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/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_MAX, 64 }; 65 66 #ifdef CONFIG_F2FS_FAULT_INJECTION 67 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 68 69 struct f2fs_fault_info { 70 atomic_t inject_ops; 71 unsigned int inject_rate; 72 unsigned int inject_type; 73 }; 74 75 extern const char *f2fs_fault_name[FAULT_MAX]; 76 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 77 #endif 78 79 /* 80 * For mount options 81 */ 82 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 83 #define F2FS_MOUNT_DISCARD 0x00000004 84 #define F2FS_MOUNT_NOHEAP 0x00000008 85 #define F2FS_MOUNT_XATTR_USER 0x00000010 86 #define F2FS_MOUNT_POSIX_ACL 0x00000020 87 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 88 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 89 #define F2FS_MOUNT_INLINE_DATA 0x00000100 90 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 91 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 92 #define F2FS_MOUNT_NOBARRIER 0x00000800 93 #define F2FS_MOUNT_FASTBOOT 0x00001000 94 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 95 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 96 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 97 #define F2FS_MOUNT_USRQUOTA 0x00080000 98 #define F2FS_MOUNT_GRPQUOTA 0x00100000 99 #define F2FS_MOUNT_PRJQUOTA 0x00200000 100 #define F2FS_MOUNT_QUOTA 0x00400000 101 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 102 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 103 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 104 #define F2FS_MOUNT_NORECOVERY 0x04000000 105 #define F2FS_MOUNT_ATGC 0x08000000 106 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 107 #define F2FS_MOUNT_GC_MERGE 0x20000000 108 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 109 110 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 111 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 112 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 113 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 114 115 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 116 typecheck(unsigned long long, b) && \ 117 ((long long)((a) - (b)) > 0)) 118 119 typedef u32 block_t; /* 120 * should not change u32, since it is the on-disk block 121 * address format, __le32. 122 */ 123 typedef u32 nid_t; 124 125 #define COMPRESS_EXT_NUM 16 126 127 /* 128 * An implementation of an rwsem that is explicitly unfair to readers. This 129 * prevents priority inversion when a low-priority reader acquires the read lock 130 * while sleeping on the write lock but the write lock is needed by 131 * higher-priority clients. 132 */ 133 134 struct f2fs_rwsem { 135 struct rw_semaphore internal_rwsem; 136 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 137 wait_queue_head_t read_waiters; 138 #endif 139 }; 140 141 struct f2fs_mount_info { 142 unsigned int opt; 143 int write_io_size_bits; /* Write IO size bits */ 144 block_t root_reserved_blocks; /* root reserved blocks */ 145 kuid_t s_resuid; /* reserved blocks for uid */ 146 kgid_t s_resgid; /* reserved blocks for gid */ 147 int active_logs; /* # of active logs */ 148 int inline_xattr_size; /* inline xattr size */ 149 #ifdef CONFIG_F2FS_FAULT_INJECTION 150 struct f2fs_fault_info fault_info; /* For fault injection */ 151 #endif 152 #ifdef CONFIG_QUOTA 153 /* Names of quota files with journalled quota */ 154 char *s_qf_names[MAXQUOTAS]; 155 int s_jquota_fmt; /* Format of quota to use */ 156 #endif 157 /* For which write hints are passed down to block layer */ 158 int alloc_mode; /* segment allocation policy */ 159 int fsync_mode; /* fsync policy */ 160 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 161 int bggc_mode; /* bggc mode: off, on or sync */ 162 int memory_mode; /* memory mode */ 163 int discard_unit; /* 164 * discard command's offset/size should 165 * be aligned to this unit: block, 166 * segment or section 167 */ 168 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 169 block_t unusable_cap_perc; /* percentage for cap */ 170 block_t unusable_cap; /* Amount of space allowed to be 171 * unusable when disabling checkpoint 172 */ 173 174 /* For compression */ 175 unsigned char compress_algorithm; /* algorithm type */ 176 unsigned char compress_log_size; /* cluster log size */ 177 unsigned char compress_level; /* compress level */ 178 bool compress_chksum; /* compressed data chksum */ 179 unsigned char compress_ext_cnt; /* extension count */ 180 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 181 int compress_mode; /* compression mode */ 182 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 184 }; 185 186 #define F2FS_FEATURE_ENCRYPT 0x0001 187 #define F2FS_FEATURE_BLKZONED 0x0002 188 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 189 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 190 #define F2FS_FEATURE_PRJQUOTA 0x0010 191 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 192 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 193 #define F2FS_FEATURE_QUOTA_INO 0x0080 194 #define F2FS_FEATURE_INODE_CRTIME 0x0100 195 #define F2FS_FEATURE_LOST_FOUND 0x0200 196 #define F2FS_FEATURE_VERITY 0x0400 197 #define F2FS_FEATURE_SB_CHKSUM 0x0800 198 #define F2FS_FEATURE_CASEFOLD 0x1000 199 #define F2FS_FEATURE_COMPRESSION 0x2000 200 #define F2FS_FEATURE_RO 0x4000 201 202 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 203 ((raw_super->feature & cpu_to_le32(mask)) != 0) 204 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 205 #define F2FS_SET_FEATURE(sbi, mask) \ 206 (sbi->raw_super->feature |= cpu_to_le32(mask)) 207 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 208 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 209 210 /* 211 * Default values for user and/or group using reserved blocks 212 */ 213 #define F2FS_DEF_RESUID 0 214 #define F2FS_DEF_RESGID 0 215 216 /* 217 * For checkpoint manager 218 */ 219 enum { 220 NAT_BITMAP, 221 SIT_BITMAP 222 }; 223 224 #define CP_UMOUNT 0x00000001 225 #define CP_FASTBOOT 0x00000002 226 #define CP_SYNC 0x00000004 227 #define CP_RECOVERY 0x00000008 228 #define CP_DISCARD 0x00000010 229 #define CP_TRIMMED 0x00000020 230 #define CP_PAUSE 0x00000040 231 #define CP_RESIZE 0x00000080 232 233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 238 #define DEF_CP_INTERVAL 60 /* 60 secs */ 239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 243 244 struct cp_control { 245 int reason; 246 __u64 trim_start; 247 __u64 trim_end; 248 __u64 trim_minlen; 249 }; 250 251 /* 252 * indicate meta/data type 253 */ 254 enum { 255 META_CP, 256 META_NAT, 257 META_SIT, 258 META_SSA, 259 META_MAX, 260 META_POR, 261 DATA_GENERIC, /* check range only */ 262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 263 DATA_GENERIC_ENHANCE_READ, /* 264 * strong check on range and segment 265 * bitmap but no warning due to race 266 * condition of read on truncated area 267 * by extent_cache 268 */ 269 META_GENERIC, 270 }; 271 272 /* for the list of ino */ 273 enum { 274 ORPHAN_INO, /* for orphan ino list */ 275 APPEND_INO, /* for append ino list */ 276 UPDATE_INO, /* for update ino list */ 277 TRANS_DIR_INO, /* for trasactions dir ino list */ 278 FLUSH_INO, /* for multiple device flushing */ 279 MAX_INO_ENTRY, /* max. list */ 280 }; 281 282 struct ino_entry { 283 struct list_head list; /* list head */ 284 nid_t ino; /* inode number */ 285 unsigned int dirty_device; /* dirty device bitmap */ 286 }; 287 288 /* for the list of inodes to be GCed */ 289 struct inode_entry { 290 struct list_head list; /* list head */ 291 struct inode *inode; /* vfs inode pointer */ 292 }; 293 294 struct fsync_node_entry { 295 struct list_head list; /* list head */ 296 struct page *page; /* warm node page pointer */ 297 unsigned int seq_id; /* sequence id */ 298 }; 299 300 struct ckpt_req { 301 struct completion wait; /* completion for checkpoint done */ 302 struct llist_node llnode; /* llist_node to be linked in wait queue */ 303 int ret; /* return code of checkpoint */ 304 ktime_t queue_time; /* request queued time */ 305 }; 306 307 struct ckpt_req_control { 308 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 309 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 310 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 311 atomic_t issued_ckpt; /* # of actually issued ckpts */ 312 atomic_t total_ckpt; /* # of total ckpts */ 313 atomic_t queued_ckpt; /* # of queued ckpts */ 314 struct llist_head issue_list; /* list for command issue */ 315 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 316 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 317 unsigned int peak_time; /* peak wait time in msec until now */ 318 }; 319 320 /* for the bitmap indicate blocks to be discarded */ 321 struct discard_entry { 322 struct list_head list; /* list head */ 323 block_t start_blkaddr; /* start blockaddr of current segment */ 324 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 325 }; 326 327 /* default discard granularity of inner discard thread, unit: block count */ 328 #define DEFAULT_DISCARD_GRANULARITY 16 329 330 /* max discard pend list number */ 331 #define MAX_PLIST_NUM 512 332 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 333 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 334 335 enum { 336 D_PREP, /* initial */ 337 D_PARTIAL, /* partially submitted */ 338 D_SUBMIT, /* all submitted */ 339 D_DONE, /* finished */ 340 }; 341 342 struct discard_info { 343 block_t lstart; /* logical start address */ 344 block_t len; /* length */ 345 block_t start; /* actual start address in dev */ 346 }; 347 348 struct discard_cmd { 349 struct rb_node rb_node; /* rb node located in rb-tree */ 350 union { 351 struct { 352 block_t lstart; /* logical start address */ 353 block_t len; /* length */ 354 block_t start; /* actual start address in dev */ 355 }; 356 struct discard_info di; /* discard info */ 357 358 }; 359 struct list_head list; /* command list */ 360 struct completion wait; /* compleation */ 361 struct block_device *bdev; /* bdev */ 362 unsigned short ref; /* reference count */ 363 unsigned char state; /* state */ 364 unsigned char queued; /* queued discard */ 365 int error; /* bio error */ 366 spinlock_t lock; /* for state/bio_ref updating */ 367 unsigned short bio_ref; /* bio reference count */ 368 }; 369 370 enum { 371 DPOLICY_BG, 372 DPOLICY_FORCE, 373 DPOLICY_FSTRIM, 374 DPOLICY_UMOUNT, 375 MAX_DPOLICY, 376 }; 377 378 struct discard_policy { 379 int type; /* type of discard */ 380 unsigned int min_interval; /* used for candidates exist */ 381 unsigned int mid_interval; /* used for device busy */ 382 unsigned int max_interval; /* used for candidates not exist */ 383 unsigned int max_requests; /* # of discards issued per round */ 384 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 385 bool io_aware; /* issue discard in idle time */ 386 bool sync; /* submit discard with REQ_SYNC flag */ 387 bool ordered; /* issue discard by lba order */ 388 bool timeout; /* discard timeout for put_super */ 389 unsigned int granularity; /* discard granularity */ 390 }; 391 392 struct discard_cmd_control { 393 struct task_struct *f2fs_issue_discard; /* discard thread */ 394 struct list_head entry_list; /* 4KB discard entry list */ 395 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 396 struct list_head wait_list; /* store on-flushing entries */ 397 struct list_head fstrim_list; /* in-flight discard from fstrim */ 398 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 399 unsigned int discard_wake; /* to wake up discard thread */ 400 struct mutex cmd_lock; 401 unsigned int nr_discards; /* # of discards in the list */ 402 unsigned int max_discards; /* max. discards to be issued */ 403 unsigned int max_discard_request; /* max. discard request per round */ 404 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 405 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 406 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int undiscard_blks; /* # of undiscard blocks */ 409 unsigned int next_pos; /* next discard position */ 410 atomic_t issued_discard; /* # of issued discard */ 411 atomic_t queued_discard; /* # of queued discard */ 412 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 413 struct rb_root_cached root; /* root of discard rb-tree */ 414 bool rbtree_check; /* config for consistence check */ 415 }; 416 417 /* for the list of fsync inodes, used only during recovery */ 418 struct fsync_inode_entry { 419 struct list_head list; /* list head */ 420 struct inode *inode; /* vfs inode pointer */ 421 block_t blkaddr; /* block address locating the last fsync */ 422 block_t last_dentry; /* block address locating the last dentry */ 423 }; 424 425 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 426 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 427 428 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 429 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 430 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 431 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 432 433 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 434 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 435 436 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 437 { 438 int before = nats_in_cursum(journal); 439 440 journal->n_nats = cpu_to_le16(before + i); 441 return before; 442 } 443 444 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = sits_in_cursum(journal); 447 448 journal->n_sits = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline bool __has_cursum_space(struct f2fs_journal *journal, 453 int size, int type) 454 { 455 if (type == NAT_JOURNAL) 456 return size <= MAX_NAT_JENTRIES(journal); 457 return size <= MAX_SIT_JENTRIES(journal); 458 } 459 460 /* for inline stuff */ 461 #define DEF_INLINE_RESERVED_SIZE 1 462 static inline int get_extra_isize(struct inode *inode); 463 static inline int get_inline_xattr_addrs(struct inode *inode); 464 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 465 (CUR_ADDRS_PER_INODE(inode) - \ 466 get_inline_xattr_addrs(inode) - \ 467 DEF_INLINE_RESERVED_SIZE)) 468 469 /* for inline dir */ 470 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 471 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 472 BITS_PER_BYTE + 1)) 473 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 474 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 475 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 477 NR_INLINE_DENTRY(inode) + \ 478 INLINE_DENTRY_BITMAP_SIZE(inode))) 479 480 /* 481 * For INODE and NODE manager 482 */ 483 /* for directory operations */ 484 485 struct f2fs_filename { 486 /* 487 * The filename the user specified. This is NULL for some 488 * filesystem-internal operations, e.g. converting an inline directory 489 * to a non-inline one, or roll-forward recovering an encrypted dentry. 490 */ 491 const struct qstr *usr_fname; 492 493 /* 494 * The on-disk filename. For encrypted directories, this is encrypted. 495 * This may be NULL for lookups in an encrypted dir without the key. 496 */ 497 struct fscrypt_str disk_name; 498 499 /* The dirhash of this filename */ 500 f2fs_hash_t hash; 501 502 #ifdef CONFIG_FS_ENCRYPTION 503 /* 504 * For lookups in encrypted directories: either the buffer backing 505 * disk_name, or a buffer that holds the decoded no-key name. 506 */ 507 struct fscrypt_str crypto_buf; 508 #endif 509 #if IS_ENABLED(CONFIG_UNICODE) 510 /* 511 * For casefolded directories: the casefolded name, but it's left NULL 512 * if the original name is not valid Unicode, if the original name is 513 * "." or "..", if the directory is both casefolded and encrypted and 514 * its encryption key is unavailable, or if the filesystem is doing an 515 * internal operation where usr_fname is also NULL. In all these cases 516 * we fall back to treating the name as an opaque byte sequence. 517 */ 518 struct fscrypt_str cf_name; 519 #endif 520 }; 521 522 struct f2fs_dentry_ptr { 523 struct inode *inode; 524 void *bitmap; 525 struct f2fs_dir_entry *dentry; 526 __u8 (*filename)[F2FS_SLOT_LEN]; 527 int max; 528 int nr_bitmap; 529 }; 530 531 static inline void make_dentry_ptr_block(struct inode *inode, 532 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 533 { 534 d->inode = inode; 535 d->max = NR_DENTRY_IN_BLOCK; 536 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 537 d->bitmap = t->dentry_bitmap; 538 d->dentry = t->dentry; 539 d->filename = t->filename; 540 } 541 542 static inline void make_dentry_ptr_inline(struct inode *inode, 543 struct f2fs_dentry_ptr *d, void *t) 544 { 545 int entry_cnt = NR_INLINE_DENTRY(inode); 546 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 547 int reserved_size = INLINE_RESERVED_SIZE(inode); 548 549 d->inode = inode; 550 d->max = entry_cnt; 551 d->nr_bitmap = bitmap_size; 552 d->bitmap = t; 553 d->dentry = t + bitmap_size + reserved_size; 554 d->filename = t + bitmap_size + reserved_size + 555 SIZE_OF_DIR_ENTRY * entry_cnt; 556 } 557 558 /* 559 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 560 * as its node offset to distinguish from index node blocks. 561 * But some bits are used to mark the node block. 562 */ 563 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 564 >> OFFSET_BIT_SHIFT) 565 enum { 566 ALLOC_NODE, /* allocate a new node page if needed */ 567 LOOKUP_NODE, /* look up a node without readahead */ 568 LOOKUP_NODE_RA, /* 569 * look up a node with readahead called 570 * by get_data_block. 571 */ 572 }; 573 574 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 575 576 /* congestion wait timeout value, default: 20ms */ 577 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 578 579 /* maximum retry quota flush count */ 580 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 581 582 /* maximum retry of EIO'ed page */ 583 #define MAX_RETRY_PAGE_EIO 100 584 585 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 586 587 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 588 589 /* dirty segments threshold for triggering CP */ 590 #define DEFAULT_DIRTY_THRESHOLD 4 591 592 /* for in-memory extent cache entry */ 593 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 594 595 /* number of extent info in extent cache we try to shrink */ 596 #define EXTENT_CACHE_SHRINK_NUMBER 128 597 598 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 599 #define RECOVERY_MIN_RA_BLOCKS 1 600 601 struct rb_entry { 602 struct rb_node rb_node; /* rb node located in rb-tree */ 603 union { 604 struct { 605 unsigned int ofs; /* start offset of the entry */ 606 unsigned int len; /* length of the entry */ 607 }; 608 unsigned long long key; /* 64-bits key */ 609 } __packed; 610 }; 611 612 struct extent_info { 613 unsigned int fofs; /* start offset in a file */ 614 unsigned int len; /* length of the extent */ 615 u32 blk; /* start block address of the extent */ 616 #ifdef CONFIG_F2FS_FS_COMPRESSION 617 unsigned int c_len; /* physical extent length of compressed blocks */ 618 #endif 619 }; 620 621 struct extent_node { 622 struct rb_node rb_node; /* rb node located in rb-tree */ 623 struct extent_info ei; /* extent info */ 624 struct list_head list; /* node in global extent list of sbi */ 625 struct extent_tree *et; /* extent tree pointer */ 626 }; 627 628 struct extent_tree { 629 nid_t ino; /* inode number */ 630 struct rb_root_cached root; /* root of extent info rb-tree */ 631 struct extent_node *cached_en; /* recently accessed extent node */ 632 struct extent_info largest; /* largested extent info */ 633 struct list_head list; /* to be used by sbi->zombie_list */ 634 rwlock_t lock; /* protect extent info rb-tree */ 635 atomic_t node_cnt; /* # of extent node in rb-tree*/ 636 bool largest_updated; /* largest extent updated */ 637 }; 638 639 /* 640 * This structure is taken from ext4_map_blocks. 