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/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_MAX, 63 }; 64 65 #ifdef CONFIG_F2FS_FAULT_INJECTION 66 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 67 68 struct f2fs_fault_info { 69 atomic_t inject_ops; 70 unsigned int inject_rate; 71 unsigned int inject_type; 72 }; 73 74 extern const char *f2fs_fault_name[FAULT_MAX]; 75 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 76 #endif 77 78 /* 79 * For mount options 80 */ 81 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 82 #define F2FS_MOUNT_DISCARD 0x00000004 83 #define F2FS_MOUNT_NOHEAP 0x00000008 84 #define F2FS_MOUNT_XATTR_USER 0x00000010 85 #define F2FS_MOUNT_POSIX_ACL 0x00000020 86 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 87 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 88 #define F2FS_MOUNT_INLINE_DATA 0x00000100 89 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 90 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 91 #define F2FS_MOUNT_NOBARRIER 0x00000800 92 #define F2FS_MOUNT_FASTBOOT 0x00001000 93 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 94 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 95 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 #define F2FS_MOUNT_NORECOVERY 0x04000000 104 #define F2FS_MOUNT_ATGC 0x08000000 105 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 106 #define F2FS_MOUNT_GC_MERGE 0x20000000 107 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 108 109 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 110 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 111 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 112 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 113 114 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 115 typecheck(unsigned long long, b) && \ 116 ((long long)((a) - (b)) > 0)) 117 118 typedef u32 block_t; /* 119 * should not change u32, since it is the on-disk block 120 * address format, __le32. 121 */ 122 typedef u32 nid_t; 123 124 #define COMPRESS_EXT_NUM 16 125 126 /* 127 * An implementation of an rwsem that is explicitly unfair to readers. This 128 * prevents priority inversion when a low-priority reader acquires the read lock 129 * while sleeping on the write lock but the write lock is needed by 130 * higher-priority clients. 131 */ 132 133 struct f2fs_rwsem { 134 struct rw_semaphore internal_rwsem; 135 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 136 wait_queue_head_t read_waiters; 137 #endif 138 }; 139 140 struct f2fs_mount_info { 141 unsigned int opt; 142 int write_io_size_bits; /* Write IO size bits */ 143 block_t root_reserved_blocks; /* root reserved blocks */ 144 kuid_t s_resuid; /* reserved blocks for uid */ 145 kgid_t s_resgid; /* reserved blocks for gid */ 146 int active_logs; /* # of active logs */ 147 int inline_xattr_size; /* inline xattr size */ 148 #ifdef CONFIG_F2FS_FAULT_INJECTION 149 struct f2fs_fault_info fault_info; /* For fault injection */ 150 #endif 151 #ifdef CONFIG_QUOTA 152 /* Names of quota files with journalled quota */ 153 char *s_qf_names[MAXQUOTAS]; 154 int s_jquota_fmt; /* Format of quota to use */ 155 #endif 156 /* For which write hints are passed down to block layer */ 157 int alloc_mode; /* segment allocation policy */ 158 int fsync_mode; /* fsync policy */ 159 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 160 int bggc_mode; /* bggc mode: off, on or sync */ 161 int discard_unit; /* 162 * discard command's offset/size should 163 * be aligned to this unit: block, 164 * segment or section 165 */ 166 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 167 block_t unusable_cap_perc; /* percentage for cap */ 168 block_t unusable_cap; /* Amount of space allowed to be 169 * unusable when disabling checkpoint 170 */ 171 172 /* For compression */ 173 unsigned char compress_algorithm; /* algorithm type */ 174 unsigned char compress_log_size; /* cluster log size */ 175 unsigned char compress_level; /* compress level */ 176 bool compress_chksum; /* compressed data chksum */ 177 unsigned char compress_ext_cnt; /* extension count */ 178 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 179 int compress_mode; /* compression mode */ 180 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 181 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 182 }; 183 184 #define F2FS_FEATURE_ENCRYPT 0x0001 185 #define F2FS_FEATURE_BLKZONED 0x0002 186 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 187 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 188 #define F2FS_FEATURE_PRJQUOTA 0x0010 189 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 190 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 191 #define F2FS_FEATURE_QUOTA_INO 0x0080 192 #define F2FS_FEATURE_INODE_CRTIME 0x0100 193 #define F2FS_FEATURE_LOST_FOUND 0x0200 194 #define F2FS_FEATURE_VERITY 0x0400 195 #define F2FS_FEATURE_SB_CHKSUM 0x0800 196 #define F2FS_FEATURE_CASEFOLD 0x1000 197 #define F2FS_FEATURE_COMPRESSION 0x2000 198 #define F2FS_FEATURE_RO 0x4000 199 200 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 201 ((raw_super->feature & cpu_to_le32(mask)) != 0) 202 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 203 #define F2FS_SET_FEATURE(sbi, mask) \ 204 (sbi->raw_super->feature |= cpu_to_le32(mask)) 205 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 206 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 207 208 /* 209 * Default values for user and/or group using reserved blocks 210 */ 211 #define F2FS_DEF_RESUID 0 212 #define F2FS_DEF_RESGID 0 213 214 /* 215 * For checkpoint manager 216 */ 217 enum { 218 NAT_BITMAP, 219 SIT_BITMAP 220 }; 221 222 #define CP_UMOUNT 0x00000001 223 #define CP_FASTBOOT 0x00000002 224 #define CP_SYNC 0x00000004 225 #define CP_RECOVERY 0x00000008 226 #define CP_DISCARD 0x00000010 227 #define CP_TRIMMED 0x00000020 228 #define CP_PAUSE 0x00000040 229 #define CP_RESIZE 0x00000080 230 231 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 232 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 233 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 234 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 235 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 236 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 237 #define DEF_CP_INTERVAL 60 /* 60 secs */ 238 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 241 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 242 243 struct cp_control { 244 int reason; 245 __u64 trim_start; 246 __u64 trim_end; 247 __u64 trim_minlen; 248 }; 249 250 /* 251 * indicate meta/data type 252 */ 253 enum { 254 META_CP, 255 META_NAT, 256 META_SIT, 257 META_SSA, 258 META_MAX, 259 META_POR, 260 DATA_GENERIC, /* check range only */ 261 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 262 DATA_GENERIC_ENHANCE_READ, /* 263 * strong check on range and segment 264 * bitmap but no warning due to race 265 * condition of read on truncated area 266 * by extent_cache 267 */ 268 META_GENERIC, 269 }; 270 271 /* for the list of ino */ 272 enum { 273 ORPHAN_INO, /* for orphan ino list */ 274 APPEND_INO, /* for append ino list */ 275 UPDATE_INO, /* for update ino list */ 276 TRANS_DIR_INO, /* for trasactions dir ino list */ 277 FLUSH_INO, /* for multiple device flushing */ 278 MAX_INO_ENTRY, /* max. list */ 279 }; 280 281 struct ino_entry { 282 struct list_head list; /* list head */ 283 nid_t ino; /* inode number */ 284 unsigned int dirty_device; /* dirty device bitmap */ 285 }; 286 287 /* for the list of inodes to be GCed */ 288 struct inode_entry { 289 struct list_head list; /* list head */ 290 struct inode *inode; /* vfs inode pointer */ 291 }; 292 293 struct fsync_node_entry { 294 struct list_head list; /* list head */ 295 struct page *page; /* warm node page pointer */ 296 unsigned int seq_id; /* sequence id */ 297 }; 298 299 struct ckpt_req { 300 struct completion wait; /* completion for checkpoint done */ 301 struct llist_node llnode; /* llist_node to be linked in wait queue */ 302 int ret; /* return code of checkpoint */ 303 ktime_t queue_time; /* request queued time */ 304 }; 305 306 struct ckpt_req_control { 307 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 308 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 309 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 310 atomic_t issued_ckpt; /* # of actually issued ckpts */ 311 atomic_t total_ckpt; /* # of total ckpts */ 312 atomic_t queued_ckpt; /* # of queued ckpts */ 313 struct llist_head issue_list; /* list for command issue */ 314 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 315 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 316 unsigned int peak_time; /* peak wait time in msec until now */ 317 }; 318 319 /* for the bitmap indicate blocks to be discarded */ 320 struct discard_entry { 321 struct list_head list; /* list head */ 322 block_t start_blkaddr; /* start blockaddr of current segment */ 323 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 324 }; 325 326 /* default discard granularity of inner discard thread, unit: block count */ 327 #define DEFAULT_DISCARD_GRANULARITY 16 328 329 /* max discard pend list number */ 330 #define MAX_PLIST_NUM 512 331 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 332 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 333 334 enum { 335 D_PREP, /* initial */ 336 D_PARTIAL, /* partially submitted */ 337 D_SUBMIT, /* all submitted */ 338 D_DONE, /* finished */ 339 }; 340 341 struct discard_info { 342 block_t lstart; /* logical start address */ 343 block_t len; /* length */ 344 block_t start; /* actual start address in dev */ 345 }; 346 347 struct discard_cmd { 348 struct rb_node rb_node; /* rb node located in rb-tree */ 349 union { 350 struct { 351 block_t lstart; /* logical start address */ 352 block_t len; /* length */ 353 block_t start; /* actual start address in dev */ 354 }; 355 struct discard_info di; /* discard info */ 356 357 }; 358 struct list_head list; /* command list */ 359 struct completion wait; /* compleation */ 360 struct block_device *bdev; /* bdev */ 361 unsigned short ref; /* reference count */ 362 unsigned char state; /* state */ 363 unsigned char queued; /* queued discard */ 364 int error; /* bio error */ 365 spinlock_t lock; /* for state/bio_ref updating */ 366 unsigned short bio_ref; /* bio reference count */ 367 }; 368 369 enum { 370 DPOLICY_BG, 371 DPOLICY_FORCE, 372 DPOLICY_FSTRIM, 373 DPOLICY_UMOUNT, 374 MAX_DPOLICY, 375 }; 376 377 struct discard_policy { 378 int type; /* type of discard */ 379 unsigned int min_interval; /* used for candidates exist */ 380 unsigned int mid_interval; /* used for device busy */ 381 unsigned int max_interval; /* used for candidates not exist */ 382 unsigned int max_requests; /* # of discards issued per round */ 383 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 384 bool io_aware; /* issue discard in idle time */ 385 bool sync; /* submit discard with REQ_SYNC flag */ 386 bool ordered; /* issue discard by lba order */ 387 bool timeout; /* discard timeout for put_super */ 388 unsigned int granularity; /* discard granularity */ 389 }; 390 391 struct discard_cmd_control { 392 struct task_struct *f2fs_issue_discard; /* discard thread */ 393 struct list_head entry_list; /* 4KB discard entry list */ 394 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 395 struct list_head wait_list; /* store on-flushing entries */ 396 struct list_head fstrim_list; /* in-flight discard from fstrim */ 397 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 398 unsigned int discard_wake; /* to wake up discard thread */ 399 struct mutex cmd_lock; 400 unsigned int nr_discards; /* # of discards in the list */ 401 unsigned int max_discards; /* max. discards to be issued */ 402 unsigned int max_discard_request; /* max. discard request per round */ 403 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 404 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 405 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 406 unsigned int discard_granularity; /* discard granularity */ 407 unsigned int undiscard_blks; /* # of undiscard blocks */ 408 unsigned int next_pos; /* next discard position */ 409 atomic_t issued_discard; /* # of issued discard */ 410 atomic_t queued_discard; /* # of queued discard */ 411 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 412 struct rb_root_cached root; /* root of discard rb-tree */ 413 bool rbtree_check; /* config for consistence check */ 414 }; 415 416 /* for the list of fsync inodes, used only during recovery */ 417 struct fsync_inode_entry { 418 struct list_head list; /* list head */ 419 struct inode *inode; /* vfs inode pointer */ 420 block_t blkaddr; /* block address locating the last fsync */ 421 block_t last_dentry; /* block address locating the last dentry */ 422 }; 423 424 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 425 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 426 427 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 428 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 429 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 430 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 431 432 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 433 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 434 435 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 436 { 437 int before = nats_in_cursum(journal); 438 439 journal->n_nats = cpu_to_le16(before + i); 440 return before; 441 } 442 443 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 444 { 445 int before = sits_in_cursum(journal); 446 447 journal->n_sits = cpu_to_le16(before + i); 448 return before; 449 } 450 451 static inline bool __has_cursum_space(struct f2fs_journal *journal, 452 int size, int type) 453 { 454 if (type == NAT_JOURNAL) 455 return size <= MAX_NAT_JENTRIES(journal); 456 return size <= MAX_SIT_JENTRIES(journal); 457 } 458 459 /* for inline stuff */ 460 #define DEF_INLINE_RESERVED_SIZE 1 461 static inline int get_extra_isize(struct inode *inode); 462 static inline int get_inline_xattr_addrs(struct inode *inode); 463 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 464 (CUR_ADDRS_PER_INODE(inode) - \ 465 get_inline_xattr_addrs(inode) - \ 466 DEF_INLINE_RESERVED_SIZE)) 467 468 /* for inline dir */ 469 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 470 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 471 BITS_PER_BYTE + 1)) 472 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 473 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 474 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 475 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 476 NR_INLINE_DENTRY(inode) + \ 477 INLINE_DENTRY_BITMAP_SIZE(inode))) 478 479 /* 480 * For INODE and NODE manager 481 */ 482 /* for directory operations */ 483 484 struct f2fs_filename { 485 /* 486 * The filename the user specified. This is NULL for some 487 * filesystem-internal operations, e.g. converting an inline directory 488 * to a non-inline one, or roll-forward recovering an encrypted dentry. 489 */ 490 const struct qstr *usr_fname; 491 492 /* 493 * The on-disk filename. For encrypted directories, this is encrypted. 494 * This may be NULL for lookups in an encrypted dir without the key. 495 */ 496 struct fscrypt_str disk_name; 497 498 /* The dirhash of this filename */ 499 f2fs_hash_t hash; 500 501 #ifdef CONFIG_FS_ENCRYPTION 502 /* 503 * For lookups in encrypted directories: either the buffer backing 504 * disk_name, or a buffer that holds the decoded no-key name. 505 */ 506 struct fscrypt_str crypto_buf; 507 #endif 508 #if IS_ENABLED(CONFIG_UNICODE) 509 /* 510 * For casefolded directories: the casefolded name, but it's left NULL 511 * if the original name is not valid Unicode, if the original name is 512 * "." or "..", if the directory is both casefolded and encrypted and 513 * its encryption key is unavailable, or if the filesystem is doing an 514 * internal operation where usr_fname is also NULL. In all these cases 515 * we fall back to treating the name as an opaque byte sequence. 516 */ 517 struct fscrypt_str cf_name; 518 #endif 519 }; 520 521 struct f2fs_dentry_ptr { 522 struct inode *inode; 523 void *bitmap; 524 struct f2fs_dir_entry *dentry; 525 __u8 (*filename)[F2FS_SLOT_LEN]; 526 int max; 527 int nr_bitmap; 528 }; 529 530 static inline void make_dentry_ptr_block(struct inode *inode, 531 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 532 { 533 d->inode = inode; 534 d->max = NR_DENTRY_IN_BLOCK; 535 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 536 d->bitmap = t->dentry_bitmap; 537 d->dentry = t->dentry; 538 d->filename = t->filename; 539 } 540 541 static inline void make_dentry_ptr_inline(struct inode *inode, 542 struct f2fs_dentry_ptr *d, void *t) 543 { 544 int entry_cnt = NR_INLINE_DENTRY(inode); 545 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 546 int reserved_size = INLINE_RESERVED_SIZE(inode); 547 548 d->inode = inode; 549 d->max = entry_cnt; 550 d->nr_bitmap = bitmap_size; 551 d->bitmap = t; 552 d->dentry = t + bitmap_size + reserved_size; 553 d->filename = t + bitmap_size + reserved_size + 554 SIZE_OF_DIR_ENTRY * entry_cnt; 555 } 556 557 /* 558 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 559 * as its node offset to distinguish from index node blocks. 560 * But some bits are used to mark the node block. 561 */ 562 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 563 >> OFFSET_BIT_SHIFT) 564 enum { 565 ALLOC_NODE, /* allocate a new node page if needed */ 566 LOOKUP_NODE, /* look up a node without readahead */ 567 LOOKUP_NODE_RA, /* 568 * look up a node with readahead called 569 * by get_data_block. 570 */ 571 }; 572 573 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 574 575 /* congestion wait timeout value, default: 20ms */ 576 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 577 578 /* maximum retry quota flush count */ 579 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 580 581 /* maximum retry of EIO'ed page */ 582 #define MAX_RETRY_PAGE_EIO 100 583 584 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 585 586 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 587 588 /* dirty segments threshold for triggering CP */ 589 #define DEFAULT_DIRTY_THRESHOLD 4 590 591 /* for in-memory extent cache entry */ 592 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 593 594 /* number of extent info in extent cache we try to shrink */ 595 #define EXTENT_CACHE_SHRINK_NUMBER 128 596 597 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 598 #define RECOVERY_MIN_RA_BLOCKS 1 599 600 struct rb_entry { 601 struct rb_node rb_node; /* rb node located in rb-tree */ 602 union { 603 struct { 604 unsigned int ofs; /* start offset of the entry */ 605 unsigned int len; /* length of the entry */ 606 }; 607 unsigned long long key; /* 64-bits key */ 608 } __packed; 609 }; 610 611 struct extent_info { 612 unsigned int fofs; /* start offset in a file */ 613 unsigned int len; /* length of the extent */ 614 u32 blk; /* start block address of the extent */ 615 #ifdef CONFIG_F2FS_FS_COMPRESSION 616 unsigned int c_len; /* physical extent length of compressed blocks */ 617 #endif 618 }; 619 620 struct extent_node { 621 struct rb_node rb_node; /* rb node located in rb-tree */ 622 struct extent_info ei; /* extent info */ 623 struct list_head list; /* node in global extent list of sbi */ 624 struct extent_tree *et; /* extent tree pointer */ 625 }; 626 627 struct extent_tree { 628 nid_t ino; /* inode number */ 629 struct rb_root_cached root; /* root of extent info rb-tree */ 630 struct extent_node *cached_en; /* recently accessed extent node */ 631 struct extent_info largest; /* largested extent info */ 632 struct list_head list; /* to be used by sbi->zombie_list */ 633 rwlock_t lock; /* protect extent info rb-tree */ 634 atomic_t node_cnt; /* # of extent node in rb-tree*/ 635 bool largest_updated; /* largest extent updated */ 636 }; 637 638 /* 639 * This structure is taken from ext4_map_blocks. 