4645 block_t blkaddr) 4646 { 4647 return f2fs_zone_is_seq(sbi, devi, blkaddr / sbi->blocks_per_blkz); 4648 } 4649 #endif 4650 4651 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4652 struct block_device *bdev) 4653 { 4654 int i; 4655 4656 if (!f2fs_is_multi_device(sbi)) 4657 return 0; 4658 4659 for (i = 0; i < sbi->s_ndevs; i++) 4660 if (FDEV(i).bdev == bdev) 4661 return i; 4662 4663 WARN_ON(1); 4664 return -1; 4665 } 4666 4667 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4668 { 4669 return f2fs_sb_has_blkzoned(sbi); 4670 } 4671 4672 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4673 { 4674 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4675 } 4676 4677 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4678 { 4679 int i; 4680 4681 if (!f2fs_is_multi_device(sbi)) 4682 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4683 4684 for (i = 0; i < sbi->s_ndevs; i++) 4685 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4686 return true; 4687 return false; 4688 } 4689 4690 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4691 { 4692 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4693 f2fs_hw_should_discard(sbi); 4694 } 4695 4696 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4697 { 4698 int i; 4699 4700 if (!f2fs_is_multi_device(sbi)) 4701 return bdev_read_only(sbi->sb->s_bdev); 4702 4703 for (i = 0; i < sbi->s_ndevs; i++) 4704 if (bdev_read_only(FDEV(i).bdev)) 4705 return true; 4706 return false; 4707 } 4708 4709 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4710 { 4711 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4712 } 4713 4714 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4715 { 4716 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4717 } 4718 4719 static inline bool f2fs_is_sequential_zone_area(struct f2fs_sb_info *sbi, 4720 block_t blkaddr) 4721 { 4722 if (f2fs_sb_has_blkzoned(sbi)) { 4723 #ifdef CONFIG_BLK_DEV_ZONED 4724 int devi = f2fs_target_device_index(sbi, blkaddr); 4725 4726 if (!bdev_is_zoned(FDEV(devi).bdev)) 4727 return false; 4728 4729 if (f2fs_is_multi_device(sbi)) { 4730 if (blkaddr < FDEV(devi).start_blk || 4731 blkaddr > FDEV(devi).end_blk) { 4732 f2fs_err(sbi, "Invalid block %x", blkaddr); 4733 return false; 4734 } 4735 blkaddr -= FDEV(devi).start_blk; 4736 } 4737 4738 return f2fs_blkz_is_seq(sbi, devi, blkaddr); 4739 #else 4740 return false; 4741 #endif 4742 } 4743 return false; 4744 } 4745 4746 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4747 { 4748 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4749 } 4750 4751 static inline bool f2fs_may_compress(struct inode *inode) 4752 { 4753 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4754 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4755 f2fs_is_mmap_file(inode)) 4756 return false; 4757 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4758 } 4759 4760 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4761 u64 blocks, bool add) 4762 { 4763 struct f2fs_inode_info *fi = F2FS_I(inode); 4764 int diff = fi->i_cluster_size - blocks; 4765 4766 /* don't update i_compr_blocks if saved blocks were released */ 4767 if (!add && !atomic_read(&fi->i_compr_blocks)) 4768 return; 4769 4770 if (add) { 4771 atomic_add(diff, &fi->i_compr_blocks); 4772 stat_add_compr_blocks(inode, diff); 4773 } else { 4774 atomic_sub(diff, &fi->i_compr_blocks); 4775 stat_sub_compr_blocks(inode, diff); 4776 } 4777 f2fs_mark_inode_dirty_sync(inode, true); 4778 } 4779 4780 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4781 int flag) 4782 { 4783 if (!f2fs_is_multi_device(sbi)) 4784 return false; 4785 if (flag != F2FS_GET_BLOCK_DIO) 4786 return false; 4787 return sbi->aligned_blksize; 4788 } 4789 4790 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4791 { 4792 return fsverity_active(inode) && 4793 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4794 } 4795 4796 #ifdef CONFIG_F2FS_FAULT_INJECTION 4797 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4798 unsigned long type, enum fault_option fo); 4799 #else 4800 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4801 unsigned long rate, unsigned long type, 4802 enum fault_option fo) 4803 { 4804 return 0; 4805 } 4806 #endif 4807 4808 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4809 { 4810 #ifdef CONFIG_QUOTA 4811 if (f2fs_sb_has_quota_ino(sbi)) 4812 return true; 4813 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4814 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4815 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4816 return true; 4817 #endif 4818 return false; 4819 } 4820 4821 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4822 { 4823 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4824 } 4825 4826 static inline void f2fs_io_schedule_timeout(long timeout) 4827 { 4828 set_current_state(TASK_UNINTERRUPTIBLE); 4829 io_schedule_timeout(timeout); 4830 } 4831 4832 static inline void f2fs_io_schedule_timeout_killable(long timeout) 4833 { 4834 while (timeout) { 4835 if (fatal_signal_pending(current)) 4836 return; 4837 set_current_state(TASK_UNINTERRUPTIBLE); 4838 io_schedule_timeout(DEFAULT_IO_TIMEOUT); 4839 if (timeout <= DEFAULT_IO_TIMEOUT) 4840 return; 4841 timeout -= DEFAULT_IO_TIMEOUT; 4842 } 4843 } 4844 4845 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 4846 struct folio *folio, enum page_type type) 4847 { 4848 pgoff_t ofs = folio->index; 4849 4850 if (unlikely(f2fs_cp_error(sbi))) 4851 return; 4852 4853 if (ofs == sbi->page_eio_ofs[type]) { 4854 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4855 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4856 } else { 4857 sbi->page_eio_ofs[type] = ofs; 4858 sbi->page_eio_cnt[type] = 0; 4859 } 4860 } 4861 4862 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4863 { 4864 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4865 } 4866 4867 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4868 block_t blkaddr, unsigned int cnt) 4869 { 4870 bool need_submit = false; 4871 int i = 0; 4872 4873 do { 4874 struct folio *folio; 4875 4876 folio = filemap_get_folio(META_MAPPING(sbi), blkaddr + i); 4877 if (!IS_ERR(folio)) { 4878 if (folio_test_writeback(folio)) 4879 need_submit = true; 4880 f2fs_folio_put(folio, false); 4881 } 4882 } while (++i < cnt && !need_submit); 4883 4884 if (need_submit) 4885 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4886 NULL, 0, DATA); 4887 4888 truncate_inode_pages_range(META_MAPPING(sbi), 4889 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4890 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4891 } 4892 4893 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4894 block_t blkaddr, unsigned int len) 4895 { 4896 f2fs_truncate_meta_inode_pages(sbi, blkaddr, len); 4897 f2fs_invalidate_compress_pages_range(sbi, blkaddr, len); 4898 } 4899 4900 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4901 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4902 4903 #endif /* _LINUX_F2FS_H */ 4904