641 * 642 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 643 */ 644 #define F2FS_MAP_NEW (1 << BH_New) 645 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 646 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 647 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 648 F2FS_MAP_UNWRITTEN) 649 650 struct f2fs_map_blocks { 651 struct block_device *m_bdev; /* for multi-device dio */ 652 block_t m_pblk; 653 block_t m_lblk; 654 unsigned int m_len; 655 unsigned int m_flags; 656 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 657 pgoff_t *m_next_extent; /* point to next possible extent */ 658 int m_seg_type; 659 bool m_may_create; /* indicate it is from write path */ 660 bool m_multidev_dio; /* indicate it allows multi-device dio */ 661 }; 662 663 /* for flag in get_data_block */ 664 enum { 665 F2FS_GET_BLOCK_DEFAULT, 666 F2FS_GET_BLOCK_FIEMAP, 667 F2FS_GET_BLOCK_BMAP, 668 F2FS_GET_BLOCK_DIO, 669 F2FS_GET_BLOCK_PRE_DIO, 670 F2FS_GET_BLOCK_PRE_AIO, 671 F2FS_GET_BLOCK_PRECACHE, 672 }; 673 674 /* 675 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 676 */ 677 #define FADVISE_COLD_BIT 0x01 678 #define FADVISE_LOST_PINO_BIT 0x02 679 #define FADVISE_ENCRYPT_BIT 0x04 680 #define FADVISE_ENC_NAME_BIT 0x08 681 #define FADVISE_KEEP_SIZE_BIT 0x10 682 #define FADVISE_HOT_BIT 0x20 683 #define FADVISE_VERITY_BIT 0x40 684 #define FADVISE_TRUNC_BIT 0x80 685 686 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 687 688 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 689 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 690 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 691 692 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 694 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 695 696 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 697 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 698 699 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 700 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 701 702 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 703 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 704 705 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 706 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 707 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 708 709 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 710 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 711 712 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 713 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 714 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 715 716 #define DEF_DIR_LEVEL 0 717 718 enum { 719 GC_FAILURE_PIN, 720 MAX_GC_FAILURE 721 }; 722 723 /* used for f2fs_inode_info->flags */ 724 enum { 725 FI_NEW_INODE, /* indicate newly allocated inode */ 726 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 727 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 728 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 729 FI_INC_LINK, /* need to increment i_nlink */ 730 FI_ACL_MODE, /* indicate acl mode */ 731 FI_NO_ALLOC, /* should not allocate any blocks */ 732 FI_FREE_NID, /* free allocated nide */ 733 FI_NO_EXTENT, /* not to use the extent cache */ 734 FI_INLINE_XATTR, /* used for inline xattr */ 735 FI_INLINE_DATA, /* used for inline data*/ 736 FI_INLINE_DENTRY, /* used for inline dentry */ 737 FI_APPEND_WRITE, /* inode has appended data */ 738 FI_UPDATE_WRITE, /* inode has in-place-update data */ 739 FI_NEED_IPU, /* used for ipu per file */ 740 FI_ATOMIC_FILE, /* indicate atomic file */ 741 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 742 FI_DROP_CACHE, /* drop dirty page cache */ 743 FI_DATA_EXIST, /* indicate data exists */ 744 FI_INLINE_DOTS, /* indicate inline dot dentries */ 745 FI_SKIP_WRITES, /* should skip data page writeback */ 746 FI_OPU_WRITE, /* used for opu per file */ 747 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 748 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 749 FI_HOT_DATA, /* indicate file is hot */ 750 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 751 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 752 FI_PIN_FILE, /* indicate file should not be gced */ 753 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 754 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 755 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 756 FI_MMAP_FILE, /* indicate file was mmapped */ 757 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 758 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 759 FI_ALIGNED_WRITE, /* enable aligned write */ 760 FI_COW_FILE, /* indicate COW file */ 761 FI_MAX, /* max flag, never be used */ 762 }; 763 764 struct f2fs_inode_info { 765 struct inode vfs_inode; /* serve a vfs inode */ 766 unsigned long i_flags; /* keep an inode flags for ioctl */ 767 unsigned char i_advise; /* use to give file attribute hints */ 768 unsigned char i_dir_level; /* use for dentry level for large dir */ 769 unsigned int i_current_depth; /* only for directory depth */ 770 /* for gc failure statistic */ 771 unsigned int i_gc_failures[MAX_GC_FAILURE]; 772 unsigned int i_pino; /* parent inode number */ 773 umode_t i_acl_mode; /* keep file acl mode temporarily */ 774 775 /* Use below internally in f2fs*/ 776 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 777 struct f2fs_rwsem i_sem; /* protect fi info */ 778 atomic_t dirty_pages; /* # of dirty pages */ 779 f2fs_hash_t chash; /* hash value of given file name */ 780 unsigned int clevel; /* maximum level of given file name */ 781 struct task_struct *task; /* lookup and create consistency */ 782 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 783 nid_t i_xattr_nid; /* node id that contains xattrs */ 784 loff_t last_disk_size; /* lastly written file size */ 785 spinlock_t i_size_lock; /* protect last_disk_size */ 786 787 #ifdef CONFIG_QUOTA 788 struct dquot *i_dquot[MAXQUOTAS]; 789 790 /* quota space reservation, managed internally by quota code */ 791 qsize_t i_reserved_quota; 792 #endif 793 struct list_head dirty_list; /* dirty list for dirs and files */ 794 struct list_head gdirty_list; /* linked in global dirty list */ 795 struct task_struct *atomic_write_task; /* store atomic write task */ 796 struct extent_tree *extent_tree; /* cached extent_tree entry */ 797 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 798 799 /* avoid racing between foreground op and gc */ 800 struct f2fs_rwsem i_gc_rwsem[2]; 801 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 802 803 int i_extra_isize; /* size of extra space located in i_addr */ 804 kprojid_t i_projid; /* id for project quota */ 805 int i_inline_xattr_size; /* inline xattr size */ 806 struct timespec64 i_crtime; /* inode creation time */ 807 struct timespec64 i_disk_time[4];/* inode disk times */ 808 809 /* for file compress */ 810 atomic_t i_compr_blocks; /* # of compressed blocks */ 811 unsigned char i_compress_algorithm; /* algorithm type */ 812 unsigned char i_log_cluster_size; /* log of cluster size */ 813 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 814 unsigned short i_compress_flag; /* compress flag */ 815 unsigned int i_cluster_size; /* cluster size */ 816 817 unsigned int atomic_write_cnt; 818 }; 819 820 static inline void get_extent_info(struct extent_info *ext, 821 struct f2fs_extent *i_ext) 822 { 823 ext->fofs = le32_to_cpu(i_ext->fofs); 824 ext->blk = le32_to_cpu(i_ext->blk); 825 ext->len = le32_to_cpu(i_ext->len); 826 } 827 828 static inline void set_raw_extent(struct extent_info *ext, 829 struct f2fs_extent *i_ext) 830 { 831 i_ext->fofs = cpu_to_le32(ext->fofs); 832 i_ext->blk = cpu_to_le32(ext->blk); 833 i_ext->len = cpu_to_le32(ext->len); 834 } 835 836 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 837 u32 blk, unsigned int len) 838 { 839 ei->fofs = fofs; 840 ei->blk = blk; 841 ei->len = len; 842 #ifdef CONFIG_F2FS_FS_COMPRESSION 843 ei->c_len = 0; 844 #endif 845 } 846 847 static inline bool __is_discard_mergeable(struct discard_info *back, 848 struct discard_info *front, unsigned int max_len) 849 { 850 return (back->lstart + back->len == front->lstart) && 851 (back->len + front->len <= max_len); 852 } 853 854 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 855 struct discard_info *back, unsigned int max_len) 856 { 857 return __is_discard_mergeable(back, cur, max_len); 858 } 859 860 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 861 struct discard_info *front, unsigned int max_len) 862 { 863 return __is_discard_mergeable(cur, front, max_len); 864 } 865 866 static inline bool __is_extent_mergeable(struct extent_info *back, 867 struct extent_info *front) 868 { 869 #ifdef CONFIG_F2FS_FS_COMPRESSION 870 if (back->c_len && back->len != back->c_len) 871 return false; 872 if (front->c_len && front->len != front->c_len) 873 return false; 874 #endif 875 return (back->fofs + back->len == front->fofs && 876 back->blk + back->len == front->blk); 877 } 878 879 static inline bool __is_back_mergeable(struct extent_info *cur, 880 struct extent_info *back) 881 { 882 return __is_extent_mergeable(back, cur); 883 } 884 885 static inline bool __is_front_mergeable(struct extent_info *cur, 886 struct extent_info *front) 887 { 888 return __is_extent_mergeable(cur, front); 889 } 890 891 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 892 static inline void __try_update_largest_extent(struct extent_tree *et, 893 struct extent_node *en) 894 { 895 if (en->ei.len > et->largest.len) { 896 et->largest = en->ei; 897 et->largest_updated = true; 898 } 899 } 900 901 /* 902 * For free nid management 903 */ 904 enum nid_state { 905 FREE_NID, /* newly added to free nid list */ 906 PREALLOC_NID, /* it is preallocated */ 907 MAX_NID_STATE, 908 }; 909 910 enum nat_state { 911 TOTAL_NAT, 912 DIRTY_NAT, 913 RECLAIMABLE_NAT, 914 MAX_NAT_STATE, 915 }; 916 917 struct f2fs_nm_info { 918 block_t nat_blkaddr; /* base disk address of NAT */ 919 nid_t max_nid; /* maximum possible node ids */ 920 nid_t available_nids; /* # of available node ids */ 921 nid_t next_scan_nid; /* the next nid to be scanned */ 922 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 923 unsigned int ram_thresh; /* control the memory footprint */ 924 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 925 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 926 927 /* NAT cache management */ 928 struct radix_tree_root nat_root;/* root of the nat entry cache */ 929 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 930 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 931 struct list_head nat_entries; /* cached nat entry list (clean) */ 932 spinlock_t nat_list_lock; /* protect clean nat entry list */ 933 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 934 unsigned int nat_blocks; /* # of nat blocks */ 935 936 /* free node ids management */ 937 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 938 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 939 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 940 spinlock_t nid_list_lock; /* protect nid lists ops */ 941 struct mutex build_lock; /* lock for build free nids */ 942 unsigned char **free_nid_bitmap; 943 unsigned char *nat_block_bitmap; 944 unsigned short *free_nid_count; /* free nid count of NAT block */ 945 946 /* for checkpoint */ 947 char *nat_bitmap; /* NAT bitmap pointer */ 948 949 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 950 unsigned char *nat_bits; /* NAT bits blocks */ 951 unsigned char *full_nat_bits; /* full NAT pages */ 952 unsigned char *empty_nat_bits; /* empty NAT pages */ 953 #ifdef CONFIG_F2FS_CHECK_FS 954 char *nat_bitmap_mir; /* NAT bitmap mirror */ 955 #endif 956 int bitmap_size; /* bitmap size */ 957 }; 958 959 /* 960 * this structure is used as one of function parameters. 961 * all the information are dedicated to a given direct node block determined 962 * by the data offset in a file. 963 */ 964 struct dnode_of_data { 965 struct inode *inode; /* vfs inode pointer */ 966 struct page *inode_page; /* its inode page, NULL is possible */ 967 struct page *node_page; /* cached direct node page */ 968 nid_t nid; /* node id of the direct node block */ 969 unsigned int ofs_in_node; /* data offset in the node page */ 970 bool inode_page_locked; /* inode page is locked or not */ 971 bool node_changed; /* is node block changed */ 972 char cur_level; /* level of hole node page */ 973 char max_level; /* level of current page located */ 974 block_t data_blkaddr; /* block address of the node block */ 975 }; 976 977 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 978 struct page *ipage, struct page *npage, nid_t nid) 979 { 980 memset(dn, 0, sizeof(*dn)); 981 dn->inode = inode; 982 dn->inode_page = ipage; 983 dn->node_page = npage; 984 dn->nid = nid; 985 } 986 987 /* 988 * For SIT manager 989 * 990 * By default, there are 6 active log areas across the whole main area. 991 * When considering hot and cold data separation to reduce cleaning overhead, 992 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 993 * respectively. 994 * In the current design, you should not change the numbers intentionally. 995 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 996 * logs individually according to the underlying devices. (default: 6) 997 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 998 * data and 8 for node logs. 999 */ 1000 #define NR_CURSEG_DATA_TYPE (3) 1001 #define NR_CURSEG_NODE_TYPE (3) 1002 #define NR_CURSEG_INMEM_TYPE (2) 1003 #define NR_CURSEG_RO_TYPE (2) 1004 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1005 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1006 1007 enum { 1008 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1009 CURSEG_WARM_DATA, /* data blocks */ 1010 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1011 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1012 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1013 CURSEG_COLD_NODE, /* indirect node blocks */ 1014 NR_PERSISTENT_LOG, /* number of persistent log */ 1015 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1016 /* pinned file that needs consecutive block address */ 1017 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1018 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1019 }; 1020 1021 struct flush_cmd { 1022 struct completion wait; 1023 struct llist_node llnode; 1024 nid_t ino; 1025 int ret; 1026 }; 1027 1028 struct flush_cmd_control { 1029 struct task_struct *f2fs_issue_flush; /* flush thread */ 1030 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1031 atomic_t issued_flush; /* # of issued flushes */ 1032 atomic_t queued_flush; /* # of queued flushes */ 1033 struct llist_head issue_list; /* list for command issue */ 1034 struct llist_node *dispatch_list; /* list for command dispatch */ 1035 }; 1036 1037 struct f2fs_sm_info { 1038 struct sit_info *sit_info; /* whole segment information */ 1039 struct free_segmap_info *free_info; /* free segment information */ 1040 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1041 struct curseg_info *curseg_array; /* active segment information */ 1042 1043 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1044 1045 block_t seg0_blkaddr; /* block address of 0'th segment */ 1046 block_t main_blkaddr; /* start block address of main area */ 1047 block_t ssa_blkaddr; /* start block address of SSA area */ 1048 1049 unsigned int segment_count; /* total # of segments */ 1050 unsigned int main_segments; /* # of segments in main area */ 1051 unsigned int reserved_segments; /* # of reserved segments */ 1052 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1053 unsigned int ovp_segments; /* # of overprovision segments */ 1054 1055 /* a threshold to reclaim prefree segments */ 1056 unsigned int rec_prefree_segments; 1057 1058 /* for batched trimming */ 1059 unsigned int trim_sections; /* # of sections to trim */ 1060 1061 struct list_head sit_entry_set; /* sit entry set list */ 1062 1063 unsigned int ipu_policy; /* in-place-update policy */ 1064 unsigned int min_ipu_util; /* in-place-update threshold */ 1065 unsigned int min_fsync_blocks; /* threshold for fsync */ 1066 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1067 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1068 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1069 1070 /* for flush command control */ 1071 struct flush_cmd_control *fcc_info; 1072 1073 /* for discard command control */ 1074 struct discard_cmd_control *dcc_info; 1075 }; 1076 1077 /* 1078 * For superblock 1079 */ 1080 /* 1081 * COUNT_TYPE for monitoring 1082 * 1083 * f2fs monitors the number of several block types such as on-writeback, 1084 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1085 */ 1086 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1087 enum count_type { 1088 F2FS_DIRTY_DENTS, 1089 F2FS_DIRTY_DATA, 1090 F2FS_DIRTY_QDATA, 1091 F2FS_DIRTY_NODES, 1092 F2FS_DIRTY_META, 1093 F2FS_DIRTY_IMETA, 1094 F2FS_WB_CP_DATA, 1095 F2FS_WB_DATA, 1096 F2FS_RD_DATA, 1097 F2FS_RD_NODE, 1098 F2FS_RD_META, 1099 F2FS_DIO_WRITE, 1100 F2FS_DIO_READ, 1101 NR_COUNT_TYPE, 1102 }; 1103 1104 /* 1105 * The below are the page types of bios used in submit_bio(). 1106 * The available types are: 1107 * DATA User data pages. It operates as async mode. 1108 * NODE Node pages. It operates as async mode. 1109 * META FS metadata pages such as SIT, NAT, CP. 1110 * NR_PAGE_TYPE The number of page types. 