640 * 641 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 642 */ 643 #define F2FS_MAP_NEW (1 << BH_New) 644 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 645 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 646 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 647 F2FS_MAP_UNWRITTEN) 648 649 struct f2fs_map_blocks { 650 struct block_device *m_bdev; /* for multi-device dio */ 651 block_t m_pblk; 652 block_t m_lblk; 653 unsigned int m_len; 654 unsigned int m_flags; 655 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 656 pgoff_t *m_next_extent; /* point to next possible extent */ 657 int m_seg_type; 658 bool m_may_create; /* indicate it is from write path */ 659 bool m_multidev_dio; /* indicate it allows multi-device dio */ 660 }; 661 662 /* for flag in get_data_block */ 663 enum { 664 F2FS_GET_BLOCK_DEFAULT, 665 F2FS_GET_BLOCK_FIEMAP, 666 F2FS_GET_BLOCK_BMAP, 667 F2FS_GET_BLOCK_DIO, 668 F2FS_GET_BLOCK_PRE_DIO, 669 F2FS_GET_BLOCK_PRE_AIO, 670 F2FS_GET_BLOCK_PRECACHE, 671 }; 672 673 /* 674 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 675 */ 676 #define FADVISE_COLD_BIT 0x01 677 #define FADVISE_LOST_PINO_BIT 0x02 678 #define FADVISE_ENCRYPT_BIT 0x04 679 #define FADVISE_ENC_NAME_BIT 0x08 680 #define FADVISE_KEEP_SIZE_BIT 0x10 681 #define FADVISE_HOT_BIT 0x20 682 #define FADVISE_VERITY_BIT 0x40 683 #define FADVISE_TRUNC_BIT 0x80 684 685 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 686 687 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 688 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 689 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 690 691 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 692 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 694 695 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 696 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 697 698 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 699 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 700 701 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 702 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 703 704 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 705 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 706 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 707 708 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 709 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 710 711 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 712 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 713 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 714 715 #define DEF_DIR_LEVEL 0 716 717 enum { 718 GC_FAILURE_PIN, 719 MAX_GC_FAILURE 720 }; 721 722 /* used for f2fs_inode_info->flags */ 723 enum { 724 FI_NEW_INODE, /* indicate newly allocated inode */ 725 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 726 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 727 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 728 FI_INC_LINK, /* need to increment i_nlink */ 729 FI_ACL_MODE, /* indicate acl mode */ 730 FI_NO_ALLOC, /* should not allocate any blocks */ 731 FI_FREE_NID, /* free allocated nide */ 732 FI_NO_EXTENT, /* not to use the extent cache */ 733 FI_INLINE_XATTR, /* used for inline xattr */ 734 FI_INLINE_DATA, /* used for inline data*/ 735 FI_INLINE_DENTRY, /* used for inline dentry */ 736 FI_APPEND_WRITE, /* inode has appended data */ 737 FI_UPDATE_WRITE, /* inode has in-place-update data */ 738 FI_NEED_IPU, /* used for ipu per file */ 739 FI_ATOMIC_FILE, /* indicate atomic file */ 740 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 741 FI_DROP_CACHE, /* drop dirty page cache */ 742 FI_DATA_EXIST, /* indicate data exists */ 743 FI_INLINE_DOTS, /* indicate inline dot dentries */ 744 FI_SKIP_WRITES, /* should skip data page writeback */ 745 FI_OPU_WRITE, /* used for opu per file */ 746 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 747 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 748 FI_HOT_DATA, /* indicate file is hot */ 749 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 750 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 751 FI_PIN_FILE, /* indicate file should not be gced */ 752 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 753 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 754 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 755 FI_MMAP_FILE, /* indicate file was mmapped */ 756 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 757 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 758 FI_ALIGNED_WRITE, /* enable aligned write */ 759 FI_MAX, /* max flag, never be used */ 760 }; 761 762 struct f2fs_inode_info { 763 struct inode vfs_inode; /* serve a vfs inode */ 764 unsigned long i_flags; /* keep an inode flags for ioctl */ 765 unsigned char i_advise; /* use to give file attribute hints */ 766 unsigned char i_dir_level; /* use for dentry level for large dir */ 767 unsigned int i_current_depth; /* only for directory depth */ 768 /* for gc failure statistic */ 769 unsigned int i_gc_failures[MAX_GC_FAILURE]; 770 unsigned int i_pino; /* parent inode number */ 771 umode_t i_acl_mode; /* keep file acl mode temporarily */ 772 773 /* Use below internally in f2fs*/ 774 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 775 struct f2fs_rwsem i_sem; /* protect fi info */ 776 atomic_t dirty_pages; /* # of dirty pages */ 777 f2fs_hash_t chash; /* hash value of given file name */ 778 unsigned int clevel; /* maximum level of given file name */ 779 struct task_struct *task; /* lookup and create consistency */ 780 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 781 nid_t i_xattr_nid; /* node id that contains xattrs */ 782 loff_t last_disk_size; /* lastly written file size */ 783 spinlock_t i_size_lock; /* protect last_disk_size */ 784 785 #ifdef CONFIG_QUOTA 786 struct dquot *i_dquot[MAXQUOTAS]; 787 788 /* quota space reservation, managed internally by quota code */ 789 qsize_t i_reserved_quota; 790 #endif 791 struct list_head dirty_list; /* dirty list for dirs and files */ 792 struct list_head gdirty_list; /* linked in global dirty list */ 793 struct task_struct *atomic_write_task; /* store atomic write task */ 794 struct extent_tree *extent_tree; /* cached extent_tree entry */ 795 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 796 797 /* avoid racing between foreground op and gc */ 798 struct f2fs_rwsem i_gc_rwsem[2]; 799 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 800 801 int i_extra_isize; /* size of extra space located in i_addr */ 802 kprojid_t i_projid; /* id for project quota */ 803 int i_inline_xattr_size; /* inline xattr size */ 804 struct timespec64 i_crtime; /* inode creation time */ 805 struct timespec64 i_disk_time[4];/* inode disk times */ 806 807 /* for file compress */ 808 atomic_t i_compr_blocks; /* # of compressed blocks */ 809 unsigned char i_compress_algorithm; /* algorithm type */ 810 unsigned char i_log_cluster_size; /* log of cluster size */ 811 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 812 unsigned short i_compress_flag; /* compress flag */ 813 unsigned int i_cluster_size; /* cluster size */ 814 }; 815 816 static inline void get_extent_info(struct extent_info *ext, 817 struct f2fs_extent *i_ext) 818 { 819 ext->fofs = le32_to_cpu(i_ext->fofs); 820 ext->blk = le32_to_cpu(i_ext->blk); 821 ext->len = le32_to_cpu(i_ext->len); 822 } 823 824 static inline void set_raw_extent(struct extent_info *ext, 825 struct f2fs_extent *i_ext) 826 { 827 i_ext->fofs = cpu_to_le32(ext->fofs); 828 i_ext->blk = cpu_to_le32(ext->blk); 829 i_ext->len = cpu_to_le32(ext->len); 830 } 831 832 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 833 u32 blk, unsigned int len) 834 { 835 ei->fofs = fofs; 836 ei->blk = blk; 837 ei->len = len; 838 #ifdef CONFIG_F2FS_FS_COMPRESSION 839 ei->c_len = 0; 840 #endif 841 } 842 843 static inline bool __is_discard_mergeable(struct discard_info *back, 844 struct discard_info *front, unsigned int max_len) 845 { 846 return (back->lstart + back->len == front->lstart) && 847 (back->len + front->len <= max_len); 848 } 849 850 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 851 struct discard_info *back, unsigned int max_len) 852 { 853 return __is_discard_mergeable(back, cur, max_len); 854 } 855 856 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 857 struct discard_info *front, unsigned int max_len) 858 { 859 return __is_discard_mergeable(cur, front, max_len); 860 } 861 862 static inline bool __is_extent_mergeable(struct extent_info *back, 863 struct extent_info *front) 864 { 865 #ifdef CONFIG_F2FS_FS_COMPRESSION 866 if (back->c_len && back->len != back->c_len) 867 return false; 868 if (front->c_len && front->len != front->c_len) 869 return false; 870 #endif 871 return (back->fofs + back->len == front->fofs && 872 back->blk + back->len == front->blk); 873 } 874 875 static inline bool __is_back_mergeable(struct extent_info *cur, 876 struct extent_info *back) 877 { 878 return __is_extent_mergeable(back, cur); 879 } 880 881 static inline bool __is_front_mergeable(struct extent_info *cur, 882 struct extent_info *front) 883 { 884 return __is_extent_mergeable(cur, front); 885 } 886 887 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 888 static inline void __try_update_largest_extent(struct extent_tree *et, 889 struct extent_node *en) 890 { 891 if (en->ei.len > et->largest.len) { 892 et->largest = en->ei; 893 et->largest_updated = true; 894 } 895 } 896 897 /* 898 * For free nid management 899 */ 900 enum nid_state { 901 FREE_NID, /* newly added to free nid list */ 902 PREALLOC_NID, /* it is preallocated */ 903 MAX_NID_STATE, 904 }; 905 906 enum nat_state { 907 TOTAL_NAT, 908 DIRTY_NAT, 909 RECLAIMABLE_NAT, 910 MAX_NAT_STATE, 911 }; 912 913 struct f2fs_nm_info { 914 block_t nat_blkaddr; /* base disk address of NAT */ 915 nid_t max_nid; /* maximum possible node ids */ 916 nid_t available_nids; /* # of available node ids */ 917 nid_t next_scan_nid; /* the next nid to be scanned */ 918 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 919 unsigned int ram_thresh; /* control the memory footprint */ 920 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 921 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 922 923 /* NAT cache management */ 924 struct radix_tree_root nat_root;/* root of the nat entry cache */ 925 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 926 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 927 struct list_head nat_entries; /* cached nat entry list (clean) */ 928 spinlock_t nat_list_lock; /* protect clean nat entry list */ 929 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 930 unsigned int nat_blocks; /* # of nat blocks */ 931 932 /* free node ids management */ 933 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 934 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 935 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 936 spinlock_t nid_list_lock; /* protect nid lists ops */ 937 struct mutex build_lock; /* lock for build free nids */ 938 unsigned char **free_nid_bitmap; 939 unsigned char *nat_block_bitmap; 940 unsigned short *free_nid_count; /* free nid count of NAT block */ 941 942 /* for checkpoint */ 943 char *nat_bitmap; /* NAT bitmap pointer */ 944 945 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 946 unsigned char *nat_bits; /* NAT bits blocks */ 947 unsigned char *full_nat_bits; /* full NAT pages */ 948 unsigned char *empty_nat_bits; /* empty NAT pages */ 949 #ifdef CONFIG_F2FS_CHECK_FS 950 char *nat_bitmap_mir; /* NAT bitmap mirror */ 951 #endif 952 int bitmap_size; /* bitmap size */ 953 }; 954 955 /* 956 * this structure is used as one of function parameters. 957 * all the information are dedicated to a given direct node block determined 958 * by the data offset in a file. 959 */ 960 struct dnode_of_data { 961 struct inode *inode; /* vfs inode pointer */ 962 struct page *inode_page; /* its inode page, NULL is possible */ 963 struct page *node_page; /* cached direct node page */ 964 nid_t nid; /* node id of the direct node block */ 965 unsigned int ofs_in_node; /* data offset in the node page */ 966 bool inode_page_locked; /* inode page is locked or not */ 967 bool node_changed; /* is node block changed */ 968 char cur_level; /* level of hole node page */ 969 char max_level; /* level of current page located */ 970 block_t data_blkaddr; /* block address of the node block */ 971 }; 972 973 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 974 struct page *ipage, struct page *npage, nid_t nid) 975 { 976 memset(dn, 0, sizeof(*dn)); 977 dn->inode = inode; 978 dn->inode_page = ipage; 979 dn->node_page = npage; 980 dn->nid = nid; 981 } 982 983 /* 984 * For SIT manager 985 * 986 * By default, there are 6 active log areas across the whole main area. 987 * When considering hot and cold data separation to reduce cleaning overhead, 988 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 989 * respectively. 990 * In the current design, you should not change the numbers intentionally. 991 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 992 * logs individually according to the underlying devices. (default: 6) 993 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 994 * data and 8 for node logs. 995 */ 996 #define NR_CURSEG_DATA_TYPE (3) 997 #define NR_CURSEG_NODE_TYPE (3) 998 #define NR_CURSEG_INMEM_TYPE (2) 999 #define NR_CURSEG_RO_TYPE (2) 1000 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1001 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1002 1003 enum { 1004 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1005 CURSEG_WARM_DATA, /* data blocks */ 1006 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1007 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1008 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1009 CURSEG_COLD_NODE, /* indirect node blocks */ 1010 NR_PERSISTENT_LOG, /* number of persistent log */ 1011 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1012 /* pinned file that needs consecutive block address */ 1013 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1014 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1015 }; 1016 1017 struct flush_cmd { 1018 struct completion wait; 1019 struct llist_node llnode; 1020 nid_t ino; 1021 int ret; 1022 }; 1023 1024 struct flush_cmd_control { 1025 struct task_struct *f2fs_issue_flush; /* flush thread */ 1026 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1027 atomic_t issued_flush; /* # of issued flushes */ 1028 atomic_t queued_flush; /* # of queued flushes */ 1029 struct llist_head issue_list; /* list for command issue */ 1030 struct llist_node *dispatch_list; /* list for command dispatch */ 1031 }; 1032 1033 struct f2fs_sm_info { 1034 struct sit_info *sit_info; /* whole segment information */ 1035 struct free_segmap_info *free_info; /* free segment information */ 1036 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1037 struct curseg_info *curseg_array; /* active segment information */ 1038 1039 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1040 1041 block_t seg0_blkaddr; /* block address of 0'th segment */ 1042 block_t main_blkaddr; /* start block address of main area */ 1043 block_t ssa_blkaddr; /* start block address of SSA area */ 1044 1045 unsigned int segment_count; /* total # of segments */ 1046 unsigned int main_segments; /* # of segments in main area */ 1047 unsigned int reserved_segments; /* # of reserved segments */ 1048 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1049 unsigned int ovp_segments; /* # of overprovision segments */ 1050 1051 /* a threshold to reclaim prefree segments */ 1052 unsigned int rec_prefree_segments; 1053 1054 /* for batched trimming */ 1055 unsigned int trim_sections; /* # of sections to trim */ 1056 1057 struct list_head sit_entry_set; /* sit entry set list */ 1058 1059 unsigned int ipu_policy; /* in-place-update policy */ 1060 unsigned int min_ipu_util; /* in-place-update threshold */ 1061 unsigned int min_fsync_blocks; /* threshold for fsync */ 1062 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1063 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1064 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1065 1066 /* for flush command control */ 1067 struct flush_cmd_control *fcc_info; 1068 1069 /* for discard command control */ 1070 struct discard_cmd_control *dcc_info; 1071 }; 1072 1073 /* 1074 * For superblock 1075 */ 1076 /* 1077 * COUNT_TYPE for monitoring 1078 * 1079 * f2fs monitors the number of several block types such as on-writeback, 1080 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1081 */ 1082 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1083 enum count_type { 1084 F2FS_DIRTY_DENTS, 1085 F2FS_DIRTY_DATA, 1086 F2FS_DIRTY_QDATA, 1087 F2FS_DIRTY_NODES, 1088 F2FS_DIRTY_META, 1089 F2FS_DIRTY_IMETA, 1090 F2FS_WB_CP_DATA, 1091 F2FS_WB_DATA, 1092 F2FS_RD_DATA, 1093 F2FS_RD_NODE, 1094 F2FS_RD_META, 1095 F2FS_DIO_WRITE, 1096 F2FS_DIO_READ, 1097 NR_COUNT_TYPE, 1098 }; 1099 1100 /* 1101 * The below are the page types of bios used in submit_bio(). 1102 * The available types are: 1103 * DATA User data pages. It operates as async mode. 1104 * NODE Node pages. It operates as async mode. 1105 * META FS metadata pages such as SIT, NAT, CP. 1106 * NR_PAGE_TYPE The number of page types. 