1111 * META_FLUSH Make sure the previous pages are written 1112 * with waiting the bio's completion 1113 * ... Only can be used with META. 1114 */ 1115 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1116 enum page_type { 1117 DATA = 0, 1118 NODE = 1, /* should not change this */ 1119 META, 1120 NR_PAGE_TYPE, 1121 META_FLUSH, 1122 IPU, /* the below types are used by tracepoints only. */ 1123 OPU, 1124 }; 1125 1126 enum temp_type { 1127 HOT = 0, /* must be zero for meta bio */ 1128 WARM, 1129 COLD, 1130 NR_TEMP_TYPE, 1131 }; 1132 1133 enum need_lock_type { 1134 LOCK_REQ = 0, 1135 LOCK_DONE, 1136 LOCK_RETRY, 1137 }; 1138 1139 enum cp_reason_type { 1140 CP_NO_NEEDED, 1141 CP_NON_REGULAR, 1142 CP_COMPRESSED, 1143 CP_HARDLINK, 1144 CP_SB_NEED_CP, 1145 CP_WRONG_PINO, 1146 CP_NO_SPC_ROLL, 1147 CP_NODE_NEED_CP, 1148 CP_FASTBOOT_MODE, 1149 CP_SPEC_LOG_NUM, 1150 CP_RECOVER_DIR, 1151 }; 1152 1153 enum iostat_type { 1154 /* WRITE IO */ 1155 APP_DIRECT_IO, /* app direct write IOs */ 1156 APP_BUFFERED_IO, /* app buffered write IOs */ 1157 APP_WRITE_IO, /* app write IOs */ 1158 APP_MAPPED_IO, /* app mapped IOs */ 1159 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1160 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1161 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1162 FS_GC_DATA_IO, /* data IOs from forground gc */ 1163 FS_GC_NODE_IO, /* node IOs from forground gc */ 1164 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1165 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1166 FS_CP_META_IO, /* meta IOs from checkpoint */ 1167 1168 /* READ IO */ 1169 APP_DIRECT_READ_IO, /* app direct read IOs */ 1170 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1171 APP_READ_IO, /* app read IOs */ 1172 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1173 FS_DATA_READ_IO, /* data read IOs */ 1174 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1175 FS_CDATA_READ_IO, /* compressed data read IOs */ 1176 FS_NODE_READ_IO, /* node read IOs */ 1177 FS_META_READ_IO, /* meta read IOs */ 1178 1179 /* other */ 1180 FS_DISCARD, /* discard */ 1181 NR_IO_TYPE, 1182 }; 1183 1184 struct f2fs_io_info { 1185 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1186 nid_t ino; /* inode number */ 1187 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1188 enum temp_type temp; /* contains HOT/WARM/COLD */ 1189 int op; /* contains REQ_OP_ */ 1190 int op_flags; /* req_flag_bits */ 1191 block_t new_blkaddr; /* new block address to be written */ 1192 block_t old_blkaddr; /* old block address before Cow */ 1193 struct page *page; /* page to be written */ 1194 struct page *encrypted_page; /* encrypted page */ 1195 struct page *compressed_page; /* compressed page */ 1196 struct list_head list; /* serialize IOs */ 1197 bool submitted; /* indicate IO submission */ 1198 int need_lock; /* indicate we need to lock cp_rwsem */ 1199 bool in_list; /* indicate fio is in io_list */ 1200 bool is_por; /* indicate IO is from recovery or not */ 1201 bool retry; /* need to reallocate block address */ 1202 int compr_blocks; /* # of compressed block addresses */ 1203 bool encrypted; /* indicate file is encrypted */ 1204 bool post_read; /* require post read */ 1205 enum iostat_type io_type; /* io type */ 1206 struct writeback_control *io_wbc; /* writeback control */ 1207 struct bio **bio; /* bio for ipu */ 1208 sector_t *last_block; /* last block number in bio */ 1209 unsigned char version; /* version of the node */ 1210 }; 1211 1212 struct bio_entry { 1213 struct bio *bio; 1214 struct list_head list; 1215 }; 1216 1217 #define is_read_io(rw) ((rw) == READ) 1218 struct f2fs_bio_info { 1219 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1220 struct bio *bio; /* bios to merge */ 1221 sector_t last_block_in_bio; /* last block number */ 1222 struct f2fs_io_info fio; /* store buffered io info. */ 1223 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1224 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1225 struct list_head io_list; /* track fios */ 1226 struct list_head bio_list; /* bio entry list head */ 1227 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1228 }; 1229 1230 #define FDEV(i) (sbi->devs[i]) 1231 #define RDEV(i) (raw_super->devs[i]) 1232 struct f2fs_dev_info { 1233 struct block_device *bdev; 1234 char path[MAX_PATH_LEN]; 1235 unsigned int total_segments; 1236 block_t start_blk; 1237 block_t end_blk; 1238 #ifdef CONFIG_BLK_DEV_ZONED 1239 unsigned int nr_blkz; /* Total number of zones */ 1240 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1241 #endif 1242 }; 1243 1244 enum inode_type { 1245 DIR_INODE, /* for dirty dir inode */ 1246 FILE_INODE, /* for dirty regular/symlink inode */ 1247 DIRTY_META, /* for all dirtied inode metadata */ 1248 ATOMIC_FILE, /* for all atomic files */ 1249 NR_INODE_TYPE, 1250 }; 1251 1252 /* for inner inode cache management */ 1253 struct inode_management { 1254 struct radix_tree_root ino_root; /* ino entry array */ 1255 spinlock_t ino_lock; /* for ino entry lock */ 1256 struct list_head ino_list; /* inode list head */ 1257 unsigned long ino_num; /* number of entries */ 1258 }; 1259 1260 /* for GC_AT */ 1261 struct atgc_management { 1262 bool atgc_enabled; /* ATGC is enabled or not */ 1263 struct rb_root_cached root; /* root of victim rb-tree */ 1264 struct list_head victim_list; /* linked with all victim entries */ 1265 unsigned int victim_count; /* victim count in rb-tree */ 1266 unsigned int candidate_ratio; /* candidate ratio */ 1267 unsigned int max_candidate_count; /* max candidate count */ 1268 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1269 unsigned long long age_threshold; /* age threshold */ 1270 }; 1271 1272 struct f2fs_gc_control { 1273 unsigned int victim_segno; /* target victim segment number */ 1274 int init_gc_type; /* FG_GC or BG_GC */ 1275 bool no_bg_gc; /* check the space and stop bg_gc */ 1276 bool should_migrate_blocks; /* should migrate blocks */ 1277 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1278 unsigned int nr_free_secs; /* # of free sections to do GC */ 1279 }; 1280 1281 /* For s_flag in struct f2fs_sb_info */ 1282 enum { 1283 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1284 SBI_IS_CLOSE, /* specify unmounting */ 1285 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1286 SBI_POR_DOING, /* recovery is doing or not */ 1287 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1288 SBI_NEED_CP, /* need to checkpoint */ 1289 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1290 SBI_IS_RECOVERED, /* recovered orphan/data */ 1291 SBI_CP_DISABLED, /* CP was disabled last mount */ 1292 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1293 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1294 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1295 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1296 SBI_IS_RESIZEFS, /* resizefs is in process */ 1297 SBI_IS_FREEZING, /* freezefs is in process */ 1298 }; 1299 1300 enum { 1301 CP_TIME, 1302 REQ_TIME, 1303 DISCARD_TIME, 1304 GC_TIME, 1305 DISABLE_TIME, 1306 UMOUNT_DISCARD_TIMEOUT, 1307 MAX_TIME, 1308 }; 1309 1310 enum { 1311 GC_NORMAL, 1312 GC_IDLE_CB, 1313 GC_IDLE_GREEDY, 1314 GC_IDLE_AT, 1315 GC_URGENT_HIGH, 1316 GC_URGENT_LOW, 1317 GC_URGENT_MID, 1318 MAX_GC_MODE, 1319 }; 1320 1321 enum { 1322 BGGC_MODE_ON, /* background gc is on */ 1323 BGGC_MODE_OFF, /* background gc is off */ 1324 BGGC_MODE_SYNC, /* 1325 * background gc is on, migrating blocks 1326 * like foreground gc 1327 */ 1328 }; 1329 1330 enum { 1331 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1332 FS_MODE_LFS, /* use lfs allocation only */ 1333 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1334 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1335 }; 1336 1337 enum { 1338 ALLOC_MODE_DEFAULT, /* stay default */ 1339 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1340 }; 1341 1342 enum fsync_mode { 1343 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1344 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1345 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1346 }; 1347 1348 enum { 1349 COMPR_MODE_FS, /* 1350 * automatically compress compression 1351 * enabled files 1352 */ 1353 COMPR_MODE_USER, /* 1354 * automatical compression is disabled. 1355 * user can control the file compression 1356 * using ioctls 1357 */ 1358 }; 1359 1360 enum { 1361 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1362 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1363 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1364 }; 1365 1366 enum { 1367 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1368 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1369 }; 1370 1371 1372 1373 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1374 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1375 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1376 1377 /* 1378 * Layout of f2fs page.private: 1379 * 1380 * Layout A: lowest bit should be 1 1381 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1382 * bit 0 PAGE_PRIVATE_NOT_POINTER 1383 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1384 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1385 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1386 * bit 4 PAGE_PRIVATE_INLINE_INODE 1387 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1388 * bit 6- f2fs private data 1389 * 1390 * Layout B: lowest bit should be 0 1391 * page.private is a wrapped pointer. 1392 */ 1393 enum { 1394 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1395 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1396 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1397 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1398 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1399 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1400 PAGE_PRIVATE_MAX 1401 }; 1402 1403 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1404 static inline bool page_private_##name(struct page *page) \ 1405 { \ 1406 return PagePrivate(page) && \ 1407 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1408 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1409 } 1410 1411 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1412 static inline void set_page_private_##name(struct page *page) \ 1413 { \ 1414 if (!PagePrivate(page)) { \ 1415 get_page(page); \ 1416 SetPagePrivate(page); \ 1417 set_page_private(page, 0); \ 1418 } \ 1419 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1420 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1421 } 1422 1423 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1424 static inline void clear_page_private_##name(struct page *page) \ 1425 { \ 1426 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1427 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1428 set_page_private(page, 0); \ 1429 if (PagePrivate(page)) { \ 1430 ClearPagePrivate(page); \ 1431 put_page(page); \ 1432 }\ 1433 } \ 1434 } 1435 1436 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1437 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1438 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1439 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1440 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1441 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1442 1443 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1444 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1445 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1446 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1447 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1448 1449 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1450 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1451 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1452 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1453 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1454 1455 static inline unsigned long get_page_private_data(struct page *page) 1456 { 1457 unsigned long data = page_private(page); 1458 1459 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1460 return 0; 1461 return data >> PAGE_PRIVATE_MAX; 1462 } 1463 1464 static inline void set_page_private_data(struct page *page, unsigned long data) 1465 { 1466 if (!PagePrivate(page)) { 1467 get_page(page); 1468 SetPagePrivate(page); 1469 set_page_private(page, 0); 1470 } 1471 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1472 page_private(page) |= data << PAGE_PRIVATE_MAX; 1473 } 1474 1475 static inline void clear_page_private_data(struct page *page) 1476 { 1477 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1478 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1479 set_page_private(page, 0); 1480 if (PagePrivate(page)) { 1481 ClearPagePrivate(page); 1482 put_page(page); 1483 } 1484 } 1485 } 1486 1487 /* For compression */ 1488 enum compress_algorithm_type { 1489 COMPRESS_LZO, 1490 COMPRESS_LZ4, 1491 COMPRESS_ZSTD, 1492 COMPRESS_LZORLE, 1493 COMPRESS_MAX, 1494 }; 1495 1496 enum compress_flag { 1497 COMPRESS_CHKSUM, 1498 COMPRESS_MAX_FLAG, 1499 }; 1500 1501 #define COMPRESS_WATERMARK 20 1502 #define COMPRESS_PERCENT 20 1503 1504 #define COMPRESS_DATA_RESERVED_SIZE 4 1505 struct compress_data { 1506 __le32 clen; /* compressed data size */ 1507 __le32 chksum; /* compressed data chksum */ 1508 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1509 u8 cdata[]; /* compressed data */ 1510 }; 1511 1512 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1513 1514 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1515 1516 #define COMPRESS_LEVEL_OFFSET 8 1517 1518 /* compress context */ 1519 struct compress_ctx { 1520 struct inode *inode; /* inode the context belong to */ 1521 pgoff_t cluster_idx; /* cluster index number */ 1522 unsigned int cluster_size; /* page count in cluster */ 1523 unsigned int log_cluster_size; /* log of cluster size */ 1524 struct page **rpages; /* pages store raw data in cluster */ 1525 unsigned int nr_rpages; /* total page number in rpages */ 1526 struct page **cpages; /* pages store compressed data in cluster */ 1527 unsigned int nr_cpages; /* total page number in cpages */ 1528 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1529 void *rbuf; /* virtual mapped address on rpages */ 1530 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1531 size_t rlen; /* valid data length in rbuf */ 1532 size_t clen; /* valid data length in cbuf */ 1533 void *private; /* payload buffer for specified compression algorithm */ 1534 void *private2; /* extra payload buffer */ 1535 }; 1536 1537 /* compress context for write IO path */ 1538 struct compress_io_ctx { 1539 u32 magic; /* magic number to indicate page is compressed */ 1540 struct inode *inode; /* inode the context belong to */ 1541 struct page **rpages; /* pages store raw data in cluster */ 1542 unsigned int nr_rpages; /* total page number in rpages */ 1543 atomic_t pending_pages; /* in-flight compressed page count */ 1544 }; 1545 1546 /* Context for decompressing one cluster on the read IO path */ 1547 struct decompress_io_ctx { 1548 u32 magic; /* magic number to indicate page is compressed */ 1549 struct inode *inode; /* inode the context belong to */ 1550 pgoff_t cluster_idx; /* cluster index number */ 1551 unsigned int cluster_size; /* page count in cluster */ 1552 unsigned int log_cluster_size; /* log of cluster size */ 1553 struct page **rpages; /* pages store raw data in cluster */ 1554 unsigned int nr_rpages; /* total page number in rpages */ 1555 struct page **cpages; /* pages store compressed data in cluster */ 1556 unsigned int nr_cpages; /* total page number in cpages */ 1557 struct page **tpages; /* temp pages to pad holes in cluster */ 1558 void *rbuf; /* virtual mapped address on rpages */ 1559 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1560 size_t rlen; /* valid data length in rbuf */ 1561 size_t clen; /* valid data length in cbuf */ 1562 1563 /* 1564 * The number of compressed pages remaining to be read in this cluster. 1565 * This is initially nr_cpages. It is decremented by 1 each time a page 1566 * has been read (or failed to be read). When it reaches 0, the cluster 1567 * is decompressed (or an error is reported). 1568 * 1569 * If an error occurs before all the pages have been submitted for I/O, 1570 * then this will never reach 0. In this case the I/O submitter is 1571 * responsible for calling f2fs_decompress_end_io() instead. 1572 */ 1573 atomic_t remaining_pages; 1574 1575 /* 1576 * Number of references to this decompress_io_ctx. 1577 * 1578 * One reference is held for I/O completion. This reference is dropped 1579 * after the pagecache pages are updated and unlocked -- either after 1580 * decompression (and verity if enabled), or after an error. 1581 * 1582 * In addition, each compressed page holds a reference while it is in a 1583 * bio. These references are necessary prevent compressed pages from 1584 * being freed while they are still in a bio. 1585 */ 1586 refcount_t refcnt; 1587 1588 bool failed; /* IO error occurred before decompression? */ 1589 bool need_verity; /* need fs-verity verification after decompression? */ 1590 void *private; /* payload buffer for specified decompression algorithm */ 1591 void *private2; /* extra payload buffer */ 1592 struct work_struct verity_work; /* work to verify the decompressed pages */ 1593 struct work_struct free_work; /* work for late free this structure itself */ 1594 }; 1595 1596 #define NULL_CLUSTER ((unsigned int)(~0)) 1597 #define MIN_COMPRESS_LOG_SIZE 2 1598 #define MAX_COMPRESS_LOG_SIZE 8 1599 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1600 1601 struct f2fs_sb_info { 1602 struct super_block *sb; /* pointer to VFS super block */ 1603 struct proc_dir_entry *s_proc; /* proc entry */ 1604 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1605 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1606 int valid_super_block; /* valid super block no */ 1607 unsigned long s_flag; /* flags for sbi */ 1608 struct mutex writepages; /* mutex for writepages() */ 1609 1610 #ifdef CONFIG_BLK_DEV_ZONED 1611 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1612 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1613 #endif 1614 1615 /* for node-related operations */ 1616 struct f2fs_nm_info *nm_info; /* node manager */ 1617 struct inode *node_inode; /* cache node blocks */ 1618 1619 /* for segment-related operations */ 1620 struct f2fs_sm_info *sm_info; /* segment manager */ 1621 1622 /* for bio operations */ 1623 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1624 /* keep migration IO order for LFS mode */ 1625 struct f2fs_rwsem io_order_lock; 1626 mempool_t *write_io_dummy; /* Dummy pages */ 1627 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1628 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1629 1630 /* for checkpoint */ 1631 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1632 int cur_cp_pack; /* remain current cp pack */ 1633 spinlock_t cp_lock; /* for flag in ckpt */ 1634 struct inode *meta_inode; /* cache meta blocks */ 1635 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1636 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1637 struct f2fs_rwsem node_write; /* locking node writes */ 1638 struct f2fs_rwsem node_change; /* locking node change */ 1639 wait_queue_head_t cp_wait; 1640 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1641 long interval_time[MAX_TIME]; /* to store thresholds */ 1642 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1643 1644 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1645 1646 spinlock_t fsync_node_lock; /* for node entry lock */ 1647 struct list_head fsync_node_list; /* node list head */ 1648 unsigned int fsync_seg_id; /* sequence id */ 1649 unsigned int fsync_node_num; /* number of node entries */ 1650 1651 /* for orphan inode, use 0'th array */ 1652 unsigned int max_orphans; /* max orphan inodes */ 1653 1654 /* for inode management */ 1655 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1656 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1657 struct mutex flush_lock; /* for flush exclusion */ 1658 1659 /* for extent tree cache */ 1660 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1661 struct mutex extent_tree_lock; /* locking extent radix tree */ 1662 struct list_head extent_list; /* lru list for shrinker */ 1663 spinlock_t extent_lock; /* locking extent lru list */ 1664 atomic_t total_ext_tree; /* extent tree count */ 1665 struct list_head zombie_list; /* extent zombie tree list */ 1666 atomic_t total_zombie_tree; /* extent zombie tree count */ 1667 atomic_t total_ext_node; /* extent info count */ 1668 1669 /* basic filesystem units */ 1670 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1671 unsigned int log_blocksize; /* log2 block size */ 1672 unsigned int blocksize; /* block size */ 1673 unsigned int root_ino_num; /* root inode number*/ 1674 unsigned int node_ino_num; /* node inode number*/ 1675 unsigned int meta_ino_num; /* meta inode number*/ 1676 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1677 unsigned int blocks_per_seg; /* blocks per segment */ 1678 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1679 unsigned int segs_per_sec; /* segments per section */ 1680 unsigned int secs_per_zone; /* sections per zone */ 1681 unsigned int total_sections; /* total section count */ 1682 unsigned int total_node_count; /* total node block count */ 1683 unsigned int total_valid_node_count; /* valid node block count */ 1684 int dir_level; /* directory level */ 1685 int readdir_ra; /* readahead inode in readdir */ 1686 u64 max_io_bytes; /* max io bytes to merge IOs */ 1687 1688 block_t user_block_count; /* # of user blocks */ 1689 block_t total_valid_block_count; /* # of valid blocks */ 1690 block_t discard_blks; /* discard command candidats */ 1691 block_t last_valid_block_count; /* for recovery */ 1692 block_t reserved_blocks; /* configurable reserved blocks */ 1693 block_t current_reserved_blocks; /* current reserved blocks */ 1694 1695 /* Additional tracking for no checkpoint mode */ 1696 block_t unusable_block_count; /* # of blocks saved by last cp */ 1697 1698 unsigned int nquota_files; /* # of quota sysfile */ 1699 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1700 1701 /* # of pages, see count_type */ 1702 atomic_t nr_pages[NR_COUNT_TYPE]; 1703 /* # of allocated blocks */ 1704 struct percpu_counter alloc_valid_block_count; 1705 /* # of node block writes as roll forward recovery */ 1706 struct percpu_counter rf_node_block_count; 1707 1708 /* writeback control */ 1709 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1710 1711 /* valid inode count */ 1712 struct percpu_counter total_valid_inode_count; 1713 1714 struct f2fs_mount_info mount_opt; /* mount options */ 1715 1716 /* for cleaning operations */ 1717 struct f2fs_rwsem gc_lock; /* 1718 * semaphore for GC, avoid 1719 * race between GC and GC or CP 1720 */ 1721 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1722 struct atgc_management am; /* atgc management */ 1723 unsigned int cur_victim_sec; /* current victim section num */ 1724 unsigned int gc_mode; /* current GC state */ 1725 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1726 spinlock_t gc_urgent_high_lock; 1727 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1728 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1729 1730 /* for skip statistic */ 1731 unsigned int atomic_files; /* # of opened atomic file */ 1732 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1733 1734 /* threshold for gc trials on pinned files */ 1735 u64 gc_pin_file_threshold; 1736 struct f2fs_rwsem pin_sem; 1737 1738 /* maximum # of trials to find a victim segment for SSR and GC */ 1739 unsigned int max_victim_search; 1740 /* migration granularity of garbage collection, unit: segment */ 1741 unsigned int migration_granularity; 1742 1743 /* 1744 * for stat information. 