1107 * META_FLUSH Make sure the previous pages are written 1108 * with waiting the bio's completion 1109 * ... Only can be used with META. 1110 */ 1111 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1112 enum page_type { 1113 DATA = 0, 1114 NODE = 1, /* should not change this */ 1115 META, 1116 NR_PAGE_TYPE, 1117 META_FLUSH, 1118 IPU, /* the below types are used by tracepoints only. */ 1119 OPU, 1120 }; 1121 1122 enum temp_type { 1123 HOT = 0, /* must be zero for meta bio */ 1124 WARM, 1125 COLD, 1126 NR_TEMP_TYPE, 1127 }; 1128 1129 enum need_lock_type { 1130 LOCK_REQ = 0, 1131 LOCK_DONE, 1132 LOCK_RETRY, 1133 }; 1134 1135 enum cp_reason_type { 1136 CP_NO_NEEDED, 1137 CP_NON_REGULAR, 1138 CP_COMPRESSED, 1139 CP_HARDLINK, 1140 CP_SB_NEED_CP, 1141 CP_WRONG_PINO, 1142 CP_NO_SPC_ROLL, 1143 CP_NODE_NEED_CP, 1144 CP_FASTBOOT_MODE, 1145 CP_SPEC_LOG_NUM, 1146 CP_RECOVER_DIR, 1147 }; 1148 1149 enum iostat_type { 1150 /* WRITE IO */ 1151 APP_DIRECT_IO, /* app direct write IOs */ 1152 APP_BUFFERED_IO, /* app buffered write IOs */ 1153 APP_WRITE_IO, /* app write IOs */ 1154 APP_MAPPED_IO, /* app mapped IOs */ 1155 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1156 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1157 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1158 FS_GC_DATA_IO, /* data IOs from forground gc */ 1159 FS_GC_NODE_IO, /* node IOs from forground gc */ 1160 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1161 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1162 FS_CP_META_IO, /* meta IOs from checkpoint */ 1163 1164 /* READ IO */ 1165 APP_DIRECT_READ_IO, /* app direct read IOs */ 1166 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1167 APP_READ_IO, /* app read IOs */ 1168 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1169 FS_DATA_READ_IO, /* data read IOs */ 1170 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1171 FS_CDATA_READ_IO, /* compressed data read IOs */ 1172 FS_NODE_READ_IO, /* node read IOs */ 1173 FS_META_READ_IO, /* meta read IOs */ 1174 1175 /* other */ 1176 FS_DISCARD, /* discard */ 1177 NR_IO_TYPE, 1178 }; 1179 1180 struct f2fs_io_info { 1181 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1182 nid_t ino; /* inode number */ 1183 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1184 enum temp_type temp; /* contains HOT/WARM/COLD */ 1185 int op; /* contains REQ_OP_ */ 1186 int op_flags; /* req_flag_bits */ 1187 block_t new_blkaddr; /* new block address to be written */ 1188 block_t old_blkaddr; /* old block address before Cow */ 1189 struct page *page; /* page to be written */ 1190 struct page *encrypted_page; /* encrypted page */ 1191 struct page *compressed_page; /* compressed page */ 1192 struct list_head list; /* serialize IOs */ 1193 bool submitted; /* indicate IO submission */ 1194 int need_lock; /* indicate we need to lock cp_rwsem */ 1195 bool in_list; /* indicate fio is in io_list */ 1196 bool is_por; /* indicate IO is from recovery or not */ 1197 bool retry; /* need to reallocate block address */ 1198 int compr_blocks; /* # of compressed block addresses */ 1199 bool encrypted; /* indicate file is encrypted */ 1200 enum iostat_type io_type; /* io type */ 1201 struct writeback_control *io_wbc; /* writeback control */ 1202 struct bio **bio; /* bio for ipu */ 1203 sector_t *last_block; /* last block number in bio */ 1204 unsigned char version; /* version of the node */ 1205 }; 1206 1207 struct bio_entry { 1208 struct bio *bio; 1209 struct list_head list; 1210 }; 1211 1212 #define is_read_io(rw) ((rw) == READ) 1213 struct f2fs_bio_info { 1214 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1215 struct bio *bio; /* bios to merge */ 1216 sector_t last_block_in_bio; /* last block number */ 1217 struct f2fs_io_info fio; /* store buffered io info. */ 1218 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1219 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1220 struct list_head io_list; /* track fios */ 1221 struct list_head bio_list; /* bio entry list head */ 1222 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1223 }; 1224 1225 #define FDEV(i) (sbi->devs[i]) 1226 #define RDEV(i) (raw_super->devs[i]) 1227 struct f2fs_dev_info { 1228 struct block_device *bdev; 1229 char path[MAX_PATH_LEN]; 1230 unsigned int total_segments; 1231 block_t start_blk; 1232 block_t end_blk; 1233 #ifdef CONFIG_BLK_DEV_ZONED 1234 unsigned int nr_blkz; /* Total number of zones */ 1235 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1236 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1237 #endif 1238 }; 1239 1240 enum inode_type { 1241 DIR_INODE, /* for dirty dir inode */ 1242 FILE_INODE, /* for dirty regular/symlink inode */ 1243 DIRTY_META, /* for all dirtied inode metadata */ 1244 ATOMIC_FILE, /* for all atomic files */ 1245 NR_INODE_TYPE, 1246 }; 1247 1248 /* for inner inode cache management */ 1249 struct inode_management { 1250 struct radix_tree_root ino_root; /* ino entry array */ 1251 spinlock_t ino_lock; /* for ino entry lock */ 1252 struct list_head ino_list; /* inode list head */ 1253 unsigned long ino_num; /* number of entries */ 1254 }; 1255 1256 /* for GC_AT */ 1257 struct atgc_management { 1258 bool atgc_enabled; /* ATGC is enabled or not */ 1259 struct rb_root_cached root; /* root of victim rb-tree */ 1260 struct list_head victim_list; /* linked with all victim entries */ 1261 unsigned int victim_count; /* victim count in rb-tree */ 1262 unsigned int candidate_ratio; /* candidate ratio */ 1263 unsigned int max_candidate_count; /* max candidate count */ 1264 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1265 unsigned long long age_threshold; /* age threshold */ 1266 }; 1267 1268 struct f2fs_gc_control { 1269 unsigned int victim_segno; /* target victim segment number */ 1270 int init_gc_type; /* FG_GC or BG_GC */ 1271 bool no_bg_gc; /* check the space and stop bg_gc */ 1272 bool should_migrate_blocks; /* should migrate blocks */ 1273 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1274 unsigned int nr_free_secs; /* # of free sections to do GC */ 1275 }; 1276 1277 /* For s_flag in struct f2fs_sb_info */ 1278 enum { 1279 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1280 SBI_IS_CLOSE, /* specify unmounting */ 1281 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1282 SBI_POR_DOING, /* recovery is doing or not */ 1283 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1284 SBI_NEED_CP, /* need to checkpoint */ 1285 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1286 SBI_IS_RECOVERED, /* recovered orphan/data */ 1287 SBI_CP_DISABLED, /* CP was disabled last mount */ 1288 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1289 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1290 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1291 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1292 SBI_IS_RESIZEFS, /* resizefs is in process */ 1293 SBI_IS_FREEZING, /* freezefs is in process */ 1294 }; 1295 1296 enum { 1297 CP_TIME, 1298 REQ_TIME, 1299 DISCARD_TIME, 1300 GC_TIME, 1301 DISABLE_TIME, 1302 UMOUNT_DISCARD_TIMEOUT, 1303 MAX_TIME, 1304 }; 1305 1306 enum { 1307 GC_NORMAL, 1308 GC_IDLE_CB, 1309 GC_IDLE_GREEDY, 1310 GC_IDLE_AT, 1311 GC_URGENT_HIGH, 1312 GC_URGENT_LOW, 1313 GC_URGENT_MID, 1314 MAX_GC_MODE, 1315 }; 1316 1317 enum { 1318 BGGC_MODE_ON, /* background gc is on */ 1319 BGGC_MODE_OFF, /* background gc is off */ 1320 BGGC_MODE_SYNC, /* 1321 * background gc is on, migrating blocks 1322 * like foreground gc 1323 */ 1324 }; 1325 1326 enum { 1327 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1328 FS_MODE_LFS, /* use lfs allocation only */ 1329 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1330 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1331 }; 1332 1333 enum { 1334 ALLOC_MODE_DEFAULT, /* stay default */ 1335 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1336 }; 1337 1338 enum fsync_mode { 1339 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1340 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1341 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1342 }; 1343 1344 enum { 1345 COMPR_MODE_FS, /* 1346 * automatically compress compression 1347 * enabled files 1348 */ 1349 COMPR_MODE_USER, /* 1350 * automatical compression is disabled. 1351 * user can control the file compression 1352 * using ioctls 1353 */ 1354 }; 1355 1356 enum { 1357 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1358 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1359 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1360 }; 1361 1362 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1363 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1364 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1365 1366 /* 1367 * Layout of f2fs page.private: 1368 * 1369 * Layout A: lowest bit should be 1 1370 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1371 * bit 0 PAGE_PRIVATE_NOT_POINTER 1372 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1373 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1374 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1375 * bit 4 PAGE_PRIVATE_INLINE_INODE 1376 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1377 * bit 6- f2fs private data 1378 * 1379 * Layout B: lowest bit should be 0 1380 * page.private is a wrapped pointer. 1381 */ 1382 enum { 1383 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1384 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1385 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1386 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1387 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1388 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1389 PAGE_PRIVATE_MAX 1390 }; 1391 1392 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1393 static inline bool page_private_##name(struct page *page) \ 1394 { \ 1395 return PagePrivate(page) && \ 1396 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1397 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1398 } 1399 1400 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1401 static inline void set_page_private_##name(struct page *page) \ 1402 { \ 1403 if (!PagePrivate(page)) { \ 1404 get_page(page); \ 1405 SetPagePrivate(page); \ 1406 set_page_private(page, 0); \ 1407 } \ 1408 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1409 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1410 } 1411 1412 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1413 static inline void clear_page_private_##name(struct page *page) \ 1414 { \ 1415 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1416 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1417 set_page_private(page, 0); \ 1418 if (PagePrivate(page)) { \ 1419 ClearPagePrivate(page); \ 1420 put_page(page); \ 1421 }\ 1422 } \ 1423 } 1424 1425 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1426 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1427 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1428 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1429 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1430 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1431 1432 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1433 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1434 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1435 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1436 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1437 1438 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1439 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1440 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1441 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1442 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1443 1444 static inline unsigned long get_page_private_data(struct page *page) 1445 { 1446 unsigned long data = page_private(page); 1447 1448 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1449 return 0; 1450 return data >> PAGE_PRIVATE_MAX; 1451 } 1452 1453 static inline void set_page_private_data(struct page *page, unsigned long data) 1454 { 1455 if (!PagePrivate(page)) { 1456 get_page(page); 1457 SetPagePrivate(page); 1458 set_page_private(page, 0); 1459 } 1460 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1461 page_private(page) |= data << PAGE_PRIVATE_MAX; 1462 } 1463 1464 static inline void clear_page_private_data(struct page *page) 1465 { 1466 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1467 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1468 set_page_private(page, 0); 1469 if (PagePrivate(page)) { 1470 ClearPagePrivate(page); 1471 put_page(page); 1472 } 1473 } 1474 } 1475 1476 /* For compression */ 1477 enum compress_algorithm_type { 1478 COMPRESS_LZO, 1479 COMPRESS_LZ4, 1480 COMPRESS_ZSTD, 1481 COMPRESS_LZORLE, 1482 COMPRESS_MAX, 1483 }; 1484 1485 enum compress_flag { 1486 COMPRESS_CHKSUM, 1487 COMPRESS_MAX_FLAG, 1488 }; 1489 1490 #define COMPRESS_WATERMARK 20 1491 #define COMPRESS_PERCENT 20 1492 1493 #define COMPRESS_DATA_RESERVED_SIZE 4 1494 struct compress_data { 1495 __le32 clen; /* compressed data size */ 1496 __le32 chksum; /* compressed data chksum */ 1497 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1498 u8 cdata[]; /* compressed data */ 1499 }; 1500 1501 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1502 1503 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1504 1505 #define COMPRESS_LEVEL_OFFSET 8 1506 1507 /* compress context */ 1508 struct compress_ctx { 1509 struct inode *inode; /* inode the context belong to */ 1510 pgoff_t cluster_idx; /* cluster index number */ 1511 unsigned int cluster_size; /* page count in cluster */ 1512 unsigned int log_cluster_size; /* log of cluster size */ 1513 struct page **rpages; /* pages store raw data in cluster */ 1514 unsigned int nr_rpages; /* total page number in rpages */ 1515 struct page **cpages; /* pages store compressed data in cluster */ 1516 unsigned int nr_cpages; /* total page number in cpages */ 1517 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1518 void *rbuf; /* virtual mapped address on rpages */ 1519 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1520 size_t rlen; /* valid data length in rbuf */ 1521 size_t clen; /* valid data length in cbuf */ 1522 void *private; /* payload buffer for specified compression algorithm */ 1523 void *private2; /* extra payload buffer */ 1524 }; 1525 1526 /* compress context for write IO path */ 1527 struct compress_io_ctx { 1528 u32 magic; /* magic number to indicate page is compressed */ 1529 struct inode *inode; /* inode the context belong to */ 1530 struct page **rpages; /* pages store raw data in cluster */ 1531 unsigned int nr_rpages; /* total page number in rpages */ 1532 atomic_t pending_pages; /* in-flight compressed page count */ 1533 }; 1534 1535 /* Context for decompressing one cluster on the read IO path */ 1536 struct decompress_io_ctx { 1537 u32 magic; /* magic number to indicate page is compressed */ 1538 struct inode *inode; /* inode the context belong to */ 1539 pgoff_t cluster_idx; /* cluster index number */ 1540 unsigned int cluster_size; /* page count in cluster */ 1541 unsigned int log_cluster_size; /* log of cluster size */ 1542 struct page **rpages; /* pages store raw data in cluster */ 1543 unsigned int nr_rpages; /* total page number in rpages */ 1544 struct page **cpages; /* pages store compressed data in cluster */ 1545 unsigned int nr_cpages; /* total page number in cpages */ 1546 struct page **tpages; /* temp pages to pad holes in cluster */ 1547 void *rbuf; /* virtual mapped address on rpages */ 1548 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1549 size_t rlen; /* valid data length in rbuf */ 1550 size_t clen; /* valid data length in cbuf */ 1551 1552 /* 1553 * The number of compressed pages remaining to be read in this cluster. 1554 * This is initially nr_cpages. It is decremented by 1 each time a page 1555 * has been read (or failed to be read). When it reaches 0, the cluster 1556 * is decompressed (or an error is reported). 1557 * 1558 * If an error occurs before all the pages have been submitted for I/O, 1559 * then this will never reach 0. In this case the I/O submitter is 1560 * responsible for calling f2fs_decompress_end_io() instead. 1561 */ 1562 atomic_t remaining_pages; 1563 1564 /* 1565 * Number of references to this decompress_io_ctx. 1566 * 1567 * One reference is held for I/O completion. This reference is dropped 1568 * after the pagecache pages are updated and unlocked -- either after 1569 * decompression (and verity if enabled), or after an error. 1570 * 1571 * In addition, each compressed page holds a reference while it is in a 1572 * bio. These references are necessary prevent compressed pages from 1573 * being freed while they are still in a bio. 1574 */ 1575 refcount_t refcnt; 1576 1577 bool failed; /* IO error occurred before decompression? */ 1578 bool need_verity; /* need fs-verity verification after decompression? */ 1579 void *private; /* payload buffer for specified decompression algorithm */ 1580 void *private2; /* extra payload buffer */ 1581 struct work_struct verity_work; /* work to verify the decompressed pages */ 1582 }; 1583 1584 #define NULL_CLUSTER ((unsigned int)(~0)) 1585 #define MIN_COMPRESS_LOG_SIZE 2 1586 #define MAX_COMPRESS_LOG_SIZE 8 1587 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1588 1589 struct f2fs_sb_info { 1590 struct super_block *sb; /* pointer to VFS super block */ 1591 struct proc_dir_entry *s_proc; /* proc entry */ 1592 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1593 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1594 int valid_super_block; /* valid super block no */ 1595 unsigned long s_flag; /* flags for sbi */ 1596 struct mutex writepages; /* mutex for writepages() */ 1597 1598 #ifdef CONFIG_BLK_DEV_ZONED 1599 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1600 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1601 #endif 1602 1603 /* for node-related operations */ 1604 struct f2fs_nm_info *nm_info; /* node manager */ 1605 struct inode *node_inode; /* cache node blocks */ 1606 1607 /* for segment-related operations */ 1608 struct f2fs_sm_info *sm_info; /* segment manager */ 1609 1610 /* for bio operations */ 1611 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1612 /* keep migration IO order for LFS mode */ 1613 struct f2fs_rwsem io_order_lock; 1614 mempool_t *write_io_dummy; /* Dummy pages */ 1615 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1616 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1617 1618 /* for checkpoint */ 1619 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1620 int cur_cp_pack; /* remain current cp pack */ 1621 spinlock_t cp_lock; /* for flag in ckpt */ 1622 struct inode *meta_inode; /* cache meta blocks */ 1623 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1624 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1625 struct f2fs_rwsem node_write; /* locking node writes */ 1626 struct f2fs_rwsem node_change; /* locking node change */ 1627 wait_queue_head_t cp_wait; 1628 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1629 long interval_time[MAX_TIME]; /* to store thresholds */ 1630 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1631 1632 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1633 1634 spinlock_t fsync_node_lock; /* for node entry lock */ 1635 struct list_head fsync_node_list; /* node list head */ 1636 unsigned int fsync_seg_id; /* sequence id */ 1637 unsigned int fsync_node_num; /* number of node entries */ 1638 1639 /* for orphan inode, use 0'th array */ 1640 unsigned int max_orphans; /* max orphan inodes */ 1641 1642 /* for inode management */ 1643 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1644 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1645 struct mutex flush_lock; /* for flush exclusion */ 1646 1647 /* for extent tree cache */ 1648 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1649 struct mutex extent_tree_lock; /* locking extent radix tree */ 1650 struct list_head extent_list; /* lru list for shrinker */ 1651 spinlock_t extent_lock; /* locking extent lru list */ 1652 atomic_t total_ext_tree; /* extent tree count */ 1653 struct list_head zombie_list; /* extent zombie tree list */ 1654 atomic_t total_zombie_tree; /* extent zombie tree count */ 1655 atomic_t total_ext_node; /* extent info count */ 1656 1657 /* basic filesystem units */ 1658 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1659 unsigned int log_blocksize; /* log2 block size */ 1660 unsigned int blocksize; /* block size */ 1661 unsigned int root_ino_num; /* root inode number*/ 1662 unsigned int node_ino_num; /* node inode number*/ 1663 unsigned int meta_ino_num; /* meta inode number*/ 1664 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1665 unsigned int blocks_per_seg; /* blocks per segment */ 1666 unsigned int segs_per_sec; /* segments per section */ 1667 unsigned int secs_per_zone; /* sections per zone */ 1668 unsigned int total_sections; /* total section count */ 1669 unsigned int total_node_count; /* total node block count */ 1670 unsigned int total_valid_node_count; /* valid node block count */ 1671 int dir_level; /* directory level */ 1672 int readdir_ra; /* readahead inode in readdir */ 1673 u64 max_io_bytes; /* max io bytes to merge IOs */ 1674 1675 block_t user_block_count; /* # of user blocks */ 1676 block_t total_valid_block_count; /* # of valid blocks */ 1677 block_t discard_blks; /* discard command candidats */ 1678 block_t last_valid_block_count; /* for recovery */ 1679 block_t reserved_blocks; /* configurable reserved blocks */ 1680 block_t current_reserved_blocks; /* current reserved blocks */ 1681 1682 /* Additional tracking for no checkpoint mode */ 1683 block_t unusable_block_count; /* # of blocks saved by last cp */ 1684 1685 unsigned int nquota_files; /* # of quota sysfile */ 1686 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1687 1688 /* # of pages, see count_type */ 1689 atomic_t nr_pages[NR_COUNT_TYPE]; 1690 /* # of allocated blocks */ 1691 struct percpu_counter alloc_valid_block_count; 1692 /* # of node block writes as roll forward recovery */ 1693 struct percpu_counter rf_node_block_count; 1694 1695 /* writeback control */ 1696 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1697 1698 /* valid inode count */ 1699 struct percpu_counter total_valid_inode_count; 1700 1701 struct f2fs_mount_info mount_opt; /* mount options */ 1702 1703 /* for cleaning operations */ 1704 struct f2fs_rwsem gc_lock; /* 1705 * semaphore for GC, avoid 1706 * race between GC and GC or CP 1707 */ 1708 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1709 struct atgc_management am; /* atgc management */ 1710 unsigned int cur_victim_sec; /* current victim section num */ 1711 unsigned int gc_mode; /* current GC state */ 1712 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1713 spinlock_t gc_urgent_high_lock; 1714 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1715 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1716 1717 /* for skip statistic */ 1718 unsigned int atomic_files; /* # of opened atomic file */ 1719 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1720 1721 /* threshold for gc trials on pinned files */ 1722 u64 gc_pin_file_threshold; 1723 struct f2fs_rwsem pin_sem; 1724 1725 /* maximum # of trials to find a victim segment for SSR and GC */ 1726 unsigned int max_victim_search; 1727 /* migration granularity of garbage collection, unit: segment */ 1728 unsigned int migration_granularity; 1729 1730 /* 1731 * for stat information. 1732 * one is for the LFS mode, and the other is for the SSR mode. 1733 */ 1734 #ifdef CONFIG_F2FS_STAT_FS 1735 struct f2fs_stat_info *stat_info; /* FS status information */ 1736 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1737 unsigned int segment_count[2]; /* # of allocated segments */ 1738 unsigned int block_count[2]; /* # of allocated blocks */ 1739 atomic_t inplace_count; /* # of inplace update */ 1740 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1741 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1742 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1743 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1744 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1745 atomic_t inline_inode; /* # of inline_data inodes */ 1746 atomic_t inline_dir; /* # of inline_dentry inodes */ 1747 atomic_t compr_inode; /* # of compressed inodes */ 1748 atomic64_t compr_blocks; /* # of compressed blocks */ 1749 atomic_t max_aw_cnt; /* max # of atomic writes */ 1750 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1751 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1752 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1753 #endif 1754 spinlock_t stat_lock; /* lock for stat operations */ 1755 1756 /* to attach REQ_META|REQ_FUA flags */ 1757 unsigned int data_io_flag; 1758 unsigned int node_io_flag; 1759 1760 /* For sysfs suppport */ 1761 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1762 struct completion s_kobj_unregister; 1763 1764 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1765 struct completion s_stat_kobj_unregister; 1766 1767 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1768 struct completion s_feature_list_kobj_unregister; 1769 1770 /* For shrinker support */ 1771 struct list_head s_list; 1772 struct mutex umount_mutex; 1773 unsigned int shrinker_run_no; 1774 1775 /* For multi devices */ 1776 int s_ndevs; /* number of devices */ 1777 struct f2fs_dev_info *devs; /* for device list */ 1778 unsigned int dirty_device; /* for checkpoint data flush */ 1779 spinlock_t dev_lock; /* protect dirty_device */ 1780 bool aligned_blksize; /* all devices has the same logical blksize */ 1781 1782 /* For write statistics */ 1783 u64 sectors_written_start; 1784 u64 kbytes_written; 1785 1786 /* Reference to checksum algorithm driver via cryptoapi */ 1787 struct crypto_shash *s_chksum_driver; 1788 1789 /* Precomputed FS UUID checksum for seeding other checksums */ 1790 __u32 s_chksum_seed; 1791 1792 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1793 1794 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1795 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1796 1797 /* For reclaimed segs statistics per each GC mode */ 1798 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1799 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1800 1801 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1802 1803 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1804 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1805 1806 #ifdef CONFIG_F2FS_FS_COMPRESSION 1807 struct kmem_cache *page_array_slab; /* page array entry */ 1808 unsigned int page_array_slab_size; /* default page array slab size */ 1809 1810 /* For runtime compression statistics */ 1811 u64 compr_written_block; 1812 u64 compr_saved_block; 1813 u32 compr_new_inode; 1814 1815 /* For compressed block cache */ 1816 struct inode *compress_inode; /* cache compressed blocks */ 1817 unsigned int compress_percent; /* cache page percentage */ 1818 unsigned int compress_watermark; /* cache page watermark */ 1819 atomic_t compress_page_hit; /* cache hit count */ 1820 #endif 1821 1822 #ifdef CONFIG_F2FS_IOSTAT 1823 /* For app/fs IO statistics */ 1824 spinlock_t iostat_lock; 1825 unsigned long long rw_iostat[NR_IO_TYPE]; 1826 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1827 bool iostat_enable; 1828 unsigned long iostat_next_period; 1829 unsigned int iostat_period_ms; 1830 1831 /* For io latency related statistics info in one iostat period */ 1832 spinlock_t iostat_lat_lock; 1833 struct iostat_lat_info *iostat_io_lat; 1834 #endif 1835 }; 1836 1837 #ifdef CONFIG_F2FS_FAULT_INJECTION 1838 #define f2fs_show_injection_info(sbi, type) \ 1839 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1840 KERN_INFO, sbi->sb->s_id, \ 1841 f2fs_fault_name[type], \ 1842 __func__, __builtin_return_address(0)) 1843 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1844 { 1845 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1846 1847 if (!ffi->inject_rate) 1848 return false; 1849 1850 if (!IS_FAULT_SET(ffi, type)) 1851 return false; 1852 1853 atomic_inc(&ffi->inject_ops); 1854 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1855 atomic_set(&ffi->inject_ops, 0); 1856 return true; 1857 } 1858 return false; 1859 } 1860 #else 1861 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1862 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1863 { 1864 return false; 1865 } 1866 #endif 1867 1868 /* 1869 * Test if the mounted volume is a multi-device volume. 1870 * - For a single regular disk volume, sbi->s_ndevs is 0. 1871 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1872 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1873 */ 1874 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1875 { 1876 return sbi->s_ndevs > 1; 1877 } 1878 1879 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1880 { 1881 unsigned long now = jiffies; 1882 1883 sbi->last_time[type] = now; 1884 1885 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1886 if (type == REQ_TIME) { 1887 sbi->last_time[DISCARD_TIME] = now; 1888 sbi->last_time[GC_TIME] = now; 1889 } 1890 } 1891 1892 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1893 { 1894 unsigned long interval = sbi->interval_time[type] * HZ; 1895 1896 return time_after(jiffies, sbi->last_time[type] + interval); 1897 } 1898 1899 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1900 int type) 1901 { 1902 unsigned long interval = sbi->interval_time[type] * HZ; 1903 unsigned int wait_ms = 0; 1904 long delta; 1905 1906 delta = (sbi->last_time[type] + interval) - jiffies; 1907 if (delta > 0) 1908 wait_ms = jiffies_to_msecs(delta); 1909 1910 return wait_ms; 1911 } 1912 1913 /* 1914 * Inline functions 1915 */ 1916 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1917 const void *address, unsigned int length) 1918 { 1919 struct { 1920 struct shash_desc shash; 1921 char ctx[4]; 1922 } desc; 1923 int err; 1924 1925 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1926 1927 desc.shash.tfm = sbi->s_chksum_driver; 1928 *(u32 *)desc.ctx = crc; 1929 1930 err = crypto_shash_update(&desc.shash, address, length); 1931 BUG_ON(err); 1932 1933 return *(u32 *)desc.ctx; 1934 } 1935 1936 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1937 unsigned int length) 1938 { 1939 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1940 } 1941 1942 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1943 void *buf, size_t buf_size) 1944 { 1945 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1946 } 1947 1948 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1949 const void *address, unsigned int length) 1950 { 1951 return __f2fs_crc32(sbi, crc, address, length); 1952 } 1953 1954 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1955 { 1956 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1957 } 1958 1959 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1960 { 1961 return sb->s_fs_info; 1962 } 1963 1964 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1965 { 1966 return F2FS_SB(inode->i_sb); 1967 } 1968 1969 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1970 { 1971 return F2FS_I_SB(mapping->host); 1972 } 1973 1974 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1975 { 1976 return F2FS_M_SB(page_file_mapping(page)); 1977 } 1978 1979 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1980 { 1981 return (struct f2fs_super_block *)(sbi->raw_super); 1982 } 1983 1984 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1985 { 1986 return (struct f2fs_checkpoint *)(sbi->ckpt); 1987 } 1988 1989 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1990 { 1991 return (struct f2fs_node *)page_address(page); 1992 } 1993 1994 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1995 { 1996 return &((struct f2fs_node *)page_address(page))->i; 1997 } 1998 1999 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2000 { 2001 return (struct f2fs_nm_info *)(sbi->nm_info); 2002 } 2003 2004 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2005 { 2006 return (struct f2fs_sm_info *)(sbi->sm_info); 2007 } 2008 2009 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2010 { 2011 return (struct sit_info *)(SM_I(sbi)->sit_info); 2012 } 2013 2014 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2015 { 2016 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2017 } 2018 2019 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2020 { 2021 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2022 } 2023 2024 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2025 { 2026 return sbi->meta_inode->i_mapping; 2027 } 2028 2029 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2030 { 2031 return sbi->node_inode->i_mapping; 2032 } 2033 2034 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2035 { 2036 return test_bit(type, &sbi->s_flag); 2037 } 2038 2039 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2040 { 2041 set_bit(type, &sbi->s_flag); 2042 } 2043 2044 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2045 { 2046 clear_bit(type, &sbi->s_flag); 2047 } 2048 2049 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2050 { 2051 return le64_to_cpu(cp->checkpoint_ver); 2052 } 2053 2054 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2055 { 2056 if (type < F2FS_MAX_QUOTAS) 2057 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2058 return 0; 2059 } 2060 2061 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2062 { 2063 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2064 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2065 } 2066 2067 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2068 { 2069 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2070 2071 return ckpt_flags & f; 2072 } 2073 2074 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2075 { 2076 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2077 } 2078 2079 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2080 { 2081 unsigned int ckpt_flags; 2082 2083 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2084 ckpt_flags |= f; 2085 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2086 } 2087 2088 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2089 { 2090 unsigned long flags; 2091 2092 spin_lock_irqsave(&sbi->cp_lock, flags); 2093 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2094 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2095 } 2096 2097 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2098 { 2099 unsigned int ckpt_flags; 2100 2101 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2102 ckpt_flags &= (~f); 2103 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2104 } 2105 2106 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2107 { 2108 unsigned long flags; 2109 2110 spin_lock_irqsave(&sbi->cp_lock, flags); 2111 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2112 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2113 } 2114 2115 #define init_f2fs_rwsem(sem) \ 2116 do { \ 2117 static struct lock_class_key __key; \ 2118 \ 2119 __init_f2fs_rwsem((sem), #sem, &__key); \ 2120 } while (0) 2121 2122 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2123 const char *sem_name, struct lock_class_key *key) 2124 { 2125 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2126 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2127 init_waitqueue_head(&sem->read_waiters); 2128 #endif 2129 } 2130 2131 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2132 { 2133 return rwsem_is_locked(&sem->internal_rwsem); 2134 } 2135 2136 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2137 { 2138 return rwsem_is_contended(&sem->internal_rwsem); 2139 } 2140 2141 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2142 { 2143 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2144 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2145 #else 2146 down_read(&sem->internal_rwsem); 2147 #endif 2148 } 2149 2150 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2151 { 2152 return down_read_trylock(&sem->internal_rwsem); 2153 } 2154 2155 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2156 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2157 { 2158 down_read_nested(&sem->internal_rwsem, subclass); 2159 } 2160 #else 2161 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2162 #endif 2163 2164 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2165 { 2166 up_read(&sem->internal_rwsem); 2167 } 2168 2169 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2170 { 2171 down_write(&sem->internal_rwsem); 2172 } 2173 2174 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2175 { 2176 return down_write_trylock(&sem->internal_rwsem); 2177 } 2178 2179 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2180 { 2181 up_write(&sem->internal_rwsem); 2182 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2183 wake_up_all(&sem->read_waiters); 2184 #endif 2185 } 2186 2187 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2188 { 2189 f2fs_down_read(&sbi->cp_rwsem); 2190 } 2191 2192 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2193 { 2194 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2195 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2196 return 0; 2197 } 2198 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2199 } 2200 2201 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2202 { 2203 f2fs_up_read(&sbi->cp_rwsem); 2204 } 2205 2206 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2207 { 2208 f2fs_down_write(&sbi->cp_rwsem); 2209 } 2210 2211 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2212 { 2213 f2fs_up_write(&sbi->cp_rwsem); 2214 } 2215 2216 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2217 { 2218 int reason = CP_SYNC; 2219 2220 if (test_opt(sbi, FASTBOOT)) 2221 reason = CP_FASTBOOT; 2222 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2223 reason = CP_UMOUNT; 2224 return reason; 2225 } 2226 2227 static inline bool __remain_node_summaries(int reason) 2228 { 2229 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2230 } 2231 2232 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2233 { 2234 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2235 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2236 } 2237 2238 /* 2239 * Check whether the inode has blocks or not 2240 */ 2241 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2242 { 2243 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2244 2245 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2246 } 2247 2248 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2249 { 2250 return ofs == XATTR_NODE_OFFSET; 2251 } 2252 2253 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2254 struct inode *inode, bool cap) 2255 { 2256 if (!inode) 2257 return true; 2258 if (!test_opt(sbi, RESERVE_ROOT)) 2259 return false; 2260 if (IS_NOQUOTA(inode)) 2261 return true; 2262 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2263 return true; 2264 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2265 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2266 return true; 2267 if (cap && capable(CAP_SYS_RESOURCE)) 2268 return true; 2269 return false; 2270 } 2271 2272 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2273 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2274 struct inode *inode, blkcnt_t *count) 2275 { 2276 blkcnt_t diff = 0, release = 0; 2277 block_t avail_user_block_count; 2278 int ret; 2279 2280 ret = dquot_reserve_block(inode, *count); 2281 if (ret) 2282 return ret; 2283 2284 if (time_to_inject(sbi, FAULT_BLOCK)) { 2285 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2286 release = *count; 2287 goto release_quota; 2288 } 2289 2290 /* 2291 * let's increase this in prior to actual block count change in order 2292 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2293 */ 2294 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2295 2296 spin_lock(&sbi->stat_lock); 2297 sbi->total_valid_block_count += (block_t)(*count); 2298 avail_user_block_count = sbi->user_block_count - 2299 sbi->current_reserved_blocks; 2300 2301 if (!