1745 * one is for the LFS mode, and the other is for the SSR mode. 1746 */ 1747 #ifdef CONFIG_F2FS_STAT_FS 1748 struct f2fs_stat_info *stat_info; /* FS status information */ 1749 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1750 unsigned int segment_count[2]; /* # of allocated segments */ 1751 unsigned int block_count[2]; /* # of allocated blocks */ 1752 atomic_t inplace_count; /* # of inplace update */ 1753 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1754 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1755 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1756 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1757 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1758 atomic_t inline_inode; /* # of inline_data inodes */ 1759 atomic_t inline_dir; /* # of inline_dentry inodes */ 1760 atomic_t compr_inode; /* # of compressed inodes */ 1761 atomic64_t compr_blocks; /* # of compressed blocks */ 1762 atomic_t max_aw_cnt; /* max # of atomic writes */ 1763 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1764 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1765 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1766 #endif 1767 spinlock_t stat_lock; /* lock for stat operations */ 1768 1769 /* to attach REQ_META|REQ_FUA flags */ 1770 unsigned int data_io_flag; 1771 unsigned int node_io_flag; 1772 1773 /* For sysfs support */ 1774 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1775 struct completion s_kobj_unregister; 1776 1777 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1778 struct completion s_stat_kobj_unregister; 1779 1780 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1781 struct completion s_feature_list_kobj_unregister; 1782 1783 /* For shrinker support */ 1784 struct list_head s_list; 1785 struct mutex umount_mutex; 1786 unsigned int shrinker_run_no; 1787 1788 /* For multi devices */ 1789 int s_ndevs; /* number of devices */ 1790 struct f2fs_dev_info *devs; /* for device list */ 1791 unsigned int dirty_device; /* for checkpoint data flush */ 1792 spinlock_t dev_lock; /* protect dirty_device */ 1793 bool aligned_blksize; /* all devices has the same logical blksize */ 1794 1795 /* For write statistics */ 1796 u64 sectors_written_start; 1797 u64 kbytes_written; 1798 1799 /* Reference to checksum algorithm driver via cryptoapi */ 1800 struct crypto_shash *s_chksum_driver; 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 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1808 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1809 1810 /* For reclaimed segs statistics per each GC mode */ 1811 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1812 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1813 1814 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1815 1816 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1817 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1818 1819 /* For atomic write statistics */ 1820 atomic64_t current_atomic_write; 1821 s64 peak_atomic_write; 1822 u64 committed_atomic_block; 1823 u64 revoked_atomic_block; 1824 1825 #ifdef CONFIG_F2FS_FS_COMPRESSION 1826 struct kmem_cache *page_array_slab; /* page array entry */ 1827 unsigned int page_array_slab_size; /* default page array slab size */ 1828 1829 /* For runtime compression statistics */ 1830 u64 compr_written_block; 1831 u64 compr_saved_block; 1832 u32 compr_new_inode; 1833 1834 /* For compressed block cache */ 1835 struct inode *compress_inode; /* cache compressed blocks */ 1836 unsigned int compress_percent; /* cache page percentage */ 1837 unsigned int compress_watermark; /* cache page watermark */ 1838 atomic_t compress_page_hit; /* cache hit count */ 1839 #endif 1840 1841 #ifdef CONFIG_F2FS_IOSTAT 1842 /* For app/fs IO statistics */ 1843 spinlock_t iostat_lock; 1844 unsigned long long rw_iostat[NR_IO_TYPE]; 1845 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1846 bool iostat_enable; 1847 unsigned long iostat_next_period; 1848 unsigned int iostat_period_ms; 1849 1850 /* For io latency related statistics info in one iostat period */ 1851 spinlock_t iostat_lat_lock; 1852 struct iostat_lat_info *iostat_io_lat; 1853 #endif 1854 }; 1855 1856 #ifdef CONFIG_F2FS_FAULT_INJECTION 1857 #define f2fs_show_injection_info(sbi, type) \ 1858 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1859 KERN_INFO, sbi->sb->s_id, \ 1860 f2fs_fault_name[type], \ 1861 __func__, __builtin_return_address(0)) 1862 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1863 { 1864 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1865 1866 if (!ffi->inject_rate) 1867 return false; 1868 1869 if (!IS_FAULT_SET(ffi, type)) 1870 return false; 1871 1872 atomic_inc(&ffi->inject_ops); 1873 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1874 atomic_set(&ffi->inject_ops, 0); 1875 return true; 1876 } 1877 return false; 1878 } 1879 #else 1880 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1881 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1882 { 1883 return false; 1884 } 1885 #endif 1886 1887 /* 1888 * Test if the mounted volume is a multi-device volume. 1889 * - For a single regular disk volume, sbi->s_ndevs is 0. 1890 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1891 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1892 */ 1893 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1894 { 1895 return sbi->s_ndevs > 1; 1896 } 1897 1898 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1899 { 1900 unsigned long now = jiffies; 1901 1902 sbi->last_time[type] = now; 1903 1904 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1905 if (type == REQ_TIME) { 1906 sbi->last_time[DISCARD_TIME] = now; 1907 sbi->last_time[GC_TIME] = now; 1908 } 1909 } 1910 1911 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1912 { 1913 unsigned long interval = sbi->interval_time[type] * HZ; 1914 1915 return time_after(jiffies, sbi->last_time[type] + interval); 1916 } 1917 1918 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1919 int type) 1920 { 1921 unsigned long interval = sbi->interval_time[type] * HZ; 1922 unsigned int wait_ms = 0; 1923 long delta; 1924 1925 delta = (sbi->last_time[type] + interval) - jiffies; 1926 if (delta > 0) 1927 wait_ms = jiffies_to_msecs(delta); 1928 1929 return wait_ms; 1930 } 1931 1932 /* 1933 * Inline functions 1934 */ 1935 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1936 const void *address, unsigned int length) 1937 { 1938 struct { 1939 struct shash_desc shash; 1940 char ctx[4]; 1941 } desc; 1942 int err; 1943 1944 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1945 1946 desc.shash.tfm = sbi->s_chksum_driver; 1947 *(u32 *)desc.ctx = crc; 1948 1949 err = crypto_shash_update(&desc.shash, address, length); 1950 BUG_ON(err); 1951 1952 return *(u32 *)desc.ctx; 1953 } 1954 1955 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1956 unsigned int length) 1957 { 1958 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1959 } 1960 1961 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1962 void *buf, size_t buf_size) 1963 { 1964 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1965 } 1966 1967 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1968 const void *address, unsigned int length) 1969 { 1970 return __f2fs_crc32(sbi, crc, address, length); 1971 } 1972 1973 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1974 { 1975 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1976 } 1977 1978 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1979 { 1980 return sb->s_fs_info; 1981 } 1982 1983 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1984 { 1985 return F2FS_SB(inode->i_sb); 1986 } 1987 1988 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1989 { 1990 return F2FS_I_SB(mapping->host); 1991 } 1992 1993 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1994 { 1995 return F2FS_M_SB(page_file_mapping(page)); 1996 } 1997 1998 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1999 { 2000 return (struct f2fs_super_block *)(sbi->raw_super); 2001 } 2002 2003 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2004 { 2005 return (struct f2fs_checkpoint *)(sbi->ckpt); 2006 } 2007 2008 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2009 { 2010 return (struct f2fs_node *)page_address(page); 2011 } 2012 2013 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2014 { 2015 return &((struct f2fs_node *)page_address(page))->i; 2016 } 2017 2018 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2019 { 2020 return (struct f2fs_nm_info *)(sbi->nm_info); 2021 } 2022 2023 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2024 { 2025 return (struct f2fs_sm_info *)(sbi->sm_info); 2026 } 2027 2028 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2029 { 2030 return (struct sit_info *)(SM_I(sbi)->sit_info); 2031 } 2032 2033 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2034 { 2035 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2036 } 2037 2038 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2039 { 2040 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2041 } 2042 2043 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2044 { 2045 return sbi->meta_inode->i_mapping; 2046 } 2047 2048 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2049 { 2050 return sbi->node_inode->i_mapping; 2051 } 2052 2053 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2054 { 2055 return test_bit(type, &sbi->s_flag); 2056 } 2057 2058 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2059 { 2060 set_bit(type, &sbi->s_flag); 2061 } 2062 2063 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2064 { 2065 clear_bit(type, &sbi->s_flag); 2066 } 2067 2068 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2069 { 2070 return le64_to_cpu(cp->checkpoint_ver); 2071 } 2072 2073 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2074 { 2075 if (type < F2FS_MAX_QUOTAS) 2076 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2077 return 0; 2078 } 2079 2080 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2081 { 2082 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2083 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2084 } 2085 2086 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2087 { 2088 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2089 2090 return ckpt_flags & f; 2091 } 2092 2093 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2094 { 2095 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2096 } 2097 2098 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2099 { 2100 unsigned int ckpt_flags; 2101 2102 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2103 ckpt_flags |= f; 2104 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2105 } 2106 2107 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2108 { 2109 unsigned long flags; 2110 2111 spin_lock_irqsave(&sbi->cp_lock, flags); 2112 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2113 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2114 } 2115 2116 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2117 { 2118 unsigned int ckpt_flags; 2119 2120 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2121 ckpt_flags &= (~f); 2122 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2123 } 2124 2125 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2126 { 2127 unsigned long flags; 2128 2129 spin_lock_irqsave(&sbi->cp_lock, flags); 2130 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2131 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2132 } 2133 2134 #define init_f2fs_rwsem(sem) \ 2135 do { \ 2136 static struct lock_class_key __key; \ 2137 \ 2138 __init_f2fs_rwsem((sem), #sem, &__key); \ 2139 } while (0) 2140 2141 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2142 const char *sem_name, struct lock_class_key *key) 2143 { 2144 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2145 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2146 init_waitqueue_head(&sem->read_waiters); 2147 #endif 2148 } 2149 2150 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2151 { 2152 return rwsem_is_locked(&sem->internal_rwsem); 2153 } 2154 2155 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2156 { 2157 return rwsem_is_contended(&sem->internal_rwsem); 2158 } 2159 2160 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2161 { 2162 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2163 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2164 #else 2165 down_read(&sem->internal_rwsem); 2166 #endif 2167 } 2168 2169 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2170 { 2171 return down_read_trylock(&sem->internal_rwsem); 2172 } 2173 2174 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2175 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2176 { 2177 down_read_nested(&sem->internal_rwsem, subclass); 2178 } 2179 #else 2180 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2181 #endif 2182 2183 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2184 { 2185 up_read(&sem->internal_rwsem); 2186 } 2187 2188 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2189 { 2190 down_write(&sem->internal_rwsem); 2191 } 2192 2193 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2194 { 2195 return down_write_trylock(&sem->internal_rwsem); 2196 } 2197 2198 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2199 { 2200 up_write(&sem->internal_rwsem); 2201 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2202 wake_up_all(&sem->read_waiters); 2203 #endif 2204 } 2205 2206 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2207 { 2208 f2fs_down_read(&sbi->cp_rwsem); 2209 } 2210 2211 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2212 { 2213 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2214 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2215 return 0; 2216 } 2217 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2218 } 2219 2220 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2221 { 2222 f2fs_up_read(&sbi->cp_rwsem); 2223 } 2224 2225 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2226 { 2227 f2fs_down_write(&sbi->cp_rwsem); 2228 } 2229 2230 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2231 { 2232 f2fs_up_write(&sbi->cp_rwsem); 2233 } 2234 2235 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2236 { 2237 int reason = CP_SYNC; 2238 2239 if (test_opt(sbi, FASTBOOT)) 2240 reason = CP_FASTBOOT; 2241 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2242 reason = CP_UMOUNT; 2243 return reason; 2244 } 2245 2246 static inline bool __remain_node_summaries(int reason) 2247 { 2248 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2249 } 2250 2251 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2252 { 2253 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2254 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2255 } 2256 2257 /* 2258 * Check whether the inode has blocks or not 2259 */ 2260 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2261 { 2262 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2263 2264 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2265 } 2266 2267 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2268 { 2269 return ofs == XATTR_NODE_OFFSET; 2270 } 2271 2272 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2273 struct inode *inode, bool cap) 2274 { 2275 if (!inode) 2276 return true; 2277 if (!test_opt(sbi, RESERVE_ROOT)) 2278 return false; 2279 if (IS_NOQUOTA(inode)) 2280 return true; 2281 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2282 return true; 2283 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2284 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2285 return true; 2286 if (cap && capable(CAP_SYS_RESOURCE)) 2287 return true; 2288 return false; 2289 } 2290 2291 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2292 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2293 struct inode *inode, blkcnt_t *count) 2294 { 2295 blkcnt_t diff = 0, release = 0; 2296 block_t avail_user_block_count; 2297 int ret; 2298 2299 ret = dquot_reserve_block(inode, *count); 2300 if (ret) 2301 return ret; 2302 2303 if (time_to_inject(sbi, FAULT_BLOCK)) { 2304 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2305 release = *count; 2306 goto release_quota; 2307 } 2308 2309 /* 2310 * let's increase this in prior to actual block count change in order 2311 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2312 */ 2313 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2314 2315 spin_lock(&sbi->stat_lock); 2316 sbi->total_valid_block_count += (block_t)(*count); 2317 avail_user_block_count = sbi->user_block_count - 2318 sbi->current_reserved_blocks; 2319 2320 if (!__allow_reserved_blocks(sbi, inode, true)) 2321 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2322 2323 if (F2FS_IO_ALIGNED(sbi)) 2324 avail_user_block_count -= sbi->blocks_per_seg * 2325 SM_I(sbi)->additional_reserved_segments; 2326 2327 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2328 if (avail_user_block_count > sbi->unusable_block_count) 2329 avail_user_block_count -= sbi->unusable_block_count; 2330 else 2331 avail_user_block_count = 0; 2332 } 2333 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2334 diff = sbi->total_valid_block_count - avail_user_block_count; 2335 if (diff > *count) 2336 diff = *count; 2337 *count -= diff; 2338 release = diff; 2339 sbi->total_valid_block_count -= diff; 2340 if (!*count) { 2341 spin_unlock(&sbi->stat_lock); 2342 goto enospc; 2343 } 2344 } 2345 spin_unlock(&sbi->stat_lock); 2346 2347 if (unlikely(release)) { 2348 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2349 dquot_release_reservation_block(inode, release); 2350 } 2351 f2fs_i_blocks_write(inode, *count, true, true); 2352 return 0; 2353 2354 enospc: 2355 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2356 release_quota: 2357 dquot_release_reservation_block(inode, release); 2358 return -ENOSPC; 2359 } 2360 2361 __printf(2, 3) 2362 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2363 2364 #define f2fs_err(sbi, fmt, ...) \ 2365 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2366 #define f2fs_warn(sbi, fmt, ...) \ 2367 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2368 #define f2fs_notice(sbi, fmt, ...) \ 2369 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2370 #define f2fs_info(sbi, fmt, ...) \ 2371 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2372 #define f2fs_debug(sbi, fmt, ...) \ 2373 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2374 2375 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2376 struct inode *inode, 2377 block_t count) 2378 { 2379 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2380 2381 spin_lock(&sbi->stat_lock); 2382 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2383 sbi->total_valid_block_count -= (block_t)count; 2384 if (sbi->reserved_blocks && 2385 sbi->current_reserved_blocks < sbi->reserved_blocks) 2386 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2387 sbi->current_reserved_blocks + count); 2388 spin_unlock(&sbi->stat_lock); 2389 if (unlikely(inode->i_blocks < sectors)) { 2390 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2391 inode->i_ino, 2392 (unsigned long long)inode->i_blocks, 2393 (unsigned long long)sectors); 2394 set_sbi_flag(sbi, SBI_NEED_FSCK); 2395 return; 2396 } 2397 f2fs_i_blocks_write(inode, count, false, true); 2398 } 2399 2400 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2401 { 2402 atomic_inc(&sbi->nr_pages[count_type]); 2403 2404 if (count_type == F2FS_DIRTY_DENTS || 2405 count_type == F2FS_DIRTY_NODES || 2406 count_type == F2FS_DIRTY_META || 2407 count_type == F2FS_DIRTY_QDATA || 2408 count_type == F2FS_DIRTY_IMETA) 2409 set_sbi_flag(sbi, SBI_IS_DIRTY); 2410 } 2411 2412 static inline void inode_inc_dirty_pages(struct inode *inode) 2413 { 2414 atomic_inc(&F2FS_I(inode)->dirty_pages); 2415 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2416 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2417 if (IS_NOQUOTA(inode)) 2418 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2419 } 2420 2421 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2422 { 2423 atomic_dec(&sbi->nr_pages[count_type]); 2424 } 2425 2426 static inline void inode_dec_dirty_pages(struct inode *inode) 2427 { 2428 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2429 !S_ISLNK(inode->i_mode)) 2430 return; 2431 2432 atomic_dec(&F2FS_I(inode)->dirty_pages); 2433 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2434 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2435 if (IS_NOQUOTA(inode)) 2436 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2437 } 2438 2439 static inline void inc_atomic_write_cnt(struct inode *inode) 2440 { 2441 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2442 struct f2fs_inode_info *fi = F2FS_I(inode); 2443 u64 current_write; 2444 2445 fi->atomic_write_cnt++; 2446 atomic64_inc(&sbi->current_atomic_write); 2447 current_write = atomic64_read(&sbi->current_atomic_write); 2448 if (current_write > sbi->peak_atomic_write) 2449 sbi->peak_atomic_write = current_write; 2450 } 2451 2452 static inline void release_atomic_write_cnt(struct inode *inode) 2453 { 2454 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2455 struct f2fs_inode_info *fi = F2FS_I(inode); 2456 2457 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2458 fi->atomic_write_cnt = 0; 2459 } 2460 2461 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2462 { 2463 return atomic_read(&sbi->nr_pages[count_type]); 2464 } 2465 2466 static inline int get_dirty_pages(struct inode *inode) 2467 { 2468 return atomic_read(&F2FS_I(inode)->dirty_pages); 2469 } 2470 2471 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2472 { 2473 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2474 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2475 sbi->log_blocks_per_seg; 2476 2477 return segs / sbi->segs_per_sec; 2478 } 2479 2480 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2481 { 2482 return sbi->total_valid_block_count; 2483 } 2484 2485 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2486 { 2487 return sbi->discard_blks; 2488 } 2489 2490 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2491 { 2492 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2493 2494 /* return NAT or SIT bitmap */ 2495 if (flag == NAT_BITMAP) 2496 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2497 else if (flag == SIT_BITMAP) 2498 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2499 2500 return 0; 2501 } 2502 2503 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2504 { 2505 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2506 } 2507 2508 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2509 { 2510 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2511 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2512 int offset; 2513 2514 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2515 offset = (flag == SIT_BITMAP) ? 2516 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2517 /* 2518 * if large_nat_bitmap feature is enabled, leave checksum 2519 * protection for all nat/sit bitmaps. 2520 */ 2521 return tmp_ptr + offset + sizeof(__le32); 2522 } 2523 2524 if (__cp_payload(sbi) > 0) { 2525 if (flag == NAT_BITMAP) 2526 return &ckpt->sit_nat_version_bitmap; 2527 else 2528 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2529 } else { 2530 offset = (flag == NAT_BITMAP) ? 2531 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2532 return tmp_ptr + offset; 2533 } 2534 } 2535 2536 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2537 { 2538 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2539 2540 if (sbi->cur_cp_pack == 2) 2541 start_addr += sbi->blocks_per_seg; 2542 return start_addr; 2543 } 2544 2545 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2546 { 2547 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2548 2549 if (sbi->cur_cp_pack == 1) 2550 start_addr += sbi->blocks_per_seg; 2551 return start_addr; 2552 } 2553 2554 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2555 { 2556 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2557 } 2558 2559 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2560 { 2561 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2562 } 2563 2564 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2565 struct inode *inode, bool is_inode) 2566 { 2567 block_t valid_block_count; 2568 unsigned int valid_node_count, user_block_count; 2569 int err; 2570 2571 if (is_inode) { 2572 if (inode) { 2573 err = dquot_alloc_inode(inode); 2574 if (err) 2575 return err; 2576 } 2577 } else { 2578 err = dquot_reserve_block(inode, 1); 2579 if (err) 2580 return err; 2581 } 2582 2583 if (time_to_inject(sbi, FAULT_BLOCK)) { 2584 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2585 goto enospc; 2586 } 2587 2588 spin_lock(&sbi->stat_lock); 2589 2590 valid_block_count = sbi->total_valid_block_count + 2591 sbi->current_reserved_blocks + 1; 2592 2593 if (!__allow_reserved_blocks(sbi, inode, false)) 2594 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2595 2596 if (F2FS_IO_ALIGNED(sbi)) 2597 valid_block_count += sbi->blocks_per_seg * 2598 SM_I(sbi)->additional_reserved_segments; 2599 2600 user_block_count = sbi->user_block_count; 2601 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2602 user_block_count -= sbi->unusable_block_count; 2603 2604 if (unlikely(valid_block_count > user_block_count)) { 2605 spin_unlock(&sbi->stat_lock); 2606 goto enospc; 2607 } 2608 2609 valid_node_count = sbi->total_valid_node_count + 1; 2610 if (unlikely(valid_node_count > sbi->total_node_count)) { 2611 spin_unlock(&sbi->stat_lock); 2612 goto enospc; 2613 } 2614 2615 sbi->total_valid_node_count++; 2616 sbi->total_valid_block_count++; 2617 spin_unlock(&sbi->stat_lock); 2618 2619 if (inode) { 2620 if (is_inode) 2621 f2fs_mark_inode_dirty_sync(inode, true); 2622 else 2623 f2fs_i_blocks_write(inode, 1, true, true); 2624 } 2625 2626 percpu_counter_inc(&sbi->alloc_valid_block_count); 2627 return 0; 2628 2629 enospc: 2630 if (is_inode) { 2631 if (inode) 2632 dquot_free_inode(inode); 2633 } else { 2634 dquot_release_reservation_block(inode, 1); 2635 } 2636 return -ENOSPC; 2637 } 2638 2639 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2640 struct inode *inode, bool is_inode) 2641 { 2642 spin_lock(&sbi->stat_lock); 2643 2644 if (unlikely(!sbi->total_valid_block_count || 2645 !sbi->total_valid_node_count)) { 2646 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2647 sbi->total_valid_block_count, 2648 sbi->total_valid_node_count); 2649 set_sbi_flag(sbi, SBI_NEED_FSCK); 2650 } else { 2651 sbi->total_valid_block_count--; 2652 sbi->total_valid_node_count--; 2653 } 2654 2655 if (sbi->reserved_blocks && 2656 sbi->current_reserved_blocks < sbi->reserved_blocks) 2657 sbi->current_reserved_blocks++; 2658 2659 spin_unlock(&sbi->stat_lock); 2660 2661 if (is_inode) { 2662 dquot_free_inode(inode); 2663 } else { 2664 if (unlikely(inode->i_blocks == 0)) { 2665 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2666 inode->i_ino, 2667 (unsigned long long)inode->i_blocks); 2668 set_sbi_flag(sbi, SBI_NEED_FSCK); 2669 return; 2670 } 2671 f2fs_i_blocks_write(inode, 1, false, true); 2672 } 2673 } 2674 2675 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2676 { 2677 return sbi->total_valid_node_count; 2678 } 2679 2680 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2681 { 2682 percpu_counter_inc(&sbi->total_valid_inode_count); 2683 } 2684 2685 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2686 { 2687 percpu_counter_dec(&sbi->total_valid_inode_count); 2688 } 2689 2690 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2691 { 2692 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2693 } 2694 2695 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2696 pgoff_t index, bool for_write) 2697 { 2698 struct page *page; 2699 unsigned int flags; 2700 2701 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2702 if (!for_write) 2703 page = find_get_page_flags(mapping, index, 2704 FGP_LOCK | FGP_ACCESSED); 2705 else 2706 page = find_lock_page(mapping, index); 2707 if (page) 2708 return page; 2709 2710 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2711 f2fs_show_injection_info(F2FS_M_SB(mapping), 2712 FAULT_PAGE_ALLOC); 2713 return NULL; 2714 } 2715 } 2716 2717 if (!for_write) 2718 return grab_cache_page(mapping, index); 2719 2720 flags = memalloc_nofs_save(); 2721 page = grab_cache_page_write_begin(mapping, index); 2722 memalloc_nofs_restore(flags); 2723 2724 return page; 2725 } 2726 2727 static inline struct page *f2fs_pagecache_get_page( 2728 struct address_space *mapping, pgoff_t index, 2729 int fgp_flags, gfp_t gfp_mask) 2730 { 2731 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2732 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2733 return NULL; 2734 } 2735 2736 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2737 } 2738 2739 static inline void f2fs_put_page(struct page *page, int unlock) 2740 { 2741 if (!page) 2742 return; 2743 2744 if (unlock) { 2745 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2746 unlock_page(page); 2747 } 2748 put_page(page); 2749 } 2750 2751 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2752 { 2753 if (dn->node_page) 2754 f2fs_put_page(dn->node_page, 1); 2755 if (dn->inode_page && dn->node_page != dn->inode_page) 2756 f2fs_put_page(dn->inode_page, 0); 2757 dn->node_page = NULL; 2758 dn->inode_page = NULL; 2759 } 2760 2761 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2762 size_t size) 2763 { 2764 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2765 } 2766 2767 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2768 gfp_t flags) 2769 { 2770 void *entry; 2771 2772 entry = kmem_cache_alloc(cachep, flags); 2773 if (!entry) 2774 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2775 return entry; 2776 } 2777 2778 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2779 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2780 { 2781 if (nofail) 2782 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2783 2784 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2785 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2786 return NULL; 2787 } 2788 2789 return kmem_cache_alloc(cachep, flags); 2790 } 2791 2792 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2793 { 2794 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2795 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2796 get_pages(sbi, F2FS_WB_CP_DATA) || 2797 get_pages(sbi, F2FS_DIO_READ) || 2798 get_pages(sbi, F2FS_DIO_WRITE)) 2799 return true; 2800 2801 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2802 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2803 return true; 2804 2805 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2806 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2807 return true; 2808 return false; 2809 } 2810 2811 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2812 { 2813 if (sbi->gc_mode == GC_URGENT_HIGH) 2814 return true; 2815 2816 if (is_inflight_io(sbi, type)) 2817 return false; 2818 2819 if (sbi->gc_mode == GC_URGENT_MID) 2820 return true; 2821 2822 if (sbi->gc_mode == GC_URGENT_LOW && 2823 (type == DISCARD_TIME || type == GC_TIME)) 2824 return true; 2825 2826 return f2fs_time_over(sbi, type); 2827 } 2828 2829 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2830 unsigned long index, void *item) 2831 { 2832 while (radix_tree_insert(root, index, item)) 2833 cond_resched(); 2834 } 2835 2836 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2837 2838 static inline bool IS_INODE(struct page *page) 2839 { 2840 struct f2fs_node *p = F2FS_NODE(page); 2841 2842 return RAW_IS_INODE(p); 2843 } 2844 2845 static inline int offset_in_addr(struct f2fs_inode *i) 2846 { 2847 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2848 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2849 } 2850 2851 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2852 { 2853 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2854 } 2855 2856 static inline int f2fs_has_extra_attr(struct inode *inode); 2857 static inline block_t data_blkaddr(struct inode *inode, 2858 struct page *node_page, unsigned int offset) 2859 { 2860 struct f2fs_node *raw_node; 2861 __le32 *addr_array; 2862 int base = 0; 2863 bool is_inode = IS_INODE(node_page); 2864 2865 raw_node = F2FS_NODE(node_page); 2866 2867 if (is_inode) { 2868 if (!inode) 2869 /* from GC path only */ 2870 base = offset_in_addr(&raw_node->i); 2871 else if (f2fs_has_extra_attr(inode)) 2872 base = get_extra_isize(inode); 2873 } 2874 2875 addr_array = blkaddr_in_node(raw_node); 2876 return le32_to_cpu(addr_array[base + offset]); 2877 } 2878 2879 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2880 { 2881 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2882 } 2883 2884 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2885 { 2886 int mask; 2887 2888 addr += (nr >> 3); 2889 mask = 1 << (7 - (nr & 0x07)); 2890 return mask & *addr; 2891 } 2892 2893 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2894 { 2895 int mask; 2896 2897 addr += (nr >> 3); 2898 mask = 1 << (7 - (nr & 0x07)); 2899 *addr |= mask; 2900 } 2901 2902 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2903 { 2904 int mask; 2905 2906 addr += (nr >> 3); 2907 mask = 1 << (7 - (nr & 0x07)); 2908 *addr &= ~mask; 2909 } 2910 2911 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2912 { 2913 int mask; 2914 int ret; 2915 2916 addr += (nr >> 3); 2917 mask = 1 << (7 - (nr & 0x07)); 2918 ret = mask & *addr; 2919 *addr |= mask; 2920 return ret; 2921 } 2922 2923 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2924 { 2925 int mask; 2926 int ret; 2927 2928 addr += (nr >> 3); 2929 mask = 1 << (7 - (nr & 0x07)); 2930 ret = mask & *addr; 2931 *addr &= ~mask; 2932 return ret; 2933 } 2934 2935 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2936 { 2937 int mask; 2938 2939 addr += (nr >> 3); 2940 mask = 1 << (7 - (nr & 0x07)); 2941 *addr ^= mask; 2942 } 2943 2944 /* 2945 * On-disk inode flags (f2fs_inode::i_flags) 2946 */ 2947 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2948 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2949 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2950 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2951 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2952 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2953 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2954 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2955 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2956 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2957 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2958 2959 /* Flags that should be inherited by new inodes from their parent. */ 2960 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2961 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2962 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2963 2964 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2965 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2966 F2FS_CASEFOLD_FL)) 2967 2968 /* Flags that are appropriate for non-directories/regular files. */ 2969 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2970 2971 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2972 { 2973 if (S_ISDIR(mode)) 2974 return flags; 2975 else if (S_ISREG(mode)) 2976 return flags & F2FS_REG_FLMASK; 2977 else 2978 return flags & F2FS_OTHER_FLMASK; 2979 } 2980 2981 static inline void __mark_inode_dirty_flag(struct inode *inode, 2982 int flag, bool set) 2983 { 2984 switch (flag) { 2985 case FI_INLINE_XATTR: 2986 case FI_INLINE_DATA: 2987 case FI_INLINE_DENTRY: 2988 case FI_NEW_INODE: 2989 if (set) 2990 return; 2991 fallthrough; 2992 case FI_DATA_EXIST: 2993 case FI_INLINE_DOTS: 2994 case FI_PIN_FILE: 2995 case FI_COMPRESS_RELEASED: 2996 f2fs_mark_inode_dirty_sync(inode, true); 2997 } 2998 } 2999 3000 static inline void set_inode_flag(struct inode *inode, int flag) 3001 { 3002 set_bit(flag, F2FS_I(inode)->flags); 3003 __mark_inode_dirty_flag(inode, flag, true); 3004 } 3005 3006 static inline int is_inode_flag_set(struct inode *inode, int flag) 3007 { 3008 return test_bit(flag, F2FS_I(inode)->flags); 3009 } 3010 3011 static inline void clear_inode_flag(struct inode *inode, int flag) 3012 { 3013 clear_bit(flag, F2FS_I(inode)->flags); 3014 __mark_inode_dirty_flag(inode, flag, false); 3015 } 3016 3017 static inline bool f2fs_verity_in_progress(struct inode *inode) 3018 { 3019 return IS_ENABLED(CONFIG_FS_VERITY) && 3020 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3021 } 3022 3023 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3024 { 3025 F2FS_I(inode)->i_acl_mode = mode; 3026 set_inode_flag(inode, FI_ACL_MODE); 3027 