__allow_reserved_blocks(sbi, inode, true)) 2302 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2303 2304 if (F2FS_IO_ALIGNED(sbi)) 2305 avail_user_block_count -= sbi->blocks_per_seg * 2306 SM_I(sbi)->additional_reserved_segments; 2307 2308 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2309 if (avail_user_block_count > sbi->unusable_block_count) 2310 avail_user_block_count -= sbi->unusable_block_count; 2311 else 2312 avail_user_block_count = 0; 2313 } 2314 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2315 diff = sbi->total_valid_block_count - avail_user_block_count; 2316 if (diff > *count) 2317 diff = *count; 2318 *count -= diff; 2319 release = diff; 2320 sbi->total_valid_block_count -= diff; 2321 if (!*count) { 2322 spin_unlock(&sbi->stat_lock); 2323 goto enospc; 2324 } 2325 } 2326 spin_unlock(&sbi->stat_lock); 2327 2328 if (unlikely(release)) { 2329 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2330 dquot_release_reservation_block(inode, release); 2331 } 2332 f2fs_i_blocks_write(inode, *count, true, true); 2333 return 0; 2334 2335 enospc: 2336 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2337 release_quota: 2338 dquot_release_reservation_block(inode, release); 2339 return -ENOSPC; 2340 } 2341 2342 __printf(2, 3) 2343 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2344 2345 #define f2fs_err(sbi, fmt, ...) \ 2346 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2347 #define f2fs_warn(sbi, fmt, ...) \ 2348 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2349 #define f2fs_notice(sbi, fmt, ...) \ 2350 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2351 #define f2fs_info(sbi, fmt, ...) \ 2352 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2353 #define f2fs_debug(sbi, fmt, ...) \ 2354 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2355 2356 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2357 struct inode *inode, 2358 block_t count) 2359 { 2360 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2361 2362 spin_lock(&sbi->stat_lock); 2363 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2364 sbi->total_valid_block_count -= (block_t)count; 2365 if (sbi->reserved_blocks && 2366 sbi->current_reserved_blocks < sbi->reserved_blocks) 2367 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2368 sbi->current_reserved_blocks + count); 2369 spin_unlock(&sbi->stat_lock); 2370 if (unlikely(inode->i_blocks < sectors)) { 2371 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2372 inode->i_ino, 2373 (unsigned long long)inode->i_blocks, 2374 (unsigned long long)sectors); 2375 set_sbi_flag(sbi, SBI_NEED_FSCK); 2376 return; 2377 } 2378 f2fs_i_blocks_write(inode, count, false, true); 2379 } 2380 2381 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2382 { 2383 atomic_inc(&sbi->nr_pages[count_type]); 2384 2385 if (count_type == F2FS_DIRTY_DENTS || 2386 count_type == F2FS_DIRTY_NODES || 2387 count_type == F2FS_DIRTY_META || 2388 count_type == F2FS_DIRTY_QDATA || 2389 count_type == F2FS_DIRTY_IMETA) 2390 set_sbi_flag(sbi, SBI_IS_DIRTY); 2391 } 2392 2393 static inline void inode_inc_dirty_pages(struct inode *inode) 2394 { 2395 atomic_inc(&F2FS_I(inode)->dirty_pages); 2396 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2397 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2398 if (IS_NOQUOTA(inode)) 2399 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2400 } 2401 2402 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2403 { 2404 atomic_dec(&sbi->nr_pages[count_type]); 2405 } 2406 2407 static inline void inode_dec_dirty_pages(struct inode *inode) 2408 { 2409 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2410 !S_ISLNK(inode->i_mode)) 2411 return; 2412 2413 atomic_dec(&F2FS_I(inode)->dirty_pages); 2414 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2415 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2416 if (IS_NOQUOTA(inode)) 2417 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2418 } 2419 2420 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2421 { 2422 return atomic_read(&sbi->nr_pages[count_type]); 2423 } 2424 2425 static inline int get_dirty_pages(struct inode *inode) 2426 { 2427 return atomic_read(&F2FS_I(inode)->dirty_pages); 2428 } 2429 2430 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2431 { 2432 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2433 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2434 sbi->log_blocks_per_seg; 2435 2436 return segs / sbi->segs_per_sec; 2437 } 2438 2439 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2440 { 2441 return sbi->total_valid_block_count; 2442 } 2443 2444 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2445 { 2446 return sbi->discard_blks; 2447 } 2448 2449 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2450 { 2451 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2452 2453 /* return NAT or SIT bitmap */ 2454 if (flag == NAT_BITMAP) 2455 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2456 else if (flag == SIT_BITMAP) 2457 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2458 2459 return 0; 2460 } 2461 2462 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2463 { 2464 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2465 } 2466 2467 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2468 { 2469 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2470 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2471 int offset; 2472 2473 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2474 offset = (flag == SIT_BITMAP) ? 2475 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2476 /* 2477 * if large_nat_bitmap feature is enabled, leave checksum 2478 * protection for all nat/sit bitmaps. 2479 */ 2480 return tmp_ptr + offset + sizeof(__le32); 2481 } 2482 2483 if (__cp_payload(sbi) > 0) { 2484 if (flag == NAT_BITMAP) 2485 return &ckpt->sit_nat_version_bitmap; 2486 else 2487 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2488 } else { 2489 offset = (flag == NAT_BITMAP) ? 2490 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2491 return tmp_ptr + offset; 2492 } 2493 } 2494 2495 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2496 { 2497 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2498 2499 if (sbi->cur_cp_pack == 2) 2500 start_addr += sbi->blocks_per_seg; 2501 return start_addr; 2502 } 2503 2504 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2505 { 2506 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2507 2508 if (sbi->cur_cp_pack == 1) 2509 start_addr += sbi->blocks_per_seg; 2510 return start_addr; 2511 } 2512 2513 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2514 { 2515 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2516 } 2517 2518 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2519 { 2520 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2521 } 2522 2523 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2524 struct inode *inode, bool is_inode) 2525 { 2526 block_t valid_block_count; 2527 unsigned int valid_node_count, user_block_count; 2528 int err; 2529 2530 if (is_inode) { 2531 if (inode) { 2532 err = dquot_alloc_inode(inode); 2533 if (err) 2534 return err; 2535 } 2536 } else { 2537 err = dquot_reserve_block(inode, 1); 2538 if (err) 2539 return err; 2540 } 2541 2542 if (time_to_inject(sbi, FAULT_BLOCK)) { 2543 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2544 goto enospc; 2545 } 2546 2547 spin_lock(&sbi->stat_lock); 2548 2549 valid_block_count = sbi->total_valid_block_count + 2550 sbi->current_reserved_blocks + 1; 2551 2552 if (!__allow_reserved_blocks(sbi, inode, false)) 2553 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2554 2555 if (F2FS_IO_ALIGNED(sbi)) 2556 valid_block_count += sbi->blocks_per_seg * 2557 SM_I(sbi)->additional_reserved_segments; 2558 2559 user_block_count = sbi->user_block_count; 2560 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2561 user_block_count -= sbi->unusable_block_count; 2562 2563 if (unlikely(valid_block_count > user_block_count)) { 2564 spin_unlock(&sbi->stat_lock); 2565 goto enospc; 2566 } 2567 2568 valid_node_count = sbi->total_valid_node_count + 1; 2569 if (unlikely(valid_node_count > sbi->total_node_count)) { 2570 spin_unlock(&sbi->stat_lock); 2571 goto enospc; 2572 } 2573 2574 sbi->total_valid_node_count++; 2575 sbi->total_valid_block_count++; 2576 spin_unlock(&sbi->stat_lock); 2577 2578 if (inode) { 2579 if (is_inode) 2580 f2fs_mark_inode_dirty_sync(inode, true); 2581 else 2582 f2fs_i_blocks_write(inode, 1, true, true); 2583 } 2584 2585 percpu_counter_inc(&sbi->alloc_valid_block_count); 2586 return 0; 2587 2588 enospc: 2589 if (is_inode) { 2590 if (inode) 2591 dquot_free_inode(inode); 2592 } else { 2593 dquot_release_reservation_block(inode, 1); 2594 } 2595 return -ENOSPC; 2596 } 2597 2598 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2599 struct inode *inode, bool is_inode) 2600 { 2601 spin_lock(&sbi->stat_lock); 2602 2603 if (unlikely(!sbi->total_valid_block_count || 2604 !sbi->total_valid_node_count)) { 2605 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2606 sbi->total_valid_block_count, 2607 sbi->total_valid_node_count); 2608 set_sbi_flag(sbi, SBI_NEED_FSCK); 2609 } else { 2610 sbi->total_valid_block_count--; 2611 sbi->total_valid_node_count--; 2612 } 2613 2614 if (sbi->reserved_blocks && 2615 sbi->current_reserved_blocks < sbi->reserved_blocks) 2616 sbi->current_reserved_blocks++; 2617 2618 spin_unlock(&sbi->stat_lock); 2619 2620 if (is_inode) { 2621 dquot_free_inode(inode); 2622 } else { 2623 if (unlikely(inode->i_blocks == 0)) { 2624 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2625 inode->i_ino, 2626 (unsigned long long)inode->i_blocks); 2627 set_sbi_flag(sbi, SBI_NEED_FSCK); 2628 return; 2629 } 2630 f2fs_i_blocks_write(inode, 1, false, true); 2631 } 2632 } 2633 2634 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2635 { 2636 return sbi->total_valid_node_count; 2637 } 2638 2639 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2640 { 2641 percpu_counter_inc(&sbi->total_valid_inode_count); 2642 } 2643 2644 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2645 { 2646 percpu_counter_dec(&sbi->total_valid_inode_count); 2647 } 2648 2649 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2650 { 2651 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2652 } 2653 2654 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2655 pgoff_t index, bool for_write) 2656 { 2657 struct page *page; 2658 2659 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2660 if (!for_write) 2661 page = find_get_page_flags(mapping, index, 2662 FGP_LOCK | FGP_ACCESSED); 2663 else 2664 page = find_lock_page(mapping, index); 2665 if (page) 2666 return page; 2667 2668 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2669 f2fs_show_injection_info(F2FS_M_SB(mapping), 2670 FAULT_PAGE_ALLOC); 2671 return NULL; 2672 } 2673 } 2674 2675 if (!for_write) 2676 return grab_cache_page(mapping, index); 2677 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2678 } 2679 2680 static inline struct page *f2fs_pagecache_get_page( 2681 struct address_space *mapping, pgoff_t index, 2682 int fgp_flags, gfp_t gfp_mask) 2683 { 2684 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2685 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2686 return NULL; 2687 } 2688 2689 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2690 } 2691 2692 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2693 { 2694 char *src_kaddr = kmap(src); 2695 char *dst_kaddr = kmap(dst); 2696 2697 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2698 kunmap(dst); 2699 kunmap(src); 2700 } 2701 2702 static inline void f2fs_put_page(struct page *page, int unlock) 2703 { 2704 if (!page) 2705 return; 2706 2707 if (unlock) { 2708 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2709 unlock_page(page); 2710 } 2711 put_page(page); 2712 } 2713 2714 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2715 { 2716 if (dn->node_page) 2717 f2fs_put_page(dn->node_page, 1); 2718 if (dn->inode_page && dn->node_page != dn->inode_page) 2719 f2fs_put_page(dn->inode_page, 0); 2720 dn->node_page = NULL; 2721 dn->inode_page = NULL; 2722 } 2723 2724 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2725 size_t size) 2726 { 2727 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2728 } 2729 2730 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2731 gfp_t flags) 2732 { 2733 void *entry; 2734 2735 entry = kmem_cache_alloc(cachep, flags); 2736 if (!entry) 2737 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2738 return entry; 2739 } 2740 2741 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2742 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2743 { 2744 if (nofail) 2745 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2746 2747 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2748 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2749 return NULL; 2750 } 2751 2752 return kmem_cache_alloc(cachep, flags); 2753 } 2754 2755 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2756 { 2757 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2758 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2759 get_pages(sbi, F2FS_WB_CP_DATA) || 2760 get_pages(sbi, F2FS_DIO_READ) || 2761 get_pages(sbi, F2FS_DIO_WRITE)) 2762 return true; 2763 2764 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2765 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2766 return true; 2767 2768 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2769 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2770 return true; 2771 return false; 2772 } 2773 2774 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2775 { 2776 if (sbi->gc_mode == GC_URGENT_HIGH) 2777 return true; 2778 2779 if (is_inflight_io(sbi, type)) 2780 return false; 2781 2782 if (sbi->gc_mode == GC_URGENT_MID) 2783 return true; 2784 2785 if (sbi->gc_mode == GC_URGENT_LOW && 2786 (type == DISCARD_TIME || type == GC_TIME)) 2787 return true; 2788 2789 return f2fs_time_over(sbi, type); 2790 } 2791 2792 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2793 unsigned long index, void *item) 2794 { 2795 while (radix_tree_insert(root, index, item)) 2796 cond_resched(); 2797 } 2798 2799 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2800 2801 static inline bool IS_INODE(struct page *page) 2802 { 2803 struct f2fs_node *p = F2FS_NODE(page); 2804 2805 return RAW_IS_INODE(p); 2806 } 2807 2808 static inline int offset_in_addr(struct f2fs_inode *i) 2809 { 2810 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2811 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2812 } 2813 2814 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2815 { 2816 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2817 } 2818 2819 static inline int f2fs_has_extra_attr(struct inode *inode); 2820 static inline block_t data_blkaddr(struct inode *inode, 2821 struct page *node_page, unsigned int offset) 2822 { 2823 struct f2fs_node *raw_node; 2824 __le32 *addr_array; 2825 int base = 0; 2826 bool is_inode = IS_INODE(node_page); 2827 2828 raw_node = F2FS_NODE(node_page); 2829 2830 if (is_inode) { 2831 if (!inode) 2832 /* from GC path only */ 2833 base = offset_in_addr(&raw_node->i); 2834 else if (f2fs_has_extra_attr(inode)) 2835 base = get_extra_isize(inode); 2836 } 2837 2838 addr_array = blkaddr_in_node(raw_node); 2839 return le32_to_cpu(addr_array[base + offset]); 2840 } 2841 2842 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2843 { 2844 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2845 } 2846 2847 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2848 { 2849 int mask; 2850 2851 addr += (nr >> 3); 2852 mask = 1 << (7 - (nr & 0x07)); 2853 return mask & *addr; 2854 } 2855 2856 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2857 { 2858 int mask; 2859 2860 addr += (nr >> 3); 2861 mask = 1 << (7 - (nr & 0x07)); 2862 *addr |= mask; 2863 } 2864 2865 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2866 { 2867 int mask; 2868 2869 addr += (nr >> 3); 2870 mask = 1 << (7 - (nr & 0x07)); 2871 *addr &= ~mask; 2872 } 2873 2874 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2875 { 2876 int mask; 2877 int ret; 2878 2879 addr += (nr >> 3); 2880 mask = 1 << (7 - (nr & 0x07)); 2881 ret = mask & *addr; 2882 *addr |= mask; 2883 return ret; 2884 } 2885 2886 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2887 { 2888 int mask; 2889 int ret; 2890 2891 addr += (nr >> 3); 2892 mask = 1 << (7 - (nr & 0x07)); 2893 ret = mask & *addr; 2894 *addr &= ~mask; 2895 return ret; 2896 } 2897 2898 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2899 { 2900 int mask; 2901 2902 addr += (nr >> 3); 2903 mask = 1 << (7 - (nr & 0x07)); 2904 *addr ^= mask; 2905 } 2906 2907 /* 2908 * On-disk inode flags (f2fs_inode::i_flags) 2909 */ 2910 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2911 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2912 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2913 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2914 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2915 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2916 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2917 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2918 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2919 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2920 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2921 2922 /* Flags that should be inherited by new inodes from their parent. */ 2923 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2924 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2925 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2926 2927 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2928 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2929 F2FS_CASEFOLD_FL)) 2930 2931 /* Flags that are appropriate for non-directories/regular files. */ 2932 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2933 2934 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2935 { 2936 if (S_ISDIR(mode)) 2937 return flags; 2938 else if (S_ISREG(mode)) 2939 return flags & F2FS_REG_FLMASK; 2940 else 2941 return flags & F2FS_OTHER_FLMASK; 2942 } 2943 2944 static inline void __mark_inode_dirty_flag(struct inode *inode, 2945 int flag, bool set) 2946 { 2947 switch (flag) { 2948 case FI_INLINE_XATTR: 2949 case FI_INLINE_DATA: 2950 case FI_INLINE_DENTRY: 2951 case FI_NEW_INODE: 2952 if (set) 2953 return; 2954 fallthrough; 2955 case FI_DATA_EXIST: 2956 case FI_INLINE_DOTS: 2957 case FI_PIN_FILE: 2958 case FI_COMPRESS_RELEASED: 2959 f2fs_mark_inode_dirty_sync(inode, true); 2960 } 2961 } 2962 2963 static inline void set_inode_flag(struct inode *inode, int flag) 2964 { 2965 set_bit(flag, F2FS_I(inode)->flags); 2966 __mark_inode_dirty_flag(inode, flag, true); 2967 } 2968 2969 static inline int is_inode_flag_set(struct inode *inode, int flag) 2970 { 2971 return test_bit(flag, F2FS_I(inode)->flags); 2972 } 2973 2974 static inline void clear_inode_flag(struct inode *inode, int flag) 2975 { 2976 clear_bit(flag, F2FS_I(inode)->flags); 2977 __mark_inode_dirty_flag(inode, flag, false); 2978 } 2979 2980 static inline bool f2fs_verity_in_progress(struct inode *inode) 2981 { 2982 return IS_ENABLED(CONFIG_FS_VERITY) && 2983 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2984 } 2985 2986 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2987 { 2988 F2FS_I(inode)->i_acl_mode = mode; 2989 set_inode_flag(inode, FI_ACL_MODE); 2990 f2fs_mark_inode_dirty_sync(inode, false); 2991 } 2992 2993 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2994 { 2995 if (inc) 2996 inc_nlink(inode); 2997 else 2998 drop_nlink(inode); 2999 f2fs_mark_inode_dirty_sync(inode, true); 3000 } 3001 3002 static inline void f2fs_i_blocks_write(struct inode *inode, 3003 block_t diff, bool add, bool claim) 3004 { 3005 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3006 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3007 3008 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3009 if (add) { 3010 if (claim) 3011 dquot_claim_block(inode, diff); 3012 else 3013 dquot_alloc_block_nofail(inode, diff); 3014 } else { 3015 dquot_free_block(inode, diff); 3016 } 3017 3018 f2fs_mark_inode_dirty_sync(inode, true); 3019 if (clean || recover) 3020 set_inode_flag(inode, FI_AUTO_RECOVER); 3021 } 3022 3023 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3024 { 3025 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3026 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3027 3028 if (i_size_read(inode) == i_size) 3029 return; 3030 3031 i_size_write(inode, i_size); 3032 f2fs_mark_inode_dirty_sync(inode, true); 3033 if (clean || recover) 3034 set_inode_flag(inode, FI_AUTO_RECOVER); 3035 } 3036 3037 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3038 { 3039 F2FS_I(inode)->i_current_depth = depth; 3040 f2fs_mark_inode_dirty_sync(inode, true); 3041 } 3042 3043 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3044 unsigned int count) 3045 { 3046 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3047 f2fs_mark_inode_dirty_sync(inode, true); 3048 } 3049 3050 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3051 { 3052 F2FS_I(inode)->i_xattr_nid = xnid; 3053 f2fs_mark_inode_dirty_sync(inode, true); 3054 } 3055 3056 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3057 { 3058 F2FS_I(inode)->i_pino = pino; 3059 f2fs_mark_inode_dirty_sync(inode, true); 3060 } 3061 3062 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3063 { 3064 struct f2fs_inode_info *fi = F2FS_I(inode); 3065 3066 if (ri->i_inline & F2FS_INLINE_XATTR) 3067 set_bit(FI_INLINE_XATTR, fi->flags); 3068 if (ri->i_inline & F2FS_INLINE_DATA) 3069 set_bit(FI_INLINE_DATA, fi->flags); 3070 if (ri->i_inline & F2FS_INLINE_DENTRY) 3071 set_bit(FI_INLINE_DENTRY, fi->flags); 3072 if (ri->i_inline & F2FS_DATA_EXIST) 3073 set_bit(FI_DATA_EXIST, fi->flags); 3074 if (ri->i_inline & F2FS_INLINE_DOTS) 3075 set_bit(FI_INLINE_DOTS, fi->flags); 3076 if (ri->i_inline & F2FS_EXTRA_ATTR) 3077 set_bit(FI_EXTRA_ATTR, fi->flags); 3078 if (ri->i_inline & F2FS_PIN_FILE) 3079 set_bit(FI_PIN_FILE, fi->flags); 3080 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3081 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3082 } 3083 3084 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3085 { 3086 ri->i_inline = 0; 3087 3088 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3089 ri->i_inline |= F2FS_INLINE_XATTR; 3090 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3091 ri->i_inline |= F2FS_INLINE_DATA; 3092 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3093 ri->i_inline |= F2FS_INLINE_DENTRY; 3094 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3095 ri->i_inline |= F2FS_DATA_EXIST; 3096 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3097 ri->i_inline |= F2FS_INLINE_DOTS; 3098 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3099 ri->i_inline |= F2FS_EXTRA_ATTR; 3100 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3101 ri->i_inline |= F2FS_PIN_FILE; 3102 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3103 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3104 } 3105 3106 static inline int f2fs_has_extra_attr(struct inode *inode) 3107 { 3108 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3109 } 3110 3111 static inline int f2fs_has_inline_xattr(struct inode *inode) 3112 { 3113 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3114 } 3115 3116 static inline int f2fs_compressed_file(struct inode *inode) 3117 { 3118 return S_ISREG(inode->i_mode) && 3119 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3120 } 3121 3122 static inline bool f2fs_need_compress_data(struct inode *inode) 3123 { 3124 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3125 3126 if (!f2fs_compressed_file(inode)) 3127 return false; 3128 3129 if (compress_mode == COMPR_MODE_FS) 3130 return true; 3131 else if (compress_mode == COMPR_MODE_USER && 3132 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3133 return true; 3134 3135 return false; 3136 } 3137 3138 static inline unsigned int addrs_per_inode(struct inode *inode) 3139 { 3140 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3141 get_inline_xattr_addrs(inode); 3142 3143 if (!f2fs_compressed_file(inode)) 3144 return addrs; 3145 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3146 } 3147 3148 static inline unsigned int addrs_per_block(struct inode *inode) 3149 { 3150 if (!f2fs_compressed_file(inode)) 3151 return DEF_ADDRS_PER_BLOCK; 3152 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3153 } 3154 3155 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3156 { 3157 struct f2fs_inode *ri = F2FS_INODE(page); 3158 3159 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3160 get_inline_xattr_addrs(inode)]); 3161 } 3162 3163 static inline int inline_xattr_size(struct inode *inode) 3164 { 3165 if (f2fs_has_inline_xattr(inode)) 3166 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3167 return 0; 3168 } 3169 3170 /* 3171 * Notice: check inline_data flag without inode page lock is unsafe. 3172 * It could change at any time by f2fs_convert_inline_page(). 3173 */ 3174 static inline int f2fs_has_inline_data(struct inode *inode) 3175 { 3176 return is_inode_flag_set(inode, FI_INLINE_DATA); 3177 } 3178 3179 static inline int f2fs_exist_data(struct inode *inode) 3180 { 3181 return is_inode_flag_set(inode, FI_DATA_EXIST); 3182 } 3183 3184 static inline int f2fs_has_inline_dots(struct inode *inode) 3185 { 3186 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3187 } 3188 3189 static inline int f2fs_is_mmap_file(struct inode *inode) 3190 { 3191 return is_inode_flag_set(inode, FI_MMAP_FILE); 3192 } 3193 3194 static inline bool f2fs_is_pinned_file(struct inode *inode) 3195 { 3196 return is_inode_flag_set(inode, FI_PIN_FILE); 3197 } 3198 3199 static inline bool f2fs_is_atomic_file(struct inode *inode) 3200 { 3201 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3202 } 3203 3204 static inline bool f2fs_is_first_block_written(struct inode *inode) 3205 { 3206 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3207 } 3208 3209 static inline bool f2fs_is_drop_cache(struct inode *inode) 3210 { 3211 return is_inode_flag_set(inode, FI_DROP_CACHE); 3212 } 3213 3214 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3215 { 3216 struct f2fs_inode *ri = F2FS_INODE(page); 3217 int extra_size = get_extra_isize(inode); 3218 3219 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3220 } 3221 3222 static inline int f2fs_has_inline_dentry(struct inode *inode) 3223 { 3224 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3225 } 3226 3227 static inline int is_file(struct inode *inode, int type) 3228 { 3229 return F2FS_I(inode)->i_advise & type; 3230 } 3231 3232 static inline void set_file(struct inode *inode, int type) 3233 { 3234 if (is_file(inode, type)) 3235 return; 3236 F2FS_I(inode)->i_advise |= type; 3237 f2fs_mark_inode_dirty_sync(inode, true); 3238 } 3239 3240 static inline void clear_file(struct inode *inode, int type) 3241 { 3242 if (!is_file(inode, type)) 3243 return; 3244 F2FS_I(inode)->i_advise &= ~type; 3245 f2fs_mark_inode_dirty_sync(inode, true); 3246 } 3247 3248 static inline bool f2fs_is_time_consistent(struct inode *inode) 3249 { 3250 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3251 return false; 3252 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3253 return false; 3254 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3255 return false; 3256 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3257 &F2FS_I(inode)->i_crtime)) 3258 return false; 3259 return true; 3260 } 3261 3262 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3263 { 3264 bool ret; 3265 3266 if (dsync) { 3267 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3268 3269 spin_lock(&sbi->inode_lock[DIRTY_META]); 3270 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3271 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3272 return ret; 3273 } 3274 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3275 file_keep_isize(inode) || 3276 i_size_read(inode) & ~PAGE_MASK) 3277 return false; 3278 3279 if (!f2fs_is_time_consistent(inode)) 3280 return false; 3281 3282 spin_lock(&F2FS_I(inode)->i_size_lock); 3283 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3284 spin_unlock(&F2FS_I(inode)->i_size_lock); 3285 3286 return ret; 3287 } 3288 3289 static inline bool f2fs_readonly(struct super_block *sb) 3290 { 3291 return sb_rdonly(sb); 3292 } 3293 3294 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3295 { 3296 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3297 } 3298 3299 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3300 { 3301 if (len == 1 && name[0] == '.') 3302 return true; 3303 3304 if (len == 2 && name[0] == '.' && name[1] == '.') 3305 return true; 3306 3307 return false; 3308 } 3309 3310 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3311 size_t size, gfp_t flags) 3312 { 3313 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3314 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3315 return NULL; 3316 } 3317 3318 return kmalloc(size, flags); 3319 } 3320 3321 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3322 size_t size, gfp_t flags) 3323 { 3324 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3325 } 3326 3327 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3328 size_t size, gfp_t flags) 3329 { 3330 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3331 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3332 return NULL; 3333 } 3334 3335 return kvmalloc(size, flags); 3336 } 3337 3338 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3339 size_t size, gfp_t flags) 3340 { 3341 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3342 } 3343 3344 static inline int get_extra_isize(struct inode *inode) 3345 { 3346 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3347 } 3348 3349 static inline int get_inline_xattr_addrs(struct inode *inode) 3350 { 3351 return F2FS_I(inode)->i_inline_xattr_size; 3352 } 3353 3354 #define f2fs_get_inode_mode(i) \ 3355 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3356 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3357 3358 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3359 (offsetof(struct f2fs_inode, i_extra_end) - \ 3360 offsetof(struct f2fs_inode, i_extra_isize)) \ 3361 3362 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3363 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3364 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3365 sizeof((f2fs_inode)->field)) \ 3366 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3367 3368 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3369 3370 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3371 3372 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3373 block_t blkaddr, int type); 3374 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3375 block_t blkaddr, int type) 3376 { 3377 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3378 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3379 blkaddr, type); 3380 f2fs_bug_on(sbi, 1); 3381 } 3382 } 3383 3384 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3385 { 3386 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3387 blkaddr == COMPRESS_ADDR) 3388 return false; 3389 return true; 3390 } 3391 3392 /* 3393 * file.c 3394 */ 3395 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3396 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3397 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3398 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3399 int f2fs_truncate(struct inode *inode); 3400 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3401 struct kstat *stat, u32 request_mask, unsigned int flags); 3402 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3403 struct iattr *attr); 3404 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3405 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3406 int f2fs_precache_extents(struct inode *inode); 3407 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3408 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3409 struct dentry *dentry, struct fileattr *fa); 3410 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3411 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3412 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3413 int f2fs_pin_file_control(struct inode *inode, bool inc); 3414 3415 /* 3416 * inode.c 3417 */ 3418 void f2fs_set_inode_flags(struct inode *inode); 3419 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3420 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3421 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3422 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3423 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3424 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3425 void f2fs_update_inode_page(struct inode *inode); 3426 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3427 void f2fs_evict_inode(struct inode *inode); 3428 void f2fs_handle_failed_inode(struct inode *inode); 3429 3430 /* 3431 * namei.c 3432 */ 3433 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3434 bool hot, bool set); 3435 struct dentry *f2fs_get_parent(struct dentry *child); 3436 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3437 struct inode **new_inode); 3438 3439 /* 3440 * dir.c 3441 */ 3442 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3443 int f2fs_init_casefolded_name(const struct inode *dir, 3444 struct f2fs_filename *fname); 3445 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3446 int lookup, struct f2fs_filename *fname); 3447 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3448 struct f2fs_filename *fname); 3449 void f2fs_free_filename(struct f2fs_filename *fname); 3450 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3451 const struct f2fs_filename *fname, int *max_slots); 3452 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3453 unsigned int start_pos, struct fscrypt_str *fstr); 3454 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3455 struct f2fs_dentry_ptr *d); 3456 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3457 const struct f2fs_filename *fname, struct page *dpage); 3458 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3459 unsigned int current_depth); 3460 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3461 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3462 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3463 const struct f2fs_filename *fname, 3464 struct page **res_page); 3465 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3466 const struct qstr *child, struct page **res_page); 3467 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3468 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3469 struct page **page); 3470 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3471 struct page *page, struct inode *inode); 3472 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3473 const struct f2fs_filename *fname); 3474 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3475 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3476 unsigned int bit_pos); 3477 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3478 struct inode *inode, nid_t ino, umode_t mode); 3479 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3480 struct inode *inode, nid_t ino, umode_t mode); 3481 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3482 struct inode *inode, nid_t ino, umode_t mode); 3483 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3484 struct inode *dir, struct inode *inode); 3485 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3486 bool f2fs_empty_dir(struct inode *dir); 3487 3488 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3489 { 3490 if (fscrypt_is_nokey_name(dentry)) 3491 return -ENOKEY; 3492 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3493 inode, inode->i_ino, inode->i_mode); 3494 } 3495 3496 /* 3497 * super.c 3498 */ 3499 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3500 void f2fs_inode_synced(struct inode *inode); 3501 int f2fs_dquot_initialize(struct inode *inode); 3502 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3503 int f2fs_quota_sync(struct super_block *sb, int type); 3504 loff_t max_file_blocks(struct inode *inode); 3505 void f2fs_quota_off_umount(struct super_block *sb); 3506 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3507 int f2fs_sync_fs(struct super_block *sb, int sync); 3508 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3509 3510 /* 3511 * hash.c 3512 */ 3513 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3514 3515 /* 3516 * node.c 3517 */ 3518 struct node_info; 3519 3520 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3521 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3522 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3523 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3524 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3525 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3526 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3527 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3528 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3529 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3530 struct node_info *ni, bool checkpoint_context); 3531 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3532 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3533 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3534 int f2fs_truncate_xattr_node(struct inode *inode); 3535 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3536 unsigned int seq_id); 3537 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3538 int f2fs_remove_inode_page(struct inode *inode); 3539 struct page *f2fs_new_inode_page(struct inode *inode); 3540 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3541 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3542 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3543 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3544 int f2fs_move_node_page(struct page *node_page, int gc_type); 3545 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3546 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3547 struct writeback_control *wbc, bool atomic, 3548 unsigned int *seq_id); 3549 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3550 struct writeback_control *wbc, 3551 bool do_balance, enum iostat_type io_type); 3552 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3553 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3554 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3555 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3556 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3557 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3558 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3559 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3560 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3561 unsigned int segno, struct