f2fs_mark_inode_dirty_sync(inode, false); 3028 } 3029 3030 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3031 { 3032 if (inc) 3033 inc_nlink(inode); 3034 else 3035 drop_nlink(inode); 3036 f2fs_mark_inode_dirty_sync(inode, true); 3037 } 3038 3039 static inline void f2fs_i_blocks_write(struct inode *inode, 3040 block_t diff, bool add, bool claim) 3041 { 3042 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3043 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3044 3045 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3046 if (add) { 3047 if (claim) 3048 dquot_claim_block(inode, diff); 3049 else 3050 dquot_alloc_block_nofail(inode, diff); 3051 } else { 3052 dquot_free_block(inode, diff); 3053 } 3054 3055 f2fs_mark_inode_dirty_sync(inode, true); 3056 if (clean || recover) 3057 set_inode_flag(inode, FI_AUTO_RECOVER); 3058 } 3059 3060 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3061 { 3062 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3063 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3064 3065 if (i_size_read(inode) == i_size) 3066 return; 3067 3068 i_size_write(inode, i_size); 3069 f2fs_mark_inode_dirty_sync(inode, true); 3070 if (clean || recover) 3071 set_inode_flag(inode, FI_AUTO_RECOVER); 3072 } 3073 3074 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3075 { 3076 F2FS_I(inode)->i_current_depth = depth; 3077 f2fs_mark_inode_dirty_sync(inode, true); 3078 } 3079 3080 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3081 unsigned int count) 3082 { 3083 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3084 f2fs_mark_inode_dirty_sync(inode, true); 3085 } 3086 3087 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3088 { 3089 F2FS_I(inode)->i_xattr_nid = xnid; 3090 f2fs_mark_inode_dirty_sync(inode, true); 3091 } 3092 3093 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3094 { 3095 F2FS_I(inode)->i_pino = pino; 3096 f2fs_mark_inode_dirty_sync(inode, true); 3097 } 3098 3099 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3100 { 3101 struct f2fs_inode_info *fi = F2FS_I(inode); 3102 3103 if (ri->i_inline & F2FS_INLINE_XATTR) 3104 set_bit(FI_INLINE_XATTR, fi->flags); 3105 if (ri->i_inline & F2FS_INLINE_DATA) 3106 set_bit(FI_INLINE_DATA, fi->flags); 3107 if (ri->i_inline & F2FS_INLINE_DENTRY) 3108 set_bit(FI_INLINE_DENTRY, fi->flags); 3109 if (ri->i_inline & F2FS_DATA_EXIST) 3110 set_bit(FI_DATA_EXIST, fi->flags); 3111 if (ri->i_inline & F2FS_INLINE_DOTS) 3112 set_bit(FI_INLINE_DOTS, fi->flags); 3113 if (ri->i_inline & F2FS_EXTRA_ATTR) 3114 set_bit(FI_EXTRA_ATTR, fi->flags); 3115 if (ri->i_inline & F2FS_PIN_FILE) 3116 set_bit(FI_PIN_FILE, fi->flags); 3117 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3118 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3119 } 3120 3121 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3122 { 3123 ri->i_inline = 0; 3124 3125 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3126 ri->i_inline |= F2FS_INLINE_XATTR; 3127 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3128 ri->i_inline |= F2FS_INLINE_DATA; 3129 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3130 ri->i_inline |= F2FS_INLINE_DENTRY; 3131 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3132 ri->i_inline |= F2FS_DATA_EXIST; 3133 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3134 ri->i_inline |= F2FS_INLINE_DOTS; 3135 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3136 ri->i_inline |= F2FS_EXTRA_ATTR; 3137 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3138 ri->i_inline |= F2FS_PIN_FILE; 3139 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3140 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3141 } 3142 3143 static inline int f2fs_has_extra_attr(struct inode *inode) 3144 { 3145 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3146 } 3147 3148 static inline int f2fs_has_inline_xattr(struct inode *inode) 3149 { 3150 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3151 } 3152 3153 static inline int f2fs_compressed_file(struct inode *inode) 3154 { 3155 return S_ISREG(inode->i_mode) && 3156 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3157 } 3158 3159 static inline bool f2fs_need_compress_data(struct inode *inode) 3160 { 3161 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3162 3163 if (!f2fs_compressed_file(inode)) 3164 return false; 3165 3166 if (compress_mode == COMPR_MODE_FS) 3167 return true; 3168 else if (compress_mode == COMPR_MODE_USER && 3169 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3170 return true; 3171 3172 return false; 3173 } 3174 3175 static inline unsigned int addrs_per_inode(struct inode *inode) 3176 { 3177 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3178 get_inline_xattr_addrs(inode); 3179 3180 if (!f2fs_compressed_file(inode)) 3181 return addrs; 3182 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3183 } 3184 3185 static inline unsigned int addrs_per_block(struct inode *inode) 3186 { 3187 if (!f2fs_compressed_file(inode)) 3188 return DEF_ADDRS_PER_BLOCK; 3189 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3190 } 3191 3192 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3193 { 3194 struct f2fs_inode *ri = F2FS_INODE(page); 3195 3196 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3197 get_inline_xattr_addrs(inode)]); 3198 } 3199 3200 static inline int inline_xattr_size(struct inode *inode) 3201 { 3202 if (f2fs_has_inline_xattr(inode)) 3203 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3204 return 0; 3205 } 3206 3207 /* 3208 * Notice: check inline_data flag without inode page lock is unsafe. 3209 * It could change at any time by f2fs_convert_inline_page(). 3210 */ 3211 static inline int f2fs_has_inline_data(struct inode *inode) 3212 { 3213 return is_inode_flag_set(inode, FI_INLINE_DATA); 3214 } 3215 3216 static inline int f2fs_exist_data(struct inode *inode) 3217 { 3218 return is_inode_flag_set(inode, FI_DATA_EXIST); 3219 } 3220 3221 static inline int f2fs_has_inline_dots(struct inode *inode) 3222 { 3223 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3224 } 3225 3226 static inline int f2fs_is_mmap_file(struct inode *inode) 3227 { 3228 return is_inode_flag_set(inode, FI_MMAP_FILE); 3229 } 3230 3231 static inline bool f2fs_is_pinned_file(struct inode *inode) 3232 { 3233 return is_inode_flag_set(inode, FI_PIN_FILE); 3234 } 3235 3236 static inline bool f2fs_is_atomic_file(struct inode *inode) 3237 { 3238 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3239 } 3240 3241 static inline bool f2fs_is_cow_file(struct inode *inode) 3242 { 3243 return is_inode_flag_set(inode, FI_COW_FILE); 3244 } 3245 3246 static inline bool f2fs_is_first_block_written(struct inode *inode) 3247 { 3248 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3249 } 3250 3251 static inline bool f2fs_is_drop_cache(struct inode *inode) 3252 { 3253 return is_inode_flag_set(inode, FI_DROP_CACHE); 3254 } 3255 3256 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3257 { 3258 struct f2fs_inode *ri = F2FS_INODE(page); 3259 int extra_size = get_extra_isize(inode); 3260 3261 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3262 } 3263 3264 static inline int f2fs_has_inline_dentry(struct inode *inode) 3265 { 3266 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3267 } 3268 3269 static inline int is_file(struct inode *inode, int type) 3270 { 3271 return F2FS_I(inode)->i_advise & type; 3272 } 3273 3274 static inline void set_file(struct inode *inode, int type) 3275 { 3276 if (is_file(inode, type)) 3277 return; 3278 F2FS_I(inode)->i_advise |= type; 3279 f2fs_mark_inode_dirty_sync(inode, true); 3280 } 3281 3282 static inline void clear_file(struct inode *inode, int type) 3283 { 3284 if (!is_file(inode, type)) 3285 return; 3286 F2FS_I(inode)->i_advise &= ~type; 3287 f2fs_mark_inode_dirty_sync(inode, true); 3288 } 3289 3290 static inline bool f2fs_is_time_consistent(struct inode *inode) 3291 { 3292 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3293 return false; 3294 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3295 return false; 3296 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3297 return false; 3298 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3299 &F2FS_I(inode)->i_crtime)) 3300 return false; 3301 return true; 3302 } 3303 3304 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3305 { 3306 bool ret; 3307 3308 if (dsync) { 3309 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3310 3311 spin_lock(&sbi->inode_lock[DIRTY_META]); 3312 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3313 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3314 return ret; 3315 } 3316 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3317 file_keep_isize(inode) || 3318 i_size_read(inode) & ~PAGE_MASK) 3319 return false; 3320 3321 if (!f2fs_is_time_consistent(inode)) 3322 return false; 3323 3324 spin_lock(&F2FS_I(inode)->i_size_lock); 3325 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3326 spin_unlock(&F2FS_I(inode)->i_size_lock); 3327 3328 return ret; 3329 } 3330 3331 static inline bool f2fs_readonly(struct super_block *sb) 3332 { 3333 return sb_rdonly(sb); 3334 } 3335 3336 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3337 { 3338 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3339 } 3340 3341 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3342 { 3343 if (len == 1 && name[0] == '.') 3344 return true; 3345 3346 if (len == 2 && name[0] == '.' && name[1] == '.') 3347 return true; 3348 3349 return false; 3350 } 3351 3352 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3353 size_t size, gfp_t flags) 3354 { 3355 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3356 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3357 return NULL; 3358 } 3359 3360 return kmalloc(size, flags); 3361 } 3362 3363 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3364 size_t size, gfp_t flags) 3365 { 3366 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3367 } 3368 3369 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3370 size_t size, gfp_t flags) 3371 { 3372 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3373 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3374 return NULL; 3375 } 3376 3377 return kvmalloc(size, flags); 3378 } 3379 3380 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3381 size_t size, gfp_t flags) 3382 { 3383 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3384 } 3385 3386 static inline int get_extra_isize(struct inode *inode) 3387 { 3388 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3389 } 3390 3391 static inline int get_inline_xattr_addrs(struct inode *inode) 3392 { 3393 return F2FS_I(inode)->i_inline_xattr_size; 3394 } 3395 3396 #define f2fs_get_inode_mode(i) \ 3397 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3398 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3399 3400 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3401 (offsetof(struct f2fs_inode, i_extra_end) - \ 3402 offsetof(struct f2fs_inode, i_extra_isize)) \ 3403 3404 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3405 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3406 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3407 sizeof((f2fs_inode)->field)) \ 3408 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3409 3410 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3411 3412 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3413 3414 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3415 block_t blkaddr, int type); 3416 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3417 block_t blkaddr, int type) 3418 { 3419 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3420 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3421 blkaddr, type); 3422 f2fs_bug_on(sbi, 1); 3423 } 3424 } 3425 3426 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3427 { 3428 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3429 blkaddr == COMPRESS_ADDR) 3430 return false; 3431 return true; 3432 } 3433 3434 /* 3435 * file.c 3436 */ 3437 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3438 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3439 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3440 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3441 int f2fs_truncate(struct inode *inode); 3442 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3443 struct kstat *stat, u32 request_mask, unsigned int flags); 3444 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3445 struct iattr *attr); 3446 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3447 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3448 int f2fs_precache_extents(struct inode *inode); 3449 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3450 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3451 struct dentry *dentry, struct fileattr *fa); 3452 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3453 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3454 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3455 int f2fs_pin_file_control(struct inode *inode, bool inc); 3456 3457 /* 3458 * inode.c 3459 */ 3460 void f2fs_set_inode_flags(struct inode *inode); 3461 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3462 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3463 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3464 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3465 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3466 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3467 void f2fs_update_inode_page(struct inode *inode); 3468 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3469 void f2fs_evict_inode(struct inode *inode); 3470 void f2fs_handle_failed_inode(struct inode *inode); 3471 3472 /* 3473 * namei.c 3474 */ 3475 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3476 bool hot, bool set); 3477 struct dentry *f2fs_get_parent(struct dentry *child); 3478 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3479 struct inode **new_inode); 3480 3481 /* 3482 * dir.c 3483 */ 3484 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3485 int f2fs_init_casefolded_name(const struct inode *dir, 3486 struct f2fs_filename *fname); 3487 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3488 int lookup, struct f2fs_filename *fname); 3489 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3490 struct f2fs_filename *fname); 3491 void f2fs_free_filename(struct f2fs_filename *fname); 3492 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3493 const struct f2fs_filename *fname, int *max_slots); 3494 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3495 unsigned int start_pos, struct fscrypt_str *fstr); 3496 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3497 struct f2fs_dentry_ptr *d); 3498 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3499 const struct f2fs_filename *fname, struct page *dpage); 3500 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3501 unsigned int current_depth); 3502 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3503 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3504 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3505 const struct f2fs_filename *fname, 3506 struct page **res_page); 3507 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3508 const struct qstr *child, struct page **res_page); 3509 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3510 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3511 struct page **page); 3512 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3513 struct page *page, struct inode *inode); 3514 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3515 const struct f2fs_filename *fname); 3516 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3517 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3518 unsigned int bit_pos); 3519 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3520 struct inode *inode, nid_t ino, umode_t mode); 3521 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3522 struct inode *inode, nid_t ino, umode_t mode); 3523 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3524 struct inode *inode, nid_t ino, umode_t mode); 3525 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3526 struct inode *dir, struct inode *inode); 3527 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3528 bool f2fs_empty_dir(struct inode *dir); 3529 3530 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3531 { 3532 if (fscrypt_is_nokey_name(dentry)) 3533 return -ENOKEY; 3534 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3535 inode, inode->i_ino, inode->i_mode); 3536 } 3537 3538 /* 3539 * super.c 3540 */ 3541 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3542 void f2fs_inode_synced(struct inode *inode); 3543 int f2fs_dquot_initialize(struct inode *inode); 3544 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3545 int f2fs_quota_sync(struct super_block *sb, int type); 3546 loff_t max_file_blocks(struct inode *inode); 3547 void f2fs_quota_off_umount(struct super_block *sb); 3548 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3549 int f2fs_sync_fs(struct super_block *sb, int sync); 3550 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3551 3552 /* 3553 * hash.c 3554 */ 3555 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3556 3557 /* 3558 * node.c 3559 */ 3560 struct node_info; 3561 3562 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3563 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3564 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3565 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3566 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3567 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3568 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3569 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3570 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3571 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3572 struct node_info *ni, bool checkpoint_context); 3573 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3574 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3575 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3576 int f2fs_truncate_xattr_node(struct inode *inode); 3577 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3578 unsigned int seq_id); 3579 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3580 int f2fs_remove_inode_page(struct inode *inode); 3581 struct page *f2fs_new_inode_page(struct inode *inode); 3582 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3583 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3584 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3585 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3586 int f2fs_move_node_page(struct page *node_page, int gc_type); 3587 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3588 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3589 struct writeback_control *wbc, bool atomic, 3590 unsigned int *seq_id); 3591 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3592 struct writeback_control *wbc, 3593 bool do_balance, enum iostat_type io_type); 3594 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3595 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3596 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3597 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3598 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3599 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3600 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3601 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3602 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3603 unsigned int segno, struct f2fs_summary_block *sum); 3604 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3605 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3606 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3607 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3608 int __init f2fs_create_node_manager_caches(void); 3609 void f2fs_destroy_node_manager_caches(void); 3610 3611 /* 3612 * segment.c 3613 */ 3614 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3615 int f2fs_commit_atomic_write(struct inode *inode); 3616 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3617 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3618 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3619 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3620 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3621 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3622 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3623 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3624 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3625 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3626 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3627 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3628 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3629 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3630 struct cp_control *cpc); 3631 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3632 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3633 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3634 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3635 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3636 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3637 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3638 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3639 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3640 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3641 unsigned int *newseg, bool new_sec, int dir); 3642 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3643 unsigned int start, unsigned int end); 3644 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3645 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3646 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3647 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3648 struct cp_control *cpc); 3649 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3650 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3651 block_t blk_addr); 3652 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3653 enum iostat_type io_type); 3654 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3655 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3656 struct f2fs_io_info *fio); 3657 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3658 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3659 block_t old_blkaddr, block_t new_blkaddr, 3660 bool recover_curseg, bool recover_newaddr, 3661 bool from_gc); 3662 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3663 block_t old_addr, block_t new_addr, 3664 unsigned char version, bool recover_curseg, 3665 bool recover_newaddr); 3666 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3667 block_t old_blkaddr, block_t *new_blkaddr, 3668 struct f2fs_summary *sum, int type, 3669 struct f2fs_io_info *fio); 3670 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3671 block_t blkaddr, unsigned int blkcnt); 3672 void f2fs_wait_on_page_writeback(struct page *page, 3673 enum page_type type, bool ordered, bool locked); 3674 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3675 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3676 block_t len); 3677 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3678 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3679 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3680 unsigned int val, int alloc); 3681 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3682 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3683 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3684 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3685 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3686 int __init f2fs_create_segment_manager_caches(void); 3687 void f2fs_destroy_segment_manager_caches(void); 3688 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3689 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3690 unsigned int segno); 3691 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3692 unsigned int segno); 3693 3694 #define DEF_FRAGMENT_SIZE 4 3695 #define MIN_FRAGMENT_SIZE 1 3696 #define MAX_FRAGMENT_SIZE 512 3697 3698 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3699 { 3700 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3701 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3702 } 3703 3704 /* 3705 * checkpoint.c 3706 */ 3707 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3708 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3709 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3710 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3711 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3712 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3713 block_t blkaddr, int type); 3714 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3715 int type, bool sync); 3716 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3717 unsigned int ra_blocks); 3718 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3719 long nr_to_write, enum iostat_type io_type); 3720 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3721 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3722 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3723 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3724 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3725 unsigned int devidx, int type); 3726 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3727 unsigned int devidx, int type); 3728 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3729 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3730 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3731 void f2fs_add_orphan_inode(struct inode *inode); 3732 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3733 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3734 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3735 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3736 void f2fs_remove_dirty_inode(struct inode *inode); 3737 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3738 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3739 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3740 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3741 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3742 int __init f2fs_create_checkpoint_caches(void); 3743 void f2fs_destroy_checkpoint_caches(void); 3744 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3745 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3746 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3747 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3748 3749 /* 3750 * data.c 3751 */ 3752 int __init f2fs_init_bioset(void); 3753 void f2fs_destroy_bioset(void); 3754 int f2fs_init_bio_entry_cache(void); 3755 void f2fs_destroy_bio_entry_cache(void); 3756 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3757 struct bio *bio, enum page_type type); 3758 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3759 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3760 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3761 struct inode *inode, struct page *page, 3762 nid_t ino, enum page_type type); 3763 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3764 struct bio **bio, struct page *page); 3765 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3766 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3767 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3768 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3769 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3770 block_t blk_addr, sector_t *sector); 3771 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3772 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3773 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3774 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3775 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3776 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3777 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3778 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3779 int op_flags, bool for_write); 3780 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3781 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3782 bool for_write); 3783 struct page *f2fs_get_new_data_page(struct inode *inode, 3784 struct page *ipage, pgoff_t index, bool new_i_size); 3785 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3786 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3787 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3788 int create, int flag); 3789 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3790 u64 start, u64 len); 3791 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3792 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3793 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3794 int f2fs_write_single_data_page(struct page *page, int *submitted, 3795 struct bio **bio, sector_t *last_block, 3796 struct writeback_control *wbc, 3797 enum iostat_type io_type, 3798 int compr_blocks, bool allow_balance); 3799 void f2fs_write_failed(struct inode *inode, loff_t to); 3800 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3801 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3802 #ifdef CONFIG_MIGRATION 3803 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3804 struct page *page, enum migrate_mode mode); 3805 #endif 3806 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3807 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3808 int f2fs_init_post_read_processing(void); 3809 void f2fs_destroy_post_read_processing(void); 3810 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3811 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3812 extern const struct iomap_ops f2fs_iomap_ops; 3813 3814 /* 3815 * gc.c 3816 */ 3817 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3818 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3819 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3820 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3821 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3822 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3823 int __init f2fs_create_garbage_collection_cache(void); 3824 void f2fs_destroy_garbage_collection_cache(void); 3825 3826 /* 3827 * recovery.c 3828 */ 3829 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3830 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3831 int __init f2fs_create_recovery_cache(void); 3832 void f2fs_destroy_recovery_cache(void); 3833 3834 /* 3835 * debug.c 3836 */ 3837 #ifdef CONFIG_F2FS_STAT_FS 3838 struct f2fs_stat_info { 3839 struct list_head stat_list; 3840 struct f2fs_sb_info *sbi; 3841 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3842 int main_area_segs, main_area_sections, main_area_zones; 3843 unsigned long long hit_largest, hit_cached, hit_rbtree; 3844 unsigned long long hit_total, total_ext; 3845 int ext_tree, zombie_tree, ext_node; 3846 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3847 int ndirty_data, ndirty_qdata; 3848 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3849 int nats, dirty_nats, sits, dirty_sits; 3850 int free_nids, avail_nids, alloc_nids; 3851 int total_count, utilization; 3852 int bg_gc, nr_wb_cp_data, nr_wb_data; 3853 int nr_rd_data, nr_rd_node, nr_rd_meta; 3854 int nr_dio_read, nr_dio_write; 3855 unsigned int io_skip_bggc, other_skip_bggc; 3856 int nr_flushing, nr_flushed, flush_list_empty; 3857 int nr_discarding, nr_discarded; 3858 int nr_discard_cmd; 3859 unsigned int undiscard_blks; 3860 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3861 unsigned int cur_ckpt_time, peak_ckpt_time; 3862 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3863 int compr_inode; 3864 unsigned long long compr_blocks; 3865 int aw_cnt, max_aw_cnt; 3866 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3867 unsigned int bimodal, avg_vblocks; 3868 int util_free, util_valid, util_invalid; 3869 int rsvd_segs, overp_segs; 3870 int dirty_count, node_pages, meta_pages, compress_pages; 3871 int compress_page_hit; 3872 int prefree_count, call_count, cp_count, bg_cp_count; 3873 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3874 int bg_node_segs, bg_data_segs; 3875 int tot_blks, data_blks, node_blks; 3876 int bg_data_blks, bg_node_blks; 3877 int curseg[NR_CURSEG_TYPE]; 3878 int cursec[NR_CURSEG_TYPE]; 3879 int curzone[NR_CURSEG_TYPE]; 3880 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3881 unsigned int full_seg[NR_CURSEG_TYPE]; 3882 unsigned int valid_blks[NR_CURSEG_TYPE]; 3883 3884 unsigned int meta_count[META_MAX]; 3885 unsigned int segment_count[2]; 3886 unsigned int block_count[2]; 3887 unsigned int inplace_count; 3888 unsigned long long base_mem, cache_mem, page_mem; 3889 }; 3890 3891 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3892 { 3893 return (struct f2fs_stat_info *)sbi->stat_info; 3894 } 3895 3896 #define stat_inc_cp_count(si) ((si)->cp_count++) 3897 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3898 #define stat_inc_call_count(si) ((si)->call_count++) 3899 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3900 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3901 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3902 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3903 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3904 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3905 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3906 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3907 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3908 #define stat_inc_inline_xattr(inode) \ 3909 do { \ 3910 if (f2fs_has_inline_xattr(inode)) \ 3911 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3912 } while (0) 3913 #define stat_dec_inline_xattr(inode) \ 3914 do { \ 3915 if (f2fs_has_inline_xattr(inode)) \ 3916 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3917 } while (0) 3918 #define stat_inc_inline_inode(inode) \ 3919 do { \ 3920 if (f2fs_has_inline_data(inode)) \ 3921 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3922 } while (0) 3923 #define stat_dec_inline_inode(inode) \ 3924 do { \ 3925 if (f2fs_has_inline_data(inode)) \ 3926 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3927 } while (0) 3928 #define stat_inc_inline_dir(inode) \ 3929 do { \ 3930 if (f2fs_has_inline_dentry(inode)) \ 3931 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3932 } while (0) 3933 #define stat_dec_inline_dir(inode) \ 3934 do { \ 3935 if (f2fs_has_inline_dentry(inode)) \ 3936 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3937 } while (0) 3938 #define stat_inc_compr_inode(inode) \ 3939 do { \ 3940 if (f2fs_compressed_file(inode)) \ 3941 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3942 } while (0) 3943 #define stat_dec_compr_inode(inode) \ 3944 do { \ 3945 if (f2fs_compressed_file(inode)) \ 3946 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3947 } while (0) 3948 #define stat_add_compr_blocks(inode, blocks) \ 3949 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3950 #define stat_sub_compr_blocks(inode, blocks) \ 3951 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3952 #define stat_inc_meta_count(sbi, blkaddr) \ 3953 do { \ 3954 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3955 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3956 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3957 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3958 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3959 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3960 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3961 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3962 } while (0) 3963 #define stat_inc_seg_type(sbi, curseg) \ 3964 ((sbi)->segment_count[(curseg)->alloc_type]++) 3965 #define stat_inc_block_count(sbi, curseg) \ 3966 ((sbi)->block_count[(curseg)->alloc_type]++) 3967 #define stat_inc_inplace_blocks(sbi) \ 3968 (atomic_inc(&(sbi)->inplace_count)) 3969 #define stat_update_max_atomic_write(inode) \ 3970 do { \ 3971 int cur = F2FS_I_SB(inode)->atomic_files; \ 3972 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3973 if (cur > max) \ 3974 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3975 } while (0) 3976 #define stat_inc_seg_count(sbi, type, gc_type) \ 3977 do { \ 3978 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3979 si->tot_segs++; \ 3980 if ((type) == SUM_TYPE_DATA) { \ 3981 si->data_segs++; \ 3982 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3983 } else { \ 3984 si->node_segs++; \ 3985 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3986 } \ 3987 } while (0) 3988 3989 #define stat_inc_tot_blk_count(si, blks) \ 3990 ((si)->tot_blks += (blks)) 3991 3992 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3993 do { \ 3994 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3995 stat_inc_tot_blk_count(si, blks); \ 3996 si->data_blks += (blks); \ 3997 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3998 } while (0) 3999 4000 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4001 do { \ 4002 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4003 stat_inc_tot_blk_count(si, blks); \ 4004 si->node_blks += (blks); \ 4005 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4006 } while (0) 4007 4008 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4009 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4010 void __init f2fs_create_root_stats(void); 4011 void f2fs_destroy_root_stats(void); 4012 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4013 #else 4014 #define stat_inc_cp_count(si) do { } while (0) 4015 #define stat_inc_bg_cp_count(si) do { } while (0) 4016 #define stat_inc_call_count(si) do { } while (0) 4017 #define stat_inc_bggc_count(si) do { } while (0) 4018 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4019 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4020 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4021 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4022 #define stat_inc_total_hit(sbi) do { } while (0) 4023 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4024 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4025 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4026 #define stat_inc_inline_xattr(inode) do { } while (0) 4027 #define stat_dec_inline_xattr(inode) do { } while (0) 4028 #define stat_inc_inline_inode(inode) do { } while (0) 4029 #define stat_dec_inline_inode(inode) do { } while (0) 4030 #define stat_inc_inline_dir(inode) do { } while (0) 4031 #define stat_dec_inline_dir(inode) do { } while (0) 4032 #define stat_inc_compr_inode(inode) do { } while (0) 4033 #define stat_dec_compr_inode(inode) do { } while (0) 4034 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4035 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4036 #define stat_update_max_atomic_write(inode) do { } while (0) 4037 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4038 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4039 