f2fs_summary_block *sum); 3562 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3563 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3564 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3565 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3566 int __init f2fs_create_node_manager_caches(void); 3567 void f2fs_destroy_node_manager_caches(void); 3568 3569 /* 3570 * segment.c 3571 */ 3572 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3573 int f2fs_commit_atomic_write(struct inode *inode); 3574 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3575 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3576 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3577 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3578 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3579 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3580 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3581 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3582 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3583 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3584 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3585 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3586 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3587 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3588 struct cp_control *cpc); 3589 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3590 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3591 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3592 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3593 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3594 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3595 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3596 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3597 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3598 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3599 unsigned int *newseg, bool new_sec, int dir); 3600 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3601 unsigned int start, unsigned int end); 3602 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3603 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3604 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3605 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3606 struct cp_control *cpc); 3607 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3608 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3609 block_t blk_addr); 3610 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3611 enum iostat_type io_type); 3612 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3613 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3614 struct f2fs_io_info *fio); 3615 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3616 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3617 block_t old_blkaddr, block_t new_blkaddr, 3618 bool recover_curseg, bool recover_newaddr, 3619 bool from_gc); 3620 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3621 block_t old_addr, block_t new_addr, 3622 unsigned char version, bool recover_curseg, 3623 bool recover_newaddr); 3624 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3625 block_t old_blkaddr, block_t *new_blkaddr, 3626 struct f2fs_summary *sum, int type, 3627 struct f2fs_io_info *fio); 3628 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3629 block_t blkaddr, unsigned int blkcnt); 3630 void f2fs_wait_on_page_writeback(struct page *page, 3631 enum page_type type, bool ordered, bool locked); 3632 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3633 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3634 block_t len); 3635 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3636 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3637 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3638 unsigned int val, int alloc); 3639 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3640 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3641 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3642 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3643 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3644 int __init f2fs_create_segment_manager_caches(void); 3645 void f2fs_destroy_segment_manager_caches(void); 3646 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3647 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3648 unsigned int segno); 3649 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3650 unsigned int segno); 3651 3652 #define DEF_FRAGMENT_SIZE 4 3653 #define MIN_FRAGMENT_SIZE 1 3654 #define MAX_FRAGMENT_SIZE 512 3655 3656 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3657 { 3658 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3659 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3660 } 3661 3662 /* 3663 * checkpoint.c 3664 */ 3665 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3666 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3667 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3668 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3669 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3670 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3671 block_t blkaddr, int type); 3672 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3673 int type, bool sync); 3674 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3675 unsigned int ra_blocks); 3676 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3677 long nr_to_write, enum iostat_type io_type); 3678 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3679 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3680 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3681 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3682 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3683 unsigned int devidx, int type); 3684 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3685 unsigned int devidx, int type); 3686 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3687 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3688 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3689 void f2fs_add_orphan_inode(struct inode *inode); 3690 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3691 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3692 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3693 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3694 void f2fs_remove_dirty_inode(struct inode *inode); 3695 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3696 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3697 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3698 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3699 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3700 int __init f2fs_create_checkpoint_caches(void); 3701 void f2fs_destroy_checkpoint_caches(void); 3702 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3703 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3704 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3705 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3706 3707 /* 3708 * data.c 3709 */ 3710 int __init f2fs_init_bioset(void); 3711 void f2fs_destroy_bioset(void); 3712 int f2fs_init_bio_entry_cache(void); 3713 void f2fs_destroy_bio_entry_cache(void); 3714 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3715 struct bio *bio, enum page_type type); 3716 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3717 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3718 struct inode *inode, struct page *page, 3719 nid_t ino, enum page_type type); 3720 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3721 struct bio **bio, struct page *page); 3722 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3723 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3724 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3725 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3726 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3727 block_t blk_addr, sector_t *sector); 3728 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3729 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3730 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3731 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3732 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3733 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3734 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3735 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3736 int op_flags, bool for_write); 3737 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3738 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3739 bool for_write); 3740 struct page *f2fs_get_new_data_page(struct inode *inode, 3741 struct page *ipage, pgoff_t index, bool new_i_size); 3742 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3743 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3744 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3745 int create, int flag); 3746 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3747 u64 start, u64 len); 3748 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3749 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3750 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3751 int f2fs_write_single_data_page(struct page *page, int *submitted, 3752 struct bio **bio, sector_t *last_block, 3753 struct writeback_control *wbc, 3754 enum iostat_type io_type, 3755 int compr_blocks, bool allow_balance); 3756 void f2fs_write_failed(struct inode *inode, loff_t to); 3757 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3758 int f2fs_release_page(struct page *page, gfp_t wait); 3759 #ifdef CONFIG_MIGRATION 3760 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3761 struct page *page, enum migrate_mode mode); 3762 #endif 3763 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3764 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3765 int f2fs_init_post_read_processing(void); 3766 void f2fs_destroy_post_read_processing(void); 3767 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3768 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3769 extern const struct iomap_ops f2fs_iomap_ops; 3770 3771 /* 3772 * gc.c 3773 */ 3774 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3775 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3776 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3777 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3778 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3779 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3780 int __init f2fs_create_garbage_collection_cache(void); 3781 void f2fs_destroy_garbage_collection_cache(void); 3782 3783 /* 3784 * recovery.c 3785 */ 3786 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3787 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3788 int __init f2fs_create_recovery_cache(void); 3789 void f2fs_destroy_recovery_cache(void); 3790 3791 /* 3792 * debug.c 3793 */ 3794 #ifdef CONFIG_F2FS_STAT_FS 3795 struct f2fs_stat_info { 3796 struct list_head stat_list; 3797 struct f2fs_sb_info *sbi; 3798 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3799 int main_area_segs, main_area_sections, main_area_zones; 3800 unsigned long long hit_largest, hit_cached, hit_rbtree; 3801 unsigned long long hit_total, total_ext; 3802 int ext_tree, zombie_tree, ext_node; 3803 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3804 int ndirty_data, ndirty_qdata; 3805 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3806 int nats, dirty_nats, sits, dirty_sits; 3807 int free_nids, avail_nids, alloc_nids; 3808 int total_count, utilization; 3809 int bg_gc, nr_wb_cp_data, nr_wb_data; 3810 int nr_rd_data, nr_rd_node, nr_rd_meta; 3811 int nr_dio_read, nr_dio_write; 3812 unsigned int io_skip_bggc, other_skip_bggc; 3813 int nr_flushing, nr_flushed, flush_list_empty; 3814 int nr_discarding, nr_discarded; 3815 int nr_discard_cmd; 3816 unsigned int undiscard_blks; 3817 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3818 unsigned int cur_ckpt_time, peak_ckpt_time; 3819 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3820 int compr_inode; 3821 unsigned long long compr_blocks; 3822 int aw_cnt, max_aw_cnt; 3823 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3824 unsigned int bimodal, avg_vblocks; 3825 int util_free, util_valid, util_invalid; 3826 int rsvd_segs, overp_segs; 3827 int dirty_count, node_pages, meta_pages, compress_pages; 3828 int compress_page_hit; 3829 int prefree_count, call_count, cp_count, bg_cp_count; 3830 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3831 int bg_node_segs, bg_data_segs; 3832 int tot_blks, data_blks, node_blks; 3833 int bg_data_blks, bg_node_blks; 3834 int curseg[NR_CURSEG_TYPE]; 3835 int cursec[NR_CURSEG_TYPE]; 3836 int curzone[NR_CURSEG_TYPE]; 3837 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3838 unsigned int full_seg[NR_CURSEG_TYPE]; 3839 unsigned int valid_blks[NR_CURSEG_TYPE]; 3840 3841 unsigned int meta_count[META_MAX]; 3842 unsigned int segment_count[2]; 3843 unsigned int block_count[2]; 3844 unsigned int inplace_count; 3845 unsigned long long base_mem, cache_mem, page_mem; 3846 }; 3847 3848 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3849 { 3850 return (struct f2fs_stat_info *)sbi->stat_info; 3851 } 3852 3853 #define stat_inc_cp_count(si) ((si)->cp_count++) 3854 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3855 #define stat_inc_call_count(si) ((si)->call_count++) 3856 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3857 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3858 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3859 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3860 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3861 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3862 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3863 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3864 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3865 #define stat_inc_inline_xattr(inode) \ 3866 do { \ 3867 if (f2fs_has_inline_xattr(inode)) \ 3868 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3869 } while (0) 3870 #define stat_dec_inline_xattr(inode) \ 3871 do { \ 3872 if (f2fs_has_inline_xattr(inode)) \ 3873 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3874 } while (0) 3875 #define stat_inc_inline_inode(inode) \ 3876 do { \ 3877 if (f2fs_has_inline_data(inode)) \ 3878 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3879 } while (0) 3880 #define stat_dec_inline_inode(inode) \ 3881 do { \ 3882 if (f2fs_has_inline_data(inode)) \ 3883 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3884 } while (0) 3885 #define stat_inc_inline_dir(inode) \ 3886 do { \ 3887 if (f2fs_has_inline_dentry(inode)) \ 3888 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3889 } while (0) 3890 #define stat_dec_inline_dir(inode) \ 3891 do { \ 3892 if (f2fs_has_inline_dentry(inode)) \ 3893 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3894 } while (0) 3895 #define stat_inc_compr_inode(inode) \ 3896 do { \ 3897 if (f2fs_compressed_file(inode)) \ 3898 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3899 } while (0) 3900 #define stat_dec_compr_inode(inode) \ 3901 do { \ 3902 if (f2fs_compressed_file(inode)) \ 3903 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3904 } while (0) 3905 #define stat_add_compr_blocks(inode, blocks) \ 3906 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3907 #define stat_sub_compr_blocks(inode, blocks) \ 3908 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3909 #define stat_inc_meta_count(sbi, blkaddr) \ 3910 do { \ 3911 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3912 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3913 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3914 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3915 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3916 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3917 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3918 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3919 } while (0) 3920 #define stat_inc_seg_type(sbi, curseg) \ 3921 ((sbi)->segment_count[(curseg)->alloc_type]++) 3922 #define stat_inc_block_count(sbi, curseg) \ 3923 ((sbi)->block_count[(curseg)->alloc_type]++) 3924 #define stat_inc_inplace_blocks(sbi) \ 3925 (atomic_inc(&(sbi)->inplace_count)) 3926 #define stat_update_max_atomic_write(inode) \ 3927 do { \ 3928 int cur = F2FS_I_SB(inode)->atomic_files; \ 3929 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3930 if (cur > max) \ 3931 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3932 } while (0) 3933 #define stat_inc_seg_count(sbi, type, gc_type) \ 3934 do { \ 3935 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3936 si->tot_segs++; \ 3937 if ((type) == SUM_TYPE_DATA) { \ 3938 si->data_segs++; \ 3939 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3940 } else { \ 3941 si->node_segs++; \ 3942 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3943 } \ 3944 } while (0) 3945 3946 #define stat_inc_tot_blk_count(si, blks) \ 3947 ((si)->tot_blks += (blks)) 3948 3949 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3950 do { \ 3951 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3952 stat_inc_tot_blk_count(si, blks); \ 3953 si->data_blks += (blks); \ 3954 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3955 } while (0) 3956 3957 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3958 do { \ 3959 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3960 stat_inc_tot_blk_count(si, blks); \ 3961 si->node_blks += (blks); \ 3962 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3963 } while (0) 3964 3965 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3966 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3967 void __init f2fs_create_root_stats(void); 3968 void f2fs_destroy_root_stats(void); 3969 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3970 #else 3971 #define stat_inc_cp_count(si) do { } while (0) 3972 #define stat_inc_bg_cp_count(si) do { } while (0) 3973 #define stat_inc_call_count(si) do { } while (0) 3974 #define stat_inc_bggc_count(si) do { } while (0) 3975 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3976 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3977 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3978 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3979 #define stat_inc_total_hit(sbi) do { } while (0) 3980 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3981 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3982 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3983 #define stat_inc_inline_xattr(inode) do { } while (0) 3984 #define stat_dec_inline_xattr(inode) do { } while (0) 3985 #define stat_inc_inline_inode(inode) do { } while (0) 3986 #define stat_dec_inline_inode(inode) do { } while (0) 3987 #define stat_inc_inline_dir(inode) do { } while (0) 3988 #define stat_dec_inline_dir(inode) do { } while (0) 3989 #define stat_inc_compr_inode(inode) do { } while (0) 3990 #define stat_dec_compr_inode(inode) do { } while (0) 3991 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3992 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3993 #define stat_update_max_atomic_write(inode) do { } while (0) 3994 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3995 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3996 