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4040 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4041 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4042 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4043 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4044 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4045 4046 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4047 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4048 static inline void __init f2fs_create_root_stats(void) { } 4049 static inline void f2fs_destroy_root_stats(void) { } 4050 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4051 #endif 4052 4053 extern const struct file_operations f2fs_dir_operations; 4054 extern const struct file_operations f2fs_file_operations; 4055 extern const struct inode_operations f2fs_file_inode_operations; 4056 extern const struct address_space_operations f2fs_dblock_aops; 4057 extern const struct address_space_operations f2fs_node_aops; 4058 extern const struct address_space_operations f2fs_meta_aops; 4059 extern const struct inode_operations f2fs_dir_inode_operations; 4060 extern const struct inode_operations f2fs_symlink_inode_operations; 4061 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4062 extern const struct inode_operations f2fs_special_inode_operations; 4063 extern struct kmem_cache *f2fs_inode_entry_slab; 4064 4065 /* 4066 * inline.c 4067 */ 4068 bool f2fs_may_inline_data(struct inode *inode); 4069 bool f2fs_sanity_check_inline_data(struct inode *inode); 4070 bool f2fs_may_inline_dentry(struct inode *inode); 4071 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4072 void f2fs_truncate_inline_inode(struct inode *inode, 4073 struct page *ipage, u64 from); 4074 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4075 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4076 int f2fs_convert_inline_inode(struct inode *inode); 4077 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4078 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4079 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4080 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4081 const struct f2fs_filename *fname, 4082 struct page **res_page); 4083 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4084 struct page *ipage); 4085 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4086 struct inode *inode, nid_t ino, umode_t mode); 4087 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4088 struct page *page, struct inode *dir, 4089 struct inode *inode); 4090 bool f2fs_empty_inline_dir(struct inode *dir); 4091 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4092 struct fscrypt_str *fstr); 4093 int f2fs_inline_data_fiemap(struct inode *inode, 4094 struct fiemap_extent_info *fieinfo, 4095 __u64 start, __u64 len); 4096 4097 /* 4098 * shrinker.c 4099 */ 4100 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4101 struct shrink_control *sc); 4102 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4103 struct shrink_control *sc); 4104 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4105 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4106 4107 /* 4108 * extent_cache.c 4109 */ 4110 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4111 struct rb_entry *cached_re, unsigned int ofs); 4112 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4113 struct rb_root_cached *root, 4114 struct rb_node **parent, 4115 unsigned long long key, bool *left_most); 4116 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4117 struct rb_root_cached *root, 4118 struct rb_node **parent, 4119 unsigned int ofs, bool *leftmost); 4120 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4121 struct rb_entry *cached_re, unsigned int ofs, 4122 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4123 struct rb_node ***insert_p, struct rb_node **insert_parent, 4124 bool force, bool *leftmost); 4125 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4126 struct rb_root_cached *root, bool check_key); 4127 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4128 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4129 void f2fs_drop_extent_tree(struct inode *inode); 4130 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4131 void f2fs_destroy_extent_tree(struct inode *inode); 4132 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4133 struct extent_info *ei); 4134 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4135 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4136 pgoff_t fofs, block_t blkaddr, unsigned int len); 4137 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4138 int __init f2fs_create_extent_cache(void); 4139 void f2fs_destroy_extent_cache(void); 4140 4141 /* 4142 * sysfs.c 4143 */ 4144 #define MIN_RA_MUL 2 4145 #define MAX_RA_MUL 256 4146 4147 int __init f2fs_init_sysfs(void); 4148 void f2fs_exit_sysfs(void); 4149 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4150 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4151 4152 /* verity.c */ 4153 extern const struct fsverity_operations f2fs_verityops; 4154 4155 /* 4156 * crypto support 4157 */ 4158 static inline bool f2fs_encrypted_file(struct inode *inode) 4159 { 4160 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4161 } 4162 4163 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4164 { 4165 #ifdef CONFIG_FS_ENCRYPTION 4166 file_set_encrypt(inode); 4167 f2fs_set_inode_flags(inode); 4168 #endif 4169 } 4170 4171 /* 4172 * Returns true if the reads of the inode's data need to undergo some 4173 * postprocessing step, like decryption or authenticity verification. 4174 */ 4175 static inline bool f2fs_post_read_required(struct inode *inode) 4176 { 4177 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4178 f2fs_compressed_file(inode); 4179 } 4180 4181 /* 4182 * compress.c 4183 */ 4184 #ifdef CONFIG_F2FS_FS_COMPRESSION 4185 bool f2fs_is_compressed_page(struct page *page); 4186 struct page *f2fs_compress_control_page(struct page *page); 4187 int f2fs_prepare_compress_overwrite(struct inode *inode, 4188 struct page **pagep, pgoff_t index, void **fsdata); 4189 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4190 pgoff_t index, unsigned copied); 4191 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4192 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4193 bool f2fs_is_compress_backend_ready(struct inode *inode); 4194 int f2fs_init_compress_mempool(void); 4195 void f2fs_destroy_compress_mempool(void); 4196 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4197 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4198 block_t blkaddr, bool in_task); 4199 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4200 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4201 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct pagevec *pvec, 4202 int index, int nr_pages, bool uptodate); 4203 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4204 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4205 int f2fs_write_multi_pages(struct compress_ctx *cc, 4206 int *submitted, 4207 struct writeback_control *wbc, 4208 enum iostat_type io_type); 4209 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4210 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4211 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4212 unsigned int c_len); 4213 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4214 unsigned nr_pages, sector_t *last_block_in_bio, 4215 bool is_readahead, bool for_write); 4216 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4217 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4218 bool in_task); 4219 void f2fs_put_page_dic(struct page *page, bool in_task); 4220 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4221 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4222 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4223 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4224 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4225 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4226 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4227 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4228 int __init f2fs_init_compress_cache(void); 4229 void f2fs_destroy_compress_cache(void); 4230 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4231 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4232 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4233 nid_t ino, block_t blkaddr); 4234 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4235 block_t blkaddr); 4236 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4237 #define inc_compr_inode_stat(inode) \ 4238 do { \ 4239 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4240 sbi->compr_new_inode++; \ 4241 } while (0) 4242 #define add_compr_block_stat(inode, blocks) \ 4243 do { \ 4244 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4245 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4246 sbi->compr_written_block += blocks; \ 4247 sbi->compr_saved_block += diff; \ 4248 } while (0) 4249 #else 4250 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4251 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4252 { 4253 if (!f2fs_compressed_file(inode)) 4254 return true; 4255 /* not support compression */ 4256 return false; 4257 } 4258 static inline struct page *f2fs_compress_control_page(struct page *page) 4259 { 4260 WARN_ON_ONCE(1); 4261 return ERR_PTR(-EINVAL); 4262 } 4263 static inline int f2fs_init_compress_mempool(void) { return 0; } 4264 static inline void f2fs_destroy_compress_mempool(void) { } 4265 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4266 bool in_task) { } 4267 static inline void f2fs_end_read_compressed_page(struct page *page, 4268 bool failed, block_t blkaddr, bool in_task) 4269 { 4270 WARN_ON_ONCE(1); 4271 } 4272 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4273 { 4274 WARN_ON_ONCE(1); 4275 } 4276 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4277 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4278 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4279 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4280 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4281 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4282 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4283 static inline void f2fs_destroy_compress_cache(void) { } 4284 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4285 block_t blkaddr) { } 4286 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4287 struct page *page, nid_t ino, block_t blkaddr) { } 4288 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4289 struct page *page, block_t blkaddr) { return false; } 4290 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4291 nid_t ino) { } 4292 #define inc_compr_inode_stat(inode) do { } while (0) 4293 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4294 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4295 unsigned int c_len) { } 4296 #endif 4297 4298 static inline int set_compress_context(struct inode *inode) 4299 { 4300 #ifdef CONFIG_F2FS_FS_COMPRESSION 4301 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4302 4303 F2FS_I(inode)->i_compress_algorithm = 4304 F2FS_OPTION(sbi).compress_algorithm; 4305 F2FS_I(inode)->i_log_cluster_size = 4306 F2FS_OPTION(sbi).compress_log_size; 4307 F2FS_I(inode)->i_compress_flag = 4308 F2FS_OPTION(sbi).compress_chksum ? 4309 1 << COMPRESS_CHKSUM : 0; 4310 F2FS_I(inode)->i_cluster_size = 4311 1 << F2FS_I(inode)->i_log_cluster_size; 4312 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4313 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4314 F2FS_OPTION(sbi).compress_level) 4315 F2FS_I(inode)->i_compress_flag |= 4316 F2FS_OPTION(sbi).compress_level << 4317 COMPRESS_LEVEL_OFFSET; 4318 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4319 set_inode_flag(inode, FI_COMPRESSED_FILE); 4320 stat_inc_compr_inode(inode); 4321 inc_compr_inode_stat(inode); 4322 f2fs_mark_inode_dirty_sync(inode, true); 4323 return 0; 4324 #else 4325 return -EOPNOTSUPP; 4326 #endif 4327 } 4328 4329 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4330 { 4331 struct f2fs_inode_info *fi = F2FS_I(inode); 4332 4333 if (!f2fs_compressed_file(inode)) 4334 return true; 4335 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4336 return false; 4337 4338 fi->i_flags &= ~F2FS_COMPR_FL; 4339 stat_dec_compr_inode(inode); 4340 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4341 f2fs_mark_inode_dirty_sync(inode, true); 4342 return true; 4343 } 4344 4345 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4346 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4347 { \ 4348 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4349 } 4350 4351 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4352 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4353 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4354 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4355 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4356 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4357 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4358 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4359 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4360 F2FS_FEATURE_FUNCS(verity, VERITY); 4361 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4362 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4363 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4364 F2FS_FEATURE_FUNCS(readonly, RO); 4365 4366 static inline bool f2fs_may_extent_tree(struct inode *inode) 4367 { 4368 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4369 4370 if (!test_opt(sbi, EXTENT_CACHE) || 4371 is_inode_flag_set(inode, FI_NO_EXTENT) || 4372 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4373 !f2fs_sb_has_readonly(sbi))) 4374 return false; 4375 4376 /* 4377 * for recovered files during mount do not create extents 4378 * if shrinker is not registered. 4379 */ 4380 if (list_empty(&sbi->s_list)) 4381 return false; 4382 4383 return S_ISREG(inode->i_mode); 4384 } 4385 4386 #ifdef CONFIG_BLK_DEV_ZONED 4387 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4388 block_t blkaddr) 4389 { 4390 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4391 4392 return test_bit(zno, FDEV(devi).blkz_seq); 4393 } 4394 #endif 4395 4396 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4397 { 4398 return f2fs_sb_has_blkzoned(sbi); 4399 } 4400 4401 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4402 { 4403 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4404 } 4405 4406 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4407 { 4408 int i; 4409 4410 if (!f2fs_is_multi_device(sbi)) 4411 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4412 4413 for (i = 0; i < sbi->s_ndevs; i++) 4414 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4415 return true; 4416 return false; 4417 } 4418 4419 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4420 { 4421 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4422 f2fs_hw_should_discard(sbi); 4423 } 4424 4425 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4426 { 4427 int i; 4428 4429 if (!f2fs_is_multi_device(sbi)) 4430 return bdev_read_only(sbi->sb->s_bdev); 4431 4432 for (i = 0; i < sbi->s_ndevs; i++) 4433 if (bdev_read_only(FDEV(i).bdev)) 4434 return true; 4435 return false; 4436 } 4437 4438 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4439 { 4440 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4441 } 4442 4443 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4444 { 4445 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4446 } 4447 4448 static inline bool f2fs_may_compress(struct inode *inode) 4449 { 4450 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4451 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4452 return false; 4453 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4454 } 4455 4456 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4457 u64 blocks, bool add) 4458 { 4459 struct f2fs_inode_info *fi = F2FS_I(inode); 4460 int diff = fi->i_cluster_size - blocks; 4461 4462 /* don't update i_compr_blocks if saved blocks were released */ 4463 if (!add && !atomic_read(&fi->i_compr_blocks)) 4464 return; 4465 4466 if (add) { 4467 atomic_add(diff, &fi->i_compr_blocks); 4468 stat_add_compr_blocks(inode, diff); 4469 } else { 4470 atomic_sub(diff, &fi->i_compr_blocks); 4471 stat_sub_compr_blocks(inode, diff); 4472 } 4473 f2fs_mark_inode_dirty_sync(inode, true); 4474 } 4475 4476 static inline int block_unaligned_IO(struct inode *inode, 4477 struct kiocb *iocb, struct iov_iter *iter) 4478 { 4479 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4480 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4481 loff_t offset = iocb->ki_pos; 4482 unsigned long align = offset | iov_iter_alignment(iter); 4483 4484 return align & blocksize_mask; 4485 } 4486 4487 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4488 int flag) 4489 { 4490 if (!f2fs_is_multi_device(sbi)) 4491 return false; 4492 if (flag != F2FS_GET_BLOCK_DIO) 4493 return false; 4494 return sbi->aligned_blksize; 4495 } 4496 4497 static inline bool f2fs_force_buffered_io(struct inode *inode, 4498 struct kiocb *iocb, struct iov_iter *iter) 4499 { 4500 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4501 int rw = iov_iter_rw(iter); 4502 4503 if (!fscrypt_dio_supported(iocb, iter)) 4504 return true; 4505 if (fsverity_active(inode)) 4506 return true; 4507 if (f2fs_compressed_file(inode)) 4508 return true; 4509 4510 /* disallow direct IO if any of devices has unaligned blksize */ 4511 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4512 return true; 4513 4514 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4515 if (block_unaligned_IO(inode, iocb, iter)) 4516 return true; 4517 if (F2FS_IO_ALIGNED(sbi)) 4518 return true; 4519 } 4520 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4521 return true; 4522 4523 return false; 4524 } 4525 4526 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4527 { 4528 return fsverity_active(inode) && 4529 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4530 } 4531 4532 #ifdef CONFIG_F2FS_FAULT_INJECTION 4533 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4534 unsigned int type); 4535 #else 4536 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4537 #endif 4538 4539 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4540 { 4541 #ifdef CONFIG_QUOTA 4542 if (f2fs_sb_has_quota_ino(sbi)) 4543 return true; 4544 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4545 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4546 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4547 return true; 4548 #endif 4549 return false; 4550 } 4551 4552 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4553 { 4554 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4555 } 4556 4557 static inline void f2fs_io_schedule_timeout(long timeout) 4558 { 4559 set_current_state(TASK_UNINTERRUPTIBLE); 4560 io_schedule_timeout(timeout); 4561 } 4562 4563 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4564 enum page_type type) 4565 { 4566 if (unlikely(f2fs_cp_error(sbi))) 4567 return; 4568 4569 if (ofs == sbi->page_eio_ofs[type]) { 4570 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4571 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4572 } else { 4573 sbi->page_eio_ofs[type] = ofs; 4574 sbi->page_eio_cnt[type] = 0; 4575 } 4576 } 4577 4578 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4579 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4580 4581 #endif /* _LINUX_F2FS_H */ 4582