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3997 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3998 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3999 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4000 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4001 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4002 4003 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4004 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4005 static inline void __init f2fs_create_root_stats(void) { } 4006 static inline void f2fs_destroy_root_stats(void) { } 4007 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4008 #endif 4009 4010 extern const struct file_operations f2fs_dir_operations; 4011 extern const struct file_operations f2fs_file_operations; 4012 extern const struct inode_operations f2fs_file_inode_operations; 4013 extern const struct address_space_operations f2fs_dblock_aops; 4014 extern const struct address_space_operations f2fs_node_aops; 4015 extern const struct address_space_operations f2fs_meta_aops; 4016 extern const struct inode_operations f2fs_dir_inode_operations; 4017 extern const struct inode_operations f2fs_symlink_inode_operations; 4018 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4019 extern const struct inode_operations f2fs_special_inode_operations; 4020 extern struct kmem_cache *f2fs_inode_entry_slab; 4021 4022 /* 4023 * inline.c 4024 */ 4025 bool f2fs_may_inline_data(struct inode *inode); 4026 bool f2fs_sanity_check_inline_data(struct inode *inode); 4027 bool f2fs_may_inline_dentry(struct inode *inode); 4028 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4029 void f2fs_truncate_inline_inode(struct inode *inode, 4030 struct page *ipage, u64 from); 4031 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4032 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4033 int f2fs_convert_inline_inode(struct inode *inode); 4034 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4035 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4036 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4037 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4038 const struct f2fs_filename *fname, 4039 struct page **res_page); 4040 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4041 struct page *ipage); 4042 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4043 struct inode *inode, nid_t ino, umode_t mode); 4044 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4045 struct page *page, struct inode *dir, 4046 struct inode *inode); 4047 bool f2fs_empty_inline_dir(struct inode *dir); 4048 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4049 struct fscrypt_str *fstr); 4050 int f2fs_inline_data_fiemap(struct inode *inode, 4051 struct fiemap_extent_info *fieinfo, 4052 __u64 start, __u64 len); 4053 4054 /* 4055 * shrinker.c 4056 */ 4057 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4058 struct shrink_control *sc); 4059 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4060 struct shrink_control *sc); 4061 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4062 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4063 4064 /* 4065 * extent_cache.c 4066 */ 4067 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4068 struct rb_entry *cached_re, unsigned int ofs); 4069 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4070 struct rb_root_cached *root, 4071 struct rb_node **parent, 4072 unsigned long long key, bool *left_most); 4073 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4074 struct rb_root_cached *root, 4075 struct rb_node **parent, 4076 unsigned int ofs, bool *leftmost); 4077 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4078 struct rb_entry *cached_re, unsigned int ofs, 4079 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4080 struct rb_node ***insert_p, struct rb_node **insert_parent, 4081 bool force, bool *leftmost); 4082 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4083 struct rb_root_cached *root, bool check_key); 4084 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4085 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4086 void f2fs_drop_extent_tree(struct inode *inode); 4087 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4088 void f2fs_destroy_extent_tree(struct inode *inode); 4089 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4090 struct extent_info *ei); 4091 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4092 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4093 pgoff_t fofs, block_t blkaddr, unsigned int len); 4094 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4095 int __init f2fs_create_extent_cache(void); 4096 void f2fs_destroy_extent_cache(void); 4097 4098 /* 4099 * sysfs.c 4100 */ 4101 #define MIN_RA_MUL 2 4102 #define MAX_RA_MUL 256 4103 4104 int __init f2fs_init_sysfs(void); 4105 void f2fs_exit_sysfs(void); 4106 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4107 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4108 4109 /* verity.c */ 4110 extern const struct fsverity_operations f2fs_verityops; 4111 4112 /* 4113 * crypto support 4114 */ 4115 static inline bool f2fs_encrypted_file(struct inode *inode) 4116 { 4117 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4118 } 4119 4120 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4121 { 4122 #ifdef CONFIG_FS_ENCRYPTION 4123 file_set_encrypt(inode); 4124 f2fs_set_inode_flags(inode); 4125 #endif 4126 } 4127 4128 /* 4129 * Returns true if the reads of the inode's data need to undergo some 4130 * postprocessing step, like decryption or authenticity verification. 4131 */ 4132 static inline bool f2fs_post_read_required(struct inode *inode) 4133 { 4134 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4135 f2fs_compressed_file(inode); 4136 } 4137 4138 /* 4139 * compress.c 4140 */ 4141 #ifdef CONFIG_F2FS_FS_COMPRESSION 4142 bool f2fs_is_compressed_page(struct page *page); 4143 struct page *f2fs_compress_control_page(struct page *page); 4144 int f2fs_prepare_compress_overwrite(struct inode *inode, 4145 struct page **pagep, pgoff_t index, void **fsdata); 4146 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4147 pgoff_t index, unsigned copied); 4148 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4149 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4150 bool f2fs_is_compress_backend_ready(struct inode *inode); 4151 int f2fs_init_compress_mempool(void); 4152 void f2fs_destroy_compress_mempool(void); 4153 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4154 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4155 block_t blkaddr); 4156 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4157 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4158 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4159 int index, int nr_pages); 4160 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4161 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4162 int f2fs_write_multi_pages(struct compress_ctx *cc, 4163 int *submitted, 4164 struct writeback_control *wbc, 4165 enum iostat_type io_type); 4166 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4167 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4168 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4169 unsigned int c_len); 4170 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4171 unsigned nr_pages, sector_t *last_block_in_bio, 4172 bool is_readahead, bool for_write); 4173 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4174 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4175 void f2fs_put_page_dic(struct page *page); 4176 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4177 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4178 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4179 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4180 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4181 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4182 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4183 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4184 int __init f2fs_init_compress_cache(void); 4185 void f2fs_destroy_compress_cache(void); 4186 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4187 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4188 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4189 nid_t ino, block_t blkaddr); 4190 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4191 block_t blkaddr); 4192 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4193 #define inc_compr_inode_stat(inode) \ 4194 do { \ 4195 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4196 sbi->compr_new_inode++; \ 4197 } while (0) 4198 #define add_compr_block_stat(inode, blocks) \ 4199 do { \ 4200 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4201 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4202 sbi->compr_written_block += blocks; \ 4203 sbi->compr_saved_block += diff; \ 4204 } while (0) 4205 #else 4206 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4207 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4208 { 4209 if (!f2fs_compressed_file(inode)) 4210 return true; 4211 /* not support compression */ 4212 return false; 4213 } 4214 static inline struct page *f2fs_compress_control_page(struct page *page) 4215 { 4216 WARN_ON_ONCE(1); 4217 return ERR_PTR(-EINVAL); 4218 } 4219 static inline int f2fs_init_compress_mempool(void) { return 0; } 4220 static inline void f2fs_destroy_compress_mempool(void) { } 4221 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4222 static inline void f2fs_end_read_compressed_page(struct page *page, 4223 bool failed, block_t blkaddr) 4224 { 4225 WARN_ON_ONCE(1); 4226 } 4227 static inline void f2fs_put_page_dic(struct page *page) 4228 { 4229 WARN_ON_ONCE(1); 4230 } 4231 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4232 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4233 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4234 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4235 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4236 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4237 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4238 static inline void f2fs_destroy_compress_cache(void) { } 4239 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4240 block_t blkaddr) { } 4241 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4242 struct page *page, nid_t ino, block_t blkaddr) { } 4243 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4244 struct page *page, block_t blkaddr) { return false; } 4245 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4246 nid_t ino) { } 4247 #define inc_compr_inode_stat(inode) do { } while (0) 4248 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4249 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4250 unsigned int c_len) { } 4251 #endif 4252 4253 static inline void set_compress_context(struct inode *inode) 4254 { 4255 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4256 4257 F2FS_I(inode)->i_compress_algorithm = 4258 F2FS_OPTION(sbi).compress_algorithm; 4259 F2FS_I(inode)->i_log_cluster_size = 4260 F2FS_OPTION(sbi).compress_log_size; 4261 F2FS_I(inode)->i_compress_flag = 4262 F2FS_OPTION(sbi).compress_chksum ? 4263 1 << COMPRESS_CHKSUM : 0; 4264 F2FS_I(inode)->i_cluster_size = 4265 1 << F2FS_I(inode)->i_log_cluster_size; 4266 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4267 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4268 F2FS_OPTION(sbi).compress_level) 4269 F2FS_I(inode)->i_compress_flag |= 4270 F2FS_OPTION(sbi).compress_level << 4271 COMPRESS_LEVEL_OFFSET; 4272 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4273 set_inode_flag(inode, FI_COMPRESSED_FILE); 4274 stat_inc_compr_inode(inode); 4275 inc_compr_inode_stat(inode); 4276 f2fs_mark_inode_dirty_sync(inode, true); 4277 } 4278 4279 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4280 { 4281 struct f2fs_inode_info *fi = F2FS_I(inode); 4282 4283 if (!f2fs_compressed_file(inode)) 4284 return true; 4285 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4286 return false; 4287 4288 fi->i_flags &= ~F2FS_COMPR_FL; 4289 stat_dec_compr_inode(inode); 4290 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4291 f2fs_mark_inode_dirty_sync(inode, true); 4292 return true; 4293 } 4294 4295 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4296 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4297 { \ 4298 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4299 } 4300 4301 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4302 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4303 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4304 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4305 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4306 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4307 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4308 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4309 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4310 F2FS_FEATURE_FUNCS(verity, VERITY); 4311 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4312 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4313 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4314 F2FS_FEATURE_FUNCS(readonly, RO); 4315 4316 static inline bool f2fs_may_extent_tree(struct inode *inode) 4317 { 4318 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4319 4320 if (!test_opt(sbi, EXTENT_CACHE) || 4321 is_inode_flag_set(inode, FI_NO_EXTENT) || 4322 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4323 !f2fs_sb_has_readonly(sbi))) 4324 return false; 4325 4326 /* 4327 * for recovered files during mount do not create extents 4328 * if shrinker is not registered. 4329 */ 4330 if (list_empty(&sbi->s_list)) 4331 return false; 4332 4333 return S_ISREG(inode->i_mode); 4334 } 4335 4336 #ifdef CONFIG_BLK_DEV_ZONED 4337 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4338 block_t blkaddr) 4339 { 4340 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4341 4342 return test_bit(zno, FDEV(devi).blkz_seq); 4343 } 4344 #endif 4345 4346 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4347 { 4348 return f2fs_sb_has_blkzoned(sbi); 4349 } 4350 4351 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4352 { 4353 return blk_queue_discard(bdev_get_queue(bdev)) || 4354 bdev_is_zoned(bdev); 4355 } 4356 4357 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4358 { 4359 int i; 4360 4361 if (!f2fs_is_multi_device(sbi)) 4362 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4363 4364 for (i = 0; i < sbi->s_ndevs; i++) 4365 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4366 return true; 4367 return false; 4368 } 4369 4370 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4371 { 4372 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4373 f2fs_hw_should_discard(sbi); 4374 } 4375 4376 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4377 { 4378 int i; 4379 4380 if (!f2fs_is_multi_device(sbi)) 4381 return bdev_read_only(sbi->sb->s_bdev); 4382 4383 for (i = 0; i < sbi->s_ndevs; i++) 4384 if (bdev_read_only(FDEV(i).bdev)) 4385 return true; 4386 return false; 4387 } 4388 4389 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4390 { 4391 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4392 } 4393 4394 static inline bool f2fs_may_compress(struct inode *inode) 4395 { 4396 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4397 f2fs_is_atomic_file(inode)) 4398 return false; 4399 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4400 } 4401 4402 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4403 u64 blocks, bool add) 4404 { 4405 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4406 struct f2fs_inode_info *fi = F2FS_I(inode); 4407 4408 /* don't update i_compr_blocks if saved blocks were released */ 4409 if (!add && !atomic_read(&fi->i_compr_blocks)) 4410 return; 4411 4412 if (add) { 4413 atomic_add(diff, &fi->i_compr_blocks); 4414 stat_add_compr_blocks(inode, diff); 4415 } else { 4416 atomic_sub(diff, &fi->i_compr_blocks); 4417 stat_sub_compr_blocks(inode, diff); 4418 } 4419 f2fs_mark_inode_dirty_sync(inode, true); 4420 } 4421 4422 static inline int block_unaligned_IO(struct inode *inode, 4423 struct kiocb *iocb, struct iov_iter *iter) 4424 { 4425 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4426 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4427 loff_t offset = iocb->ki_pos; 4428 unsigned long align = offset | iov_iter_alignment(iter); 4429 4430 return align & blocksize_mask; 4431 } 4432 4433 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4434 int flag) 4435 { 4436 if (!f2fs_is_multi_device(sbi)) 4437 return false; 4438 if (flag != F2FS_GET_BLOCK_DIO) 4439 return false; 4440 return sbi->aligned_blksize; 4441 } 4442 4443 static inline bool f2fs_force_buffered_io(struct inode *inode, 4444 struct kiocb *iocb, struct iov_iter *iter) 4445 { 4446 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4447 int rw = iov_iter_rw(iter); 4448 4449 if (!fscrypt_dio_supported(iocb, iter)) 4450 return true; 4451 if (fsverity_active(inode)) 4452 return true; 4453 if (f2fs_compressed_file(inode)) 4454 return true; 4455 4456 /* disallow direct IO if any of devices has unaligned blksize */ 4457 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4458 return true; 4459 /* 4460 * for blkzoned device, fallback direct IO to buffered IO, so 4461 * all IOs can be serialized by log-structured write. 4462 */ 4463 if (f2fs_sb_has_blkzoned(sbi)) 4464 return true; 4465 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4466 if (block_unaligned_IO(inode, iocb, iter)) 4467 return true; 4468 if (F2FS_IO_ALIGNED(sbi)) 4469 return true; 4470 } 4471 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4472 return true; 4473 4474 return false; 4475 } 4476 4477 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4478 { 4479 return fsverity_active(inode) && 4480 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4481 } 4482 4483 #ifdef CONFIG_F2FS_FAULT_INJECTION 4484 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4485 unsigned int type); 4486 #else 4487 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4488 #endif 4489 4490 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4491 { 4492 #ifdef CONFIG_QUOTA 4493 if (f2fs_sb_has_quota_ino(sbi)) 4494 return true; 4495 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4496 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4497 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4498 return true; 4499 #endif 4500 return false; 4501 } 4502 4503 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4504 { 4505 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4506 } 4507 4508 static inline void f2fs_io_schedule_timeout(long timeout) 4509 { 4510 set_current_state(TASK_UNINTERRUPTIBLE); 4511 io_schedule_timeout(timeout); 4512 } 4513 4514 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4515 enum page_type type) 4516 { 4517 if (unlikely(f2fs_cp_error(sbi))) 4518 return; 4519 4520 if (ofs == sbi->page_eio_ofs[type]) { 4521 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4522 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4523 } else { 4524 sbi->page_eio_ofs[type] = ofs; 4525 sbi->page_eio_cnt[type] = 0; 4526 } 4527 } 4528 4529 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4530 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4531 4532 #endif /* _LINUX_F2FS_H */ 4533